Grassland Plant Community Composition Research Articles

Pollinator-mediated effects of landscape-scale land use on grassland plant community composition and ecosystem functioning - seven hypotheses.

Environmental change is disrupting mutualisms between organisms worldwide. Reported declines in insect populations and changes in pollinator community compositions in response to land use and other environmental drivers have put the spotlight on the need to conserve pollinators. While this is often motivated by their role in supporting crop yields, the role of pollinators for reproduction and resulting taxonomic and functional assembly in wild plant communities has received less attention. Recent findings suggest that observed and experimental gradients in pollinator availability can affect plant community composition, but we know little about when such shifts are to be expected, or the impact they have on ecosystem functioning. Correlations between plant traits related to pollination and plant traits related to other important ecosystem functions, such as productivity, nitrogen uptake or palatability to herbivores, lead us to expect non-random shifts in ecosystem functioning in response to changes in pollinator communities. At the same time, ecological and evolutionary processes may counteract these effects of pollinator declines, limiting changes in plant community composition, and in ecosystem functioning. Despite calls to investigate community- and ecosystem-level impacts of reduced pollination, the study of pollinator effects on plants has largely been confined to impacts on plant individuals or single-species populations. With this review we aim to break new ground by bringing together aspects of landscape ecology, ecological and evolutionary plant-insect interactions, and biodiversity-ecosystem functioning research, to generate new ideas and hypotheses about the ecosystem-level consequences of pollinator declines in response to land-use change, using grasslands as a focal system. Based on an integrated set of seven hypotheses, we call for more research investigating the putative pollinator-mediated links between landscape-scale land use and ecosystem functioning. In particular, future research should use combinations of experimental and observational approaches to assess the effects of changes in pollinator communities over multiple years and across species on plant communities and on trait distributions both within and among species.

Soil warming duration and magnitude affect the dynamics of fine roots and rhizomes and associated C and N pools in subarctic grasslands.

The response of subarctic grassland's below-ground to soil warming is key to understanding this ecosystem's adaptation to future climate. Functionally different below-ground plant organs can respond differently to changes in soil temperature (Ts). We aimed to understand the below-ground adaptation mechanisms by analysing the dynamics and chemistry of fine roots and rhizomes in relation to plant community composition and soil chemistry, along with the duration and magnitude of soil warming. We investigated the effects of the duration [medium-term warming (MTW; 11 years) and long-term warming (LTW; > 60 years)] and magnitude (0-8.4 °C) of soil warming on below-ground plant biomass (BPB), fine root biomass (FRB) and rhizome biomass (RHB) in geothermally warmed subarctic grasslands. We evaluated the changes in BPB, FRB and RHB and the corresponding carbon (C) and nitrogen (N) pools in the context of ambient, Ts < +2 °C and Ts > +2 °C scenarios. BPB decreased exponentially in response to an increase in Ts under MTW, whereas FRB declined under both MTW and LTW. The proportion of rhizomes increased and the C-N ratio in rhizomes decreased under LTW. The C and N pools in BPB in highly warmed plots under MTW were 50 % less than in the ambient plots, whereas under LTW, C and N pools in warmed plots were similar to those in non-warmed plots. Approximately 78 % of the variation in FRB, RHB, and C and N concentration and pools in fine roots and rhizomes was explained by the duration and magnitude of soil warming, soil chemistry, plant community functional composition, and above-ground biomass. Plant's below-ground biomass, chemistry and pools were related to a shift in the grassland's plant community composition - the abundance of ferns increased and BPB decreased towards higher Ts under MTW, while the recovery of below-ground C and N pools under LTW was related to a higher plant diversity. Our results indicate that plant community-level adaptation of below ground to soil warming occurs over long periods. We provide insight into the potential adaptation phases of subarctic grasslands.

Wind Power Increases the Plant Diversity of Temperate Grasslands but Decreases the Dominance of Palatable Plants

As an important clean energy source, the scale and quantity of wind power have steadily increased under the background of global change. The construction and operation of wind power facilities have massive impacts on grassland microclimates. However, the effect of wind power operation on the plant community composition is still unclear. To investigate this issue, we selected wind farms in 6 meadow grasslands and 6 typical steppes in the central region of Inner Mongolia, the province with the largest scale of grassland wind power operations in China. At these sites, we conducted field sample surveys to obtain species information, measure plant biomass, calculate plant diversity, and take soil samples to determine soil nutrients. The results showed that wind power operation significantly reduced the dominance of Poaceae and Cyperaceae plants in both types of grasslands and significantly increased the Shannon diversity of meadow grasslands. The inconsistent responses at each experimental site led to a nonsignificant overall effect of wind power operation on the plant beta diversity. In addition, wind power operation significantly increased plant biomass in meadow grasslands. Wind power operation did not change the soil total carbon, total nitrogen, ammonium nitrogen, or nitrate nitrogen. On the basis of the results, we suggest strengthening the long-term monitoring of temperate grassland plant community composition in wind farms, and replanting of community-building species could be done at appropriate times.

