LONGITUDINAL PATTERNS IN STREAM CHANNEL GEOMORPHOLOGY …€¦ · LONGITUDINAL PATTERNS IN STREAM...

LONGITUDINAL PATTERNS IN STREAM CHANNEL GEOMORPHOLOGY AND

AQUATIC HABITAT IN THE LUQUILLO MOUNTAINS OF PUERTO RICO

Andrew Stephen Pike

A DISSERTATION

in

Earth and Environmental Science

Presented to the faculties of the University of Pennsylvania in partial fulfillment of the

requirements for the degree of Doctor of Philosophy

2008

___________________________________

Supervisor of Dissertation – Dr. Frederick N. Scatena

___________________________________

Graduate Group Chairperson – Dr. Gomaa Omar

ACKNOWLEDGEMENTS

This dissertation would not have been possible with the help and support of

numerous mentors, colleagues, and friends, all of whom deserve recognition here. First

and foremost, I would like to thank my advisor, Fred Scatena. Although his expectations

were high and his demands were challenging, his perpetual guidance from the field to the

office has been unparalleled. With his iron drive to help me succeed, and his deep breadth

of knowledge of the natural world, it has truly been an honor to have studied under Fred.

Second, I would like to thank each member of my graduate committee. I extend special

thanks to Art Johnson, for first launching me onto the scientific research course, and for

his continued mentorship throughout the years; I most certainly would not have made it

this far without him. I thank Todd Crowl for arranging for me to spend an enjoyable and

beneficial semester in his laboratory at Utah State University. I thank Dana Tomlin for

his lucid intellect and always enlightening discussions. Lastly, I am deeply appreciative

of Gomaa Omar not only for his support in handling logistical issues, but also for sharing

his unique talent of bringing out the buried spiritual essence in scientific work.

The field work required to survey tropical mountain rainforest streams was

inherently enjoyable, but it was not without its challenges. Yet I had the pleasure of

working with people who knew how to get a difficult job done, and have a good time

doing it. The assistance from the following people made four summers of stream surveys

possible, and all the more pleasant: Erik Drew, Tamara Heartsill-Scalley, Emmanuelle

Humblet, Kris Johnson, Kunal Mandal, José Marcial, Amanda Moyer, Pablo Piña, and of

course, Jennifer Schiffner.

ii

Housing and logistical support while in the field was made possible by the El

Verde Field Station (University of Puerto Rico), the Sabana Research Center (United

States Forest Service), and the International Institute for Tropical Forestry (IITF).

My colleagues on the Biocomplexity project were critical in the formulation of

this dissertation. I especially thank Katie Hein for her hard work collecting ecological

data, her insight throughout the project, and for two excellent field seasons that made this

collaborative project work. I also thank Drs. Ellen Wohl, Jorge Ramirez, Alan Covich,

Felipe Blanco, and all others involved with the project for their intellectual contributions.

I would like to thank numerous people around the department of Earth and

Environmental Science at Penn. A special thanks for Karen Taylor for her support in all

matters of life. Also, thanks to all the fellow grad students for their camaraderie, and to

all the faculty who I’ve studied with and learned from over the past years.

Funding for this dissertation was provided from the National Science Foundation

Biocomplexity Grant (NSF #030414) —Rivers, Roads, and People: Complex Interactions

of Overlapping Networks in Watersheds. Additional support was provided by the

University of Pennsylvania, and the Long-Term Ecological Research (LTER) Network.

iii

ABSTRACT

LONGITUDINAL PATTERNS IN STREAM CHANNEL GEOMORPHOLOGY AND

AQUATIC HABITAT IN THE LUQUILLO MOUNTAINS OF PUERTO RICO

Andrew S. Pike

Frederick N. Scatena

The hydrologic, geomorphic, and ecological dynamics of tropical montane

streams are poorly understood in comparison to many temperate and/or alluvial rivers.

Yet as the threat to tropical freshwater environments increases, information on the

dynamics of relatively pristine streams is important for understanding landscape

evolution, managing and conserving natural resources, and implementing stream

restoration. This dissertation characterizes the geomorphology and hydrology of five

adjacent watersheds draining the Luquillo Experimental Forest (LEF) in northeastern

Puerto Rico, and discusses implications on aquatic habitat. I performed several

interrelated studies, including: 1) formulating a geographic information systems (GIS)

framework to estimate hydrologic parameters from topographic information and

hydrologic records, 2) developing a method to determine active stream channel

boundaries (“bankfull” stage) that allows for comparison of channel geometry on the

basis of flow-frequency, 3) decoupling the relative influences of lithologic and hydraulic

controls on channel morphology using an extensive field-based stream survey and

analysis of stream profiles, channel geometry, and sediment dynamics, 4) linking

network- and pool-scale geofluvial dynamics to the abundance of migratory fish and

iv

shrimp through a collaborative analysis combining geomorphic surveys and aquatic

faunal sampling. This research indicates that these streams have some properties

resembling both temperate montane and alluvial rivers. Similar to low-gradient rivers

where floodplains mark channel boundaries, the active channel stage in these streams is

defined by the incipient presence of woody vegetation and soil development. Systematic

basin-scale geomorphic patterns are well-developed despite apparent non-fluvial and

lithologic control on local channel morphology. This implies that strong fluvial forces are

sufficient to override channel boundary resistance; a feature common in self-forming

“threshold” alluvial channels. Migratory aquatic fauna abundances are influenced by a

variety of geomorphic factors such as barrier waterfalls and suitable headwater habitat,

and are consequently highly variable and patchy. These results stand in contrast to the

notion that aquatic communities mirror systematic geomorphic gradients, but rather

acknowledges the influences of multiscale geomorphic processes. Ultimately, this

research provides baseline information on physical and biological processes in relatively

unaltered tropical streams and can be used to inform further studies that document human

interactions with stream networks.

v

TABLE OF CONTENTS ACKNOWLEDGEMENTS ii

ABSTRACT iv

LIST OF TABLES x

LIST OF FIGURES xii

CHAPTER 1: General Introduction

Introduction 1

Chapter Outlines 6

References 9

CHAPTER 2: Application of Digital Terrain Analysis to Model Surface

Water Flow in the Luquillo Mountains of Northeastern Puerto Rico

Abstract 15

Introduction 16

Study Area 17

Digital Elevation Model (DEM) Construction 18

Stream Network Extraction 22

Rainfall, Runoff, and Discharge 24

Conclusion 27

Acknowledgements 27

References 28

CHAPTER 3: Defining a Bankfull Analog for Tropical Montane Streams

Using Riparian Features

Abstract 31

vi

Introduction 33

Study Area 38

Northeastern Puerto Rico 38

Regional Stream Gages 41

Riparian Vegetation 42

Methods 50

Field Surveys 50

Estimation of Flow-Frequency 51

Multivariate Regression Trees 55

Effective Discharge 56

Results 60

Riparian Vegetation 60

Bankfull Discharge and Effective Discharge 61

Multivariate Regression Trees 61

Discussion 70

Riparian Features 70

Bankfull and Effective Discharge 73

Applicability to other stream systems 76

Conclusion 77

Acknowledgements 78

References 79

CHAPTER 4: Longitudinal Patterns in Stream Channel Geomorphology in

the Tropical Mountain Streams of the Luquillo Mountains, Puerto Rico

vii

Abstract 88

Introduction 89

Study Area 97

Methods 105

Results 109

Longitudinal Profiles 109

Hydraulic Geometry 113

Grain Size 116

Stream Power 127

Discussion 130

Conclusion 136

Acknowledgements 137

References 138

CHAPTER 5: Multiscale Linkages Between Geomorphology and Aquatic

Habitat in a Tropical Montane Stream Network, Puerto Rico

Abstract 149

Introduction 151

Study Area 156

Stream Community 161

Methods 164

Field Methods 164

Pool Length and Spacing 168

Statistical Analyses 168

viii

Principal Components Analysis 168

Non-Parametric Multiplicative Regression 169

Stepwise Multiple Linear Regression 171

Results 171

Species Distribution in Geomorphic Space 171

Longitudinal Trends of Species 183

Influence of Reach and Pool-Scale Geomorphology 192

Pool Size and Spacing 195

Discussion 195

Landscape Scale Patterns 195

Reach and Pool-Scale Patterns 201

Conclusion 204

References 205

CHAPTER 6: Conclusions and Future Research

Summary and Conclusions 216

Future Research 218

APPENDIX 222

Site Information 227

Baseflow Channel Geometry 234

Active Channel Geometry 241

Grain Size and Pebble Count Data 248

Additional Biocomplexity Pool Information 255

INDEX 259

ix

LIST OF TABLES Table 3.1 Physiographic site information for the selected study gages.

Table 3.2 Riparian features (vegetation, substrate, and soil

characteristics) that were recorded at each survey point.

Table 3.3 For each study USGS gage, the time span of the discharge

record, flow parameters, and at station width and depth hydraulic

geometry coefficients and exponents.

Table 3.4 The median height above water table, unit discharge, flow

frequency, and recurrence interval of zones defined by the multivariate

regression tree analysis.

Table 3.5 The discharge, unit discharge, flow frequency, and

recurrence interval associated with the bankfull stage, the effective

discharge, and the bankfull analog at each study stream gage.

Table 4.1 Downstream hydraulic geometry coefficients and exponents

divided by watershed.

Table 5.1 Species identified during trapping and electrofish sampling.

Table 5.2 Principal component eigenvalues for 58 geomorphic

variables, using data from 113 pools throughout the study basins.

Table 5.3 Principal component eigenvalues for 20 geomorphic

variables measured for pools in the Quebrada Prieta.

Table 5.4 Landscape scale variables used to predict species

presence/absence using a NPMR 2-parameter model.

44

52

54

68

69

115

166

173-174

181

184

x

Table 5.5 Internal validation statistics for NPMR models of species

presence/absence.

Table 5.6 Stepwise Multiple Linear Regression model output and

goodness of fit showing the relative influence of geomorphic variables

on predicting decapod abundance at the reach-scale.

Table 5.7 Stepwise Multiple Linear Regression model output and

goodness of fit showing the relative influence of geomorphic variables

on predicting decapod abundance in the Quebrada Prieta.

190-191

193

194

xi

LIST OF FIGURES Figure 2.1 The process of extracting a stream network from contour

data.

Figure 2.2 Comparison of the USGS stream network to the DEM

generated stream network at 6ha drainage area threshold for the Río

Mameyes.

Figure 2.3 Spatial distribution of mean annual rainfall, mean annual

runoff, and mean annual discharge within the Río Mameyes drainage

basin according to elevation-based regression equations.

Figure 3.1 Location map of the selected study USGS gages in and

around the Luquillo Experimental Forest in Northeastern Puerto Rico.

Figure 3.2 Photographs of study USGS gages.

Figure 3.3 Illustration of the vertical zonation of vegetation types at a

cross section at Río Mameyes near Sabana, alongside a hydrograph

representative of the flow regime and flood disturbance.

Figure 3.4 The flow duration, sediment transport, and relative

effectiveness curves using data from the Río Mameyes near Sabana

gage, and the corresponding discharge at which different vegetation

types occur.

Figure 3.5 Box plots of the flow frequency for surveyed data points at

alluvial sites that have been partitioned into clusters based on

vegetation, substrate, and soil characteristics using a multivariate

regression tree technique.

20

23

26

43

45-46

48

59

63-64

xii

Figure 3.6 Box plots of the flow frequency for surveyed data points at

mid-elevation and steepland sites that have been partitioned into

clusters based on vegetation, substrate, and soil characteristics using a

multivariate regression tree technique.

Figure 4.1 Location map of the 238 surveyed reaches in the study

watersheds and the regional topography, geology, and land cover in

northeastern Puerto Rico.

Figure 4.2 Longitudinal profiles of the main stem of each river

highlighting the relationship between local profile shape and lithology.

Figure 4.3 Downstream hydraulic geometry relationships between

active channel discharge, width, depth, and velocity using data from

all surveyed reaches.

Figure 4.4 Upstream views of typical reaches throughout the basins.

Figure 4.5 Grain size distributions for all measured clasts, grouped by

lithology.

Figure 4.6 The percentage of megaboulders in the channel as a

function of the adjacent hillslope steepness.

Figure 4.7 Relationship between active channel discharge and

dimensionless shear stresses and critical shear stresses.

Figure 4.8 Relationship between median grain size, drainage area, and

slope.

65-66

98-99

110-111

114

117-118

120

121

123-124

126

xiii

Figure 4.9 Downstream changes in elevation, drainage area, median

grain size, stream power, and the ratio of stream power to coarse grain

size along the main stem profile of each river.

Figure 5.1 Location map of 33 sample sites located in the Rio

Mameyes, Rio Espiritu Santo, and Rio Fajardo.

Figure 5.2 Plot of pools in geomorphic space, with species presence

indicated.

Figure 5.3 Plot of pools in geomorphic space, with decapod relative

abundance and proportional abundance indicated.

Figure 5.4 Plot of Quebrada Prieta pools in geomorphic space, with

decapod relative abundance and proportional abundance indicated.

Figure 5.5 An example of a non-parametric multiplicative regression

response curve applied to map the distribution of species.

Figure 5.6 Longitudinal profiles of estimated probability of

occurrence for each sampled species.

Figure 5.7 Map showing the location of 7 stream segments where

pools length and spacing were measured, as well as relationships

between drainage area, pool length, and pool spacing.

128-129

157-158

175-176

178-179

182

185-186

187-188

196-197

xiv

CHAPTER 1

GENERAL INTRODUCTION

INTRODUCTION

A common theme in river research and aquatic and riparian ecology is the role of

self-organizing principles that give rise to systematic basin-scale patterns. Decades of

research have identified several similar morphologic features in alluvial rivers worldwide

that are formed in response to the flow regime and sediment transport capacity. These

include a floodplain defining the bankfull discharge that occurs with a recurrence interval

of approximate 1-3 years (Wolman and Miller 1960), well-graded concave upward

longitudinal profiles (Hack 1957), well-developed downstream hydraulic geometry

(Leopold and Maddock 1953), and systematic changes in grain size (Leopold et al. 1964).

These features are a result of the tendency of rivers to adjust to the frequency and

magnitude of flows, and to create gradients that most effectively route water and

sediment through the network. Furthermore, some theory in aquatic ecology is based on

the observation that aquatic habitats, populations, and communities often change

systematically downstream in response to imposed gradients in channel morphology and

consequent changes in canopy openness, light, substrate size, and stream flow (Vannote

et al. 1982). A careful investigation of these morphologic and ecological patterns can

yield insight into landscape evolution, inform management of natural resources, and

provide guidelines for stream restoration.

However, these established linkages between geomorphic patterns and fluvial

processes may not hold in mountainous and bedrock streams. Recent research on the

morphology of mountain streams demonstrates their complexity and implicates several

1

different controls on their morphology, including: tectonic and structural influences

(Whipple 2004, VanLaningham et al. 2006), bedrock (Snyder et al. 2003), storm pulses

(Gupta 1988), non-fluvial processes such as landslides/debris flows (Brummer and

Montgomery 2003, Stock and Dietrich 2006) and glaciers (Wohl et al. 2004). Other

studies have demonstrated the characteristic morphology of mountain streams (Grant et

al. 1990, Montgomery and Buffington 1997, Wohl and Merritt 2001), the lack of

consistent bankfull indicators (Radecki-Pawelick 2002), their complicated hydraulic

geometry (Wohl 2004), the complexity of sediment transport (Blizard and Wohl 1998,

Lenzi et al. 2006, Torizzo and Pitlick 2004), as well as the distribution of sediment within

mountain drainages (Pizzuto 1995, Golden and Spring 2006).

Of all mountain streams in the world, drainages in humid tropical montane areas

are among the most extreme fluvial environments (Gupta 1988). The high rates of erosion

and dramatically dissected landscapes prevalent in the world’s tropical mountainous

regions are testament to the power of these rivers. In comparison to the many temperate

montane and/or alluvial rivers and streams that have been studied worldwide, tropical

montane streams have several unique characteristics that may structure their morphology:

A combination of steep slopes, high mean annual rainfall, and intense tropical storms

generate an energetic and powerful flow regime (Gupta 1995). The absence of past and

present glaciation excludes glacial landforms, such as U-shaped valleys and coarse

moraine deposits, that are prevalent in some temperate montane basins. Relatively high

rates of chemical and physical weathering rapidly denude tropical landscapes and may

affect rates of channel-sediment diminution and patterns of downstream fining (Brown et

al. 1995, White et al. 1998, Rengers and Wohl 2007). Frequent landslides triggered by

2

heavy rains introduce pulses of coarse sediment to the channels and strongly link fluvial

and colluvial forces (Larsen et al. 1999). Large woody debris that is common in

temperate streams is rapidly decomposed in the tropics, despite frequent inputs from

surrounding mature forests and hurricanes (Covich and Crowl 1990). Periodic high-

magnitude floods associated with hurricanes and other tropical disturbances effectively

rework boulder channels (Gupta 1975, Scatena and Larsen 1991). Yet the channel

morphology that is sculpted by fluvial and non-fluvial processes in tropical montane

environments is generally unknown. The relatively few studies that have specifically

addressed the morphology of tropical mountain streams have shown that the conflicting

lithologic and hydraulic controls complicate the development of expected downstream

morphologic patterns (Lewis 1969, Ahmad et al. 1993, Wohl 2005).

Montane streams are also an integral part of the ecological web of many humid

tropical islands. Biodiversity in tropical island streams is generally low in comparison to

continental streams due to physical isolation (Covich 1988, Smith et al. 2003). The

species that do inhabit tropical island streams have distinct habitat preferences and most

fish and decapods in these streams have a diadromous life cycle, requiring direct linkages

between freshwater and salt water to breed and feed. Consequently, the geomorphology

of the river channel, imposed by the interplay between fluvial and tectonic processes, is

critical to understanding habitat formation and the consequent distribution of aquatic

fauna throughout the stream network. Several theories have been proposed to describe the

linkages between geomorphology and aquatic organisms in streams, emphasizing the

roles of systematic longitudinal gradients (Vannote et al. 1982), patchiness and

heterogeneity (Pringle et al. 1988, Townsend 1989), hydraulics (Statzner and Higler

3

1986), geomorphic disturbance (Montgomery 1999), multiscale habitat formation (Wu

and Loucks 1995, Poole 2002), and river network structure (Benda et al. 2004). Although

these theories apply to many stream systems worldwide, it is not known whether their

predictions of community distributions hold in tropical island streams where migratory

aquatic fauna interact with short, steep-gradient, frequently-flooded, bedrock and

boulder-lined channels that are punctuated by waterfalls.

As the threat to tropical freshwater streams increases through dam-building and

landuse changes (Holmquist et al. 1997, Pringle and Scatena 1998, Gupta and Ahmad

1999, Pringle et al. 2000, March et al. 2003, Anderson-Olivas et al. 2006, Greathouse et

al. 2006), it is critical to understand the dynamics of geomorphic processes and the

consequent response on stream channel morphology and aquatic biota in relatively

unaltered streams. Toward this end, this dissertation focuses on such geomorphic and

ecological patterns and processes in tropical mountain streams.

Specifically, I investigate the longitudinal variations in channel morphology and

aquatic biota in bedrock and alluvial streams draining the Luquillo Mountains of

northeastern Puerto Rico; a relatively old, subtropical island landscape that is subject to a

high frequency of atmospheric and hillslope disturbances, as well as steep elevational and

climatic gradients (Scatena 1995). Five adjacent watersheds are considered in this

research: Río Blanco, Río Espiritu Santo, Río Fajardo, Río Mameyes, and Río Sabana.

The watersheds are similar physiographically, but are different in size, geology, and land

cover. A combination of high-quality geographic information systems (GIS) coverages,

long-term hydrologic records, and extensive field-based survey data allows for high-

resolution spatial analyses of geomorphic patterns and processes.

4

This dissertation is also an integral part of a National Science Foundation-

supported Biocomplexity study to understand complex interactions between roads, rivers,

and people (NSF #030414). This Biocomplexity project is a multidisciplinary

collaboration addressing linkages between fluvial geomorphology, aquatic biology, and

human recreation in Luquillo streams. Researchers on this project have conducted

specialized yet complementary studies at common field sites (road-river crossings) with

the intent of generating holistic conclusions on how the structure of both river and road

networks facilitates flows of both aquatic organisms and human recreation. In this light,

this research provides both a conceptual and quantitative understanding of the hydrology,

geomorphology, and physical processes of the stream network of northeastern Puerto

Rico that can be used as a baseline for complementary studies addressing aquatic biota

and human influences.

The chapters presented in this dissertation are organized as journal articles. This

dissertation research is organized into four interrelated studies with the following goals:

1) Develop a Geographic Information Systems (GIS) template to quantify topographic

features, map the stream network, and estimate hydrologic parameters across the

landscape, 2) Define an active channel boundary in tropical montane streams, analogous

to bankfull in alluvial streams, based on characteristics of the riparian zone to be used as

a basis to compare channel geometry, 3) Decouple lithologic and hydraulic controls on

channel morphology through analysis of downstream patterns in channel form, and 4)

Determine the influence of local and network-scale geomorphology on the distribution of

migratory aquatic fauna.

5

CHAPTER OUTLINES

Chapter 2 develops a Geographic Information Systems (GIS) template to analyze

topographic features, map the stream network, and estimate hydrologic parameters

(rainfall, runoff, and discharge) in the region. This chapter also generates independent

variables (drainage area, slope, discharge) to be used in subsequent chapters. The process

involves (1) creation of a hydrologically correct Digital Elevation Model (DEM), (2)

defining the stream network using flow accumulation models using a drainage area

threshold, and (3) developing relationships to estimate mean annual rainfall, runoff, and

discharge from topographic factors. The result is a spatial framework to describe

hydrological variables for 10 m × 10 m cells within the stream networks.

Chapter 3 develops a method to define an active channel boundary for tropical

montane stream channels that is analogous to the bankfull stage in alluvial rivers. The

bankfull stage of a river channel is an important geomorphologic boundary, as it marks

the channel-forming discharge that occurs at a consistent flow-frequency throughout the

stream network, yet it is rarely identifiable in steep mountain channels that lack

floodplains. However, other features along the channel margins may be used to mark

bankfull stage. For example, assuming that there is a functional relationship between the

frequency of flooding and the establishment of vegetation, the characteristics of near-

channel vegetation and associated substrate and soil development may indicate

hydrogeomorphic conditions. By correlating the relative elevation of different types of

vegetation and other riparian features with the known magnitude and frequency of flows

that inundate that elevation, we can determine whether or not these features are a reliable

indicator of the channel-forming discharge. This chapter quantifies such relationships

6

between flow-frequency and riparian features at a series of long-term stream gages to

find the characteristics of vegetation, soil, and substrate that mark the active channel and

correspond to the bankfull stage in adjacent alluvial channels. The active channel

boundary determined here is used in the next chapter as the basis to compare cross-

sectional channel geometry at a known flow-frequency throughout the study basins.

Chapter 4 investigates potential lithologic and hydraulic controls on stream

channel morphology. Using a comprehensive dataset of surveyed stream cross-sections

capturing reaches along the entire longitudinal gradient from the headwaters to the

estuary, I analyze channel profiles and subsequent longitudinal changes in channel cross-

sectional geometry, grain size, stream power, and shear stresses. Such analyses of

downstream geomorphic trends yield insight into the evolution and self-organization of

the stream network. If local lithologic factors such as multiple rock types, resistant

channel boundaries, and coarse sediment delivery from landslides dominate the form of

the river, the river profile should be segmented, display poorly developed hydraulic

geometry, and a have seemingly random pattern of grain sizes. Conversely, if the high

unit discharge and associated stream power of the energetic tropical flow regime are

sufficient to overcome lithologic resistance and mobilize coarse sediment, the

downstream changes in channel morphology should have systematic trends similar to

those in many alluvial rivers.

Chapter 5 links the geomorphology of the stream network and local-scale physical

habitat with the distribution of migratory fish and decapods. Extensive geomorphic

surveys of pools, complemented with intensive biological sampling at the same sites,

were used to correlate geomorphic features of the stream channel to the presence and

7

abundances of fish and decapods. At the landscape scale, spatial patterns of the presence

and absence of all major macrofauna were analyzed. We expect that the steepness of the

stream channel may hinder the upstream migration of some species so that they display

distinct yet discontinuous longitudinal patterns. At the reach-scale and pool-scale, where

primarily decapod species are known to be present, we correlate local geomorphic

features of the pool with abundance. We expect that the decapods seek out optimal

habitat, but that the natural variability in the geomorphic environment at these scales

gives rise to patchiness in their abundances.

Lastly, Chapter 6 summarizes the key results of the previous chapters, synthesizes

the relationships between hydrologic, geomorphic, and ecologic flows, and discusses

further avenues of research based on additional questions that this research poses.

8

REFERENCES Ahmad R, Scatena FN, Gupta A. 1993. Morphology and sedimentation in Caribbean

montane streams: examples from Jamaica and Puerto Rico. Sedimentary Geology

85: 157-169.

Anderson-Olivas EA, Freeman MC, Pringle CM. 2006. Ecological consequences of

hydropower development in Central America: Impacts of small dams and water

diversion on neotropical stream fish assemblages. River Research and

Applications 22: 397-411.

Benda L, Poff NL, Miller D, Dunne T, Reeves G, Pess G, Pollock M. 2004. The Network

Dynamics Hypothesis: How channel networks structure riverine habitats.

BioScience 54: 413-427.

Blizard CR, Wohl EE. 1998. Relationships between hydraulic variables and bedload

transport in a subalpine channel, Colorado Rocky Mountains, U.S.A.

Geomorphology 22: 359-371.

Brown ET, Stallard RF, Larsen MC, Raisbeck GN, Yiou F. 1995. Denudation rates

determined from the accumulation of in situ-produced 10Be in the Luquillo

Experimental Forest, Puerto Rico. Earth and Planetary Science Letters 129: 193-

202.

Brummer CJ, Montgomery DR. 2003. Downstream coarsening in headwater channels.

Water Resources Research 39(10): 1-12.

9

Covich AP. 1988. Geographical and historical comparisons of neotropical streams: biotic

diversity and detrital processes in highly variable habitats. Journal of the North

American Benthological Society 7: 361-386.

Covich AP, Crowl TA. 1990. Effects of hurricane storm flow on transport of woody

debris in a rain forest stream (Luquillo Experimental Forest, Puerto Rico).

International Symposium on Tropical Hydrology and Fourth Caribbean Islands

Water Resources Congress. American Water Resources Association: 197-205.

Golden LA, Spring GS. 2006. Channel geometry, median grain size, and stream power in

small mountain streams. Geomorphology 78: 64-76.

Grant GE, Swanson FJ, Wolman MG. 1990. Pattern and origin of stepped-bed

morphology in high-gradient streams, western Cascades, Oregon. Geological

Society of America Bulletin. 102(3): 340-352.

Greathouse EA, Pringle CM, McDowell WH, Holmquist JG. 2006. Indirect upstream

effects of dams: consequences of migratory consumer extirpation in Puerto Rico.

Ecological Applications 16: 339-352.

Gupta A. 1975. Stream characteristics in eastern Jamaica, an environment of seasonal

flow and large flood. American Journal of Science 275: 825-847.

Gupta A. 1988. Large floods as geomorphic events in the humid tropics. In: Flood

Geomorphology. Baker VR, Kochel RC, Patton PC (eds). Wiley: New York, NY;

301-315.

10

Gupta A. 1995. Magnitude, frequency, and special factors affecting channel form and

processes in the seasonal tropics. In: Natural and Anthropogenic Influences in

Fluvial Geomorphology. Costa JE, Miller AJ, Potter KW, Wilcock PR (eds).

Geophysical Monograph 89. American Geophysical Union: Washington DC; 125-

136.

Gupta A, Ahmad R. 1999. Geomorphology and the urban tropics: building an interface

between research and usage. Geomorphology 31: 133-149.

Hack JT. 1957. Studies of longitudinal stream profiles in Virginia and Maryland. United

States Geological Survey Professional Paper 294-B; 1-97.

Holmquist JG, Schmidt-Gengenbach JM, Yoshioka BB. 1998. High dams and marine-

freshwater linkages: effects on native and introduced fauna in the Caribbean.

Conservation Biology 12: 621-630.

Larsen MC, Torres-Sanchez AJ, Concepcion IM. 1999. Slopewash, surface runoff and

fine-litter transport in forest and landslide scars in humid-tropical steeplands,

Luquillo Experimental Forest, Puerto Rico. Earth Surface Processes and

Landforms 24: 481-502.

Lenzi MA, Mao L, Comiti F. 2006. When does bedload transport begin in steep boulder-

bed streams? Hydrological Processes 20: 3517-3533.

Leopold LB, Maddock JT. 1953. The hydraulic geometry of stream channels and some

physiographic implications. United States Geologic Survey Professional Paper

252.

Leopold LB, Wolman MG, Miller JP. 1964. Fluvial Processses in Geomorphology. W.H.

Freeman and Company: San Francisco, CA.

11

Lewis LA. 1969. Some fluvial characteristics of the Manati Basin, Puerto Rico. Annals of

the Association of American Geographers 59: 280-293.

March JG, Benstead JP, Pringle CM, Scatena FN. 2003. Damming tropical island

streams: problems, solutions and alternatives. BioScience 53: 1069-1078.

Montgomery DR, Bufffington JM. 1997. Channel-reach morphology in mountain

drainage basins. Geological Society of America Bulletin 109(5): 596-611.

Montgomery DR. 1999. Process Domains and the River Continuum. Journal of the

American Water Resources Association 35: 397-410.

Pizzuto JE. 1995. Downstream fining in a network of gravel-bedded rivers. Water

Resources Research 31: 753-759.

Poole GC. 2002. Fluvial landscape ecology: addressing uniqueness within the river

discontinuum. Freshwater Biology 47: 641-600.

Pringle CM, Naiman RJ, Bretschko G, Karr JR, Oscwood MW, Webster JR, Welcomme

RL, Winterbourn MJ. 1988. Patch dynamics in lotic systems: the stream as a

mosaic. Journal of the North American Benthological Society 7: 503-524.

Pringle CM, Scatena FN. 1998. Factors affecting aquatic ecosystem deterioration in Latin

America and the Caribbean. In: Tropical Managed Ecosystems: New perspectives

on Sustainability. Hatch U, Swisher ME (eds). Oxford University Press: New

York.

Pringle CM, Freeman M, Freeman B. 2000. Regional effects of hydrologic alterations on

riverine macrobiota in the New World: Tropical-temperate comparisons.


12

Radecki-Pawlik A. 2002. Bankfull discharge in mountain streams: theory and practice.

Earth Surface Processes and Landforms 27: 115-123.

Rengers F, Wohl E. 2007. Trends of grain sizes on gravel bars in the Rio Chagres,

Panama. Geomorphology 83: 282-293.

Scatena FN, Larsen MC. 1991. Physical aspects of Hurricane Hugo in Puerto Rico.

Biotropica 23: 317-323.

Scatena FN. 1995. Relative scales of time and effectiveness of watershed processes in a

tropical montane rain forest of Puerto Rico. In: Natural and Anthropogenic

Influences in Fluvial Geomorphology. Costa JE, Miller AJ, Potter KW, Wilcock

PR (eds). Geophysical Monograph 89. American Geophysical Union: Washington

D.C.; 103-111.

Smith GC, Covich AP, Brasher AMD. 2003. An ecological perspective on the

biodiversity of tropical island streams. BioScience 53: 1048-1051.

Snyder NP, Whipple KX, Tucker GE, Merritts DJ. 2003. Importance of a stochastic

distribution of floods and erosion thresholds in the bedrock river incision

problem. Journal of Geophysical Research 108: DOI: 10.1029/2001JB00165

Stock JD, Dietrich WE. 2006. Erosion of steepland valleys by debris flows. Geological

Society of America Bulletin 118: 1125-1148.

Statzner B, Higler B. 1986. Stream hydraulics as a major determinant of benthic

invertebrate zonation patterns. Freshwater Biology 16: 127-139.

Torizzo M, Pitlick J. 2004. Magnitude-frequency of bed load transport in mountain

streams in Colorado. Journal of Hydrology 290: 137-151.

13

Townsend CR. 1989. The patch dynamics concept of stream community ecology. Journal

of the North American Benthological Society 8: 36-50.

White AF, Blum AE, Schulz MS, Vivit DV, Stonestorm DA, Larsen M, Murphy SF,

Eberl D. 1998. Chemical weathering in a tropical watershed, Luquillo mountains

Puerto Rico: I. Long-term versus short-term weathering fluxes. Geochimica et

Cosmochimica Acta 62(2): 209-226.

Whipple KX. 2004. Bedrock rivers and the geomorphology of active orogens. Annual

Review of Earth and Planetary Sciences 32: 151-185.

Wohl E, Merritt DM. 2001. Bedrock channel morphology. Geological Society of

America Bulletin 113: 1205-1212.

Wohl E, 2004. Limits of downstream hydraulic geometry. Geology 32(10): 897-900.

Wohl E, Kuzma KN, Brown NE. 2004. Reach-scale channel geometry of a mountain

river. Earth Surface Processes and Landforms 29: 969-981.

Wolman MG, Miller JP. 1960. Magnitude and frequency of forces in geomorphic

processes. Journal of Geology 68: 54-74.

Wu JG, Loucks OL. 1995. From balance of nature to hierarchical patch dynamics: A

paradigm shift in ecology. Quarterly Review of Biology 70: 439-466.

VanLaningham S, Meigs A, Golfinger C. 2006. The effects of rock uplift and rock

resistance on river morphology in a subduction zone forearc, Oregon, USA. Earth

Surface Processes and Landforms 31: 1257-1279.

Vannote RL, Minshall GW, Cummins KW, Sedell JR, Cushing CE. 1982. The River

Continuum Concept. Canadian Journal of Fisheries and Aquatic Sciences 37: 130-

137.

14

CHAPTER 2

APPLICATION OF DIGITAL TERRAIN ANALYSIS TO MODEL SURFACE WATER FLOW IN THE LUQUILLO MOUNTAINS OF NORTHEASTERN PUERTO

RICO*

A.S. Pike

ABSTRACT

Digital terrain analysis was applied to estimate hydrologic parameters in basins

draining the Luquillo Mountains of Northeastern Puerto Rico. A 10m resolution Digital

Elevation Model (DEM) was interpolated from 10m elevation contour lines and used as

the template for hydrologic analysis. A high drainage density stream network,

representing perennial streams, including previously unmapped 1st order streams, was

extrapolated from the DEM. Similarly, for each 10m grid cell within the DEM, rainfall,

runoff, and mean annual discharge were estimated using regression equations derived

from long-term rainfall and stream flow gages. The result is a simple, reproducible spatial

framework that researchers and managers can use to estimate hydrologic conditions in the

region.

* Published as: Pike AS. 2006. Application of digital terrain analysis to estimate hydrological variables in the Luquillo Mountains of Puerto Rico. In: Climate Variability and Change–Hydrological Impacts (Proceedings of the Fifth FRIEND World Conference held at Havana, Cuba, November 2006), IAHS Publ. 308, 81-85.

15

INTRODUCTION

Digital terrain analysis can be used to derive a wealth of information about the

morphology and hydrodynamics of a land surface. When coupled with the spatial

distribution of basic hydrologic variables, such as rainfall and runoff, digital terrain

analysis is a powerful tool for estimating stream network parameters and analyzing

drainage basin characteristics (Montgomery et al. 1998). However, digital terrain analysis

is often underutilized in tropical drainage basins due to a scarcity of appropriate data.

This paper details how digital terrain analysis is used to model surface water flow in the

region draining the Luquillo Experiment Forest (LEF), a montane subtropical rainforest

in northeastern Puerto Rico.

Extensive research has been conducted in the LEF in the fields of fluvial

geomorphology (Scatena 1989), aquatic biology (Covich et al. 1996), and human-river

interactions (González-Cabán and Loomis 1998). Each of these disciplines demand

spatially explicit information on annual rainfall and streamflow. In mountain drainage

basins, hydrologic and geomorphic processes are key drivers of biological processes and

ecological integrity, which in turn influence the human and economic valuation of the

river.

While there has been work incorporating Geographic Information Systems (GIS)

to predict drought and low-flows in the LEF (García-Martinó et al. 1996a), a simple

organized framework to estimate hydrologic parameters that is reproducible between

researchers in the area is lacking. Unfortunately, estimates of elevation, slope, and

drainage areas from topographic maps may not always be accurate. The river network

portrayed on United States Geological Survey (USGS) maps do not represent all streams

16

of research value; many studied 1st and even 2nd order streams in the region are not

plotted on drainage basin scale maps.

Therefore, this paper aims to develop an initial organized framework to estimate

hydrologic parameters for all 10m x 10m cells within a stream network that can act as a

template for future research on the stream in the area. The process involves 1) creation of

a hydrologically correct Digital Elevation Model (DEM), 2) extraction of the stream

network using a drainage area threshold, and 3) estimation of mean annual rainfall,

runoff, and discharge from elevation data.

STUDY AREA

The Luquillo Mountains in northeastern Puerto Rico are characterized by rugged

terrain and steep gradients in elevation and climate. Over a distance of 10 to 20km, the

mountain range rises from sea level to an elevation of 1075m. Mean annual rainfall

increases with elevation from approximately 1500mm/yr at the coast to >4500mm/yr at

the highest elevations (García-Martinó et al. 1996b).

The climate is characterized as humid tropical maritime, and is influenced by both

northeasterly trade winds and local orographic effects. The principal weather systems

affecting climate are convective storms, easterly waves, cold fronts, and tropical storms

(van der Molen 2002). Rainfall events at mid-elevations are generally small (median

daily rainfall 3mm/day) but numerous (267 rain days per year) and of relatively low

intensity (<5mm/hr) (Schellekens et al. 1999). At mid-upper elevations, the majority of

streamflow results from direct surface runoff in the form of saturated overland flow or

17

through shallow (>30cm depth) soil macropores, as no significant groundwater sources

exist on the steep slopes (Schellekens et al. 2004).

Five principal basins drain the Luquillo Experimental Forest: Río Blanco, Río

Espiritu Santo, Río Fajardo, Río Mameyes, and Río Sabana. Streams draining the mid-

upper elevations (>100m) are relatively pristine, surrounded by protected forest, and are

laterally confined by steep valley walls. In contrast, streams flowing across the broad

alluvial coastal plain characterizing the lower elevations readily migrate laterally, and

many are physically altered to attend to local needs. Intensive agriculture and

urbanization has resulted in many lowland 1st order streams being rerouted for irrigation

canals or artificially channelized (Clark & Wilcock 2000). Similarly, the main stems of

all major rivers in the region have either a dam or water intake device to withdraw water

for municipal use (March et al. 2003).

DIGITAL ELEVATION MODEL (DEM) CONSTRUCTION

The main principle of digital terrain analysis is that an abundance of topographic

information is contained within elevation contour lines (elevation, geomorphic position,

slope, etc.) such that a continuous landscape surface can be generated from these

contours. Surface water flow can be routed across this surface under the assumption that

water flows downslope according to principles of least energy, i.e. water follows the path

of steepest descent (Jenson and Domingue 1988). Using this simple rule, the drainage

network of a landscape can be extracted.

A high-resolution Digital Elevation Model (DEM) is critical for terrain and

hydrologic analysis. While a 30m for the entire island of Puerto Rico exists, this is not

18

sufficient resolution to model the complex topographic structure of the Luquillo

mountains, in particular for 1st order streams that typically have an active channel width

of < 10m. Therefore a 10m x 10m grid cell resolution DEM was constructed for the

region of northeastern Puerto Rico for the purpose of digital terrain analysis.

Many Geographical Information Systems (GIS) packages are available that

provide the necessary tools and algorithms to generate a hydrologically correct DEM

from contour data. These include the ArcGIS Spatial Analyst, ArcHydro (Maidment

2002), TauDEM (Tarboten 2000), and GRASS open-source GIS (Neteler and Mitasova

2002). The general procedure employed by these packages is as follows (Fig. 2.1):

a. Conversion of Contour Lines to Triangulated Irregular Network (TIN)

b. Conversion of TIN to Digital Elevation Model (DEM) raster grid

c. Fill sinks in DEM to create a hydrologically correct surface

d. Calculation of Flow Direction Grid

e. Calculation of Flow Accumulation Grid

f. Designation of Stream Channel Threshold from Flow Accumulation Grid

As the basis for the DEM, 10m elevation contours from the United States

Geological Survey (Seiders 1971) were used. Contours were converted to a Triangulated

Irregular Network (TIN) surface, which is constructed by triangulating a set of vertices

that are connected with a series of edges to form a network of triangles. The edges of

TINS form contiguous, non-overlapping triangular facets, and can be used to capture the

position of linear features that play an important role in a surface, such as ridgelines or

stream courses (Wise 1998). TIN surfaces are advantageous for representing a landscape

in that they are variable resolution, with more triangles where the topography is more

19

Figure 2.1 The process of extracting a stream network from contour data. Contour

TIN DEM Flow Direction Flow Accumulation Vector Stream Network. The

area illustrated is known locally as ‘Puente Roto’, on the Río Mameyes.

10m ContoursTriangulated Irregular

Network (TIN)10m Digital Elevation

Model (DEM)

Flow Direction Grid Flow Accumulation Grid Vector Stream Network0 500 m

¯

20

complex. However, algorithms to route flow and extract stream networks from TINs are

less widely available than raster surface algorithms, given the complex structure of the

TIN.

For the purpose of flow routing, the TIN was interpolated to a raster DEM at 10m

x 10m grid cell resolution using ArcGIS Spatial Analyst. Because interpolation of the

input TIN surface occurs at regular intervals, some loss of information in the output raster

should be expected. How well the raster represents the TIN is dependent on the resolution

of the raster, and the degree and interval of the TIN surface variation (Wise 1998).

Generally, as the resolution is increased, the output raster more closely represents the

TIN surface. In the Luquillo Mountains, 10m resolution is sufficient to capture many

elements of the TIN, and is an appropriate resolution for digital terrain analysis (Pike

2001).

The resulting raw DEM generated from contour lines often contains topographic

sinks that create problems in simple hydrologic models. These include possible negative

(below sea-level) values, and pits and depressions that ultimately act as sinks for flow

(Tarboten et al. 1991). They are the product of both digital interpolation errors and

natural features of the landscape (lakes, depressions, etc.), and are easily corrected by

raising negative values and filling sinks to ensure continuous hydrologic flow. Similarly,

in flat terrain, fine-scale features such as river meanders and river course may not be well

constrained by the contour lines. If uncorrected, rivers flowing over low-relief surfaces

may have overly straight flow paths, or may appear as a series of banded lines. For better

accordance with known river paths, the river can be forced into a DEM such that river

21

cells are lowered by several meters to ensure that they are topographically lower than

surrounding cells (Maidment 2002).

Flow direction was calculated according to the simple D8 algorithm, whereby

flow is routed to the adjacent cell of steepest descent, or greatest elevation drop (Jenson

and Domingue 1988). A flow direction matrix is computed where each grid cell is

assigned a value (1-8), corresponding to the eight cardinal directions, that routes that flow

to the appropriate adjacent cell.

The flow accumulation grid was computed using the flow direction grid to sum

the total number of upslope cells contributing to a given cell. Cells can be weighted so

that the accumulated surface represents the sum of upslope weights. For example, if the

weight is a constant 100m2 area for each 10m cell, then the accumulation represents the

drainage area. Similarly, if the weight for each cell is the yearly runoff, then the

accumulated surface will represent mean annual discharge.

STREAM NETWORK EXTRACTION

To extract a stream network from a DEM, a drainage area threshold must be

applied to the flow accumulation surface (Tarboten et al. 1998). The threshold represents

the critical drainage area that distinguishes perennial from ephemeral streams; grid cells

that exceed the threshold represent streams with year-round flow. The threshold for

streams in the lower elevations, or tabonuco forest zone, of the Luquillo Experimental

Forest has been determined to be approximately 6ha (Scatena 1989).

The resulting map of the perennial stream network shows a much higher drainage

density than the USGS stream network (Fig. 2.2). For the Río Mameyes, the resulting

22

±

2,000

Meters

±

2,000

Meters

Figure 2.2 Comparison of the USGS stream network (a) to the DEM generated stream

network at 6ha drainage area threshold (b) for the Río Mameyes. Lines widths are scaled

according to stream order. While only the Río Mameyes is illustrated (to show fine scale

features), other drainage basins show a similar comparison between USGS and DEM

generated stream networks.

a) b)

23

total length of the stream network at 6ha drainage area is 133km, compared to a total

length of 70km for the USGS stream network. This increased length is due to the

inclusion of a large number of previously unmapped 1st and 2nd order streams that do not

appear at the resolution of the 1:20,000 scale USGS map. Similarly, the large amount of

1st order streams changes the total stream ordering of the network. The mouth of the Río

Mameyes is a 4th order stream according to the USGS map, but is a 5th order stream

according to the derived stream network.

While the resulting stream network accurately represents the path of major stream

channels (as they were forced to follow the mapped lines), some of the smaller streams

may not be accurately represented. This is generally a problem in the lowland areas;

stream paths in flat terrain may follow artifacts of the surface interpolation procedures

rather than real topographic features. Similarly, small streams have been diverted in the

lowland agricultural fields and urbanizations, so that even if the flow paths were

represented correctly by stream network, the digital estimation and reality may not agree.

Therefore, for the purpose of mapping small 1st and 2nd order streams, the digitally

extracted stream network should be used with caution when outside of natural, valley

confined upland streams.

RAINFALL, RUNOFF, AND DISCHARGE

The high spatial variability of rainfall in the LEF, especially its relationship with

elevation, suggests that simple scaling of drainage area with discharge may not apply to

this landscape. A small drainage in the uplands will have distinctly more runoff than a

comparable sized drainage basin in the lowlands. To estimate the spatial distribution in

24

rainfall, runoff and discharge, the following regression equations estimated from long-

term rain and stream gages were used (García-Martinó et al. 1996b):

P = 2300 + 3.8h -.0016h2 n = 17, r2 = 0.91, P < 0.001 Eq. 2.1

R = 4.26havg + 360 n = 9, r2 = 0.77, P = 0.002 Eq. 2.2

Discharge can be estimated from runoff by multiplying by drainage area:

Q = 3.17x10-5DA(4.26havg + 360) n = 9, r2 = 0.97, P < 0.001 Eq. 2.3

where: P = mean annual rainfall (mm/yr), R = mean annual runoff (mm/yr), Q = mean

annual discharge (m3/s), h = elevation (masl), havg = weighted average elevation (m), and

DA = drainage area (km2).

The resulting maps of mean annual rainfall, runoff, and discharge for the Río

Mameyes are shown (Fig. 2.3). Note that rainfall and runoff very closely resemble the

elevation structure, while discharge shows a similar pattern as the flow accumulation

grid. This is due to the fact that both rainfall and runoff are based on elevation, while

discharge is a strong function of drainage area.

The “weighted upstream elevation”, used in calculating runoff, is an accumulated

function. That is, it is the sum the elevation of all upslope cells, divided by the number of

accumulated cells. This accounts for the fact that basins at higher elevations have greater

mean annual runoff than corresponding basins of equal area at lower elevations.

More complex models exist to estimate the spatial distribution of rainfall and

runoff, such as PRISM (Daly et al. 2003), a rainfall models incorporating aspect,

windward/leeward orographic affects, and coastal advection, and TOPMODEL (Bevin et

al. 1995), a rainfall-runoff model based on topographic properties. While these models

are a better predictor of rainfall and runoff on the scale of the island of Puerto Rico, the

25

±

2,000

Meters

Dischargem3/s

5.0

0.0

±

2,000

Meters

Runoffmm/yr

4300

800

±

2,000

Meters

Rainfallmm/yr

4500

2300

Figure 2.3 Spatial distribution of a) mean annual rainfall, b) mean annual runoff, and c)

mean annual discharge within the Río Mameyes drainage basin according to elevation-

based regression equations.

a) b) c)

26

simple regression-based approach mentioned here is sufficient to make accurate

predictions of rainfall and runoff on the windward steep slopes of the Luquillo

Mountains.

CONCLUSIONS

The stream network and rainfall, runoff, and mean annual discharge can be

accurately estimated at 10m spatial resolution according to a simple DEM-based process

for basins draining the Luquillo Experimental Forest (LEF). The estimates are best

applied to stream in the forested upland regions, as anthropogenic activity on the lowland

rivers have altered stream channel courses and hydrologic budgets. However, the

simplicity of this DEM-based approach allows any researcher knowledgeable in GIS and

working in the regional area to estimate key hydrological parameters.

ACKNOWLEDGEMENTS

The author thanks Dr. Fred Scatena for advice on the manuscript, and Dr. Lena

Tallaksen for strengthening comments. Funding for this study was provided by the

National Science Foundation Biocomplexity Grant (NSF #030414)—Rivers, Roads, and

People: Complex Interactions of Overlapping Networks in Watersheds.

27

REFERENCES Beven KJ, Lamb R, Quinn PF, Romanowicz R, Freer J. 1995. TOPMODEL. In:

Computer Models of Watershed Hydrology. Singh VP (ed). Water Resources

Publications: Highlands Ranch, CO; 627-668.

Clark JJ, Wilcock PR. 2000. Effects of land-use change on channel morphology in

northeastern Puerto Rico. Geological Society of American Bulletin 112: 1763-

1777.

Covich AP, Crowl TA, Johnson SL, Pyron M. 1996 Distribution and abundance of

tropical freshwater shrimp along a stream corridor: response to disturbance.


Daly C, Helmer EH, Quiñones M. 2003. Mapping the climate of Puerto Rico,

Vieques, and Culebra. International Journal of Climatology 23: 1359-1381.

García-Martinó AR, Scatena FN, Warner GS, Civco DL. 1996a. Statistical low flow

estimation using GIS analysis in humid montane regions in Puerto Rico. Water

Resources Bulletin 32: 1259-1271.

García-Martinó A.R, Warner GS, Scatena FN, Civco DL. 1996b. Rainfall, runoff, and

elevation relationships in the Luquillo Mountains of Puerto Rico. Caribbean

Journal of Science 32: 413-424.

González-Cabán A., Loomis J. 1998. Economic benefits of maintaining ecological

integrity of Río Mameyes, in Puerto Rico. Ecological Economics 21: 63- 75.

28

Jenson SK, Domingue JO. 1988. Extracting topographic structure from digital

elevation data for geographic information system analysis. Photogrammetric

Engineering and Remote Sensing. 54: 1593-1600.

Maidment DR. 2002. ArcHydro: GIS for water resources. ESRI Press: Redlands, CA.

March JG, Benstead JP, Pringle CM, Scatena FN. 2003 Damming tropical island

streams: problems, solutions, and alternatives. BioScience 53: 1069-1078.

Montgomery DR, Dietrich WE, Sullivan K. 1998. The role of GIS in watershed

analaysis. In: Landform Monitoring, Modelling, and Analysis. Lane SN,

Richards KS, and Chandler JH (eds). Wiley: West Sussex, England; 241-261.

Neteler M, Mitasova H. 2002. Open Source GIS: A GRASS GIS Approach. Kluwer

Academic Press: Boston, Dordrecht.

Pike RJ. 2001. “Topographic fragments” of geomorphometry, GIS, and DEMs. In:

DEMS and Geomorphology, Geographic Information Systems Asssociation

(Japan) Special Publication. 5th International Conference on Geomorphology,

Chuo University: Tokyo, Japan; 1: 34-35.

Scatena FN. 1989. An introduction to the physiography and history of the Bisley

Experimental Watersheds in the Luquillo Mountains of Puerto Rico. United

States Department of Agricultural Forest Service General Technical Report

SO-72.

Schellekens J, Scatena FN, Bruijnzeel LA, Wickel AJ. 1999. Modelling rainfall

interception by a lowland tropical rain forest in north eastern Puerto Rico. Journal

of Hydrology 225: 168-184.

29

Schellekens J, Scatena FN, Bruijnzeel LA, van Dijk AIJM, Groen MMA, van

Hoogezand RJP. 2004. Stormflow generation in a small rain-forest catchment

in the Luquillo Experimental Forest, Puerto Rico. Hydrologic Processes 18:

503-530.

Seiders VM. 1971. Geologic map of the El Yunque quadrangle, Puerto Rico. United

States Geological Survey Miscellaneous Geologic Investigations Map I-658;

1:20,000 scale.

Tarboton DG, Bras RL, Rodriguez-Iturbe I. 1991. On the extraction of channel

networks from digital elevation data. Hydrologic Processes 5: 81-100.

Tarboton DG. 2000. Terrain analysis using digital elevation models (TauDEM). Utah

Water Research Laboratory, Utah State University, Logan, Utah, USA.

van der Molen MK. 2002. Meteorological impacts and land use change in the

maritime tropics. Ph.D. Thesis. Vrije Universiteit: Amsterdam, Netherlands.

Wise SM. 1998. The Effect of GIS Interpolation Errors on the Use of Digital

Elevation Models in Geomorphology. In: Landform Monitoring, Modelling,

and Analysis. Lane SN, Richards KS, Chandler JH (eds). Wiley: West Sussex,

England; 139-164.

30

CHAPTER 3

DEFINING A BANKFULL ANALOG FOR TROPICAL MONTANE STREAMS USING RIPARIAN FEATURES*

A.S. Pike and F.N. Scatena

ABSTRACT

The bankfull stage of a river channel is an important geomorphologic and

ecological boundary, but is rarely identifiable in steep mountain channels. This study

defines a ‘bankfull’ zone in tropical mountain streams that is based on statistically

defined combinations of riparian features that occur at the same flood frequency as the

bankfull stage and the effective discharge in adjacent alluvial channels. The relative

elevation of riparian vegetation, soil, and substrate characteristics were surveyed at nine

(9) stream gages in and around the Luquillo Experimental Forest in Northeastern Puerto

Rico. The corresponding discharge, flow frequency, and recurrence intervals associated

with these features was determined from a partial duration series analysis using long-term

15-minute resolution discharge records. Survey data indicate that mosses and short

grasses dominate at a stage often inundated by sub-effective flows. Herbs first occur at

elevations along the channel margin associated with intermediate discharges that

correspond to the threshold for sediment mobilization. Near-channel woody shrubs and

trees establish at elevations with a less frequent discharge that is coincident with the

effective discharge of bedload sediment transport. A multivariate regression tree

technique was then used to identify the characteristic features within the riparian zone

that are inundated at similar flow frequencies as the bankfull stage in alluvial channels. * Submitted to the Journal of Hydrology in January 2008. All data analysis, interpretation of results, and manuscript writing were done by author of this dissertation.

31

Our data demonstrate that in alluvial channels, both the bankfull stage (as marked by a

floodplain) and the channel-forming (effective) discharge are associated with the

presence of fine-grained substrate and soil, and tall, mature woody vegetation. In

montane reaches that lack a floodplain, a zone that is characterized by the incipient

presence of soil, woody shrubs, and trees has the same flow frequency as the bankfull

discharge of the alluvial channels. The bankfull discharge based on these riparian features

in steepland sites has an average exceedance probability between 0.09% and 0.30%, and

a recurrence interval between 40 and 90 days (based on 15-min resolution discharge

data). We conclude that flows with similar frequencies influence the establishment of

riparian vegetation, soil development, and substrate characteristics along channel margins

in similar ways. Thus, riparian features can be used as an indicator of hydrogeomorphic

site conditions to identify active-channel boundaries that occur at a constant flow

frequency throughout the study basins.

32

INTRODUCTION

The bankfull discharge and morphology of river channels are fundamental

concepts in fluvial geomorphology, hydrology, and stream ecology. Bankfull stage is

loosely defined as where stream water flow begins to flow out of the main channel and

into the adjacent floodplain (Williams 1978). It is the most common morphologic

measure used in comparing spatial variations in channel morphology for downstream

hydraulic geometry studies (Leopold and Maddock 1953) and is thought to correspond to

the effective, or channel-forming, discharge in alluvial reaches (Wolman and Miller

1960, Andrews 1980). Bankfull stage is also an important hydrologic and ecological

boundary that can mark the extent of flood zones and riparian forests (Williams 1978,

Andrews 1980, Radecki-Pawlick 2002). Consequently, it is also used as a central design

parameter in stream restoration projects and land use planning (Rosgen 1994). Although

the utility of bankfull conditions is widely acknowledged, the difficulty of consistently

identifying bankfull conditions across streams and rivers is equally recognized, even in

alluvial channels where a floodplain is present (Navratil et al. 2006). The task of

identifying bankfull flows is even more difficult in mountainous environments that do not

have readily distinguishable alluvial floodplains. This study describes a field based

statistical technique that can be used to characterize a bankfull analog in tropical montane

stream channels that has the same flow frequency as the bankfull discharge in adjacent

alluvial channels.

Many definitions have been coined to capture the range and complexity of

bankfull conditions (see Radecki-Pawlick 2002 for a comprehensive review). The

multitude of definitions are based on features in alluvial channels that fall into four

33

general categories: 1) morphologic features, 2) geometric features, 3) bank and sediment

features, and 4) riparian features (Harrelson et al. 1994). Morphologic indicators are

marked by the elevation of the top of depositional surfaces that correspond to the

boundary in the channel between net sediment transport and sediment deposition

(Wolman and Leopold 1957, Woodyer 1968, Williams 1978). Geometric features are

primarily based on the change in slope that occurs along the cross-section from the

channel to the banks (Wolman 1955, Riley 1972). Bank and sediment features include

changes in particle size or the extent of undercutting under dense root mats (Leopold and

Skibitzke 1967). Lastly, riparian features such as stain lines and vegetation are often used

to mark bankfull in steep streams where other features are not present. Stain lines on

large boulders that are marked by fine sediment may indicate either bankfull or the last

large flood (Harrelson et al. 1994), and a sharp break in the vegetation density, a change

in the type of vegetation, or the lower limit perennial vegetation (usually trees) can also

approximate bankfull (Williams 1978).

Although the bankfull concept is widely used in alluvial streams, it does not

necessarily apply to channels in mountain landscapes (Radecki-Pawelik 2002). In these

environments, the morphologic and geometric features typically used to mark bankfull

conditions rarely exist in the steep, confined, ‘v’-shaped valleys. Many researchers have

estimated ‘bankfull conditions’ in mountain streams using field observations of high-flow

features in riparian zones, including: the boundary of the active-scour zone (Montgomery

et al. 2001), flow-deposited organic debris and changes in the grain size of surface

sediment (Wohl and Wilcox 2005), changes in bank-gradient and channel geometry

(Wohl et al. 2004), the presence of perennial vegetation (Radecki-Pawlik 2002), or a

34

combination of all factors. However, the flow-frequency associated with these various

reference features is rarely quantified in mountain stream studies and many of them may

reflect the influence of the last large flood rather than the steady forcing from more

frequent bankfull conditions. Consequently, it is more appropriate to refer to these high-

water marks as a ‘reference’ discharge (Wohl and Wilcox 2005), rather than using the

term ‘bankfull discharge’—a term that implies a discharge with a specific recurrence

interval (Andrews 1980). Yet there is no common consensus on what either ‘bankfull

discharge’ or ‘reference discharge’ are in mountain streams, nor how often these high-

flow events occur.

