Impacts of an invasive herbivore on indigenous forests


Corresponding author. E-mail


  1. Invasive herbivores can have large negative impacts on natural ecosystems. Management of invasive populations often requires frequent, broadscale, expensive control, which must be justified by demonstrating progress in achieving conservation objectives. This study evaluates benefits of regular extensive control of an invasive herbivore and develops an alternative strategy based on damage thresholds.
  2. We carried out replicated experimental management of brushtail possums, Trichosurus vulpecula, in three areas of native forest in New Zealand. Each area included a site that had extensive possum control for 10 years, prior to and during the 5-year study, and a paired site with no control. We measured indices of possum browse on c.2400 possum ‘preferred’ and c.1200 ‘non-preferred’ trees, and an index of possum abundance, at the beginning and end of the experiment.
  3. Extensive control was effective in reducing possum browse on preferred tree species. Reductions in browse led to increased foliage cover and decreased probability of tree mortality. The probability of browse on an individual tree decreased with increasing amounts of possum-preferred foliage on nearby trees but increased with tree size and with increasing levels of browse on nearby trees. At one site where possum control ceased prematurely, foliage cover decreased, reducing benefits from earlier control.
  4. Synthesis and applications. Our study provides evidence that sustained, extensive control of invasive herbivores can result in significant conservation benefits to susceptible tree species, and that both impacts and benefits can be measured using data typically collected in herbivore impact studies. Furthermore, it shows how local factors such as forest composition can influence the impact of herbivory, how this can be included in large-scale assessments of the benefits of pest control and how site- and species-specific damage thresholds can be derived for improving pest management.


Invasive mammals are a key driver of global change through impacts on native biodiversity, agriculture and human health (Mack et al. 2000). In many countries, their control is conducted over large areas, often frequently, and at high cost. Management agencies usually monitor the effect of control on the target species (e.g. Reddiex et al. 2006; Clayton & Cowan 2010). However, there is limited knowledge of the relationships between pest abundance and resulting damage (e.g. Hone 2007), making it difficult to estimate the potential value of intervention or to use damage thresholds for deciding when to control pests. Often the actual benefits of control are not measured directly or at all (e.g. Reddiex & Forsyth 2006; Clayton & Cowan 2010). Impacts on native biota can be difficult to quantify, especially for generalist foragers affecting many species, and for impacts that accumulate slowly over time. Nevertheless, some forest impacts with these characteristics have been quantified for invasive herbivores ranging from deer (e.g. Tanentzap et al. 2009) to invertebrates (e.g. Gandhi & Herms 2010). Our aim was to assess benefits for native forests from large-scale control of invasive brushtail possums Trichosurus vulpecula in New Zealand, using data typical of that collected in herbivore impact studies worldwide (e.g. Makhabu, Skarpe & Hytteborn 2006; Kamler et al. 2010; Nugent et al. 2010).

Brushtail possums were introduced to New Zealand in 1837 to establish an export fur trade (Clout & Ericksen 2000). They became widespread, colonizing most forested areas, and are highly abundant, with a recent estimate of c.30 million possums (Warburton, Cowan & Shepherd 2009). High population densities in New Zealand are due to a combination of suitable habitat and a lack of competitors, predators and parasites (Clout & Ericksen 2000). Possums are nocturnal, predominantly arboreal folivores, supplementing their diet with high-energy, non-foliar foods when available (Nugent et al. 2000). Selective browsing by possums has been shown to alter forest composition, with tree species preferred by possums becoming locally extinct and less preferred species becoming relatively more abundant (Campbell 1990; Owen & Norton 1995). This change can be rapid with defoliation of forest canopies and subsequent canopy collapse occurring within 20 years of colonization by possums (Payton 2000).

Current management of possums on public conservation lands ranges from localized, intensive and continuous suppression to very low densities, to large-scale (mean, 7178 ha; Veltman & Westbrooke 2011) aerial poisoning operations at intervals varying from three to 4 years (Parkes & Murphy 2003). There are published case studies of the effectiveness of possum control for reducing impacts on vegetation (e.g. Nugent et al. 2002; Urlich & Brady 2006); however, there are few examples assessing the effectiveness of extensive possum control with a controlled experimental design including replicated treatments (but see Nugent et al. 2010; Duncan et al. 2011).

