SEARCH

SEARCH BY CITATION

Keywords:

  • apex predators;
  • devil facial tumor disease;
  • mesopredator release hypothesis;
  • spotlighting;
  • Tasmania;
  • depredadores apicales;
  • enfermedad de tumor facial;
  • hipótesis de liberación de mesodepredador;
  • lampareo;
  • Tasmania

Abstract

  1. Top of page
  2. AbstractResumen
  3. Introduction
  4. Methods
  5. Results
  6. Discussion
  7. Acknowledgments
  8. Supporting Information
  9. Literature Cited
  10. Supporting Information

As apex predators disappear worldwide, there is escalating evidence of their importance in maintaining the integrity and diversity of the ecosystems they inhabit. The largest extant marsupial carnivore, the Tasmanian devil (Sarcophilus harrisii) is threatened with extinction from a transmissible cancer, devil facial tumor disease (DFTD). The disease, first observed in 1996, has led to apparent population declines in excess of 95% in some areas and has spread to more than 80% of their range. We analyzed a long-term Tasmania-wide data set derived from wildlife spotlighting surveys to assess the effects of DFTD-induced devil decline on populations of other mammals and to examine the relative strength of top–down and bottom–up control of mesopredators between 2 regions with different environmental conditions. Collection of the data began >10 years before DFTD was first observed. A decrease in devil populations was immediate across diseased regions following DFTD arrival, and there has been no indication of population recovery. Feral cats (Felis catus) increased in areas where the disease was present the longest, and feral cat occurrence was significantly and negatively associated with devils. The smallest mesopredator, the eastern quoll (Dasyurus viverrinus), declined rapidly following DFTD arrival. This result suggests the species was indirectly protected by devils through the suppression of larger predators. Rainfall deficiency was also a significant predictor of their decline. Environmental variables determined the relative importance of top–down control in the population regulation of mesopredators. In landscapes of low rainfall and relatively higher proportions of agriculture and human settlement, top–down forces were dampened and bottom–up forces had the most effect on mesopredators. For herbivore prey species, there was evidence of population differences after DFTD arrival, but undetected environmental factors had greater effects. The unique opportunity to assess population changes over extensive temporal and spatial scales following apex predator loss further demonstrated their role in structuring ecosystems and of productivity in determining the strength of top–down control.

Cascadas Tróficas Después de la Declinación Inducida por Enfermedad de un Depredador Apical, el Demonio de Tasmania

Resumen

A medida que desaparecen depredadores apicales globalmente, cada vez hay mayor evidencia de su importancia en el mantenimiento de la integridad y diversidad de los ecosistemas que habitan. El mayor marsupial carnívoro existente, el demonio de Tasmania (Sarcophilus harrisii) esta amenazado de extinción por un cáncer transmisible, la enfermedad de tumor facial (ETF). La enfermedad, observada por primera vez en 1996 ha conducido a declinaciones poblacionales aparentes que superan el 95% en algunas áreas y se ha extendido a más de 80% de su área de distribución. Analizamos datos de largo plazo, y de toda Tasmania, derivados de muestreos con lámpara para evaluar los efectos de la declinación de demonios inducida por ETF sobre poblaciones de otros mamíferos y para examinar la robustez relativa del control arriba–abajo y abajo–arriba de mesodepredadores en 2 regiones con diferentes condiciones ambientales. La recolección de datos comenzó >10 años antes de que se observará ETF por primera vez. Un decremento en las poblaciones de demonios fue inmediato después del arribo de ETF en las regiones infestadas, y no hay indicios de la recuperación de la población. Los gatos ferales (Felis catus) incrementaron en áreas donde la enfermedad permaneció por más tiempo, y la ocurrencia de gatos ferales estuvo significativa y negativamente asociada con los demonios. El mesodepredador más pequeño, Dasyurus viverrinus,declinó rápidamente después del arribo de ETF. Este resultado sugiere que la especie fue protegida indirectamente por demonios mediante la supresión de depredadores mayores. La deficiencia de precipitación también fue un predictor significativo de su declinación. Las variables ambientales determinaron la importancia relativa del control arriba–abajo en la regulación de poblaciones de mesodepredadores. En paisajes con baja precipitación y proporciones relativamente altas de agricultura y asentamientos humanos, los actoress arriba–abajo fueron amortiguadas y los factores abajo–arriba tuvieron el mayor efecto sobre los mesodepredadores. Para especies herbívoras presa, hubo evidencia de diferencias poblacionales después del arribo de ETF, pero factores ambientales no detectados tuvieron efectos mayores. La oportunidad única de evaluar cambios poblacionales en escalas temporales y espaciales extensivas después de la pérdida de depredadores ápice demostró su papel en la estructuración de ecosistemas y de la productividad en la determinación de la robustez del control arriba–abajo.


Introduction

  1. Top of page
  2. AbstractResumen
  3. Introduction
  4. Methods
  5. Results
  6. Discussion
  7. Acknowledgments
  8. Supporting Information
  9. Literature Cited
  10. Supporting Information

Apex predators play a substantial role in maintaining the integrity of ecological systems (Terborgh et al. 1999; Ritchie & Johnson 2009) but are disappearing at unprecedented rates from geographically and taxonomically diverse ecosystems (Pace et al. 1999; Dalerum et al. 2009). The loss of an ecologically functional apex predator can result in wide-ranging effects and has led to the homogenization and simplification of food webs globally (Estes et al. 2011). Effects range from the transformation of predator guilds (e.g., Berger & Gese 2007), species extinctions in diverse taxa, including avian fauna (Crooks & Soulé 1999) and ground-dwelling marsupials (Johnson et al. 2007), and the modification of plant communities (e.g., Beschta & Ripple 2009).

