Impact of disease on devil survival rates
The arrival of DFTD at the Freycinet peninsula in June 2001 triggered an immediate and steady decline in apparent survival rates of adults and subadults in the population, which was predicted well by the increase in disease prevalence in the population over time. DFTD had the most evident impact on the average apparent survival rates of older individuals. Apparent survival of adults marked-as-adults halved in the year following the detection of DFTD and continued to decline effectively to zero as DFTD prevalence in the population increased. The variable apparent survival rates of younger adults likely reflects spatial variation in disease impact as it spread southwards down the peninsula, with the survival rate of individuals in some locations remaining unaffected until the disease progressed through the area. The change in survival rates through time, differed with age and marking group in a manner consistent with the observation that DFTD affected older animals first, but inconsistent with the possible action of environmental variables operating similarly across all survival rates.
The finding from the multistate recapture models that DFTD negatively impacted on the apparent survival of subadults in this study was unexpected. Only five diseased subadults have ever been captured in this population, all of which were caught in the final sampling period (July 2006) and all of which were excluded from the analyses presented here. In fact, it was thought that either there was a lengthy latency period associated with DFTD or that subadults were somehow excluded from DFTD infection and thus, that the continued survival of this group might ensure population persistence or act as a buffer to population extinction (Hawkins et al. 2006). Although the negative impact of DFTD on this group to date has not been as great as that observed for adults, the relationship between increasing disease prevalence and decreasing subadult survival rates detected in this study has diminished these possibilities.
We found that older individuals were the first in the population to be infected and succumb to DFTD; younger adults followed; and subadults only became infected once the majority of adults had died. This was evidenced by the decline in the age of infected individuals following the detection of DFTD into the population and also by the different pattern of infection rates (transition rates) of adults and subadults in the years following DFTD detection. In view of this, the time frame of this study may have been insufficient to detect the full effect of DFTD in the population, particularly its impact on subadult survival.
The observation that DFTD is first detected in older individuals is an interesting one. It may simply be related to a lower force of infection due to low prevalence when the disease first enters the population. The average age of infection by a pathogen is inversely related to the force of infection (Grenfell & Anderson 1985). Alternatively, older individuals may be behaviourally more susceptible to acquiring disease, due to increased exposure to opportunities for transmission (Altizer et al. 2003). DFTD is believed to be transmitted via biting during feeding or sexual interactions (Pearse & Swift 2006). Thus, the pattern observed here may indicate that older individuals engage in aggressive or sexual encounters more frequently and predominantly with other adults. Older animals may also be more physiologically susceptible to disease due to higher levels of stress-related hormones associated with reproductive activity (Bronson 1989). Males of other dasyurid species are well-known to suffer stress-related immunological dysfunction following breeding (Boonstra 2005). However, we detected no evidence that males were particularly prone to DFTD infection or suffered higher mortality than females.
One important limitation of the mark–recapture methods employed in this study is that estimates of apparent survival do not differentiate between death and permanent emigration. If DFTD has disrupted movement patterns and resulted in increased emigration rates from diseased areas, then our survival rates will be biased low (as we are estimating the combined effects of both processes on survival). Biologically plausible explanations, relating to local population density, resource availability and available mating opportunities exist that could explain either increased or decreased dispersal in diseased areas (Dieckmann, O’Hara & Weisser 1999). Little is known about dispersal patterns or the mechanisms driving dispersal decisions in Tasmanian devils. Although they are likely to follow the normal mammalian (and dasyurid) pattern of predominantly male-biased natal dispersal of newly independent juveniles, an indication of such a pattern in terms of strongly sex-biased recapture rates was not detected in this study.
Multistate analyses allow robust estimates of apparent survival for individuals of different disease states. Apparent survival of diseased devils was effectively zero, because no diseased devils survived the year-long interval between sampling periods, confirming that DFTD is an aggressive, consistently fatal disease. From the regular trapping trips conducted at this study site the longest period between recaptures for a diseased animal was 6 months, while the longest known period between recaptures for any diseased animal from all state-wide disease monitoring efforts is 9 months (Hawkins et al. 2006).
