Do Exotic Vertebrates Structure the Biota of Australia? An Experimental Test in New South Wales

Abstract

From 1993 to 2001, we conducted a series of experiments in a mixed grassland–woodland system in central New South Wales (NSW) to quantify the interactions between red foxes and their prey and competitors. Foxes were removed from two areas around the perimeter of Lake Burrendong, and data were collected from these areas and a nearby untreated area before, during, and after the period of fox control. The arrival of rabbit hemorrhagic disease (RHD) in 1996 provided an opportunity to examine the interactive effects of controlling foxes and rabbits. In this landscape, typical of central NSW, (a) the fox population was not affected by a large reduction in the abundance of rabbits, or vice versa; (b) the cat population declined in areas where foxes were removed after the large RHD-induced reduction in rabbit numbers, but there was no consistent response to the removal of foxes; (c) the abundance of some macropod species increased in response only to the combined removal of rabbits and foxes; (d) there were no consistent changes in the abundances of bird species in response to the removal of either foxes or rabbits, but there were clear habitat differences in bird species richness; and (e) there was likely to be an increase in woody plant species after the large reduction in rabbit populations by RHD. We conclude that (a) long-term field experiments (more than 3 years) are required to quantify the indirect consequences of controlling foxes and rabbits, and (b) single manipulations, such as fox control or rabbit control, are not necessarily sufficient for the conservation of remnant woodland communities in southeastern Australia.

Appendices

Appendix 1

Methods used to Assess Changes in the Abundance of Mammal and Bird Species; Diet Analyses for Foxes, Cats, and Wedge-tailed Eagles; and Methods used to Assess Vegetation Changes with the Arrival of Rabbit Hemorrhagic Disease (RHD) at Lake Burrendong, New South Wales.

Spotlight Counts of Mammal Species (Macropods, Rabbits, Foxes, Cats)

From November 1994 to December 2001, counts were conducted for each 200-m section of a 35.6-km transect; they were repeated over 3 consecutive nights for each session. This transect covered all three areas (Control, Removal 1, Removal 2) and was designed to obviate any repeated counting of animals. An index of abundance for each species was taken as the average number of individuals recorded over the 3 nights. Data were obtained on the relative abundance of these species over an 11-month period before the fox treatments were initiated (T1: six spotlight sessions), for 4 years during the fox removal phase (T2 + T3: 34 sessions), and for a 26-month period to December 2001 when there was no fox baiting (T4: eight sessions). There were 15 spotlight sessions in the 22-month period (T1 + T2) of generally high rabbit numbers prior to the arrival of rabbit hemorrhagic disease (RHD), followed by 33 sessions during the 5 years and 3 months of low rabbit numbers (T3 + T4).

Index of Bird Abundance

Monthly surveys to monitor bird abundance were run from October 1995 to July 1996 (during T2). For each survey, the more common bird species were counted from a slow-moving vehicle over 3 consecutive days. Species were defined as “more fox-prone” if they nest on or near the ground, or feed mostly at ground level; species potentially vulnerable to foxes that are either very small or do not spend the majority of time feeding on the ground were classified as “less fox-prone.” There were four transects covering the Control area (21 km)—one covering Removal 1 (7.6 km) and two covering Removal 2 (7 km). On most occasions, each transect was surveyed in the early morning and again in the evening. For some of the earlier surveys, not all transects were complete and not all were covered both morning and evening, so an appropriate correction factor was applied to account for the shorter distances. The number of individuals of each species, irrespective of distance from the center of the transect, was recorded for 200-m sections along each transect. The surveys recommenced in October 1996 and continued to January 2001 (T3, T4). For this period, the frequency of surveys was reduced to once a quarter, and all bird species were recorded. Each 200-m section was classified according to two types of vegetation structure. Grassy–open woodland contained sections that were dominated by either grassland or open eucalypt woodland, whereas dense woodland was dominated by Callitris or by dense eucalypt woodland. The Control area contained 15.6 km of grassy–open woodland, Removal 1 contained 5.4 km, and Removal 2 contained 2.6 km; the remaining subsections in each area were dense woodland.

