1. Introduction
Tropical dry forests are of great ecological importance but are also highly threatened and have received little attention focusing on the impact of their loss in the context of climate change [
1]. Forests offer solutions for surface cooling due to the ability of trees to capture and redistribute the energy of the Sun [
2]. Forests can exert a strong influence on local land surface temperature based on biophysical mechanisms as they generally have a lower surface albedo and higher evapotranspiration than unvegetated areas [
3]. However, these environmental functions can be degraded and even lost if forests are not managed sustainably and responsibly.
Changes in land use have great potential to influence temperature extremes [
4]. Indeed, the processes of deforestation directly lead to changes in the physical characteristics of the land surface, which can substantially alter the fluxes of heat, carbon, and moisture in atmospheric circulation and climate [
5,
6,
7]. Globally, different biomes are experiencing changes in the Land Surface Temperature (LST) distributions driven by extreme weather events and land surface changes [
8]. As the loss of forest cover increases, greater absorption of shortwave radiation from the Sun is generated, releasing longwave radiation after a few hours and producing increased wind chill [
8]. In tropical areas, deforestation significantly suppresses evapotranspiration and moderately increases albedo, exerting a warming influence on the global climate [
9,
10]. Deforestation also causes warming effects during the day and cooling at night [
11]. In addition, changes in land use due to the growth of urban, periurban, and rural areas are generating notable increases in LST levels as urbanization processes reduce natural vegetation, thereby increasing impervious areas and albedo alteration and causing urban residents to be exposed to a greater risk of heat stress than rural residents. This effect is known as the Urban Heat Island [
12].
Globally, estimates indicate that between 1982 and 2016, the largest net loss of tree canopy in the tropical region occurred in the tropical dry forest biome (−8%), while the largest net loss of tree canopy occurred in the tropical dry forest biome (−2%) [
13]. The main causes of disturbance in these forests are associated with commodity-driven deforestation, silviculture, shifting agriculture, and wildfire [
14]. In South America, the area of natural tree cover decreased by 16%, while the use of grasslands, croplands, and plantations increased considerably [
15]. Recent global studies highlighted the tropical forests of Brazil and the Chaco of Argentina, Paraguay, and Bolivia as the most threatened forest ecosystems in the world due to recent agricultural expansion. In Bolivia, the area identified with the highest growth in deforestation is the Chiquitano Forest [
16,
17] due to the rapid expansion of the mechanized agricultural and livestock frontier [
18,
19]. Currently, the drivers of the rapid and recent increase in deforestation in the Chiquitano Forest are the combined effect of soy expansion, the consolidation of corporate land holdings, and the increase of degraded soils in productive areas [
19]. In the Department of Santa Cruz, where the main extension of the Chiquitano Forest in Bolivia is located, the annual rate of deforestation reached −0.16% until 2019 [
20], mainly caused by an expansion of cattle ranching, medium- and large-scale mechanized agriculture, and small-scale agriculture [
18,
19]. Despite this factor, 70.7% of the forests are of great importance in terms of forest area in the region, with three non-fragmented stands of more than two million hectares that represent the largest proportion with 57.9% [
21]. If the current trend of forest loss in Santa Cruz continues, it is expected that by 2050 the forest area will be reduced from 25.1 million to 12.8 million hectares, mainly in the Chiquitania region [
20].
Based on remotely sensed temperature data, recent research conducted in a lowland metropolitan area of Bolivia [
22] found that daytime temperatures of forested areas showed a difference of −1.1 °C in relation to agricultural/livestock use and −2.4 °C in comparison with urban areas, while the average nighttime temperature of fragmented forests showed a dissimilarity of 0.1 °C with agricultural/livestock use and −0.7 °C with urban areas. However, studies on the effects of deforestation on land surface temperature in Chiquitania are nonexistent. The conversion of forests to cropland in the Chiquitania region corresponds to morphological changes in the vegetation, leading to a decrease in canopy conductance and, consequently, to an increase of approximately 2 °C in local temperature, as well as slight nocturnal cooling [
23].
This study was conducted to identify the effects of recent deforestation on land surface temperature in the Chiquitania region. Specifically, the three aims of this study were (a) to determine annual trends in forest cover loss; (b) to determine annual trends in the daytime and nighttime LST (°C) in forested and deforested areas; (c) to establish the relationship between intact forested areas and areas with deforestation processes and daytime and nighttime LST (°C) in Chiquitania. We hypothesized that the increase in temperature at ground level is related to a reduction in tree cover, which would demonstrate that conservation is key to risk and hazard management. This research will help decision makers formulate risk prevention strategies for natural disasters, adapt to climate change, and establish public policies that can help improve land-use planning, thus avoiding the progressive advance of land-use change.
