Heterogeneity of soil and plant N and C associated with individual plants and openings in North American shortgrass steppe

Plant and Soil 138: 247-256. 1991 0 1991 Kl~iwer Acatleinic Publishers. Printed in the Netherlands PLSO 9111

Heterogeneity of soil and plant N and C associated with individual plants and openings in North American shortgrass steppe

PAUL B. HOOK', INGRID C. BURKE',' and WILLIAM K. LAUENROTH'.' 'Department of Range Science. 2~epurtnzent of Forest and Wood Sciences and ' ~ u t u r a l Resource Ecology Laboratory, Colorado State University, Fort Collins, C O 80-523, U S A

Received 28 February 1991. Revised July 1991

Key words: Bouteloua gracilis, grassland, nitrogen mineralization, respiration, small-scale hetero- geneity, soil organic matter

Abstract

Small-scale spatial heterogeneity of soil organic matter (SOM) associated with patterns of plant cover can strongly influence population and ecosystem dynamics in dry regions but is not well characterized for semiarid grasslands. We evaluated differences in plant and soil N and C between soil from under individual grass plants and from small openings in shortgrass steppe. In samples from 0 to 5 cm depth, root biomass, root N, total and mineralizable soil N, total and respirable organic C, C : N ratio, fraction of organic C respired, and ratio of respiration to N mineralization were significantly greater for soil under plants than soil from openings. These differences, which were consistent for two sites with contrasting soil textures, indicate strong differentiation of surface soil at the scale of individual plants, with relative enrichment of soil under plants in total and active SOM. Between-microsite differences were substantial relative to previously reported differences associated with landscape position and grazing intensity in shortgrass steppe. We conclude that microscale heterogeneity in shortgrass steppe deserves attention in investigation of controls on ecosystem and population processes and when sampling to estimate properties at plot or site scales.

Introduction

Small-scale spatial heterogeneity of soil organic matter and plant cover can strongly modify biogeochemical processes and plant population dynamics. This spatial variability also presents challenges for estimation of element concen- trations and availability to plants and microbes at plot. field, and site scales. Small-scale hetero- geneity of soil properties may be particularly important in dry regions, which are characterized by low plant cover. Soil beneath and between plants may be subject to different rates of ero- sion, deposition, and above- and belowground litter inputs. Contrasts between microsites undcr shrub canopies and bare or grass-covered open-

ings are often striking in arid and semiarid shrub- lands. Differences in soil chemical and physical properties in soils beneath and between shrubs in dry regions are well documented (Burke, 1989; Burke et al., 1989; Charley and West, 1975, 1977; Schlesinger et al., 1990). Schlesinger et al. (1990) proposed that development of shrub- interspace heterogeneity of soils is central to major changes in ecosystems undergoing deserti- fication.

Differences in soil properties between grass- covered soil and bare openings in semiarid grass- lands are not wcll characterized. Openings are usually smaller and less conspicuous in semiarid grasslands than in shrublands, but low and patchy cover of grasses promotes analogous con-

248 Hook et al.

trasts between bare and plant-covered soil. De- scription of heterogeneity of soil associated with patchiness of grass cover is important for analyz- ing the influence of low plant cover on soil fertility and seedling dynamics in semiarid grass- lands and for understanding changes in ecosys- tems that may result from changes in plant lifeform and cover patterns. In addition, knowl- edge of the magnitude of systematic differences in soil propel-ties under plants and in openings and relative area of plant cover and openings is needed to select appropriate sampling schemes to estimate large-scale averages.

In the shortgrass steppe of northcentral Col- orado. basal cover of all plants ranges from 20 to 40%, of which 85-95% is the perennial bunch- grass blue grama (Boureloua gracilis [H.B.K.] Lag. ex Griffiths; nomenclature follows Great Plains Flora Association [1986]) (Milchunas et al., 1989). Most of the surface is bare or litter- covered soil, which occurs as small openings between grass plants. Most openings are 0.05- 0.30 m across and are several centimeters lower than soil under plants. Biologically active organic matter is concentrated in the surface 1 cm of soil under B. gracilis plants (Woods, 1989) and is susceptible to redistribution by wind and rain. Rapid decrease in root density with distance from B. grncilis plants (Coffin and Lauenroth, 1992) may also contribute to horizontal hetero- geneity of soil organic matter (SOM). We hy- pothesized that soil from openings would have lower quantities of SOM, mineralizable N, and respirable C than soil from under B. graci1i.s plants.