In the absence of traditional depositional and morphologic bankfull indicators in

mountain streams, the occurrence of riparian vegetation may be used as a marker of flow

frequency. There is often a consistently observable vertical zonation in the type and

structure of riparian vegetation along a gradient from the active channel to the banks. It

has been shown that such patterns in riparian vegetation are strongly influenced and

maintained by the natural flow regime of a river (Poff et al. 1997). The active channel is

physically the harshest environment for terrestrial vegetation because it has the highest

frequency of flooding and scouring (Naiman et al. 1998, Swanson et al. 1998). Woody

plants may be mechanically broken by the force of floodwaters, uprooted by erosion of

the substrate in which they are rooted, or unable to establish in their seedling stage

(Bendix and Hupp 2000). Since periodic flooding is an important physical determinant of

the establishment and growth of many riparian plants, there should be a functional

relationship between flood hydrology and riparian plant community patterns (Chopin et

al. 2002). In aseasonal humid tropical environments where vegetation is abundant and

35

rapidly colonizes disturbed surfaces, streamside vegetation should closely reflect the

flood disturbance regime. Therefore, vegetation and corresponding surficial features

(substrate type, degree of soil development, organic matter) can presumably be used as

indicators for particular hydrogeomorphic site conditions (Hupp and Osterkamp 1985).

Furthermore, it is likely that some combination of riparian vegetation and corresponding

features can approximate bankfull on the basis of flow-frequency.

Bankfull discharge has been shown to correspond to the effective, or channel-

forming, discharge in alluvial streams (Wolman and Miller 1960, Andrews 1980). The

effective discharge is defined as the discharge that transports the most sediment over time

(Leopold et al. 1964). The concept of effective discharge assumes a balance between the

frequency and magnitude of flows and the corresponding amount of sediment transport.

Flows of low magnitude are common, but transport little to no bed sediment. Conversely,

large floods exert tremendous fluvial energy, transport large amounts of sediment,

deposit sediment on floodplains, and have the potential to completely reform the stream

channel. However, these high-magnitude flows are infrequent so that their effective

contribution to geomorphic work over time is negligible. Between the tranquility of

baseflow and the ferocity of large floods, there exists a relatively frequent, moderately

high magnitude channel-maintaining flow that effectively transports sediment through the

channel. The effective discharge often corresponds with the bankfull stage in alluvial

channels, where the morphology of the channel depends on the magnitude and frequency

of sediment-transporting flows (Wolman and Miller 1960).

Textbook scenarios of bankfull recurrence intervals assert that the bankfull stage

is exceeded by a flood occurring every 1-3 years (Leopold et al. 1964, Harrelson et al.

36

1994, Knighton 1998). This number has been used as the basis for many stream

restoration projects to construct channels with dimensions in balance with the natural

flow regime (Rosgen 1994). This recurrence interval has also been used in many

downstream hydraulic geometry studies (Wohl et al. 2004, Wohl and Wilcox 2005),

where longitudinal changes in bankfull channel width, depth, and velocity are driven by

the magnitude of this channel-forming discharge (Leopold and Maddock 1953).

Furthermore, analysis of downstream hydraulic geometry requires that both the bankfull

and effective discharge occur at a constant frequency throughout a basin. Identifying a

common marker of the channel-forming discharge with equal flow-frequency along the

course of a river is essential to make meaningful comparisons of channel cross-sectional

geometry.

This study investigates relationships between the flow regime and riparian

features in the tropical montane streams of Northeastern Puerto Rico to ultimately find a

consistent indicator of bankfull stage. We have four primary objectives. First, we

quantify the flow-frequency associated with the first occurrence of different types of

riparian vegetation, soil development, organic debris, and substrate sizes at a series of

long-term stream gages. Second, we determine combinations of these features that are

inundated at similar flow frequencies throughout the stream network. Third, we compare

these riparian features in montane reaches to bankfull stage and effective discharge in

adjacent alluvial channels. Lastly, we characterize the properties of those features that

mark an analog of bankfull stage in these tropical steepland streams.

37

STUDY AREA

Northeastern Puerto Rico

This study was conducted in the streams draining the Luquillo Mountains in

Northeastern Puerto Rico. The Luquillo Mountains rise steeply from sea-level to over

1000m in elevation over a distance of 15km-20km. The climate is maritime subtropical

and is influenced by convective storms, easterly waves, cold fronts, tropical storms and

hurricanes (Scatena 1995). Mean annual temperatures at mid-elevations are 26ºC, and

range from an average of 22ºC in the winter to 30ºC in the summer (Ramirez and

Melendez-Colom 2003). Mean annual rainfall increases with elevation from

approximately 1500mm per year at the coast to >4500mm/yr at elevations above 1000 m

(Garcia-Martino et al. 1996). Rainfall events at mid-elevations are generally small

(median daily rainfall 3mm/day) and numerous (267 rain days per year) (Schellekens et

al. 1999). High-intensity rainfall events and floods can occur in any given month.

Hurricanes are common between August through October and typically bring high daily

rainfall in excess of 200mm/day (Heartsill-Scalley et al. 2007). The maximum recorded

daily rainfall and runoff in the region are in excess of 600mm/day (Scatena and Larsen

1991).

Luquillo streams drain a landscape that is tectonically active, disturbed by

periodic tropical storms and hurricanes, and prone to massive landsliding (Larsen and

Torres-Sanchez 1998). These streams, as well as montane streams in the Greater Antilles

in general, have steep gradients, channels lined with coarse boulder-sized sediment,

numerous bedrock cascades, and abrupt waterfalls. Their morphologies have been called

“flood dominated” and like many mountain streams, traditional depositional forms built

38

by sand, pebbles, cobbles, and boulders are found sporadically or only at the lowest

gradient reaches (Gupta 1975, Gupta 1988, Ahmad et al. 1993). It has also been

cautioned that the standard descriptions of alluvial channel form and behavior are not

necessarily adequate for these rivers (Ahmad et al. 1993). Nevertheless, the following

areas are considered to have a similar combination of geological, tectonic, and climatic

conditions that influence channel morphology (Gupta 1988): 1) River valleys of East

Asia, especially Taiwan and the Philippines, 2) Upland areas of Vietnam, Sumatra, Java,

and Burma, 3) Humid areas of the Indian subcontinent, 4) Madagascar and neighboring

parts of coastal East Africa, 5) North and northeastern Australia, 6) Central and South

American highlands and other Caribbean islands.

The intense tropical rains, steep slopes, and rapid runoff generation in the

Luquillo Mountains create an extremely flashy flow regime (Schellekens et al. 2004).

High-magnitude, but short-lived, floods occur sporadically throughout the year. Peak

discharges can be approximately 1000 times greater than baseflow in Luquillo and other

Caribbean streams (Gupta 1995). The average unit discharge of baseflow is

0.02m3/s/km2, whereas the highest peak unit discharge measured by regional USGS

stream gages is 19.7m3/s/km2. However, high-flow hydrographs are short and stormflow

runoff is quickly flushed through the system such that the streams return to baseflow

within hours, even after the largest events. These large floods are primarily driven by

atmospheric disturbances that occur throughout the year rather than by seasonal events

(such as snowmelt) that are common in temperate basins. Consequently, flood discharges

that are close to the annual peak are often experienced independently several times in a

year (Scatena et al. 2004).

39

The morphology of Luquillo stream channels, as well as the composition of the

channel bed, are directly related to the underlying lithology (Ahmad et al. 1993). There

are three dominant lithologies in the study region: volcaniclastics, granodiorite, and

coastal plain alluvium (Seiders 1971). The streams draining volcaniclastics are steep and

typically have a bed composed of large boulders (up to several meters in diameter),

interspersed with finer cobbles and gravels, as well as sporadic bedrock outcrops.

Although the volcaniclastic rocks weather to deep clayey saprolite, the channels are

relatively devoid of sands, silts and clays because these sediments are quickly transported

out of the mountains as suspended load in floods. These channels are generally situated at

the bottom of steeply-walled bedrock valleys and lack floodplains along the channel

margins.

In contrast to areas underlain by the volcaniclastic rocks, streams draining

granodiorite are almost entirely composed of sand and large case-hardened boulders that

can be several meters in diameter. This granodiorite bedrock has one of the highest

documented weathering rates in the world (White et al. 1998). The sand-beds of these

channels are constantly mobile, even at low flows, and are readily reworked during high

flows. These reaches do have some bars and depositional surfaces within the channels,

but they still generally lack a continuous or well-defined floodplain. Like the streams

draining the volcaniclastic bedrock, they are lined with large immobile boulders, even

though the fluvial transport capacity is considered to exceed the sediment supply (i.e. a

‘supply limited’ environment), (Larsen 1997). Many of the larger boulders in both

volcaniclastic and granodioritic channels were apparently delivered to the channels by

landslides and are not readily transported by fluvial processes. Furthermore, some of the

40

largest boulders have not moved in the years of modern observation, and have been

estimated to only be mobile in a 500-year flood or larger.

Streams flowing across the coastal plain alluvium typically have a comparatively

gentle gradient compared to channels on bedrock, and have beds composed of cobbles

and gravels. These lowland streams have several overflow and bench surfaces that are

typical of many alluvial channels, including a morphological bankfull surface (Gupta

1975). The lowest alluvial surface is often an in-channel partially vegetated inset deposit

or bar composed of cobbles (Clark 1997). A slightly higher inset floodplain surface

marks the bankfull stage, and is generally between 1m to 1.5m above the baseflow water

level, and approximately 1.5 to 3 m above the channel thalweg. A discontinuous terrace

at an elevation between 1.75 and 3 m above the floodplain, occurs throughout the coastal

plain, and is most evident on the cutbank side of the channel. There is also a distinctive

higher terrace approximately 9 m above the river that occurs sporadically along the

length of the channel. These terraces are thought to be remnant alluvial surfaces from

Puerto Rico’s pre-agricultural period before 1830 A.D. (Clark and Wilcock 2000).

Regional Stream Gages

Nine long-term stream gaging stations in and around the Luquillo Mountains that

are maintained by the United States Geological Survey (USGS) were selected as sites for

this study (Figure 3.1). Selection criteria included: 1) currently operating gages, with 2)

greater than 10 years of record, and 3) instantaneous (i.e. 15 min) discharge records. The

gages are located in five (5) adjacent watersheds: Río Blanco, Río Espiritu Santo, Río

Fajardo, Río Mameyes, and Río Sabana (Figure 3.1). The sites range from small 1st order

headwater streams to larger 3rd and 4th order streams, with corresponding drainage basin

41

areas ranging from 0.3 km2 to 39 km2. Basic site information is available in Table 3.1.

The reaches near the stream gages are pictured in Figure 3.2.

The gaged reaches were divided into three physiographic categories: lowland (0-

50m), mid-elevation (50-150m), and steepland (>150m). The lowland reaches are low-

gradient, slightly meandering channels that flow across an alluvial coastal plain. These

reaches are unconfined by valley walls, and are typically accompanied by an adjacent

floodplain, high flow channels, and/or terrace deposits. The mid-elevation reaches are

moderate gradient channels that are located upstream of where the mountains grade to the

coastal plain, but downstream of the major cascades and waterfalls. They flow typically

within confining steep valley walls, and have a coarse boulder and bedrock channels. The

steepland reaches are high gradient channels located in the bottom of deeply incised “v-

notched” shaped valleys, and are often confined by nearly vertical valley walls. Cascades,

waterfalls, and long step-pool sequences composed of large boulders are common in

these steepland streams.

Riparian Vegetation

The vegetation of the Luquillo Mountains is typical of a humid tropical rainforest,

with greater tree species diversity, vegetation density, and productivity than most

temperate forests. There are over 225 tree species in the Luquillo Experimental Forest,

and four major forest types (Lugo and Scatena 1995). Although tree species diversity is

higher than in montane temperate forests, the diversity of understory herbs, ferns, grasses,

and shrubs is typically less (Arnold 1996).

Luquillo vegetation is dense and productive due to the tropical climate and year-

round growing season, and covers virtually any surface that is not frequently disturbed.

42

Figu

re 3

.1 L

ocat

ion

map

of t

he se

lect

ed st

udy

gage

s in

and

arou

nd th

e Lu

quill

o Ex

perim

enta

l For

est i

n N

orth

east

ern

Puer

to R

ico.

98

2

6

4

5

3

1 7

04

2Km

Luqu

illo

Expe

rimen

tal F

ores

t

Rio

Esp

iritu

San

to

Rio

Mam

eyes

Rio

Sab

ana

Rio

Faj

ardo

Rio

Bla

nco

Puer

to R

ico

670’W

6630’W

660’W

6530’W

180’N1830’N

43

Tab

le 3

.1 P

hysi

ogra

phic

info

rmat

ion

for t

he se

lect

ed st

udy

gage

s.

U

SGS

Gag

e #

Gag

e N

ame

Cha

nnel

Ty

pe

Geo

logy

B

ed S

ubst

rate

El

evat

ion

(m)

Dra

inag

e A

rea

(km

2 )

Rea

ch

Slop

e (m

/m)

Med

ian

Gra

in

Size

, d50

(m

)

Mea

n D

isch

arge

(m

3 /s)

1 50

0638

00

Rio

Esp

iritu

San

to n

r Rio

Gra

nde

Low

land

A

lluvi

um

Cob

ble,

Gra

vel,

Bed

rock

12

2

2.4

0.01

1 0.

127

1.69

2 50

0710

00

Rio

Faj

ardo

nr F

ajar

do

Low

land

A

lluvi

um

Cob

ble,

Gra

vel

4

2

38.

8 0.

008

0.09

0

2.

05

3 50

0642

00

Rio

Gra

nde

nr E

l Ver

de

Low

land

A

lluvi

um

Cob

ble,

Gra

vel,

Bed

rock

50

1

9.0

0.01

2 0.

177

1.10

4 50

0657

00

Rio

Mam

eyes

at M

amey

es

Low

land

A

lluvi

um

Gra

vel,

San

d, S

ilt

5

34.

9 0.

002

0.03

8

2.

43

5 50

0655

00

Rio

Mam

eyes

nr S

aban

a M

id-

Ele

vatio

n V

olca

nicl

astic

B

ould

er, C

obbl

e,

Bed

rock

84

1

7.9

0.01

5 0.

159

1.55

6 50

0670

00

Rio

Sab

ana

at S

aban

a M

id-

Ele

vatio

n V

olca

nicl

astic

C

obbl

e, G

rave

l

79

1

0.3

0.01

3 0.

068

0.59

7 50

0634

40

Que

brad

a S

onad

ora

nr E

l Ver

de

Ste

epla

nd

Vol

cani

clas

tic

Bou

lder

375

2.6

0.

233

0.48

3

0.

22

8 50

0749

50

Que

brad

a G

uaba

nr N

agua

bo

Ste

epla

nd

Gra

nodi

orite

S

and,

Bou

lder

640

0.3

0.

102

0.01

9

0.

02

9 50

0750

00

Rio

Icac

os n

r Nag

uabo

S

teep

land

G

rano

dior

ite

San

d, B

ould

er

61

6

3

.3

0.02

0 0.

019

0.39

44

Figure 3.2 Photographs of study gages. Note the abundance of vegetation along the

channel margins, the bankfull forms of the alluvial sites (#1-4), and the absence of

bankfull forms in mid-elevation and steepland sites (#5-9).

45

Figu

re 3

.2 c

ont.

46

Moreover, the active stream channels are one of the only geomorphic features

consistently devoid of vegetation. A vegetation transect along a fluvial disturbance

gradient from the middle of the channel into the adjacent forest follows a consistent

pattern. Cushion mosses colonize in-channel boulders, whereas herbs, ferns, and grasses

grow along channel margins, and woody shrubs and trees establish on higher, less

frequently flooded surfaces (Figure 3.3). Vegetation stature similarly increases with

stage. Short-stature vegetation grows along the channel and tall closed-canopy woody

vegetation and tall grasses grow on the banks and hillslopes.

Although there are consistent vegetation patterns in every reach, not all the types

of vegetation are present everywhere, because the abundance of certain species can also

be influenced by land-use legacies and light availability (Heartsill-Scalley and Aide 2003,

Brown et al. 2006). In areas surrounded by forests, the riparian understory vegetation is

mainly composed of shrubs, herbs, and ferns (Scatena 1990, Heartsill-Scalley 2005).

Riparian zones surrounded by pastures and mixed land-uses are commonly dominated by

grasses, vines, and bare soil. Mosses and lichens that require shade are more common in

steepland streams having ample canopy cover. Conversely, wider lowland channels have

a greater amount of incident light and consequently have a greater abundance of grasses.

Furthermore, unlike many arid and semi-arid riparian forests, there is no distinct riparian

forest community in the headwater streams of the Luquillo Mountains (Heartsill-Scalley

2005). Riparian forests along many alluvial streams in arid and semi-arid regions often

have a unique composition and greater productivity than the surrounding vegetation due

to increased availability of water. Yet in the continually humid climate of the Luquillo

Mountains, both riparian and non-riparian forests have ample moisture availability and

47

Figu

re 3

.3 C

ross

sec

tion

at R

ío M

amey

es n

ear

Saba

na, a

mid

-ele

vatio

n si

te. T

here

is a

ver

tical

zon

atio

n of

veg

etat

ion

type

s, fr

om

mos

ses

to h

erbs

to g

rass

es to

shr

ubs

to tr

ees.

The

vege

tatio

n re

flect

s th

e flo

w re

gim

e, a

nd h

ydro

grap

h on

the

right

is u

sed

to v

isua

lly

com

pare

the

inu

ndat

ion

perio

ds o

f ea

ch v

eget

atio

n ty

pe.

Not

e th

e se

vera

l in

tra-a

nnua

l flo

ods

reac

hing

eac

h su

rfac

e. H

ydro

grap

h

disc

harg

e da

ta is

15-

min

ute

reso

lutio

n fo

r the

yea

r 200

3.

48

are consequently similar in composition, but can be different in structure and biomass

(Scatena and Lugo 1995).

Although there is no distinct Luquillo riparian forest community, some tree

species are more abundant along the streams. Valley floors are typically dominated by

palms, herbs, and by light-gap colonizing species, whereas the dominant hardwoods are

confined to more stable ridges (Scatena 1990). Native species commonly found alongside

the steepland streams include: Guarea glabra (alligatorwood), Pterocarpus officinalis

(dragonsblood tree), Inga vera (river koko), and Prestoea montana (sierra palm). Non-

native tree species are common along lowland to mid-elevation streams and are generally

associated with reforesting former agricultural land (O’Connor et al. 2000, Brown et al.

2006). Common non-natives found alongside the streams are: Syzygium jambos (rose

apple), Spathodea campanulata (african tulip tree), Mangifera indica (mango), and

Bambusa spp. (bamboo).

Following a disturbance (flood, treefall gap, hurricane), grasses and herbs can

begin colonizing within days and are typically well established within weeks to months

(Scatena et al. 1996). Likewise, early successional trees can become established within a

year. Given this rapid establishment of vegetation, it is assumed in this study, and

supported by our observations over the years, that there is a general balance between the

frequency and magnitude of floods and the vegetation and soil features adjacent to the

stream channel. Small floods frequently cover in-channel and side-channel boulders that

are habitat for cushion mosses and lichen. Intermediate-magnitude floods inundate

channel bars and low-lying benches, mobilize coarse sediment, and disturb the substrate

occupied by herbs and grasses. Larger floods can have sufficient power to flatten in-

49

channel vegetation, particularly grasses and shrubs, but rarely uproot trees. It is only the

rarest and largest floods, like those observed during Hurricane Hugo (Scatena and Larsen

1990), that uproot mature riparian trees, scour the banks, and completely rework the

channel morphology. These observations indicate that vegetation structure is a highly

sensitive indicator of flow-frequency and that differences in vegetation near the active

channel can be used to define flow regimes and flow frequencies.

METHODS

Our general method involved surveying the relative elevation of different

vegetation types and riparian features at a series of long-term stream gages, and relating

the elevation of each survey point to the corresponding discharge and flow-frequency for

that gage. Multivariate regression techniques were then used to statistically partition the

survey points into groups that maximized the difference in average flow-frequency. This

created several “zones” of equal flow-frequency that are identified by distinct soil,

substrate, and vegetation characteristics. Lastly, it was determined which of these zones

was analogous to the bankfull stage and effective discharge of adjacent alluvial channels,

based on a comparison of flow-frequency.

Field Surveys

Approximately 8 to 10 transects spanning from the channel into the adjacent

forest were surveyed in the immediate vicinity of each USGS gage. Along each transect

we surveyed the elevation of the moss, herb, grass, shrub, and tree closest to the water

level. This allowed us to capture the boundary of incipient vegetation growth and the

minimum flow-frequency associated with the establishment of each type of vegetation.

50

At each survey point we also noted the vegetation height and the accompanying substrate,

soil development, leaf litter abundance, and degree of canopy cover (Table 3.2).

Although vegetation type and vegetation height are not strictly independent, they were

separated because both short and tall communities of grasses and herbs were prevalent at

some lowland reaches.

The surveys were made during baseflow conditions in June 2006. The elevation

of each survey point was measured relative to the USGS stream gage, using a Sokkia

Total Station and reflector prism. A total of 309 points were surveyed at the nine stream

gages, or approximately 34 points per reach. Water surface slope was also surveyed in the

field by surveying the height of the water level throughout the length of the reach.

Estimation of Flow-Frequency

For each of the surveyed points, the corresponding discharge was determined

using the gage’s most current stage-discharge rating curve. The corresponding flow-

frequency of each discharge value was determined using two different metrics: flow-

duration and recurrence intervals. Flow-duration is the amount of time that a given

discharge threshold is met or exceeded and is a metric of the total duration that a surface

is inundated by water. The recurrence interval is the average number of individual

occurrences that exceed a threshold discharge and indicates the average time between

events. Both measures are needed to understand both the extent and regularity of high-

flows. Within this manuscript, flow-frequency and exceedance probability are

synonymous with flow-duration, and corresponding recurrence intervals are given where

applicable.

51

Table 3.2 Riparian features that were recorded at each survey point. Vegetation,

substrate, and soil characteristics were divided into numerical categories for multivariate

statistical analysis.

Riparian Features

Vegetation Type 0 - mosses 1 - herbs, ferns 2 - grasses 3 - shrubs 4 - trees (continuous cover

on boulder or bedrock)

(saplings included)

(both short and tall)

(woody stem, <2.5cm dbh)

(woody stem, >2.5cm dbh)

Vegetation Height 0 - short 1 - short/medium 2 - medium 3 - medium/tall 4 - tall

(<30cm) (30-60cm) (60-90cm) (90-120cm) (>120cm, includes some grasses)

Substrate 0 - soil, clay 1 - sand, silt 2 - gravel 3 - cobble, boulder 4 - bedrock

(0 - 1/256mm) (1/256 - 2mm) (2 - 64mm) (> 64mm)

Soil 0 - none 1 - discontinuous 2 - continuous (bare rock and/or

no soil) (some soil and/or some bare rock)

(stable, developed accumulation of soil)

Leaf Litter 0 - none 1 - discontinuous 2 - continuous (no litter) (litter present in

small patches) (continuous litter present)

Canopy Cover 0 - none 1 - partial shade 2 - full shade 3 - canopy tree (full light, no

canopy cover) (under canopy, but receives direct incident light)

(under closed canopy)

(a canopy dominant tree)

52

Flow-duration curves were constructed from the USGS approved discharge

records for each gage. While both daily records and instantaneous (15-minute) discharge

records were available, the instantaneous data were used because they capture the

magnitude and timing of peak flows in the flashy hydrologic regime of the Luquillo

Mountains. The Log-Pearson Type III (LP3) distribution was fit to the flow-duration

curves in this analysis (Water Resources Council 1981, Goodwin 2004). The LP3

distribution is given by the following probability density function and cumulative

distribution functions:

( )ε)λ(yexpΓ(β)

ε)(yλpdf(y)1ββ

−−−

=−

Eq. 3.1

( )dyε)λ(yexpΓ(β)

ε)(yλcdf(y)y

ε

1ββ

∫ −−−

=−

Eq. 3.2

where: y = ln(Q), Q = discharge (m3/s), and:

yσβ

=λ 2

yC2⎟⎟⎠

⎞⎜⎜⎝

⎛=β

λβ

−μ=ε y (Eqs. 3.3, 3.4, 3.5)

LP3 distribution parameters (λ, β, ε) for each gage were estimated from the

sample mean (μy), standard deviation (σy), and skew coefficient (Cy) of the natural

logarithm-transformed discharge records (Table 3.3).

Recurrence intervals were calculated using partial duration flood series (where the

entire hydrograph is considered) rather than annual maximum series (one peak discharge

value per year). This approach was preferred because of the abundance of intra-annual

floods that can modify riparian vegetation. Intra-annual floods were counted as long as

they were independent events (i.e. not part of the same rainfall event or influenced by

53

Tab

le 3

.3 T

he ti

me

span

of t

he d

isch

arge

reco

rd, t

he s

ampl

e m

ean

(μ),

stan

dard

dev

iatio

n (σ

), an

d sk

ew c

oeff

icie

nt (C

) of t

he n

atur

al

loga

rithm

-tran

sfor

med

dis

char

ge r

ecor

d (in

m3 /s

) us

ed f

or th

e Lo

g-Pe

arso

n Ty

pe I

II d

istri

butio

n, a

nd th

e co

effic

ient

s an

d ex

pone

nts

for a

t-sta

tion

wid

th (w

) and

dep

th (h

) hyd

raul

ic g

eom

etry

rela

tions

hips

of t

he fo

rm (w

=c1Q

b and

h=c

2Qf ).

D

istr

ibut

ion

Stat

istic

s (1

5-m

inut

e da

ta in

m3 /s

)

Hyd

raul

ic G

eom

etry

Mom

ents

Coe

ffici

ents

Ex

pone

nts

Gag

e N

ame

Star

t En

d

µy

σy

Cy

c 1

c 2

b

f

Rio

Esp

iritu

San

to n

r Rio

Gra

nde

7/27

/199

4 8/

20/2

006

-0

.21

0.94

1.

24

11

.4

0.36

0.

28

0.35

R

io F

ajar

do n

r Faj

ardo

10

/1/1

986

9/30

/200

6

-0.1

3 1.

10

0.74

14.4

0.

24

0.34

0.

22

Rio

Gra

nde

nr E

l Ver

de

8/16

/199

0 8/

20/2

006

-0

.68

1.00

1.

16

11

.4

0.33

0.

19

0.30

R

io M

amey

es a

t Mam

eyes

8/

1/19

97

6/1/

2006

0.23

0.

94

1.04

15.4

0.

32

0.17

0.

24

Rio

Mam

eyes

nr S

aban

a 10

/1/1

990

6/1/

2006

0.00

0.

76

1.24

12.7

0.

34

0.19

0.

34

Rio

Sab

ana

at S

aban

a 10

/1/1

990

8/20

/200

6

-1.3

8 1.

08

0.73

12.3

0.

33

0.17

0.

29

Que

brad

a S

onad

ora

nr E

l Ver

de

10/1

/199

4 8/

20/2

006

-2

.42

1.17

0.

73

4.

3 0.

58

0.28

0.

32

Que

brad

a G

uaba

nr N

agua

bo

6/23

/199

2 6/

1/20

06

-4

.65

0.86

1.

47

5.

5 0.

36

0.33

0.

29

Rio

Icac

os n

r Nag

uabo

7/

21/1

992

6/1/

2006

-1.3

3 0.

70

1.64

8.4

0.35

0.

34

0.31

54

saturation from previous storms). We followed general guidelines for identifying

independent peaks set forth by Lang et al. (2002), which suggest that independent flows

must be separated by a minimum of 5 days and accompanied by a drop below 75% of the

lower peak. However, due to the short duration of floods in these flashy streams, we

considered events to be independent if they were separated by at least 24 hours, and also

had a 75% drop between peaks.

Multivariate Regression Trees

Multivariate regression trees (MRT) are a statistical technique that can be used to

predict relationships between a response variable and multiple environmental

characteristics (De’ath 2002). MRT forms clusters or groups by repeated splitting of the

data, with each split defined by a simple rule based on the predictor variables. The splits

are chosen to minimize the dissimilarity of data within clusters, or maximize the

differences between clusters. The groups or clusters formed by MRT are defined by a

simple splitting of one environmental variable at a time, generating an intuitive and easily

interpreted decision tree. We used MRT to define groups of equal flow frequencies based

on the environmental variables measured in the field (Table 3.2). The flow-frequency of

each surveyed point was used as the response variable, after a logarithm-transformation

was used to reduce skewness and achieve a normal distribution. Splits were based on the

vegetation type, vegetation height, substrate size, soil development, leaf litter, canopy

cover, and reach location. This procedure ultimately generated clusters of riparian

features that occur at distinct flow frequencies.

Separate regression trees were performed for the data collected at the alluvial sites

and the mid-elevation/steepland sites. One reach, Quebrada Guaba near Naguabo, was

55

removed from the MRT analysis due to anomalous flow-frequencies. Further supporting

its removal from this analysis, this gage also has the smallest drainage area, a closed

canopy that reduces the regeneration of channel side vegetation, the most flashy

hydrograph, subsurface drainage through sandy substrate, and riparian features that occur

above the boundary of apparent common floods.

Effective Discharge

The effective, or channel-forming discharge, and its flow frequency, was

estimated for the alluvial and mid-elevation reaches so that this frequency could be

compared to frequencies of the clusters formed by MRT in the steepland channels. The

effective discharge is defined as the discharge that transports the most sediment over time

and is quantified as the discharge where the product of the frequency of discharge and the

magnitude of sediment transport (the relative effectiveness curve) is a maximum

(Wolman and Miller 1960). Sediment transport here is quantified by the bedload

discharge, because the gravel-, cobble-, and boulder-bedded channel form of these rivers

is fundamentally determined by the bedload rather than suspended sediment (Knighton

1998). Bedload discharge is usually estimated using either a bedload sediment rating

curve or similar threshold-based function of fluvial discharge (Emmett and Wolman

2001, Torizzo and Pitlick 2004). This approach has been challenged by some authors

(Lenzi et al. 2006a) because treating sediment transport as a continuous function of water

discharge does not consider the variation in sediment supply over time, the impulsive and

pulsating nature of sediment discharge, and the dramatic increase in transport when some

discharge thresholds are passed. However, empirical bedload transport formulas based on

56

discharge have been shown to accurately predict sediment yields in these and other

steepland drainages in Puerto Rico (Simon and Guzman-Rios 1990), and were used here.

The bedload sediment transport curve for each alluvial reach was estimated

according to the Meyer-Peter and Muller (1948) relationship. The Meyer-Peter & Muller

sediment transport equation relates sediment yield to channel width, unit boundary shear

stress, and critical shear stress:

2/3c2/3s )τ(τ

)gρ1(s8wQ −

−= for )cτ(τ − > 0, Qs = 0 otherwise Eq. 3.6

where: Qs = sediment yield (m3/s), w = channel width (m), s = sediment density ratio

(dimensionless, 2.65), g = acceleration due to gravity (9.8m/s2), ρ = specific weight of

water (1000kg/m3) τ = unit boundary shear stress (Pa), τc = critical boundary shear stress

(Pa)

The unit boundary shear stress and critical boundary shear stress can be estimated

according to the following relationships, assuming steady, uniform flow:

gRSρ=τ (depth-slope product) Eq. 3.7

50c*

c gd)1s( ρ−τ=τ Eq. 3.8

where: R = hydraulic radius (m), S = slope (m/m), τ*c = critical dimensionless shear stress

(0.030 for alluvial streams, 0.045 for mid-elevation streams) following a positive

relationship between slope and τ*c discussed in (Mueller et al., 2005), d50 = median grain

size (m)

Data on the width and hydraulic radius at varying discharges were obtained from

the USGS measurements. The following at-a-station hydraulic geometry power

relationships were estimated for each site:

57

b1Qcw = Eq. 3.9

f2QcR = ` Eq. 3.10

where: Q = discharge (m3/s), and c1,c2, b, and f are coefficients and exponents empirically

derived from a logarithm-transformed linear regression.

The estimates for unit boundary shear stress, critical boundary shear stress, and

hydraulic geometry relationships for width and depth were substituted into the Meyer-

Peter and Muller relationship. This yields an equation relating the bedload discharge (Qs,

m3/s) to the flow discharge, median particle size, and slope:

2/350

*c

f22/3

b150s )gd)1s(S)Qc(g(

g)1s(8Qc)S,d,Q(Q ρ−τ−ρρ−

= Eq. 3.11

The relative effectiveness function, Ф, is the product of the probability density

function (LP3 flow duration curve), pdf(Q), and the bedload sediment transport curve, Qs

(Figure 3.4). This relative effectiveness function represents the amount of sediment

transported over time, and the effective discharge is the discharge where this function is a

maximum (i.e. derivative of the function is 0), such that:

Ф = pdf(Q)*Qs Eq. 3.12

0dQdΦ Qeffective = Eq. 3.13

The effective discharge was estimated for only the alluvial and mid-elevation

streams because our initial analysis and other studies (Torizzo and Pitlick 2004, Lenzi et

al. 2006a) indicate these sediment transport equations were not considered appropriate for

the boulder-lined steepland streams. Parameters used for each site in the calculation of

58

Figure 3.4 The effective discharge occurs at the maximum of the relative effectiveness

curve that is generated by multiplying the flow duration and sediment transport functions.

In this illustration, using data from the Río Mameyes near Sabana gage, the effective

discharge is roughly coincident with the presence of woody shrubs and trees.

59

the flow duration curves, bedload sediment transport equation, and the relative

effectiveness function are listed in Table 3.3.

RESULTS

Riparian Vegetation

The average elevation of the first occurrence of the different riparian vegetation

types (mosses, grasses, herbs, shrubs, trees) display a consistent zonation with elevation

along the cross-sectional profile of the channel, as is illustrated for one of the sites in

Figure 3.3. Canopy cover, the abundance of leaf litter, and soil development also increase

from the channel to the adjacent forest. Our surveys and observations in other streams in

the region indicate that this zonation is best developed along channels that have open or

partially open canopies where there is sufficient light for grasses and herbaceous

vegetation to establish and also sufficient shade for the development of cushion mosses.

While local environmental conditions (e.g. light, substrate, hydraulic shielding) can

constrain the establishment of vegetation at any particular location, the average elevation

of the first occurrence of different vegetation forms, litter cover, and soil development is

consistently related to the frequency of flow inundation both within and between sites.

Moreover, mosses, herbs, and grasses start establishing at elevations that are inundated

weekly or monthly, and are slightly above the baseflow water level. Shrubs and trees are

present at higher stages where they are inundated, at least briefly, several times a year.

The average elevation of the first occurrence of the different vegetation types is

also related to sediment transporting flows and the effective discharge, as illustrated by

data from the one of the Rio Mameyes sites (Figure 3.4). Mosses and short grasses

60

dominate at a stage below the threshold of sediment transport but above the most frequent

flows (the peak in the probability density function of discharge). That is, they first occur

at stages that are frequently inundated by sub-effective flows. Herbs first occur at a stage

associated with intermediate discharges that are around the threshold for sediment

mobilization. Woody shrubs and trees establish at a less frequent discharge that is

coincident with the effective discharge of bedload sediment. The greatest variation in the

vegetation types occurs around the threshold for sediment mobilization, where grasses,

herbs, shrubs, and trees commonly occur together.

Bankfull and Effective Discharge

The median exceedance probability of the bankfull (morphologic) discharge at the

four alluvial sites was found to be 0.16% and had a corresponding median recurrence

interval of 50 days (Table 3.5). The median exceedance probability of the calculated

effective discharges was 0.20% and had a corresponding median recurrence interval of 39

days. The flow-frequency of both bankfull and effective discharge for the alluvial sites

were not significantly different (Student’s t-test, P > 0.1). This confirms the assumption

that the bankfull stage is coincident with the channel-forming discharge in this region.

For comparison, the average flow-frequency of the mean annual discharge among all

sites is 23%; far more frequent than both the bankfull and effective discharge.

Multivariate Regression Trees

The data at the alluvial sites was first partitioned into two clusters: tall vegetation

(120cm in height or greater), and short vegetation (stature shorter than 120cm) (Figure

3.5). The cluster defined by short vegetation has a median exceedance probability of

8.5% and is inundated more frequently than the tall vegetation cluster (0.60%). Within

61

both clusters, the data were further split into two divisions based on the same

characteristics: clay sized substrate (including soil), and substrate coarser than clay. Thus,

the data for alluvial sites were effectively divided into four clusters: A1) short vegetation

with coarse substrate (median exceedance probability = 14.2%), A2) short vegetation

with clay/soil (0.96%), A3) tall vegetation with coarse substrate (0.84%), and A4) tall

vegetation with clay/soil (0.30%). Based on a comparison of means from logarithm-

transformed data (using Tukey’s HSD test), clusters A1 and A4 were found to be

significantly different from each other and clusters A2 and A3 (P < 0.01). Clusters A2

and A3 were not significantly different from each other (P > 0.1). Similarly, the

frequency of bankfull discharge, effective discharge, and the cluster defined by tall

vegetation growing on soil or clay (A4) were not significantly different (P > 0.1). This

suggests that in these alluvial streams, the morphologic bankfull stage and the effective

channel forming discharge is coincident with the presence of tall vegetation and soil

development along the banks.

The vegetation and substrate data for the steepland sites separated into clusters

that were based on similar factors as the data for alluvial sites (Figure 3.6). The MRT

analysis first split the data into two clusters based on the presence of soil. Continuous and

discontinuous soil formed one cluster (0.25%), and the absence of soil formed the other

(1.73%). Moreover, the cluster with soil had a lower flow-frequency than the cluster

without soil. These two clusters were both further partitioned by vegetation type:

presence of trees and shrubs, and presence of only herbs, mosses, and grasses. Hence, the

data for steepland sites were divided into the following four clusters: S1) no soil and

herbs/mosses/grasses (2.74%), S2) no soil and trees/shrubs (0.21%), S3) soil and

62

Figure 3.5 Box plots of the flow frequency for surveyed data points at alluvial sites that

have been partitioned into clusters based on vegetation, substrate, and soil characteristics

using a multivariate regression tree technique (see section 4.3). The means of the

logarithm-transformed flow-frequency data in cluster A4 (tall vegetation and soil),

bankfull discharge, and effective discharge (all marked by an asterisk) are not

significantly different.

63

Figure 3.5 cont.

64

Figure 3.6 Box plots of the flow frequency for surveyed data points at mid-elevation and

steepland sites that have been partitioned into clusters based on vegetation, substrate, and

soil characteristics using a multivariate regression tree technique. Based on a comparison

of flow-frequency, the means of the logarithm-transformed data in clusters S2 (soil

absent / shrubs and trees), and S4 (soil present / shrubs and trees) are not significantly

different from the means of logarithm-transformed data in clusters A4 (tall vegetation and

soil), bankfull discharge, and effective discharge (all marked by asterisk) at adjacent

alluvial sites that are shown in Figure 5.

65

Figure 3.6 cont.

66

herbs/mosses/grasses (1.14%), and S4) soil and trees/shrubs (0.13%). Further subdivision

and partitioning in both models did not create additional clusters that were statistically

different.

It is also important to acknowledge that the location of the reach was not chosen

as a split within these models, therefore the clusters are applicable to all sites. The model

coefficient of determination (r2) for the MRT, based on cross-validation, for alluvial sites

is 0.54, indicating that 54% of the variance among all surveyed points was explained by

the division into these clusters. For steepland sites, the model r2 is 0.45. However, at any

given site, the division of data into these clusters accounted for a higher proportion of the

variance (average at-site r2 = 0.63). This indicates that although aggregating data between

sites introduces error that is not necessarily present at a given site, the clusters can be

compared among sites.

To determine the characteristic features of the steepland sites that are analogous to

the bankfull stage in alluvial streams, the flow frequency of each of the steepland clusters

(S1-S4) were systematically tested against the clusters developed for the bankfull flow-

frequency in the alluvial sites (Table 3.4). The comparison of means of the log-

transformed flow frequency between each zone using Tukey’s HSD test indicated that the

steepland zone S2 (no soil, trees/shrubs), zone S4 (soil, trees/shrubs), zone A4 (tall

vegetation and clay/soil, for alluvial sites), bankfull discharge, and effective discharge are

not significantly different (P > 0.1). On the basis of flow-frequency, the zone defined by

soil development and the presence of woody shrubs and trees in steepland sites (zone S4)

was most analogous to the bankfull stage in alluvial streams. The bankfull stage for the

alluvial streams occurs on average 1.3m (±0.09m, 1 S.D.) above the baseflow water level,

67

Tab

le 3

.4 T

he m

edia

n he

ight

abo

ve w

ater

tabl

e, u

nit d

isch

arge

, flo

w fr

eque

ncy,

and

recu

rren

ce in

terv

al o

f eac

h zo

ne d

efin

ed b

y th

e

mul

tivar

iate

regr

essi

on tr

ee a

naly

sis.

Zone

s A

1-A

4 ar

e ba

sed

on s

urve

y da

ta p

oint

s on

the

4 al

luvi

al s

tream

gag

es, w

here

as z

ones

S1-

S4 a

re b

ased

on

data

from

the

4 st

eepl

and

gage

s.

Zone

R

ipar

ian

Feat

ures

# su

rvey

po

ints

#

Gag

es

Hei

ght

abv

Bas

eflo

w

(m)

Med

ian

Uni

t Q

(m3 /s

/km

2 )

Med

ian

Freq

uenc

y (%

tim

e)

Med

ian

Rec

urre

nce

(day

s)

A1

Sho

rt V

eget

atio

n, C

oars

e S

ubst

rate

84

4

0.25

0.

08

14.2

%

8 A

2 S

hort

Veg

etat

ion,

Soi

l/Cla

y 19

4

0.56

0.

64

0.96

%

19

A3

Tall

Veg

etat

ion,

Coa

rse

Sub

stra

te

21

4 0.

71

0.74

0.

84%

21

A

4 Ta

ll V

eget

atio

n, S

oil/C

lay

29

4 0.

99

1.41

0.

30%

39

S

1 N

o S

oil,

Mos

s/H

erbs

79

4*

0.

25

0.41

2.

74%

10

S

2***

N

o S

oil,

Tree

s/S

hrub

s 20

4*

0.

80

2.01

0.

21%

41

S

3 S

oil,

Mos

s/H

erbs

18

4*

0.

32

0.88

1.

14%

14

S

4***

S

oil,

Tree

s/S

hrub

s 39

4*

0.

92

2.51

0.

13%

55

B

ankf

ull

4

4 1.

27

2.29

0.

16%

50

E

ffect

ive

4

4 1.

20

1.61

0.

20%

39

* da

ta fr

om Q

. Gua

ba re

mov

ed fr

om a

naly

sis

***

zone

not

sig

nica

ntly

diff

eren

t fro

m b

ankf

ull i

n al

luvi

al c

hann

els

base

d on

a T

ukey

's H

SD

test

, α =

0.0

5 68

Tab

le 3

.5 T

he d

isch

arge

(Q, m

3 /s),

unit

disc

harg

e (U

nit Q

, m3 /s

/km

2 ), flo

w fr

eque

ncy

(Fre

q., %

tim

e), a

nd re

curr

ence

inte

rval

(Rec

.,

days

) ass

ocia

ted

with

the

bank

full

stag

e (f

or a

lluvi

al si

tes)

, the

eff

ectiv

e di

scha

rge

(for

low

land

and

mid

-ele

vatio

n si

tes)

, and

the

bank

full

anal

og th

at is

def

ined

in th

is m

anus

crip

t.

Ban

kful

l

Effe

ctiv

e

Ban

kful

l Ana

log

Gag

e N

ame

Q

U

nit

Q

Freq

. R

ec.

Q

U

nit

Q

Freq

. R

ec.

Zo

ne

Q

Un

it Q

Fr

eq.

Rec

.

Rio

Esp

iritu

San

to n

r Rio

Gra

nde

61

.5

2.7

0.14

%

42

25

.7

1.1

0.51

%

16

A

4 33

.7

1.5

0.34

%

26

Rio

Faj

ardo

nr F

ajar

do

96

.9

2.5

0.08

%

90

65

.3

1.7

0.16

%

46

A

4 55

.4

1.4

0.21

%

43

Rio

Gra

nde

nr E

l Ver

de

39

.7

2.1

0.18

%

57

63

.2

3.3

0.09

%

98

A

4 30

.1

1.6

0.27

%

48

Rio

Mam

eyes

at M

amey

es

46

.2

1.3

0.31

%

27

54

.0

1.5

0.24

%

33

A

4 38

.0

1.1

0.42

%

30

Rio

Mam

eyes

nr S

aban

a

---

---

---

---

37

.1

2.1

0.12

%

50

S

4 75

.4

4.2

0.03

%

92

Rio

Sab

ana

at S

aban

a

---

---

---

---

6.

9 0.

7 0.

70%

18

S4

23.4

2.

3 0.

09%

48

Que

brad

a S

onad

ora

nr E

l Ver

de

--

- --

- --

- --

-

---

---

---

---

S

4 6.

7

2.6

0.27

%

40

Que

brad

a G

uaba

nr N

agua

bo

--

- --

- --

- --

-

---

---

---

---

S

4 2.

0

6.6

0.03

%

346

Rio

Icac

os n

r Nag

uabo

---

---

---

---

--

- --

- --

- --

-

S4

9.2

2.

8 0.

16%

50

69

has a corresponding unit discharge of 2.3m3/s/km2 (±0.62), a flow frequency of 0.16%

(±0.09), and an average recurrence interval of 50 days (±0.31). The analogous steepland

zone that corresponds to the first occurrence of soil and woody vegetation occurs on

average 0.92m (±0.31) above the baseflow water level, has a corresponding unit

discharge of 2.5 m3/s/km2 (±0.85), a flow frequency of 0.13% (±0.10), and an average

recurrence interval of 55 days (±23).

To determine the consistency of this bankfull analog throughout the stream

network, the bankfull analog was compared across all the sites (Table 3.5). For 8 of the 9

gaged channels, the bankfull analog occurs at a flow frequency ranging from 0.03% and

0.42% (median = 0.24%), with a recurrence interval between 26 and 92 days (median =

46 days). There was no significant relationship between drainage area and flow-

frequency of the bankfull analog (r2 = 0.15, P > 0.1), suggesting that it does not vary

systematically with area throughout the stream network. The only site that is dramatically

different in terms of flow frequency is Quebrada Guaba, with a flow frequency of 0.03%

and a recurrence interval of 346 days. This is also the smallest stream, and suggests that

the technique of flow frequency estimation based on riparian features may not work for

the 1st order channels. However, as evidence by the other 8 reaches, the bankfull analog is

consistent in flow frequency throughout the stream network, albeit with local variation.

DISCUSSION

Riparian Features

The multiple regression trees developed for both the alluvial and steepland sites

generated several statistically significant clusters of points, based on the first occurrence

70

of riparian features, that have different average flow-frequencies. The alluvial sites were

partitioned on the basis of vegetation height (tall vs. short), and substrate size (soil/clay

vs. coarser substrate). The steepland sites were partitioned on the basis of soil

development (present or absent) and vegetation type (moss, herbs, grass, shrubs, trees).

The partitioning of both upland and alluvial sites suggests that the first occurrence of

short-stature vegetation (herbs, mosses, and grasses) occurs on coarse substrates and in

areas that are inundated by flows of moderate frequency and intensity. Moreover, the

zones defined by these features are associated with the mean annual discharge and sub-

effective flows. In contrast, tall, mature, woody vegetation and soil development are

related to less frequent flows of higher-magnitude that approximate the effective

discharge. Furthermore, these features are related to the bankfull and effective flows of

alluvial streams and can be used to identify a bankfull analog in steepland streams.

Because this analog occurs at a relatively constant flow-frequency throughout the stream

network, this analog can be used to determine channel boundaries at ungaged reaches in

the region.

The characteristics of the two zones that are based on riparian features that

estimate bankfull discharges at both the alluvial and steepland sites are remarkably

similar. In the alluvial sites, bankfull stage is associated with the first occurrence of tall

vegetation (shrubs, trees and some grasses) and clay substrate/soil. In the steepland sites,

the bankfull analog is also marked by the first occurrence of woody shrubs/trees and soil

development. However, the MRT analysis indicates that in the alluvial sites, vegetation

height was found to be more important than vegetation type in the cluster divisions.

Conversely, in the steepland sites, vegetation type was more important than vegetation

71

height. This difference is due to the large proportion of different grass species found at

open alluvial sites. Despite being all considered the same vegetation type, the grasses

differ in height (some greater than 120cm) according to their proximity to the channel, so

that vegetation height more strongly reflects flow frequency than does vegetation type. In

contrast, the type of vegetation was more important in forming clusters for steepland sites

because there is less dominance of one vegetation type over the rest at these sites. This

difference in vegetation between alluvial and steepland reaches is driven by the fact that

the alluvial reaches are higher order streams, lower in elevation, have greater incident

light, and are generally surrounded by non-forest land-use (mostly pastures and rural

development areas).

However, it appears that this analog does not apply as well to small 1st order

streams. In these small channel and swales, the riparian features are more influenced by

local factors than by fluvial disturbance. It should also be noted that there is a large

amount of variability within any given zone defined by riparian features. Flow

frequencies within a cluster can span an order or two of magnitude, which represent

drastically different flood magnitudes. The large degree of natural variability is

responsible for this variance. Although the first occurrence of vegetation that was

surveyed along each transect is primarily influenced by the frequency of flooding, the

exact stage where the vegetation grows relative to the stream channel is also influenced

by local factors such as hydraulic shielding by boulders, differences in light, and

substrate stability. These small differences in height translate into a larger difference in

flow-frequency and generate a large degree of natural noise. Fortunately, the repeated

measure of different vegetation types in a reach reduces this variation and provides a

72

reliable estimate of the corresponding flow frequency. This analysis also indicates that it

is valid to identify the high-flow riparian features at gaging stations using the techniques

identified in this paper and then identifying them at non-gaged reaches, much like

bankfull morphology is identified in alluvial reaches. Field identification at non-gaged

sites is reasonably accurate and precise because the first occurrence of woody vegetation

and soil is easily recognizable.

Bankfull and Effective Discharge

The recurrence intervals in this study that are associated with both the bankfull

discharge in alluvial channels and the bankfull analog in steepland channels are between

40 and 90 days. This range is significantly more frequent than commonly reported values

of 1-3 years (Wolman and Miller 1960, Dunne and Leopold 1978, Rosgen 1994,

Knighton 1998). The difference is due both to the methodology used and the flashy

nature of these streams. Recurrence intervals presented here were calculated according to

a partial duration series analysis using 15-minute instantaneous discharge data, rather

than an annual maximum series and/or daily discharge records used in many studies. By

definition, the annual maximum series used in these classic publications is drawn from

one annual peak per year, thus forcing recurrence intervals greater than or equal to 1 year.

While annual maximum series analysis may be pertinent for large temperate basins, it

fails to capture the intra-annual flows that are responsible for structuring the vegetation in

and adjacent to the channels in these flashy and relatively small streams. Although some

of these publications acknowledge that bankfull discharge can be observed “several times

per year” (Rosgen 1994) in many rivers, little guidance is given to assess the recurrence

interval of multiple flows per year. A partial duration series captures the many flow-

73

events over the bankfull threshold found in these flashy streams because it allows for

multiple floods each year. Had an annual maximum series been used in this study,

bankfull stage would be exceeded by the peak flow in every year (recurrence interval of

1.0 year). Furthermore, if daily discharge data (rather than 15-min data) had been used in

this study to calculate recurrence intervals (using partial duration series), the frequent

short-lived peaks would be damped such that only one or two events per year would have

exceeded the bankfull threshold (recurrence interval = 0.5 to 1 year).

Despite differences in recurrence intervals between this study and others, the

flow-duration values of bankfull reported in this study are comparable to other studies.

We report average bankfull flow durations of ~0.27%, or approximately 1 day of

inundation per year. For comparison, the bankfull duration of streams in England and

Wales is reported to be 0.60% (2 days per year) (Nixon 1959), and between 0.4% to 3.0%

(1.5 to 11 days per year) in mountain streams in Colorado and Wyoming (Andrews

1980). Dunne and Leopold (1978) asserted that bankfull flow duration often varies

between 1.3% and 4.5%, with an average of 2.1% (8 days per year). The bankfull stage

here is inundated slightly less total time than the rivers mentioned in these other studies,

although there are more floods of bankfull magnitude per year. This suggests that

bankfull formation among these different rivers is related more to the total amount of

time bankfull stage is exceeded rather than the timing between floods.

The similarity of bankfull on the basis of duration among very different rivers

suggests common organizing principles of these systems. This organizing principle may

be related to the energy expenditure and geomorphic effectiveness of large floods. Costa

and O’Connor (1995) assert that the geomorphic effectiveness and total energy

74

expenditure of a flood is dependent on both the duration and peak stream power per unit

area. Flood peaks in mountain streams are brief (hours in duration) and strong, reflecting

rapid movement of water down steep hillslopes and channels (Swanson et al. 1998).

While these floods are intense, they are short-lived so that several events of bankfull

magnitude each year may be required for channel and riparian zone maintenance. In

contrast, floods in large lowland basins often inundate the floodplain for up to several

days, but may only occur once a year. The floods in the Luquillo Mountains are

characterized as having a short duration, but high peak stream power, and thus

intermediate total energy expenditure. Consequently, they may have a similar

geomorphic effectiveness as floods in other basins that are longer in duration, but have

lower peak unit stream power.

However, these geomorphically effective flows are not necessarily channel-

forming in the steepland boulder and bedrock-lined reaches. The channel-forming

discharge in bedrock mountain rivers may be higher than the effective discharge of

sediment transport due to the high threshold of stream power required for bedrock

incision and movement of large-sized boulders (Costa and O’Connor 1995). There is also

discussion in the literature that the effective discharge is not necessarily a discrete value,

but rather a range of flow-events that are responsible for the greatest amount of

geomorphic work (Goodwin 2004). Here we are only using effective discharge as a guide

to the magnitude of the flow-frequency of the channel-altering events. Some researchers

have even posited that there are two dominant discharge ranges for steep mountain rivers:

a relatively frequent flow responsible for maintaining channel forms, and a more

infrequent high flow responsible for large-scale channel shaping (Lenzi et al. 2006b). On

75

this token, an extremely rare multi-year recurrence flood in these streams may be

responsible for bedrock incision and mass sediment movement, but it is the more frequent

effective discharge that consistently affects riparian vegetation, defines channel

boundaries, and is responsible for channel-maintenance.

The same floods that are geomorphically effective are also responsible for

maintaining riparian vegetation patterns, as evidenced by the relationship between the

effective discharge in alluvial and mid-elevation channels and the boundary of woody

vegetation in this study. The channel-maintaining discharge based on bedload transport is

at the same magnitude as the threshold for woody vegetation, as both are apparently

influenced by similar periodic flows. Since these two concepts are so tightly linked, the

bankfull analog can be used as a constant marker of effective flows. Downstream

hydraulic geometry requires that the bankfull and effective discharge occur at a constant

frequency throughout a basin (Leopold and Maddock 1953). Because the bankfull analog

defined in this study marks a boundary of equal flow frequency and is also consistent

with the supposed channel-maintaining discharge, it can be used as a channel boundary

marker for downstream hydraulic geometry studies to compare channel cross-section

geometry throughout the stream network.

Applicability to other stream systems

The results of this study are directly applicable to streams in the Luquillo region

of Puerto Rico. However, the techniques of quantifying flow frequencies associated with

different types of vegetation and riparian features used here should be applicable in a

range of environments. It is expected that streams with both a similar flashy flow-regime

76

and a humid tropical climate with similar vegetation types will have analogous riparian

features occurring at flow-frequencies consistent with those determined in this study.

Riparian features have also been noted as indicators of ‘bankfull’ in a variety of

temperate montane systems, including the South Island of New Zealand (Wohl and

Wilcox 2005), the Rocky Mountains (Wohl et al. 2004), the Pacific Northwest

(Montgomery and Gran 2001), and the alpine region of Poland (Radecki-Pawlik 2002).

Each of these streams are in different physiographic provinces. Some are humid

environments with rapidly growing vegetation, whereas others, such as parts of the

Rocky Mountains, are semi-arid environments with limited vegetation. It would be

spurious to assume that the features that approximate bankfull found in this study occur at

the same flow-frequency as those in temperate mountain streams. However, it does

indicate that similar features based on soil, substrate, and vegetation do apply to other

systems and that local/regional flow frequency zones can be identified using techniques

outlined here. Nevertheless, similar studies are needed on other gaged streams to

determine the flow frequency associated with such riparian features in different

environments.

CONCLUSIONS

In the study area of the Luquillo Mountains, we used a network of stream gages to

determine a zone of constant flow-frequency, based on riparian features, in steepland

streams where a bankfull stage was absent. The results indicated that in these steepland

streams, the discharge associated with the average first occurrence of soil and woody

vegetation has the same flow-frequency as the bankfull discharge of adjacent alluvial

77

streams. Likewise, the elevations associated with the first occurrence of woody

vegetation in alluvial streams can be similarly used to estimate bankfull. The results also

indicate that bankfull stage, effective discharge, and presence of tall vegetation and a

clayey substrate all occurred at a stage associated with a discharge that is exceeded the

same amount of the time. Thus, throughout the stream network, high-flow riparian

features such as the presence of soil development and perennial vegetation can provide a

common benchmark of flow-frequency. Furthermore, the general approach of surveying

the first occurrence of riparian features and using multivariate statistical analysis linking

these occurrences to 15-minute flow duration can provide an internally consistent

framework for identifying flow frequencies within a region.

ACKNOWLEDGMENTS

The authors thank Ellen Wohl, Tamara Heartsill-Scalley, Jerry Mead, and

Douglas Jerolmack for their strengthening comments on earlier versions of this

manuscript. We also thank the International Institute of Tropical Forestry for logistical

support. Funding for this study was provided by the National Science Foundation

Biocomplexity Grant (NSF #030414)—Rivers, Roads, and People: Complex Interactions

of Overlapping Networks in Watersheds.

78



85: 157-169.

Andrews ED. 1980. Effective and bankfull discharges of streams in the Yampa River

basin, Colorado and Wyoming. Journal of Hydrology 46: 311-330.

Arnold AE. 1996. Influence of microenvironment on growth and nutrient dynamics of

three herbaceous species in the Bisley Experimental Watersheds of Luquillo

Experimental Forest, Puerto Rico. M.S. Thesis, University of Tennessee,

Knoxville.

Bendix J. 1999. Stream power influence on southern California riparian vegetation.

Journal of Vegetation Science 10: 243-252.

Bendix J, Hupp CR. 2000. Hydrological and geomorphological impacts on riparian plant

communities. Hydrological Processes 13: 2977-2990.

Brown KA, Scatena FN, Gurevitch J. 2006. Effects of an invasive tree on community

structure and diversity in a tropical forest in Puerto Rico. Forest Ecology and

Management 226: 145-152.

Chopin DM, Beschta RL, Shen HW. 2002. Relationships between flood frequencies and

riparian plant communities in the upper Klamath basin, Oregon. Journal of the

American Water Resources Association 38: 603–617.

Clark JJ. 1997. Effects of land use change on northeastern Puerto Rican rivers. PhD

Dissertation. Johns Hopkins University.

79


northeastern Puerto Rico. GSA Bulletin 112: 1763-1777.

