Definition of agroforestry

The World Agroforestry Centre defines agroforestry as the cultivation of trees in association with crops and livestock either in a spatial mixture or in a temporal sequence. ^{Garrett (2009)} describes AFS as a land use practice where woody perennials interact biologically with crops and livestock in the same space, with the fundamental purpose to diversify and optimize productivity by taking into account the principle of sustainability. Recently, ^{Nair et al. (2008)} defined agroforestry as landscape management involving trees and shrubs interacting with crops and livestock under an integral scheme in temporal or sequential phases, offering a wide variety of benefits and services.

Despite some differences in the exact definition of agroforestry, all of them recognize that integrating trees and shrubs with other elements of agriculture (i. e. crops and livestock) can increase soil fertility, reduce soil erosion, improve water and air quality, favor biodiversity, enhance aesthetic appeal, increase carbon capture and storage, and reduce methane emissions from livestock (^{Ganry, Feller, Harmand, & Guibert, 2001}; ^{Harvey & González-Villalobos, 2007}; ^{Montagnini, 2006}; ^{Montagnini et al., 2015}; ^{Nair et al., 2008}; ^{Nair, Kumar, & Nair, 2009}; ^{Palm et al., 2005}; ^{Pandey, 2002}; ^{Schroth & Sinclair, 2003}; ^{Schroth et al., 2004}). In addition, ^{Shibu (2009)} recognized that these diverse ecosystem services vary at different spatial scales (local, regional, or global) (Table 1).

Services from agroforestry systems

Carbon sequestration. Carbon sequestration is the capture and storage of atmospheric carbon into carbon sinks (e. g. oceans, vegetation, or soils) through physical and biological processes (^{Ibrahim, Villanueva, & Mora, 2005}). Incorporating trees and shrubs into AFS can markedly increase carbon sequestration compared with other systems like monoculture pastures. Besides storing important amounts of carbon in aboveground biomass, they can also store greater amounts of carbon in belowground biomass (^{Nair et al., 2009}). ^{Pandey (2002)} declared that, in the context of carbon cycling, agroforestry is important for two reasons: 1) the tree component captures atmospheric carbon through photosynthesis and stores it underground, behaving as a carbon sink; and, 2) agroforestry reduces deforestation of temperate and tropical forests related to clearing for agriculture.

A wide range of studies (^{Albrecht & Kandji, 2003}; ^{Ibrahim et al., 2005}; ^{Montagnini et al., 2015}; ^{Nair et al., 2009}; ^{Palm et al., 2005}; ^{Schroth et al., 2004}; ^{Shibu, 2009}) support the concept that AFS is a unique opportunity to increase carbon reserves in the terrestrial biosphere, even if AFS were not originally designed for that purpose. ^{Dixon (1995)} estimated that a total area of 585 to 1,215 million ha implementing agroforestry management (in Africa, Asia, and America) has the potential to capture and store 1.1 to 2.2 Pg of carbon in vegetation and soils, over a period of 50 years. However, ^{Nair et al. (2009)} estimated that the world’s agroforestry area amounts to 1,023 million ha. Using the mean carbon sequestration values of ^{Dixon (1995)} and a stable agroforestry area, we estimate a potential carbon capture and storage of 1.9 Pg in a period of 50 years (^{Nair et al., 2009}).

There is also enormous potential for additional carbon sequestration from a huge area of croplands and degraded pastures in the tropics, which could be transformed by agroforestry management practices (^{Shibu, 2009}). However, carbon sequestration potential is related to the kind of system, species, spatial design, age, geographic placement, environmental factors, and management practices (^{Ibrahim et al., 2005}; ^{Shibu, 2009}). This variability inherent in AFS and the lack of uniform methodologies have hindered estimations of carbon sequestration of AFS. For example, (^{Nair et al., 2009}) found that above and below ground biomass carbon sequestration of AFS varied from 0.29 Mg·ha^-1·yr^-1 in a forage bank in West Africa, to 15.2 Mg·ha^-1·yr^-1 in mixed plots in Puerto Rico. In addition, an estimated 1.25 Mg·ha^-1 of soil carbon is stored in alley cropping AFS in southern Canada, in comparison with 173 Mg·ha^-1 in silvopastoral systems on the atlantic coast of Costa Rica (^{Nair et al., 2009}). In general, agroforestry systems in arid, semiarid, and degraded zones have less carbon sequestration potential than those in wet and fertile zones; and, temperate zones have less potential than tropical zones (^{Nair et al., 2009}). These factors affect the C storage capacity in different stocks within the system: biomass C ranges from 9 to 54 % of total C, and soil C ranges from 46 to 91 % of total C in the system.Table 2 shows the characteristics (i. e. age, species, and soils) of selected AFS in Mexico, and their C storage capacity.