The objective of this study was to characterize heterogeneity of soil and plant nitrogen and organic carbon associated with individual B. gracilis plants and openings in shortgrass steppe. The questions we addressed include: 1) What fraction of the soil is made up of open-

ings and what is their size distribution? 2) Are standing stocks of total and available N

and total and respirable organic C higher in soil under individuals of B. gracilis than in openings?

3) Do C : N ratios and C turnover rates, esti- mated by the ratio of respirable C to total C , differ between soil from under plants and soil from openings'?

4) Are patterns of small-scale spatial hetero- geneity of soil properties and spatial patterns of plant cover and openings consistcnt be- tween two sites with soils of different texture?

5) How large are differences between soil from under plants and soil from openings relative to differences among topographic positions, physiographic units, and land uses?

Methods

Study urea

We sampled soils and cover patterns at the Cen- tral Plains Experimental Range (CPER) of the United States Department of Agriculture- Agricultural Research Service, located in north- central Colorado (USA), approximately 60 km northeast of Fort Collins (40" 49' N, 107" 47' W). Mean annual precipitation is 309 mm, and mean monthly temperatures range from -3°C in January to 22°C in July (Parton and Greenland, 1987). Land surfaces are level to gently rolling, late-Pleistocene to mid-Holocene alluvial de- posits modified by Holocene eolian deposition (Davidson, 1988). Typical soils are Aridic Ar- giustolls with sandy loam surface horizons (Yon- ker et al., 1988). Vegetation is typical of the shortgrass steppe, with B. grncilis accounting for approximately 90% of basal cover on most sites (Milchunas et al., 1989). A number of grasses, succulents, sub-shrubs, and forbs account for the remainder of the cover. Moderate grazing by cattle occurs over most of the CPER; stocking is approximately 0.95 yearling month ha- ' and is adjusted to achieve approximately 40% reinoval of aboveground biomass of forage grasses by the end of the (,-month grazing season (Bement, 1969; Klipple and Costello, 1960). This grazing intensity is less than the estimated maximum rate of forage removal that can be sustained without - - degrading pastures to unsatisfactory condition I

and reducing forage production (SO%, Klipple and Costello, 1960).

We sampled two sites on nearly level, alluvial surfaces, a clay loam site ('CL', Township 10 N, Range 66 W. west 112 Section 22 E) and a sandy loam site ('SL', Township ION, Range 66 W, west 112 Section 23 E). We selected sites for

Small-scale spatial lzeterogeneity of N rznd C 249

contrasting soil texture, similar management, and large extent of uniform soil and vegetation. Vegetation at both sites is dominated by B. gracilis. At the clay loam site, the perennial grasses Buchloe cinctyloides (Nutt.) Engelm. and Agropyrotz smithii Rydb. are also abundant and tend to alternate with B. gracilis in patches several meters wide. The shrub Atriplex canes- cells (Pursh) Nutt. and the sub-shrub Eriogotzum effusurn Nutt. are abundant at the sandy loam site. The succulent O p u n t ~ c ~ polyacantha Haw. and the sub-shrubs Arterr l~~ia frigida Willd.,

I Chrysothumtzus nauseosus (Pall.) Britt., and Gutierrez~a sarothrae (Pursh) Britt. and Rushy account for a small fraction of cover at both sites. Both sites have been grazed moderately by cattle from November through April since 1970.

Field sampling

We sampled cover patterns and soils separately using two sets of randomly located transects. At

o transects each site, we located 8 cover samplin, and 5 soil sampling transects in a 100-m wide by 900-m long band. We excluded areas not domi- nated by B . gracilis.

We estimated the proportion of the surface in openings and their size distributions using 25-m transects and a line-intercept method. We de- fined openings as transect segments greater than 0.01 m long without live plant cover. To sample small and large openings evenly, we sampled three sections of each transect at three different levels of resolution. From 0 to 1 m, we measured all openings to the nearest 0.01 m; we also re- corded length of segments with plant cover at this scale. Froin 1 to 5 m, we recordcd sizes of openings over 0.1 m, and from 5 to 25 m, we recorded sizes of openings over 0.2 m.

We sampled soil from 2 types of microsites: the centers of openings and the centers of adja- cent B. gracilis plants. Along each 2-m widc soil sampling transect, we sampled the first five open- ings with a minimum diameter greater than 0.15 m, the approximate median size of open- ings; we excluded openings less than I m from shrubs. Soil was sampled to a depth of 0.05 m with a 0.05-m diameter coring device. The five cores from each type of microsite were compo- sited for each transect and placed on ice for

transport to the laboratory. At each pair of openings and plants, we estimated the difference in elevation of the surface of the mineral soil under plants and in openings to the nearest nim.

