The climatic risk of Amazonian protected areas is driven by climate velocity until 2050

RESEARCH ARTICLE The climatic risk of Amazonian protected areas is driven by climate velocity until 2050 Calil Torres-AmaralID1,2*, Luciano Jorge Serejo dos AnjosID1,3, Ima Ce´ lia Guimarães Vieira4, Everaldo Barreiros de Souza1 1 Postgraduate Program in Environmental Science—PPGCA, Institute of Geosciences, Meteorology Faculty, Federal University of Para´ —UFPA, Bele´m, Para´ , Brazil, 2 Postgraduate Program in Ecology and Conservation, State University of Mato Grosso, Nova Xavantina, Mato Grosso, Brazil, 3 Campus Parauapebas, Federal Rural University of the Amazon, Parauapebas, Para´, Brazil, 4 Botany Department, Museu Paraense Emilio Goeldi, Bele´m, Para´ , Brazil * calil.torresamaral@gmail.com Abstract Changes in species distribution in response to climate change might challenge the territorial boundaries of protected areas. Amazonia is one of the global regions most at risk of devel- oping long distances between current and future analogous climates and the emergence of climate conditions without analogs in the past. As a result, species present within the net- work of Protected Areas (PAs) of Amazonia may be threatened throughout the 21st century. In this study, we investigated climate velocity based on future and past climate-analogs using forward and backward directions in the network of PAs of Amazonia, in order to assess the climatic risk of these areas to climate change and verify their effectiveness in maintaining the current climate conditions. Using current (1970–2000) and future (2041–

and ILs as conservation tools, they are not immune to the effects of climate change and new PLOS ONE PLOS ONE | https://doi.org/10.1371/journal.pone.0286457 June 22, 2023 1 / 14 a1111111111 a1111111111 a1111111111 a1111111111 a1111111111 OPEN ACCESS Citation: Torres-Amaral C, Anjos LJSd, Vieira ICG, de Souza EB (2023) The climatic risk of Amazonian protected areas is driven by climate velocity until 2050. PLoS ONE 18(6): e0286457. https://doi.org/ 10.1371/journal.pone.0286457 Editor: Gabriel Parra-Henao, Universidad Cooperativa de Colombia, COLOMBIA Received: August 8, 2022 Accepted: May 11, 2023 Published: June 22, 2023 Copyright: © 2023 Torres-Amaral et al. This is an open access article distributed under the terms of the Creative Commons Attribution License, which permits unrestricted use, distribution, and reproduction in any medium, provided the original author and source are credited. Data Availability Statement: All relevant data are within the paper and its Supporting Information files. Funding: We are grateful to CAPES (http://www. capes.gov.br/) for granting the master scholarship to CTA (88882.461690/2019-01). Pro´-Reitoria de Pesquisa e Po´s-Graduac¸ão/Universidade Federal do Para´ (PROPESP/UFPA-http://www.propesp. ufpa.br/) - PAPQ (Programa de Apoio à Publicac¸ão Qualificada) for payment of the publication fee, awarded to EBS. The funders had no role in study

management strategies, specific to each area and that allow adaptation to global changes, will be necessary. Introduction Climate change is predicted to become the most serious threat to biodiversity in the 21st cen- tury [1]. Among the numerous and multifaceted dimensions of climate change, there is the redistribution of climatic conditions over large spatial extensions [2, 3]. Given the spatial scale of this phenomenon, it’s impact on biodiversity is commonly related to the macroecological effect of reducing (or expanding in some cases) the geographic distribution of a species [4]. This impact directly affects the static way in which biodiversity conservation strategies are cur- rently developed. Despite the demarcation of protected areas (PAs) and indigenous lands (ILs) being considered the most efficient and economical line of defense in the global effort to pro- tect biodiversity [5–7], and although its effectiveness in controlling deforestation and other anthropogenic pressures of land use and land cover is recognized [8], even the natural commu- nities existing in these territories are not immune to climate change. Across the Amazon, an extensive network of 6983 ILs and 1170 PAs expands over 4.5 mil- lion km2 and occupies 53.6% of the biome [9]. The carbon stock of the Amazon’s PAs and ILs alone is enough to destabilize or contribute to the stabilization of the Earth’s atmosphere [10]. However, whilst PAs are subject to chronic disturbances such as deforestation and forest deg-

radation [11], evidence suggests that climate change may also limit their efficacy [12]. Ensuring the continued relevance and effectiveness of the PA network across Amazonia is therefore a critical challenge for biodiversity conservation in face of climate change [13, 14]. The spatial redistribution of biodiversity is considered one of the main ecological responses to climate change [15]. This phenomenon is likely to be particularly pronounced across the tropics where species have narrow climatic niches, making them less able to acclimatize or adapt [16], probably due to evolution under stable climatic regimes [17]. Evidence at regional [18] and continental [19] scales suggests that many tropical plant species are already shifting their geographic distributions due to warming from lowlands to Andean slopes. Among the metrics for assessing the climatic risk of biota to climate change [2], climate velocity stands out as a metric that describes the spatial change of climate variables over time, resulting in a rate of change (km/year), which from a biological perspective describes the speed that an organism needs to migrate to maintain constant climatic conditions [20]. This climatic risk assessment metric has been increasingly used in conservation planning [21], and since it uses only climate data, it is becoming indispensable for megadiverse ecosystems with absence of specific information, such as the Amazonia. Analog-based climate velocities represent the physical distance between a location and the

location where a similar climate will be found in the future. This concept can be employed in two temporal directions: forward velocity–which reflects the distance from a grid cell in a cur- rent climate to the nearest cell with similar projected future climate; and backward velocity– which reflects the distance from projected future climate cells back to the closest cell with a similar current climate [20]. The ability to calculate two different metrics allows the extraction of two ecological implications from the same cell. Forward velocity reflects the shortest dis- tance that an organism in the current landscape must migrate to track a consistent climate condition. A high forward velocity indicates a potential risk for climatically-adapted species, as they may be unable to migrate at the necessary speed (km/year). Moreover, a climate that lacks a future analog is classified as a "disappearing climate" [3, 12]. Species inhabiting cells with PLOS ONE Amazonian protected areas and climate velocity PLOS ONE | https://doi.org/10.1371/journal.pone.0286457 June 22, 2023 2 / 14 design, data collection and analysis, decision to publish, or preparation of the manuscript. Competing interests: The authors have declared that no competing interests exist.

