pure.au.dkpure.au.dk/portal/files/121439036/ordonez_GEB_2015.pdf.docx · Web viewFour...
Transcript of pure.au.dkpure.au.dk/portal/files/121439036/ordonez_GEB_2015.pdf.docx · Web viewFour...
Ordonez & Svenning – Glacial-interglacial legacies in functional diversity
Title: Geographic patterns in the ratio of realized to potential functional diversity are linked
to glacial-interglacial climate stability and accessibility
Running title: Glacial-interglacial legacies in functional diversity
Authors: Alejandro Ordonez1 and Jens-Christian Svenning1
1Section for Ecoinformatics and Biodiversity, Department of Bioscience, Aarhus University,
Ny Munkegade 114, DK-8000 Aarhus C, Denmark
Corresponding author: Alejandro Ordonez. Section for Ecoinformatics and Biodiversity,
Department of Bioscience, Aarhus University, Ny Munkegade 114, DK-8000 Aarhus C.,
Denmark. Email: [email protected]
Manuscript types: Research papers
Number of words in the abstract: 298/300
Number of words in the manuscript: 5233/5000 (Introduction through the Biosketch,
excluding abstract, references and figure legends)
Number of references: 49/50
Number of figures and tables: 4 Figures and 3 Tables [Total 7/8]
Author contributions. A.O. and J-C.S. developed the idea, analyzed the results and
contributed equally to writing the manuscript. A.O. compiled the datasets, developed and
performed the statistical modeling.
Key words: climatic anomalies, disequilibrium dynamics, functional traits, historical effects,
paleoclimate, time lags, Europe, Climate velocity, Last Glacial Maximum, functional traits
1
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
Ordonez & Svenning – Glacial-interglacial legacies in functional diversity
ABSTRACT
Aim: Late Quaternary climate change can be an important determinant of large-scale species
richness patterns, but it is currently unknown if its effects also extend to functional diversity
(FD) and thus potentially ecosystem functioning. This study assessed whether deficits in
contemporary FD relative to expectations based on contemporary environmental regions and
species richness, are associated with glacial-interglacial climate stability and accessibility
from glacial refugia.
Location: Europe
Methods: Atlas data on distributions of plant species and functional trait information were
used to estimate functional richness (FRich) and dispersion (FDisp) for ~50×50 km grid cells
across Europe. Maximum expected FD was calculated using a both a quantile regression and
a null-model approach, with species richness and contemporary environmental regions as
constraints. Then, the proportion of the potential FD present in a grid cell – the realized-to-
potential FD ratio (R/P) – was estimated. Spatial autoregressive modeling with information-
theoretic multi-model selection was used to estimate the explanatory importance and
predictive ability of glacial-interglacial climate instability and accessibility to recolonization
from glacial refugia for R/P ratios.
Results: Minimum and median R/P ratios were 58% and 74% for FRich and 82% and 93% for
FDisp, showing that contemporary realized FD is often lower than expectations from species
richness and contemporary environmental regions. This deficit is significantly associated
with glacial-interglacial Quaternary climate change (FRich-R2: 0.52; FDisp-R2: 0.41). FRich and
FDisp R/P ratios high climatic instability (fast temperature and precipitation velocities) and
proximity to the major glacial temperate refuge regions (high accessibility).
2
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
Ordonez & Svenning – Glacial-interglacial legacies in functional diversity
Main conclusions: Realized macro-scale FD patterns appear to be in moderate
disequilibrium with contemporary richness and environmental regions. Importantly, small
R/P ratios are associated with low glacial-interglacial climatic instability and proximity to
glacial refugia. These findings suggest that future climate change may elicit long-term
functional disequilibria and thus also have long-term consequences for ecosystem
functioning.
INTRODUCTION
Studies assessing the effects of climate on diversity have focused on evaluating how
contemporary or historical conditions shape species ranges (Svenning & Skov, 2007b;
Normand et al., 2011; Boucher-Lalonde et al., 2014) and species richness (Svenning & Skov,
2007a; Svenning et al., 2010; Boucher-Lalonde et al., 2014). A largely unexplored dimension
of the climate-diversity relation is the effect of historical climate on functional diversity
across large regions, despite the importance of functional diversity for community and
ecosystem functioning (Chapin III et al., 2000; Hooper et al., 2005). For the purpose of
clarity, we consider functional diversity (FD hereafter) as the overall range as well as the
dispersion of trait values in a region.
Evidence indicates that paleoclimatic changes have filtered regional species pools based on
climate-related traits (Brodribb & Hill, 2004; Svenning et al., 2010) and shaped dispersal trait
distribution patterns within and among continents (Normand et al., 2011; Kissling et al.,
2012), suggesting that there could also be effects on FD. Additionally, the relation between
species richness and the range in functional trait values (Cornwell et al., 2006; Laliberté &
Legendre, 2010) indicate that historical effects on species richness are likely, just for this
reason, to influence FD and any other correlate of richness. However, historical legacies in
species richness can be independent of past changes in species composition (Algar et al.,
3
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
Ordonez & Svenning – Glacial-interglacial legacies in functional diversity
2011) and thereof break the richness-FD relation given that changes in the number of species
would not result in changes of the community mean, range or dispersion of trait values.
Either way, the existence and nature of long-lasting paleoclimatic effects on FD would have
significant implications for what to expect in terms of species, community and ecosystem
responses to future climatic changes (Chapin III et al., 2000; Hooper et al., 2005) and for the
development of traits-based approaches to forecast global vegetation functioning (Van
Bodegom et al., 2012).
Contemporary biotic, environmental and anthropogenic disturbances can affect the FD of a
community (Mouillot et al., 2013), restricting the occupied trait space (lower richness or
range) and reducing the deviation of species trait values from the community mean (lower
dispersion or variation). By analogy, paleoclimatic disturbances such as glaciations should
have also lowered the realized FD from its potential value based on contemporary
environmental regions and species’ richness. Assessing to what extent realized FD values
deviate from their potential maxima given current regional environment and species richness,
and the degree to which these deviations relate to historical disturbances offers a way to
quantify historical legacies in contemporary FD patterns.
We combine species’ distribution data with high-resolution contemporary and historical
environmental information to evaluate whether late-Quaternary climate has left a lasting
imprint on contemporary patterns of FD. Using the European flora as a study system and
building on Svenning et al. (2010)’s approach for species richness, we evaluated the FD
realized-to-potential ratio (R/P) of angiosperms in Europe as a way to determine the extent to
which historical climate has determined FD in the study region. We specifically evaluated the
idea that R/P ratios may be associated with late-Quaternary glacial-interglacial climate
change (measured as climate change velocity since the Last Glacial Maximum (LGM) ~
4
68
69
70
71
72
73
74
75
76
77
78
79
80
81
82
83
84
85
86
87
88
89
90
91
Ordonez & Svenning – Glacial-interglacial legacies in functional diversity
21,000 years BP; Sandel et al., 2011) and/or accessibility to postglacial re-colonization from
glacial refugia (Normand et al., 2011) .
We expect that R/P ratios will decrease with faster climate velocities and larger distances to
glacial refugia. Our working hypothesis is that limited extinction pruning of the species pool
in climatically stable areas (slow velocity and high accessibility) promote species coexistence
and facilitate long-term species survival (Sandel et al., 2011), and allows higher rates of
immigration in the proximity of highly accessible areas (Normand et al., 2011). This in turn
would result in a more comprehensive sampling of the surviving regional trait pool
(increasing richness or range). It would also reduce functional dispersion, as species would
have more time for competition-driven filtering resulting in an even spacing among co-
occurring taxa and a loss of species with environmentally marginal trait combinations. In the
light of our findings, we also discuss the possible effects of historical factors on FD diversity
patterns in contemporary and future ecosystem functioning.
