MEPS 583:15-34 (2017)  -  DOI: https://doi.org/10.3354/meps12286

ABSTRACT: Models that incorporate complex interactions are useful tools for estimating the relative importance of ecological processes in changing ecological communities. Multispecies Markov chain models (MMCMs) use transition probabilities to explore how complex interactions influence community structure. However, the accuracy of MMCMs over different spatial and temporal scales is not well known and likely influenced by the scales over which species interactions differ. Here we explore how MMCM predictions vary with scale in the Gulf of Maine intertidal by quantifying transition probabilities at increasing temporal (annual, biannual, seasonal) and spatial (local, meso-, regional) scales. Model predictions were accurate at local scales, even with data from coarse sampling frequencies (annual), suggesting short-term variations in ecological processes do not strongly affect stable-stage community composition over these time scales. Models were less accurate when incorporating spatial variation beyond local scales. Geographic differences in species persistence, mortality, and colonization probabilities reduced MMCM predictive capacity. Our work demonstrates that MMCMs can predict local community structure, even with data from coarse sampling intervals, and may also be useful for identifying regional differences in ecological processes that might be responsible for shifts in community structure.

KEY WORDS: Multispecies Markov chain · Rocky intertidal · Gulf of Maine · Spatial variation · Temporal variation