Ecosystem productivity affects top–down and bottom–up population regulation of species (Elmhagen et al. 2010), although the extent to which these 2 forces affect biological diversity in an ecosystem remains contentious (Terborgh et al. 1999; Brashares et al. 2010; Estes et al. 2011). Elmhagen et al. (2010) describe interference ecosystems, a concept that brings together 2 hypotheses regarding the relative strengths of top–down and bottom–up control in determining ecosystem structure. The mesopredator release hypothesis (MRH) explains the increase in smaller predator populations when they are released from competition from larger predators, which in turn can affect their prey species (Soule et al. 1988), and the exploitation ecosystems hypothesis (EEH) (Oksanen et al. 1981) considers population regulation a consequence of top–down consumption, position of the organism in the trophic structure, and bottom–up productivity (Oksanen 1990). Evidence for diminished mesopredator release following top predator loss has been demonstrated in environments with relatively low productivity (Elmhagen & Rushton 2007). The complexity of interactions among ecosystem productivity, number of trophic levels, and interference competition in predator guilds and the relative strengths of top–down and bottom–up control in shaping ecosystem structure highlight the importance of considering the effects of both top–down and bottom–up forces in studies of trophic cascades.

A natural experiment, involving the progressive disease-induced decline of the apex mammalian predator across the island state of Tasmania, Australia, provides a unique opportunity to assess the relative strengths of top–down (MRH) and bottom–up control on populations of mesopredators and prey. The Tasmanian devil (Sarcophilius harrisii) (males approximately 8 kg, females approximately 6 kg), the world's largest extant marsupial carnivore and now the apex predator following the extinction of the thylacine (Thylacinus cynocephalus), is threatened with extinction from a fatal transmissible cancer, devil facial tumor disease (DFTD). This disease has spread progressively from east to west, is present across most of the devil's current range, and has caused population declines of >95% in areas where the disease has been present the longest. There is no indication of population recovery in any areas where the disease is present (Hawkins et al. 2006; McCallum et al. 2007, 2009).

In contrast to mainland Australia, Tasmania retains a nearly intact mammalian fauna. A complex multilevel carnivore guild has 3 functional levels of predators: the devil, the similar-sized native marsupial spotted-tailed quoll (Dasyurus maculatus) (males approximately 4 kg, females approximately 2 kg) and introduced feral cat (Felis catus) (males approximately 4.5 kg, females approximately 3kg), and the smaller eastern quoll (Dasyurus viverrinus) (males approximately 1 kg, females approximately 0.8 kg) (Jones 1997). There is evidence of both current (diet overlap) and evolutionary-scale (competitive character displacement) competition among the 3 marsupial carnivores (Jones 1997; Jones & Barmuta 1998). Spotted-tailed quolls are expected to experience the most extensive competition due to dietary overlap with both the larger devil and the smaller eastern quoll (Jones & Barmuta 1998). Feral cats may also compete with spotted-tailed quolls due to their similar size and prey range, which comprises small- and medium-sized mammals (Dickman 1996). These larger mesopredators may subject the eastern quoll to intense competition and predation (Jones et al. 2004) (Supporting Information). The Tasmanian pademelon (Thylogale billardierii) and the common brushtail possum (Trichosurus vulpecula) are key prey species of the devil (Jones & Barmuta 1998), although there is no mortality data that can be used to infer the effects of devils on these species. The variable nature of Tasmania's environment, including rainfall gradients, which affects vegetation communities and human land use and occupation, may affect ecosystem productivity and top–down and bottom–up regulation of species.

We considered the effects on mammalian fauna of the progressive DFTD-associated decline of the Tasmanian devil in the context of the variable Tasmanian ecosystem. We used a long-term, broadscale data set of spotlight counts of wildlife to assess whether the removal of top–down effects from devils results in the competitive release of mesopredators (MRH) and subsequent effects on mesopredator and devil prey species. We also considered whether environmental factors affect the strength of top–down effects for regulating mesopredator populations.

Methods

  1. Top of page
  2. AbstractResumen
  3. Introduction
  4. Methods
  5. Results
  6. Discussion
  7. Acknowledgments
  8. Supporting Information
  9. Literature Cited
  10. Supporting Information

Data Collection

Annual statewide spotlighting surveys were initiated in Tasmania in 1975 (Fig. 1) to assess the effects of culling for meat, fur, and crop protection on browsing herbivores, the Tasmanian pademelon (males approximately 8 kg, females approximately 4 kg), Bennett's wallaby (Macropus rufogriseus) (males approximately 15 kg, females approximately 11 kg), and common brushtail possum (approximately 3 kg). All other mammals sighted incidentally were also recorded (Driessen & Hocking 1992). In 1985, the collection methods were standardized and spatial extent of the survey effort increased (Southwell & Fletcher 1993); thus, we used data collected between 1985 through 2008. The 170 individual spotlighting transects were grouped into 29 districts by proximity, and there were 3–8 transects in each district (Hocking & Driessen 1992). All transects within a district were surveyed on the same night beginning 30–40 min after sunset. Each 10-km-long transect was driven at a constant speed of 25 km/h (Hocking & Driessen 1992). Districts were generally surveyed within 2 months beginning in mid-November. Surveys were not conducted in heavy rain, strong winds, or fog. The data are raw counts and have high levels of internal variability, partly due to observer bias and differences in vehicles and weather conditions (Driessen & Hocking 1992). Nonetheless, a comprehensive data set such as this, in which mammals have been monitored over extensive spatial and temporal scales, is rare and extremely valuable in ecological studies. The data set predates the emergence of DFTD by over a decade.

image

Figure 1. Map of Tasmania showing individual spotlight survey transects within each of 4 regions representing different year ranges of arrival of devil facial tumor disease (DFTD) (late, mid, disease-free, early). Years are range of arrival of DFTD in each region.

Download figure to PowerPoint

Data Organization

Dates of DFTD arrival for each transect were obtained from the date of first recorded DFTD in the area or extrapolated from confirmed records of DFTD-infected devils in surrounding regions and expected patterns of disease spread (McCallum et al. 2007). In the absence of systematic disease monitoring, disease may have predated these estimates.