Costa JE, O’Connor JE. 1995. Geomorphically effective floods. In: Natural and

Anthropogenic Influences in Fluvial Geomorphology. Costa JE, Miller AJ, Potter

KW, Wilcock PR (eds). American Geophysical Union Geophysical Monograph

89: Washington, DC; 45-56.

De’ath G. 2002. Multivariate regression trees: a new technique for modeling species-

environment relationships. Ecology 83: 1105-1117.

Dunne T, Leopold LB. 1978. Water in environmental planning. W.H. Freeman and

Company: San Francisco, CA.

Emmett WW, Wolman MG. 2001. Effective discharge and gravel-bed rivers. Earth


García-Martinó AR, Warner GS, Scatena FN, Civco DL. 1996. Rainfall, runoff, and



Goodwin P, 2004. Analytical solutions for estimating effective discharge. Journal of

Hydraulic Engineering 130: 729-738.


flow and large floods. American Journal of Science 275: 825-847.



301-315.

80





136.

Harrelson CC, Rawlins CL, Potyondy JP. 1994. Stream channel reference sites: an

illustrated guide to field technique. USDA Forest Service Rocky Mountain Forest

and Range Experiment Station General Technical Report RM-245: Fort Collins,

CO.

Heartsill-Scalley T, Aide TM. 2003. Riparian vegetation and stream condition in a

tropical agricultural-secondary forest mosaic. Ecological Applications 13: 225-

234.

Heartsill-Scalley T. 2005. Characterization of riparian-zone vegetation and litter

production in tropical, montane rainforest along an environmental gradient. PhD

Dissertation. Utah State University.

Heartsill-Scalley T, Scatena FN, Estrada C, McDowell WH, Lugo AE. 2007. Disturbance

and long-term patterns of rainfall and throughfall nutrient fluxes in a subtropical

wet forest in Puerto Rico. Journal of Hydrology 333: 472-485.

Hupp CR, Osterkamp WR. 1985. Bottomland vegetation distribution along Passage

Creek, Virginia, in relation to fluvial landforms. Ecology 66: 670-681.

Knighton D. 1998. Fluvial Forms & Processes: A New Perspective. Oxford University

Press: New York, NY.

81

Lang M, Ouarda TBMJ, Bobee B. 1999. Towards operational guidelines for over-

threshold modeling. Journal of Hydrology 225: 103-117.

Larsen MC. 1997. Tropical geomorphology and geomorphic work: a study of

geomorphic processes and sediment and water budgets in montane humid-tropical

forested and developed watersheds, Puerto Rico. PhD Disseration. University of

Colorado.

Larsen MC, Torres-Sanchez AJ, 1998. The frequency and distribution of recent

landslides in three montane tropical regions of Puerto Rico. Geomorphology 24:

309-331.

Lenzi MA, Mao L, Comiti F. 2006a. When does bedload transport begin in steep boulder-

bed streams?. Hydrological Processes 20: 3517-3533.

Lenzi MA, Mao L, Comiti F. 2006b. Effective discharge for sediment transport in a

mountain river: Computation approaches and geomorphic effectiveness. Journal




252.

Leopold LB, Wolman MG, Miller JP. 1964. Fluvial Processes in Geomorphology. W.H.

Freeman and Company: San Francisco, CA.

Leopold LB, Skibitzke HE. 1967. Observations on unmeasured rivers. Geografiska

Annaler 49: 247-255.

82

Lugo AE, Scatena FN. 1995. Ecosystem-level properties of the Luquillo Experimental

Forest with emphasis on the Tabonuco Forest. In: Tropical Forests: Management

and Ecology. Lugo AE, Lowe C (eds). Ecological Studies 112: New York, NY;

59-108.

Meyer-Peter E, Muller R. 1948. Formulas for bedload transport. In: Proceedings of the

Second Meeting of the International Association of Hydraulic Structures

Research: Delft, The Netherlands; 39-64.

Montgomery DR, Gran KB. 2001. Downstream variations in the width of bedrock

channels. Water Resources Research 37: 1841-1846.

Mueller ER, Pitlick J, Nelson JN. 2005. Variation in the reference Shields stress for bed

load transport in gravel-bed streams and rivers. Water Resources Research 41,

W04006, doi:10.1029/2004WR003692.

Naiman RJ, Fetherston KL, McKay S, Chen J. 1998. Riparian forests. River Ecology and

Management: Lessons from the Pacific Coastal Ecoregion. Naiman RJ, Bilby RE

(eds). Springer-Verlag: New York; 289-323.

Navratil O, Albert MB, Hérouin E, Gresilllon JM. 2006. Determination of bankfull

discharge magnitude and frequency: comparison of methods on 16 gravel-bed

river reaches. Earth Surface Processes and Landforms 31: 1345-1363.

O’Connor PJ, Covich AP, Scatena FN, Loope LL. 2000. Non-indigenous bamboo along

headwater streams of the Luquillo Mountains, Puerto Rico: leaf-fall, aquatic leaf

decay, and patterns of invasion. Journal of Tropical Ecology 16: 499-516.

83

Poff NL, Allan JD, Bain MB, Karr JR, Prestegaard KL, Richter BD, Sparks RE,

Stromberg JC. 1997. The natural flow regime. A paradigm for river conservation

and restoration. Bioscience 47: 769-784.

Radecki-Pawlik A. 2002. Bankfull discharge in mountain streams: theory and practice.


Ramirez A, Melendez-Colom E. 2003. Meteorological summary for El Verde Field

Station: 1975-2003. Available at http://luq.lternet.edu/data.

Riley SJ. 1972. A comparison of morphometric measures of bankfull. Journal of

Hydrology 17: 23-31.

Rosgen DL. 1994. A classification of natural rivers. Catena 22: 169-199.

Scatena FN. 1990. Selection of riparian buffer zones in humid tropical steeplands. In:

Research Needs and Applications to Reduce Erosion and Sedimentation in

Tropical Steeplands (Proceedings of the Fiji Symposium, June 1999). IAHS-

AIHS Publication 192, International Association of Hydrological Sciences:

Washington, DC; 35-47.





D.C.; 103-111.



84

Scatena FN, Lugo AE. 1995. Geomorphology, disturbance, and the soil and vegetation of

two subtropical wet steepland watersheds of Puerto Rico. Geomorphology 13:

199-213.

Scatena FN, Moya S, Estrada C, Chinea JD. 1996. The first five years in the

reorganization of aboveground biomass and nutrient use following Hurricane

Hugo in the Bisley Experimental Watersheds, Luquillo Experimental Forest,

Puerto Rico. Biotropica 28: 424-440.

Scatena FN, Planos-Gutiérrez EO, Schellekens J. 2004. Natural disturbances and the

hydrology of humid tropical forests. In: Forests, Water, and People in the Humid

Tropics. Bonell M, Bruijnzeel LA (eds). Cambridge University Press: Cambridge,

UK; 5-28.


interception by a lowland tropical rain forest in northeastern Puerto Rico. Journal


Schellekens J, Scatena FN, Bruijnzeel LA, van Dijk AIJM, Groen MMA, van Hogezand

RJP. 2004. Stormflow generation in a small rainforest catchment in the Luquillo

Experimental Forest, Puerto Rico. Hydrological Processes 18: 505-530.



1:20,000 scale.

85

Simon A, Guzman-Rios S. 1990. Sediment discharge from a montane basin, Puerto Rico:

Implications of erosion processes and rates in the humid tropics. In: Research

Needs and Applications to Reduce Erosion and Sedimentation in Tropical

Steeplands (Proceedings of the Fiji Symposium, June 1999). IAHS-AIHS

Publication 192, International Association of Hydrological Sciences: Washington,

DC; 35-47.

Swanson FJ, Johnson SL, Gregory SV, Acker SA. 1998. Flood disturbance in a forested

mountain landscape: interactions of land use and floods. BioScience 48: 681-689.


streams in Colorado. Journal of Hydrology 290: 137-151.

U.S. Water Resources Council, 1981. Guidelines for determining flood flow frequency.

U.S. Water Resources Council Bulletin 17B, 28 pp.

Williams PG. 1978. Bankfull discharge of rivers. Water Resources Research 13: 1141-

1154.

White AF, Blum AE, Schulz MS, Vivit DV, Stonestrom DA, Larsen M, Murphy SF,

Eberl D. 1998. Chemical weathering in a tropical watershed, Luquillo Mountains,

Puerto Rico: I. Long-term versus short-term weathering fluxes - implications for

hydrologic control of differences in absolute rate. Geochimica et Cosmochimica

Acta 62: 229-226.

Wohl E, Kuzma J, Brown NE. 2004. Reach-scale channel geometry of a mountain river.


Wohl E, Wilcox A. 2005. Channel geometry of mountain stream in New Zealand. Journal


86

Wolman MG. 1955. The natural channel of Brandywine Creek, Pennsylvania. United

States Geological Survey Professional Paper 271.

Wolman MG, Leopold LB, 1957. River flood plains: Some observation on their

formation. United States Geological Survey Professional Paper 282: 87-107.



Woodyer KD. 1968. Bankfull frequency in rivers. Journal of Hydrology 6: 114-14.

87

CHAPTER 4

LONGITUDINAL PATTERNS IN STREAM CHANNEL GEOMORPHOLOGY IN THE TROPICAL MONTANE STREAMS OF THE LUQUILLO MOUNTAINS,

PUERTO RICO*

A.S. Pike, F.N. Scatena, and E.E. Wohl

ABSTRACT

An extensive dataset of 238 surveyed reaches in five adjacent watersheds draining

the Luquillo Mountains in northeastern Puerto Rico was used to examine downstream

changes in channel geometry, grain size, stream power, and shear stress along humid

tropical mountain streams. Surveyed data were used to compare the influences of

lithologic and hydraulic forces in shaping channel morphology. The Luquillo Mountains

are a steep landscape composed of volcaniclastic and igneous rocks that exert a strong

localized lithologic influence on the stream channels. Longitudinal profiles are generally

graded but have convexities and protrusions that reflect the influence of multiple rock

types. Non-fluvial processes, such as landslides along steep hillslopes (>12°), deliver

coarse sediment (>2000 mm) to the channels and may locally determine channel gradient

and geometry. Median grain size is strongly related to drainage area and slope, and

coarsens in the headwaters before fining in the downstream reaches; a pattern associated

with a mid-basin transition between colluvial and fluvial processes. However, the streams

also have strong hydraulic forcing due to high unit discharge. Downstream hydraulic

geometry relationships between discharge, width and velocity (although not depth) are

well developed for all watersheds (exponents are 0.33 for width, 0.12 for depth, and 0.55

* Submitted to Earth Surface Processes and Landforms in March 2008. All data analysis, interpretation of results, and manuscript writing were done by author of this dissertation.

88

for velocity). Stream power displays a mid-basin maximum in all basins, although the

ratio of stream power to coarse grain size (indicative of hydraulic forcing) increases

downstream. Excess dimensionless shear stress at bankfull flow wavers around the

threshold for sediment mobility of moderate-sized grains within a coarser matrix, and

does not vary systematically with bankfull discharge; a feature common in self-forming

“threshold” alluvial channels. The results suggest that although there is apparent bedrock

and lithologic control on local reach-scale channel morphology, strong fluvial forces

acting over time have been sufficient to override boundary resistance and give rise to

systematic basin-scale patterns.

89

INTRODUCTION

Recent advances in understanding the linkages between tectonics and surface

processes have spurred interest in the evolution of mountain and bedrock streams

(Whipple 2004, Bishop 2007). Mountain stream channels have complex morphologies

and a number of studies implicate several different controls on their development,

including: tectonic and structural (VanLaningham et al. 2006), bedrock (Snyder et al.

2003), storm pulses (Gupta 1988), and non-fluvial processes such as landslides/debris

flows (Brummer and Montgomery 2003, Stock and Dietrich 2006) and glaciers (Wohl et

al. 2004). Other studies have demonstrated the characteristic morphology of mountain

streams (Grant et al. 1990, Montgomery and Buffington 1997, Wohl and Merritt 2001),

their hydraulic geometry (Wohl 2004), the complexity of sediment transport (Blizard and

Wohl 1998, Lenzi et al. 2004, Torizzo and Pitlick 2004), and the distribution of sediment

within mountain drainages (McPherson 1971, Grimm et al. 1995, Pizzuto 1995,

Constantine et al. 2003, Golden and Spring 2006). Of all montane streams, those in the

tropics are among the most extreme fluvial environments in the world (Gupta 1988). A

combination of steep slopes, high mean annual rainfall, and intense tropical storms

generate an energetic and powerful flow regime. The high rates of erosion and

dramatically dissected landscapes prevalent in the world’s tropical mountainous regions

attest to the power of these rivers. Yet the channel morphology that is sculpted by fluvial

processes in tropical montane environments is generally unknown. This paper

investigates controls on mountain stream channel morphology in the Luquillo Mountains

of Puerto Rico, a tectonically active landscape with varying bedrock and structural

90

controls that is rapidly eroding due to extremely wet tropical conditions, frequent intense

storms, and a high susceptibility to mass-wasting.

Montane streams in both tropical and temperate environments share some

common characteristics. Steep gradients, tectonic activity, and multiple rock types yield

resistant channel boundaries that are dominated by bedrock and coarse clasts (Grant et al.

1990). Vertical valley walls and confined channel boundaries inhibit floodplain

development and may locally determine channel width (Montgomery and Gran 2001,

Finnegan et al. 2005). Longitudinal profiles are typically segmented by knickpoints and

waterfalls. There is often high boundary roughness, intense turbulence, high entrainment

rates and stochastic bedload movement (Wohl et al. 2004).

However, some tropical mountain streams may have unique features that vary

from their temperate counterparts. The absence of glaciation excludes glacial landforms,

such as u-shaped valleys and coarse moraine deposits, that are prevalent in some

temperate montane basins. Relatively high rates of chemical and physical weathering

rapidly denude tropical landscapes and may affect rates of channel-sediment diminution

and patterns of downstream fining (Brown et al. 1995, White et al. 1998, Rengers and

Wohl 2007). Frequent landslides triggered by heavy rains introduce pulses of coarse

sediment to the channels and strongly link fluvial and colluvial forces (Larsen et al.

1999). Large woody debris that is common in temperate streams is rapidly decomposed

in the tropics, despite high inputs from surrounding mature forests and hurricanes

(Covich and Crowl 1990). Rapid runoff production generates flashy, frequent, short-

duration floods (Schellekens et al. 2004, Niedzialek and Ogden 2005), and periodic high-

91

magnitude floods associated with hurricanes and other tropical disturbances effectively

rework boulder channels (Gupta 1975, Scatena and Larsen 1991).

The relatively few studies that have addressed the underlying controls structuring

the morphology of tropical mountain streams demonstrate the influence of a variety of

factors. Ahmad et al. (1993) and Gupta (1995) concluded that the lithology of many

streams in the Caribbean plays a strong role in locally determining channel morphology,

dictating the course of the river, and governing the distribution of large boulders. These

streams, similar to many mountain streams, commonly have bedrock-lined channels

whereas traditional depositional forms built by sand, gravels, cobbles, and boulders are

only found sporadically. Lewis (1969) demonstrated that local lithologic factors, such as

bed material cohesion and channel constriction, influenced at-station hydraulic geometry

in the Río Manati of north-central Puerto Rico. However, it was also demonstrated that

consistent scaling of downstream hydraulic geometry was developed across multiple

lithologies. In the streams of Jamaica and Puerto Rico, Gupta (1975) emphasized the role

of high discharge relative to drainage area as a key hydraulic control shaping channel

morphology. Similar characteristics were noted in the Río Chagres in Panama, where

hydraulic controls due to notably high unit discharge are apparently sufficient to override

lithologic controls and develop a basin with well-developed downstream hydraulic

geometry (Wohl 2005). This study also utilized extensive basin-scale field

reconnaissance to quantify downstream patterns in tropical mountain stream morphology,

and contended that similar high-quality field surveys in different tropical regions are

essential to further knowledge about these systems.

92

Geomorphic features of rivers that reflect underlying controls include longitudinal

profiles, hydraulic geometry, grain sizes, and the spatial distribution of fluvial energy

expenditure and shear stresses. Many alluvial rivers develop systematic changes in slope,

channel geometry, and grain size from their headwaters to the coast in response to

changes in discharge and sediment yield (Paola and Seal 1995). These changes result in

many well-known basin-scale patterns such as concave-upward longitudinal profiles and

progressive downstream fining, whereby adherence or significant deviations from the

theoretical patterns reflect the relative importance of lithologic and hydraulic controls.

Longitudinal profiles of rivers often reflect the lithologic and tectonic controls on

channel development (Kirby and Whipple 2001). A theoretical profile of a graded stream

has a smoothly concave-upward shape; steep in the headwaters and flat near the mouth

(Hack 1957). A river of this form has achieved an assumed balance between the erosion

from fluvial processes and the resistance from lithologic and tectonic forces. Deviations

from this idealized grade, such as changes in concavity (Seidl et al. 1994) and the

presence of segmentation/knickpoints (Crosby and Whipple 2006, Goldrick and Bishop

2007) can indicate the influence of non-fluvial forces. For example, faults and structural

barriers may confine a river and constrain slope (Whipple 2004), multiple bedrock units

with varying resistance to river incision often create slope breaks, protrusions, and/or

knickpoints (Wohl et al. 1994), and landslides/debris flows can locally constrain the

channel gradient and concavity (Grant et al. 1990, Stock and Dietrich 2003).

Downstream hydraulic geometry (DHG) characterizes systematic downstream

changes in channel geometry as power-law relationships with discharge, and may be used

to quantify the influence of fluvial controls on channel form (Leopold and Maddock

93

1953). DHG has successfully described river patterns worldwide in many physiographic

environments. However, it is intended to describe changes in self-forming alluvial rivers

that readily adjust their geometry in response to changes in discharge and sediment

transport. The ubiquity of DHG in these self-forming rivers has been explained from a

combination of basic hydraulic and sediment transport processes (e.g., Singh 2003,

Parker et al. 2007). However, the complicated hydraulics and sediment transport

processes associated with boulder- and bedrock-armored channels in many mountain

rivers may confound these relationships. Consistent power-law relations in downstream

channel geometry have been observed in some mountain rivers, even though these

streams alter their morphology at longer time scales than most alluvial rivers (Molnar and

Ramirez 2002, Wohl and Wilcox 2005). In fact, mountain rivers with well-developed

DHG tend to have an above-average ratio of total stream power (a measure of hydraulic

driving forces) to coarse grain size (a measure of boundary resistance) (Wohl 2004). In

contrast, mountain rivers that are strongly controlled by geologic rather than hydraulic

controls will often display poorly defined DHG (Wohl et al. 2004).

The distribution of grain sizes throughout the stream network can also yield

insight into the underlying lithologic controls. Grain size in the stream channel is largely

dependent on the underlying bedrock, the input from hillslopes, and the mechanisms of

weathering and transporting clasts. In fluvial systems where the bed material is readily

mobile, there is often a balance between discharge, sediment transport, and slope.

Consequently, grain size often declines with increasing drainage area such that the largest

grains are found in the in headwater channels and smaller grains are found in lower

reaches (Paola and Seal 1995, Rice 1999, Constantine et al. 2003). In steep montane

94

catchments where landslides introduce large pulses of coarse (and potentially immobile)

sediment, this balance is upset, and grain-size patterns are often discontinuous (Grimm et

al. 1995, Pizzuto 1995, Brummer and Montgomery 2003). The downstream pattern in

grain sizes indicates the relative influence of coarse material deposited from hillslope

processes and the ability of the channel to transport sediment.

Many river networks also tend towards an assumed optimal state of energy

expenditure throughout their evolution such that certain indices of energy expenditure are

either constant or linear along the river profile (Molnar and Ramirez 2002). Nonlinearity

in stream power, whereby energy expenditure is concentrated in specific reaches rather

than uniformly dispersed, can indicate underlying geologic control (Graf 1983, Lecce

1997). Similarly, many stream networks have a mid-basin maximum in stream power, the

location of which is dependent on slope, the flow regime, and the structure of the basin

(Knighton 1999). Large gradients in bed stress or energy expenditure also yield gradients

in sediment flux, causing certain parts of the river to erode and others to deposit

sediments in an effort to remove these gradients. In bedrock- and boulder-lined channels

where coarse sediment is not readily mobile, the ability of the channels to adjust their

morphology to remove these gradients in energy expenditure may be hindered.

Lastly, the downstream trend in boundary shear stress at bankfull discharge

provides insight into sediment mobility and the relative stability of channels. The Shields

parameter, a dimensionless bed shear stress that is expressed as a ratio of slope, depth,

and size of the bed material, is a quantitative indicator of flow competence and is strongly

related to alluvial channel form (Dade and Friend 1998, Dade 2000). In many self-

forming alluvial channels, the Shields parameter at bankfull flow does not vary

95

systematically throughout a basin. Assuming a constant critical dimensionless shear

stress, this lack of scaling between the Shields stress and bankfull discharge implies that

many alluvial channels are at the threshold for incipient sediment mobility. However, in

gravel- and boulder-lined mountain channels, both the Shields parameter and critical

dimensionless shear stress often vary widely throughout the basin, depending on flow

resistance (Mueller et al. 2005). If the flow regime in montane channels is sufficient to

mobilize the bulk of the sediment, we would expect the Shields parameter to be

consistently higher the critical dimensionless shear stress; lower if the sediment is too

coarse to transport. Furthermore, if the controls on sediment transport shift from non-

fluvial forces upstream to fluvial forces downstream, we would expect the excess

dimensionless shear stress to increase with bankfull discharge.

In this study, we present results from an extensive field survey of a tropical

mountain stream network. We compare channel profiles and subsequent downstream

changes in cross-sectional geometry, grain size, and channel energetics in five adjacent

watersheds in the Luquillo Mountains of northeastern Puerto Rico. This comprehensive

dataset allows us to test two alternate hypotheses on the potential controls on stream

channel morphology. First, if local lithologic factors such as multiple rock types, resistant

channel boundaries, and coarse sediment delivery from landslides dominate the form of

the river, we would expect the channel profile to be segmented, display poorly developed

hydraulic geometry, have seemingly random pattern of grain sizes, and have insufficient

stream power and boundary shear stress to mobilize available sediment. Conversely, if

the high unit discharge and associated stream power of the energetic tropical flow regime

are sufficient to overcome lithologic resistance and mobilize coarse sediment, then we

96

would expect downstream changes in channel morphology to have systematic trends

similar to those found in many alluvial rivers.

STUDY AREA

The Luquillo Mountains in northeastern Puerto Rico rise steeply from sea-level to

over 1000m in elevation over a distance of 15 to 20 km. They are characterized by steep

slopes, rugged peaks, and highly dissected valleys. The landscape is composed of several

lithologies and a variety of land cover. The streams have their headwaters in the Luquillo

Experimental Forest (LEF), a 113 km2 protected forest reserve under the management of

the United States Forest Service. The study area consists of five adjacent watersheds

draining the LEF: Río Blanco, Río Espiritu Santo, Río Fajardo, Río Mameyes, and Río

Sabana (Figure 4.1). The watersheds are similar physiographically, although they vary in

size, lithology, and land cover. Drainage areas of these watersheds are 72 km2, 92 km2,

67 km2, 44 km2, and 35 km2, respectively. All of the watersheds, except for the Río

Sabana, reach the upper-most ridges of the Luquillo Mountains.

The humid subtropical maritime climate is influenced by both northeasterly trade

winds and local orographic effects that interact to form steep gradients in precipitation.

Mean annual rainfall increases with elevation from approximately 1500 mm per year at

the coast to >4500 mm/yr at elevations above 1000 m (García-Martinó et al. 1996). The

principal weather systems affecting climate are convective storms, easterly waves, cold

fronts, and tropical storms (van der Molen 2002). Rainfall is a near-daily occurrence

(Schellekens et al. 1999), and high-intensity rainfall events and floods can occur in any

given month. Hurricanes and tropical storms are common from August through October,

97

Figure 4.1 Location map of northeastern Puerto Rico. Shown are the 238 surveyed

reaches in 5 adjacent watersheds and the regional topography, geology, and land cover.

98

Figu

re 4

.1 c

ont.

99

and typically bring high daily rainfall in excess of 200 mm/day (Heartsill-Scalley et al.

2007); the maximum recorded daily rainfall is >600 mm/day (Scatena and Larsen 1991).

The streams of the Luquillo Mountains have been classified as “flood dominated”

channels that have a hydrologic setting similar to other montane environments in the

Greater Antilles and regions along active tectonic zones in the humid tropics (Gupta

1988, Ahmad et al. 1993). Floods are intense and peak discharges can be 1000 times

greater than baseflow. The unit discharge at baseflow is approximately 0.02 m3/s/km2,

whereas the highest peak unit discharge ever recorded at a regional stream gage was

19.7 m3/s/km2 (United States Geological Survey, updated 2006). Peak-flow hydrographs

are short-lived and typically have a duration of less than one hour. Stormflow runoff is

quickly flushed through the system such that the streams return to baseflow within 24

hours of large events. Large floods are driven by storm events, as opposed to the seasonal

floods associated with snowmelt in many temperate mountain streams. Consequently,

discharges that are close to the annual peak are often experienced independently several

times in a year (Scatena et al. 2004).

The Luquillo Mountains were formed by early Tertiary volcanism and tectonic

uplift associated with oceanic island-arc subduction. The landscape consists of several

dominant lithologies: volcaniclastics, plutonic intrusions and dikes, contact

metamorphics, and alluvium (Seiders 1971a, Briggs and Anguilar-Cortés 1980). The

volcaniclastic rocks, comprised of marine-deposited volcanic sediments of late

Cretaceous age, form the bulk of the Luquillo Mountains. They include units of

sandstones, siltstone, mudstones, breccias, conglomerates, tuff, and lava, that are

complexly faulted and steeply tilted (>30º). A Tertiary quartz diorite (granodiorite)

100

batholith underlays the southern side of the study area. It outcrops in an area of

approximately 24 km2, and is drained largely by the Río Blanco watershed, but also by

small parts of the upper Río Espiritu Santo and Río Mameyes watersheds. It is rapidly

eroding at an estimated denudation rate of 25-50 m/million yr; one of the highest

documented weathering rates of silicate rocks on the Earth’s surface (Brown et al. 1995,

White et al. 1998). A 1-2 km zone of contact metamorphism surrounds the granodiorite.

These contact metamorphosed volcaniclastic rocks (hornfels facies) exhibit greater

hardness than both their unmetamorphosed equivalents and the granodiorite (Seiders

1971b). Because of their relative resistance to erosion, these rocks form steep cliffs and

the tallest peaks in the region. Several vertical dikes traverse the volcaniclastic rocks,

mainly at lower elevations (<150 m). The mountains are fringed by a lowland coastal

plain composed of Quaternary alluvium.

Past climates in the region are thought to be similar to the present, due to both

comparable elevation and location of the mountain range in the subtropical maritime belt

(Graham and Jarzen 1969). Pollen assemblages and plant microfossils of subtropical and

warm-temperate communities found in sedimentary sequences on the island suggest that

the mountains of Puerto Rico had a subtropical climate in the Oligocene, with additional

cooler higher elevation environments not present today. However, the Luquillo

Mountains’ climate is generally considered to have oscillated around a humid subtropical

state over time, without glaciers or dramatic changes (Scatena 1998).

The main stream channels and rivers are relatively old; on the order of tens of

millions of years. The emplacement of the plutonic rocks and supposed uplift of the

Luquillo fault block occurred in the Eocene (Cox et al. 1977, Kesler and Sutter 1979),

101

although significant erosion of the landscape and initial formation of the modern stream

network probably did not occur until the Miocene (Monroe 1980). Stratigraphic records

preserved on the northwestern side of the island suggest that no surface streams delivered

clastic sediment into the sea until this time period. Since then, the stream network has

been continually formed by a humid climate, and the main upland channels that are

confined by steep valleys walls have presumably not migrated significantly.

All five watersheds currently drain protected primary forest in their upper

elevations, mature secondary forest at mid-elevations, and both abandoned (reforesting)

agricultural fields/grazing pastures and scattered urbanized developments along the

coastal plain. Each river flows through a mangrove-lined estuary before reaching the

coast. However, the land cover in the region has been continually changing since the

Spanish colonization of the island in the late 17th century. Many low-elevation areas

(<300 m) of northeastern Puerto Rico were cleared for agriculture between 1830 and

1950. This caused an estimated 50% increase in runoff, and an order of magnitude

increase in sedimentation on the coastal plain (Clark and Wilcock 2000). Subsequent land

clearance on steep valley slopes resulted in widespread erosion and landslides that

delivered a large load of coarse sediments to the river. Large portions in the upper

elevations of the LEF were never deforested during the 19th and 20th centuries due to

government protection, steep slopes, and high rainfall (Scatena 1989). Since 1950,

urbanization and reforestation of former agricultural land in low-lying areas has resulted

in elevated storm runoff and decreased sedimentation, allowing transport of previously

deposited coarse alluvial sediment in coastal plain streams (Clark and Wilcock 2000, Wu

et al. 2007).

102

Hillslopes are typically steep, in excess of 30° in many headwater areas, and are

consequently strongly linked to channel processes (Scatena and Lugo 1995). Landslides

are the dominant process that physically weathers the regolith and delivers sediment to

the channel (Simon and Guzman-Rios 1990, Larsen et al. 1999). Other hillslope

weathering processes such as sheetwash, soil creep, and treefall-induced mass movement

are prevalent but less important to the total sediment yield (Larsen 1997). There is an

abundance of clay in the deeply-weathered soil and thick saprolite that is derived from

both volcaniclastic and granodiorite bedrock (Frizano et al. 2002, Schellekens 2004).

However, there is typically little fine sediment that persists in the streams channels

(Simon and Guzman-Rios 1990). Flood-discharges quickly wash fine sediment from the

channel, and the streams are generally clear within a day of a large storm.

Triggered by intense rains, landslides are common at upper elevations,

particularly on areas underlain by granodiorite, and on hillslopes that exceed 12° gradient

(Larsen and Torres-Sánchez 1998). Forested areas underlain by granodiorite rocks

experience twice as many landslides as comparable areas underlain by volcaniclastics

(43 landslides/km2/100yr on granodiorite; 21 landslides/km2/100yr on volcaniclastics)

(Larsen 1997). Landslides are capable of delivering very large boulders to the stream

channels, and the corresponding volume of material transported is substantial

(700 Mg/km2/yr on granodiorite; 480 Mg/km2/yr on volcaniclastics). The large majority

(80-90%) of total sediment delivery to the streams is attributed to landslides, and the

associated pulse of sediment delivery can locally alter the channel morphology. Since

1979, there have been numerous sliding events, and two large landslides have temporarily

103

dammed permanent streams, persisting for a few weeks before being removed by

stormflow (personal observation by F.N. Scatena).

The first-order drainage network consists of a dense, dendritic network of small

ephemeral channels that range from leaf-filled swales to mossy cobble-lined channels

that become active only during large rainfall events (Scatena 1989, Schellekens 2004).

Larger first-order perennial streams have channels dominated by boulders in steeply

sloped reaches, and clay and soil-lined channels in reaches with more gentle slopes.

Second- and third-order streams have steep-gradient reaches, exposed bedrock channels,

matrices of large boulders interspersed with finer sediment, and periodic waterfalls (up to

30 m in height). Many of the upland streams are characterized by cascade and step-pool

morphologies, whereas the lower reaches are plane bed and pool-riffle sequences (sensu

Grant et al. 1990, Montgomery and Buffington 1997, Trainor and Church 2003).

Structural control of channel pattern is apparent in many places as witnessed by

rectangular stream bends at fault intersections, streams following bedrock joints, and

knickpoints at lithologic boundaries. Due to rapid decomposition, these channels lack the

large coarse woody debris dams that influence the morphology of many channels in

humid temperate environments (Covich and Crowl 1990).

Fourth- and fifth-order streams occur only at lower elevations, flowing across the

coastal plain as relatively gentle gradient pool-riffle sequences. Alluvial inset deposits,

high-flow channels, floodplains, and terraces are common features in these lower reaches

(Ahmad et al. 1993, Clark and Wilcock 2000). These larger lowland streams are

relatively straight, are not constricted by bedrock, and have laterally migrating high-flow

104

channels that indicate that the alluvial channels adjust in response to varying discharge

and sediment supply.

METHODS

A total of 238 stream cross-sections (see Appendix) were surveyed in the

summers of 2003-2006; 11 in the Río Blanco, 88 in the Río Espiritu Santo, 31 in the Río

Fajardo, 91 in the Río Mameyes, and 17 in the Río Sabana (Figure 4.1). Cross-section

locations were chosen to capture the entire range of elevation, drainage areas, and

substrate type. The surveyed cross-sections are located on 1st to 5th order streams, have

drainage areas between 0.1 km2 to 79 km2, and are located on average approximately

every 30 m in elevation and 500 m in distance along the channel.

Cross-sections were surveyed at a straight uniform section within a reach.

Relative distance and elevation were measured at evenly spaced intervals along a transect

spanning from vegetated bank to bank, using a Sokkia Total Station Laser Theodolite

(Set 530). The most meaningful comparison of cross-sectional geometry is at the bankfull

stage that also correlates to the effective discharge of sediment and occurs at a constant

flow frequency throughout the basin (Leopold and Maddock 1953, Wolman and Miller

1960). However, the absence of floodplains in the mountainous reaches confounds the

identification of bankfull stage. In place of bankfull in steepland streams, cross-sections

extended to the boundary of the active channel that is marked by the edge of perennial

woody vegetation (shrubs and trees) and incipient soil development. A previous analysis

of flow-frequency at gaged stream reaches indicates that this active channel boundary

coincides with a flood discharge that occurs at the same frequency of both the bankfull

105

and effective discharge in adjacent alluvial channels (Pike and Scatena, in review). Using

these riparian features as markers, active channel width, average depth, and cross-

sectional area were calculated for each cross-section. Channel slope was similarly

measured as the difference in elevation of the water surface over 10 uniformly spaced

points spanning approximately five channel widths upstream of the cross-section.

Active-channel (bankfull) discharge at each cross-section was estimated by a

regional equation based on long-term stream gage data (Pike 2006). At nine stream gages

(with at least 10 years of record) in the study watersheds, the active channel discharge (as

marked by the first occurrence of woody shrubs, trees, and soil) corresponds to a flood

that is exceeded 0.16% of the time, and has an average unit discharge of 2.2 m3/s/km2

(Pike and Scatena, in review). In this region, rainfall and runoff increase with elevation

due to the precipitation gradient; higher elevation basins have more runoff than low

elevation basins of comparable size. Thus, active channel discharge was best estimated as

a function of drainage area multiplied by a linear model relating runoff to average basin

elevation:

)406.0*h0042.0DA(Q avgAC += n = 9, r2 = 0.97, P < 0.001 Eq. 4.1

where: QAC is the active channel discharge in m3/s, DA is drainage area in km2,

and havg is the average upstream elevation in m. Both of these variables are estimated for

each reach using a 10m Digital Elevation Model (DEM) and a GIS-based flow

accumulation algorithm. Longitudinal profiles were also constructed from a 10 m DEM.

Concavity was estimated along the main stem based on the relationship between slope

and drainage area, using points spaced every 10 m drop in elevation:

θkDAS = Eq. 4.2

106

where: θ is the concavity index, k is a steepness coefficient, DA is drainage area

(km2), and S is slope (m/m).

Elevation, drainage area, slope, and active-channel discharge (Equation 4.1) were

also estimated along the main-stem profile at a series of points where the streams

intersect 10m contour lines; that is, at every 10m drop in elevation.

Downstream hydraulic geometry relationships were calculated by least-squares

log-linear regressions between active channel discharge and channel geometry

measurements. Active channel discharge (Q) correlates with active channel width (w),

average flow depth (d), and mean velocity (v), such that: w=c1Qb, d=c2Qf, v=c3Qm

(Leopold and Maddock 1953). By conservation of mass, the product of the coefficients

(c1,c2,c3) and sum of the exponents (b,f,m) must equal 1; c1*c2*c3=1, b+f+m=1. Mean

velocity was calculated as the active channel discharge divided by the channel cross-

sectional area (QAC/A). Since this indirect calculation of velocity is a function of

discharge, the strength of the regression equation was estimated as the correlation

between the logarithms of active channel discharge and cross-sectional area.

Grain size in the active channel was estimated using a modified Wolman Pebble

Count method (Wolman 1954). Approximately 100 clasts were selected randomly by

pacing across the width of the stream. The median diameter of each clast was measured,

and classified into the following seven size categories: bedrock (no size), megaboulder

(>2000 mm), boulder (256-2000 mm), cobble (64-256 mm), gravel (2-64 mm), sand

(.063-1 mm), and fines (silt/clay, 0.001–0.063 mm). From these grain size measurements,

we determined the median grain size (d50), coarse grain size (d84), and the percent of

bedrock exposed within the active channel.

107

Sediment mobility was calculated using the survey data and estimates of shear

stresses. Sediment is considered mobile when the dimensionless boundary shear stress

exceeds the dimensionless critical shear stress. Boundary shear stress, τ (N/m), was

calculated as:

ρgRSτ = Eq. 4.3

where: ρ is water density (1000 kg/m3), g is acceleration due to gravity (9.8 m/s2), R is

hydraulic radius (m), S is slope (m/m).

Dimensionless shear stress, τ*, or Shields stress, required to mobilize the coarse

sediment was estimated as:

( ) 50s gdρρττ*

−= Eq. 4.4

where: ρs is sediment density (2650 kg/m3), d50 is the median grain size (in m).

Dimensionless critical shear stress, τ*c, was estimated from an equation

developed for steep gravel-bed rivers by Mueller et al. (2005), that relates the

dimensionless reference shear stress (assumed to be critical) to slope (S):

021.0S18.2τ*c += Eq. 4.5

Typically, critical dimensionless shear stress is assumed to be constant throughout

the basin. However, the aforementioned equation shows the critical dimensionless shear

stress varies by accounting for excess bed roughness in steep reaches with large grains, as

supported from data from numerous steep gravel-bed rivers.

Total stream power per unit channel length, Ω (W/m), is defined as:

ρgQSΩ = Eq. 4.6

108

Stream power was calculated for each reach using the survey data and estimated

active-channel discharge.

RESULTS

Longitudinal Profiles

The longitudinal profiles of each of the five rivers have unique shapes that are

related to their underlying geology. A theoretical graded concave-upward profile is

steepest in the headwaters, and displays a systematic downstream decline in slope. Yet

the profiles here are segmented by a series of convex protrusions and slope breaks that

deviate from a systematic grade. For example, the volcaniclastic headwaters of the Río

Fajardo and the contact metamorphic upper reaches of the Río Mameyes display a

traditionally concave shape (Figure 4.2). Similarly, the alluvial reaches are well-graded

and have few slope breaks. However, local factors also shape the profile. For example,

the steep streams on volcaniclastic rocks have knickpoints that generally correspond to

bedrock faults identified on USGS 1:20,000 geologic maps (Seiders 1971a, Briggs and

Anguilar-Cortez 1980). Also, locally exposed outcrops create small convexities in the

longitudinal profiles. Most striking are the anomalously convex profiles of granodiorite

streams. The headwaters of granodioritic Río Blanco are unusually flat before cascading

steeply down the side of the batholith and leveling out along the alluvial coastal plain

(Figure 4.2). The inflection point where the stream sharply steepens occurs at the edge of

a non-glacial hanging valley. This flat form is also seen in the headwaters of the Río

Espiritu Santo.

109

Figure 4.2 Longitudinal profiles of the main stem of each river highlighting the

relationship between local profile shape and lithology. Drainage areas as well as major

mapped faults (vertical bars) are indicated. Slope-area plots (inset), using points spaced

every 10m in elevation along the main stem, are shown to indicate changes in the

concavity index (θ).

110

Figure 4.2 cont.

111

Where the stream flows across one lithology, the longitudinal profile is locally

graded and has a traditional concave pattern. Where the stream flows over two or more

rock types, there is often a slope break at the contact, especially where the adjoining

rocks have varying resistance to erosion. The boundary between contact metamorphic

rocks and other lithologies, as on the main stem of the Río Mameyes and Río Espiritu

Santo, is accompanied by a pronounced convexity (Figure 4.2). Furthermore, slight

changes in the composition of the volcaniclastic rocks, from a sandstone unit to a

mudstone unit, are often the site of waterfalls and/or steep gradients (personal

observation). Similar notable breaks occur where upland alluvial formations, typically

terrace deposits, merge with volcaniclastics, such as a knickpoint on the Río Fajardo that

occurs at the boundary between a mid-elevation structural bench and the surrounding

volcaniclastics. Also, the lowest elevation waterfall in the region occurs at the transition

where the Río Sabana flows across a locally exposed volcaniclastic formation

approximately 7 km from its headwaters.

The concavity index (θ) of the main stem of each river profile, calculated from

slope-area relationships, is related to both the underlying rock type, and drainage area.

Concavity values for headwater portions of the profiles (DA < 10 km2) are starkly

different between areas underlain by granodiorite and those by volcaniclastics.

Concavities on volcaniclastics (Rio Fajardo, Rio Mameyes, and Rio Blanco) are in the

low to moderate range (θ = 0.15 to 0.42). On granodiorite and other lithologies (Rio

Blanco and Rio Espiritu Santo), concavities values are high (θ = 0.99 to 1.21) along the

gently-sloped reaches and convex (θ = -0.75 to -1.74) where the channel steepens. There

is a break in the slope-area relationship at approximately 10km2 along each profile.

112

Consequently, lowland and alluvial portions of the profiles (DA > 10km2), are all

strongly concave (θ = 1.07 to 4.65).

Hydraulic Geometry

Downstream hydraulic geometry relationships for the active channel width,

hydraulic radius, and mean velocity were calculated using the cross-sectional data and

estimated active channel discharge (Figure 4.3). The coefficient of determination (r2) for

these hydraulic geometry relationships are 0.71 for width, 0.21 for depth, and 0.66 for

velocity (as determined by the relationship with cross-sectional area). Discharge displays

a strong power-law relation with both width and velocity, but not depth. The active

channel systematically widens in the downstream direction, despite potential constriction

from bedrock outcrops and confined valley walls. Yet the streams do not deepen

substantially downstream. Instead, they display strong local variation. Comparably deep

pools and shallow riffles are observed in both headwater and lowland reaches.

Downstream hydraulic geometry exponents for all basins are 0.33 for width, 0.12

for depth, and 0.55 for velocity. With increasing discharge, width increases at

approximately three times the rate of depth. This implies that the width/depth ratio

similarly increases in the downstream direction, and that the channel form changes from a

triangular ‘v’-shape (low w/d ratio) with in the headwaters to a more rectangular (high

w/d ratio) form near the mouth. Velocity increases at a rapid rate of change in the

downstream direction resulting in a mean cross-sectional velocity 100 times greater in the

lower reaches than in the headwaters during a flood at active-channel discharge.

DHG relationships for individual watersheds show general consistency among

basins (Table 4.1). The width exponents range from 0.24 to 0.37, the depth exponents

113

Figure 4.3 Downstream hydraulic geometry relationships between active channel

discharge, width, depth, and velocity using data from all surveyed reaches.

114

Table 4.1 Downstream hydraulic geometry divided by watershed. Coefficients,

exponents, and coefficient of determination (r2) are between the active channel discharge

and each corresponding channel geometry variable.

width depth velocity Watershed c1 b r2 c2 f r2 c3 m *r2 n Blanco 4.0 0.37 0.65 0.6 0.09 0.05 0.4 0.54 0.54 11 Espiritu Santo 5.9 0.30 0.63 0.6 0.13 0.22 0.3 0.57 0.55 88 Fajardo 4.9 0.35 0.73 0.9 0.02 0.02 0.2 0.62 0.73 31 Mameyes 5.2 0.35 0.80 0.6 0.13 0.33 0.3 0.52 0.77 91 Sabana 7.7 0.24 0.50 0.4 0.17 0.30 0.3 0.58 0.57 17 ALL 5.4 0.33 0.71 0.6 0.12 0.21 0.3 0.56 0.66 238 * r2 for velocity relationships estimated from discharge vs. cross-sectional area

115

from 0.02 to 0.17, and the velocity exponents from 0.52 to 0.62. The differences in

coefficients and exponents are not strongly correlated to basin-scale factors such as

catchment size or geology. For each watershed, the r2 values for width and velocity

relationships are > 0.5, but less than 0.5 for depth relationships. Consequently, DHG is

considered well-developed for all the watersheds.

For comparison, average DHG exponents for many alluvial rivers world wide are

0.5 for width, 0.4 for depth, and 0.1 for velocity (Park 1977). DHG exponents in

mountain streams deviate slightly from the world average by having a lower width

exponent and greater velocity exponent, with average values of 0.36 for width, 0.38 for

depth, and 0.20 for velocity (Wohl 2004). The Luquillo streams have a width exponent

comparable to other mountain streams, but both the lowest known depth exponent and the

highest velocity exponent for a mountain stream.

Grain Size

The grain size of bed material varies widely throughout the watersheds, but is

clearly related to the underlying rock type (Figure 4.4). For example, long stretches of

step-pool sequences composed of boulders up to several meters in diameter are present in

a steep upland tributary of the Río Espiritu Santo underlain by volcaniclastic rock. In

contrast, the headwaters of the Río Blanco are composed mostly of mobile sand that is

weathered from granodiorite interspersed with large boulders. A typical lowland reach is

composed of cobbles and gravels derived both from the surrounding alluvium and

transported from upper reaches. The largest clasts observed in the river channels are slabs

of volcaniclastic rock and granodiorite corestones that reach 15m in diameter. These are

so large and immobile that they are hydraulically indistinguishable from bedrock.

116

Figure 4.4 Upstream views of typical reaches throughout the basins. The grain size

varies with both the lithology and the position along the stream profile. The average

channel width / median grain size, d50, for each reach are: a) 13.2m / 480mm, b) 6.8m /

1mm, c) 11.0m / 60mm, d) 32.3m / 150mm, e) 14.0m / 330mm, f) 25.4 m / 70mm.

117

Figu

re 4

.4 c

ont.

118

Unique distributions of grain sizes are observed on different lithologic types.

Using the pebble-count data from each study reach, all measured grains (excluding

bedrock) were sorted into logarithmically distributed bins (2φ intervals) for each major

rock type (Figure 4.5). Streambed material on volcaniclastic rocks has a high frequency

of cobble and boulder sized-sediment (64-1028 mm), but also contains lesser proportions

of large boulders (>1028 mm) and gravel. Field observations suggest that different

volcaniclastic formations have varying proportions of large boulders that are dependent

on the formation thickness. Contact metamorphic rocks display a distribution with fewer

large boulders and more sand than their unmetamorphosed equivalents. Mafic dikes have

a high proportion of large boulders, which is unique given the lower elevation.

Granodiorite streams have a bimodal grain-size distribution composed primarily of sand

and large boulders. Alluvial streams contain an abundance of cobble and gravel-sized

grains.

Megaboulders (boulders > 2000 mm diameter) are a relatively common feature in

the channels. Presumably, the megaboulders are corestones that are weathered directly

from bedrock along fracture planes, and subsequently deposited into the stream channels

by landslides. The abundance of megaboulders in the channel correlates with the slope of

the adjacent hillsides. Where the adjacent hillslope exceeds a threshold of 12º, there is

potential for landslides and possible deposition of megaboulders into the stream channel

(Larsen and Torres-Sánchez 1998). Conversely, below this hillslope threshold, the

hillside is relatively stable. Among 45 reaches having shallow adjacent hillslopes that do

not exceed 12º, no megaboulders were observed (Figure 4.6). Of the remaining 193

reaches adjacent to steep hillsides exceeding the 12º threshold, about half (89 reaches,

119

Figure 4.5 Grain size distributions for all measured clasts, grouped by lithology.

120

Figure 4.6 The percentage of megaboulders (boulders > 2000m) is associated with

adjacent hillslope; they are present when the hillslope exceed 12º. This is also the slope

threshold for landslides (Larsen and Torrez-Sánchez 1998) – the process that presumably

delivers these large boulders to the channel.

121

46%) have megaboulders present. Thus, megaboulders are only found in about half of

these reaches that are considered landslide-prone areas based on slope and lithology.

However, not all reaches in landslide-prone areas have megaboulders.

Most of the reaches have a relatively stable framework of large boulders that do

not appear to move, as evidenced by a thick moss-covering and occasional tree growth on

the boulders’ surfaces. Yet within this matrix, there is an abundance of smaller loosely-

packed gravel and sands that is transported during floods (personal observation). Average

dimensionless shear stress in these channels varies considerably but generally decreases

as a function of active-channel channel discharge (Figure 4.7a). Critical dimensionless

shear stress, as estimated by equation 4.5, also decreases with increasing active-channel

discharge (Figure 4.7b). This suggests that many of the headwater reaches have

additional flow resistance and form drag due to large boulders, which leads to an

increased threshold to initial sediment motion. Excess dimensionless shear stress (τ/τc)

does not vary systematically with active-channel discharge (Figure 4.7c). The average

dimensionless boundary shear stress at the active channel discharge exceeds the

dimensionless critical shear stress to mobilize the d50 in approximately 45% of the study

reaches (Figure 4.7c). That is, the median sediment size in approximately half of the

surveyed reaches can potentially be mobilized during the active channel flood that occurs

several times per year. The lack of scaling of excess sheer stress with discharge also

suggests that at the active channel discharge, the channels are generally at the sediment

transport threshold—similar to many alluvial channels.

Given the variety of grain sizes and range of rock types, the pattern of grain size

throughout the basin is not immediately apparent. The median diameter (d50) of particle

122

Figure 4.7 Relationship between active channel discharge and dimensionless shear

stresses and critical shear stresses. a) Dimensionless boundary shear stress is negatively

correlated with active channel discharge, rather than constant. b) Dimensionless critical

shear stress (a function of slope) also decreases with active channel discharge c) Excess

shear stress (as the ratio of dimensionless shear stress to critical shear stress) does not

vary systematically downstream with discharge, suggesting that the channels waver

around the threshold for sediment transport. Approximately 45% of the reaches have

presumably mobile substrate (τ* > τ*c) at the active channel discharge.

123

Figure 4.7 cont.

124

size at a reach correlates poorly with drainage area (r2 = .002, P = 0.83), and significantly

but only moderately with either slope (r2 = .20, P < 0.0001) or stream power (r2 = 0.30,

P < 0.0001). However, a log-linear plot between the ratio of d50 to drainage area (d50:DA)

and slope (sensu Hack 1957) yields a stronger and highly significant correlation (r2 = .74,

P < 0.0001) (Figure 4.8). A similar significant relationship was found for the d84 data (r2

= 0.72, P < 0.0001).

55.050

DAd

007.0S ⎟⎠

⎞⎜⎝

⎛= or

8.1

50 007.0SDAd ⎟

⎠⎞

⎜⎝⎛= Eq. 4.7

57.084

DAd

0026.0S ⎟⎠

⎞⎜⎝

⎛= or

75.1

84 0026.0SDAd ⎟

⎠⎞

⎜⎝⎛= Eq. 4.8

These two relationships state that grain size is a function of both drainage area and slope.

At a given drainage area, grain size is proportional to 1.8 power of slope. Conversely, if

slope remains constant, then grain size is directly proportional to drainage area.

Equation 4.7 was used to estimate median grain size along the profile of the river,

using drainage area and slope derived from the 10m DEM (Figure 4.9). The predicted

grain size function is jaggedly shaped due to local variations in slope. Yet the resulting

pattern shows a distinct downstream coarsening from the headwaters to a mid-basin

maximum. This is followed by subsequent downstream fining. Typically, cobble and

small boulder-sized sediment in the headwaters are replaced by very large boulders

(>2000 mm) in the mid to upper elevation steepland channels. Along the main stem, this

maximum typically occurs at approximately 5km from the headwaters, or approximately

25% to 33% of the distance of the main stem. As the slope declines towards the lowland

reaches, the sediment gradually fines as well.

125

Figure 4.8 The highly significant relationship (r2 = 0.74, P < 0.0001) between median

grain size (d50), drainage area, and slope. Data from this study (dark circles) are shown

alongside data representing humid streams in Maryland and Virginia from Hack (1957,

white circles) where this relationship was first published. The solid line is a least-squares

regression used as the basis for estimating grain size for a given drainage area and slope

(Equation 4.7), and dotted lines represent the 95% prediction interval. Outliers from this

study (crosses) were removed from regression on the basis of uncertainty of either

average grain size estimation or slope measurement.

126

The measured grain size data for reaches along the main stem show general

agreement with the predicted grain size (Figure 4.9). The data best display the coarsening

to fining trend in those rivers where surveyed reaches span a range of slopes and drainage

areas (Río Espiritu Santo, and Río Sabana). On other rivers, such as the Río Fajardo, the

measured grain sizes do not vary as widely because the surveyed reaches had comparable

slopes.

The largest discrepancy between the actual grain size and the predicted grain size

occurs in the steepest reaches (Figure 4.9). Here, the grain size function predicts boulders

in excess of 10m. These reaches are generally waterfalls and cascades where the largest

sediment is practically indistinguishable from bedrock, or is bedrock itself, and

consequently cannot be quantified by a measurable diameter. Field observations did

indicate notably larger grains in the mid-basin steep reaches. Data from waterfalls and

cascades on tributaries also show that the largest grains are found in the steepest reaches

with moderate drainage areas.

Stream Power

Total stream power along the main stem displays a peaked pattern in the

downstream direction (Figure 4.9). Given that total stream power is the product of

discharge and slope, stream power is low in the headwaters where slopes are steep but

there is minimal discharge. It is similarly low near the mouth where there is high

discharge but gentle slopes. Thus, stream power peaks at an intermediate distance where

a combination of adequate discharge and steep slopes generate maximum power. In these

streams, the location of this stream power peak typically occurs at mid to upper

elevations and at a downstream distance approximately a quarter to a third of the profile

127

Figure 4.9 Downstream changes in elevation, drainage area, median grain size, stream

power, and the ratio of stream power to coarse grain size along the main stem profile of

each river. 95% prediction intervals are shown in gray for the trends in grain size. Both

median grain size and stream power display a peaked pattern, the location of which

coincides with steep reaches having ample discharge. The ratio of stream power to coarse

grain size increases in the downstream direction, suggesting the increased relative

influence of hydraulic forces over lithologic controls downstream. A threshold of 10,000

W/m/m differentiates alluvial conditions from bedrock controls.

128

Figu

re 4

.9 c

ont.

129

length. The magnitude of the peak varies according to the watershed, with the larger and

steepest watersheds (such as the Río Blanco) having greater maximum stream power.

The ratio of total stream power to the coarse grain size (Ω/d84, W/m/m) shows the

relative influence of hydraulic forces (as stream power) to lithologic resistance (coarse

grain size). While, both stream power and grain size display a similar peaked trend in the

downstream direction, the ratio between the two shows an opposing trend. Along the

main stems, this ratio generally shows a positive trend in the downstream direction

(Figure 4.9), indicating the relative dominance of stream power in the lowland reaches,

and strong resistance by coarse grains in the headwater reaches. Using a threshold of

10,000 W/m/m to differentiate between supposed alluvial conditions and lithologic

controls (Wohl 2004), it is apparent that the transition from strong lithologic control to

more alluvial condition occurs approximately a third to a half of the profile length.

Furthermore, those watersheds having granodiorite substrate in the headwaters (Rio

Blanco and Rio Espiritu Santo) have a Ω/d84 in excess of this threshold, indicating they

have alluvial conditions in their sandy headwater reaches.

DISCUSSION

The results presented above indicate that the streams of the Luquillo Mountains

have an intricate connection between the underlying lithologic and hydraulic controls,

and the resulting profile shape, grain size distribution, channel geometry, and channel

energetics. Local-scale geologic factors such as the rock type, exposed in-channel

outcrops, and bedrock faults are seemingly dominant in determining the shape of the

longitudinal profiles. Different lithologies correspond to local variations in profile slope

130

and concavity. They also weather into unique particle sizes, and are associated with

specific channel geometries. Large immobile boulders are deposited in the channel by

landslides. Yet increasing discharge and hydraulic controls on sediment transport

override these lithologic influences and give rise to basin-scale patterns. This is

evidenced in that channel geometry and grain size are also strongly related to slope and

discharge (Figures 4.3, 4.8, and 4.9). Here we discuss the importance of each of these

basin-scale patterns and the implications for the dynamics of tropical mountain streams.

The longitudinal profile of each of the study rivers, although generally concave-

upward, display fragmented patterns consistent with lithologic control. The rivers have

slope breaks, knickpoints, and profile convexities that correlate with different rock types

and structural features of the underlying bedrock. The relative strength of different

bedrock types during fluvial incision can yield such segmented profiles (Brocard and van

der Beek 2006). Chemical and physical weathering, as well as debris flows, are the

dominant processes of bedrock incision in these rivers, as noted in other mountainous

drainages (Stock and Dietrich 2006). Many of the common processes of river incision

into bedrock, notably plucking, macro-abrasion, wear, and cavitation (Whipple 2004) are

rarely observed.

The concavity of the longitudinal profiles is related to underlying bedrock and

hillslope processes. Whipple (2004) discusses potential controls on channel concavity.

Low concavities (<0.4), such as that are seen in the headwaters of those rivers draining

volcaniclastics, are associated with short, steep drainages importantly influenced by

debris flows. High concavities (0.7-1.0), such as the gentle-gradient sandy-bed headwater

reaches on granodiorite, are associated with fully alluvial conditions. Convex profiles

131

(negative concavity), seen along granodiorite streams, are typically associated with

abrupt knickpoints owing either to pronounced along-stream changes in substrate

properties (VanLaningham 2003) or to spatial or temporal differences in rock uplift rate

(Whipple 2004). Lastly, extreme concavities (>1.0), present along the lowland and

alluvial reaches, are associates with transitions from incisional to depositional conditions.

Thus, along streams draining volcaniclastics, the transition from low concavity to high-

concavity at approximately 10 km2 marks a transition from dominant colluvial and

hillslope processes incisional channels to a depositional alluvial-type channels. Brummer

and Montgomery (2003) noted a similar break in the slope-area relationships at a

drainage area of 10 km2 in some coastal temperate streams, contending that the associated

change in concavity reflected a shift from dominant colluvial processes in headwater

channels to alluvial processes in downstream channels.

Downstream hydraulic geometry is considered well developed in all of these

basins, despite the influence of non-fluvial processes, differences in lithology, and local

structural features. Mountain rivers are considered to have well-developed DHG when

the coefficient of determination (r2) between discharge and at least two of the three

hydraulic variables is 0.5 or greater (Wohl 2004). The high r2 values for width (0.71) and

velocity (0.66) relationships in the Luquillo streams satisfy this criteria, so that the DHG

for the stream network is considered well-developed.

The well-developed hydraulic geometry can be attributed to the strong influence

of fluvial forces over lithologic resistance, as reflected in the ratio of stream power to

grain size. Wohl (2004) found that mountain streams have well developed DHG when the

ratio of total stream power to the coarse grain size, Ω/d84, is greater than 10,000 W/m/m.

132

Above this threshold, the river presumably has enough power to rework the coarse

sediment and adjust channel parameters in response to downstream changes in discharge.

