Decline in the prevalence of chytridiomycosis in frog populations in North Queensland, Australia K. R. McDONALD I, D. MENDEZ', R. MULLER', A. B. FREEMAN 1 and R. SPEARE','
In the early 19905 stream-associated amphibian populations in tropical upland North Queensland experienced severe declines resulting in extinction of three species, local elimination of four species, marked reductions in one species and apparently no declines in other species, Chytridiomycosis, a disease due to the amphibian chytrid fungus, Batrachochylrium dendrobatidis, was the likely cause of this epidemic, We conducted a monitoring study for chytridiomycosis in four species of frogs in North Queensland from October 1998 to October 2002 by collecting specimens in the field and using histology of removed digits to diagnose chytridiomycosis, Chytridiomycosis was diagnosed in 112 (7,1%) of the 1 578 specimens and prevalence was significantly associated with season and altitude, with higher prevalences in winter and above 300 metres altitude, A multivariate model adjusting for potential confounding effects arising from the sampling process demonstrated a significant decline in the time trend of prevalence of chytridiomycosis, The study supports the hypothesis that B, dendrobatidis becomes endemic after the initial epidemic wave, Since the surviving species of stream-associated frog, Utoria genimaculata, has increased to pre-decline numbers, the decline in prevalence of chytridiomycosis is evidence of a changed pathogen-host relationship. The reasons for this change are speculative but could be due to an increase in innate host resistance in response to selection pressure by B. dendrobatidis or to lower rainfall associated with an EI Nino effect. These findings justify management strategies that assist susceptible amphibian species to survive an initial epidemic wave of chytridiomycosis.
Key words: Batrachochytrium dendrobatidis, Amphibian chytrid, Chytridiomycosis, Amphibian declines, Utoria, Nyctimystes, Taudaclylus, Emerging infectious diseases,
INTRODUCTION
AN
epidemic wave of declines of frog populations occurred between 1978 and 1993 in coastal Queensland. Australia, and spread from south to north at roughly 100 km per year (Laurance et al. 1996; Laurance et al. 1997). The starL of the epideluic in Queensland was in the Conondale Rang'es, south-east Queensland with the disappearance of two species, the Southern Gastric Brooding Frog Rheabatrachus silus and the Southern Day Frog Taudactylus diumus, and the last detected decline was at Big 'L.blcland in north Queensland, 1 500 km north, with the extinction of the Sharp Snouted Day Frog Taudactylus acutirostris and local extinction of three other species at the site (Nycti",ystes dayi, Lit01'ia nannatis, L. rheocola) (Hines et al. 1999; McDonald and Alford 1999). In central Queensland the Northern Gastric Brooding Frog R. vitellinus also became extinct (McDonald 1990; McDonald and Alford 1999; McDonald 2002). In north Queensland declines were most obvious above 300 metres altitude, while in central and southeastern Queensland declines occurred above 150 metres and above sea level respectively(McDonald 1990; Richards et al. 1993; McDonald and AlfDl'd 1999). Apart from the four frog species that became extinct, the numbers of many other species in the same area of upland coastal Qlleensland declined drastically (McDonald and Alford 1999; McDonald 2002). Two species, the Armoured Mistfrog L. larica and the Mountain Mistfrog L.
nyalwlensis have not been located since 1990 and are also presumed extinct (Richards et al. 1993; McDonald 2002). Of the stream-dwelling frogs surviving in upland far North Queensland the Green-Eyed Tree Frog L. genimaculata remained at low density for 5 years after the declines of the early 1990s and then increased to its current level which is similar to the pre-decline population (McDonald 2002; McDonald, unpub!.). Two other species, the Waterfall Frog L. nannatis and the Common Mistfrog L. "heacola disappeared above 400 metres altitude during the epidemic declines in the early 1990s, while populations persisted below this level (McDonald and Alford 1999). The upper altitude of these species has increased over the intervening decade, and specimens have recently been seen up to 650 metres, but not uniformly (McDonald 2002; Retallick, pel's. comm. 2002; McDonald, unpubl.). Other species near the stream environments, Mixophyes spp. and L. lesueurii, appear not to have had population declines.
