Cascaded neural networks improving fish species prediction accuracy: the role of the biotic information

Abstract

Species distribution is the result of complex interactions that involve environmental parameters as well as biotic factors. However, methodological approaches that consider the use of biotic variables during the prediction process are still largely lacking. Here, a cascaded Artificial Neural Networks (ANN) approach is proposed in order to increase the accuracy of fish species occurrence estimates and a case study for Leucos aula in NE Italy is presented as a demonstration case. Potentially useful biotic information (i.e. occurrence of other species) was selected by means of tetrachoric correlation analysis and on the basis of the improvements it allowed to obtain relative to models based on environmental variables only. The prediction accuracy of the L. aula model based on environmental variables only was improved by the addition of occurrence data for A. arborella and S. erythrophthalmus. While biotic information was needed to train the ANNs, the final cascaded ANN model was able to predict L. aula better than a conventional ANN using environmental variables only as inputs. Results highlighted that biotic information provided by occurrence estimates for non-target species whose distribution can be more easily and accurately modeled may play a very useful role, providing additional predictive variables to target species distribution models.

Endemic Freshwater Fish Range Shifts Related to Global Climate Changes: A Long-Term Study Provides Some Observational Evidence for the Mediterranean Area

“…The present research has shown fish distributional changes in a context of climate warming; these results, based on observational data, can be useful for planning future management scenarios [78], also in light of the fact that the climate models predict further strong impacts on freshwater fish [64]. In particular, the possible conservation strategies that can be suggested for fish biodiversity conservation under climate-induced impacts are: (i) the maintenance of ecological flow, whereas the natural variability of the flow rate is indispensable for the conservation of habitat suitable for the life of freshwater fishes [3]; (ii) the restoration of river connectivity in order to provide opportunity, to fish species with low thermal tolerance, to move towards thermal optimum considering, however, the possibility of maintaining barriers in the event of a need to prevent the rise of exotic species; (iii) the establishment of protection areas where total autochthonous assemblages are preserved and thus play a key role in maintaining biodiversity; (iv) improvement of water quality in polluted rivers is needed to reduce additional anthropogenic stressors for resilient fish populations to adapt to climate change; (v) the restoration of riparian areas to provide shade and have directly cooling effect on river temperature [79].…”

An ecologically constrained procedure for sensitivity analysis of Artificial Neural Networks and other empirical models

“…In this respect, the ecological variables that can be taken into account are often characterized by complex and non-linear dependencies [7]. Ecological models have been increasingly applied in the management and conservation of freshwater fish communities, especially to predict spatial patterns of fish occurrence [8], [9]. In particular, Artificial Neural Networks (ANNs) modeling has proved to be a valuable method in order to assess whether predictable relationship between environmental descriptors and fish species richness exist in small stream environments [10], [11], [12].…”

Predicting Fishing Footprint of Trawlers From Environmental and Fleet Data: An Application of Artificial Neural Networks

“…Thus, each MPN2 has a single output neuron. CMPN has been applied in other ecological studies when a correlation exists among output variables (Franceschini et al, 2018).…”

“…This structural characteristic of MPN can be used to evaluate the relative influence of the input variables: if the values of a single input variable are changed (while those for the other input variables are kept the same) it is possible to observe the related effects on the outputs. Therefore, if this procedure is applied, for each input variable, by perturbing the input values at defined levels, the comparisons of the effects on the output values can be used to assess the relative influence (importance) of the input variables (Scardi and Harding, 1999;Franceschini et al, 2018). Hence, to evaluate the contributions of the independent variables to the prediction accuracies, a set of perturbation levels [initial values 5 ±0.1, ±0.2, ±0.3, ±0.4, and ±0.5] was applied, and the related effects on the output values were measured as the mean square errors (MSE) of the CMPN outputs with respect to the values in the test datasets.…”

“…The method is based on artificial neural networks (ANN), a group of modeling techniques that imitate the functioning of the human brain to analyze large and complex datasets characterized by non-linear relationships among variables, internal redundancy and noise (Lek and Guégan, 1999). We designed a cascaded multilayer perceptron network (CMPN hereafter - Watts and Worner, 2008) that is a sequence of two multilayer perceptron networks (MPN) in which the output from one MPN becomes the input to another MPN (Franceschini et al, 2018). MPN represents the simplest and most widely used ANN architecture FIGURE 1 | Frequencies of the number of trawlers by country and length class equipped with the VMS (green) and without VMS (red).…”