Laborc~tory incubations and ntzrrlyses

Soil was sieved through a 2-mm mesh sieve; soil was at field water contcnt (3-7%), which was nearly air-dry. Soil water content was deter- mined by drying a 10-g subsample to constant mass at 105°C. Bulk density was estimated from the mass of sieved soil and total sample volume and was used to convert data from a soil mass basis to volumetric units. The mass of soil material removed by sieving did not exceed 1% of sieved soil mass. Field capacity of sieved soil was estimated by the pressure plate method (Klute, 1986). Sand, silt, and clay fractions were determined by the hydrometer method (Bouyoucos, 1962). Plant material was sorted into 4 categories: live roots, dead roots and other belowground litter, tillers and leaves. Live roots were distinguished by their light color and high tensile strength. Plant material was dried to con- stant mass at 5j°C, weighed, and ground to pass a 20-mesh sieve in a Wiley mill. Total N (TN) content of plant and soil subsamples was de- termined using a micro-Kjeldahl digestion (Nel- son and Son~mers, 1980) followed by colorimct- ric analysis for NH, (EPA, 1979). Total organic carbon (TOC) of finely ground soil subsainples was estimated using wet oxidation-diffusion (Snyder and Trofymow, 1984).

Potential net N mineralization and respiration were estimated in 30-day, aerobic incubations of soil subsamples. We performed two sets of incu- bations to evaluate the sensitivity of estimates to removal of belowground litter. In the first set, we incubated 25 g of sieved soil. In the second set, we mixed 25 g of sieved soil with the same amount of belowground litter as was sieved frorn the sample initially. Belowground litter was mostly dead roots but included other particulate SOM retained by the sieve; added litter was not dried or ground. Soil samples frorn under plants received from 0.2 to 0.4g additional litter and opening san~ples received from 0.01 to 0.02g. Soil samples were placed in 125-mL erlenrneyer flasks, brought to field capacity, placed in 0.9 L

250 H o o k et ul.

jars with 5 tnL of 2-M NaOH trap for CO,, and 0.20 - 1 incubated for 30 days at 25°C (Schimel et a]. , 0.18

1985a). Soil samples were extracted for determi- 0.16

nation of initial and final NH,-N and NO,-N 0.14

concentrations using 2-M KC1 amended with 0.12 0

5 p g g 1 phenyl-mercuric-acetate to retard mi- , o.lo

crobial activity. Extracts were analyzed using o.os standard colorimetric methods on a Lachat au- 0,06 toanalyzer (EPA, 1979). Net N mineralization 0.04 was calculated as the difference between final 0.02 and initial NH,-N plus NOI-N concentrations.

0 1-5 5-1 0 10-20 20-50 > 50

Statistical analysis Opening size (cm)

We analyzed data by analysis of variance using a split-plot factorial design (Montgomery, 1984). We treated sites as blocks. Use of two sites with contrasting soils permitted limited inference about the consistency of microsite effects be- tween sites but was not intended to permit infer- ence about patterns associated with specific tex- tures. We tested data for heterogeneity of var- iance using Bartlett's test (Montgomery, 1984). Where variances were heterogeneous, we report results of analysis of variance of rank-trans- formed data (Conover, 1980). We used Wilcox- on's rank sum test to evaluate between-site dif- ferences in the proportion of surface arca in openings of different sizes (Conover, 1980). We performed all analyses using SAS statistical soft- ware (SAS Institute Inc., 1988) and a signifi- cance level of 0.05.

Results

Approximately half of the surface area was in openings over 0.01 m across (Fig. I ) ; the remain-

Fig. 1. Fraction of surface in five size-classes of openings at two sites at the Central Plains Experimental Range. Open- ings were defined as segments longer than 0.01 m without live plant cover on line transects. Error bars represent ? 1 stan- dard error of the mean.

ing area was mostly covered by clusters of grass tillers separated by less than 0.01 m. 0. poly- acarztha cladodes and dung accounted for less than 2% of the surface area. The fraction of the surface in openings did not differ significantly between CL and SL sites (48 * 5% and 62 4%, respectively). However, the fraction of the area made up of openings at least as large as those sampled for soil analyses ( > 0.15 m) was signifi- cantly greater at the SL than at the CL site (35 vs 21Rl), and the smallcst openings (0.01-0.05 m) represented significantly more of thc surface at the CL than at the SL site (10 vs 5%).

