Primate conservation in the Arc of Deforestation: a case study of Vieira's titi monkey Plecturocebus vieirai

Abstract

Fifty years of deforestation in the Arc of Deforestation have put at risk species survival, ecosystem services and the stability of biogeochemical cycles in Amazonia, with global repercussions. In response, we need to understand the diversity, distribution and abundance of flagship species groups, such as primates, which can serve as umbrella species for broad biodiversity conservation strategies and help mitigate climate change. Here we identify the range, suitable habitat areas and population size of Vieira's titi monkey Plecturocebus vieirai and use it as an emblematic example to discuss biodiversity conservation and climate change mitigation in one of the largest deforestation frontiers. Our findings show that deforestation for agriculture and cattle-ranching expansion is the major threat to P. vieirai and is responsible for present (56%) and projected (14%) reductions in habitat area and population size. We also found that human-driven climate change affects the P. vieirai niche negatively, triggering habitat degradation and further population decline even inside protected areas. Primate watching can be a profitable alternative to forest exploitation on private, public or Indigenous lands in the Arc of Deforestation and is a way to shift the traditional, predatory extraction of natural resources from Amazonia towards sustainable land use based on biodiversity conservation at local, regional and global scales, local people's welfare and climate change mitigation. New models of land use and income generation are required to protect the unique natural and human heritages of the Arc of Deforestation and the life-supporting ecosystem services and products provided by Amazonia.

Introduction

Amazonia, the largest tropical forest, is paramount as a functioning ecosystem for human welfare. It covers 5.3 million km² across nine South American countries and provides life-supporting resources such as food, water and medicines, regulates weather and rainfall, and stores carbon (Fearnside, 2003, 2017; Soares-Filho et al., 2006; Malhi et al., 2008; Hubau et al., 2020; Harris et al., 2021). Nonetheless, 5 decades of continuous and unregulated expansion of agriculture and settlements have transformed southern Amazonia, adjacent ecotonal forests and the northern Cerrado into the largest global deforestation frontier (Kirby et al., 2006; FAO, 2016; Silva et al., 2019; Montibeller et al., 2020), known as the Arc of Deforestation, significantly affecting the functioning of the Amazonian ecosystem (Nobre et al., 2016; Gatti et al., 2021).

There are 52 primate species in the Arc of Deforestation. The majority are endemic to this region and approximately half are threatened or near threatened with extinction (Costa-Araújo et al., 2021b). At the same time, there is a scarcity of baseline data on these primates, including a significant taxonomic deficit (Costa-Araújo et al., 2019; Byrne et al., 2021), which hampers the protection of these species (Rylands & Mittermeier, 2014). An understanding of species diversity, distribution and abundance is therefore necessary to clarify and quantify extinction risks and to provide a scientific foundation for the conservation of the poorly studied primates endemic to the Arc of Deforestation.

Investment in research into and the conservation of primates is especially important as they play a vital role in ecosystem functioning (Peres et al., 2016; Trolliet et al., 2016; Heymann et al., 2017), and they are the most vulnerable vertebrate group globally (Arroyo-Rodríguez et al., 2013, 2017), with c. 60% of all primate species threatened to some degree (Estrada et al., 2017). Moreover, primates are valuable as flagship species (Dietz et al., 1994; Chapman et al., 2020), and so the protection of their habitats in the Arc of Deforestation will not only contribute to their conservation but also provide protection to other species endemic to this region, contributing to biodiversity conservation with impacts at local, regional and global scales.

Here we identify the areas of suitable habitat, delimit the geographical distribution, estimate the population size and assess the conservation status of Vieira's titi monkey Plecturocebus vieirai (Plate 1). This species was recently discovered in the Arc of Deforestation (Gualda-Barros et al., 2012) but is known only from its pelage colour and occurrence in three localities and is therefore considered Data Deficient (Alonso et al., 2018; IUCN, 2021). Using our findings, we discuss the threats to primates and the opportunities for biodiversity conservation coupled with climate change mitigation and income generation in the Arc of Deforestation using P. vieirai as an emblematic example, a species that we assess to be Critically Endangered based on the IUCN Red List criteria (IUCN, 2019).

