Differences in Soil and Leaf Litterfall Nitrogen Dynamicsfor Five Forest PlantationsStith T. Gower* and Yowhan Son

ABSTRACTVegetation can influence N cycling in forest soils; however, it is

difficult to isolate the positive feedback of vegetation on N cyclingbecause other factors are often not held constant. The objective ofthis study was to measure and compare leaf litterfall N and in situand laboratory N mineralization rates for 28-yr-old northern red oak(Quercus rubra L.), European larch (Larix decidua Miller), easternwhite pine (Pinus strobus L.), red pine (P. resinosa Alton), and Nor-way spruce [Picea abies (L.) Karst.) plantations on a similar soil insouthwestern Wisconsin. Average seasonal soil NO 5 and NHJ concen-trations (mg kg -1) were 3.9 and 3.4 for red oak; 7.7 and 5.8 forEuropean larch; 5.4 and 6.7 for white pine; 4.9 and 5.1 for red pine;and 5.2 and 6.2 for Norway spruce, respectively. Annual in situ netN mineralization in the top 20 cm of mineral soil differed significantly(P < 0.01) among species and ranged from 46 kg ha-'yr-'for Norwayspruce to 117 kg ha -' yr~' for European larch. Annual in situ nitri-fication differed significantly (P < 0.001) among species and com-prised from 42 (red oak) to 95% (European larch) of the total annualnet N mineralized. Laboratory net N mineralization rates also differedsignificantly among the five species. Average annual leaf litterfall Ncontent for a 2-yr period ranged from 26 kg ha-1 yr-1 for white pineto 40 kg ha-1 yr -' for Norway spruce but was not correlated to annualnet N mineralization. We suggest that leaf litterfall lignin/N may bean important positive feedback mechanism that influences N miner-alization.

ONE CHARACTERISTIC of vegetation that can ulti-mately influence soil processes is how long trees

S.T. Gower, Dep. of Forestry, University of Wisconsin, Madison,WI 53706; and Y. Son, Dep. of Plant, Soil and EnvironmentalStudies, Univ. of Maine, Orono, ME 04469. Received 29 Sept.1991. 'Corresponding author.

Published in Soil Sci. Soc. Am. J. 56:1959-1966 (1992).

retain their foliage (i.e., leaf lifespan) because thereare intrinsic relationships between biochemical char-acteristics of foliage and leaf lifespan. For example,foliage N concentration is inversely related to leaf life-span for tree species in natural forests (Gower andRichards, 1990; Reich et al., 1991). Also, Vogt et al.(1986) reviewed detritus production patterns for worldforests and concluded that litterfall N content was greaterfor deciduous than evergreen forests in warm and coldtemperate environments. Coley (1988) reported a strongpositive correlation between concentrations of im-mobile defense compounds, such as lignin and tan-nins, and leaf lifespan.

These relationships are important because litterquality exerts a strong feedback on ecosystem processes(Pastor et al., 1984). It is widely known that decom-position and nutrient mineralization occur more rap-idly in litter with a low C/N ratio (Edmonds, 1980;Berg and Staaf, 1981; Berg and Ekbohm, 1983; Pas-tor et al., 1984; Stohlgren, 1988). Horner et al. (1988)elucidated the potential influence that C-based sec-ondary metabolites such as lignin and tannins can haveon decomposition in terrestrial ecosystems. It is dif-ficult, however, to clearly demonstrate the importanceof plant processes on soil N dynamics unless the dif-ferent vegetation types occur on a homogeneous soiland in a similar climate. The physiologically basedrelationship between biochemical characteristics of fo-liage and leaf lifespan, and the positive relationshipbetween litter quality and net N mineralization, to-gether suggest that N mineralization may be inverselyrelated to leaf lifespan. Others have suggested, how-Abbreviations: LSD, least significant difference; ANOVA, analysisof variance; OLM, general linear model.

1960 SOIL SCI. SOC. AM. J., VOL. 56, NOVEMBER-DECEMBER 1992

ever, that species have a relatively minor impact onecosystem-level processes (O'Neill et al., 1986; Alienand Hoekstra, 1989).

To date, much of the research that has examined Nmineralization beneath different tree species is con-founded because the sites were selected on the basisof species without controlling for disturbance history,parent material, or natural adaptation of species tosites of differing fertility (Pastor et al., 1984; Nadel-hoffer et al., 1983, 1984; Matson, 1990). In fact,there are very few well-designed studies that have ex-amined the effects of trees (excluding N fixers) on soilprocesses (Binkley and Valentine, 1991). The objec-tive of this study was to examine the influence ofspecies with different leaf lifespahs on leaf litterfalland soil N dynamics. In this study we made use of a28-yr-old experimental forest containing replicated"blocks of five species with different leaf longevities;all blocks were established at the same time and, withineach block, all five species were established on a sim-ilar soil. This common garden experimental design(e.g., similar soil and climate conditions in each block)allowed us to examine the direct effect of tree specieson soil and litterfall N dynamics.