Below this threshold, combinations of low stream power or large grain sizes inhibit the

river from developing strong DHG relationships. The average Ω/d84 ratio of all surveyed

reaches in the streams of the Luquillo Mountains is approximately 14,000 W/m/m, or

slightly above the threshold. Although the average grain size is very large, so is the

average stream power. Thus, the combination of high discharge and steep slopes

generates sufficient stream power to overcome lithologic controls and adjust channel

geometry accordingly over time.

Clark and Wilcock (2000) noted a DHG reversal in the lowland alluvial reaches in

some of these streams. Values of channel width, depth, and velocity either decreased or

were constant in the downstream direction in the lowland reaches. This hydraulic

geometry reversal trend was found on coastal plain alluvial reaches along the main stem,

or approximately the lower 33% of the main stem profile length. The authors attribute

this reversal to historic and modern land-use changes. Apparently, the shift from forest to

agriculture to urbanization over 400 years altered the sediment supply and flow regime.

Net aggradation of sediment during periods of land-clearance and recent net degradation

from heightened runoff due to urbanization have altered the balance that maintains

channel geometry. However, our data confirms that this reversal is strictly confined to the

lowland alluvial reaches. Hydraulic geometry remains relatively well-developed at the

basin-scale that spans four orders of magnitude in discharge.

The poor correlation between depth and discharge across all watersheds suggests

that local factors are a strong determinant of channel form. Bedrock outcrops, scour

133

pools, and the accumulation of large boulders can all locally determine depth. To

compensate for the small increase in depth, these streams increase drastically in velocity

with increasing downstream discharge. We document the largest downstream velocity

exponent reported for a mountain stream. This sizeable downstream increase in velocity

may be a result of the basin physiography and variations in flow resistance. These island

streams generally have shorter and more truncated profiles than streams on continental

land masses. Yet the short coastal plain is still relatively steep so that the flood waters

flow rapidly to the ocean with minimal resistance. Despite having faster average flow, the

downstream reaches are not the most energetic. Rather, upland streams that have lower

average velocity, but greater slope, shear stress, and flow resistance, expend the greatest

amount of energy.

The peaked distribution of grain size along the profiles of these rivers stands in

contrast to a common systematic downstream fining trend in many alluvial rivers

(Pizzuto 1995, Paola and Seal 1995). However, a similar systematic headwater

coarsening pattern has been noted in several mountain basins in western Washington

(Brummer and Montgomery 2003). In both western Washington and the Luquillo

Mountains, grain size and stream power maxima occur at approximately the same

location as the transition from debris-flow and landslide dominated channels to fluvially

dominated channels. This suggests that a tendency for downstream coarsening may be

ubiquitous in headwater reaches of mountain drainages where debris flow processes set

the channel gradient. Apparently, when landslides dominate the transport and routing of

sediment in low-order headwater channels, a coarsening trend occurs. Downstream fining

occurs as fluvial forces override colluvial forces as the driving sediment transport process

134

in high-order alluvial channels. These observations suggest that basin-wide trends in d50

are also in part hydraulically influenced by variations in stream power, as well as by

landslide deposits.

There is a complex interaction between profile slope, grain size, drainage area and

lithology, as noted by Hack (1957, 1960). Data from this study follow a similar

relationship between these three variables (Figure 4.8) as data from temperate piedmont

streams in Maryland and Virginia (Hack 1957). The streams in Luquillo display the same

adjustment between the grain size, drainage area, and slope as more gentle gradient

streams in a very different physiographic region. The same basic relationship holds even

though the Luquillo streams have steeper slopes and consequently greater d50:DA ratios.

Furthermore, rock type does not factor into this relationship, so that reaches on all

lithologies display the same relationship among the three variables.

The casual mechanisms associated this relationship (i.e. whether slope is

influenced by both the size of the sediment and discharge, or whether slope and discharge

determine the grain size) is seemingly time-scale dependent (sensu Schumm and Lichty

1965). Alluvial channels can adjust slope in response to transport capacity and sediment

supply such that slope is a dependent variable related to water and sediment discharge,

and grain size. Yet in the steep headwater bedrock channels where non-fluvial forces

dominate, slope is generally imposed by lithology, and becomes an independent variable

over the timescales of channel geometry adjustment. Channel sediment is shaped by

persistent short-term fluvial and colluvial processes that organize the bed surface upon a

slope set by longer-term erosion processes (Scatena 1995).

135

Despite the abundance of large boulders throughout the basin, much of the

interstitial bed-material among these boulders is readily mobile. The lack of correlation

between excess dimensionless shear stress and discharge (Figure 4.7) suggests that

throughout the stream network, these are “threshold channels” that are capable of

mobilizing moderately-sized sediment during bankfull floods. The constancy of excess

shear stress is a feature commonly associated with alluvial channels that can readily

adjust slope (Dade 2000). However, Luquillo streams are evidently adjusted to be

threshold channels, despite a geologically-imposed slope. Yet on longer time scales, the

profile slope of these upland channels changes over the course of drainage network

evolution. The upland channels adjust slope to the underlying lithology and consequently

influence the type of sediment that is delivered to the channels. The combination of these

processes and scales suggest that the resulting channel morphology is not exclusively

controlled by a single factor.

CONCLUSION

The morphology of the stream channels in the Luquillo Mountains are influenced

by a combination of both local lithologic controls and strong hydraulic forces. Slope and

grain size in many headwater areas are imposed by properties of the underlying lithology

and coarse sediment delivery by landslides. Longitudinal profiles and concavity are

strongly related to lithologic boundaries. At the reach-scale, non-fluvial factors such as

bedrock outcrops, knickpoints, and fault bends locally affect the channel morphology.

Hillslopes are strongly linked to channel dynamics and colluvial processes are dominant

in many headwater areas.

136

Within the framework set upon by local and non-fluvial constraints, there are

many basin-scale patterns that indicate the river functions similar to a fully alluvial river.

The presence of strongly developed hydraulic geometry relationships, grain size patterns

organized to slope and discharge, and high stream power relative to channel resistance

indicate the influence of overruling fluvial forces. Furthermore, excess dimensionless

shear stress at bankfull wavers around the threshold for sediment mobility indicating the

river is able to systematically transport sediment and organize its own morphology. These

basin-scale patterns attest to the ability of the forceful flow regime generated by the

humid tropical climate to sculpt mountainous streams that share some commonalities

with alluvial rivers.

ACKNOWLEDGEMENTS

The authors would like to thank Doug Jerolmack for his strengthening comments

on an earlier version of this manuscript. We also thank the International Institute of

Tropical Forestry for logistical support. Funding for this study was provided by the

National Science Foundation Biocomplexity Grant (NSF #030414)—Rivers, Roads, and

People: Complex Interactions of Overlapping Networks in Watersheds.

137



85: 157-169.

Bishop P. 2007. Long-term landscape evolution: linking tectonics and surface processes.


Blizard CR, Wohl EE. 1998. Relationships between hydraulic variables and bedload

transport in a subalpine channel, Colorado Rocky Mountains, U.S.A.


Briggs RP, Anguilar-Cortés E. 1980. Geologic map of the Fajardo and Cayo Icacos

quadrangles. United States Geological Survey Miscellaneous Field Studies Map I-

1153; 1:20,000 scale.

Brocard GY, van der Beek PA. 2006. Influence of incision rate, rock strength, and

bedload supply on bedrock river gradients and valley-flat widths: Field-based

evidence and calibrations from western Alpine rivers (southeast France).

Geological Society of America Special Paper 398: 101-126.

Brown ET, Stallard RF, Larsen MC, Raisbeck GN, Yiou F. 1995. Denudation rates

determined from the accumulation of in situ-produced 10Be in the Luquillo

Experimental Forest, Puerto Rico. Earth and Planetary Science Letters 129: 193-

202.

Brummer CJ, Montgomery DR. 2003. Downstream coarsening in headwater channels.

Water Resources Research 39(10): 1-12.

138


northeastern Puerto Rico. Geological Society of American Bulletin 112(12):

1763-1777.

Constantine CR, Mount JF, Florsheim JL. 2003. The effects of longitudinal differences in

gravity mobility on the downstream fining pattern in the Cosumnes River,

California. Journal of Geology 111: 233-241.





Cox DP, Marvin RF, M’Gonigle JW, McIntyre DH, Rogers CL. 1977. Potassium-argon

geochronology of some metamorphic, igneous and hydrothermal events in Puerto

Rico and the Virgin Islands. United States Geological Survey Journal of Research

5: 689-703.

Crosby BT, Whipple KX. 2006. Knickpoint initiation and distribution within fluvial

networks: 236 waterfalls in the Waipaoa River, North Island, New Zealand.


Dade WB, Friend PF. 1998. Grain size, sediment-transport regime and channel slope in

alluvial rivers. Journal of Geology 106: 661-675.

Dade WB. 2000. Grain size, sediment transport and alluvial channel pattern.


139

Finnegan NJ, Roe G, Montgomery DR, Hallet B. 2005. Controls on the channel width of

rivers: implications for modeling fluvial incision of bedrock. Geology 33: 229-

232.

Frizano J, Johnson AH, Vann DR, Scatena FN. 2002. Soil phosphorus fractionation

during forest development on landslides scars in the Luquillo Mountains, Puerto

Rico. Biotropica 34(1): 17-26.



Journal of Science 32(4): 413-424.

Golden LA, Spring GS. 2006. Channel geometry, median grain size, and stream power in

small mountain streams. Geomorphology 78: 64-76.

Goldrick G, Bishop P. 2007. Regional analysis of bedrock stream long profiles:

evaluation of Hack’s SL form, and formulation and assessment of an alternative

(the DS form). Earth Surface Processes and Landforms 32: 649-671.

Graf WL. 1983. Downstream changes in stream power in the Henry Mountains, Utah.

Annals of the Association of American Geographers 73: 373-387.

Graham A, Jarzen DM. 1969. Studies in neotropical paleobotany. I. The Oligocene

communities of Puerto Rico. Annals of the Missouri Botanical Garden 56: 308-

357.



Society of America Bulletin 102(3): 340-352.

140

Grimm MM, Wohl EE, Jarrett RD. 1995. Coarse-sediment distribution as evidence of an

elevation limit for flash flooding, Bear Creek, Colorado. Geomorphology 14: 199-

210.


flow and large floods. American Journal of Science 275: 825-847.



301-315.





136.

Hack JT. 1957. Studies of longitudinal stream profiles in Virginia and Maryland. United

States Geological Survey Professional Paper 294-B; 1-97.

Hack JT. 1960. Interpretation of erosional topography in humid temperate regions.

American Journal of Science 258-A: 80-97.

Heartsill-Scalley T, Scatena FN, Estrada C, McDowell WH, Lugo AE. 2007. Disturbance

and long-term patterns of rainfall and throughfall nutrient fluxes in a subtropical

wet forest in Puerto Rico. Journal of Hydrology 333: 472-485.

Kesler SE, Sutter JF. 1979. Compositional evolution of intrusive rocks in the eastern

Greater Antilles island arc. Geology 7: 197-200.

141

Kirby E, Whipple K. 2001. Quantifying differential rock-uplift rates via stream profile

analysis. Geology 29: 415-418.

Knighton AD. 1999. Downstream variation in stream power. Geomorphology 29: 293-

306.

Larsen MC. 1997. Tropical geomorphology and geomorphic work: a study of

geomorphic processes and sediment and water budgets in montane humid-tropical

forested and developed watersheds, Puerto Rico. PhD thesis, University of

Colorado, Boulder, CO.

Larsen MC, Torres-Sánchez AJ. 1998. The frequency and distribution of recent

landslides in three montane tropical regions of Puerto Rico. Geomorphology 24,

309-331.

Larsen MC, Torres-Sánchez AJ, Concepción IM. 1999. Slopewash, surface runoff and

fine-litter transport in forest and landslide scars in humid-tropical steeplands,

Luquillo Experimental Forest, Puerto Rico. Earth Surface Processes and

Landforms 24: 481-502.

Lenzi MA, Mao L, Comiti F. 2004. Magnitude-frequency analysis of bed load data in an

Alpine boulder bed stream. Water Resources Research 40: W07201.

DOI:10.1029/2003WR002961



252; 1-57.

Lecce SA. 1997. Nonlinear downstream changes in stream power on Wisconsin’s Blue

River. Annals of the Association of American Geographers 87: 471-486.

142

Lewis LA. 1969. Some fluvial characteristics of the Manati Basin, Puerto Rico. Annals of

the Association of American Geographers 59: 280-293.

McPherson HJ. 1971. Downstream changes in sediment character in a high energy

mountain stream channel. Arctic and Alpine Research 3: 65-79.

Molnar P, Ramirez JA. 2002. On downstream hydraulic geometry and optimal energy

expenditure: case study of the Ashley and Taieri Rivers. Journal of Hydrology

259: 105-115.

Monroe WH. 1980. Some tropical landforms of Puerto Rico. United States Geological

Survey Professional Paper 1159; 1-93.



Montgomery, DR, Gran KB. 2001. Downstream variations in the width of bedrock

channels. Water Resources Research 37: 1841-1846.

Mueller ER, Pitlick J, Nelson JM. 2005. Variation in the reference Shields stress for bed

load transport in gravel-bed streams and rivers. Water Resources Research 41:

W04006. DOI:10.1029/2004WR003692

Niedzalek JL, Ogden FM. 2005. Runoff production in the Upper Río Chagres watershed,

Panama. In: The Río Chagres, Panama: a multidisciplinary profile of a tropical

watershed. Harmon RS (ed). Springer: Dordrecht, The Netherlands; 149-168.

Paola C, Seal R. 1995. Grain size patchiness as a cause of selective deposition and

downstream fining. Water Resources Research 31: 1395-1408.

Park CC. 1977. World-wide variations in the width of bedrock channels: an analysis and

some observations. Journal of Hydrology 33: 133-146.

143

Parker G, Wilcock PR, Paola C, Dietrich WE, Pitlick J. 2007. Physical basis for quasi-

universal relations describing bankfull hydraulic geometry of single-thread

gravel-bed rivers. Journal of Geophysical Research Earth Surface 112: F04005.

DOI: 10.1029/2006JF000549

Pike AS. 2006. Application of digital terrain analysis to estimate hydrological variables

in the Luquillo Mountains of Puerto Rico. In: Climate Variability and

Change–Hydrological Impacts (Proceedings of the Fifth FRIEND World

Conference held at Havana, Cuba, November 2006). Demuth S, Gustard A,

Planos E, Scatena F, Servat E (eds). IAHS Publication 308. IAHS Press:

Wallingford, UK; 81-85.

Pike, AS., Scatena, F.N. 2008. A bankfull analog for tropical montane streams based on

riparian features. Journal of Hydrology. In review.

Pizzuto JE. 1995. Downstream fining in a network of gravel-bedded rivers. Water

Resources Research 31: 753-759.

Rengers F, Wohl E. 2007. Trends of grain sizes on gravel bars in the Rio Chagres,

Panama. Geomorphology 83: 282-293.

Rice S. 1999. The nature and controls on downstream fining within sedimentary links.

Journal of Sedimentary Research 69: 32-39.

Scatena FN. 1989. An introduction to the physiography and history of the Bisley

Experimental Watersheds in the Luquillo Mountains of Puerto Rico. United States

Department of Agricultural Forest Service General Technical Report SO-72; 1-22.

144





D.C.; 103-111.

Scatena FN. 1998. Climate change and the Luquillo Experimental Forest of Puerto Rico:

assessing the impacts of various climate change scenarios. American Water

Resources Association TPS 98(2): 91-96.



Scatena FN, Lugo AE. 1995. Geomorphology, disturbance, and the soil and vegetation of

two subtropical wet steepland watersheds of Puerto Rico. Geomorphology 13:

199-213.

Scatena FN, Planos-Gutiérrez EO, Schellekens J. 2004. Natural disturbances and the

hydrology of humid tropical forests. In: Forests, Water, and People in the Humid

Tropics. Bonell M, Bruijnzeel LA (eds). Cambridge University Press: Cambridge,

UK; 5-28.

Schumm SA, Lichty RW. 1965. Time, space and causality in geomorphology. American





145

Schellekens J, Scatena FN, Bruijnzeel LA, van Dijk AIJM, Groen MMA, van Hogezand

RJP. 2004. Stormflow generation in a small rainforest catchment in the Luquillo

Experimental Forest, Puerto Rico. Hydrological Processes 18: 505-530.

Seiders VM. 1971a. Geologic map of the El Yunque quadrangle, Puerto Rico. United


1:20,000 scale.

Seiders VM. 1971b. Cretaceous and lower Tertiary stratigraphy of the Gurabo and El

Yunque quadrangles, Puerto Rico. United States Geological Survey Bulletin

1294-F; 1-58.

Seidl MA, Dietrich WE, Kirchner JW. 1994. Longitudinal profile development into

bedrock: an analysis of Hawaiian channels. Journal of Geology 102: 457-474.

Simon A, Guzman-Rios S. 1990. Sediment discharge from a montane basin, Puerto Rico:

Implications of erosion processes and rates in the humid tropics. In: Research

Needs and Applications to Reduce Erosion and Sedimentation in Tropical

Steeplands (Proceedings of the Fiji Symposium, June 1999). IAHS-AIHS

Publication 192, International Association of Hydrological Sciences: Washington,

DC; 35-47.

Singh VP. 2003. On the theories of downstream hydraulic geometry. International

Journal of Sediment Research 18: 196-218.

Snyder NP, Whipple KX, Tucker GE, Merritts DJ. 2003. Importance of a stochastic

distribution of floods and erosion thresholds in the bedrock river incision

problem. Journal of Geophysical Research 108: DOI: 10.1029/2001JB00165

146

Stock JD, Dietrich WE. 2006. Erosion of steepland valleys by debris flows. Geological

Society of America Bulletin 118: 1125-1148.


streams of Colorado. Journal of Hydrology 290: 137-151.

Trainor K, Church M. 2003. Quantifying variability in stream channel morphology.

Water Resources Research 39(9): 1248 1259.

VanLaningham S, Meigs A, Golfinger C. 2006. The effects of rock uplift and rock

resistance on river morphology in a subduction zone forearc, Oregon, USA. Earth


van der Molen MK. 2002. Meteorological impacts and land use change in the maritime

tropics. PhD thesis, Vrije Universiteit, Amsterdam, The Netherlands.

Whipple KX. 2004. Bedrock rivers and the geomorphology of active orogens. Annual

Review of Earth and Planetary Sciences 32: 151-185.

White AF, Blum AE, Schulz MS, Vivit DV, Stonestorm DA, Larsen M, Murphy SF,

Eberl D. 1998. Chemical weathering in a tropical watershed, Luquillo Mountains

Puerto Rico: I. Long-term versus short-term weathering fluxes. Geochimica et

Cosmochimica Acta 62(2): 209-226.

Wohl E. 2004. Limits of downstream hydraulic geometry. Geology 32: 897-900.

Wohl E. 2005. Downstream hydraulic geometry along a tropical mountain river. In: The

Río Chagres, Panama: a multidisciplinary profile of a tropical watershed. Harmon

RS (ed). Springer: Dordrecht, The Netherlands; 169-188.

147

Wohl E, Greenbaum N, Schick AP, Baker VR. 1994. Controls on bedrock channel

incision along Nahal Paran, Israel. Earth Surface Processes and Landforms 19: 1-

13.

Wohl E, Merritt DM. 2001. Bedrock channel morphology. Geological Society of

America Bulletin 113: 1205-1212.

Wohl E, Kuzma KN, Brown NE. 2004. Reach-scale channel geometry of a mountain

river. Earth Surface Processes and Landforms 29: 969-981.

Wohl E, Wilcox A. 2005. Channel geometry of mountain stream in New Zealand. Journal


Wolman MG. 1954. A method of sampling coarse river-bed material. Transactions of the

American Geophysical Union 35: 951-956.



Wu W, Hall CAS, Scatena FN. 2007. Modelling the impact of recent land-cover changes

on the stream flows in northeastern Puerto Rico. Hydrological Processes 21:

2944-2956.

148

CHAPTER 5

MULTISCALE LINKAGES BETWEEN GEOMORPHOLOGY AND AQUATIC HABITAT IN A TROPICAL MONTANE STREAM NETWORK PUERTO RICO*

A.S. Pike, C.L. Hein, F.N. Scatena, T.A. Crowl, J.F. Blanco, and A.P Covich

ABSTRACT

Linkages between stream geomorphology and migratory aquatic fauna at the

network and pool scales were investigated in three tropical montane watersheds draining

the Luquillo Mountains in Northeastern Puerto Rico. A total of 113 pools at 33 sites,

capturing the full range of stream size and geomorphic conditions in the region, were

physically surveyed and sampled for fish, shrimp, crabs, and snails. In addition, similar

data at 49 pools in a 1st order headwaters stream were used to investigate local-scale

variability in geomorphology and decapod abundance. Principal components analysis

(PCA) identified two geomorphic gradients that account for 59% of the variance in

among-pool variation at the landscape scale, a longitudinal gradient and a substrate

* This chapter contains work and analysis (including a combined dataset) developed in collaboration with Catherine Hein of Utah State University. Portions of this chapter will be submitted for publication in two separate, though complementary, journal articles. One of these is lead-authored by the author of this dissertation, and the other is lead-authored by Catherine Hein.

1. A.S. Pike, C.L. Hein, F.N. Scatena, T.A. Crowl, J.F. Blanco, A.P. Covich. Multiscale linkages between geomorphology and aquatic habitat in a tropical montane stream network, Puerto Rico.

2. C Hein, C.L., A.S. Pike, J.F. Blanco, F.N. Scatena, K.R. Sherrill, A.P. Covich, and T.A. Crowl. Influence of road networks on the community structure of diadromous fauna in tropical island streams.

Several analyses presented in this chapter are modified from the second publication (in preparation), on which I am a second author. These include Table 5.1, Figure 5.2, Table 5.4, Figure 5.3, Figure 5.6, and Table 5.5. These tables and figures illustrate concepts on the distribution of the presence/absence of species that are discussed in Hein et al. (in prep). However, the modified analyses presented in this chapter are unique, due to use of different statistical methods and models on the same dataset, and represent the work of the author of this dissertation. Because they intend to illustrate similar concepts and conclusion to those drawn in Hein et al. (in prep), the presentation of these results and figures are referenced appropriately in the text and captions, as Hein et al. (in prep). Further data analysis, interpretation of results, and all manuscript writing were done by the author of this disseration

149

gradient, and correlations with species presence/absence indicate strong habitat

preferences along these gradients. Non-parametric multiplicative regression (NPMR)

models of species presence/absence estimating the probability of occurrence along the

longitudinal profiles confirmed these patterns and indicate abrupt transitions in

community composition at waterfalls. Predatory fish (gobies, mullets, sleepers, eels)

occupied reaches below waterfalls that hinder their upstream migration, whereas decapod

(atyid and palaemonid shrimp, and crabs) were more common in the headwater reaches

above waterfalls, suggesting that longitudinal gradients are more important than pool-

scale geomorphic and hydraulic factors in governing species distributions at the

landscape scale. At the network-scale above waterfalls, where predatory fish are absent,

the main geomorphic gradients determined by PCA do not adequately predict the

abundance of decapods. At the pool-scale, Stepwise Multiple Linear Regression (SMLR)

models indicate that pool size and substrate characteristics influence decapod

abundances. Furthermore, pool-size and pool-spacing were found to vary predictably

with drainage area. However, gradient analysis shows that the geomorphic features

structuring aquatic habitat do not always vary systematically throughout the stream

network, and are rather patchy at all scales. Our results contrast with the River

Continuum Concept, which argues that stream assemblages vary predictably along steam

size gradients. Rather, our findings are more consistent with more recent ecological

theories (Process Domains Concept, Network Dynamics Hypothesis and Hierarchical

Patch Dynamics perspective) that address naturally occurring breaks in the geomorphic

and network continuum and recognize the importance of each stream’s hierarchical

pattern of habitat transitions.

150

INTRODUCTION

Understanding how geomorphic processes influence the spatial distribution and

abundance of aquatic fauna throughout a stream network are important aspects of stream

ecology (Gorman and Karr 1978, Angermeier and Karr 1983, Power et al. 1988, Newson

and Newson 2000, Walters et al. 2003). Overlapping habitat preferences of interacting

organisms not only determine community composition, but their interactions can also

structure food webs and affect ecological function (Reagan and Waide 1996).

Information on habitat preferences of species and the geomorphic processes structuring

stream habitat are critical for developing conservation targets to protect riverine and

coastal ecosystems, particularly in ecologically sensitive tropical streams. As the threat to

tropical freshwater ecosystems increases through dam-building and landuse changes

(Holmquist et al. 1997, Pringle and Scatena 1998, Pringle et al. 2000, March et al. 2003,

Anderson-Olivas et al. 2006, Greathouse et al. 2006), knowing how aquatic organisms

respond to geomorphic gradients in relatively pristine river networks provides the

baseline information needed for conservation and restoration efforts.

Several theories have been proposed to describe the linkages between the

geomorphology of streams and the spatial distribution of aquatic organisms. These

emphasize the roles of systematic longitudinal gradients (Vannote et al. 1981), patchiness

and heterogeneity (Pringle et al. 1988, Townsend 1989, Crowl et al. 1997), hydraulics

(Statzner and Higler 1986), geomorphic disturbance (Montgomery 1999), multiscale

habitat formation (Wu and Loucks 1995, Poole 2002), and river network structure (Benda

et al. 2004). Although they apply to many stream systems worldwide, it is not known

whether their predictions of species distributions hold in tropical island streams where

151

migratory aquatic fauna interact with short, steep-gradient, frequently-flooded channels

that are punctuated by waterfalls. To test whether some specific predictions of these

theories hold in tropical mountain streams, this study builds on a prior analysis of

geomorphic controls on food web structure (Hein et al., in prep) by including geomorphic

gradients at the drainage network and pool scales in the streams of the Luquillo

Mountains of Puerto Rico, and investigates the consequent response on the distribution

and abundance of fish and decapods.

The River Continuum Concept (RCC) relates longitudinal variations in aquatic

communities to systematic downstream changes in river systems (Vannote et al. 1981),

and is arguably the most influential perspective in stream ecology. The work was

primarily concerned with streams in humid temperate areas and predicts that aquatic

habitat, food sources, and populations shift downstream in response to longitudinal

changes in channel morphology, canopy openness, light, substrate size, and stream flow.

The RCC posits these physical variables present a gradational continuum of habitat

conditions that control aquatic community composition from small headwater streams to

large floodplain rivers. Although the RCC has been an effective framework for

understanding stream attributes within large drainage networks, longitudinal relationships

in Puerto Rico streams may be obscured by local factors within parts of networks and

may not apply to smaller drainage basins (Covich et al. 1996, Greathouse and Pringle

2006).

Other researchers have promoted the concept of patch dynamics to characterize

geomorphic patterns and processes in heterogeneous stream environments (Pringle et al.

1988, Townsend 1989). Many biological communities are influence by the division of

152

landscapes into patches maintained by either disturbance or spatial transition in the

processes creating and maintaining habitats. This approach has been useful for comparing

conditions and communities within and between patches and at spatial scales ranging

from microhabitats within pools to heterogeneous reaches. However, a key problem in

applying patch dynamics concepts in montane watersheds is the lack of ability to predict

areas in the network that are characterized by different patch-forming processes.

Still, other researchers argue that neither the RCC nor patch dynamics explicitly

address the spatial structure of geomorphic controls on aquatic habitat. Statzner and

Higler (1986) suggested that hydraulic forces regulate communities on a world-wide

scale. From the source to the mouth of a stream, zones of transition in ‘stream hydraulics’

occur, to which the general patterns of stream faunal assemblages can be related.

Montgomery (1999) proposed the Process Domain Hypothesis (PDC), which

hypothesizes that spatial variability in geomorphic processes governs stream habitat and

disturbance regimes that influence ecosystem structure and dynamics. Process domains

are predictable areas of the landscape within which distinct geomorphic processes operate

and thereby impart spatial variability to lotic communities at the landscape scale.

Identification of these processes can provide a mechanistic understanding of the

distribution of habitats and stream biota predicted by the river continuum and patch

dynamics model.

Recent work has attempted to unify both the concept of patches and longitudinal

continuums at coarse spatial scales. The “Hierarchical Patch Dynamics” perspective (Wu

and Loucks 1995, Poole 2002), and “Network Dynamics Hypothesis” (Benda et al. 2004)

progress beyond the RCC’s linear perspective by considering the stream system as a non-

153

linear network. These models have been developed in montane landscapes and describe

each river network as a discontinuum comprised of a longitudinal series of alternating

stream segments with different geomorphological structures. Each confluence in the

stream network punctuates the discontinuum because the sudden change in stream

characteristics creates a gap in the expected systematic pattern of downstream transitions.

Furthermore, the hierarchical patch dynamics perspective asserts that differences in

geomorphic processes structuring aquatic habitat at varying scales give rise to patchiness,

so dividing otherwise systematic gradients in aquatic habitat into a series of multiscale

patches. This discontinuum view recognizes general trends in habitat characteristics

along the longitudinal profiles, but understands the importance of each stream’s

hierarchical pattern of habitat transitions.

Although the RCC and these other hypotheses of geomorphic and biological

linkages have been effectively applied to many stream systems worldwide, mountainous

tropical island streams have several unique physical characteristics that may affect

whether they conform to such predictions (Smith et al. 2003). First, tropical islands tend

to have short, steep drainages, often entering the ocean as low or mid-order streams. They

typically display few river forms (sensu Rosgen 1994) as they flow from the mountains

steeply to the coast. Consequently, they do not vary as a continuum from headwaters to

very large rivers, and longitudinal patterns could differ from continental drainages by

being truncated (Greathouse and Pringle 2006). Second, humid tropical streams often

have frequent torrential flows. Intense tropical precipitation generates stochastic, high

power flow regime, with large floods occurring several times per year. Organisms must

be able to tolerate such a disturbance regime and may seek out habitat that provides

154

refuge from swift flow. Third, the steep channels display strong heterogeneity of

geomorphic and hydraulic conditions from reach to reach. The channels have complex

downstream hydraulic geometry and grain size relationships (see Chapter 4), and the

mountainous reaches generally lack floodplains that provide critical ecological and

biogeochemical linkages in many rivers. Furthermore, longitudinal profiles are typically

punctuated by waterfalls, which may pose a barrier to upstream migration.

The aquatic organisms in tropical montane streams may be especially sensitive to

geomorphic gradients because most freshwater species are diadromous, requiring

migration between freshwater and salt waters to complete their life cycles (Covich 1988).

Freshwater assemblages of fish and macroinvertebrates on tropical islands are generally

dominated by migratory freshwater shrimps, fish, eels, and snails (Bass 2003, Smith et al.

2003). The freshwater shrimps spend adulthood in the headwaters and release eggs

during high flow to be transported to the saline waters of the estuary (March 1998,

Benstead et al. 2000). As juveniles or post-larvae, they migrate upstream. Most tropical

island fishes are catadromous and spend their adulthood in the rivers but migrate to the

estuary to spawn, after which the juveniles migrate upstream to feed (Covich and

McDowell 1996). Since all these species they must ascend the river network, hydrologic

and geomorphic barriers may be important in determining their distributions.

In this study, we investigate relationships between geomorphology and aquatic

fauna in the streams draining the Luquillo Mountains in northeastern Puerto Rico. This

study complements a companion paper (Hein et al., in prep) that assesses changes in

community composition at the landscape scale, and biological interactions between

species. Expanding upon those results, this study investigates the biological response to

155

key geomorphic gradients that structure aquatic habitat at the network and pool scales.

We have four primary objectives. First, we identify the key geomorphic gradients present

at the network and pool scales in three relatively pristine tropical montane stream

networks. Second, we investigate how influential these gradients are to determining the

distribution of species. Third, we use the best landscape-scale predictors of species

presence/absence to map the longitudinal distribution of species. Fourth, at pools where

decapods are present, we develop models to predict their relative abundances based on

local-scale pool geomorphology. We ultimately compare our results with predictions of

existing conceptual models of the longitudinal variations in aquatic habitat.

STUDY AREA

This study was conducted in the streams draining the Luquillo Mountains in

Northeastern Puerto Rico. The Luquillo Mountains rise steeply from sea-level to over

1000m in elevation over a distance of 15 to 20km. The streams have their headwaters in

the Luquillo Experimental Forest (LEF), a 113km2 protected forest reserve under the

management of the United States Forest Service. The study area consists of three adjacent

watersheds draining the LEF: Río Mameyes, Río Espiritu Santo, and Río Fajardo (Figure

5.1). The watersheds are similar physiographically, although they vary in size. Drainage

areas of these watersheds are 44km2, 92km2, and 67km2, respectively.

The climate is maritime subtropical, with mean annual temperatures ranging from

an average of 22ºC in the winter to 30ºC in the summer (Ramirez and Melendez-Colom

2003). Due to the tropical climate, stream temperature displays little longitudinal and

temporal variation. Rainfall is near-daily occurrence, and mean annual rainfall is

156

Figure 5.1 a) Location of 33 sample sites located in three adjacent watersheds draining

the Luquillo Mountains in northeastern Puerto Rico. Sites were chosen to span the

longitudinal gradient from the headwaters to near the coast. The location of the first

major waterfalls, representing a barrier to aquatic migration, are also shown. b) At each

sample site, between 2 and 4 replicate pools were surveyed, allowing comparison of the

variability among pools within a reach, and pools between reaches. c) An additional 49

pools in Quebrada Prieta, a 1st order headwater stream, are used in this study to compare

the variability of pools within a homogenous reach.

157

Figure 5.1 cont.

158

approximately 3500mm/yr at mid-elevations (Garcia-Martino et al. 1996). The climate is

weakly seasonal, and high-intensity rainfall events and floods can occur in any given

month (Schellekens et al. 1999).

Similar to montane streams in the Greater Antilles in general, Luquillo streams

have steep gradients, channels lined with coarse boulder-sized sediment, numerous

bedrock cascades, and abrupt waterfalls (up to 30m in height) (Ahmad et al. 1993). First-

order perennial streams have bouldery channels in steeply sloped reaches, and clay and

soil-lined channels in reaches with more gentle slopes. Second and third-order streams

have steep gradient reaches, exposed bedrock channels, large boulders, and periodic

waterfalls. Many of the upland streams are characterized by cascade and step-pool

morphologies, whereas the lower reaches are plane bed and pool-riffle sequences (sensu

Grant et al. 1990, Montgomery and Buffington 1997, Trainor and Church 2003). Due to

rapid decomposition, these channels lack the large coarse woody debris dams that create

aquatic habitat in many channels in humid temperate environments (Covich and Crowl,

1990). Fourth and fifth-order streams occur only at lower elevations along the coastal

plain as gentle gradient pool-riffle sequences.

There are three dominant lithologies underlying the study region: volcaniclastics,

granodiorite, and coastal plain alluvium (Seiders 1971). The morphology of both the

stream channel and the banks, as well as the composition of the channel bed, are directly

related to the underlying lithology (Ahmad et al. 1993). Streams draining volcaniclastics

are steep and typically have a bed composed of large boulders (up to several meters in

diameter), interspersed with finer cobbles and gravels, as well as sporadic bedrock

outcrops (see Chapter 4). In contrast, streams draining granodiorite are almost entirely

159

composed of sand and large case-hardened boulders that can be several meters in

diameter. Streams flowing across the coastal plain alluvium typically have a

comparatively gentle gradient compared to channels on bedrock, and have channel beds

composed of cobbles, gravels, and fines

All three watersheds currently drain protected primary forest in their headwaters,

mature secondary forest at lower elevations, and both abandoned (reforesting)

agricultural fields/grazing pastures and urbanized developments along the coastal plain.

Each river flows through a mangrove-lined estuary before reaching the coast. Despite

some development along the coastal plain, these watersheds are considered among the

most pristine in Puerto Rico (Santos-Román et al. 2003). Both the Rio Espiritu Santo and

Rio Fajardo have a low-head dam spanning the main stem at low elevations, and the Rio

Mameyes has a non-dam water intake device. However, these obstructions do not pose a

significant barrier to upstream migration (March et al. 2003).

Water quality often plays a strong role in species distributions, but the quality of

water in the study watersheds is relatively pure. Water chemistry data indicate that major

cations and anion concentrations do not exceed water quality standards (McDowell and

1994, Scatena et al. in review). Islandwide, water quality is generally covariate with land

use and discharge (Santos-Román et al. 2003). Forested upland watersheds have

relatively low nutrient concentrations (McDowell and Asbury 1994). In contrast, nutrient-

rich sewage effluent and/or agricultural runoff often released into lower elevation streams

creates poorer water quality conditions near both urbanized and agricultural/pasture areas

(Scatena 2001). Although there is some development at lower elevations, the high

discharge of these rainforest streams significantly dilutes the relatively small amount of

160

effluent delivered to the channels, and water quality does not change appreciably along

the length of the Rio Mameyes (Ortiz-Zayas et al. 2005). Furthermore, the presence of

freshwater snails—a sensitive bioindicator of water quality—at lower and mid elevations

in all of the study streams confirms the strong chemical integrity of these rivers (Blanco

and Scatena 2006). Consequently, it is assumed that any variations in water quality

between study sites is minor and that it should correlate with the landcover of the

upstream catchment.

Stream Community

Similar to other pan-tropical islands, the stream community in the Luquillo

mountains consists predominantly of diadromous fishes including gobies, sleepers, and

eels, atyid and palaemonid shrimp, freshwater crabs, and neritid snails. All of these

species are require direct linkages between fresh and salt water to complete their life

cycles. Diadromous migrations of fishes are thought to be a response to difference in

aquatic productivity between oceanic and stream habitats. Gross et al. (1988) notes that

catadromy is prevalent at tropical latitudes because fresh waters are more productive

relative to the ocean and fish consequently migrate upstream to feed.

One species of river goby (Awaous Tajasica) and two species of sleepers

(Gobiomorus dormitor and Eleotris pisonis) are common at low elevations (Covich and

McDowell 1996). These gobies ascend rivers as juveniles and return to brackish waters to

spawn. Both the river goby A. tajasica and bigmouth sleeper G. dormitor are bottom-

dwelling fish and feed on juvenile shrimp and aquatic insects. The spinycheek sleeper, E.

pisonis, is found in shallow, muddy, and sandy bottom streams, and prefers estuarine

environments and low-elevation tributaries with little to moderate salinity.

161

The mountain mullet (Agonostomus monticola) is an omnivorous fish found at

low to intermediate elevations and less steeply sloped channels. Mountain mullets move

downstream to spawn in brackish waters but spend most of their adult lives in fresh

water. They are well suited for life in turbulent mountainous streams with deep pools and

feed on shrimp, insects, and algae. Another major predator at low to mid-elevations is the

American eel (Anguilla rostrata). Eels primarily feed on shrimp, aquatic insects, and

small fishes, and typically grow to 50-60cm in length. Worldwide, fifteen species of eels

all share the same basic life cycle: they spawn at sea and after long-distance migrations

from the ocean (specifically from the Sargasso Sea in the Atlantic Ocean for American

and European eels), juveniles ascend upstream to specific areas for feeding (Erdman et al.

1984). The sirajo goby, Sicydium plumieri, is common throughout the streams,

particularly at higher elevations (Keith 2003). These fish are herbivorous and actively

graze on periphyton. Juveniles migrate upstream from the seas and grow to 15cm as

adults. They can climb up the edges of steep waterfalls using their specialized mouth

parts and pelvic fin sucking discs to ascend to high elevations (Covich and McDowell

1996). In Hawaii, similar gobies have been observed climbing vertical waterfalls up to

350m high (Schoenfuss and Blob 2003).

Atyid shrimp are the most abundant large consumers in these streams (Crowl et

al. 2000). There are three primary species of Atya: A. lanipes, A. innocuous, and A.

scabra. They are considered both scrapers and filter feeders, feeding on leaf litter, algae

and lichen, and suspended particles in flowing water (Covich and McDowell 1996). The

common A. lanipes and A. innocuous spend most of their time near the substrate in pools,

whereas the rare A. scabra prefer riffles. It is thought that atyid shrimp prefer stream

162

channels with large boulders in which they can penetrate crevices to feed and escape

larger predators (Covich et al. 1996). A fourth species, Xiphocaris elongata (Yellow-

nosed shrimp), is also very common throughout the streams. They are shredders, using

small saw-like pinchers to shred pieces of leaves (Crowl et al. 2006). They are very

active swimmers and spend most of their time in the water column. Juveniles migrate

upstream in masses along the wetted channel margin and they are able to climb sheer

waterfalls (March et al. 2003). Palaemonid shrimp, Macrobrachium spp., are widespread

throughout Caribbean and gulf coasts of the neotropics (Bass 2003, Covich et al. 2006),

and several species of this genus inhabit the Luquillo streams. Macrobrachium carcinus

(bigclaw river shrimp) is one of the largest (Covich and McDowell 1996), and are

generally found in upper portions of rivers where currents are rapid. Macrobrachium

species are omnivorous; usually feeding on detritus and on smaller shrimps. An

additional four species, M. acanthurus, M. crenulatum, M. heterochirus, and M.

faustinum have been reported in the streams (Covich et al. 2006). Females produce larvae

that must reach brackish water to complete their development before returning upstream

to grow. They can spend short amounts of time outside of water provided that the

relatively humidity is high, and can consequently climb waterfalls along channel margins

(Covich et al. 2000).

The amphibious crab, Epilobocera sinuatifrons, completes its life cycle entirely in

fresh water–a unique life cycle for crabs (Covich and McDowell 1996). They are usually

found in gravel beds and leaf packs in streams, although adults can move on land. They

are widespread in greater Antilles between 100m and 300m. Lastly, the neritid snail,

Neritina virginea, is common at low elevations in larger order rivers. They feed on lichen

163

and algae on boulders, and migrate slowly upstream in masses along wetted boulders and

channel margins (Schneider and Lyons 1993, Blanco and Scatena 2005). Damaged shells

indicate they occasionally fall prey to predatory fish. They require hydrologic

connectivity between the sea and the estuary; they are absent in Puerto Rican rivers that

have active sand-bars at the coast blocking access to the ocean (Blanco and Scatena

2006). They are also highly sensitive to water quality changes, and their migratory

passages may be blocked by poorly constructed road crossings (Blanco and Scatena

2007).

METHODS

Field Methods

In order to capture the landscape, reach, and pool-scale influences of

geomorphology on aquatic biota, we utilized two datasets in this study. The first dataset,

developed and sampled in collaboration with Catherine Hein and Dr. Todd Crowl of Utah

State University, consists of complementary geomorphic survey data and biological

sampling at 113 pools located at 33 sites in three watersheds (Figure 5.1). Due to the

breadth of these sites throughout the stream network, this data was used to assess the

influence of landscape and reach scale geomorphology. The second dataset consists

geomorphic and biological surveys at a series of 49 pools in a 1km reach in Quebrada

Prieta, a mid to upper elevation 1st order headwater channel. Relationships between the

geomorphology and the abundance of decapods using this second dataset shows the

influence of local-scale pool geomorphology on decapod abundance, as the reach is

relatively homogenous and does not vary along any landscape-scale gradients. This

164

dataset was collected by Coralys Ortiz Maldonado and María Ocasio, two undergraduate

interns that were supervised by the authors of this paper as part of the Research

Experience for Undergraduates (REU) program, and is described in further detail in

Ortiz-Maldonado (2005).

We physically surveyed the stream channel and sampled for fish, shrimp, crabs,

and snails at 33 sites in three watersheds: Rio Espiritu Santo, Rio Mameyes, and Rio

Fajardo. Sites were selected to capture the range of geomorphic conditions present

throughout the stream network, ranging from upper-elevation headwater streams

(drainage area, 0.1km2) to lowland rivers (58km2). At each site, two to five pools were

sampled for a total of 113 pools. Pools were surveyed and sampled during the summer

months (June-August) of 2004-2006.

A combination of trapping and electrofishing methods were used to sample the

fish, decapods, and snails (Hein et al., in prep). Trapping was done according to the

procedures developed by the Luquillo LTER program (http://luq.lternet.edu). All

individuals were identified to the species level, except juvenile Macrobrachium, which

were excluded from this analysis (Table 5.1).

To describe the geomorphic environment of each sampling reach, we measured

and estimated a total of 57 geomorphic, hydraulic, and topographic variables using a

combination of field measurements and GIS analysis (see Appendix). In the field, cross-

sections were surveyed at each pool at a uniform section. Relative distance and elevation

were measured at evenly spaced intervals along a transect spanning from vegetated bank

to bank. Channel top width, average depth, and cross-sectional area at both baseflow and

active-channel (bankfull) stage were calculated for each cross-section. Pool length, and

165

T

able

5.1

Spe

cies

iden

tifie

d du

ring

trap

and

elec

trofis

h sa

mpl

ing.

Fre

quen

cy is

the

perc

enta

ge o

f site

s w

here

a s

peci

es w

as o

bser

ved,

with

the

corr

espo

ndin

g ra

nk. (

Mod

ified

from

Hei

n et

al.,

in p

rep)

.

Cod

e Sp

ecie

s C

omm

on N

ame

Fam

ily

Life

His

tory

Fr

eq.

Freq

. R

ank

agom

on

Ago

nost

omus

mon

ticol

a M

ount

ain

mul

let

Mug

ilidae

C

atad

rom

ous

48.5

%7

angr

os

Ang

uilla

rost

rata

A

mer

ican

eel

A

ngui

lloid

ae

Cat

adro

mou

s 57

.6%

5

atyi

nn

Aty

a in

noco

us

Inno

cuou

s fre

shw

ater

sh

rimp

Aty

dae

Am

phid

rom

ous

39.4

%11

atyl

an

Aty

a la

nipe

s S

pinn

ing

fresh

wat

er

shrim

p A

tyda

e A

mph

idro

mou

s42

.4%

8

atys

ca

Aty

a sc

abra

R

ough

back

fres

hwat

er

shrim

p A

tyda

e A

mph

idro

mou

s18

.2%

15aw

ataj

A

wao

us ta

jasi

ca

Riv

er g

oby

Gob

iidae

C

atad

rom

ous

36.4

%13

elep

is

Ele

otris

pis

onis

S

piny

chee

k sl

eepe

r E

leot

ridae

C

atad

rom

ous

18.2

%15

epis

in

Epi

lobo

cera

sin

uatif

rons

P

uerto

Ric

an fr

eshw

ater

cr

ab

Pse

udot

help

husi

dae

Pota

drom

ous

51.5

%6

gobd

or

Gob

iom

orus

dor

mito

r B

igm

outh

sle

eper

E

leot

ridae

C

atad

rom

ous

42.4

%8

mac

aca

Mac

robr

achi

um a

cant

huru

s C

inna

mon

rive

r shr

imp

Pal

aem

onid

ae

Am

phid

rom

ous

9.1%

17m

acca

r*

Mac

robr

achi

um c

arci

nus*

B

igcl

aw ri

ver s

hrim

p P

alae

mon

idae

A

mph

idro

mou

s81

.8%

1m

accr

e*

Mac

robr

achi

um c

renu

latu

m*

Stri

ped

river

shr

imp

Pal

aem

onid

ae

Am

phid

rom

ous

42.4

%8

Mac

fau

Mac

robr

achi

um fa

ustin

um

Big

arm

rive

r shr

imp

Pal

aem

onid

ae

Am

phid

rom

ous

78.8

%2

mac

het*

Mac

robr

achi

um h

eter

ochi

rus*

C

asca

de ri

ver s

hrim

p P

alae

mon

idae

A

mph

idro

mou

s39

.4%

11si

cplu

S

icyd

ium

plu

mie

ri S

irajo

gob

y G

obiid

ae

Am

phid

rom

ous

63.6

%4

xipe

lo

Xip

hoca

ris e

long

ata

Yel

low

nos

e sh

rimp

Aty

dae

Am

phid

rom

ous

75.8

%3

nerv

ir N

eriti

na v

irgin

ea

Virg

in N

erite

sna

il N

eriti

dae

Am

phid

rom

ous

30.3

%14

*G

roup

ed to

geth

er a

s mac

CC

H in

som

e an

alys

es

166

variation of depth of five points along the thalweg were also measured. Channel slope

was similarly measured as the difference in elevation of the water surface over 10

uniformly spaced points spanning approximately five channel widths throughout the

reach.

Both baseflow and active-channel (bankfull) discharge at each cross-section was

estimated by a regional elevation/drainage area based equation derived from long-term

stream gage data (Pike 2006). Combining discharge estimates with channel

measurements, we estimated several hydraulic variables: stream power, unit stream

power, and the Darcy-Weisbach friction factor (see chapter 4). Grain size in the channel

was estimated using a modified Wolman Pebble Count method (Wolman, 1954).

Approximately 100 clasts were selected randomly by pacing across the width of the

stream. The median diameter of each clast was measured, and classified into the

following seven size categories: megaboulder (>2000mm), boulder (256-2000mm),

cobble (64-256mm), gravel (2-64mm), sand (.063-1mm), and fines (silt/clay, 0.001–

0.063 mm). Bedrock was also included in the count. From these grain size measurements,

we determined the median grain size (d50), coarse grain size (d84), and the percent of

bedrock exposed along the perimeter of the active channel.

Landscape-scale variables were estimated using a 10m digital elevation model

(DEM) and flow accumulation algorithms (Pike 2006). These included the distance from

the headwaters and the coast, drainage area, map slope, and the largest downstream drop

(an indicator strongly correlated with waterfall height). Furthermore, land-cover

characteristics (percent of agriculture/pasture, forest, and urbanization) were estimated

167

for both the upstream catchment and the downstream zone of influence affecting the

migratory corridor.

Pool Length and Spacing

Since pools are the primary habitat for most fish and shrimp in these streams,

quantifying the size and spacing of these pools varies across the landscape is needed to

predict their distribution. Seven stream segments, between 0.5km and 2.5km in length,

were surveyed. These segments spanned a range of stream size, from 1st order headwaters

streams to larger low-elevation main river channels. We walked the distance of each

segment, measuring the length of every pool with a tape measure, as well as measuring

the spacing between pools. The drainage area of the reach was estimated as the average

of the drainage area at the bottom and top ends of the reach.

Statistical Analyses

Principal Components Analysis

We quantified geomorphic differences between pools using principal components

analysis (PCA) (PC-ORD 4 software; MjM Software Design, Glendale Beach, OR,

USA). PCA is a technique used to reduce multidimensional data sets to a fewer number

of variables. It is particularly useful to identify key gradients in large data sets when

many of the measured variables are covariate. PCA is mathematically defined as an

orthogonal linear transformation that transforms the data to a new coordinate system such

that the greatest variance by any projection of the data comes to lie on the first coordinate

(called the first principal component), the second greatest variance on the second

coordinate, and so on. PCA involves the calculation of the eigenvalue decomposition of a

data set to weight a linear combination of variables that form the principal components.

168

This technique relies on least squares methods, and as such, requires that input variables

be relatively normal in their distribution. Prior to analysis, all non-normal input variables

were transformed using a Box-Cox transform (Box and Cox 1964) to achieve

approximate normality. PCA was used here to reduce the 57 geomorphic variables, many

of which are related and covariate, into a reduced number of factors that reflect the

primary gradients describing the geomorphic environment. To assess geomorphic

linkages to aquatic biota, we then related the presence/absence, and relative abundance of

species to these principal components.

Non-parametric Multiplicative Regression (NPMR)

We predicted the probability of species occurrence at unsampled reaches

throughout the stream network using a habitat model based on non-parametric

multiplicative regression (NPMR) (Hyperniche 1.0 software, MjM Software Design,

Glendale Beach, OR, USA). NPMR is a multivariate regression in which the response is

estimated from a multiplicative combination of all predictors, rather than the usual least-

squares multiple linear regression or generalized additive models. The multiplicative

approach captures the natural complexity of species distribution along multiple

environmental gradients. If any of the predictor variables are not conducive to the

presence of the species, then the multiplicative model predicts that species response is

also zero. It simply represents the axiom that organisms must simultaneously meet all

environmental challenges or die (McCune 2006). Furthermore, NPMR estimates response

curves based on one or more humped-shaped functions (Gaussian or sigmoid curves),

which allows the model to capture non-linear responses to environmental gradients and

threshold boundaries. NPMR is an ideal technique for ecological abundance data with

169

zero as a natural lower bound, and has been used effectively to predict abundance of fish

populations (Wedderburn et al. 2007), and the distribution of lichens and forest

populations (McCune 2006).

We developed models for each species using presence/absence data at the site-

scale, rather than pool-scale. This was done for two reasons. First, aggregating pool-scale

presence/absence data to the site scale decreases the likelihood of incorrectly classifying

a species as absent and consequently creates better models. Second, results of the model

were applied to a GIS grid where only landscape-scale factors can be calculated at non-

surveyed sites. The following 13 landscape-scale factors were used: elevation, drainage

area, active channel discharge, distance to headwaters and coast, reach slope, maximum

downstream drop, stream power, maximum downstream stream power, and the

percentage of agriculture, forest, and urbanization upstream and downstream. Each

individual model that was fitted to each species based on landscape scale factors was

applied to a 10m-resolution GIS grid. Values from the grid were extracted at regular

intervals, and plotted as a longitudinal profile.

For each species, a two-parameter model with the best goodness-of-fit was

chosen. Using more than two predictor variables did not add additional goodness-of-fit at

the expense of model complexity (based on a data:predictor ratio minimum and

improvement criterion). With presence/absence data, goodness of fit is determined

through the Bayes factor, log-likelihood ratios (Log-B), a descriptive statistic of “weight

of evidence”. Values of logB reported here are based on internally cross-validatation,

where the point being estimated is excluded from the model to reduce overfitting.

Significance of relationship is determined based on performance over the “naïve model”

170

(McCune 2006), which assumes that the probability of occurrence for all points is the

same at all sites and is the average probability that species.

Stepwise Multiple Linear Regression (SMLR)

The role of reach and pool-scale geomorphic variables on decapod abundance was

also investigated using Stepwise Multiple Linear Regression (SMLR). SMLR is a

technique for choosing the variables to include in a multiple linear regression model.

Forward stepwise regression starts with no model terms, and adds the most statistically

significant term (the one with the highest F statistic or lowest p-value), one step at a time,

until no added variables appreciably increase the residual sum of squares. Although it is

recognized that both the choice of stepwise procedure (forward, backward) and

inclusion/exclusion of starting variables can affect the outcome of the stepwise regression

procedure, we used all the measured geomorphic variables. The 57 Box-Cox transformed

geomorphic variables were used as inputs to predict the logarithm-transformed relative

abundance of each decapod species at pools where they were present. For the Quebrada

Prieta dataset, the 20 geomorphic variables measured at each pool were similarly used as

inputs to predict decapod abundances.

RESULTS

Species Distribution in Geomorphic Space at Network and Pool-Scales

For all 113 pools, the 57 geomorphic variables were reduced to 3 significant

principal component axes that explain a total of 59% of the variance. The first axis

explains 33% of the variance and the second axis explains the remaining 26% of the

variance. A third significant axis explains 9% of the variance, but was not considered in

171

this analysis. Other principal components individually explained less than 6% of the

residual variance, and were also not considered further.

Based on strength of the eigenvalues for each variable, we determined which

geomorphic variables most importantly contribute to each principal component, and

interpreted each axis accordingly (Table 5.2). Principal component 1 is interpreted as a

“longitudinal axis” as it is positively correlated with drainage area (eigenvalue = 0.22),

active channel discharge (0.22), elevation (0.17), distance from the headwaters (0.22) and

the length (0.21), area (0.22), and volume (0.22) of the pool. Principal component 2 is

interpreted as a “hydraulic axis”. It is most strongly correlated with median grain size

(eigenvalue = 0.22), shear stress (0.18), and stream power (0.23), and negatively

correlated with the proportion of fine sediment (-0.16), and proportion of agricultural

land cover in the upstream catchment (-0.19).

Pools plotted along these “longitudinal” and “hydraulic” axes fall into three

general clusters (Figure 5.2). Pools plotting in the upper left hand side of the ordination

have low values of PC1 and higher values of PC2. They represent steep, gradient

headwater pools that have high stream power and large boulders. In contrast, pools

plotting on the right hand side of the ordination (high values of PC1, decreasing values of

PC2) represent mid-to-low elevation pools along the main stem. The cluster of pools

plotting in the middle of the ordination, having intermediate values of PC1 and low

values of PC2, represent lowland tributaries that are of intermediate drainage area, but are

generally low gradient, have low stream power, and contain a higher proportion of fine

sediment.

172

Table 5.2 Eigenvalues for 58 geomorphic variables, indicating the relative influence of

each variable on principal components 1 and 2. Note that axes are better correlated with

landscape-scale variables rather than pool scale factors. (Modified from Hein et al., in

prep).

173

Table 5.2 cont.

Description Unit PC1 PC2 Elevation m -0.17 0.15 Average Basin Elevation m -0.02 0.22 Drainage Area km2 0.22 0.02 Distance from Ocean km -0.11 0.13 Distance from Headwaters km 0.22 0.01 Slope m/m -0.18 0.14 Maximum Downstream Drop m -0.16 0.14 % Agriculture Cover Downstream % 0.09 -0.08 % Agriculture Cover Upstream % 0.06 -0.19 % Forest Cover Downstream % -0.12 0.12 % Forest Cover Upstream % -0.08 0.19 % Urban Cover Downstream % 0.05 -0.07 % Urban Cover Upstream % 0.12 -0.14 Pool Maximum Depth m 0.14 0.11 Coefficient of Variation for Depth unitless 0.01 0.04 Pool Length m 0.21 -0.01 Pool Surface Area m2 0.22 0.03 Pool Volume m3 0.21 0.05 Baseflow Discharge m3/s 0.21 0.09 Baseflow Width m 0.20 0.07 Baseflow Wetter Perimeter m 0.20 0.08 Baseflow Hydraulic Radius m 0.12 0.10 Baseflow Maximum Depth along Cross-section m 0.11 0.12 Baseflow Cross-sectional Area m2 0.18 0.10 Baseflow Velocity m/s 0.14 0.04 Baseflow Width/Depth Ratio unitless 0.12 -0.01 Active Channel Discharge m3/s 0.22 0.07 Active Channel Width m 0.20 0.08 Active Channel Wetter Perimeter m 0.20 0.10 Active Channel Hydraulic Radius m 0.14 0.11 Active Channel Maximum Depth along Cross-section m 0.11 0.13 Active Channel Cross-sectional Area m2 0.19 0.11 Active Channel Velocity m/s 0.18 0.02 Active Channel Width/Depth Ratio unitless 0.14 0.03 d16, Fine grain size mm 0.00 0.22 d50, Median grain size mm -0.01 0.20 d84, Coarse grain size mm -0.05 0.17 Geometric Mean of Grain Size mm -0.02 0.22 % Bedrock % -0.01 0.11 % Megaboulder % -0.03 0.15 % Boulder % -0.07 0.08 % Cobble % 0.04 0.00 % Gravel % 0.04 -0.12 % Sand % 0.04 -0.13 % Fines % 0.01 -0.17 Simpson Diversity Index of Grain Size Categories -0.06 -0.02 Grain Size Sorting Coefficient -0.02 -0.16 Grain Size Skew Coefficient -0.04 0.05 Baseflow Shear Stress Pa -0.14 0.18 Active Channel Shear Stress Pa -0.15 0.18 Critical Shear Stress Pa -0.02 0.20 Baseflow Stream Power W/m 0.06 0.24 Active Channel Stream Power W/m 0.07 0.23 Baseflow Unit Stream Power W/m2 -0.02 0.23 Active Channel Unit Stream Power W/m2 -0.01 0.22 Baseflow Darcy-Weisbach Friction Factor -0.16 0.03 Active Channel Darcy-Weisbach Friction Factor -0.19 0.06

174

Figure 5.2 Plot of pools in geomorphic space, with species presence indicated by circles

(large circles = presence, dots = absence). Principal component 1 is interpreted as a

measure of longitudinal position (distance downstream, drainage area, and pool size),

whereas principal component 2 is strongly influenced by stream power and grain size.

