Traditional Agroforestry Systems and Conservation of Native Plant Diversity of Seasonally Dry Tropical Forests

Traditional agroforestry systems (TAFS), which integrate crops with wildlife, are important reservoirs of human culture and technical experiences with a high capacity for biodiversity conservation. Our study aimed to evaluate the capacity of TAFS to conserve the floristic diversity of tropical dry forests (TDF) in the Tehuacán-Cuicatlán Valley, Mexico. We compared TAFS and TDF by measuring their forest cover, floristic composition, and structure, in addition to documenting the motivations of people to maintain native vegetation in their agricultural fields. We conducted a restricted randomized sampling of perennial plant species, including nine sites of TAFS and nine of TDF to determine the alpha, beta, and gamma diversity. Furthermore, we conducted semi-structured interviews with peasants who managed the agricultural plots we studied. We also performed workshops with people of the communities where surveys were performed. Our findings show that TAFS can maintain, on average, 68% of the species (95% of them native to the region) and 53% of the abundance of individuals occurring in the adjacent TDF. TAFS harbour 30% (39 species) of plants endemic to Mexico. Total species richness of TDF and TAFS were similar, as well as the effective number of species or communities estimated for the alpha, beta, and gamma diversity, but differed in the abundance of individuals. The high species turnover recorded in TDF (72%) and TAFS (74%) has profound implications for conservation, suggesting that it would be necessary to maintain several sites in order to conserve the regional diversity of native vegetation. Material, non-material, and regulatory contributions were reported to be the reason that peasants take into account maintaining natural vegetation. TAFS associated with TDF in the region (also called “Apancles”) contain an important richness, diversity, and endemism of components of natural ecosystems, as well as provide multiple socio-ecological contributions. These systems could represent a viable alternative to reconcile biological conservation with social well-being.


Introduction
Traditional forms of rural life are commonly able to satisfy basic peasants households' needs by using natural ecosystems and biodiversity while conserving them [1][2][3][4]. Among the strategies for such purposes, traditional agroforestry systems (TAFS) are outstanding. These systems deliberately

Study Area
The Tehuacán-Cuicatlán Valley is one of the main reservoirs of the biocultural richness of Mexico, with a human cultural history of approximately 12,000 years [7], and early archaeological signs of agriculture in association with forest management [7,[41][42][43][44]. In 1998, the Mexican government decreed the Tehuacán-Cuicatlán Biosphere Reserve, with the purpose to maintain the regional biodiversity shared among the states of Oaxaca and Puebla. More recently, in 2018, the UNESCO inscribed the "Tehuacán-Cuicatlán Valley: originary habitat of Mesoamerica" within the World Heritage List as a Mixed Heritage of Humanity (both cultural and natural), due to the extraordinary value of the regional natural ecosystems and biodiversity, the unique cultural history, and the diversity of traditional Mesoamerican cultures.
This study was conducted in three communities of Mazatec and Cuicatec origin, in the "Cañada Oaxaqueña" region, in the municipality of San Juan Bautista Cuicatlán. This territory has a semi-arid and very dry climate (annual mean temperature and precipitation of 25 • C and 485 mm, respectively), Sustainability 2020, 12, 4600 4 of 27 being the confluence zone of several rivers that allow the presence of riparian vegetation, columnar cacti forests, and TDF. Our studies were conducted in the communities of Santiago Quiotepec, San Juan Bautista Cuicatlán, and Santiago Dominguillo, where local people practice primary activities like agriculture, raising goats, and planting fruit trees and gathering of non-timber forest products [45,46].
The Apancle: A Traditional Agroforestry System Associated with Tropical Dry Forest The term "Apancle" or "Apantle" is the local name for the TAFS associated with the TDF of the Cañada Oaxaqueña region. The name derives from "Apantli", a Náhuatl term referring to the irrigation channels [47] that make agriculture possible in the dry zone studied. It is also the name given to the irrigated agricultural systems found in practically all the communities of the region.
The Apancle is a small-scale traditional agricultural system, carried out in crop fields of 1-3 ha on average (1.66 ± 0.55 ha, min. 0.69, max. 2.38), in areas irrigated by permanent rivers, springs, and seasonal streams. There, people cultivate native varieties of maize ("blanco", "negrito" or "prieto", "pinto", "amarillo"), beans ("delgadito", "mosquito"), and squash ("támala"), mainly destined for direct consumption by households and partly used for barter, presents to friends, and commercialization at local level. Fruit trees cultivated there (lemon, Citrus aurantifolia; sapodilla, Manilkara zapota; mango, Mangifera indica; jobo, Spondias purpurea; annona Annona reticulata) are directed mainly to regional markets and partly to direct consumption. In the Apancle, the domesticated plants are integrated with remnants of native vegetation, live fences, forest cover patches, vegetation islands, and fringes, as well as isolated trees of the TDF. In these TAFS, people make use of organic fertilizers (dung from goats, cows, and bats, and ash from home stoves) combined with chemical fertilizers (urea and other nitrogen compounds, and diammanic phosphate). This means a relatively low economic investment in external inputs, since people select and store their seeds for the following agricultural cycle, and make use of the labour of family members (except during seed sowing, when they pay two people for two days for sowing 0.5 ha). Land rotation without using fire is undertaken in one-to two-year cycles (in the community of Quiotepec). Mechanization is at a low level, mainly using manual tools like shovels, sticks, "talacho" (a hand tool used for digging), machetes, "chicole" (a long stick with a basket in the tip for fruit harvesting), and ploughs; tractors are most commonly rented and used only for preparing the land. After harvest, people allow cattle and goats to enter the agricultural fields to feed on straw, and therefore the Apancle should be considered an agrosilvopastoral system, according to Nair [48].