high forward velocity or disappearing climates are ultimately under threat of extinction. Alter- natively, backward velocity reflects the shortest distance that an organism in the current land- scape would have to migrate to colonize a particular newly suitable habitat given the projected future climate. A high backward velocity indicates a potential risk for sites on holding species adapted to its future climate. Accordingly, a future climate that lacks a current climate analog is classified as a "novel climate". Therefore, cells with high backward velocity or novel climates are ultimately under threat of becoming unsuitable habitats, with consequent effects on ecosys- tem function and services. Whilst forward velocity measures the distance from a single origin to multiple future destinations, backward velocity implies the distance between multiple origin points to a single future destination. Consequently, the use of these two metrics in planning will result in a contrasting focus on species/populations (forward velocity) or sites (backward velocity) [20, 21]. The ecological interpretation of the effects of climate velocity is often associated with the movement of species, which relies on the assumption that species will not be able to acclimate or adapt to shifting climate and thus need to move to track suitable climates. Observed evi- dence from long-term forest dynamics in Costa Rica, Panama and Malaysia indicated that trees have declining growth rates in response to regional increases on temperatures [22, 23].

Further, evidence from a long-term drought experiment in the Amazon indicated no change in the overall thermal sensitivity of net photosynthesis or foliar respiration rates in response to reduced precipitation [24]. Both studies indicate that individuals may be unable to acclimate physiological processes to changing climate. Similarly, long-term inventory plots have indi- cated that long generation times of tree communities are lagging behind increases in moisture stress [25]. By contrast, a long-term collection of morphometric data on a nonmigratory understory bird community within Amazonia suggested that other components of biodiversity are potentially already responding to climate change through morphological shifts [26]. In this work, the potential ecological implications of the climate change velocity metric will be dis- cussed assuming the premise that most species will not be able to acclimatize and, therefore, will respond to climate change through spatial changes. To understand the underlying causes that determine the climate risk of different areas, cli- matic velocities provide two valuable pieces of information. For example, for areas that have analogous climates, the magnitude of velocity is determined by the distance between analogous climates. In this context, climate velocities depend, to an extent, on the geography of the study area, which was not investigated in this study. Alternatively, for areas that have no-analog cli- mates, climate risk is determined by the magnitude of the climatic variables used, as their val-

ues escape the historical variability of the study region. As the multivariate definition of climate analog was used, there is no single temperature or precipitation threshold that deter- mines the presence of no-analog climates. Rather, the definition of no-analog is a result of the lack of equivalent precipitation and temperature values within the historical variability of all cells within the study area. The Amazon has been ranked amongst the global regions most at risk of developing new climatic conditions [3], and developing the greatest distances between current and future (2100) equivalent temperatures [27]. Previous velocity studies carried out at macroclimatic scale and considering only temperature, have overestimated velocities in regions close to the equator, driven by the dominance of flat areas [28]. The only visualization [20] of the spatial distribution of analog-based climate velocity for the Amazon to date is at macroclimatic scale (50x50 km). The application of finer scales such as meso (1-10km), topo (10m-1km) or micro- climatic (<50m) is therefore essential to identify refuges and other areas where conservation may facilitate the persistence of biodiversity under climate change, even within potentially unsuitable regions [29]. Using the 5km x 5km scale, other studies [12, 30] have sought climatic PLOS ONE Amazonian protected areas and climate velocity PLOS ONE | https://doi.org/10.1371/journal.pone.0286457 June 22, 2023 3 / 14

analogs using forward velocity, first across the entire Brazilian Amazon incorporating defores- tation scenarios, and then focusing on the edge of the PA network. However, the variation in magnitude of climate risks across PAs were not calculated, nor were the backward velocity cal- culated. More recently, forward velocity was associated with ecological niche models, but as one of several measures to identify priority areas for mammals [31]. In view of the high rates of exposure to climate change previously projected across the Ama- zon, and given its global importance as a hotspot for biodiversity [32], we ask: 1) What is the climatic risk to the network of PAs and ILs across the Amazon in terms of climate redistribu- tion that will occur until 2050 under the IPCC SSP245 scenario?; 2) Will the highest levels of climate risk be inside or outside of the PAs and ILs?; 3) Will the climate risk levels be higher in ILs or PAs? We work with the hypotheses that: the Amazon biome will be exposed to high rates of forward and backward velocities and will have a large area without climatic analogs; however, significant differences in climate risk between PAs, ILs and UAs, will be found. These differences are expected to infer different ecological implications across Pas, ILs and UAs, and consequently necessitate the implementation of different conservation strategies across the Amazon. Material and methods Study area We used the biogeographic boundary outlined by the Amazon Network of Georeferenced