In 2004, the New Zealand Department of Conservation (DOC) set up a management experiment to determine whether survival of tree species browsed by possums improved at sites with a history of extensive control every 4–5 years. The experiment was designed to test three hypotheses (Fig. 1): (1) There is a direct relationship between possum abundance and resulting damage: lower possum density results in (a) reduced levels of browse, (b) increased tree foliage and (c) lower mortality of possum-preferred species. (2) Impacts are heterogeneous because possums browse (a) more in patches with a large number (or biomass) of preferred tree species or conversely (b) less in patches with more non-preferred species (i.e. patch selection). There is (c) a higher likelihood of browse on larger trees that are more likely to have den sites and new foliage exposed to high levels of sunlight (i.e. tree selection). (3) Trees require a minimum amount of foliage to survive; trees browsed to below this threshold will not survive.

Figure 1.

Management actions and outcomes, processes relating to browse on ‘indicator’ trees, and available data. Trap-Catch Index (TCI) is a measure of possum abundance, Foliage Cover Index (FCI) is a measure of canopy biomass, Foliar Browse Index (FBI) is the proportion of leaves showing evidence of browse by possums. Tree size is indexed by diameter at breast height (DBH).

We analysed the data to test these hypotheses and to determine whether there are damage thresholds that could be used to specify when control of possums is required for environmental management.

Materials and methods

Field Methods

Paired sites, one subject to extensive control (treatment) and one with no control (non-treatment), were located at Coromandel (treatment: 2300 ha, 36°56′S, 175°37′E; non-treatment: 1200 ha, 36°58′S, 175°39′E) and Urewera (850 ha, 38°17′S, 177°11′E; 850 ha, 38°14′S, 177°11′E) in the North Island, and Haast in the South Island (1300 ha, 44°03′S, 168°47′E; 1300 ha, 44°06′S, 168°39′E). Each pair of sites had similar topography, altitude and forest composition (assessed using the Recce method; Hurst & Allen 2007). Extensive possum control was carried out at: Coromandel, 1995, 1999 and 2003 prior to the study and during the study in 2006; Haast, 1995–96 and 2000–01 prior to the study, but the next control operation was delayed until after the study period; and Urewera, 1993–94, partially in 1996–97, 1998–99 and 2002–03 prior to the study, and during the study in 2006–07.

Experimental protocols were based on techniques used by DOC to assess the abundance of possums (Trap-Catch Index (TCI); National Possum Control Agency 2010) and their impacts on tree canopies (Foliar Browse Index (FBI) and Foliage Cover Index (FCI); Payton, Pekelharing & Frampton 1999) and survival. Additional data were collected to characterize foraging patterns at a local scale because previous studies found significant site differences in possum impacts (Duncan et al. 2011). The experimental design included measurements of tree size using diameter at breast height (DBH), local availability of other food sources and possum browse at the patch scale (Fig. 1).

In each area, two tree species with foliage preferred by possums and one with foliage not preferred (Owen & Norton 1995; Sweetapple 2003; Sweetapple, Fraser & Knightbridge 2004) were selected as ‘indicators’ (Table 1) to assess the outcomes of extensive possum control. A criterion for selecting species was that they were relatively common and widespread throughout the area. Non-preferred species acted as experimental controls for non-selective disturbances such as storms, drought and earthquakes. Smaller numbers of a third highly preferred species were selected from sparse populations at Coromandel (Dysoxylum spectabile) and Haast (Fuchsia excorticata).

Table 1. Preferred and non-preferred tree species at each site, annual tree mortality at non-treatment and treatment sites [sample size n of trees that were tagged at the start of the study (Time1) and relocated during the remeasurement period (Time 2)]. Trees with unrecorded browse data (FCI and/or FBI), or that could not be relocated, at Time 2 were not used for analysis
AreaSpeciesAbbreviationPossumNon-treatment sitesTreatment sites P a
preferredMortality (%) n Mortality (%) n
  1. a

    Bold text indicates statistical significance (at the 5% level).

