TOBY GARDNER
The majority of the recent reductions in the Earth's biodiversity can be attributed to direct human impacts on the environment. An increasing number of studies over the last decade have reported declines in amphibian populations in areas of pristine habitat. Such reports suggest the role of indirect factors and a global effect of human activities on natural systems. Declines in amphibian populations bear significant implications for the functioning of many terrestrial ecosystems, and may signify important implications for human welfare. A wide range of candidates have been proposed to explain amphibian population declines. However, it seems likely that the relevance of each factor is dependent upon the habitat type and species in question, and that complex synergistic effects between a number of environmental factors is of critical importance. Monitoring of amphibian populations to assess the extent and cause of declines is confounded by a number of ecological and methodological limitations.
Centre for Ecology, Evolution and Conservation
Department Biological Sciences
University East Anglia
Norwich
NR4 7TJ
United Kingdom
Human alteration of the global environment namely through habitat modification, agricultural practices, anthropogenically induced climate change, and atmospheric pollutants has triggered what is widely regarded as the sixth major extinction event in the history of life (CHAPIN ET AL., 2000). The extent of loss of biological diversity, and alterations in the distribution of organisms shows considerable variance both with respect to geographic location but also with respect to the ecological and taxonomic characteristics of the species involved. Usually proximal factors such as habitat destruction or modification are easily identified as the responsible cause of local losses of biodiversity, and as such local reductions in biodiversity are most frequently observed across the taxonomic spectrum. Like most terrestrial species amphibians are threatened foremost by habitat destruction (SALA et al., 2000). However, in the past few decades amphibian populations have been threatened by other incompletely understood factors in areas that are perceived to be intact from human disturbance (WALDMAN & TOCHER, 1998, ALFORD & RICHARDS, 1999, CAREY et al., 2001). The suspicion that amphibians are suffering from an unprecedented and abnormally high rate of decline even in protected areas was first voiced at the First World Herpetology Conference in September of 1989, in Canterbury, England (BARINGA, 1990), although it is clear that widespread concern existed long before this (BURY, 1999). The seriousness with which the scientific community recognised this problem was reflected in the rapid organisation of a NRC sponsored workshop in February of 1990 in Irvine USA, and in light of the perturbing reports presented (BLAUSTEIN & WAKE, 1990, WAKE, 1991), the establishment of a special task force on declining amphibian populations (DAPTF), allied with the Species Survival Commission of the IUCN. During the subsequent decade neither the scale of the problem nor the widespread concern expressed in both the scientific (WAKE, 1998), and public (BLAUSTEIN & WAKE, 1995, MORELL, 2001) community has seen any abatement. In order to tackle any problem in ecology it is essential that one is aware of the present level of understanding of its scale, diagnostic characteristics, and methodologies appropriate to its resolution. This review complements previous similar efforts (e.g. WALDMAN & TOCHER, 1998, ALFORD & RICHARDS, 1999) by exploring many important advances in the last two years. Edward Wilson recently described conservation biology as the 'intensive care ward of ecology' (WILSON, 2000), and as such a conservation biologist who lacks an up-to-date appreciation of their field is failing the prescription of this definition in inadvertently advocating inefficient, repetitive, or even counterproductive research. This review seeks to provide such a revision, dealing in turn with; (1) the ecological and human importance of amphibians in natural ecosystems, (2) evidence for population declines and caveats in their interpretation, (3) the range of candidates which have been proposed to explain such declines, and (4) some challenges presently facing conservation biologists in resolving and preventing amphibian declines.
A world-wide decline of amphibian populations could have a significant and detrimental impact on both natural ecosystems and human welfare. Amphibians are integral components of many ecosystems, often constituting the highest fraction of vertebrate biomass (BURTON & LIKENS, 1975, BEEBEE, 1996). Their conspicuous role is noted to be of particular importance in tropical forests, where in acting as both predator and prey species, they play a key role in trophic dynamics (TOFT, 1980, BLAUSTEIN et al., 1994c). Their high collective biomass, alongside their high digestion and production efficiencies (WOOLBRIGHT, 1991) go someway to explaining their potential importance in such 'functions' as the maintenance ecosystem energetics and carbon flow (PEARMAN, 1997) - namely through the maintenance of arthropod abundance (GUYER, 1990), and the provision of an critical prey base for higher order predators, such as arachnids, snakes, and birds (GUYER, 1990, WOOLBRIGHT, 1991, DUELLMAN & TRUEB, 1994). In identifying the functional significance of amphibians its is clearly of relevance to understand whether species diversity per se plays a unique role over and above species identity - i.e. are a few specific and perhaps more abundant frog species sufficient to maintain the natural integrity and productivity of the ecosystem? In view of the limitations on the world's resources for the conservation of biodiversity, it would seem sensible to identify the functionally important amphibian species or 'guilds' in order to prioritise concern and subsequent potential conservation action following a reported decline in number. However such an approach could be very dangerous, as it is often extremely difficult if not impossible to identify the functional role or contribution of many species (CHAPIN et al., 2000). Some (limited) empirical evidence exists to offer an explicit justification for the functional importance of species richness per se (LOREAU et al., 2001). However, a more convincing argument is that differences in the environmental tolerances of many species that may be functionally analogous to dominant species can provide critical insurance or resilience for the system in the face of climate change or altered disturbance patterns (WALKER, 1995, WALKER et al., 1999, NAEEM, 1998). In light of our ignorance of the ecology of most amphibians, and the growing domination of natural systems by human activities (VITOUSEK et al., 1997), it seems that an attitude of concern towards all populations that could be potentially at risk is prudent, if not essential. It is important to stress that concerns about the functional importance of species in no way detracts from the importance of other values humans can attach to biodiversity - including cultural, existence and intrinsic values (DOLMAN, 2000).