Pools plotted on these principal component axes cluster into 3 groups: steep headwater

channels, lowland channels along the main stem, and lowland tributaries. Species (see

Table 5.1 for species codes) show distinct habitat preferences along these two axes:

predatory fish are present in lowland channels, atyid shrimp and crabs thrive in headwater

streams, and palaemonid shrimp occupy most pools. (Modified from Hein et al., in prep)

175

Figure 5.2 cont.

176

Plots of the presence and absence of fish and decapod species show that they have

distinct preferences along these “longitudinal” and “hydraulic” axes (Figure 5.2). In

general, major predatory fish (mullets and eels), as well as gobies and sleepers, plot on

the right and lower center of the ordination, indicating their preference for both lowland

main channels and lowland tributaries. Although not clearly evident in this figure, the

pools that these fish species occupy are also below barrier waterfalls that exceed 4m in

height. In contrast to most fish species, the sirajo goby, S. plumieri, inhabits pools that

span a broad longitudinal range, but are absent in lowland tributaries. Among atyid

shrimp, Atya lanipes is strongly confined to and present in most headwater pools,

whereas the other two Atya species occupy some lowland pools in addition to headwater

pools. Xiphocaris is present in virtually all headwater pools, but also occupies some

lowland pools as well. Palaemonid shrimp are present at most pools, reflecting their

ubiquity in these streams, although M. faustinum displays a slight affinity for lower

elevation pools. Lastly, crabs are present in some lowland streams, but are more common

in headwater channels. These patterns indicate that predatory fish and most decapods

(especially atyid shrimp) generally occupy contrasting pools across the landscape; fish in

lowland channels and tributaries, and decapods in steep headwater pools.

At the reach-scale, the relative abundance and proportional abundance of decapod

species were also plotted along the PCA axes (Figure 5.3). Among the pools were species

were present, Principal component 1 (“longitudinal” axis) was well correlated with the

relative abundance of crabs (r2 = 0.51), Macrobrachium spp (r2 = 0.27), and Xiphocaris

(r2 = 0.13). Similarly, principal component 1 correlated with the proportional abundance

of M. faustinum (r2 = 0.32). Among the pools were each decapod species was present, no

177

Figure 5.3 Plot of pools in geomorphic space, with decapod relative abundance (catch

per unit effort) and proportional abundance (% of species in pool) indicated by circles

(size of circle is proportional to abundance, dots = absence). Correlations between

abundance, at the pools where each decapod species was present, and the principal

component axes are shown.

178

Figure 5.3 cont.

179

other significant correlations between principal components and decapod abundances

were observed. Among the pools that were above waterfalls, the only significant

relationship between these geomorphic gradients and decapod abundance was a positive

correlation between the Principal Component 1 and the relative abundance (r2 = 0.22) and

proportional abundance of M. faustinum (r2 = 0.30). At the scale of the headwater stream

network, no further significant relationships between principal components and decapod

abundances were observed.

At the pool-scale, PCA of the pools within the reach of the Quebrada Prieta

reduced 20 geomorphic variables to 2 significant principal components (Table 5.3). The

first principal component explains 26% of the variance, whereas the second principal

component explains 15%, for a total of 41%. The first principal component is interpreted

to be a measure of pool size, as it is most strongly influenced by pool volume (eigenvalue

= 0.39), and depth (0.38). The second principal component is interpreted to be a measure

of substrate suitability, as it the most influential factors are the proportion of gravel (0.42)

and leaves (0.42) composing the substrate.

Correlations between the abundance of Atya, Xiphocaris, and Macrobrachium and

principal components indicate that there are weak patterns along these two gradients

(Figure 5.4). Atya relative abundance is weakly correlated with principal component 2 (r2

= 0.22), indicating a slight preference for pools with gravel and leaves. Xiphocaris was

significantly correlated with both principal components 1 and 2 (r2PC1 = 0.18, r2

PC2 =

0.20), reflecting the influence of both pool size and leaf availability for these shredders.

Similarly, both principal components were modestly correlated with the proportional

abundance of Macrobrachium (r2PC1 = 0.18, r2

PC2 = 0.21), but not correlated the

180

Table 5.3 Eigenvalues for 20 geomorphic variables measured for pools in the Quebrada

Prieta, indicating the relative influence of each variable on principal components 1 and 2.

Based on the strength of the eigevectors, principal component 1 is interpreted as a

measure of pool size, and principal component 2 is interpreted as an indicator of substrate

suitability (% gravel and % leaves).

Quebrada Prieta Data Description Unit PC1 PC2 Distance from Headwaters m 0.17 0.15 Compass Direction degrees -0.02 -0.22 Active Channel Width m 0.22 -0.02 Maximum Width m -0.11 -0.13 Maximum Depth m 0.22 -0.01 Pool Area m2 -0.18 -0.14 Pool Volume m3 -0.16 -0.14 Average Depth m 0.09 0.08 Standard Deviation of Depth m 0.06 0.19 Coefficient of Variance of Depth n/a -0.12 -0.12 % Silt % -0.08 -0.19 % Sand % 0.05 0.07 % Gravel % 0.12 0.14 % Cobble % 0.14 -0.11 % Boulder % 0.01 -0.04 % Leaf % 0.21 0.01 % Organic Matter % 0.22 -0.03 % Open Substrate % 0.21 -0.05 # Pool Entrances # 0.21 -0.09 # Pool Exits # 0.20 -0.07

181

Figure 5.4 Plot of Quebrada Prieta pools in geomorphic space, with decapod relative

abundance (catch per unit effort) and proportional abundance (% of species in pool)

indicated by circles (size of circle is proportional to abundance, dots = absence).

Correlations between abundance, at the pools where each decapod species was present,

and the principal component axes are shown.

182

proportional abundance of Atya (r2PC1 = 0.05, r2

PC2 = 0.07) and Xiphocaris (r2PC1 = 0.05,

r2PC2 = 0.04).

Longitudinal Trends of Species

The NPMR habitat models identified two landscape scale variables for each

species that best combined to make a multivariate response curve (Table 5.4). An

example is shown in Figure 5.5a, where the probability of eels, A. rostrata, was predicted

as a function of both elevation, and the maximum downstream drop in elevation in a

reach. Applying this multivariate response curve to unsampled reaches throughout the

streams network shows a strong spatial trend: eels are common in low to mid-elevation

streams but abrupt drop out near the location of the first waterfall (Figure 5.5b). Those

species that had models influenced by the maximum downstream drop (A. monticola, A

rostrata, A. lanipes) also had the best fits (e.g. highest logB), indicating that their spatial

distribution is highly pronounced and influenced by waterfalls. Other species with

relatively high goodness of fit statistics (A. tajasica, G. dormitor, and X. elongata) were

predicted by either slope or stream power. These are both variables that distinguish

headwater streams from lowland reaches. In contrast, species models that were best

predicted by landcover variables generally had lower goodness of fit statistics suggesting

that land use is not a major factor in these relatively pristine watersheds.

The NPMR habitat models demonstrate that most species have a pronounced

spatial distribution along the longitudinal profiles, either occurring in lowland reaches or

in the headwaters (Figure 5.6). For many species, the location in the stream network

where they drop out is typically abrupt. The estimated longitudinal trends show that

predatory fish drop out at the boundary of the first waterfall on the main stem. Few are

183

Table 5.4 Landscape scale variables used to predict species presence/absence using a

NPMR 2-parameter model. Goodness of fit is determined through log-likelihood ratios

(logB), a descriptive statistic of “weight of evidence”, where higher values indicate a

stronger model. (Modified from regression tree models presented in Hein et al., in prep).

NPMR 2-PARAMETER MODELS Response Eval Ave Predictor Predictor Variable logB Size Variable 1 Tolerance Variable 2 ToleranceAgomon 7.5 3.7 discharge 1.09E+01 max_drop 9.10E-01angros 7.9 3.6 elevation 1.65E+02 max_drop 4.55E-01atyinn 1.1 9.4 discharge 1.82E+01 agr_downstream 4.00E+01atylan 8.3 4.2 reach_slope 3.85E-02 max_drop 9.10E-01atysca 1.3 2.7 max_drop 9.10E-01 agr_downstream 1.50E+01awataj 5.3 4.5 dist_to_ocean 2.14E+03 reach_slope 3.85E-02elepis 2.5 13.1 agr._downstream 4.50E+01 forest_upstream 2.02E+01episin 2.7 8.2 discharge 1.82E+01 forest_upstream 8.07E+00gobdor 5.7 5.3 reach_slope 5.14E-02 forest_upstream 4.03E+00maccar 2.9 3.4 reach_slope 3.85E-02 agr_downstream 1.00E+01maccre 2.5 4.9 dist_to_ocean 4.29E+03 stream_power 5.88E+03macfau 3.0 2.8 max_drop 9.10E-01 stream_power 2.94E+03machet 3.3 5.8 dist_to_headwater 4.00E+03 discharge 7.27E+00sicplu 2.5 4.3 reach_slope 6.42E-02 stream_power 2.94E+03xipelo 5.3 4.2 discharge 1.45E+01 stream_power 5.88E+03nervir 4.2 3.0 reach_slope 2.57E-02 urb_downstream 5.00E+00

184

Figure 5.5 a) Non-parametric multiplicative regression response curve for eels (Anguilla

rostrata). The curve shows a primary response along the maximum downstream drop

axis; eels are most likely present where the downstream drop < 4m, and are absent above

this threshold. Additionally, there is a slight secondary species response along the

elevation axis. Similar curves were constructed for each species, described in Table 5.4.

b) Eel distribution in the Rio Mameyes as estimated by the NPMR species response

curve. The response, based on elevation and maximum downstream drop, was applied

throughout the watershed. Streamlines are scaled according to the estimated probability

of occurrence. Observed presence (gray circles) and absence (white circles) at study sites

are also shown. The map shows eels are present along the main stem and at low

elevation, but drop out abruptly near the first waterfall and are consequently absent in the

upper-elevation headwaters. (Modified from regression tree models presented in Hein et

al., in prep)

185

Figu

re 5

.5 c

ont.

186

Figure 5.6 Longitudinal views of estimated probability of occurrence for each species.

The longitudinal profile of the river (above) shows the location of sample site (circles)

along the main stem (solid line) and tributaries (dashed line). For each species, the

probability of occurrence (between 0 and 1) along the main stem is indicated by the solid

line, whereas the probability of occurrence on the tributaries is indicated by the dashes

lines. Observed presences (black circles) and absences (white circles) at sample sites are

also shown. As an example, moving upstream along the longitudinal profile, eels are

expected to be present along the main stem until the first waterfall at approximately

11km, where the probability abruptly decreases to 0. Observed presences confirm this

trend. Additionally, eels are expected to drop out along mid to upper elevation tributaries,

where the dotted lines connected to the main stem rapidly decrease. (Modified from Hein

et al., in prep)

187

Figure 5.6 cont.

188

present in the adjoining tributaries. The herbivorous goby, S. plumieri, is common on the

main stem at low elevation, and at upper elevations, and typically absent on low to mid-

elevation tributaries. The atyid shrimp, A. innocuous, A. lanipes, and X. elongata are

absent along the main stem below waterfalls, but are present at upper elevations and

along tributaries. In general, predatory fish and atyid shrimp have contrasting

longitudinal trends and rarely co-occur. High-elevation sites above large waterfalls are

dominated by decapods whereas lowland streams are dominated by fish.

Predatory shrimp (Macrobrachium spp.) show a different trend. M. carcinus is

present at most sites except the uppermost pools, whereas M. crenulatum and M.

faustinum are most commonly present on mid-elevation tributaries. Crabs and snails

display contrasting longitudinal patterns. Snails, N. virginea are common along the main

stem at low elevations, whereas crabs, E. sinuafrons, are present along the main stem at

mid-to-high elevations and along tributaries.

Internal cross-validation (estimating a value at a site with that point removed)

shows that all multiplicative models showed significant improvement over the naïve

model (Table 5.5). Models predicting the presence of species that displayed pronounced

longitudinal variation (A. monticola, A. rostrata, A. lanipes, X. elongata) had the best

percentage of improvement over the naïve model, and the most statistically significant.

For ubiquitous species (S. plumieri, M. carcinus) that do not display strong longitudinal

distributions, and for rare species (A. scabra, E. pisonis), the model predictions were not

as strong.

189

Table 5.5 Internal validation statistics for NPMR models, where the point being

estimated is excluded from the model. All models are significant (p > 0.05) over the

naïve model where the probability of a species being “present” is constant across sites.

(Modified from regression tree models presented in Hein et al. 2008)

190

Tab

le 5

.5 c

ont.

N

MP

R 2

-PA

RA

ME

TER

MO

DE

L IN

TER

NA

L V

ALI

DA

TIO

N

(ALL

SIT

ES

)

Ove

rall

Ove

rall

Naï

ve

Est

. E

st.

Impr

oved

N

ot Im

prov

ed

Impr

ovem

ent

N

ame

xR2

Pre

sent

A

bsen

t p

p>na

ïve

p<na

ïve

P

A

Om

iss

Com

mis

%

O

ddsR

at

logB

av

eB

Chi

Sq

p

agom

on

0.71

17

15

0.

53

16

16

15

14

2 1

90.6

9.

7 7.

5 1.

72

34.5

0.

0000

angr

os

0.74

19

13

0.

59

22

10

19

10

0 3

90.6

9.

7 7.

9 1.

76

36.3

0.

0000

atyl

an

0.97

12

18

0.

40

12

18

12

18

0 0

100.

0 n/

a 8.

3 1.

89

38.1

0.

0000

atyi

nn

0.18

13

20

0.

39

19

14

10

11

3 9

63.6

1.

8 1.

1 1.

08

5.3

0.02

17

at

ysca

0.

15

6 24

0.

20

17

13

5 12

1

12

56.7

1.

3 1.

3 1.

11

6.0

0.01

43

aw

ataj

0.

61

12

20

0.38

15

17

11

16

1

4 84

.4

5.4

5.3

1.47

24

.5

0.00

00

el

epis

0.

35

6 27

0.

18

9 24

5

23

1 4

84.8

5.

6 2.

6 1.

20

11.7

0.

0006

epis

in

0.36

18

13

0.

58

21

10

17

9 1

4 83

.9

5.2

2.7

1.22

12

.6

0.00

04

go

bdor

0.

50

13

18

0.42

15

16

11

14

2

4 80

.6

4.2

5.7

1.53

26

.2

0.00

00

m

acca

r 0.

21

24

6 0.

80

22

8 20

4

4 2

80.0

4.

0 2.

9 1.

25

13.3

0.

0003

mac

cre

0.26

14

19

0.

42

16

17

9 12

5

7 63

.6

1.8

2.5

1.19

11

.5

0.00

07

m

acfa

u 0.

20

24

7 0.

77

19

12

18

6 6

1 77

.4

3.4

3.1

1.25

14

.0

0.00

02

m

ache

t 0.

33

13

20

0.39

22

11

13

11

0

9 72

.7

2.7

3.3

1.26

15

.0

0.00

01

si

cplu

0.

28

22

11

0.67

16

17

14

9

8 2

69.7

2.

3 2.

5 1.

19

11.6

0.

0007

xipe

lo

0.69

25

8

0.76

22

11

22

8

3 0

90.9

10

.0

5.4

1.45

24

.6

0.00

00

ne

rvir

0.54

9

21

0.30

13

17

9

17

0 4

86.7

6.

5 4.

2 1.

38

19.5

0.

0000

NO

TES

: S

ampl

e un

its w

ith e

mpt

y ne

ighb

orho

ods

wer

e ex

clud

ed fr

om th

ese

calc

ulat

ions

. xR

2 =

cros

s-va

lidat

ed p

seud

o-R

-squ

ared

. -9

9.99

99=

mis

sing

if <

2 n

onem

pty

neig

hbor

hood

s.

Ove

rall

pres

ent =

num

ber o

f "pr

esen

t" ca

ses

(non

-zer

o or

exc

eedi

ng b

inar

y cu

toff

valu

e).

Ove

rall

abse

nt =

num

ber o

f "ab

sent

" cas

es (z

ero

or le

ss th

an o

r equ

al to

bin

ary

cuto

ff va

lue.

) N

aive

p =

nai

ve e

stim

ate

of th

e pr

obab

ility

of o

ccur

renc

e =

prop

ortio

n pr

esen

t. E

st p

>nai

v =

num

ber o

f cas

es p

redi

cted

mor

e lik

ely

than

ave

rage

(nai

ve p

) to

be "p

rese

nt"

Est

p<n

aiv

= nu

mbe

r of c

ases

pre

dict

ed le

ss li

kely

than

ave

rage

(nai

ve p

) to

be "p

rese

nt"

Impr

oved

p =

num

ber o

f "pr

esen

t" ca

ses

with

est

imat

ed p

roba

bilit

y >

naiv

e p.

Im

prov

ed a

= n

umbe

r of "

abse

nt" c

ases

with

est

imat

ed p

roba

bilit

y <

naiv

e p.

N

ot im

prov

ed o

mis

s =

Err

or o

f om

issi

on: s

peci

es p

rese

nt b

ut e

st.p

<=

naiv

e p

Not

impr

oved

com

mis

= E

rror

of c

omm

issi

on: s

peci

es a

bsen

t but

est

.p >

= na

ive

p Im

prov

emen

t % =

Per

cent

of c

ases

with

est

imat

es im

prov

ed o

ver n

aive

mod

el.Im

prov

emen

t odd

s ra

tio =

Per

cent

of c

ases

with

est

imat

es im

prov

ed o

ver n

aive

m

odel

.Odd

s of

impr

ovem

ent =

pro

porti

on o

f im

prov

ed/(1

- pr

opor

tion

impr

oved

) 9

999.

9=in

f. lo

gB =

log1

0(B

ayes

fact

or)

0

to 0

.5 --

not

wor

th m

ore

than

a b

are

men

tion

0

.5 to

1 --

sub

stan

tial

1

to 2

--

stro

ng

> 2

--

dec

isiv

e av

eB =

ave

rage

con

tribu

tion

of e

ach

case

to th

e B

ayes

fact

or =

10^

(logB

/N)

Chi

Sq

= 2*

ln(B

ayes

fact

or) =

Dev

ianc

e co

mpa

ring

mod

el to

nai

ve m

odel

p

= pr

obab

ility

of t

ype

I err

or fr

om c

hi-s

quar

e di

strib

utio

n, d

.f.=1

191

Influence of Reach and Pool-Scale Geomorphology

SMLR identified a combination of reach and pool-scale variables that best predict

decapod species abundances at the pools where they are present (Table 5.6). Adjusted

correlation coefficients (r2adj) for 3-term models ranged from 0.30 to 0.66. Reach-scale

variables such as pool length, distance to the ocean, and the maximum downstream drop

were the first significant terms for all species except Atya lanipes. These variables are

negatively correlated with abundance, reflecting the trend that abundances for these

species are generally greater in the headwater reaches than in the relatively few larger

streams they inhabit. Other reach-scale factors such as the proportion of forest upstream

and downstream, and stream power were similarly important. Pool-scale variables,

including the grain size sorting coefficient, and the proportion of cobbles and boulders

were significant factors for Atya and Macrobrachium species—shrimp that are known to

dwell in a complex matrix of well-sorted coarse substrate on the channel bottom (Covich

and McDowell 1996). For Xiphocaris—shrimp that spend much of their time

swimming—abundance was negatively correlated with the stream power of the pool.

SMLR similarly identified strictly the pool-scale variables that best predict Atya,

Xiphocaris, and Macrobrachium abundance using the Quebrada Prieta data (Table 5.7).

SMLR models had adjusted correlation coefficients between 0.51-0.53. Atya and

Xiphocaris abundance were most strongly correlated with pool area. The distance from

the headwaters negatively correlated with Atya abundance, but was positively correlated

with Macrobrachium abundance. Furthermore, the number of pool entrances was

negatively correlated with both Atya and Xiphocaris abundances.

192

Table 5.6 Stepwise Multiple Linear Regression model output and goodness of fit

showing the relative influence of geomorphic variables on predicting decapod abundance.

Note the combination of both reach-scale variables (stream power, forest cover) and

pool-scale variables (substrate).

STEPWISE MULTIPLE LINEAR REGRESSION (REACH-SCALE) Response Species Predictor Variable Sign r2

adj n atylan Grain Sorting Coefficient + 0.17 49 % Forest Downstream + 0.25 Width (baseflow) - 0.30 episin Pool Length - 0.50 30 Wetted Perimeter (active channel) - 0.57 % Forest Upstream + 0.66 macCCH Distance to ocean - 0.27 60 Maximum Downstream Drop + 0.37 % Cobble + 0.41 macfau Distance to ocean + 0.34 52 % Forest Downstream - 0.49 % Boulder + 0.55 xipelo Maximum Downstream Drop + 0.51 68 Unit Stream Power - 0.55 Pool Volume - 0.58

193

Table 5.7 Stepwise Multiple Linear Regression model output and goodness of fit

showing the relative influence of geomorphic variables on predicting decapod abundance

in the Quebrada Prieta. Note the influence of pool size and substrate characteristics.

STEPWISE MULTIPLE LINEAR REGRESSION (POOL-SCALE) Response Species Predictor Variable Sign r2

adj n atylan Pool Area + 0.14 49 Distance from Headwaters - 0.49 Pool Entances - 0.53 mac Distance from Headwaters + 0.30 22 Max Width + 0.40 % Organic Matter + 0.53 xipelo Pool Area + 0.39 49 % Gravel + 0.47 Pool Entrances - 0.51

194

Pool Size and Spacing

The length of 101 pools at seven stream segments spanning a range of stream

sizes were measured and plotted against drainage area (Figure 5.7a). Pool length

increases as a power function of drainage area (r2 = 0.71, Figure 5.7b), from an average

of 2.5m in length at 0.15km2 drainage area, to 50m at 25km2. The spacing between pools

similarly increase as a power function of drainage area (Figure 5.7c). However, this

relationship is less robust (r2 = 0.21) due a large degree of variance within each segment.

Moreover, the rate of increase between pool length and drainage area (exponent = 0.54) is

greater the rate of increase between pool spacing and drainage area (0.27). Consequently,

the ratio of average pool spacing to average pool length decreases with drainage area

(Figure 5.7d). In headwater reaches, pools are spaced approximately 4 times of their

length apart. In the lowermost reaches, pools are spaced closer together, approximately

one pool length apart. Furthermore, the percentage of total segment length that is

classified as a pool increases with drainage area (Figure 5.7e). Pools compose between

15-20% of the length of headwater reaches, whereas they cover up to 35-50% in lowland

reaches along the main stem having greater drainage area.

DISCUSSION

Landscape Scale Patterns

Principal components analysis of the geomorphology of all pools illustrates the

dominance of one primary geomorphic gradient in the Luquillo streams: longitudinal

position within the stream network. Along this gradient, several key habitat features

similarly vary, including pool length, channel size, and the location relative to waterfalls.

195

Figure 5.7 a) Map showing the location of 7 stream segments where pool length and

spacing were measured b) pool length increases as a power function with discharge c)

pool spacing similarly increases as a power function with discharge, but at half the rate d)

consequently, the ratio of pool spacing to pool length decreases with drainage area, and e)

the % of pool in reach increases with drainage area.

196

Figure 5.7 cont.

197

The biological response to this geomorphically imposed gradient is evident in the

longitudinal patterns of species presence/absence. Predatory fish show a preference for

lowland main channels and tributaries, whereas atyid shrimp and crabs showing an

affinity for headwater pools.

The linkages between geomorphology and species preference that are evident

through principal components analysis are similar to those noted by Hein et al. (in prep)

using a different ordination technique. By quantifying the differences among pools on the

basis of their community composition (rather than by geomorphology), using a non-

metric multidimensional scaling ordination technique, this metric of community

composition difference similarly correlated strongly with longitudinal position. The

results from the ordination in Hein et al. (in prep) clearly demonstrate that there are three

general communities: one consisting of mullets and eels, another consisting of gobies,

and the last consisting of atyid shrimp. The location of waterfalls were found to be more

important in determining the distribution of these communities than either local pool-

scale geomorphic or local hydraulic variables. Furthermore, regression tree analysis

presented in Hein et al. (in prep) demonstrates that the presence and absence of many

species can be predicted by a simple split of whether or not a site is above a waterfall.

Access to upper reaches is limited for some migratory species as waterfalls act as a

barrier for most fish. All predatory fishes are limited to areas of the stream network

below 4m vertical drops, whereas shrimps, river crabs, and the sirajo goby are common at

sites above waterfalls.

The NPMR analysis presented here confirms these patterns. Since the distribution

of species is best determined by landscape scale variables, NPMR is an effective way to

198

map the species throughout these stream networks. The longitudinal probability curves

for each species (Figure 5.5.5) further reinforce the pattern of fish-dominated

assemblages in the lowlands shifting to decapod dominance in the headwaters. They also

show that transitions between these two communities are typically abrupt. Furthermore,

the species maps illustrate that fish species richness increases with catchment areas and

stream size, whereas decapod species richness apparently decreases. In actuality, decapod

species richness is relatively constant throughout the network; all species of shrimp must

be present below waterfalls to migrate upstream, but typically only in their juvenile stage.

The absence of adult decapods in lower reaches has been explained as a response to fish

predation (Covich and McDowell 1996, Greathouse and Pringle 2006). The headwaters

provide a critical refuge from fish predators, and contain an abundant supply of leaf-litter

and particulate organic matter to sustain large population shrimps throughout adulthood.

Thus, no new species are added to the headwaters (they only shift from juveniles to

adults), although species do drop out in the upstream direction.

Such biological zonation is common in rivers with abrupt ecological transition or

barriers to fish movement (Rahel and Hubert 1991). The concept of waterfalls as fish

barriers, and consequently as boundaries that fragment aquatic communities, has been

noted in the literature for decades (Stuart 1962, add others). Waterfall-induced (8-12m

high) longitudinal zonation was noted in an inland tropical rainforest river in southern

Mexico, with continual addition of species downstream and little species deletion

(Rodiles-Hernandez et al. 1999). Waterfalls greater than 3m high in rivers on the South

Island of New Zealand effectively divided the fish community between native galaxiids

and non-native trout (Townsend and Crowl 1991). On the steep volcanic Comoros islands

199

off the southeastern coast of Africa, waterfalls of 15m height divided the fish population

such that only adult river gobies inhabit the fast-flowing waters above waterfalls, whereas

catadromous eels are restricted below the waterfalls (Balon and Bruton 1994). Gilliam et

al. (1993) found a similar pattern in steep drainages on the Caribbean island of Trinidad

where species drop out proceeding upstream, and the community becomes truncated at

barrier waterfalls. On the other hand, in some similar tropical streams that lack waterfalls,

such as those along Caribbean coast of Central America, longitudinal position is also an

important factor in structuring fish-assemblages, but the lack of waterfalls prevent abrupt

transitions in aquatic communities (Winemiller and Leslie 1992, Esselman et al. 2006).

Although species distributions are clearly related to the longitudinal changes in

the stream network, the abrupt shifts in communities owning to waterfalls stand in

contrast to the systematic continuum predicted by the RCC. However, some predictions

of the RCC on the distribution of food resources and the consequent biotic response have

been found to hold in the Luquillo streams to a mixed degree (Ortiz-Zayas et al. 2005,

Greathouse and Pringle 2006). These include shifts in stream metabolism and functional

feeding groups, which are tightly linked to instream productivity, light availability, and

turbulence, and consequently have implications for ecosystem functioning. Stream

metabolism, as reflected by the ratio of primary productivity to community respiration

(P/R), is thought to shift from heterotrophic (P/R < 1) in the headwaters to autotrophic

(P/R >1) further downstream, following the trend in riparian shading, algal production,

detrital inputs, and upstream organic matter transport. Ortiz-Zayas et al. (2005) found that

Luquillo streams displayed a contrasting pattern: all reaches were strongly heterotrophic

until the coastal plain reaches just above the estuary. This was attributed to high rates of

200

respiration, suppressed periphyton production due to low light, and large inputs of detrital

carbon from the surrounding mature forest. Intensive herbivory by decapods prevents

biomass accumulation expected at intermediate stream orders (March et al. 2002).

The RCC further predicts that in response to the available food resources,

functional feeding groups will shift predictably downstream from a dominance of

shredders, scrapers, and filterers in the headwaters, to grazers, collectors, and predators

further downstream. Greathouse and Pringle (2006) found that predictions held for

scrapers, shredders, and predators in the Luquillo streams, while collector-filterers

showed a trend opposite to RCC predictions. This collector-filterer trend may be a result

of the prevalence of snails at lower elevations, or may be explained by fish predation

affecting distributions of filter-feeding shrimp. However, the general theme is consistent;

longitudinal distributions of functional feeding generally groups follow longitudinal

patterns in basal resources, but are interrupted by abrupt barriers.

Reach and Pool-Scale Patterns

Some have viewed such abrupt interruptions simply as adjustments to the original

RCC (Bruns et al. 1984, Minshall et al. 1985), whereas others have argued that they serve

as the basis for a new view of a river as a “discontinuum” (Perry and Schaeffer 1987,

Townsend 1989, Rice et al. 2001, Poole 2002). In essence, river discontinuum

perspectives highlight the nonuniform or patchy distribution of habitats and therefore

emphasize habitat heterogeneity, expressed at the scale of meters to kilometers.

Despite the dominance of longitudinal position in structuring aquatic-

communities, local-scale factors do influence species distribution and abundances at the

reach, pool, and microhabitat scales. PCA identified an additional geomorphic gradient of

201

grain size and stream power among the pools that is in part independent of longitudinal

position. Based on presence/absence, some fish and decapod species evidently prefer

pools that have high stream power and large boulders or pools having low stream power

and a higher proportion of fine sediment. However, this geomorphic gradient was not

significant in determining the abundance of decapod species. Thus, at the scale of the

headwater stream network above waterfalls, there are no systematic geomorphic

gradients that determine decapod abundance (see Fig. 5.3). In fact, decapod abundance

varies strongly between adjacent pools, and a particular species may be abundant in one

but absent in the other. This patchiness in decapod abundance may be the result of natural

variation in micro-habitat, or indicative that other factors beyond the hydrologic and

geomorphic variables measured in this study determine their abundance.

However, SLMR models identified reach and pool-scale variables that are

important in determining abundances of decapods in the pools where they are present.

The variables selected were consistent with the known microhabitat preferences of

decapods. For example, filter feeding Atya prefer a cobble substrate to attach to as they

feed. Palaemonid shrimp prefer a complex substrate of boulders with crevices to dwell,

and their abundances were consequently strongly determined by grain size factors. In

contrast, Xiphocaris often swim in the water column, and abundances were found to be

negatively correlated with high stream power that may create a turbulent swimming

environment. Further, most of these decapod species were more abundant in reaches that

had a high proportion of forested land cover in their downstream migratory corridor.

At the pool-scale within the Quebrada Prieta, the geomorphic environment

consists of gradients of pool size and substrate suitability. Although decapod abundances

202

vary modestly along these gradients, pool characteristics are often patchy throughout the

reach. The large degree of variability in decapod abundance between pools in a reach

reflects patchiness of the geomorphic environment. Furthermore, within the Quebrada

Prieta, it has been shown that Atya and Xiphocaris abundance relationships can be

influenced by biotic interactions such as avoiding predatory Macrobrachium, and locally

adjusting their abundance in pools to an optimal density (Covich et al. 1996, Papella,

thesis). Furthermore, microhabitat characteristics within a pool may be responsible for

pool-scale abundances. For example, within a pool, shrimp are known to have distinct

depth and velocity preferences (Scatena and Johnson 2001). With the addition of large

boulders that alter flow and create a complex microhabitat, shrimp abundances are

expected to be patchy even within pools.

Such heterogeneity also arises because of the perception of scale, in which fluvial

landforms are hierarchically organized from valley segments to stream bed particles

(Frissell et al. 1986). The importance of landscape scale factors over reach and pool-scale

variables in determining species distribution in the Luquillo streams seemingly stands in

contrast to predictions of patch dynamics. Yet the idea of patchy and multiscale habitat

formation and its related heterogeneity is often related to longitudinal position. The

Process Domains Concept (Montgomery 1999) contends that fundamental differences in

multiscale landscape processes dictate differences in the community structure of aquatic

fauna. In the Luquillo streams, the patches that act as local habitat (e.g. geomorphic

features such as deep pools, boulder crevices, fine substrate) in the as well as hydraulic

variables (velocity, turbulence, etc) do vary somewhat predictably along the longitudinal

gradient (see Chapter 4). For example, large boulders that structure complex

203

microhabitats in headwater pools are deposited by landslides that are most common along

steep hillslopes. Furthermore, grain size patterns are related to slope and drainage area.

Other habitat features such as the size and frequency of pools, and average flow velocity

also increase downstream with drainage area and mean annual discharge. Thus, as

predicted by the PDC, the geomorphic processes structuring patchy habitat vary also

along the longitudinal gradient and mirror changes in aquatic communities.

CONCLUSION

In conclusion, the longitudinal patterns of aquatic assemblages observed in the

Luquillo streams are best explained by hypotheses that incorporate the natural

discontinuous patterns present in stream networks and consider that different factors

operate at different scales. Hierarchy theory asserts that different geomorphic processes

acting at the landscape, reach, and pool scales give rise to both local patchiness and

basin-scale patterns. No hypothesis specifically addresses discontinuities in community

structure created by waterfalls, but both the Network Dynamics Hypothesis and Process

Domains Concept generally capture the relationships between multiscale geomorphic

processes and aquatic communities. The natural breaks in the river continuum imposed

by geomorphic processes and network structure in these tropical island streams are

apparently most influential in determining the distribution of fish and macroinvertebrates.

However, at the pool and reach scales, decapod abundances are highly variable, and

reflect a complex interaction of geomorphic patchniness and biotic interactions.

204



85: 157-169.

Anderson-Olivas EA, Freeman MC, Pringle CM. 2006. Ecological consequences of

hydropower development in Central America: Impacts of small dams and water

diversion on neotropical stream fish assemblages. River Research and


Angermeier PL, Karr JR. 1983. Fish communities along environmental gradients in a

system of tropical streams. Environmental Biology of Fishes 9: 117-135.

Balon EK, Bruton MN. 1994. Fishes of the Tatinga River, Comoros, with comments on

freshwater amphidromy in the goby Sicyopterus lagocephalus. Ichthyological

exploration of freshwaters 5: 25-40.

Bass D. 2003. A comparison of freshwater macroinvertebrate communities on small

Caribbean islands. BioScience 53: 1094-1100.

Benda L, Poff NL, Miller D, Dunne T, Reeves G, Pess G, Pollock M. 2004. The Network

Dynamics Hypothesis: How channel networks structure riverine habitats.


Benstead JP, March JG, Pringle CM. 2000. Estuarine larval development and upstream

post-larval migration of freshwater shrimps in two tropical rivers of Puerto Rico.


205

Blanco JF, Scatena FN. 2005. Floods, habitat hydraulics, and upstream migration of

Neritina virginea (Gastropods: Neritidae) in Northeastern Puerto Rico. Caribbean


Blanco JF, Scatena FN. 2006. Hierarchical contribution of river-ocean connectivity,

water chemistry, hydraulics, and substrate to the distribution of diadromous snails

in Puerto Rican streams. Journal of the North American Benthological Society 25:

82-98.

Blanco JF, Scatena FN. 2007. The spatial arrangement of Nertina virginea (Gastropoda:

Neritidae) during upstream migration under a road crossing. River Research and

Applications.

Box GEP, Cox DR. 1964. An analysis of transformations. Journal of Royal Statistical

Society, Series B 26: 211-246.

Covich AP. 1988. Geographical and historical comparisons of neotropical streams: biotic

diversity and detrital processes in highly variable habitats. Journal of the North

American Benthological Society 7: 361-386.





Covich AP, McDowell WH. 1996. The stream community. In: The food web of a tropical

rain forest. Reagan DP, Waide RB (eds). University of Chicago Press: Chicago,

IL; 433-459.

206

Covich AP, Crowl TA, Johnson SL, Pyron M. 1996. Distribution and abundance of

tropical freshwater shrimp along a stream corridor: response to disturbance.


Covich AP, Crowl TA, Scatena FN. 2000. Linking habitat stability to floods and

droughts: effects on shrimp in montane streams, Puerto Rico. Verhandlungen

Internationale vereinigung für Theoretische und Angewandte Limnologie 27: 1-5.

Covich AP, Crowl TA, Heartsill-Scalley T. 2006. Effects of drought and hurricane

disturbances on headwater distributions of palaemonid river shrimp

(Macrobrachium spp.) in the Luquillo Mountains, Puerto Rico. Journal of the

North American Benthological Society 25: 99-107.

Crowl TA, Townsend CR, Bouwes N, Thomas H. 1997. Scales and causes of patchiness

in stream invertebrate assemblages: top-down predator effects?. Journal of the

North American Benthological Society 16: 277-285.

Crowl TA, Welsh V, Heartsill-Scalley T, Covich AP. 2006. Effects of different types of

conditioning on rates of leaf-litter shredding by Xiphocaris elongata, a

Neotropical freshwater shrimp. Journal of the North American Benthological

Society 25: 198-208.

Erdman DS. 1984. Exotic fishes in Puerto Rico. In: Distribution, biology, and

management of exotic fishes.Courtney WR Jr., Stauffer JR Jr. (eds). Johns

Hopkins University Press: Baltimore, MD; 162-176.

207

Esselman PC, Freeman MC, Pringle CM. 2006. Fish-assemblage variation between

geologically defined regions and across a longitudinal gradient in the Monkey

River Basin, Belize. Journal of the North American Benthological Society 25:

142-156.

Fausch KD, Torgersen CE, Baxter CV, Li HW. 2002. Landscapes to riverscapes:

Bridging the gap between research and conservation of stream fishes. BioScience

52: 483–498.

Frissell CA, Liss WJ,Warren WJ,Hurley MD. 1986. A hierarchical framework for stream

habitat classification: Viewing streams in a watershed context. Environmental




Journal of Science 32(4): 413-424.

Gilliam JF, Fraser DF, Alkins-Koo M. 1993. Structure of a tropical stream fish

community: a role for biotic interactions. Ecology 74: 1856-1870.

Gorman OT, Karr JR. 1978. Habitat structure and stream fish communities. Ecology 59:

507-515.



Society of America Bulletin. 102(3): 340-352.

Greathouse EA, Pringle CM. 2006. Does the river continuum concept apply on a tropical

island? Longitudinal variation in a Puerto Rican stream. Canadian Journal of

Fisheries and Aquatic Sciences 63: 134-152.

208

Greathouse EA, Pringle CM, McDowell WH, Holmquist JG. 2006. Indirect upstream

effects of dams: consequences of migratory consumer extirpation in Puerto Rico.

Ecological Applications 16: 339-352..

Gross MR, Coleman RM, McDowall RM. 1988. Aquatic productivity and the evolution

of diadromous fish migration. Science 239: 1291-1293.



301-315.





136.

Hein CL, Pike AS, Blanco JF, Scatena FN, Sherrill KR, Covich AP, Crowl TA. In

preparation. Influence of road networks on the community structure of

diadromous fauna in tropical island streams.

Hoeinghaus DJ, Winemiller KO, Taphorn DC. 2004. Compositional change in fish

assemblages along the Andean piedmont–Llanos floodplain gradient of the Río

Portuguesa, Venezuela. Neotropical Ichthyology 2: 85-92.

Holmquist JG, Schmidt-Gengenbach JM, Yoshioka BB. 1998. High dams and marine-

freshwater linkages: effects on native and introduced fauna in the Caribbean.

Conservation Biology 12: 621-630.

209

Ibanez C, Oberdorff T, Teugels G, Mamononekene V, Lavoué S, Fermon Y, Paugy D,

Toham AK. 2007. Fish assemblage structure and function along environmental

gradients in rivers of Gabon (Africa). Ecology of Freshwater Fish 16: 315-334.

Keith P. 2003. Biology and ecology of amphidromous Gobiidae of the Indo-Pacific and

the Caribbean regions. Journal of Fish Biology 63: 831-847.

March JG, Benstead JP, Pringle CM, Scatena FN. 1998. Migratory drift of larval

freshwater shrimps in two tropical streams, Puerto Rico. Freshwater Biology 40:

261-273.

March JG, Pringle CM, Townsend MJ, Wilson AI. 2002. Effects of freshwater shrimp

assemblages on benthic communities along an altitudinal gradient of a tropical

island stream. Freshwater Biology 47: 377-390,

March JG, Benstead JP, Pringle CM, Scatena FN. 2003. Damming tropical island

streams: problems, solutions and alternatives. BioScience 53: 1069-1078.

McCune BC. 2006. Non-parametric habitat models with automatic interactions. Journal

of Vegetation Science 17: 819-830.

McDowell WH, Asbury CE. 1994. Export of carbon, nitrogen, and major ions from three

tropical montane watersheds. Limnology and Oceanography 39: 111-125.



Montgomery DR. 1999. Process Domains and the River Continuum. Journal of the


210

Newson MD, Newson CL. 2000. Geomorphology, ecology, and river channel habitat:

mesoscale approaches to basin-scale challenges. Progress in Physical Geography

24: 195-217.

Ortiz-Maldonado C. 2005. Influence of habitat morphology on shrimp communities in a

headwater stream: Quebrada Prieta, Puerto Rico. Undergraduate honors thesis.

University of Puerto Rico: Rio Piedras, PR.

Ortiz-Zayas JR, Lewis WM, Saunders JF, McCutchan JH, Scatena FN. 2005. Metabolism

of a tropical rainforest stream. Journal of the North American Benthological

Society 24: 769-783.

Perry JA, Schaeffer DJ. 1987. The longitudinal distribution of riverine benthos: A river

discontinuum? Hydrobiologia 148: 257–268.

Poff NL. 1997. Landscape filters and species traits: Towards mechanistic understanding

and prediction in stream ecology. Journal of the North American Benthological

Society 16: 391-409.

Poole GC. 2002. Fluvial landscape ecology: addressing uniqueness within the river

discontinuum. Freshwater Biology 47: 641-600.

Pouilly M, Barrera S, Rosales C. 2006. Changes of taxonomic and trophic structure of

fish assemblages along an environmental gradient in the Upper Beni watershed

(Bolivia). Journal of Fish Biology 68: 137-156.

Power MR, Stout RJ, Cushing CE, Harper PP, Hauer FR, Matthews WJ, Moyle PB,

Statzner B, Waid de Badgen IR. 1988. Biotic and abiotic controls in river and

stream communities. Journal of the North American Benthological Society 7:

456-479.

211

Pringle CM, Naiman RJ, Bretschko G, Karr JR, Oscwood MW, Webster JR, Welcomme

RL, Winterbourn MJ. 1988. Patch dynamics in lotic systems: the stream as a

mosaic. Journal of the North American Benthological Society 7: 503-524.

Pringle CM, Scatena FN. 1998. Factors affecting aquatic ecosystem deterioration in Latin

America and the Caribbean. In: Tropical Managed Ecosystems: New perspectives

on Sustainability. Hatch U, Swisher ME (eds). Oxford University Press: New

York.

Pringle CM, Freeman M, Freeman B. 2000. Regional effects of hydrologic alterations on

riverine macrobiota in the New World: Tropical-temperate comparisons.


Pyron M, Covich AP. 2003. Migration patterns, densities, and growth of Neritina

punctulata snails in Rio Espiritu Santo and Rio Mameyes, Northeastern Puerto

Rico. Caribbean Journal of Science 39: 338-347.

Rahel FJ, Hubert WA. 1991. Fish assemblages and habitat gradients in a Rocky

Mountain–Great Plains stream: biotic zonation and additive patterns of

community change. Transactions of the American Fisheries Society 120: 219-332.

Ramirez A, Melendez-Colom E. 2003. Meteorological summary for El Verde Field

Station: 1975-2003. Available at http://luq.lternet.edu/data

Ramírez A, Pringle CM, Douglas M. 2006. Temporal and spatial patterns in stream

physicochemistry and insect assemblages in tropical lowland streams. Journal of

the North American Benthological Society 25: 108-125.

Reagan DP, Waide RB. 1996. The Food Web of a Tropical Rain Forest. University of

Chicago Press: Chicago, IL.

212

Rice SP, Greenwood MT, Joyce CB. 2001. Tributaries, sediment sources, and the

longitudinal organization of macroinvertebrate fauna along river systems.

Canadian Journal of Fisheries and Aquatic Sciences 58: 828–840.

Rodiles-Hernández R, Díaz-Pardo E, Lyons J. 1999. Patterns in the species diversity and

composition of the fish community of the Lacanja River, Chiapas, Mexico.

Journal of Freshwater Ecology 14: 455-468.

Rosgen DL. 1994. A Classification of Natural Rivers. Catena 22: 169-199.

Santos-Román DM, Warner GS, Scatena FN. 2003. Multivariate analysis of water quality

and physical characteristics of selected watersheds in Puerto Rico. Journal of the


Scatena FN. 2001. Ecological rhythms and the management of humid tropical forests:

examples from the Caribbean National Forest, Puerto Rico. Forest Ecology and


Scatena FN, Johnson SL. 2001. Instream-flow analysis for the Luquillo Experimental

Forest, Puerto Rico: methods and analysis. United States Department of

Agriculture Forest Service General Technical Report IITF-GTR-11: 30pp.

Scatena FN, Marcial JM, Hein CL, Pike AS, Gonzalez-Caban A, McDowell WH. In

revision. Habitat preferences and water quality impacts of recreational swimming

in the Luquillo Mountains, Puerto Rico. Submitted to River Research and

Applications.




213

Schlosser IJ. 1991. Stream fish ecology: A landscape perspective. BioScience 41: 704–

712.

Schneider DW, Lyons J. 1993. Dynamics of upstream migration in two species of

tropical freshwater snails. Journal of the North American Benthological Society

12: 3-16.

Schoenfuss HL, Blob RW. 2003. Kinematics of waterfall climbing in Hawaiian

freshwater fishes (Gobiidae): vertical propulsion at the aquatic-terrestrial

interface. Journal of Zoology 261: 191-205.



1:20,000 scale.

Smith GC, Covich AP, Brasher AMD. 2003. An ecological perspective on the

biodiversity of tropical island streams. BioScience 53: 1048-1051.

Statzner B, Higler B. 1985. Questions and Comments on the River Continuum Concept.

Canadian Journal of Fisheries and Aquatic Sciences 42: 1038-1044.

Statzner B, Higler B. 1986. Stream hydraulics as a major determinant of benthic

invertebrate zonation patterns. Freshwater Biology 16: 127-139.

Stuart TA. 1962. The leaping behavior of salmon and trout at falls and obstructions.

Freshwater and Salmon Fisheries Research 28: 1-46.

Thorp JH, Thoms MC, Delong MD. 2006. The riverine ecosystem synthesis:

biocomplexity in river networks across space and time. River Research and


214

Trainor K, Church M. 2003. Quantifying variability in stream channel morphology.

Water Resources Research 39(9): 1248 1259.

Townsend CR. 1989. The patch dynamics concept of stream community ecology. Journal

of the North American Benthological Society 8: 36-50.

Townsend CR, Crowl TA. 1991. Fragmented population structure in a native New

Zealand fish: an effect of introduced brown trout?. Oikos 61: 347-354.

Walters DM, Leigh DS, Freeman MC, Freeman BJ, Pringle CM. 2003. Geomorphology

and fish assemblages in a Piedmont river basin, U.S.A. Freshwater Biology 48:

1950-1970.

Ward JV, Tockner K, Arscott DB, Claret C. 2002. Riverine landscape diversity.

Freshwater Biology 47: 517–539.

Wedderburn SD, Walker KF, Zampatti BP. 2007. Habitat separation of Craterocephalus

(Atherinidae) species and populations in off-channel areas of the lower River

Murray, Australia. Ecology of Freshwater Fish 16: 442-449.

Wiens JA. 2002. Riverine landscapes: Taking landscape ecology into the water.

Freshwater Biology 47: 501–515.

Winemiller KO, Leslie MA. 1992. Fish assemblages across a complex, tropical

freshwater/marine ecotone. Environmental Biology of Fishes 34: 29-50.

Wu JG, Loucks OL. 1995. From balance of nature to hierarchical patch dynamics: A

paradigm shift in ecology. Quarterly Review of Biology 70: 439-466.

Vannote RL, Minshall GW, Cummins KW, Sedell JR, Cushing CE. 1982. The River

Continuum Concept. Canadian Journal of Fisheries and Aquatic Sciences 37: 130-

137.

215

CHAPTER 6

CONCLUSIONS AND FUTURE RESEARCH

SUMMARY AND CONCLUSIONS

This purpose of this dissertation was to quantify the longitudinal patterns in

stream channel morphology in a tropical mountain stream network and assess the relative

influences of contrasting hydraulic and lithologic forces that sculpt channel morphology,

as well as to address how the hydrologic and geomorphologic gradients present in the

stream network influence aquatic habitat.

Several studies were presented here to address these research aims. The first

(Chapter 2) developed and expanded GIS-based relationships to estimate hydrologic

parameters throughout that watersheds. These included estimates of mean annual rainfall,

runoff, and discharge based on elevation and flow accumulation models, as well as an

expanded map of the stream network. The spatial framework that was discussed in this

chapter was used a template to develop hydrologic, geomorphic, and ecological patterns

in subsequent chapters.

Chapter 3 developed and used a technique to calibrate high-flow riparian features

to long-term gage records for the purpose of determining an active-channel boundary

marker. The general approach of surveying the first occurrence of riparian features and

using multivariate statistical analysis to link these occurrences to 15-minute flow duration

provided an internally consistent framework for identifying flow frequencies within the

region. In the Luquillo streams, it was found that the incipient presence of soil and woody

shrubs and trees can be used as an indicator of hydrogeomorphic site conditions to

216

identify active-channel boundaries that occur at a constant flow frequency throughout the

stream network. Furthermore, it was found that the total duration of time that bankfull

stage is exceeded in the Luquillo streams is similar to that in many temperate streams, but

that floods of bankfull magnitude occur much more frequently in these streams. This

study concluded that flows with similar frequencies influence the establishment of

riparian vegetation, soil development, and substrate characteristics along tropical stream

channel margins in similar ways to those of temperate and alluvial rivers. Lastly, the

method developed in this study was essential to identifying a common marker of flow

frequency as a basis to compare channel geometry in subsequent chapter.

The next study (Chapter 4) highlighted the central geomorphologic question of

the dissertation. The key conclusion drawn from this study is that although there are

apparent non-fluvial and lithologic controls on local channel morphology, strong fluvial

forces are sufficient to override boundary resistance and give rise to systematic basin-

scale patterns. Lithologic influence is evident in non-uniformly graded longitudinal

profiles and prevalence of large (presumably immobile) boulders that are delivered from

hillslope weathering. Since the time scale of slope adjustment is vastly greater than the

adjustment of channel geometry, lithology imposes a slope upon which hydraulic forces

sculpt the channel on shorter time scales. Yet the dominance of hydraulic influences are

evident through well developed downstream hydraulic geometry, and the apparent ability

of the channels to mobilize coarse sediment throughout the watershed. In an almost

paradoxical sense, the stream network displays many lithologic features similar to other

mountain streams, but the intense hydrologic regime gives rise to a threshold channel that

shares some similarities with fully alluvial rivers.

217

The last study (Chapter 5) addressed the ecological implications of the patterns in

geomorphology. The conclusions drawn from this study highlight the importance of the

scale at which ecological processes are observed. At the basin-scale, waterfalls hinder the

upstream migration of some species, and consequently segment the stream community.

Yet above these waterfalls where decapods are present, local-scale geomorphology

influence decapod abundance. However, at these local scales, the geomorphic

environment is patchy, as the variation in pool characteristics varies as strongly from pool

to pool, as from reach to reach. This natural variability in the geomorphic environment

present at these scales gives rise to a similar patchiness in the abundance of biota.

FUTURE RESEARCH

This dissertation provides baseline information about the stream environment to

pursue further avenues of geomorphology research. First, the technique of quantifying the

flow frequency associated with high-flow riparian features using long-term stream gage

records should be applied in a variety of stream that have different flow regimes. Using

this technique in other environments, is there a similar relationship between riparian

vegetation and flow frequency? Does this technique provide consistent results in streams

throughout the island of Puerto Rico? Within other mountainous streams in the humid

tropics? It is expected that stream with both a similar flashy flow regime and humid

tropical climate with similar vegetation types will have analogous riparian features

occurring at flow-frequencies consistent with those determined in this study.

Another further extension of this technique would be to investigate linkages

between the frequency of flooding and the incipient presence of different vegetation

218

species. In this dissertation, vegetation is grouped into broad categories (mosses and

lichens, herbs, grasses, shrubs, and trees), yet there may be differences in the flooding

tolerance of individual species that comprise these categories. Furthermore, more in-

depth studies of in-channel mosses, herbs, and shrubs may provide additional information

about the frequency of sub-effective flows.

The geomorphology discussion in this research poses further questions about the

long-term adjustment of these channels, and implications for landscape evolution. One

issue that necessitates future research is to address site-scale linkages between channel

hydrology and sediment transport. The concept of sediment transport is critical to the

understanding of channel morphology, yet is difficult to precisely quantify in streams

with large boulders. There are few established analytical tools to estimates rates of

sediment transport, and also a lack of data in steep mountain streams to calibrate such

tools. For example, although the sediment transport equations used in this research

suggest that the median and coarse grains in many headwater reaches are mobile, are the

largest boulders mobile? To answer such a question would require developing new

techniques to assess sediment transport in streams with large boulders, and also modify

existing techniques that have been developed on gravel-bed streams to collect empirical

data on the movement of these boulders.

Several questions about the evolution of the stream network remain. Most

notably, how old are these channels? And have the steepland reaches migrated over

geologic time? As mentioned, streams in Puerto Rico have assumed to flow since the

uplift of the Island during the Cretaceous period, and sedimentary deposits across the

island suggest that clastic sediment was delivered to the ocean by streams during the

219

Miocene. However, it is not known whether the streams draining the mountain regions

have had relatively similar paths through geological time, or whether these steepland

streams have adjusted their course through ridge migration. Although it may not be

possible to answer these questions directly, they highlight the importance of further

understanding the history of this landscape. Deep insight into landscape evolution may be

gathered from the use geochronological techniques to evaluate the thermal and

mechanical history of the granodiorite rocks, and to constrain uplift rates over time.

Another pressing question above the evolution of this landscape is whether

waterfalls migrate as slope adjusts through long-term degradation? Field observations

indicate that the location of waterfalls are influenced by a variety of lithologic and

tectonic factors. Some waterfalls occur at lithologic boundaries and changes in

volcaniclastic units, others occur along faults and at locally exposed bedrock outcrops,

and others, particularly those flowing across granodiorite, exploit natural joints in the

rock. Although waterfalls generally occur at higher elevations (>300m), several

anomalous waterfalls are present at lower elevations, suggesting that waterfalls do not

necessarily correlate with elevated peneplain surfaces. In some rivers, waterfalls migrate

headwater over geologic time, whereas others are more permanent and act as a local

base-level for the upstream drainage basin. A detailed survey of the waterfalls through

the Luquillo Mountains, coupled with models of bedrock incision, could illuminate the

evolution of these features. If waterfalls are migrating headward, it would imply that the

long-term tendency of these streams is to minimize gradients in energy expenditure. In

contrast, if the waterfalls tend to be more permanent features, this may imply that the

220

underlying lithology will always exert strong control on the slope of these stream

channels.

Lastly, further linkages between geomorphology and aquatic habitat should be

investigated at the microhabitat scale. The nature of the surveying methods in this study

did not allow for such fine-scale habitat assessment. However, the hydraulics and

distribution of sediment within an individual pool may be just as complex as their

distribution throughout the basin. Do the fish and shrimp living seek out optimal

hydraulic and geomorphologic features within a pool that may influence their abundance?

And if so, how do these vary within the pool? And how does the flow regime of frequent

short-duration floods alter the geomorphology of specific pools? An investigation into

these questions at the microhabitat scale could provide further insight into the complex

interactions between geomorphology, hydraulics, and aquatic habitat.

Ultimately, this research provides baseline information on physical and biological

processes in relatively unaltered tropical streams and can be used to inform such further

studies that document human interactions with stream networks.

221

APPENDIX DESCRIPTION OF VARIABLES

Site Information

Watershed – The watershed where the cross-section was located.

Biocomplexity ID – The site identification name for sites included in the Biocomplexity

study. Blank indicates a cross-section that was not included in the Biocomplexity

study.

Cross-Section ID – The identification label assigned to each cross-section. Non-

Biocomplexity sites are labeled by a letter corresponding to the watershed and a

number corresponding to the order in which the cross-section was measured. For

sites included in the Biocomplexity study, the label is prefaced by the letter (U or

D) indicated whether the cross-section was located upstream or downstream of

the road-river crossing, followed by a number indicating the sequential ordering

of the cross-section from the road-river crossing. Those that are labeled “Bridge”

indicate a cross-section taken at a pool directly at the road-river crossing.

PR Datum N – Northing coordinate (m) of the cross-section according to the following

coordinate system used for the Puerto Rico DEM:

NAD_1927_Lambert_Conformal_Conic

Projection: Lambert_Conformal_Conic

False_Easting: 152400.304801

False_Northing: 0.000000

Central_Meridian: -66.433333

Standard_Parallel_1: 18.033333

222

Standard_Parallel_2: 18.433333

Latitude_Of_Origin: 17.833333

GCS_North_American_1927

PR Datum E – Easting coordinate (m) of the cross-section according the coordinate

system described above.

Main Stem – Cross-sections along a tributary or the main stem of the river (0 = tributary,

1 = main stem)

Rock Type – Rock type underlying the channel at each cross-section (VC =

Volcaniclastic, GD = Granodiorite, AL = Alluvium, DK = Mafic Dike)

Formation – Formation of the rock type underlying the channel at each cross-section

(Kt =Tabonuco formation, Kf = Frailes formation, Kh = Hato Puerco Formation,

Tqd = Granodiorite, Qa = Quaternary alluvium, Tkmi = Mafic Dike)

h – Elevation of the cross section (m a.s.l.)

havg – Average elevation of the contributing basin upstream of the cross-section (m a.s.l.)

