| Randomly investigating sampling points by setting plots mechanically in Wupao Forestry Farm,Yingchun Forestry bureau, Wandashan Mt., Heilongjiang Province, China, during the late winters of 2002 and 2003, and selecting 14 types of winter habitat factors experientially on the basis of previous studies and the status of human activities, we studied how habitat fragmentation factors impacted the distribution of red deer (Cevus elaphus xanthopygus Milne-edwards,1967) by designing 35 transects, 716 plots and 3 580 smaller plots with the application of GIS (Geographic Information System), GPS (Global Position System), Multivariate statistic analysis (Component analysis, Logistic modeling ), Bayes decision and subordination degree of fuzzy mathematics, the number of red deer tracks found (amounted to 245 in 112 plots of them) and then drew conclusions as following:1. The result demonstrated that other ungulate's disturbance, slope, altitude, logging points, nearest distance from settlements, nearest distance from farmlands, nearest distance from forest roads, nearest distance from abandoned logging roads, ecotope types were significant impact factors for habitats of red deer, while the abundance of food, snow depths, cover classes, aspect of slope, positions of slopes weren't significant.2. It was found that human activity factors were major general factors and predictable model of red deer distribution during the two late winters was: log (Pi/1-Pi) = -1.963 + 0.816PC1 -0.262PC2 -0.039PC3, 0.2 was determined to be optimal cutoff point for probability of activity, total correctly classified rate 71.1%, correctly classified rate for present plots 58.9%, correctly classified rate for absent plots 73.3%. So it was believed that model may reflect red deer habitat distribution during late winter on the whole through component analysis and binary logistic analysis.3. It was found that the area of potentially suitable habitat for red deer was 104.7km2 accounting for 66.7 percent of the total area and that of the unsuitable habitat was 50.9km2 accounting for 33.3 percent of the total area, through Bayes decision and GIS ecotope graph layer calculation, when study was emphasized on natural, physical and biological landscape factors and not human activities which demonstrated that more than half of the study area could be converted into suitable habitat if managementmeasures were taken timely.4. Regularities of several spatial factors were revealed under 14 distance classes (230~3 220m) by analyzing spatial heterogeneity of the factors of red deer habitat through adopting spatial autocorrelation analysis, which showed that most factors (89%) autocorrelation was significant within near distance classes (230~920m), but all factors showed significant similarities or dissimilarities at a different extent impacting the survive of red deer interactively; Spatial distribution characters of multiple scale and class patch on the number distribution of red deer was revealed through semivariance analysis and fractal analysis, which indicated that the dimension of the number of red deer tracks (D = 1. 926+1. 360) approaching 2 meant spatial heterogeneity within small scale.5.The results revealed that the suitable area was 14.81 km2 ( only 9.52%) including many patches isolated and fragmented seriously, marginally suitable area was 9.57km2 (only 6.15% ) that can be converted into suitable area through improvement of habitat, less suitable area was 130.05km2 (83.58%) that limited the movement of red deer because of integrated effect of many factors in this area, so it can't be converted into suitable area in short time, suitable area was 1.17km2 (0.75%) including settlements, farmlands, abandoned logging roads, forestry roads and so on. It was believed that many human activity landscapes played an important role on the deprival of ecosystem function of red deer, although they were smaller.
|
PDF Full Text Request