Multiscale Effects of Slash-and-Burn Agriculture Across the Tropics: Implications for the Sustainability of an Ancestral Agroecosystem

Abstract

Slash-and-burn agriculture (SBA) is critical to maintaining rural peoples’ livelihoods. Yet, it causes environmental degradations that challenge its sustainability. Such degradations are often underestimated, as they are usually assessed at the local (stand) scale, overlooking larger-scale impacts. Here, we drew upon existing SBA and landscape ecology knowledge to assess the multiscale abiotic and biotic effects of SBA. This agroecosystem involves four stages (slashing of vegetation, burning of vegetation, farming, and forest recovery) but the SBA research is biased towards biotic impacts, especially during forest recovery. Despite its importance for key abiotic (e.g., soil fertility) and biotic (e.g., species richness) attribute recovery, this stage is typically too short (<10 years) to compensate for the environmental degradation caused by the previous stages. Successional and landscape ecology theory suggests that such compensatory dynamics can promote SBA sustainability in landscapes dominated by old-growth forests. Yet, when old-growth forest loss exceeds certain boundaries, abiotic and biotic SBA impacts may compromise the conservation value and sustainability of this ancient agroecosystem. We highlight that SBA sustainability should be comprehensively assessed by including landscape-scale variables (e.g., percent old-growth forest cover) that may be key for maintaining biodiversity patterns and processes in landscapes where SBA is practiced.

1. Introduction

Slash-and-burn agriculture (SBA) has shaped tropical ecosystems for millennia, but its practice entails environmental impacts that challenge its sustainability [1]. When properly designed and managed, SBA can be sustainable, especially if practiced in highly forested regions with a low population density [2,3,4,5]. However, this ancestral agroecosystem is increasingly expanding in densely populated tropical regions, where millions of poor people depend on it for their livelihood—a socioeconomic context that can push local people to unsustainably use resources [6,7,8,9]. Thus, understanding the abiotic and biotic impacts of SBA is urgently needed, especially if we wish to promote win–win scenarios for biodiversity and humans.

Despite the key role played by SBA for people’s livelihood, our understanding of the environmental impacts of SBA remains limited. For one, although this agroecosystem involves four major stages (i.e., slashing of vegetation, burning of vegetation, farming, and forest recovery) (Figure 1), most studies have only assessed the effects of a single stage (e.g., burning of vegetation or forest recovery) [10,11]. More importantly, SBA not only causes environmental changes at the local (stand) scale, as it is usually studied [12], but its effects can be noticed beyond the limits of the cultivated field, shaping abiotic and biotic factors at the landscape scale [13]. Here, we filled these knowledge gaps to have a more comprehensive understanding of the environmental impacts of SBA across the world’s tropics.

Our main aim was to use existing SBA and landscape ecology knowledge to assess the multiscale abiotic and biotic effects of the slash-and-burn cycle. In particular, we first carried out a systematic review of studies assessing the abiotic and biotic effects of SBA. From this starting point, we identified how the studies were distributed across the tropics, the types of forests and stages that were studied within the slash-and-burn cycle, as well as the scale of analysis and response variables (abiotic or biotic attributes) that were evaluated in each study. As described below, we recovered 154 studies, but almost all (153 studies, 99%) evaluated local (or stand)-scale impacts, probably because SBA typically operates at this scale. Therefore, we first assessed the impacts at the local scale for each stage within the slash-and-burn cycle (Figure 1). Yet, as SBA promotes the creation of complex landscape mosaics composed of different land uses (old-growth forests, secondary forests with various degrees of development, burned lands, and agricultural fields), this agroecosystem can also shape environmental patterns and processes at the landscape scale [13]. Thus, we also compiled emerging evidence from landscape ecology research [14,15,16,17,18,19,20,21] to identify potential impacts of SBA at the landscape scale, particularly considering the mosaic of land uses created by SBA in each territory. At both scales, we highlight emerging patterns regarding the abiotic and biotic effects of SBA, and highlight the most important applied implications, especially those aimed at promoting the sustainability of tropical ecosystems and SBA itself.

2. Literature Search

To compile the available studies on SBA effects across the tropics, on 10 August 2023, we searched for studies indexed in the Scopus and Web of Science databases. We used the following search string: (“agricultur*” OR “cultiv*” OR “farm” OR “crop”) AND TITLE-ABS-KEY (“tropic*”) AND TITLE-ABS-KEY (“slash-and-burn” OR “shifting cultivation*” OR “swidden” OR “subsist*” OR “human fire*” OR “wildfire*” OR “anthropogenic fire*” OR “anthropic fire*” OR “human-induced fire*” OR “human-caused fire*” OR “man-made fire”). This search produced 2404 studies, spanning from 1951 to 2023. We then filtered the studies using the following criteria. First, the studies should have been carried out in lowland (<1100 m.a.s.l.) tropical xeric shrublands and woodlands, tropical pluviseasonal forests, and/or tropical rainforests, i.e., between the latitudes of 23.5° N (the Tropic of Cancer) and 23.5° S (the Tropic of Capricorn). We focused on lowland forests to avoid confounding factors (e.g., steep slopes and temperature reductions) associated with altitude. Second, SBA should encompass annual or semi-permanent subsistence crops (i.e., mixed crops of bean, maize, pumpkin, cassava, upland rice, or banana), so we excluded studies on monocultures for commercial purposes (e.g., oil palm, coconut, cashew, and home gardens). Third, the studies should have assessed at least one abiotic (e.g., physicochemical characteristics of the soil and climate) or biotic (e.g., ecological patterns and processes) response to SBA. Fourth, we excluded all studies that evaluated the effect of SBA related to other productive activities such as livestock grazing. In fact, we excluded all studies not explicitly stating that SBA was the type of disturbance analyzed. After this filtering, we performed a careful reading of 187 independent studies of SBA effects in the tropics, which included 154 case studies and 33 reviews (Figure 2) published between 1986 and 2023 (Figure 3).