METHODS
Distribution data. We focused on the European flora because its current ecology and
historical distribution are relatively well understood (Ellenberg, 1988; Bennett et al., 1991),
making it an ideal system for critical interpretation of the historical legacies of glacial-
interglacial changes in macro-scale diversity patterns. We used plant species distribution data
from Atlas Florae Europaeae (Jalas & Suominen, 1994-1999), which maps the distribution of
European flora on an equal-area mapping unit of c. ~50km×50km (AFE grid cells). All cells
in the former Soviet Union were excluded due to the heterogeneous and sometimes limited
sampling effort in this region. All taxa below the species level were collapsed to the
corresponding species. The AFE covers approximately 20% of the plant species in Europe,
completely covering the clades treated so far. We only included angiosperms (n=3904
5
92
93
94
95
96
97
98
99
100
101
102
103
104
105
106
107
108
109
110
111
112
113
114
115
Ordonez & Svenning – Glacial-interglacial legacies in functional diversity
species) in this study, excluding ferns, fern allies and gymnosperms to avoid possible
divergent effects on FD by including these phylogenetically distant groups.
Functional diversity estimation. Four eco-morphological traits associated to important plant
ecological strategies (Westoby et al., 2002; Chave et al., 2009) and the way plants respond to
environmental factors such as resources and disturbances (Lavorel & Garnier, 2002) were
used to estimate FD: specific leaf area (SLA, cm2*g-1), seed weight (SWT, mg), maximum
stem height (Hmax, m) and stem tissue density (WD, kg*m-3). SLA was used to represent the
trade-off between construction costs and carbon fixation (leaf economics spectrum; Wright et
al., 2004). SWT was used to represent a trade-off between seed size and seed production
(Moles & Westoby, 2006). Hmax was used to indicate the adult light niche (Moles et al.,
2009). WD was used to represent the trade-off between volumetric growth rates and mortality
rates or construction costs (wood economics spectrum; Chave et al., 2009). Trait values for
each taxon were initially determined using multiple published sources (Appendix S1).
Remaining missing values (SLA=80%, Hmax=69% SWT=54% and WD=15%) were filled
using a Multivariate Imputation Chained Equations approach (Buuren & Groothuis-
Oudshoorn, 2011). Before the analyses, all traits were log-transformed (as they were log-
normally distributed) and standardized (mean=0 and SD=1) to ensure that trait contribution to
FD metrics is scale independent.
We estimated two complementary FD metrics to describe grid level functional composition:
richness (Villéger et al., 2008) and dispersion (Laliberté & Legendre, 2010). Functional
richness (FRich) measures the range of the trait spectrum of the species pool based on the
multivariate trait space filled (volume) by the community as described by a multivariate
convex hull. In comparison, functional dispersion (FDisp) measures the spread of functional
types with respect to the functional space local center. While the FRich provided a measure of
the niche space occupied by co-occurring species, FDisp measured niche segregation among
6
116
117
118
119
120
121
122
123
124
125
126
127
128
129
130
131
132
133
134
135
136
137
138
139
140
Ordonez & Svenning – Glacial-interglacial legacies in functional diversity
coexisting species. We also evaluated these moments for individual traits, defining trait
richness as the trait value range and functional dispersion as the standard deviation of the trait
values.
Realized-to-potential functional diversity ratio determinants. FD and species richness
(number of species per AFE grid cell) showed a significant relation (Fig. 1), as indicated by
Pearson correlations with Dutilleul et al. (1993) geographically effective degrees of freedom
(FRich-ρ=0.75 and FDisp-ρ=-0.16). Furthermore, the relation between species richness and FRich
or FDisp was consistent across environmentally distinct regions (Fig. 1). For this reason, we
used a combination of species richness and contemporary environmental regions to calculate
the expected maximum FD.
Two distinct methods are used to estimate the maximum potential FD based on species
richness and contemporary environmental regions for both multivariate and single traits:
Quantile regression and a null-community ensemble approach. These estimates represent the
expected FRich or FDisp if only contemporary species richness and regional environmental
conditions constrained FD, i.e., in the absence of historical influences. Based on this we
calculated realized-to-potential ratios in FD (R/P ratio) to represent how much of the possible
functional space and dispersion is actually currently realized.
First, we used an additive quantile regression smoothing approach with species richness and
Metzger et al. (2005)’s environmental regions of Europe as predictors. This approach
provided a regionally environmentally constrained estimate of FD on the basis of species
richness per AFE grid cell. Quantile regressions were fitted to estimate the 99% percentile
(τ=0.99) as it provides a conservative assessment of the expected maximum potential FD
limits due to the number of species and broad-scale cotemporary environmental conditions.
7
141
142
143
144
145
146
147
148
149
150
151
152
153
154
155
156
157
158
159
160
161
162
163
Ordonez & Svenning – Glacial-interglacial legacies in functional diversity
After fitting this quantile regression model, we then projected it back to the study area to
estimate the maximum potential FD per AFE grid cell.
Second, we constructed 1000 null communities for each AFE grid cell by randomizing the
occurrence matrix, maintaining the original species richness, while restricting the pool of
species to be sampled to those within the same Metzger et al. (2005) environmental region.
This approach assumes that each species has an equal chance of occurring within each grid
cell contained within a given environmental region, so a deviation from this expectation can
be attributed to other factors beyond contemporary effects of species richness and regional
environmental conditions. We then proceeded to estimate the maximum FRich and FDisp for
each grid cell null community to get an estimate of highest potential FD per AFE grid cell.
Statistical analyses. We used spatial regressions with information-theoretic multi-model
selection to test the association between R/P ratios and historical predictors. Two types of
complementary variables (Appendix S1) were used to measure historical climatic instability
and postglacial accessibility to re-colonization: spatiotemporal rates of late-Quaternary
glacial-interglacial climate change (hereafter referred as climatic instability) and accessibility
to postglacial re-colonization from LGM refugia (hereafter referred only as accessibility).
While velocity represents the climatic instability of an AFE grid, accessibility represents how
easily each AFE grid could be reached by migration from glacial refugia.
Climatic instability was measured as mean annual temperature and precipitation change
velocities over the last 21,000 years (measured in km×decade-1 and caluclated as in Sandel et
al., 2011). Accessibility was measured as the inverse of the summed distance from each AFE
grid cell to all LGM refuge cells (measured in km-1 as in Svenning et al., 2010). Possible
glacial refugia were defined based on the minimum climatic requirements for cool-temperate
trees: GDD ≥800°C, mean temperature of the coldest month ≥-15°C, and summer
8
164
165
166
167
168
169
170
171
172
173
174
175
176
177
178
179
180
181
182
183
184
185
186
187
Ordonez & Svenning – Glacial-interglacial legacies in functional diversity
precipitation ≥50 mm sensu Leroy and Arpe (2007). It is important to highlight that our
accessibility estimate is qualitatively similar to climatic analog-based estimates of climatic
refugia (Ohlemüller et al., 2012) and highly correlated to the mean of Normand et al.
(2011)’s species-specific accessibility estimates for 1016 European species of herbaceous and
woody plants using hind-casted species distribution models (Spearman’s ρ=0.92; pDutilleul-
corrected=0.001). Furthermore, Normand et al. (2011)’s study also estimated that a large fraction
of European species were likely displaced into refugia wholly or primarily located south of
the 46°N latitude during the LGM, supporting our usage of this area as refuge regions.