Coromandel Weinmannia silvicola WEISILYes0·182250·282130·525
Olearia rani OLERANYes2·172210·94238 0·014
Dysoxylum spectabile DYSSPEYes6·20840·0 21 0·008
Knightia excelsa KNIEXCNo0·252370·092060·359
Haast Weinmannia racemosa WEIRACYes0·02290·412470·957
Schefflera digitata SCHDIGYes15·5320012·16197 0·040
Fuchsia excorticata FUCEXCYes21·2010219·73 30·5
Nothofagus menziesii NOTMENNo0·602020·572120·5
Urewera Weinmannia racemosa WEIRACYes6·872180·11237 <0·001
Beilschmiedia tawa BEITAWYes0·332290·402490·621
Knightia excelsa KNIEXCNo0·112220·122110·5

Preliminary power analysis indicated that 200 trees of each species per site would be required to have an 80% chance of detecting a treatment effect in annual tree survival where the difference in annual mortality is ≥2% and the interval between measurements is ≥4 years (B. Reddiex unpublished data). Initial measurements (Time 1) and remeasurements (Time 2) were carried out in 2004 and 2009, respectively, at Coromandel and Haast, and in 2006 and 2010 at Urewera. Sampling transects, consisting of 5–15 plots at 50-m intervals, were located randomly in each site with randomly selected bearings. Transects and plot centres were marked and GPS coordinates recorded. Within 20 m of each plot centre, up to four trees of each indicator species at least 10 m apart where possible were permanently marked using numbered tags. Trees were included if they measured ≥10 cm DBH (1·4 m above the ground), and if the canopy was visible from the ground.

Browse damage was recorded using Foliar Browse Index (FBI) in five categories, 0–4, indicating that 0%, 1–25%, 26–50%, 51–75% or >75% of leaves, respectively, showed evidence of possum browse. Foliage cover was recorded using Foliage Cover Index (FCI), the fraction of sky occluded by leaves, observed from beneath the centre of the tree crown looking up. Occlusion was recorded in 10% categories, that is, 0·05 = 0–10%, 0·15 = 10–20%, etc. For each indicator tree, DBH, FCI and FBI were recorded at Time 1, and again at Time 2 for those trees still alive. At Time 2 the status (alive or dead) of each tree was recorded.

At each plot, the closest indicator tree to the plot centre was selected as the ‘focal tree’. The DBH, Alive/Dead status and FBI of all trees (with DBH ≥ 10 cm) within 5 m of the focal tree were measured (irrespective of species). These ‘neighbourhood trees’ were used to characterize a patch of forest that could be selected by foraging possums.

Possum Trap-Catch Index (TCI) was obtained at each site as near to Times 1 and 2 as possible following the standard protocol of randomly located lines within each area, each line consisting of 10 or 20 traps set for three nights (National Possum Control Agency 2010). TCI lines were located independently of the transects for measuring browse impacts. DOC measured TCI on treatment sites after each control operation; we used DOC trap lines that overlapped our study area to calculate mean TCI values (6–18 lines per site). In the non-treatment areas, we measured TCI using 7–10 lines per site, with some additional DOC measurements at Haast.

Analytical Methods

Although there was no possum control at the Haast treatment site between Times 1 and 2, we assumed it to be notionally ‘treated’ when testing for treatment effects. This is just within the likely period of reduced possum abundance, taking into account their expected recovery rate (Morgan & Hickling 2000).

Treatment effects on possum browse (FBI), canopy biomass (FCI) and tree mortality

Comparisons were carried out between treatment and non-treatment sites in each of the three areas, at Time 1 and Time 2 separately. Distributions of FBI values for each indicator species at each site and time were compared using Fisher's exact test for count data (Hypothesis 1a). To account for the low frequency of higher FBI values, we also compared the proportion of trees in the non-zero browse categories (‘any browse’) and the proportion of trees in FBI category 2 or higher (‘moderate-to-severe browse’). A one-sided binomial proportions test was used to determine whether the proportion of trees in each category was greater at the non-treatment site.