Plate 1 Vieira's titi monkey Plecturocebus vieirai. Photo: F. Reis.

Study area

The study area is located in the Tapajós–Xingu interfluve, a region within the Arc of Deforestation that is especially susceptible to land-use change (Laurance et al., 2002; Fig. 1). The Arc of Deforestation extends across southern Amazonia, ecotonal forests and the northern Cerrado from its eastern edge in the states of Pará and Maranhão to the state of Acre in the west (Fearnside et al., 2009; Silva et al., 2019). This region accounts for almost half of the total global land-use changes during 1990–2015 (FAO, 2016). Although sustainable development was expected for the Tapajós–Xingu region 2 decades ago (Nepstad et al., 2002), it became an epicentre of deforestation as a result of logging and slash-and-burn clearance for agriculture and cattle ranching as well as legal and illegal gold mining, affecting even protected areas and Indigenous lands (Printes, 2017; Montibeller et al., 2020).

Fig. 1 The Tapajós–Xingu interfluve, southern Amazonia, Brazil, showing the location of fires, illegal mining, hydroelectric dams as of 2020 (RAISG, 2020), protected areas and Indigenous lands (MMA, 2018).

Methods

Data collection

We collected new occurrence records in field expeditions within the Tapajós–Xingu interfluve during 2015–2019 to model areas of suitable habitat and to delimit the geographical distribution of P. vieirai. We reviewed the literature and examined specimens housed in museum collections in Brazil, the USA and Europe for additional occurrence records. To estimate the size and density of the P. vieirai population, we surveyed four areas during 2016–2018 using the linear transect method (Buckland et al., 1993). In each area we established a transect of 5 km covering mature and secondary forests. Each transect was surveyed at a constant speed of c. 1.25 km/h twice daily, at 07.00–11.00 and 14.00–18.00 by DAS, LFS and RMA. We noted the number of individuals and the perpendicular distance between the group and the transect using a GPS and a measuring tape; we considered each of the two daily surveys to be independent (Peres & Cunha, 2011). The four forest areas surveyed are on the east bank of the middle Teles Pires River in the municipalities of Cláudia and Sinop, Mato Grosso State, Brazil, c. 50 km from the the type locality of P. vieirai.

Data analysis

We used the occurrence records, and environmental variables from WorldClim 2.1 (Fick & Hijmans, 2017) and CliMond (Kriticos et al., 2012), to model habitat suitability and delimit the geographical distribution of P. vieirai. We selected only spatially independent records (n = 33) from our dataset (n = 99 records), applying a threshold of 10 km between records. We eliminated autocorrelated environmental variables to avoid model overfitting (Pearson's correlation test r > 0.80, P < 0.05; Supplementary Table 1; Callegari-Jacques, 2003; Mateo et al., 2013). We converted the 11 environmental variables selected to a 2.5-min scale using the raster package (Hijmans & Etten, 2012) in R 4.1 (R Core Team, 2018). We modelled habitat suitability using four algorithms adequate for the type of data available (presence and pseudo-absence records), using the Biomod2 package (Thuiller et al., 2016) in R: artificial neural networks (Ripley, 1996), generalized boosted models (Friedman, 2001), random forest (Breiman, 2001) and maximum entropy (Phillips et al., 2006). We selected these four algorithms to avoid model overfitting that could result from the use of algorithms suitable for presence-only and presence and true absence records from long-term surveys (Andrade et al., 2020; Silva et al., 2020). For each algorithm we established five datasets, each composed of 10,000 background records randomly distributed throughout the study area. We then used 70% of the records for training and 30% for evaluating model fitting. In total we performed 200 runs (four algorithms, 10 runs of cross-validation, five sets of random background points), with 1,000 iterations each. To check the accuracy of models we used true skill statistical (TSS) analysis (Allouche et al., 2006) and the area under the curve value (AUC) of the receiver operating characteristic curve, incorporating a binomial probability as a null model (Phillips et al., 2006). The AUC and TSS values vary from 0 to 1 based on the specificity and sensitivity of the species’ response to environmental variables. We selected models with AUC > 0.7 and TSS > 0.4 (Buisson et al., 2010) using the mean suitability value of each grid (2.5-min scale) and the minimum omission method (Silva et al., 2017, 2020). To identify the environmental variables that best explain the occurrence of the species we used the jackknife test (Phillips et al., 2006).

