Temporal variation in effect sizes in a long-term, split-plot field experiment

Ecological field experiments initiate successional and evolutionary changes among resident species, yet effect sizes are often reported as if they were constants. Few ecological studies have addressed their questions through long-term, experimental approaches, and many questions remain unanswered regarding temporal patterns in ecological effect sizes. We document temporal variation in effect sizes in response to pulse and press manipulations in a long-term factorial field experiment at Nash's Field, England. The experiment comprises seven treatments applied in a split-plot design to test the single and interactive effects of herbivory by insects, molluscs, and rabbits, liming, nutrient limitation (applied as press experiments), competition (exclusion of grasses or herbs with specific herbicides), and seed limitation (pulse experiments) on plant community dynamics. The response of all vascular plant species was followed for two decades. High species richness was positively related to the minus-grass herbicide in the first decade and negatively related to both nitrogen addition and the abundance of dominant species in both decades. Many significant effects appeared quickly, but some large effects were not detected until year 15. Press experiments produced some long-lasting effects, but effect sizes changed due to both idiosyncratic "year effects" and secular trends. For pulse experiments, most effects, including positive and negative responses to herbicide application and the invasion of most of the sown species, disappeared quickly. However, some endured or grew monotonically, such as the invasion of two sown species that benefited from particular combinations of the press treatments. The fastest effects to appear were the responses from established species. Many of these responses were negative, likely resulting from reduced niche dimensionality and competitive exclusion by new dominant species. Contrarily, one of the largest community-level effects took well over a decade to appear: the natural invasion by one species, which responded to a four-way interaction between experimental treatments. The insights gained from individual effects increased with the duration of the lag before their first appearance, drawing attention to the importance of long-term, manipulative field experiments. This experiment also reinforces the point that factorial experiments are the most insightful way to explore ecological interactions.