Chytridiomycosis, a disease caused by the amphibian chytrid fungus, IJatmchachytriu1II dendrabatidis, was identified as the cause of these declines (Berger et al. 1998). The disease also caused epidemics in wild amphibians in Panama, EcuadOl~ Venezuela, USA, Spain and New Zealand (Berger et al. 1998; Lips 1999; Ron and Merino 2000; Bosch et al. 200 I; Fellers et al. 2001; Waldman et al. 2001; Bradley et al. 2002; Green et al. 2002; Bonaccorso et al. 2003), and once it had appeared in an area,
'Environmcnt Protcction Agency, Qucensland Parks and Wildlife Selvice, Atherton, Quecnsl~nd, Australia 4883. 'Amphibian Diseases Group, School of Public Ilcahh and Tropical Medicine, .I amcs Cook University, Townsville, Queensland, Australia 4811, 'COrres]londing author: Associate I'rofessor Rick Speare, School of Public Health and Ti'opicat Mcdicine, James Cook Univcrsity, 'Ibwllsville, Qlleens[and, 4811. Email: Richard.s]
[email protected] PACIFIC CONSERVATION BIOLOGY Vol. ll: 114-20. Surrcy Bcatty & Som, Sydney. 200.5,
, Amtr~ha
McDONALD ETAL.: CHYTRIDIOMYCOSIS IN NORTH QUEENSLAND
chytridiomycosis was detected thereafter. Batrachochytriurn dend1'Obatidis is a chytrid fungus and appears to be capable of existing in the environment independently of amphibian hosts (Longcore et al. 1999; Johnson and Speare 2003). Since B. dend1'Obatidis can exist without susceptible species by being either in the environment or in reservoir species, it has the potential to drive the susceptible species to extinction. The question whether amphibians that survive the initial epidemic wave can adapt to B. dendrobatidis is an important onc, as an increase in innate host resistance may be the critical strategy for survival of the species. The relationship between pathogen and host is a dynamic one and adaptations can occur in pathogen virulence and host resistance (Fenner 1998). Selection for a more resistant host and lower pathogen virulence was shown for rabbits and myxoma virus (Fennel' 1998; Zuniga 2002). Innate resistance in a host can be tested in the laboratory by infection experiments using standard conditions and inoculation doses. In the wild, innate resistance can only be indirectly measured using prevalence. Although environmental factors appeal' to play a role in the impact of chytridiomycosis on amphibians, only temperature has been identified as a significant factor (Berger 200 I). This paper monitors prevalence in amphibians over a period roughly 5 to 9 years after the initial epidemic decline and provides evidence that B. dendrobatidis has become endemic and that the relationship between amphibian host and fungal pathogen has shifted in favour of the host.
METHODS Monitoring and surveys of rainforest dwelling frogs has been carried out in the Wet Tropics of Queensland since 1992 across a range of altitudes and latitudes where there has been historical records of declining species (Richards et al. 1993; McDonald and Alford 1999; McDonald 2002). Prior to this study monitoring and survey visits to sites had been at irregular intervals and frequency. Three major sites (Big Tableland (l5.7'S, 145.3'E elevation 600 m), Topaz (17.4'S, 145.7'E elevation 640 m) and Tully (l7.7'S, 145.6'E elevation 100 m») and 34 minor sites were visited at irregular intervals during the year (4-6 weeks for the major sites) from October 1998 to October 2002 provided access was possible. Frogs (L. genirnaculata, L. nannotis, L. Theocola and Nyctimystes dayi) were caught on transects, toe tips removed for numbering of frogs, and these examined for chytridiomycosis (see below). Frogs were numbered by toe clipping (Ferner 1979) when first caught only; hence examination of status for chytridiomycosis in recaptures was not possible. Sample sizes varied for each site and frog
115
species because of the low population numbers or absence of frogs in upland areas, difficulty of locating frogs in upland sites in winter, markrecapture populations having a high percentage of recaptures and hence no samples available for histological examination, and sites being more fi'equently visited in spring and summer as frogs were easier to locate especially species episodically present such as the Lace-Eyed Tree Frog N. dayi. The Queensland Wet Tropics is under a monsoonal influence with 75% of annual rainfall between November and April. For this papel; Uwinter" is defined as May to September (cooler and dryer months) and usummer" as October to April (warmer and wetter months). uEndemic" is used as an epidemiological term used to describe levels of infection due to a microorganism that is established in a defined place and host population. DIAGNOSIS OF CHYTRIDIOMYCOSIS Histology of toe tips removed by toe clipping was used to diagnose chytridiomycosis (Berger et al. 1999). Clipped toes were instantly fixed upon collection in 90% ethanol, before being decalcified for 48 to 36 hours, at room temperature, in a 10% formic acid solution. Decalcification is a necessary step to soften the bone tissues to allow routine sectioning of the specimen. After