Banca de QUALIFICAÇÃO: MIZAEL CIRINEU DA SILVA

Uma banca de QUALIFICAÇÃO de MESTRADO foi cadastrada pelo programa.
DISCENTE : MIZAEL CIRINEU DA SILVA
DATA : 16/12/2019
HORA: 14:00
LOCAL: ICA Sala 5
TÍTULO:

Modelling the potential distribiution of Swietenia macrophylla King in the western Brazilian Amazon.

PALAVRAS-CHAVES:

Mahogany, Maxent algorithm, Environmental niche, Conservation.

PÁGINAS: 30
GRANDE ÁREA: Ciências Agrárias
ÁREA: Recursos Florestais e Engenharia Florestal
RESUMO:

Considering the importance of efficiency and optimization in endangered species conservation processes of extinction or degree of environmental vulnerability, the potential distribution techniques of species, can serve as the basis of decision making and defining priority conservation areas, with a view that there is great diversity in the distribution of natural resources worldwide, it is essential to define areas with higher characterization of appropriate environmental niche target species. Thus, this paper aims to achieve the prediction of areas with environmental niche statistically conducive to the establishment and perpetuation of the species Swietenia macrophylla King. (Brazilian mahogany) in the municipalities Boca do Acre, Manoel Urbano, Pauini and Sena Madureira. The forest inventory data used in this study, They come from Forest Management Area (MPA) belonging to AGROCORTEX company, located in southern Amazonas state, making border with the state of Acre on the left bank of the Purus River. The predictive variables used in modeling will be derived version Worldclim database 2 and Ambdata database, the National Institute for Space Research (INPE). The variables will be cut and resampled area for the cities of interest. Reduction and choice of variables used in modeling will be applied Principal Component Analysis, in the characterization of variables that have greater variability of the total data. The model will be processed in MaxEnt algorithm, version 3.4.1, configured to use 70% of trigger points will be used in the test stage and internal validation algorithm with processing thirty replications to generate the average model of species distribution, output Logic. It will be realized external validation of the model used 30% of the total points, calculating failure rate and hit rate. After confirming the possibility of using the model, the output image is processed using the maximum cutoff limit (threshold), and torque map prepared with pixel values ranging from 0 to 1, with the occurrence of propriety class ranging from 0 - 0.5, average from 0.51 to 0.75, and high> 0.75 pixels. From the result of principal component analysis, 14 variables were selected, which together account for 95.52% of the variability in the data. The model proposed disposed of 0.9946 AUC value with AUC of standard deviation of 0.0007, with a maximum threshold value of 0.4802, the p-values were all highly significant. The variable with the greatest gain alone is the bio2, which therefore statistically have more useful information by itself. The variable that most decreases the gain, the model, when it is omitted is the soil. From the modeling had an answer, a total of approximately 540.60 square kilometers in area with a high probability of occurrence of the species, potential for medium area the figure was 506.99 km² in the four target municipalities. The class with pixel value to 0.5 (low), had more significant value in terms of quantity of 96,876.12 square kilometers.

MEMBROS DA BANCA:
Presidente - 142.060.489-91 - JOSE NATALINO MACEDO SILVA - Oxford
Interno - 2411802 - RODRIGO GERONI MENDES NASCIMENTO
Interno - 4143218 - JOAO OLEGARIO PEREIRA DE CARVALHO
Externo à Instituição - MOISES MOURAÕ DE OLIVEIRA JUNIOR - EMBRAPA
Externo à Instituição - LEANDRO SCHLEMMER BRASIL - UFPA