Evaluation of caridean ecological distribution in the Ubatuba region, southeastern Brazilian coast using unsupervised machine learning technique

We used the unsupervised machine learning technique to evaluate the environmental factors responsible for modulating the spatial and seasonal distribution of caridean shrimps from a southeastern region of the Brazilian coast. Samplings were collected from seven transects with an artisanal shrimp fishery boat with two double-rig nets. Samplings occurred every month from October 2008 to September 2009. The most frequently captured species were Exhippolysmata oplophoroides, Leander paulensis, and Nematopalaemon schmitti. The highest abundance of shrimps occurred in autumn at the II, III, and V transects, which present a higher amount of coarse sediment and biodetritic fragments on the bottom. During autumn, the temperatures were the highest and salinity values were the lowest. Data evaluation indicated efficiency in the visualization of interactions of different shrimp species and environmental data. This kind of sediment may be allowing shrimps to burrow in shelters that prevent predation. The seasons with high temperatures and low salinities can offer better conditions for the establishment of the studied species, despite the fact that there is no hypothesis to prove it. Additionally, the higher abundance of such shrimps coincides with vegetal debris deposition, which could serve as food and provide protection for these shrimps. In this region, the vegetation matter deposited at the bottom of the bay comes from the Atlantic Forest. Overall, the preservation of the coastal forest strongly influences the abundance of this taxon, as it provides protection and food for these shrimps.