**What Is Threatening Forests in Protected Areas? A Global Assessment of Deforestation in Protected Areas, 2001–2018**

**Christopher M. Wade, Kemen G. Austin, James Cajka, Daniel Lapidus, Kibri H. Everett, Diana Galperin, Rachel Maynard and Aaron Sobel**


Received: 2 April 2020; Accepted: 8 May 2020; Published: 12 May 2020

**Abstract:** The protection of forests is crucial to providing important ecosystem services, such as supplying clean air and water, safeguarding critical habitats for biodiversity, and reducing global greenhouse gas emissions. Despite this importance, global forest loss has steadily increased in recent decades. Protected Areas (PAs) currently account for almost 15% of Earth's terrestrial surface and protect 5% of global tree cover and were developed as a principal approach to limit the impact of anthropogenic activities on natural, intact ecosystems and habitats. We assess global trends in forest loss inside and outside of PAs, and land cover following this forest loss, using a global map of tree cover loss and global maps of land cover. While forests in PAs experience loss at lower rates than non-protected forests, we find that the temporal trend of forest loss in PAs is markedly similar to that of all forest loss globally. We find that forest loss in PAs is most commonly—and increasingly—followed by shrubland, a broad category that could represent re-growing forest, agricultural fallows, or pasture lands in some regional contexts. Anthropogenic forest loss for agriculture is common in some regions, particularly in the global tropics, while wildfires, pests, and storm blowdown are a significant and consistent cause of forest loss in more northern latitudes, such as the United States, Canada, and Russia. Our study describes a process for screening tree cover loss and agriculture expansion taking place within PAs, and identification of priority targets for further site-specific assessments of threats to PAs. We illustrate an approach for more detailed assessment of forest loss in four case study PAs in Brazil, Indonesia, Democratic Republic of Congo, and the United States.

**Keywords:** protected areas; deforestation; tree cover loss; global forest

#### **1. Introduction**

Protected Areas (PAs) are a key strategy for safeguarding global biodiversity and ecosystem services. As of 2018, there were more than 230,000 terrestrial PAs worldwide, protecting 14.9% of the earth's surface and inland waters outside of Antarctica [1], and 5.2% of global tree cover (Figure 1). The extent of PAs has increased substantially since the 1990s, and under the Convention on Biological Diversity, nations committed to further increasing the land area in PAs to 17% by 2020. Despite this, PAs are under increasing threat from anthropogenic activities, including encroachment for settlements, agriculture, mining, logging, and poaching and bushmeat hunting [2]. Worldwide almost one-third of PAs are under intense human pressure, determined as the combined influence of built environments,

agriculture, human population, and transportation infrastructure [3]. Moreover, less than half of PAs are free of any human pressure, and this pressure has increased since the 1990s [4].

**Figure 1.** Location of Protected Areas (PAs) based on World database on Protected Areas [5].

Globally, between 2001 and 2012, 3% of PA forests and 5% of all forests were converted to other land cover types [6]. While the lower rate of forest loss in PAs relative to the global average may suggest that PAs are effective in preventing some, if not all, forest loss, several studies have shown that PAs are preferentially located in areas that have a lower risk of deforestation [7–10]. The locations of PAs are biased towards areas with lower potential agricultural revenues and limited access, in order to minimize conflict with extractive industries and thus reduce the cost of acquisition and establishment [11]. Nevertheless, studies controlling for these confounding factors generally demonstrate that PAs do provide additional protection beyond what would have been expected in the absence of their designation [12,13].

Here, we examine global tree cover loss in PAs globally over 2001–2018, providing the most up-to-date report on forest conversion trends in PAs. Previous analyses of forest loss have been restricted to national or regional scales (e.g., References [8,14]), and/or have not been recently updated (e.g., Reference [7]). We highlight regions and countries where PAs are succeeding and failing to prevent forest loss, a proxy for their ability to safeguard intact habitats and protect other provisioning and regulating ecosystem services. Next, we examine the land cover following tree cover loss in PAs, and variations in follow-up land cover over space and time. In particular, we measured the magnitude (extent) of forest to agricultural land conversion in PAs, as agriculture has been shown to be a dominant driver of deforestation in the tropics [15,16] and globally [17]. Our analysis provides useful information about what may be causing forest loss in PAs and informs the development of PA management and enforcement strategies that are tailored to the agents of change on the ground. Our global assessment should be considered a screening tool to identify priority regions for further detailed investigation of threats to PAs.

#### **2. Materials and Methods**

To conduct our analysis, we took advantage of three recently published, or recently updated, spatially explicit datasets (Table 1), (1) protected areas from the World Database on Protected Areas (WDPA) [5], (2) 30 m resolution Global Forest Change (GFC) data representing tree cover loss annually, from 2001–2018 [18], and (3) 300 m resolution land cover maps for the years 2005, 2010, and 2015 from the European Space Agency–Climate Change Initiative (ESA-CCI) [19].


**Table 1.** Summary of spatial datasets used.

<sup>1</sup> World Database on Protected Areas; <sup>2</sup> Global Forest Change; <sup>3</sup> European Space Agency-Climate Change Initiative.

From the WDPA, we excluded marine PAs, PAs which have been proposed but not formally designated, and those without spatial information (e.g., only provided as point data). We examined forest loss trends in a given PA beginning the year after which it was formally designated, and in the ~10% of cases where the establishment year was not provided, we assumed that the PA had been established prior to 2001. We included all PA types in our analysis (Supplementary Table S1), including those designated in the International Union for Conservation of Nature (IUCN) categorization system as "Not Applicable", "Not Assigned", and "Not Reported". These categories include some important types of protection, including lands managed by indigenous communities in Brazil and United Nations Educational, Scientific and Cultural Organization (UNESCO) Biosphere Reserves in Guatemala. However, these may be left uncategorized according to the IUCN typology due to reporting errors. We isolate trends in PAs which are more strictly protected according to the IUCN categorization system (strict nature reserves, wilderness areas, and national parks), in recognition of the fact that less stringent PA categories may support forest management and other types of sustainable land use change that would result in forest loss. Within IUCN category IV, it is recognized that active management, or modifications to the ecosystem (e.g., halting natural succession, providing supplementary food, or artificially creating habitats) will take place. Specifically, IUCN Category IV sites allow sustainable management of natural resources to maintain culturally defined ecosystems with unique biodiversity, but are not designed for industrial harvest levels [5], and Category VI areas allow the sustainable use of natural resources to promote ecosystem services. In the case where PA polygons overlapped, we assumed the most stringent level of protection. We converted the vector shapefile to a raster grid with spatial resolution of 30 m.

The GFC dataset [18] mapped tree cover loss, defined as the conversion from forest to non-forest, during the 2000–2018 period. We refer to this as forest loss, under the assumption that loss in PAs is predominantly natural forest loss as opposed to loss of planted trees or plantations. We restrict our analysis to areas with greater than or equal to 50% canopy cover in the year 2000 [18]. We tabulated forest loss through three time periods: 2001–2004, 2005–2009, and 2010–2014. We then categorized each forest loss pixel to a land cover class in the year immediately following each of these periods: 2005, 2010, and 2015, respectively. Previous research demonstrated that the land cover following forest loss does not change substantially in a 1–10-year period after the loss occurred. Our approach is based on a period of 1–4 years after forest loss [20]. For a given PA, we excluded any tree cover loss that occurred prior to the year of PA establishment. We additionally report loss in PAs from 2015 to 2018, though we cannot assign a follow-up land cover to this loss, as the most recent land cover map is from the year 2015.