The second most recognised importance of amphibians is their potential role as indicators of global environmental health and resilience (BLAUSTEIN & WAKE, 1990, BARINGA, 1990, DIAMOND, 1996). They inhabit both aquatic and terrestrial habitats, and are thus exposed to aquatic and terrestrial pollutants - to which they are particularly sensitive due to their highly permeable skin (DUELLMAN & TRUEB, 1994). Furthermore they are 'exposed' to a large proportion of the biotic environment, due to their roles as both herbivores during larval stages and carnivores as adults, making them potentially good indicators of changes in both floristic and faunal community composition - possibly induced through environmental stress. As BARINGA (1990) states, the fact that amphibians as a group have remained largely unchanged since the era of the dinosaurs, highlights the potentially disastrous consequences for humans and other species if their suspected demise continues unabated. Finally, from a purely anthropocentric perspective amphibians represent a storehouse of pharmaceutical products waiting to be exploited fully (BLAUSTEIN & WAKE, 1995). Some compounds already extracted are presently being used as painkillers and in the treatment of traumas such as burns and heart attacks, whilst many more undoubtedly await discovery.
Although serious recognition of the potential problem of declining amphibians was not afforded until the last decade, individual anecdotal reports of population declines have been known since the late 18th century (BURY, 1999). However it was during the 1980's and early 1990's that the observations of more dramatic and scientifically credible declines were made. Such declines include notable individual examples such as the Golden toad (Bufo periglenes) and Harlequin frog (Atelopus varius) (CRUMP et al., 1992, POUNDS & CRUMP, 1994), the Cascades frog (Rana cascade) (FELLERS & DROST, 1993), and the Yellow and Red-legged frogs (Rana muscosa and Rana aurora) (BLAUSTEIN & WAKE, 1990). All of these declines have occurred in areas considered largely intact from human interference, which explains their common citation in justifying concern for the viability of other seemingly well protected amphibian populations. Further evidence for the apparent vulnerability of the class Amphibia as a whole comes from reports of population declines across whole communities of amphibians at the regional level, also in relatively pristine areas; the Central Valley of California (DROST & FELLERS, 1990, FISHER & SCHAFFER, 1996), the montane forests of Eastern Australia (LAURANCE et al., 1996), and the Monteverde cloud forest of Costa Rica (LIPS, 1998, LIPS, 1999). It is hard to draw global conclusions from such varied examples, although a number of commonly occurring factors or attributes can be identified (LIPS, 1998, ALFORD & RICHARDS, 1999, CAREY, 2000, CAREY et al., 2001, MIDDLETON et al., 2001). These factors include:
Due to the heavily skewed distribution of amphibian biologists towards North America, Europe, and Australia it is possible that a number of these common attributes are at least partly artifacts of research bias. To remove some of this bias and view the declining amphibian problem from a more global perspective, it is necessary to collate information from across many sites and many species.
In attempting to draw global conclusions or patterns about a particular ecological phenomenon (such as population declines) from across different studies, one is commonly faced with two main problems; the inaccessibility of many research reports, and the extreme variability in monitoring techniques used - from the purely anecdotal to the scientifically rigorous. A number of recent reviews have attempted such a difficult collaboration in order to view the plight of amphibians from the widest possible perspective - both at spatial and temporal scales, the two most noteworthy of which are those of ALFORD & RICHARDS (1999) and HOULAHAN et al. (2000). ALFORD & RICHARDS (1999) considered 85 time series of amphibian populations spanning the period between 1951-1997, and following regression analysis concluded that more populations correlated negatively against time than would be expected under their null hypotheses of 'normal' population fluctuations, with 67% of relationships being negative. However, they found no evidence that the proportion of populations decreasing changed over time - i.e. what one would expect if the proposed agents of decline where becoming more prevalent or intense. As the authors themselves admit, it is impossible to draw firm conclusions about the global status of amphibian populations due to variance in the size of data sets, their methodological origin, and the inter-specific variance in population dynamics which renders their null model far from optimal for all amphibian species. HOULAHAN et al. (2000) made by far the most exhaustive attempt to date in collating data from 936 populations of 157 species from 6 continents, for studies of between 2 and 31 years duration. Although their results identify marked temporal variation in the speed of the decline, and spatial variation as to its extent, a definite negative relationship is clearly evident, adding perhaps the first real quantitative 'weight' to the declining amphibian phenomenon. Criticism has recently been raised as to the validity of the statistical averaging methods used by Houlahan and colleagues (ALFORD et al. 2001; but see HOULAHAN et al. (2001), although re-analysis under the alterative methodology (ALFORD et al., 2001) still concluded that an overall population decline existed and disagreed only in the shape of the relationship - with the more recent interpretation identifying a increase in the rate of declines in the last decade. A number of serious inadequacies exist in the study by HOULAHAN et al. (2000) - for example it includes only four studies from Latin America, despite the fact that this continent hosts about half the worlds amphibian species (DUELLMAN, 1999). However, a recent synthesis of published and unpublished (>95% of the total) work from both Central and South America adds strength to the evidence for a global decline (YOUNG et al., 2001). In summarising 118 monitoring projects, population declines were found to be widespread, occurring in 13 countries, with 40 cases of recent extinction or regional extirpation affecting 30 genera and nine families of amphibians.