METHODSStudy Site and Experimental Design

The study was conducted at the Coulee ExperimentalForest (43 °52'N, 91 °51'W) in La Crosse County, Wis-consin. The forest was established in 1960 to study theeffects of landscape position and forest species compositionon runoff and soil erosion (Sartz, 1978). The topographyof the experimental forest is characterized by broad ridgesdissected by gently to strongly sloping valleys. Soils of thisregion have formed primarily from windblown silt (loess)overlying dolomite or sandstone.

The experimental design was a randomized block withfour replicate blocks; for some unknown reason, red oakwas not planted in one block. Two of the blocks are locatedon a ridgetop and the other two blocks are located on mid-and lower-slope positions. Maximum distance between thefour blocks is ~2 km. All species plots (experimental unit)within a block (replicate) were planted on the same soilseries. The ridgetop soils belong to the Fayette series whilethe soils for the other two blocks belong to the Dubuqueseries. The only difference between the two soil series isthe Fayette soil has a deeper loess cap than the Dubuquesoil; however, the loess cap for both soils exceeds the root-

ing depth. The Dubuque and Fayette soil are classified asfine-silty, mixed, mesic Typic Hapludalfs (Hole, 1976).

Within each block, all five species were planted on thesame soil, aspect, slope, and elevation. A mixed deciduousforest originally covered all four experimental blocks butthis forest was removed and the land was used for pasturefor an unknown period of time before the plantations wereestablished (A. Hagen, Wisconsin Dep. of Natural Re-sources, 1992, personal communication). Prior to planting,the ground for all four replicate blocks was prepared witha small tractor with a 2-m-wide angle dozer blade. Eachspecies was randomly assigned to a plot in each block.Bareroot seedlings from a nearby nursery were planted with2 by 2 m spacing in 45 by 45 m plots. Stand structuralcharacteristics for the replicated plantations are summarizedin Table 1.

To demonstrate a species effect on soil processes, it isimportant that the initial soil conditions within each blockbe as uniform as possible. Although we do not have his-torical data conclusively demonstrating that the soil wassimilar for the species plots within each block, we offerseveral lines of indirect evidence that supports this as-sumption. First, the soil series, aspect, slope, and elevationfor all species plots within a block are similar. Second, allfour blocks were subject to the same past land use practicesand plantation establishment methods. Third, the averagesoil texture and bulk density did not differ significantlyamong the five species (Son and Gbwer, 1992). Fourth, theaverage total Kjeldahl N concentration (based on 6—8 cores)for the upper 30 cm of soil differed by <0*02 g kg-1 be-tween the two blocks located on the ridgetop (A. Hagen,1992, personal communication). No standard errors wereprovided for the means for each block, but based on similaranalyses that we have conducted at this site (Son and Gower,1992) we are confident that these differences are not sta-tistically significant. Last, the random assignment of spe-cies to a plot within each block gives each species an equalchance of being assigned to some unknown gradient withineach block.

Nitrogen Availability and Soil AnalysisMany methods have been used to assess soil N miner-

alization but no one method is used universally and all onlyprovide an index of N availability (Binkley and Hart, 1989).We used the buried-bag incubation method (Eno, 1960) toestimate soil net N mineralization because it is widely usedand is sensitive to differences in soil microclimate (Binkleyand Hart, 1989; Binkley and Vitousek, 1989). This latterpoint is important because leaf area index, which stronglyinfluences radiation interception, differed by more than two-fold among the five species (Gower and Norman, 1991;

Table 1. Select stand structural characteristics for the 28-yr-old (1989 age) plantation-grown tree species.

Species

Red oak

European larch

White pine

Red pine

Norway spruce

Heightm

13.2

21.0(2.1)16.5(0.9)15.2(0.9)15.5(1.2)

MeanDBHf

cm9.3

(5.1)21.3(5.1)20.6(5.4)16.6(3.1)15.4(5.6)

Basal areairfha-1

11.5(2.5)38.8(0.5)44.1(5.5)44.9(1.9)25.7(2.9)

Treesno. ha"1

2033(422)1045

(60)1248(251)2032

(56)1917(153)

AbovegroundbiomassMgha-1

68.9(38.8)191.0

(3.5)174.5(28.3)163.8(11.3)168.5(17.9)

Forest floorbiomassMgha-1

8.7(0.5)37.5(2.6)33.0(3.5)42.8(6.0)24.7(3.0)

LABm'm-2

4.5(1.8)5.1

(0.1)7.4

(1.0)6.2

(0.4)10.2(1.8)

t DBH, diameter at breast height.$ LAI, leaf area index; data from Gower et al. (1992).§ One standard error of the mean is in parentheses, data based on 10 dominant or codominant trees from each plantation.

