What is the risk of overcollecting for translocation? An opportunistic assessment of a wingless grasshopper

Translocation is an increasingly used tool in conservation management, but there is a risk that source populations are overcollected. The risk depends critically on the detection probability and the source population size. We quantified this risk for a wingless grasshopper population in a patch of remnant habitat in suburban Melbourne that was condemned to be cleared for housing development. We collected ∼2000 grasshoppers in five samples spread over 1 month and used the results to estimate the initial population size (∼3400) with high confidence. Despite our perception of substantially depleting the population, we removed only an estimated 60%, and this relatively high fecundity (∼50 eggs per lifetime) annual species had recovered by the following year to near its original density. Wild‐to‐wild translocation is likely to be a low‐cost and effective strategy in the conservation of many invertebrates, and our findings highlight the feasibility of using natural source populations.


INTRODUCTION
Translocations can be a powerful tool in conservation biology to establish new populations and genetically rescue existing populations (e.g., Griffith et al., 1989;Weeks et al., 2011).Vertebrate translocations are increasing in frequency, and detailed protocols have been established to decide when and how to do them and how to monitor their success.Yet, a recent literature review of 292 vertebrate translocation studies found that only 11% of studies made any assessment of the impact on the source population (Mitchell et al., 2022) despite a long-term recognition of the need for such assessments due to the potential risks of overcollection (Kirkwood, 2013).Invertebrate translocations are less common (Seddon et al., 2014), even though the smaller size and higher fecundity of many invertebrates make translocation a highly attractive conservation tool for these animals (Bellis et al., 2019).
In 2021, we undertook a large-scale translocation of a wingless grasshopper, Vandiemenella viatica (Morabidae) (Yagui et al., 2023).This annual species has been declining around metropolitan Melbourne because of habitat loss.It is ecologically similar to the endangered Key's Matchstick grasshopper (Keyacris scurra) and thus our translocations served as a test case to establish methods for translocating similar endangered species.We sourced the grasshoppers from a patch of remnant woodland vegetation in outer-suburban Melbourne that had been sold for housing development.Thus, we collected without regard to maintaining the source population and indeed aimed to collect as much of this population as possible ("mitigation translocation" Bradley et al., 2022;Kirkwood, 2013) for translocation to other remnant and revegetation sites (Yagui et al., 2023).We estimated the initial population density from our capture data with a Bayesian hierarchical model.We used this information to assess the impact of our collecting on the source population and the effort we would have needed to remove it completely.We then resurveyed the population the following year after the next generation had matured.

Study species
The Larapuna matchstick, V. viatica, is a member of an endemic Australian grasshopper family (Morabidae), all of which are wingless (Key, 1976).It is found on low to medium forbs and shrubs throughout southern South Australia, Victoria, and coastal areas of Tasmania.It is an annual species, hatching in early-mid summer, overwintering as a late-stage nymph, and maturing around the beginning of September (spring), with the adult population disappearing in December (summer).

Study site and collection method
The study site (Figure 1) was a block of 8.7 ha remnant vegetation corresponding to the Ecological Vegetation Class (EVC) of Grassy Dry Forest (EVC 22), located in the suburb of Diamond Creek in northeastern Melbourne.We surveyed 10 patches of suitable grasshopper habitat (Figure 1) summing to approximately 6.3 ha.We visited the site five times in August and September of 2021 under sunny weather conditions and resurveyed in the same months in 2022 (Table 1).The grasshoppers were late-instar nymphs (females) or adults (some males) at this time.
They are usually detected when they hop, and to collect the grasshoppers with a pooter, we walked around slowly, scanning the ground immediately at our feet, and gently disturbing the vegetation.One of us was initially inexperienced at collecting the species and thus was counted as 0.5 people on day 1, 0.75 on day 2, and 1 person thereafter.

