Alleles In Space (AIS): Computer Software for the Joint Analysis of Interindividual Spatial and Genetic Information

Genetic analyses of natural populations have historically relied on statistical procedures based on the concept that distinct “populations” of a species exist across a landscape. Invariably, commonly used analyses reduce to approaches that treat collections of individuals (“populations”) as independent/causative variables and allele frequencies as dependent/response variables. Examples of these procedures include Wright's F_ST and its variants (Excoffier et al. 1992; Nei 1973; Slatkin 1995; Weir and Cockerham 1984), contingency table procedures (Raymond and Rousset 1995; Roff and Bentzen 1989), and measures of genetic distances among populations (e.g., Nei 1972, 1978; Reynolds et al. 1983). These analyses qualitatively or explicitly test null hypotheses of homogeneity of allele frequencies between or among populations.

Although almost universally applied, the analyses mentioned above are not necessarily appropriate in many situations. For example, highly mobile organisms such as large mammals or birds can occupy continuous habitats over large spatial scales. Plants may also occupy large continuous habitats, as can species inhabiting marine or aquatic systems. In these cases, objectively designating groups of individuals at population levels for use in genetic analyses may prove difficult, if not impossible. Clearly, an important consideration in these situations is the spatial extent of the “populations” that are chosen for analyses. If groups of organisms are defined over larger than appropriate spatial scales, resulting measures of genetic differentiation may actually provide ambiguous or misleading results (Miller et al. 2002).

To address many of these issues, I have developed a new software package entitled “Alleles In Space” (AIS). This program, rather than implementing methodology that relies on arbitrary groupings of individuals, instead has the ability to perform joint analyses of interindividual spatial and genetic information that can be applied at virtually any spatial scale. These approaches specifically lend themselves to analyses of genetic data when one or a few individuals are sampled from large numbers of collection sites. Moreover, the program is designed to handle a wide variety of genetic data types, including codominant marker systems, dominant marker systems, and DNA sequences. Thus AIS will likely be useful for elucidation of patterns in diverse study types ranging from local analyses of genetic structure, phylogeographical studies, and studies encompassing aspects of the emerging field of landscape genetics (Manel et al. 2003).

Program Description

Alleles In Space has a simple graphical interface that runs under any 32-bit Windows operating system (95/98/ME/NT/XP). A Pentium III processor with at least 64MB RAM is recommended. An approximately 4MB self-extracting installation file containing the executable program file, sample datasets, and documentation (in portable document format [PDF] format and a Windows help file) can be downloaded free of charge from http://www.marksgeneticsoftware.net. Two separate input files are used to perform analyses. One data file contains sets of spatial coordinates for each observation in the dataset, while the second contains genetic data for each individual analyzed. Once input files have been selected, users may specify any number of different options for the analyses they wish to perform. Following the analyses, new windows are displayed that contain text-based representations of analysis results and graphical depictions of the analyses (when appropriate). All text and graphics created by the program can be copied to the Windows clipboard and inserted in other electronic documents.

Analyses Implemented in AIS

Alleles In Space performs a number of different analyses that can be used to detect or characterize patterns of spatial genetic structure. For example, it can perform simple Mantel tests (Mantel 1967) to evaluate correlations between genetic and geographical distances of sampled individuals. Likewise, AIS can perform a generalized form of spatial autocorrelation analysis (Cliff and Ord 1973; Sokal and Oden 1978a,b) that permits detection of genetic structure and allows for inferences to be made about spatial scales over which the genetic structure occurs (Barbujani 2000; Clark and Richardson 2002; Manel et al. 2003).

Alleles In Space also implements a novel procedure based on the statistical concept of aggregation. Aggregation indices are commonly used in ecological studies to characterize spatial distributions of individuals across landscapes (Clark and Evans 1954; Hopkins and Skellam 1954; Pielou 1977) and have been widely used as measures of forest stand structure (Pommerening 2002), specifically with respect to describing the presence of either random, clumped, or uniform spatial distributions of individuals. AIS uses a modification of the aggregation index of Clark and Evans (1954) to perform an allelic aggregation index analysis (AAIA) that provides a basis for testing the null hypothesis that each allele at a locus is distributed at random across a landscape (i.e., no aggregation or genetic structure) relative to the aggregation of the actual organisms sampled for analysis purposes.

The analyses described above (Mantel tests, spatial autocorrelation analyses, and AAIA) provide a basis for determining if, on average, nonrandom patterns of genetic diversity exist over a landscape. However, over large spatial scales, considerable variation may exist in patterns of genetic structure due to vicariance or barriers to gene flow (Manel et al. 2003). Thus AIS includes two different procedures that may hold utility for researchers conducting phylogeographical analyses or other landscape-scale explorations of patterns of genetic diversity and structure. First, the program contains routines that implement Monmonier's algorithm (Monmonier 1973). This geographical regionalization procedure is increasingly being used to detect the locations of putative barriers to gene flow by iteratively identifying sets of contiguous, large genetic distances along connectivity networks (Doupanloup et al. 2002; Manel et al. 2003; Manni et al. 2004). In AIS, a Delaunay triangulation (Brouns et al. 2003; Watson 1992) is used to generate the connectivity network among collection sites. After analyses, a graphical representation of putative “barriers” inferred by the algorithm is superimposed over the connectivity network to assist with rapid identification of important geographical features reflected by the genetic dataset. A text-based representation of the search procedure is provided that contains quantitative information about detected barriers from each analysis.

This program was written primarily to assist with the analysis and interpretation of data from projects funded by the U.S. Bureau of Reclamation (Cooperative Agreement no. 1425-02-FC-10-8730) and U.S. Geological Survey (contract no. 03WRSA0535). I am grateful for their support, as well as the support and interest of many additional individuals who provided me with feedback on this program and its documentation.

References