DA – Drainage area of the contributing basin upstream of the cross-section (km2)

Qmean – Mean annual discharge at the cross-section, based on equation 2.3 using drainage

area and average upstream elevation (m3/s)

Distdown – Distance along the stream from the cross-section to coast (m)

Distup – Distance along the stream from the cross-section to the headwaters (m)

S – Stream gradient (m/m)

Hillslopemax – Maximum adjacent hillslope gradient within a 50m radius of the cross-

section (°)

223

Baseflow Channel Geometry

Q bf – Discharge at baseflow, Q50 (m3/s), based on a modified equation 2.3

w bf – Baseflow channel width (m)

R bf – Baseflow hydraulic radius (m)

dmax bf – Maximum depth along the cross-section at baseflow (m)

P bf – Wetted perimeter at baseflow (m)

A bf – Cross-sectional area at baseflow (m2)

v bf – Average velocity through the cross-section at baseflow (m/s)

w/d bf – Width to depth ratio of the cross-section at baseflow (m/m)

τ bf = Boundary shear stress at baseflow, based on equation 4.3

Ω bf = Stream power at baseflow, based on equation 4.6

Active Channel Geometry

Q ac – Active-channel discharge, Q99.84 (m3/s), based on equation 4.1

w ac – Active-channel width (m)

R ac – Active-channel hydraulic radius (m)

dmax ac – Maximum depth along the cross-section of the active-channel discharge (m)

P ac – Wetted perimeter of the active-channel (m)

A ac – Cross-sectional area at the active-channel discharge (m2)

v ac – Average velocity through the cross-section at the active-channel discharge (m/s)

w/d ac – Width to depth ratio of the cross active-channel (m/m)

τ ac = Boundary shear stress at the active-channel discharge, based on equation 4.3

Ω ac = Stream power at the active-channel discharge, based on equation 4.6

224

Grain Size and Pebble Count Data

d16 – Fine grain-size fraction, coarser than 16% of the sample (mm)

d50 – Median grain size, coarser than 50% of the sample (mm)

d84 – Coarse grain-size fraction, coarser than 84% of the sample (mm)

dmax – Maximum grain size (mm)

Simpson’s Index – Simpson’s diversity index characterizing diversity in grain size,

based on categorical pebble count data.

Sorting Index – Grain size sorting index, calculated as 1684 dd

Bedrock – Percentage of the pebble count that included bedrock

Megaboulder – Percentage of the pebble count that included megaboulder-sized clasts

(>2000 mm)

Boulder – Percentage of the pebble count that included boulder-sized clasts (256-2000

mm)

Cobble – Percentage of the pebble count that included cobble-sized clasts (64-256 mm)

Gravel – Percentage of the pebble count that included gravel-sized sediment (2-64 mm)

Sand – Percentage of the pebble count that included sand-sized sediment (0.063-1 mm)

Fines – Percentage of the pebble count that included fine sediment (0.001-0.063 mm)

225

Additional Biocomplexity Pool Information

Drop – Maximum downstream drop in elevation according to a 10m DEM (m)

Pdepmax – Maximum pool depth at baseflow (m)

Pdep cv – Coefficient of variation of depth at baseflow based on six measurements

spanning the length of the pool (dimensionless)

Plength – Length of the pool (m)

P area – Surface area of the pool at baseflow (m2)

P volume – Volume of the pool at baseflow based on the product of length and surface

area (m3)

agr d – Proportion of agricultural (deforested non-urban) land cover in the

downstream corridor

agr u – Proportion of agricultural (deforested non-urban) land cover in the upstream

basin

fst d– Proportion of forested land cover in the downstream corridor

fst u – Proportion of forestedl land cover in the upstream basin

urb d – Proportion of urban land cover in the downstream corridor

urb u – Proportion of urban land cover in the upstream basin

226

Wat

ersh

edBi

ocom

plex

ity I

DCr

oss-

Sect

ion

IDPR

Dat

um N

PR D

atum

EM

ain

Stem

Roc

k Ty

peFo

rmh

h avg

DA

Qm

ean

dist

dow

ndi

stup

SH

illsl

ope m

ax

Mam

eyes

M1

59,7

15

222,

885

1AL

Qa

733

234

.5

6.19

2848

1463

90.

002

0.

0M

amey

esM

259

,085

222,

525

1AL

Qa

1036

730

.6

6.02

3812

1367

50.

002

7.

4M

amey

esM

358

,725

222,

615

1AL

Qa

1041

124

.7

5.38

4189

1329

80.

008

17

.6M

amey

esM

455

,845

223,

965

1VC

Kt70

474

20.4

5.

08

8269

9217

0.01

6

28.3

Mam

eyes

M5

55,3

95

223,

965

1VC

Kt72

483

19.8

5.

03

8756

8731

0.01

6

34.0

Mam

eyes

M6

55,1

25

224,

325

1VC

Kt80

503

17.8

4.

69

9394

8093

0.01

5

46.1

Mam

eyes

M7

58,1

85

222,

795

1AL

Qa

2041

824

.2

5.37

4933

1255

40.

008

0.

2M

amey

esM

857

,915

223,

065

1AL

Qa

2042

224

.0

5.36

5376

1211

10.

008

27

.5M

amey

esM

956

,385

224,

145

1VC

Kt57

463

21.1

5.

14

7576

9911

0.01

7

29.8

Mam

eyes

M10

56,3

85

224,

145

0VC

Kt58

188

0.5

0.

05

7594

1752

0.08

4

14.0

Mam

eyes

M11

56,7

45

224,

055

1VC

Kt50

448

22.2

5.

24

7101

1038

60.

019

11

.9M

amey

esM

1257

,105

223,

875

1VC

Kt35

433

23.2

5.

32

6511

1097

60.

015

21

.9M

amey

esM

1353

,325

221,

625

0VC

Kt49

676

80.

6

0.22

1488

517

670.

313

31

.7M

amey

esM

1453

,055

221,

445

0VC

Kt60

579

60.

5

0.20

1525

313

990.

299

41

.6M

amey

esM

1553

,145

221,

175

0VC

Kt66

082

60.

4

0.16

1555

610

950.

148

27

.5M

amey

esM

1653

,235

221,

895

0VC

Kt46

071

31.

2

0.43

1467

219

790.

129

33

.2M

amey

esM

1755

,125

223,

515

0VC

Kt15

828

70.

1

0.02

9387

942

0.30

8

28.3

Mam

eyes

M18

55,3

05

223,

245

0VC

Kt18

846

60.

8

0.20

9624

2402

0.10

8

36.2

Mam

eyes

M19

55,3

05

222,

705

0VC

Kt31

138

00.

1

0.01

1020

241

90.

200

28

.3M

amey

esM

2055

,125

222,

705

0VC

Kt32

352

60.

2

0.06

1033

212

280.

238

42

.2M

amey

esM

2154

,855

222,

795

0VC

Kt36

058

10.

3

0.09

1046

915

580.

222

27

.3M

amey

esM

2257

,285

223,

605

1VC

Kt30

429

23.5

5.

34

6204

1128

30.

012

23

.9M

amey

esM

2354

,495

224,

595

0VC

Kf10

333

20.

9

0.16

1038

025

010.

213

26

.6M

amey

esM

2454

,495

224,

415

1VC

Kt11

353

615

.8

4.41

1055

769

300.

034

51

.6M

amey

esM

2554

,585

224,

685

1VC

Kf10

052

216

.8

4.59

1025

072

370.

030

28

.0M

amey

esM

2654

,315

224,

415

1VC

Kt12

053

715

.7

4.41

1074

567

420.

044

39

.9M

amey

esM

2752

,695

221,

985

0VC

Kt48

760

90.

2

0.05

1447

394

90.

250

30

.0M

amey

esM

2852

,335

221,

805

0VC

Kt47

575

02.

4

0.93

1481

926

680.

316

40

.7M

amey

esM

2952

,875

222,

615

0VC

Kf33

564

71.

6

0.54

1376

628

850.

168

28

.7M

amey

esM

3052

,785

222,

705

0VC

Kf32

067

84.

7

1.64

1365

438

330.

132

43

.4M

amey

esM

3153

,235

223,

515

1VC

Kf21

659

011

.8

3.62

1246

950

180.

095

37

.3M

amey

esM

3252

,875

223,

335

1VC

Kf25

055

36.

4

1.85

1297

844

870.

061

30

.0M

amey

esM

3352

,875

223,

335

0VC

Kf25

065

25.

1

1.72

1298

045

070.

066

23

.2M

amey

esM

3451

,795

222,

435

1G

DTq

d44

064

02.

1

0.70

1479

826

670.

064

13

.5M

amey

esM

3551

,885

222,

435

1G

DTq

d43

063

63.

1

1.01

1461

728

470.

047

30

.6M

amey

esM

3651

,885

222,

345

0G

DTq

d44

063

90.

9

0.29

1476

020

430.

118

14

.2

App

endi

x. S

ite

Info

rmat

ion

227

Wat

ersh

edBi

ocom

plex

ity I

DCr

oss-

Sect

ion

IDPR

Dat

um N

PR D

atum

EM

ain

Stem

Roc

k Ty

peFo

rmh

h avg

DA

Qm

ean

dist

dow

ndi

stup

SH

illsl

ope m

ax

Mam

eyes

M37

50,9

85

222,

075

1VC

Kf53

171

51.

1

0.39

1579

016

750.

179

42

.0M

amey

esM

3851

,165

222,

345

0VC

Kf51

661

40.

1

0.05

1558

267

90.

220

29

.1M

amey

esM

3951

,615

222,

255

1G

DTq

d45

566

71.

7

0.57

1504

924

150.

090

30

.9M

amey

esM

4053

,685

225,

045

0VC

Kf26

734

70.

1

0.01

1152

542

50.

238

30

.9M

amey

esM

4153

,685

224,

955

0VC

Kf27

035

40.

1

0.01

1155

049

80.

198

30

.3M

amey

esM

4253

,595

224,

865

0VC

Kf27

046

50.

3

0.08

1162

612

550.

164

24

.1M

amey

esM

4360

,255

223,

335

1AL

Qa

332

834

.9

6.20

2059

1542

80.

002

0.

0Es

pirit

u Sa

nto

ES1

54,4

05

217,

395

0VC

Kh35

073

32.

7

1.00

1538

034

310.

208

28

.6Es

pirit

u Sa

nto

ES3

54,0

45

218,

115

0VC

Kh52

578

10.

9

0.36

1622

925

630.

071

20

.4Es

pirit

u Sa

nto

ES4

54,0

45

218,

565

0VC

Kh61

081

40.

8

0.32

1669

820

940.

144

25

.1Es

pirit

u Sa

nto

ES5

53,7

75

218,

835

0VC

Kh71

688

70.

4

0.16

1720

115

910.

194

23

.1Es

pirit

u Sa

nto

ES6

54,7

65

217,

035

0VC

Kh26

069

13.

1

1.09

1484

439

670.

132

28

.2Es

pirit

u Sa

nto

ES7

58,3

65

217,

845

1AL

Qa

1746

022

.4

5.43

8539

1309

10.

011

11

.8Es

pirit

u Sa

nto

ES8

54,2

25

217,

665

0VC

Kh42

077

12.

3

0.90

1573

730

740.

192

30

.9Es

pirit

u Sa

nto

ES9

55,6

65

220,

275

0VC

Kh29

267

42.

3

0.80

1387

029

790.

225

30

.3Es

pirit

u Sa

nto

ES10

55,4

85

220,

365

0VC

Kh32

068

22.

3

0.79

1405

927

900.

183

23

.3Es

pirit

u Sa

nto

ES11

56,7

45

214,

965

0AL

Qa

4050

819

.6

5.20

1193

311

352

0.01

2

20.4

Espi

ritu

Sant

oES

1256

,745

214,

695

0D

KTk

mi

5051

919

.1

5.18

1224

811

037

0.02

6

8.2

Espi

ritu

Sant

oES

1352

,335

215,

955

0VC

Kh45

475

41.

9

0.72

1860

029

550.

247

27

.0Es

pirit

u Sa

nto

ES14

57,1

05

216,

675

1D

KTk

mi

5054

215

.8

4.46

1131

310

317

0.02

1

17.3

Espi

ritu

Sant

oES

1556

,655

216,

585

1D

KTk

mi

6855

615

.3

4.42

1200

796

240.

039

19

.4Es

pirit

u Sa

nto

ES16

56,2

95

216,

225

1VC

Kh80

598

13.8

4.

26

1273

688

950.

016

7.

8Es

pirit

u Sa

nto

ES17

53,8

65

216,

765

1VC

Kh44

073

46.

5

2.46

1576

758

640.

115

32

.6Es

pirit

u Sa

nto

ES18

54,4

95

217,

665

0VC

Kh37

951

10.

4

0.10

1551

210

560.

078

27

.4Es

pirit

u Sa

nto

ES19

54,7

65

217,

215

0VC

Kh27

346

70.

5

0.13

1487

316

950.

145

16

.9Es

pirit

u Sa

nto

ES20

53,3

25

216,

945

1VC

Kh52

075

45.

9

2.28

1641

052

210.

040

16

.8Es

pirit

u Sa

nto

ES21

52,9

65

217,

035

1VC

Kh54

576

65.

3

2.06

1694

046

910.

030

23

.5Es

pirit

u Sa

nto

ES22

54,4

05

217,

395

0VC

Kh35

073

32.

7

1.00

1538

034

310.

104

28

.6Es

pirit

u Sa

nto

ES23

61,2

45

217,

035

1AL

Qa

137

964

.2

13.0

1

4905

1838

00.

001

0.

0Es

pirit

u Sa

nto

ES24

63,8

55

219,

195

1AL

Qa

026

795

.0

14.0

3

233

2305

20.

001

0.

7Es

pirit

u Sa

nto

ES25

51,7

95

218,

655

1VC

Kh72

080

22.

9

1.21

1952

321

080.

029

21

.2Es

pirit

u Sa

nto

ES26

51,7

05

218,

745

1G

DTq

d72

179

72.

5

1.03

1961

620

150.

031

18

.4Es

pirit

u Sa

nto

ES27

51,7

05

218,

745

1G

DTq

d73

077

61.

3

0.52

1972

819

030.

037

10

.4Es

pirit

u Sa

nto

ES28

54,5

85

217,

215

0VC

Kh29

472

22.

7

1.02

1508

937

220.

111

30

.9Es

pirit

u Sa

nto

ES29

54,2

25

216,

675

1VC

Kh33

972

96.

6

2.49

1536

562

660.

227

39

.8Es

pirit

u Sa

nto

ES30

57,2

85

216,

765

1VC

Kh40

539

15.9

4.

46

1101

610

614

0.01

7

26.4

Espi

ritu

Sant

oES

3158

,095

217,

305

1AL

Qa

2346

522

.1

5.42

9429

1220

20.

009

28

.3

228

Wat

ersh

edBi

ocom

plex

ity I

DCr

oss-

Sect

ion

IDPR

Dat

um N

PR D

atum

EM

ain

Stem

Roc

k Ty

peFo

rmh

h avg

DA

Qm

ean

dist

dow

ndi

stup

SH

illsl

ope m

ax

Espi

ritu

Sant

oES

3258

,815

217,

935

1AL

Qa

1042

932

.0

7.26

8006

1362

50.

011

22

.4Es

pirit

u Sa

nto

ES33

55,7

55

216,

585

1VC

Kh10

462

812

.7

4.12

1343

781

930.

033

22

.1Es

pirit

u Sa

nto

ES34

55,1

25

216,

765

1VC

Kh15

367

211

.3

3.90

1432

673

050.

200

35

.1Es

pirit

u Sa

nto

ES35

59,8

05

217,

575

1AL

Qa

441

233

.4

7.30

6841

1479

00.

001

19

.7Es

pirit

u Sa

nto

ES36

59,8

95

217,

575

1AL

Qa

641

233

.4

7.30

6798

1483

20.

001

16

.5Es

pirit

u Sa

nto

ES37

52,2

45

217,

575

1VC

Kh63

079

04.

2

1.69

1799

036

400.

168

41

.4Es

pirit

u Sa

nto

ES38

52,5

15

217,

125

1VC

Kh57

378

24.

6

1.84

1750

341

270.

075

21

.8Bl

anco

B149

,365

220,

815

1G

DTq

d62

068

73.

2

1.14

1792

729

470.

009

15

.0Bl

anco

B249

,455

220,

905

1G

DTq

d62

568

83.

1

1.10

1804

828

260.

009

12

.8Bl

anco

B349

,545

220,

815

1G

DTq

d63

068

92.

7

0.96

1816

527

090.

009

1.

8Bl

anco

B449

,905

220,

725

1G

DTq

d63

069

31.

9

0.68

1855

223

220.

009

25

.0Bl

anco

B549

,005

220,

995

1G

DTq

d61

268

94.

0

1.41

1748

433

900.

112

23

.1Bl

anco

B649

,995

220,

545

0G

DTq

d64

069

60.

1

0.05

1886

558

30.

027

21

.1Bl

anco

B748

,375

219,

825

0G

DTq

d48

370

24.

1

1.50

1731

040

030.

063

31

.5Bl

anco

B848

,105

219,

195

0G

DTq

d47

576

63.

7

1.45

1728

735

680.

125

41

.8Bl

anco

B945

,945

219,

375

0G

DTq

d14

846

42.

5

0.62

1521

130

040.

098

21

.9Bl

anco

B10

45,4

05

220,

905

1AL

Qa

3058

427

.1

8.23

1335

179

610.

011

0.

0Bl

anco

B11

46,0

35

220,

815

1G

DTq

d51

657

21.3

7.

20

1407

572

370.

121

25

.0Sa

bana

S154

,225

226,

755

1AL

Qa

110

394

6.6

1.

39

8602

5094

0.03

0

16.3

Saba

naS2

53,9

55

226,

665

1AL

Qa

121

404

6.3

1.

36

9013

4683

0.04

1

15.9

Saba

naS3

53,5

05

226,

845

1AL

Qa

151

403

3.0

0.

64

9491

4205

0.09

0

24.0

Saba

naS4

53,5

95

226,

755

0AL

Qa

148

423

3.0

0.

68

9461

3735

0.07

5

23.6

Saba

naS5

53,5

95

226,

755

1AL

Qa

140

412

6.1

1.

33

9338

4358

0.05

4

19.6

Saba

naS6

53,2

35

226,

935

1AL

Qa

180

423

2.6

0.

59

9833

3863

0.06

7

32.2

Saba

naS7

53,0

55

227,

025

1AL

Qa

203

436

2.5

0.

57

1014

435

520.

097

32

.2Sa

bana

S854

,765

226,

575

0AL

Qa

100

204

2.0

0.

23

8026

2262

0.01

7

19.7

Saba

naS9

54,7

65

226,

035

0AL

Qa

110

233

1.0

0.

13

8617

1671

0.02

9

21.2

Saba

naS1

055

,395

226,

485

1AL

Qa

9031

710

.4

1.78

6975

6721

0.01

3

25.1

Saba

naS1

155

,035

226,

665

1AL

Qa

100

326

9.9

1.

75

7632

6064

0.01

1

9.9

Saba

naS1

254

,495

226,

125

0VC

Kf14

121

80.

1

0.01

8993

669

0.11

0

20.3

Saba

naS1

355

,845

226,

395

1AL

Qa

8030

611

.2

1.86

6444

7252

0.01

8

25.0

Saba

naS1

456

,475

226,

485

1AL

Qa

3830

011

.5

1.89

5674

8023

0.07

7

41.0

Saba

naS1

556

,655

226,

665

1AL

Qa

3029

611

.8

1.91

5444

8252

0.01

2

11.8

Saba

naS1

654

,315

228,

105

0AL

Qa

160

273

2.5

0.

38

1024

724

910.

031

27

.8Sa

bana

S17

53,8

65

228,

555

0AL

Qa

190

272

1.2

0.

17

1117

415

640.

017

29

.2Fa

jard

o19

4 Br

idge

1U54

,225

234,

945

1AL

Qa

521

755

.2

6.79

3974

2210

60.

002

0.

0Fa

jard

o19

4 Br

idge

Brid

ge54

,405

235,

125

1AL

Qa

420

858

.1

6.88

3676

2240

40.

002

0.

0

229

Wat

ersh

edBi

ocom

plex

ity I

DCr

oss-

Sect

ion

IDPR

Dat

um N

PR D

atum

EM

ain

Stem

Roc

k Ty

peFo

rmh

h avg

DA

Qm

ean

dist

dow

ndi

stup

SH

illsl

ope m

ax

Faja

rdo

194

Brid

ge1D

54,4

95

235,

215

1AL

Qa

420

858

.1

6.88

3557

2252

30.

002

0.

0Fa

jard

o97

7 Br

idge

2U51

,705

230,

355

1AL

Qa

5027

738

.2

5.82

1126

014

820

0.00

7

7.0

Faja

rdo

977

Brid

ge1U

51,6

15

230,

265

1AL

Qa

5027

937

.9

5.80

1141

914

661

0.00

7

1.2

Faja

rdo

977

Brid

geBr

idge

51,7

05

230,

535

1AL

Qa

4927

738

.3

5.83

1113

214

948

0.00

7

1.8

Faja

rdo

977

Brid

ge1D

51,7

05

230,

715

1AL

Qa

4327

638

.3

5.83

1089

315

186

0.00

7

2.4

Faja

rdo

Char

co F

rio1D

48,4

65

228,

195

1AL

Qa

9050

211

.0

2.91

1718

988

910.

017

3.

7Fa

jard

oCh

arco

Frio

Brid

ge48

,465

227,

925

1AL

Qa

100

505

11.0

2.

90

1743

286

480.

012

3.

6Fa

jard

oCh

arco

Frio

1U48

,555

227,

925

1AL

Qa

100

505

11.0

2.

90

1747

486

050.

012

0.

0Fa

jard

oCh

arco

Frio

2U48

,555

227,

925

1VC

Kf10

050

610

.9

2.90

1754

185

390.

012

0.

7Fa

jard

oCh

arco

Frio

3U48

,735

227,

835

1VC

Kf10

050

710

.9

2.90

1769

283

880.

012

13

.6Fa

jard

oCh

arco

Frio

4U48

,735

227,

655

1VC

Kf10

051

010

.8

2.89

1793

781

430.

012

17

.5Fa

jard

oCh

arco

Frio

5U48

,825

227,

565

1VC

Kf11

051

610

.7

2.88

1812

679

540.

020

24

.8Fa

jard

oCh

arco

Frio

7U48

,825

227,

385

1VC

Kf11

051

910

.6

2.87

1828

477

960.

020

17

.8Fa

jard

oLa

Tin

aja

3U49

,185

226,

305

1VC

Kf15

055

99.

3

2.72

1952

765

530.

060

43

.3Fa

jard

oLa

Tin

aja

2U49

,185

226,

575

1VC

Kf14

055

29.

6

2.75

1928

068

000.

060

36

.3Fa

jard

oLa

Tin

aja

1U49

,095

226,

575

1VC

Kf13

054

79.

7

2.77

1918

568

950.

034

30

.7Fa

jard

oRiv

er B

ar2U

51,4

35

229,

815

0VC

Kf60

234

2.6

0.

34

1196

828

290.

029

18

.8Fa

jard

oRiv

er B

ar1U

51,3

45

229,

905

0VC

Kf60

233

2.6

0.

34

1191

028

870.

029

24

.2Fa

jard

oRiv

er B

ar1D

51,3

45

229,

905

0VC

Kf57

232

2.6

0.

34

1186

629

310.

029

22

.6Fa

jard

oRiv

er B

ar2D

51,3

45

229,

995

0VC

Kf50

232

2.6

0.

34

1179

230

060.

029

22

.6Fa

jard

oSm

all T

ribut

ary

2U48

,915

226,

755

0VC

Kf14

023

90.

3

0.03

1903

289

00.

116

25

.3Fa

jard

oSm

all T

ribut

ary

1U48

,915

226,

755

0VC

Kf13

023

70.

3

0.03

1900

791

40.

116

25

.0Fa

jard

oSm

all T

ribut

ary

1D48

,915

226,

755

0VC

Kf13

023

70.

3

0.03

1897

994

30.

116

25

.0Fa

jard

oSm

all T

ribut

ary

2D48

,915

226,

845

0VC

Kf12

423

70.

3

0.03

1895

197

10.

116

21

.2Fa

jard

oU

pper

Faj

ardo

Trib

2U50

,715

225,

585

0VC

Kf35

851

80.

4

0.12

2233

313

570.

185

34

.7Fa

jard

oU

pper

Faj

ardo

Trib

1U50

,715

225,

585

0VC

Kf36

051

80.

4

0.12

2235

313

370.

185

34

.8Fa

jard

oU

pper

Faj

ardo

3U50

,355

225,

675

1VC

Kf33

057

95.

4

1.63

2190

141

780.

025

39

.6Fa

jard

oU

pper

Faj

ardo

2U50

,175

225,

765

1VC

Kf32

057

55.

5

1.66

2170

243

770.

019

37

.6Fa

jard

oU

pper

Faj

ardo

1U50

,085

225,

945

1VC

Kf32

057

25.

6

1.68

2145

146

290.

019

31

.7M

amey

esAn

gelit

o Tr

ail

1U54

,495

224,

415

1VC

Kt11

353

615

.8

4.41

1055

769

300.

071

51

.6M

amey

esAn

gelit

o Tr

ail

1D54

,585

224,

505

1VC

Kt11

053

615

.8

4.41

1047

570

120.

071

48

.5M

amey

esAn

gelit

o Tr

ail

2D54

,585

224,

685

1VC

Kf10

052

216

.8

4.59

1025

072

370.

030

28

.0Es

pirit

u Sa

nto

Back

way

Sch

ool

2U58

,455

215,

775

0AL

Qa

2043

224

.1

5.51

9136

1415

00.

007

13

.9Es

pirit

u Sa

nto

Back

way

Sch

ool

1U58

,545

215,

865

0AL

Qa

2042

824

.4

5.52

8957

1432

90.

007

25

.7Es

pirit

u Sa

nto

Back

way

Sch

ool

1D58

,545

215,

865

0AL

Qa

2042

824

.4

5.52

8933

1435

30.

007

25

.7Es

pirit

u Sa

nto

Back

way

Sch

ool

2D58

,545

215,

955

0AL

Qa

2042

824

.4

5.52

8878

1440

70.

007

18

.2Es

pirit

u Sa

nto

Behi

nd B

risas

2U60

,255

216,

675

0D

KTk

mi

440

026

.2

5.58

6056

1722

90.

001

0.

0

230

Wat

ersh

edBi

ocom

plex

ity I

DCr

oss-

Sect

ion

IDPR

Dat

um N

PR D

atum

EM

ain

Stem

Roc

k Ty

peFo

rmh

h avg

DA

Qm

ean

dist

dow

ndi

stup

SH

illsl

ope m

ax

Espi

ritu

Sant

oBe

hind

Bris

as1U

60,2

55

216,

675

0D

KTk

mi

340

026

.2

5.58

5996

1728

90.

001

0.

0Es

pirit

u Sa

nto

Behi

nd B

risas

BRID

GE

60,3

45

216,

675

0D

KTk

mi

340

026

.2

5.58

5976

1730

90.

001

0.

0M

amey

esBi

sley

32U

53,5

95

224,

865

0VC

Kf27

746

70.

3

0.08

1165

412

270.

174

26

.1M

amey

esBi

sley

31U

53,5

95

224,

865

0VC

Kf27

046

50.

3

0.08

1162

612

550.

174

24

.1M

amey

esBi

sley

31D

53,5

95

224,

865

0VC

Kf26

945

70.

4

0.08

1160

212

790.

174

24

.9M

amey

esBi

sley

32D

53,5

95

224,

865

0VC

Kf26

445

50.

4

0.09

1157

813

030.

174

27

.0Es

pirit

u Sa

nto

Dav

ila3U

57,4

65

217,

035

0VC

Kh35

345

4.8

0.

89

1042

762

810.

012

5.

4Es

pirit

u Sa

nto

Dav

ila2U

57,5

55

217,

035

0VC

Kh35

344

4.8

0.

89

1036

963

400.

012

20

.5Es

pirit

u Sa

nto

Dav

ila1U

57,5

55

216,

945

0VC

Kh30

343

4.8

0.

89

1030

164

080.

012

22

.7Es

pirit

u Sa

nto

Dav

ila1D

57,6

45

216,

945

0VC

Kh30

342

4.8

0.

89

1026

364

460.

012

24

.3Es

pirit

u Sa

nto

Dav

ila2D

57,6

45

216,

945

0VC

Kh30

341

4.8

0.

89

1020

465

040.

012

28

.1M

amey

esEa

st P

eak

2U51

,885

220,

815

0VC

Kt67

073

20.

3

0.11

1603

091

10.

088

20

.4M

amey

esEa

st P

eak

1U51

,885

220,

905

0VC

Kt66

172

90.

3

0.11

1596

897

30.

130

20

.4M

amey

esEa

st P

eak

1D51

,975

220,

905

0VC

Kt65

772

80.

3

0.12

1592

510

160.

141

18

.7M

amey

esEa

st P

eak

2D51

,975

220,

905

0VC

Kt65

472

80.

3

0.12

1591

110

300.

141

18

.7Es

pirit

u Sa

nto

El V

erde

3U57

,105

216,

585

1D

KTk

mi

5054

215

.8

4.46

1141

610

215

0.02

1

47.3

Espi

ritu

Sant

oEl

Ver

de2U

57,1

05

216,

675

1D

KTk

mi

5054

215

.8

4.46

1136

310

267

0.02

1

33.9

Espi

ritu

Sant

oEl

Ver

de1U

57,1

05

216,

765

1D

KTk

mi

5054

115

.8

4.46

1128

910

341

0.02

1

9.8

Espi

ritu

Sant

oEl

Ver

de1D

57,1

05

216,

765

1D

KTk

mi

5054

115

.8

4.46

1123

510

395

0.02

1

21.2

Espi

ritu

Sant

oEl

Ver

de2D

57,1

05

216,

855

1D

KTk

mi

5054

015

.8

4.46

1117

710

454

0.02

1

20.9

Espi

ritu

Sant

oES

Rt.

3 P

aint

ball

BRID

GE

60,2

55

217,

125

1AL

Qa

540

534

.0

7.32

6223

1540

80.

001

0.

0Es

pirit

u Sa

nto

ES W

ater

fall

3U54

,045

216,

675

1VC

Kh41

073

26.

6

2.48

1559

960

310.

294

40

.6Es

pirit

u Sa

nto

ES W

ater

fall

2U54

,045

216,

675

1VC

Kh40

173

26.

6

2.48

1556

560

650.

294

43

.7Es

pirit

u Sa

nto

ES W

ater

fall

1U54

,135

216,

675

1VC

Kh37

673

16.

6

2.48

1549

361

380.

455

46

.1Es

pirit

u Sa

nto

ES W

ater

fall

1D54

,225

216,

675

1VC

Kh33

972

96.

6

2.49

1536

562

660.

227

39

.8Es

pirit

u Sa

nto

Jim

enez

2U58

,275

217,

755

1VC

Kh20

461

22.3

5.

43

8695

1293

60.

011

16

.3Es

pirit

u Sa

nto

Jim

enez

1U58

,365

217,

755

1VC

Kh20

460

22.4

5.

43

8610

1302

10.

011

2.

5Es

pirit

u Sa

nto

Jim

enez

1D58

,455

217,

845

1VC

Kh13

460

22.4

5.

43

8487

1314

40.

011

25

.8Es

pirit

u Sa

nto

Jim

enez

Wat

erfa

l2U

55,5

75

220,

275

0VC

Kh31

067

72.

3

0.80

1400

128

480.

112

22

.3Es

pirit

u Sa

nto

Jim

enez

Wat

erfa

l1U

55,5

75

220,

275

0VC

Kh30

967

62.

3

0.80

1396

228

870.

091

22

.3Es

pirit

u Sa

nto

Jim

enez

Wat

erfa

l1D

55,6

65

220,

275

0VC

Kh29

267

42.

3

0.80

1387

029

790.

250

30

.3Es

pirit

u Sa

nto

Jim

enez

Wat

erfa

l2D

55,8

45

220,

275

0VC

Kh25

566

62.

4

0.81

1368

331

660.

270

33

.4M

amey

esJu

an D

iego

3U53

,325

221,

625

0VC

Kt51

377

20.

6

0.22

1493

917

120.

370

30

.4M

amey

esJu

an D

iego

2U53

,325

221,

625

0VC

Kt50

176

90.

6

0.22

1490

517

470.

370

30

.4M

amey

esJu

an D

iego

1U53

,325

221,

715

0VC

Kt49

073

11.

1

0.39

1484

518

070.

182

37

.0M

amey

esJu

an D

iego

BRID

GE

53,3

25

221,

715

0VC

Kt48

573

01.

1

0.40

1481

518

370.

182

39

.0M

amey

esJu

an D

iego

1D53

,325

221,

715

0VC

Kt48

373

01.

1

0.40

1480

518

470.

182

39

.0

231

Wat

ersh

edBi

ocom

plex

ity I

DCr

oss-

Sect

ion

IDPR

Dat

um N

PR D

atum

EM

ain

Stem

Roc

k Ty

peFo

rmh

h avg

DA

Qm

ean

dist

dow

ndi

stup

SH

illsl

ope m

ax

Mam

eyes

La C

oca

Trai

l2U

53,8

65

222,

705

0VC

Kt35

560

30.

7

0.21

1296

317

770.

143

26

.7M

amey

esLa

Coc

a Tr

ail

1U53

,865

222,

795

0VC

Kt35

060

00.

7

0.21

1291

818

210.

147

26

.7M

amey

esLa

Coc

a Tr

ail

1D53

,865

222,

795

0VC

Kt33

859

70.

7

0.21

1285

418

850.

217

26

.3M

amey

esLa

Coc

a Tr

ail

2D53

,865

222,

885

0VC

Kt33

059

40.

7

0.21

1281

419

250.

156

28

.1M

amey

esLa

Lin

guen

te2U

54,8

55

222,

795

0VC

Kt36

958

60.

3

0.09

1051

315

140.

146

25

.0M

amey

esLa

Lin

guen

te1U

54,9

45

222,

795

0VC

Kt35

457

90.

3

0.09

1044

515

820.

146

27

.3M

amey

esLa

Lin

guen

te1D

54,9

45

222,

885

0VC

Kt34

057

00.

3

0.09

1036

216

640.

146

28

.3M

amey

esLa

Lin

guen

te2D

55,0

35

222,

885

0VC

Kt33

156

60.

3

0.09

1032

417

030.

146

28

.3M

amey

esLa

Maq

uina

2U55

,305

223,

245

0VC

Kt17

046

30.

8

0.20

9550

2476

0.16

2

34.9

Mam

eyes

La M

aqui

na1U

55,3

05

223,

155

0VC

Kt19

146

70.

8

0.20

9639

2388

0.16

2

36.2

Mam

eyes

La M

aqui

naBR

IDG

E55

,305

223,

245

0VC

Kt18

046

50.

8

0.20

9584

2442

0.16

2

36.2

Mam

eyes

La M

aqui

na1D

55,3

05

223,

335

0VC

Kt17

046

30.

8

0.20

9540

2486

0.16

2

32.6

Mam

eyes

La M

aqui

na2D

55,3

05

223,

335

0VC

Kt16

443

81.

0

0.22

9510

2516

0.16

2

34.7

Mam

eyes

La M

ina

Falls

2U52

,335

221,

715

0VC

Kt49

875

92.

3

0.89

1488

126

060.

213

31

.3M

amey

esLa

Min

a Fa

lls1U

52,3

35

221,

715

0VC

Kt50

576

02.

3

0.89

1493

925

480.

213

29

.9M

amey

esLa

Min

a Fa

lls1D

52,3

35

221,

805

0VC

Kt48

675

72.

3

0.90

1485

726

300.

213

31

.3M

amey

esLa

Veg

a2U

(ne

w)

58,1

85

222,

795

1AL

Qa

2041

824

.2

5.37

4943

1254

40.

008

1.

6M

amey

esLa

Veg

a2U

(ol

d)58

,185

222,

795

1AL

Qa

2041

824

.2

5.37

4987

1250

00.

008

5.

8M

amey

esLa

Veg

a1U

58,0

95

222,

885

1AL

Qa

2041

924

.1

5.36

5106

1238

10.

008

10

.6M

amey

esLa

Veg

aRO

AD58

,185

222,

615

1AL

Qa

2041

424

.5

5.37

4776

1271

10.

008

0.

0M

amey

esLa

Veg

a1D

58,2

75

222,

615

1AL

Qa

1941

424

.5

5.38

4664

1282

30.

008

0.

0M

amey

esLa

Veg

a2D

58,4

55

222,

525

1AL

Qa

1541

324

.5

5.38

4497

1299

00.

008

0.

0Es

pirit

u Sa

nto

Las

Tres

T (

1)2U

57,9

15

219,

285

0VC

Kh49

192

1.2

0.

14

9148

2919

0.04

2

35.3

Espi

ritu

Sant

oLa

s Tr

es T

(1)

1U58

,005

219,

285

0VC

Kh41

188

1.3

0.

14

9114

2953

0.04

2

28.0

Espi

ritu

Sant

oLa

s Tr

es T

(1)

1D58

,005

219,

285

0VC

Kh40

188

1.3

0.

14

9061

3005

0.04

2

20.0

Espi

ritu

Sant

oLa

s Tr

es T

(1)

2D58

,095

219,

375

0VC

Kh38

181

1.4

0.

15

9009

3058

0.04

2

15.9

Espi

ritu

Sant

oLa

s Tr

es T

(2)

2U57

,465

219,

465

0VC

Kh80

213

1.0

0.

12

9711

2356

0.06

5

11.2

Espi

ritu

Sant

oLa

s Tr

es T

(2)

1U57

,555

219,

465

0VC

Kh80

209

1.1

0.

12

9651

2416

0.06

5

14.9

Espi

ritu

Sant

oLa

s Tr

es T

(2)

1D57

,555

219,

375

0VC

Kh80

208

1.1

0.

13

9622

2444

0.06

5

24.8

Espi

ritu

Sant

oLa

s Tr

es T

(2)

2D57

,555

219,

375

0VC

Kh77

205

1.1

0.

13

9588

2478

0.06

5

30.1

Mam

eyes

Mam

eyes

Rt.

31U

59,8

95

223,

065

1AL

Qa

533

034

.8

6.20

2532

1495

50.

002

0.

0M

amey

esM

amey

es R

t. 3

1D60

,075

223,

155

1AL

Qa

432

934

.9

6.20

2289

1519

80.

002

0.

0Es

pirit

u Sa

nto

Prie

ta3U

54,6

75

217,

305

0VC

Kh29

248

40.

5

0.12

1496

716

020.

163

26

.3Es

pirit

u Sa

nto

Prie

ta2U

54,6

75

217,

215

0VC

Kh28

046

80.

5

0.13

1492

716

410.

163

26

.3Es

pirit

u Sa

nto

Prie

ta1U

54,7

65

217,

215

0VC

Kh27

646

70.

5

0.13

1487

816

900.

163

21

.3Es

pirit

u Sa

nto

Prie

ta1D

54,7

65

217,

125

0VC

Kh26

046

10.

5

0.13

1481

117

570.

163

15

.5Es

pirit

u Sa

nto

Prie

ta2D

54,8

55

217,

125

0VC

Kh25

045

90.

5

0.13

1478

817

800.

163

23

.7

232

Wat

ersh

edBi

ocom

plex

ity I

DCr

oss-

Sect

ion

IDPR

Dat

um N

PR D

atum

EM

ain

Stem

Roc

k Ty

peFo

rmh

h avg

DA

Qm

ean

dist

dow

ndi

stup

SH

illsl

ope m

ax

Mam

eyes

Puen

te R

oto

2U55

,035

224,

505

1VC

Kt90

507

17.6

4.

67

9669

7818

0.01

6

37.0

Mam

eyes

Puen

te R

oto

1U/B

RID

GE

54,9

45

224,

595

1VC

Kt90

508

17.5

4.

66

9780

7707

0.01

6

26.9

Mam

eyes

Puen

te R

oto

1D55

,035

224,

505

1VC

Kt90

507

17.6

4.

67

9584

7903

0.01

6

37.1

Mam

eyes

Puen

te R

oto

2D55

,125

224,

415

1VC

Kt90

505

17.7

4.

67

9468

8019

0.01

6

41.4

Mam

eyes

Q. A

non

2U58

,365

224,

595

0AL

Qa

3587

1.0

0.

06

6726

1930

0.01

1

3.6

Mam

eyes

Q. A

non

1U/B

RID

GE

58,3

65

224,

595

0AL

Qa

3586

1.0

0.

06

6688

1969

0.01

1

4.9

Mam

eyes

Q. A

non

1D58

,365

224,

505

0AL

Qa

3586

1.0

0.

06

6664

1993

0.01

1

4.9

Mam

eyes

Q. A

non

2D58

,365

224,

505

0AL

Qa

3585

1.0

0.

06

6630

2027

0.01

1

4.9

Espi

ritu

Sant

oSo

nado

ra3U

54,4

95

217,

305

0VC

Kh31

372

42.

7

1.01

1519

036

210.

286

30

.7Es

pirit

u Sa

nto

Sona

dora

2U54

,585

217,

215

0VC

Kh30

372

32.

7

1.02

1515

636

550.

286

32

.1Es

pirit

u Sa

nto

Sona

dora

1U54

,585

217,

215

0VC

Kh29

472

22.

7

1.02

1508

937

220.

111

30

.9Es

pirit

u Sa

nto

Sona

dora

1D54

,675

217,

125

0VC

Kh28

069

53.

1

1.09

1500

338

080.

144

22

.2Es

pirit

u Sa

nto

Sona

dora

2D54

,675

217,

125

0VC

Kh28

069

43.

1

1.09

1496

538

470.

144

22

.2Es

pirit

u Sa

nto

Toro

nja

2U54

,855

217,

215

0VC

Kh27

032

40.

1

0.02

1476

264

60.

132

32

.6Es

pirit

u Sa

nto

Toro

nja

1U54

,855

217,

215

0VC

Kh26

032

00.

1

0.02

1473

167

70.

132

31

.0Es

pirit

u Sa

nto

Toro

nja

1D54

,945

217,

215

0VC

Kh25

131

90.

1

0.02

1470

270

60.

132

27

.3Es

pirit

u Sa

nto

Toro

nja

2D54

,945

217,

125

0VC

Kh24

531

70.

1

0.02

1467

073

80.

132

27

.3

233

Wat

ersh

edBi

ocom

plex

ity I

DCr

oss-

Sect

ion

IDQ

bf

w b

f R

bf

dm

ax b

f P

bf

A b

f v

bf

w/d

bf

τ b

f Ω

bf

Mam

eye s

M1

0.71

8.2

0.27

0.50

8.4

2.3

0.

3

30

.0

6

16.8

Mam

eyes

M2

0.72

19.3

0.82

1.67

19.9

16.4

0.0

23.4

19

16.9

Mam

eyes

M3

0.67

24.7

0.37

0.60

25.0

9.2

0.

1

67

.5

28

50

.5

M

amey

esM

40.

6514

.9

0.

47

0.

94

15

.2

7.

2

0.1

31.4

76

105.

4

Mam

eyes

M5

0.65

10.4

0.47

0.78

10.9

5.1

0.

1

22

.2

73

10

1.4

M

amey

esM

60.

6118

.2

0.

48

0.

96

18

.4

8.

9

0.1

37.8

69

88.0

Mam

eyes

M7

0.67

19.0

0.25

0.39

19.2

4.8

0.

1

75

.5

19

50

.6

M

amey

esM

80.

6722

.2

0.

30

0.

59

22

.6

6.

9

0.1

73.1

23

50.6

Mam

eyes

M9

0.66

13.1

0.50

1.01

13.5

6.8

0.

1

26

.1

86

11

2.6

M

amey

esM

100.

003.

3

0.

12

0.

21

3.

3

0.

4

0.0

28.0

97

3.8

Mam

eyes

M11

0.67

11.1

1.05

2.21

12.9

13.6

0.0

10.6

194

122.

6

Mam

eyes

M12

0.67

15.0

0.47

1.00

15.3

7.2

0.

1

31

.8

69

98

.3

M

amey

esM

130.

034.

7

0.

80

1.

52

6.

3

5.

0

0.0

5.9

2,44

2

97.2

Mam

eyes

M14

0.03

5.4

0.45

0.84

5.8

2.6

0.

0

12

.1

1,

306

83

.7

M

amey

esM

150.

026.

0

0.

20

0.

33

6.

1

1.

3

0.0

29.4

297

33.8

Mam

eyes

M16

0.06

3.8

0.49

1.15

5.5

2.7

0.

0

7.

8

61

9

76

.5

M

amey

esM

170.

002.

8

0.

10

0.

19

2.

9

0.

3

0.0

28.2

300

6.5

Mam

eyes

M18

0.03

5.0

0.11

0.32

5.4

0.6

0.

0

43

.9

12

1

27

.3

M

amey

esM

190.

001.

8

0.

04

0.

17

1.

9

0.

1

0.0

44.6

79

2.7

Mam

eyes

M20

0.01

1.8

0.10

0.18

1.9

0.2

0.

0

17

.9

23

6

18

.4

M

amey

esM

210.

014.

7

0.

24

0.

56

5.

2

1.

2

0.0

19.6

518

26.0

Mam

eyes

M22

0.67

20.7

0.21

0.50

21.1

4.5

0.

1

98

.1

24

76

.0

M

amey

esM

230.

022.

9

0.

23

0.

50

3.

3

0.

8

0.0

12.8

479

39.4

Mam

eyes

M24

0.58

17.5

0.78

1.91

18.4

14.4

0.0

22.5

263

196.

8

Mam

eyes

M25

0.60

15.2

0.48

0.88

15.8

7.5

0.

1

31

.9

13

9

17

6.7

M

amey

esM

260.

5813

.9

0.

29

0.

77

14

.3

4.

1

0.1

48.6

125

254.

4

Mam

eyes

M27

0.01

2.5

0.04

0.10

2.5

0.1

0.

1

57

.0

10

7

16

.8

M

amey

esM

280.

139.

1

0.

30

0.

83

9.

6

2.

9

0.0

29.7

943

404.

9

Mam

eyes

M29

0.07

2.5

0.10

0.20

2.6

0.3

0.

3

24

.9

16

3

12

2.8

M

amey

esM

300.

233.

9

0.

21

0.

31

4.

0

0.

8

0.3

19.2

267

295.

1

Mam

eyes

M31

0.49

10.5

1.21

3.00

13.2

16.0

0.0

8.7

1,13

1

455.

9

Mam

eyes

M32

0.25

12.2

0.83

1.37

12.8

10.6

0.0

14.7

499

148.

4

Mam

eyes

M33

0.24

14.0

0.50

0.88

14.4

7.3

0.

0

27

.8

32

7

15

3.2

M

amey

esM

340.

107.

2

0.

34

0.

70

7.

9

2.

7

0.0

21.5

211

59.8

Mam

eyes

M35

0.14

8.9

0.53

0.99

9.5

5.0

0.

0

16

.9

24

4

64

.3

M

amey

esM

360.

047.

2

0.

12

0.

31

7.

4

0.

9

0.0

60.7

137

46.5

App

endi

x. B

asef

low

Ch

ann

el G

eom

etry

234

Wat

ersh

edBi

ocom

plex

ity I

DCr

oss-

Sect

ion

IDQ

bf

w b

f R

bf

dm

ax b

f P

bf

A b

f v

bf

w/d

bf

τ b

f Ω

bf

Mam

eye s

M37

0.05

5.8

0.70

1.11

6.8

4.8

0.

0

8.

2

1,

230

94

.5

M

amey

esM

380.

012.

6

0.

06

0.

14

2.

6

0.

1

0.0

45.8

121

13.3

Mam

eyes

M39

0.08

1.7

0.15

0.37

2.0

0.3

0.

3

11

.4

12

9

69

.4

M

amey

esM

400.

001.

0

0.

09

0.

19

1.

1

0.

1

0.0

11.5

203

2.7

Mam

eyes

M41

0.00

1.4

0.06

0.20

1.5

0.1

0.

0

23

.2

11

4

2.

9

M

amey

esM

420.

014.

2

0.

11

0.

28

4.

3

0.

5

0.0

39.8

170

16.7

Mam

eyes

M43

0.71

10.0

0.43

0.78

10.5

4.5

0.

2

23

.3

10

16

.7

Es

pirit

u Sa

nto

ES1

0.14

11.4

0.65

1.24

12.0

7.8

0.

0

17

.5

1,

329

28

6.6

Es

pirit

u Sa

nto

ES3

0.05

6.0

0.07

0.18

6.1

0.5

0.

1

80

.9

51

35

.4

Es

pirit

u Sa

nto

ES4

0.05

6.0

0.06

0.33

6.2

0.4

0.

1

94

.8

89

63

.6

Es

pirit

u Sa

nto

ES5

0.02

4.5

0.17

1.44

9.5

1.6

0.

0

27

.2

31

6

44

.7

Es

pirit

u Sa

nto

ES6

0.15

7.3

0.32

0.61

7.6

2.5

0.

1

22

.5

42

0

19

6.8

Es

pirit

u Sa

nto

ES7

0.69

22.0

0.21

0.45

22.1

4.6

0.

2

10

5.8

22

72

.3

Es

pirit

u Sa

nto

ES8

0.13

5.7

0.16

0.30

5.8

0.9

0.

1

35

.6

30

2

23

9.3

Es

pirit

u Sa

nto

ES9

0.11

3.9

0.14

0.46

4.3

0.6

0.

2

27

.2

31

6

24

4.6

Es

pirit

u Sa

nto

ES10

0.11

6.7

0.33

0.98

7.8

2.5

0.

0

20

.5

58

7

19

7.3

Es

pirit

u Sa

nto

ES11

0.68

10.2

0.40

0.84

13.3

5.3

0.

1

25

.6

48

82

.5

Es

pirit

u Sa

nto

ES12

0.68

---

---

---

---

---

---

---

---

---

Espi

ritu

Sant

oES

130.

104.

7

0.

08

0.

37

4.

9

0.

4

0.3

59.9

190

245.

0

Espi

ritu

Sant

oES

140.

5911

.6

0.

25

0.

56

12

.7

3.

2

0.2

46.4

5212

3.9

Es

pirit

u Sa

nto

ES15

0.59

5.2

0.15

0.53

5.6

0.8

0.

7

35

.1

56

223.

4

Espi

ritu

Sant

oES

160.

5817

.7

0.

27

0.

53

17

.9

4.

8

0.1

66.7

4189

.1

Es

pirit

u Sa

nto

ES17

0.35

7.2

0.29

0.83

8.2

2.3

0.

1

25

.1

32

438

9.2

Es

pirit

u Sa

nto

ES18

0.01

4.2

0.27

0.53

4.6

1.2

0.

0

15

.8

20

310

.1

Es

pirit

u Sa

nto

ES19

0.02

3.0

0.11

0.21

3.1

0.3

0.

0

27

.7

15

423

.3

Es

pirit

u Sa

nto

ES20

0.32

15.6

0.23

0.42

16.1

3.7

0.

1

67

.8

91

126.

4

Espi

ritu

Sant

oES

210.

2911

.1

0.

34

0.

65

11

.2

3.

8

0.1

32.9

100

85.9

Espi

ritu

Sant

oES

220.

148.

2

0.

54

0.

99

9.

2

5.

0

0.0

15.1

557

143.

3

Espi

ritu

Sant

oES

231.

5736

.1

1.

45

2.

49

36

.7

53

.1

0.

0

24

.9

17

18.1

Espi

ritu

Sant

oES

241.

4850

.0

2.

37

3.

40

50

.5

11

9.6

0.

0

21

.1

27

17.0

Espi

ritu

Sant

oES

250.

173.

6

0.

10

0.

17

3.

6

0.

4

0.5

35.9

2949

.3

Es

pirit

u Sa

nto

ES26

0.15

2.1

0.14

0.20

2.1

0.3

0.

5

14

.8

43

44.6

Espi

ritu

Sant

oES

270.

071.

1

0.

16

0.

27

1.

2

0.

2

0.4

6.5

5926

.8

Es

pirit

u Sa

nto

ES28

0.14

7.7

0.80

1.69

9.3

7.4

0.

0

9.

6

87

215

5.0

Es

pirit

u Sa

nto

ES29

0.35

14.5

0.55

2.38

18.6

10.3

0.0

26.2

1233

776.

4

Espi

ritu

Sant

oES

300.

5915

.2

0.

47

1.

13

16

.1

7.

6

0.1

32.1

7796

.3

Es

pirit

u Sa

nto

ES31

0.69

17.7

0.54

1.08

18.5

10.0

0.1

32.6

4760

.2

235

Wat

ersh

edBi

ocom

plex

ity I

DCr

oss-

Sect

ion

IDQ

bf

w b

f R

bf

dm

ax b

f P

bf

A b

f v

bf

w/d

bf

τ b

f Ω

bf

Espi

ritu

Sant

oES

320.

9114

.5

0.

41

0.

88

14

.9

6.

2

0.1

34.9

4496

.9

Es

pirit

u Sa

nto

ES33

0.56

13.0

0.32

0.70

13.4

4.3

0.

1

40

.6

10

518

4.3

Es

pirit

u Sa

nto

ES34

0.54

11.7

0.29

0.73

12.0

3.5

0.

2

40

.7

56

41,

056.

9

Espi

ritu

Sant

oES

350.

9024

.0

1.

60

2.

82

25

.1

40

.1

0.

0

15

.0

18

10.4

Espi

ritu

Sant

oES

360.

9016

.6

0.

66

1.

33

17

.0

11

.3

0.

1

25

.0

8

10.4

Espi

ritu

Sant

oES

370.

248.

3

0.

58

1.

11

8.

7

5.

0

0.0

14.3

956

395.

7

Espi

ritu

Sant

oES

380.

267.

0

0.

47

0.

91

7.

3

3.

4

0.1

15.0

343

191.

7

Blan

coB1

0.16

5.3

0.21

0.65

5.8

1.2

0.

1

25

.1

18

13.6

Blan

coB2

0.15

5.6

0.16

0.23

5.7

0.9

0.

2

35

.7

13

13.1

Blan

coB3

0.13

3.0

0.30

0.51

3.2

1.0

0.

1

9.

8

26

11.5

Blan

coB4

0.09

4.1

0.18

0.31

4.2

0.8

0.

1

22

.8

15

8.1

Blan

coB5

0.20

5.2

0.12

0.27

5.5

0.7

0.

3

43

.8

13

021

3.9

Bl

anco

B60.

011.

6

0.

10

0.

17

1.

7

0.

2

0.0

16.2

251.

9

Bl

anco

B70.

217.

2

0.

08

0.

25

7.

3

0.

6

0.3

86.9

5112

8.5

Bl

anco

B80.

208.

9

0.

15

0.

39

9.

0

1.

4

0.1

58.6

186

250.

1

Blan

coB9

0.08

3.0

0.09

0.16

3.0

0.3

0.

3

32

.0

90

76.0

Blan

coB1

01.

1125

.7

0.

32

0.

71

26

.0

8.

3

0.1

80.8

3512

2.6

Bl

anco

B11

0.99

12.3

1.06

2.14

14.8

15.7

0.1

11.6

1256

1,17

8.5

Sa

bana

S10.

1716

.3

0.

14

0.

32

16

.3

2.

3

0.1

114.

8

4249

.8

Sa

bana

S20.

173.

3

0.

20

0.

43

3.

4

0.

7

0.2

16.0

8367

.4

Sa

bana

S30.

0811

.1

0.

22

0.

76

11

.9

2.

6

0.0

51.1

191

70.2

Saba

naS4

0.09

9.9

0.14

0.34

10.0

1.5

0.

1

68

.4

10

662

.3

Sa

bana

S50.

1618

.6

0.

27

0.

86

19

.6

5.

3

0.0

68.9

143

87.5

Saba

naS6

0.07

11.0

0.36

0.88

11.6

4.1

0.

0

30

.9

23

448

.7

Sa

bana

S70.

073.

2

0.

32

0.

50

3.

4

1.

1

0.1

10.0

302

67.4

Saba

naS8

0.02

5.3

0.10

0.22

5.3

0.5

0.

0

52

.7

17

3.4

Saba

naS9

0.01

2.5

0.17

0.33

2.7

0.5

0.

0

14

.3

50

3.7

Saba

naS1

00.

208.

1

0.

75

1.

58

9.

4

7.

1

0.0

10.8

9425

.5

Sa

bana

S11

0.20

8.9

0.41

0.69

9.1

3.7

0.

1

21

.7

45

22.0

Saba

naS1

20.

001.

2

0.

02

0.

02

1.

2

0.

0

0.0

81.5

160.

7

Sa

bana

S13

0.21

4.6

0.48

0.99

5.8

2.8

0.

1

9.

6

83

36.1

Saba

naS1

40.

2113

.2

0.

26

0.

71

13

.8

3.

5

0.1

51.4

193

157.

9

Saba

naS1

50.

2115

.6

0.

20

0.

55

16

.1

3.

2

0.1

78.9

2223

.8

Sa

bana

S16

0.04

3.8

0.18

0.29

3.9

0.7

0.

1

20

.9

56

12.3

Saba

naS1

70.

023.

4

0.

16

0.

20

3.

5

0.

5

0.0

21.4

263.

1

Fa

jard

o19

4 Br

idge

1U0.

6412

.9

1.

36

3.

13

14

.6

19

.8

0.

0

9.

5

25

11.6

Faja

rdo

194

Brid

geBr

idge

0.63

9.6

0.21

0.39

9.8

2.1

0.

3

45

.2

4

11.4

236

Wat

ersh

edBi

ocom

plex

ity I

DCr

oss-

Sect

ion

IDQ

bf

w b

f R

bf

dm

ax b

f P

bf

A b

f v

bf

w/d

bf

τ b

f Ω

bf

Faja

rdo

194

Brid

ge1D

0.63

14.5

0.54

0.90

14.8

7.9

0.

1

27

.1

10

11.4

Faja

rdo

977

Brid

ge2U

0.62

8.6

0.68

1.20

9.1

6.2

0.

1

12

.6

48

43.7

Faja

rdo

977

Brid

ge1U

0.62

12.6

0.49

0.89

12.8

6.3

0.

1

25

.6

35

43.6

Faja

rdo

977

Brid

geBr

idge

0.62

20.9

0.14

0.31

21.0

2.9

0.

2

14

9.5

10

43.7

Faja

rdo

977

Brid

ge1D

0.62

23.4

0.14

0.36

23.5

3.3

0.

2

16

8.0

10

43.7

Faja

rdo

Char

co F

rio1D

0.38

11.2

0.57

0.89

11.8

6.7

0.

1

19

.6

94

62.4

Faja

rdo

Char

co F

rioBr

idge

0.38

10.5

0.21

0.34

10.8

2.2

0.

2

50

.7

25

45.6

Faja

rdo

Char

co F

rio1U

0.38

12.2

0.72

1.20

12.6

9.1

0.

0

16

.9

87

45.6

Faja

rdo

Char

co F

rio2U

0.38

12.0

0.54

1.05

12.3

6.6

0.

1

22

.4

64

45.5

Faja

rdo

Char

co F

rio3U

0.38

6.1

0.77

1.59

7.5

5.7

0.

1

7.

9

92

45.5

Faja

rdo

Char

co F

rio4U

0.38

17.9

1.00

2.75

19.7

19.7

0.0

18.0

120

45.5

Faja

rdo

Char

co F

rio5U

0.38

11.3

0.80

1.55

11.9

9.5

0.

0

14

.1

15

874

.8

Fa

jard

oCh

arco

Frio

7U0.

