In 1984 breeding Sand Martins declined to 27% of their 1983 numbers at a large colony in Scotland. A population crash of similar magnitude occurred over much of central Scotland and over Britain as a whole.T h e decline in numbers was probably attributable to severe drought conditions experienced overwinter in the Sahel zone of Africa. There was a significant reduction in the mean keel-length of Sand Martins captured at the main study colony in 1984 compared with individuals measured in both 1982 and 1983, suggesting that selection for small body-size had occurred during the period of high overwinter mortality.Few studies have reported natural selection acting on body-size in bird populations during periods of high mortality (Bumpus 1899, Boag & Grant 1981, Johnston & Fleischer 1981. Exceptional climatic conditions (Johnston & Fleischer 1981)) such as droughts (Boag & Grant 1981, may increase avian mortality rates substantially: one study of body-size changes in a bird population after drought conditions revealed selection for large individuals (Boag & Grant 1981. This paper reports a dramatic population crash of Sand Martins Riparia riparia breeding in central Scotland, the decrease in numbers probably being caused by drought conditions encountered overwinter. A reduction in mean body-size occurred over the period of high overwinter mortality, and plausible selective advantage of small body size during a drought is proposed.
MethodsDuring an investigation into the evolutionary significance of body-size in breeding Sand Martins in central Scotland between 1982 and 1984, about 900 individuals were ringed at Barbush sand quarry near Dunblane (Grid Reference N N 787026). Measurements of wing-length (maximum chord, measured to the nearest 0.5 mm), keel-length (measured to the nearest 0.1 mm), and body-mass (to the nearest 0.1 g) were made on 423 adults over the three years. Keel-lengths were measured three times upon each capture of an individual by the author alone, and a mean value recorded. During the study period, Sand Martin breeding colonies in the Stirling area were surveyed, counts of occupied burrows being made in late June and early July, the time when most pairs were rearing first broods. Details of the study area and counting methods are outlined in Jones (1986). Eight sites counted in all three years of the study, which were relatively undisturbed by quarrying activities or by river flooding, were used to estimate population changes in the Stirling area between 1982 and 1984.
ResultsOf the two potential measures of body-size used, keel-length was judged to be a better indicator of size than wing-length. Keel-length reached its maximum size at or soon after fledging: there was no significant difference between mean keel-lengths in a sample of fledgling and adult carcasses ( t 3 4 = 0.65, P > 0.5). Wing-length continued to increase after fledging, however, and over adult life (of 27 birds controlled in two *Present address: Department of Zoology, University of Bristol, Woodland Road, Bristol BS8 1 UG.