4. Discussion
Deforestation in Bolivia is growing at an alarming rate. Recent estimates indicate that the country lost a total of seven million hectares by 2021, 86% of which will be concentrated in the department of Santa Cruz [
19]. In addition, the advance of deforestation has been associated in recent years with areas destined for forest use and exploitation [
20], as well as legally protected areas [
20]. The expansion of soybean and cattle ranching has been the main direct cause of deforestation [
19]. Our analyses sought to identify changes in forest cover in sites located in protected areas and indigenous territories using the MODIS product MCD12Q1.006, as well as changes in areas known to be the fastest growing areas of deforestation in Chiquitania. While the area of Nueva Esperanza presented the highest concentration of cropland, we found an evident reduction of natural cover in urban areas (Concepción and San Ignacio), as well as in the areas of El Cerro/California and Santa Ana/Buena Vista. If this deforestation trend continues, it is estimated that forest could disappear completely by 2050 [
20].
Assessments of vegetation cover status, changes, and processes are important components of global change research programs and are topics of considerable societal importance [
46]. Spectral vegetation indices (e.g., NDVI) are among the most widely used satellite data products and provide metrics for climate, hydrological, and biogeochemical studies, land cover change detection, phenology, and natural resource management [
47]. In addition, these indices are important for the analysis of vegetation growth and decrease trends [
48]. Our results were mainly based on a comparison of vegetation trends in forested and deforested areas. Using NDVI, we detected continuous vegetation growth in intact forests (Noel Kempff, Monteverde, and San Rafael), which showed no evidence of forest degradation processes in these areas. The explanation for these results is that high NDVI values are mainly related to the density of green leaves in a given area, making them a good indicator of vegetation cover and vitality [
49]. Our assessments considered a long time series (2001–2020) to ensure that the results would not be influenced by seasonal and interannual variability in the forests of the Chiquitania region [
50]. In contrast, we found sites with negative and statistically significant trends in NDVI values in three of the five sampling sites (El Cerro/California Mennonite colonies, Santa Ana/Buena Vista ranches, and San Ignacio), where a decrease in forest cover was recorded at different sites in Chiquitania. This result was expected, especially considering the current scenario of accelerated land-use change processes in the region [
19]. In addition, there was evidence of a negative trend marked by decreases of the NDVI in the Tucabaca forests. These results are very close to those of San Rafael, a forested site with a positive trend, which suggests disturbances due to illegal logging; however, ecological disturbance processes (e.g., changes in stomatal conductance) that cause physiological weakening and tree mortality could not be ruled out. However, the results for Tucabaca should be interpreted with caution and field evaluations should be carried out to identify the factors causing this degradation.
Land Surface Temperature (LST) monitoring using remote sensors offers several advantages over a wide observation range, including easy access and strong spatial continuity. Multiple studies have been conducted globally in recent years to determine LST [
3,
8,
51,
52], which is a key parameter for obtaining a more direct measure of surface conditions when analyzing surface–climate interactions [
33,
52]. Due to these qualities, we were able to determine, over a long period of time (2001–2020), the diurnal and nocturnal temperatures at the local level in Chiquitania. Of the 10 study sites, the highest diurnal LST levels were found in Nueva Esperanza, a completely deforested area. In addition, in the Cerro/California Mennonite colony area, which lost its remaining forest cover, a notable increase in diurnal LST was identified from 2007 onwards, which coincides with the decline in the trend of photosynthetic activity of the vegetation presented in
Figure 3. In addition, we were able to compare the differences in LST between forested and non-forested areas. Our results showed that the average values of daytime LST between forested sampling sites varied by up to 2.7 °C and those between deforested sites varied by up to 4.7 °C. However, when comparing forested and deforested sites, the difference in the averages was 3.1 °C. Differences in the diurnal temperature range are known to be smaller for forested areas and larger for non-forested areas [
53], because LST increases in deforested areas may be exacerbated by changes in roughness length, which impede energy dissipation via sensible or latent heat fluxes [
54]. In contrast, nocturnal heating is caused by the release of stored thermal energy during the day [
3]. However, comparisons of nighttime LST averages for the ten sampling sites did not show notable differences and, for this reason, nighttime LST could not be considered a good indicator of the differences between forested and deforested areas in Chiquitania. Nevertheless, our data help highlight the large differences between daytime and nighttime LST averages, which for the forest was 7.1 °C and for the deforested areas was 10.2 °C. This information is valuable, because it highlights the importance of forests in regulating temperatures in the Chiquitania region.