Socio-Environmental Information [33] (RAISG, September 2020, http://raisg.socioambiental. org/) (S1 Fig) to define the study region of the Amazon. This region covers approximately 7 million square kilometers and is characterized by a high diversity of ecoregions (53) [34]. The area has predominantly low topographical diversity, with some isolated areas of elevations dis- tributed across eight countries (Bolivia, Brazil, Colombia, Ecuador, Guyana, Peru, Suriname, and Venezuela) and one French territory (French Guiana). To define the network of protected areas (PAs) and Indigenous Lands (ILs), we used spatial data from World Database on Pro- tected Areas (WDPA, September 2020, https://www.protectedplanet.net) [35]. This database includes terrestrial and marine PAs administered by national, regional, or international enti- ties, representing an ecologically based protection system for all countries throughout the world. We considered all terrestrial PAs across the Amazon, totaling 837 PAs covering 3,111,929 km2. We excluded PAs defined by international entities such as RAMSAR sites, bio- sphere reserves, and world heritage sites. Overall, PAs and ILs cover about 44.8% of the study area, of which 42.7% are ILs. Climate data We used temperature and precipitation data from WorldClim, a repository that provides spa- tially interpolated monthly climate data for global land areas at a high spatial resolution (approximately 1x1 km) [36]. To apply the climate velocity metric and make inferences about

the spatial distribution of living organisms [29, 37], we used a 10x10 km resample provided by WorldClim. At this scale, the climate is determined by factors ranging from large-scale conti- nental circulation patterns to the local variability of the terrain, which is fundamental for our analysis. We selected mean annual temperature (Bio1) and annual precipitation (Bio12) as the bioclimatic variables because they represent biologically relevant annual trends. Mean annual temperature is an important indicator variable for a set of other temperature-based metrics [38] and produces projections with relatively lower levels of uncertainty compared to precipi- tation [39]. Total annual precipitation on the other hand, is crucial for understanding the func- tioning of the Amazon ecosystem [40]. The current climatic conditions were based on the PLOS ONE Amazonian protected areas and climate velocity PLOS ONE | https://doi.org/10.1371/journal.pone.0286457 June 22, 2023 4 / 14

average values recorded from a reference climatic period of 1970–2000 [36]. To represent future climate data, we used an ensemble of 5 GCMs (S1 Table) selected from the World Cli- mate Research Program Coupled Model Intercomparison Project phase 6 (CMIP6) for the period 2041–2060 under a greenhouse gas concentration scenario SSP245. This scenario pre- dicts an average trajectory within the range of future paths where trends continue their histori- cal patterns without substantial deviations. This condition has the potential to reflect a minimum-gain trajectory of climate governance, where societies rapidly reduce emissions but not enough to limit warming to below 2˚C. In practice, some countries may reach and increase in temperature of 3˚C by the end of the 21st century–, which is above the objective established in the Paris Agreement of 2015 [41, 42]. Forward and backward velocities We used an algorithm based on analogous climates to calculate the forward and backward cli- matic velocities [43, 44]. The algorithm determined the trajectories with the least climatic dis- similarity between current (annual averages) and future projections, using the least-cost path configuration. To define the specific analogous climate for each cell, we examined the set of values covered by the standard deviation of monthly historical variability of average tempera- ture and total annual precipitation of each cell between 1970 and 2000. We measured distances between similar climates using two climatic variables Bio1 and Bio12 by first measuring the

distances for each variable and then averaging them. Thus, for each pixel in the study area we calculated the lowest cost path to a cell with similar climatic conditions. A greater distance indicated a greater hazard for the cell in terms of spatial redistribution of climatic conditions. To calculate the forward velocity, we determined the future analog that corresponded to the present values of the climatic variables. Similarly, we calculated the path with the lowest cost between a cell with a climate projected for the future and another with similar conditions in the present to calculate the backward velocity. Cells for which no correspondence could be found in future projections represented “disappearing climates” or “novel climates” when the new climate fell outside the historical annual variability of any other region of the Amazon [3]. Climate risk assessment We calculated forward and backward climate velocities for all pixels within PAs and ILs using baseline (1970–2000) and mid-21st century climate (2041–2060) data. We then classified each protected area based on a joint characterization of forward and backward velocities, dividing the range of velocity values into low, moderate, and high using three equal area quantiles. This resulted in a nine-quadrant bivariate plot (S3 Fig). PAs without analogs were those that lacked analogs in any cell. We compared the median velocity of the entire network of PAs and ILs to the median speed in UAs to evaluate the effectiveness of the spatial arrangement of the PAs

and ILs system in reducing climatic velocity. Similarly, we compared how median climatic velocities differed between PA and IL categories. We also performed cell-level analyses to com- pare value variability across different scales of analysis. To quantify the spatial extent of no- analog cells, we counted the number of no-analog cells in each of the PAs and ILs and com- pared the extensions of no-analogs between UAs, ILs and PAs. Finally, we compared the geo- graphic distribution of the region identified as no-analogous with the distributions of climate deltas (i.e., the difference between present and future mean temperature (2050–1990) classified as low, medium and high) and the historical annual variability (i.e., 1 standard deviation) of precipitation and temperature (from 1970–2000) across cells (S2 Fig). This allowed us to dem- onstrate the spatial overlap between areas with higher climate deltas and lower historical cli- mate variability and future no-analog cells. PLOS ONE Amazonian protected areas and climate velocity PLOS ONE | https://doi.org/10.1371/journal.pone.0286457 June 22, 2023 5 / 14