Soil texture and bulk density did not differ significantly between microsites (Table 1); how- ever, bulk density was consistently lower undcr plants than in openings at the CL site. Average soil rnicrorelief between plants and openings was 3 cm. Soil had significantly higher sand content,

7irble I . Mean hulk dcnsity, microrelief, and sand, silt and clay fractions of soil at two sites at thc Central Plains Experimental Range. Surface soil was sampled to 0.05 m depth under Bo~itelolrrr grucilis plants and in small openings between plants. Microrelief was cstimated as the difference in soil surface elevation betwccn plant-covered micrositcs and adjacent openings. C L =clay loarn site; SL = sandy loam site. Numbers in parentheses are standard errors of the means

Site Microsite Bulk Microrelief Sand Silt Clay density (cm) (S/Z) (g ig) (gig) (Mglm' )

C L Plant l.OY(0.07) 2.6(0.3) 0.39(0.03) 0.25(0.03) 0.30(0.01) C L Opening 1.29(0.04) 0.39(0.03) 0.24(0.02) 0.37(0.02) SL Plant 1.56(0.06) 3.1(0.5) 0.66(0.07) 0.16(0.04) O.lS(0.02) SL Opening 1 .59(0.13) 0.67(0.07) 0 . 15(0.05) 0 . lS(,0.02)

Small-scale spatial heterogeneity of N and C 251

bulk density. and microrelief at the SL than at the C L site (Table 1).

Soil and plant N and C pools were significantly greater in plant-covered microsites than in open- ings (Figs. 2 and 3). Soil T O C and TN were 1.4 and 1 .2 times higher under plants than in open- ings (Figs. 2a, b). In contrast, plant biomass was over 20 times higher in plant-covered microsites than in openings, causing an equally large differ- ence in biomass N between microsites (Fig. 3). Soil TN accounted for approximately 60% of the differencc in total N (plant plus soil TN) be-

Fig. 2. Standing stocks of soil organic carbon (TOC) (a) and total nitrogen (TN) (b) and mass ratio of C to N in the top O.Oim of soil (c) at two sites at the Central Plains Cx-

1 2 - 11 - 1 0 - 9 -

Z 0 8 1 7

6 ~ cn 5 - 4 -

3 - 2 - 1 0

perimental Kangc. Site and mlcrosite codes arc: C L , clay loam site; SL, sandy loam site; P. plant-covercd microsite; 0,

CL-P CL-0 SL-P SL-0

C !

opening. Error bars represent tl standard crror of the mean.

,

18 16 14 12 10 8 6 4 2 0

CL-P

I

I I I

SL-P SL-0

Fig. 3. Standing stocks of plant biomass (a) and N (b) at two sites at thc Central Plains Experimental Range. Above- ground biomass and biomass in the top 0.05 m of soil were sampled at microsites with Bouteloua gracilis plants and in small openings between plants. Site and microsite codcs are: CL. clay loam site; SL, sandy loam site; P, plant covered microsite: 0. opening. Error har:r reprcscnt +1 standard error of the means of total biomass and N.

tween microsites. Soil C : N ratios were slightly but significantly lower in openings than under plants (Fig. 2c). Initial total extractable inorganic N did not differ significantly between micrositcs, but NH,-N concentrations were significantly lower and NO,-N concentrations wcre signifi- cantly highcr in soil from openings than soil from under plants (Table 2). Soil TN was significantly higher and C : N ratio significantly lower for the

Table 2. Inorganic N concentrations (g m ' to 0.0.5 m depth) in KC1 extracts of soil samplcs from two sites at the Central Plains Experimental Range. Surface soil was sampled to 0.0.5 m depth under Borrrelor~a gracilis plants and in small openings between plants. C L = clay loam sitc; SL = sandy loam site. Numbers in parentheses arc standard errors of the means

Sitc Micrositc NO,-N NH,-N Tolal

C L Plant 0.12(0.02) 0.33(0.0h) 0.46(0.07) cL Opening 0.24(0.01) 0.23(0.OI) 0.47(0.01) S L Plant 0. 12(0.02) 0.21(0.02) 0.33(0.02) S L Opening 0.3h(0.05) 0.09(0.03) 0.41(0.05)

252 Hook ct al.

SL sitc than the C L site (Figs. 2b, c). No signifi- cant site-by-microsite interactions were present for TN, T O C . C : N , or inorganic N.

Potential net N mineralization and C0,-C production were significantly higher for soil from under plants than soil from openings (Figs. 4a, b). Respiration was significantly higher in subsamples amended with belowground litter, but net N mineralization was unaffected. In- creases in respiration with added litter were greater for soil from under plants than soil from openings, consistent with the larger amounts of litter added.

