Density, Dispersion, and Population Structure in Drosophila Pseudoobscura

Wild D. pseudoobscura flies were captured, marked, and released in relatively low numbers at nine separate, centrally located sites in each of two Colorado, U.S.A., locations in midsummer 1970. The flies were marked by spraying with micronized dusts which fluoresce in characteristic colors under ultraviolet light. The dusts are harmless, they mark the flies well, and they are not transferred from one fly to another. The trapping design extending out from the central release points contained additional traps on eight evenly spaced radii. Data on the number of captured, unmarked flies and recaptured, marked flies in the central and outlying traps provided maximum likelihood estimates of adult density and dispersion. These data were adjusted for the presence of a sibling species, D. lowei, by means of an accurate morphological technique which permits classification of the two kinds of males. Preliminary analyses indicated that dispersal estimates made during an evening activity period 1 day after release of marked flies pertain mostly to dispersion during the intervening morning activity period, when no traps were exposed. It was necessary to base estimates of both density and dispersion on trapping days following days when only a few or no traps had been exposed in the habitat. The minimum attractive radius of a trap was approximately 46 m. The estimates of density at both locations were similar, averaging 0.38 flies/100 m². An independent and much simpler estimate of density, based on the number of flies captured at the center trap, gave inconsistent results. This was attributed to accidental interference with normal fly activity. The mean distance (d) of marked flies from the release points after 1 day (presumably after one morning activity period) was 176 and 202 m at the two locations. The mean—squared distance (d²°) was 97 and 146, respectively, in units of 400 m². The estimates for the first location are more reliable because those data were more homogeneous and extensive. The patterns of the observed dispersion agreed well with Brownian motion expectations on the basis of comparing first— and second—order moments of distance dispersed. This permitted the construction of graphs which depict the spread of released flies with an increasing number of activity periods. An activity parameter °, the standard deviation of dispersed flies along one direction in a two—dimensional environment, was estimated from @ under the assumption of Brownian motion (°b . ₘ.) and without this assumption (°D . F .). These estimates refer only to the observed dispersion. At the first location the values were 141 and 139 m, respectively; at the second they were 162 and 171. Closer inspection of the observed dispersal patterns, accompanied by further analyses of the data, revealed a departure from Brownian motion and probably serious underestimates of the true dispersal rates. A certain proportion of flies dispersed rapidly to points outside the experimental area during the morning activity period following an evening release. Using an estimate of 95.3% daily survival of marked flies, 2—day dispersal data suggested that more than half the flies had moved beyond the experimental area in two morning activity periods. The earlier experiments on density and dispersion of D. pseudoobscura, conducted by Dobzhansky and Wright in California, were re—analyzed by the present methods. Midsummer density of wild flies in Colorado was one—seventeenth that on Mt. San Jacinto in southern California, and three—fifths that at Mather in the Sierra Nevada of central California. Comparison of °d . f. estimates indicated that the wild Colorado flies dispersed at a rate approximately 50% greater than the laboratory flies released in California, even though the latter probably had more opportunity for dispersion. This difference may result partly from an adaptive strategy for greater dispersion of flies in low density habitats, and partly from a fundamentally different behavior of wild and laboratory flies. The large values of °d . f . (which are underestimates due to dispersion beyond the experimental areas) in both Colorado and California suggest that natural populations of D. pseudoobscura are not broken up into a number of very small breeding units within which allelic variation could be stored by genetic drift. Wright's panmictic circle concept yielded estimates of effective population size (Nₑ) between 1,000 and 10,000 in both Colorado and California. According to existing theory, these values are not large enough to maintain the allelic diversity known to exist for certain enzyme loci. The need for more sophisticated experiments to determine which components of dispersion are most closely associated with distribution of emergence sites of parents and offspring is discussed.