A combination of hydrodynamical and statistical modelling reveals non-stationary climate effects on fish larvae distributions

Abstract

Biological processes and physical oceanography are often integrated in numerical modelling of marine fish larvae, but rarely in statistical analyses of spatio-temporal observation data. Here, we examine the relative contribution of inter-annual variability in spawner distribution, advection by ocean currents, hydrography and climate in modifying observed distribution patterns of cod larvae in the Lofoten -Barents Sea. By integrating predictions from a particle-tracking model into a spatially explicit statistical analysis, the effects of advection and the timing and locations of spawning are accounted for. The analysis also includes other environmental factors: temperature, salinity, a convergence index and a climate threshold determined by the North Atlantic Oscillation (NAO). We found that the spatial pattern of larvae changed over the two climate periods, being more upstream in low NAO years. We also demonstrate that spawning distribution and ocean circulation are the main factors shaping this distribution, while temperature effects are different between climate periods, probably due to a different spatial overlap of the fish larvae and their prey, and the consequent effect on the spatial pattern of larval survival. Our new methodological approach combines numerical and statistical modelling to draw robust inferences from observed distributions and will be of general interest for studies of many marine fish species.

Scale of inference: on the sensitivity of habitat models for wide‐ranging marine predators to the resolution of environmental data

“…Moreover, non‐stationarity in animal–environmental interactions through space and time, including behavioural plasticity in movement strategies, diminishes the predictive capacity of presence‐availability models constructed using climatological fields (Hidalgo et al , Dodge et al , Schmidt et al ). Many species distribution models for wide‐ranging species show evidence of poor extrapolation through space and time (Elith and Leathwick , Torres et al ).…”

“…Coarse-scale climatological models projected onto coarse-scale climatological fields appear likely to lead to more severe prediction error than averaging predictions made on contemporaneous environmental data fields, particularly where animals respond strongly to the contemporaneous physical environment. Moreover, non-stationarity in animal-environmental interactions through space and time, including behavioural plasticity in movement strategies, diminishes the predictive capacity of presence-availability models constructed using climatological fields (Hidalgo et al 2012, Dodge et al 2014, Schmidt et al 2014. Many species distribution models for wide-ranging species show evidence of poor extrapolation through space and time (Elith andLeathwick 2009, Torres et al 2015).…”

Modelling dendritic ecological networks in space: an integrated network perspective

“…For example, the effects of ocean currents on larval dispersal (Hidalgo et al . ); patterns of dissolved oxygen in estuaries (Rathbun ); animal movement along habitat corridors (Castellón & Sieving ); and plant (Spooner et al . ) and animal dispersal along road networks (Brock & Kelt ).…”

Disentangling the mechanisms behind climate effects on zooplankton

“…Individual-based particle tracking models are considered a valuable tool to incorporate the role of advection (23), but although the results of such models commonly are compared and calibrated with observation data, they are rarely directly used in statistical analyses of observation data (but see refs. [24][25][26].…”