Exploring the effects of pasture trees on plant community patterns

AbstractQuestionsWe aim to answer three questions: (a) what are the effects of canopy shading of different tree species on seed germination of eight understorey grassland species? (b) how is plant species’ occurrence in understorey communities affected by differences in canopy cover and does that depend on tree species composition? and (c) are there mechanistic links between the effects of trees on species’ germination and their occurrence in the understorey plant community?LocationSemi‐natural wooded pastures in the biosphere reserve Östra Vätterbranterna, southern Sweden.MethodsIn this study, we examined the germination of eight grassland plant species in a seed sowing experiment under natural conditions in the field. Seeds were sown beneath and outside the canopy of two tree species within 48 plots split up by four wooded pasture sites. We combined observed germination responses with a plant community survey to assess the effects of canopy cover in relation to tree species composition on plant community responses. We analysed these data in relation to species’ seed mass and vegetative shade tolerance.ResultsShade‐tolerant species germinated better beneath compared to outside tree canopies, without any clear advantage of large‐seeded over small‐seeded species. As expected, species’ shade tolerance was also positively related to canopy cover within the understorey plant community. Importantly, we found strong tree species‐specific effects of canopy shading on the species’ germination response, but not on their presence within the plant community. However, optimal canopy cover conditions for germination and for the mature plants differed across grassland species and depended on tree species.ConclusionsOur results show that different tree species play key ecosystem engineering roles in shaping wooded grassland plant community composition at the germination stage. Management practices favouring specific tree species may therefore be highly relevant for targeted biodiversity conservation of wooded semi‐natural grasslands.

Defoliation and arbuscular mycorrhizal fungi shape plant communities in overgrazed semiarid grasslands.

Overgrazing substantially contributes to global grassland degradation by decreasing plant community productivity and diversity through trampling, defoliation, and removal of nutrients. Arbuscular mycorrhizal (AM) fungi also play a critical role in plant community diversity, composition, and primary productivity, maintaining ecosystem functions. However, interactions between grazing disturbances, such as trampling and defoliation, and AM fungi in grassland communities are not well known. We examined influences of trampling, defoliation, and AM fungi on semiarid grassland plant community composition for 3yr, by comparing all combinations of these factors. Benomyl fungicide was applied to reduce AM fungal abundance. Overgrazing typically resulted in reduced dominance of Stipa Krylovii, contributing to degradation of typical steppe grasslands. Our results indicated trampling generally had little effect on plant community composition, unless combined with defoliation or AM fungal suppression. Defoliation was the main component of grazing that promoted dominance of Potentilla acaulis over Stipa krylovii and Artemisia frigida, presumably by alleviating light limitation. In non-defoliated plots, AM fungi promoted A.frigida, with a concomitant reduction in S.krylovii growth compared to corresponding AM suppressed plots. Our results indicate AM fungi and defoliation jointly suppress S.krylovii biomass; however, prolonged defoliation weakens mycorrhizal influence on plant community composition. These findings give new insight into dominant plant species shifts in degraded semiarid grasslands.

Landscape history confounds the ability of the NDVI to detect fine‐scale variation in grassland communities