3810

.0

0.

44

1.

01

10

.3

4.

5

0.1

22.7

8774

.7

Fa

jard

oLa

Tin

aja

3U0.

3611

.8

0.

52

1.

32

13

.0

6.

8

0.1

22.5

310

214.

7

Faja

rdo

La T

inaj

a2U

0.37

9.0

1.66

3.03

13.4

22.2

0.0

5.4

973

215.

4

Faja

rdo

La T

inaj

a1U

0.37

11.8

0.91

1.63

12.9

11.7

0.0

13.0

300

121.

7

Faja

rdo

Riv

er B

ar2U

0.03

2.8

0.28

0.57

3.4

1.0

0.

0

9.

7

80

9.4

Faja

rdo

Riv

er B

ar1U

0.03

8.7

0.58

0.95

9.6

5.6

0.

0

14

.9

16

59.

4

Fa

jard

oRiv

er B

ar1D

0.03

9.4

0.50

0.95

10.7

5.3

0.

0

18

.8

14

09.

4

Fa

jard

oRiv

er B

ar2D

0.03

6.2

0.21

0.54

6.5

1.4

0.

0

29

.3

60

9.4

Faja

rdo

Smal

l Trib

utar

y2U

0.00

3.0

0.29

0.51

3.2

0.9

0.

0

10

.4

32

83.

8

Fa

jard

oSm

all T

ribut

ary

1U0.

003.

2

0.

23

0.

46

3.

4

0.

8

0.0

13.9

261

3.8

Faja

rdo

Smal

l Trib

utar

y1D

0.00

2.6

0.29

0.58

3.5

1.0

0.

0

8.

8

33

03.

8

Fa

jard

oSm

all T

ribut

ary

2D0.

004.

8

0.

43

0.

85

5.

2

2.

3

0.0

11.1

490

3.8

Faja

rdo

Upp

er F

ajar

doTr

ib2U

0.02

1.6

0.40

0.89

2.9

1.1

0.

0

4.

0

72

227

.6

Fa

jard

oU

pper

Faj

ardo

Trib

1U0.

022.

0

0.

44

0.

98

3.

5

1.

5

0.0

4.5

805

27.6

Faja

rdo

Upp

er F

ajar

do3U

0.22

7.0

0.64

1.33

7.8

5.0

0.

0

10

.9

15

552

.8

Fa

jard

oU

pper

Faj

ardo

2U0.

229.

7

0.

87

1.

77

11

.7

10

.2

0.

0

11

.1

16

542

.0

Fa

jard

oU

pper

Faj

ardo

1U0.

2210

.2

0.

97

1.

47

11

.7

11

.4

0.

0

10

.5

18

442

.5

M

amey

esAn

gelit

o Tr

ail

1U0.

5816

.0

0.

50

1.

04

16

.3

8.

1

0.1

32.2

347

407.

6

Mam

eyes

Ange

lito

Trai

l1D

0.58

20.0

1.54

3.41

22.4

34.6

0.0

12.9

1079

407.

6

Mam

eyes

Ange

lito

Trai

l2D

0.60

15.2

0.48

0.88

15.8

7.5

0.

1

31

.9

13

917

6.7

Es

pirit

u Sa

nto

Back

way

Sch

ool

2U0.

6912

.7

0.

36

0.

52

13

.1

4.

7

0.1

35.4

2445

.7

Es

pirit

u Sa

nto

Back

way

Sch

ool

1U0.

6911

.7

0.

39

0.

63

12

.2

4.

7

0.1

30.4

2545

.7

Es

pirit

u Sa

nto

Back

way

Sch

ool

1D0.

6910

.0

0.

50

0.

81

10

.2

5.

1

0.1

20.1

3345

.7

Es

pirit

u Sa

nto

Back

way

Sch

ool

2D0.

698.

0

0.

39

0.

72

8.

3

3.

2

0.2

20.5

2645

.7

Es

pirit

u Sa

nto

Behi

nd B

risas

2U0.

6810

.7

0.

47

0.

74

10

.9

5.

1

0.1

22.9

57.

9

237

Wat

ersh

edBi

ocom

plex

ity I

DCr

oss-

Sect

ion

IDQ

bf

w b

f R

bf

dm

ax b

f P

bf

A b

f v

bf

w/d

bf

τ b

f Ω

bf

Espi

ritu

Sant

oBe

hind

Bris

a s1U

0.68

7.5

0.41

0.74

7.8

3.2

0.

2

18

.5

5

7.9

Espi

ritu

Sant

oBe

hind

Bris

asBR

IDG

E0.

6814

.8

0.

66

1.

32

15

.6

10

.4

0.

1

22

.4

8

7.9

Mam

eyes

Bisl

ey 3

2U0.

014.

1

0.

30

0.

53

4.

5

1.

3

0.0

13.8

507

17.6

Mam

eyes

Bisl

ey 3

1U0.

014.

2

0.

32

0.

53

4.

6

1.

5

0.0

12.9

554

17.7

Mam

eyes

Bisl

ey 3

1D0.

015.

4

0.

31

0.

77

6.

5

2.

0

0.0

17.7

522

18.5

Mam

eyes

Bisl

ey 3

2D0.

012.

0

0.

30

0.

58

2.

5

0.

8

0.0

6.6

516

18.6

Espi

ritu

Sant

oD

avila

3U0.

105.

6

0.

41

0.

75

6.

4

2.

6

0.0

13.6

4812

.3

Es

pirit

u Sa

nto

Dav

ila2U

0.10

6.4

0.18

0.32

6.5

1.2

0.

1

36

.0

21

12.2

Espi

ritu

Sant

oD

avila

1U0.

107.

0

0.

19

0.

26

7.

1

1.

3

0.1

36.8

2212

.2

Es

pirit

u Sa

nto

Dav

ila1D

0.10

5.4

0.21

0.37

5.5

1.1

0.

1

26

.2

24

12.2

Espi

ritu

Sant

oD

avila

2D0.

107.

3

0.

18

0.

27

7.

4

1.

3

0.1

41.2

2112

.2

M

amey

esEa

st P

eak

2U0.

013.

1

0.

23

0.

44

3.

4

0.

8

0.0

13.8

194

12.9

Mam

eyes

East

Pea

k1U

0.02

2.9

0.34

0.69

3.3

1.1

0.

0

8.

5

43

320

.4

M

amey

esEa

st P

eak

1D0.

021.

5

0.

16

0.

30

1.

7

0.

3

0.1

9.5

217

22.5

Mam

eyes

East

Pea

k2D

0.02

4.3

0.56

1.21

6.6

3.7

0.

0

7.

7

77

722

.5

Es

pirit

u Sa

nto

El V

erde

3U0.

599.

3

0.

30

1.

02

8.

7

2.

6

0.2

30.7

6312

3.9

Es

pirit

u Sa

nto

El V

erde

2U0.

5911

.6

0.

68

1.

53

13

.1

8.

9

0.1

17.0

143

123.

9

Espi

ritu

Sant

oEl

Ver

de1U

0.59

19.1

0.32

0.72

19.6

6.4

0.

1

58

.8

68

123.

9

Espi

ritu

Sant

oEl

Ver

de1D

0.59

24.3

0.38

0.81

24.0

9.0

0.

1

64

.7

79

123.

9

Espi

ritu

Sant

oEl

Ver

de2D

0.59

14.7

0.64

1.31

11.1

7.2

0.

1

22

.9

13

512

3.9

Es

pirit

u Sa

nto

ES R

t. 3

Pai

ntba

llBR

IDG

E0.

9025

.0

0.

52

0.

99

25

.3

13

.2

0.

1

48

.1

6

10.5

Espi

ritu

Sant

oES

Wat

erfa

ll3U

0.35

6.3

0.50

0.90

7.6

3.8

0.

1

12

.6

14

481,

000.

9

Espi

ritu

Sant

oES

Wat

erfa

ll2U

0.35

6.1

0.90

1.61

7.1

6.4

0.

1

6.

8

25

941,

001.

2

Espi

ritu

Sant

oES

Wat

erfa

ll1U

0.35

14.1

1.03

2.00

16.4

16.9

0.0

13.7

4581

1,54

8.7

Es

pirit

u Sa

nto

ES W

ater

fall

1D0.

3514

.5

0.

57

2.

35

17

.2

9.

8

0.0

25.4

1272

776.

4

Espi

ritu

Sant

oJi

men

ez2U

0.69

13.0

0.37

0.79

13.3

4.9

0.

1

35

.5

39

74.5

Espi

ritu

Sant

oJi

men

ez1U

0.69

22.9

0.25

0.50

21.3

5.3

0.

1

92

.8

27

74.5

Espi

ritu

Sant

oJi

men

ez1D

0.69

30.9

0.53

1.18

30.1

15.9

0.0

58.4

5774

.5

Es

pirit

u Sa

nto

Jim

enez

Wat

erfa

l2U

0.11

5.8

0.23

0.53

6.1

1.4

0.

1

25

.4

25

012

1.2

Es

pirit

u Sa

nto

Jim

enez

Wat

erfa

l1U

0.11

4.7

0.49

0.79

5.1

2.5

0.

0

9.

7

43

498

.7

Es

pirit

u Sa

nto

Jim

enez

Wat

erfa

l1D

0.11

3.3

0.15

0.45

3.7

0.6

0.

2

21

.5

37

727

2.2

Es

pirit

u Sa

nto

Jim

enez

Wat

erfa

l2D

0.11

10.4

0.82

2.60

12.9

10.6

0.0

12.7

2175

296.

4

Mam

eyes

Juan

Die

go3U

0.03

9.0

0.38

0.69

9.2

3.5

0.

0

23

.9

13

6611

4.1

M

amey

esJu

an D

iego

2U0.

0311

.0

0.

59

0.

98

11

.4

6.

7

0.0

18.6

2150

114.

8

Mam

eyes

Juan

Die

go1U

0.06

6.4

0.35

0.55

7.0

2.5

0.

0

18

.1

63

098

.6

M

amey

esJu

an D

iego

BRID

GE

0.06

4.4

0.34

0.53

4.6

1.6

0.

0

13

.0

60

498

.7

M

amey

esJu

an D

iego

1D0.

064.

1

0.

25

0.

43

4.

4

1.

1

0.1

16.6

441

98.9

238

Wat

ersh

edBi

ocom

plex

ity I

DCr

oss-

Sect

ion

IDQ

bf

w b

f R

bf

dm

ax b

f P

bf

A b

f v

bf

w/d

bf

τ b

f Ω

bf

Mam

eye s

La C

oca

Trai

l2U

0.03

6.0

0.37

0.64

6.7

2.5

0.

0

16

.3

51

639

.5

M

amey

esLa

Coc

a Tr

ail

1U0.

033.

5

0.

33

0.

61

4.

1

1.

3

0.0

10.5

479

40.9

Mam

eyes

La C

oca

Trai

l1D

0.03

5.0

0.36

0.83

6.2

2.3

0.

0

13

.9

76

960

.9

M

amey

esLa

Coc

a Tr

ail

2D0.

034.

5

0.

24

0.

38

4.

8

1.

2

0.0

18.7

370

44.1

Mam

eyes

La L

ingu

ente

2U0.

016.

3

0.

88

1.

46

8.

3

7.

3

0.0

7.2

1265

16.9

Mam

eyes

La L

ingu

ente

1U0.

013.

9

0.

30

0.

57

4.

1

1.

2

0.0

13.2

424

17.2

Mam

eyes

La L

ingu

ente

1D0.

012.

2

0.

31

0.

55

2.

9

0.

9

0.0

7.0

451

17.6

Mam

eyes

La L

ingu

ente

2D0.

013.

9

0.

22

0.

48

4.

1

0.

9

0.0

17.7

316

17.8

Mam

eyes

La M

aqui

na2U

0.03

2.1

0.51

1.02

3.7

1.9

0.

0

4.

1

81

241

.0

M

amey

esLa

Maq

uina

1U0.

032.

4

0.

13

0.

25

2.

5

0.

3

0.1

18.1

211

40.8

Mam

eyes

La M

aqui

naBR

IDG

E0.

034.

0

0.

10

0.

14

4.

0

0.

4

0.1

40.1

159

40.9

Mam

eyes

La M

aqui

na1D

0.03

2.1

0.13

0.51

2.8

0.4

0.

1

16

.6

20

141

.0

M

amey

esLa

Maq

uina

2D0.

032.

4

0.

12

0.

20

2.

5

0.

3

0.1

19.1

199

44.2

Mam

eyes

La M

ina

Falls

2U0.

138.

1

0.

36

0.

67

8.

4

3.

1

0.0

22.2

764

263.

7

Mam

eyes

La M

ina

Falls

1U0.

1310

.1

0.

65

1.

02

11

.4

7.

5

0.0

15.5

1367

263.

2

Mam

eyes

La M

ina

Falls

1D0.

135.

4

0.

44

0.

83

5.

9

2.

6

0.0

12.3

922

265.

7

Mam

eyes

La V

ega

2U (

new

)0.

6714

.1

0.

61

0.

92

14

.4

8.

7

0.1

23.3

4650

.2

M

amey

esLa

Veg

a2U

(ol

d)0.

6717

.2

0.

34

0.

55

17

.3

5.

9

0.1

50.1

2650

.2

M

amey

esLa

Veg

a1U

0.67

16.7

0.48

0.93

17.4

8.4

0.

1

34

.5

36

50.2

Mam

eyes

La V

ega

RO

AD0.

6711

.2

0.

98

1.

87

12

.5

12

.2

0.

1

11

.4

74

50.1

Mam

eyes

La V

ega

1D0.

6711

.7

0.

41

0.

74

12

.0

4.

9

0.1

28.9

3050

.1

M

amey

esLa

Veg

a2D

0.67

16.9

0.53

0.73

17.1

9.0

0.

1

31

.9

40

50.1

Espi

ritu

Sant

oLa

s Tr

es T

(1)

2U0.

013.

8

0.

13

0.

26

4.

2

0.

5

0.0

30.2

524.

9

Es

pirit

u Sa

nto

Las

Tres

T (

1)1U

0.01

4.6

0.20

0.36

4.7

1.0

0.

0

22

.6

84

4.9

Espi

ritu

Sant

oLa

s Tr

es T

(1)

1D0.

018.

7

0.

66

1.

22

9.

4

6.

2

0.0

13.1

276

4.9

Espi

ritu

Sant

oLa

s Tr

es T

(1)

2D0.

012.

9

0.

15

0.

25

3.

2

0.

5

0.0

19.1

635.

0

Es

pirit

u Sa

nto

Las

Tres

T (

2)2U

0.01

3.2

0.35

0.64

3.7

1.3

0.

0

9.

1

22

47.

2

Es

pirit

u Sa

nto

Las

Tres

T (

2)1U

0.01

3.8

0.14

0.25

3.8

0.5

0.

0

26

.9

90

7.3

Espi

ritu

Sant

oLa

s Tr

es T

(2)

1D0.

017.

0

0.

41

1.

17

7.

7

3.

2

0.0

17.0

263

7.3

Espi

ritu

Sant

oLa

s Tr

es T

(2)

2D0.

012.

9

0.

31

0.

64

3.

3

1.

0

0.0

9.3

199

7.3

Mam

eyes

Mam

eyes

Rt.

31U

0.71

14.9

0.83

1.61

15.4

12.7

0.1

18.0

1916

.8

M

amey

esM

amey

es R

t. 3

1D0.

7116

.0

1.

09

1.

78

16

.7

18

.2

0.

0

14

.7

26

16.8

Espi

ritu

Sant

oPr

ieta

3U0.

022.

5

0.

27

1.

08

3.

9

1.

1

0.0

9.0

436

24.1

Espi

ritu

Sant

oPr

ieta

2U0.

021.

8

0.

14

0.

21

2.

1

0.

3

0.1

13.1

219

26.2

Espi

ritu

Sant

oPr

ieta

1U0.

022.

1

0.

26

0.

46

2.

6

0.

7

0.0

8.0

419

26.2

Espi

ritu

Sant

oPr

ieta

1D0.

027.

1

0.

67

1.

08

8.

5

5.

7

0.0

10.6

1074

26.8

Espi

ritu

Sant

oPr

ieta

2D0.

022.

0

0.

25

0.

43

2.

3

0.

6

0.0

7.9

406

26.8

239

Wat

ersh

edBi

ocom

plex

ity I

DCr

oss-

Sect

ion

IDQ

bf

w b

f R

bf

dm

ax b

f P

bf

A b

f v

bf

w/d

bf

τ b

f Ω

bf

Mam

eye s

Puen

te R

oto

2U0.

6110

.1

0.

65

1.

35

10

.9

7.

1

0.1

15.6

9992

.9

M

amey

esPu

ente

Rot

o1U

/BRID

GE

0.61

19.3

0.62

1.12

19.6

12.1

0.1

31.2

9492

.9

M

amey

esPu

ente

Rot

o1D

0.61

6.9

0.62

1.08

7.6

4.7

0.

1

11

.1

95

92.9

Mam

eyes

Puen

te R

oto

2D0.

6116

.2

0.

64

1.

25

16

.8

10

.8

0.

1

25

.2

98

93.0

Mam

eyes

Q. A

non

2U0.

002.

2

0.

26

0.

49

2.

6

0.

7

0.0

8.4

290.

1

M

amey

esQ

. Ano

n1U

/BRID

GE

0.00

2.9

0.33

0.44

3.4

1.1

0.

0

8.

9

36

0.1

Mam

eyes

Q. A

non

1D0.

005.

9

0.

77

1.

21

6.

8

5.

2

0.0

7.7

840.

1

M

amey

esQ

. Ano

n2D

0.00

2.7

0.36

0.73

3.4

1.2

0.

0

7.

4

39

0.1

Espi

ritu

Sant

oSo

nado

ra3U

0.14

6.7

0.68

1.35

8.6

5.8

0.

0

9.

9

19

0039

7.7

Es

pirit

u Sa

nto

Sona

dora

2U0.

1414

.3

0.

79

1.

44

16

.0

12

.6

0.

0

18

.2

22

0739

8.0

Es

pirit

u Sa

nto

Sona

dora

1U0.

147.

9

0.

74

1.

69

10

.2

7.

5

0.0

10.8

804

155.

0

Espi

ritu

Sant

oSo

nado

ra1D

0.15

6.5

0.32

0.77

7.1

2.2

0.

1

20

.5

44

721

3.9

Es

pirit

u Sa

nto

Sona

dora

2D0.

157.

9

0.

27

0.

58

8.

2

2.

2

0.1

29.5

377

214.

1

Espi

ritu

Sant

oTo

ronj

a2U

0.00

1.6

0.28

0.49

2.2

0.6

0.

0

5.

7

36

42.

4

Es

pirit

u Sa

nto

Toro

nja

1U0.

001.

7

0.

06

0.

11

1.

7

0.

1

0.0

28.8

762.

5

Es

pirit

u Sa

nto

Toro

nja

1D0.

003.

2

0.

34

0.

60

3.

6

1.

2

0.0

9.4

438

2.6

Espi

ritu

Sant

oTo

ronj

a2D

0.00

4.2

0.15

0.27

4.4

0.6

0.

0

28

.6

19

02.

7

240

Wat

ersh

edBi

ocom

plex

ity I

DCr

oss-

Sect

ion

IDQ

ac

w a

c R

ac

dm

ax a

c P

ac

A a

c v

ac

w/d

ac

τ a

c Ω

ac

Mam

eye s

M1

62.1

323

.5

0.

96

2.

04

25

.4

24

.4

2.

6

11

.5

23

1,

459

M

amey

esM

259

.59

23.2

2.35

3.73

25.8

60.6

1.0

6.2

55

1,39

9

Mam

eyes

M3

52.5

929

.7

0.

96

1.

34

30

.3

29

.1

1.

8

22

.2

73

3,

984

M

amey

esM

448

.85

32.3

0.69

1.70

33.0

22.8

2.1

19.0

111

7,88

2

Mam

eyes

M5

48.2

626

.1

0.

93

1.

85

27

.1

25

.1

1.

9

14

.1

14

5

7,

539

M

amey

esM

644

.81

20.6

1.12

1.75

21.4

24.0

1.9

11.8

162

6,45

6

Mam

eyes

M7

52.3

328

.0

0.

80

1.

10

28

.5

22

.9

2.

3

25

.5

61

3,

964

M

amey

esM

852

.20

29.0

0.99

1.51

30.2

29.9

1.7

19.2

75

3,95

4

Mam

eyes

M9

49.5

224

.7

1.

04

2.

02

25

.7

26

.8

1.

8

12

.3

17

8

8,

493

M

amey

esM

100.

604.

0

0.

42

0.

61

4.

5

1.

9

0.3

6.6

343

497

Mam

eyes

M11

50.7

620

.0

1.

07

2.

90

22

.4

24

.0

2.

1

6.

9

19

7

9,

360

M

amey

esM

1251

.67

24.4

0.88

1.79

25.2

22.2

2.3

13.6

130

7,59

9

Mam

eyes

M13

2.07

5.7

0.98

2.02

7.8

7.6

0.

3

2.

8

3,

002

6,

357

M

amey

esM

141.

868.

5

0.

80

1.

54

9.

4

7.

5

0.2

5.5

2,33

7

5,43

6

Mam

eyes

M15

1.50

8.0

0.48

0.71

8.3

3.9

0.

4

11

.3

69

0

2,

176

M

amey

esM

164.

025.

6

0.

84

1.

99

8.

0

6.

8

0.6

2.8

1,06

3

5,08

8

Mam

eyes

M17

0.20

3.8

0.43

0.66

4.3

1.9

0.

1

5.

8

1,

308

61

4

M

amey

esM

181.

936.

8

0.

27

0.

58

7.

6

2.

1

0.9

11.7

287

2,05

1

Mam

eyes

M19

0.11

4.1

0.25

0.47

4.3

1.1

0.

1

8.

7

49

2

22

0

M

amey

esM

200.

576.

1

0.

45

0.

78

7.

0

3.

1

0.2

7.9

1,04

3

1,32

8

Mam

eyes

M21

0.84

5.6

0.56

1.06

6.8

3.8

0.

2

5.

3

1,

219

1,

821

M

amey

esM

2251

.88

22.4

0.58

0.95

23.3

13.5

3.9

23.6

66

5,89

7

Mam

eyes

M23

1.64

5.7

0.48

1.12

6.5

3.1

0.

5

5.

1

99

8

3,

429

M

amey

esM

2441

.93

26.8

1.20

2.75

28.3

33.9

1.2

9.7

404

14,1

35

M

amey

esM

2543

.70

18.9

0.98

1.58

19.8

19.4

2.3

12.0

287

12,7

98

M

amey

esM

2641

.87

15.9

0.93

1.60

17.3

16.1

2.6

9.9

405

18,2

57

M

amey

esM

270.

474.

7

0.

47

0.

71

5.

2

2.

4

0.2

6.6

1,14

8

1,16

3

Mam

eyes

M28

8.60

11.5

1.22

2.12

13.7

16.7

0.5

5.4

3,77

9

26,6

29

M

amey

esM

295.

087.

4

0.

56

0.

94

8.

7

4.

9

1.0

7.9

926

8,36

9

Mam

eyes

M30

15.2

915

.7

0.

79

1.

41

16

.9

13

.4

1.

1

11

.1

1,

024

19

,865

Mam

eyes

M31

34.0

724

.8

1.

46

4.

40

30

.9

45

.1

0.

8

5.

6

1,

363

31

,833

Mam

eyes

M32

17.5

416

.0

0.

95

1.

76

16

.8

15

.9

1.

1

9.

1

57

0

10

,559

Mam

eyes

M33

16.0

318

.4

0.

66

1.

19

18

.8

12

.5

1.

3

15

.4

42

9

10

,417

Mam

eyes

M34

6.53

8.4

0.90

1.51

10.0

9.0

0.

7

5.

6

56

5

4,

083

M

amey

esM

359.

4611

.2

0.

80

1.

50

12

.3

9.

9

1.0

7.5

373

4,39

8

Mam

eyes

M36

2.76

8.1

0.63

0.94

9.1

5.7

0.

5

8.

6

72

3

3,

181

App

endi

x. A

ctiv

e C

han

nel

Geo

met

ry

241

Wat

ersh

edBi

ocom

plex

ity I

DCr

oss-

Sect

ion

IDQ

ac

w a

c R

ac

dm

ax a

c P

ac

A a

c v

ac

w/d

ac

τ a

c Ω

ac

Mam

eye s

M37

3.59

7.0

0.90

1.57

8.5

7.6

0.

5

4.

5

1,

570

6,

285

M

amey

esM

380.

435.

2

0.

54

0.

93

5.

9

3.

2

0.1

5.6

1,17

0

921

Mam

eyes

M39

5.31

2.4

0.34

0.75

3.1

1.1

5.

0

3.

2

30

1

4,

695

M

amey

esM

400.

102.

4

0.

39

0.

79

3.

0

1.

2

0.1

3.0

915

227

Mam

eyes

M41

0.13

2.7

0.31

0.59

3.0

0.9

0.

1

4.

6

60

1

24

5

M

amey

esM

420.

786.

2

0.

48

0.

89

7.

0

3.

3

0.2

7.0

768

1,25

9

Mam

eyes

M43

62.3

114

.8

0.

76

1.

36

15

.6

11

.8

5.

3

10

.9

18

1,

463

Es

pirit

u Sa

nto

ES1

9.27

13.1

1.06

1.87

14.5

15.4

0.6

7.0

2,17

3

18,9

47

Es

pirit

u Sa

nto

ES3

3.33

11.0

0.21

0.47

11.8

2.5

1.

4

23

.4

14

5

2,

309

Es

pirit

u Sa

nto

ES4

2.91

11.3

0.36

0.94

13.0

4.6

0.

6

12

.1

50

3

4,

112

Es

pirit

u Sa

nto

ES5

1.49

6.5

0.84

2.48

12.2

10.2

0.1

2.6

1,60

0

2,83

9

Espi

ritu

Sant

oES

610

.18

12.0

0.64

1.20

12.6

8.0

1.

3

10

.0

82

8

13

,187

Espi

ritu

Sant

oES

752

.37

24.5

0.68

1.00

25.1

17.1

3.1

24.4

71

5,46

6

Espi

ritu

Sant

oES

88.

296.

6

0.

49

0.

70

7.

0

3.

4

2.4

9.4

924

15,6

48

Es

pirit

u Sa

nto

ES9

7.48

12.7

0.89

1.77

14.4

12.8

0.6

7.2

1,96

0

16,4

99

Es

pirit

u Sa

nto

ES10

7.37

10.0

0.80

2.00

12.8

10.2

0.7

5.0

1,44

1

13,2

67

Es

pirit

u Sa

nto

ES11

49.7

214

.5

0.

75

1.

77

22

.6

16

.9

2.

9

8.

2

91

6,

030

Es

pirit

u Sa

nto

ES12

49.3

824

.0

0.

82

1.

51

25

.8

21

.2

2.

3

15

.9

21

1

12

,682

Espi

ritu

Sant

oES

136.

657.

5

0.

35

0.

97

10

.2

3.

5

1.9

7.7

837

16,0

97

Es

pirit

u Sa

nto

ES14

42.3

419

.5

0.

61

1.

10

20

.9

12

.8

3.

3

17

.8

12

8

8,

875

Es

pirit

u Sa

nto

ES15

41.9

110

.0

0.

35

1.

05

11

.4

4.

0

10.4

9.6

133

15,8

74

Es

pirit

u Sa

nto

ES16

40.1

225

.5

0.

52

1.

15

26

.6

14

.0

2.

9

22

.1

81

6,

197

Es

pirit

u Sa

nto

ES17

22.8

112

.6

0.

51

1.

25

14

.2

7.

2

3.2

10.1

575

25,7

17

Es

pirit

u Sa

nto

ES18

0.97

6.0

0.74

1.39

7.3

5.4

0.

2

4.

3

56

7

74

0

Es

pirit

u Sa

nto

ES19

1.23

4.7

0.56

0.99

5.6

3.2

0.

4

4.

7

80

2

1,

756

Es

pirit

u Sa

nto

ES20

21.0

718

.0

0.

73

1.

04

18

.8

13

.8

1.

5

17

.4

28

9

8,

302

Es

pirit

u Sa

nto

ES21

19.0

312

.3

0.

61

1.

26

13

.8

8.

3

2.3

9.8

179

5,62

4

Espi

ritu

Sant

oES

229.

2713

.2

0.

75

1.

77

15

.1

11

.3

0.

8

7.

5

76

5

9,

473

Es

pirit

u Sa

nto

ES23

128.

3336

.1

1.

45

2.

49

36

.7

53

.1

2.

4

14

.5

17

1,

474

Es

pirit

u Sa

nto

ES24

145.

2750

.0

2.

37

3.

40

50

.5

11

9.6

1.

2

14

.7

27

1,

669

Es

pirit

u Sa

nto

ES25

11.1

36.

5

0.

44

0.

74

7.

2

3.

2

3.5

8.8

126

3,19

3

Espi

ritu

Sant

oES

269.

444.

0

0.

41

0.

67

4.

5

1.

8

5.2

6.0

125

2,89

4

Espi

ritu

Sant

oES

274.

812.

4

0.

48

1.

11

3.

4

1.

6

3.0

2.2

173

1,74

8

Espi

ritu

Sant

oES

289.

4314

.0

1.

15

2.

89

19

.4

22

.2

0.

4

4.

8

1,

250

10

,282

Espi

ritu

Sant

oES

2923

.05

29.0

1.11

3.60

34.1

37.9

0.6

8.1

2,48

0

51,3

81

Es

pirit

u Sa

nto

ES30

42.3

922

.0

1.

37

2.

50

23

.7

32

.6

1.

3

8.

8

22

4

6,

908

Es

pirit

u Sa

nto

ES31

52.2

119

.0

1.

03

1.

67

20

.3

20

.8

2.

5

11

.4

89

4,

529

242

Wat

ersh

edBi

ocom

plex

ity I

DCr

oss-

Sect

ion

IDQ

ac

w a

c R

ac

dm

ax a

c P

ac

A a

c v

ac

w/d

ac

τ a

c Ω

ac

Espi

ritu

Sant

oES

3270

.59

21.7

0.73

1.48

22.3

16.4

4.3

14.7

78

7,51

9

Espi

ritu

Sant

oES

3338

.64

21.0

0.48

1.06

21.8

10.4

3.7

19.8

157

12,6

57

Es

pirit

u Sa

nto

ES34

36.3

521

.0

1.

06

1.

92

22

.1

23

.4

1.

6

11

.0

2,

075

71

,313

Espi

ritu

Sant

oES

3571

.31

26.0

2.05

3.47

27.5

56.3

1.3

7.5

24

819

Espi

ritu

Sant

oES

3671

.32

22.0

1.11

2.03

22.6

25.0

2.8

10.9

13

819

Espi

ritu

Sant

oES

3715

.61

17.3

1.11

2.31

18.5

20.7

0.8

7.5

1,83

7

25,7

35

Es

pirit

u Sa

nto

ES38

16.9

410

.3

0.

79

1.

43

10

.7

8.

5

2.0

7.2

583

12,4

98

Bl

anco

B110

.57

8.0

0.84

1.64

9.5

8.0

1.

3

4.

9

73

91

3

Bl

anco

B210

.20

8.0

1.04

1.58

9.4

9.8

1.

0

5.

0

90

88

2

Bl

anco

B38.

905.

7

0.

77

1.

48

6.

7

5.

2

1.7

3.9

67

770

Blan

coB4

6.32

6.8

0.73

1.26

7.7

5.6

1.

1

5.

4

63

54

6

Bl

anco

B513

.09

7.1

0.70

1.07

8.5

5.9

2.

2

6.

6

76

7

14

,344

Blan

coB6

0.49

4.0

0.42

0.97

4.8

2.1

0.

2

4.

1

11

1

12

9

Bl

anco

B713

.91

17.0

0.80

1.31

17.5

13.9

1.0

12.9

491

8,58

0

Blan

coB8

13.3

517

.0

0.

88

1.

53

18

.1

15

.9

0.

8

11

.1

1,

076

16

,375

Blan

coB9

5.99

7.0

0.19

0.51

7.4

1.4

4.

3

13

.8

18

0

5,

732

Bl

anco

B10

77.6

131

.0

0.

35

0.

75

34

.9

12

.3

6.

3

41

.6

39

8,

584

Bl

anco

B11

67.2

614

.0

1.

05

2.

30

16

.9

17

.7

3.

8

6.

1

1,

243

79

,975

Saba

naS1

13.6

520

.6

0.

64

0.

92

20

.8

13

.4

1.

0

22

.4

18

8

3,

999

Sa

bana

S213

.27

11.3

0.37

0.95

11.6

4.3

3.

1

11

.9

14

8

5,

355

Sa

bana

S36.

3220

.1

0.

48

1.

26

22

.3

10

.7

0.

6

16

.0

42

4

5,

583

Sa

bana

S46.

6412

.3

0.

70

1.

08

13

.7

9.

6

0.7

11.4

512

4,86

1

Saba

naS5

13.0

122

.8

0.

83

1.

61

25

.0

20

.6

0.

6

14

.2

43

8

6,

898

Sa

bana

S65.

7716

.0

1.

02

1.

94

17

.6

18

.0

0.

3

8.

2

67

4

3,

800

Sa

bana

S75.

498.

9

0.

86

1.

69

9.

9

8.

5

0.6

5.3

813

5,20

1

Saba

naS8

2.48

10.3

0.53

0.90

10.5

5.6

0.

4

11

.4

88

41

5

Sa

bana

S91.

395.

8

0.

59

1.

13

6.

7

4.

0

0.3

5.1

169

395

Saba

naS1

018

.00

11.0

1.16

2.45

13.1

15.1

1.2

4.5

146

2,26

6

Saba

naS1

117

.58

13.0

0.71

1.19

13.4

9.5

1.

9

10

.9

78

1,

933

Sa

bana

S12

0.07

5.9

0.21

0.60

6.4

1.4

0.

1

9.

9

22

9

75

Sa

bana

S13

18.8

918

.0

0.

53

1.

67

20

.1

10

.6

1.

8

10

.8

92

3,

273

Sa

bana

S14

19.1

914

.3

0.

54

1.

05

15

.2

8.

2

2.3

13.6

407

14,4

79

Sa

bana

S15

19.4

522

.0

0.

43

0.

91

22

.7

9.

8

2.0

24.2

49

2,20

6

Saba

naS1

63.

897.

3

0.

37

0.

70

7.

6

2.

8

1.4

10.4

115

1,19

6

Saba

naS1

71.

804.

7

0.

34

0.

47

4.

9

1.

7

1.1

10.0

56

297

Faja

rdo

194

Brid

ge1U

72.7

213

.8

1.

70

3.

70

16

.1

27

.4

2.

7

3.

7

31

1,

326

Fa

jard

o19

4 Br

idge

Brid

ge74

.27

56.0

0.40

1.03

56.6

22.8

3.3

54.4

7

1,

354

243

Wat

ersh

edBi

ocom

plex

ity I

DCr

oss-

Sect

ion

IDQ

ac

w a

c R

ac

dm

ax a

c P

ac

A a

c v

ac

w/d

ac

τ a

c Ω

ac

Faja

rdo

194

Brid

ge1D

74.2

815

.9

1.

02

1.

50

16

.6

17

.0

4.

4

10

.6

19

1,

354

Fa

jard

o97

7 Br

idge

2U59

.98

12.7

0.91

1.80

13.6

12.3

4.9

7.1

64

4,20

0

Faja

rdo

977

Brid

ge1U

59.7

125

.4

0.

65

1.

39

25

.9

16

.9

3.

5

18

.3

46

4,

181

Fa

jard

o97

7 Br

idge

Brid

ge60

.00

36.5

0.63

1.00

37.1

23.3

2.6

36.3

44

4,20

2

Faja

rdo

977

Brid

ge1D

60.0

427

.9

0.

39

0.

67

28

.0

10

.9

5.

5

41

.9

27

4,

204

Fa

jard

oCh

arco

Frio

1D27

.80

17.8

0.67

1.35

18.6

12.6

2.2

13.2

111

4,57

6

Faja

rdo

Char

co F

rioBr

idge

27.7

413

.9

0.

62

0.

99

16

.1

10

.0

2.

8

14

.0

74

3,

335

Fa

jard

oCh

arco

Frio

1U27

.73

14.6

1.09

1.78

15.5

17.0

1.6

8.2

132

3,33

4

Faja

rdo

Char

co F

rio2U

27.7

218

.0

0.

78

1.

59

18

.6

14

.5

1.

9

11

.3

94

3,

332

Fa

jard

oCh

arco

Frio

3U27

.69

14.8

0.93

2.74

16.6

15.5

1.8

5.4

112

3,32

9

Faja

rdo

Char

co F

rio4U

27.6

120

.0

1.

56

3.

59

22

.7

35

.3

0.

8

5.

6

18

7

3,

320

Fa

jard

oCh

arco

Frio

5U27

.45

15.6

0.96

2.02

16.3

15.6

1.8

7.7

189

5,44

1

Faja

rdo

Char

co F

rio7U

27.3

414

.0

0.

61

1.

36

14

.4

8.

8

3.1

10.3

121

5,41

8

Faja

rdo

La T

inaj

a3U

25.7

614

.8

0.

81

1.

84

16

.4

13

.4

1.

9

8.

0

48

0

15

,223

Faja

rdo

La T

inaj

a2U

26.1

111

.7

1.

76

3.

67

16

.5

29

.1

0.

9

3.

2

1,

036

15

,338

Faja

rdo

La T

inaj

a1U

26.3

013

.7

1.

21

2.

17

15

.3

18

.5

1.

4

6.

3

39

9

8,

688

Fa

jard

oRiv

er B

ar2U

3.60

4.4

0.48

1.01

5.4

2.6

1.

4

4.

4

13

4

1,

016

Fa

jard

oRiv

er B

ar1U

3.62

9.7

0.84

1.36

11.0

9.3

0.

4

7.

1

23

7

1,

019

Fa

jard

oRiv

er B

ar1D

3.62

11.2

0.72

1.33

13.1

9.4

0.

4

8.

4

20

2

1,

020

Fa

jard

oRiv

er B

ar2D

3.63

7.8

0.72

1.29

8.9

6.4

0.

6

6.

1

20

4

1,

022

Fa

jard

oSm

all T

ribut

ary

2U0.

353.

5

0.

49

0.

83

4.

0

2.

0

0.2

4.3

555

400

Faja

rdo

Smal

l Trib

utar

y1U

0.36

4.4

0.64

1.12

5.2

3.3

0.

1

3.

9

72

8

40

4

Fa

jard

oSm

all T

ribut

ary

1D0.

364.

2

0.

54

1.

18

6.

0

3.

3

0.1

3.5

613

404

Faja

rdo

Smal

l Trib

utar

y2D

0.36

6.9

0.72

1.39

7.7

5.6

0.

1

5.

0

81

7

40

4

Fa

jard

oU

pper

Faj

ardo

Trib

2U1.

102.

4

0.

53

1.

65

4.

7

2.

5

0.4

1.4

965

2,00

4

Faja

rdo

Upp

er F

ajar

doTr

ib1U

1.10

2.3

0.59

1.54

4.7

2.8

0.

4

1.

5

1,

073

2,

003

Fa

jard

oU

pper

Faj

ardo

3U15

.37

11.5

1.10

2.24

12.8

14.1

1.1

5.1

266

3,70

4

Faja

rdo

Upp

er F

ajar

do2U

15.6

414

.0

1.

05

2.

37

16

.5

17

.4

0.

9

5.

9

19

8

2,

956

Fa

jard

oU

pper

Faj

ardo

1U15

.83

11.7

1.26

2.00

13.6

17.1

0.9

5.9

238

2,99

2

Mam

eyes

Ange

lito

Trai

l1U

41.9

324

.5

0.

85

1.

74

27

.4

23

.3

1.

8

28

.9

59

2

29

,277

Mam

eyes

Ange

lito

Trai

l1D

41.9

424

.8

1.

67

3.

90

27

.6

46

.1

0.

9

14

.8

1,

169

29

,284

Mam

eyes

Ange

lito

Trai

l2D

43.7

018

.9

0.

98

1.

58

19

.8

19

.4

2.

3

19

.3

28

7

12

,798

Espi

ritu

Sant

oBa

ckw

ay S

choo

l2U

53.5

615

.2

0.

80

1.

11

15

.9

12

.8

4.

2

18

.9

53

3,

539

Es

pirit

u Sa

nto

Back

way

Sch

ool

1U53

.71

12.5

0.76

1.08

13.4

10.2

5.3

16.4

50

3,54

9

Espi

ritu

Sant

oBa

ckw

ay S

choo

l1D

53.7

213

.0

0.

78

1.

29

13

.4

10

.4

5.

1

16

.7

52

3,

550

Es

pirit

u Sa

nto

Back

way

Sch

ool

2D53

.73

14.4

0.70

1.29

15.2

10.6

5.1

20.7

46

3,55

1

Espi

ritu

Sant

oBe

hind

Bris

as2U

54.6

817

.0

0.

66

1.

15

17

.4

11

.5

4.

8

25

.8

8

633

244

Wat

ersh

edBi

ocom

plex

ity I

DCr

oss-

Sect

ion

IDQ

ac

w a

c R

ac

dm

ax a

c P

ac

A a

c v

ac

w/d

ac

τ a

c Ω

ac

Espi

ritu

Sant

oBe

hind

Bris

a s1U

54.6

814

.6

0.

62

1.

27

15

.2

9.

5

5.8

23.4

7

63

3

Es

pirit

u Sa

nto

Behi

nd B

risas

BRID

GE

54.6

817

.5

1.

07

1.

90

18

.5

19

.8

2.

8

16

.4

12

63

3

M

amey

esBi

sley

32U

0.78

6.7

0.50

0.92

7.2

3.6

0.

2

13

.5

84

9

1,

326

M

amey

esBi

sley

31U

0.78

4.7

0.50

0.80

5.4

2.7

0.

3

9.

5

84

6

1,

336

M

amey

esBi

sley

31D

0.82

5.8

0.68

1.37

7.8

5.3

0.

2

8.

4

1,

162

1,

400

M

amey

esBi

sley

32D

0.83

4.7

0.40

1.08

5.5

2.2

0.

4

11

.8

68

1

1,

412

Es

pirit

u Sa

nto

Dav

ila3U

8.88

8.7

0.78

1.46

10.0

7.8

1.

1

11

.2

91

1,

045

Es

pirit

u Sa

nto

Dav

ila2U

8.89

7.7

0.70

0.94

8.1

5.7

1.

6

11

.0

82

1,

046

Es

pirit

u Sa

nto

Dav

ila1U

8.90

9.2

0.77

1.03

9.9

7.6

1.

2

11

.9

91

1,

047

Es

pirit

u Sa

nto

Dav

ila1D

8.91

8.0

0.50

0.85

8.3

4.2

2.

1

15

.9

59

1,

048

Es

pirit

u Sa

nto

Dav

ila2D

8.92

9.2

0.64

0.86

9.8

6.3

1.

4

14

.3

76

1,

049

M

amey

esEa

st P

eak

2U0.

997.

7

0.

37

0.

84

8.

2

3.

1

0.3

20.5

321

850

Mam

eyes

East

Pea

k1U

1.06

4.2

0.40

0.94

4.9

2.0

0.

5

10

.5

51

5

1,

352

M

amey

esEa

st P

eak

1D1.

082.

8

0.

41

0.

80

3.

4

1.

4

0.8

7.0

561

1,48

8

Mam

eyes

East

Pea

k2D

1.08

5.0

0.73

1.63

7.7

5.7

0.

2

6.

9

1,

016

1,

492

Es

pirit

u Sa

nto

El V

erde

3U42

.33

18.3

0.75

2.04

20.2

15.0

2.8

24.5

156

8,87

3

Espi

ritu

Sant

oEl

Ver

de2U

42.3

414

.9

1.

08

2.

34

17

.5

18

.9

2.

2

13

.8

22

7

8,

874

Es

pirit

u Sa

nto

El V

erde

1U42

.34

25.3

0.89

1.55

26.4

23.5

1.8

28.4

186

8,87

5

Espi

ritu

Sant

oEl

Ver

de1D

42.3

526

.0

0.

77

1.

30

27

.5

21

.2

2.

0

33

.7

16

2

8,

877

Es

pirit

u Sa

nto

El V

erde

2D42

.37

16.2

0.81

1.69

14.2

11.5

3.7

19.9

171

8,88

1

Espi

ritu

Sant

oES

Rt.

3 P

aint

ball

BRID

GE

71.6

133

.2

0.

83

1.

49

33

.7

28

.1

2.

6

39

.9

10

83

6

Es

pirit

u Sa

nto

ES W

ater

fall

3U22

.94

10.1

0.88

1.77

11.9

10.5

2.2

11.5

2,52

5

66,1

92

Es

pirit

u Sa

nto

ES W

ater

fall

2U22

.95

9.1

0.99

2.11

10.3

10.2

2.2

9.3

2,85

2

66,2

16

Es

pirit

u Sa

nto

ES W

ater

fall

1U22

.97

18.2

1.32

2.79

21.7

28.6

0.8

13.8

5,87

8

102,

435

Es

pirit

u Sa

nto

ES W

ater

fall

1D23

.05

28.7

1.17

3.77

33.9

39.5

0.6

24.6

2,60

2

51,3

81

Es

pirit

u Sa

nto

Jim

enez

2U52

.34

16.1

0.79

1.36

16.7

13.2

4.0

20.4

85

5,62

9

Espi

ritu

Sant

oJi

men

ez1U

52.3

725

.9

0.

77

1.

13

26

.5

20

.4

2.

6

33

.6

83

5,

631

Es

pirit

u Sa

nto

Jim

enez

1D52

.38

33.2

0.81

1.56

34.8

28.2

1.9

41.0

87

5,63

3

Espi

ritu

Sant

oJi

men

ez W

ater

fal2

U7.

4510

.6

0.

61

1.

07

8.

9

5.

4

1.4

17.3

670

8,16

5

Espi

ritu

Sant

oJi

men

ez W

ater

fal1

U7.

468.

0

0.

56

1.

14

8.

5

4.

8

1.6

14.2

500

6,65

2

Espi

ritu

Sant

oJi

men

ez W

ater

fal1

D7.

487.

9

0.

47

0.

99

8.

7

4.

1

1.8

16.7

1,15

9

18,3

55

Es

pirit

u Sa

nto

Jim

enez

Wat

erfa

l2D

7.56

17.3

0.94

3.15

20.5

19.3

0.4

18.4

2,49

5

20,0

41

M

amey

esJu

an D

iego

3U2.

0510

.2

0.

74

1.

17

10

.9

8.

1

0.3

13.7

2,70

0

7,45

8

Mam

eyes

Juan

Die

go2U

2.07

12.6

1.01

1.55

13.6

13.7

0.2

12.5

3,66

6

7,50

8

Mam

eyes

Juan

Die

go1U

3.66

7.4

0.70

1.05

8.6

6.0

0.

6

10

.6

1,

244

6,

523

M

amey

esJu

an D

iego

BRID

GE

3.66

5.4

0.66

1.03

6.3

4.2

0.

9

8.

2

1,

178

6,

531

M

amey

esJu

an D

iego

1D3.

679.

0

0.

43

0.

87

10

.2

4.

4

0.8

21.1

765

6,54

3

245

Wat

ersh

edBi

ocom

plex

ity I

DCr

oss-

Sect

ion

IDQ

ac

w a

c R

ac

dm

ax a

c P

ac

A a

c v

ac

w/d

ac

τ a

c Ω

ac

Mam

eye s

La C

oca

Trai

l2U

1.96

8.6

0.43

0.85

9.3

4.0

0.

5

20

.0

60

1

2,

742

M

amey

esLa

Coc

a Tr

ail

1U1.

975.

2

0.

51

1.

02

6.

2

3.

2

0.6

10.3

735

2,84

5

Mam

eyes

La C

oca

Trai

l1D

1.99

5.6

0.67

1.34

7.5

5.0

0.

4

8.

3

1,

425

4,

242

M

amey

esLa

Coc

a Tr

ail

2D2.

005.

0

0.

57

0.

84

5.

8

3.

3

0.6

8.8

873

3,07

2

Mam

eyes

La L

ingu

ente

2U0.

827.

8

0.

99

1.

84

10

.1

10

.0

0.

1

7.

9

1,

426

1,

182

M

amey

esLa

Lin

guen

te1U

0.84

5.1

0.48

0.90

5.7

2.7

0.

3

10

.7

68

6

1,

208

M

amey

esLa

Lin

guen

te1D

0.86

2.7

0.43

0.83

3.8

1.6

0.

5

6.

2

62

0

1,

241

M

amey

esLa

Lin

guen

te2D

0.87

6.6

0.47

0.94

7.1

3.3

0.

3

14

.0

67

3

1,

255

M

amey

esLa

Maq

uina

2U1.

952.

5

0.

58

1.

33

4.

5

2.

6

0.7

4.2

929

3,09

3

Mam

eyes

La M

aqui

na1U

1.93

3.7

0.35

0.59

4.0

1.4

1.

4

10

.7

55

3

3,

070

M

amey

esLa

Maq

uina

BRID

GE

1.94

5.0

0.54

0.75

5.7

3.1

0.

6

9.

2

86

5

3,

083

M

amey

esLa

Maq

uina

1D1.

954.

6

0.

43

1.

05

5.

9

2.

5

0.8

10.7

681

3,09

4

Mam

eyes

La M

aqui

na2D

2.14

5.7

0.36

0.60

6.1

2.2

1.

0

16

.0

56

6

3,

403

M

amey

esLa

Min

a Fa

lls2U

8.26

10.3

0.71

1.16

10.7

7.6

1.

1

14

.5

1,

476

17

,302

Mam

eyes

La M

ina

Falls

1U8.

2510

.8

1.

00

1.

52

12

.7

12

.7

0.

6

10

.8

2,

089

17

,265

Mam

eyes

La M

ina

Falls

1D8.

339.

1

0.

57

1.

38

10

.4

5.

9

1.4

16.0

1,19

4

17,4

41

M

amey

esLa

Veg

a2U

(ne

w)

52.3

317

.2

0.

87

1.

35

17

.8

15

.5

3.

4

19

.7

66

3,

934

M

amey

esLa

Veg

a2U

(ol

d)52

.32

19.8

0.65

0.93

20.1

13.1

4.0

30.5

49

3,93

3

Mam

eyes

La V

ega

1U52

.29

21.0

0.68

1.26

22.0

14.9

3.5

31.0

51

3,93

1

Mam

eyes

La V

ega

RO

AD52

.47

14.6

1.23

2.47

16.2

19.9

2.6

11.8

93

3,94

5

Mam

eyes

La V

ega

1D52

.48

21.3

0.43

1.00

21.7

9.3

5.

7

49

.8

32

3,

945

M

amey

esLa

Veg

a2D

52.5

217

.7

0.

84

1.

09

18

.2

15

.2

3.

5

21

.1

63

3,

948

Es

pirit

u Sa

nto

Las

Tres

T (

1)2U

1.50

7.4

0.29

0.56

8.0

2.3

0.

7

25

.8

11

8

62

3

Es

pirit

u Sa

nto

Las

Tres

T (

1)1U

1.53

6.5

0.42

0.70

6.7

2.8

0.

5

15

.3

17

5

63

5

Es

pirit

u Sa

nto

Las

Tres

T (

1)1D

1.53

10.1

1.07

1.85

11.3

12.1

0.1

9.4

444

637

Espi

ritu

Sant

oLa

s Tr

es T

(1)

2D1.

613.

4

0.

33

0.

51

4.

0

1.

3

1.2

10.6

135

669

Espi

ritu

Sant

oLa

s Tr

es T

(2)

2U1.

325.

0

0.

56

1.

12

5.

8

3.

3

0.4

8.8

359

839

Espi

ritu

Sant

oLa

s Tr

es T

(2)

1U1.

355.

3

0.

39

0.

61

5.

6

2.

2

0.6

13.7

246

859

Espi

ritu

Sant

oLa

s Tr

es T

(2)

1D1.

358.

1

0.

68

1.

57

9.

0

6.

2

0.2

11.9

436

861

Espi

ritu

Sant

oLa

s Tr

es T

(2)

2D1.

384.

1

0.

35

0.

79

4.

7

1.

6

0.9

12.0

220

880

Mam

eyes

Mam

eyes

Rt.

31U

62.2

416

.0

1.

12

2.

01

16

.8

18

.9

3.

3

14

.2

26

1,

463

M

amey

esM

amey

es R

t. 3

1D62

.28

35.0

0.71

2.19

45.7

32.4

1.9

49.2

17

1,46

4

Espi

ritu

Sant

oPr

ieta

3U1.

129.

5

0.

56

2.

19

9.

8

5.

5

0.2

17.0

893

1,78

8

Espi

ritu

Sant

oPr

ieta

2U1.

238.

5

0.

48

0.

95

6.

0

2.

9

0.4

17.7

766

1,96

5

Espi

ritu

Sant

oPr

ieta

1U1.

236.

0

0.

55

1.

46

6.

2

3.

5

0.4

10.9

884

1,97

1

Espi

ritu

Sant

oPr

ieta

1D1.

279.

3

0.

90

1.

56

10

.9

9.

8

0.1

10.3

1,43

5

2,02

2

Espi

ritu

Sant

oPr

ieta

2D1.

276.

0

0.

49

1.

25

6.

9

3.

4

0.4

12.2

785

2,02

8

246

Wat

ersh

edBi

ocom

plex

ity I

DCr

oss-

Sect

ion

IDQ

ac

w a

c R

ac

dm

ax a

c P

ac

A a

c v

ac

w/d

ac

τ a

c Ω

ac

Mam

eye s

Puen

te R

oto

2U44

.58

12.3

0.95

1.85

13.4

12.7

3.5

13.0

145

6,79

2

Mam

eyes

Puen

te R

oto

1U/B

RID

GE

44.5

327

.6

0.

84

1.

64

28

.2

23

.7

1.

9

32

.8

12

8

6,

784

M

amey

esPu

ente

Rot

o1D

44.6

012

.8

0.

65

1.

62

14

.0

9.

1

4.9

19.7

99

6,79

5

Mam

eyes

Puen

te R

oto

2D44

.68

16.8

1.02

1.70

17.9

18.2

2.5

16.5

155

6,80

8

Mam

eyes

Q. A

non

2U0.

773.

3

0.

48

0.

99

4.

3

2.

0

0.4

7.0

52

84

Mam

eyes

Q. A

non

1U/B

RID

GE

0.78

4.4

0.46

0.79

5.1

2.4

0.

3

9.

4

50

85

M

amey

esQ

. Ano

n1D

0.79

7.0

1.02

1.71

8.3

8.5

0.

1

6.

9

11

1

86

M

amey

esQ

. Ano

n2D

0.80

3.1

0.58

1.23

4.5

2.6

0.

3

5.

3

63

87

Es

pirit

u Sa

nto

Sona

dora

3U9.

4110

.7

1.

01

2.

31

13

.4

13

.6

0.

7

10

.6

2,

834

26

,365

Espi

ritu

Sant

oSo

nado

ra2U

9.42

18.0

1.32

2.33

20.2

26.7

0.4

13.7

3,69

6

26,3

92

Es

pirit

u Sa

nto

Sona

dora

1U9.

4313

.1

0.

98

2.

65

16

.7

16

.4

0.

6

13

.4

1,

072

10

,282

Espi

ritu

Sant

oSo

nado

ra1D

10.1

414

.8

0.

58

1.

50

15

.9

9.

2

1.1

25.6

817

14,3

17

Es

pirit

u Sa

nto

Sona

dora

2D10

.16

10.3

0.75

1.21

10.9

8.1

1.

2

13

.8

1,

054

14

,336

Espi

ritu

Sant

oTo

ronj

a2U

0.17

5.0

0.36

0.91

5.6

2.0

0.

1

13

.9

46

0

21

5

Es

pirit

u Sa

nto

Toro

nja

1U0.

173.

5

0.

35

0.

56

3.

9

1.

4

0.1

9.9

457

224

Espi

ritu

Sant

oTo

ronj

a1D

0.18

4.5

0.63

1.16

5.4

3.4

0.

1

7.

2

81

2

23

1

Es

pirit

u Sa

nto

Toro

nja

2D0.

187.

0

0.

41

0.

69

7.

3

3.

0

0.1

17.2

524

238

247

Wat

ersh

edBi

ocom

plex

ity I

DCr

oss-

Sect

ion

IDd1

6d5

0d8

4dm

axSi

mps

on's

In

dex

Sort

ing

Inde

xBe

droc

kM

egab

ould

erBo

ulde

rCo

bble

Gra

vel

Sand

Fine

s

Mam

eye s

M1

144

192

---

4.26

13.9

1.2

0.0

12.0

27.5

28.7

24.0

6.6

Mam

eyes

M2

2187

457

500

3.01

4.7

0.0

0.0

26.7

20.7

46.5

3.4

2.6

Mam

eyes

M3

2010

851

570

03.

335.

00.

00.

031

.922

.138

.06.

71.

2M

amey

esM

427

160

614

1800

3.20

4.8

6.0

0.0

41.8

16.3

32.6

2.7

0.5

Mam

eyes

M5

2932

812

0827

004.

586.

425

.411

.929

.98.

520

.34.

00.

0M

amey

esM

615

4624

680

03.

444.

124

.60.

013

.110

.744

.37.

40.

0M

amey

esM

718

110

451

500

3.70

5.0

0.0

0.0

26.3

29.9

32.3

6.0

5.4

Mam

eyes

M8

2311

254

360

03.

054.

90.

00.

035

.119

.040

.85.

20.

0M

amey

esM

911

9849

810

004.

416.

733

.60.

020

.114

.221

.67.

53.

0M

amey

esM

1014

4638

545

03.

215.

215

.90.

018

.28.

048

.90.

09.

1M

amey

esM

1183

188

601

1000

3.63

2.7

24.2

0.0

32.3

32.3

8.9

0.0

2.4

Mam

eyes

M12

1988

173

500

3.04

3.0

8.4

0.0

4.2

42.0

37.0

8.4

0.0

Mam

eyes

M13

2510

523

380

02.

223.

063

.61.

53.

815

.913

.60.

01.

5M

amey

esM

1469

169

568

1500

3.72

2.9

26.2

0.0

27.9

32.8

12.3

0.0

0.8

Mam

eyes

M15

2712

456

070

02.

824.

51.

80.

036

.718

.343

.10.

00.

0M

amey

esM

1652

430

1170

1500

2.68

4.7

54.4

6.7

24.4

4.4

10.0

0.0

0.0

Mam

eyes

M17

4437

591

313

003.

564.

618

.58.

143

.59.

720

.20.

00.

0M

amey

esM

1833

306

1185

1700

4.20

6.0

9.1

14.9

35.5

12.4

25.6

0.0

2.5

Mam

eyes

M19

2292

604

650

3.23

5.3

21.3

0.0

34.0

3.2

38.3

0.0

3.2

Mam

eyes

M20

3518

262

195

03.

434.

212

.40.

040

.020

.027

.60.

00.

0M

amey

esM

2136

370

853

900

3.51

4.9

31.8

5.6

37.4

4.7

19.6

0.9

0.0

Mam

eyes

M22

2116

161

112

003.

