A global assessment of the impact of individual protected areas on preventing forest loss

Highlights

• About 71% of the protected areas worldwide contributed to preventing forest loss.

• Only 30% of forest loss in protected areas have been prevented.

• PAs situated in regions with higher pressure of forest loss prevented more forest loss.

• PAs allow fewer uses of forest resources performed better than PAs allow more uses.

• Private PAs performed similarly to public PAs in preventing forest loss.

Abstract

Globally, the number and extent of terrestrial protected areas (PAs) are expanding rapidly. Nonetheless, their impacts on preventing forest loss and the factors influencing the impacts are not well understood, despite the critical roles of forests in biodiversity conservation, provision of ecosystem services, and achievement of the United Nations' Sustainable Development Goals. To address this important knowledge gap, we quantified the impacts of 54,792 PAs worldwide on preventing forest loss from 2000 to 2015, and assessed important landscape and management factors affecting the impacts of PAs. Although the majority (71.4%) of the PAs contributed to preventing forest loss, only 30.5% of forest loss in the PAs have been prevented. PAs with higher rates of forest loss in their surrounding regions, located at lower elevations, within a few hours of travel from the nearest city, with higher agricultural productivity, and permission for fewer human uses were better able to prevent forest loss. Impacts on preventing forest loss were similar regardless of whether the PAs were privately or publicly owned. Our findings highlight the potential benefits of strict protections, involving private entities in the establishment of PAs, and situating PAs in areas exposed to high risks of forest loss to enhance the capacity to combat global forest loss.

Introduction

Under the auspices of the Convention of Biological Diversity (CBD), the international community has designated >4 million km² – an area larger than India – of new land as PAs over the past decade (UNEP-WCMC and IUCN, 2020). While PA coverage has increased rapidly, there is little evidence of significant transformations that PAs have made to conservation outcomes. Moreover, growing evidence from around the globe points to a widespread degradation and under-resourcing of existing PAs (Mascia and Pailler, 2011; Watson et al., 2014). Even some globally renowned PAs have experienced significant wildlife habitat loss (Geldmann et al., 2019; Liu et al., 2001) and collapse of species populations (Brodie and Waterhouse, 2012).

These concerns highlight the urgency for a comprehensive global evaluation of the impacts of PAs on desired conservation outcomes, such as preventing forest loss (IUCN World Park Congress, 2014). Curbing global forest loss is essential for biodiversity conservation (Betts et al., 2017; Pimm et al., 2014), provision of ecosystem services (Curran and Trigg, 2006; Watson et al., 2018), as well as achieving a number of United Nations' Sustainable Development Goals (Gregersen et al., 2017; United Nations, 2015). Establishing PAs is one of the most common approaches used to prevent forest loss (Andam et al., 2008). Given that current PAs cover >15% of global land surface (UNEP-WCMC and IUCN, 2020), an important question is “Have PAs reduced forest loss around the world?” However, previous global-scale analyses of the performance of PAs have often focused on their spatial overlaps with biodiversity hotspots (Rodrigues et al., 2004; Runge et al., 2015; Venter et al., 2014), management capabilities (Geldmann et al., 2018; Leverington et al., 2010), and the changes in land cover (Heino et al., 2015) or human disturbances (Jones et al., 2018) within PAs, while the impacts of PAs on forest loss worldwide are less well quantified. Existing studies about the impact of PAs on forest loss were mostly conducted at local or regional scales and did not provide a global perspective (Andam et al., 2008; Bowker et al., 2017; Yang et al., 2019) (see Table S1 in Supplementary Information for a list of publications on PAs' impacts on preventing forest loss). In addition, previous evaluations were conducted primarily using parametric regression (e.g., Armenteras et al., 2006; Gaveau et al., 2007) or direct comparison of forest loss inside and outside PAs (e.g., Bruner et al., 2001; Songer et al., 2009) (Table S1). These methods are often biased due to misspecification of functional form or poor comparability between forest inside and outside PAs (Andam et al., 2008; Coetzee, 2017). There have been some meta-analyses (e.g., Geldmann et al., 2013; Oldekop et al., 2016) concerning the impacts of PAs on forest loss at the global scale (Table S1), however, they considered a relatively small number of PAs (<5000 PAs, Table S1) and the case studies included in the meta-analysis primarily rely on regressions or direct inside-outside comparisons to evaluate the impacts. These limitations restricted the ability of meta-analyses to reliably assess the impact of PAs worldwide on preventing forest loss.

