Community composition varies in response to many different factors. For example, spatial variation may reflect differences in colonization history and abiotic conditions, while temporal variation may reflect interannual weather patterns and successional dynamics. When spatial and/or temporal variation are considerable, it may be difficult to detect or understand the effects of management actions. Therefore, we examined the relative importance of these sources of variation, and why they differ among sites.
We analyzed compositional data from 39 grassland sites from the Nutrient Network, a globally distributed collaborative experiment. At each site, nitrogen (N), phosphorus (P), and potassium plus micronutrients (K+) were applied in a factorial design to 24 plots (8 treatments; 1 replicate plot per block; 3 blocks). Composition was assessed annually for 4 years (pre-treatment plus 3 years post-treatment) in each plot. Sites were analyzed separately due to large compositional differences among them. Composition was quantified using Bray-Curtis distance. The dissimilarity matrix was partitioned and the variation associated with each term (block, treatment, year, and interactions) expressed as a percentage. How the variation was partitioned among sites was examined via Principal Component Analysis, and related to potential explanatory variables. Treatment effects were assessed using PERMANOVA.
Results/Conclusions
Most of the variation in composition (>80%) was explained by two PCs, one strongly related to temporal variation (year) and the other to spatial variation (block, block:treatment). Temporal variation (PC1) was strongly positively related to elevation, nitrogen deposition, and climate, but spatial variation (PC2) was not related to any of the explanatory variables. Treatment effects were much less pronounced than spatial and temporal variation in composition: only 20% of sites exhibited compositional differences among fertilizer treatments. N affected composition significantly at 12 sites, compared to 8 sites for P and 3 sites for K+.
This study demonstrates that temporal and spatial variation in composition often greatly exceed the variation associated with fertilizer treatments, and can confound the detection of treatment-related compositional changes. Managers need to understand the relative importance of these sources and incorporate them into monitoring strategies, such as when deciding how many plots to establish, whether to use permanent or temporary plots, and how many years to monitor them. Future work should examine whether temporal and treatment-related compositional change reflect directional change, divergence, or convergence.