Distributions of FCI values for each indicator species at each site and time were compared using a one-sided t-test (Hypothesis 1b). The relationship between FBI and FCI was assessed using data pooled across treatment and area (where applicable) for each preferred species at Time 1.

Annual mortality of each tree species was compared between treatment and non-treatment sites using a one-sided t-test for the difference of two proportions (Hypothesis 1c). The observed annual mortality (m) for each tree species (by area and treatment) was calculated using N1(1−m)yrs=N2 where Nt is the number of trees alive in the sample at time t and yrs is the number of years between observations.

We examined whether tree mortality differed depending on FCI at Time 1 (Hypothesis 3). Due to the low number of recorded tree deaths, FCI values were grouped into three categories (low = <0·4; moderate = 0·41–0·7; high = 0·71–1·0) and pooled across treatment and non-treatment sites. Distributions of the FCI groups were compared, by species, between trees alive and dead at Time 2, using Fisher's exact test for count data.

Hierarchical models of possum browse and tree mortality

We used hierarchical models to generalize the relationships between browse and/or tree mortality and a range of covariates across tree species and sites. First, we modelled no browse, FBI = 0, vs. any browse, FBI > 0, on the possum-preferred ‘focal trees’ at Time 2 against the covariates: (1) Treatment/non-treatment, (2) PrefBio: sum of the biomass of preferred neighbourhood trees, excluding the focal tree (Hypothesis 2a), (3) NPrefBio: sum of the biomass of non-preferred neighbourhood trees surrounding the focal tree (Hypothesis 2b), (4) PrefEat: index of level of use of the forest patch by possums, calculated as the amount of preferred biomass consumed at each plot, that is, FBI × Biomass for each neighbourhood tree, summed across all neighbourhood trees in the plot (Hypothesis 2a), (5) DBH: index of size of focal tree (Hypothesis 2c) and (6) TCI: Trap-Catch Index at Time 2.

A complete set of covariate data were available from Time 2, and browse levels on the focal tree at Time 2 were assumed to depend on the current state, not the state 5 years ago. Each tree's biomass was calculated using an allometric relationship between DBH and biomass for New Zealand trees, where Biomass = 0·0406 × DBH1·53 (Richardson et al. 2009). Tree species additional to the indicator species were defined as being either preferred or non-preferred by possums based on published results (Owen & Norton 1995; Sweetapple 2003; Sweetapple, Fraser & Knightbridge 2004), possum browse from this study, and expert opinion (P. Sweetapple, pers. comm.).

For the seven main indicator species, both preferred and non-preferred, we investigated the relationship of annual mortality for the ‘focal trees’ against the covariates for their initial state and subsequent cumulative impacts: FCI at Time 1 (Hypothesis 3), DBH at Time 1 (Hypothesis 2c), treatment (Hypothesis 1) and average TCI from Times 1 and 2 (Hypothesis 1).

Browse and mortality were modelled using a Bayesian hierarchical approach, where for each covariate the coefficients for each tree species are distributed with a mean and variance from a higher-level ‘hyper-distribution’ (Royle & Dorazio 2008). This enabled us to model all species together, yet was flexible enough to allow the relationship between the covariates and FBI to differ between species. A range of models were specified separately for browse and mortality with various combinations of covariates mentioned above. The natural log of all continuous covariates (PrefBio, NPrefBio and PrefEat) was taken after adding a small value to accommodate zero values where applicable (1 for PrefBio and NPrefBio, 0·1 for PrefEat). To assist with convergence, covariates were standardized to have mean = 0 and SD = 1. Models were fitted using OpenBUGS 3·1 (Lunn et al. 2009), and compared using Deviance Information Criterion (DIC; Spiegelhalter et al. 2002).


Abundance of Possums

Mean Trap-Catch Index (TCI) was generally higher on the non-treatment sites in all years, the exception being Urewera in 2010 (Table S1, Supporting Information). There was large variation in TCI between sites, reflecting the differences in forest composition (and hence food availability and carrying capacity) between pairs of sites, as well as treatment effects within pairs and variability in TCI between years within sites.