More obscure and perhaps counter-intuitive examples of the deleterious impacts of habitat modification exist. For example, in the case of the Natterjack toad (Bufo calamita) in Britain, where the removal of modification (grazing) on shrub heathland led to the encroachment of tall vegetation, thus allowing the entrance of the more successful competitor Bufo bufo - the Common Toad (BEEBEE, 1977). Such examples serve to emphasise the fragility of many ecological systems to what we may perceive to be minimal human intervention. However, habitat destruction and modification although of prime concern, are usually easily to isolate, and therefore if possible to rectify. It is the proposal that amphibian declines in largely pristine areas of the world are the result of more indirect and complex reasons that is cause for exceptional concern (WAKE, 1998, WALDMAN & TOCHER, 1998, CAREY, 2000, CAREY et al., 2001).
A number of other factors serve to shed doubt on the responsibility of UV-B in global declines. Most of the field experimental studies cited above have been conducted in shallow, clear high altitude ponds, largely in high latitude locations such as North America and south-eastern Australia. UV-B radiation is largely absorbed in the first few centimetres of the water column (NAGLE & HOFER, 1997), and the depth of penetration is negatively correlated against the dissolved organic carbon content (CRUMP et al., 1999) - factors which suggest that UV-B radiation is unlikely to be a problem in bottom laying species, or in forest (and especially tropical) species. An initial lack of evidence for significant increases in UV-B at tropical or sub-tropical latitudes since the mid-1970's (MADRONICH & GRUJII, 1993) further diminished its perceived importance in the decline of tropical amphibian populations (e.g. CRUMP et al., 1992, LIPS, 1998). However, recent remote sensing analysis (MIDDLETON et al., 2001) has identified increases in both annual and daily levels of UV-B exposure (average and maximum) between 1978-1998 at Central and South American sites where amphibian population declines have been recorded (e.g. POUNDS & CRUMP, 1994, LIPS, 1998, LIPS, 1999, POUNDS et al., 1999). Although this work reports only correlative rather then causative evidence, it identifies the value and urgency for further field studies on the effects of UV-B radiation on amphibian populations.
Although some of the above evidence is convincing in showing an effect of high UV-B radiation on embryo mortality and larval survival, the ecological significance of such a phenomenon at the population level is far from clear, and equally difficult to assess (ALFORD & RICHARDS, 1999). For example there may be density dependent compensation effects, through the enhanced fitness of competing individuals that survive high levels of UV-B exposure. The potential indirect effects of enhanced UV-B on amphibian dynamics, such as changes in water chemistry and food supplies, are even less well known (ALFORD & RICHARDS, 1999). Finally it is likely that unimodal experiments manipulating only levels of UV-B are inadequate, and that the crucial agent of decline could be in the interaction of UV-B with other key environmental stresses (see below).
Acidification and toxins
The existence of amphibian extinctions and population declines in what are
otherwise seen to be some of the most pristine environments on Earth has
led to the frequent suggestion that atmospheric pollutants may act as indirect
agents of decline (LIPS, 1998, CAREY et al., 2001). Such pollutants could
originate from neighbouring and foreign agricultural depositions, as well
as from factory emissions of industrialised nations, and are able to travel
vast distances and persist for considerable periods of time. Recent analysis
of remote sensing data in Puerto Rico has shown spatial correlations between
urban and agricultural pollutants and amphibian population declines (STALLARD,
2001).
One of the most acknowledged remote impacts of human activity is increased acidity of rainfall, a phenomenon of great potential importance in light of the importance of the annual water regime to amphibians. Increased acidity of ground and pond water is suspected to have both lethal and sub-lethal effects on amphibian populations through a number of factors; enhanced embryo and larval mortality, reduced egg and larval growth, reduced reproductive output, delayed hatching times, reduced adult body size, alterations in geographic distribution, and alterations in predator-prey ratios through indirect effects on plant growth and pH sensitive competitors and predators (FREDA & DUNSON, 1986, WALDMAN & TOCHER, 1998, ALFORD & RICHARDS, 1999). Some field and laboratory work has provided evidence for such detrimental effects, for example; reduced ion exchange and larval growth in the Wood Frog Rana sylvatica (FREDA & DUNSON, 1986), and a significant reduction in range size of Natterjack Toad Bufo calamita following long term acidification of many British ponds (BEEBEE et al., 1990). Evidence also exists for the potential role major agricultural pollutants, namely pesticides and herbicides in precipitating amphibian declines (BERRIL et al., 1994, HECNAR, 1995, RUSSELL et al., 1995), with early embryonic stages being particularly vulnerable (CAREY & BRYANT, 1995). Alongside cases of direct mortality (CAREY & BRYANT, 1995), empirical evidence has also identified important indirect influences of agricultural pollution on populations, e.g. through altered recruitment and predator response behaviours (COOKE, 1971, BRIDGES & SEMLITSCH, 2000). The marked interspecific variation of amphibians in their susceptibility to pesticides alongside geographic variation in their deposition is highlighted as a potential explanation for the heterogeneous nature of observed declines at both taxonomic and geographic scales (BRIDGES & SEMLITSCH, 2000).
Although it is clear from the above examples that an increase in environmental acidity or other pollutants can have a negative effect on some amphibian populations, the actual physiological mechanisms remain unclear (CAREY et al., 1999). Furthermore, and perhaps most crucially, it is once again not evident what the overall consequences of such environmental changes at the population level would be, and there is very little data implicating contaminants on the recent catastrophic population declines (ALFORD & RICHARDS, 1999). However, as in the case for enhanced UV-B levels, it is possible that the critical role of environmental contamination in population declines comes from its interaction with other causative agents (CAREY et al., 2001, STALLARD, 2001; see below).