Each 200-m section of the bird transects was assigned as having high or low rabbit abundance pre-RHD. The transect route was similar to that used for the spotlight counts of mammals; 136 sections were included in both types of survey, and birds were monitored on a further 42 sections. Data for the first 15 sessions from the 136 sections in common with the spotlight route were used to identify localities with high or low rabbit numbers—that is, parts of the study area with characteristically few rabbits versus rabbit-prone locations. Sections with more than a total of 50 rabbits during the first 15 sessions were defined as having high numbers (that is, an average of approximately 17 rabbits per spotlight km during T1 and T2). In addition, any section with less than a total of 50 rabbits that was between two sections of high numbers was also regarded as high. The remaining 42 sections unique to the daytime bird survey were allocated as having high or low rabbit abundance by comparing the number of rabbits counted during the bird surveys with those sections that were also covered by the spotlight counts. There was little, if any, scope for the arrival of RHD to change the abundance of rabbits in localities where they were scarce during T1 and T2 but high potential for disease-induced change in localities with high rabbit numbers.

From November 1996 to April 2001 (T3 and most of T4), we recorded the number and species of birds visiting thirty-six 0.25-ha circular plots over a 10-min period. Plots were located in each of four vegetation types within the Control area and the Removal 1 area as follows: grassland n = 5, 5; open eucalypt woodland n = 5, 4; dense eucalypt woodland n = 4, 4; Callitris woodland n = 4, 5. The area of each plot was determined by the visibility within Callitris, the most dense vegetation type. All plots were monitored once in the early morning during each survey, which extended over a 3-day period. All plots were surveyed twice a season (that is, eight times per year), for a total of 32 surveys. The plot surveys were designed to include bird species not likely to be adequately recorded during the vehicle-based transects.

Index of Change in Bird Population Densities Due to Fox and Rabbit Treatments

To examine the effects of fox removal on bird populations, an index of change was calculated for each species, i, using transect data from the Control area (C) and Removal 1 area (R) for two time periods: fox removal (T2 + T3) and after the return of foxes (T4). Morning and evening counts for each survey were summed and divided by the total distance covered. For each period, τ, the data were combined from all surveys to calculate the average number of individuals per kilometer (km⁻¹) for each species in the Control area, d _iC (τ), and the number in Removal 1, d _iR (τ).

For the Control area, the instantaneous rate of change of each species between the time periods was:

$$ r_{iC} = \ln [d_{iC} ({\rm T}4)/d_{iC}({\rm T}2 + {\rm T}3)] $$

(1)

For Removal 1, it was:

$$ r_{iR} = \ln [d_{iR} ({\rm T}4)/d_{iR}({\rm T}2 + {\rm T}3)] $$

(2)

Because bird numbers may have changed for other reasons, such as climate, it is necessary to compare the rates of change in Removal 1 with Control. We computed an index of relative rate of change (I _Fi ) as the difference in instantaneous rates between fox treatment and control:

$$ I_{Fi} = r_{iR} - r_{iC} $$

(3)

A negative value for I _Fi means bird species i declined faster or increased more slowly compared to the control, and hence was disadvantaged by the return of foxes to the Removal 1 area—that is, they had initially benefited from the removal of foxes.

Similarly, to examine the effects of rabbit removal, average indices of abundance for each bird species were calculated for the time periods before RHD (T2) and after the arrival of RHD (T3 + T4). Using the classifications based on 200-m sections of rabbit spotlight counts, we compared localities of low rabbit abundance (L) with those of high rabbit abundance (H). Thus, for locations with low rabbit abundance during T2 (our effective control for rabbit removal by RHD), the rate of change for bird species i was:

$$ r_{iL} = \ln [d_{iL} ({\rm T}3 + {\rm T}4)/d_{iL}({\rm T}2)] $$

(4)

For locations with high rabbit abundance during T2, it was

$$ r_{iH} = \ln [d_{iH} ({\rm T}3 + {\rm T}4)/d_{iH}({\rm T}2)] $$

(5)

The relative change in rates between areas of low and high rabbit numbers (I _Ri ) was:

$$ I_{Ri} = r_{iH} - r_{iL} $$

(6)

A positive value for I _Ri means that bird species i increased faster or declined more slowly compared to the control and hence benefited from the reduction in rabbits due to RHD.

Diets of Foxes and Cats

From July 1994 to June 1997, cat and fox scats were collected on random walks throughout the study area and the prey remains identified. The percentage volume of each food item in each scat was estimated visually and the mean percentage volume calculated for each season (Molsher 1999; Molsher and others 1999). Holling (1959) type II functional responses were fitted for predation by foxes on rabbits and predation by cats on rabbits, assuming that the percentage volume of rabbit material in scats was directly proportional to the intake of rabbits. To allow for a continuously changing population, the seasonal index of rabbit abundance was taken as the area under the curve (AUC) of rabbit counts over time, using spotlight data from the entire site because there was no evidence that the fox treatment affected rabbit abundance (see Results). Maximum likelihood estimation was used to fit the functional response curves (Hilborn and Mangel 1997).