3. Overview of Reviewed Studies

Our literature review found that studies on SBA effects have mostly focused on Neotropical forests (n = 81, 52%), while less emphasis has been placed on Asian (n = 44, 29%) and continental African forests (n = 18, 12%) (Figure 3 and Figure 4A). Madagascar (n = 6, 4%) and Oceania (n = 5, 3%) were the least-studied regions. In fact, many tropical countries from Southeast Asia (e.g., Myanmar, Philippines, and Vietnam) and Oceania (Papua New Guinea) lack research on the topic. Therefore, the available evidence to date is notably biased towards the Neotropical realm, which can limit inferences for other continents.

We also found that the research was biased towards tropical rainforests (61% of studies; Figure 4B), was largely focused on the forest recovery stage (82% of studies; Figure 4C), and mainly assessed the effects on biotic variables (42% of studies; Figure 4D). These biases are likely related to the growing interest in understanding how biological populations, species assemblages, and ecosystem properties change after disturbance cessation. In fact, this has been one of the most important topics in ecology and conservation biology since the dawn of the 20th century, with field studies increasing exponentially since the 1970s [24]. Such importance is justified, at least partially, by the economic growth, rural exodus, and urbanization, as these processes prompted the transition from net forest loss to net forest gain (forest transition phenomenon) in some tropical regions [25], fueling interest in the prediction of the successional trajectories and conservation value of regenerating forests, and in restoring biodiversity and ecosystem functioning and services in human-modified landscapes [11,13,19,26,27].

4. Local-Scale Impacts of SBA

The SBA cycle involves four different stages. Therefore, to gain a deeper understanding of the local impacts of SBA, it is necessary to assess the abiotic and biotic effects of each one of these stages separately.

4.1. Slashing of Vegetation

SBA initiates with the slashing of vegetation in a relatively small (<1 to 5 ha) stand of old-growth forest (Figure 1). Thus, the first and most noticeable impact of SBA is the removal of tree cover. This implies the creation of an open area with profound alteration of the abiotic conditions. In particular, these open areas will show a large increase in temperature and a decrease in humidity, both on the soil surface and in the air [28,29] (Figure 5a). These microclimatic changes can reduce the primary production of both the forest and agricultural systems [30] and, as discussed below, they can also have a myriad of cascading (indirect) effects on biodiversity patterns and processes, especially on organisms and ecological processes that evolved in shaded environments and that do not tolerate forest clearing [29,31,32]. By virtually removing all the aboveground vegetation, forest clearing also produces a major reduction in carbon stocks. Some carbon is retained in the remaining deadwood, but this pool is rapidly depleted due to the rapid decay of fallen branches. Indeed, at this stage, the transference of carbon from the forest to the atmosphere is achieved through the fragmentation and accelerated decomposition of dead plant material [33,34]. Therefore, the slashing of vegetation causes important microclimatic changes in the cleared stands and promotes a large emission of carbon stored in the logged forest to the atmosphere (Figure 5a).

The biotic impacts of this SBA stage have been poorly studied. In fact, we did not find a single study on the effects of slashing on animals. However, the faunistic effects of forest loss have been widely studied in the context of other anthropogenic activities, such as deforestation [14] and clear-cutting [35], and thus, they can be inferred to some extent from these investigations. For example, as most animals can move to favorable environments when logging occurs, they are probably less affected by slashing than plants (Figure 5b). The arrival of generalist species and open-area specialists may result in open areas exposed to SBA having a higher number of bird [36] and bee [37] species than old-growth forests. However, even if these events do not immediately occur, animal populations can collapse if the impacts of slashing the vegetation accumulate and become more pronounced over time [38].

In strong contrast, plants are almost completely removed from the site at this stage (Figure 5b). This is not a minor issue, as the tropics can host hundreds of tree species in just one hectare [39], and most woody species are shade-tolerant and do not tolerate the conditions that prevail in open areas [40]. Also, considering the restricted global distribution and rarity of most tropical tree species, most of them (>40,000 species) may qualify as globally threatened following the IUCN criteria [41,42]. Therefore, even when causing the loss of a few hectares of forest, SBA could have a disproportionately high impact on tropical trees, especially on rare and threatened species, which is an important topic for future research.

