Effects of spring precipitation on a temperate arable collembolan community analysed using Principal Response Curves

We hypothesized that changes in the pattern of spring precipitation would alter the species composition of a temperate soil-surface collembolan community. Experimental manipulation of precipitation during a drought in southern England was used to test this hypothesis in spring 1997. Replicated plots in a field of spring peas received spring drought (plots shielded from rainfall), reference (actual) rainfall and spring irrigation during the first 2 weeks of May. Shielding plots extended the existing drought to 58 days. Thereafter all plots received natural precipitation only, which in June considerably exceeded the long-term average for the site. Redundancy Analysis (RDA) and the recently-developed method of Principal Response Curves analysis (PRC) were used to summarise and test statistically changes in community composition under the three precipitation treatments, using counts of soil-surface species obtained by suction sampling. Irrigation decreased counts of Bourletiella hortensis, but had a positive impact on all other species studied, whilst the effect of spring drought was negative. Effects of spring precipitation persisted for the 97-day duration of the study despite the unusually wet summer. The possibility that changes in patterns of rainfall in northern Europe would promote some species as crop pests is proposed, but not supported by our findings. We use the collembolan data to illustrate some advantages of PRC analysis over RDA and other ordination techniques for analysing community responses to experimental treatments.