Spatial autoregressive (SAR) error models (autoregressive process only in the error term)
were used to describe the statistical relation between R/P ratios and historical predictors.
These have been demonstrated to the most reliable statistical specification regardless of the
process generating the spatial autocorrelation or the model selection criteria (Kissling & Carl,
2008). Neighborhood distances in the neighborhood structure were established using the
semi-variogram range, based on the residuals of an ordinary least squares (OLS) multiple-
regression with full set of predictors. The SAR modeling was used to examine the association
between historical climate and R/P ratios for both multivariate and single traits across all of
Europe. Similar analyses were done for the regions north and south of the 46oN latitude
separately, as these represent areas where species richness and distributions patterns have
shown to be associated with either dispersal limitation (Svenning & Skov, 2007b; Normand
et al., 2011) or ecological tolerance (mainly cold and drought; Leroy & Arpe, 2007). We
expect that R/P ratios in regions north of the 46oN latitude are smaller and show a stronger
association with historical predictors than areas south of this boundary.
We estimated SAR models for all possible combinations of the three historical predictors and
used information-theoretic model selection and multi-model inference to provide robust
estimates of their importance (Burnham & Anderson, 2002). We estimated Akaike weights
9
188
189
190
191
192
193
194
195
196
197
198
199
200
201
202
203
204
205
206
207
208
209
210
211
212
Ordonez & Svenning – Glacial-interglacial legacies in functional diversity
(wAIC) calculated following Burnham and Anderson (2002). Relative importance of each
historical predictor was determined by summing the wAIC of all the models where the variable
of interest was included. Then, we determined the best fitting model, as the model with the
lowest ΔAIC (a measure of the probability that a given model was the best model for the
observed data within the set of candidate models). We also estimated model fit using
Nagelkerke’s (1991) pseudo-R2. Using this model, we further assessed which were the most
important historical predictor variables (contrast of regressions coefficients), considering both
their overall importance and their directional relation to the R/P ratios.
Contemporary climatic conditions could also modulate R/P ratios beyond their effects on
species richness. For this reason, all spatial regression analyses were also performed using
historical predictors and mean annual temperature and annual precipitation as predictors.
Contemporary predictors were selected due to it strong association diversity patterns in the
region (cf. Svenning & Skov, 2007b; Svenning & Skov, 2007a). These results are reported
in….
It is possible that the estimated effects of historical predictors on R/P ratios could simply
reflect spatial structure in the variables. We investigated this possibility using several
approaches. Firstly, we estimated how strong predictive power the historical variables confer
beyond the calibration region, using geographically independent validation. To do this, we
divided the study region in two (along the 10°E meridian) defining two training and testing
datasets (east training – west testing, and west training – east testing). Then we fitted a SAR
model on the training half of the data and then predicted the R/P ratios values in the testing
half. The predictive ability was determined as the Spearman correlation between predicted
and observed values in the test dataset. This approach allowed us to measure the predictive
ability of historical predictors and determine a possible causal link between R/P ratios and
historical climate.
10
213
214
215
216
217
218
219
220
221
222
223
224
225
226
227
228
229
230
231
232
233
234
235
236
237
Ordonez & Svenning – Glacial-interglacial legacies in functional diversity
Secondly, as we found intercept-only SAR models to have high explanatory power for R/P
ratios in both FRich andf FDisp (see Results), we compared the explanatory power of intercept-
only SAR models fitted to the observed R/P ratios to those fitted on predicted to R/P ratios
using an ordinary least squares (OLS) multiple regression model. The implemented OLS
model was based on all supported historical variables as predictors. The predicted values
from this model by definition only depend on the historical variables; analyzing their pattern
using an intercept-only SAR model therefore tells us to what extent such historically-
determined variation may be captured by a purely spatial SAR.. Additionally, we partitioned
the variation resulting from sets of three OLS models that used either historical predictors,
space or both as predictor variables (Borcard et al., 1992). With this, we assessed the
individual and joint contributions of historical predictors, space as explanatory factors of R/P
ratios.
RESULTS
Realized functional richness (FRich) and dispersion (FDisp) estimates based on multiple-traits
(Fig. 2A, B) and single traits (Supplementary material S2) exhibited strong geographic
patterns. Areas of high FD were observed in regions with distinct contemporary
environmental conditions, notably areas with high topographic variability (e.g., the Pyrenees
and Alps), and high seasonality and annual precipitation (Scandinavia). Importantly, FD
values were higher in areas south of 46°N, the northernmost limit of the major glacial refugia
for temperate species (indicated by the dashed black line in Fig. 2), notably in Southern
Europe (with parts of the Iberian Peninsula as the main exception for FRich). FDisp also
achieved high values in some coastal areas of formerly glaciated areas of Scandinavia.
Furthermore, realized FD was lower than expected FD based on a random sampling of
species within the same environmental region (significantly lower FRich and FDisp in 75% and
79% of the evaluated AFE grid cells).
11
238
239
240
241
242
243
244
245
246
247
248
249
250
251
252
253
254
255
256
257
258
259
260
261
262
Ordonez & Svenning – Glacial-interglacial legacies in functional diversity
These trends were much weaker for the potential FD, which exhibited a generalized smoother
south to north gradient in FRich, with just as high values in most of Central Europe as in
Southern Europe (Fig. 2C), and a generalized intermediate FDisp across all Europe with
pockets of high FDisp in the Mediterranean region (Iberian and Balkans Peninsula) and
Scandinavia (Fig. 2D). These quantile regressions based patterns were qualitatively
consistent with predictions based on null models (Appendix S2). For clarity, the primary
results discussed hereafter are based on quantile-regression based predictions of potential FD.
Results for the null model approach are presented as supplementary material (Appendix S2).
The R/P ratios ranged between 58% to 100% for FRich and 82% to 100% for FDisp (Fig. 2E, F).
Similarly, minimum R/P ratios in the range and variability of single-traits ranged between
30% and 80% (Appendix S2). Such relatively high R/P ratios are not unexpected given the
strong correlation between species richness on realized FD (see methods). Nevertheless, given
that realized FD levels were missing as much as 42 % or 18% of the potential maximum for
FRich and FDisp respectively, other factors besides richness and environmental regions may also
be of importance for current FD patterns.
Concerning the role of individual traits in driving R/P ratios, seed weight and wood density
were the traits with strongest support as possible determinants of FRich R/P ratios. Overall,
areas of large seed weight range deficit (lower seed sizes than expected) and stem density
range deficit (wood density is lower than expected) had the lowest FRich R/P ratios (SWT
R2NK: 0.51 WD R2
NK: 0.49). Meanwhile, stem height and seed weight were the traits with the
largest support as possible determinants of FDisp R/P ratios as areas of large deficits in stem
height (lower mean canopies than expected) and seed weight (lower seed sizes than expected)
trait variability had he lowest FDisp R/P ratios (Hmax R2NK: 0.597 SWT R2
NK: 0.58).
12
263
264
265
266
267
268
269
270
271
272
273
274
275
276
277
278
279
280
281
282
283
284
285
Ordonez & Svenning – Glacial-interglacial legacies in functional diversity
As for realized FD, the dominant geographic pattern in R/P ratios was a contrast between
Southern and Northern Europe, with significant lower values in the latter (Fig. 3).
Geographic patterns of R/P ratios show Southern (including the Pyrenees) and Central
Europe as the areas with the highest R/P ratios for FRich (Fig. 2C) and FDisp (Fig. 2F).
