Canadian private protected areas are located in regions of higher vertebrate species richness than government protected areas

The development and exploitation of land by humans, now estimated to be 50% of the Earth’s terrestrial surface (Hooke et al. 2012), is widely recognized as the leading cause of worldwide declines in biodiversity (Maxwell et al. 2016; Johnson et al. 2017; Sanchez-Bayo and Wyckhuys 2019). One solution to mitigate these declines is the creation of protected areas. The International Union for Conservation of Nature (IUCN) defines protected areas as “a clearly defined geographical space, recognised, dedicated and managed, through legal or other effective means, to achieve the long-term conservation of nature with associated ecosystem services and cultural values” (Dudley 2008). In 1992, the United Nations set a worldwide goal of having 17% of all terrestrial land protected by 2020 (United Nations Convention on Biological Diversity 1992). In Canada, approximately 12.1% of terrestrial land and 13.8% of marine environments are currently covered by protected areas (Environment and Climate Change Canada 2019a). As of 2015, over 95% of these protected areas were under government jurisdiction and governance (Environment and Climate Change Canada 2016). However, the effectiveness of government protected areas for conserving biodiversity has been debated. One study found that, in 9 of 15 ecozones, randomly generated plots were over 4 times as likely to have more endangered species than government protected areas (Deguise and Kerr 2006). Conversely, a recent meta-analysis found that government-run marine protected areas had higher fish biomass compared with unprotected waters (Ban et al. 2014). Similarly, evidence suggests that government protected areas in Canada have higher species richness and species at risk richness than randomly selected parcels of similar-size land (Schuster et al. 2019).

Nongovernment protected areas could also play an important role in protecting biodiversity. A recent study provided evidence that Indigenous-managed lands in Australia, Brazil, and Canada contained higher species richness than both IUCN categories I–VI protected areas and a random sample of land (Schuster et al. 2019). Conservation agreements, which are legal agreements that limit the use of land for the purpose of conservation, are one method by which protected areas can be secured by nongovernmental organizations. In the United States, conservation agreements had a lower human modification index than a random sample of land in the same region (Fouch et al. 2019). Additionally, while conservation agreements in the United States Rocky Mountains represented under 5% of total land area, they increased the protected area coverage of 10% of ecosystems within the region (Graves et al. 2019). There is less information, however, on the contribution of private protected areas to biodiversity conservation outside the United States, despite there being many active nongovernmental organizations that secure and manage protected areas around the world.

In this study, we examined estimates of the number of native terrestrial vertebrate species (amphibians, birds, mammals, and reptiles) and species at risk (SAR) in protected areas created by Canadian governments and the Nature Conservancy of Canada (NCC) located in the southern half of Canada (Fig. 1). NCC is a not-for-profit organization founded in 1962 with the goal of “creating a legacy for future generations by conserving important natural areas and biological diversity across all regions of Canada” (Nature Conservancy of Canada 2020c). Over the course of its existence, NCC has contributed to the conservation of over 14 million ha of habitat (Nature Conservancy of Canada 2019c), has directly protected over 2 million ha of land (includes fee simple ownership, conservation agreements, land transferred to partners and resource rights directly secured by NCC), and currently has over 348 470 ha of land under their direct management (fee simple and conservation agreements). For over a decade, NCC has used a quantitative prioritization procedure to acquire protected areas (Arnold and Kraus 2009). While this procedure is not based on standardized species diversity data collected over a broad geographic area, their prioritization does attempt to target areas of high biodiversity for land securement. Thus, we hypothesized that NCC protected areas in Canada would have higher biodiversity estimates than randomly sampled pieces of land, but similar biodiversity compared with government protected areas. Following this hypothesis, we predicted that total native vertebrate species richness and SAR richness on NCC protected areas would be higher than or equal to that of government protected areas and would be higher than a random sample of land.