195.

40.

00.

044

.319

.126

.79.

90.

0M

amey

esM

2375

502

1582

2400

3.12

4.6

0.0

24.8

46.8

10.1

17.4

0.0

0.9

Mam

eyes

M24

2310

056

222

003.

014.

94.

91.

233

.113

.544

.82.

50.

0M

amey

esM

2521

114

578

1400

3.93

5.3

25.0

0.0

29.5

9.8

30.3

5.3

0.0

Mam

eyes

M26

---

---

---

---

---

---

---

---

---

---

---

---

---

Mam

eyes

M27

8734

971

011

002.

632.

99.

10.

056

.620

.210

.14.

00.

0M

amey

esM

2825

231

658

2.12

5.2

65.4

0.0

17.9

3.8

10.3

2.6

0.0

Mam

eyes

M29

4236

877

420

002.

564.

34.

14.

157

.110

.222

.42.

00.

0M

amey

esM

3093

294

719

1200

2.83

2.8

5.4

3.1

50.4

27.9

13.2

0.0

0.0

Mam

eyes

M31

9134

070

514

002.

522.

855

.60.

027

.411

.13.

41.

70.

9M

amey

esM

3228

146

648

1200

4.37

4.8

23.3

3.8

27.1

17.3

26.3

2.3

0.0

Mam

eyes

M33

3417

059

412

003.

594.

27.

90.

038

.629

.118

.95.

50.

0M

amey

esM

3424

9746

013

003.

624.

413

.80.

023

.221

.039

.91.

40.

7M

amey

esM

3520

7328

816

002.

743.

853

.70.

08.

112

.223

.62.

40.

0M

amey

esM

3646

345

707

1300

2.40

3.9

5.6

0.0

58.9

12.9

22.6

0.0

0.0

App

endi

x. G

rain

Siz

e an

d P

ebbl

e C

oun

t D

ata

248

Wat

ersh

edBi

ocom

plex

ity I

DCr

oss-

Sect

ion

IDd1

6d5

0d8

4dm

axSi

mps

on's

In

dex

Sort

ing

Inde

xBe

droc

kM

egab

ould

erBo

ulde

rCo

bble

Gra

vel

Sand

Fine

s

Mam

eye s

M37

2389

199

800

3.10

2.9

46.0

0.0

6.2

21.2

24.8

1.8

0.0

Mam

eyes

M38

2917

463

010

002.

714.

73.

90.

045

.312

.538

.30.

00.

0M

amey

esM

3921

158

573

700

3.84

5.2

8.7

0.0

35.6

30.8

13.5

11.5

0.0

Mam

eyes

M40

189

254

635

3.52

22.4

0.0

0.0

19.4

40.8

21.4

18.4

0.0

Mam

eyes

M41

2512

738

117

783.

863.

90.

04.

825

.737

.117

.115

.20.

0M

amey

esM

4238

229

785

1626

3.05

4.5

0.0

7.2

38.1

39.2

15.5

0.0

0.0

Mam

eyes

M43

1610

222

991

43.

103.

80.

00.

021

.941

.032

.44.

80.

0Es

pirit

u Sa

nto

ES1

---

---

---

---

---

---

---

---

---

---

---

---

---

Espi

ritu

Sant

oES

364

203

610

1397

2.65

3.1

0.0

0.0

47.5

34.7

17.8

0.0

0.0

Espi

ritu

Sant

oES

412

740

612

4525

402.

743.

10.

016

.353

.819

.210

.60.

00.

0Es

pirit

u Sa

nto

ES5

7630

583

513

972.

513.

30.

00.

054

.224

.121

.70.

00.

0Es

pirit

u Sa

nto

ES6

7619

353

312

192.

652.

63.

05.

950

.533

.76.

90.

00.

0Es

pirit

u Sa

nto

ES7

2511

435

673

73.

463.

70.

00.

033

.025

.233

.08.

70.

0Es

pirit

u Sa

nto

ES8

7623

548

316

512.

652.

50.

02.

952

.027

.517

.60.

00.

0Es

pirit

u Sa

nto

ES9

5173

718

2918

293.

686.

012

.020

.042

.07.

019

.00.

00.

0Es

pirit

u Sa

nto

ES10

203

673

1829

1829

3.13

3.0

12.8

20.2

48.9

14.9

3.2

0.0

0.0

Espi

ritu

Sant

oES

1111

178

551

1829

4.19

7.0

8.6

3.2

32.3

30.1

16.1

9.7

0.0

Espi

ritu

Sant

oES

1230

127

351

610

1.44

3.4

82.8

0.0

3.2

8.6

5.4

0.0

0.0

Espi

ritu

Sant

oES

1351

381

1278

1981

3.47

5.0

4.1

15.6

43.4

13.1

23.8

0.0

0.0

Espi

ritu

Sant

oES

1462

279

610

1575

2.95

3.1

30.3

0.0

46.5

7.1

16.2

0.0

0.0

Espi

ritu

Sant

oES

1520

478

713

9527

432.

902.

60.

035

.844

.213

.73.

23.

20.

0Es

pirit

u Sa

nto

ES16

8030

573

713

212.

163.

00.

00.

062

.523

.413

.30.

80.

0Es

pirit

u Sa

nto

ES17

112

213

9722

614.

7237

.40.

021

.920

.312

.527

.318

.00.

0Es

pirit

u Sa

nto

ES18

812

755

291

43.

008.

20.

00.

040

.412

.838

.50.

97.

3Es

pirit

u Sa

nto

ES19

8943

210

0615

242.

623.

40.

014

.257

.513

.210

.40.

04.

7Es

pirit

u Sa

nto

ES20

6539

483

815

242.

373.

62.

70.

060

.419

.812

.64.

50.

0Es

pirit

u Sa

nto

ES21

5117

861

096

53.

003.

50.

00.

041

.735

.018

.44.

90.

0Es

pirit

u Sa

nto

ES22

7748

314

8525

913.

404.

410

.112

.146

.520

.211

.10.

00.

0Es

pirit

u Sa

nto

ES23

---

---

---

---

---

---

---

---

---

---

---

---

---

Espi

ritu

Sant

oES

24--

---

---

---

---

---

---

---

---

---

---

---

---

-Es

pirit

u Sa

nto

ES25

112

729

661

03.

1317

.20.

00.

024

.641

.84.

928

.70.

0Es

pirit

u Sa

nto

ES26

11

11

1.00

1.0

0.0

0.0

0.0

0.0

0.0

100.

00.

0Es

pirit

u Sa

nto

ES27

11

11

1.00

1.0

0.0

0.0

0.0

0.0

0.0

100.

00.

0Es

pirit

u Sa

nto

ES28

5171

121

3430

483.

446.

50.

037

.231

.911

.719

.10.

00.

0Es

pirit

u Sa

nto

ES29

254

1295

1880

3251

3.10

2.7

48.5

18.4

21.4

7.8

3.9

0.0

0.0

Espi

ritu

Sant

oES

3014

647

012

1924

382.

792.

95.

814

.455

.811

.59.

62.

90.

0Es

pirit

u Sa

nto

ES31

1512

761

015

244.

006.

340

.06.

014

.017

.019

.04.

00.

0

249

Wat

ersh

edBi

ocom

plex

ity I

DCr

oss-

Sect

ion

IDd1

6d5

0d8

4dm

axSi

mps

on's

In

dex

Sort

ing

Inde

xBe

droc

kM

egab

ould

erBo

ulde

rCo

bble

Gra

vel

Sand

Fine

s

Espi

ritu

Sant

oES

3238

152

381

813

4.00

3.2

33.0

0.0

24.0

24.0

16.0

2.0

1.0

Espi

ritu

Sant

oES

3344

146

506

1549

3.01

3.4

0.0

0.0

31.6

28.4

38.9

1.1

0.0

Espi

ritu

Sant

oES

3415

240

612

7027

433.

792.

920

.815

.837

.523

.32.

50.

00.

0Es

pirit

u Sa

nto

ES35

---

---

---

---

---

---

---

---

---

---

---

---

---

Espi

ritu

Sant

oES

36--

---

---

---

---

---

---

---

---

---

---

---

---

-Es

pirit

u Sa

nto

ES37

4020

381

324

383.

474.

50.

08.

838

.231

.418

.62.

90.

0Es

pirit

u Sa

nto

ES38

2519

186

414

223.

275.

85.

60.

044

.918

.725

.25.

60.

0Bl

anco

B11

11

726

1.12

1.0

0.0

0.0

2.8

0.0

2.8

94.4

0.0

Blan

coB2

11

189

1.65

1.0

0.0

0.0

0.0

0.0

11.5

76.0

12.5

Blan

coB3

11

1513

81.

923.

90.

00.

00.

00.

917

.068

.913

.2Bl

anco

B41

11

11.

021.

00.

00.

00.

00.

00.

098

.91.

1Bl

anco

B51

11

2692

1.26

1.0

3.5

7.9

0.0

0.0

0.0

88.6

0.0

Blan

coB6

11

110

671.

351.

07.

80.

06.

90.

00.

085

.30.

0Bl

anco

B71

17

891.

832.

669

.20.

00.

00.

05.

025

.80.

0Bl

anco

B810

230

595

936

583.

253.

10.

012

.845

.326

.512

.03.

40.

0Bl

anco

B91

8986

419

054.

6829

.40.

014

.726

.510

.818

.627

.52.

0Bl

anco

B10

111

429

211

433.

8917

.10.

00.

027

.234

.014

.621

.42.

9Bl

anco

B11

101

330

4978

4978

3.36

7.0

43.6

26.7

6.9

16.8

5.9

0.0

0.0

Saba

naS1

2816

059

412

003.

154.

60.

00.

042

.225

.726

.65.

50.

0Sa

bana

S244

320

694

2800

2.35

4.0

0.0

0.0

58.7

17.5

22.2

1.6

0.0

Saba

naS3

135

313

6945

004.

2237

.00.

018

.638

.512

.413

.03.

114

.3Sa

bana

S494

410

787

2000

2.00

2.9

0.0

3.3

67.8

14.0

14.0

0.8

0.0

Saba

naS5

3140

913

5435

003.

236.

60.

018

.345

.86.

324

.64.

20.

7Sa

bana

S618

246

734

2500

3.19

6.4

0.0

4.9

47.3

9.3

26.4

9.3

2.7

Saba

naS7

3436

672

016

002.

064.

60.

00.

065

.26.

823

.53.

80.

8Sa

bana

S80

3119

390

03.

3013

8.8

0.0

0.0

15.5

10.0

44.5

3.6

26.4

Saba

naS9

033

324

800

4.95

180.

00.

00.

019

.520

.420

.416

.823

.0Sa

bana

S10

1559

433

500

3.97

5.4

28.0

0.0

17.8

11.0

35.6

2.5

5.1

Saba

naS1

10

2731

990

04.

3825

.80.

00.

019

.010

.232

.722

.415

.6Sa

bana

S12

1710

355

780

03.

185.

80.

00.

036

.913

.638

.81.

09.

7Sa

bana

S13

3532

380

330

003.

914.

822

.95.

339

.711

.516

.83.

80.

0Sa

bana

S14

188

648

1100

4.28

25.5

11.0

4.2

31.4

6.8

29.7

16.9

0.0

Saba

naS1

521

8752

828

003.

515.

013

.30.

028

.612

.441

.04.

80.

0Sa

bana

S16

1944

166

1100

1.77

2.9

0.0

0.0

13.5

11.5

73.1

0.0

1.9

Saba

naS1

710

3512

360

01.

973.

40.

00.

09.

96.

969

.310

.93.

0Fa

jard

o19

4 Br

idge

1U0.

0520

100

200

2.77

44.7

0.0

0.0

0.0

19.2

34.3

1.0

45.5

Faja

rdo

194

Brid

geBr

idge

4070

152

500

2.04

1.9

0.0

0.0

8.0

27.0

64.0

1.0

0.0

250

Wat

ersh

edBi

ocom

plex

ity I

DCr

oss-

Sect

ion

IDd1

6d5

0d8

4dm

axSi

mps

on's

In

dex

Sort

ing

Inde

xBe

droc

kM

egab

ould

erBo

ulde

rCo

bble

Gra

vel

Sand

Fine

s

Faja

rdo

194

Brid

ge1D

0.05

1.0

8325

02.

7140

.80.

00.

01.

010

.137

.45.

146

.5Fa

jard

o97

7 Br

idge

2U1.

060

145

450

3.54

12.0

10.0

0.0

3.0

26.0

40.0

21.0

0.0

Faja

rdo

977

Brid

ge1U

4090

210

410

3.04

2.3

0.0

0.0

17.0

37.0

40.0

6.0

0.0

Faja

rdo

977

Brid

geBr

idge

5010

016

242

02.

411.

80.

00.

09.

048

.042

.01.

00.

0Fa

jard

o97

7 Br

idge

1D40

9015

258

02.

741.

90.

00.

014

.040

.043

.03.

00.

0Fa

jard

oCh

arco

Frio

1D32

100

198

850

3.29

2.5

5.0

0.0

14.9

36.6

37.6

5.9

0.0

Faja

rdo

Char

co F

rioBr

idge

---

---

---

---

---

---

---

---

---

---

---

---

---

Faja

rdo

Char

co F

rio1U

5085

150

600

2.72

1.7

5.2

0.0

7.2

40.2

44.3

3.1

0.0

Faja

rdo

Char

co F

rio2U

3060

100

520

3.50

1.8

30.0

0.0

6.0

19.0

39.0

6.0

0.0

Faja

rdo

Char

co F

rio3U

4495

502

1150

4.09

3.4

34.0

0.0

20.0

17.0

24.0

5.0

0.0

Faja

rdo

Char

co F

rio4U

3070

130

850

3.52

2.1

21.0

0.0

9.0

25.0

41.0

4.0

0.0

Faja

rdo

Char

co F

rio5U

3880

142

750

3.09

1.9

15.5

0.0

7.8

32.0

43.7

1.0

0.0

Faja

rdo

Char

co F

rio7U

3090

274

2300

4.04

3.0

20.0

4.0

13.0

27.0

34.0

2.0

0.0

Faja

rdo

La T

inaj

a3U

4080

200

2300

3.47

2.2

23.7

6.2

1.0

23.7

41.2

4.1

0.0

Faja

rdo

La T

inaj

a2U

4015

523

740

01.

232.

490

.00.

02.

05.

03.

00.

00.

0Fa

jard

oLa

Tin

aja

1U40

8015

080

02.

781.

947

.00.

02.

019

.032

.00.

00.

0Fa

jard

oRiv

er B

ar2U

1.0

7034

397

04.

1818

.50.

00.

031

.021

.027

.06.

015

.0Fa

jard

oRiv

er B

ar1U

0.05

8593

245

005.

6413

6.5

0.0

16.0

23.0

11.0

19.0

12.0

19.0

Faja

rdo

Riv

er B

ar1D

1.0

4018

017

503.

4213

.40.

03.

011

.018

.047

.06.

015

.0Fa

jard

oRiv

er B

ar2D

5.0

6575

011

003.

1912

.20.

03.

034

.011

.042

.01.

09.

0Fa

jard

oSm

all T

ribut

ary

2U50

105

400

1300

3.17

2.8

0.0

5.0

25.0

30.0

40.0

0.0

0.0

Faja

rdo

Smal

l Trib

utar

y1U

5017

063

214

503.

043.

60.

03.

142

.325

.828

.90.

00.

0Fa

jard

oSm

all T

ribut

ary

1D60

170

732

1300

3.39

3.5

0.0

7.1

26.3

41.4

22.2

2.0

1.0

Faja

rdo

Smal

l Trib

utar

y2D

7018

077

813

504.

083.

315

.05.

033

.025

.022

.00.

00.

0Fa

jard

oU

pper

Faj

ardo

Trib

2U40

8023

157

03.

172.

445

.50.

010

.915

.826

.70.

01.

0Fa

jard

oU

pper

Faj

ardo

Trib

1U50

9014

040

03.

021.

742

.40.

04.

032

.321

.20.

00.

0Fa

jard

oU

pper

Faj

ardo

3U50

9016

038

03.

561.

821

.80.

06.

932

.734

.74.

00.

0Fa

jard

oU

pper

Faj

ardo

2U30

8017

060

03.

092.

412

.00.

06.

038

.040

.04.

00.

0Fa

jard

oU

pper

Faj

ardo

1U54

110

230

700

3.50

2.1

15.0

0.0

18.0

39.0

28.0

0.0

0.0

Mam

eyes

Ange

lito

Trai

l1U

5114

037

712

192.

872.

70.

00.

023

.536

.340

.20.

00.

0M

amey

esAn

gelit

o Tr

ail

1D38

114

356

1778

3.61

3.1

14.7

2.9

19.6

21.6

41.2

0.0

0.0

Mam

eyes

Ange

lito

Trai

l2D

2111

457

814

003.

935.

325

.00.

029

.59.

830

.35.

80.

0Es

pirit

u Sa

nto

Back

way

Sch

ool

2U1

1338

140

2.36

6.2

0.0

0.0

0.0

5.2

60.0

20.0

14.8

Espi

ritu

Sant

oBa

ckw

ay S

choo

l1U

113

5130

52.

357.

10.

00.

01.

85.

460

.718

.813

.4Es

pirit

u Sa

nto

Back

way

Sch

ool

1D13

7615

245

72.

443.

50.

00.

03.

839

.450

.05.

81.

0Es

pirit

u Sa

nto

Back

way

Sch

ool

2D1

2551

114

1.88

7.1

0.0

0.0

0.0

8.0

70.5

16.1

5.4

Espi

ritu

Sant

oBe

hind

Bris

as2U

2517

843

261

02.

704.

154

.50.

017

.919

.63.

63.

60.

9

251

Wat

ersh

edBi

ocom

plex

ity I

DCr

oss-

Sect

ion

IDd1

6d5

0d8

4dm

axSi

mps

on's

In

dex

Sort

ing

Inde

xBe

droc

kM

egab

ould

erBo

ulde

rCo

bble

Gra

vel

Sand

Fine

s

Espi

ritu

Sant

oBe

hind

Bris

a s1U

1316

545

776

22.

716.

056

.20.

014

.312

.412

.43.

81.

0Es

pirit

u Sa

nto

Behi

nd B

risas

BRID

GE

138

574

1143

4.87

24.0

12.5

0.0

30.8

11.5

24.0

7.7

13.5

Mam

eyes

Bisl

ey 3

2U51

127

610

1397

3.32

3.5

0.0

5.9

28.7

25.7

38.6

1.0

0.0

Mam

eyes

Bisl

ey 3

1U51

152

391

762

3.16

2.8

0.0

0.0

34.2

31.5

31.5

0.9

1.8

Mam

eyes

Bisl

ey 3

1D51

305

838

1143

3.21

4.1

0.0

9.6

45.8

16.9

25.3

2.4

0.0

Mam

eyes

Bisl

ey 3

2D76

184

813

1651

2.83

3.3

0.0

0.0

42.6

36.2

20.2

1.1

0.0

Espi

ritu

Sant

oD

avila

3U36

178

406

737

3.29

3.4

0.0

0.0

38.1

33.3

20.0

8.6

0.0

Espi

ritu

Sant

oD

avila

2U78

216

406

762

2.98

2.3

6.1

0.0

33.0

45.2

13.0

2.6

0.0

Espi

ritu

Sant

oD

avila

1U1

7622

945

73.

6815

.10.

00.

014

.934

.732

.714

.93.

0Es

pirit

u Sa

nto

Dav

ila1D

676

330

533

3.27

7.2

0.0

0.0

22.0

25.0

43.0

10.0

0.0

Espi

ritu

Sant

oD

avila

2D13

102

305

610

3.24

4.9

0.0

0.0

28.0

26.0

40.0

5.0

1.0

Mam

eyes

East

Pea

k2U

120

345

716

263.

5921

.40.

06.

850

.734

.28.

20.

00.

0M

amey

esEa

st P

eak

1U1

127

268

1168

4.06

16.4

0.0

0.0

31.5

52.1

16.4

0.0

0.0

Mam

eyes

East

Pea

k1D

11

135

61.

581.

076

.70.

01.

90.

00.

021

.40.

0M

amey

esEa

st P

eak

2D1

127

991

41.

8816

.769

.90.

05.

82.

91.

919

.40.

0Es

pirit

u Sa

nto

El V

erde

3U25

229

449

1575

3.87

4.2

28.4

0.0

20.2

32.1

18.3

0.9

0.0

Espi

ritu

Sant

oEl

Ver

de2U

3814

648

376

24.

153.

620

.40.

031

.119

.425

.23.

90.

0Es

pirit

u Sa

nto

El V

erde

1U25

203

483

914

3.06

4.4

0.0

0.0

44.1

25.5

25.5

4.9

0.0

Espi

ritu

Sant

oEl

Ver

de1D

5124

155

715

753.

313.

35.

92.

031

.742

.610

.96.

90.

0Es

pirit

u Sa

nto

El V

erde

2D15

243

293

517

022.

432.

50.

011

.559

.319

.59.

70.

00.

0Es

pirit

u Sa

nto

ES R

t. 3

Pai

ntba

llBR

IDG

E0

5126

748

34.

2373

.00.

00.

015

.825

.730

.75.

921

.8Es

pirit

u Sa

nto

ES W

ater

fall

3U57

762

1727

2667

3.41

5.5

9.3

15.7

48.1

11.1

12.0

3.7

0.0

Espi

ritu

Sant

oES

Wat

erfa

ll2U

782

914

1321

1321

1.25

1.3

89.0

4.0

7.0

0.0

0.0

0.0

0.0

Espi

ritu

Sant

oES

Wat

erfa

ll1U

181

2337

3048

3048

2.08

4.1

63.6

27.1

1.9

4.7

1.9

0.9

0.0

Espi

ritu

Sant

oES

Wat

erfa

ll1D

254

1295

1880

3251

3.10

2.7

48.5

18.4

21.4

7.8

3.9

0.0

0.0

Espi

ritu

Sant

oJi

men

ez2U

101

432

1119

3404

3.27

3.3

44.1

0.0

13.7

28.4

9.8

3.9

0.0

Espi

ritu

Sant

oJi

men

ez1U

2511

435

673

73.

363.

70.

00.

035

.827

.435

.81.

10.

0Es

pirit

u Sa

nto

Jim

enez

1D13

545

711

5228

703.

992.

929

.80.

025

.427

.214

.92.

60.

0Es

pirit

u Sa

nto

Jim

enez

Wat

erfa

l2U

8663

513

9717

784.

364.

019

.422

.233

.313

.012

.00.

00.

0Es

pirit

u Sa

nto

Jim

enez

Wat

erfa

l1U

3845

711

6819

053.

435.

542

.59.

428

.34.

714

.20.

90.

0Es

pirit

u Sa

nto

Jim

enez

Wat

erfa

l1D

5173

718

2918

293.

686.

012

.020

.042

.07.

019

.00.

00.

0Es

pirit

u Sa

nto

Jim

enez

Wat

erfa

l2D

5023

599

117

783.

434.

446

.08.

018

.017

.011

.00.

00.

0M

amey

esJu

an D

iego

3U41

114

381

914

3.76

3.0

21.8

0.0

27.7

16.8

33.7

0.0

0.0

Mam

eyes

Juan

Die

go2U

3717

810

1615

243.

695.

20.

013

.028

.023

.035

.01.

00.

0M

amey

esJu

an D

iego

1U25

7650

012

703.

444.

411

.06.

019

.019

.045

.00.

00.

0M

amey

esJu

an D

iego

BRID

GE

4415

275

813

723.

764.

122

.20.

034

.317

.226

.30.

00.

0M

amey

esJu

an D

iego

1D25

178

762

1549

3.02

5.5

0.0

2.7

42.3

19.8

33.3

1.8

0.0

252

Wat

ersh

edBi

ocom

plex

ity I

DCr

oss-

Sect

ion

IDd1

6d5

0d8

4dm

axSi

mps

on's

In

dex

Sort

ing

Inde

xBe

droc

kM

egab

ould

erBo

ulde

rCo

bble

Gra

vel

Sand

Fine

s

Mam

eye s

La C

oca

Trai

l2U

5117

876

213

723.

973.

96.

57.

426

.928

.729

.60.

90.

0M

amey

esLa

Coc

a Tr

ail

1U54

203

1054

1626

3.72

4.4

2.0

12.0

34.0

30.0

22.0

0.0

0.0

Mam

eyes

La C

oca

Trai

l1D

2923

510

6718

293.

956.

111

.56.

736

.515

.427

.91.

90.

0M

amey

esLa

Coc

a Tr

ail

2D51

216

782

1118

3.55

3.9

15.0

0.0

40.0

22.0

22.0

1.0

0.0

Mam

eyes

La L

ingu

ente

2U13

546

999

1778

3.58

8.9

26.0

4.0

42.0

3.0

15.0

0.0

10.0

Mam

eyes

La L

ingu

ente

1U19

7648

371

13.

745.

016

.30.

028

.811

.537

.55.

80.

0M

amey

esLa

Lin

guen

te1D

2510

283

816

514.

525.

717

.12.

724

.318

.930

.64.

51.

8M

amey

esLa

Lin

guen

te2D

1976

610

1016

2.96

5.7

8.2

0.0

30.9

14.5

46.4

0.0

0.0

Mam

eyes

La M

aqui

na2U

1364

192

762

3.81

3.9

29.6

0.0

5.6

24.0

32.8

8.0

0.0

Mam

eyes

La M

aqui

na1U

1351

127

381

2.80

3.2

47.6

0.0

3.8

10.5

34.3

2.9

1.0

Mam

eyes

La M

aqui

naBR

IDG

E1

1385

279

1.84

9.2

72.0

0.0

0.0

6.0

13.0

6.0

3.0

Mam

eyes

La M

aqui

na1D

2510

237

166

03.

883.

86.

20.

031

.023

.031

.94.

43.

5M

amey

esLa

Maq

uina

2D19

127

305

965

4.60

4.0

21.5

0.0

20.6

23.4

26.2

7.5

0.9

Mam

eyes

La M

ina

Falls

2U77

222

786

1626

3.83

3.2

30.0

2.0

33.0

23.0

9.0

3.0

0.0

Mam

eyes

La M

ina

Falls

1U15

750

813

5125

403.

662.

926

.014

.041

.010

.09.

00.

00.

0M

amey

esLa

Min

a Fa

lls1D

5123

578

317

783.

713.

939

.14.

326

.113

.916

.50.

00.

0M

amey

esLa

Veg

a2U

(ne

w)

1215

238

168

63.

695.

50.

00.

029

.931

.825

.213

.10.

0M

amey

esLa

Veg

a2U

(ol

d)31

152

385

610

3.16

3.5

0.0

0.0

30.0

39.0

27.0

4.0

0.0

Mam

eyes

La V

ega

1U25

127

279

724

3.01

3.3

0.0

0.0

14.0

48.6

26.2

8.4

2.8

Mam

eyes

La V

ega

RO

AD38

102

242

635

2.76

2.5

0.0

0.0

16.3

43.3

38.5

1.9

0.0

Mam

eyes

La V

ega

1D51

152

279

508

2.85

2.3

0.0

0.0

23.5

50.0

20.6

0.0

5.9

Mam

eyes

La V

ega

2D28

178

330

838

3.08

3.4

0.0

0.0

28.7

46.1

10.4

13.9

0.9

Espi

ritu

Sant

oLa

s Tr

es T

(1)

2U1

7633

063

54.

1619

.70.

00.

029

.021

.029

.05.

016

.0Es

pirit

u Sa

nto

Las

Tres

T (

1)1U

038

305

610

5.41

78.1

12.4

0.0

18.1

21.9

23.8

7.6

16.2

Espi

ritu

Sant

oLa

s Tr

es T

(1)

1D13

7668

616

763.

307.

30.

03.

933

.010

.741

.77.

82.

9Es

pirit

u Sa

nto

Las

Tres

T (

1)2D

038

178

635

4.48

59.6

0.0

0.0

11.9

25.7

27.7

11.9

22.8

Espi

ritu

Sant

oLa

s Tr

es T

(2)

2U1

1933

094

04.

2318

.20.

00.

020

.211

.836

.116

.815

.1Es

pirit

u Sa

nto

Las

Tres

T (

2)1U

119

176

737

3.85

13.3

15.0

0.0

12.5

9.2

44.2

10.0

9.2

Espi

ritu

Sant

oLa

s Tr

es T

(2)

1D1

191

356

1067

3.35

18.9

0.0

0.0

47.3

15.2

15.2

7.1

15.2

Espi

ritu

Sant

oLa

s Tr

es T

(2)

2D1

5441

915

494.

9420

.50.

02.

022

.419

.426

.515

.314

.3M

amey

esM

amey

es R

t. 3

1U1

152

267

483

3.21

16.3

0.0

0.0

19.1

46.4

17.3

17.3

0.0

Mam

eyes

Mam

eyes

Rt.

31D

138

140

635

3.10

11.8

0.0

0.0

8.0

18.0

44.0

30.0

0.0

Espi

ritu

Sant

oPr

ieta

3U25

857

218

1730

482.

282.

70.

024

.560

.85.

97.

80.

01.

0Es

pirit

u Sa

nto

Prie

ta2U

5145

799

214

733.

324.

40.

017

.845

.820

.613

.12.

80.

0Es

pirit

u Sa

nto

Prie

ta1U

7630

563

591

42.

772.

90.

00.

050

.030

.412

.72.

04.

9Es

pirit

u Sa

nto

Prie

ta1D

024

858

411

683.

4810

8.1

0.9

2.7

40.7

18.6

8.0

0.9

28.3

Espi

ritu

Sant

oPr

ieta

2D0

3217

640

63.

9659

.30.

00.

05.

931

.419

.611

.831

.4

253

Wat

ersh

edBi

ocom

plex

ity I

DCr

oss-

Sect

ion

IDd1

6d5

0d8

4dm

axSi

mps

on's

In

dex

Sort

ing

Inde

xBe

droc

kM

egab

ould

erBo

ulde

rCo

bble

Gra

vel

Sand

Fine

s

Mam

eye s

Puen

te R

oto

2U25

6415

281

32.

602.

411

.10.

07.

128

.353

.50.

00.

0M

amey

esPu

ente

Rot

o1U

/BRID

GE

5115

953

710

163.

623.

39.

90.

032

.427

.928

.80.

90.

0M

amey

esPu

ente

Rot

o1D

3217

853

015

243.

854.

139

.63.

618

.918

.018

.01.

80.

0M

amey

esPu

ente

Rot

o2D

1764

323

787

3.06

4.4

50.5

0.0

9.0

18.0

17.1

5.4

0.0

Mam

eyes

Q. A

non

2U0

1556

192

2.36

74.6

0.0

0.0

0.0

5.0

50.5

4.0

40.6

Mam

eyes

Q. A

non

1U/B

RID

GE

025

5916

01.

7876

.90.

00.

00.

02.

071

.03.

024

.0M

amey

esQ

. Ano

n1D

00

135

261

2.21

116.

10.

00.

00.

031

.310

.10.

058

.6M

amey

esQ

. Ano

n2D

014

6521

42.

7480

.40.

00.

00.

08.

745

.27.

738

.5Es

pirit

u Sa

nto

Sona

dora

3U76

305

1981

3505

4.10

5.1

5.2

20.9

33.9

23.5

16.5

0.0

0.0

Espi

ritu

Sant

oSo

nado

ra2U

7433

098

618

543.

393.

60.

015

.040

.030

.015

.00.

00.

0Es

pirit

u Sa

nto

Sona

dora

1U51

711

2134

3048

3.44

6.5

0.0

37.2

31.9

11.7

19.1

0.0

0.0

Espi

ritu

Sant

oSo

nado

ra1D

152

483

1321

2438

3.18

2.9

0.0

25.2

41.1

28.0

5.6

0.0

0.0

Espi

ritu

Sant

oSo

nado

ra2D

7625

482

320

574.

443.

313

.313

.324

.831

.417

.10.

00.

0Es

pirit

u Sa

nto

Toro

nja

2U1

7643

794

05.

4320

.922

.90.

022

.915

.219

.010

.59.

5Es

pirit

u Sa

nto

Toro

nja

1U1

9552

894

03.

6423

.00.

00.

039

.613

.226

.417

.03.

8Es

pirit

u Sa

nto

Toro

nja

1D1

3833

011

683.

6318

.20.

00.

022

.38.

742

.710

.715

.5Es

pirit

u Sa

nto

Toro

nja

2D0

1335

616

764.

9784

.30.

00.

018

.019

.821

.621

.618

.9

254

Wat

ersh

edBi

ocom

plex

ity I

DCr

oss-

Sect

ion

IDD

rop

Pdep

max

Pdep

cv

Plen

gth

Pare

aPv

olum

eag

r d

agr

ufs

t d

fst

uur

b d

urb

u

Faja

rdo

194

Brid

ge1U

0.5

4.25

0.

27

60.0

77

2

1,

190

1.

00

0.

43

-

0.

49

-

0.

08

Fa

jard

o19

4 Br

idge

Brid

ge0.

5

--

---

---

---

---

-1.

00

0.

43

-

0.

49

-

0.

08

Fa

jard

o19

4 Br

idge

1D0.

5

2.

12

0.20

19

5.0

2,82

8

1,54

1

1.00

0.42

-

0.47

-

0.11

Faja

rdo

977

Brid

ge2U

1.0

2.02

0.

46

33.0

28

4

20

5

0.96

0.40

0.03

0.56

0.01

0.04

Faja

rdo

977

Brid

ge1U

1.0

1.20

0.

39

44.0

55

6

27

8

0.96

0.40

0.03

0.56

0.01

0.04

Faja

rdo

977

Brid

geBr

idge

1.0

---

---

56.0

1,

172

16

5

0.96

0.40

0.03

0.56

0.01

0.04

Faja

rdo

977

Brid

ge1D

1.0

0.40

0.

16

85.0

1,

989

27

8

0.96

0.40

0.03

0.56

0.01

0.04

Faja

rdo

Char

co F

rio1D

2.0

0.91

0.

30

30.0

33

6

20

2

0.97

0.04

0.02

0.96

0.01

0.00

Faja

rdo

Char

co F

rioBr

idge

2.0

---

---

---

---

---

0.97

0.04

0.02

0.96

0.01

0.00

Faja

rdo

Char

co F

rio1U

2.0

2.30

0.

46

41.0

50

0

37

4

0.97

0.04

0.02

0.96

0.01

0.00

Faja

rdo

Char

co F

rio2U

2.0

1.92

0.

24

53.0

63

6

35

0

0.97

0.04

0.02

0.96

0.01

0.00

Faja

rdo

Char

co F

rio3U

2.0

1.87

0.

27

23.0

14

0

13

2

0.97

0.04

0.02

0.96

0.01

0.00

Faja

rdo

Char

co F

rio4U

2.0

3.05

0.

36

61.0

1,

092

1,

200

0.

97

0.

04

0.

02

0.

96

0.

01

0.

00

Fa

jard

oCh

arco

Frio

5U2.

0

2.

13

0.41

47

.0

531

445

0.

97

0.

04

0.

02

0.

96

0.

01

0.

00

Fa

jard

oCh

arco

Frio

7U2.

0

1.

66

0.52

48

.0

480

216

0.

97

0.

04

0.

02

0.

96

0.

01

0.

00

Fa

jard

oLa

Tin

aja

3U4.

0

3.

60

0.39

43

.5

513

297

0.

91

-

0.

08

1.

00

0.

01

0.

00

Fa

jard

oLa

Tin

aja

2U4.

0

3.

60

0.48

22

.0

198

488

0.

93

-

0.

07

1.

00

0.

01

0.

00

Fa

jard

oLa

Tin

aja

1U2.

5

3.

30

0.50

39

.0

460

456

0.

93

-

0.

06

1.

00

0.

01

0.

00

Fa

jard

oRiv

er B

ar2U

2.0

0.64

0.

40

13.5

37

13

0.

95

0.

15

0.

03

0.

84

0.

02

0.

01

Fa

jard

oRiv

er B

ar1U

2.0

1.42

0.

63

10.5

91

59

0.

95

0.

15

0.

03

0.

84

0.

02

0.

01

Fa

jard

oRiv

er B

ar1D

2.0

1.20

0.

46

13.5

14

2

72

0.

96

0.

15

0.

03

0.

84

0.

02

0.

01

Fa

jard

oRiv

er B

ar2D

2.0

0.56

0.

13

6.0

37

8

0.

96

0.

15

0.

03

0.

84

0.

01

0.

01

Fa

jard

oSm

all T

ribut

ary

2U3.

5

0.

58

0.28

3.

0

9

3

0.

94

-

0.

05

1.

00

0.

01

-

Fa

jard

oSm

all T

ribut

ary

1U3.

5

0.

51

0.19

3.

5

11

3

0.94

-

0.05

1.00

0.01

-

Faja

rdo

Smal

l Trib

utar

y1D

3.5

0.72

0.

34

5.0

19

5

0.

94

-

0.

05

1.

00

0.

01

-

Fa

jard

oSm

all T

ribut

ary

2D3.

5

1.

15

0.34

4.

0

19

9

0.94

-

0.05

1.00

0.01

-

Faja

rdo

Upp

er F

ajar

doTr

ib2U

6.0

0.81

0.

49

5.5

9

6

0.79

-

0.20

1.00

0.01

-

Faja

rdo

Upp

er F

ajar

doTr

ib1U

6.0

0.83

0.

43

5.0

10

8

0.

79

-

0.

20

1.

00

0.

01

-

Fa

jard

oU

pper

Faj

ardo

3U6.

0

1.

37

0.32

39

.0

273

196

0.

81

-

0.

19

1.

00

0.

01

-

Fa

jard

oU

pper

Faj

ardo

2U6.

0

2.

06

0.57

18

.0

175

184

0.

82

-

0.

18

1.

00

0.

01

-

Fa

jard

oU

pper

Faj

ardo

1U6.

0

2.

21

0.36

36

.0

364

411

0.

83

-

0.

17

1.

00

0.

01

-

M

amey

esAn

gelit

o Tr

ail

1U3.

0

2.

17

0.21

31

.0

496

251

0.

13

-

0.

79

1.

00

0.

09

0.

00

M

amey

esAn

gelit

o Tr

ail

1D3.

0

3.

45

0.30

54

.0

1,08

0

1,86

8

0.13

-

0.79

1.00

0.09

0.00

Mam

eyes

Ange

lito

Trai

l2D

3.0

---

---

---

---

---

0.13

-

0.79

1.00

0.09

0.00

Espi

ritu

Sant

oBa

ckw

ay S

choo

l2U

0.5

1.59

0.

57

43.5

55

2

20

4

0.96

0.19

-

0.65

0.04

0.16

Espi

ritu

Sant

oBa

ckw

ay S

choo

l1U

0.5

1.59

0.

34

78.0

91

3

36

7

0.96

0.20

-

0.64

0.04

0.16

Espi

ritu

Sant

oBa

ckw

ay S

choo

l1D

0.5

1.45

0.

51

35.0

35

0

17

7

0.96

0.20

-

0.64

0.04

0.16

Espi

ritu

Sant

oBa

ckw

ay S

choo

l2D

0.5

1.02

0.

40

61.5

48

9

19

9

0.96

0.20

-

0.64

0.04

0.16

Espi

ritu

Sant

oBe

hind

Bris

as2U

0.1

1.27

0.

33

27.3

29

2

13

9

0.88

0.25

-

0.57

0.12

0.17

App

endi

x. A

ddit

ion

al B

ioco

mpl

exit

y P

ool I

nfo

rmat

ion

255

Wat

ersh

edBi

ocom

plex

ity I

DCr

oss-

Sect

ion

IDD

rop

Pdep

max

Pdep

cv

Plen

gth

Pare

aPv

olum

eag

r d

agr

ufs

t d

fst

uur

b d

urb

u

Espi

ritu

Sant

oBe

hind

Bris

a s1U

0.1

0.81

0.

37

12.0

90

38

0.

87

0.

25

-

0.

57

0.

13

0.

17

Es

pirit

u Sa

nto

Behi

nd B

risas

BRID

GE

0.1

1.76

0.

47

42.5

63

1

44

1

0.87

0.25

-

0.57

0.14

0.17

Mam

eyes

Bisl

ey 3

2U3.

0

0.

60

0.46

3.

2

13

4

0.11

-

0.81

1.00

0.08

-

Mam

eyes

Bisl

ey 3

1U3.

0

0.

68

0.28

3.

2

13

5

0.11

-

0.81

1.00

0.08

-

Mam

eyes

Bisl

ey 3

1D3.

0

0.

75

0.49

3.

8

21

8

0.11

-

0.81

1.00

0.08

-

Mam

eyes

Bisl

ey 3

2D3.

0

0.

62

0.20

4.

5

9

3

0.

11

-

0.

81

1.

00

0.

08

-

Es

pirit

u Sa

nto

Dav

ila3U

2.0

0.83

--

-22

.3

125

58

0.84

0.18

-

0.73

0.16

0.09

Espi

ritu

Sant

oD

avila

2U2.

0

0.

44

0.33

9.

0

58

10

0.

84

0.

18

-

0.

73

0.

16

0.

09

Es

pirit

u Sa

nto

Dav

ila1U

2.0

0.59

0.

28

30.7

21

5

41

0.

86

0.

18

-

0.

73

0.

14

0.

09

Es

pirit

u Sa

nto

Dav

ila1D

2.0

0.47

0.

22

12.0

65

14

0.

87

0.

18

-

0.

73

0.

14

0.

09

Es

pirit

u Sa

nto

Dav

ila2D

2.0

0.39

0.

31

11.3

82

15

0.

88

0.

18

-

0.

73

0.

12

0.

09

M

amey

esEa

st P

eak

2U7.

0

0.

52

0.32

6.

3

20

19

0.

08

-

0.

87

1.

00

0.

05

-

M

amey

esEa

st P

eak

1U7.

0

0.

68

0.29

12

.5

36

25

0.08

-

0.87

1.00

0.05

-

Mam

eyes

East

Pea

k1D

7.0

0.55

0.

26

7.3

11

10

0.08

-

0.87

1.00

0.05

-

Mam

eyes

East

Pea

k2D

7.0

1.27

0.

27

3.0

13

17

0.08

-

0.87

1.00

0.05

-

Espi

ritu

Sant

oEl

Ver

de3U

2.0

1.03

--

-13

.8

128

36

0.81

0.04

-

0.95

0.19

0.01

Espi

ritu

Sant

oEl

Ver

de2U

2.0

1.79

0.

35

32.5

37

7

28

9

0.81

0.04

-

0.95

0.19

0.01

Espi

ritu

Sant

oEl

Ver

de1U

2.0

0.96

0.

39

20.6

39

3

13

1

0.82

0.04

-

0.95

0.18

0.02

Espi

ritu

Sant

oEl

Ver

de1D

2.0

1.95

0.

51

38.0

92

3

34

3

0.83

0.04

-

0.95

0.17

0.02

Espi

ritu

Sant

oEl

Ver

de2D

2.0

1.80

0.

19

24.0

35

4

17

2

0.83

0.04

-

0.94

0.17

0.02

Espi

ritu

Sant

oES

Rt.

3 P

aint

ball

BRID

GE

0.1

2.50

0.

37

200.

0

5,

000

2,

632

-

0.

15

-

0.

77

1.

00

0.

08

Es

pirit

u Sa

nto

ES W

ater

fall

3U9.

0

1.

25

0.41

8.

6

54

33

0.

46

-

0.

40

1.

00

0.

13

-

Es

pirit

u Sa

nto

ES W

ater

fall

2U9.

0

2.

62

0.18

12

.0

73

77

0.47

-

0.40

1.00

0.13

-

Espi

ritu

Sant

oES

Wat

erfa

ll1U

9.0

2.76

0.

41

27.0

38

1

45

5

0.47

-

0.40

1.00

0.13

-

Espi

ritu

Sant

oES

Wat

erfa

ll1D

9.0

3.10

0.

60

15.5

22

5

15

2

0.48

-

0.39

1.00

0.14

-

Espi

ritu

Sant

oJi

men

ez2U

0.5

1.19

0.

28

30.5

39

7

14

9

0.93

0.12

-

0.82

0.07

0.06

Espi

ritu

Sant

oJi

men

ez1U

0.5

1.00

0.

37

94.0

2,

155

49

5

0.93

0.12

-

0.82

0.07

0.06

Espi

ritu

Sant

oJi

men

ez1D

0.5

2.20

0.

51

55.0

1,

699

87

7

0.93

0.12

-

0.82

0.07

0.06

Espi

ritu

Sant

oJi

men

ez W

ater

fall

2U5.

0

1.

55

0.47

16

.2

94

22

0.44

-

0.52

1.00

0.05

0.00

Espi

ritu

Sant

oJi

men

ez W

ater

fall

1U5.

0

1.

00

0.69

17

.0

80

42

0.44

-

0.51

1.00

0.05

0.00

Espi

ritu

Sant

oJi

men

ez W

ater

fall

1D5.

0

0.

80

0.71

8.

0

26

5

0.45

-

0.51

1.00

0.05

0.00

Espi

ritu

Sant

oJi

men

ez W

ater

fall

2D5.

0

2.

60

0.69

8.

2

85

87

0.

46

-

0.

49

1.

00

0.

05

0.

00

M

amey

esJu

an D

iego

3U5.

0

1.

01

0.66

9.

2

83

32

0.

08

-

0.

86

1.

00

0.

06

-

M

amey

esJu

an D

iego

2U5.

0

1.

01

0.27

8.

9

98

60

0.

08

-

0.

86

1.

00

0.

06

-

M

amey

esJu

an D

iego

1U5.

0

0.

75

0.55

3.

9

25

10

0.

09

-

0.

86

1.

00

0.

06

-

M

amey

esJu

an D

iego

BRID

GE

5.0

0.73

0.

44

3.5

15

5

0.

09

-

0.

86

1.

00

0.

06

-

M

amey

esJu

an D

iego

1D5.

0

0.

77

0.44

5.

2

21

6

0.09

-

0.86

1.00

0.06

-

Mam

eyes

La C

oca

Trai

l2U

5.0

0.81

0.

38

8.5

51

21

0.10

-

0.83

1.00

0.07

-

Mam

eyes

La C

oca

Trai

l1U

5.0

0.68

0.

42

4.3

15

6

0.

10

-

0.

83

1.

00

0.

07

-

M

amey

esLa

Coc

a Tr

ail

1D5.

0

0.

87

0.36

4.

5

23

10

0.

10

-

0.

83

1.

00

0.

07

-

256

Wat

ersh

edBi

ocom

plex

ity I

DCr

oss-

Sect

ion

IDD

rop

Pdep

max

Pdep

cv

Plen

gth

Pare

aPv

olum

eag

r d

agr

ufs

t d

fst

uur

b d

urb

u

Mam

eye s

La C

oca

Trai

l2D

5.0

0.49

0.

23

3.0

14

3

0.

10

-

0.

83

1.

00

0.

07

-

M

amey

esLa

Lin

guen

te2U

5.0

1.48

0.

40

6.5

41

47

0.13

-

0.79

1.00

0.09

-

Mam

eyes

La L

ingu

ente

1U5.

0

0.

60

0.11

4.

2

16

5

0.13

-

0.79

1.00

0.09

-

Mam

eyes

La L

ingu

ente

1D5.

0

0.

56

0.25

4.

0

9

4

0.

13

-

0.

78

1.

00

0.

09

-

M

amey

esLa

Lin

guen

te2D

5.0

0.50

0.

33

5.2

20

5

0.

13

-

0.

78

1.

00

0.

09

-

M

amey

esLa

Maq

uina

2U4.

0

1.

12

0.56

7.

0

15

13

0.

14

-

0.

76

1.

00

0.

10

-

M

amey

esLa

Maq

uina

1U4.

0

0.

35

0.77

5.

6

13

2

0.14

-

0.76

1.00

0.10

-

Mam

eyes

La M

aqui

naBR

IDG

E4.

0

0.

14

---

2.5

10

1

0.

14

-

0.

76

1.

00

0.

10

-

M

amey

esLa

Maq

uina

1D4.

0

0.

52

0.94

3.

8

8

1

0.

14

-

0.

76

1.

00

0.

10

-

M

amey

esLa

Maq

uina

2D4.

0

0.

46

0.36

1.

9

5

1

0.

14

-

0.

76

1.

00

0.

10

-

M

amey

esLa

Min

a Fa

lls2U

7.0

1.02

0.

58

6.3

51

19

0.08

-

0.86

1.00

0.06

0.01

Mam

eyes

La M

ina

Falls

1U7.

0

2.

97

0.75

9.

5

96

71

0.

08

-

0.

86

1.

00

0.

06

0.

01

M

amey

esLa

Min

a Fa

lls1D

7.0

0.83

0.

47

8.7

47

23

0.08

-

0.86

1.00

0.06

0.01

Mam

eyes

La V

ega

2U (

new

)1.

0

1.

02

0.16

66

.0

931

574

0.

32

0.

05

0.

46

0.

94

0.

22

0.

01

M

amey

esLa

Veg

a2U

(ol

d)1.

0

0.

64

---

34.5

59

3

20

5

0.32

0.05

0.46

0.94

0.22

0.01

Mam

eyes

La V

ega

1U1.

0

1.

06

---

47.5

79

3

40

1

0.32

0.05

0.46

0.94

0.22

0.01

Mam

eyes

La V

ega

RO

AD1.

0

1.

87

0.46

35

.0

392

428

0.

35

0.

06

0.

41

0.

93

0.

24

0.

01

M

amey

esLa

Veg

a1D

1.0

0.83

--

-30

.0

351

146

0.

35

0.

06

0.

41

0.

93

0.

24

0.

01

M

amey

esLa

Veg

a2D

1.0

0.83

0.

32

115.

0

1,

944

1,

038

0.

39

0.

06

0.

35

0.

93

0.

26

0.

01

Es

pirit

u Sa

nto

Las

Tres

T (

1)2U

2.0

0.36

0.

51

4.5

17

2

0.

57

0.

29

0.

20

0.

63

0.

23

0.

08

Es

pirit

u Sa

nto

Las

Tres

T (

1)1U

2.0

0.68

0.

34

15.0

69

14

0.

57

0.

29

0.

20

0.

63

0.

23

0.

08

Es

pirit

u Sa

nto

Las

Tres

T (

1)1D

2.0

1.36

0.

30

13.0

11

3

81

0.

58

0.

28

0.

20

0.

63

0.

22

0.

09

Es

pirit

u Sa

nto

Las

Tres

T (

1)2D

2.0

0.33

0.

31

5.2

15

3

0.

58

0.

28

0.

20

0.

63

0.

22

0.

10

Es

pirit

u Sa

nto

Las

Tres

T (

2)2U

4.0

0.63

0.

23

8.4

27

11

0.52

0.37

0.21

0.58

0.27

0.05

Espi

ritu

Sant

oLa

s Tr

es T

(2)

1U4.

0

0.

29

0.19

7.

1

27

4

0.53

0.36

0.21

0.58

0.26

0.07

Espi

ritu

Sant

oLa

s Tr

es T

(2)

1D4.

0

0.

78

0.67

9.

0

63

28

0.

53

0.

36

0.

21

0.

58

0.

26

0.

07

Es

pirit

u Sa

nto

Las

Tres

T (

2)2D

4.0

0.59

0.

33

5.3

15

6

0.

53

0.

34

0.

21

0.

59

0.

26

0.

06

M

amey

esM

amey

es R

t. 3

1U0.

4

1.

65

0.26

65

.0

965

826

0.

15

0.

15

-

0.

79

0.

85

0.

07

M

amey

esM

amey

es R

t. 3

1D0.

4

1.

78

0.42

52

.0

832

944

0.

02

0.

15

-

0.

78

0.

98

0.

07

Es

pirit

u Sa

nto

Prie

ta3U

7.0

0.60

--

-1.

4

3

1

0.

50

-

0.

36

1.

00

0.

14

-

Es

pirit

u Sa

nto

Prie

ta2U

7.0

0.48

0.

46

4.0

7

1

0.50

-

0.36

1.00

0.14

-

Espi

ritu

Sant

oPr

ieta

1U7.

0

0.

57

0.16

2.

4

5

2

0.

50

-

0.

36

1.

00

0.

14

-

Es

pirit

u Sa

nto

Prie

ta1D

7.0

1.14

0.

28

6.8

48

39

0.50

-

0.36

1.00

0.14

-

Espi

ritu

Sant

oPr

ieta

2D7.

0

0.

87

0.45

8.

4

17

5

0.51

-

0.35

1.00

0.14

-

Mam

eyes

Puen

te R

oto

2U2.

5

2.

03

0.25

55

.0

556

388

0.

14

-

0.

77

1.

00

0.

09

0.

00

M

amey

esPu

ente

Rot

o1U

/BRID

GE

2.5

1.78

0.

35

63.0

1,

216

76

4

0.14

-

0.77

1.00

0.10

0.00

Mam

eyes

Puen

te R

oto

1D2.

5

1.

61

0.26

76

.0

524

360

0.

14

-

0.

76

1.

00

0.

10

0.

00

M

amey

esPu

ente

Rot

o2D

2.5

1.75

--

-65

.0

1,05

3

700

0.

14

-

0.

76

1.

00

0.

10

0.

00

M

amey

esQ

. Ano

n2U

1.3

0.64

0.

51

17.5

39

12

0.

72

0.

77

-

0.

19

0.

28

0.

04

M

amey

esQ

. Ano

n1U

/BRID

GE

1.3

1.26

0.

77

26.0

75

29

0.

73

0.

76

-

0.

19

0.

27

0.

05

257

Wat

ersh

edBi

ocom

plex

ity I

DCr

oss-

Sect

ion

IDD

rop

Pdep

max

Pdep

cv

Plen

gth

Pare

aPv

olum

eag

r d

agr

ufs

t d

fst

uur

b d

urb

u

Mam

eye s

Q. A

non

1D1.

3

2.

07

0.37

17

.0

100

89

0.73

0.75

-

0.20

0.27

0.05

Mam

eyes

Q. A

non

2D1.

3

0.

88

0.36

12

.0

32

15

0.73

0.75

-

0.20

0.27

0.05

Espi

ritu

Sant

oSo

nado

ra3U

7.0

1.71

--

-12

.8

86

74

0.49

-

0.37

1.00

0.14

-

Espi

ritu

Sant

oSo

nado

ra2U

7.0

2.29

0.

39

9.5

136

120

0.

49

-

0.

37

1.

00

0.

14

-

Es

pirit

u Sa

nto

Sona

dora

1U7.

0

2.

00

0.32

12

.0

95

91

0.50

-

0.36

1.00

0.14

-

Espi

ritu

Sant

oSo

nado

ra1D

7.0

1.13

0.

30

7.2

47

16

0.50

-

0.36

1.00

0.14

-

Espi

ritu

Sant

oSo

nado

ra2D

7.0

1.15

0.

45

8.5

67

19

0.50

-

0.36

1.00

0.14

-

Espi

ritu

Sant

oTo

ronj

a2U

7.0

0.47

0.

27

2.0

3

1

0.51

-

0.35

1.00

0.15

-

Espi

ritu

Sant

oTo

ronj

a1U

7.0

0.24

0.

17

3.1

5

0

0.51

-

0.34

1.00

0.15

-

Espi

ritu

Sant

oTo

ronj

a1D

7.0

0.71

0.

13

6.2

20

8

0.

51

-

0.

34

1.

00

0.

15

-

Es

pirit

u Sa

nto

Toro

nja

2D7.

0

0.

43

0.29

5.

8

24

4

0.52

-

0.34

1.00

0.15

-

258

INDEX active-channel, 5-7, 19, 32, 35, 50, 105-107, 109, 113-115, 122-123, 165, 167, 170, 172,

174, 181, 193, 216, 217

alluvium, 40-41, 100-101, 116, 159-160

Atya, 177, 180, 183, 192, 202-203

bankfull, 1-2, 5-7, 31-37, 41, 45, 50, 61-63, 65, 67, 70-71, 73-78, 89, 95-96, 105-106,

136-137, 166-167, 217

Caribbean, 39, 92, 163, 200

crabs, 149-150, 161, 163, 165, 175, 177, 189, 198

Digital Elevation Model (DEM), 6, 15, 17-21, 23, 106, 125, 167

downstream hydraulic geometry (DHG), 1, 2, 7, 33, 37, 57-58, 76, 88, 90, 92-94, 96,

107, 113-116, 132-133, 137, 155, 217

eels, 150, 155, 161-162, 177, 183, 185, 187, 198, 200

effective discharge, 31, 36-37, 50, 56, 58-63, 65, 67, 71, 73, 75-76, 78, 105, 106

fish, 3, 7, 8, 149-150, 152, 155, 161-162, 164-165, 168, 170, 175, 177, 183, 189, 198-201

flow-frequency, 6-7 31-33, 35-37, 50, 51, 55-56, 61-63, 65, 67, 70-73, 75-78, 86, 105,

217-218

Geographic Information Systems (GIS), 4-6, 16, 19, 21, 27, 106, 165, 170, 216

grain size, 1, 7, 10, 34, 57, 88-89, 93-96, 107-108, 116-117, 119-120, 122, 125-128, 130-

132, 134-137, 155, 167, 172, 174-175, 192, 194, 202, 204, 219

granodiorite, 40, 100-101, 103, 109, 112, 116, 119, 130-131, 159, 220

landscape evolution, 1, 7, 91, 96, 136, 219

259

landslides, 2, 7, 11, 40, 88, 90-91, 93, 95-96, 102-103, 119, 121-122, 131, 134-136, 204

longitudinal profiles, 1, 88, 91, 93, 106, 109-110, 112, 131, 136, 150, 154-155, 170,

183, 187, 217

Macrobrachium, 163, 165, 177, 180, 189, 192, 203

megaboulders, 107, 119, 121-122, 167, 174

metamorphic, 100, 109, 112, 119

pools, 7-8, 42, 104, 113, 116, 134, 149-153, 156-157, 159, 162, 164-165, 167, 170-175,

177-178, 180-182, 189, 192-196, 198, 201, 204

riffles, 104, 113, 159, 162

riparian vegetation, 31-32, 35-37, 42, 53, 60, 76, 217

River Continuum Concept (RCC), 150, 152-154, 200-201

sediment transport, 1-2, 31, 34, 36, 56-61, 75, 90, 94, 96, 122-123, 131, 134, 219

snails, 149, 155, 161, 163, 165, 189, 201

shear stress, 7, 57-60, 93, 95-96, 108, 122-123, 134, 136-137, 172, 174

stream power, 7, 10, 75, 88-89, 94-96, 108-109, 125, 127-130, 132-135, 137, 167, 170,

172, 174-175, 183, 192-193, 202

United States Geological Survey (USGS), 16, 22-24, 39, 41, 50, 51, 53, 57, 109

volcaniclastic, 40, 88, 100-101, 103, 109, 112, 116, 119, 131-132, 159, 220

water quality, 160-161, 164

Xiphocaris, 163, 177, 180, 183, 192, 202-203

260

LONGITUDINAL PATTERNS IN STREAM CHANNEL GEOMORPHOLOGY …€¦ · LONGITUDINAL PATTERNS IN STREAM...

Documents

Transcript of LONGITUDINAL PATTERNS IN STREAM CHANNEL GEOMORPHOLOGY …€¦ · LONGITUDINAL PATTERNS IN STREAM...