provisioned with yolk (see Sec. 5.5.3). The tadpoles
hatch with enough stored energy to develop and trans-
form without eating, but they grow larger before trans-
forming if food is available (Crump 1989b).
5.5.5. Reproductive Phenology
Relationships between reproductive cycles and climate
are an important area for future study. Changes in the
seasonal breeding cycles of British amphibians are
correlated with long-term warming trends (Beebee
1995). In the tropics, many amphibians and reptiles
reproduce seasonally (e.g., Fitch 1973, Duellman and
Trueb 1986, Guyer 1986, Siegel and Ford 1987), but
patterns vary according to seasonality of the environ-
ment. The Cloud Forest Anole, for example, inhabits
relatively aseasonal areas and normally produces eggs
year-round with no apparent reduction in rate (Fitch
1973, J. A. Pounds, pers. obs.). The Gray Lichen Anole
inhabits strongly seasonal environments on the Pacific
slope and produces few or no eggs during the dry sea-
son. The climate trends at Monteverde imply an in-
crease in seasonality manifested as increasingly severe
dry seasons (see Sec. 5.4.3). Because the Cloud Forest
Anole has disappeared from the more seasonal, west-
ern parts of its distribution (see Sees. 5.4.1, 5.6.1), it
would be valuable to examine whether the normally
aseasonal reproduction of this species has become
more seasonal in areas where populations remain. The
climate trends may also force pool-breeding frogs and
toads to breed later in the year than normal. Increased
soil-moisture deficits or an unusual lowering of the
water table might delay the formation of temporary
pools when the rainy season begins.
5.6. Habitat UseHabitat complexity is hierarchical. A natural environ-
ment is "like a checkerboard of habitats" in which each
square has "its own checkerboard ... of component
subhabitats" (MacArthur 1972). This pattern of mosa-
ics-within-mosaics repeated at different spatial scales
is especially complex in a tropical forest. Major patches
can be defined in relation to bodies of water (well-
drained areas vs. streams, pools, and swamps of vari-
ous sizes) and components of the disturbance mosaic
(mature-phase forest and canopy gaps of different sizes
and ages). These patches may be subdivided into strata,
each containing a mosaic of microhabitats.
5.6.1. Aquatic Habitats
Amphibians. About 60% of the amphibian species at
Monteverde, all of them frogs and toads, are ordinarilyassociated with streams, pools, or swamps during at
least part of the year. Although the pattern largely
reflects aquatic breeding habits, three terrestrial-
breeding species (the Tilaran Rain Frog, Salmon-
bellied Rain Frog, and Middle American Stream Frog)
frequent stream margins, possibly for reasons of water
economy or prey availability. Harlequin Frogs that
inhabit the moist cloud forest (Fig. 5.3) are often ac-
tive far from the watercourses in which they breed,
whereas frogs of the same species that inhabit the
drier, more seasonal areas of the Pacific slope are typi-
cally found within a few meters of stream margins
(Crump and Pounds 1989). The difference seems to
reflect the constraints of moisture availability, al-
though food availability might also be a factor. Along
a gallery-forest stream in the Rio Lagarto drainage
(1140 m), the abundance of arthropod prey and the
predictability of interpatch differences in prey abun-
dance were greater in the dry season than in the wet
season (Crump 1988). No data exist to compare prey
availability at the stream margins to that in the adja-
cent forest.
Observations at the same site documented the re-
sponse of Harlequin Frogs to the El Nino warm epi-
sodes of 1982-83 and 1986-87 (Pounds and Crump
1994). The thermal signal of the former event was
stronger than any in the preceding century (Graham
and White 1988). During the 1983 dry season, as
stream flow diminished and moisture availability
became increasingly patchy, Harlequin Frogs shifted
their home ranges into the remaining wet areas, es-
pecially near waterfalls. The pattern of dispersion
became highly clumped, and the observed density of
frogs decreased as many sought refuge in damp crev-
ices. Because frogs hidden in crevices included a dis-
proportionately large number of females, the opera-
tional sex ratio (observed in exposed areas) was
strongly male biased. Despite the warm, dry weather,
many Harlequin Frogs survived; they were present in
large numbers in subsequent years. The population
fared worse during the 1986-87 El Nino, which had
a greater impact on local weather than the 1982-83
event (Pounds and Crump 1994; see Sec. 5.4.3). In
March 1987, when the warm, dry conditions were
near their peak, the observed density (1 frog per 0.5
m of stream) was a record high, 4.4 times greater than
that predicted from 1982-83 patterns (Fig. 5.11A).
The 1987 census was the last before the population
disappeared. In May 1988, density was a record low
(1 frog per 40 m of stream); by June, it had fallen to
zero.
Suspecting that the high density in March 1987
might be a clue to why the population had crashed,
Pounds and Crump (1994) considered two hypoth-
eses. First, if recruitment had been high in 1986, the166 Amphibians and Reptiles