a thorough rinsing under running tap watel~ toes were histologically processed, embedded in paraffin, sectioned at 5 ~m, and stained with routine haematoxylin and eosin technique (Culling 1963). Since the chytrid fungus is mainly found in the stratum corneUlll, and to a lesser extent in the outermost layers of the stratum granulosum of the epidermis (Berger et ai. 1999), it was of m£tior importance to ensure that the longest length possible of skin was examined. Therefore, toes were embedded horizontally with either their ventral or lateral surface downwards. For each specimen four representative serial sections from the middle of each sample were examined under a compound microscope at X 200 and X 400 magnifications by the same observer (DM). The number of toes (from I to 5) available for histological examination varied from animal to animal depending on the collection number assigned to the frog in the field. To ensure that specimens were tested "blind", with DM having no knowledge of host details, all samples were recoded before any processing took place. Results were initially reported in foul' categories defIned as: "negative" [intact epidermis. no zoosporangia seen], "suspicious negative" [intact epidermis, a single suspicious stlucture which could not obviously be declared to be a zoosporangium], ususpicious positive" [mostly intact epidermis, one or more areas of sloughing
116
PACIFIC CONSERVATION BIOLOGY
keratin and/or hyperkeratosis, one or more suspicious structures which could not be
convincingly diagnosed as zoosporangia], and "positive" [damaged epidermis, and unequivocal zoosporangia). A conservative approach was adopted to the classification with only results with classical appearances being placed in the "positive>! category. For statistical analysis results
were dichotomized to 1) Negatives ("negatives" and "suspicious negatives))), and 2) Positives ("suspicious positives" and "positives"), Histological diagnosis depends on pattern recognition and matching and can be complicated by various factors, In our experience, most zoosporangia are usually found grouped into clusters, so in the case of a mild infection, unless the sections taken coincide with affected areas, animals with chytridiomycosis may have this diagnosis missed. Furthermore, the uneven spread of the infection on the affected amphibian body in light infections (Berger et al. 1999; Pessier et at. 1999) may result in the zoosporangia being present on only one or two of the digits. A positive result relates to the whole animal and can be determined by the presence of a single zoosporangium on the slide (one toe tip suffices). An histologically negative result cannot extend to the whole animal, and is only an inference of the absence of zoosporangia in the sections of sample material examined. Therefore, the more negative toes that are examined, the greater the likelihood that the status of the whole animal is negative. Regardless of the number of toes examined, an histologically negative result will always remain only an indication of the negative status of the whole animal. These factors explain the high specificity and lower sensitivity of
histological diagnosis (Berger 2001), and mean that prevalence is liable to be underestimated. STATISTICAL METHODS Associations of categorical variables and prevalence of chytridiomycosis were assessed by means of exact versions of chi-square tests. For nominal variables or exact trend tests for ordinal items. Log-linear modelling was lIsed to check for higher dimensional correlations. The prevalence trend over the years was adjusted for potential confounders by logistic regression modelling. In the modelling process, "dummy" coding was used for all categorical variables, i.e., all odds ratios are expressed relative to the first category (of each variable) which therefore constitutes the "baseline". For all analyses, a p-value of below 0.05 was regarded as statistically significant. RESULTS
The four major species studied had similar prevalences of chytridiomycosis in asymptomatic frogs with average prevalences for the species between 5.3% and 7.9% (Table 1). Chytridiomycosis was diagnosed in 112 (7.1 %) of the 1 578 specimens collected between 1998 and 2002. The prevalence of chytridiomycosis was found to be significantly associated with the season and the altitude, with higher prevalences in winter and at altitudes over 300 metres (Table 1). No significant time trend over the years could be found in this bivariate analysis. One significant three-way interaction betvveen the season, the altitude of collection and the
Table 1. Bivariate Associations between Chitrid Prevalence and EnvironmcntalfTemporal Variables (n Variable
CategOl),
Chitrid Prevalence in % (number pos. out of n)
=
I 578).
p - value Exact Chi 2 test 0.65
Site
Thlly Big lableland 'Ibpaz Othcr
6.5% 8.0% 9.4% 6.9%
(31/477) (14/174) (13/139) (54/788) 0.62
Species
L. llannotis L. genimaculala L. rheocola
N. dayi Other
6.4% (16/250) 7.8% (50/648) 7.9% (30/380) 5.3% (14/262) 4.7% (2/43)
Season