To date, there has been only one global empirical analysis about the impacts of PAs on vegetation loss (Joppa and Pfaff, 2011). This study evaluated PAs' impacts by comparing land cover change of protected land pixels to unprotected counterparts with similar characteristics (e.g., elevation and slope) using the matching approach (Ferraro and Hanauer, 2014; Rubin, 1973), and thereby addressing the limitations associated with direct inside-outside comparison or regression. Despite the methodological merit of this study, high uncertainty remained because the land cover maps used in the study had a coarse resolution (1 km/pixel) and inconsistent classification schemes, which might have failed to capture many small-scale forest changes occurred during the short evaluation period (2000 to 2005). Furthermore, this analysis treated the entire network of PAs in one country, instead of each individual PA, as the evaluation unit. Important questions that require a global analysis at the individual PA level remain unanswered. For example, what proportion of the world's PAs reduced forest loss and where were these PAs located relative to human pressure and landscape features?

Furthermore, previous studies evaluating the impacts of PAs primarily focus on quantifying PAs' impacts while factors influencing PAs' impacts are less well investigated, especially at the global scale. For example, previous studies show that the establishment of PAs is primarily driven by land availability and acquisition cost (Baldi et al., 2017). As a result, the distribution of PAs worldwide is biased toward remote areas with low population density and potential for agriculture production, where PAs may least prevent land conversion (Joppa and Pfaff, 2009). In addition to location-associated factors, there are debates over the effectiveness of PAs which allow some human uses of the natural resources and PAs that are owned by private entities. As compared to strictly protected PAs which exclude local inhabitants from access to natural resources, multiple-use PAs aim to achieve both social and conservation goals through allowing some sustainable uses by humans to meet their livelihood demands (Pfaff et al., 2014; Roe and Elliott, 2006). Although strictly protected areas legally permit fewer human uses, the social conflicts associated with strict protection and inadequate management capacity of PAs may comprise their effectiveness. A few regional studies (Ferraro et al., 2013; Nelson and Chomitz, 2011) also show that more strictly protected PAs are not necessarily more effective in reducing human disturbances. Land tenure is known to have a profound impact on land cover change (Hora et al., 2018; Yang et al., 2015) and a growing number of private PAs are recognized and reported to national and international databases (Bingham et al., 2017). Private PAs can complement the state-owned PAs to increase the coverage and connectivity of PAs but they are often believed to have less capacity than PAs owned by governments (Bingham et al., 2017). The performance of private PAs in achieving conservation goals remains unclear and requires empirical evaluations. Understanding the influences of these landscape and management factors on the performance of PAs is critical for effective planning and management of PAs. Armed with this knowledge, conservation practitioners can design strategies accordingly to regulate the factors and enhance the ability of PAs to achieve conservation goals. However, quantitative studies on the influences of those factors on PAs' performance in reducing forest loss at the global scale are rare.

Here we addressed those and other related questions by evaluating the 16-year impacts (2000 to 2015) of 54,792 forested PAs worldwide on forest loss rate using a matching approach. To understand the location-associated factors influencing a PA's ability to prevent forest loss, we assessed the relationships of PAs' impact on forest loss rate with four landscape features, including surrounding forest loss rate, travel time to the nearest urban area, elevation, and agricultural productivity. We also addressed the debates about the performance of PAs that allow some human uses or are owned by private entities by evaluating the influence of protection level (strictly protected versus multi-use) and ownership (public versus private) of PAs on their impacts on preventing forest loss.