Results The comparison between the medians of forward and backward velocities across the Amazon showed a statistically significant difference (W = 1435894876, p-value < 0.001) between the backward velocity (10.9 km /year) and the forward (7.6 km/year). When comparing the veloci- ties inside and outside the PAs and ILs (Fig 1), the medians are significantly different in both forward (W = 715900341, p < 0.001) and backward (W = 300464189, p < 0.001). For forward velocities, the median of protected areas (9 km/year) was higher than that of non-protected areas (6 km/year). On the other hand, for backward velocities, the median of UAs (12.8 km/ year) was higher than that of PAs (9 km/year) and ILs (9 km/year). The comparison of the median climatic velocities between PAs and ILs (Fig 1C, 1D) showed a significant difference between the categories for both forward (W = 166466959, p-value < 0.001) and backward (W = 52366546, p-value < 0.001). At both velocities, the median of ILs Fig 1. Comparison of forward and backward velocities between protected and unprotected areas (A-B) and Indigenous Lands and Protected Areas (C-D). The dotted red line represents the median value of each of the velocities over the entire Amazon biome. https://doi.org/10.1371/journal.pone.0286457.g001 PLOS ONE Amazonian protected areas and climate velocity PLOS ONE | https://doi.org/10.1371/journal.pone.0286457 June 22, 2023 6 / 14

was higher than that of PAs. The median forward velocity in PAs is 7 km/year, while in ILs it reaches 13 km/year. The backward velocity is higher for both categories, being 7.8 km/year in PAs and 12.8 km/year in ILs. If we consider the presence of no-analog climates (Table 1), our results show that approxi- mately one third (33.5%) of the PAs and ILs do not have a climatic analog in any of the metrics, and 33% of them are due to new climates (no-analog for backward velocities) and 0.5% are for disappearing climates (no-analog for forward velocities). The PAs also varied in relation to the presence of no-analog climates at the cell scale (Fig 2). About 3.2% of the PAs presented at least one cell that did not have a climatic analogs with respect to forward velocity, and almost 45% of the PAs with respect to backward velocity. The velocity-based climatic risk assessment shows that the extent of areas with high back- ward-only velocity (32.4%) is greater than that of forward (5.85%). This trend stands out if we Table 1. Spatial extent and cover percent of disappearing and novel climates in unprotected areas, protected areas and indigenous lands. Protection category Climatic risk Cover across Amazon (%) Area (km2) UAs Disappearing 0.29 20677.66 UAs Novel 25.44 1774348.06 PAs and ILs Disappearing 0.15 10702.90 PAs and ILs Novel 17.70 1235069.07 PAs Disappearing 0.1522 10617.44 PAs Novel 11.96 834745.27 ILs Disappearing 0.0073 515.73 ILs Novel 6.26 436616.21 Amazon (all area) - 100 6974270.54 https://doi.org/10.1371/journal.pone.0286457.t001

Fig 2. Spatial distribution of forward and backward climate velocities at the cell level. Map source: Amazon biogeographical limit [9]; protected areas and indigenous land [35]. https://doi.org/10.1371/journal.pone.0286457.g002 PLOS ONE Amazonian protected areas and climate velocity PLOS ONE | https://doi.org/10.1371/journal.pone.0286457 June 22, 2023 7 / 14

consider the large number of PAs and ILs with low forward velocities (31.3%), in relation to backward (16.3%) (as defined by the quantile with the lowest values within a range of velocities previously divided into three equal areas). In addition, a higher percentage of PAs and ILs experience low forward and backward combined velocities (11.4%) than high combined veloc- ities (3.1%). However, the smaller number of PAs classified with high velocities may underesti- mate the real threat to which these areas will be subjected to (Fig 3). Discussion When comparing climate velocities across Amazonia we find that the network of PAs and ILs will be less at risk to the impacts of backward velocity compared to UAs, emphasizing the importance of these areas as tools in conservation. In contrast, for forward velocity impacts, the PA network will be slightly more exposed than UAs–indicating that the current spatial arrangement of the PAs and ILs network is still not optimal for minimizing the impacts of cli- mate redistribution. Despite this difference, both PAs, ILs and UAs will be exposed to high rates of climatic velocity, particularly in terms of backward velocity, which will cause greater effects in magnitude considering a median value across the entire Amazon. Previous studies have already classified Amazonia as having a high risk of developing new climatic conditions [3], developing greater distances between current and future (2100) equivalent temperatures [28] and high forward and backward velocities [20]. These projections are much greater than

the median latitudinal migration rate of 1.69 km/year observed in most ecosystems and Fig 3. Velocity-based climatic risk assessment of each protected area within the Amazon biome. We assessed the velocity-based climatic risk of each PA within the Amazon based on nine categories derived from the classification of the gradient of values of forward and backward velocities in three equal quantiles of area (in the up-right panel). No-analog PAs and ILs are indicated in black and are not included in the bivariate panel. In (c) we see the biogeographic boundary of the Amazon with PAs in black and ILs in grey. Map source: Amazon biogeographical limit [9]; protected areas and indigenous land [30]. https://doi.org/10.1371/journal.pone.0286457.g003 PLOS ONE Amazonian protected areas and climate velocity PLOS ONE | https://doi.org/10.1371/journal.pone.0286457 June 22, 2023 8 / 14

taxonomic groups studied so far [45]. Our results are in line with previously projected climate velocities, however, our projections also reveal the types and magnitudes of particular threats within the Amazon, especially with respect to PAs and ILs. The potential impacts of both forwards and backwards velocities will be greater on ILs than on PAs. This means that species within ILs, rather than PAS, will be further away from their climate analogs by 2050, and will therefore have to migrate at higher annual velocities, whilst their former climatic envelopes will not become suitable for other colonizing species. Given the correlation between velocities based on analogs and geographic characteristics [37], we suggest that this pattern is the result of the arrangement between the distribution of climatic velocities and the spatial arrangement of the ILs. However, further studies are needed to iden- tify which spatial features determine this pattern. The same striking pattern of backward velocity risk can be seen in the spatial extent of areas without analogous climates. There is a large distribution of novel climates throughout the cen- tral region of the Amazon, making this area unsuitable as a habitat for species that cannot acclimate or adapt, and thus, indicating potential effects on the ecosystem functions and ser- vices of floodplain forests [46]. In contrast, few areas have been identified as disappearing cli- mates. The majority of areas with novel and/or disappearing climates are in UAs, however,