Predation
Biotic interactions amongst and between species can play a critical role
in determining their relative local abundance, distribution and population
dynamics (RICKELFS & SCHLUTER, 1993, HUSTON, 1994). Although perhaps
more appropriately considered as human induced habitat modification, the
introduction of exotic predators to amphibian environments has been implicated
as the factor responsible for many population declines, including the collapse
of whole communities (FELLERS & DROST, 1993, FISHER & SCHAFFER,
1996, HECNAR & M'CLOSKEY, 1996a). Aside from such co-incidental evidence,
experimental manipulations of predator and amphibian distributions provide
firm support as to their devastating effect on amphibian populations. Two
good examples are the significant reduction in survival of the endangered
Red-legged frog (Rana aurora) in California, following the introduction
of the two larval predators - Mosquitofish (Gambusia affinis), and
Bullfrogs (Rana catesbeiana) (LAWLER et al., 1999), and secondly,
a severe impact of both Mosquitofish, and a crayfish (Procambarus clarki)
on the eggs and larvae of the Californian newt, Taricha torosa (GAMRADT
& KATS, 1996). Although the introduction of exotic predators such as
the above is considered to be a prime cause of population decline across
North America (FISHER & SCHAFFER, 1996), their role is usually easily
identified, and as such seems unlikely to be a global factor, especially
in largely pristine tropical areas.
Disease
The remote nature of many amphibian population declines, in addition to
the frequent observations of larval and adult growth abnormalities, has
led to the perhaps unsurprising and widespread implication of disease (CAREY,
1993, CAREY, 2000). In particular the wave-like pattern of population decline
across the range of many threatened species seem to implicate the role of
a biotically induced agent - as observed in both the Atlantic forest of
Brazil (HEYER et al., 1988), the Eastern montane forests of Australia (LAURANCE
et al., 1996), and the forests of Panama and Costa Rica (LIPS, 1998, LIPS,
1999). Perhaps the most confident proposition as to the culpability of disease
in precipitating the collapse of an entire amphibian community is in Australia,
where 14 endemic species have decreased by more than 90% in the last 15
years (LAURANCE et al., 1996). The authors note the extreme virulence of
the disease as being evidence of its potentially exotic nature, and report
histological changes in infected tissue of diseased individuals as being
consistent with viral infection. (LIPS, 1998, LIPS, 1999) identified a fungal
infection found on dead individuals as being the most likely cause of population
decline in the forests of Panama, between 1993 and 1997. Furthermore she
notes that the similarity in timing of declines, regional climatic factors,
frog phylogenies, and clinical symptoms, as being suggestive of the same
causal factor being responsible for catastrophic amphibian declines noted
in nearby Costa Rica (POUNDS & CRUMP, 1994, LIPS, 1998). Support for
this theory comes from (Berger et al., 1998)who identified the same chytridomycete
(Chytridiomycota: Chytridiales) fungus on dead anurans from forests of both
Central America, and also of Queensland, Australia - adding some further
strength to the argument of (LAURANCE et al., 1996), although the disease
origin differs. The same fungus has been more recently identified to the
species level (LONGORE et al., 1999), and an increasing number of reports
have confirmed its presence in all of the six continents that are inhabited
by amphibians (DASZAK et al., 1999, CAREY, 2000). The most recent report
of the expanding geographic distribution of this species, documents its
arrival in Europe where it is implicated as being responsible for the disappearance
of the Common Midwife toad (Alytes obstretricans) from more than
85% of its breeding sites in Spain (BOSCH et al., 2001). The relatively
sudden observation of catastrophic declines in such disparate areas of the
world is suggestive of either a recent increase in virulence, or decrease
in amphibian immuno-activity, perhaps due to a key interaction with a changing
global climate - although the potential mechanisms behind any such interaction
are poorly understood (CAREY, 2000; see below). In addition to a decrease
in amphibian immuno-activity or an increase in pathogenic virulence, an
increase in the level of pathogenic activity could be affecting amphibian
populations through changes in the food supply or competitive ability of
species (CAREY et al., 2001). Further convincing evidence as to the role
of disease in population declines comes from the Pacific north-west of America,
where a different species of fungus, Saprolegina ferax (a globally
distributed fish pathogen), has been implicated as responsible for declines
in the Boreal toad Bufo boreas, through increased egg mortality (BLAUSTEIN
et al., 1994b, KIESECKER & BLAUSTEIN, 1997).
There is clearly enough convincing evidence to support the two facts that disease agents can be highly detrimental to amphibian fitness and survival, and that furthermore, evidence of them can be found in many areas where catastrophic declines have occurred. However, as for most if not all agents of decline, it is very difficult to attribute what contribution they make to the overall population dynamics of declining amphibians. Strong circumstantial evidence exists as to the role of disease in mass declines such as that observed in Australia (LAURANCE et al., 1996). However, when experimental proof is difficult to obtain, it is easy to argue for competing hypotheses which may produce equally parsimonious statistical comparisons of a potential agent of decline against the spatial distribution of population declines (ALFORD & RICHARDS, 1997, HERO & GILLESPIE, 1997). Despite the ubiquitous presence of a large range of competing hypotheses to explain any one population decline, it is crucial to the progress of science that plausible hypotheses are voiced, if only for their heuristic value in targeting future research and formulating further, refined hypotheses (LAURANCE et al., 1997). As noted below, in the case of disease it is even more likely than in other agents of decline, that interactions of disease vectors with other environmental factors plays a crucial role in determining their impact on amphibian populations (CAREY, 2000).
Climate and weather
As discussed already, amphibians are particularly sensitive to changes in
their external environment, both due to their biphasic lifestyle in existing
as both aquatic larvae and terrestrial adults, and due to their highly permeable
skins. Perhaps the most important component of the abiotic environment to
both amphibian fitness and population dynamics is the maintenance of a stable
and predictable water-temperature regime (POUNDS & CRUMP, 1994, LIPS,
1998). Many amphibians are subject to both water and temperature sensitive
physiological limitations on locomotive and reproductive activities. As
a consequence of this the balancing of evaporative water loss against direct
absorption through the skin is a critical functional attribute, as has been
observed in the Marine toad Bufo marinus (POUNDS & CRUMP, 1994).