Parameter estimation
Detection of individuals can be modeled as a Bernoulli trial where the number of captures c in each survey time t can be considered as drawn from a binomial distribution c t ∼ Binom(p t , N t ) with N t the population size at time t and p t the probability of detection at time t (Royle & Dorazio, 2008).The initial population size N 0 is a Poissondistributed latent variable with a mean and variance of λ, that is, N 0 ∼ Poiss(λ) and, because we sampled without replacement, for t > 0:   =  0 − −1 ∑ 0   (see also Smart et al., 2020).We fitted models to estimate N 0 and p t using a Markov chain Monte Carlo (MCMC) Gibbs sampler approach in JAGS v4.3.1 via the rjags package v4.14 in R v4.3.1.We ran three model chains for 30,000 iterations, with the first 20,000 iterations discarded as a burn-in to allow convergence of the MCMC chains.We thinned the remaining samples by a factor of 2, resulting in 5000 samples per chain for postprocessing.We calculated the potential scale reduction factors to assess convergence (Gelman & Rubin, 1992).The time to remove the entire population, , was then determined as: where   is the proportion of remaining individuals, set to 1/N 0 (Smart et al., 2020).
F I G U R E 1 Study site and approximate collection locations with an indication of relative abundance at these locations.

RESULTS
The number of grasshoppers collected and survey effort each day is summarized in Table 1 All MCMC chains converged (potential scale reduction factors ≤ 1.02).The total population size prior to collec-tion (N 0 ) was estimated as 3371 (95% CI 3052-3796) with a mean per-individual capture probability of 0.17 (95% CI 0.14-0.20),which in our study combines probability of encounter (we did not sample the entire area each time) with probability of detection.The number of days to remove all but one of the grasshoppers  was estimated to be 45.5 (95% CI 35.9-52.9).Accounting for the mean number of person-hours per day that we collected (14 h), this amounts to 26.5 days of continuous collecting.