Importantly, tropical forests are home to other plant growth forms and deadwood-dependent bryophytes and fungi that can also be lost at this stage, at least in the slashed area. For example, the abundance of saprotrophic fungi (which support plant growth by increasing nutrient availability in the soil) can decrease in cleared areas [33]. As lianas, vines, epiphytes, and epiphyllous species depend on tree cover for their establishment and growth [43,44,45], they are also lost at this stage. These groups are extremely diverse but insufficiently known, and their extinction is relatively silent, as they are often overlooked in most diversity assessments. For example, lianas can account for approx. 25% of the woody stem density (abundance) and species diversity (species richness) in many tropical forests [46,47], and among the 22 largest epiphyte families, there are >50,000 species that only grow in the Neotropics, with Orchidaceae being by far the richest one [48]. Taken together, these plant growth forms play critical roles in ecosystem functioning, providing water, food, shelter, connectivity, and unique microhabitats that are of key relevance for maintaining forest diversity [44,49]. They also deliver important services to humans, including carbon sequestration, and providing food, fibers, and aesthetic and spiritual values [45,50]. Therefore, at this stage, SBA can cause important biotic impacts, with potential consequences for human wellbeing.

It is important to note, however, that all these abiotic and biotic changes not only affect the slashed area. They can also affect the surrounding forest through the so-called ‘edge effects’ [51]). Although none of the reviewed studies evaluated biotic and/or abiotic changes at the forest edge surrounding the slashed area, this topic has been widely studied in forest fragments surrounded by land exposed to SBA, so the existing knowledge on edge effects can be applied in the context of SBA. We know, for example, that the strength and penetration of edge effects are higher in recently created edges, especially when bordered by high-contrast open-area matrices, such as those created by SBA [29,51,52,53,54]. These effects include microclimatic changes at forest edges (e.g., increased temperature and decreased humidity), especially in tropical forests as they receive higher solar radiation than temperate ones [29,52]. The consequences of these climatic changes for biotic assemblages and ecosystem functioning depend on the thermosensitivity of individuals and species, but in general, they can explain many of the changes in ecological patterns and processes that are recorded at forest edges [29]. Among the other changes [54,55], there is evidence that, compared to forest interior, forest edges can show (i) a reduced density of fungal fruiting bodies; (ii) increased tree mortality; (iii) invasion of disturbance-adapted species; (iv) a reduced canopy height; and (v) a lower species richness of several taxa. The penetration distance of each of these abiotic and biotic changes is highly variable, but in treeless matrices like those created by SBA, most of them may penetrate 100 m or more into the forest [55]. Therefore, this topic merits additional attention, as it is frequently overlooked in studies on SBA.

4.2. Burning of Vegetation

Once slashed, the downed vegetation is left to dry and it is subsequently burned. This implies that carbon emissions will peak at this stage (Figure 5a). For example, burning 1 ha of dense forest can result in exorbitant greenhouse gas emissions, contributing to global warming [56]. In particular, if only the carbon from aboveground living biomass is considered, burning 1 ha of old-growth tropical forest could cause the emission of 70 to 300 Mg of carbon [57,58,59]. However, this trend depends on the quantity (e.g., fuel load tons per ha) and quality (e.g., fine vs. slightly coarse litter) of the plant biomass, as well as on the climatic conditions, including wind speed and air humidity, during the burning of vegetation [60,61,62,63]. For example, it may be more difficult for high-caliber biomass to be completely consumed by fire when wind speeds are low [64], leading to the emission of more incompletely oxidized products such as methane [65]. The emission of greenhouse gases derived from SBA in the Amazon between 2000 and 2011 was responsible for approximately 1% of the global emissions of greenhouse gases [66].

The effect of fire on soil fertility has also been amply studied. The evidence indicates that burning promotes a rapid incorporation of nutrients into the soil [67,68,69] (Figure 5a) without affecting the soil physical properties involved in the hydrological and erosion responses such as water repellency and aggregate stability [70]. However, the incorporated nutrients are rapidly lost through leaching and surface runoff [71] and are absorbed by cultivated plants [72,73]. Therefore, the benefit of fire in incorporating nutrients into the soil is probably only short-term (i.e., a few months after fire) [61,74,75,76] and may involve the early release of heavy metals such as mercury (Hg) into the soil during the first year after the fire [75].

Another important abiotic effect of fires is the short-lasting but intense increase in the soil surface temperature. This increase can kill weeds and pests (the desired outcomes at this stage). However, the increase in soil temperature can also result in some cascading negative effects associated with biological properties of the soil. In fact, we found that fire effects on bacterial and fungal communities have been the focus of post-fire microbial biomass studies [72]. For example, high-intensity fires can alter soil surface and subsurface properties (e.g., pH), consequently affecting the persistence of soil microbial communities. In other words, fire induced high rates of microbial mortality due to the heating of the topsoil and the abrupt increase in the pH immediately after burning [72]. However, fire does not appear to be harmful to other living organisms that may be beneficial to plants and crops, such as arbuscular mycorrhizal fungi in the soil [77], which play critical roles in post-fire nutrient cycling (see below).