Aside from detrimental effects of disrupting this balance (i.e. through
desiccation), at the individual level, the water regime in particular can
play a vital role in many other aspects of amphibian ecology, including;
(1) determination of phenological patterns of reproductive activity (WELLS,
1977, AICHINGER, 1987, GASCON, 1991), (2) determination of the spatial distribution
of community assemblages (INGLER & VORIS, 1993), and (3) in the provision
of suitable breeding sites and conditions (e.g.
PYBURN, 1970).
The suspected role of alterations in the annual water regime of amphibians in global population declines, follows increasing recognition of gradual changes in the global climate due to human activities. There has been a discernible human influence on world temperatures during the last century, with average temperatures projected to increase by between 1.4 and 5.8°C by 2100, with considerably greater regional variation (IPCC, 2001). One consequence of this that is relevant here is a projected increase in activity of the tropical hydrological cycle, with the prediction of erratic and frequently severe weather patterns (GRAHAM, 1995, IPCC, 2001). The effect of climatic change on ecological systems has been observed at all levels, from population and life history alterations, to shifts in geographic range, and subsequent changes in community composition resulting in disruption of ecosystem structure and function (HUGHES, 2000, MCCARTY, 2001). It is therefore perhaps unsurprising that many changes in the population dynamics of amphibians, organisms which are so closely coupled with their environment, have been attributed to changing climatic and weather patterns.
A number of notable reports documenting multiple amphibian declines have implicated the potential role of synchronously observed climatic extremes, and in particular, periods of reduced or abnormally distributed rainfall (CORN & FOGELMAN, 1984, HEYER et al., 1988, BERVEN, 1990, CRUMP et al., 1992, FELLERS & DROST, 1993, STEWART, 1995). Owing to the sensitive response of amphibian breeding cycles it is easily conceivable that a simple shift in the commencement of the wet season in seasonal environments could either trigger premature spawning and subsequent desiccation of eggs, or if early rains are abnormally intense, the flooding of breeding ponds, and an equally disastrous loss of an entire breeding attempt (WELLS, 1977, CRUMP et al., 1992). POUNDS & CRUMP (1994) executed a detailed analysis of the infamous declines in number of Golden toad and Harlequin frog populations in the Monteverde cloud forests of Costa Rica, and concluded that coincidentally low periods of rainfall during phases of population decline were at least in part responsible. It was clear that depletions in number of the Harlequin frog populations (for which demographic data was available) matched climatic records of reduced rainfall during both the 1982-1983 and 1986-1987 El Ninô induced drought periods. The potential role of long-term warming and increased intensity of precipitation patterns, when coupled with intense warm periods of El Ninô - Southern Oscillation cycles, has been noted to be of severe consequence for many biological communities (MCCARTY, 2001), and in light of the above, particularly so for amphibians (POUNDS, 2001). With reference to the example of (Crump et al., 1992), it has been recently calculated that the effect of El Ninô events in Central America is expected to be through severe drought periods rather than increased rainfall (HOLMGREN et al., 2001). Although the juxtaposition of the timing and extent of population declines in Harlequin frogs with the timing and intensity of periods of drought suggests that they are causally linked, it is much more difficult to identify either the environmental variable that is of crucial ecological significance, and further, the exact mechanism by which that change acts to reduce amphibian populations. (POUNDS et al., 1999) in a further analysis of the situation in the Monteverde cloud forests, isolate a perhaps rather unintuitive climatic variable as being closely correlated with not only amphibian declines but also with demographic changes in many other taxa, including birds and reptiles. The climatic variable is that of decreasing dry season mist frequency, which suggests that the important water-related mechanism affecting amphibian populations is likely to be an increase in desiccating conditions affecting egg hatching in non-aquatic species, alongside individual survival, rather than changes in reproductive phenology and breeding behaviour. Human induced climate change can be implicated here, as dry season mist frequency is negatively correlated with sea surface temperatures of the equatorial Pacific, which have increased dramatically since the mid-1970's (STILL et al., 1999). Such examples highlight the importance of studying the environment of a species under threat so as to identify the ecologically important variables, and allow an assessment of future population stability through the parameterisation of predictive ecological models - both verbal and mathematical (MCCARTY, 2001).
Aside from the above, a number of other mechanisms have been suggested to explain the potential role of climatic change in precipitating amphibian population declines. A shift in rainfall patterns could result in a change in availability of breeding sites, a reduction in which could increase levels of competition and predation, and even vulnerability to disease, resulting in a reduced overall reproductive output for that year (DONNELLY & CRUMP 1998). An increased frequency of drought periods, coupled with increased temperatures, have also been identified as having potentially severe effects on leaf litter species which don't congregate to breed, through alteration of their arthropod prey base and an increase soil desiccation (DONNELLY & CRUMP, 1998). Finally, there is evidence of changes in spring spawning times of amphibian species in England, showing that amphibian reproductive cycles are highly sensitive to climate warming, with possible long-term consequences for population dynamics through alterations of biotic interactions (BEEBEE, 1995). However, as with all the potential agents of amphibian decline listed above, changes in climatic patterns cannot always be found to explain observed declines (LAURANCE, 1996, ALEXANDER & EISCHEID, 2001). Furthermore, due to the close coupling of amphibian population dynamics to their ecological environments, it is likely that any climatic change would affect amphibians through interactions with other biotic and abiotic factors, to which both the external climate and amphibians themselves are closely linked.
