Figure 5.11. Observed densities and operational sex
ratios of Harlequin Frogs (Atelopus varius) along a gallery-
forest stream on Monteverde's Pacific slope during El
Nino climate oscillations. The plotted data are from
weekly censuses during the 1982-83 event along a 200-m
section of the stream. A, Number of frogs per 200 m. The
curve is for a quadratic regression. The number of frogs
observed in late March 1987, near the peak of the 1986-
87 El Nino, was 403, or 4.4 times greater than that
predicted by the regression. B, Proportion of frogs that
were males. The curve is for a 4° polynomial regression.
The proportion in late March 1987 was 0.239, or 60%
lower than that predicted by the regression. The bizarre
snapshot of frog density and sex ratio in 1987 is from the
last census before this population crashed and disap-
peared (from Pounds and Crump 1994).
population might have been unusually large in 1987.
Second, if the warm, dry conditions caused individu-
als to leave drying crevices and gather in remaining
wet areas, observability might have been unusually
high. These hypotheses differ in their predictions con-
cerning the sex ratio. Whereas the first predicts a
normal (male-biased) operational sex ratio, the sec-ond predicts a ratio that is less male biased than nor-
mal. In accordance with the second hypothesis, the
proportion of frogs that were males (0.29) was 60%
lower than that predicted from 1982-83 patterns (Fig.
5.11B). The excess of females, a pattern that had not
been observed previously, suggested that both sexes
had gathered at the stream margins and that males,
which are normally more exposed than females, had
already begun to decline. The population appeared
to be responding to the extreme conditions of mois-
ture and temperature shortly before it crashed.
At the same site, Bransford's Litter Frogs, which
survived the 1987 crash, migrate into the stream bed
during the dry season. During the wet months of 1993
(May-December), no frogs were found in the stream
bed, compared to an average of 2 frogs per 100 m
transect in the adjacent gallery forest and overgrown
guayaba plantations (J. A. Pounds, unpubl. data). Dur-
ing the dry season, there were 47 frogs per 100 m
transect in the stream bed and none in the forest and
plantations. Future studies should examine the impacts
of climate fluctuations and trends on this and other
populations of rain frogs (Eleutherodactylus). Climate
change may be reducing the numbers of these frogs in
the Luquillo Mountains of Puerto Rico (Stewart 1995,
R. Joglar, P. Burrowes, and N. Rios, unpubl. data).
Studies of rain frogs may also reveal why some
species of anurans persisted in the Monteverde area
after the 1987 crash while others vanished. These ter-
restrial breeders were as likely to disappear as aquatic
breeders (see Sec. 5.5.1). Nevertheless, species not
dependent on bodies of water (i.e., rain frogs exclu-
sive of stream-dwelling forms) were less likely to dis-
appear than species associated with aquatic habitats
(Pounds et al. 1997). Population structure may be a
key factor in this pattern. In contrast to the islandlike
distribution of many anurans associated with bodies
of water, forest-dwelling rain frogs tend to be distrib-
uted more continuously over the landscape (Stewart
and Woolbright 1996). The comparatively large num-
ber of local populations per area may reduce vulner-
ability to stochastic mechanisms of extinction in the
wake of deterministic declines (e.g., the loss of small
populations due to random fluctuations; Gilpin and
Soule 1986) and may increase the chances that pre-
viously occupied sites will be recolonized (Travis
1994). Tests of these hypotheses should examine both
terrestrial-breeding rain frogs and aquatic-breeding
species. The latter differ in the number of local popu-
lations per area and quantitative characteristics of
dispersal. They may differ accordingly in susceptibil-
ity to regional extinction.Reptiles. In contrast to the pattern for amphibians,
only 4% of reptile species are associated with aquatic167 Amphibians and Reptiles