DISCUSSION
According to the IUCN guidelines for reintroductions and other conservation translocations, any potential adverse effects on the source population should be evaluated (IUCN/SSC, 2013).Most threat assessments of endangered insects, including orthopterans, tend to focus on external threats.These may include habitat damage and the impact of invasive species that directly interact with the threatened species or alter their habitat, such as the impact of invasive plants in the case of insect herbivores (Hochkirch et al., 2008;Samways & Lockwood, 1998;Wagner & Van Driesche, 2010).Populations of insects are typically viewed as having life histories that allow for a rapid recovery once weather conditions and resources are available (Dempster, 1963).This makes many endangered insect species particularly suitable for translocations given that populations may readily establish in areas where suitable habitat has been established and adequate management is in place, and it means that source populations can recover quickly.Despite this, there are very few studies assessing the impact of intensely collecting threatened insects on the persistence of source populations, although removal sampling has been used to assess the overall health of threatened insects, including orthopterans (e.g., Schori et al., 2020).In contrast, there is substantial research on this issue in the agricultural sector, where pest insect collection has been widely used in attempts to suppress populations and eradicate incursions in new areas.In these mass trapping attempts, insects are usually lured to traps with various semiochemicals and suppression/eradication is typically successful only when the pests are at a low density in isolated populations (El-Sayed et al., 2006).Due to their flightless nature, Larapuna grasshoppers often exist as a series of isolated and genetically differentiated populations, even when these are in close geographic proximity (Hoffmann et al., 2023).However, their relatively high density and reproductive output likely provides Larapuna grasshoppers with the capacity to recover rapidly and avoid population collapse.Other threatened orthopterans also show quite rapid recovery of populations when the habitat is suitable (Hochkirch et al., 2008).
The high local abundance at our site, at around one grasshopper every 19 m 2 (but with local densities up to ca. 20 m −2 ), meant we could obtain accurate estimates of population size and probability of detection.The statistical removal model rapidly gives robust estimates once removal has begun and can be used in real time to guide stopping decisions.Given that we left ∼1300 individuals, that they have an even sex ratio (Yagui et al., 2023) and lay around 50 eggs per female (unpublished data), we could expect > 32,000 recruits in the next generation.Consistent with this, by the following year (and hence after one generation), population density appeared to have fully recovered (Table 1) (though our sampling was not as extensive in the latter year).Thus, for this grasshopper and similarly fecund and abundant invertebrates, the risk of overcollecting for translocation appears low.
A locally common species such as V. viatica could be used to experimentally test the effects of different collecting strategies, such as changing the ratio of males versus females collected (White and colleagues mostly collected males for their cytological work and did not report local extinctions) or collecting at different life-cycle stages.The extremely high density of Larapuna matchsticks at our study site and others we have surveyed makes them a functionally important component of the ecosystem as prey for other animals.Lizards (Lampropholis guichenoti) were commonly observed during the study and Larapuna matchsticks were the only obvious insect prey available for them.On other occasions, we observed magpies and spiders feeding on them.Invertebrate translocations can have long-term benefits for insectivorous species at the translocation sites, but consideration must be given to the short-term effects of collection for insectivores at the source populations.
The Bayesian hierarchical model assumes that we could have had successes up to the estimated current population size, N t , in a given trial.The area of the property that we searched in each trial varied and thus our estimate of the per-individual capture probability is the combination of encountering an individual and then of detecting it.We can alternatively model the collection process as a hyperbolic functional response foraging following Kooijman (2010, p. 33).In this model, the capture rate ḣ = ḣ  ( ḣ ∕ Ḟ+) , where ḣ is the maximum capture rate, Ḟ is the searching rate (#/m 2 ), and N is the density of grasshoppers (#/m 2 ).We can maximally capture Larapuna matchsticks at around ḣ = 12 min −1 , moving at around 1 km h −1 and scanning an area of ∼1 m 2 , leading to Ḟ = 16.7 m 2 min −1 .Given our estimate of N 0 and our total search area of 6.3 ha, the initial density of grasshoppers was 0.05 m −2 .Modeling collection using the functional response equation for ḣ above, we obtain the dashed curve in Figure 2 which suggests that the number of person-days to remove all but one of the grasshoppers is 21.4 (cross in Figure 2) corresponding well with our estimate of  of ∼19 person-days.
A more detailed analysis of this kind, better quantifying search areas and times, could help untangle detection and encounter probabilities.
Our findings add to the reasons for a high potential for success in using translocation to recover invertebrate populations that are locally common but threatened by habitat loss and fragmentation (Yagui et al., 2023).We could safely collect a large number of individuals from the source population at the maximum possible rate without threatening the population, due to the combination of fecundity, generation time, and detectability of our study species.Our study species has high fecundity and a short generation time compared to most vertebrates but is not particularly extreme for an invertebrate.There is a general trade-off between fecundity and vagility in grasshoppers, and in insects more generally due to the costs of building and maintaining flight muscles (Chang et al., 2021;Tigreros & Davidowitz, 2019).Thus, it may be that many insects threatened due to their limited colonizing ability are inherently of high fecundity and hence more amenable to translocation.
Although the risks of translocation programs to source populations may in general be lower for invertebrates, each translocation must be assessed on a case-by-case basis.Where sample sizes are sufficient, the Bayesian hierarchical modeling method can be employed to dynamically assess population size and detectability in similar systems.This will allow adjustments to be made "on-thefly" to minimize impact on the source population.Such adjustments may include returning a certain proportion of individuals captured, varying the sex ratio or stage (egg/nymph/adult) collected or ceasing collection from that population, thereby enhancing the overall success of translocations for conservation.

A C K N O W L E D G M E N T S
This work was supported by an Australian Research Council Discovery Grant DP190100990 to MRK and AAH.We thank James Maino for facilitating access to the property and Gayee Maino for her help in the field.This work was undertaken during the COVID pandemic when travel restrictions were in place, and we obtained travel authorizations from the Victorian Government to undertake the work.We declare no conflict of interest.All data are available within the paper and the code for the paper can be found at (TBA).

D ATA AVA I L A B I L I T Y S TAT E M E N T
The data that support the findings of this study are available in their entirety within the manuscript.All codes for the analyses are provided as Supplementary Material.

R E F E R E N C E S
Estimation of population decline with collecting time from a Hollings type II foraging model taking density, effort, and maximum harvest rate as arguments.Circular symbols represent the estimated remaining grasshoppers at each time from the Bayesian hierarchical model.Dashed line represents one remaining individual which meets the estimate population decline at 21 days compared to the Bayesian model best estimate of time to complete harvest ( = 19 days) indicated by the triangle.
Summary of grasshopper collections and resurveys.
TA B L E 1