Abstract The NDVI is a remotely sensed vegetation index that is frequently used in ecological studies. There is, however, a lack of studies that evaluate the ability of the NDVI to detect fine‐scale variation in grassland plant community composition and species richness. Ellenberg indicators characterize the environmental preferences of plant species—and community‐mean Ellenberg values have been used to explore the environmental drivers of community assembly. We used variation partitioning to test the ability of satellite‐based NDVI to explain community‐mean Ellenberg nutrient (mN) and moisture (mF) indices, and the richness of habitat‐specialist species in dry grasslands of different ages. The grasslands represent a gradient of decreasing soil nutrient status. If community composition is determined by the responses of individual species to the underlying environmental conditions and if, at the same time, community composition determines the optical characteristics of the vegetation canopy, then positive relationships between the NDVI and mN and mF are expected. Many grassland specialists are intolerant of nutrient‐rich soils. If specialist richness is negatively related to soil‐nutrient levels, then a negative association between the NDVI and specialist richness is expected. However, because grassland community composition is not only influenced by abiotic variables but also by other spatial and temporal drivers, we included spatial variables and grassland age in the statistical analyses. The NDVI explained the majority of the variation in mF, and also contributed to a substantial proportion of the variation in mN. However, variation in specialist richness and the lowest values of mN were explained by grassland age and spatial variables—but were poorly explained by the NDVI. Synthesis and applications. The NDVI showed a good ability to detect variation in plant community composition, and should provide a valuable tool for assessing fine‐scale environmental variation in grasslands or for monitoring changes in grassland habitat properties. However, because the concentration of grassland specialists not only depends on environmental variables but also on the age and spatial context of the grasslands, the NDVI is unlikely to allow the identification of grasslands with high numbers of specialist species.

Responses of plant diversity and primary productivity to nutrient addition in a Stipa baicalensis grassland, China

Nutrient addition can affect the structure and diversity of grassland plant communities, thus alter the grassland productivity. Studies on grassland plant community composition, structure and diversity in response to nutrient addition have an important theoretical and practical significance for the scientific management of grassland, protection of plant diversity and the recovery of degraded grassland. A randomized block design experiment was conducted with six blocks of eight treatments each: control (no nutrient addition) and K, P, N, PK, NK, NP, and NPK addition. We evaluated plant composition, height, coverage, density, and aboveground biomass to estimate primary productivity and plant diversity. Results showed that all treatments increased primary productivity significantly (P<0.05) with the exception of the K and the NPK treatments had the greatest effect, increasing aboveground biomass 2.46 times compared with the control (P<0.05). One-way ANOVA and factorial analysis were used for the species richness, Shannon-Wiener index, Pielou index and aboveground biomass, and the relationships between the diversity indices and aboveground biomass were determined through linear regression. We found that fertilization altered the community structure; N (but not P or K) addition increased the proportion of perennial rhizome grasses and significantly reduced that of perennial forbs (P<0.05), thus it presented a trend of decrease in species richness, Shannon-Wiener and Pielou indexex, respectively. Only the main effects of N had significant impacts on both the diversity indices and the aboveground biomass (P<0.05), and the interactions between N-P, N-K, P-K and N-P-K could be neglected. With fertilization, plant diversity (correlation coefficient, −0.61), species richness (−0.49), and species evenness (–0.51) were all negatively linearly correlated with primary productivity. The correlations were all significant (P<0.01). Scientific nutrient management is an effective way to improve grassland productivity, protect the plant diversity as well as recover the degraded grassland.

The effect of soil pH on persistence of seeds of grassland species in soil

Soil pH is one of the main factors determining grassland plant community composition. However, its effects on seed persistence are not well understood, and it is not clear whether soil pH influences seed persistence, either directly or indirectly through microbial pathogens. This study examined the soil seed bank along a natural pH gradient from acidic to calcareous grassland in the Peak District National Park, UK. In addition, a seed burial experiment was performed using three species and including a fungicide treatment. Seeds of a calcicole species (Scabiosa columbaria), a calcifuge (Hypericum pulchrum) and a third species growing over a wide range of soil pH (Campanula rotundifolia) were buried, and seed germination, viability and damage (both physical and fungal) were recorded. Increasing soil pH was correlated with decreased total and grass seed abundance in the seed bank, and with a decline in the probability that H. pulchrum seeds persisted, mainly due to physical damage. In soils with pH higher than 5.6, fungicide increased seed persistence of all three species. Viability of intact, undamaged seeds and germination were not affected by soil pH. Decreasing soil pH was correlated with an increased proportion of persistent seeds. Results suggest that acid soils are associated with increased seed persistence, the size and longevity of grassland seed banks decrease as soil pH increases, and that pH influences seed persistence by an indirect effect mediated by microbial pathogens, while germination is not influenced by pH.

