## Introduction

Biodiversity is declining at an alarming rate as a result of habitat loss, overexploitation, and climate change1,2. To address this challenge, 196 countries have signed the Convention on Biological Diversity, which aims to halt biodiversity loss by 2020 by reducing direct harm, increasing protected areas, mitigating climate, and reducing global carbon emissions. With global energy demand projected to increase by 14–33% by the year 20353, these commitments have contributed to an exponential increase in the development of renewable electricity sources (e.g., wind, solar, biomass, hydropower4). Concomitantly, unconventional fossil fuels such as shale gas, are being developed worldwide (International Energy Agency; iea.org), and presented as less GHG intensive alternatives to coal-fired electricity generation5. The development of new renewable electricity has been dominated by distributed energy resources (e.g. wind, solar, small hydropower) that are assumed to be more environmentally benign than traditional, large scale technologies, such as large dams or coal-fired plants6. For example, the widespread adoption of renewable energy has resulted in spatially distributed interconnected networks of facilities that can range from small (i.e., single wind turbines or small hydropower dams, <5 or 10 MW installed capacity) to large wind and solar farms7,8.

Distributed energy resources pose interesting challenges to land use planning when the goal is to solve a grid-based energy problem9. For any given resource type, individual distributed energy resource facilities may have small physical footprints compared to single large facilities, but cumulatively may require substantially more infrastructure (roads, powerlines) per unit of energy produced (e.g., many small hydropower projects with little to no water storage capacity vs. a single large dam flooding vast areas)10,11. Similarly, shale gas extraction and transportation is characterized by a dense network of well pads, access roads, seismic lines, and pipelines which can cumulatively affect large geographic areas12,13. As such, both renewable distributed energy resources and shale development may result in similar spatial footprints that can be difficult to evaluate or integrate into land use planning, as well as conservation planning.

Given the rapid increase in renewable energy and shale gas development worldwide, there is an increasing need to understand the potential for cumulative environmental and biodiversity impacts of such technologies14,15,16. Policy-makers and ecologists alike recognize the need for strategic planning for renewable energy and unconventional fossil fuel development to navigate the complex trade-offs between reducing global GHG emissions, minimizing local impacts to biodiversity and human health, while meeting growing energy demands17,18,19,20,21. Thus, identifying optimal energy portfolios that meet economic, environmental and social constraints requires not only a stakeholder-engaged process that addresses social acceptance of emerging technologies22,23, but also the development of analytical frameworks that integrate multiple targets and consider the cumulative effects of energy development24,25,26,27. While many studies have focused on identifying optimal sites of renewable energy resources based on physical attributes of their target area and technical specifications of a particular technology, only few attempted to expand the optimization criteria to biotic, economic, or social factors28,29,30,31.

Renewable and unconventional energy development in North America has been characterized by an exponential increase in electricity generating capacity from wind and solar power (15-fold and 24-fold increase since 2000, respectively; US Department of Energy, energy.gov), as well as extraction of natural gas from shale (7-fold increase in gas production since 200714). Until recently, the Province of British Columbia (BC), Canada, was undergoing a shift from large energy production infrastructure to a system of numerous small dispersed electricity production facilities installed and operated by private corporations32; though this focus has changed recently with the approval of a large dam on the Peace River (Site C; 1100 MW installed capacity capable of producing up to 5100 GWh per year; https://www.sitecproject.com), small distributed energy sources are still being developed throughout the province. As such, BC can serve as an example for other jurisdictions aiming at meeting their energy demand via small, distributed renewable sources. Simultaneously large shale gas deposits are being rapidly developed in BC (half of Canada’s shale gas reserves, estimated at ~1200 trillion cubic feet33). However, neither form of new energy development is currently regulated by land use planning that accounts for the potential for regional biodiversity impacts concomitant with meeting GHG emissions targets. This is an important gap, as British Columbia harbors some the last pristine ecosystems and large mammal assemblages in North America, free-flowing salmon rivers, while many ecosystems and iconic species are under threat from land conversion and other forms of human impacts, including energy development34,35,36.