Methane emissions. Methane (CH₄) is one of the products of fermentation and metabolism of carbohydrates in the rumen. Methane, carbon dioxide (CO₂), and nitrous oxide (NO₂) represent the main greenhouse gases produced by ruminants (^{Pinares-Patiño, Muetzel, Molnano, Hunt, & Clark, 2010}). Eighty million tons of CH₄ are produced annually in the world (^{Eckard, Grianger, & de Klein, 2010}) with ruminants responsible for 30 % of these emissions. Methane has a global warming potential 23 times greater than CO₂, and represents a loss of between 5 to 18 % of gross energy consumed by ruminants; however, this figure is more pronounced in animals fed diets high in fiber and low in quality (^{Eckard et al., 2010}).

Ruminant grazing systems in the tropics are characterized by low forage availability and quality at certain times of year. Feeding systems based on cellulose leads to increased acetic acid production, releasing a large amount of metabolic hydrogen (H₂) which is used as substrate for the production of CH₄. In comparison with diets high in starch, ruminal fermentation favors the production of propionate (^{Pinares-Patiño et al., 2010}).

In the tropics there are a diversity of trees and shrubs, such as Leucaena leucocephala (Lam.) de Wit, Piscidia piscipula (L.) Sarg., Pithecellobium saman (Jacq.) Benth., Guazuma ulmifolia Lam., Enterolobium cyclocarpum (Jacq.) Griseb., Gliricidia sepium (Jacq.) Walp., Sesbania sesban (L.) Merr. that have great potential for feeding ruminants (^{Ferrer et al., 2007}; ^{Pinares-Patiño et al., 2010}). These tree species have a wide range of secondary metabolites such as coumarins, phenols, condensed tannins and saponins, which can significantly reduce enteric CH₄ emissions (^{Dias-Moreira et al., 2013}; ^{Silivong, Xaykham, Aloun, & Preston, 2012}). Condensed tannins (CT) have the ability to form complexes with proteins and carbohydrates, making them less digestible and fermentable in the rumen, thereby reducing the formation of H₂ specific substrates for methane formation and substantially reducing enteric CH₄ formation (^{Pinares-Patiño et al., 2010}). Additionally, condensed tannins possess the ability to significantly reduce the bacterial population of the rumen protozoa which are responsible for the formation of substrates for the synthesis of CH₄. Several studies (^{Abdalla et al., 2007}; ^{Berra, Finster, & Valtorta, 2009}; ^{Dias-Moreira et al., 2013}) argue that through the manipulation of protozoan and bacterial populations, secondary metabolites of trees and shrubs can reduce CH₄ emissions by 25.7 % to 36.9 % (Table 3). Steroidal saponins are an abundant compound in some tropical trees and shrubs. Their main mode of action on the modification of rumen fermentation is through defaunation of protozoa which are directly responsible for CH₄. These compounds are present in the foliage and fruit of species such as Sapindus saponaria L., P. saman, G. sepium, and E. cyclocarpum, which have been intensively studied (^{Hess et al., 2004}; ^{Silivong et al., 2012}). Reports show a reduction in CH₄ emissions ranging from 10.5 % in sheep (^{Hess et al., 2004}) to 70 % in goats (^{Silivong et al., 2012}) (Table 3).