Interaction effects amongst environmental factors
Frequently, separation of the almost myriad of current hypotheses for amphibian
declines in any one situation can be almost impossible, although some recent
advances have been made using GIS technology to compare spatial patterns
of decline with spatial patterns of potentially causal agents (DAVIDSON
et al., 2001). However, as noted above, most of the studies to date that
have considered a single causal mechanism behind amphibian declines have
invoked a critical interaction between multiple factors (ALFORD & RICHARDS,
1999, CAREY et al., 2001, MIDDLETON et al., 2001). Such acceptance means
that despite the urgency of explaining observed declines, it is important
to realise that interacting suites of environmental change could produce
complex effects that are often difficult or even inappropriate to isolate
(ADAMS, 1999).
A number of recent experimental and observational studies offer support to the importance of interaction and synergistic effects between different hypothetical agents of decline. Increased UV-B exposure has been shown to increase the susceptibility of some amphibian species to disease (KIESECKER & BLAUSTEIN, 1995, KIESECKER & BLAUSTEIN, 1997). Furthermore an increase in UV-B can act synergistically with reduced pH levels to reduce embryo survival, when each factor alone is shown to have no significant effect (LONG et al., 1995). Normally harmless diseases may increase their effective virulence under increased environmental pollution by contaminants such as pesticides (CAREY & BRYANT, 1995, BRIDGES & SEMLITSCH, 2000), and even different diseases themselves can be seen to act in concert in order to produce a detrimental effect (CUNNINGHAM et al., 1996). Temperature and water pH have been shown to interact to increase the detrimental effect of pathogenic fungi on reproductive success and survival in amphibians (BEATTIE et al., 1991, BANKS & BEEBEE, 1998). An interaction between a changing environment and either the virulence and distribution of a pathogen or the immuno-activity of amphibians may not be sufficient to increase mortality directly. However, through differential responses of both different amphibian species and predators it may significantly alter the competitive and predatory dynamics resulting in a shift in the species composition or abundance rank (KIESECKER & BLAUSTEIN, 1999). The presence of carbaryl pesticide has been shown to dramatically increase the level of predation stress felt by the Gray Tree frog Hyla versicolor with mortality being found to be 2-4 times greater when individuals were subject to predatory cues in addition to the pesticide (RELYEA & MILLS, 2001).
Although the mechanisms are not well understood (CAREY, 2000, CAREY et al., 2001), it is likely that global climate change can interact importantly with virtually all local environmental factors with respect to their effect on amphibian populations (POUNDS, 2001). For example an increase in drought events and the subsequent loss of many ponds and breeding sites, could greatly exacerbate the effects of local predators (DROST & FELLERS, 1996). Also an increase in temperature can increase the volatility of potentially harmful chemical deposits, the aerial concentration of which may then be increased due to a reduced frequency of rainfall events (POUNDS & CRUMP, 1994). Due to the sensitivity of many ecological systems to climatic change (MCCARTY, 2001), it is likely that alterations of key environmental variables such as rainfall patterns and temperature, have the effect of reducing or even removing an important constraint on the potential of many agents of decline, both with respect to their geographic distribution but also in their physiological or biotic effect (e.g. UV-B, pH, disease) (POUNDS, 2001). Just how complex an effect such changes in climatic parameters can precipitate has been illustrated in a very recent study by KIESECKER et al., (2001). Their findings illustrate that climatic induced reductions in water depth at amphibian oviposition sites have caused a high level of mortality in embryos, by increasing their exposure to UV-B radiation, and consequently their vulnerability to infection by disease. The implication of this is that elevated sea surface temperatures in the tropical Pacific, which drive large scale climatic patterns, could be the precursor for many pathogen-mediated amphibian declines world-wide (KIESECKER et al., 2001). One common theme with respect to the implication of synergistic effects in amphibian population declines is that the direct or proximate mechanism which increases mortality is thought to frequently be disease following immosuppression (CAREY, 1993, CAREY, 2000).
It should be clear from the above discussion that observed amphibian population declines seem unlikely to be the result of a small number of independent global agents, but rather the complex interaction of local effects in the context of varying regional influences and global climatic change. In order to study the existence of such effects in natural populations, and thus elucidate the relative stability and integrity of such populations, well-planned programs of observation and experimentation are needed (ALFORD & RICHARDS, 1999, CAREY, 2000). Furthermore, in light of the importance of both abiotic but also biotic interactions, it is important to gain an understanding of the interactions of the populations under study with both other species (amphibians and other potential competitors and predators), and their physical environment. Finally in order to identify the existence of a real population decline with confidence, development of specific 'null' hypotheses or models is needed to describe how amphibian populations behave in the absence of external pressures (ALFORD & RICHARDS, 1999, MARSH, 2001).
In studying the proposed phenomenon of global declining amphibian populations, there are perhaps three main questions in which uncertainty remains; (1) How to determine real declines from natural population fluctuations?, (2) Whether human induced agents can be isolated as the potential cause of the decline?, and (3) Whether global agents are responsible for the majority of observed declines?. In light a growing recognition of the important implications of the mounting extinction crisis (see above), we cannot afford to be either complacent or conservative in our approach towards answering such central questions. In view of this there is a desperate need for comprehensive monitoring studies on amphibian populations world-wide (BLAUSTEIN et al., 1994c, WAKE, 1998, YOUNG et al., 2001). As can be seen from recent compilations by ALFORD & RICHARDS (1999) and HOULAHAN et al. (2000), existing studies exhibit a notable disparity with respect to length, scope, and detail. In order to draw firm conclusions at both the local and global level, it is imperative that future studies build upon previous work, and where possible incorporate recent advances in our understanding of amphibian species and their population dynamics. It is therefore instructive here to draw attention to a number of considerations, both biological and methodological, which are central to planning amphibian-monitoring programs.