Meanwhile, regions covered by the large northern ice sheets during the LGM showed the
lowest R/P ratios for FRich and FDisp (areas inside the red outline in Fig. 2C and F). The
similarity between FRich and FDisp R/P ratios in northern Europe and Scandinavia was not
surprising given the correlation between both components (FRich- FDisp: ρ=0.17, pDutilleul-
corrected<0.001).
Considering each historical climatic variable separately (single variable models in Table 1),
the R/P ratios showed negative associations to climatic instability (R/P ratios decreased as
velocity increase; Fig. 4A-B and D-E) and positive associations to postglacial accessibility
(R/P ratios increased with proximity to the main temperate glacial region; Fig. 4 C, F). If
contemporary factors (mean annual temperature and precipitation) were also evaluated
(Appendix S3), R/P ratios showed a negative association with mean annual temperature in
the case of FRich; and a positive associations to annual precipitation and mean annual
temperature in the case of FDisp. These associations were consistent, but weaker for single-
trait-based FD metrics with the exception of a negative effect of accessibility on SLA
dispersion (Appendix S2 and S3). Furthermore, the direction of the relation between historic
predictors and R/P ratios was similar in areas north (blue lines; Fig. 4) and south (orange
lines; Fig. 4) of the 46°N major temperate refuge boundary. This was also the case for annual
precipitation, but not mean annual temperature, where relations in areas south of the 46°N
boundary were flat and non-significant (Appendix S3).
ANCOVAS indicated that areas north of the 46°N glacial limit of climatic refugia had lower
ratios than those south of this limit (see also Fig. 3 and Appendix S3) and that the relation
13
286
287
288
289
290
291
292
293
294
295
296
297
298
299
300
301
302
303
304
305
306
307
308
309
310
Ordonez & Svenning – Glacial-interglacial legacies in functional diversity
between R/P ratios and environmental factors differed between regions (Table 2), with the
FDisp relations with historical factors being noticeably weaker in Southern Europe (Fig. 4).
The different linear combinations of the three historical predictors produced R2NK greater or
equal to 0.5 for FRich or 0.4 FDisp (Table 1). These trends were consistent after including mean
annual temperature and annual precipitation as predictor variables (Appendix S3), showing
that including contemporary variables did not add predictor power. When summarized across
all models, statistical support (wAIC) for an effect of temperature instability was 1 and 0.999 in
the case of for FRich and FDisp, respectively, while that for accessibility was 0.935 of FRich and
0.999 for FDisp. The support for precipitation instability was much lower, with wAIC=0.656 for
FRich and wAIC=0.497 for FDisp. The importance of temperature instability and accessibility was
also observed after including mean annual temperature and annual precipitation as predictor
variables (Appendix S3). This was further supported by information-theoretic multi-model
inference as temperature instability wAIC was 1 for both FRich and FDisp, while accessibility wAIC
was 0.99 and 1 for FRich and FDisp respectively.
Information-theoretic multi-model inference showed differences in the most important
variables associated with R/P ratios for individual traits’ range and deviation (Appendix S2).
For trait ranges, temperature velocity highest relative importance score for SLA (wAIC=0.968),
SWT (wAIC=0.990) and Hmax (wAIC=1.0), while accessibility was the most important variable
for WD (wAIC=0.791). For trait dispersion, accessibility had the highest relative importance
score for SLA (wAIC=0.952), SWT (wAIC=1.0) and Hmax (wAIC=0.999), while precipitation
velocity was the most important variable for WD (wAIC=0.589).
For both FRich and FDisp a model explaining R/P ratios in terms of temperature velocity and
accessibility had the lowest ΔAIC and highest pseudo-R2 (Table 1). The model fit of this
model, measured as R2NK, was 0.521 for FRich or 0.413 FDisp (Table 3). Standardized
14
311
312
313
314
315
316
317
318
319
320
321
322
323
324
325
326
327
328
329
330
331
332
333
334
Ordonez & Svenning – Glacial-interglacial legacies in functional diversity
coefficients from the best model showed the importance of both temperature instability and
accessibility, as well as an increase in R/P ratios with increasing climatic instability (negative
coefficient for temperature and precipitation velocity; Table 3) and increasing accessibility to
postglacial re-colonization (positive coefficient for accessibility; Table 3). When
contemporary and historical variables were both included as predictors, the model with the
lowest ΔAIC and highest pseudo-R2 for FRich included the full set of predictors, while for FDisp
it included annual mean temperature, annual precipitation, temperature instability and
accessibility (Appendix S3). Furthermore, the importance of contemporary and historical
variables in these models were comparable in terms of standardized coefficients (Appendix
S3).
The SAR model based on temperature velocity and accessibility had a high predictive ability
of R/P ratios as indicated by geographically independent validation. Models trained east of
10°E longitude accounted for 39,8% and 44,2% of the variation in FRich and FDisp of areas west
of this longitude, respectively, and models trained west of 10°E longitude accounted for
56,5% and 51,3% of the variation in FRich and FDisp in areas east of this longitude.
An intercept-only SAR model was not supported by information-theoretic model selection
when compared to models including climatic instability and/or accessibility (FRich: wAIC=0 and
FDisp: wAIC=0; Table 1). Still, it had relatively high explanatory power, just 1.9% for FRich and
1.4% of FDisp lower than the most supported model, i.e., with all three predictors. This was
also the case after including mean annual temperature and annual precipitation as predictor
variables (Appendix S3); indicating that the unexplained variability cannot be accounted by
including spatially autocorrelated contemporary predictors. This likely reflects how the SAR
model was able to capture much of the historical relations in the spatial error component.
Evidence for this was the very high predictive ability of intercept-only SAR models fitted to
R/P ratio predicted values from an OLS multiple regression using the historical predictors
15
335
336
337
338
339
340
341
342
343
344
345
346
347
348
349
350
351
352
353
354
355
356
357
358
359
Ordonez & Svenning – Glacial-interglacial legacies in functional diversity
(R/P FRich R2NK= 0.89; R/P FDisp R2
NK= 0.94). Moreover, when observed R/P ratio explained
variation was partitioned using a OLS model into individual and shared components of
history and space, the shared contribution of both factors accounted for almost half of the
total explain variation (FRich: 15% and FDisp: 11%) and almost the same variability explained
by solely by history (FRich: 12% and FDisp: 10%) or space (FRich: 9% and FDisp: 8%) alone.
DISCUSSION
Our results suggest that plant assemblages in Europe often do not fully realize their potential
maximum FRich and FDisp, given contemporary regional environmental conditions and local
species richness, hence resulting in a “functional deficit” for regions across Europe. This
limited realization of the potential FD suggests that contemporary regional environment and
sampling of the regional species-pool are not the sole determinants of contemporary FD
patterns. This was further indicated by the importance of historical predictors as explanatory
variables of R/P ratios even after including contemporary environmental predictors.
Furthermore, we show that functional deficits in richness and dispersion are statistically
negatively related to Quaternary glacial-interglacial climate instability and positively related
to postglacial accessibility, suggesting functional disequilibria despite the last glaciation
ending ~11,500 years ago. These associations are consistent with the hypothesis that glacial
extinction dynamics and postglacial migration lags shown for European plants (Svenning &
Skov, 2007b; Normand et al., 2011) have an effects on plant FD and appear to supplement
the possible effects of random sampling from the current regional trait/species pool and local
habitat filtering by contemporary regional environmental factors.