Soil fertility. It is indisputable that decreasing vegetation cover has caused a reduction in nutrient cycling and soil fertility (^{Sanginga et al., 2003}). As deforestation and ecosystem deterioration in the tropics has occurred in part due to land clearing for conventional livestock raising, strategies to help mitigate its negative impacts on the environment have been developed. Indeed, modern agroforestry was developed as a strategy to improve sustainability of agroecosystems (^{Nair et al., 2008}); however, mismanagement can lead to system fragmentation, as has been the case in traditional AFS.

Forest ecosystems are closed and efficient systems (^{Petit, Casanova, & Solorio, 2009}). They have high rates of return and low rates of losses, and are thus self-sustaining. On the other hand, many conventional agroecosystems (e. g., monocrops) are open or permeable, with relatively low rates of return and high rates of losses. AFS are positioned between these two extremes, with more efficient nutrient cycling than conventional agricultural systems and similar productivity to forest ecosystems. ^{Nair et al. (2008)} claimed that the difference between AFS and other agricultural land use practices lies in the transference or recovery of nutrients into the system from one component to another, and the possibility of managing the system or its components to increase nutrient recycling rates without affecting total productivity.

The role of AFS in improvement and stabilization of long-term soil productivity and sustainability is well documented (^{Nair et al., 2008}; ^{Schroth & Sinclair, 2003}). Inclusion of nitrogen fixing trees and crops is a common practice in tropical AFS; however, non-fixing trees can also improve the physical, chemical, and biological characteristics of soils through organic material shed and nutrient cycling (^{Petit-Aldana, Uribe-Valle, Casanova-Lugo, Solorio-Sánchez, & Ramírez-Avilés, 2012}). Trees have deep root systems which serve as an underground net through which nutrients can be captured from deep in the soil profile. These nutrients are returned to the soil via leaf litter, increasing the nutrient recycling efficiency of the system (^{Allen et al., 2004}). Furthermore, trees have long life cycles, increasing capture of nutrients before and after the crop cycle that might otherwise be lost. In locations where shifting cultivation practices include planting trees to improve the fallow cycle, trees can minimize soil fertility decline (^{Ganry et al., 2001}; ^{Sanginga et al., 2003}). Soil inorganic N, aerobic mineralization, and biomass can be significantly higher after crop rotation with N fixing trees than with non N fixing trees and/ or pastures (^{Harmand & Balle, 2001}). For herbaceous fallows, greater organic material accumulation in soils, nutrient storage in biomass, and greater density and vertical distribution of roots help maintain nutrient reserves by reducing leaching and/or “pumping” nutrients from the deepest soil layers to the soil surface (Nair et al., 2008).

^{Uribe and Petit (2007)} analyzed the influence of vegetation cover (L. leucocephala, Mucuna pruriens (L.) DC. and secondary vegetation) and crop rotation intervals on soil fertility restoration of crop lands in Yucatán, Mexico. Their results indicated that short crop rotation intervals contribute to restoring soil fertility, depending on the species. Leucaena leucocephala enhanced K, Ca, and Mg content; M. pruriens increased NO₃ content; and, secondary vegetation increased organic matter content.

Biodiversity conservation. The urgent need for designing effective strategies for biodiversity conservation has attracted enormous worldwide attention. Scientists and politicians are increasingly aware of agroforestry’s role in biodiversity conservation in both tropical and temperate regions. The importance of agroforestry in biodiversity conservation has been highlighted recently by numerous authors (^{Harvey, González, & Somarriba, 2006}; ^{McNeely, 2004}; ^{Schroth et al., 2004}). ^{Shibu (2009)} indicated that agroforestry plays five fundamental roles in biodiversity conservation: a) provides habitat to perturbation tolerant species; b) permits preservation of germplasm of sensitive species; c) helps to reduce habitat destruction, providing more productive and sustainable alternatives to conventional (i. e. modern) agricultural systems that require increasingly more land; d) supplies connectivity through corridors between habitat remnants, creating an integrating net that increases the conservation of flora and fauna; and e) provides services in erosion control and groundwater recharge, preventing habitat degradation and loss.