Treatment Effects on Possum Browse, Canopy Biomass and Tree Mortality

For Hypothesis 1a, non-preferred species consistently had zero or near zero probabilities of suffering any browse in all sites (Fig. 2) and were never recorded as having ‘moderate-to-severe browse’. There was no evidence of a difference in the distributions of Foliar Browse Index (FBI) for all non-preferred species between treatments at initial measurement (Time 1) or at remeasurement (Time 2) (Fisher's P; Table S2, Supporting Information).

Figure 2.

Distribution of the Foliar Browse Index (FBI) for possum-preferred species (rows 1 and 2) and non-preferred species (row 3) at the paired non-treatment (light bars) and treatment (dark bars) sites at Coromandel, Haast and Urewera at Time 1 (2004 or 2006) and Time 2 (2009 or 2010). FBI categories: 0, 1, 2, 3 and 4 correspond to 0%, 1–25%, 26–50%, 51–75% or more than 75% of leaves showing evidence of possum browse. See Table 1 for full species names.

For all preferred tree species at Coromandel and Urewera, there was strong evidence for a difference in the distributions of FBI between treatment and non-treatment sites (Fig. 2 and Fisher's P; Table S2, Supporting Information), as well as differences in the probability of ‘any browse’ and ‘moderate-to-severe browse’ at both Times 1 and 2 (Table S2, Supporting Information). For Haast, there was moderate evidence of a difference in the distributions of FBI between treatment and non-treatment sites for Weinmannia racemosa and F. excorticata at Time 1, but no evidence of a difference for any preferred species at Time 2. There was moderate evidence that all preferred species in the non-treatment site had a higher probability of ‘any browse’ at Time 1, but weak evidence for W. racemosa only at Time 2, and no evidence for a higher probability of moderate-to-severe browse for any species at Time 1 or Time 2.

For Hypothesis 1b, most possum-preferred species had higher mean Foliage Cover Index (FCI) values at Times 1 and 2 on treatment sites compared with non-treatment sites (Fig. 3). Despite strong evidence for a positive effect of treatment on FCI for many species and/or sites, the magnitudes of the differences with non-treatment sites were often small (Table S3, Supporting Information).

Figure 3.

Distribution of the Foliage Cover Index (FCI) for possum-preferred species (rows 1 and 2) and non-preferred species (row 3) at paired non-treatment (light bars) and treatment (dark bars) sites at Coromandel, Haast and Urewera at Time 2 (2009 or 2010). FCI, which is the fraction of sky occluded by leaves, was recorded in 10% categories, i.e. 0·05 = 0–10% occlusion, 0·15 = 10–20% occlusion, etc. See Table 1 for full species names.

There was a negative relationship between FCI and FBI for nearly all the possum-preferred species, indicating a negative impact of browse on canopy health at Times 1 and 2: P< 0·001 for all species, except Beilschmiedia tawa (P = 0·756, Time 1; P = 0·018, Time 2) and Schefflera digitata (P = 0·411, Time 2; Fig. S1, Supporting Information). There was no detectable relationship for the non-preferred species (Knightia excelsa, Nothofagus menziesii) due to negligible levels of possum browse.

For non-preferred species, overall tree mortality was low (≤0·6%), with no evidence of a treatment effect. For preferred species, the results were mixed (Hypothesis 1c; Table 1). At Coromandel, mortality of Olearia rani (and the additional preferred species, D. spectabile) was significantly higher in the untreated area, yet nearly all Weinmannia silvicola trees survived in both areas. At Haast, there was a strong treatment effect on mortality of S. digitata, but no difference for W. racemosa (low mortality) or F. excorticata (high mortality). At Urewera, there was a strong treatment effect for W. racemosa, but no difference for B. tawa.

When pooled across treatment and non-treatment sites, there was strong evidence of a negative relationship between mortality and categorical values of FCI at Time 1 for all preferred species except W. silvicola at Coromandel and W. racemosa at Haast (Hypothesis 3; Table 2). At Haast, there was moderate evidence of a negative relationship between canopy biomass and mortality for the non-preferred species N. menziesii, although the number of trees with highest mortality (‘Low FCI’) was very small (= 7). (Mortality rates, by FBI, are shown in Table S4, Supporting Information)

Table 2. Annual tree mortality by FCI at Time 1, pooled across treatments. Number of trees shown in parentheses
AreaSpeciesaLow FCI (0–0·4)Moderate FCI (0·41–0·7)High FCI (0·71–1·0) P b
  1. a

    Non-preferred species are italicized. See Table 1 for full species names.