### Spatial conservation priorities in BC

When the four datasets (large mammals, small-bodied vertebrates, freshwater fishes, and intact landscapes) were considered individually, we found that the spatial conservation priorities identified were geographically distinct (Fig. 3a,b,c,d). We found that the fifth scenario, which combined all vertebrate species and undisturbed landscapes, provided a balance between the other four scenarios (Fig. 3e). After discounting existing protected areas, which had the highest Zonation scores (>0.9), the top spatial conservation priority areas for the large mammal-only scenario (Zonation rank 0.7–0.9; i.e., top 20% conservation priority area not currently protected from development) were concentrated in the Southern Rockies and North-Central BC (Fig. 3a); priorities for small-bodied vertebrates were concentrated in Central and Southern BC, Southern Vancouver Island, and valley bottoms in the Rockies, and Coast Mountains (Fig. 3). In contrast, priorities for conserving undisturbed landscapes without considering vertebrate species distributions were concentrated in Northwestern BC and the Central Coast (Fig. 3d), and priorities for freshwater fishes were more dispersed across geographic regions (Fig. 3b). Overall, prioritizing any given dataset performed poorly for the other datasets, suggesting limited overlap in biogeography and low potential for surrogacy (i.e., protecting a particular taxa conferring protection to other species62; Fig. 4a,b,c,d). For example, prioritizing small-bodied vertebrates failed to provide a conservation benefit to large mammals or intact landscapes, and provided only a moderate benefit for freshwater fishes (Fig. 4c). Similarly, areas identified as important for large, wide-ranging carnivores and ungulates did not capture species rich areas for either small-bodied vertebrates or freshwater fishes (Fig. 4a). The prioritization scenario (5) that included all four datasets provided a balance in representing all biodiversity features included in the analysis (Fig. 4e), although it performed less well for large mammals and intact landscapes.

#### Overlap between energy development and conservation priorities

We found that the three energy technologies were geographically segregated from one another, with shale development (extraction) occurring in relatively low elevation and flat areas of Northeastern BC, run-of-river hydro on streams in the coastal mountains and southern Rocky Mountains, and wind farms at higher elevations in the northern Rockies and central-southern BC, as well as Vancouver Island (Fig. 1). According to BC Hydro estimates, the 66 potential run-of-river hydro projects included in our analyses would require 2,858 km of new powerlines and 174 km of new roads, and a physical footprint of the actual projects of 579.5 ha (range = 2.3–12.9 ha/project). The 87 wind farms in our dataset would need 5,460 km new powerlines and 451 km of new roads, and would cover 1984 ha (range = 9–96 ha/project). The two shale gas basins are much larger (~3,900,000 ha) relative to the spatial footprint of potential renewable energy (wind farms, run-of-river hydro) development. Currently, shale gas extraction occurs on ~2,700,000 ha in Northeastern BC, with a total gas production of ~59 billion m3 in 2018 (with an increasing trend from ~32 billion m3 in 2007; http://www2.gov.bc.ca/gov/content/industry/natural-gas-oil/statistics). Of this amount, approximately 3% (~1.6 billion m3) would be needed to power the four potential natural gas-fired plants under consideration in this study (for a total of ~5,500 GWh/year), assuming that 0.2862 m3 of natural gas is required to produce 1 kWh of electricity (US Energy Information Administration; www.iea.gov). This highlights that the footprint of natural gas electrification plants represents a relatively small contribution to the overall development footprint of shale gas (as most of the shale gas is exported, not used in electricity generation in BC), as well as the substantial uncertainty that exists in determining the specific areas that would be disturbed from extracting natural gas specifically for electricity production. Shale gas extraction activities in the two shale basins, Montney and Horn River, had moderate overlap with large mammals (proportion of cells with rank > 0.7 = 0.43; Fig. 2, Figure S1, Table 1), and little overlap with small-bodied vertebrates, fish, or landscape disturbance (proportion of cells with rank > 0.7 = 0.07–0.29; Fig. 2, Figure S1, Table 1).

