Evaluation of averted loss gains under Victorian biodiversity offset policy

Biodiversity offsetting seeks to balance losses in biodiversity from development with offset actions that generate biodiversity gains. In Victoria, Australia, native vegetation offset policy aims to achieve No Net Loss (NNL) in biodiversity, meaning gains from offsetting must be at least equivalent to development losses. Gains may be generated from measures that protect the existing biodiversity from loss (averted loss gains), and those that restore or enhance the biodiversity (improvement gains). However, previous work has suggested the Victorian offset policy overestimates averted loss gains, allowing a larger development impact for the same offset. Here, we undertake a detailed examination of five impact‐offset exchanges permitted under Victorian offset policy in 2019 and 2020, comparing gains under the policy to those resulting from plausible counterfactuals. We find that the projected gains from averted loss were likely to be overestimated by between 3 and 240 times, compared to the counterfactuals. If this level of over‐estimation holds throughout the scheme, it would mean that up to 86% of offsets purchased do not compensate for permitted impacts to native vegetation, thwarting Victoria's NNL policy objective. We discuss the key policy issues and make recommendations to address over‐estimating averted loss gains.


| INTRODUCTION
Globally, biodiversity is declining faster than at any time in human history (IPBES, 2019a).This loss is largely attributed to human activities, with urban, industrial, and agricultural development being major drivers (IPBES, 2019b;McLellan et al., 2014).A range of policies have been implemented by governments to help reduce this loss, including biodiversity offsetting.Biodiversity offsets are regulatory or voluntary instruments that seek to balance losses from development impacts through the generation of measurable "gains" in biodiversity in another location that are additional to existing land management requirements and conservation targets (BBOP, 2012a;IUCN, 2014;Maron, Gordon, et al., 2015).They are used in over 80 different countries worldwide, including separate schemes in all Australian states and territories (Maron et al., 2018;Maron, Bull, et al., 2015).
In this paper, an "impact" refers to an action that causes losses in biodiversity (e.g., development; see S2, Supplementary Information).Offset policies generally combine the loss from a development impact and the gain from an offset to achieve at least No Net Loss (NNL, see S1 and S2) of biodiversity within some time horizon (Gibbons et al., 2015;IUCN, 2014).NNL is achieved when the gain in biodiversity from an offset is equivalent to the loss from the development impact (Bull et al., 2016).Evaluating this impact-offset exchange, and the achievement of NNL, necessitates a biodiversity accounting system to measure losses and gains in the same units, so their equivalence can be determined (Peterson et al. 2018).Due to the multi-dimensional nature of biodiversity, biodiversity metrics can be complex and may target specific biota and/or vegetation communities, as defined by the scope of the policy (Maron et al., 2016).
In Victoria, the State Government introduced biodiversity offsetting into policy in 2002 (DNRE 2002;DEPI, 2013;DELWP, 2017a), making it the first offsetting scheme in Australia and one of the earliest in the World, along with the US Compensatory Mitigation and Wetland Banking program (NRC, 2001).Under Victorian policy, offsets are required to achieve "no net loss in biodiversity as a result of the removal, destruction or lopping of native vegetation" (DTP, 2023, Clauses 12.01-2S, 52.16, and 52.17).Like some offsetting policies, gains from offsetting in Victoria can be generated from measures that both protect the existing biodiversity from loss (averted loss gains) and from the restoration or enhancement of the existing biodiversity (improvement gain) (BBOP, 2012b;DELWP, 2017b;Dorrough et al., 2019;Gardner et al., 2013).Studies show that biodiversity is in continual decline in Victoria (CES, 2013;CES, 2018;DSE, 2008).Under Victorian offset policy, the averted loss gains assume that the existing biodiversity at a given site would also decline without an offset, and therefore, any losses averted are attributable to the offset itself.The assumed trajectory of biodiversity that would have occurred without the offset taking place is referred to as the "counterfactual." Figure 1 illustrates how losses and gains in biodiversity may be measured against a declining baseline value of biodiversity, shown as either a debiting (for loss) or crediting (for gain) baseline (see S2) (Bull et al., 2016;Maron et al., 2018).While the development impact is typically immediate and certain, gains from offsetting are predicted to occur over an offset management period and can carry significant uncertainty (Bekessy et al., 2010;Dorrough et al., 2019).Multiple assumptions must be made about the projected decline in biodiversity without the offset (averted loss) and the improved biodiversity value with the offset (improvement).If these projections are implausible, equivalence between the impact and the offset, and an NNL outcome, may not be achieved (Gordon et al., 2015;Maron, Bull, et al., 2015;Maseyk et al., 2020).Maron, Bull, et al. (2015) examined the assumptions of gain from averted loss in biodiversity offsetting policies across all six Australian states and two territories, finding that on average crediting baselines assumed rates of loss of 0.4%-4.2%per annum that were up to five times higher than recent rates of woody native vegetation extent loss in Australia (estimated to be 0%-0.5% per annum).They found that Victorian offset policy assumed one of the largest background declines in biodiversity at 2%-4.2% per annum.Maron, Bull, et al. (2015) did not provide a detailed review of each offset policy, and the study was limited be the availability of data on losses in native vegetation condition and non-woody native vegetation.
More recently, the Victorian Auditor-General's Office (VAGO) completed an audit of offsetting native vegetation loss on private land in Victoria (VAGO, 2022).The VAGO concluded Victoria is not achieving its NNL objective, citing illegal clearing, inadequate support for Local Government Areas (LGAs) implementing the policy, limitations of biodiversity assessment tools, and inadequate monitoring and reporting by DEECA as the key policy  issues.Gains from offsetting that are not an accurate reflection of what occurs with and without the offset also have implications on Victoria's achievement of NNL.
Here, we undertake a detailed review of the projected gains for averted loss of biodiversity under five impactoffset exchanges that were permitted under current NNL offset policy in Victoria, Australia, (DELWP, 2017a) to understand the extent to which gains in these offsets are overestimated and the implications of this for each impact-offset exchange.We identify the gains awarded for averted loss in each offset and compare them against available data on the background declines of native vegetation condition and extent in Victoria (from which we can estimate counterfactuals).By examining real-world impact-offset case studies in detail, we can identify inconsistencies between the counterfactuals and the application of gain scoring methodology in Victorian policy, and examine whether the NNL policy objective was likely to have been achieved through each impact-offset exchange.