Litter accumulation drives grassland plant community composition and functional diversity via leaf traits

Direct physical interference from litter is an important mechanism driving changes in the diversity and composition of plant communities dominated by perennial species. While the consequences of litter for species diversity and composition are well known, the mechanisms shaping how litter acts as a filter on leaf and shoot functional traits to alter community structure and functional diversity remain unclear. These questions were examined using a large field experiment in native mixed-grass prairie grasslands in the brown and dark brown soil zones in central Saskatchewan, Canada. Litter additions ranging from 0 to 2,290 g m−2 were applied to plots in a response surface design, and responses including community structure, above-ground functional diversity, and community-weighted mean values for leaf functional traits were measured. Species richness, functional richness, and functional dispersion declined with increasing litter mass, but functional evenness increased. Community composition became more variable because increased litter acted as a filter on the perennial species previously present in the plots; the remaining species in a given plot were the most litter-tolerant subset of the original community in the plot. Increased litter mass affected graminoid and forb species differently. Increased litter mass selected for graminoid species characterized by greater height, leaf area, specific leaf area (SLA), leaf thickness, and leaf tissue density. In contrast, increased litter selected for forb species characterized by larger, thicker leaves, lower SLA, and lower leaf tissue density. Forbs with cauline leaves were also selected for over acaulescent growth forms. Physical interference by litter likely reduces grassland diversity and alters species composition in diverse grassland communities by forcing convergence in community trait structure, but not convergence in species composition.

Invertebrate, not small vertebrate, herbivory interacts with nutrient availability to impact tallgrass prairie community composition and forb biomass

The effects of herbivores and their interactions with nutrient availability on primary production and plant community composition in grassland systems is expected to vary with herbivore type. We examined the effects of invertebrate and small vertebrate herbivores and their interactions with nutrient availability on grassland plant community composition and aboveground biomass in a tallgrass prairie ecosystem. The abundance of forbs relative to grasses increased with invertebrate herbivore removals. This increase in forb abundance led to a shift in community composition, where invertebrate removals resulted in greater plant species evenness as well as a divergence in composition among plots. In contrast, vertebrate herbivore removals did not affect plant community composition or aboveground biomass. Nutrient additions alone resulted in a decrease in plant species richness and an increase in the abundance of the dominant grass, but the dominant grass species did not greatly increase in abundance when nutrient additions were combined with invertebrate removals. Rather, several subdominant forbs came to dominate the plant community. Additionally, the combined nutrient addition and invertebrate herbivore removal treatment increased forb biomass, suggesting that invertebrate herbivores suppress the responses of forb species to chronic nutrient additions. Overall, the release of forbs from invertebrate herbivore pressure may result in large shifts in species composition, with consequences for aboveground biomass and forage quality due to altered grass:forb ratios in grassland systems.

Sensitivity of grassland plant community composition to spatial vs. temporal variation in precipitation

Climate gradients shape spatial variation in the richness and composition of plant communities. Given future predicted changes in climate means and variability, and likely regional variation in the magnitudes of these changes, it is important to determine how temporal variation in climate influences temporal variation in plant community structure. Here, we evaluated how species richness, turnover, and composition of grassland plant communities responded to interannual variation in precipitation by synthesizing long-term data from grasslands across the United States. We found that mean annual precipitation,(MAP) was a positive predictor of species richness across sites, but a positive temporal relationship between annual precipitation and richness was only evident within two sites with low MAP. We also found higher average rates of species turnover in dry sites that in turn had a high proportion of annual species, although interannual rates of species turnover were surprisingly high across all locations. Annual species were less abundant than perennial species at nearly all sites, and our analysis showed that the probability of a species being lost or gained from one year to the next increased with decreasing species abundance. Bray-Curtis dissimilarity from one year to the next, a measure of species composition change that is influenced mainly by abundant species, was insensitive to precipitation at all sites. These results suggest that the richness and turnover patterns we observed were driven primarily by rare species, which comprise the majority of the local species pools at these grassland sites. These findings are consistent with the idea that short-lived and less abundant species are more sensitive to interannual climate variability than longer-lived and more abundant species. We conclude that, among grassland ecosystems, xeric grasslands are likely to exhibit the greatest responsiveness of community composition (richness and turnover) to predicted future increases in interannual precipitation variability. Over the long-term, species composition may shift to reflect spatial patterns of mean precipitation; however, perennial-dominated systems will be buffered against rising interannual variation, while systems that have a large number of rare, annual species will show the greatest temporal variability in species composition in response to rising interannual variability in precipitation.

