Dendrocephalus brasiliensis (Crustacea: Anostraca) hatching egg capacity in different water treatments

Abstract

When considering the threats against biodiversity, biological invasion is an important element, as alien species alter the composition and functioning of the ecosystems, so much so that the invasion from exotic species is considered the top cause of global biodiversity loss. Considering how invasive species pose significant threats against local biodiversity, it’s necessary to develop technologies that allow the use, including for ecotoxicological tests, of native species. Dendrocephalus brasiliensis (Crustacea: Anostraca) is a species that presents economical potential, with high nutritional value, so important in aquaculture, as well as high sensibility to several toxic substances, which allows its use as a scientific tool in toxicity studies. Therefore, it’s important to study this species and develop new methodologies that could enable its use to the detriment of exotic species. Therefore, this thesis aimed to provide a new approach on how to perform hatching studies for Dendrocephalus brasiliensis species, evaluate the effects of using different mediums such as Dimethylsulfoxide (DMSO), Glycerol, Reconstituted (RW), and Natural water (NW) and their effects on the hatching rate; the effects of controlling/not the mediums pH, and the effect of saline buffers use over the cyst hatch. The results indicate the use of Natural and/or Reconstituted Water as a preferential medium for Dendrocephalus brasiliensis cultures, buffered (as a tool to provide hatching homogeneity) in a range of 7.3 to 8 (being 8 the recommended in the literature for the species), using cysts without previous dormancy break attempt (pre-treatment). Also, to automatize and facilitate cyst processing for Dendrocephalus brasiliensis using computer vision as a tool. To evaluate the viability of automating cyst recognition and counting using domain-specific object detection techniques based on computer vision. Then, we trained two state-of-the-art object detection methods, YOLOv3 (You Only Look Once) and Faster R-CNN (Region-based Convolutional Neural Networks), on the DBrasiliensis data set, which was also created for this study, to compare them under both cyst detection and counting tasks. We concluded that the proposed approach using YOLOv3 is adequate to detect and count Dendrocephalus brasiliensis cysts. The stated results and considerations provided by this study allowed us to provide important considerations that can be applied to improve Dendrocephalus brasiliensis studies and production.