We also evaluated species diversity based on NCC’s methods of securement. Generally, NCC secures land through either conservation agreements (also known as conservation easements, covenants, or servitudes depending on provincial jurisdiction) or direct acquisition (known as “fee simple”). Conservation agreements are legal agreements that exist between a private landowner and NCC that permanently limits, but does not entirely prevent, land-use activities that are incompatible with the protection of targeted conservation values. In contrast, fee simple properties are secured and managed solely by NCC (Nature Conservancy of Canada 2019a). We hypothesized that fee simple properties would have both higher total biodiversity and SAR diversity than conservation agreements in part because fee simple ownership does not limit the scope of the species or habitats being targeted for conservation and also because there is a much lower overall degree of residential and commercial activity on fee simple properties. Following this hypothesis, we predicted more total and SAR species would be expected to be present on fee simple properties than on conservation agreements properties. Finally, fee simple properties can either be secured via direct purchase or donation. We hypothesized that properties directly purchased would be in regions that have higher total and SAR biodiversity than properties received through donations because NCC would more likely invest limited financial resources in purchasing fee simple properties that are identified of greater priority through their rigorous planning and prioritization procedure. In contrast, while potential properties offered through donation are still subjected to the same evaluation process, there is still an opportunistic element to their acquisition. Following this hypothesis, we predicted that there will be higher total and SAR species diversity on purchased protected areas than on donated protected areas.

NCC protected areas ( $\overline{x}$  = 187 ha, SD = 1347, range = 0.034–54 612, n = 1958) were, on average, an order of magnitude smaller than government protected areas ( $\overline{x}$  = 4988, SD = 39 064 ha, range = 0.01–1 378 067, n = 6267). Within NCC protected areas, conservation agreement properties ( $\overline{x}$  = 214 ha, SD = 642, range = 0.038–12 306, n = 637) were similar in size compared with fee simple properties ( $\overline{x}$  = 174 ha, SD = 1579, range = 0.034–54 612, n = 1321) and, within fee simple properties, donated properties ( $\overline{x}$  = 108 ha, SD = 547, range = 0.034–8964, n = 325) were similar in size to purchased properties ( $\overline{x}$  = 142 ha, SD = 497, range = 0.079–8735, n = 925).

Comparing species diversity between NCC protected areas, government protected areas, and the random sample of land, our best supported model included protected area type and size as fixed effects and ecoregion as a random effect (ΔAIC of next best model =10.71; Table 1). Government protected areas were predicted to be located in areas with 4% (i.e., 7 species; Fig. 2A) fewer species than NCC protected areas of the same size (coefficient = −0.179, 95% CI = −0.207, −0.150; Table 1). The random sample had the lowest species diversity compared with both protected area types, with 6% (i.e., 11 species; Fig. 2A) fewer species than areas where NCC properties were located, when controlling for size and ecoregion (coefficient = −0.277, 95% CI = −0.308, −0.246; Table 1). Models examining SAR richness yielded similar results (Fig. 2B; Table S1).

For total species richness within NCC properties, the best-supported model included NCC protected area type (conservation agreement, fee simple) as a fixed effect and ecoregion as a random effect (ΔAIC of next best model = 8.69; Table 2, Fig. 2C). However, fee simple properties were predicted to be in areas that have only 2% (i.e., 4 species) lower species diversity than conservation agreements while controlling for ecoregion (coefficient for fee simple protected areas = −0.115, 95% CI = −0.171, −0.058; Fig. 2C, Table 2). Within NCC fee simple properties, the best supported models included NCC property type (donated, purchased) but the next best supported model was an intercept only model with a ΔAIC of 1.46 (Table 3, Fig. 2E). Models examining SAR richness yielded similar results for both protected area type and whether the property was donated or purchased (Figs. 2D and 2F; Tables S2 and S3).

Our study provides evidence that, while controlling for ecoregion and size, NCC protected areas tend to be located in areas of higher species richness when compared with government protected areas and that both of these protected area types are located in areas of higher species richness and SAR species richness than randomly sampled land. We also provide evidence that conservation agreement properties have a higher total species and SAR richness compared with properties that are solely managed by NCC (i.e., fee simple) and that donated fee simple properties have similar diversity compared with directly acquired fee simple properties. Overall, our work provides some evidence for the notion that nongovernmental agencies make significant contributions to the protection of biodiversity at a national level.