Direct and indirect effects of roads and road vehicles on the plant community composition of calcareous grasslands

Exposure of plants to vehicle exhaust emissions and road-induced changes to soil biogeochemistry and hydrology can lead to shifts in plant composition in calcareous grasslands. Mixed effects models were used to identify relationships between plant community composition and a suite of measured and modelled environmental variables along transects away from roads at eight calcareous grasslands. Ellenberg pH, moisture and nitrogen (N) scores increased nearer roadsides, however, only Ellenberg N scores were associated with their respective measured or modelled values highlighting NO2 deposition as a likely driver of change. Forb abundance and diversity increases nearer roadsides were also associated with NO2 deposition, with increases seen in the abundance and diversity of typical edge species rather than species characteristic of calcareous grasslands. Grazing, removal of invasive species and the use of barriers to intercept transport-derived air pollution may help to reduce the detrimental effects of roads across these diverse but threatened landscapes.

Fifty years of change in Central European grassland vegetation: Large losses in species richness and animal-pollinated plants

There is growing concern that biodiversity loss in European agricultural landscapes is having negative effects on functional trait diversity. Long-term studies examining vegetation changes from the period before agricultural industrialisation are however rare. Here, we ask how management intensification and increased nutrient input initiated in the 1950/1960s have altered grassland plant community composition, species diversity and functional trait composition using comprehensive datasets from five floodplain regions (plus one protected reference region) in northern Germany. Sites with available historical relevés and vegetation maps (1950/1960s, 1990s) were resampled in 2008 to facilitate the analysis of a period spanning four to five decades.Plant community composition changed tremendously in all study regions during the 50year period, which was related to increasing Ellenberg indicator values for nutrient availability. Species richness at the plot-level fell by 30–50% over the period, and losses in functional diversity were equally large. A non-formal comparison with the results from the protected reference study region indicates that the changes may mostly be attributable to local nutrient input rather than to supra-regional climate change. Our results indicate a consistent trend towards much more species-poor communities dominated by mow-tolerant, N-demanding competitive grasses, whereas species with more ruderal strategies, species flowering early in the season and, in particular, insect-pollinated herbs have all decreased. The substantial loss of nectar-producing grassland herbs is likely to have negative effects on the abundance of pollinating insects, with consequences for the grassland animal communities. This highlights the growing need for adequate grassland management schemes with low N input to preserve high-nature-value grassland.

Plant community response to loss of large herbivores: comparing consequences in a South African and a North American grassland

Loss of biodiversity poses one of the greatest threats to natural ecosystems throughout the world. However, a comprehensive understanding of the impacts of species losses from upper trophic levels is still emerging. Here we compare the impacts of large mammalian herbivore species loss on grassland plant community structure and composition in a South African and North American grassland. Herbaceous plant communities were surveyed at sites without large mammalian herbivores present and at sites with a single species of herbivore present in both locations, and additionally at one site in South Africa with multiple herbivore species. At both the North American and South African locations, plant communities on sites with a single herbivore species were more diverse and species rich than on sites with no herbivores. At the multi-herbivore site in South Africa, plant diversity and richness were comparable to that of the single herbivore site early in the growing season and to the no herbivore site late in the growing season. Analyses of plant community composition, however, indicated strong differences between the multi-herbivore site and the single and no herbivore sites, which were more similar to each other. In moderate to high-productivity ecosystems with one or a few species of large herbivores, loss of herbivores can cause a significant decrease in plant diversity and richness, and can have pronounced impacts on grassland plant community composition. In ecosystems with higher herbivore richness, species loss may also significantly alter plant community structure and composition, although standard metrics of community structure may obscure these differences.

Plant Community and Soil Microbial Carbon and Nitrogen Responses to Fire and Clipping in a Southern Mixed Grassland

Disturbances, such as fire and grazing, play important roles in determining grassland plant community composition and soil microbial dynamics, as well as regulating the flows of carbon (C) and nitrogen (N) between the two groups of organisms. In a mixed grassland of the southern Great Plains, we tested the hypotheses that spring-season fire would increase the absolute biomass and relative proportion of C4 grasses in the plant community, and decrease soil microbial biomass N, thereby increasing microbial C∶N ratios. We also tested the hypothesis that clipping (to simulate grazing) would reduce effects of fire, with a greater reduction of fire effect corresponding to an increased frequency of clipping. Contrary to our hypothesis, C4 grasses showed no significant treatment responses. Treatment effects were limited to C3 grasses, and clipping was more important than fire in terms of effects on plant community composition. However, because of its greater capacity to reduce aboveground litter, fire had the greater impact on soil microbial C. Contrary to the hypothesized outcome, no significant effects of disturbance on soil microbial N were observed. This suggests that control of N cycling in this ecosystem is primarily microbial in nature, though dependent on inputs of plant C via litter. Interactions between fire and clipping were observed in litter mass, highlighting the importance of litter inputs for plant–soil nutrient feedbacks.

