Ecological niche models and the geography of biological invasions: a review and a novel application - Invasive Plants: Ecological and Agricultural Aspects - page 34

(GARP) [25-27]. GARP is a genetic algorithm that uses multiple unique niche
modeling methods (e.g., logistic regression, bioclimatic rules) and environ-
mental datasets to model various factors that govern distribution potential
throughout the range of a species. Genetic algorithms are a solution-optimiza-
tion technique, loosely akin to evolution by natural selection, wherein a set of
possible solutions to a problem is formed, and, through a series of iterations,
the solutions are modified and tested until the best solution is found. GARP
uses such a process to compose, evaluate and produce a set of rules that
approximates the ecological niche of a species. GARP models can be import-
ed into geographic information systems (GIS) and visualized as maps that rep-
resent hypotheses for the potential distribution of the organism(s) under study.
When projected onto another region, the potential distribution of the species in
that location can be estimated. GARP has been used to model the distributions
of both native and invasive species, and its predictions are generally more
robust than other niche modeling techniques [13, 28, 29]. See [25-27] for a
more detailed explanation of GARP and its application.
We confined our analysis to South America and the continental 48 States of
the US. We used 12 WorldClim [30] environmental data layers as modeled
niche dimensions, including an elevation layer and 11 bioclimatic datasets that
summarize temperature and precipitation aspects of climate (Tab. 1). These
layers are typical of those commonly used to produce niche models with
GARP.
Occurrence data, in the form of state, county and year of infestation for
invasive populations of RIFA in the US were provided by the National
Agricultural Pest Information System [21]. We obtained 771 distributional
points, each defined as the latitude-longitude center-point of counties (deter-
mined by ArcGIS 8.3) within which RIFA is established. 71 collection loca-
Table 1. WorldClim v1.2 [30] environmental data layers used to develop ecological niche models for
the red imported fire ant (RIFA, Solenopsis invicta Buren). All layers had a pixel resolution of
300 km 2
Layer
Elevation
Mean annual temperature
Mean diurnal temperature range (mean of monthly (max temp-min temp))
Isothermality (mean diurnal range/temperature annual range)
Temperature seasonality (σ*100)
Maximum temperature of warmest month
Minimum temperature of coldest month
Temperature annual range (max temp of warmest month-min temp of coldest month)
Annual precipitation
Precipitation of wettest month
Precipitation of driest month
Precipitation seasonality (coefficient of variation)