| METHODS
To review the plausibility of gains from averted loss in Victoria, we provide a brief overview of Victoria's NNL offset policy and recent market activity, and review five impact-offset exchange case studies permitted under this policy between March 2019 and May 2020.We identify the predicted gains from averted loss from each offset that contribute to achieving NNL under Victorian offset policy, which provides the assumed trajectories of biodiversity without the offsets.We compare these trajectories against two available counterfactuals on the average background decline in native vegetation condition and extent in Victoria.
In this study, we do not review the development impact and determination of loss, or the plausibility of projected gains for improvement in biodiversity, under each impact-offset exchange presented in the paper.In O'Brien and Gordon (2020), we discuss the current limitations to a meaningful evaluation of improvement gain under Victoria's offsetting policy, including the incompatibility of Victoria's offset gain scoring metrics and native vegetation assessment methodologies.

| Victorian offset policy
The Victorian State Government first introduced biodiversity offsets into policy in 2002 to compensate for unavoidable impacts to native vegetation (NRE 2002).The Victorian offset policy, its objective, assessment methodology, and biodiversity metrics, have been adjusted periodically since its first introduction.Native vegetation loss (from the development impact) and gain (from the offset) in the current policy are measured in General and Species Habitat Units (HUs) (DELWP 2017a(DELWP & 2017b; see Box 1), where native vegetation is classified as either a patch or scattered tree (see S2). Box 1 describes how General and Species HUs of loss and gain are determined in accordance with the policy.A fundamental component of these metrics is the use of habitat hectares (HHa): a site-based measure of native vegetation condition and extent, where the habitat score represents the current condition of vegetation relative to a benchmark for the relevant native vegetation community type, known as an Ecological Vegetation Class (EVC) in Victoria (DEECA, 2023c;DSE, 2004;Parkes et al., 2003).Further details are provided in the Supplementary Information (see S2) and O'Brien and Gordon (2020).
HUs of loss depend on the current condition (habitat score) and extent (hectares) of native vegetation to be impacted (HHa of loss, Figure 1a, Equation (1)).HUs of gain depend on the projected averted loss and improvement in condition (gain score) and extent of native vegetation for the offset (HHa of gain, Equation ( 2)) over a 10-year offset management period (t 2 -t 1 , Figure 1b).Via the General or Species Landscape Factor (see S2), they BOX 1 Victorian offset policy biodiversity metrics (DELWP 2017a(DELWP & 2017b)).
General or Species HUs required to offset an impact  Gordon, 2020).General and Species offsets must also meet the offset attribute requirements (see S3), including providing for at least the same number of large trees being impacted, and locating a General offset within the same Catchment Management Authority (CMA) or LGA as the impact.
Multipliers are applied to determine the General and Species offset requirements (Box 1).A multiplier of 1.5 or 2 is applied to general offsets and species offsets, respectively.The use of multipliers in offset policy is often intended to manage uncertainty, by increasing the offset requirement to balance risks of the predicted gains not being achieved (Bull et al., 2013;Bull et al., 2017).The Victorian offset policy states that a "higher multiplier is applied to species offsets to account for the greater loss in biodiversity value (where there is a significant impact to rare or threatened species), and the increased risk to biodiversity if the gains are not achieved."(DELWP, 2017a, p. 16).How the multipliers of 1.5 and 2 have been determined is however not explicit in the policy (DELWP, 2017a).

| Victorian gain scoring methodology
Equation (2) in Box 1 shows how General or Species HUs are generated from an offset.A HHa of gain is a site-based measure of the projected gain in biodiversity value (the "gain score") from an offset over its extent (hectares).The landscape factor is drawn from statewide modeled biodiversity and threatened species data and does not incorporate assumptions of gain from offsetting.The gain score is therefore the most important component in Equation ( 2) when evaluating the plausibility of gains from offsetting under Victorian offset policy.Under the policy, the gain score is the aggregate of points awarded to an offset under four types of gain: prior management gain, security gain, maintenance gain, and improvement gain (see S2) (DELWP, 2017b;O'Brien & Gordon, 2020).Table 1 summarizes each type of gain as per Victoria's Native vegetation gain scoring manual-Version 2 ("gain scoring manual") (DELWP, 2017b).This includes the range of points available under each type according to the offset land tenure and land use entitlements, in-perpetuity protection, offset type, the habitat (condition) score at the beginning of the 10-year offset management period, and the proposed management activities to be implemented over the 10-year offset management period.
As described in the Victorian gain scoring manual, the maintenance gain is awarded for averting losses in native vegetation condition, while the improvement gain is awarded for improving native vegetation condition (Table 1).Although not stated explicitly, the prior management and security gains are also awarded for averting loss in native vegetation from offsetting.Prior management gain rewards landowners on freehold land for maintaining native vegetation at an offset site prior to the offset occurring (Table 1), which, by definition, may not be attributable to the offset itself.The security gain is awarded for increasing the administrative protection on the land title to permanently restrict allowable land uses in the offset to conservation activities only.In addition to the maintenance gain, the security gain therefore also accounts for averted loss in native vegetation condition, and potentially native vegetation extent, from foregoing existing land use entitlements.The maintenance, prior management, and security gains collectively represent the gains from averted losses under Victorian offset policy.
It is important to note that the Victorian gain scoring manual is not explicit regarding the extent to which the prior management and security gains reflect averted losses in vegetation condition and/or averted losses in vegetation extent (DELWP, 2017b).Our analysis of all three averted loss gains therefore includes a review of the background decline in both native vegetation condition and extent in Victoria.