To assess temporal changes in mammal populations, we combined individual transects into regions on the basis of 4-year periods of DFTD arrival. Analysis at the transect and spotlight-district level was impractical. For some species, particularly carnivores, which occur at lower densities, data were too sparse across individual transects to detect population changes. Data aggregated by spotlighting district had inherent problems because all transects within a district were surveyed once a year on the same night; thus, differences among districts could have been confounded by the effect of environmental conditions, such as moonlight and rainfall. Aggregating the data into larger regions reduced some of this variation; it allowed us to average the environmental variables across multiple districts and nights and thus to detect long-term trends in populations. We named these combined regions by the time of DFTD arrival (early, mid, late, or disease-free) (Fig. 1). Due to the consistent pattern of devil population decline following DFTD arrival, these DFTD arrival regions reflect local devil populations and time since suppression of apex predators began.

Because the DFTD arrival regions covered a broad area, ecological and environmental differences among the early and mid regions differed substantially. The mid region was drier (77% of transects had <900 mm of rainfall) than the early region (38% of transects had <900 mm of rainfall) and had more farmland (14% more open vegetation per transect on average) and a greater intensity of human settlement (almost double the number of total property addresses as a proxy) than the early region. These differences allowed us to conduct a second analysis in which we assessed the strength of top–down and bottom–up mechanisms in determining mesopredator populations in different environments. In this analysis, we used the species data for each individual survey of a transect and transect environmental variables to examine the factors associated with the localized occurrence of mesopredators within the early and mid regions. We accounted for pseudoreplication in transects surveyed each year by assigning transect as a random effect in our models.

We obtained both gridded statewide annual rainfall totals for 1984–2009 and the standard mean annual rainfall (1961–1990) from the Australian Bureau of Meteorology. Using the midpoint of each transect and a cubic convolution in ArcGIS (ESRI, Redlands, California), we determined the annual rainfall value for each spotlight transect. This procedure interpolates a weighted average from the values of the nearest 16-grid cells. We used these values to calculate the deviation of a given year from the mean annual rainfall for each transect. We used mean annual rainfall for all transects within the region each year to calculate rainfall deviation for each DFTD arrival region as a whole. We also calculated a 1-year deviation lag in rainfall because for some species the effects of rainfall variation on population vital rates, such as juvenile survivorship, may not be evident immediately.

Data for spotted-tail quolls were too sparse for analyses, likely due to their cryptic nature and low densities. The southern brown bandicoot (Isoodon obseulus) (females approximately 620 g; males approximately 890 g), eastern barred bandicoot (Perameles gunnii) (990 g), and the long-nosed potoroo (Potorous tridactylis) (females approximately 1020 g; males approximately 1180 g) (van Dyck & Strahan 2008) were also detected infrequently. Because they are all native marsupials of a similar size and within the prey range of larger mesopredators, we combined them into a category of medium mammals.

Statistical Analyses

To describe the effects of DFTD arrival and rainfall deviation on species populations, we fitted generalized additive models (GAMs) with a log-link function to data for individual species with R (R Development Core Team 2011 ) (packages mgcv and ggplot2). We used GAMs as a tool to visualize changes in counts of species over time because GAMs do not assume any specific functional form a priori between the predictor variables and the link function and fit a smooth curve through the data (Zuur et al. 2009). We then fitted generalized linear models (GLMs) to investigate whether observed changes were associated with time before and after DFTD arrival. We used a Poisson error structure with a log-link function or a negative binomial error if there was evidence of overdispersion. We generated a dummy variable for DFTD absence (0) and presence (1). This variable was included in the model as a DFTD by year interaction without a main effect. Year was centered to zero at the time of DFTD arrival to ensure the resulting model did not have a discontinuity at the time of DFTD arrival. Changing this variable also allowed for lagged relations between changes in populations and DFTD arrival. We also incorporated annual rainfall deviation into the models as an explanatory variable. An appropriate rainfall-deviation metric of immediate year or lagged by 1 year was chosen for each individual species on the basis of the univariate model that yielded the lowest Akaike information criterion (AIC) value (Burnham & Anderson 2002; Rhodes et al. 2006). We used weights (wi) derived from AIC corrected (AICc) for small sample size to evaluate support for alternate models in the candidate set (Burnham & Anderson 2002).

As a plausibility check of our approach, we analyzed devils first. A range of independent information (Lachish et al. 2007; McCallum et al. 2009) shows that rapid declines in devil numbers occurred soon after disease arrival. If our methods did not detect this decline, it would mean they could not reliably detect changes in other species following the arrival of DFTD.

To assess the importance of top–down and bottom–up effects in the early and mid-DFTD arrival regions at the level of an individual transect, we used generalized linear mixed models (GLMMs) with a Poisson error distribution and log-link function. We assessed whether mesopredator populations could be predicted by populations of competitively dominant species or their prey and whether mesopredator populations were affected by environmental variables with the nlme library in R (version 2.11.0). We used mixed models with transect as a random factor to account for temporal pseudoreplication resulting from surveying the same transects each year. We ran separate analyses on the early and mid-DFTD regions because they varied in land use, mean annual rainfall, and intensity of human settlement. To quantify these differences, we estimated environmental variables for each transect with ArcGIS (version 9.2) by demarcating a 2-km buffer around each individual transect. In addition to rainfall, we obtained estimates of the percentage of open vegetation (lacking trees and complex vegetation structure; made up mostly of agricultural land) and, as a proxy for the extent of human settlement, the number of property addresses (Information and Land Services 2004). To ensure simplicity and reduce overfitting, we limited the number of species used as explanatory variables to 3. These species were prey and competitor species we considered likely to have the largest effect on the mesopredator species being assessed. We incorporated rainfall deviation, number of properties, and percentage of open vegetation estimated for each transect into the models as covariates. We fitted a set of alternative models to occurrence data of the given species and based model selection on AICc. The relative effect of each explanatory variable was quantified by summing the weights of all models containing the variable (Burnham & Anderson 2002; Rhodes et al. 2006).

