INTRODUCTION
As human population reaches nine billion, demands for food, energy and fibre will put unprecedented pressure on the world's remaining original forests (World Business Council for Sustainable Development 2014). These pressures are playing out in Alberta, Canada, where both forestry and energy development threaten the ecological integrity of Alberta's boreal forest. While protected areas are a cornerstone of biodiversity conservation (Secretariat of the Convention on Biodiversity 2014), on their own they are insufficient to maintain biodiversity (Butchart et al. Reference Butchart, Walpole, Collen, van Strien, Scharlemann, Almond, Baillie, Bomhard, Brown and Bruno2010). New instruments such as offsets are required to maintain biodiversity on working landscapes that are prioritized for commodity production (Convention on Biological Diversity 2009).
Offsets are management actions, such as avoided habitat loss or habitat restoration, which result in biodiversity benefits that can be sold to developers to compensate for the impacts of development. Working forest lands alternate between states of development, restoration, and maturation. In order to protect biodiversity, it is necessary to understand how incentives in an offset market affect the dynamics of habitat loss and restoration on these landscapes. At the same time, in order to ensure political feasibility and public buy-in, the economic and ecological trade-offs of offset policy rules must be made explicit (McShane et al. Reference McShane, Hirsch, Trung, Songorwa, Kinzig, Monteferri, Mutekanga, Thang, Dammert, Pulgar-Vidal, Welch-Devine, Brosius, Coppolillo and O’Connor2011; Harrington et al. Reference Harrington, Safirova, Coleman, Houde and Finkel2014; Jenkins Reference Jenkins2014). In this paper, we model the behaviour of developers in an offset market and evaluate the economic and biodiversity trade-offs associated with alternative offset policies in Alberta's boreal forest.
Over the next 50 years, Canada's oil production is expected to increase by 75%, primarily due to the development of Alberta's oil sands, which cover 141000 km2 of boreal forest. By 2035, oil sands production will reach 5 million barrels per day (Alberta Energy Resources Conservation Board 2011), leading to habitat loss and fragmentation over large scales (Schneider et al. Reference Schneider, Stelfox, Boutin and Wasel2003, Reference Schneider, Hauer, Adamowicz and Boutin2010; Schneider & Dyer Reference Schneider and Dyer2006; Dyer et al. Reference Dyer, Grant, Lesack and Weber2008; Grant et al. Reference Grant, Angen and Dyer2013; Habib et al. Reference Habib, Farr, Schneider and Boutin2013; Wilson et al. Reference Wilson, Liebezeit and Loya2013; Mahon et al. Reference Mahon, Bayne, Sólymos, Matsuoka, Carlson, Dzus, Schmiegelow and Song2014). Although most impacts from energy and forestry are temporary, features such as roads, seismic lines and well pads may persist as long as 60 years, creating significant risks for biodiversity, as well as threatened species such as grizzly bear and caribou (Gosselin et al. Reference Gosselin, Naeth, Hrudey, Plourde, Therrien, Van Der Kraak and Xu2010; Environment Canada 2012; Alberta Energy Regulator and Canadian Environmental Assessment Agency 2013).
In 2008, Alberta released the Land-use Framework (Government of Alberta 2008), a policy for land use decision-making that recommends the use of offsets to address the loss of biodiversity and other natural values from development. To date, there is no policy or guidance for offsets in Alberta, however companies are investing in voluntary offsets and are increasingly required to provide offsets as a condition of approval for large oil sands projects. Both companies and regulators have asked government to develop a consistent policy framework for offsets (Alberta Energy Regulator and Canadian Environmental Assessment Agency 2013).