  2. b

    Bold text indicates statistical significance (at the 5% level).

CoromandelWEISIL2·6% (8)0·2% (392)0·0% (38)0·147
OLERAN4·7% (28)1·3% (406)2·5% (25) 0·008
DYSSPE16·3% (34)1·2% (51)0·0% (20) <0·001
KNIEXC 0·0% (9)0·3% (199)0·1% (235)0·389
HaastWEIRAC0·0% (2)0·2% (337)0·3% (137)0·637
SCHDIG18·5% (189)10·7% (197)3·9% (11) <0·001
FUCEXC34·9% (77)3·9% (28)NA (0) <0·001
NOTMEN 6·5% (7)0·5% (317)0·4% (90) 0·023
UreweraWEIRAC9·0% (114)0·8% (311)0·0% (30) <0·001
BEITAW3·9% (11)0·2% (279)0·2% (188) 0·011
KNIEXC 0·0% (2)0·2% (133)0·1% (298)0·526

At the species+site level, there was evidence of a threshold effect on annual mortality for browse and tree canopy cover (Fig. 4). A rapid shift to increased mortality occurred when the proportion of trees with moderate-to-severe browse was above 0·05, and when the mean FCI was below 0·5 (Fig. 4).

Figure 4.

Relationship between the annual mortality for indicator tree species and the proportion of trees with moderate-to-severe browse (FBI>1) at Time 1 (left) and mean foliage cover (FCI) at Time 1 (right) for all species+site combinations. The dotted lines indicate observed thresholds in levels of FBI and FCI where there is a rapid shift from near zero to high annual tree mortality.

Hierarchical Models of Possum Browse and Tree Mortality

The mean number of neighbourhood trees in each 5 m radius plot was 8·9 (95% CI = 2–18·7), with a mean of 4·3 (95% CI = 0–12) possum-preferred trees and 4·5 (95% CI = 0–13) non-preferred trees. The mean number of preferred trees per plot was similar, irrespective of the focal tree species; however, plots with S. digitata and O. rani as the focal tree had higher numbers of non-preferred neighbourhood trees (7·3 and 6·2, respectively), and subsequently a lower proportion of preferred trees (Fig. S2, Supporting Information).

The browse model with the lowest DIC included PrefBio, PrefEat, DBH and Treatment as predictors of possum browse on focal trees (Table S5, Supporting Information), consistent with Hypotheses 2a and c. Inclusion of Treatment and PrefEat gave the greatest reductions in DIC, followed by PrefBio and DBH. Including NPrefBio resulted in a worse model (as measured by DIC) than comparable models without it, contrary to Hypothesis 2b. The model with TCI at Time 2 performed much worse than the model with Treatment.

The probability of browse on focal trees increased with decreasing PrefBio, increasing PrefEat and increasing DBH (Fig. 5).There was lower probability of browse on focal trees at treatment sites for all species except S. digitata, consistent with the treatment effect observed for focal trees (Table S6, Supporting Information) but not with the observations for the complete sample of indicator trees (Table S2, Supporting Information).

Figure 5.

Mean coefficients (Beta) and 95% CIs for the hierarchical browse model with the lowest DIC value (Table S6), for each preferred species, and the average across species (mu). Beta < 0 indicates a negative relationship; beta > 0 indicates a positive relationship. See Table 1 for full species names.

The best model for annual mortality included FCI and Treatment, consistent with Hypotheses 1c and 3 (Table S7, Supporting Information). A strong negative relationship between mortality and FCI was apparent for all preferred and non-preferred species (Fig. 6). The relationship was weakest for S. digitata, which had the highest overall mortality, with deaths occurring over a wide range of FCI values. There was generally lower mortality on treatment areas (Fig. 6), but no evidence of a treatment effect for S. digitata when FCI was included in the model despite the significant difference observed with most indicator trees (Table 2).

