1996; Berger et al. 1998), where they have spawned a
recent debate (Alford and Richards 1997, Hero and
Gillespie 1997, Laurance et al. 1997) One point of
agreement is that some anuran declines in seemingly
undisturbed highlands cannot easily be attributed to
pathogen outbreaks. The same applies to the declines
of anoline lizards at Monteverde (see Sec. 5.4.1). The
implication is that at least one other factor has been
important in these and similar cases.
Of the potential changes in ecological context men-
tioned above, climate change has received the least
attention. Unusual weather could exert indirect ef-
fects in at least two ways (Pounds and Crump 1994).
First, mist and cloud-water deposition punctuating
long dry periods might deliver contaminants at criti-
cal concentrations (the "climate-linked contaminant
pulse hypothesis"). Second, extreme conditions of
moisture or temperature could increase the probabil-
ity of pathogen outbreaks (the "climate-linked epi-
demic hypothesis"). These conditions might reduce
foraging success of host animals (Stewart 1995) or
otherwise weaken them and suppress their immune
responses, thereby increasing vulnerability to in-
fection (Carey 1993). Also, a lowland microparasite
could respond to altered climatic gradients by mov-
ing up mountain slopes with its insect vectors (Dob-
son and Carper 1992). If it encounters host popula-
tions that lack innate resistance, it could drive them
to extinction even where physical conditions re-
main within their range of tolerance. Finally, un-
usual weather might lead to an irruption of lethal
microparasites that are ordinarily present in small
numbers. An increase in the abundance of vectors or
a change in the spatial distribution of hosts might
favor rapid population growth in these natural agents
of mortality. The 1987 crash of amphibian popula-
tions at Monteverde coincided with warm, dry
weather associated with the 1986-87 El Nino (a peri-
odic warming of surface waters in the equatorial Pa-
cific; Graham and White 1988). Shortly before the
crash, Harlequin Frogs at one site underwent an un-
precedented shift in behavior, apparently in response
to desiccating conditions (see Sec. 5.6.1). In human
populations, recent outbreaks of cholera, malaria,
dengue, and hantavirus pulmonary syndrome appear
to be associated with climate oscillations related to
El Nino (Sprigg 1996).
The hypothesis that unusual weather has played a
role in population declines in highland areas is plau-
sible only if there are long-term trends suggestive of
climate change. Laurance (1996) discarded this hy-
pothesis for Australia after an analysis of monthly
precipitation data. Although rainfall totals were
often below average prior to declines, they showed no
major trends. Rainfall totals, however, may be less im-
portant biologically than temporal patterns in daily
precipitation (Stewart 1995). At Monteverde (1540 m),
there is a strong 26-year trend toward dry seasons
that are more severe (J. A. Pounds, M. P. L. Fogden,
and J. H. Campbell, unpubl. data). Variability of daily
rainfall has increased, leading to drier extremes with-
out producing detectable trends in monthly or yearly
averages. Days with no measurable precipitation
have become more frequent and have increasingly
coalesced into dry periods. Stream-flow minima have
also declined. Fluctuations in sea-surface temperature
(i.e., the signal of El Nino) account for much of the
interannual variation in these patterns, but there is
evidence of long-term drying trends after the effects
of El Nino are taken into account.A model of climate change on tropical mountains. These
climate trends suggest a change in the advective pro-
cesses that account for most precipitation during the
dry season (Pounds 1997). As moisture-laden trade
winds meet the windward (Caribbean) slope of the
Cordillera de Tilaran and flow upward, they cool
adiabatically, producing a large orographic cloud
bank (see Chap. 2, Physical Environment). I hypoth-
esize that atmospheric warming has raised the mean
height at which condensation begins and thereby has
increased the average altitude at the base of this cloud
bank. Higher clouds should deposit less moisture di-
rectly onto vegetation. Because they may pass over the
cordillera with reduced turbulence and drag, they are
also less likely to produce low-intensity precipitation
(mist) before dissipating on the leeward (Pacific) side,
and deposit less moisture directly onto vegetation.
Temperature patterns during the dry season are
consistent with this lifting-cloud-base model. Despite
a nocturnal warming trend, the average daily maxi-
mum for dry days has declined, implying that there
is an increase in the number of days with cloud cover
but no measurable precipitation. Clouds, which re-
duce radiative heat losses at night and the intensity
of solar radiation during the day, are principal modu-
lators of local temperatures in the tropics (Nieuwolt
1977).
Biological patterns also follow the model's predic-
tions (J. A. Pounds, M. P. L. Fogden, and J. H. Camp-
bell, unpubl. data). Many premontane breeding bird
species have invaded lower-montane habitats, whereas
some lower-montane species have retreated up the
mountain slopes. In multiple regression analyses, the
number of dry-season days with no measurable pre-
cipitation accounts for most of the year-to-year varia-
tion in the rate of invasion by premontane bird species.
The number of dry days is also strongly correlated
with abundance in lizard populations that have de-
clined and disappeared (see Sec. 5.4.1). The driest162 Amphibians and Reptiles