Although we used IUCN range maps to estimate vertebrate species richness, several studies have shown these maps can overestimate ranges or lead to misclassification on the presence/absence of species (Di Marco et al. 2017; Ramesh et al. 2017). Specifically, range maps have been shown to overestimate species at a spatial resolution of under 200 km (Hurlbert and Jetz 2007). This is important to note given the average size of NCC protected areas (187 ha) is smaller than this resolution, meaning there may be an over estimation of species diversity in these protected areas. While we considered the use of these alternative data sets, such as citizen science data or range maps generated by the Committee on the Status of Endangered Wildlife in Canada, we found that they were less complete than IUCN range maps (Di Marco et al. 2017), had issues with accuracy at small spatial resolution, and (or) had sampling bias (Hurlbert and Jetz 2007; Perkins 2020) and were, therefore, not suitable for our analyses. Although local data on species diversity is available for some protected areas, there are no standardized data available for NCC or government protected areas making these local data unsuitable for our broad analysis. Therefore, at this time, the IUCN maps, with over 75 000 animals (IUCN 2020), appear to represent the most complete data source available (Di Marco et al. 2017). Use of this data set also provides the advantage of conducting standardized comparisons with other areas of the world. That said, potential inaccuracies within the IUCN data set suggest that our results should be considered with some caution. Another, perhaps equally cautionary note, is that we have not included invertebrate diversity in this assessment because range maps or abundance data for this hyper-diverse taxon are nonexistent.

One explanation why NCC protected areas would have been created in regions with higher species richness than government protected areas could be related to the motivation behind their creation or acquisition. Historically, federally run protected areas in Canada have been created with multiple goals, one of which is to conserve biodiversity, but these could also include conserving locations of cultural importance as well for public recreation and education (Dearden and Dempsey 2004). Although the National Parks Act was created in 1930, it was not until 1969 that policies on the creation of national parks were first updated to prioritize ecological needs and not until 1988 that this was made law (Dearden and Dempsey 2004). This is in contrast with NCC, whose sole goal since their 1962 inception has been focused on the protection of biodiversity (Nature Conservancy of Canada 2019b).

Of the 15 federal and provincial government organizations responsible for planning and securing protected areas in Canada, 13 have created a systematic framework for creating and managing protected areas; however, as of 2016, these strategies have only been partially implemented (Environment and Climate Change Canada 2016). Because the creation of government protected areas can take years (Hume 2008), it is possible that the outcomes of these strategies will not be seen for a number of years ahead. Even in cases where protected areas have been created by government organizations, the priority behind their creation is not always to maximize the number of species protected. For example, a top priority of 80% of government organizations is representation of all ecological regions (Environment and Climate Change Canada 2016) rather than optimizing species richness across a jurisdiction. In contrast, NCC has been implementing formal conservation plans since the late 1990s (Arnold and Kraus 2009) and prioritizes nationally and provincially significant land, habitats for at risk species, and land that increases connectivity between protected areas (Nature Conservancy of Canada 2020a).

Another plausible explanation for NCC protected areas being located where species richness is higher than government protected areas is that nongovernmental organizations typically have fewer stakeholders to consult and requirements to fulfill during the land acquisition process. Since 1988, public participation has been included in the creation and modification procedures of federal protected areas (Dearden and Dempsey 2004). While often valuable, public participation can also delay, modify, or even halt the creation of protected areas (Bendickson 2009). In addition, federal protected areas need to be approved by the House of Commons and Senate (Bendickson 2009). In contrast, although NCC always engages in discussions with neighbouring landowners, they are not legally required to get external input when creating a protected area (Nature Conservancy of Canada 2020a). NCC does follow external guidelines as part of the Natural Areas Conservation Program, which is a federally funded program to aid nongovernmental organizations in establishing protected areas (Environment and Climate Change Canada 2018). Under this program, organizations are audited every five years to measure their effectiveness at achieving their program goals and progress that has been made (Nature Conservancy of Canada 2020a). Thus, nongovernmental organizations have the potential to take advantage of opportunities quicker than a government agency. For example, when land became available next to the Waterton Lakes National Park, NCC was able to act quickly, partly because they did not require public discussion or approval from Parliament (Nature Conservancy of Canada 2020d).