Net N mineralization as a fraction of soil T N was significantly higher for the C L site than the

z 0.03

.- - E .; 0.01

z m

0 P P-L

Clay 0 0-L P P-L 0 0 - L -- ---

loam Sandy loam

SL site but showed n o significant microsite or litter-amendment effccts o r interactions (Fig. 4c). C0 , -C production as a fraction of T O C was significantly higher for soil from under plants than soil from openings, and was significantly higher for litter-amended subsamples than un- amended subsamples (Fig. 4d). The ratio of respiration t o T O C was slightly but significantly higher for the S L site than the CL site. The ratio of C 0 , - C production to net N mineralization was significantly lower for soil from openings than soil from under plants (Fig. 4e); a signifi- cant rnicrosite-by-sitc interaction reflected a larger between-microsite difference for the CL site.

18 16 14

12

10 8 6 4 2

0 P P-L 0 0 - L P P-L 0 0 - L

Clay loam Sandy loam

Fig. 4. N mineralized and C respired during 30-d, ;~erobic incubations of soil from 0 to 0.05 m depth at a sandy loam site and a clay loam sitc a t the Central Plains Experimental Range: net N mineralized (a) and C0, -C produced (b) during incubations; ratios of net N rninel-alized to total N (c), COI-C produced to total organic carbon (d), and C 0 , - C produced to net N mineralized ( e ) . Microsite codes are: P, plant-covered microsite; 0 , opening. 'L' indicates subsamples amended with additional helowground litter. Error bnrs represent 2 1 standard error of the means.

Small-scale spatial heterogcjzeity of N and C 253

Discussion

Boutcloun gracilis plants and small openings each cover a significant proportion of the soil surface in moderately grazed shortgrass steppe at the CPER, and the area associated with each con- tains different amounts of nutrients for plant growth and substrates for biogeochemical re- actions. Soil TOC, TN, C : N ratio, potential net N mineralization, C0,-C production, ratio of C0 , -C production to TOC, and ratio of C0,-C production to net 1V mineralization were all high- er for soil undcr plants than soil between plants. 'Thcse patterns indicate that soil under individual grass plants is enriched in total and bioJogically activc SOM (estimated by potentially miner- alizable N and respirable C) rclativc to soil in openings (Parton et al.. 1987). Two processes are most likely to contribute to the generation of thesc patterns: spatially variable primary produc- tion and redistribution of surface soil. There is evidencc for both processes, but our data do not permit us to determinc their relative importance.

We infer from our root biomass data and previous studies of B. gracilis root distribution (Coffin and Lauenroth, 1991; Lce, 1990; Mil- chunas and Lauenroth, 1989) that root produc- tion is lower in openings than undcr plants, evcn within 0.05 to 0.1 m from plants. Greater pro- duction of above- and bclowground litter, leach- ates. and cxudatcs under plants than in opcnings should contribute larger amounts of labilc or- ganic compounds to soil undcr plants and raise TOC, respirable C , and C : N. In addition, these patterns of production may result in plant- mediated transfcr of N from soil in openings to soil under plants. Thus spatial patterns of plant growth may control spatial patterns of both C and N in soil, C directly through photosynthesis and belowground production and N indircctly through translocation from locations of 1V uptake to areas of greatest root production. The balance of these two processes may control differenccs in C : N ratio under and between plants.

Rcdistribution of surface soil from openings to around individuals of B. gracilis may also con- tribute to between-microsite diffcrences in total and active SOM. We have observed accumula- tion of soil matcrial around bunchgrasses during thc coursc of several experiments at the CPER

(personal observation), and our microrelief data are consistent with the interpretation that soil accumulates in plant-covered microsites. Rates of erosion by heavy rains are low at the CPER (H Weagraff, personal communication, 1990), and eolian sediment traps 1 m above the ground have collected very little material (D Doehring, personal communication, 1990). We infer that most soil movement by wind or water occurs over short distances and results in slight erosion of openings and deposition of soil material around plants. Similar results have been ob- tained in scmiarid shrublands (Coppinger et al., 1991). Bccause both active and total SOM arc strongly concentrated in the upper 1 to 3 cm of soil beneath individual B. gracilis plants in shortgrass steppe (Woods, 1989; Woods, person- al communication, 1990), erosion following death of a plant could create a ncw opening with low TOC, TN, respirable C and mineralizable 1V.