It is known that anomalies of maximum LST capture spatial patterns that are associated with droughts and heat waves on the land surface [
8]. In our study, both forested areas and those with deforestation processes in Chiquitania showed interannual anomalies of temperature increases in the daytime LST averages for the years 2002, 2010–2012, and 2019–2020, with the latter increases being higher than the others. In addition, in terms of nighttime LST, similar patterns were recorded for 2002 and 2015. In the Amazon area, generalized anomalies in LST values and large-scale directional changes towards higher temperatures were found in 2005 and 2010 [
8], where large and severe droughts were recorded [
55]. However, by 2020 different severe to extreme meteorological megadroughts events were recorded at the continental level. One of these events occurred in southern Brazil and Paraguay [
56,
57] and was also observed in Bolivia. In addition, on a local scale, drought events in the Chiquitania region were identified based on different climate stations, showing significant trends of increased frequency and intensity [
21]. Further studies are needed to understand the relationship between heat waves and droughts, which could improve our understanding of the impacts of land-use change on the local climate.
Recent studies indicate that in some lowland areas of Bolivia, temperatures tend to increase mainly in urban and agricultural/livestock areas [
19]. In our research on Chiquitania, it was shown that there is a tendency for an increase in daytime LST values, which is associated with deforestation processes, independent of the differences between the climate regions (humid and dry sub-humid areas) of the sampling sites. The trend of an increase in diurnal LST was statistically significant in four of the five areas with deforestation, mainly in San Ignacio, which registered the highest value in the Mann–Kendall test. However, in the sampling sites with forests, there were no statistically significant trends in daytime LST values, because tropical forests can exert a strong influence on local LST. As such, forests generally have a lower surface albedo due to absorbing more shortwave radiation during the day and higher evapotranspiration compared with areas with bare vegetated surfaces [
3,
58]. Nevertheless, nighttime LST values corresponded to another scenario, with a significant increase observed for the two forested areas (Noel Kempff and Bajo Paraguá) and two urban areas (San Ignacio and Concepción). Although an increasing trend in areas with deforestation was expected, new questions emerged regarding nighttime LST trends in forests. A possible explanation for this phenomenon is the presence of a large number of granitic rocky outcrops (inselbergs) in the Noel Kempff and Bajo Paraguá areas [
54], which concentrate the heat during the day and radiate heat to the surrounding forest during the night [
59,
60]. However, further research is needed in this area.
Globally, deforestation in tropical regions is causing strong warming between 0.38 ± 0.02 [
52] and 2.4 ± 0.10 °C [
3] (LST values). Our results showed a statistically significant correlation between NDVI and LST at three sites for diurnal values and two sites for nocturnal values. This result is in addition to the trend of 0.1 °C per decade observed for the air temperature increase rate [
61]. This result is concerning because the Chiquitania region is home to hundreds of thousands of people who carry out productive livestock and agricultural activities. The increase in LST values in the Chiquitania region could lead to a series of effects that have not been evaluated, including migration, economic production, and human capital to impacts on biodiversity [
62]. Given that forest loss is at the heart of both global and local warming, initiatives to reduce deforestation must remain a priority [
63]. Climate change adaptation strategies based on maintaining forest integrity are required to mitigate the increase in LST in the Chiquitania region [
64], where the conservation of existing forests and restoration of those lost to deforestation should be prioritized [
65].
Protected areas continue to be the main defense against forest cover loss and the best strategy for maintaining the ecological integrity of forests [
66]. Bolivia has approximately 130 protected areas, of which approximately one-third are located in the Department of Santa Cruz [
67]. Most of these protected sites contain ecosystems of high conservational value and, in many cases, intact forests [
68]. However, there is direct and indirect pressure around these areas owing to the rapid expansion of the agricultural and livestock frontier [
19,
69]. In the Chiquitania region, this expansion has caused a loss of forest connectivity [
19], a reduction of critical ecosystems [
70,
71], and a loss of habitats of key and priority species for conservation [
72]. If this scenario continues in these protected areas, up to half of forest area harbors could be lost in the next 30 years [
68].
Reducing current deforestation trends in Chiquitania is a high priority. It is necessary to redefine norms and public policies regarding forest conservation, aimed at territorial management at multiple scales. Undoubtedly, revising the design criteria for land-use plans, enforcing the economic and social functions of private and communal property, and promoting the protection and restoration of conservation easements will improve opportunities for thermal regulation and the maintenance of ecological functionality at the landscape scale in the Chiquitania region. It is also necessary to establish a monitoring system to evaluate changes in natural vegetation cover, especially temperature and precipitation, at regional and local scales. With these changes, there will be greater opportunities to increase socio-ecological resilience to the impacts of climate change.