Interestingly, our results did not provide support for the hypothesis that fee simple properties being in areas with higher vertebrate species richness than conservation agreement properties. In contrast, the top model predicted an average of 4 additional species and 0.4 additional species at risk on conservation agreement properties compared to fee simple properties. One possible explanation for this result is that conservation agreements are, in theory, able to protect diverse habitat types specifically targeted as a result of the quantitative land prioritization procedure but that can be difficult to directly acquire, either because there is little to no available habitat to purchase, such as ranch lands (Bendickson 2009), or the cost of acquisition is high, such as habitats near urban developments (Morris 2011). Whatever the reason, we view this result as positive news for many land trust organizations because conservation agreements are an effective way to designate conservation land and can make up a significant portion of a protected area portfolio. At NCC, for example, conservation agreements represent 40% of protected areas.

Our model comparing types of fee simple properties provided no support for the hypothesis predicting that estimated vertebrate species richness would be higher on directly purchased property compared with donated property. The reason why we proposed this hypothesis was because there are more opportunities to be selective about purchasing specific properties, whether it be through individuals or organizations approaching NCC to sell their property or through standard real estate channels. In contrast, NCC may accept land donations that are of lesser conservation importance because the conservation benefit relative to the financial cost remains extremely high. However, NCC receives many more requests for land donations than they are capable of accepting. In this way, there is, to some extent, a selection process such that properties that are offered through donations still go through a prioritization procedure (Nature Conservancy of Canada 2020b). Therefore, donated fee simple properties form a very important component of land holdings for NCC and likely for other nongovernmental land trusts.

Although we used broad patterns of species richness as our metric of “effectiveness”, there are a number of other potentially useful approaches. One possibility is an estimate of “human development” using a combination of distance to population centres, proximity to roads, and accessibility of the protected area to humans (Fouch et al. 2019). The degree of connectivity between protected areas has also been used to assess effectiveness of protected areas (Bargelt et al. 2020), although this relies on the assumption that higher connectivity enhances dispersal and promotes long-term population persistence for all species that are included in the analysis. There are also other biological indicators that may be informative. For example, abundance has been used to assess protected areas where the data are available (e.g., African mammals; Craigie et al. 2010). Abundance can be used to examine changes over time, which provides the potential to use it in a before-and-after design (Craigie et al. 2010). Another potential biological indicator which has been used to assess biodiversity is beta diversity (Bergamin et al. 2017). Challenges that arise with these measures, however, revolve around data availability and accuracy, particularly when analyzing effectiveness of protected areas at a national scale.

Our results highlight the important role of private protected areas for land conservation. We also show that conservation agreements and donated fee simple properties are key tools for the success of nongovernmental organizations. While our work emphasizes the positive contributions of a nongovernmental organization, we note that our results are based on broad patterns of species diversity based on range maps and do not depict the actual number of species located in each study location. However, our general findings do suggest that it will be important that multiple governance bodies, including Indigenous groups (Schuster et al. 2019), continue to establish and manage protected areas for effectively conserving biodiversity. It will also be important for organizations that create protected areas to consider how they identify areas of high biodiversity, especially with the limitations of current large-scale biodiversity data sets (Hurlbert and Jetz 2007). While nongovernment organizations may have more flexibility for targeting specific areas or habitats, governments have the ability to secure much larger protected areas and all possible approaches will be needed to stem global biodiversity loss.

We thank Michael Bradstreet for his valuable insight and for assistance with data collection and analysis. Claire Kremen and Angela Brennan provided valuable guidance during all stages of the project. This work was supported by The W. Garfield Weston Foundation. We also thank the editor and two anonymous reviewers whose comments improved the manuscript.