^{Harvey and González-Villalobos (2007)} studied bird and bat assemblages from undisturbed forests; two AFS, cacao (Theobroma cacao L.) and banana (Musa sp.); and plantain monocultures at an indigenous reservation in Talamanca, Costa Rica. They found that AFS maintained bat and bird assemblages that were as (or more) species-rich, abundant and diverse as forests, and had the same basic suite of dominant species. However, the species composition of these assemblages was highly modified. On the other hand, the plantain monocultures had highly modified and disparate assemblages of both birds and bats.

Homegardens are another type of agroforestry system that have been studied for conservation values, and are well known for their high species diversity. Many ecologists consider their structure and functionality to closely resemble those of natural forests (^{Kabir & Webb, 2009}). ^{Kumar and Nair (2004)} reported species richness of tropical homegardens ranging from 27 in Sri Lanka to 602 in West Java, Indonesia. Furthermore, in tropical zones where agriculture has eliminated forest cover, homegardens and AFS are refuges for many species. For instance, in Bangladesh, where forest cover is less than 10 % of original cover, homegardens (which are implemented by at least 20 million households) are an important source of conservation. ^{Kabir and Webb (2009)} recorded the floristic and structural diversity of 402 homegardens from six regions of Bangladesh, and found 419 species, 59 % of which were native, and six with some conservation priority status.

Crop combinations and spatial arrangements of AFS influence density and species diversity of insect populations. ^{Brandle, Hodges, and Zhou (2004)} reported a high density and diversity of insects in windbreaks. They associated this diversity with the heterogeneous windbreak profile, which provides a wide variety of microhabitats at every life cycle stage, and resource availability, including hosts, prey, pollen, and nectar. AFS provide an appropriate habitat for wildlife due to the high complexity of landscape structure and composition. Windbreaks, tree barriers, and riparian damping zones offer a woody habitat for wildlife in landscapes dominated by agriculture (^{Harvey et al., 2006}; ^{Harvey & González-Villalobos, 2007}).

^{Da Silva, da Gama-Rodrigues, da Gama-Rodrigues, Machado, and Baligar (2009)} compared the distribution of meso- and macro-fauna communities in soils and leaf litter between a cacao AFS and a natural forest in south Bahia, Brasil. Results suggested that a high diversity of plants in agroforestry systems and natural forest provide more microhabitats and heterogeneity in the leaf litter, and thus greater biodiversity in soils. They concluded that these AFS have beneficial effects for soil fauna communities, and can be used as a strategy for their conservation.

Habitat loss and fragmentation are major threats to biodiversity (^{Harvey et al., 2006}; ^{Harvey & González-Villalobos, 2007}; ^{Schroth et al., 2004}). In Latin America, land use change from natural forests to pasture has caused changes in the size and distribution of forest remnants, loss of biodiversity, and water pollution (^{Harvey, Alpízar, Chacón, & Madrigal, 2005}). Remnant landscapes are a mosaic of forest fragments spread over pasture or crops lands; however, tree cover of these agricultural systems is abundant and with different spatial arrangements including forest patches, riparian forests, isolated trees, living fences, and windbreak tree barriers. These trees can be remnants of original forests, a result of natural regeneration, or planted by farmers (^{Martínez-Encino, Villanueva-López, & Casanova-Lugo, 2013}; ^{Montagnini et al., 2015}).

From a biodiversity conservation point of view, AFS can provide habitat, feeding sites, perch, and biological corridors for plants and animals (^{Ibrahim et al., 2005}). Studies have evaluated the role of trees in silvopastoral systems (a type of AFS) for fauna and flora biodiversity conservation, species population maintenance, and impact on ecological processes in agricultural landscapes (Table 4). For example, ^{Camargo, Ibrahim, Somarriba, Finegan, and Current (2000)} noted that silvopastoral systems can be structurally variable and floristically diverse depending on their origin, which can include remnants, natural regeneration, or sown. The importance of remnants depends upon their structure, composition, management, and spatial arrangement in the agricultural landscape. ^{Enríquez-Lenis, Sáenz, & Ibrahim (2007)} plant diversity and landscape heterogeneity positively affect bird richness and abundance in agricultural landscapes.