Results

  1. Top of page
  2. AbstractResumen
  3. Introduction
  4. Methods
  5. Results
  6. Discussion
  7. Acknowledgments
  8. Supporting Information
  9. Literature Cited
  10. Supporting Information

Population Changes Associated with DFTD Arrival

Clear regional and temporal relations were evident in several species in the early and mid-DFTD regions (Figs. 2 & 3). Results from the mid-DFTD region should be interpreted with caution because disease arrival was as late as 2003 and 5 years may not be sufficient to detect population-level changes if there are lags in response of vital rates to declining devil populations. Thus, we do not present results for the late region, although the GAM plots for the late region and the disease-free region are shown in Figures 2 and 3.

image

Figure 2. Mean number of animals per transect from 1985 to 2008 and the generalized additive models (GAMs) for each predator species for all devil facial tumor disease (DFTD) arrival regions (early, mid, late, absent) (vertical lines, period of disease arrival; shaded areas, 95% CIs for GAM models; points, actual values).

Download figure to PowerPoint

image

Figure 3. Mean number of animals per transect from 1985 to 2008 and the generalized additive models (GAMs) for each devil and mesopredator prey species for all devil facial tumor disease (DFTD) arrival regions (early, mid, late, absent) (vertical lines, period of disease arrival; shaded areas, 95% CIs for GAM models; points, actual values).

Download figure to PowerPoint

Declines in devil populations occurred rapidly following recorded disease arrival across both regions (Fig. 2 & Table 1). DFTD led to detectable population changes within 1 year of disease detection in the early region and within 1–3 years in the mid region (Fig. 4 & Table 1). In the mid region, the 3-year lag model had the greatest wi, although this model did not have substantially more support than models with a shorter lag time. By 2003 the disease had spread across the entire region and decline in devil populations was evident. Rainfall deviation had little effect on devil populations in either region.

Table 1. Results of generalized linear models assessing the effects of the arrival of devil facial tumor disease (DFTD) on other species in areas where the disease has been present the longest (early DFTD region, 1996–1999) and where disease arrived between 2000 and 2003 (mid-DFTD region).*
Top-order carnivoresDevil and mesopredator prey
competing models∆AICcawibcompeting models∆AICcawib
  1. a

    Abbreviations: ∆AICc, change in small-sample corrected Akaike information criterion.

  2. b

    Model weights calculated from the AICc.

  3. c

    The number before year is the lag time from the first year of disease detection in the region until a population change is detected.

early-DFTD regionearly-DFTD region
Tasmanian devil  Tasmanian pademelon  
DFTD.1 yearc0.000.41DFTD + rainfall0.000.51
DFTD.1 year + rainfall1.050.24DFTD.1 year + rainfall1.660.22
DFTD1.800.16null25.530.00
null50.020.00Brushtail possum  
Feral cat  DFTD0.000.17
year0.000.21DFTD + rainfall.1 year0.030.16
DFTD.5 years0.740.14rainfall.1 year0.280.14
DFTD.3 years1.310.11year0.620.12
DFTD.2 years1.340.11DFTD.1 year + rainfall. 1 year1.250.09
DFTD.1 year1.460.10null7.300.00
null5.210.02European rabbit  
Eastern quoll  null0.000.21
DFTD.3 years + rainfall.1 year0.000.65DFTD.5 years0.360.18
DFTD.5 years + rainfall.1 year2.470.19Year0.570.16
null23.140.00DFTD.3 years1.770.09
   Medium mammals  
   Year0.000.19
   rainfall.1 year0.030.19
   Null0.310.16
mid-DFTD regionmid-DFTD region
Tasmanian devil  Tasmanian pademelon  
DFTD.3 years0.000.34DFTD + rainfall.1 year0.000.27
DFTD.1 year1.240.18DFTD0.300.23
DFTD1.640.15DFTD.1 year1.810.11
DFTD.2 years2.110.12Null22.150.00
Null17.590.00Brushtail possum  
Feral cat  DFTD0.000.42
null0.000.42DFTD.1 year2.000.15
rainfall.1 year2.490.12Null2.950.10
Eastern quoll  European rabbit  
DFTD + rainfall0.000.77Year0.000.28
DFTD.1 year + rainfall2.970.17Rainfall1.350.14
null14.060.00DFTD.5 years1460.13
   Null8.180.00
   Medium mammals  
   Rainfall0.000.32
   DFTD + rainfall1.350.16
   Null16.740.00
image

Figure 4. The best-fitting generalized linear model (GLM) for the mean number of animals per transect for devils and eastern quolls. The best model, the linear model for feral cats, was not significantly different from models containing a DFTD arrival variable, so the second-best model was used to illustrate devil facial tumor disease arrival affects (vertical lines, period of disease arrival; shaded areas, 95% CIs for GLMs; points, actual values).

Download figure to PowerPoint

Responses of feral cats to devil population decline varied across the disease-arrival regions. Populations increased after DFTD arrival in the early region (Figs. 2, 4, & Table 1). GLM models that included a DFTD arrival variable (with varying lag times) received almost 70% of the total weight in the model set (not all models are shown in Table 1), although a simple linear trend with year was the single most strongly supported model (20% of the weight) (Table 1). In the mid region, a complex temporal pattern in cat populations cannot be explained by either rainfall or DFTD arrival. The lack of clear effects may be due to a lagged response to devil decline, as indicated from results for the early region. Rainfall deviation did not affect changes in feral cat populations in either early or mid regions.

Changes in eastern quoll populations were the most strongly associated with DFTD arrival of any species, aside from devils. In the early region, eastern quolls declined within 3 years of DFTD arrival and subsequent devil decline. This model, which incorporated rainfall deviation lagged by 1 calendar year, carried 65% of the weight (Fig. 4 & Table 1). In the mid region, however, the population decline of eastern quolls occurred simultaneously with DFTD arrival. Eastern quoll populations appeared to decline during the same period that feral cat populations increased (Fig. 2). Rainfall deficiency was also important in this region and may have more biological effect than the lagged deviation in the early region because mean annual rainfall was lower.

For Tasmanian pademelons, 1 of 2 devil prey species analyzed, rainfall deviation metrics were associated with increased populations, whereas DFTD arrival resulted in a slight decrease in populations in both regions (Table 1). Rainfall deviations had much more effect on pademelon populations in the early region than in the mid region. A similar pattern was seen in brushtail possums, the second of the devil prey species analyzed. The arrival of DFTD led to small declines of brushtail possums in both regions, although this relation was less pronounced in the early region; alternative models containing the rainfall deviation and the simple time trend model had similar support (Table 1). Other variables that could have a large effect on abundance of these prey species, such as culling intensity, could not be incorporated in models due to a lack of systematic data.

