Submitted by Jorge Silva (jorgelmsilva@ufrrj.br) on 2016-10-20T18:25:56Z No. of bitstreams: 1 2016 - Cristina Fernandes do Amarante.pdf: 776756 bytes, checksum: 758e9e88a96110176bd0dff013f8c6ac (MD5) / Made available in DSpace on 2016-10-20T18:25:56Z (GMT). No. of bitstreams: 1 2016 - Cristina Fernandes do Amarante.pdf: 776756 bytes, checksum: 758e9e88a96110176bd0dff013f8c6ac (MD5) Previous issue date: 2016-03-03 / CNPq / This study aimed to assess the factors involved in the determination of ecological events, including parasite aggregation, abundance, and dominance, via multivariate analysis using fish as a biological model. For this purpose, we used a database consisting of 3,746 fish specimens collected on the coast of Rio de Janeiro and from the Guandu river and evaluated their characteristics, habits, and associated parasites. Parasites were grouped into seven higher taxa: Nematoda, Monogenea, Cestoda, Trematoda, Acanthocephala, Hirudinea, and Crustacea. Aggregation was calculated using the dispersion index; abundance was determined by calculating the total number of parasites; and dominance was calculated using the Berger-Parker index. Data normality was tested using the Shapiro-Wilk test; group comparisons, when necessary, were performed using the Wilcoxon test or the Kruskal-Wallis test. Adjustments were made in the linear regression models for parasite aggregation in each parasite taxon evaluated and in the mixed-effect models to evaluate abundance and dominance using Poisson and Gaussian distributions, respectively. The factors inherent to the hosts, including sex, size, habitats, formation of schools, eating habits, and aquatic environments, were tested as potential explanatory variables of these ecological events. In addition, the parasite development stage and parasite taxa were included as explanatory variables in the analysis of the determinants of aggregation and dominance, respectively. In the mixed-effects model, the ecological events were considered an outcome variable, and fish species were considered a random variable. Regression coefficients were calculated in the study of aggregation, and parasite dominance and prevalence rate were estimated in the study of abundance. Significance was determined using confidence intervals (CI) and the Wald test. All calculations were performed using R software with a confidence interval of 95%. Statistically significant associations between levels of parasite aggregation and explanatory variables (factors inherent to the host and the percentage of larvae) varied according to the parasite taxon. The coefficient of determination was lower than 60% in all of the adjusted models, indicating that factors other than those analyzed may be associated with the level of parasite aggregation. Parasite abundance was significantly associated with sex, formation of schools, habitats, and host length, which were considered risk factors. The numerical dominance of parasites was significantly associated with host length and the taxa Nematoda, Trematoda, Monogenea, Hirudinea, and Crustacea. In addition, our results indicate that the mixed-effects model was more parsimonious compared with the classical model, underscoring the importance of choosing a statistical model that takes into consideration the nature of the data to avoid spurious results, especially when autocorrelations data were not considered. In general, our results point to the need for developing studies with more comprehensive databases and larger samples of parasite species, in which data on the life cycles of fish species and parasite taxa can be analyzed more thoroughly. / Este trabalho teve como objetivo avaliar fatores envolvidos na determina??o dos eventos ecol?gicos: agrega??o, a abund?ncia e a domin?ncia parasit?ria por meio de an?lises multivariadas, utilizando peixes como modelo biol?gico. Para tanto, foi utilizado um banco de dados composto por 3.746 esp?cimes de peixes, suas caracter?sticas e h?bitos e seus parasitos, provenientes do litoral do Rio de Janeiro e do Rio Guandu. Os parasitos foram agrupados em sete t?xons superiores: Nematoda, Monogenea, Cestoda, Trematoda, Acanthocephala, Hirud?nea e Crust?cea. A agrega??o foi calculada atrav?s do ?ndice de dispers?o; a abund?ncia pela quantidade total de parasitas computados; e a domin?ncia atrav?s do ?ndice de Berger Parker. A normalidade dos dados foi testada pelo teste de Shapiro Wilk e a compara??o de grupos, quando necess?ria, foi realizada pelo teste de Wilcoxon ou Kruskal-Wallis. Foram ajustados modelos de regress?o linear para a agrega??o parasit?ria para cada t?xon parasit?rio estudado e modelos de efeitos mistos para an?lise dos determinantes da abund?ncia e da domin?ncia, com distribui??o de Poisson e gaussiana, respectivamente. Os fatores inerentes aos hospedeiros tais como: sexo, tamanho, habitat, forma??o de cardumes, h?bitos alimentares e ambiente aqu?tico foram testados como poss?veis vari?veis explicativas desses eventos ecol?gicos. Al?m desses, na an?lise dos determinantes da agrega??o e domin?ncia foram inclu?dos os fatores est?dio de desenvolvimento do parasita e os t?xons parasit?rios como vari?veis explicativas, respectivamente. Os eventos ecol?gicos foram considerados vari?veis desfecho e nos modelos mistos a esp?cie de peixe como vari?vel aleat?ria. Foram calculados os coeficientes de regress?o nos estudos de agrega??o e domin?ncia parasit?ria e estimada a raz?o de preval?ncia no estudo de abund?ncia. A signific?ncia foi verificada por meio o intervalo de confian?a (IC) e do teste de Wald. Todos os c?lculos foram realizados utilizando o software R, com um IC de 95%. As associa??es estatisticamente significativas entre o n?vel de agrega??o parasit?ria e as vari?veis explicativas (fatores inerentes ao hospedeiro e propor??o de larvas) variaram conforme o t?xon parasit?rio. O coeficiente de determina??o foi inferior a 60%, em todos os modelos ajustados, indicando que outros, fatores al?m dos analisados, devam estar relacionados ao grau de agrega??o parasit?ria. A abund?ncia parasit?ria mostrou-se associada significativamente ?s vari?veis: sexo, forma??o de cardumes, habitat e comprimento do hospedeiro, que se comportaram como poss?veis fatores de risco. A domin?ncia num?rica de parasitos esteve associada significativamente ao comprimento do hospedeiro e aos t?xons Nematoda, Trematoda, Monogenea, Hirud?nea e Crust?cea. Os resultados mostraram que os modelos mistos foram mais parcimoniosos em rela??o aos modelos cl?ssicos, refor?ando a import?ncia da escolha de um modelo estat?stico mais adequado ? natureza dos dados, evitando-se resultados esp?rios, principalmente quando n?o se leva em conta a autocorrela??o dos dados..De um modo geral, os resultados deste estudo sinalizam para a necessidade de pesquisas com bancos de dados mais abrangentes e com amostras maiores de esp?cies de parasitos, nos quais as caracter?sticas do ciclo biol?gico das esp?cies de peixes e dos t?xons parasit?rios possam ser mais detalhados nas an?lises.