| The Victorian offsets market
The Victorian Native Vegetation Credit Register (NVCR) regulates and administers trading of General and Species HUs between the offset credit owners and purchasers (e.g., permit holders) seeking to meet their offset requirements under Victorian offset policy for permitted impacts to native vegetation (DELWP, 2017a).As of June 30, 2021, there were 388 offset sites registered on the NVCR protecting 19,843 hectares of land, with new sites being registered each year (see S4; DELWP, 2021).Figure 2 shows the growth in sales and value of the Victorian offsets market since 2008.In 2022, the NVCR administered 1056 offset trades from registered offset sites, for a total value of $19,010,911.84(exc GST).These trades included a total of 128.974General HUs and 31.703Species HUs, with 1790 large trees (DEECA, 2023a).

| Impact-offset exchange case studies
The five impact-offset exchange case studies presented in this study were made available at the T A B L E 1 Types of gain available on freehold and crown land (DELWP, 2017b).All five case studies involved the exchange of General HUs to offset a permitted loss of native vegetation between March 2019 and May 2020, for a total value of $78,637 (exc GST).They do not reflect all the possible loss-gain scenarios under the Victorian policy, however, they are five examples of a loss-gain exchange from five of the 10 CMA areas in Victoria: North Central, Goulburn Broken, Corangamite, Port Phillip and Westernport (now known as Melbourne Water) and West Gippsland (Figure 3).Within proximity to Melbourne, 76% of all General HUs sold since 2017 (when HUs were introduced under the 2017 Guidelines) were purchased from offsets in these five CMAs (DEECA, 2023a).

Type of gain
Table 2 summarizes the native vegetation attributes of the impact and offset(s) for each case study, with one impact involving two offsets.Further information on the case studies is provided in the supplementary information (see S5). Table 2 shows how the equivalence between the General HUs of loss and the General HUs of gain (in bold) was determined in accordance with the 2017 Guidelines.In these case studies, impacts to native vegetation occurred on public and private land with all corresponding offset sites being located on private land.
All five case studies relate to the loss and gain of General HUs comprising woody native vegetation at the impact and offset sites and met the "like-for-like" offsetting requirements in accordance with Victorian offsetting policy (see S3).Note that these "like-for-like" requirements do not include a requirement to balance impacts to native vegetation with offsets comprising the same vegetation communities (EVCs) or vegetation types (e.g.patch or scattered tree) (DELWP, 2017a).Accordingly, the EVCs impacted were not represented in the corresponding offset in any of the case studies presented in this paper (Table 2).In case studies 2, 3, and 4, the scattered trees impacted were also not represented in the offset.Case study 2 involved two offsets (labeled 2.1 and 2.2) to meet the like-for-like requirements: one to meet the General HU requirement (0.184 General HUs) located in the Port Phillip and Westernport CMA, and another to meet the large tree requirement (0.001 General HUs with seven large trees) from an offset in the Goulburn Broken CMA.This is a suitable offsetting option under Victorian policy (DELWP, 2017a).

| The counterfactual
All case study offsets presented in this paper are located on private land and comprise gains in woody native vegetation types only (Table 2).As the Victorian gain scoring manual is not explicit regarding the extent to which gains from averted loss reflect avoided losses in vegetation condition and/or avoided losses in vegetation extent, counterfactuals are required for background decline that includes both the condition and extent of woody native vegetation on private land.Accordingly, two counterfactuals have been estimated based on Victorian Government data on native vegetation decline: (1) the state average background decline in (woody and nonwoody) native vegetation condition on private land in Victoria; (2) the state average background decline in woody native vegetation extent in Victoria.These two counterfactuals provide estimates for what would have occurred without the offsets taking place, and therefore, an estimate of what loss was averted due to the offset itself.More detail on each counterfactual is given below.We note this data does not account for site-specific offset details, such as location (e.g., peri-urban or rural), other vegetation attributes, land use histories, and other factors that could alter the counterfactual.The counterfactuals presented in this paper are average estimates over the state, and therefore the estimates of vegetation loss averted from each offset are representative of the average averted loss over many offsets.Other limitations on the available data used for each counterfactual are outlined in the supplementary information (see S6).