For the prey species of mesopredators, medium mammals and European rabbits (Oryctolagus cuniculus), data

were highly variable and unmeasured environmental variables probably had more effect on population fluctuations than devil decline. Medium mammal populations showed an overall declining trend across all regions, and rainfall deviation was more important in the drier mid region than in the early region. For rabbits, a decline in the mid and late regions following the introduction of rabbit calicivirus disease (RCD) in 1996 (Tasmanian Year Book 1998) was evident (Fig. 3). Decline from calicivirus was less pronounced in the early region, where the mean number of rabbits per transect was much lower.

Bottom–Up and Top–Down Effects

The extent of top–down and bottom–up control for regulating mesopredator populations varied between the 2 DFTD arrival regions (Table 2). Top–down control was much more important in the early region, which overall had much less agricultural land and human settlement and higher rainfall than the mid region. Here, the strong negative association of feral cats with devils suggests effective top–down regulation of cats. This variable received 96% of the weight within the candidate model set. Prey availability was less important in predicting feral cat populations; models with rabbit populations received 27% of the weight and medium mammals 64% of the weight. In this region, feral cats were more common in agricultural areas, but their presence was not strongly affected by human settlement.

Table 2. Three generalized linear mixed models that were the best predictors of mesopredator occurrence in the early- and mid-devil facial tumor disease (DFTD) arrival regionsa
     Model explanatory variables 
        RainfallPercentAddress 
DFTDModel∆AI InterceptTDERMMdeviationopen vegetationpoints 
regionrankCcwi(SE)(SE)(SE)(SE)(SE)(SE)(SE)σ
  1. a

    Abbreviations: wi, Akaike weights corrected for small sample size; FC, feral cat; TD, Tasmanian devil; ER, European rabbit; MM, medium mammals; σ, standard deviation of the random effect.

  2. b

    Sum of weights of all candidate models containing the variable.

Feral cat
Early10.000.29−3.30 (0.27)−0.39 (0.16) −0.30(0.20)−1.03 (0.48)2.10 (0.59) 0.68
 20.510.23−3.30 (0.27)−0.38 (0.16) −0.30 (0.20)−1.02 (0.48)1.85 (0.62)0.00 (0.00)0.65
 30.900.19−3.30 (0.27)−0.41 (0.16)  −1.01 (0.48)1.97 (0.59) 0.68
 null250.00        
 relative importance of variableb9627641009941 
Mid10.000.16−3.30 (0.29)0.11 (0.07)0.03 (0.02)0.23 (0.14)−0.36 (0.40)2.17 (0.60)0.00 (0.00)0.39
 20.040.15−3.24 (0.29) 0.03 (0.02)0.22 (0.14)−0.30 (0.40)2.21 (0.60)0.00 (0.00)0.39
 30.260.14−3.24 (0.29)0.11 (0.07) 0.23 (0.14)−0.37 (0.40)2.23 (0.61)0.00 (0.00)0.41
 null1340.00        
 relative importance of variableb4560619910094 
        rainfallpercentaddress 
     TDFCMMdeviationopen vegetationpoints 
     (SE)(SE)(SE)(SE)(SE)(SE) 
Eastern quoll
Early10.000.35−1.91 (0.27)0.16 (0.04)−0.36 (0.21)0.20 (0.09)1.25 (0.30) −0.02 (0.01)1.58
 21.370.18−1.94 (0.27)0.17 (0.05) 0.20 (0.09)1.26 (0.30) −0.02 (0.01)1.58
 31.990.13−1.96 (0.41)0.16 (0.05)−0.36 (0.21)0.20 (0.09)1.24 (0.30)0.18 (1.09)−0.02 (0.01)1.58
 null520.00        
 relative importance of variableb9968801002391 
Mid10.000.32−0.50 (0.45)  0.37 (0.07)−0.03 (0.26)−2.68 (1.08)−0.00 (0.00)1.19
 21.590.15−0.52 (0.45)0.04 (0.05) 0.37 (0.07)−0.03 (0.26)−2.69 (1.08)−0.00 (0.00)1.19
 31.790.13−0.50 (0.45) −0.06 (0.13)0.37 (0.07)−0.03 (0.26)−2.65 (1.08)−0.00 (0.00)1.19
 null760.00        
 relative importance of variableb37371001008972 

In contrast, in the mid region feral cat populations were better predicted by bottom–up factors of prey availability than top–down pressure from devils (Table 2). Here, the weight for models incorporating devil populations was much lower (45%) than for the early region, and there was a slightly positive relation. Feral cats were more strongly associated with their prey in this region; rabbits and medium mammals received 60% and 61% of the model weights, respectively. Further evidence for bottom–up control of cat populations can be seen in the GAM plots (Figs. 2 & 3). The mid region supported much higher densities of the major prey of feral cats (i.e., European rabbits) (Fig. 3), and the decline in rabbits due to the release of rabbit calicivirus disease in 1996 was synchronous with a decline in feral cats in this region. This was not evident in the early region, where there were much lower densities of rabbits. Density of human settlement and agricultural land were also important predictors of feral cat populations and may be related to prey population density.

Top–down effects were very important in explaining eastern quoll populations in the early region. Eastern quolls were strongly and positively associated with the occurrence of devils, but where feral cats were present, eastern quolls were much less likely to occur. Models containing these variables received 99% and 68% of the weights, respectively (Table 2). Eastern quolls were also strongly and negatively associated with the density of human settlement. In contrast to the early region, populations of eastern quolls in the mid region were less affected by top-down control. In this region neither the competitively dominant feral cats nor devils were strongly linked with eastern quoll populations; each variable received 37% of the weight. Instead, bottom–up control from agriculture and human settlement were both strongly and negatively linked with eastern quoll populations in this region. For both regions the populations of medium mammals and eastern quolls were positively associated, probably due to their similar ecological requirements and habitat, and rainfall deviations were important.

