Dr. En Ciencias

Modelling rodent pest distributions in Mexico
Víctor Sánchez-Cordero* and Saralina García Zepeda
Departamento de Zoología, Instituto de Biología, Universidad nacional Autónoma de México, Apartado Postal 70-153,
México, D.F. 04510, MEXICO
*Corresponding author, email: victors@ibiologia.unam.mx
Abstract. Ecological niche models based on museum specimen data were generated to providepredictions of geographical
distributions of native rodents known to be agricultural pests in Mexico. By overlapping rodent pest distributions,
we found significant correlation between predicted presence of rodent species and crop damage in four main
crops widely distributed nationwide. This study provides a first step for generating risk maps for crop damage due to
rodent pests in agricultural regions.Museum collections may thus provide essential information for expanding these
analyses to other pests.

Introduction
Mexico is an extremely varied country with a rich crop
diversity. Historically, Mexico has mainly focused on agriculture,
which presently occupies almost half of its area,
producing corn, beans, sorghum, and sugarcane, among
other crops. Mexico has a diverse rodent fauna,which has
been reported as pests in crops (de Ita 1992). Most of
these rodent species share the life-history traits of frequent
litters, short gestation periods, post-partum oestrous, and
aseasonal reproduction (Nowack 1991). In a previous
study, we used novel techniques for modelling ecological
niches of 17 rodent pests to determine potential species
distributions, and showed that cropdamage was related
significantly to the predicted presence of rodent pests in the
Mexican state of Veracruz (Sánchez-Cordero and Martínez-
Meyer 2000). Here, we expand these analyses at a national
level in four main and widely distributed crops. This
approach can be used as a first approximation for generating
risk maps in crop damage due to rodent pests, as well
as other pests, in agriculturalregions in the country.
Material and methods
Distribution data for each species were obtained from the
mammal collections of the University of Kansas Natural
History Museum (KU), The Field Museum of Natural
History (FMNH) and the Colección Nacional de Mamíferos,
Instituto de Biología, Universidad Nacional Autónoma
de México (CNMA). Species names and taxonomic
arrangements followedaccepted authorities. Locality data
were geo-referenced to the nearest 10–3 degree by direct
consultation of maps, and reduced to unique latitude–longitude
combinations. The four thematic geographical coverages
used (annual mean temperature, annual mean
precipitation, elevation, potential vegetation) consisted of
raster grids (5 × 5 km pixels), obtained from Conabio
(2002).
Rodent pestdistributions
Ecological niches and potential geographical distributions
were modelled using the genetic algorithm for ruleset
prediction (GARP) (Stockwell and Peters 1999; see
BIODI 2002). Specifically, GARP relates ecological characteristics
of known occurrence points to those of points
randomly sampled from the rest of the study region,
seeking to develop a series of decision rules that bestsummarise those factors that are associated with the
species’ presence (Stockwell and Peters 1999). GARP
includes several distinct algorithms for niche modelling in
an artificial-intelligence-based approach. Occurrence
points are divided evenly into training and test data sets. It
works in an iterative process of rule selection, evaluation,
testing, and incorporation or rejection, choosing amethod
from a set of possibilities (e.g. logistic regression, bioclimatic
rules) applied to the training data. Then a rule is
developed or evolved. Predictive accuracy is then evaluated
based on 1250 points resampled from the test data
and 1250 points sampled randomly from the study region
as a whole. The change in predictive accuracy from one
iteration to the next is used to evaluate...