within the PA and IL network, most novel/disappearing climates are spatially distributed in PAs. Nevertheless, it is important to note that when considering PAs and ILs who’s entire ter- ritory will be without a climate analog across Amazonia by 2050, ILs are the majority, reinforc- ing the pattern of high climate risk found in IL’s. Based on spatial distribution of forward and backward climate velocities at the cell level, our results show that the backward velocity is gradually distributed from values with lower speeds in some parts of high elevation areas towards the edges of Amazonia, becoming higher in the lowland interiors. In central Amazonia, along the entire floodplain region of the Ama- zon River, there is a large extension of climatic conditions that are not analogous to the histori- cal variability of the rest of the biome. The forward velocity is higher along a diagonal extending from the central-western region to the southeastern region, but presents some higher velocity hotspots in parts of eastern Amazon. Under the assumption that species are unable to acclimate/adapt to changing climate [16], these results are a cause for concern, given the significant role of PAs, and especially ILs, on forest conservation, which reduces forest carbon emissions and contributes to climate change mitigation [47]. Many studies have shown that Amazonian PAs and ILs act as buffers for the advance of the deforestation frontier, reducing deforestation [11], degradation [46] and fire occurrence [48]. These results imply that despite the important role of Pas and ILs in contain-

ing changes in land use and cover, they are not immune to climate change, and therefore, cli- mate adaptation measures are needed to increase their resilience. Adaptive management strategies [49] can be applied in view of the impacts on the distribu- tion of species projected by the climate velocity metric. In areas with high forward velocity, cli- mate adaptation strategies are recommended, and in extreme cases, assisted migration [21, 50]. Conservation strategies that prioritize the protection and connectivity of climatically het- erogeneous landscapes are also suggested, either through the expansion of existing reserves or the creation of new corridors that facilitate movement in response to climate change [51]. In areas with high backward velocity, intervention through assisted migration is indicated, either in the context of conservation of species with restricted distributions or on a large scale, such as movement of populations of widely distributed species in regular reforestation operations [52]. Actions focused on monitoring invasive alien species that may occupy empty niches may also be necessary [21]. PLOS ONE Amazonian protected areas and climate velocity PLOS ONE | https://doi.org/10.1371/journal.pone.0286457 June 22, 2023 9 / 14

However, comparative analyzes such as those carried out in this work (UAs vs PAs; ITs vs PNAs) are not without limitations. For example, they can mask the real risk that different areas will be subjected to in terms of actual climate velocity values. In our analysis, exposure to moderate and low velocities means a spectrum of 0–13 km/year on forward velocity and 0–21 km/year for backward velocity, as defined by the classes using quantile method. In previous research, PAs in North America were classified as having high climate velocity when they obtained values above 10 km/year [53]. Therefore, here, even the classification of moderate velocities, depending on the dispersion capacity and sensitivity of the species, may indicate high risk. In the same way, whilst the backward velocity for PAs revealed a lower median value than that of UAs, this value was still 12 km/year, which may imply severe threats to some species. Finally, it should be noted that this work does not take into account the potential influences of changes in land use and cover on species dispersion. However, deforestation and degrada- tion can influence the dispersion of species, both via the formation of barriers that restrict the movement of species between analogous climates [31], as well as potentiating the effects of cli- mate change on local [54], regional [55] and global scales [56]. Also, while any environmental variable can be utilized to calculate climate velocity, temperature is the most commonly used

variable in research studies, as it has a significant influence on species distribution at the land- scape level. The bioclimatic variables mean annual temperature and total annual precipitation were selected as they represent biologically relevant trends on species distributions. However, it should be also noted that while annual mean values are better suited for predicting changes across the entire range of a species, extreme temperature values are appropriate for predicting a species’ tolerance limits, which represents a promising approach for future studies [21]. The methodological limitations inherent to analyzes based on climate change metrics and the degree of uncertainty of climate projections, both of which are extensively debated should also be kept in mind [57, 58]. Conclusion The effects of backward velocity will be greater in magnitude considering the median across the entire Amazon biome. This striking pattern of backward velocity occurs both for climatic conditions with and without future analogs, with emphasis on a large expanse of no-analogs across central Amazonia. This indicates that high temperatures and changes in precipitation will exceed the historical variability that occurs in the biome, which may have effects on eco- system functioning and services of floodplain forests. Despite this, the network of Protected Areas will be less exposed to the impacts of the backward velocity than unprotected areas emphasizing the importance of these areas as a tool in conservation. In contrast, for forward

velocity impacts, the PA network will be slightly more exposed than unprotected areas indicat- ing that the current spatial arrangement of the PA network is still not the most suitable for minimizing the impacts of climate redistribution. Furthermore, at both velocities, the potential impacts will be greater on ILs than on PAs. However, risks to PAs should also be emphasized as the range of no-analogs is greater in this land category. Therefore, despite the important role of PAs in reducing land use changes, they are not immune to the effects of climate change, and new management strategies, specific to each area and that allow adaptation to global changes, will be necessary. Despite the methodological limitations inherent to analyzes based on climate change metrics [57], it is urgent to refine information about the effect of global changes on biota in the tropics, With the input provided here, we hope to further improve our abilities to identify the different needs for adaptive management of PAs in the face of climate PLOS ONE Amazonian protected areas and climate velocity PLOS ONE | https://doi.org/10.1371/journal.pone.0286457 June 22, 2023 10 / 14

