In the early 1970s, stimulated by the theoretical work Robert MacArthur, there was a great deal of interest in the role of interspecific interactions, especially competition, in the organization of communities of coexisting species. Non-experimental studies, many involving comparisons of multiple and geographically separated assemblages of desert rodents and ants (e.g. Rosenzweig and Winakur 1969; Brown 1973, 1975; Brown and Leiberman 1973; Rosenzweig 1973; M'Closkey 1976, 1978; Davidson 1977 a,b) produced results consistent with the hypothesis that competition between closely related, ecologically similar species played a major role in determining the composition of communities. At that time I was working on desert rodents and Dinah Davidson was my graduate student studying desert ants. Comparing and discussing our data for the two taxa, we came up with the idea that these distantly related organisms, which are the predominate consumers of seed in most North American deserts, might also compete.
Originally conducted in the Sonoran Desert, northwest of Tucson, in 1973, with the support of the International Biological Program (IBP), Davidson and I set up experiments to test the hypothesis that rodents and ants compete for the seeds of deserts plants. We set up eight 0.1-ha circular plots, giving two replicates of a simple 2 X 2 factorial design: rodents were removed by fencing and trapping, ants were removed by poisoning, both rodents and ants were removed by a combination of the preceding procedures, and there were two unmanipulated control plots.
These experiments produced three important results. First, ants increased approximately twofold on plots where rodents had been removed and rodents appeared to have increased in the absence of ants. Second, the removal of rodents led to major changes in the annual plants: principally an increase in population densities of large-seeded species. Third, the increase of large-seeded annuals plants on rodent-removed plots was accompanied by a decrease in the abundance of small-seeded species and, after a lag of at least two years, by a decrease in the ants that had initially increased in response to rodent removal.
As informative as these results were, the experiments suffered from several important problems. Plots were too small to support good numbers of rodents or ant colonies, so in the summer of 1976, Davidson and I set up a second set of experiments. In addition to repeating the study of competition between rodents and ants, we also wanted to test for interspecific competition within each taxon. We elected to do the work in the Chihuahuan Desert near Portal, Arizona.
The site was set up between July and October 1977, using a combination of graduate and undergraduate students, and local workers. A barbed-wire fence was installed to exclude livestock from the entire 20-ha site, then 24 experimental plots, each 0.25 ha in area (50 X50 m) were set up. The plots were surrounded by fencing: 1/4 inch (6.25 mm) wire mesh, buried 20 cm in the ground, extending 60 cm above ground, topped on the outside with a 15-cm strip of aluminum flashing, supported by 1/2 inch (1.25 cm) metal posts at intervals of approx. 2 meters for a total of 4.8 km of fencing.
Before starting manipulations, plots were censused for baseline, pretreatment data. In July, ants were counted and then poisoned to remove either all species or just the two largest and dominant granivores (Pogonomyrmex rugosus and Po. barbatus) from designated plots. During July, August and September, rodents were trapped with large holes in the fences so that all species had free access. In October, holes were reduced and either all species of rodents or the selected kangaroo rat species (Dipodomys spp.) were removed from designated plots.