Figure 6.

Mean coefficients (beta) and 95% CIs for the best hierarchical tree mortality model (Table S8), for each species and the overall mean (mu). See Table 1 for full species names.

The best model where DBH was included (‘Mortality = FCI+logDBH’) showed that larger trees with relatively lower FCI had a higher probability of mortality, which is consistent with Hypothesis 2c. However, the results were equivocal, including weak evidence for a negative relationship between mortality and size for S. digitata (Fig. S3, Supporting Information).


Outcomes of Possum Control and Impacts of Possums on Tree Species

A management experiment was used to test hypotheses derived from a process model for the impacts of possum browse in native forests in New Zealand (Fig. 1). The results show that extensive possum control at intervals of 4–5 years was typically effective in lowering possum impacts on preferred tree species. In general, in treated areas, there was less browse on preferred tree species (Hypothesis 1a), higher foliage biomass (Hypothesis 1b) and decreased mortality of at least one preferred tree species per area (Hypothesis 1c).

The negative relationship between the probability of any browse on a preferred tree and the amount of preferred biomass within a 5-m radius (Fig. 5) was contrary to Hypothesis 2a that possums will browse more in patches dominated by preferred tree species. The absence of a relationship between the probability of browse on preferred trees and the amount of nearby non-preferred biomass (Hypothesis 2b; Table S5, Supporting Information) suggests possums had no difficulty finding isolated preferred trees. The positive relationship between the probability of browse and tree size indicates that large trees may be important for possums as a food source (Hypothesis 2c; Fig. 5) and may contribute to the higher mortality rates of larger trees.

The annual mortality rates measured for W. racemosa spanned a similar range to the values reported by Bellingham, Stewart & Allen (1999) for sites with a long-term presence of possums. Their highest rate of 6·6% p.a. was recorded at Pureora, which matches the maximum rate recorded for W. racemosa in this study at the non-treatment site at Urewera (6·87% p.a.; Table 1). For all preferred tree species, there was strong evidence of a threshold in foliage biomass below, which trees have much reduced survival (Hypothesis 3; Fig. 4). As expected, mortality of non-preferred tree species was low at all sites, and the observed rates of 0·60% and 0·57% p.a. for N. menziesii (Table 1) were within the range of values reported for this species by Bellingham, Stewart & Allen (1999).

Additional Factors Affecting Assessments of Possum Impacts on Susceptible Trees

Levels of possum browse on all trees in the immediate neighbourhood of ‘focal trees’ provided data for the hierarchical browse model used to evaluate hypothesis 2 on patch and tree selection by possums (Table S6, Supporting Information). Unlike the hierarchical model of tree mortality (Table S7, Supporting Information), the hierarchical browse model was conditional on trees surviving to the end of the study. Consequently, where there was very high mortality, for example S. digitata at Haast, it is possible that the probability of browse was lower on the treatment site, but because a large proportion of browsed trees on the non-treatment site had died (Table 1), it appeared that browse was higher on the treatment site (Fig. 2).

Possum control did not take place as planned during the study period at the Haast treatment site. Increased possum abundance at this site is likely to be why treatment differences in browse at the initial measurement (Time 1) were not maintained at remeasurement (Time 2) for any preferred species (Table S2, Supporting Information), although we note the measured increase was small (Table S1, Supporting Information). Depending on control effectiveness and demographics at the site (i.e. single vs. double breeding within a season and/or compensatory breeding), possum populations may recover to their pre-control densities within 10 years (Veltman & Pinder 2001). Therefore, partial recovery would be expected with the 5-year period of this study. The discontinuation of possum control at Haast was unfortunate; however, difficulties with maintaining large-scale management experiments are not uncommon (Walters 2007). Had this also occurred at one of the other two areas, it is unlikely the effectiveness of extensive possum control could have been assessed adequately. Nevertheless, the results from Haast suggest that, to be effective, control operations need to be maintained so that possum populations remain suppressed to low densities.