Using ArcGIS 10.1 (Esri, Redlands, USA), we delimited the geographical distribution of P. vieirai from the occurrence records, the ensemble model, knowledge of the distribution of congeneric, neighbouring species and considering that rivers restrict the occurrence of primate species in Amazonia (Mourthé et al., 2022). Finally, we overlaid the species distribution with raster layers of vegetation classes (Souza et al., 2020), elevation (which is a constraint for species occurrence; Jarvis et al., 2008) and forest cover (Soares-Filho et al., 2006), to extract the area of suitable habitat available for P. vieirai in the present (2020) and in the future (2044). We used a threshold of 24 years, representing three P. vieirai generations (Veiga et al., 2011; Defler & García, 2012), as a baseline to estimate the availability of suitable habitat in the future and to assess the species' conservation status. We calculated the extent of occurrence and the area of suitable habitat (Brooks et al., 2019) in 2020 and in 2044 using the R package red (Cardoso, 2017, 2018). We validated the area of suitable habitat using our dataset of location records overlain on a binary map of presence or absence predicted from the environmental variables.

We used the data from our linear transect surveys to estimate population density and sighting rates with Distance 7.1 (Thomas et al., 2010). We carried out this analysis by pooling all sightings from the four survey areas using the hazard-rate model, a simple polynomial adjustment and 45 m as the effective sighting distance (χ ² = 0.80; df = 3; P = 0.84); the final model was selected using the Akaike information criterion. We used the oldest available satellite data of forest cover over the species range (1985; Souza et al., 2020) as a conservative baseline to extract an approximate estimate of the original habitat area of P. vieirai.

Results

We gathered 96 occurrence records of P. vieirai in addition to the three records available from the species description: 90 from fieldwork, four from the literature (Miranda et al., 2014; Vendramel, 2016) and two from specimens stored in the Museu Paraense Emílio Goeldi (MPEG 246, 21837; Supplementary Table 2). All records are from 150–410 m altitude and within the boundaries of Amazonian forests of the Tapajós domain, except for three records that fall within the boundaries of the Cerrado (Instituto Brasileiro de Geografia e Estatística, 2019) but were obtained in the Amazonia–Cerrado ecotonal forest (Silva et al., 2019). We also gathered a new record of Plecturocebus moloch on the north bank of the Cristalino River (9.552926° S, 55.928568° W; Fig. 2), which is the southernmost documented record of this species in the Tapajós–Xingu interfluve and helped us to delimit the geographical distribution of P. vieirai, which is expected to occur on the opposite bank of the river according to our modelling (see below).

Our combined results from habitat modelling and occurrence records show that P. vieirai occurs in an area of 226,054 km² and is limited in the east by the Xingu River, in the north by the Irirí River, in the west by the Irirí, Cristalino and Teles Pires rivers and extending southwards to the upper Teles Pires and Xingu rivers (Fig. 2). The linear transect surveys totalled 1,192 km of trails walked, with 78 group encounters and 86 individuals observed in the four areas, resulting in a mean density of 0.03 groups/ha and 0.07 individuals/ha (Table 1), with mean group sizes ranging from 1.5 (95% CI 1.0–4.2) to 2.62 (95% CI 2.13–3.23).

Fig. 2 Geographical distribution and potential habitat of Plecturocebus vieirai in the Tapajós–Xingu interfluve: (a) records of P. vieirai and Plecturocebus moloch, and habitat suitability modelled using environmental variables; (b) suitable forest habitat area in 2020; (c) modelled potentially suitable habitat area in 2044. AOH, area of habitat (based on environmental suitability and forest availability).

Table 1 Survey effort, number of sightings, sighting rates and density estimates (per ha) for Vieira's titi monkey Plecturocebus vieirai obtained during 3 years (2016–2018) of systematic transect surveys in the southern part of the species range, Cláudia and Sinop municipalities, Mato Grosso State, Brazil.

¹ Locality names as per Supplementary Table 2.