| Counterfactual 1 (average native vegetation condition decline)
Table 3 summarizes the estimated rates of background decline in native vegetation condition on private land across Victoria.This data was drawn from satellite imagery and land use data since 1994, and was published by the Victorian Government (CES, 2018;DELWP, 2020;DSE, 2008).Estimates from all periods show that remaining native vegetation on private land is either declining or at risk of declining.It was reported that observed declines were predominantly due to native vegetation condition loss from existing entitled land uses, unmanaged threats beyond legislative obligations (e.g., environmental weeds), removal of native vegetation exempt from requiring a permit (e.g., fences and fire protection), and grassy native vegetation extent loss from conversion to more intensive agriculture (DSE, 2008;DELWP 2015;CES, 2018;DELWP, 2020).Importantly, losses of native vegetation that were authorized under a formal regulatory and/or planning process and subject to offsetting requirements were considered to result in a neutral outcome for native vegetation under these estimates, as were natural and semi-natural fluctuations (e.g., wildfire, timber harvesting, and passive regeneration), and therefore, were not included in the decline rates under this counterfactual.The increased losses in native vegetation estimated on public land in 2015 were partly due to the inclusion of impacts from controlled management regimes such as recently planned burns (DELWP 2015).
Additional unpublished background data relevant to these published reports were also sourced from the then Victorian Department of Environment, Land, Water and Planning (DELWP) to support these estimates for the purposes of this study ( personal communication, Biodiversity Division, DELWP, 2021).The Victorian Government acknowledges that these estimates include broad assumptions on land use and management, have poorly quantified levels of uncertainty, and estimate that the figures drawn from the published reports may be under or overestimated by up to 20%.The data also contains other limitations for use as a counterfactual in this study, including not differentiating between woody and nonwoody condition loss.As this is the only available data on native vegetation condition decline in Victoria, we used the data with a range of assumptions to provide a counterfactual estimate of background decline for this offset evaluation.These assumptions are detailed in the supplementary information (see S6).

| Counterfactual 2 (native vegetation extent decline)
Table 4 summarizes the estimated rates of background decline in woody native vegetation extent in Victoria over two recent 10-year periods.This data was published by the Victorian Government and used satellite imagery to estimate native vegetation extent loss in five-year intervals since 1990 (CES, 2018).The data differentiates native vegetation extent loss between different vegetation types, including woody and non-woody native vegetation.It does not however differentiate between woody native vegetation extent loss on public versus private land, nor discriminate between the different types of loss under regulatory and planning processes (e.g., entitled land uses, illegal clearance, or permitted losses) that may or may not be subject to offsetting requirements.For the purposes of this study, we conservatively assume all woody native vegetation extent loss occurs on private land and is not subject to offsetting.This is likely to over-estimate background decline in woody native vegetation extent on private land in Victoria resulting in larger estimates of gain from avoided loss (see S6).

| RESULTS
Table 5 provides a summary of the gain values awarded to each case study offset under the averted loss gains (including prior management, security, and maintenance) and improvement gains.These values have been drawn from the offset assessment reports and gain calculations prepared prior to the commencement of each offset.In accordance with the Victorian gain scoring manuals (DELWP, 2017b;DSE, 2006), these gains are relative to the habitat (condition) score of each offset at the beginning of its 10-year offset management period, existing land use entitlements and the proposed offset management targets to be achieved during this period.These T A B L E 3 Estimated per annum loss in woody and non-woody native vegetation condition on private land in Victoria (DSE 2008;CES 2018;DELWP 2020).