Our data suggest that patterns of demand by microbial populations for N and organic C rela- tive to supply differ between microsites. Thc greater supply of organic substrate to microbes in soil under plants was evident in higher poten- tial rates of respiration and net N mincralization, but con~parison of respiration to N mineraliza- tion revealed that N supply and C supply did not diffcr proportionally. The ratio of C 0 2 - C pro- duction to nct N mineralization was higher in soil from under plants than soil from openings; this indicates proportionally greater N immobiliza- tion potential undcr plants. This is consistent with higher inputs of C-rich O M under plants than in openings. Organic C turnover, indexcd by the ratio of potential respiration to TOC, is also higher under plants than in openings; N turnover cannot be estimated without an esti- mate of gross N mincralization.

The inference that SOM quantity and quality vary significantly at the microsite scalc docs not appear to depend strongly on soil tcxture or inclusion of belowground litter in incubated soil. The direction and approximate magnitude of differences between microsites were consistent betwccn the two sites, despite largc differences in texturc and TN. A significant microsite-by-site intcraction, which could indicate a difference in microsite effccts between sites. was found only for the ratio of C0,-C production to net N

254 Hook et al.

mineralization. However, this statistical inter- action reflected only a slight between-site differ- ence in the magnitude of the microsite effect; the ratio of respiration to N mineralization was much larger under plants than in openings at both sites. Although patterns of microsite properties did not differ between sites, the relative area of large and small openings did differ significantly between sites. If soil properties vary with open- ing size, pasture-scale estimates of soil N and C fractions may need to account for differences in the size distributions as well as the total area of openings.

We found 20% to 50% mean differences in TOC, TN, and net N mineralization and 115% to 150% differences in respirable C associated with presence or absence of individual B. gracilis plants at scales of 0.1 to 0.2 rn. Other studies at the CPER reported differences between pairs of topographic positions, physiographic units, or

land management units from 0% to 170% for TOC, TN, net N mineralization, and respiration in surface soil; these comparisons correspond to scales from 10 m to 10 km (Table 3). Thus mi- croscale variability resulting from patterns of plant cover may be larger or smaller than mac- roscale variability resulting from geomorphic and pedogenic processes and land management prac- tices depending on the soil properties, sites, and scales considered. Only relative differences in SOM between cultivated and native steppe sites appear to be consistently larger than differences between microsites.

Spatial heterogeneity of SOM and nutrient pools at scales related to the sizes of individual plants appears to be a general characteristic of ecosystems in arid and semiarid regions (Burke, 1989; Burke et al., 1989; Klemmedson, 1989; Schlesinger et al., 1990). Spatial scale and mag- nitude of heterogeneity associated with indi-

Tuble 3. Comparison of relative variation in T O C or SOM, respirable C. T N , and mineralizable N arnong scales and sources of variation in the shortgrass steppe. Rclative variation was calculated as [I00 x (high mean -low mean)/(low mean)] for each pair of means. The numbers of means (microsites, locations, o r land uses) compared to generate each reported range of relative variation is indicated in parentheses in the first column; the types of means compared are reported in footnotes. Scale is the approximate order of ~nagnitudc distance between locations salnplcd to generate means. If respiration and nct N mineralization data were reported for incubations of different lengths, data for incubations nearest to 30 days were used

Comparison Approximate Depth Percent variation Source scale (m) sampled (cm)

T O C or C,L.l> TN NllIln SOM

Plant vs opening (4)"

Physiographic units ( 3 ( 1 ) ~

7'opographic positions (3)

Topographic positions (30)&

Topographic positions ( G ) ~

Topographic positions (4)'

Heavily grazed vs ungrazed (4)'

Native vs cultivated (12)"

115-153 19-27 24-52 This study

Yonkcr et al.. 1088

2-50 Schimel et al., 1985b 23-93

Yonker et al.. 1988

5-32 1-22 2-25 Woods and Schuman, 1988

10-135 Milchunas e t al.. 1992 (Submitted)

20-40 Milchunas et al . , 1992 (Submitted)

129-396 19-1 00 104-343 Woods and Schuman, 1988

" T w o nlicrosite typcs compared at two sites. " Six physiographic units compared at five topographic positions. ' Fivc topographic positions co~nparcd across six physiogr;~phic units. "Three topographic positions compared at two native range sites. ' T w o topographic positions compared within two grazing regimes. 'Adjacent long-tern~ ungrazed and heavily grazed treatments compared at two topographic positions. "djacent native range and cultivated sites compared a t three topographic positions on two hillslopes