Numerous countries have developed, or are in the process of developing, polices for biodiversity offsets. Many of these policies are informed by international standards and best practices that have evolved to ensure no-net-loss of biodiversity on a project-by-project basis (Business and Biodiversity Offsets Programme 2009; Gardner et al. Reference Gardner, Von Hase, Brownlie, Ekstrom, Pilgrim, Savy, Theo Stephens, Treweek, Ussher, Ward and Ten Kate2013). The principles of permanence and additionality are fundamental to best practices for no-net-loss. However important trade-offs are embedded in these principles. While offset policies should be guided by standards for biodiversity protection (Brownlie & Botha Reference Brownlie and Botha2009; Pilgrim et al. Reference Pilgrim, Brownlie, Ekstrom, Gardner, von Hase, Kate, Savy, Stephens, Temple, Treweek, Ussher and Ward2013) the selection of targets such as no-net-loss should also be based on trade-offs which themselves are functions of policy rules (Ferraro & Pattanayak Reference Ferraro and Pattanayak2006).
The principle of permanence requires that biodiversity benefits from offsets last at least as long as project impacts. In practice, permanence often leads to a requirement for an easement or other deed restricting mechanism to protect benefits in perpetuity (eftec [Economics for the Environment Consultancy] & IEEP [Institute for European Environmental Policy] 2010). However, in Alberta, most forest land is publically owned and managed by the Crown. Land is allocated in temporary dispositions for specific purposes, and there is no legal mechanism for private parties to secure and transfer long-term biodiversity benefits. Furthermore, offsets cannot be protected, even in the near term, from future disturbance, since under Alberta's Surface Rights Act oil and gas companies have rights to access and disturb land in order to develop subsurface leases (Wilman Reference Wilman1994). Thus, in Alberta's forests, offsets must be temporary and policy rules must be responsive to the exploration and development of energy reserves.
Permanence may not be ecologically desirable on working landscapes where the benefits of particular offset sites change over time as sites become isolated as a result of neighbouring development, or as habitat use changes (Cabeza & Moilanen Reference Cabeza and Moilanen2001, Reference Cabeza and Moilanen2003; Van Teeffelen et al. Reference Van Teeffelen, Vos and Opdam2012; Bull et al. Reference Bull, Suttle, Singh and Milner-Gulland2013). This is particularly true in the boreal forest, which consists of stands of various ages and successional stages whose spatial distribution is determined by large-scale disturbances such as fire (Simberloff Reference Simberloff2001; Jentsch et al. Reference Jentsch, Beierkuhnlein and White2002). Shifting non-permanent ecological reserves have been recommended as a conservation strategy to emulate disturbance dynamics in boreal ecosystems (Cumming et al. Reference Cumming, Burton and Klinkenberg1996). Over time, both human and natural disturbance interact to influence the successional dynamics of the forest and the availability of different habitat types. In this context temporary offsets can be viewed as positive management interventions in the successional dynamics of forest stands to maintain biodiversity at a landscape scale.
Additionality requires that offsets create biodiversity benefits over and above what would have occurred under a business as usual scenario. Because of asymmetric information, only land managers know the true probability of habitat conversion for a particular site. Therefore, offsets from avoided loss can lead to habitat securement on sites not actually threatened by development with no real additional biodiversity benefit (Quétier et al. Reference Quétier, Regnery and Levrel2014). As a result, many programmes prioritize habitat restoration over avoided loss in order to ensure additionality and no-net-loss (Bekessy et al. Reference Bekessy, Wintle, Linenmayer, McCarthy, Colyvan and Burgman2010; eftec & IEEP 2010; Van Teeffelen et al. Reference Van Teeffelen, Opdam, Wätzold, Hartig, Johst, Drechsler, Vos, Wissel and Quétier2014).