2015 2020
Native vegetation condition loss in habitat hectares per annum values are summed to determine the total Gain Score for each offset (Table 5).
The averted loss gains assume that, without each offset, the habitat (condition) score will decline over the 10-year offset management period by the same number of gain points awarded in Table 5.Under the combined total for the three averted loss gains, the assumption is that the habitat score for each case study offset would have declined by 13.8-25.2habitat points, or 23.4%-31.1%,over the same 10-year period without the offset (Table 5).If this rate of decline was consistent over the 10-year period, this assumed background decline equates to a loss of 1.38-2.52habitat points, or 2.3%-3.1%,per annum.
While the gains awarded for averted loss imply an averted loss in native vegetation condition (being a proportion of the habitat score), the Victorian gain scoring manual does not explicitly state whether the prior management or security gains reflect an averted loss in condition and/or extent (DELWP, 2017b).Accordingly, our analysis of the averted loss gains from case studies 1-5 (as summarized in Table 5) reviews the averted loss gains against two available counterfactuals: (1) the background decline in (woody and non-woody) native vegetation condition on private land in Victoria (Table 3, Figure 4); and, (2) the background decline in woody native vegetation extent on public and private land in Victoria (Table 4, Figure 5).
Figure 4 compares the rates of assumed decline in native vegetation without the offsets (averted loss) from case studies 1-5 with Counterfactual 1 (the estimated background decline in native vegetation woody and nonwoody conditions on private land, Table 3).The assumed gain from averted loss under the maintenance, security, and prior management gains from case study offsets 1-5 are 2.8-5.0 times (276%-504%) higher than Counterfactual 1.The projected gains from averted loss under the maintenance gain alone are however more consistent, with the maintenance gain for case studies 1, 2.2, 3, and 5 being approximately 8%-80% higher, and the maintenance gain for case studies 2.1 and 4 being approximately 40%-60% under Counterfactual 1.
Figure 5 compares the rates of assumed decline in native vegetation without the offsets (averted loss) from case studies 1-5 with Counterfactual 2 (the background decline in woody native vegetation extent on public and private land, Table 4).Figure 5 shows that, even with the conservative assumption that all the loss of woody native vegetation extent during the 1995-2005 and 2005-2015 periods is assumed to have occurred on private land in Victoria, and was not subject to offsetting, the averted loss gains (prior management, security, and maintenance gains) in case studies 1-5 over the 10-year offset management periods remain 33-239 times higher than Counterfactual 2. Under the maintenance gain alone, the assumed gains from averted loss are still 5-85 times higher than this Counterfactual on native vegetation extent decline.
Notwithstanding the uncertainties associated with Counterfactuals 1 and 2, Figures 4 and 5 illustrate significant discrepancies between the background decline in native vegetation in Victoria and the assumed averted loss from the case study offsets presented in this paper.
Tables 6 and 7 compare Counterfactuals 1 and 2 with the averted loss and total (including improvement) gains of each case study offset and calculate the percentage by which each total gain exceeds each counterfactual.Tables 6 and 7 show that the projected gains in native vegetation from case study offsets 1-5 are over-estimated by approximately 39-69% when using Counterfactual 1 (average native vegetation condition decline in Victoria, see Table 3) to estimate averted loss from each offset, or by 60-86% when using Counterfactual 2 (average native vegetation extent decline in Victoria, see Table 4).

