Numbers of plant, insect and land bird species on nineteen remote islands in the Southern Hemisphere

Abstract

The familiar species-area equation S=cAz may not account for as much of the variation in the numbers of species on islands as it has been held to. The undue emphasis on area as a factor probably results from early workers not considering as many factors as possible in their regression analyses. In the case of 19 remote islands in the Southern Hemisphere, island area and elevation, latitude, mean temperature of coldest month and distance of island from mainland, nearest land and nearest land to west were used to consider variation in number of plant species with the technique of multiple regression. A nonlinear model explained 83% of the variation, with temperature and area accounting for 71% of the variation in number of plant species. The above factors with number of plant species were then used to consider variation in number of insect species on the islands. The linear model was superior, accounting for 91% of the variation. Number of plant species accounted for 73% of the variation in number of insect species. Variation in numbers of passerine bird species and total land bird avifauna was considered using all of the above nine factors. The linear model was superior in both cases. Number of plant species accounts for 60% and 71% of the variation respectively. Various suggestions are made to account for the unexpected result of number of insect species making little contribution to the last two regressions.