change, providing decision-makers with more personalized data for the conservation of biodi- versity in the Amazon. Supporting information S1 Fig. Referenced territorial limits. Location of the Amazon biogeographic boundary (based on RAISG) and protected areas (based on WDPA). (TIF) S2 Fig. Maps of the difference between projected and current climate variables (climatic delta) and historical annual variability. Spatial distribution of climate deltas (2050–1990) of temperature and precipitation and spatial distribution of historical annual variability of tem- perature and precipitation. (TIF) S3 Fig. Nine-quadrant bivariate plot with a joint characterization of forward and backward velocities, where the range of velocity values is divided into low, moderate, and high cate- gories using three equal area quantiles. (TIF) S1 Table. General circulation models (GCM) used for projections of bioclimatic variables. 5 General circulation models (GCM) from CMIP6 used to assess the future distribution of cli- mate conditions. A multi-model ensemble was generated based on all the GCMs. (XLSX) Acknowledgments We are grateful to Emma Docherty for review the English language in the manuscript and V. H.F Gomes, M. Adami and D.R. Galbraith for their valuable suggestions throughout the devel- opment of this work. Author Contributions Conceptualization: Calil Torres-Amaral, Luciano Jorge Serejo dos Anjos, Ima Ce´lia Gui- marães Vieira, Everaldo Barreiros de Souza. Data curation: Calil Torres-Amaral, Luciano Jorge Serejo dos Anjos.

Formal analysis: Calil Torres-Amaral, Luciano Jorge Serejo dos Anjos. Funding acquisition: Calil Torres-Amaral, Everaldo Barreiros de Souza. Investigation: Calil Torres-Amaral, Luciano Jorge Serejo dos Anjos. Methodology: Calil Torres-Amaral, Luciano Jorge Serejo dos Anjos. Project administration: Calil Torres-Amaral, Ima Ce´lia Guimarães Vieira, Everaldo Barreiros de Souza. Resources: Calil Torres-Amaral. Software: Calil Torres-Amaral, Luciano Jorge Serejo dos Anjos. Supervision: Calil Torres-Amaral, Luciano Jorge Serejo dos Anjos, Ima Ce´lia Guimarães Vieira, Everaldo Barreiros de Souza. Validation: Calil Torres-Amaral. Visualization: Calil Torres-Amaral, Luciano Jorge Serejo dos Anjos. PLOS ONE Amazonian protected areas and climate velocity PLOS ONE | https://doi.org/10.1371/journal.pone.0286457 June 22, 2023 11 / 14

Writing – original draft: Calil Torres-Amaral. Writing – review & editing: Calil Torres-Amaral, Luciano Jorge Serejo dos Anjos, Ima Ce´lia Guimarães Vieira, Everaldo Barreiros de Souza. References 1. Pecl G, Araujo MB, Bell JD, Blanchard J, Bonebrake TC, Pecl GT, et al. Biodiversity redistribution under climate change: Impacts on ecosystems and human well-being Publication Details. Science (80-) [Inter- net]. 2017; 355(6332):1–9. Available from: http://ro.uow.edu.au/smhpapers/4629 2. Garcia RA, Cabeza M, Rahbek C, Arau´ jo MB. Multiple dimensions of climate change and their implica- tions for biodiversity. Science (80-). 2014; 344(6183). https://doi.org/10.1126/science.1247579 PMID: 24786084 3. Williams JW, Jackson ST, Kutzbach JE. Projected distributions of novel and disappearing climates by 2100 AD. PNAS. 2007; 104(14):5738–42. https://doi.org/10.1073/pnas.0606292104 PMID: 17389402 4. Bellard C, Cleo Bertelsmeier, Paul Leadley, Wilfried Thuillerand FC. Impacts of climate change on the future of biodiversity. Ecol Lett. 2012; 15:365–77. https://doi.org/10.1111/j.1461-0248.2011.01736.x PMID: 22257223 5. Balmford A, Bruner A, Cooper P, Costanza R, Farber S, Green RE, et al. Ecology: Economic reasons for conserving wild nature. Science (80-). 2002; 297(5583):950–3. 6. Bruner AG, Gullison RE, Rice RE, Da Fonseca GAB. Effectiveness of parks in protecting tropical biodi- versity. Science (80-). 2001; 291(5501):125–8. https://doi.org/10.1126/science.291.5501.125 PMID: 11141563

7. Rodrigues A J A, MI B, L B, M B, RM C, et al. Effectiveness of the global protected area network in rep- resenting species diversity. Nature. 2004;640–3. https://doi.org/10.1038/nature02422 PMID: 15071592 8. Thomas CD, Gillingham PK. The performance of protected areas for biodiversity under climate change. Biol J Linn Soc. 2015; 115(3):718–30. 9. RAISG. Amazoˆ nia Sob Pressão. ISA-Inst. São Paulo; 2020. 42 p. 10. Walker W, Baccini A, Schwartzman S, Rı´os S, Oliveira-miranda MA, Augusto C, et al. Forest carbon in Amazonia: the unrecognized contribution of indigenous territories and protected natural areas. Carbon Manag. 2014; 3004:479–85. 11. Jusys T. Changing patterns in deforestation avoidance by different protection types in the Brazilian Amazon. PLoS One. 2018; 13(4):1–16. https://doi.org/10.1371/journal.pone.0195900 PMID: 29689071 12. Feeley KJ, Silman MR. Disappearing climates will limit the efficacy of Amazonian protected areas. Divers Distrib. 2016; 22(11):1081–4. 13. Groves CR, Game ET, Anderson MG, Cross M, Enquist C, Ferdaña Z, et al. Incorporating climate change into systematic conservation planning. Biodivers Conserv. 2012; 21(7):1651–71. 14. Hannah L. A global conservation system for climate-change adaptation: Special section. Conserv Biol. 2010; 24(1):70–7. 15. Hoegh-Guldberg O. DJ, Taylor M, Bindi M, Brown S, Camilloni I, et al. Impacts of 1.5˚C Global Warming on Natural and Human Systems. In: Global Warming of 15˚C. 2018. p. 175–312. 16. Feeley KJ, Rehm EM, Machovina B. perspective: The responses of tropical forest species to global cli-