In this paper, we model a market for temporary biodiversity offsets in order to understand the dynamics of habitat loss and restoration on Alberta's public forest land. While temporary offsets have been developed to address non-permanence for forest carbon sinks (Marland et al. Reference Marland, Fruit and Sedjo2001; Sedjo & Marland Reference Sedjo and Marland2003), few studies have explored the implications of temporary offsetting for biodiversity. We define temporary offsets as sites that are either restored or restricted from development for a given period, but which may be developed in the future. The temporary offset market aggregates multiple buyers and sellers in every period to meet regional no-net-loss constraints. Previous studies that have examined offsets have assumed that they can be spatially targeted to optimize biodiversity outcomes (see for example Kiesecker et al. Reference Kiesecker, Copeland, Pocewicz and McKenney2010; Schneider et al. Reference Schneider, Hauer, Farr, Adamowicz and Boutin2011; Habib et al. Reference Habib, Farr, Schneider and Boutin2013; Moilanen Reference Moilanen2013; Van Teeffelen et al. Reference Van Teeffelen, Opdam, Wätzold, Hartig, Johst, Drechsler, Vos, Wissel and Quétier2014). However, optimized strategies are not necessarily feasible in market settings where the locations of offsets are determined by autonomous transactions between private entities.
The market model shows the trajectory of development and restoration in response to changing price signals about the value of land and is used to evaluate the impact of offset policy rules on economic costs and biodiversity benefits. Policy treatments include eligibility of avoided loss versus restoration actions; time lags for counting biodiversity benefits from restoration; and geographic constraints on offset trades to reflect ecological and administrative boundaries. The results are disaggregated by watershed to understand distributional effects. The analysis highlights the assumptions and trade-offs embedded in offset policy rules, as well as key uncertainties related to ecological dynamics that must be addressed to maintain biodiversity under temporary offsetting.
METHODS
The study area covers 434000 km2 of public forest land in Northern Alberta, Canada. The majority of the area (381000 km2) is in the boreal natural region with the rest from the parkland and foothill natural regions of Alberta (Natural Regions Committee 2006). The natural regions are classified into eight natural subregions based on soil type, vegetation, topography, and disturbance history, all of which are influential factors for determining the composition of species (Alberta Sustainable Resource Development et al. 2005). The study area includes five sub-watersheds (Lower Peace, Upper Peace, Lower Athabasca, Upper Athabasca, and North Saskatchewan) which serve as provincial land use planning regions (Government of Alberta 2008).
The study area is divided into 167713 sections (each comprising 259 ha) based on the Alberta Township System grid (Alberta Geological Survey 2009). Each section represents a potential development or offset site and, at any given time, is allocated to energy development, forestry development, or offsets. Two types of offsets are considered: avoided loss offsets created from the delay or cancellation of projects; and restoration offsets created from restoration of disturbed forest. The decision of whether to develop a particular site or use it as an offset is based on a comparison between the net present value (hereafter NPV) of development and the price of an offset. If the NPV is greater than the offset price then the site will be developed, otherwise development will be delayed or cancelled, and an avoided loss offset created. In general, higher value sites will be developed and lower value sites will become offsets.
Under the hypothetical offset policy, forestry and energy companies are required to obtain offsets prior to development. Offset credits must be sufficient to cover the full impact of development on biodiversity. Once projects are complete the disturbed sites may be restored with credits from restoration sold as offsets to another party. With all land uses requiring offsets, the offset policy imposes a landscape-level no-net-loss constraint on biodiversity. Although the requirement for forest companies to obtain offsets may seem inconsistent with their initial allocation of harvesting rights, the assumption recognizes the contribution of both forestry and energy activities to regional biodiversity loss and is necessary to determine the most efficient allocation of development on the landscape. The assumption does not alter the economic cost of the policy, which is defined by the loss in development value and is independent of the underlying distribution of initial rights to disturb (Montgomery Reference Montgomery1972).
Equivalence
The equivalence of biodiversity losses and gains is assessed with a biodiversity intactness index which measures the response of species to disturbance and can be scaled from a site to a region (Alberta Biodiversity Monitoring Institute 2014). Biodiversity intactness is a weighted average of individual species intactness, measured as the deviation between predicted abundance given current levels of human disturbance and predicted abundance under a reference condition with the effects of human disturbance statistically removed (Nielson et al. Reference Nielsen, Bayne, Scheick, Herbers and Boutin2007). The index is scaled between 0 and 100, with 100 representing expected abundance under reference condition (Alberta Biodiversity Monitoring Institute 2014). The index provides a snapshot of the condition of a site in terms of its capacity to support biodiversity and is used to quality adjust the area of habitat lost or gained in an offset trade. Like habitat hectares (Parkes et al. Reference Parkes, Newell and Cheal2003; McCarthy et al. Reference McCarthy, Parris, Van Der Ree, McDonnell, Burgman, Williams, McLean, Harper, Meyer, Hahs and Coates2004) biodiversity intactness is difficult to interpret, as it does not convey information about risks to persistence for individual species.