The annual firm energy gained from developing all 66 run-of-river hydro projects would be 4,762 GWh/year. These potential locations and their infrastructure had high overlap with high value conservation areas for small-bodied vertebrates (proportion of cells with rank > 0.7 = 0.7; Figs. 2, S1; Table 1). In contrast, when prioritization was based on currently undisturbed areas or large mammals, there was less overlap between run-of-river hydro infrastructure and areas of high conservation value (0.1 and 0.05 proportion of cells with rank > 0.7, respectively). Run-of-river hydro projects that could be built outside high value conservation areas across the five prioritization scenarios could generate up to ~4,000 GWh/year of electricity (Fig. 5). In contrast, potential wind farm locations had less overlap with high value conservation areas compared to run-of-river hydro across all prioritization scenarios (proportion of cells with Zonation rank > 0.7 = 0.06–0.36; Figure S1, Table 1). The 87 wind farms that could be developed for < \$150/MWh have the potential to generate a total of 35,428 GWh/year of new electricity. Of these, ~6,000–24,000 GWh/year can be developed in low conservation value areas, depending on the species and disturbance-specific conservation prioritization scenario, with the highest output and lowest overlap for the undisturbed landscapes scenario (Fig. 5). However, both run-of-river and wind farms have the potential generate very little energy in low conservation value areas when all datasets are considered simultaneously (combined prioritization scenario; Figs. 3e, 5). Because the shale gas production was considered cumulatively across the two shale gas basins (rather than a well-by-well, or lease basis), it was not possible to make a similar comparison to estimate the amount of gas (and associated energy) produced in low conservation value areas.

## Discussions

Greenhouse gas emissions (carbon dioxide, methane, nitrous oxide, etc.) from human activities are the main driver of current climate warming, with energy and transportation sectors accounting for more than half of global emissions63,64. As expected, our analysis confirms that GHG emissions were much greater for shale gas extraction and combustion compared to the two renewable energy technologies, but the differences in overlap between shale gas, run-of-river hydro, and wind farms with regional conservation priorities (Figure S1; Table 1) highlights a trade-off for biodiversity conservation between global and regional scales. In our analyses, there were substantial spatial disparities between the top conservation priorities when each species group and intact landscapes were considered individually (Fig. 3a,b,c,d). While recent studies have evaluated geographic differences in sitting of new energy projects13,65, studies that focus either on GHG emissions or biodiversity impacts may miss important trade-offs between global and regional conservation targets and priorities. Such underlying trade-offs are important, yet rarely quantified or acknowledged, when thinking about avoiding or minimizing overlap between energy development and regional conservation priorities.

In British Columbia, shale gas extraction is concentrated in two basins in the northeastern part of the province, a region of low vertebrate species richness, but with low levels of existing disturbance and moderate conservation value for freshwater fishes and large mammals (Figs. 1, 3). Many species with large space requirements (e.g., grizzly bear (Ursus arctos), caribou (Rangifer tarandus)), that also have high value to local communities, could be impacted by habitat fragmentation associated with shale development14,15, and additional shale development has high potential for cumulative effects at the population level66. Globally, shale gas formations are estimated to cover >20 million km233, and overlap with areas of high aquatic and terrestrial biodiversity14. At the same time, the GHG emissions from shale gas development are similar to those of conventional natural gas, though when used for electricity generation, GHG emissions are half of those from coal-fired plants5. Thus, without global emission-reduction measures (e.g., carbon capture and storage, alternative transportation fuels), electricity production from shale gas will increase global GHG emissions and contribute to long term temperature rise5, impacting biodiversity worldwide67,68.

Both run-of-river hydro and wind energy technologies had much lower lifetime GHG emissions than shale gas extraction and combustion, but they differed in their potential spatial overlap with areas of high conservation priority across species groups and intact landscape metrics (Fig. 2, Figure S1). Run-of-river hydro development has the highest development potential on montane streams (e.g., Coast Mountains, Southern BC Rockies; Fig. 1). We found that potential powerlines and roads associated with run-of-river hydro facilities in these regions are often concentrated in valley bottoms38, which have high species richness and house unique terrestrial vertebrate communities, thus increasing overlap with high priority conservation areas for small bodied vertebrates. The Southern Rockies also have high conservation value for large carnivores and ungulates, further highlighting potential conflicts with species sensitive to terrestrial habitat fragmentation and disturbance. In contrast to run-of-river hydro, potential wind farms are concentrated in coastal areas (northern Vancouver Island) and at higher elevation in the southern Interior of BC and the Central Rockies (Fig. 1), although their associated powerlines also tend to converge in valley bottoms. Regardless of the prioritization scenario, wind farm locations had relatively low overlap with high conservation value areas (Table 1), suggesting that potential wind energy development presents less conflict with terrestrial and aquatic biodiversity than the other energy technologies considered here. Overall, selecting renewable energy project locations is more flexible compared to shale development. While we focused on a particular set of renewable energy projects in this study, additional wind and small hydropower sites could be identified to alleviate regional conservation concerns (likely with added economic costs); shale development does not afford this flexibility, as it has to overlap with the geography of shale deposits.