The detection of real population declines which are deserving of concern, from purely natural population fluctuations can pose a serious problem in monitoring programs. It is essential that we understand the natural levels of variability inherent in amphibian populations, so as not to invoke unnecessary conservation and management action - a result that could severely compromise support for conservation in other situations (PECHMANN et al., 1991). An understanding of the levels of variability inherent in population dynamics is central to calculating both the statistical power of a monitoring program (MARSH, 2001), and the level of extinction risk from stochastic events (LEIGH, 1981, ENGEN & SAETHER, 1998, MARSH, 2001). Furthermore, an appreciation of population variability is fundamental to understanding the processes that drive population fluctuations (SEMLITSCH et al., 1996).
There exist both temporal and spatial aspects of amphibian population dynamics that can serve to confound the attempts of many monitoring programs to elucidate real declines in number. A number of ecological variables have been proposed as predictors of variability in amphibian populations, including; habitat type (WILLIAMS & HERO, 1998, WILLIAMS & HERO, 2001), reproductive mode and density dependent processes (SEMLITSCH et al., 1996, ALFORD & RICHARDS, 1999, MARSH, 2001), rainfall, taxanomic family, and latitude (MARSH, 2001). Firstly, there is strong evidence of intraspecific density dependence in many amphibian populations (BERVEN, 1990, PECHMANN et al., 1991, MEYER et al., 1998, ALFORD & RICHARDS, 1999). Crucial life history factors that appear to be regulated by density dependence include larval survival, larval size and time to metamorphosis. In a highly heterogeneous environment such as a forest, variance in such factors could produce seemingly chaotic fluctuations in population size (TURNER, 1962, BERVEN, 1990, PECHMANN et al., 1991). In the wood frog (Rana sylvatica) (Berven, 1990)recorded variation in Ro (the net population replacement rate) between 0.009 - 7.49 over only 7 years, and as monitoring programs are rarely longer than this (BLAUSTEIN et al., 1994c, ALFORD & RICHARDS, 1999) it is easy to see how a short term population decline may be interpreted with unwarranted concern. Secondly, it is possible that at the level of the population, density dependent effects following fluctuations in resource levels may override the effect of any density independent environmental stress factors that may act to reduce juvenile or adult survival. However, as the judgement of 'natural' levels of stability in biological populations is exceedingly difficult to make (CONNELL & SOUSA, 1983), it is consequently difficult to identify the ecological significance of any such contribution to mortality or reduction in reproductive success to overall population dynamics, even though they may be non-trivial. Aside from density dependence, a second important consideration of the temporal dynamics of amphibians is in the fact that fluctuations in breeding aggregations may be much greater than fluctuations in total population size, due to intra-population variance in breeding behaviour (PECHMANN et al., 1991). This point is of particular relevance, as due to severe logistical constraints most censuses of amphibians and especially frogs and toads (Amphibia: Anura), are conducted on aggregations at breeding sites (ALFORD & RICHARDS, 1999; see below).
Aside from temporal considerations one must also take into account the spatial aspects of amphibian population dynamics in making any conclusions about population stability or integrity. It is becoming increasingly recognised that many amphibian populations often exist in a metapopulation structure, where regional processes affecting extinction and colonisation of habitat patches, play a dominant role in determining local species assemblage composition and population size (HANSKI & GILPIN, 1991, HECNAR & M'CLOSKEY, 1996b, ALFORD & RICHARDS, 1999, MARSH & TRENHAM, 2000). It is important to recognise that from the perspective of a monitoring program effective habitat 'patches' which are subject to such regional influences can represent the breeding sites or transects under human surveillance, and with respect to the actual amphibian population are often not ecologically distinct. The importance of regional processes to the persistence of local populations means that local extinction can occur due to essentially stochastic factors that are unrelated to the local environmental (abiotic or biotic) quality (SJOGREN, 1991, MARSH, 2001). Local amphibian populations are predisposed to stochastic extinctions due to the susceptibility of a peak-breeding attempt to climatic conditions (i.e. droughts or floods), their relatively short life spans (MARSH & TRENHAM, 2000), and their philopatric behaviour (WALDMAN & TOCHER, 1998). However, the essential point is that although devastating reductions in population size may be observed at a particular monitoring site, at the regional spatial scale the species may be perfectly healthy, adding doubt to the extrapolation of many population censuses to conclusions about the viability of an entire species. In order to confidently assess the stability of an amphibian population, and attribute a reason to any observed decline, it is important that such spatial factors are considered (see below).