Native harvester ants threatened with widespread displacement exert localized effects on serpentine grassland plant community composition

Seed‐harvesting ants can influence the abundance and distribution of plant species through both the selective harvesting of seeds and the construction of nutrient‐rich nest mounds, but the relative contributions of these two mechanisms have not been addressed by previous studies. Furthermore, the impact of ant seed harvesting in California serpentine grasslands remains unresolved because of divergent results from several previous experiments. This study investigates the influence of harvester ants on serpentine grassland plant species composition by examining two potential signatures of seed harvesting ants on plant community composition: species composition on versus off ant nest mounds, and species abundance as a function of distance from nest mounds. Of the 28 plant species identified in this study, 22 exhibited spatial patterns consistent with effects of seed harvesting, nest construction, or both. Although most species showed significant gradients in abundance with distance from a nest, there were no clear relationships between plant species distributions and previously reported harvester ant seed foraging patterns. Harvester ant nest mounds supported plant communities that were distinct from the surrounding serpentine grassland, with notably higher densities of legumes and invasive annual grasses. Comparison of our results with those of previous studies indicates that the patterns we observed are generally consistent over time, but affect more species and a larger fraction of the grassland than previously reported. Unaffected areas of the grassland seem likely to serve as important refuges for some plant species.

Host plant differences in arbuscular mycorrhizae: Extra radical hyphae differences between an invasive forb and a native bunchgrass

Arbuscular mycorrhizae affect grassland plant community composition and host plant nutrient uptake, and can mediate shifts in competitive outcome between plant species. Centaurea maculosa, an invasive forb from Eurasia, dominates more than 4 million hectares in the Rocky Mountain region of North America. We examined the role of AM for phosphorus (P) acquisition from a distant source for C. maculosa and Festuca idahoensis, a native bunchgrass. Plants were grown individually in pots divided by a barrier that either excluded plant roots and AM hyphae, or only plant roots. In the half of the pot without a plant, 1 of 3 P treatments was applied: no P, phosphate rock (PR) or triple superphosphate (TSP), applied at a rate of 144 mg P kg−1 soil. After 14 weeks of growth, C. maculosa was twice as large as F. idahoensis, and neither species’ biomass was affected by barrier type. Phosphorus fertilizer, and especially PR, moved across the barrier to the plant side of the pot. Tissue P concentration for C. maculosa was highest with the PR treatment, and was not affected by the barrier type. In contrast, F. idahoensis tissue P concentration did not vary with barrier or P treatments. There was more AM extra radical hyphae (ERH) associated with C. maculosa than F. idahoensis, suggesting that C. maculosa provides more carbon for the AM fungi, resulting in greater ERH production, ERH soil exploration and potential for soil nutrient pool exploitation. Although not tested in this study, differences between host plants may be the result of different physiological characteristics of the host plant or differences in AM fungal species that colonize the invader, with different fungal species accessing P from different distances.

Rainfall, Biomass Variation, and Community Composition in the Park Grass Experiment

We used data on grassland plant community composition over a 90—yr period in the history of the Park Grass Experiment, England to look for relationship between variation in composition and annual variation in rainfall and biomass. This was investigated by regressions of biomass and rainfall on each other, and of these variables separately on each of three different measures of variation in plant community composition. Two of these measures, principal components analysis scores based on variation in species abundance and the ratio by mass of nongrass/grass species, showed significant relationships with biomass variation on many experimental plots, although relationships with rainfall were relatively slight or nonexistent. The third measure employed similarity indices to detect changes in species composition in response to variation in biomass, but failed to find any. Biomass was significantly increased by rainfall on all plots. We propose that variation in community composition was more closely related to biomass variation than to rainfall because rainfall selectively favored the grasses in the community, which we believe competed asymmetrically (for light) with the other species when rainfall was high. The severity of this competition would depend upon biomass more directly than upon rainfall, although it is rainfall that enhanced grass growth. In effect, asymmetric competition magnified the effect of rainfall on community composition.