| DISCUSSION
In Victoria, biodiversity offsets are required to achieve "no net loss in biodiversity as a result of the removal, destruction, or lopping of native vegetation" (DTP, 2023).Gains in native vegetation from offsetting may be generated from averting loss, as well as improving biodiversity, at an offset site.
In this study, we identified the four types of gain (prior management, security, maintenance, and improvement) awarded to offsets in five impact-offset exchange case studies permitted under Victorian offset policy.We evaluated the gains attributed to averted loss (prior management, security, and maintenance) against two available counterfactuals: the average loss of native vegetation condition (1) and woody native vegetation extent (2) on private land in Victoria.There are a range of limitations to the data available to estimate background declines (see S6) and we were constrained to use average loss rates across Victoria to inform our counterfactuals, whereas the counterfactuals for any given location could vary from these average rates.Thus, the counterfactuals should be considered indicative only of the true native vegetation loss that was or is being averted from each case study offset.
Notwithstanding these limitations, Figures 4 and 5 reveal that the projected gains from averted loss in all case study offsets presented here are likely to be significantly overestimated (3 to more than 200 times higher) when compared to either counterfactual.Over-estimating gains from averted loss permits a larger impact in exchange for the same offset.This will result in a net loss of native vegetation in Victoria and further exacerbate biodiversity decline (Gordon et al., 2015;Maron, Gordon, et al., 2015;Salzman & Ruhl, 2010).This over-estimation means that approximately 39%-86% of the offsets purchased (worth $30,668-$67,638) in these five impact-offset exchange case studies, did not balance the permitted impacts to native vegetation (Tables 6 and 7).The multiplier of 1.5 applied to all General HUs of loss under Victorian policy is inadequate to account for even the lower end of this range.The five permitted impact-offset exchanges presented in this study therefore likely resulted in a net loss of native vegetation, rather than the NNL policy objective.
As the averted loss gains for all offsets under Victorian offsetting policy are determined using the same methodology outlined in the Victorian gain scoring F I G U R E 5 Gains from averted loss in case studies offsets 1-5 over 10 years, as a percentage loss of the habitat score, compared to Counterfactual 2: woody native vegetation extent loss (in hectares) on public and private land in Victoria over two 10-year periods.manual, all impact-offset exchanges permitted under the current policy are likely to be inadequate to balance permitted losses.If the averted loss gains for all offsets in Victoria have been overestimated by a similar extent to the offsets examined in this case study (39%-86%), this over-estimation would have had the potential to result in the sale of $7.5-$16.3millionworth of non-compensatory offsets in 2022.
When buying offsets from a third-party provider in accordance with Victorian offset policy, offset purchasers may have a false confidence that their development will achieve NNL in biodiversity.In reality, they are likely to also be buying offsets that are resulting in a net loss of biodiversity.This would lead to a lack of trust in the Victorian offsets market, in addition to a net loss of biodiversity.
As shown in Figures 4 and 5, the security and prior management gains largely contribute to this overestimation of gains in all five case studies examined, resulting in inadequate offsets.Given that the prior management gain is not directly attributable to the offset itself, and the security gain may be a double up of the maintenance gain, our view is that they should both be removed from the Victorian gain scoring methodology.Minimum native vegetation condition and in-perpetuity site security requirements for offsets under Victorian offset policy already capture the intention of these gains (DELWP, 2017a).Removing both the prior management and security gains will reduce the overestimation of offset gains, however, it could also make offsets more expensive and may be politically challenging to implement.