Sampling Design
To evaluate the capacity of TAFS for plant diversity conservation, we compared the vegetation maintained in these systems with areas of TDF. We measured the vegetation cover maintained in the agricultural plots through the different types of agroforestry practices, as well as the floristic composition, structure (abundance, density, frequency), the spatial arrangement of components, and motivations of people to maintain plants and vegetation in their agricultural fields.
We adapted sampling methods developed for analysing TAFS in other vegetation types of the region by Moreno-Calles et al. [15][16][17], Vallejo et al. [14,38,39] and Campos-Salas et al. [40]. We conducted a restricted randomized sampling of perennial species, including nine sites of TAFS and nine of TDF. For selecting the TAFS plots studied, we firstly identified several plots, numbered these, selected three from each community by randomly choosing numbers, and then asked for permission to sample from the various landowners. TDF sites were selected considering similar topographic, soil, and geomorphological conditions to TAFS. In each site of TDF, we sampled a total area of 500 m 2 by sampling five squares of 10 × 10 m (100 m 2 ) each, randomly located in an area of 1.5 ha (the average area of agricultural plots studied), separating each 100 m 2 by at least 20 m. In each sampling square, we counted all individuals of the different species of trees and columnar cacti (abundance), registering for each tree its height and diameter at breast height ≥2 cm (DBH, approximately at 1.3 m). Each 100 m 2 square was subdivided into four 5 × 5 m (25 m 2 ) nested squares, one of which was randomly selected for recording height and two perpendicular diameters of shrubs, lianas (woody climbers), rosetophyllous Sustainability 2020, 12, 4600 5 of 27 plants (of the genera Agave and Hechtia), and globose cacti. Previously, we sampled vegetation in nine sites of TAFS with a total size in each plot the same area sampled in the TDF sites (i.e., 500 m 2 ). In each site, we also included five 100 m 2 squares (some 10 × 10 m and others 20 × 5 m, according to the form of the vegetation patches), where we measured the growth forms referred to. Each agricultural plot required a specific design of the sampled area, sectorizing and proportionally representing every type of agroforestry practice recorded; for defining such strategies, we mapped the distribution of the vegetation areas inside the plot, with the help of Google Earth Pro ® (Figure 1). Sampled TAFS included plots with different levels of management intensity, which was considered to be higher in plots with lower vegetation cover and higher energy invested in management, complexity of tools used, and crop production [49,50].
Sustainability 2020, 12, x FOR PEER REVIEW 5 of 27 m). Each 100 m 2 square was subdivided into four 5 × 5 m (25 m 2 ) nested squares, one of which was randomly selected for recording height and two perpendicular diameters of shrubs, lianas (woody climbers), rosetophyllous plants (of the genera Agave and Hechtia), and globose cacti. Previously, we sampled vegetation in nine sites of TAFS with a total size in each plot the same area sampled in the TDF sites (i.e., 500 m 2 ). In each site, we also included five 100 m 2 squares (some 10 × 10 m and others 20 × 5 m, according to the form of the vegetation patches), where we measured the growth forms referred to. Each agricultural plot required a specific design of the sampled area, sectorizing and proportionally representing every type of agroforestry practice recorded; for defining such strategies, we mapped the distribution of the vegetation areas inside the plot, with the help of Google Earth Pro ® (Figure 1). Sampled TAFS included plots with different levels of management intensity, which was considered to be higher in plots with lower vegetation cover and higher energy invested in management, complexity of tools used, and crop production [49,50]. In all sites sampled, we conducted ethnobotanical collection and preparation of herbarium specimens, and took photographs. Based on these materials, we then documented the cultural value of the plants occurring in the studied areas. We identified and processed all samples and the voucher In all sites sampled, we conducted ethnobotanical collection and preparation of herbarium specimens, and took photographs. Based on these materials, we then documented the cultural value of the plants occurring in the studied areas. We identified and processed all samples and the voucher specimens were deposited to the herbaria the National Herbarium of Mexico MEXU and the IBUG from the University of Guadalajara, Mexico (acronyms according to Thiers [51]). The taxonomic identification was carried out rigorously by experts in the field, in addition to the support by specialists in some plant groups. Particularly helpful for identifying plant specimens was the collection of the project "Flora del Valle de Tehuacán-Cuicatlán". For estimating and comparing plant diversity in the different settings, as far as possible, we carried out vegetation sampling under similar environmental conditions to the sites where the TAFS were sampled, particularly elevation and slope inclination. For TAFS, on average, 652 ± 79.73 m (min. 590, max. 770) and 19.33 ± 9.81 • (min. 0, max. 30), respectively, while for TDF 771 ± 163.23 m (min. 625, max. 1071) and 19.78 ± 9.93 • (min. 5, max. 32).
To document the motivations of local people for conserving native vegetation, we conducted semi-structured interviews with the peasants who managed the agricultural plots studied, as well as workshops with groups of people of the communities where the research was conducted. In the interviews, we asked questions about the history of each crop field, socio-economic aspects of the production units, aspects of agricultural management to characterize the level of intensification, and forms of managing vegetation and criteria for making decisions about the maintenance of wild species in the agricultural plots (Supplementary File S2). We classified the benefits of the vegetation maintained within agricultural plots following the proposal of Díaz et al. [52] into three main topics: (i) material contributions-substances, objects, or other tangible elements of nature that directly sustain the existence of people; (ii) nonmaterial contributions-effects of nature on subjective or psychological aspects that sustain the quality of life of people and that provide opportunities for recreation, inspiration, spiritual experiences, or social cohesion; and (iii) regulating contributions-structural and functional aspects of organisms and ecosystems that modify the environmental conditions experienced by people and that regulate the generation of material and nonmaterial contributions. The study was conducted from September 2017 to August 2019.

