Alain Belaud

Stochastic models that predict trout population density or biomass on a mesohabitat scale

Neural networks and multiple linear regression models of the abundance of brown trout (Salmo trutta L.) on the mesohabitat scale were developed from combinations of physical habitat variables in 220 channel morphodynamic units (pools, riffles, runs, etc.) of 11 different streams in the central Pyrenean mountains. For all the 220 morphodynamic units, the determination coefficients obtained between the estimated and observed values of density or biomass were significantly higher for the neural network (r2 adjusted= 0.93 and r2 adjusted=0.92 (p<0.01) for biomass and density respectively with the neural network, against r2 adjusted=0.69 (p<0.01) and r2 adjusted = 0.54 (p<0.01) with multiple linear regression). Validation of the multivariate models and learning of the neural network developed from 165 randomly chosen channel morphodynamic units, was tested on the 55 other channel morphodynamic units. This showed that the biomass and density estimated by both methods were significantly related to the observed biomass and density. Determination coefficients were significantly higher for the neural network (r2 adjusted =0.72 (p<0.01) and 0.81 (p<0.01) for biomass and density respectively) than for the multiple regression model (r2 adjusted=0.59 and r2 adjusted=0.37 for biomass and density respectively). The present study shows the advantages of the backpropagation procedure with neural networks over multiple linear regression analysis, at least in the field of stochastic salmonid ecology.

Modelling of microhabitat used by fish in natural and regulated flows in the river Garonne (France)

Abstract The aim of our study was to compare the microhabitat used by three fish species: brown trout ( Salmo trutta L.), European minnow ( Phoxinus phoxinus L.) and stone loach ( Barbatula barbatula L.), in natural and regulated flows of a section of the river Garonne (France). Six Artificial Neural Network (ANN) models were set up, one for each fish species in each flow condition. Models were run and tested with 1107 observations obtained by point abundance sampling performed by electrofishing. Each model had thirteen independent environmental variables (distance from the bank, water depth, water velocity, percentage of different substratum fractions defined as large boulders, small boulders, large pebbles, small pebbles, gravels, sand, mud and bedrock, flooded vegetation cover, and presence or absence of ‘blockage’ which is one or several pieces of wood providing shelter), and one dependent variable (fish density for the considered population). A cross-validation testing procedure (leave-one-out bootstrap) was performed to validate the ANN models. Finally, we used a method based on the first partial derivatives of the networks output with respect to each input to focus on the sensitivity of some of the variables selected. During the training phase, all models were judged satisfactory with Mean Squared Errors (MSE) ranging from 0.40 to 1.93, and Performance Indexes (PIs) from 60 to 89%. After the testing procedure, MSE ranged between 1.53 and 8.23, and PIs between 51 and 80%. With the exception of brown trout in regulated flow, patterns of microhabitat use obtained revealed that fish densities were highly connected to one major influencing variable: water depth for brown trout and stone loach, and water velocity for European minnow, other variables accounting for lower individual contributions. Analysis of the partial derivatives brought into relief some differences when comparing microhabitat use in natural and regulated flows for some of the variables tested, and no differences when comparing others. The results are discussed with regard to the biology and the ecology of each fish species at microhabitat and macrohabitat scales, and according to the relationship between microhabitat utilization and microhabitat availability.