Snlall-scule spatial heterogeneity of N and C 255

vidual plants depend on spatial patterns of biomass. which influence litter inputs. soil move- ment, microclimate, and interactions between nutrient uptake and SOM quality. Spatial pat- terns of biomass are influenced by the lifeform of the dominant species, resource availability, dis- turbance, and herbivory (Coffin and Lauenroth, 1988; Milchunas and Lauenroth, 1989). The small-scale heterogeneity we observed at the CPER probably results from the morphology of the dominant grass species and climatic con- straints on plant cover rather than from direct effects of grazing on community structure. R. gracilis is highly tolerant of grazing by cattle and dominates grazed and ungrazed sites at the CPER; after 47 y of controlled grazing, B. gracilis basal cover and total plant basal cover are larger and cover of sub-shrubs is less in heavily grazed pastures than in exclosures (Mil- chunas et al., 1989). Furthermore, horizontal distribution of roots and tillers is more even in heavily grazed pastures than ungrazed sites at the CPER (Milchunas and Lauenroth, 1989). If spatial patterns of biomass control distribution of C and N, we would expect horizontal evenness of C and N to increase along a gradient of increasing grazing intensity in shortgrass steppe. We would also predict that our observations for moderately grazed pastures represent an inter- mediate dcgree of soil heterogeneity that is less than for ungrazed sites and greater than for heavily grazed sites. Our interpretation contrasts with results of Schlesinger et al. (1990), who reported increased heterogencity of soil N in a Chihuahuan Desert ecosystem in which grazing by domestic livestock led to decreased cover of black grama grass (Bouteloua eriopoda [Torr.] Torr) and increased dominance by shrubs.

Small-scale variability of SOM and N may have important effects on plant detnography and biogeochemical processes in shortgrass steppe. Plant-induced heterogeneity in soil resources can have significant feedbacks to plant establish- ment, growth and survival (Franco and Nobel, 1989; Penridge and Walker, 1986), and small- scale spatial patterns of plant distribution can influence rates of physical and biogeochemical processes that control ecosystem C and N bal- ance (Schlesinger et al., 1990). The demonstra- tion that heterogeneity of SOM and nutrient

pools associated with individual bunchgrass plants and openings is large relative to land- scape-scale variability reinforces the need to con- sider small-scale phenomena in ecological re- search in semiarid grasslands.

Acknowledgements

We wish to thank Mark Lindquist for field and laboratory assistance. This is a contribution of the Shortgrass Steppe Long-Term Ecological Re- search Program (NSF BSR-8612105 and BSR- 90 11 659) and the Colorado Agricultural Experi- ment Station (157661).

References

Bemeot R E 1969 A stocking-rate guide for beef production on blue grarna range. J . Range Manage. 22. 83-86.

Bouyoucos G J 1962 Hydrometer method improved for making particle size analyses of soils. Agron. J . 54, 464- 465.

Burke I C 1989 Control of nitrogen mineralization in :I

sagebrush steppe landscape. Ecology 70, 1 1 15-1 126. Burke 1 C, Reiners W A and Scllimel D S 1989 Organic

matter turnover in a sagebrush steppe landscape. Biogeo- chemistry 7. 11-31.

Charley J L and Weht N E 1975 Plant-induced soil chemical patterns in some shrub-dominated semi-desert ecosystems of Utah. J . Ecol. 6.3. 945-963.

Charlcy J L and West N E 1977 Micro-patterns of nitrogen mineralizalion activity in soils of some shrub-dominated semi-dcserl ecosystems of Utah. Soil Biol. Biochem. 9. 357-365.

Coflin D P and Lauenroth W K 1988 The effects o i disturb- ance size and frequency on a shortgrass plant community. Ecology 09. 1609-1617.

Coffin D P and Lauenroth W K 1991 Effects of competition on spatial distribution of roots of blue grama. J . Rangc Manage. 11. 68-71.

Conover W J 1980 Practical Nonparametric Statistics. 2nd ed. Wiley. New York. 193 p.

Coppinger K D, Reiners W A . Burke 1 C and Olson R K 1991 Net erosion on a sagebrush steppe landscape as detcrrnined by '"'cs distribution. Soil Sci. Soc. Am. J , 55, 254-258.

Davidson J M 1988 Surficial Geology and Quarternary His- tory of the Central Plains Experimental Range, Colorado. M. Sc. Thesis, Colorado State University. Fort Collins, C O . 87 p.

EPA 1979 Methods for chcrnical analysis of water and w:istes. EPA-60014-79-020. March 1979, lJ.S. Environmental Pro- tection Agency. U.S. Governnlent Printing Oflice. Washington. DC.