Water conservation. The environmental services provided by AFS in relation to water cycles have been insufficiently studied (^{Beer et al., 2003}). Trees in AFS influence the water cycle by increasing water interception from rain and clouds, modifying transpiration and soil water retention, reducing runoff, and increasing soil infiltration. ^{Bharati, Lee, Isenhart, and Schultz (2002)} reported that infiltration on land cultivated with corn or soybean, or under pastures, was five times less than under riparian furrows cultivated with a wide variety of herbaceous and tree species. This suggests that these more diversified systems can prevent runoff and the loss of nutrients.

Trees in AFS can reduce leaching losses and contamination of groundwater reserves by tannins or other environmental and human hazardous substances. As a result of reduced runoff and leaching, micro-basins with good forest or AFS cover produce high quality water (^{Chikowo, Mapfumo, Nyamugafata, & Giller, 2004}). With conventional agricultural systems, typically less than 50 % of applied N and P is utilized by crops. In consequence, excess fertilizer moves beyond the reach of plants through runoff or leaching to deeper soil layers, contaminating groundwater reserves and decreasing water quality (^{Udawatta, Garrett, & Kallenbach, 2010}). Therefore, agroforestry management practices can be used as a strategy to provide clean water (^{Shibu, 2009}). For instance, in coffee production areas of Costa Rica where large quantities of N are applied, nitrate loss through leaching was reduced by the presence of Eucalyptus deglupta Blume, probably due to increased total evaporative demand and nitrate uptake during the dry season (^{Ávila et al., 2004}).

In Brazil, ^{Nepstad et al. (1994)} observed that under extreme drought, soil water availability at 2 to 8 m was less in a degraded pasture than in the forest (310 versus 380 mm, respectively), which could be associated with greater organic matter content and ground cover in the forest. A decrease in water availability in degraded pasture soils indicates that there is less water stored in these soils than in forested soils, and consequently there is less water infiltration to the aquifer in forested soils. At the end of the dry season, the forest stored approximately 770 mm of water in the first 8 m of soil, compared with less than 400 mm in degraded pasture soils, which indicates that water scarcity can be a critical factor in arid and semiarid landscapes dominated by pastures. Studies in Costa Rica (^{Ríos, 2006}) showed that surface runoff was significantly greater in degraded pastures (42 % of rainfall) compared with forage banks with woody perennials (3 %), secondary forests (6 %), and pastures with a high density of trees (12 %). This confirms that land use practices with high tree cover are beneficial for water capture.

Air quality conservation. Interest in AFS use as shelterbelts for improving air quality has received considerable attention (^{Tyndall & Colletti, 2007}). Trees and shrubs employed as barriers have been utilized to reduce odor emissions, mainly in regions with a high concentration of livestock. Plant species can act as buffers by filtrating particles from air currents, and removing dust, gas, and microbes (^{Abbasi & Khan, 2000}). Strategically designed shelterbelts can be an efficient way to mitigate the problem in a socioeconomically responsible manner (^{Tyndall & Colleti, 2007}). Nevertheless, the capacity of shelterbelts to reduce odors depends on their structure (i. e. height, length, width, and density of barrier). For instance, short shelterbelts will only intercept some of the odor that is in contact with the trees, in contrast to taller shelterbelts which have the possibility of covering a greater quantity of the odor source (^{Tyndall & Colleti, 2007}).

Future research needs

With a growing global population, decreasing natural resources, and a changing climate, the alternatives for sustainable land use, such as AFS, not only generate the interest of academics, but also of the general public. Further research is required to accurately assess the potential of various forms of agroforestry in the tropics. For example, studies on the role of trees for biological enrichment based livestock systems in large areas of monoculture pastures, and the effect of forage diversity on reducing ruminant methane emissions under grazing conditions are needed. In addition, further studies are required to assess the short-, medium- and long-term economic viability of these systems. To overcome these and other limitations, much more basic and applied research is required to help elucidate the elementary processes that drive or limit the use of agroforestry systems. In particular, it is necessary to investigate the interactions between the various components of the system, which are the basis for the design, evaluation and management of AFS.