GOWER & SON: SOIL AND LEAF LITTERFALL NITROGEN DYNAMICS 1961

Gower et al., 1992). Furthermore, estimates of N miner-alization from the buried-bag method correlate well withaboveground productivity (Pastor et al., 1984; Nadelhofferet al., 1983,1985). We also conducted one 30-d laboratoryincubation (e.g., similar temperature and moisture condi-tions) during the summer to examine the direct influenceof litter quality on soil N mineralization.

In Situ MineralizationTwo soil cores (5-cm i.d. by 20-cm depth) were taken

at five random locations in each plantation for each of thefour blocks (three blocks for red oak). Sample dates were27 June, 17 Aug., 2 Oct., 19 Nov., 1990 and 13 Apr. and28 May, 1991. One core was placed in a gas-permeablepolyethylene bag, tied, placed back into its original holeand incubated for =45 d (114 d for winter incubation and30 d for the last incubation). The second core was returnedto the laboratory and initial NH$ and NO 3 concentrationand soil moisture were determined. All soil samples werestored at 4 °C until processed, which was usually within 24h and never >48 h.

In the laboratory, the forest floor and mineral soil fromeach core were thoroughly mixed and ~ 15 g (fresh weight)was placed in 100 mL of 2 M KC1, shaken for 1 h on anorbital shaker and extracted for an additional 23 h. Theextract was filtered and analyzed for NH$ and NO 3 con-centrations with a Lachat continuous-flow ion analyzer(Lachat, 1986, 1987). Soil moisture was determined froma second subsample; the subsample was weighed, dried at105 °C to a constant weight, and reweighed. Field-incu-bated samples were processed and extracted using the samemethod as described above. We discarded any incubatedsoil sample that had been disturbed (e.g., removed fromthe hole) or torn by animals. Except for two species duringone incubation period (see below), we generally did notdiscard more than 1 of the 20 cores for a species for eachincubation period. Three and four of the five in situ coresfor two blocks of Norway spruce and three of the five insitu cores for two replicate blocks of red pine were partiallyto completely removed from their hole during the first in-cubation in 1990. We did not include these cores in theanalysis and for this incubation period we weighted theaverage value for each red pine and Norway spruce plotvalue by the number of undisturbed cores. Net mineraliza-tion was calculated as the difference between NH$ +NO 3 concentrations of the incubated and initial soil sam-ples; net nitrification was calculated as the difference be-tween the NO 3 concentrations of the incubated and initialsoil samples.

Two soil cores per block were taken to determine bulkdensity. The soil was dried at 105 °C to a constant massand soil bulk density was determined as the ratio of themass of dry soil to the volume of the soil (Blake and Hartge,1986). The average corresponding bulk density for eachplantation and soil depth (20 cm) were used to calculate Nmineralization rate per unit area. Annual net N mineraliza-tion was calculated as the sum of net N mineralization forthe six incubation periods.

Laboratory IncubationTo determine if differences in soil N mineralization among

the five species were related to soil microclimate or litterquality, we conducted a 30-d laboratory incubation study.On 27 June 1990, a third subsample (15 g wet weight) fromeach of the five soil samples from each species plot in thefour blocks was placed in a 125-mL polyethylene bottle.The soil moisture of each sample was adjusted to fieldcapacity using deionized, distilled water. The mouth of thebottle was sealed with gas-permeable plastic and incubatedfor 30 d at 20 °C. We chose a soil moisture of field capacity

and a soil temperature of 20 °C to simulate optimal soilenvironmental conditions. Samples were extracted using themethod described above. Net N mineralization and nitrifi-cation calculations follow those described above.

