Predicting demersal fish species distributions in the Mediterranean Sea using artificial neural networks

Predicting the occurrence of economically important demersal fish in a multispecies marine environment can be of considerable value to fisheries management and protection of biodi- versity. Here, 2 predictive modelling principles were utilised, artificial neural network (ANN) and discriminant function analysis (DFA), to develop presence/absence models for 3 species (anglerfish Lophius budegassa; hake Merluccius merluccius; red mullet Mullus barbatus) in the Mediterranean Sea. ANN-based models of demersal fish distribution outperformed conventional models and attained better recognition and prediction performance. Results indicated the ability of ANN's to pre- dict presence more accurately than DFA when tested against independent field data. More precisely, sensitivity values obtained using DFA were 62.1% for anglerfish, 5.8% for hake and 59.8% for red mullet whereas using ANN were 75, 71 and 72.9% respectively. The accuracy of test data was 79.6% for anglerfish, 49.5% for hake and 83.3% for red mullet using DFA and 83.7, 83.3 and 85.6% respec- tively using a back-propagation ANN. After learning from a set of selected patterns, the neural net- work (NN) models displayed a relatively high demersal fish classification accuracy, which was con- sistent with present understanding of the aggregating effects of the examined variables on these species' distribution. Predicting presence or absence was found to be easier for red mullet and anglerfish than for hake. The present results also suggested that the main processes modulating the occurrence of anglerfish, hake and red mullet in the NE Mediterranean Sea can be approximated by linear functions only to a limited extent. Due to their ability to mimic non-linear systems, ANNs proved far more effective in modelling the distribution of these species in the marine ecosystem. The main results and the ANN potential to predict suitable habitat profiles and structural characteristics of species assemblages are discussed.

[1]  Sovan Lek,et al.  Stochastic models that predict trout population density or biomass on a mesohabitat scale , 1996, Hydrobiologia.

[2]  Hal S. Stern,et al.  Neural networks in applied statistics , 1996 .

[3]  Simon Ferrier,et al.  Evaluating the predictive performance of habitat models developed using logistic regression , 2000 .

[4]  I. Dimopoulos,et al.  Role of some environmental variables in trout abundance models using neural networks , 1996 .

[5]  R. Kronmal,et al.  Discriminant functions when covariances are unequal and sample sizes are moderate , 1977 .

[6]  Geoffrey E. Hinton,et al.  Learning internal representations by error propagation , 1986 .

[7]  Michele Scardi,et al.  Advances in neural network modeling of phytoplankton primary production , 2001 .

[8]  Charles E. McCulloch,et al.  MULTIVARIATE ANALYSIS IN ECOLOGY AND SYSTEMATICS: PANACEA OR PANDORA'S BOX? , 1990 .

[9]  Brian D. Ripley,et al.  Neural Networks and Related Methods for Classification , 1994 .

[10]  S. Lek,et al.  The use of artificial neural networks to assess fish abundance and spatial occupancy in the littoral zone of a mesotrophic lake , 1999 .

[11]  Tomaso A. Poggio,et al.  Regularization Theory and Neural Networks Architectures , 1995, Neural Computation.

[12]  K. Mann,et al.  Physical oceanography, food chains, and fish stocks: a review , 1993 .

[13]  Sovan Lek,et al.  Predicting fish distribution in a mesotrophic lake by hydroacoustic survey and artificial neural networks , 1999 .

[14]  S. Georgakarakos,et al.  Artificial neural networks as a tool for species identification of fish schools , 1996 .

[15]  Elie Bienenstock,et al.  Neural Networks and the Bias/Variance Dilemma , 1992, Neural Computation.

[16]  Ron Kohavi,et al.  A Study of Cross-Validation and Bootstrap for Accuracy Estimation and Model Selection , 1995, IJCAI.

[17]  S. Lek,et al.  The use of artificial neural networks to predict the presence of small‐bodied fish in a river , 1997 .

[18]  A. T. C. Goh,et al.  Back-propagation neural networks for modeling complex systems , 1995, Artif. Intell. Eng..

[19]  K. Stergiou,et al.  Biology and fisheries of eastern Mediterranean hake (M. merluccius) , 1995 .

[20]  David G. Reid,et al.  Identifying the effects of oceanographic features and zooplankton on prespawning herring abundance using generalized additive models , 1997 .

[21]  M. Scardi.  Artificial neural networks as empirical models for estimating phytoplankton production , 1996 .

[22]  G. David Garson,et al.  Interpreting neural-network connection weights , 1991 .

[23]  G. Potamias,et al.  Species distribution in the southern Aegean sea based on bottom-trawl surveys , 1999 .