Biodiversity intactness was statistically modelled from cross-section data collected by the Alberta Biodiversity Monitoring Institute for 55 common bird species and 82 common vascular plants. The statistical model is based on methods described in Alberta Biodiversity Monitoring Institute (2014). Explanatory variables include location (latitude and longitude), the percentage area disturbed at each site, and the percentage area classified as lowland. Dummy variables were included to distinguish disturbances that regenerate into native vegetation from those that permanently remove soils and require vegetation replacement. The distinction between disturbance types allows for better differentiation between the impact of temporary features (such as cut-blocks and seismic lines) and semi-permanent features (such as roads and well sites).
The statistical model was used to predict biodiversity intactness for each section of land for a given land-use trajectory. The incremental change in disturbance associated with land-use change in a given period is used to calculate the loss or gain in intactness. Biodiversity credits from avoided loss are equal to the level of intactness of the offset site at the time an activity is delayed or cancelled. This assumes that the site would have been developed in the absence of an offset. In reality, sites are developed with a probability that is known only to developers. Restoration is assumed to remove all human disturbance from a site and return it to reference condition after a time lag. After the lag, biodiversity credits are equal to the difference between predicted intactness prior to restoration and reference condition.
Net present value and demand for development
NPV models for the forestry and energy sectors were developed to generate the demand for development under a base case (no offset) policy. NPV models for energy include the conventional oil, natural gas, in situ and mineable oil sands subsectors. Capacity constraints in the energy sector, including availability of rigs and upgrading capacity, create a scheduling problem of when to develop reserves and where to optimally deploy rigs each year. Similarly mill capacities, road infrastructure, and sustained yield constraints create a harvest scheduling problem for forestry.
NPVs for forestry and energy are assumed to be independent, so the scheduling of these activities can be considered separately. Forestry activities were scheduled to maximize the NPV of harvest under provincial regulations, mill demands and sustained yield constraints based on methods in Hauer et al. (Reference Hauer, Cumming, Schmiegelow, Adamowicz, Weber and Jagodinsk2010a ). The optimal schedule of energy activities was determined by maximizing the NPV of energy development under capacity and demand constraints based on methods in Hauer et al. (Reference Hauer, Adamowicz and Jagodzinski2010b ). Expected oil and gas production for each section of land was developed from ultimate potential maps, which show the amount of recoverable oil and gas for each geological strata (Alberta Energy and Utilities Board 2007). Development profiles for each well were derived from statistics on historic drilling effort and success rates (Alberta Energy and Utilities Board 2007), and exploration and development cost information obtained from Alberta Department of Energy (2007a , b , c ) and the Petroleum Services Association of Canada (2008). Revenues were calculated by multiplying projections of oil and gas production for each site by forecasted oil and gas prices (GLJ Petroleum Consultants 2009a , b ). All NPVs are discounted using a 4% discount rate (Heaps & Pratt Reference Heaps and Pratt1989).