Fire also damages seeds in the soil seed bank, thus limiting forest recovery. A few experimental studies have reported negative relationships between fire and the soil seed bank [10,78,79,80], with the proportion of viable seeds and the number of rare species being reduced after a fire. This highlights the low resistance of the soil seed bank to this agricultural method. Similarly, many invertebrate and vertebrate species (e.g., small rodents, reptiles, and amphibians) that live underground should also be negatively impacted by fire (Figure 5b). Yet, to our knowledge, this topic has not yet been investigated.

4.3. Farming

Once burned, open stands are sown and cultivated for up to 3–4 years [5,81,82] (Figure 1). The farming stage starts with land preparation, which may include the removal of unburned vegetation residues, soil levelling, and furrow creation for the sowing of commonly cultivated crops in the region [81,83]. In some regions, this stage can also include other practices such as tillage and weeding, and even some external inputs, such as fertilizers and pesticides. Therefore, the environmental impacts will depend on the practices used at this stage. However, as the farming stage lasts longer than the previous ones, its cumulative effects at the local scale can be very important, especially if it involves agricultural practices, such as tillage and the use of agrochemicals, which can lead to a significant decline in biodiversity and soil health.

Due to the very high solar exposure of croplands, the first and most noticeable abiotic impact of farming is an increase in temperature and loss of humidity, both in the soil and in the air [81] (Figure 5a). Tillage contributes to this environmental desiccation process by pulling out roots and removing woody debris from the soil surface, as these practices increase soil aeration and erosion [84,85,86,87]. Tillage also promotes nutrient loss through leaching. However, some of these abiotic changes can be reduced as crops grow. For example, crop roots stabilize the soil, preventing erosion, while leaf cover minimizes nutrient leaching [73]. Furthermore, crops provide some cover to the soil, decreasing the temperature and evapotranspiration, and maintaining the soil moisture to some extent [87] (Figure 5a). Nevertheless, the ability of crops to protect soil is often limited, as the proportion of area occupied by cultivated plants is small compared to the bare soil area, and soils are therefore frequently exposed to high solar radiation [88,89,90]. Therefore, cultivated areas usually experience deterioration in both physical (e.g., water retention capacity) and chemical (e.g., nitrogen content) soil properties over the farming stage [72].

Soil erosion and nutrient leaching are big challenges in areas exposed to SBA. Although some agricultural lands contain a high concentration of nutrients [91], as discussed above, they are rapidly lost in a few years [72,88,89,90] (Figure 5a). Importantly, after a fire, the availability of nutrients in the soil not only benefits cultivated plants, but also native plants. Therefore, management strategies, such as manual weed removal (weeding) are required to enhance crop growth and avoid weed (i.e., all nonplanted species) proliferation in agricultural fields [81,92]. This is a common practice that involves pruning or removing tree stumps, roots, and seedlings of opportunistic species to ensure crop production and prevent weed infestation [92]. When a crop is weeded, the cut plant material covers the soil, helping to prevent excessive drying and temperature fluctuations, as well as promoting organic matter decomposition [73]. This is particularly important in nutrient-depleted soils, where the decomposition of weeded material can contribute to replenishing essential nutrients, supporting subsequent crop growth and improving overall soil health. However, weeding is labor intensive and cannot fully replenish the soil nutrients needed to ensure crop growth. Therefore, farmers have increasingly turned to the application of external inputs, such as fertilizers, herbicides, and pesticides [93].

The use of agrochemicals in SBA can cause significant damage to the environment and human health. Although SAB does not typically use large quantities of these agrochemicals [94], or at least not on a large scale, some governments provide these types of products to farmers to increase their productivity. For example, the ‘Fertilizantes para el Bienestar’ (Fertilizers for Wellbeing) program in Mexico, which began in 2019 in Guerrero state and in 2023 reached national coverage, have distributed thousands of tons of highly soluble fertilizers (i.e., diammonium phosphate and urea) annually so that local communities (especially indigenous ones) could fertilize their fields [95]. The impact of this governmental program would not be so negative to environmental and human health if they were accompanied by personalized supervision on the correct use of these agrochemicals, as well as monitoring of their impacts. Yet, to our knowledge, these two crucial aspects were not considered in the program, so it is likely that local communities misused or overused these fertilizers. For example, in 2023, the government distributed 141,000 tons of these fertilizers in the Yucatan peninsula region, where the high solubility in water of these fertilizers combined with the presence of a karstic limestone terrain with shallow soils and high infiltration rates likely reduced the effectiveness of these fertilizers while polluting underground water bodies (eutrophication), negatively affecting ancient aquatic ecosystems such as cenotes (sinkholes), a critical issue that has not been yet studied. In addition to fertilizers, the use of herbicides and pesticides such as glyphosate and paraquat has also increased in the tropics, representing an increasing threat to natural ecosystems and human health [96]. These chemicals can persist in the soil and be transported to water bodies, where they can cause harm to aquatic life and pollute drinking water [97]. Worryingly, increasing reliance on external inputs such as fertilizers, herbicides, and pesticides can create a negative feedback loop, where the natural fertility and biological control mechanisms are reduced, requiring even more intensive management practices [98]. This not only limits the immediate agricultural productivity but also the sustainability of SBA and the broader environment at large.