Litterfall NitrogenLeaf litterfall was collected for 2 yr from the three rep-

licate blocks that contained all five species. Six to eight 40by 40 by 4 cm litter traps were randomly placed in eachplantation on 2 Apr. 1988. Litterfall was collected every 6wk during the growing season and less frequently duringthe winter because the litter traps were often covered bysnow. We were careful to collect the major pulse of foliagein the fall before the litter screens were covered by snow.Red oak and European larch leaf litterfall estimates basedon the litter traps were significantly lower than estimatesbased on site-specific regression equations that we devel-oped to estimate total foliage biomass. We speculate thatthe foliage of the two deciduous species was blown out ofthe litter traps, since leaf litterfall estimates for the threeevergreen species were similar to new foliage mass esti-mates. Therefore, we used our allometric equations to es-timate annual leaf litterfall mass for the two deciduousspecies. The equations used to estimate total foliage masswere based on 10 destructively sampled trees for each spe-cies; coefficients of determination (r2) for red oak and Eu-ropean larch equations were 0.848 and 0.907, respectively(Gower et al., 1992). The leaf litterfall estimates were ad-justed downward to account for the weight loss in foliagethat occurs prior to leaf senescence using species-specificrates that we observed (Son and Gower, 1991). Estimatesof leaf litterfall using this approach were within 14% ofleaf litterfall estimates based on eight 28-cm-diam. by 18-cm-high plastic litter traps that were placed in each decid-uous plantation during the second year of the study (Gower,1990, unpublished data). Average leaf lifespan was calcu-lated as total foliage mass/average leaf litterfall mass for a2-yr period (1988-1989) for the three evergreen conifersand as the number of months between leaf-out and leafsenescence for red oak and European larch. We used al-lometric equations that we developed for each of the threeevergreen species to estimate total foliage mass; r2 for whitepine, red pine, and Norway spruce equations were 0.944,0.947, and 0.960, respectively (Gower et al. 1992).

Litterfall was sorted by tissue type (foliage, woody, andmiscellaneous), dried at 70 °C to a constant mass, andweighed. Leaf litter was composited across collection datesfor each plot within each year and ground in a Wiley mill.Approximately 0.3 g of tissue was digested in a H2SO4-Li2SC)4 mixture (Parkinson and Alien, 1975) and analyzedcolorimetrically on a Lachat continuous-flow ion analyzerfor total N (Lachat, 1988). Annual leaf litterfall N contentfor each plot was calculated as the product of annual leaflitterfall mass and N concentration.

Statistical AnalysisFor all analyses, the species plot was considered to be

the experimental unit and the litter trap or mineralizationcores were subsamples (Hurlbert, 1984). Annual leaf lit-terfall characteristics (e.g., N concentration, mass, and Ncontent) for the 2-yr period were analyzed using a split-plotblock design with species and block as main effects andyear as a subeffect. If a significant (P < 0.05) species effectwas detected, we used an LSD test to separate means foreach year. A repeated-measures analysis of variance wasused to examine seasonal soil water and NH$ and NO 3pools and net N mineralization and nitrification patterns.Annual mineralization and nitrification were calculated asthe sum of the six incubation periods during the study. Thiscalculation resulted in a single estimate for these two

1962 SOIL SCI. SOC. AM. J., VOL. 56, NOVEMBER-DECEMBER 1992

processes; therefore, we used a two-way ANOVA with spe-cies and block as main effects and treated the nonexistentfourth block of red oak as missing data. Regression analysiswas used to examine the relation between soil N mineral-ization and litterfall N or soil N characteristics. Statisticalanalyses were performed using GLM and ANOVA proce-dures in SAS (SAS Institute, 1988).

RESULTS AND DISCUSSIONLitterfall Nitrogen Dynamics

Leaf litterfall N concentration ranged from 6.4 gkg-1 for red pine to 12.6 g kg-1 for European larch(Table 2). Except for Norway spruce, leaf litterfall Nconcentration was significantly greater (P < 0.05) in1988 than in 1989. Nitrogen concentration of leaf lit-terfall differed significantly (P < 0.05) among the fivespecies in 1989, but not in 1988 (Table 2). Analysisof variance suggested that the block effect was notsignificant (P = 0.15) for litterfall N concentration.

Leaf litterfall mass did not differ significantly (P <0.05) among the five species in 1988 or 1989. Analy-sis of variance results suggested the block effect wasnot significant for litterfall mass (P = 0.50) or Ncontent (P = 0.28). Averaged across the 2 yr, leaflitterfall mass ranged from 3210 for white pine to 4460kg ha-1 yr-1 for red pine (Table 2). Leaf litterfall Ncontent did not differ among species and the 2 yraverage ranged from 26 for white pine to 40 kg ha-1

yr-1 for Norway spruce and European larch (Table 2).Based on the split-plot analysis, year effect was notsignificant for leaf litterfall mass (P = 0.23) but wassignificant for leaf litterfall concentration (P < 0.001)and N content (P < 0.001).