Data Analyses
We estimated the average diversity at the local level of each site (alpha diversity; α), the total diversity at the regional level of the Cañada Oaxaqueña (gamma diversity; γ), and the relationship between both, which reflects the change in species composition (beta diversity; β = γ/α) [53], indicating the number of effective communities, which can range from 1 to N (nine in this case).
For the beta diversity, we estimated the Sørensen pairwise dissimilarity index (β sor = b + c/2a + b + c), where a represents the total number of species that occur in both sites, b represents the total number of species that occur in the neighbouring site but not in the focal site, and c represents the total number of species that occur in the focal site but not in the neighbouring site [54]. We used this index to describe the spatial differentiation and the differences in species richness between communities, as well as to obtain the total beta diversity expressed as a percentage. Furthermore, we explored the partitioning of the spatial turnover and nestedness of species assemblages proposed by Baselga [55]. The spatial turnover (β sim = min (b, c)/a + min (b, c)) estimated the replacement of some species by others, while the nestedness (β nes = β sor − β sim ) identified which of the biotas of sites with smaller numbers of species were subsets of the biotas at richer sites [55].
We estimated the effective numbers of species (also called Hill numbers) as measures of "true" diversity of order q = 0, 1, and 2 [56] for perennial species, including trees, shrubs, lianas, cacti (columnar and globose), and rosetophyllous plants (of the genera Agave and Hechtia). The q exponent determines the sensitivity of the index to relative species abundance, that is, the influence that rare, typical, or dominant species may have in the estimation of diversity [56][57][58]. When q = 0 ( 0 D, diversity of order 0) the abundance of species does not influence the value of q, thus providing disproportionate weight to rare species, and the obtained value is equivalent to the species richness. When q = 1 ( 1 D, diversity of order 1) all species have a weight proportional to their abundance in the community (it is, therefore, one of the best parameters to estimate diversity), and is equivalent to the exponential of Shannon's entropy index calculated with the natural logarithm ( 1 D = exp H'). The Hill number of order 1 can be therefore interpreted as the number of typical species in the community. When q = 2 ( 2 D, diversity of order 2) the abundant species have higher influence and rare species are discounted; Sustainability 2020, 12, 4600 7 of 27 hence, this diversity can be interpreted as the number of dominant species in the community, and is equivalent to the inverse value of Simpson's dominance index ( 2 D = 1/D).
We used t-tests to assess statistically significant differences between TDF and TAFS sites in terms of abundance of individuals, richness, and species diversity. In addition, we calculated the evenness factor (EF = 2 D/ 0 D), which indicated how equitably the abundances of species were distributed (the closer to 1, the more the community was equitable) [59]. We also estimated the indexes of Shannon's entropy (H') and Simpson's dominance (D; the closer to 1, the community had higher dominance and was less diverse).
The analyses were performed using R statistical software (v. 3.6.3; R Development Core Team) with the package entropart [60].