Regarding the biotic impacts of the farming stage, tillage can negatively affect soil fauna by destroying worm channels, termite and ant galleries, and root canals (Figure 5b). This disturbance may cause the population decline of invertebrates and microorganisms essential for soil health [99]. However, as stated above, growing crops can provide some protective cover to the soil, improving conditions for thermosensitive organisms, such as bacteria and fungi, which play a crucial role in decomposing organic matter and recycling nutrients [72,100]. Weeding can also negatively affect bee and ant populations, disrupting local plant–pollinator–disperser networks [101]. However, weeding also has some positive biotic effects by increasing food availability for earthworms, which break down plant material, helping to replenish essential nutrients and improving the soil structure and soil aeration [73]. The use of agrochemicals can also reduce populations of beneficial microorganisms and invertebrates, such as pollinators and natural predators, causing pest outbreaks and limited pollination of crops [97,99,102]. In fact, the conversion of forests to agricultural lands can reduce invertebrate diversity [103]. The diversity of birds [99] and mammals [104] can also be lower in agricultural lands than in forests. Therefore, if a sufficient amount of forest is not maintained in the landscape [14], SBA can have significant negative impacts on biodiversity.

Finally, we want to call attention to hunting, which is a widespread but often overlooked land use, especially in the context of SBA. Hunting often overlaps spatially with other land uses, such as agriculture, as some game species (e.g., white-tailed deer, peccaries, and turkeys) are attracted to agricultural land [105]. Thus, although we did not find any studies quantifying its impact, we have witnessed, both in Mexico and Brazil, that local farmers often come to their croplands with a gun to hunt some game species or kill mammals that are considered undesirable (e.g., coatis and raccoons) because they could damage the crops. Thus, an indirect impact of the farming stage is to encourage hunting (Figure 5b), whose effects can be particularly deleterious in those regions that already show signs of defaunation as a result of other threats, such as forest loss. Certainly, this topic is worth investigating and should not be overlooked in future studies.

4.4. Land Abandonment and Forest Recovery

When crop productivity decreases, cropland is abandoned to allow for forest recovery (Figure 1). The ecological drivers and consequences of this process, as well as the successional trajectories of second-growth forests, have been extensively studied (Figure 4), leading to significant theoretical and empirical advances with considerable practical implications [11,13,24,26,106,107].

In tropical forests, there are three fundamental sources of natural forest recovery: seed rain, soil seed bank, and resprouting from surviving plants [26]. Therefore, alterations to these sources of regeneration may result in delayed or even arrested forest recovery [13,24]. For instance, considering that up to 90% of tropical tree species depend on animals to disperse their seeds [108], regeneration through seed rain may be limited in defaunated regions, such as those exposed to SBA where forest loss and/or overhunting can reduce the seed-dispersing fauna [13,109,110,111,112,113]. Similarly, the low resistance of the soil seed bank to fire can limit its contribution to forest regeneration in areas exposed to SBA [10,78,79,80,114]. Consequently, in some regions, forest recovery can mostly depend on plant resprouting [115,116,117,118], which limits the genetic diversity of growing populations [119,120]. In summary, the regeneration process after SBA will be highly variable among ecosystems and regions depending on the effect of SBA on the sources of natural regeneration [13,121].

If the forest regeneration process is not interrupted, forest recovery can have important positive abiotic and biotic effects, which are generally opposite to the three previous stages of environmental degradation (Figure 5). Firstly, plant resprouting and the establishment of pioneer plant species increases tree cover in the area [116,117,122]. The tree basal area recovers rapidly with forest age, especially in more humid forests [27,123,124,125,126,127,128,129,130,131,132,133,134,135,136]. This makes it possible to reverse some of the negative abiotic changes described above. For example, the microclimatic conditions for plant regeneration improve as succession progresses, with soil and air temperatures in the forest interior decreasing significantly [137,138,139] (Figure 5a). In fact, in as little as 54 months, secondary forests can exhibit microenvironmental conditions similar to those of interior areas of old-growth forests [140].

Furthermore, some of the carbon emitted during the slashing and burning of vegetation stages is captured at this stage [126,141,142,143] (Figure 5a). Nevertheless, SBA is far from being carbon neutral, as the original old-growth forest stored more carbon than secondary forests can store [27,144]. Indeed, within 120 years after land abandonment, secondary forests can capture up to 90% of the aboveground biomass of their old-growth values [11,27,142,145,146], but secondary forests are rarely allowed to recover to such an advanced age [147]. In most regions, they are only abandoned for 10–15 years before being exposed to a new slash-and-burn cycle [148,149]. Therefore, although the average net carbon uptake in young secondary forests can be up to 20 times higher than in old-growth forests [27,144], evidence from the Amazon, Borneo, and Central Africa indicates that regrowing secondary forests are only able to counterbalance around one-quarter (21–34%) of the carbon emissions from the old-growth forest loss [150].