Annual leaf litterfall N content for the plantationsin this study were similar to values reported for otherforests in the Lake States. Nadelhoffer et al. (1985)reported an annual leaf litterfall N content of 30, 23,21, 12, and 28 kg ha-1 yr-1 for natural red oak, redmaple, white pine, red pine, and white spruce forests,respectively, in southcentral Wisconsin. Perala andAlban (1982) reported an annual total (leaf + woody)litterfall N content of 38, 54, 45, and 50 kg ha-1 yr-1

for 40-yr-old aspen (Populus tremuloides Michaux),white spruce [P. glauca (Moench.) Voss], red pine,and jack pine (P. banksiana Lambert) plantations, re-spectively, planted on a loam soil, compared with 33,41, 34, and 41 kg ha"1 yr"1 for the same speciesplanted nearby on a sand.

In this study, leaf litterfall N content did not differsignificantly (P = 0.20) between deciduous and ev-ergreen plantations. Vogt et al. (1986) summarizedleaf litterfall N content for evergreen and deciduousforests for each climatic regime and concluded litter-fall N content was greater for deciduous than for ev-ergreen forests in warm and cold temperate climates.One possible explanation for the disparity between theresults from this study and Vogt et al. (1986) may bethat this study is not representative of nutrient cyclingpatterns for evergreen and deciduous forests. Our re-sults, however, are consistent with other studies thathave examined litterfall N content for evergreen anddeciduous tree species that were planted on similarsoils. For example, annual total litterfall N content(kg ha"1) was slightly less for aspen (36) than forthree evergreen conifers (44) in Minnesota (Perala andAlban, 1982). Binkley and Valentine (1991) also didnot find a significant difference in annual litterfall Ncontent (kg ha-1) among green ash (63), white pine(77), and Norway spruce (83) plantations growing ona similar soil in Connecticut. Miller (1984) also con-cluded that, when evergreen and deciduous forests areplanted on a similar soil, litterfall N content is similarfor these two groups. We speculate that the differencein results from common garden experimental designsvs. those based on literature reviews may be attributedto the fact that, in natural forests, deciduous tree spe-cies commonly occur on more fertile soils and cyclemore N in litterfall than evergreen forest species (Gosz,1981; Nadelhoffer et al., 1983, 1985).

Seasonal Soil Nitrogen DynamicsSeasonal soil moisture and NO 3 and NH^ concen-

tration patterns are shown in Fig. 1. Except for redoak, soil moisture generally did not differ significantly(P < 0.05) among species. Soil NO 3 and NH^ con-centrations periodically differed significantly (P < 0.05)among the five species. In general, soil NO 3 concen-tration was greatest in European larch and lowest inred oak plantations and soil NH^ concentration wassignificantly smaller in red oak than in the other fourspecies. Average soil NO 3 and NH^ concentrations(mg kg-1), respectively, were 3.9 and 3.4 for red oak,7.7 and 5.8 for European larch, 5.4 and 6.7 for whitepine, 4.9 and 5.1 for red pine, and 5.2 and 6.2 forNorway spruce. These seasonal average soil NOa and

^ concentrations are similar to values reported for

Table 2. Annual leaf litterfall mass, N concentration, and N content for the five species in 1988 and 1989.

Species

Red oak

European larch

White pine

Red pine

Norway spruce

Masskg ha-' yr-1

3720 tt(290)3660

(30)3130(360)4020(330)5480(400)

1988N

concentrationgkg-1

9.6(1-3)12.6(1.6)9.9

(0.8)8.0

(1.0)9.6

(0.4)

Ncontent- kg ha-'

36(12)46(4)31(4)32(6)53(4)

Massyr-1-

4190(860)3700

(40)3290(470)4910(690)2730(480)

1989N

concentrationgkg-'7.5 be

(0.7)8.9 ab

(1.4)6.6 be

(0.8)6.4 c

(0.3)10.2 a

(0.6)

Ncontent- kg ha-'

31(12)33(5)22(5)31(5)28(7)

Massyr-'-

3960(640)3680

(60)3210(270)4460(400)4100(680)

AverageN

concentrationgkg-1

8.6(0.8)10.7(1.2)8.2

(0.9)7.2

(0.6)9.9

(0.3)

Ncontent

kgha-'yr-'34(8)40(5)26(4)32(4)40(6)

t Species mean and one standard error (in parentheses) are based on the three blocks.t Values within a column without letters did not differ (P > 0.05) among species.

