Use of temporal variation in population and climatic synchrony to identify causal mechanisms

Background/Question/Methods

Synchrony in population fluctuations over long distances is often attributed to correlations in environmental stochasticity, but identifying the climatic factors generating synchrony remains difficult. Analyses frequently focus on finding similar synchrony distance-decay relationships between the population and a climatic variable, but dispersal and heterogeneity in population dynamics can make interpretation difficult. Comparing temporal fluctuations in synchrony of both the population and suspected climatic factors may provide further evidence for causal relationships. Here, we study temporal variation in synchrony of gypsy moth (Lymantria dispar) populations (S_GM) across the northeastern United States from 1975-2009. We compared these values with temporal variation in synchrony of total precipitation in June (S_June), which has been previously identified as a potential synchrony generating mechanism. Finding no previous theoretical study, we examined a simple metapopulation model to support the relationships we found. In this model, the dynamics of the 100 local populations were governed by identical second order autoregressive model. The populations were linked by correlated environmental stochasticity, but the strength of the correlation (ρ_t) fluctuated through time. We varied the cycle length, amplitude, and average value of fluctuations, as well as the population autoregressive parameters governing the population dynamics.

Results/Conclusions

Temporal variation in the synchrony of gypsy moth S_GM was correlated with synchrony in S_June of the previous year. Additional relationships between S_GM and outbreak phase, and between S_June and population density suggested that S_June may influence the timing of gypsy moth outbreaks. No relationship was found between S_June and total precipitation in June, indicating that neither drought nor particularly wet years were disproportionately responsible for the observed S_June patterns.

The simulation model supported these results, though the strength of the relationships were dependent on the autoregressive parameters used. Parameters producing strong oscillatory behavior led to a strong relationship between outbreak phase and population synchrony, but no relationships with ρ_t. As tendency towards oscillatory behavior was reduced, outbreak phase had less of an effect, and correlations between ρ_t and both population synchrony and population size increased in strength. Autoregressive parameters fit to the actual gypsy moth population data fell in the middle, where all three relationships would be expected to occur at moderate levels. Our results support previous studies linking synchrony in precipitation with synchrony in gypsy moth populations, and may provide an alternative method of identifying potential causal synchrony mechanisms in other species.