Offset market model
The NPV models run for 50 years, scheduling activities for each sector over 10 five-year time intervals. The outputs provide a base case (no offset) optimal schedule of activities for forestry (xf* it ) and energy (xe* it ) for each section of land i in each period t. The optimal schedule determines the demand for offsets in the offset market model. The offset market model is a linear program which chooses the optimal schedule of sections to develop for forestry (xit f ) and energy (xit e ) and to restore (xit r ) or avoid (xit a ) in each period in order to maximize the NPV of profits. Specifically, the programming problem is expressed as:
\begin{eqnarray}
{\rm Max NPV}\;&{\rm = }&\;\sum\limits_{i,t} {NPV_{it}^f \times x_{it}^f } + \sum\limits_{i,t} {NPV_{it}^e \times x_{it}^e } \nonumber\\
&& - \sum\limits_{i,t} {C_t^r x_{it}^r }\end{eqnarray}
subject to:
\begin{equation}
&&\sum\limits_i {\beta _{it}^{} (x_{it}^f + x_{it}^e } - (x_{it - 1}^f + x_{it - 1}^e - x_{it - 1}^r ))\nonumber\\
&& \le \sum\limits_i {(\beta _{it}^{} (x_{it - L}^r ) - \beta _{it}^{} (x_{it}^a ))} \;\forall t;\\
\end{equation}
\begin{equation}
&&\sum\limits_{i,t} {x_{it}^{e,f} } \le \sum\limits_{i,t} {x_{it}^{e*,f*} } \;\forall i,t;\\
\end{equation}
\begin{equation}
&&\sum\limits_i {x_{it}^r } \le \sum\limits_i {x_{i,0}^{f,e} + } \sum\limits_i {\sum\limits_{l = 1}^{t - 1} {(x_{i,l}^{f,e} - x_{i,l}^r } )} \;\forall t;\\
\end{equation}
\begin{equation}
&&\sum\limits_i {x_{it}^f } + \sum\limits_i {x_{it}^e } + \sum\limits_i {x_{it}^r } + \sum\limits_i {x_{it}^a } \le \overline X \;\forall t.
\end{equation}
Equation 1 is the NPV of forestry and energy activities less restoration costs (Cr ), which are assumed to be C$ 25000 ha−1 (C$ 1 = US$0.92, September 2014) (Alberta Department of Sustainable Resource Development, personal communication 2009). Equation 2 is the no-net-loss constraint on biodiversity intactness β(xt ), which is a function of land use at time t and must hold in aggregate for the forest in every period. The left-hand side of the equation shows the loss in biodiversity intactness due to forestry and energy development. Losses are the result of incremental changes in disturbance from one period to the next. The right-hand side of the equation shows the credits obtained from restoration and avoided loss, with L representing the time lag over which restoration benefits are counted. The model does not restrict offset sites from future development. This reflects the inability to constrain subsurface development on public lands. However future disturbance of restored sites shows up as a loss on the left-hand side of the equation and requires an offset.
Equation 3 specifies that the schedule of development for energy and forestry activity at any location cannot exceed the demand for development at that location under the base case obtained from the NPV models. This means that if a site is selected for energy or forestry development in a particular period, then the area of forest harvested, and the number of wells drilled in each section, are the same as under the unconstrained optimal schedule. The difference between the offset market model and the NPV model is in the timing in which sections are developed, since, in any period, some sites which would have been developed are set aside. Equations 4 and 5 are boundary conditions. Equation 4 requires that the amount of land restored in any period does not exceed the amount of land available for restoration (the number of sites available for restoration at t = 0 plus the cumulative sites developed up to time t less past cumulative restoration). Equation 5 requires that the total amount of land allocated in any period does not exceed the total amount of land in the study area.
There are six five-year periods in the offset market model covering 30 years. While this is shorter than the 50-year time horizon applied in the NPV models, it is sufficient to understand the impact of different offset policies on the trajectory of costs and biodiversity outcomes. To reduce the number of decision variables in the linear programming model, the section level data were aggregated into land classes using seven criteria, including: planning region and natural subregion boundaries; grizzly habitat range map boundaries; earliest period of development; energy and forest land values; and biodiversity intactness values (Weber et al. Reference Weber, Farr, Hauer, Nemes and Perger2011). This resulted in 3433 aggregated decision variables.