GOWER & SON: SOIL AND LEAF LITTERFALL NITROGEN DYNAMICS 1963

Red oakEuropean larch

ED White pineH Red pineD Norway spruce

L

June 27 Aug. 17 Oct. 2 Nov. 19 April 13 May 281990 1991

June 27 Aug. 17 Oct. 2 Nov. 19 April 13 May 281990 1991

June 27 Aug. 17 Oct. 21990

Nov. 19 April 13 May 281991

Fig. 1. Seasonal percentage of soil moisture and NO, andNH J concentration in the top 20 cm of forest floor plus soilfor 28-yr-old red oak, European larch, white pine, red pine,and Norway spruce plantations. Vertical bar is one standarderror for replicate plots.

other forest soils in the Lake States (Nadelhoffer etal., 1983).

Based on an adjusted Greenhouse-Geiser statisticfrom the repeated-measures analysis, we concludedthat nitrification was significantly affected by species(P < 0.01) and incubation date (P < 0.001) but therewas no species x incubation date interaction (P >0.05). In situ mineralization was significantly influ-enced by species (P < 0.05) and incubation date (P< 0.05) but the interaction between these two factorswas not significant (P > 0.05). Net N mineralizationwas variable within a species and during the year (Fig.2). Standard errors were generally < 20% of the mean,except in late fall when they were much greater. Pas-tor et al. (1984) reported similar standard errors fornatural forests in southern Wisconsin and also notedunexplainably greater standard errors for one or twoincubation periods. Net N mineralization for all fivespecies was greatest in the spring and late spring whensoil moisture was high (Fig. 1) and very low in thewinter. For example, average net N mineralization forall five species for summer (June-August), fall (Au-gust-October), late fall (October-November), winter(November-April), spring (April-May), and late spring(May-June) were 20.5, 23.1, 18.6, 1.6, 33.0, and

100

80

2J3 601.5 'enE •*

• Red oak^ European larchH White pine5 Red pineD Norway spruce

Jun-Aug Aug-Oct Oct-Nov Nov-Apr Apr-May May-Jun

Fig. 2. Seasonal patterns of net N mineralization for 28-yr-oldred oak, European larch, white pine, red pine, and Norwayspruce plantations. Net N mineralization for red oak in winteris 0.11 (± 0.05). Vertical bar is one standard error.

28.5 fjig kg-1 d-1, respectively. These results are con-sistent with other studies in southern Wisconsin; Na-delhoffer et al. (1983, 1984) and Pastor et al. (1984)both reported high rates in the spring and little or nonet N mineralization during the winter. In southwest-ern Wisconsin, forest soils commonly freeze to «1 meach winter (Sartz et al., 1977).

Net mineralization rates were low for red oak dur-ing the entire year except for the spring incubationperiod. Except during the early spring incubation (April-May 1991), European larch and white pine generallyhad greater net N mineralization rates than the otherthree species. The small number of samples collectedwithin a species plot might have contributed to thelack of significant differences for some dates. We didnot observe a pattern between seasonal net N miner-alization rates and species phenology or timing of lit-terfall senescence; however, Wedin and Tilman (1990)reported that seasonal nutrient uptake or belowgroundorganic matter turnover dynamics may explain sea-sonal N mineralization patterns.

Annual Nitrogen MineralizationWe observed a significant species effect on annual

net N mineralization (P < 0.01) and nitrification (P< 0.001). Analysis of variance results suggest theblock effect was significant for annual net nitrification(P < 0.01) and almost significant for annual net Nmineralization (P = 0.06). Annual net N mineraliza-tion in the top 20 cm of forest floor plus mineral soilfor Norway spruce, red pine, red oak, white pine, andEuropean larch were 46, 51, 55, 87 and 117 kg ha-1

yr-1, respectively (Fig. 3). In general, annual net Nmineralization rates estimated in this study were sim-ilar to values reported for other Wisconsin forestsgrowing on a similar soil. For example, based on pre-vious estimates of annual net N mineralization usingthe buried-bag approach, we calculated annual net Nmineralization rates of 53 to 111 kg ha-1 yr-1 fordeciduous and 32 to 80 kg ha-1 yr-1 for evergreenforests growing on Alfisols in southern Wisconsin(Nadelhoffer et al., 1983; Pastor et al., 1984). Weobserved a substantially lower annual net N mineral-ization rate (55 kg ha^yr-1) in the red oak plantation,however, compared with values of 94 to 120 kg ha-1

1964 SOIL SCI. SOC. AM. J., VOL. 56, NOVEMBER-DECEMBER 1992

nJCon

x3cgato

o05 Norway

spruce

Fig. 3. Annual net N mineralization and nitrification for redoak, European larch, white pine, red pine, and Norwayspruce. Annual net N mineralization or nitrification rateswith same letters do not differ (P > 0.05) among the fivespecies. Vertical bar is one standard error.

