Nonlinear ecological processes driving the distribution of marine decapod larvae

Abstract

The complexity of the natural processes lead to many nonlinear interacting factors that in uence the distribution and survival of marine pelagic species, particularly in their larval phase. The management of these ecosystems require techniques that unveil those interactions by studying the system globally, including all relevant variables and combining both community and environmental data in a single step. Specifically, we apply an unsupervised neural network, the Self-Organizing Map (SOM), to a combined dataset of environmental and decapod larvae community data from the Balearic sea, obtained in two years with contrasting environmental scenarios, as an Exploratory Data Analysis (EDA) technique that provides a global and more detailed view of both the environmental processes and their in uence on the distribution of such planktonic community. We examine the parental in uence on the initial larval distribution by aggregating data by adult habitat, which also increments the signal to noise ratio (mean data patterns over noise due to outliers or measurement errors), and consider the distribution of larvae by development stage (as a proxy of age and hence of potential dispersion). The joined study of parental e ect, drifting or concentration events determined by dynamical processes in the whole water column, and lifespan, draws the possible paths followed by larvae, and highlights the more in uencing variables in their distribution. Investigation of the di erent aspects of dynamic height (absolute values, gradients or edges and correlations) clari ed the e ect of the oceanographic processes on decapods' larvae.