As the forest recovers, certain essential soil conditions also show signs of improvement [151,152]. Soil recovery is initially driven by the formation of aggregates from regular inputs of organic material into the soil (i.e., leaves, branches, and other debris that fall to the ground) [125,153,154]. The formation of soil aggregates also drives the recovery of other physical properties, such as porosity and soil compactness, two important attributes linked to an increased infiltration and water-holding capacity in recently abandoned areas [151]. Soil decompaction results from the penetration of pioneer roots below the soil surface [155]. Concurrently, the stabilization of the soil surface occurs with the establishment of vegetation, which serves to protect against erosion and facilitate the retention of litter within the topsoil layer [125].

Litter production plays a pivotal role during early successional stages by replenishing soil nutrients and supporting plant growth in abandoned lands that have been depleted of nutrients due to prior agricultural use [11,88,156,157,158,159,160,161,162] (Figure 5a). While some young secondary forests may benefit from increased soil fertility resulting from the previous burning stage [163,164], pioneer plants usually cause a gradual depletion of nutrients [12,165]. However, pioneer species also tend to have high concentrations of leaf nutrients, which are essential for soil recovery [11,166]. Also, the establishment of fast-growth nitrogen-fixing species (e.g., those from the Fabaceae family) can promote the recovery of some essential nutrients, such as nitrogen [145,167,168,169].

As litter decomposes, it undergoes a series of chemical and biological transformations that result in the release of essential nutrients back into the soil, thereby contributing to the restoration of fertility in abandoned, nutrient-poor lands [170]. The process of litter decomposition is influenced by a multitude of abiotic and biotic factors, including air humidity and temperature, substrate quality, and decomposer community composition [67,170,171,172,173,174,175,176,177,178]. For instance, elevated temperatures accompanied with high humidity can accelerate decomposition rates [178]. Furthermore, litter quality, which is determined by its nutrient content and structural composition, influences the decomposition rates, with high-quality litter decomposing at faster rates than low-quality litter [179] These abiotic factors play a key role in litter decomposition and can be critical for restoring essential soil nutrients [170,180].

Soil fauna, which includes a variety of invertebrates (e.g., nematodes, ants, and beetles), also plays a fundamental role in litter decomposition [176,181]. For example, by consuming microorganisms (e.g., saprophytic fungi and bacteria), soil nematode communities release nitrogen, promoting plant growth in secondary forests [174,182]. Ant colonization can create and maintain soil nutrient heterogeneity at early successional stages, greatly stimulating microbial growth [183]. Dung beetles also recover rapidly alongside forest succession, enabling the decomposition of dung and organic matter in secondary forests that were previously exposed to SBA [184]. Earthworms also play an important role in restoring soil quality after SBA. These animals increase soil aeration and water infiltration, and from their feces, they also create soils rich in nutrients that are essential for plant growth. Some studies have shown that earthworm activity is evident during the first year of land abandonment [176,185]. During this initial stage, secondary forests may contain a mix of native and exotic earthworm species, while native species dominate in late-successional forests [175]. Some forests can also support depleted and low-density earthworm assemblages even after almost a decade of abandonment [186]. Therefore, the activity of these animals and their contribution to increasing soil fertility can be highly variable among ecosystems and regions.

Microorganisms (e.g., fungi and bacteria) also help convert organic matter into mineral nutrients—a fundamental process that improves soil structure and nutrient retention capacity [187]. Although some microbial communities appear to be adapted to SBA [91], soil fertility recovery depends on the diversity of these communities, which increases as the forest structure becomes more complex [176,188]. Arbuscular mycorrhizal fungi are particularly valuable after SBA, as they facilitate nutrient uptake by plants and enhance plant resistance to environmental stresses [189]. These fungi appear to recover rapidly in abandoned fields [189,190,191,192,193,194]. However, some species can be lost during agricultural conversion [195]. Therefore, the recovery of soil quality depends on dynamic feedback between vegetation, environmental conditions, and soil biota [107,196].

Regardless of the decomposition mechanism, the soil carbon concentration typically increases with soil organic matter content, and thus, with forest age [11,126,145,155,166,170,197,198,199]. However, it is difficult to predict how long it will take before full recovery, although the evidence suggests that it is a potentially slow process [133,200,201,202]. Phosphorus recovery is also slow because it depends on atmospheric deposition and mineral weathering, as well as the number of previous agricultural cycles [157,165,167,203]. However, nitrogen recovery can be faster due to organic deposition and biological fixation by legume–rhizobia symbiosis [141,168,169,203]. However, the full recovery of soil fertility is a gradual process that depends on the continuous input of leaves and roots and their subsequent biological transformation, which will probably take decades to complete [152,160,170,204,205,206].