Offset policy scenarios
Based on consultation with the Government of Alberta, seven offset policy scenarios were developed (Weber et al. Reference Weber, Farr, Hauer, Nemes and Perger2011). The scenarios (Table 1) are based on combinations of eligibility rules, time lags, and geographic trading restrictions. Three eligibility rules for offsets were considered. Scenario 1 only credits restoration activities, while Scenario 2 only credits avoided loss. The remaining policies allow for both restoration and avoided loss (Scenarios 3 to 7; Table 1). The impact of time lags on the mix of offset types is evaluated by comparing five-year and 20-year time lags (Scenarios 3 and 4, respectively; Table 1). Scenarios 5–7 consider the impacts of geographic trading restrictions. Scenario 5 limits offset trades to regional planning area boundaries, reflecting the principle that offsets should compensate communities for losses of biodiversity. Scenario 6 limits trades to natural subregion boundaries in order to reflect similarity in the underlying ecological processes that support species. Scenario 7 adds a grizzly habitat constraint to the natural subregion constraint to examine the effect of nesting individual habitat requirements for threatened species within a more general trading area.
Offset policy scenarios analysed with combinations of policy rules for eligible offsets (restoration and avoided loss), time lags for counting biodiversity benefits from restoration (five and 25 years; na = not applicable), and geographic offset trading constraints based on watershed, natural subregion, and grizzly habitat boundaries.
The equations in the offset market model were implemented in parts to reflect the different policy scenarios. For example, to examine a restoration-only policy, the last term of Equation 2 was dropped. The constraints were also decomposed into regional or subregional constraints, based on geographic trading restrictions.
RESULTS
Offset policy costs
The total economic cost of the offset policy is defined as the difference between the NPV of resource development under the base case (no-offset) scenario and the NPV of development under the offset policy as determined by substituting the optimal schedule obtained by solving the market model (Equations 1–5) into Equation 1. The C$ 299000 million base case NPV (Table 2) represents the maximum value that could be obtained from resource development in the absence of offset requirements. Restoration only was the most expensive offset policy, resulting in a reduction in NPV of C$ 115000 million, or 38.4% relative to the base case (Scenario 1; Table 2). Avoided loss only was the next most expensive option, with a cost of C$ 4500 million or 1.5% of baseline NPV (Scenario 2; Table 2). Costs were lowest under Scenario 3 (C$ 500 million or 0.2% of baseline NPV), which is the most flexible of all of the policies as it includes both restoration and avoided loss offsets and allows for forest-wide offset trading.
Comparison of net present values and offset policy costs under base case and alternative offset policy scenarios (C$ millions). Policy scenarios include eligibility rules for restoration and avoided loss offsets, five and 20 year time lags for counting biodiversity benefits from restoration, and geographic offset trading restrictions based on regional planning boundaries (sub-watersheds), natural subregions, and grizzly habitat constraints. The % difference from base case is derived from the ratio of net present values under the offset policy to the net present value of the base case. na = not applicable.
When both avoided loss and restoration offsets were permitted, increasing the time lag for crediting restoration benefits from five to 20 years increased policy costs from C$ 500 million to C$ 4000 million (Scenarios 3 and 4; Table 2). The cost of Scenario 4 was similar to the cost of Scenario 2 (avoided loss only) and reflects the fact that companies almost completely substitute avoided loss for restoration offsets as the latter become relatively more expensive. Geographic restrictions had a minimal impact on cost. Restricting offset trades to land-use planning regions added an additional cost of only C$ 27 million over 30 years relative to Scenario 3 (Scenario 5; Table 2). Similarly restricting trades to natural subregions (Scenario 6; Table 2) and within grizzly habitat ranges (Scenario 7; Table 2) only increased costs slightly.
Biodiversity outcomes
We assessed the change in average biodiversity intactness over the study area for each of the scenarios (Table 3). For the base case, the intactness index dropped 3.6% over 30 years, from 86.8 to 83.7 (Table 3). For restoration only (Scenario 1; Table 3), average intactness remained constant by definition because of the no-net-loss constraint and the accounting rules for restoration credits.