The combined prioritization scenario (5) suggests that that energy development planning in British Columbia could consider both species-rich areas for terrestrial vertebrates, freshwater fishes, and intact landscapes simultaneously, but individual prioritizations are useful for evaluating potential overlap with areas of high conservation values for species groups that have high importance for local communities (e.g., freshwater fishes, Fig. 4b), and intact landscapes (Fig. 4d).

### The need for cross-scale evaluation of biodiversity impacts from energy development

In spite of the worldwide boom in renewable energy development, as well as shale gas extraction, the trade-offs between protecting biodiversity at local or regional scales versus measures aimed at protecting biodiversity globally remain largely unquantified. The exponential increase in renewable energy development worldwide (including solar and geothermal, not considered in this study4), has recently been accompanied by evaluations of impacts on fish and wildlife. Studies of impacts of wind energy on biodiversity have focused on bird and bat mortalities at wind turbines69,70,71, leading to changes in mitigation measures required to reduce such impacts, such as decreasing the rotation speed, placing facilities outside migration routes72, or developing best management practices for sitting wind farms that minimize impacts to wildlife73,74. Fewer studies have investigated the biodiversity impacts of run-of-river hydro75,76. Similarly, the potential for impacts from shale development on fish and wildlife has only been studied superficially, and surprisingly large gaps exist in our knowledge of impacts to biodiversity, particularly around ground and surface water contamination from fracturing fluids, terrestrial fragmentation from infrastructure, and cumulative impacts14. Our study provides a platform for future studies to incorporate uncertainty in renewable and unconventional energy development locations, and their potential biodiversity impacts, while simultaneously considering their contribution to GHG emissions, as well as economic competitiveness. A natural extension of this work would be to identify portfolios of renewable and shale development projects that minimize trade-offs between energy production and local impacts on species of conservation concern, as well as species and landscapes of high social and cultural value for settler and First Nation communities. The three metrics considered here (GHG, cost and species conservation) provide contrasting perspectives on energy development, but the range of potential metrics for investigating trade-offs between renewable and fossil fuels is considerably greater. In particular, the current study does not acknowledge social aspects, such as attitudes towards different types of energy development, or cultural, recreational, and spiritual values of settler and First Nation communities. Tackling such additional complexity and identifying favored energy solutions requires thoughtful engagement with diverse groups in a participatory decision-making framework.

At a global scale, lowering the GHG emissions of the energy sector is a critical step towards stabilizing global temperatures63, and requires a fundamental transformation of the global energy production system. While some progress has been made at reducing GHG emissions among developed countries since the 1990’s (e.g., −20% in the European Union, no change in the United States, +19% in Canada), the energy sector remains the greatest source GHG emissions globally (~30%). Along with increased efficiency and lower consumption rates, the Fifth Assessment Report of IPCC recommends that the efforts to reduce the GHG emissions from the energy sector focus on transitioning from coal-generated electricity to renewable electricity, as well as natural gas-powered plants (providing that upstream emissions from natural gas extraction become minimal)63. Concomitantly, some jurisdictions and energy sectors have been proactive in incorporating potential biodiversity impacts in distributed renewable energy planning (e.g., wind energy and solar energy in the U.S., onshore and offshore wind in the EU). However, for jurisdictions where energy planning relies largely on economic-only metrics, challenges remain to environmental assessment and planning for meeting energy demands from many small renewable energy facilities.

Development projects are generally considered individually in the environmental impact assessment process77, an approach that does not account for the potential cumulative impacts of multiple facilities across spatial scales or multiple energy sources78. In contrast, strategic-level assessment approaches evaluating the potential for cumulative impacts from many energy projects relative to pre-existing economic activities and future development24, must incorporate conservation and biodiversity as quantified objectives along with lifecycle GHG emissions and cost of energy. Our study complements existing strategic planning approaches to energy development and cumulative impact assessment (e.g.25,26,27,79), and provides a perspective that cuts across energy sectors and build from well-established systematic conservation planning principles. Evaluating the potential for biodiversity impacts from energy portfolios across multiple scales, species groups, and jurisdictions complements local and regional environmental assessments for individual energy projects, and can act as an important filter to inform strategic decisions for land allocation and energy policies.