Two possible factors might cause the estimated paleoclimate-related deficit on FD: 1) sorting
according to climatic tolerance by past climatic conditions (Svenning, 2003; Bhagwat &
Willis, 2008) and 2) postglacial migration lags (Normand et al., 2011) coupled to links
16
360
361
362
363
364
365
366
367
368
369
370
371
372
373
374
375
376
377
378
379
380
381
382
383
Ordonez & Svenning – Glacial-interglacial legacies in functional diversity
between dispersal ability or immigration potential and plant traits (Bhagwat & Willis, 2008;
Nogués-Bravo et al., 2014). Glacial cooling or direct glaciation are likely to have removed
particular taxa or functional groups from a given area in non-random fashion depending on
their traits (e.g., via links to cold and drought tolerance; Svenning, 2003). It might have also
caused changes in species and community mean trait values depending on climatic historical
effects (e.g., glaciations filters could make species more cold-adapted than expected from the
current regional environment alone, by lowering/increasing the mean of some trait axis; e.g.,
Bhagwat & Willis, 2008). In comparison, migration lags are likely non-random with respect
to traits (Normand et al., 2011; Nogués-Bravo et al., 2014) and could explain the failure to
colonize climatically suitable areas as glacial cover receded and climate warmed since the
LGM.
The above factors are known to also influence broad-scale species richness patterns in Europe
(Svenning & Skov, 2007a; Fløjgaard et al., 2011), but also elsewhere (Blach-Overgaard et
al., 2013; Rakotoarinivo et al., 2013). These historical legacies on species richness could
potentially permeate to FRich and FDisp via the relations between FD and species richness. For
example, the lower realized than potential species richness shown in the European tree flora
by Svenning et al. (2010) is likely to have imposed an additional, indirect historical constraint
to the size and dispersion of the realized functional space of areas covered by ice during the
LGM.
The observed patterns in deviations in FD from their current potential are at least partially
consistent with differential impacts of the Quaternary glacial cooling in different parts of
Europe. The small R/P ratios in boreal areas may reflect both the slow and non-random
colonization from climatic refugia in Central and/or Eastern Europe (Leroy & Arpe, 2007;
Bhagwat & Willis, 2008) and the glacial climatic filtering of nearby boreal refuge regions.
Meanwhile, the small R/P ratios in the center of the Iberian Peninsula may reflect a failure to
17
384
385
386
387
388
389
390
391
392
393
394
395
396
397
398
399
400
401
402
403
404
405
406
407
408
Ordonez & Svenning – Glacial-interglacial legacies in functional diversity
re-colonize after extirpation during Pleistocene glacial periods (Mijarra et al., 2009),
probably due to migratory barriers posed by the Pyrenees, the aridity in interior Iberian
Peninsula and the Mediterranean Sea (Svenning et al., 2008). In comparison, the otherwise
high R/P ratios in Southern Europe may be the result of more stable climate than in areas
further north and the high immigration possibilities into the region from LGM accessible
areas in the Italian and Balkan peninsulas (Medail & Diadema, 2009; Abellán & Svenning,
2014); resulting in the occurrence of relatively stable assemblages locally and/or regionally
over extended periods of time (Rodríguez, 2006) and may thereby reduced the effects on
LGM-present climatic variability on FD.
Intercept-only SAR models have high model fits, with only a small increase in explanatory
power when including only historical predictors or both contemporary and historical
predictors. This could be taken to indicate that associations with historical predictors, while
significant, are small enough to be negligible. However, several lines of evidence indicate
that such an interpretation would be misguided. Firstly, geographical independent validations
indicate that the estimated associations between R/P ratios and historical drivers have strong
predictive power beyond the calibration region. Perhaps more importantly, intercept-only
SAR models fitted not to the original R/P values, but to R/P values from OLS models with
historical predictors, have high explanatory power, meaning that the spatial error term in the
SAR models is capable of capturing much of a pattern determined by such spatially
structured variables. A reason for why much of the relations to the historical drivers are
captured in an intercept-only model is that they are broad-scale spatially structured variables,
as expected from the underlying mechanisms, such as large-scale migration dynamics (cf.
Bjorholm et al., 2006; Normand et al., 2011).
The estimated constraints on R/P ratios linked to glacial-interglacial climate instability and
accessibility suggest lags in climate tracking of >10,000 years. However, other spatially
18
409
410
411
412
413
414
415
416
417
418
419
420
421
422
423
424
425
426
427
428
429
430
431
432
433
Ordonez & Svenning – Glacial-interglacial legacies in functional diversity
structured contemporary variables (namely temperature and precipitation) account for some
of the observed variability in R/P ratios even after the effects via species richness have been
considered. This points to a high potential for vegetation dynamics in response to future
climate change to also involve substantial disequilibria (Svenning & Sandel, 2013). This
disequilibrium will not just be in species distributions and species richness, but also in FD,
with likely consequences for ecosystem functioning due to changes in trait composition. Such
changes would occur directly via filtering of unsuitable traits or its absence due to migration
lags; and indirectly, via the effects on FD resulting from changes in species richness that
would reshuffle the local trait composition. Establishing the strength and prevalence of links
between FD and ecosystem functioning would help to predict how future changes in the later
as a result of legacies of contemporary environmental-changes on FD.
CONCLUSION
Our results suggest that realized macro-scale FD are often below the potential set by
contemporary environmental regions and species richness. The geographic patterns in these
deviations can be linked to glacial-interglacial climate change and postglacial accessibility.
Hence, glacial climatic sorting and lagged postglacial re-colonization appear to not only
affect species distributions (Normand et al., 2011) and species richness (Svenning et al.,
2010) in Europe, but also affect FD. As a result, plant assemblages realize less of their
potential FD where glacial cooling was strong and postglacial accessibility has been low.
These results point to the potential for long-term disequilibria in FD and highlight the need to
incorporate historical constraints on FD and the mechanisms driving them into predictive
models of future vegetation-related ecosystem functioning. This finding is important, as such
disequilibrium effects are likely to be much stronger over the much shorter (50-200 years)
time scales typically considered for future climate change (Svenning & Sandel, 2013). Future
studies should aim to evaluate the mechanisms and timescales of potential lags to provide a
19
434
435
436
437
438
439
440
441
442
443
444
445
446
447
448
449
450
451
452
453
454
455
456
457
458
Ordonez & Svenning – Glacial-interglacial legacies in functional diversity
better understanding of the resistance and resilience of FD to future changes in climatic
conditions.
Acknowledgements. This article is a contribution of the HISTFUNC project (ERC Starting
Grant 310886 to J-C.S.). We thank the Atlas Florae Europaeae project for access to the
distribution data. The LEDA project and the Millennium seed database (Kew Botanical
garden) are acknowledged for contributing trait data. The authors thank David Curie, Jeremy
Kerr, Oliver Schweiger and Véronique Boucher-Lalonde for providing comments on earlier
versions of this manuscript.
Biosketch: Alejandro Ordonez and Jens-Christian Svenning are interested in understanding
how tolerance to past and current changes in climatic conditions, long-term migration lags
and lineage biogeography impose lasting legacies in species distribution and richness patterns
and shape contemporary functional diversity patterns and ecosystem functions.
References
Abellán, P. & Svenning, J.-C. (2014) Refugia within refugia – patterns in endemism and
genetic divergence are linked to Late Quaternary climate stability in the Iberian
Peninsula. Biological Journal of the Linnean Society, 113, 13-28.
Algar, A.C., Kerr, J.T. & Currie, D.J. (2011) Quantifying the importance of regional and
local filters for community trait structure in tropical and temperate zones. Ecology,
92, 903-914.