Considering the data on past and projected future forest cover (Soares-Filho et al., 2006; Souza et al., 2020) and our new range delimitation for P. vieirai, in 1985 forests covered 90% of the range of P. vieirai (204,191 km²) but in 2044 only 30% (54,731 km²) of these forests will remain. Our modelling shows that suitable habitat (Table 2) for P. vieirai is currently 56% of the original area and that 58% of this is within protected areas and Indigenous lands. By 2044, according to our model, only 14% of the original P. vieirai habitat will remain, 88% within protected areas and Indigenous lands.

Table 2 Extent of occurrence, suitable habitat, suitable habitat in protected areas and forest cover for P. vieirai in 2020 and after three P. vieirai generations (2044), with actual (2020) or predicted habitat loss (2044).

¹Relative to original forest cover in 1985.

²Relative to suitable habitat.

We found that variables related to rainfall (specifically the wettest quarter and the driest month) explain 60% of the likelihood of P. vieirai occurrence (AUC = 0.999, TSS = 0.996) and therefore these variables are the best indicator of habitat suitability for this species (Supplementary Table 3).

Discussion

Our findings indicate that habitat loss as a result of conversion of forest into large-scale agricultural monocultures and pastures for cattle ranching is the main threat to P. vieirai. Our model showed that because of deforestation the habitat of P. vieirai has been reduced by 56% and only 14% will remain by 2044. We expect a concomitant decline in the population size of P. vieirai. We assess P. vieirai to be Critically Endangered (Costa-Araújo et al., 2022) based on IUCN criterion A3bc (IUCN, 2019), given that our estimates suggest a projected population size reduction of ≥ 80% within the next three P. vieirai generations (A3), and that these estimates are based on an index of abundance that is appropriate for the taxon (b) and a decline in the area of occupancy of > 80% (c).

Additionally, decreases in the suitability of any remaining habitat because of climate change compound the threats to the long-term survival of P. vieirai. Greenhouse gas emissions from human activities are causing extended dry seasons, increased frequencies of droughts and forest fires and reductions in rainfall, forest biomass and net primary production in southern Amazonia (Jiménez-Muñoz et al., 2016; Fearnside, 2017; Brando et al., 2020; Sales et al., 2020; IPCC, 2021). Conversely, we found that precipitation has a positive relationship with habitat suitability for P. vieirai, and forest biomass and primary production also have a positive relationship with titi monkey occurrence (Costa-Araújo et al., 2021a). Therefore, climate (IPCC, 2021) and forest (Soares-Filho et al., 2006) changes projected for southern Amazonia will negatively affect the biotic and abiotic dimensions of the P. vieirai niche in the next quarter of a century, further exacerbating the threats to the species’ long-term survival.

Habitat loss from deforestation and degradation because of climate change have distinct implications for the conservation of P. vieirai populations at the northern and southern extents of the species’ range. Deforestation and forest fires are and will remain high (Soares-Filho et al., 2006; Brando et al., 2020) in the southern portion of the range of P. vieirai, where most forests are on private lands and where there are few and only small protected areas. Although deforestation has been and is expected to continue to be low (Soares-Filho et al., 2006) in the northern portion of the range of P. vieirai because of the existence of an extensive system of protected areas and Indigenous lands, this habitat is not safe from degradation driven by climate change.

Therefore, the populations in the southern portion of the range of P. vieirai require immediate conservation efforts. Beyond protecting primate populations (Paim et al., 2019), the demarcation of Indigenous lands and establishment of public and private protected areas in the Tapajós–Xingu interfluve would safeguard other endemic species and protect representative tracts of Amazonia and the Amazonia–Cerrado ecotone, a unique and almost entirely unprotected ecosystem (Nepstad et al., 2006; Marques et al., 2019). These protected areas would also contribute to reducing the extent of land-use change and fire regimes in this region, maintain carbon stocks and sinks and generate income through REDD+ projects (Nogueira et al., 2018). For the same reasons, the effective protection of reservas legais (an area of 80% that legally must be preserved as forest in all private land holdings in Brazilian Amazonia) needs to be enforced and private landowners rewarded appropriately (Schielein & Börner, 2018).

