POTENTIAL DISTRIBUTION OF 20 PINE SPECIES IN MEXICO

Abstract

Conifer species largely depend on climatic variables, but their distribution is modified by climate change. The objectives in this study were to model the potential distribution of 20 Pinus species in Mexico, determine which climatic variables influence their distribution and establish their bioclimatic profile. Our hypothesis was that bioclimatic variables predict the potential distribution of coniferous species. A total of 10 222 presence records paired with 19 bioclimatic and three topographic variables were analyzed with the MaxEnt algorithm to generate potential distribution model of each species. Model validation was determined assessing their response curves (omission /commission analysis) and the sensitivity of their Receiver Operated Curve (ROC), area under the curve (AUC), and Jackknife tests to measure each variable effect. The ROC test showed that species with few records either overestimate or underestimate the prediction. The bioclimatic and topographic variables with the greatest contribution in the models were altitude, average temperature of the warmest quarter (Bio10) and annual average temperature (Bio1), both in °C; the variables related to low temperature, low precipitation and terrain aspect did not explain the species distribution. The species with the greatest potential surface area, with a 0.70 probability were: P. montezumae, P. devoniana and P. pseudostrobus with 14 744.8, 14 436.1 and 11 594.8 potential km2 . The generated prediction models are reliable, since their AUC values were higher than 0.90. The model’s adjustment dependent on the number of records of the bioclimatic variables that constitute it.