The maintenance gain alone is somewhat more consistent with the available data on native vegetation condition loss in Victoria, however, it is also likely an overestimate when compared to the available data on native vegetation extent loss (CES, 2018;DELWP, 2020;DSE, 2008).To allow a more robust evaluation of offsetting under Victorian offset policy, greater transparency is required regarding the underlying assumptions for averted loss, including the source of these assumptions and whether they are reflective of a decline in native vegetation condition or extent, or a combination of both.
In the absence of empirical data to inform averted loss gains, Maseyk et al. (2020) support the use of recent average background rates of loss at similar sites (i.e., sites in the same region, with similar habitat types, and subject to similar threats) to estimate averted loss in extent.They also emphasize the need to exclude gains from avoiding losses that would trigger an offset under existing legislative or policy controls.Counterfactual 1 excludes these types of losses.However, as per the limitations of Counterfactual 2, recent rates of native vegetation extent decline do not include condition decline, and may not discriminate between the different types of loss under regulatory and planning processes and, therefore, whether any of the loss was already subject to offsetting.With clear limitations in the available data on background decline in native vegetation in Victoria, the collation of more targeted data to reform the Victorian gain scoring methodology will provide greater accuracy to the assumed gains from averted loss and a stronger commitment to evidence-based policy.
All five impact-offset case studies presented here involved impacts to native vegetation communities that already experience higher rates of loss, including grassy woodland, box ironbark forest, and scattered trees (CES, 2013;CES, 2018).Due to the limited like-for-like requirements under Victorian offset policy, these impacts were not offset with gains in the same EVCs and may have contributed to a disproportionate loss in these communities at greater risk.The disproportionate loss of vegetation communities across Victoria is a significant threat to biodiversity, as the diversity of our remaining habitats is likely to decline.Over-simplifying biodiversity attributes under NNL accounting systems may ultimately negate the biodiversity policy objectives being sought.This study has shown that averted loss gains from offsets in five permitted impact-offset exchange case studies in Victoria are likely to be overestimated, leading to a net loss of biodiversity.We highlight the need for an urgent review into Victoria's gain scoring methodology as the over-estimation of averted loss gains further compounds the issues identified in the Victorian Auditor-General's report into Victoria's NNL policy (VAGO, 2022).
While our study examines the Victorian offset policy, Maron, Bull, et al. (2015) found that on average all Australian offset policies are likely to overestimate gains from averted loss.The extent to which this is an issue in offset policies around the globe has significant implications for global biodiversity decline.If governments are committed to achieving NNL in biodiversity from offsetting, correcting this over-estimation is paramount, and this could include addressing potential biodiversity losses from past impact-offset exchanges.
T A B L E 7 Percentage over-estimation of averted loss gains from case study offsets 1-5 when compared to counterfactual 2. Department of Energy, Environment and Climate Action (formerly DELWP).This raw data is not publicly available, however, an anonymized subset data relevant to the analysis is provided in the manuscript.