We use the land cover type following forest loss to categorize the cause of deforestation. We acknowledge that subsequent land cover is only a proxy for the complex and dynamic causes of deforestation, but more detailed investigation of these underlying causes is not possible at the

scale of our analysis. We reclassified the 22 land cover categories presented in the European Space Agency Climate Change Initiative (ESA-CCI) land cover dataset to seven categories (Supplementary Table S2) [21]. Our reclassification schema consolidated forest categories (e.g., broadleaf tree cover, needleleaf tree cover), shrubland categories (e.g., shrubland, mosaic herbaceous cover), grassland, and 'other' land cover types (e.g., urban, bare land, water bodies, snow cover). We retained three separate agriculture categories: cropland (including both rainfed and irrigated cultivated crops), mosaic cropland (>50% cropland mixed with trees, shrubs, and herbaceous cover), and mosaic vegetation (<50% cropland and >50% mixed trees, shrubs, and herbaceous cover). The mapped cropland land cover categories have reported accuracies of 73%–89%, except for the mosaic vegetation category. This land cover type has a reported accuracy of just 59%, largely due to commissions of the other cropland categories. Notably, ESA-CCIs agriculture category includes areas used for crop cultivation but does not include areas used for livestock grazing (pasture land or managed grasslands). ESA-CCI does include a grassland category, but does not differentiate natural grasslands from managed grasslands, as this is difficult at global scales using mid-resolution satellite imagery [22]. We resampled the landcover dataset to a 30 m pixel raster grid, matching the resolution of the forest loss map.

To gauge the robustness of our approach to classifying the land cover following forest loss using global-scale data, we compared our results to two previous studies which investigated drivers of deforestation using nationally and regionally specific datasets (Supplementary Table S3). We aggregated several land cover categories in order to facilitate comparison according to Supplementary Table S3. In Indonesia, the authors of Reference [15] found that about 67% of deforestation nationally was followed by agriculture and 40% of deforestation events in PAs were followed by agriculture. We found a similar proportion of forest loss to agriculture nationally (61%) and in PAs (38%). In South America, the authors of Reference [16] reported that 20% of deforestation was followed by agriculture and another 69% followed by pasture lands from 2001 to 2005. We estimated that 44% of forest loss was followed by agriculture, and another 44% was followed by grassland. The differences in Brazil may be partially explained by the challenge of differentiating natural grasslands from managed grassland or pastureland. The results of these robustness checks provide confidence that we can broadly track agriculture as a driver of forest loss using the global land cover dataset, though differentiating pastureland as a driver of forest loss remains difficult with currently available land cover maps (Supplementary Tables S4 and S5). We address the implications of this challenge in more detail in the Discussion Section.

We examined case studies of forest loss in PAs in four countries: Brazil, Indonesia, Democratic Republic of Congo, and the US (Supplementary Figure S1). We selected these countries because each has some of the highest forest loss globally and we wanted to include representation across continents and biomes (Supplementary Table S4). In each case, we investigated forest loss trends in a PA with one of the highest rates of forest loss nationally. We visualized forest cover and loss for each selected PA, examined the Landsat time series from 2000 to 2016 in Google Earth, and available high-resolution satellite imagery in the PA over the study period. We do not aim to provide a systematic validation of our approach to tracking forest loss and following land cover in PAs, but rather use these case studies to explore areas of concern in more detail with higher resolution satellite imagery.

#### **3. Results**

#### *3.1. Ongoing Forest Loss in Protected Areas*

From 2001 to 2018, 12.2% of global forest area (401.3 million hectares (Mha) of 3289.4 Mha) and 4.1% of protected forest area (25.5 Mha of 628.1 Mha) experienced forest loss. Total forest loss generally increased continuously from 2001 to 2018, with a notable spike in 2016 likely due to a spike in forest fires [23]. Forest loss in PAs followed a strikingly similar trend, suggesting that PAs are not exempt from the underlying climatic and macroeconomic forces that drive forest loss globally (Figure 2).

**Figure 2.** comparison of tree cover loss worldwide, and within PAs: 2001–2018 (% of tree cover lost). The distance between the global trend and PA trend is representative of (1) the effectiveness of PAs promoting natural resource conservation, and (2) the impact of location bias of PAs [9,11].

South and Central America are responsible for the largest proportion, 32%, of forest loss in PAs over the study period, followed by North America (20%), Eastern Europe (18%), and Africa and the Middle East (12%). Forest loss in PAs increased across several regions over the study period, including Eastern Europe, Southeast Asia, Africa, and the Middle East, and in particular, in South and Central America (Figure 3). Specifically, Brazil is found to be the largest contributor to this increase in tree cover loss over time, with exceptionally high amounts of tree cover loss in 2016–2017. We also find that no regions experienced a substantial decline in forest loss in PAs over 2001–2018. Total and proportional forest loss by country is shown in Figure 4 (omitting countries with less than 1000 ha of tree cover in PAs) and presented in Supplementary Table S4.

**Figure 3.** Top 10 countries with tree cover loss in PAs, by Union for Conservation of Nature (IUCN) category from 2001–2018. Bars represent tree cover loss in Mha, percentage is proportion of total PA tree cover lost in each country between 2001–2018. (See Supplementary Table S4 for full list of country-level results).

**Figure 4.** Top: total tree cover loss in PAs by country from 2001–2018 (Mha). Bottom: percentage of tree cover lost within PAs from 2001–2018.

Forest loss by IUCN PA classification follows expected trends, with stricter categories of PA experiencing less loss than categories which allow some form of sustainable use (Figure 5, which shows annual PA tree cover loss by country (left) and by IUCN Category (right) with trendlines shown for reference). Categories Ia (Strict nature reserve), Ib (Wilderness Area), and III (National Monuments) have low annual forest loss and no noticeable trend over time. On the other hand, less strict categories, including IV (Habitat and Species Management Areas) and VI (Protected area with sustainable use of natural resources), experienced higher, and somewhat increasing, forest loss during the period between 2001 and 2018. However, PAs which do not fit into the IUCN classification scheme have both the highest amounts of tree cover loss, and the highest rate of increase in tree cover loss over time, but without more detailed management information, we cannot determine whether this is sanctioned clearing. Concerningly, National Parks (category II) should also be very strictly protected, but forest loss in these PAs doubled over the 2001–2018 period.

**Figure 5.** Left: historical trend in tree cover loss in PAs by region. Right: historical trend in tree cover loss by IUCN Category.

#### *3.2. Land Cover Following Forest Loss in Protected Areas*

Globally, across all PAs, shrubland is the dominant land cover following forest loss over 2001–2014, comprising almost half (47%) of all observations. Shrublands comprise a broad land cover category that could include re-growing forest, agricultural fallows, or pasture lands in some regional contexts. The proportion of forest loss in PAs followed by agriculture, including both cropland and mosaic cropland, is 22% (Figure 6). Another 14% of forest loss in PAs is followed by mosaic vegetation (which has the potential to be interspersed with small scale agriculture), and 6% by grassland. The remaining 11% of forest loss is followed by 'other' land uses including urban areas, water bodies, and bare areas. The proportion of forest loss followed by shrubland is the only category that significantly increased over 2001–2014, from about 35% in 2001 to more than 50% in 2014. On the other hand, mosaic cropland and grassland categories have decreased over the study period (Figure 6).