Discussion

  1. Top of page
  2. AbstractResumen
  3. Introduction
  4. Methods
  5. Results
  6. Discussion
  7. Acknowledgments
  8. Supporting Information
  9. Literature Cited
  10. Supporting Information

Our results suggest that the decline of the Tasmanian devil, the apex mammalian predator in the Tasmanian ecosystem, has had a substantial effect on terrestrial mammal fauna, a result consistent with evidence from apex predator loss in other ecosystems (e.g., Crooks & Soulé 1999; Johnson et al. 2007). At a regional scale, disease arrival and consequent decline in devil populations affected mammalian mesopredator populations and potentially reduced ecosystem complexity, although the relative strength of top–down and bottom–up forces was affected by environmental variation. In areas of more recent DFTD arrival, which are drier and have greater anthropogenic influences, a bottom–up mechanism appeared to be driving mesopredator populations and mesopredator release was diminished. This demonstrates the importance of studying trophic cascades and interspecific relations across heterogeneous environments.

Our results provide strong evidence that the loss of devils from the ecosystem may trigger a cascade of competitive release and suppression among the size-structured mesopredator guild. There was compelling evidence to support the assertion that devils can exert top–down control on feral cat populations. First, in the region where the disease had been present the longest, there was some evidence that feral cats were increasing at a faster rate subsequent to DFTD arrival and corresponding devil decline, indicating mesopredator release. Second, the strong negative association of the occurrence of cats and devils on transects indicated a suppressive effect of devils on cat populations. As devil populations declined and cats increased, the smaller mesopredator, the eastern quoll, declined, and there was no sign of population recovery, which lead to concerns for its conservation.

Some important outcomes for conservation arise from this result. First is the consequence of increased predation by cats on populations of native mammals. Competitive release of cats following apex predator decline has been linked with declines in prey species in ecosystems in Australia and in North America (Smith & Quin 1996; Crooks & Soulé 1999) and with the decline in eastern quolls in this study. A second outcome is that if devils can suppress populations of feral cats, then this apex predator could be used as a conservation tool to suppress cat abundance or activity and thus limit the effects of cat predation on native fauna, as has been proposed for the dingo (Canis lupus dingo) (Dickman et al. 2009). It may be beneficial to reintroduce devils to parts of their former mainland Australian range (Wroe & Johnson 2003). A third and more general implication of our work is that ours is one of the first examinations of a trophic cascade resulting from a disease-induced decline of an apex predator. Carnivores are the mammalian order with the second highest proportion of species threatened by infectious diseases or parasites (Pedersen et al. 2007). We therefore expect there will be many other examples of disease-induced declines of apex predators that will have pervasive effects throughout the remainder of the ecosystem.

Ecosystem productivity affects the complexity of food webs and the relative strength of top–down and bottom–up regulation (Elmhagen & Rushton 2007; Elmhagen et al. 2010). Our results show a clear disparity in species-specific responses to devil decline across 2 ecologically different regions. This highlights the importance of accounting for localized and large-scale environmental factors when assessing trophic cascades. The strength of top–down control of cats by devils appeared to be affected by environmental factors that may affect productivity and the abundance of cat prey species. Top–down control was strongest in northeastern Tasmania, where the disease has been present the longest and where rainfall and forest cover are high. In the mid region, which overall has less rainfall, more agricultural land, and denser human settlement, the feral cat population was higher overall, was not affected by devil populations, and was best explained by factors that affected food resource levels for cats (i.e., rainfall, vegetation type, prey abundance, and human presence). Human-modified landscapes can support elevated populations of rabbits and rodents, and there was evidence that higher prey densities affected feral cat populations in agricultural landscapes. A sharp decline in cats was evident following the illegal release of rabbit calicivirus in the mid north of Tasmania in 1996 (Tasmanian Year Book 1998), which devastated their most common prey species, the European rabbit. Human modification of landscapes has a pervasive effect on top–down and bottom–up structuring of ecosystems (Elmhagen & Rushton 2007). Urbanization and agricultural expansion, including associated habitat fragmentation, can increase populations of mesopredators, such as red foxes (Vulpes vulpes), through the provision of food resources (Prange & Gehrt 2004; Elmhagen & Rushton 2007).

In contrast to the larger introduced mesopredator, the native eastern quoll declined following the large DFTD-associated decline in devil populations in northeastern Tasmania. On individual transects, eastern quolls were strongly and positively associated with devils and were detected less frequently when feral cats were present. Thus, eastern quolls may be indirectly protected by devils because devils suppress cats. Feral cats prey on similar-sized northern quolls (Dasyurus hallucatus) on mainland Australia (e.g., Oakwood 2000). Cats have coexisted with all species of quolls across Australia for about 200 years since European settlement (Abbott 2002), apparently without definitive or direct evidence of detrimental population-level effects on quolls. It could be that the synergistic effects of prolonged drought and increased cat populations leading to more intensive predation by cats is causing population decline in quolls. The regulatory mechanism affecting eastern quoll populations was more complex in the mid region, where, similar to cats, their relation with larger carnivores was considerably less accentuated. The strong support for environmental factors affecting eastern quoll populations suggested the role of bottom–up forces in driving population trends. Human settlement and large-scale agriculture may be having detrimental effects on this species despite a preference for foraging in open grasslands (Jones & Barmuta 2000) and a diet at lower elevations largely comprised of pasture invertebrates (Blackhall 1980). Maintaining populations of the devil in the Tasmanian ecosystem, at densities sufficient to exert top–down control, may protect populations of smaller mesopredators, at least under some environmental conditions.

We did not find clear explanations for populations changes of devil and mesopredator prey species. Some potentially important variables that may affect abundance, such as lethal control programs for pademelons and possums, could not be included in the models because of the lack of systematic data. Culling intensity is a significant factor in the population variation observed in these species at regional scales (Driessen & Hocking 1992).