As progression to death is so rapid, many of the individuals in the population may have acquired disease and died without this transition ever appearing as ‘diseased’ in our data set. In addition, those individuals that were captured with DFTD during trapping periods other than the winter period analysed in this study (n = 36 in this case) could also not be included in this analysis. These missed infections partly explain the somewhat counter-intuitive result that apparent survival of healthy individuals declined with disease prevalence. An additional possibility is that some asymptomatic devils were misclassified as healthy individuals but were in fact diseased, thereby depressing the apparent survival estimates of the healthy group. Although DFTD tumours are distinctive and thorough examinations for symptomatic signs were undertaken, it is possible that early stage lesions were misdiagnosed. Also, the latent period is unknown but could be as long as 12 months. Some devils were inevitably infected at the time they were trapped but had not yet developed visible tumours. The inevitable outcome of these missed and possibly misdiagnosed infections would be that the negative impacts of disease on apparent survival and infection rates were underestimated in this study.
No differences in the impact of DFTD on males and females were detected. Males and females appear to be equally susceptible to DFTD infection indicating that transmission of DFTD within and between the sexes occurs with equal probability. This study also did not detect any difference in apparent survival rates of males and females. This is in contrast to the only other study of mortality rates in Tasmanian devils, which reported higher mortality rates for males than females (Pemberton 1990). The estimates from that study, however, were obtained using cohort-based approaches with smaller sample sizes than recommended for those methods (Caughley 1977).
Impact of disease prevalence on transition rates
The clear positive relationship found in this study between increasing disease prevalence and the transition rate to the infected class is expected from an infectious disease. Estimating the force of infection of a disease, which is the rate of acquisition of the infection for a susceptible host (Heisey, Joly & Messier 2006), is crucial to understanding infectious disease dynamics. Transition rates obtained from multistate models are a compound measure of the probability of becoming infected (showing symptoms) and surviving to be captured and may be regarded as a conservative indicator of the force of infection in the population. Since the detection of DFTD in this population, the total infection rate (all transitions of healthy individuals to disease states) has followed a steadily increasing trend over time. Moreover, as discussed above, true infection rates are probably considerably higher than is indicated by the estimates obtained here. These observations provide strong evidence that the force of DFTD infection in this population is increasing and that the current epidemic is not subsiding. Particularly disturbing is the rapid and ongoing increase in the infection rate of subadults. A continuation in the trend observed (Fig. 4) will soon see subadults become more likely to develop into diseased adults than healthy adults. This scenario could constitute a major change to this species’ life history (typically multiple breeding in a 5–6-year life span) as diseased devils may not survive long enough to rear a litter successfully (devils become independent at 9 months of age).
Impact of disease on population growth rate
The decrease in estimated survival rates has resulted in a marked ongoing decline in the population growth rate. The population growth rate declined immediately following disease arrival despite disease prevalence being extremely low at this time (only 3·5%) and continued to decline rapidly, reaching a 30% decline after just 3 years. This is extremely relevant for the management of disease-free populations, highlighting the need to implement strategies swiftly following disease detection to prevent population decline. Of particular concern is the decline of almost 50% per year in the adult segment of the population.
Epidemiological theory suggests that a single-host pathogen is unlikely to drive its host to extinction, if disease transmission follows a density-dependent process, because disease maintenance and spread will not be possible when populations are sufficiently reduced (below a population threshold) (McCallum & Dobson 1995; de Castro & Bolker 2005). Indeed, a recent population projection model for devils showed that devils would be unlikely to be driven to extinction by a density-dependent disease (Bradshaw & Brook 2005).
If disease transmission is not density-dependent, however, thresholds for disease maintenance will not occur and population extinction is possible (de Castro & Bolker 2005). Although more field data are required to determine the transmission dynamics of DFTD, data from Mt William National Park in the far north-east of the state, suggest that any threshold population density for DFTD persistence is very low. At this site, DFTD prevalence remains high (33%) despite a reduction in population size from 7·14 individuals per km2 to just 0·18 individuals per km2 (unpubl. data). Whether the negative impacts of DFTD on survival detected in this study can be compensated for remains to be investigated. Our results suggest, however, that in the absence of population thresholds for disease maintenance and strong mechanisms of demographic compensation, local population extinction seems likely.