Additionally, PA tree cover loss varies by regions (Figure 7). Early in the study period, South and Central America, Africa and the Middle East, and North America experienced the greatest proportion of tree cover loss followed by shrubland (42%, 19%, and 18% of global total, respectively) (shown in Figure 7). In later periods, the share of tree cover loss followed by shrubland declined in South and Central America, and North America (to 28% and 12%, respectively), while it continued to increase in Africa and the Middle East and Southeast Asia (from 19% in 2001 to 23% in 2014, and from 5% in 2001 to 23% in 2014, respectively).

**Figure 6.** The proportion of forest loss within PAs followed by each land cover category, globally from 2001 to 2018, with the first and last year proportion labeled for reference.

**Figure 7.** Calculated regional distribution of tree cover loss followed by shrubland from 2001 to 2014 (Mha).

#### *3.3. Case Studies*

We identified four case studies to illustrate varying drivers of land conversion, based on those countries and PAs with significant tree cover loss. For these PAs, in Brazil, Indonesia, Democratic Republic of Congo, and the United States, we examined high spatial resolution orthoimages to develop a more detailed understanding of the land cover following loss in these cases. This is not intended as a systematic validation of our analysis, but rather an illustration of how the global analysis can be followed by more detailed investigation with higher resolution satellite imagery.

Brazil's Triunfo do Xingu Environmental PA (IUCN Category V) has recently been noted as a hotspot of deforestation due to pasture expansion, with more than 14,000 hectares of protected land converted to pasture over a six month period in 2018, and over 350,000 ha converted since 2006 [24]. Our analysis found that from 2001 to 2018, over 560,000 ha had experienced tree cover loss. Based on our global analysis, almost 40% of loss in this PA is followed by shrubland and grassland, and another 40% by mosaic agriculture from 2001 to 2015. Using high-resolution imagery from Google Earth, we observed that the forest loss in this PA appears to be organized along roads and settlements and in rectilinear configurations characteristic of agriculture and pastureland, with substantial grassland cover. This configuration suggests that indeed much of the grassland and shrubland cover following loss is managed for livestock grazing and emphasizes the challenge of distinguishing managed and unmanaged grasslands [22,25].

In 2016, just two PAs hosted 40% of forest loss in Indonesian PAs: Tanjung Puting National Park (IUCN Category II) had tree cover loss of about 470 km2 of 3250 km<sup>2</sup> of total tree cover, and Sebangau National Park (IUCN Category II) which had tree cover loss of about 460 km<sup>2</sup> of 5700 km<sup>2</sup> of total tree cover, both peat forest PAs in the Central Kalimantan region. As the majority of this loss occurred after 2015, we do not have results from our global analysis about the subsequent land cover. However, more detailed examination of loss patterns in Sebangau National Park (Figure 8) suggests that forest loss is generally followed by grassland or shrubland. This largely conforms to findings from previous research highlighting the important role of fires, generally anthropogenic in origin but unintentionally impacting large expanses of peat forests, in driving deforestation across Central Kalimantan since 2015 [15].

**Figure 8.** Historical trends in tree cover loss within PAs across selected countries from 2001 to 2018, note the y-axis is not consistent across each graph (Mha).

Democratic Republic of Congo has the fifth highest rate of forest loss in PAs and experienced a steadily increasing rate of forest loss in PAs over 2001–2018. The majority of this loss occurred in PAs categories without an IUCN category ("Not applicable"). We examined loss in the Sankuru Nature Reserve, which was created in 2007 to protect Bonobo habitat and is managed by local communities [26]. Our global analysis found that more than 90% of forest loss in Sankuru was followed by crop land, including mosaic agriculture (in total we found that 1100 km2 of 26,700 km<sup>2</sup> of tree cover was lost). Our detailed examination of imagery on google earth confirmed that the majority of the land cover in areas of loss was small-scale agriculture along roads and near urban areas.

Between 2001 and 2018, 11.4% of global forest loss in PAs occurred in the US, where loss remained relatively stable over time (Figure 8). We examined forest loss trends in Nowitna National Wildlife Refuge (IUCN Category IV) in Alaska, which regularly experiences wildfires and associated forest loss. Between 2001 to 2015, we found that out of about 5700 km2 of forest cover within this PA, more than 1200 km<sup>2</sup> of forest cover was lost. Our global analysis found that nearly all the forest loss in this PA was followed by shrubland and grassland (99.6%). High-resolution imagery suggests that forest loss in Nowitna does appear to be caused by wildfires, which leave burn scars and are followed by a mosaic vegetation dominated by shrubland and grassland categories (Figure 9).

**Figure 9.** Selected PA imagery showcasing deforestation events across case studies: (**A**) Sebangau National Park in Indonesia, (**B**) Triunfo do Xingu Environmental PA in Brazil, (**C**) Sankuru Nature Reserve in Democratic Republic of Congo, and (**D**) Nowitna National Wildlife Refuge in the United States. Images were collected from Google Earth and represent the years 2018 or 2019. Zoomed out images on the left are from the United States Geological Survey (USGS) and the National Aeronautics and Space Administration's (NASA) Landsat program, while zoomed in images on the right are satellite imagery products from Digital Globe.

#### **4. Discussion**

Between 2001 and 2018, two trends took place simultaneously. The absolute area of protected tree cover increased due to countries designating additional land as PAs. Conversely, the annual rate of tree cover loss inside PAs nearly doubled during this time period. The highest loss in tree cover within PAs occurred in 2016, when 0.44% of protected forests experienced loss. Globally, it seems that forests in PAs face the same economic, natural, and social pressures as non-protected forests, as shown by the consistency in trends of forest loss between the two forests categories (Figure 2). This is despite PAs being in areas which should experience fewer human pressures of deforestation [9,11].

In the tropics, the extent of forest loss in PAs increased notably over the study period, and occurred largely in Indonesia, Brazil, and the Democratic Republic of Congo. Global land cover maps demonstrate that shrublands and grasslands were the dominant land cover following forest loss in PAs in the tropics. Our case study analysis demonstrated that this loss corresponds to areas impacted by fires, for example in Indonesian peat lands, and may also correspond to pasture land, for example in Brazil. In many countries in the tropics, agriculture was also a dominant land cover following forest loss, particularly in Sub-Saharan Africa.

In the northern hemisphere, the United States, Canada, and Russia contribute large total amounts of protected tree cover loss. Unlike in the tropics, forest loss in PAs in these countries did not increase noticeably over time. But similarly, shrublands were also the dominant land cover following forest loss over the study period. It is likely that most of this loss corresponds to natural occurrences such as fire, pests, or storm blowdown, which have been shown to be dominant drivers of deforestation in these regions outside of PAs [17]. Indeed, our US case study identified wildfire as a dominant driver of loss.

PAs that do not fit within IUCN categorization schema, which comprise roughly one-third of all PAs, have the highest rates of tree cover loss. These PAs include indigenous lands and UNESCO reserves. There is evidence that indigenous land tenure recognition is effective at preventing deforestation [27]. On the other hand, the majority of uncategorized PAs do not have any active management authority. It is possible that a lack of clear authority over these PAs may be one reason for higher rates of tree cover loss in these uncategorized PAs as a whole.

An important limitation of our assessment of land cover following forest loss in PAs is the reliance on a global mid-resolution land cover dataset. We used a global approach to allow for direct comparisons across regions, and to identify specific regions (within and across countries) where further investigation is needed. However, global land cover datasets do not necessarily address land use and may struggle, for example, to differentiate grazing and pasture lands from shrublands or grasslands [22]. This limits our ability to reliably track pasture expansion into PAs in some geographies where it is important, including in Brazil. Also, mosaic land cover classes such as shrubland/mosaic natural vegetation according to the global map could actually be agroforestry or mixed cropland/agroforestry. This limits our ability to track small-scale and mixed agriculture classes in geographies where those are dominant land cover transitions, such as Central Africa.