From our data set, we could not distinguish between behavioral shifts and demographic processes as the mechanism underlying observed changes in the spotlighting counts of species. Behavioral shifts, including temporal partitioning and habitat use, may explain some of the observed changes in populations due to increasing or decreasing detectability. Behavioral changes are linked with fitness (Morris et al. 2009) and over time will likely translate to changes in abundance.

The spotlight data are the only broadscale, longitudinal wildlife monitoring data set available over the period spanning DFTD emergence. Data were sufficient to show clear population changes of devils, feral cats, and eastern quolls but not of the rare and cryptic spotted-tailed quoll. We also could not account for large avian predators. Future efforts to assess ecosystem effects of devil decline will benefit from implementing methods that specifically target cryptic carnivores, small mammals, and birds.

Our results are consistent with assertions that DFTD causes rapid and severe population decline in devils (Lachish et al. 2007; McCallum et al. 2009) that results in substantially reduced apex predator populations within a few years. There was no indication of population recovery, and with no effective management options at present that could aid recovery of wild populations, the trophic cascades in the Tasmanian ecosystem resulting from loss of devils are likely to continue. Of concern is the potential endangerment of prey species of feral cats, which include eastern quolls and small native mammals.

Globally, the continuing decline of apex predators is reshaping ecosystems into more simplified and homogenized states (Estes et al. 2011). As apex predator loss occurs concurrently with habitat degradation and fragmentation and climate change, the importance of accounting for environmental drivers to tease out complex relations cannot be understated. The changes occurring in Tasmanian ecosystems are consistent with this global trend. There is a substantial risk of a state shift to an alternative system driven by invasive species that could cause very high extinction rates from which ecological recovery would be difficult (Wallach et al. 2010). Australia leads the world in mammalian species extinctions (Short & Smith 1994). Until now, Tasmania has been a refuge for mammals that have been driven to extinction on mainland Australia, probably by introduced predators. This example provides further support for the strong role of predators in structuring ecosystems and for the role of productivity in determining the strength of top–down control (Elmhagen & Rushton 2007). In ecosystems throughout the world, apex predators play a keystone role, and their preservation and restoration is important for building resilience into natural ecosystems and the conservation of many vulnerable species.

Acknowledgments

  1. Top of page
  2. AbstractResumen
  3. Introduction
  4. Methods
  5. Results
  6. Discussion
  7. Acknowledgments
  8. Supporting Information
  9. Literature Cited
  10. Supporting Information

We acknowledge the Department of Primary Industries, Parks, Water and the Environment and the Bureau of Meteorology for allowing us to use their data for this study.

Supporting Information

  1. Top of page
  2. AbstractResumen
  3. Introduction
  4. Methods
  5. Results
  6. Discussion
  7. Acknowledgments
  8. Supporting Information
  9. Literature Cited
  10. Supporting Information

A summary of the hypothesized species interactions (Appendix S1) is available online. The authors are solely responsible for the content and functionality of these materials. Queries (other than absence of the material) should be directed to the corresponding author.