Results
The TAFS studied can maintain on average 71% of the families, 66% of the genera, and 68% of the perennial species (95% of them native to the region), and 53% of the abundance of individuals occurring in the neighbouring TDF. The percentage of species is one of the highest recorded in TAFS of other vegetation types studied in the region, as shown in Figure 2. TAFS also provide multiple benefits to local societies, among which the most outstanding were shade, firewood, edible fruit, medicine, wood for making tools and fences, and ornamental and ritual uses. When q = 2 ( 2 D, diversity of order 2) the abundant species have higher influence and rare species are discounted; hence, this diversity can be interpreted as the number of dominant species in the community, and is equivalent to the inverse value of Simpson's dominance index ( 2 D = 1/D). We used t-tests to assess statistically significant differences between TDF and TAFS sites in terms of abundance of individuals, richness, and species diversity. In addition, we calculated the evenness factor (EF = 2 D/ 0 D), which indicated how equitably the abundances of species were distributed (the closer to 1, the more the community was equitable) [59]. We also estimated the indexes of Shannon's entropy (H') and Simpson's dominance (D; the closer to 1, the community had higher dominance and was less diverse).
The analyses were performed using R statistical software (v. 3.6.3; R Development Core Team) with the package entropart [60].

Results
The TAFS studied can maintain on average 71% of the families, 66% of the genera, and 68% of the perennial species (95% of them native to the region), and 53% of the abundance of individuals occurring in the neighbouring TDF. The percentage of species is one of the highest recorded in TAFS of other vegetation types studied in the region, as shown in Figure 2. TAFS also provide multiple benefits to local societies, among which the most outstanding were shade, firewood, edible fruit, medicine, wood for making tools and fences, and ornamental and ritual uses.