We used a case study approach to gauge availability and usefulness of the additional information available via high-resolution imagery from Google Earth to identify and track drivers of forest loss in PAs. This was not intended as a validation of the global approach, but rather an exploration of the potential utility of this emerging technology for more in-depth examination of drivers of forest loss in hotspots of deforestation or priority conservation areas. We found that the spatial and temporal resolution of the imagery available on Google Earth helped inform possible reasons for forest loss, including wildfire, small-scale agriculture, and pasturelands.

Finally, the ESA-CCI dataset represents land cover at 300 m resolution, so pixels with a small proportion of a given land cover category may not be represented, even if they are identified in the 30 m resolution tree cover loss dataset (also discussed in Reference [20]). This will impact our results in areas with highly heterogenous land cover, or small and isolated deforestation events such as targeted logging operations. Recent studies report that logging is the most common driver of loss in intact, but not necessarily protected, forests globally [28]. Logging is difficult to detect via satellite imagery because in many cases, sufficient canopy cover remains following logging that land cover is still classified as forest. High-spatial resolution and frequent satellite imagery may be able to detect the most evident indications of logging, including access roads, skid trails, and tree fall gaps. However, research suggests that these may comprise as little as 20% of the total area impacted by logging activities [29]. Because we use a relatively coarse resolution land cover map, very small-scale or ephemeral forest disturbances—even isolated tree cover loss events in the GFC loss map—will be reported as followed by forest cover. Future research with higher resolution imagery could support investigation of the role of logging in PAs globally.

Despite limitations, by calculating tree cover loss at the PA level, we now have a comprehensive global dataset that can be used to compare outcomes across PAs, to identify the specific characteristics of PAs which limit the rate of tree cover loss over time, and to evaluate the impact of PAs on reducing tree cover loss. There is a growing literature aimed at measuring the impact of human pressure on PAs [3]. This dataset can complement future studies which aim to assess the impacts of socioand macro-economic factors on ecosystem degradation within PAs. Also, as the land use sector is increasingly recognized for its important role in stabilizing future climate, this research can inform assumptions of land available for agriculture. Given that agriculture is occurring in PAs in some regions, despite their designation, researchers and modelers may not want to assume that all protected land will remain in a natural state to more accurately represent land cover dynamics globally.

#### **5. Conclusions**

Though PAs are a key strategy for safeguarding global biodiversity and ecosystem services, they remain under threat from a range of direct and indirect drivers of forest loss [2–4,6]. We found that between 2001 and 2018, global PAs lost 25.5 Mha of forest, or 4.1% of their forested area. This study aimed to improve our understanding of why this loss occurred by examining the land cover following forest loss in PAs. We found that shrubland was the dominant land cover following forest loss in PAs and became increasingly dominant over the study period. This may reflect the fact that the shrubland category encompasses a range of land cover types and land uses, including burned and regenerating forests, fallow lands, and possibly pasture lands, that have been shown to have extensive impacts in key deforestation hotpots globally [15–17]. Agriculture was not the most prominent land cover following forest loss events in PAs globally, but agriculture was shown to be prominent in key geographies—many in Sub-Saharan Africa, including Nigeria, Ghana, and Côte d'Ivoire (Supplementary Table S4). Our analysis improves our understanding of the causes of forest loss in PAs globally and at regional/national scales and can be used to broadly inform strategies to improve PA management and enforcement that are tailored to these agents of change.

**Supplementary Materials:** The following are available online at http://www.mdpi.com/1999-4907/11/5/539/s1: Table S1: IUCN Definition of Protected Area Classifications, Table S2: Reclassification schema to simplify the ESA-CCI land cover dataset, Table S3: Comparison of land cover following deforestation results from this study and previous studies [15,16]. Shown as percent of total deforested area. Figure S1: Location of case studies. Table S4: Total tree cover estimates (1000 ha), tree cover loss estimates from 2001–2018 (1000 ha), relative tree cover loss from 2001 to 2018 (%), tree cover loss followed by agricultural land type (industrial, mosaic, and total) from 2001 to 2014 (1000 ha), and proportion of total tree cover loss followed by agriculture (%) within PAs at the national level. Supplementary analysis on agricultural suitability and tree cover loss. Figure S2: Relationship between agricultural suitability and land cover class following deforestation event within all IUCN categories, and within more stringent IUCN categories (categories Ia, Ib, and II). Figure S3: Agricultural suitability map.

**Author Contributions:** Each author contributed substantially to the completion of this work. Conceptualization was performed by C.M.W., A.S., K.G.A., and D.G.; data curation, investigation and software development were performed by J.C., K.H.E., and R.M.; the formal analysis was completed by C.M.W. and J.C.; the methodology was developed by C.M.W., A.S., K.G.A., J.C., and D.L.; project administration at RTI was overseen by C.M.W. and D.L.; project administration at US EPA was overseen by D.G. and A.S.; supervision was completed by D.L., A.S., and D.G.; data validation was completed by C.M.W. and K.H.E.; visualization was completed by C.M.W., K.G.A., and J.C.; the original draft paper was written by C.M.W., K.G.A., and J.C.; the review and editing was completed by C.M.W., K.G.A., J.C., D.L., K.H.E., D.G., and A.S. All authors have read and agreed to the published version of the manuscript.

**Funding:** This research was funded by the US Environmental Protection Agency (EPA) (Contract EP-C-16-021).

**Conflicts of Interest:** The authors declare no conflict of interest. The views expressed in this publication are those of the authors and do not necessarily represent the views of policies of the US Environmental Protection Agency.

#### **References**


© 2020 by the authors. Licensee MDPI, Basel, Switzerland. This article is an open access article distributed under the terms and conditions of the Creative Commons Attribution (CC BY) license (http://creativecommons.org/licenses/by/4.0/).

### *Article* **The Impact of COVID-19 on the Management of European Protected Areas and Policy Implications**

**James McGinlay, Vassilis Gkoumas, Jens Holtvoeth, Ruymán Federico Armas Fuertes, Elena Bazhenova, Alessandro Benzoni, Kerstin Botsch, Carmen Cabrera Martel, Cati Carrillo Sánchez, Isabel Cervera, Guillermo Chaminade, Juliana Doerstel, Concepción J. Fagundo García, Angela Jones, Michael Lammertz, Kaja Lotman, Majda Odar, Teresa Pastor, Carol Ritchie, Stefano Santi, Mojca Smolej, Francisco Soriano Rico, Holly Waterman, Tomasz Zwijacz-Kozica, Andreas Kontoleon, Panayiotis G. Dimitrakopoulos and Nikoleta Jones**


Received: 29 September 2020; Accepted: 13 November 2020; Published: 18 November 2020 -

**Abstract:** The COVID-19 pandemic led to many European countries imposing lockdown measures and limiting people's movement during spring 2020. During the summer 2020, these strict lockdown measures were gradually lifted while in autumn 2020, local restrictions started to be re-introduced as a second wave emerged. After initial restrictions on visitors accessing many Nature Protected Areas (PAs) in Europe, management authorities have had to introduce measures so that all users can safely visit these protected landscapes. In this paper, we examine the challenges that emerged due to COVID-19 for PAs and their deeper causes. By considering the impact on and response of 14 popular European National and Nature Parks, we propose tentative longer-term solutions going beyond the current short-term measures that have been implemented. The most important challenges identified in our study were overcrowding, a new profile of visitors, problematic behavior, and conflicts between different user groups. A number of new measures have been introduced to tackle these challenges including information campaigns, traffic management, and establishing one-way systems on trail paths. However, measures to safeguard public health are often in conflict with other PA management measures aiming to minimize disturbance of wildlife and ecosystems. We highlight three areas in which management of PAs can learn from the experience of this pandemic: managing visitor numbers in order to avoid overcrowding through careful spatial planning, introducing educational campaigns, particularly targeting a new profile of visitors, and promoting sustainable tourism models, which do not rely on large visitor numbers.

