Short time series obscure compensatory dynamics in ecological communities.

The degree of synchronous versus compensatory dynamics among species is crucial for determining the stability of ecological communities. Although robust quantification of species synchrony requires long-term observations, empirical studies are often based on short time series. Here we explore the effects of time series length on species synchrony by combining spectral analysis, dynamical community models and empirical plant community data. Our theoretical analyses show that competition contributes to decreasing species synchrony over long timescales but causes increases in synchrony over short timescales. As a result, species synchrony tends to decrease with time series length. In model communities, species synchrony calculated from long time series decreases with species diversity and competition, whereas that calculated from short time series increases with diversity and competition. Empirical analyses of >2,000 time series of plant communities support these theoretical predictions. Our analyses demonstrate that both species synchrony itself and its relationship with species richness can exhibit opposite patterns, depending on the length of time series, challenging the implicit assumption in ecological studies that observational length should not qualitatively alter patterns of interest. Our findings help reconcile results from theoretical and empirical studies on synchrony and have implications for sampling design.