Bennett, K.D., Tzedakis, P.C. & Willis, K.J. (1991) Quaternary refugia of north European
trees. Journal of Biogeography, 18, 103-115.
Bhagwat, S.A. & Willis, K.J. (2008) Species persistence in northerly glacial refugia of
Europe: a matter of chance or biogeographical traits? Journal of Biogeography, 35,
464-482.
20
459
460
461
462
463
464
465
466
467
468
469
470
471472473
474
475
476
477
478
479
480
481
482
483
Ordonez & Svenning – Glacial-interglacial legacies in functional diversity
Bjorholm, S., Svenning, J.C., Baker, W.J., Skov, F. & Balslev, H. (2006) Historical legacies
in the geographical diversity patterns of New World palm (Arecaceae) subfamilies.
Botanical Journal of the Linnean Society, 151, 113-125.
Blach-Overgaard, A., Kissling, W.D., Dransfield, J., Balslev, H. & Svenning, J.C. (2013)
Multimillion-year climatic effects on palm species diversity in Africa. Ecology, 94,
2426-2435.
Borcard, D., Legendre, P. & Drapeau, P. (1992) Partialling out the spatial component of
ecological variation. Ecology, 73, 1045-1055.
Boucher-Lalonde, V., Kerr, J.T. & Currie, D.J. (2014) Does climate limit species richness by
limiting individual species’ ranges? Proceedings of the Royal Society B: Biological
Sciences, 281
Brodribb, T. & Hill, R.S. (2004) The rise and fall of the Podocarpaceae in Australia—a
physiological explanation. The Evolution of Plant Physiology, 381.
Burnham, K.P. & Anderson, D.R. (2002) Model selection and multi-model inference: a
practical information-theoretic approach. Springer.
Buuren, S. & Groothuis-Oudshoorn, K. (2011) MICE: Multivariate imputation by chained
equations in R. Journal of statistical software, 45
Chapin III, F.S., Zavaleta, E.S., Eviner, V.T., Naylor, R.L., Vitousek, P.M., Reynolds, H.L.,
Hooper, D.U., Lavorel, S., Sala, O.E., Hobbie, S.E., Mack, M.C. & Díaz, S. (2000)
Consequences of changing biodiversity. Nature, 405, 234-242.
Chave, J., Coomes, D., Jansen, S., Lewis, S.L., Swenson, N.G. & Zanne, A.E. (2009)
Towards a worldwide wood economics spectrum. Ecology Letters, 12, 351-366.
Cornwell, W.K., Schwilk, D.W. & Ackerly, D.D. (2006) A trait-based test for habitat
filtering: Convex hull volume. Ecology, 87, 1465-1471.
21
484
485
486
487
488
489
490
491
492
493
494
495
496
497
498
499
500
501
502
503
504
505
506
507
Ordonez & Svenning – Glacial-interglacial legacies in functional diversity
Dutilleul, P., Clifford, P., Richardson, S. & Hemon, D. (1993) Modifying the t test for
assessing the correlation between two spatial processes. Biometrics, 305-314.
Ellenberg, H. (1988) Vegetation ecology of central Europe. Cambridge University Press.
Fløjgaard, C., Normand, S., Skov, F. & Svenning, J.C. (2011) Deconstructing the mammal
species richness pattern in Europe - towards an understanding of the relative
importance of climate, biogeographic history, habitat heterogeneity and humans.
Global Ecology and Biogeography, 20, 218-230.
Hooper, D.U., Chapin, F.S., Ewel, J.J., Hector, A., Inchausti, P., Lavorel, S., Lawton, J.H.,
Lodge, D.M., Loreau, M., Naeem, S., Schmid, B., Setala, H., Symstad, A.J.,
Vandermeer, J. & Wardle, D.A. (2005) Effects of biodiversity on ecosystem
functioning: A consensus of current knowledge. Ecological Monographs, 75, 3-35.
Jalas, J. & Suominen, J. (1994-1999) Atlas Florae Europaeae. Distribution of vascular plants
in Europe, Vol. 1-12. Helsinki: Societas Biologica Fennica Vanamo, 1964-1999.
Kissling, W.D. & Carl, G. (2008) Spatial autocorrelation and the selection of simultaneous
autoregressive models. Global Ecology and Biogeography, 17, 59-71.
Kissling, W.D., Eiserhardt, W.L., Baker, W.J., Borchsenius, F., Couvreur, T.L.P., Balslev, H.
& Svenning, J.-C. (2012) Cenozoic imprints on the phylogenetic structure of palm
species assemblages worldwide. Proceedings of the National Academy of Sciences,
109, 7379-7384.
Laliberté, E. & Legendre, P. (2010) A distance-based framework for measuring functional
diversity from multiple traits. Ecology, 91, 299-305.
Lavorel, S. & Garnier, E. (2002) Predicting changes in community composition and
ecosystem functioning from plant traits: revisiting the Holy Grail. Functional
Ecology, 16, 545-556.
22
508
509
510
511
512
513
514
515
516
517
518
519
520
521
522
523
524
525
526
527
528
529
530
531
Ordonez & Svenning – Glacial-interglacial legacies in functional diversity
Leroy, S.A.G. & Arpe, K. (2007) Glacial refugia for summer-green trees in Europe and
south-west Asia as proposed by ECHAM3 time-slice atmospheric model simulations.
Journal of Biogeography, 34, 2115-2128.
Medail, F. & Diadema, K. (2009) Glacial refugia influence plant diversity patterns in the
Mediterranean Basin. Journal of Biogeography, 36, 1333-1345.
Metzger, M.J., Bunce, R.G.H., Jongman, R.H.G., Mucher, C.A. & Watkins, J.W. (2005) A
climatic stratification of the environment of Europe. Global Ecology and
Biogeography, 14, 549-563.
Mijarra, J.M.P., Barron, E., Manzaneque, F.G. & Morla, C. (2009) Floristic changes in the
Iberian Peninsula and Balearic Islands (south-west Europe) during the Cenozoic.
Journal of Biogeography, 36, 2025-2043.
Moles, A.T. & Westoby, M. (2006) Seed size and plant strategy across the whole life cycle.
Oikos, 113, 91-105.
Moles, A.T., Warton, D.I., Warman, L., Swenson, N.G., Laffan, S.W., Zanne, A.E., Pitman,
A., Hemmings, F.A. & Leishman, M.R. (2009) Global patterns in plant height.
Journal of Ecology, 97, 923-932.
Mouillot, D., Graham, N.A.J., Villeger, S., Mason, N.W.H. & Bellwood, D.R. (2013) A
functional approach reveals community responses to disturbances. Trends in Ecology
& Evolution, 28, 167-177.
Nagelkerke, N.J.D. (1991) A Note on a General Definition of the Coefficient of
Determination. Biometrika, 78, 691-692.
Nogués-Bravo, D., Pulido, F., Araújo, M.B., Diniz-Filho, J.A.F., García-Valdés, R.,
Kollmann, J., Svenning, J.-C., Valladares, F. & Zavala, M.A. (2014) Phenotypic
correlates of potential range size and range filling in European trees. Perspectives in
Plant Ecology, Evolution and Systematics, 16, 219-227.
23
532
533
534
535
536
537
538
539
540
541
542
543
544
545
546
547
548
549
550
551
552
553
554
555
556
Ordonez & Svenning – Glacial-interglacial legacies in functional diversity
Normand, S., Ricklefs, R.E., Skov, F., Bladt, J., Tackenberg, O. & Svenning, J.-C. (2011)
Postglacial migration supplements climate in determining plant species ranges in
Europe. Proceedings. Biological sciences / The Royal Society, 278, 3644-53.