Changes in average biodiversity intactness over 30 years under alternative offset policy scenarios, illustrating the effects of avoided loss and restoration offsets and five and 20 year time lags for crediting restoration benefits to biodiversity. Biodiversity intactness is the difference between predicted species abundance for each site under future land use scenarios and reference condition. The intactness index is scaled between 0 and 100, with 100 representing abundance equal to that expected under reference condition (Alberta Biodiversity Monitoring Institute 2014).
Scenarios 2–4 (Table 3), which include avoided loss, showed a decline in intactness over time. This is because avoided loss results in the creation of credits in areas that would not have been developed under the base case. With a decline in intactness of between 2% and 2.2%, avoided loss policies still resulted in an improvement in biodiversity relative to the base case, but they did not achieve no-net-loss.
Regional distribution of offset costs and benefits
We disaggregated the results by land-use planning region to determine distributional effects (Table 4). The North Saskatchewan planning region, which contains less than 0.03% of the study area, was omitted from this analysis, as was Scenario 4. For the base case, NPV was highest in the Lower Athabasca (C$ 183 000 million; Table 4), which contains the most valuable oil sands deposits. NPVs in the other three regions ranged from C$42000 million in the Lower Peace, to C$32000 million in the Upper Athabasca (Table 4). The cost of offsets as a percentage of base NPV is highest for the Lower Peace under all of the scenarios.
Regional distribution of offset policy outcomes, including the initial and final net present values (NPVs) of development (in C$ millions), the total number of Alberta Township System sections of land (each comprising 259 ha) in which development was delayed (avoided loss), and total area restored (ha) for four sub-watersheds over 30 years.
The regional burden of cost in terms of absolute changes in NPV shifted from the Lower Athabasca under a restoration-only policy (C$65682 million, Scenario 1; Table 4) to the Lower Peace under the avoided-loss policy (C$2566, Scenario 2; Table 4). Scenarios 3, 5, 6 and 7 (Table 4), which allow for both avoided loss and restoration, further reduced the burden for the Lower Athabasca, which bore the least cost in terms of both absolute and relative changes in NPV for each of these scenarios. For Scenarios 2, 3, 6, and 7 (Table 4), the Lower Peace bore the highest cost burden, both in terms of absolute and relative cost; Scenario 5, which imposes trading constraints by land-use planning region, resulted in the least regional disparity in cost with the highest burden in the Upper Athabasca.
The total number of sections in avoided loss, as well as the area restored for each region over 30 years, was highest in the Lower Peace for all scenarios (Table 4). Compared to forest-wide trading (Scenario 3; Table 4), regional trade restrictions (Scenario 5; Table 4) reduced restoration effort in the Lower Peace (–41%) and increased restoration effort in the other regions. Natural subregion and grizzly habitat constraints (Scenarios 6 and 7; Table 4) led to a reduction in avoided loss across all regions; restoration decreased in the Upper Athabasca and Upper Peace, and increased in the Lower Peace.
DISCUSSION
The analysis demonstrates that policies that adhere to strict additionality can be costly to implement. Economic efficiency requires that the value of biodiversity conserved under an offset policy exceeds cost, and also that the policy with the highest benefit-cost ratio is selected (Hahn et al. Reference Hahn, Olmstead and Stavins2003). The highest cost policy (Scenario 1) is 226 times more costly than the least cost policy (Scenario 3). However, the biodiversity improvement under Scenario 1 is only 2.25 times higher than the improvement under Scenario 3. From a cost-benefit perspective, the results would seem to favour scenarios which include avoided loss offsets. Still, without a better understanding of how to interpret changes in biodiversity intactness, it is not clear which policy is preferable.
In addition to being the most costly policy, the restoration-only scenario resulted in the most unequal distribution of policy costs. When avoided loss was allowed, forest-wide trading resulted in most of the offset effort directed towards the Lower Peace watershed to facilitate development of higher valued oil sand deposits in the Lower Athabasca. Communities in the Lower Athabasca bear most of the environmental impact of development, while communities in the Lower Peace bear most of the economic burden of the offset policy. Regional trading restrictions redistribute the economic burden at a relatively low cost. The cost of geographic trading restrictions is small because the boreal forest is still relatively intact, with large homogeneous areas that are easily substituted.
