Revisiting global biomes: abiotic controls of global and inter-continental biome distribution
Student name: Ms Dina Nethisa Rasquinha
Guide: Dr Joachim Michael Schmerbeck
Year of completion: 2013
Host Organisation: National Centre for Biological Sciences, Tata Institute of Fundamental Research, Bangalore
Supervisor (Host Organisation): Dr Mahesh Sankaran
Abstract: The study is based on using high resolution global gridded data sets and vegetation
maps to understand the abiotic influences of global biome distributions and intercontinental
differences across biomes. Analyses were conducted at a resolution of 5
arc-minutes (10km) to determine how global biomes are distributed across bioclimatic
and edaphic gradients. A classification accuracy of 92.26% and an out-of-bag error
rate of 3.74 % were obtained using random forests model. This high accuracy indicates
that distribution of Olson’s biomes can be well explained by the utilized bioclimatic and
edaphic variables. Annual precipitation and potential evapotranspiration were found
to be the most important variables that differentiate between biomes.Inter-continental
differences were also detected using discriminant analysis and random forests with
selected bioclimatic variables .Classification accuracy of random forests were found to
be much better than discriminant analysis for all biomes. Rather than mean
temperature or precipitation the range in temperature/ precipitation was found to be
the key factor discriminating between continents within biomes.
Key words: biomes, inter-continental differences, random forests, discriminant
analysis, bioclimatic
