Wagner da Costa Moreira, PhD candidate
Member of the Canadian Network for Aquatic Ecosystem Services (CNAES).
B.Sc. and M.Sc. in Statistics, with experience in:
Jan 2012 Ph.D candidate in Biology, Université du Québec à Montréal (UQÀM), Canada.
Supervisors: Dr. Pedro R. Peres-Neto, Dr. Nigel Lester.
2002-2004 M.Sc. in Statistics. Federal University of Minas Gerais (UFMG), Brazil.
Thesis Title (translated to english): Evaluation (via Monte Carlo simulation) of the influence of rare and common species in Bayesian estimators of the total number of distinct species in finite populations and quadrat sampling.
Supervisor: Dr. Sueli A. Mingoti.
1997-2002 B.Sc. (Hons. Equivalent) in Statistics. Federal University of Rio Grande do Norte (UFRN), Brazil.
Thesis Title (translated to english): Application of Principal Components Analysis in a semi-arid vegetation study.
Supervisor: Dr. Aristotelino Monteiro Ferreira.
The analysis of species – habitat relationships has always been a central goal in ecology. It has become a central framework to explore, understand and tackle specific questions about the intricacies and mechanisms underlying species distributional patterns in space and time. The savoir-faire generated by ecological modelling and its quantification of species-environment relationships is critical for conservation planning and ecosystem/population management. The general objective of this thesis is to improve quantitative methods to: (i) build a framework for multispecies data based on generalized linear models; (ii) species distribution modelling and linking these two approaches to the (iii) quantification of fish productivity into an ecosystem service perspective. These links can contribute to major scientific underpinnings related to the research of species-habitat relationships, while consisting of ecosystem services when promoting information about the processes underlying these relationships. The investigation of interconnected quantitative frameworks to link environmental, spatial and biotic interactions should bring to light a greater understanding of the key agents structuring biodiversity and how they interact to provide the delivery of aquatic ecosystem services, while clarifying about the actions that should be taken to mitigate the loss of these services in face of increasing human impacts. The first chapter develops a framework and evaluate its performance via simulations for modelling multispecies data based on generalized linear models (logistic and Poisson). To this date, the most used tool for multi-species modelling is based on linear regression, which is known to have undesirable properties for presence-absence and abundance data. Despite its popularity, logistic regression has extremely bad behaviour in terms of parameter estimation properties under a high number of covariates and when covariates are missing. To remediate this issue, I propose a modified Poisson model that can model presence-absence data and that provides reliable estimates in these two circumstances. Moreover, I evaluated the performance of a robust estimate for the coefficient of determination (R2) for GLMs, which allows for the first time an appropriate variation partitioning scheme, a widely used tool in ecology. The second chapter presents a comparison among a number of SDM approaches in terms of their predictive performance and explanatory power. These models will be contrasted by using empirical data and developed for six species of economically important freshwater fishes in north-temperate lakes of Ontario. The modelling routine for Chapter 2 will be done taking into account two classes of models (i) incorporating information about the occurrence of other fish species as predictors (i.e. using abiotic + biotic parameters); (ii) using solely the information about the fish community to predict the occurrence of a particular species (i.e. using biotic parameters). The third chapter looks at the feasibility of using information about species ranges predicted via SDMs to identify areas of optimal productivity across the landscape. The knowledge obtained from Chapter 2 will be used to determine the range of brook trout, lake trout, northern pike, smallmouth bass, largemouth bass and walleye, generating a map of distribution ranges that will be crossed with information about productivity for these species across the province of Ontario. These productivity estimates will be computed using an adaptation of the TOHA (thermal-optical habitat area) model, and this base model will serve as a starting point for the development of species-specific productivity models related to parameters that reflect the habitat preferences of each species studied. These new productivity models will bring new insights about the importance of optical and thermal parameters in community biomass composition.
Marinho-Soriano, E., Fonseca, P.C., Carneiro, M.A.A. & Moreira, W.S.C. 2006. Seasonal variation in the chemical composition of two tropical seaweeds. Bioresource Technology, 97 (18): 2402-2406.
Marinho-Soriano, E., Moreira, W.S.C. & Carneiro, M. A. A. 2006. Some aspects of the growth of Gracilaria birdiae (Gracilariales, Rhodophyta) in an estuary in northeast Brazil. Aquaculture International, 14 (4): 327-336.
Marinho-Soriano, E., Morales, C. & Moreira, W.S.C. 2002. Cultivation of Gracilaria (Rhodophyta) in shrimp pond effluents in Brazil. Aquaculture Research, 33: 1081-1086.
Marinho-Soriano, E., Silva, T.S.F. & Moreira, W.S.C. 2001. Seasonal variation in the biomass and agar yield from Gracilaria cervicornis and Hydropuntia cornea from Brazil. Bioresource Technology, 77: 115 – 120.