256 Small-scale spatial heterogeneity o,f N and C

Franco A C and Nobel P S 1989 Effect of nurse plants on the microhabitat and growth of cacti. J . Ecol. 77. 870-886.

Great Plains Flora Association 1'186 Flora of the Grcat Plains. University Press of Kansas, Lawrence. KS. 1392 p.

Klemmedson S O 1989 Soil organic matter in arid and semiarid ecosystems: Sources, accumulation, and distribu- tion. Arid Soil Res. Rehab. 3. 99-114.

Klipple G E and Costello D F 1960 Vegetation and cattle ~ ~

responses to different intcnsities of grazing on shortgrass ranges on the central Great Plains. Tech. Bull. 1216, U.S. Department of Agriculture-Agricultural Research Service. U.S. Government Printing Office. Washington. DC.

Klute A 1986 Water retention: Laboratory methods. In Methods of Soil Analysis, Part I: Physical and Miner- alogical Methods. Agronomy Monograph No. 9. 2nd edi- tion. Ed . A Klute. pp 335-362. American Socicty of Agronomy, Soil Science Society of America. Madison. W1.

Lee C A I990 The Spatial Distribution of Roots for Threc Plant Lifct'orrns in the Shortsrass Steppe. M. Sc. Thesis, Colorado State IJniversity, Fort Collins, CO. 114 p.

Milchunas D G and Laucnroth W K 1989. Three-dimensional distribution of plant bio~nass in relation lo grazing and topography in the shortgrass stcppe. Oikos 55. 82-86.

Milchunas D G . Lauenrvth W K and Chapman P L 1992 Abiotic and biotic control, and direct and iridirect effects of large herbivores on demography of opportunistic species in scniiarid grassland. Oecolopia (Berlin) (Submitted).

Milchunas D G . Lauenroth W K. Chapman P L and Kazcm- pour M K 1989 Effects of grazing, topography. and precipi- tation on the structure of a semiarid grassland. Vegelatio 80. 1 1 2 3 .

Montgomery D C 1984 Design and Analysis of Experi~ncnts. Wilcy. New York. 538 p.

Nelson 1) W and So~nmcrs L E I980 Total nitrogen analysis for soil and plant tissues. J . Assoc. Off. Anal. C h e ~ n . 63. 770-778.

Parton W and Greenland D 1987 Central Plains Experimen-

tal Range Site. Colorado. 111 The Climates of the Long- Term Ecological Rescarch Sites. E d . D Greenland. pp 23-27, Occasional Paper No. 14, Institute of Arctic and Alpine Research, University of Colorado. Boulder, CO.

Parton W S, Schimel D S. Colc C V and Ojima D S 1987 Analysis of factors controlling soil organic matter levels in Great Plains grasslands. Soil Sci. Soc. Am. J . 51, 1173- 1179.

Penridge L K and Walker J 1986 Effect of neighboring trees on eucalypt growth in a semi-arid woodland in Australia. J . Ecol. 74, 925-936.

SAS Institute Inc. 198s SASISTAT User's Guide, Rclease 6.03 cd. SAS Institute Inc.. Cary, NC. 1028 p.

Schlesin~er W H . Reynolds J F, Cunningham G L, Huen- nekc L F. Jarrell W M , Virginia R A and Whitford W G 1990 Biological feedbacks in global desertification. Science (Washington. DC) 11A7. 1043-1048.

Schimel D S . Coleman D C and Horton K A 1985a Soil organic Inattcr dynamics in paired rangeland and cropland toposequcnces in North Dakota. Gcoderma 36, 201-214.

Schimel D, Stillwell M A and Woodmansee R G 1985h Biogeochemistry of C , N, and P in a soil catena of the shortarass steppc. Ecology 66, 276-282.

Snyder J D and Trofymow J A 1984 A rapid accurale wet oxidation diffusioil procedure for determining organic a ~ ~ d inorganic cal-hon in plant ant1 soil samples. Commun. Soil Sci. Plant Anal. 15, 587-597.

Wootls L E 1989 Active organic mattcr distribution in thc surface I5 cm of ~~ndistui-bed and cultivated soil. Biol. Fertil. Soils 8, 1171278.

Woods L E and Schuman G E 1988 Cultivation and slopc position effects on soil organic matter. Soil Sci. Soc. Am. J . 52. 1371-1376.

Yonker C M. Scliimcl D S. Paroussis E and Heil R D 1988 Patterns of organic ci~rbon accumulation in a semiarid shortgrasss steppe. Colorado. Soil Sci. Soc. Am. J . 52. 478-483.