Ohlemüller, R., Huntley, B., Normand, S. & Svenning, J.-C. (2012) Potential source and sink
locations for climate-driven species range shifts in Europe since the Last Glacial
Maximum. Global Ecology and Biogeography, 21, 152-163.
Rakotoarinivo, M., Blach-Overgaard, A., Baker, W.J., Dransfield, J., Moat, J. & Svenning, J.-
C. (2013) Palaeo-precipitation is a major determinant of palm species richness
patterns across Madagascar: a tropical biodiversity hotspot. Proceedings of the Royal
Society B: Biological Sciences, 280
Rodríguez, J. (2006) Structural continuity and multiple alternative stable States in Middle
Pleistocene European mammalian communities. Palaeogeography,
Palaeoclimatology, Palaeoecology, 239, 355-373.
Sandel, B., Arge, L., Dalsgaard, B., Davies, R.G., Gaston, K.J., Sutherland, W.J. & Svenning,
J.-C. (2011) The influence of late Quaternary climate-change velocity on species
endemism. Science, 334, 660-664.
Svenning, J.-C. (2003) Deterministic Plio-Pleistocene extinctions in the European cool-
temperate tree flora. Ecology Letters, 6, 646-653.
Svenning, J.-C. & Skov, F. (2007a) Ice age legacies in the geographical distribution of tree
species richness in Europe. Global Ecology and Biogeography, 16, 234-245.
Svenning, J.-C. & Skov, F. (2007b) Could the tree diversity pattern in Europe be generated
by postglacial dispersal limitation? Ecology Letters, 10, 453-460.
Svenning, J.-C. & Sandel, B. (2013) Disequilibrium vegetation dynamics under future
climate change. American Journal of Botany,
24
557
558
559
560
561
562
563
564
565
566
567
568
569
570
571
572
573
574
575
576
577
578
579
580
Ordonez & Svenning – Glacial-interglacial legacies in functional diversity
Svenning, J.-C., Normand, S. & Skov, F. (2008) Postglacial dispersal limitation of
widespread forest plant species in nemoral Europe. Ecography, 31, 316-326.
Svenning, J.-C., Fitzpatrick, M.C., Normand, S., Graham, C.H., Pearman, P.B., Iverson, L.R.
& Skov, F. (2010) Geography, topography, and history affect realized-to-potential
tree species richness patterns in Europe. Ecography, 33, 1070-1080.
Van Bodegom, P., Douma, J., Witte, J., Ordoñez, J., Bartholomeus, R. & Aerts, R. (2012)
Going beyond limitations of plant functional types when predicting global ecosystem–
atmosphere fluxes: exploring the merits of traits‐based approaches. Global Ecology
and Biogeography, 21, 625-636.
Villéger, S., Mason, N.W.H. & Mouillot, D. (2008) New multidimensional functional
diversity indices for a multifaceted framework in functional ecology. Ecology, 89,
2290-2301.
Westoby, M., Falster, D.S., Moles, A.T., Vesk, P.A. & Wright, I.J. (2002) Plant ecological
strategies: Some leading dimensions of variation between species. Annual Review of
Ecology and Systematics, 33, 125-159.
Wright, I.J., Reich, P.B., Westoby, M., Ackerly, D.D., Baruch, Z., Bongers, F., Cavender-
Bares, J., Chapin, T., Cornelissen, J.H.C., Diemer, M., Flexas, J., Garnier, E., Groom,
P.K., Gulias, J., Hikosaka, K., Lamont, B.B., Lee, T., Lee, W., Lusk, C., Midgley, J.J.,
Navas, M.L., Niinemets, U., Oleksyn, J., Osada, N., Poorter, H., Poot, P., Prior, L.,
Pyankov, V.I., Roumet, C., Thomas, S.C., Tjoelker, M.G., Veneklaas, E.J. & Villar,
R. (2004) The worldwide leaf economics spectrum. Nature, 428, 821-827.
25
581
582
583
584
585
586
587
588
589
590
591
592
593
594
595
596
597
598
599
600
601
602
603
Ordonez & Svenning – Glacial-interglacial legacies in functional diversity
Additional references to data sources used in this study may be found in Appendix S1 at
[URL]. Results to complementary analyses may be found in Appendix S2 at [URL].
Additional information.
Supplementary material S1. Description of the trait imputation procedure and the
estimation of the maximum potential functional diversity
Supplementary material S2. Null model and single-traits based estimations of realized,
potential, and association between realized/potential functional diversity ratios with historical
climatic predictors.
Supplementary material S3. Association between realized/potential functional diversity
ratios with both contemporary and historical climatic predictors.
26
604
605
606
607
608
609
610
611
612
613
614
Ordonez & Svenning – Glacial-interglacial legacies in functional diversity
Figure legends:
Figure 1. Changes in realized function al richness (top) and dispersion (bottom) as a function
of richness and broad scale environmental regions. Colors indicate Metzger et al. (2005)
environmental regions for Europe. Lines show the potential maximum FD estimated using a
quantile regression constrained on current regional environment and species richness (see
methods).
Figure 2. Realized (A, B) potential, (C, D) and realized/potential ratios (E, F) of plant
functional richness (left column) and dispersion (right column) per ~50×50km AFE grid cell
across Europe. Estimates of potential maximum functional diversity based on
environmentally constrained quantile regressions on species richness (see methods). The
thick-dashed black line indicates the boundary for the major Southern Europe glacial refuge
region (latitudes<46°N). Areas outlined in light grey solid lines indicate the maximum extent
(21000yrs ago ~LGM) of the Barents-Kara, British and Scandinavian ice-sheets.
Figure 3. Scatter plots of realized/potential ratios (R/P ratios, y-axis) of plant functional
richness (top row) and dispersion (bottom row) vs. late-Quaternary climate change in the
form of log-transformed annual mean temperature velocity (left column), log-transformed
total annual precipitation velocity (center column), and accessibility to postglacial re-
colonization (right column). Potential functional richness and dispersion estimated as in Fig,
1. Spatial autoregressive (SAR) model fits are presented for all Europe (black lines), and for
areas north (blue points and lines) and south (orange points and lines) of the boundary for the
major Southern Europe glacial refuge region (46°N). Model fits are presented in Table 1 and
A.
Figure 4. Scatter plots of realized/potential ratios (R/P ratios, y-axis) of plant functional
richness (top row) and dispersion (bottom row) vs. late-Quaternary climate change in the
27
615
616
617
618
619
620
621
622
623
624
625
626
627
628
629
630
631
632
633
634
635
636
637
638
Ordonez & Svenning – Glacial-interglacial legacies in functional diversity
form of log-transformed annual mean temperature velocity (left column), log-transformed
total annual precipitation velocity (center column), and accessibility to postglacial re-
colonization (right column). Potential functional richness and dispersion estimated as in Fig,
1. Spatial autoregressive (SAR) model fits are presented for all Europe (black lines), and for
areas north (blue points and lines) and south (orange points and lines) of the boundary for the
major Southern Europe glacial refuge region (46°N). Model fits are presented in Table 1 and
2.
28
639
640
641
642
643
644
645
646
Ordonez & Svenning – Glacial-interglacial legacies in functional diversity
Figure 1. Changes in realized function al richness (top) and dispersion (bottom) as a function of richness and environmental conditions. Colors indicate Metzger et al. (2005) environmental regions for Europe. Lines show the potential maximum FD estimated using a quantile regression constrained on current regional environment and species richness (see methods).