**Keywords:** biodiversity conservation; conflict; national parks; management; pandemic; public health; wellbeing

#### **1. Introduction**

Nature Protected Areas (PAs) are important because of their high biodiversity value and the socio-economic benefits they provide for people [1]. In addition to their crucial role in biodiversity conservation, most PAs in Europe are recognized as multifunctional landscapes providing multiple benefits and ecosystem services. These may range from provisioning services, for example in working farmed landscapes, to regulating services, from water and air quality regulation to carbon storage, but also a very wide range of cultural ecosystem service benefits [2] ranging from psychological restoration [3–5] and improved physiological health [6–8] to better social relations [9–12], and spiritual development [13,14]. Whilst, to a certain extent, many of these benefits may be available to people from a local urban and peri-urban green space, countries designate a suite of high-quality nature protected areas that are of exceptional quality for biodiversity conservation, and provision of the above provides a wide range of benefits to people, and high profile sites such as National Parks are typically extremely popular with visitors. These protected landscapes, therefore, have a crucial role in improving physical and mental health [15], assisting in the improvement of people's wellbeing [16,17], and protecting local social and cultural values [18].

Europe is the region with the largest number of PAs internationally [19]. European PAs are of various sizes with overlapping designations such as the Ramsar Convention, the NATURA 2000 network, the Emerald network, and nationally designated parks [20]. PAs are a significant source of income for local communities living inside or near their boundaries. This is in part due to the high number of visitors they attract [21]. In Europe, Schägner et al. [22] estimated that 449 national parks attract over 2 billion visitors with a total value of €14.5 billion annually. These estimates represent only a fraction of the actual value of tourism in European Protected Areas considering that Europe has over 100,000 Protected Areas [23].

The COVID-19 pandemic led many European countries to impose lockdown measures to limit people's movement [24]. These measures were aimed at reducing the spread of the virus but also decreased significantly the number of people visiting outdoor spaces [25], including PAs, particularly those located in more remote areas, as is often the case for many larger PAs. In European countries, where strict lockdown restrictions were imposed, a reduction in visitor numbers was initially observed (e.g., [26,27]). Likewise, the number of visitors increased rapidly as soon as these restrictions were eased [27–29].

These changes in visitor numbers are expected to have posed significant management issues for PAs in Europe. As we write this paper (October 2020), lockdown measures that had been lifted during the past summer are gradually being re-introduced in several European countries. Central and regional government authorities are looking into ways of containing the virus' transmission, focusing significantly on measures that are enforced locally. During the lockdown, most people only had access to green spaces near where they lived and access to more remote sites was limited to those living nearby. In the case of popular and high-profile PAs, such as national and regional parks, coming out of strict lockdown in late spring and summer 2020 resulted in a steep increase in visitor numbers. Therefore, the management authorities in these areas needed to introduce new measures to enable all users to visit them safely. However, a key issue for popular nature Protected Areas is that the requirements of nature conservation and public health safety may, at times, be difficult to balance.

In normal times, PA staff manage to channel visitors in a way that minimizes disturbance of more sensitive species, such as ground-nesting birds in the nesting season, or fragile ecosystems, such as high altitude montane habitats. This normally means encouraging visitors to spend most of their time in less sensitive locations, often leading to the creation of busier 'honeypot sites', where crowds may occur. However, social distancing regulations that have been implemented in response to COVID-19 involve avoiding crowded locations, so visitors spread out more evenly across the PA, thus increasing the likelihood of human disturbance of species and habitats. In addition to the challenges presented solely by the increase in visitor numbers, visitor behavior can also conflict with landscape and nature conservation.

In this paper, we present an early analysis of how COVID-19 has impacted European PAs so far. We focus on two key issues: (a) the challenges that COVID-19 presented to the management of PAs in Europe, especially on the ability of PAs to perform their functions of conserving nature and providing nature-based benefits to visitors, and (b) review indicative measures that have been implemented across different parks in the region to tackle these challenges. In the discussion section we analyze the problems and their deeper causes, consider what lessons can be learned from the COVID-19 pandemic, and propose tentative longer-term solutions going beyond the current short-term measures.

#### **2. Materials and Methods**

In order to capture the key challenges faced by PAs due to COVID-19 and the measures applied to address them, we followed a two-stage approach: we initially surveyed the existing limited academic and grey literature and websites of park authorities to identify measures that had been introduced by park authorities across Europe during the first months of the pandemic. We then organized two workshops with key informants from the management bodies of a selected sample of European nature PAs.

Our study sample was selected based on three broad criteria: (a) high profile sites were targeted which are popular with visitors (including sites which have been awarded the EUROPARC Federation Charter of Sustainable Tourism) as they were more likely to prove very popular after the lockdown was eased and so particularly challenging to manage regarding the trade-off between visitor management and nature conservation; (b) geographical spread across a diverse range of European countries from West to East, North to South; (c) varying national regulations on COVID-19.

For the selection of the final participant PAs (Table 1), we decided to narrow down our research to eight countries which reflected a range of government responses to the pandemic from strict lockdown to softer measures (criterion c). The eight countries were the UK, Spain, Italy, Estonia, Germany, Poland, Slovenia, and Sweden. During the first months of the pandemic, the UK, Spain, and Italy were three of the most badly affected countries in Europe by the virus and strict lockdown restrictions were imposed restricting significantly people's movement [30–32]. Germany also had a large number of cases but the death rate was lower compared to other countries and restrictions were less severe compared to the UK, Spain, and Italy [33,34]. Estonia, Poland, and Slovenia had a lower fatality rate [35] and restrictions on movement were also imposed [36–39] but were eased earlier compared to other countries. Finally, Sweden was the only country in our study which did not impose strict lockdown restrictions compared to the rest of the European countries. Although no strict measures were imposed in Sweden, it is considered that a large majority of people followed social distancing recommendations while the most vulnerable self-isolated voluntarily [40].

After determining the sample of eight countries, a call was announced inviting park authorities to participate in workshops focusing on the impacts of COVID-19. Two workshops were organized (co-hosted with the EUROPARC Federation) three months apart (1 July and 6 October 2020) with invited representatives of the management of 14 selected sites in order to explore with them the challenges that parks face due to the pandemic. All sites were included in the study, as they are all popular visitor destinations, they spread across multiple European countries, and they face new challenges due to COVID-19. The 14 sites also represent two different types of designations according to the IUCN categories (II and V). (IUCN Categories are described in detail on the IUCN website: https://www.iucn.org/theme/protected-areas/about/protected-area-categories). Details of the participating parks are given in Table 1 below. Fourteen attendees participated in the first workshop representing 13 parks and 16 attendees in the second workshop also representing 13 parks.