Time
U R E 1 Measuring (a) loss in biodiversity from a development impact, against a debiting baseline, and (b) gain in biodiversity from an offset due to averted loss and improvement in biodiversity, against a crediting baseline, over an offset management period (see S2, Supplementary Information, for more information).

¼
Habitat Hectare of Loss † Â General or Species Landscape Factor Â Multiplier ^^ð1Þ General or Species HUs available from an offset ¼ Habitat Hectare of Gain ‡ Â General or Species Landscape Factor^ð 2Þ † Habitat Score Â extent ‡ Gain Score Â extent ^Adjusted modeled scores for SBV (General) or HIS (Species) ^^1.5 (General offset) or 2 (Species offset) HUs = Habitat Units SBV = Strategic Biodiversity Value HIS = Habitat Importance Score also depend on modeled scores of the strategic importance of the vegetation based on Victoria's biodiversity (Strategic Biodiversity Value score for General HUs) and/or habitat for rare or threatened plants and animals (Habitat Importance Score for Species HUs) (O'Brien & is gain resulting from management commitments beyond existing obligations under legislation to improve the current vegetation quality Freehold Proportionate points of gain for each relevant habitat attribute and proposed actions to improve quality over 10-year management period (e.g., weed control) of the relevant purchasers (the proponent responsible for the impact on biodiversity) and credit owners (who are delivering the offset) for each exchange.Data from each exchange has been drawn from detailed impact and offset assessment reports, offset gain calculations, and offset trading agreements, which are not publicly available.To protect confidentiality, all references specific to the development projects, purchasers, and credit owners from each case study have been removed.

F
I G U R E 2 Annual sales of biodiversity units traded through the NVCR from 2008 to 2022 (DEECA, 2023a).F I G U R E 3 The Victorian Catchment Management authority areas (Data Source: DPC, 2023).
T A B L E 6 Percentage over-estimation of averted loss gains from case study offsets 1-5 when compared to counterfactual 1.
T A B L E 5 Summary of gains awarded to case study offsets.