29
647648649650651
652
653
Ordonez & Svenning – Glacial-interglacial legacies in functional diversity
Figure 2. Realized (A, B) potential, (C, D) and realized/potential ratios (E, F) of plant functional richness (left column) and dispersion (right column) per ~50×50km AFE grid cell across Europe. Estimates of potential maximum functional diversity based on environmentally constrained quantile regressions on species richness (see methods). The thick-dashed black line indicates the boundary for the major Southern Europe glacial refuge region (latitudes<46°N). Areas outlined in light red solid lines indicate the maximum extent (21000yrs ago ~LGM) of the Barents-Kara, British and Scandinavian ice-sheets.
30
654655656657658659660
661662
663
Ordonez & Svenning – Glacial-interglacial legacies in functional diversity
Figure 3. Contrast between areas north (white) and south (grey) of the 46oN latitude, a limit representing the areas where diversity and distributions patterns have shown to be associated with either dispersal limitation (northern regions) or ecological tolerance (southern regions). Contras between regions show significant differences for functional richness (T(1476)=-20.76, p < 0.001) and dispersion (T(1346)=-17.95, p < 0.001) with lower values in the northern regions.
31
664665666667668
669
670
Ordonez & Svenning – Glacial-interglacial legacies in functional diversity
Figure 4. Scatter plots of realized/potential ratios (R/P ratios, y-axis) of plant functional richness (top row) and dispersion (bottom row) vs. late-Quaternary climate change in the form of log-transformed annual mean temperature velocity (left column), log-transformed total annual precipitation velocity (center column), and accessibility to postglacial re-colonization (right column). Potential functional richness and dispersion estimated as in Fig, 1. Spatial autoregressive (SAR) model fits are presented for all Europe (black lines), and for areas north (blue points and lines) and south (orange points and lines) of the boundary for the major Southern Europe glacial refuge region (46°N). Model fits are presented in Table 1 and 2.
32
671672673674675676677678679
680
681
Ordonez & Svenning – Glacial-interglacial legacies in functional diversity
Table 1. Comparison of spatial autoregressive (SAR) models explaining the realized/potential functional diversity ratios of plant functional diversity (functional richness and functional diversity), based on individual and all pair-wise combinations of three historical climate instability variables. Models order along increasing ΔAIC. Potential functional diversity was calculated using quantile regressions on current regional environment and species richness (see Methods). Predictor variables are annual mean temperature velocity (AMTVEL; in km/decade), Total annual precipitation velocity (TAPVEL; in km/decade), and accessibility to postglacial re-colonization (ACC; in km-1).
Functional RichnessΔAIC wAIC R2
NK
AMTVEL-ACC 0.000 0.497 0.521AMTVEL-TAPVEL-ACC 0.007 0.496 0.521AMTVEL 9.847 0.004 0.518AMTVEL-TAPVEL 10.084 0.003 0.519ACC 68.007 0.000 0.504TAPVEL-ACC 69.439 0.000 0.505Intercept only 78.326 0.000 0.502TAPVEL 79.809 0.000 0.502
Functional DispersionΔAIC wAIC R2
NK
AMTVEL-ACC 0.000 0.503 0.413AMTVEL-TAPVEL-ACC 0.024 0.497 0.413AMTVEL 21.132 0.000 0.406AMTVEL-TAPVEL 21.141 0.000 0.407TAPVEL-ACC 21.616 0.000 0.406ACC 27.645 0.000 0.404TAPVEL 48.248 0.000 0.398Intercept only 54.119 0.000 0.396The table shows model rescaled Akaike’s Information Criterion (∆AIC), Akaike weights (wAIC, probability that the best model includes a given variable or variable combinations), and model fit based on Nagelkerke’s pseudo-R2 (R2
NK)
33
682683684685686687688689
Ordonez & Svenning – Glacial-interglacial legacies in functional diversity
Table 2. Analysis of covariance comparing R/P ratios in functional richness and dispersion between areas north and south of the boundary for the major Southern Europe glacial refuge region (latitudes<46°N). R/P ratios estimated as the proportion of the potential functional diversity (calculated using a environmentally constrained quantile regressions on species richness, see methods) currently observed on each AFE grid cell. Regressions coefficients, standard errors (in parenthesis) and p-values presented.
Functional RichnessTemperature
VelocityPrecipitation
velocityAccessibility
Coefficient P Coefficient P Coefficient P
Intercept 0.839(0.007) ***
0.829(0.012) ***
0.766(0.02) ***
Historical climate -0.0112(0.00250) ***
-0.0023(0.0026) NS
0.0059(0.0016) ***
North-South 0.0273(0.0091) ***
0.0369(0.018) *
0.152(0.0494) ***
North-South and Historical climate
-0.0136(0.0037) ***
0.0012(0.0042) NS
-0.0077(0.0028) ***
R2NK 0.524 0.504 0.504
AIC -6045 -5959 -6081k 6 6 6
LL 3028 2985 3046Moran’s I -0.004 NS -0.001 NS -0.001 NS
Functional Dispersion
Intercept 0.922(0.003) ***
0.903(0.005) ***
0.886(0.007) ***
Historical climate -0.0066(0.0012) ***
-0.0053(0.0012) ***
0.003(0.0001) ***
North-South 0.0193(0.004) ***
0.0409(0.008) ***
0.0732(0.0188) ***
North-South and Historical climate
0.0021(0.0018) NS
0.0056(0.002) ***
-0.0037(0.0011) ***
R2NK 0.412 0.407 0.404
AIC -8845 -8827 -9005k 6 6 6
LL 4428 4420 4508Moran’s I -0.014 NS -0.012 NS -0.008 NS
Explained model variance based on Nagelkerke’s pseudo-R2 (R2NK), Akaike’s Information Criterion (AIC),
number of free parameters (k), model log-likelihood (LL), and Moran’s I and its significance level as defined by permutation tests (n=1000 permutations). Significance levels: ***P < 0.001; **P < 0.01; *P < 0.05. NS not significant
34
690691692693694695
696
Ordonez & Svenning – Glacial-interglacial legacies in functional diversity
Table 3. Best spatial autoregressive models (SAR) explaining European plants realized/potential differences (R/P ratios) in multivariate functional richness and dispersion, as a function of historical climatic conditions. Predictor variables are annual mean temperature velocity (AMTVEL; in km/decade), Total annual precipitation velocity (TAPVEL; in km/decade), and accessibility to postglacial re-colonization (ACC; in km-1). Potential functional diversity calculated using an environmentally constrained quantile regression on species richness (see methods). Best model selected from all possible models explaining functional richness and dispersion on the bases of all possible lineal combinations of historical climatic variables (see Table 1 for ΔAIC measures). Both regressions coefficients and standardized coefficients (in parentheses) are presented, and if not value is included the variable was not selected in the best model.
Functional Richness Functional DispersionCoefficient P Coefficient P
(Intercept) 0.798 *** 0.899 ***AMTVEL -0.017 -0.005 ***
(-0.255) *** (-0.175)TAPVEL
ACC 0.004 *** 0.002 ***(0.257) (0.315)
R2NK 0.521 0.413
AIC -6034 -8849k 5 5LL 3022 4430Moran’s I -0.004 NS -0.014 NSExplained model variance based on Nagelkerke’s pseudo-R2 (R2
NK), Akaike’s Information Criterion (AIC), number of free parameters (k), model log-likelihood (LL), and Moran’s I and its significance level as defined by permutation tests (n=1000 permutations). Significance levels: ***P < 0.001; **P < 0.01; *P < 0.05. NS not significant.
35
697698699700701702703704705706707
708