The first workshop took place during the first COVID-19 peak when a potential 6-month pandemic was considered likely. At the time of the second workshop, the second peak had begun and it appeared that a 12-month crisis or longer was probable.


**Table 1.** Nature Protected Areas (PAs) participating in the workshops.

\* attended first workshop only, \*\* attended second workshop only.

The first workshop focused on the main positive and negative impacts of the PAs on their local communities prior to the COVID-19 pandemic, and on the main challenges that COVID-19 had presented to them, with a particular focus on changes in visitor numbers and the impact on local people. Participants were then asked about the measures they had implemented to cope with the impact of COVID-19 on the PA and local people, and about their plans to reduce or manage the negative impacts in the future. The second workshop then focused more on the tensions and conflicts between stakeholder groups created by the PA and the COVID-19 pandemic. Participants were asked whether

COVID-19 had caused new or increased existing tensions and conflicts in their PA (e.g., between local people, between visitors, or between visitors and local people). Finally, it covered any research done by the PA management in the past that could explain why tensions or conflicts had emerged (for example, a social impact assessment).

Both workshops were recorded and facilitated by experienced colleagues who took notes during the workshop. Online polls were also conducted during the workshops to obtain quantitative data using the Zoom video-conferencing platform, and further qualitative comments were collected regarding the above discussion topics using Mentimeter.

The notes and the recordings were then analyzed by researchers experienced in qualitative data analysis to identify key emergent themes focusing on two main broad topic areas: (a) the challenges in the management of the PAs due to the pandemic and (b) actions to overcome these challenges. The findings of both qualitative and quantitative data analysis are presented below in Section 3.

#### **3. Results**

#### *3.1. Challenges Due to COVID-19 in European Nature Parks*

A range of challenges caused by COVID-19 was identified by participants (Table 2). In all 14 parks, an increase in visitors was observed especially during the summer compared to the same period the previous year. An increase in weekday visitors was also noted in certain parks. The increase in visitors led to overcrowding incidents and park authorities had to introduce very quickly new social distancing measures and recommendations that would ensure that all users were able to safely enjoy the area. In countries with strict lockdown restrictions (UK, Italy, Spain), the initial low visitation numbers (due to strict travel regulations) were followed by a significant increase in visitors during the summer. In the case of Swedish sites, an increase in visitors was noticed at the beginning of the pandemic (as no restrictions on movement were imposed), which continued throughout the summer months. Similarly, at the German sites, where the movement of people within specific regions was not significantly restricted, a gradual increase in visitors was observed from the beginning of the health emergency in the country. It should be noted that some participant parks experienced an almost 100% increase in visitors on certain days relative to expectations for that time of year. A possible explanation for this increase suggested by the participants was that people felt safer in outdoor and more remote locations, such as the ones protected by National Parks and nature reserves across Europe, compared to indoor and urban spaces. Furthermore, in some countries, the weather was relatively mild in spring 2020 (UK, Germany, and Sweden), which may also have resulted in a significant increase in visitors where people's movement was allowed.

A second important challenge was that the increase in visitors was often combined with incidents of problematic behavior, which is defined here as behavior that conflicts with either the conservation aims of the PA or the widely accepted social norms of behavior within the local communities around the PA. Although irresponsible behavior does occur in PAs, our analysis revealed that such issues became more frequent during the first months of the pandemic.

During the second workshop, park authorities were asked to specify which behaviors had become more frequent at their sites during the pandemic period (Figure 1). Parking and road congestion was the most frequently mentioned issues, which was partly due to the greatly increased number of tourists visiting the parks and a tendency to drive rather than commute by public transport or join organized groups with coaches to reach the parks. Illegal parking became a common problem in several PAs, with rangers having to routinely monitor for illegal behavior. A second important problem mentioned by respondents was linked to waste management issues, such as littering. Workshop participants noted that the observed increase in problematic behavior may be linked to a different profile of people visiting the areas due to COVID-19, who were unaware of the main regulations that are in place and of widely accepted norms of behavior in conservation areas. This was noted across most sites irrespective of the geographical region. Finally, the mountainous parks in the sample expressed concern over increased visits from inexperienced hikers in the winter and the anticipated increased demand for first-aid provisions and search and rescue missions. This is expected both due to a decrease in or ban on guided tours due to social distancing regulations and due to more visitors attempting mountain hiking as an alternative to traveling abroad or elsewhere domestically.




**Figure 1.** Frequency of problematic behavior during the first seven months of the pandemic mentioned by park authorities in the workshops.

Overcrowding incidents and irresponsible behavior also led to conflicts between local residents, and between locals and visitors at several sites. Conflicts between residents tended to arise in cases where behavior, which otherwise would be considered unremarkable or unproblematic, was perceived to contravene emergency restrictions and social distancing recommendations, or where attempts to modify behavior to conform to the new situation led to new conflicts and tension that did not normally arise. Examples of this included local people going for a walk or cycling in their local area and either being criticized for this or not being recognized as local.

Conflicts between visitors and locals arose over either local norms of behavior between locals and visitors, such as inconsiderate parking or littering or in some areas, because visitors were considered 'transmitters' of the virus and locals would prefer the government to have restricted access to the park until they felt safe. Incidents of vandalism and placement of signs on the road stating that visitors were not welcome were also reported.

Another important issue that was mentioned by several parks was the cancellation of or changes to educational and cultural activities, such as school visits, guided tours, and festivals. Although demand for such activities was high during the summer, due to new social distancing rules, several park authorities decided to limit the number of people participating in these activities or cancel them entirely to limit virus transmission. Apart from the wider social and economic impact of having reduced environmental education and cultural activities at the different sites, a related issue was people choosing to go on their own in the park instead of with a guided tour. This also led to a higher number of cars trying to access the PAs (instead of groups using coaches or public transportation).

Collectively, the coincidence of large increases in visitor numbers, attempts to social distance and avoid crowds, lack of availability of organized tours, as well as new types of visitors less aware of the susceptibility to disturbance of many natural systems, all contributed to increasing the risk and extent of disturbance in remoter more sensitive areas of PAs, as well as more general threats to the tranquility, quality, and integrity of the protected landscapes.

#### *3.2. COVID-19 Measures in Order to Overcome the New Challenges*

Different measures were introduced in the parks in order to address the new challenges (Table 3). To address overcrowding, several PAs proceeded with measures limiting access for visitors with different levels of restrictions depending on the virus transmission rates. These measures were often guided by the restrictions imposed at a higher level of administration, either by a regional or central government. In Sweden for example, no restrictions were imposed as the national guidelines did not limit people's movement. On the contrary, at the UK sites, these measures tended to be stricter compared to Sweden (e.g., complete closure of facilities and parking areas) but as lockdown measures were eased, a larger number of visitors were able to access these protected landscapes.

As visitor numbers increased, a key task for all park authorities was to introduce new measures in order to ensure social distancing. A variety of tools were introduced including a one-way system on popular and narrow paths; restricting the number of participants on guided tours; restricting the number of people allowed within facilities (e.g., restaurants, visitor centers, restrooms), and counting the number of visitors entering the area. New measures to maintain good hygiene and limit the spread of the virus were also introduced in PAs. These included enhanced cleaning and waste disposal measures, such as placing hand sanitizers in key locations, cleaning toilet facilities and frequently touched surfaces regularly, and banning cash payments (allowing only contactless payments). Protection of staff was also a key priority with the provision of PPE equipment and installing plexiglass barriers in customer-facing facilities such as restaurants and visitors' centers. A mobile application was also used at one site, which assisted in people having an overview of how many users were on a trail at the same time and reminded them of the current recommendations due to COVID-19.


