CLIMATE DRIVEN POPULATION FLUCTUATIONS IN RAIN-FOREST FROGS

A deme of Eleutherodactylus coqui was followed from 1979 to 1993 at El Verde, Puerto Rico, to determine seasonal and annual variation in numbers and activity patterns. All visible frogs and predatory spiders in a 50 x 2 m transect in the forest were counted for three consecutive nights semi-monthly for two years, then annually or biannually thereafter for a total of 255 evening counts. Ten all-night counts were made at five different times of the year to determine time of maximal activity during the night. Population size varied seasonally, with numbers increasing from June until December followed by a gradual decline until May. The number of adults varied from 1 to 29/100 m(2), whereas the number of juveniles varied from 0 to 221/100 m(2). The maximum single count of all frogs was 244. Counts of >100 juveniles occurred during October through January in the years 1979 to 1982, and in 1989. A marked drop in the numbers of frogs occurred in 1984; from 1979 to 1983, 3-50% of the counts yielded greater than or equal to 15 adults whereas the maximum count from 1984 until 1989 was 11 adults. The drop in numbers was correlated with an increased number of periods of days with less than or equal to 3 mm of rain. Over the period 1979 to 1989, the number of frogs observed was negatively correlated with the longest dry period during the previous year. Population size began to decrease in 1983 and never regained prior levels although numbers were increasing early in 1989 before Hurricane Hugo. Juveniles apparently cannot survive extensive drought, and extended dry periods may be lethal to adults who are inhibited from feeding because of potential desiccation. Predatory ctenid spider populations crashed two years following the decline of frog populations, then disappeared following the hurricane as did other arthropod predators. Rather than total monthly or annual rainfall, it is the distribution of the rain that is important to these subtropical wet forest species.