Law enforcement, the establishment of public protected areas and the demarcation of Indigenous lands rely on governmental administration, but changes in policies and demonstrated inaction and a lack of regulation have resulted in increasing threats to Amazonian forests, biodiversity and traditional peoples in Brazil (Ferrante & Fearnside, 2019; Begotti & Peres, 2020). Moreover, the establishment of private protected areas and the protection of reservas legais are unattractive to landowners profiting from the conversion of forests into agricultural landscapes in the Tapajós–Xingu region (Crouzeilles et al., 2012; Printes, 2017). The model of income generation and the policies in Brazil that permit unsustainable extraction of natural resources in the Arc of Deforestation needs to shift towards the protection of the livelihoods of local people, biodiversity, climate change mitigation and the maintenance of the ecosystem services provided by Amazonia (Moutinho et al., 2016; Nobre et al., 2016; Carvalho et al., 2019).

In this context, private landowners are the key stakeholders (Nepstad et al., 2002; Soares-Filho et al., 2006, Fearnside, 2017) for biodiversity conservation and the protection of representative tracts of remaining forests in the Arc of Deforestation, which would contribute to climate change mitigation (Fearnside, 2003, 2009; Nogueira et al., 2018). The owners of large farms are responsible for 70% of deforestation in private areas in Amazonia (Fearnside, 2017) but could change their model of land use, which is currently putting at risk the ecosystem services and the resources provided by the biome (Lovejoy & Nobre, 2018). The official custodians of natural resources in protected areas (traditional and native peoples, local and federal governments) are also key conservation stakeholders (Nepstad et al., 2006).

As an important tool for the conservation of species and habitats (Russon & Wallis, 2014), we believe that primate-watching tourism could be a viable, sustainable alternative to monoculture agriculture and cattle-ranching expansion in the Arc of Deforestation. Firstly, there is a wealth of opportunities for primate watching in the Arc of Deforestation, with 52 species of primates (Costa-Araújo et al., 2021b). Secondly, primate-watching services on private lands, public protected areas and Indigenous lands are permitted under Brazilian environmental regulations. Thirdly, the logistics to facilitate ecotourism in the Arc of Deforestation are met by a network of roads, airports and lodging facilities. Additionally, regulations for the creation of private protected areas in Brazil are compatible with primate-watching tourism and biodiversity protection (Crouzeilles et al., 2012). This scenario would facilitate the establishment of primate-watching activities in protected areas, Indigenous lands and private areas within the Arc of Deforestation, a region where logistic capacity (one of the main constraints on ecotourism; Pegas & Castley, 2014) exists.

For example, a shift from agriculture to primate watching focused on P. vieirai in the private lands of the Tapajós–Xingu interfluve would contribute to reducing habitat loss in the southern portion of the range of P. vieirai and to mitigating climate change, in turn contributing to protecting P. vieirai populations in the northern part of its range. It is equally important to elect politicians committed to biodiversity conservation, the welfare of traditional peoples and climate change mitigation (Silva, 2005; Carvalho et al., 2019), so that all key stakeholders can work together to improve outcomes for natural areas in the largest global deforestation frontier (Moutinho et al., 2018).

Conclusion

Deforestation, mainly driven by expansion of monoculture agriculture and cattle ranching, is a ubiquitous threat to P. vieirai, responsible for the past, present and projected future habitat losses and consequent decreases in the populations of this species. These types of land use in the Arc of Deforestation have contributed directly to climate change, another major threat to P. vieirai, which we expect to cause habitat degradation and population declines even within protected areas.

Primate watching could be a profitable and sustainable alternative to the exploitation and overexploitation of natural resources on private, public and Indigenous lands and could be a way to shift from the current land-use model in the Arc of Deforestation towards biodiversity conservation. In the long term, the forests protected for and by tourism will serve as carbon stocks and sinks, contributing to the mitigation of climate change.

The traditional model of predatory extraction of natural resources in the Arc of Deforestation is pushing primate species to the brink of extinction and transforming Amazonia into a source, rather than a sink, of carbon emissions. An alternate model of land use and income generation is needed if we are to protect the unique natural and human heritages of Amazonia and its life-supporting ecosystem services and products.