Geographical variability in the controls of giant kelp biomass dynamics

Abstract Aim Coastal marine environments experience a wide range of biotic and abiotic forces that can limit and punctuate the geographical range and abundance of species through time. Determining the relative strengths and nonlinear effects of these processes is vital to understanding the biogeographical structures of species. There has been an ongoing discussion concerning the relative importance of these processes in controlling the dynamics of giant kelp, Macrocystis pyrifera , an important structure‐forming species on shallow reefs in the eastern Pacific. We used novel spatial time‐series that span nearly three decades to determine the dominant drivers of giant kelp canopy biomass and the temporal and spatial scales over which they operate across the dominant range of the giant kelp in North America. Location Near‐shore areas from Año Nuevo, California, to the USA /Mexico border. Methods We employed empirical orthogonal functions to elucidate the primary drivers of giant kelp canopy biomass across space and time and then fit generalized additive and linear models to determine the nonlinear effect and relative importance of each of these potential drivers along the c . 1500‐km study region over a 25‐year period. Results Wave disturbance, nitrate availability and the state of the North Pacific Gyre Oscillation were the most important environmental predictors of giant kelp canopy biomass, explaining 24.5%, 12.7% and 6.1% of the variance, respectively. Environmental drivers of canopy biomass exhibited profound spatial differences in relative effect sizes. Nonlinear effect shapes of each potential biomass driver were determined, which explained these spatial differences. Main conclusions These large‐scale analyses help to reconcile the local‐scale conclusions of canopy biomass dynamics across the California coastline and show that these dynamics differ predictably in space and time in accordance with local and regional differences in environmental drivers. By characterizing the nonlinear effects of these drivers, we identified spatio‐temporal patterns of processes that cannot be detected by remote sensing.