Regarding the impact of the forest regeneration stage on vertebrate animals and trees, there is ample evidence that these groups gradually recover alongside secondary succession, at least in terms of taxonomic diversity [13,207,208] (Figure 5b). In fact, integrating information from 56 chronosequences distributed across the Neotropics, Rozendaal et al. (2019) demonstrated that secondary forests take a median time of 50 years to recover the tree species richness of the old-growth forest, although 80% of the species recovered after 20 years. However, the full recovery of species composition can take centuries [208]. Flying vertebrates such as bats and birds may recover faster than trees (Figure 5b), as different-aged secondary forests can be taxonomically similar to old-growth forests [209,210,211,212], probably because secondary forests are used as supplementary habitats by many forest-dwelling species [36,213]. However, some functional groups, such as nectarivorous bat and bird species and small frugivores, may use these secondary forests more frequently as insectivorous and large frugivorous species usually prefer old-growth forests [209,210,214,215]. Similar successional changes in the functional composition have been recorded for bee communities, a group largely responsible for pollination in tropical forests [37].

5. Impacts of SBA at the Landscape Scale

As a shifting form of agriculture, the impacts of SBA logically go beyond the limits of the cultivated field in a given year but rather shape the abiotic and biotic environment across much larger scales (Figure 6). Indeed, SBA generates complex landscape mosaics composed of old-growth forest, different-aged secondary forests, agricultural fields, human settlements, and areas burned by accidental fires caused by SBA [13,216,217,218,219]. Therefore, as demonstrated by decades of landscape ecology research, the landscape-scale impacts of SBA will largely depend on the area occupied by each of these land cover types in the landscape (i.e., landscape composition) and, to a lesser extent, on their spatial distribution and physiognomy (i.e., landscape configuration) [13,14,15,16,18,19,21,220].

Although ecological patterns and processes in agricultural landscapes may be driven by both landscape composition and configuration variables [16], the evidence suggests that composition variables are usually more important [14,221,222,223]. In landscapes exposed to SBA, the most important compositional variable is the percentage of remaining forest cover as it shapes the abiotic and biotic variables that are critical to maintaining the integrity of natural ecosystems and agricultural production systems [13,14,19,36,220]. In particular, forest cover is critical for regulating and preventing floods and soil erosion [224,225], as the tree canopy, litter, and roots reduce rainfall kinetic energy and surface runoff velocity [226]. Forest cover is also related to other ecosystem services, such as climate regulation [28,227]. For example, as mentioned above, forest cover reduces air and soil temperatures both directly through evapotranspirational cooling and by shading surfaces, and indirectly by storing carbon and helping to prevent emissions of pollutants that increase greenhouse gas concentrations. Forest cover might also play an important role in determining rainfall, so forest loss could promote an aridification process in some agricultural landscapes [28,30,228]. This is not trivial, as the interconnection between forest loss, forest desiccation, and elevated temperatures renders forests increasingly vulnerable to prolonged and more frequent fires [28,229]. Therefore, the landscape-scale abiotic impacts of SBA will largely depend on the amount of forest cover that is preserved at the landscape scale.

The biotic impacts of landscape forest loss have been well documented around the world. Forest loss reduces the availability of resources for forest-dependent species, negatively impacting many taxa [220,221,222,230,231,232] (e.g., bats, bees, terrestrial mammals, birds, dung beetles, and tropical trees). Indeed, the loss of forest cover in general [223,233,234], and of old-growth forests in particular [104,235], are considered the most important threats to global biodiversity. However, the effect of forest loss is not always linear, as species extinction can accelerate in landscapes that have reached a certain forest loss boundary (i.e., ‘extinction threshold’) [236], usually <10–30% remaining forest cover [237]. Since higher thresholds (30–50%) have also been documented in tropical ecosystems [14,232,238,239], we could expect SBA to be relatively more biodiversity-friendly in landscapes dominated by old-growth forest cover (i.e., >50% remaining forest cover) (Figure 6). As argued in previous studies [13,14,19,240], this intermediate level of deforestation should also be optimal to maintain most ecological patterns and processes, as well as the delivery of ecosystem services (Figure 6).

The composition of the anthropogenic matrix can also drive abiotic and biotic variables in agricultural landscapes. For example, the negative edge effects described above (e.g., increasing temperature, decreasing humidity, and species extirpation at forest edges) can be attenuated by allowing secondary forest growth in the matrix surrounding the old-growth forest fragments, as this secondary forest decrease the contrast with the forest [53,241,242]. Thus, secondary forests in landscapes exposed to SBA can increase the quality of the anthropogenic matrix, positively affecting wildlife [243,244]. For instance, secondary forests can support resources and refuge for many forest-dependent species [213,222,245] and can also facilitate successful movement between forest patches [213]. However, relatively young (<10 years) secondary forests can be less suitable than old-growth forests for shade-tolerant plant species and for many forest-specialist animals, making the persistence of these species in secondary forests (sinks) dependent on sources of individuals from old-growth forests [13]. This suggests that the conservation value and sustainability of SBA may be limited in landscapes where slash-and-burn cycles are shortened (i.e., shorter fallow periods), as this practice decreases the quality of the anthropogenic matrix by increasing the percentage of the landscape covered by young secondary forests.

