Purchase this article with an account.
Joaquin De Juan, Noemi Martinez-Ruiz, Jose L Girela, Bassima Boughlala, David Gil; Outer retinal parameters studied with Artificial Intelligence methods predict teleost predatory behavior. Invest. Ophthalmol. Vis. Sci. 2014;55(13):2958.
Download citation file:
© ARVO (1962-2015); The Authors (2016-present)
Teleots is a successful vertebrate group, constituting more than half of vertebrate species. Its retinal structure is determined more by ecological and ethological factors, imposed for the visual system, than for belonged to a given taxonomic group. Previously we observed that telosts species with the most retinal spinules were also the most predatory and vice versa. The aim of this work was to compare several morphometrics retinal parameters with trophic and predatory behavior, in twelve species of teleosts.
The study was performed on twelve species of teleosts from different habitats. Light-adapted fishes were sacrificed and their retinas processed for optical and transmission electron microscopy studies. Thickness (µm) of Outer Nuclear Layer (ONL), Inner Nuclear Layer (INL), Outer Plexiform Layer (OPL), Inner Plexiform Laye (IPL), and density of nuclei (N/100 µm2) in ONL and INL were measured in semithin vertical sections. The number of spinules (SpN) and synaptic ribbons (SRN) per cone pedicle were counted, using thin vertical sections. Finally, species were classified into four groups according to their trophic and piscivorous levels from FishBase data, and others references. The relationship between retinal parameters and trophic and predatory levels, were studied using Factorial Analyses and Decision Tree, one powerful Artificial Intelligence method for classification and prediction.
The SpN per cone pedicle varies in a wide range between <4 and >8. Fishes with higher spinule number were also the more predatory and vice versa. A factorial analyses groups in the first factor the following parameters: SpN and SRN, trophic level values, thickness of OPL and nuclei density in ONL. Decision Tree method has been carried out using the following attributes: thickness of OPL, ONL, INL, IPL, and SpN and SRN per pedicle. The results showed that thickness of OPL and ONL were attributes that participated in the first level in the classification process with accuracy ranging between 75% and 98%.
(1) The amount of spinules per pedicle, synaptic ribbons, thickness of OPL and ONL correlates positively with carnivore and predatory behavior. (2) Decision Tree method accurately predicts the trophic and piscivorous behavior of telosts. (3) In sum, higher morphometric parameters of the outer retina mean higher piscivorous behaviors.
This PDF is available to Subscribers Only