The high cost of the restoration scenario is due to requirement for companies to obtain biodiversity credits prior to development, which delays projects. The sharp decrease in NPV is caused by discounting, which puts more weight on current relative to future development values. A corollary is that increasing the time lag for crediting biodiversity benefits to reflect actual ecological recovery trajectories has a decreasing impact on cost; the first periods of delay have the most significant impact on NPV. Policies that include avoided loss cost less because companies can buy time and continue to develop while restoration produces desired biodiversity results. However, not all offsets are additional.
The no-net-loss associated with the restoration scenario is an artefact of the accounting rules. Time lags are used to assess when biodiversity benefits are established, however there is often little relationship between time lags and actual biodiversity recovery. To reduce the costs of project delay, offset credits from restoration are often sold in advance of the establishment of biodiversity benefits with uncertainties related to restoration success and the timing of benefits ignored or waived through mitigation ratios (Maron et al. Reference Maron, Hobbs, Moilanen, Matthews, Christie, Gardner, Keith, Lindenmayer and Mcalpine2012). Restoration failure and long ecological recovery times have resulted in the poor performance of restoration offsets for both wetlands (Moreno-Mateos et al. Reference Moreno-Mateos, Power, Comın and Yockteng2012; Bendor Reference Bendor2009) and biodiversity (Curran et al. Reference Curran, Hellweg and Beck2014).
Restoration-only policies may have unintended negative consequences since they devalue existing habitat. Maintaining habitat is necessary to reduce biodiversity loss in the short run while restored sites recover. However, when both avoided loss and restoration are eligible, increased time lags for crediting restoration benefits can discourage early investment in restoration and fail to ensure long-term availability of recovered landscapes to address future development needs. Therefore, policies which include both avoided loss and restoration must consider the extent to which the two activities are substitutes and over what time horizon. This is particularly important from a regional perspective since the regionally disaggregated results show that policy rules affect the distribution of restoration and avoided loss both within and between regions, which could lead to local hot spots.
CONCLUSION
Temporary offsets are positive management actions that alter the successional dynamics of forests in order to maintain biodiversity. We model a temporary offset market for Alberta's boreal forest to understand the economic and biodiversity outcomes and key uncertainties associated with temporary offsets. Restoration-based policies are more costly than avoided loss because of the delay in establishing biodiversity benefits. To ensure political feasibility, governments may have to consider the distribution of policy costs and benefits between regions, particularly when the initial endowment of resources is unequally distributed. For this landscape offset, effort can be redistributed between regions with little impact on aggregate outcomes.
Offset standards emphasize no-net-loss, permanence and additionality. However, on public lands, permanence is not feasible. Therefore it is necessary to understand the dynamics of habitat loss and restoration with temporary offsets. Similarly, in order to achieve additionality, offset programmes often prioritize restoration over avoided loss. Restoration and avoided loss have different impacts on biodiversity risk; restoration increases future biodiversity with uncertain probability, while avoided loss preserves current biodiversity but reduces the amount of early restoration. Neither approach achieves no-net-loss over the short term except on paper, however policy rules affect the risk profile. Right now, the relationship between restoration and biodiversity recovery is not well understood. To maintain biodiversity, more monitoring and research is needed to understand restoration success and biodiversity recovery rates, and to incorporate this information into offset design.
The insights from this analysis would not necessarily carry over on highly-fragmented agricultural landscapes where there is significant habitat loss and where sites are less substitutable. On agricultural lands, ownership is private and more habitat conversion is permanent. In this case, permanent offsets are more desirable and there are legal mechanisms to secure habitat in perpetuity. The disparity in cost between avoided loss and restoration offsets could shift if landowners are required to give up development options in perpetuity. However, principles such as additionality, which make sense for individual projects, can in aggregate have large economic consequences and result in unintended biodiversity risks at a landscape level. Therefore, it is better to understand the trade-offs associated with policy rules than to rely on generalized principles such as additionality for offset policy design.