120

s-, 01i

«J

100

80

60

40

Y = -29.04 + 18.50 (X)r2 = 0.73

Europeanlarch

Red pine• Norway spruce

Table 3. Average NH J and NO j produced during a laboratoryincubation at 20 °C.

Species

Red oakEuropean larchWhite pineRed pineNorway spruce

NHJ

3.0 (1.0)tabt0.8 (0.8)b0.4 (0.9)b4.0 (0.7)ab5.4 (l.l)a

NOjmg kg-1 soil (30 d)-1

1.2 (0.2)c10.2 (3.0)ab15.2 (3.9)a2.2 (2.1)c4.2 (2.5)b

NHJ + NO,

4.2 (1.2)c11.0 (2.3)ab15.6 (3.8)a6.2 (1.8)bc9.6 (2.6)bc

t The number in parentheses is one standard error of the mean.| Average values in the same column followed by the same letter are not

significantly different.

120

c1Z

100

80

60

40

•^Europeanlarch

Y = 124.0 - 15.1 (X)r2 = 0.90

Whitepine

Norwayspruce

0 1 2 3 4 5 6

Leaf longevity (years)Fig. 5. Relation of annual net N mineralization to leaf longevity

for European larch, white pine, red pine, and Norway spruce.Red oak was not included in the analysis for reasons providedin the text.

Average initial NO 1 (mg kg )jFig. 4. Relation of annual net N mineralization to average

initial NO 3 content for the top 20 cm of soil for red oak,European larch, white pine, red pine, and Norway spruce.

yr-1 for natural forests dominated by black, white,and red oak (Quercus velutina Lam., Q. alba L., andQ. rubra L., respectively) in Wisconsin (Nadelhofferet al., 1983, 1984).

Annual net nitrification differed significantly (P <0.001) among the five species and ranged from 23 kgha-1 yr-1 for red oak and Norway spruce to 111 kgha-1 yr-1 for European larch. Averaged across thefive species, 65% of the total annual net N mineralizedwas nitrified. Nitrification comprised 61 to 84% ofannual net mineralization for natural forests growingon Alfisols in southern Wisconsin (Pastor et al., 1984;Nadelhoffer et al., 1984). European larch had thehighest nitrification (95%) followed by white pine, redpine, Norway spruce, and red oak. Based on the fiveplantation species in this study, we observed a signif-icant positive correlation (r2 = 0.78, P = 0.05) be-tween annual net N mineralization and average initialNO 3 concentration (Fig. 4).

The two-fold difference in annual net N mineral-ization and nitrification rates for five species plantedon the same soil suggests that vegetation influencedsoil N dynamics. Gosz (1981) suggested annual net Nmineralization rates range from 50 to 100 kg ha"1 yr-1

for conifer forests and 100 to 300 kg ha-1 yr"1 for

deciduous forests. In this study, annual net N miner-alization averaged 85 and 66 kg ha-1 yr-1 for decid-uous and evergreen species, respectively, but was notstatistically significant (P > 0.05). The nonsignificantdifference in annual litterfall N content, yet two-folddifference in annual net N mineralization, among thefive species raises the interesting question of whatfactors are controlling soil N dynamics for tree speciesplanted on a similar soil in a similar climate. If mi-croclimate differences among the five species wereimportant factors, then N mineralization and nitrifi-cation rates measured in the laboratory for the fivespecies should have been similar. Laboratory and insitu mineralization rates, however, exhibited similarinterspecies patterns. For example, soil from the redoak plantation had the lowest net N mineralizationrates, whereas soil from European larch plantationshad significantly greater (P < 0.06) net N mineral-ization rates than Norway spruce and red pine (Table3). Also, soil from the European larch and red oakplantations exhibited the highest and lowest nitrifica-tion rates, respectively, both in the field and labora-tory. These data suggest litter quality may be animportant factor controlling N mineralization in theseplantations.

We did not observe a significant correlation be-tween net N mineralization and previously measured(Son and Gower, 1992) soil characteristics such astotal soil N content (P = 0.4), mineral soil N con-centration (P — 0.9), soil organic C (P = 0.4), or

GOWER & SON: SOIL AND LEAF LITTERFALL NITROGEN DYNAMICS 1965

soil pH (P = 0.7) for the five species. Nadelhofferet al. (1983) and Pastor et al. (1984) also did notobserve correlations between annual soil N mineral-ization and these characteristics for different tree spe-cies in Wisconsin.