Diet of Wedge-tailed Eagles

Three nests of wedge-tailed eagles were monitored in the study area during the 1996–97 breeding season, and more nests were located progressively over the following years: 10 during 1997–98, 12 during 1998–99, 13 during 1999–2000, 14 during 2000–01, and 15 during 2001–02. During the breeding season, the remains of food material (orts) were collected from around each nest site and from within the nest once the young (eyases) had fledged. Orts were sorted into bone, feather, and regurgitated pellet material. For data presented here, each pellet collected at the height of eyas provisioning in October 1999 and in October 2001 was pulled apart, and the presence of bone, feather, and hair material was identified to genus. A more detailed analysis of orts collected over the entire study will be presented elsewhere.

Vegetation Changes with the Arrival of Rabbit Hemorrhagic Disease

In December 1995, permanent 30 × 2 m belt transects were established to monitor long-term vegetation change in the following four habitats: open eucalypt (Eucalyptus albens) woodland, C. glaucophylla woodland, Allocasuarina verticillata woodland, and grassland. The transects were placed in two of the experimental areas: one set of three transects per habitat was in the Control area; the other set of three transects per habitat was in Removal 2. We monitored the transects in January of 1996, 1997, 1998, 1999, and 2001. Changes in the cover of herbaceous plant species were assessed along the center line of the belt transects, and we used a series of 10 contiguous quadrats (each 3 m long by 2 m wide) to track changes in shrub and tree density. A comparison of the data collected in 1996 (prior to arrival of RHD; T2) to the data from subsequent years (after the arrival of RHD; T3 and T4) shows the changes in vegetation related to the disease-induced reduction in rabbit abundance.

Impact of Herbivores on Grassland and Woodland Vegetation during T3 and T4

The impact of various herbivore species on vegetation in grassland and open eucalypt woodland was assessed using a series of 25 × 25 m exclusion plots, established in February 1997 (Allcock 2002; Allcock and Hik 2003, 2004). There were four grazing treatments, replicated in four locations: two in Removal 1 and two in Control. The four treatments were as follows: unfenced, stock fence (sheep and cattle excluded), kangaroo fence (macropods and livestock excluded), and complete exclosure (macropods, livestock, and rabbits excluded).

Standing biomass was measured in the exclusion plots at the end of each growing season in 1997, 1998, 2000, and 2001. We collected 12 randomly placed samples from 7-cm–diameter circular plots and bulked them to get estimates for each 25 × 25 m plot (Allcock 2002; Allcock and Hik 2003, 2004). In April 1998, we planted 20 individuals of E. albens and C. glaucophylla into each exclusion plot. We monitored browse events, growth, and survival of these seedlings until March 2001 (see details in Allcock and Hik 2004).

Establishment and Changes in Distribution of Woodland Tree Species

Long-term changes in tree canopy cover were estimated from aerial photographs of the entire study area taken in January 1965 and November 1995. Closely matching grids with 500 × 500 m cells were superimposed on both aerial photos. Ten cells were chosen at random from the 93 cells contained within the Control, Removal 1, and Removal 2 areas. Each of the 10 selected cells was subdivided with a 7 × 7 grid, giving 49 intersecting points per cell. The resulting 490 points on each aerial photo were classified as tree, grassland (no tree canopy), or undecided (for example, where the canopy boundary was indistinct). In January 2004, points assessed as having tree canopy in 1995 but not in 1965 were visited and the type of canopy identified as (a) Eucalyptus spp., (b) Callitris spp., or (c) other (with the species identified where possible).

Long-term Effects on Birds of Broad-scale Changes in the Plant Community

Species diversity of birds was calculated using rarefaction (Heck and others 1975; Krebs 1999; Hughes and others 2001), which compares the observed species numbers between habitats that have different sample sizes. A curve of species number against sample size of individuals counted was estimated by averaging randomizations of the observed accumulation curve using software in Krebs (1999). We compared bird species richness among the four vegetation types (grassland, open eucalypt woodland, dense eucalypt woodland, Callitris woodland) using data from the plot surveys.

Appendix 2

Index of Change, I _Fi , for Bird Species Prone to Fox Predation that Were Recorded at Lake Burrendong, New South Wales.

a. Within Grassy–Open Woodland

b. Within Dense Woodland

Appendix 3

Index of Change, I _Fi , for Ground-feeding Bird Species in Response to a Decline in Rabbit Abundance at Lake Burrendong, New South Wales

Table 5