The natural variability and complexity that is in inherent in both temporal and spatial amphibian population dynamics has already been highlighted above. It is crucial to recognise that such factors introduce serious practical considerations and caveats in the construction, execution, and analysis of amphibian monitoring programs. A direct consequence of such natural levels of variability is that the failure to find a significant decline in number of a particular population may frequently not be due to a lack of real decline, but rather to a lack of statistical power (GIBBS, 1995, REED & BLAUSTEIN, 1995, HAYES & STEIDL, 1997, ALFORD & RICHARDS, 1999, MARSH, 2001). The statistical power of a test for a population decline can be defined as the probability of rejecting the null hypotheses of no decline given that the null hypothesis is false and the alternative hypothesis of a declining population is true. Calculation of power requires knowledge of a number of factors, namely the sample size, the desired alpha level for avoiding Type I errors, the natural variance in sample size, and the effect size (PETERMAN, 1990). The value of conducting a prospective power analysis (HAYES & STEIDL, 1997) is severely limited by the high level of uncertainty inherent in the last two factors; (1) the natural coefficient of variability in amphibian population size - in particular in the context of the specific monitoring approach being used, and (2) the level of population decline (effect) which bears ecological significance for the future stability of an amphibian population. Uncertainty in these values produces an equal level of uncertainty in level of power calculated (GIBBS, 1995). Attempts should be made to calculate the confidence in intervals associated with estimates of power, and furthermore there may be considerable merit in using Bayesian approaches to estimate levels of uncertainty (HILBORN & MANGEL, 1997, WADE, 2000). It is suggested here that in light of the serious logistical and financial limitations imposed on many, if not most amphibian monitoring projects, the 'guestimating' of such variables is a dangerous game as it may render void many otherwise valuable projects which are lacking in apparent statistical rigour (and crucially lacking in ability to expand the project's sample size to achieve a satisfactory level of power). Except in situations where the species under surveillance is well studied, it may be of greater ecological significance to take a comprehensive approach to monitoring which incorporates a number of key ecological, as well as methodological considerations. This will hopefully achieve an increased understanding of the environmental requirements and population dynamics of the specific focal species, and afford greater confidence in any data interpretation. This does not nullify the clear value of prospective power analysis, but rather suggests that there is a great deal of merit in carefully considered monitoring projects which do not yet hold the minimum level of information needed to make such a preliminary analysis worthwhile. Identified below are some of the considerations deemed central to amphibian population monitoring.
For reasons emphasised earlier the two main problems facing monitoring projects are the logistical constraints on their temporal and spatial focus. Clearly in order to elucidate real declines from stochastic fluctuations, a long time series is highly favourable, although as seen from recent literature reviews few studies are longer than 5 years, and even less are more than 10 (ALFORD & RICHARDS, 1999, HOULAHAN et al., 2000, YOUNG et al., 2001). It is important to note however that an increase in the length of the study period will undoubtedly increase the perceived level of variability in population size and distribution due to the incorporation of a greater range of environmental conditions (PECHMANN & WILBUR, 1994, MARSH, 2001). Secondly, in light of a commonly metapopulation structure and the critical role of processes such as emigration and colonisation in amphibian populations, a regional monitoring perspective is important in order to distinguish overall regional declines from local (and essentially stochastic) extinctions (MARSH & TRENHAM, 2000, MARSH, 2001). For most amphibians this requires an appreciation as to the importance of the spatial arrangement, and degree of isolation between different breeding sites. It is appropriate to note here that according to a recent review of techniques used to quantify amphibian populations, most attempts focus on direct or indirect (e.g. vocal calls, egg masses) counts at breeding sites (ALFORD & RICHARDS, 1999)- utilising the fact that most species congregate en masse to breed (BEEBEE, 1996). It thought that the population dynamics of a species are determined primarily by recruitment processes occurring at breeding ponds, and that such a focus can accurately determine the cause of any local or regional decline (MARSH & TRENHAM, 2000). However, such an exclusive focus carries a number of caveats in data interpretation; (1) Variation in population size at breeding ponds can as well be due to variation in breeding behaviour as to actual variation in population number (PECHMANN et al., 1991), (2) It is often impossible to clearly distinguish variation in population size from simply variation in the size of breeding aggregations - ie the degree of 'openness' of the population (MCARDLE & GASTON, 1993). Both such measures represent useful information but it is important to note that they are not synonymous. (3) In terms of adult survival and distribution an exclusive focus on breeding sites ignores the potential importance of the intervening terrestrial habitat which may be of ecological significance (MARSH & TRENHAM, 2000). A final note with respect to natural variability in amphibian populations is interspecific or taxonomic variance. As emphasised earlier different species exhibit different levels of susceptibility to different agents of decline (e.g. DROST & FELLERS, 1996). Accordingly, any attempt to assess the stability or vulnerability of an amphibian fauna at any one regional site should consider not only the breadth of species present, but also a number of different populations of each. With relevance to all levels of variability that can serve to confound attempts to identify declining populations, a high number of intra-annual repeat visits to each monitoring site (especially during the peak breeding season) can add important, if not essential strength to the results (ALFORD & RICHARDS, 1999).
Alongside data on the population dynamics of the focal species, it is important to gain an appreciation of the differential importance of key ecological variables in both the biotic and abiotic environment (e.g. climate, water quality, floral composition, and predator abundance). Such a multidimensional approach to monitoring helps to identify any potential agents of decline, but also to helps predict any secondary effects or feedbacks following a potential change in the structure of the amphibian community. This information, when integrated into demographic data of the amphibian populations can be invaluable in building a null model to predict the range and patterns of population behaviour in the absence of external pressures or agents of decline (ALFORD & RICHARDS, 1999). Such null models can be used to reduce the subjectivity and ambiguity that often surrounds the evidence for a population decline (POUNDS et al., 1997).
A final note of caution in studying the declining amphibian phenomenon needs
to be emphasised. Although both scientific (WAKE, 1998), and public (MORELL,
2001) opinion recognises the severity of declining amphibian populations, it
is important to maintain a broad appreciation of other conservation problems
and priorities (HALLIDAY, 2001). Two points should be considered at this junction.
Firstly, amphibian declines have occurred in a number of pristine habitats and
protected areas - removed from areas of direct human impact. This evidence bears
serious implications for the effectiveness of the protected areas approach to
conservation, and as such the study of amphibian populations should be integrated
wherever possible into the wider context of conservation science and action.
Secondly, it is important in world where resource allocation to conservation
biology is seriously inadequate, that an focus or even over-emphasis on amphibians
does not eclipse the equally worrying status of many other taxanomic groups
(e.g. GIBBONS et al., 2000, GROOMBRIDGE & JENKINS, 2000) from both the scientific
and public eye.