At the beginning of the study we speculated thatannual net N mineralization would be inversely relatedto leaf lifespan; however, this relationship was onlysignificant if we excluded red oak (Fig. 5). Red oakmay not fit the relationship because it has very schler-ophyllous foliage and a high lignin concentration, rel-ative to many other temperate species with short leaflifespans. The lignin concentration of red oak foliage(25%) is greater than sugar and red maple (Acer rub-rum L.)(10.1-12.5%), white ash (12.2%), paper birch(Betula papyrifera Marshall)(14.5%), pin cherry(Prunus pensylvanica L.f.)(19.3%) but similar toAmerican beech (Fagus grandifolia Ehrh.) (24.1%)(Melillo et al., 1982; Aber et al., 1984). Foliage lig-nin concentration is important because leaf litter de-composition is negatively correlated to ligninconcentration of leaf litterfall (Meentemeyer, 1978)and leaf litterfall lignin/N ratio (Schlesinger and Hasey,1981; Melillo et al., 1982).

Using lignin percentage values for the same speciesfrom a similar soil (red oak, 24.8%; white pine, 22.5%;and red pine, 21.9%; Aber et al., 1984) or the verysimilar species western larch (Lara: occidentalis Nutt.),12.6% (Gower et al., 1989) for European larch andwhite spruce [P. glauca (Moench)Voss], 26.8% (C.Wessman, 1991, personal communication) for Nor-way spruce and leaf litterfall mass and N content datafrom this study, we observed a significant inverse re-lationship (r2 = 0.78, P < 0.05) between annual netN mineralization and leaf litterfall lignin/N ratio forthe five species (Fig. 6). Lignin concentration, unlikeN, for a particular species does not vary greatly fromregion to region (Ausmus, 1973) and therefore it isprobably adequate to use lignin concentration for thesame or similar species. This positive correlation sug-gests a positive feedback between vegetation and soilprocesses; this idea is not new. Pastor et al. (1984)reported an inverse relationship between litter C/N andannual net mineralization for eight forests in Wiscon-sin, but these forests were growing on very differentsoils (e.g., Histisols, Entisols, Inceptisols, Alfisols,

120

_otsN

1'<Ue •'sz

aAet̂

100

80

60

40

*\Europeanlarch 3.38 (X)

,Red oak" ^.Red pine

Norway spruce

10 20 30 40Leaf litterfall ligninrnitrogen

Fig. 6. Relation of annual net N mineralization to leaf litterfalllignin/N for red oak, European larch, white pine, red pine,and Norway spruce.

and Spodosols). Because more than one of the fivemajor factors that influences soil development (Jenny,1980) differed in the studies by Pastor et al. (1984)and Nadelhoffer et al. (1984), these studies do notclearly demonstrate a vegetation effect on soil N min-eralization. We are only aware of one other study thathas compared N mineralization rates beneath vegeta-tion that was planted on a relatively homogeneoussoil. Wedin and Tilman (1990) measured N mineral-ization rates beneath different species of grasses andprovided direct evidence of the rapid and large influ-ence different species of grasses can have on soil Ndynamics. We do not have adequate pretreatment datato conclusively demonstrate that soil N was similarwithin each block, but several indirect lines of evi-dence suggest large differences were unlikely. More-over, the empirical relationship that we observedbetween annual net N mineralization and leaf litterquality is functionally based and consistent with thecurrent understanding of factors controlling decom-position.

In conclusion, leaf litterfall N content did not differsignificantly among five species with different leaflifespans but annual net N mineralization and nitrifi-cation rates differed among the five species. The re-sults from this study demonstrate that tree species canmodify soil N mineralization rates in a relatively shorttime relative to rates of soil development. Additionalresearch is required to determine if there is a consis-tent relationship between leaf lifespan and leaf litter-fall quality (i.e., lignin/N). If such a general relationshipdoes exist, it would provide a simple, yet functionallybased, approach to incorporating a positive feedbackmechanism between forest vegetation and soil N cy-cling into species-generic ecosystem process models.

ACKNOWLEDGMENTSThis study was made possible by support from Mclntire-

Stennis and University of Wisconsin Graduate School grantsto S.T. Gower. Mr. Adrian E. Hagen, Wisconsin Depart-ment of Natural Resources, allowed us to conduct researchat the Coulee Experimental Forest and provided valuableinformation on the history of the plantations. Karin Fass-nacht, Dan Olson, Paul Plesh, and Jessica Tausend pro-vided help in the field and laboratory. Dr. Eric Nordheimand Dennis Hiese, Department of Biometry and Statisticsof the College of Agriculture and Life Sciences of the Uni-versity of Wisconsin, Madison, provided valuable discus-sion on data analysis.

1966 SOIL SCI. SOC. AM. J., VOL. 56, NOVEMBER-DECEMBER 1992