Some unofficial possum control, for example fur harvesting, probably occurred at some sites. This might have contributed to the observed declines in average possum abundance on the non-treatment sites, and to increased variability of possum abundance and impacts within sites where public access was localized. Patchiness in the distribution of possums post-control can also occur at the scale of several hectares (Fraser & Coleman 2005), either through inadvertent gaps in coverage by aerial baiting or through reaggregation of surviving possums. It is possible that localized areas of high possum abundance may coincide with transects for monitoring trees, which can result in estimates of possum density being higher than the area-wide estimate.

The effect of treatment appears to be a better predictor of browse and mortality than possum abundance as measured by the Trap-Catch Index (TCI). Measurements of TCI are subject to sampling variation, especially when small numbers of lines are used. In addition, possum abundance is likely to vary with or without possum control, while variation in abundance will exist between sites due to forest composition (Efford 2000), so that equal TCI measurements from different sites may correspond to different levels of possum browse impacts (Payton 2000). As a result, using possum TCI to indicate when to carry out control may not be appropriate unless the appropriate TCI threshold has been tailored for a specific site.

Management Applications

Herbivore impacts on forests and woodlands have been assessed using indices of browse damage and canopy condition in many studies worldwide (e.g. Nugent et al. 2002; Makhabu, Skarpe & Hytteborn 2006), and a reduction in herbivore density usually results in reduced impacts (Kamler et al. 2010). This study supports the view that browse by an invasive herbivore can decrease canopy biomass and increase tree mortality rates and that herbivore control can have measurable benefits for susceptible tree species.

That we were able to measure differences in tree mortality between treatment and non-treatment sites over an interval of only 5 years shows that the effect size can be large. The results suggest that regular control of invasive herbivores is effective in decreasing tree mortality. However, effects may not persist long term (as evidenced by the reduced benefits at Haast where possum control did not continue as planned), and control at regular intervals is required.

A key finding is that such differences could be inferred using only the type of data typically collected during routine herbivore impact monitoring (e.g. browse damage and canopy cover), at a broadscale. These metrics can be measured relatively quickly, robustly and inexpensively to identify areas where high levels of tree mortality are expected to occur, negating the need for direct observations of mortality over long time frames.


Heterogeneity of herbivore densities and their impacts among sites and years is a characteristic pattern of plant–herbivore systems (Hone 2007). Our results support previous findings that equal herbivore densities may result in different impacts on the same species (Bee et al. 2009; Duncan et al. 2011). As a result, site and species differences must be considered when planning for herbivore control. Management actions can also be based on damage thresholds: tree browse and canopy cover can be measured relatively quickly and robustly using simple measures such as Foliar Browse Index (FBI) and Foliage Canopy Index (FCI), respectively. These metrics (FCI especially) can be used to rapidly identify sites where high levels of tree mortality are likely to occur, negating the need for direct observations of mortality over long time frames. However, managers need to be aware that observed mortality is a result of multiple interacting processes, and not just browse by invasive herbivores. Site- and species-specific information such as those collected in this study could be used in conjunction with a mechanistic browse model to estimate the proportion of mortality attributable to herbivores, and hence which sites will benefit most from herbivore control (e.g. Holland et al., in press).

Clearly, tree mortality is only one component of forest dynamics. Forest management also needs to consider the balance between mortality and recruitment, which can also be impacted by invasive herbivores (Tanentzap et al. 2009), and between the chronic impacts of herbivory and acute impacts such as abiotic disturbances (e.g. Hurst et al. 2011).


This work was funded by the New Zealand Department of Conservation (DOC) and the Animal Health Board. We thank DOC, Te Waimana Kaaku Tribal Executive and Ngāti Koura Tūhoe iwi for permission to access the sites and for help with the study. We thank the considerable efforts of all those that carried out the field work (N. Fea, D. Hurst, M. Bridge, R. Heyward, C. Brausch, K. Pullen, C. Stowe, S. Hough, S. Whitford and W. Chin, R. Clayton, K. Ladley, K. Borkin, P. Horton, F. Thomson, T. Thurley, J. Pari, T. Rochford, A. Perfect and D. Ruth), M. Robinson and H. De méringo for data entry, and C. Veltman and P. Cowan for valuable discussions and comments on earlier drafts.