**Table 3.** Measures to tackle new challenges due to COVID-19.

Regarding the increase in problematic behavior at some sites, which was partly attributed to the different profiles of new users, most park authorities recognized that it was necessary to inform and educate people on permissible activity within the PA and on responsible behavior. Several information campaigns were initiated by the park authorities. These included leaflets informing people of key regulations at key entrance points and also clear signage promoting the dispersal of visitors from the main car parks. Information about regulations was also widely promoted via the websites of the PA authorities and also via social media such as Facebook and Twitter.

The number of rangers patrolling the PAs was also increased in several parks whilst one PA noted that rangers preferred to patrol in pairs as they anticipated more hostile responses from visitors, who were asked to comply with social distancing regulations. Another PA liaised with the local police to enhance their presence in the PA area. Similar measures were introduced to tackle illegal parking, insufficient car parking capacity, and traffic incidents. Several park authorities also informed people of parking and traffic issues via social media. In one park, the parking fee was increased in order to discourage visitors while in other parks, the possibility of introducing a parking fee charge is currently being considered in order to reduce obstructive and illegal parking and manage traffic.

Another key challenge that emerged from our analysis refers to conflicts between in- and out-of-area users in certain PAs. This was mainly because several lockdown restrictions had a geographical component to them with people only allowed to travel up to a certain distance (such as a 5 mile (8 km) limit in Wales, UK) or allowed to travel only within specific regions (Germany and Spain). Although no measures were recorded to tackle this specific challenge, the combined measures mentioned above aimed to reduce conflicts as management authorities tried to reduce issues of overcrowding and problematic behavior, which tended to cause local communities to complain. Regarding the broader social conflicts among residents of the different PAs or the economic impacts of discouraging visitors, no measures were noted at the local level, as such, issues would normally be beyond the remit of park authorities, falling rather to state entities such as the police or government.

As far as the reduction of educational activities is concerned, this is regarded as one of the most important challenges for park authorities. One park mentioned that they have reverted to online learning instead of face-to-face educational activities, while in most parks, some activities have resumed but with a reduced number of participants.

Finally, at this stage of the pandemic, it should be noted that PA management authorities were largely focused on coping with the short-term impacts of the pandemic, with few comments made on its long-term implications and its impacts on the management of the PAs, which are still largely unclear.

#### **4. Discussion**

Although strict restrictions for COVID-19 were eased during summer 2020 across Europe it is clear that the pandemic is not over at the time of writing this paper. On the contrary, in September 2020, Europe entered a second wave of the pandemic and indeed such pandemics might become more frequent in the future [41]. Thus, similar to other parts of the world, it is important to reflect on what has happened in these first months of the pandemic and propose ways that will facilitate the long-term management of such PAs in times of public health crises and associated restrictions and uncertainty [42]. There are two broad categories of management challenges: visitor number management and visitor behavior management.

COVID-19 so far has had significant impacts on the management of PAs across the world. In the United States, an increase in visitors was observed in outdoor spaces creating a number of challenges similar to the ones identified in this paper for Europe [42]. Conversely, in other parts of the world, different concerns have been raised with African PAs seeing a significant reduction of tourism in wildlife reserves [43] leading to reduced financial resources for park authorities and raising concerns about illegal practices.

The increase in visitor numbers to European PAs during the pandemic comes as awareness of PAs has been increasing over time [44] and people are increasingly visiting areas of natural beauty in order to improve their wellbeing [16,17,45]. PAs benefit physical and mental health [45–47] by providing people with the opportunity to come closer to nature [48–51]. In addition, a significant increase in users of outdoor spaces [52] has also been documented during the pandemic that appears to be motivated by people trying to find relatively remote places where they felt safe from the virus.

Indeed, Nature Parks are promoted as a national and regional asset, and so it is not therefore irrational or unreasonable for people to choose to visit such locations when advised to avoid crowded and indoor spaces by the Government. Additionally, travel restrictions have reduced alternative options for people to travel to, such as urban areas or destinations abroad. However, the potential conflict between the rights of visitors to access a national asset versus the right of local residents to be safe in their local area during a pandemic adds an additional dimension to existing tensions between visitors and local residents over access to and competition for local resources.

As noted, this increase in visitors has led to the emergence of new conflicts or exacerbation of existing tensions due to overcrowding incidents and problematic behavior by visitors in several European PAs. Park authorities across Europe had to react quickly to these challenges and introduced several measures. These tools aimed to manage the number of visitors in PAs whilst also accommodating new social distancing measures. Our review revealed that the severity and extent of these measures varied across locations but overall a significant effort has been invested by management authorities to face the challenges brought by the pandemic.

Regarding conflicts, the designation of PAs in Europe has often resulted in conflicts of interest between diverse users and local residents [53,54], especially as competition for space has intensified [55] due to increased tourism [56,57]. Thus, democratizing access to PAs and minimizing disturbance to natural systems was a major challenge for European PAs before the pandemic. Overcrowding in PAs [58–60] often resulted in increased noise levels and disturbance of tranquility [61] causing disruptions in the life of local communities and distortion of human ties [55,62] as well as disturbing wildlife and ecosystems. As a result, the pandemic intensified tensions between locals and visitors in many cases. People living inside or in close proximity to PAs felt under threat from growing crowds of tourists both because of the potential transmission of the virus [44] but also because of overcrowding incidents interrupting how people enjoy nature.

In the past, this has meant focusing people's attention on honeypot sites with high visitor capacity and low sensitivity to disturbance and creating buffer zones around PAs. COVID-19 and the need for social distancing make this solution problematic under the current circumstances. Indeed, well-established measures to minimize disturbance to wildlife by clustering visitors in parks, and social distancing measures to keep people apart, appear to be in conflict and need to be balanced against each other to establish a satisfactory equilibrium between nature protection and public health protection. The pandemic has in effect reduced safe limits in terms of visitor densities at popular sites thereby increasing pressure in less busy locations, but with other risks such as increased disturbance of wildlife and ecosystems, degradation of the quality of natural spaces for other visitors and local residents, and possibly also greater public safety risks, as people visit less closely managed locations where accident risks are greater.

This brings us to our first policy recommendation for the long-term management of European PAs. Future solutions will require careful spatial planning which also takes into consideration issues of social equity in accessing PAs [45]. Incidents of overcrowding can be controlled by the careful distribution of visitors within a PA (both temporal and spatial). This solution would minimize the need to reduce the number of visitors (thus having also a minimum impact on the local economy). We should note, however, that this approach may be difficult to apply in certain areas where a significant part of the land is privately owned and so access rights are limited. Furthermore, some areas of habitat, ecosystems, and wildlife are more sensitive to human disturbance than others (such as areas of ground-nesting birds). Issues of public safety, protecting disturbance-sensitive environments, and land access rights, therefore need to be balanced.

Many management authorities are heavily constrained and have limited space to distribute visitors within the territory. As well as local management strategies, this problem can be alleviated in the future by increasing provision through the designation of additional PAs which are established on land which is publicly accessible. Such a solution would be in accordance with the new EU Biodiversity Strategy, where it has been announced that 30% of land and 30% of waterways will be protected by 2030 [19].