mate change: acclimate, adapt, migrate, or go extinct? Front Biogeogr. 2012; 4(2). 17. Perez TM, Stroud JT, Feeley KJ. Thermal trouble in the tropics. Science (80). 2016; 351(6280):1392–3. 18. Feeley KJ, Davies SJ, Perez R, Hubbell SP, Foster RB. Directional changes in the species composition of a tropical forest. Ecology. 2011; 92(4):871–82. https://doi.org/10.1890/10-0724.1 PMID: 21661550 19. Feeley KJ, Malhi Y, Zelazowski P, Silman MR. The relative importance of deforestation, precipitation change, and temperature sensitivity in determining the future distributions and diversity of Amazonian plant species. Glob Chang Biol. 2012; 18(8):2636–47. 20. Carroll C, Lawler JJ, Roberts DR, Hamann A. Biotic and Climatic Velocity Identify Contrasting Areas of Vulnerability to Climate Change. PLoS One. 2015; 10(10). https://doi.org/10.1371/journal.pone. 0140486 PMID: 26466364 21. Brito-Morales I, Garcı´a Molinos J, Schoeman DS, Burrows MT, Poloczanska ES, Brown CJ, et al. Cli- mate Velocity Can Inform Conservation in a Warming World. Trends Ecol Evol. 2018; 33(6):441–57. https://doi.org/10.1016/j.tree.2018.03.009 PMID: 29716742 PLOS ONE Amazonian protected areas and climate velocity PLOS ONE | https://doi.org/10.1371/journal.pone.0286457 June 22, 2023 12 / 14

22. Clark D.A., Piper S.C., Keeling C.D. & Clark D.B. Tropical rain forest tree growth and atmospheric car- bon dynamics linked to interannual temperature variation during 1984–2000. Proceedings of the National Academy of Sciences. 2003; USA, 100, 5852–5857. https://doi.org/10.1073/pnas. 0935903100 PMID: 12719545 23. Feeley K.J., Wright S.J., Supardi M.N.N., Kassim A.R. & Davies S.J. Decelerating growth in tropical for- est trees. Ecology Letters. 2003; 10, 461–469. 24. Docherty Emma M. et al. Long-term drought effects on the thermal sensitivity of Amazon forest trees. Plant, Cell & Environment, v. 46, n. 1, p. 185–198, 2023. https://doi.org/10.1111/pce.14465 PMID: 36230004 25. Esquivel-Muelbert A, Baker TR, Dexter KG, Lewis SL, Brienen RJW, Feldpausch TR, et al. Composi- tional response of Amazon forests to climate change. Glob Chang Biol. 2019; 25(1):39–56. https://doi. org/10.1111/gcb.14413 PMID: 30406962 26. Jirinec Vitek et al. Morphological consequences of climate change for resident birds in intact Amazonian rainforest. Science Advances, v. 7, n. 46, p. eabk1743, 2021. https://doi.org/10.1126/sciadv.abk1743 PMID: 34767440 27. Wright SJ, Muller-Landau HC, Schipper J. The future of tropical species on a warmer planet. Conserv Biol. 2009; 23(6):1418–26. https://doi.org/10.1111/j.1523-1739.2009.01337.x PMID: 20078642 28. Loarie SR, Duffy PB, Hamilton H, Asner GP, Field CB, Ackerly DD. The velocity of climate change. Nature [Internet]. 2009; 462(7276):1052–5. Available from: https://doi.org/10.1038/nature08649 PMID: 20033047

29. Carroll C, Roberts DR, Michalak JL, Lawler JJ, Nielsen SE, Stralberg D, et al. Scale-dependent comple- mentarity of climatic velocity and environmental diversity for identifying priority areas for conservation under climate change. Glob Chang Biol. 2017; 23(11):4508–20. https://doi.org/10.1111/gcb.13679 PMID: 28267245 30. Feeley KJ, Rehm EM. Amazon’s vulnerability to climate change heightened by deforestation and man- made dispersal barriers. Glob Chang Biol. 2012; 18(12):3606–14. 31. Ribeiro BR, Sales LP, Loyola R. Strategies for mammal conservation under climate change in the Ama- zon. Biodivers Conserv [Internet]. 2018; 27(8):1943–59. Available from: https://doi.org/10.1007/ s10531-018-1518-x 32. Hubbell SP, He F, Condit R, Borda-De-A´ gua L, Kellner J, Ter Steege H. How many tree species are there in the Amazon and how many of them will go extinct? Light Evol. 2009; 2:107–26. 33. RAISG. Amazoˆ nia Sob Pressão. ISA-Inst. São Paulo; 2020. 42 p. 34. Dinerstein E, Olson D, Joshi A, Vynne C, Burgess ND, Wikramanayake E, et al. An Ecoregion-Based Approach to Protecting Half the Terrestrial Realm. Bioscience. 2017; 67(6). https://doi.org/10.1093/ biosci/bix014 PMID: 28608869 35. UNEP-WCMC (2020), The World Database on Protected Areas (WDPA) statistics. Cambridge, UK: UNEP- WCMC. Accessed on: 22/9/2020. 36. Fick SE, Hijmans RJ. WorldClim 2: new 1-km spatial resolution climate surfaces for global land areas. Int J Climatol. 2017; 37(12):4302–15. 37. Stralberg D, Carroll C, Pedlar JH, Wilsey CB, Mckenney DW, Nielsen SE. Macrorefugia for North Amer-