Importantly, there is a close relationship between landscape-scale forest cover and the spatial configuration of the remaining forest [223]. For instance, the number of forest patches (i.e., degree of forest fragmentation) and forest edge density usually peak in landscapes with 20–30% forest cover, whereas mean inter-patch isolation distance increases exponentially in landscapes with <20–30% forest cover [223]. Although these configurational patterns may affect ecological patterns and processes [16], there is increasing evidence suggesting that the forest spatial configuration is not as important as the forest cover, and that most configurations should protect most forest species if enough forest is present in the landscape [14,223,246,247,248]. Nevertheless, biodiversity conservation seems to be more effectively achieved by maintaining a larger number of smaller forest patches than by maintaining a smaller number of larger forest patches [223,247,248]. Therefore, SBA will likely have a less detrimental impact on biodiversity if crops are distributed in a way that preserves many patches of old-growth forest across the landscape, as this spatial configuration increases the likelihood of covering a larger range of landscape environmental heterogeneity, and with this, a greater variety of species [14,246,249]. This is an interesting hypothesis to be tested in future studies.

In summary, SBA generates complex landscape mosaics, comprising a variety of land uses. Landscape ecology research suggests that the conservation value and sustainability of these landscape mosaics is contingent upon their spatial composition, exhibiting the highest values in landscapes that are predominantly covered by old-growth forest, and a relatively high value in landscapes with an intermediate percentage of old-growth forest cover [19] (Figure 6). Nevertheless, one may reasonably argue that the conservation value and sustainability of SBA may decline significantly in landscapes that have lost most (>70%) of the original old-growth forests [13,14,19], as this compositional pattern results in considerable abiotic (e.g., environmental stress) and biotic (e.g., loss of forest species) impacts. As the delivery of ecosystem services is contingent upon a human presence to receive them, it may reach a peak in landscapes with an intermediate deforestation level [19] (Figure 6). However, increasing human population pressures in the landscape may subsequently result in a decline in the provision of ecosystem services (Figure 6).

6. Conclusions and Implications for the Sustainability of Tropical Ecosystems and SBA Itself

For decades, ecological studies have promoted SBA as a sustainable and low-impact agricultural practice. However, our review reveals that the abiotic and biotic impacts of SBA should not be underestimated, as they threaten the sustainability of this ancient agroecosystem. Despite the relevance of the issue, we found that SBA effects are typically assessed at the local (stand) scale, overlooking impacts at larger scales. This limitation is worrying, as this agriculture practice can influence environmental variables on a broader scale [220]. Furthermore, the existing information on the topic is focused on the forest recovery stage, which can have important positive abiotic and biotic effects. However, these positive effects do not always reverse the damages caused by the first stages of the slash-and-burn cycle, since fallow times are often too short (<10 years) to recover the environmental conditions (e.g., climate, soil fertility, biodiversity, and biomass) of the original forest.

Five major take-home lessons can be highlighted from this review. First, slashing of vegetation is particularly damaging, as it promotes the loss of forest cover, which can have a disproportionately high impact on tropical plants, especially rare and threatened species. Second, the burning stage is also harmful, especially due to the emission of greenhouse gases and their contribution to global warming. Although fire can increase soil fertility, this is short-term benefit as nutrients are rapidly lost through leaching, surface runoff, and absorption by plants. Third, the impact of the farming stage depends on the management practices, but it can be particularly damaging when farmers use agrochemicals and reduce the time of use of farming areas, as this practice results in the conversion of more old-growth and secondary forests. Four, the forest recovery stage is critical to restore several ecosystem parameters, but secondary forests cannot completely replace the functions of old-growth forests, for instance, in terms of biodiversity conservation and climate regulation. Finally, the impacts of SBA are not limited to the crop area, but can shape ecological patterns and processes in the surrounding landscape. Although these landscape-scale impacts depend on the spatial structure of the landscape mosaic, the amount of old-growth forest cover is likely the most important variable determining the landscape-scale impacts of SBA.

In summary, the sustainability of SBA is challenged by its significant environmental impacts, which threaten the integrity of tropical ecosystems and the practice’s own viability. Preserving a substantial amount of forest cover is crucial to mitigate these impacts and ensure the resilience of natural and agricultural systems. Such preservation of forest cover can be achieved by implementing strategies that improve soil fertility and extend fallow periods. These strategies have been increasingly studied in agroecology [250,251], such as the combination of perennial elements with crops (agroforestry)—a strategy that can provide important benefits to the local population (e.g., wood and non-wood products) and the environment (e.g., soil quality restoration, biodiversity conservation, and carbon sequestration) [252]. Moreover, integrating syntropic farming practices could enhance ecosystem regeneration and productivity, promoting long-term sustainability while reducing the dependency on destructive methods like slash-and-burn [253]. Forest loss can also be achieved by increasing sustainable non-agricultural occupation (e.g., tourism activities) and planning population density in rural communities [6].

We do not want to finish this review without recalling that most of the reviewed studies were carried out at the local scale. This bias suggests that our assessment is likely conservative, as the impact of SBA should be more detrimental if extended over larger spatial extents. Therefore, future research should integrate the emerging knowledge and methods from landscape ecology (e.g., drone monitoring and remote sensing) to assess the effects of SBA across entire landscape mosaics [254]. This is an interesting avenue for future research that should be pursued to have a more comprehensive understanding of the impact of SBA across multiple scales.