Another important issue we highlight in our study is that the increased number of visitors was also accompanied in certain cases by an increase in contentious or problematic behavior by PA users. Our study revealed that this is probably at least in part linked to a new profile of visitors coming to these areas during the pandemic. A possible explanation is that people coming from more urban areas having a limited sense of nature connectedness and may not necessarily have developed norms of environmental behavior and are unaware of the recommendations for responsible use of the park. Consequently, a new approach may be needed, targeting new groups visiting parks around Europe, while also managing conflicts between different users. Cultural sensitization and education of new visitors are necessary, possibly supported by enforcement of regulations. Although changing behavior is an extremely challenging task, it could be an opportunity to identify which tools are the most efficient in terms of altering people's behavior, when they come to visit a PA. Indeed, the arrival of a new profile of visitors could provide a window of opportunity for PA management to reach new audiences. Whilst some new visitors may only value PAs temporarily for the access to outdoor space they provide during the pandemic and then return to their former practices, a proportion of them may well be open to new experiences in natural spaces and be open to developing a greater sense of connection to nature.

As documented in our study, park authorities have invested significant time in introducing measures to manage overcrowding and visitor behavior during the COVID-19 pandemic. However, a lot of these authorities have limited powers to enforce certain measures, such as fines for illegal parking. Also, in many cases, government guidelines on COVID-19 may overrule regulations introduced by management authorities of PAs. Therefore, a second recommendation is that closer collaboration is needed between park authorities and more centralized institutions in order to propose measures, which ensure that PAs continue to benefit the wellbeing of local communities and visitors.

Tackling overcrowding and resolving conflicts in PAs necessitates also finding a balance between local economic development and the wellbeing of locals. Recreational activities within PAs are a major source of income for locals [63]. Sustainable tourism has been at the core of management plans for many PAs in Europe [64] allowing local communities to maintain an income [65,66], which often compensates for economic losses due to the PA designation. However, several PAs currently function at a maximum visitor capacity during peak periods. Reducing visitor demand to manage COVID-19 outbreaks would also imply a reduction in income for local communities, especially for those working in the hospitality and recreation sectors and, therefore, may necessitate the development of lower-impact and higher quality tourism experiences for visitors, combined with a more holistic rural development policy to reduce reliance solely on tourism.

Our third policy recommendation is, therefore, that local economies in PAs cannot rely on models with a maximum visitor capacity in order to be sustainable and must avoid scenarios of 'over-tourism'. The number of visitors needs to be managed to a level where economic benefits continue to flow but the well-being of locals is safeguarded while visitors' experience remains satisfactory. There are various tools to conduct carrying capacity studies and model alternative scenarios of visitor numbers. Furthermore, overcrowding can be an issue both for locals and visitors and thus studies should be exploring the views of these different users when determining capacity levels [58–60]. A wide application of such assessments in PAs, taking also into consideration social distancing measures, would allow management authorities to specify the optimum number of visitors in a PA but also the distribution of these visitors within the PA [67].

The challenges presented by the pandemic will persist possibly for several years as there are multiple impacts of this pandemic that will be experienced in the future, and indeed future pandemics remain a possibility. Consequently, it is important that park authorities carefully consider these challenges in order to manage PAs in a sustainable and resilient way. As noted, at this stage in the pandemic, PA management authorities were largely focused on coping with the short-term impacts of the pandemic, with few comments made about the longer-term implications of the pandemic and its impacts on the management of the PAs, which still are largely unclear. These long-term impacts will depend on a wide range of factors such as the medium to long-term impact on the economy and the future actions of governance actors.

#### **5. Conclusions**

This study has explored the impacts and challenges that the COVID-19 pandemic presented to a range of European nature PAs and their local communities, and the measures they implemented to mitigate those impacts and associated conflicts. We have also considered the lessons that might be learned from the pandemic experience to inform longer-term PA management, particularly where the pandemic exacerbated existing tensions, such as between local people and visitors.

The pandemic crisis has made the job of PA management more complex and shifted the balance of priorities in trying to achieve a balance between on the one hand nature and landscape conservation, and on the other maintaining accessibility for the visiting public. Indeed the large influxes of visitors during the periods when lockdown regulations were more relaxed demonstrates the importance of such landscapes to people and their well-being, all the more so during this period of the health crisis. Focusing people's attention on outdoor recreation and its many benefits makes logical sense whilst social distancing measures are needed but presents challenges to PA managers in managing disturbance. In consequence, PA management bodies and local communities will need support during such crises where their income becomes more variable and unpredictable, social relations become strained, and nature conservation more difficult to manage.

Overcrowding incidents and an increase in problematic behavior and use of the PAs by visitors were the most significant challenges and also led to an increase in conflicts between locals and visitors. As a response to these challenges, park authorities were quick to respond and find ways to tackle them. Through the use of social media, education campaigns, and a number of other tools, they have tried to

keep the PAs open while keeping visitors and locals safe, assisting also in the recovery of the local economy. However, as European communities are now experiencing a second wave of the pandemic, it is important that longer-term solutions are introduced by management authorities. In consequence, careful management of the spatial distribution of visitors in PAs might be necessary for the future along with educational campaigns targeting groups with a new profile of visitors which has emerged during the pandemic. Thus, although COVID-19 has introduced many challenges for PAs in Europe, it can also be seen as an opportunity to promote new and more sustainable ways to manage protected landscapes.

The pandemic has proved longer-lasting than anticipated, and enhanced global mobility may mean that such a health crisis may become more frequent in the future and so lessons from this pandemic may be worth learning for the longer term, even if the situation normalizes somewhat as the pandemic subsides. Indeed some of the conflicts such as between locals and visitors were pre-existing and exacerbated by the crisis, and the impacts of greater visitor numbers and new types of visitors may constitute a warning at a time when the popularity of nature Protected Areas is increasing and governments actively seek to encourage people to visit them. The advent of new types of visitors to the PAs studied, whilst presenting problems, also presents an opportunity to engage new audiences and foster a sense of connectedness to nature among a broader spectrum of the public.

**Author Contributions:** Conceptualization, V.G., J.M., N.J., J.H., A.K. and P.G.D.; Methodology, J.M., P.G.D., N.J., V.G., and J.H.; Formal analysis, V.G., J.M., N.J.; Investigation, V.G., J.M., N.J., A.K., J.H., P.G.D., R.F.A.F., E.B., A.B., S.S., K.B., C.C.M., C.C.S., F.S.R., G.C., I.C., J.D., M.L., C.J.F.G., A.J., K.L., M.S., T.P., M.O., C.R., H.W., T.Z.-K.; Writing—original draft preparation, V.G., J.M., and N.J.; Writing—review and editing, V.G., J.M., N.J., A.K., J.H., P.G.D., R.F.A.F., E.B., A.B., S.S., K.B., C.C.M., C.C.S., F.S.R., G.C., I.C., J.D., M.L., C.J.F.G., A.J., K.L., M.S., T.P., M.O., C.R., H.W., T.Z.-K.; Visualization N.J.; Supervision N.J., and P.G.D.; Project administration, N.J.; Funding acquisition, N.J. All authors have read and agreed to the published version of the manuscript.

**Funding:** The project has received funding from the European Research Council (ERC) under the European Union's Horizon 2020 research programme (Project FIDELIO, grant agreement no. 802605).

**Conflicts of Interest:** The authors declare no conflict of interest

#### **References**


**Publisher's Note:** MDPI stays neutral with regard to jurisdictional claims in published maps and institutional affiliations.

© 2020 by the authors. Licensee MDPI, Basel, Switzerland. This article is an open access article distributed under the terms and conditions of the Creative Commons Attribution (CC BY) license (http://creativecommons.org/licenses/by/4.0/).

*Article*