ican trees and songbirds: Climatic limiting factors and multi-scale topographic influences. Glob Ecol Bio- geogr. 2018; 27 (6)(April 2017):1–14. 38. Garcı´a-Lo´ pez JM, Allue´ C. Modelling future no-analogue climate distributions: A world-wide phytocli- matic niche-based survey. Glob Planet Change [Internet]. 2013; 101:1–11. 39. Flato G. Evaluation of climate models. Clim Chang 2013 Phys Sci Basis Work Gr I Contrib to Fifth Assess Rep Intergov Panel Clim Chang. 2014; 9781107057:741–866. 40. Eltahir EA, Bras RL. Precipitation recycling in the Amazon basin. Q J R Meteorol Soc. 1994; 120 (518):861–80. 41. Eyring V, Bony S, Meehl GA, Senior CA, Stevens B, Stouffer RJ, et al. Overview of the Coupled Model Intercomparison Project Phase 6 (CMIP6) experimental design and organization. Geosci Model Dev. 2016; 9(5):1937–58. 42. O’Neill BC, Tebaldi C, Van Vuuren DP, Eyring V, Friedlingstein P, Hurtt G, et al. The Scenario Model Intercomparison Project (ScenarioMIP) for CMIP6. Geosci Model Dev. 2016; 9(9):3461–82. 43. Garcı´a Molinos J, Schoeman DS, Brown CJ, Burrows MT. VoCC: An r package for calculating the velocity of climate change and related climatic metrics. Methods Ecol Evol. 2019; 10(12):2195–202. 44. Hamann A, Roberts DR, Barber QE, Carroll C, Nielsen and SE. Velocity of climate change algorithms for guiding conservation and management. Glob Chang Biol. 2015; 21(2):997–1004. https://doi.org/10. 1111/gcb.12736 PMID: 25310933 PLOS ONE Amazonian protected areas and climate velocity PLOS ONE | https://doi.org/10.1371/journal.pone.0286457 June 22, 2023 13 / 14

45. Chen IC, Hill JK, Ohlemu¨ller R, Roy DB, Thomas CD. Rapid range shifts of species associated with high levels of climate warming. Science (80-). 2011; 333(6045):1024–6. https://doi.org/10.1126/ science.1206432 PMID: 21852500 46. Abril G, Martinez JM, Artigas LF, Moreira-Turcq P, Benedetti MF, Vidal L, et al. Amazon River carbon dioxide outgassing fuelled by wetlands. Nature. 2014; 505(7483):395–8. https://doi.org/10.1038/ nature12797 PMID: 24336199 47. Walker WS, Gorelik SR, Baccini A, Aragon-osejo JL, Josse C, Meyer C. The role of forest conversion, degradation, and disturbance in the carbon dynamics of Amazon indigenous territories and protected areas. PNAS. 2020; 117(6):3015–25. https://doi.org/10.1073/pnas.1913321117 PMID: 31988116 48. Nepstad D, Schwartzman S, Bamberger B, Santilli M, Ray D, Schlesinger P, et al. Inhibition of Amazon Deforestation and Fire by Parks and Indigenous Lands. Conserv Biol. 2006; 20(1):65–73. https://doi. org/10.1111/j.1523-1739.2006.00351.x PMID: 16909660 49. Hannah L, Ikegami M, Hole DG, Seo C, Butchart SHM, Peterson AT, et al. Global Climate Change Adaptation Priorities for Biodiversity and Food Security. PLoS One. 2013; 8(8). https://doi.org/10.1371/ journal.pone.0072590 PMID: 23991125 50. Thomas CD. Translocation of species, climate change, and the end of trying to recreate past ecological communities. Trends Ecol Evol [Internet]. 2011; 26(5):216–21. Available from: https://doi.org/10.1016/j. tree.2011.02.006 PMID: 21411178

51. Ackerly DD, Loarie SR, Cornwell WK, Weiss SB, Hamilton H, Branciforte R, et al. The geography of cli- mate change: Implications for conservation biogeography. Divers Distrib. 2010; 16(3):476–87. 52. Pedlar JH, McKenney DW, Aubin I, Beardmore T, Beaulieu J, Iverson L, et al. Placing forestry in the assisted migration debate. Bioscience. 2012; 62(9):835–42. 53. Batllori E, Miller C, Sean P, Max AP. Potential relocation of climatic environments suggests high rates of climate displacement within the North American protection network. Glob Chang Biol. 2017;(Janu- ary):1–12. https://doi.org/10.1111/gcb.13663 PMID: 28211141 54. Laurance WF, Camargo JLC, Luizão RCC, Laurance SG, Pimm SL, Bruna EM, et al. The fate of Ama- zonian forest fragments: A 32-year investigation. Biol Conserv. 2011; 144(1):56–67. 55. Laurance WF, Bruce Williamson G. Positive feedbacks among forest fragmentation, drought, and cli- mate change in the Amazon. Conserv Biol. 2001; 15(6):1529–35. 56. Werth D, Avissar R. The local and global effects of Amazon deforestation. J Geophys Res. 2002;107 (D20). 57. Iba´ ñez I, Clark JS, Dietze MC, Feeley K, Hersh M, LaDeau S, et al. Predicting biodiversity change: Out- side the climate envelope, beyond the species-area curve. Ecology. 2006; 87(8):1896–906. https://doi. org/10.1890/0012-9658(2006)87[1896:pbcotc]2.0.co;2 PMID: 16937626 58. Arau´jo, Miguel B.; Peterson A. Townsend. Uses and misuses of bioclimatic envelope modeling. Ecol- ogy, v. 93, n. 7, p. 1527–1539, 2012. https://doi.org/10.1890/11-1930.1 PMID: 22919900 PLOS ONE

Amazonian protected areas and climate velocity PLOS ONE | https://doi.org/10.1371/journal.pone.0286457 June 22, 2023 14 / 14