Literature Cited

  1. Top of page
  2. AbstractResumen
  3. Introduction
  4. Methods
  5. Results
  6. Discussion
  7. Acknowledgments
  8. Supporting Information
  9. Literature Cited
  10. Supporting Information
  • Abbott, I. 2002. Origin and spread of the cat, Felis catus, on mainland Australia, with a discussion of the magnitude of its early impact on native fauna. Wildlife Research 29:5174.
  • Berger, K. M., and E. M. Gese. 2007. Does interference competition with wolves limit the distribution and abundance of coyotes? Journal of Animal Ecology 76:10751085.
  • Beschta, R. L., and W. J. Ripple. 2009. Large predators and trophic cascades in terrestrial ecosystems of the western United States. Biological Conservation 142:24012414.
  • Blackhall, S. 1980. Diet of the eastern native-cat, Dasyurus viverrinus (Shaw), in southern Tasmania. Australian Wildlife Research 7:191197.
  • Brashares, J. S., L. R. Prugh, C. J. Stoner, and C. W. Epps. 2010. Ecological and conservation implications of mesopredator release. Pages 221240 in J. Terborgh and J. A. Estes, editors. Trophic cascades: predators, prey and the changing dynamics of nature. Island Press, Washington, D.C.
  • Burnham, K., and D. Anderson. 2002. Model selection and multimodal inference: a practical information and theoretic approach. 2nd edition. Springer-Verlag, New York.
  • Crooks, K. R., and M. E. Soulé. 1999. Mesopredator release and avifaunal extinctions in a fragmented system. Nature 400:563566.
  • Dalerum, F., E. Z. Cameron, K. Kunkel, and M. J. Somers. 2009. Diversity and depletions in continental carnivore guilds: implications for prioritizing global carnivore conservation. Biology Letters 5:3538.
  • Dickman, C. R. 1996. Impact of exotic generalist predators on the native fauna of Australia. Wildlife Biology 2:185195.
  • Dickman C. R., A. S. Glen, and M. Letnic. 2009. Reintroducing the dingo: Can Australia's conservation wastelands be restored? Pages 238269 in M. W. Hayward and M. J. Somers, editors. Reintroduction of top-order predators. Wiley-Blackwell, Oxford, United Kingdom.
  • Driessen, M., and G. Hocking. 1992. Review and analysis of spotlight surveys in Tasmania: 1975–1990. Scientific report 92/1. Department of Parks, Wildlife and Heritage, Hobart, Tasmania.
  • Elmhagen, B., G. Ludwig, S. P. Rushton, P. Helle, and H. Linden. 2010. Top predators, mesopredators and their prey: interference ecosystems along bioclimatic productivity gradients. The Journal of Animal Ecology 79:785794.
  • Elmhagen, B., and S. Rushton. 2007. Trophic control of mesopredators in terrestrial ecosystems: Top-down or bottom-up? Ecology Letters 10:197206.
  • Estes, J. A., et al. 2011. Trophic downgrading of planet Earth. Science 333:301306.
  • Hawkins, C. E., et al. 2006. Emerging disease and population decline of an island endemic, the Tasmanian devil Sarcophilus harrisii. Biological Conservation 131:307324.
  • Hocking, G., and M. Driessen. 1992. Tasmanian spotlight survey manual. Department of Parks, Wildlife and Heritage, Hobart, Tasmania.
  • Information and Land Services. 2004. LIST Address point dataset. DPIPWE Tasmania, Australia.
  • Johnson, C. N., J. L. Isaac, and D. O. Fisher. 2007. Rarity of a top predator triggers continent-wide collapse of mammal prey: dingoes and marsupials in Australia. Proceedings of the Royal Society B: Biological Sciences 274:341346.
  • Jones, M. E. 1997. Character displacement in Australian dasyurid carnivores: size relationships and prey size patterns. Ecology 78:25692587.
  • Jones, M. E., and L. A. Barmuta. 1998. Diet overlap and abundance of sympatric dasyurid carnivores: A hypothesis of competition? Journal of Animal Ecology 67:410421.
  • Jones, M. E., and L. A. Barmuta. 2000. Niche differentiation among sympatric Australian dasyurid carnivores. Journal of Mammalogy 81:434447.
  • Jones, M. E., G. C. Smith, and S. M. Jones. 2004. Is anti-predator behaviour in Tasmanian eastern quolls (Dasyurus viverrinus) effective against introduced predators? Animal Conservation 7:155160.
  • Lachish, S., M. Jones, and H. McCallum. 2007. The impact of disease on the survival and population growth rate of the Tasmanian devil. Journal of Animal Ecology 76:926936.
  • McCallum, H., M. Jones, C. Hawkins, R. Hamede, S. Lachish, D. Sinn, N. Beeton, and B. Lazenby. 2009. Transmission dynamics of Tasmanian devil facial tumor disease may lead to disease-induced extinction. Ecology 90:33793392.
  • McCallum, H., D. Tompkins, M. Jones, S. Lachish, S. Marvanek, B. Lazenby, G. Hocking, J. Wiersma, and C. Hawkins. 2007. Distribution and impacts of Tasmanian devil facial tumor disease. EcoHealth 4:318325.
  • Morris, D. W., B. P. Kotler, J. S. Brown, V. Sundararaj, and S. B. Ale. 2009. Behavioral indicators for conserving mammal diversity. Annals of the New York Academy of Sciences 1162:334356.
  • Oakwood, M. 2000. Reproduction and demography of the northern quoll, Dasyurus hallucatus, in the lowland savanna of northern Australia. Australian Journal of Zoology 48:519539.
  • Oksanen, L., S. Fretwell, J. Arruda, and P. Niemela. 1981. Exploitation ecosystems in gradients of primary productivity. American Naturalist 118:240261.
  • Oksanen, T. 1990. Exploitation ecosystems in heterogeneous habitat complexes. Evolutionary Ecology 4:220234.
  • Pace, M. L., J. J. Cole, S. R. Carpenter, and J. F. Kitchell. 1999. Trophic cascades revealed in diverse ecosystems. Trends in Ecology & Evolution 14:483488.
  • Pedersen, A. B., et al. 2007. Infectious diseases and extinction risk in wild mammals. Conservation Biology 21:12691279.
  • Prange, S., and S. D. Gehrt. 2004. Changes in mesopredator-community structure in response to urbanization. Canadian Journal of Zoology 82:18041817.
  • R Development Core Team. 2011. R: A language and environment for statistical computing. R Foundation for Statistical Computing, Vienna, Austria.
  • Rhodes, J. R., T. Wiegand, C. A. McAlpine, J. Callaghan, D. Lunney, M. Bowen, and H. P. Possingham. 2006. Modeling species’ distributions to improve conservation in semiurban landscapes: Koala case study. Conservation Biology 20:449459.
  • Ritchie, E. G., and C. N. Johnson. 2009. Predator interactions, mesopredator release and biodiversity conservation. Ecology Letters 12:982998.
  • Short, J., and A. Smith. 1994. Mammal decline and recovery in Australia. Journal of Mammalogy 75:288297.
  • Smith, A. P., and D. G. Quin. 1996. Patterns and causes of extinctions and decline in Australian conilurine rodents. Biological Conservation 77:243267.
  • Soule, M. E., D. T. Bolger, C. A. Allison, J. Wright, M. Sorice, and S. Hill. 1988. Reconstructed dynamics of rapid extinctions of chaparral-requiring birds in urban habitat islands. Conservation Biology 2:7592.
  • Southwell, C., and M. Fletcher. 1993. Repeatability and standardisation of roadside spotlight counts of mammals in Tasmania. Australian Mammalogy 16:7375.
  • Tasmanian Year Book. 1998. Rabbit calicivirus in Tasmania Australian. Bureau of Statistics, Hobart.
  • Terborgh, J., J. Estes, P. Paquet, K. Ralls, D. Boyd-Heger, B. Miller, and R. Noss. 1999. The role of top carnivores in regulating terrestrial ecosystems. Wild Earth 9:4256.
  • van Dyck, S., and R. Strahan, editors. 2008. Mammals of Australia. New Holland Press.
  • Wallach, A. D., C. N. Johnson, E. Ritchie, and A. J. O'Neill. 2010. Predator control promotes invasive dominated ecological states. Ecology Letters 13:10081018.
  • Wroe, S., and C. Johnson. 2003. Bring back the devil? Nature Australia 27:84.
  • Zuur, A. F., E. N. Ieno, N. J. Walker, A. A. Saveliev, and G. M. Smith. 2009. Mixed effects models and extensions in ecology with R. Springer Science+Business, New York.

Supporting Information

  1. Top of page
  2. AbstractResumen
  3. Introduction
  4. Methods
  5. Results
  6. Discussion
  7. Acknowledgments
  8. Supporting Information
  9. Literature Cited
  10. Supporting Information

Disclaimer: Supplementary materials have been peer-reviewed but not copyedited.

FilenameFormatSizeDescription
cobi12152-sup-0001-Appendix.docx52KAppendix S1

Please note: Wiley Blackwell is not responsible for the content or functionality of any supporting information supplied by the authors. Any queries (other than missing content) should be directed to the corresponding author for the article.