Types of Agroforestry Practices
In the study area, we found the following types of agroforestry practices: (i) remnants of native vegetation, that are areas of TDF with different degrees of conservation, from those well conserved and with high ecological integrity (even higher than some adjacent forest areas where the raising of goats and firewood extraction are practiced) to those with highly modified structure (most commonly favouring the abundance of useful plants); (ii) forest cover patches are formed by native species of TDF (i.e., "guajes", Leucaena spp. and jobo or "ciruelas", Spondias purpurea), wild species from other plant communities (i.e., "coyul", Acrocomia mexicana and sapodilla or "chicozapote", Manilkara zapota from the tropical moist forest), or exotic species (i.e., lemon and mango), and people promote the abundance of individuals for establishing orchards of useful species with high commercial value. These patches represent forest cover, but no remnants of TDF; (iii) live fences are mainly composed of species with the capacity for rooting and sprouting of branches, like Cactaceae, Burseraceae,

Types of Agroforestry Practices
In the study area, we found the following types of agroforestry practices: (i) remnants of native vegetation, that are areas of TDF with different degrees of conservation, from those well conserved and with high ecological integrity (even higher than some adjacent forest areas where the raising of goats and firewood extraction are practiced) to those with highly modified structure (most commonly favouring the abundance of useful plants); (ii) forest cover patches are formed by native species of TDF (i.e., "guajes", Leucaena spp. and jobo or "ciruelas", Spondias purpurea), wild species from other plant communities (i.e., "coyul", Acrocomia mexicana and sapodilla or "chicozapote", Manilkara zapota from the tropical moist forest), or exotic species (i.e., lemon and mango), and people promote the abundance of individuals for establishing orchards of useful species with high commercial value. These patches represent forest cover, but no remnants of TDF; (iii) live fences are mainly composed of species with the capacity for rooting and sprouting of branches, like Cactaceae, Burseraceae, Fabaceae, and Rhamnaceae, or other spiny plants used with the purpose of delimiting and protecting crops against livestock. Here, these species are also valued because they provide edible fruit, shade, and organic matter to enrich soils; (iv) isolated trees, which are commonly relatively large and multi-purpose individual trees, especially those providing shade, fruits, fodder, or wood (i.e., "mezquite", Prosopis laevigata; "matagallina", Quadrella incana; and "cardón", Pachycereus weberi); (v) vegetation fringes that we recorded in only one field mainly composed of species of the genus Agave, placed perpendicular to the slope inclination, forming terraces that contribute to retaining soil and humidity [61]; and (vi) vegetation islands (recorded in only one field) are a small vegetation patch, which is maintained since it has plants providing shade, microclimate regulation, and edible fruit ( Figure 3).
Sustainability 2020, 12, x FOR PEER REVIEW  8 of 27 Fabaceae, and Rhamnaceae, or other spiny plants used with the purpose of delimiting and protecting crops against livestock. Here, these species are also valued because they provide edible fruit, shade, and organic matter to enrich soils; (iv) isolated trees, which are commonly relatively large and multipurpose individual trees, especially those providing shade, fruits, fodder, or wood (i.e., "mezquite", Prosopis laevigata; "matagallina", Quadrella incana; and "cardón", Pachycereus weberi); (v) vegetation fringes that we recorded in only one field mainly composed of species of the genus Agave, placed perpendicular to the slope inclination, forming terraces that contribute to retaining soil and humidity [61]; and (vi) vegetation islands (recorded in only one field) are a small vegetation patch, which is maintained since it has plants providing shade, microclimate regulation, and edible fruit ( Figure 3). In the Apancles analysed, we recorded a higher proportion of agricultural area (54.02%) compared to forest cover, considering as forest cover the area occupied by the sum of all types of agroforestry practices. The average percentage of forest cover in Apancles was 45.98%. We In the Apancles analysed, we recorded a higher proportion of agricultural area (54.02%) compared to forest cover, considering as forest cover the area occupied by the sum of all types of agroforestry practices. The average percentage of forest cover in Apancles was 45.98%. We documented a gradient of management intensity in the varying percentage of forest cover that TAFS maintain, which ranged from 11.11% (in "Rinconada" Cuicatlán) to 89.29% (in "La Cañadita" Quiotepec) ( Table 1). We also documented correspondence between the amount of vegetation cover and the capacity for biodiversity conservation in the Apancles, as shown in Figure 4. The most frequent types of agroforestry practice were live fences and remnants of native vegetation, occurring in 89% of the sampling sites, then isolated trees in 78% of the sites, forest cover patches in 44% of the sites, and finally vegetation islands and fringes, which were recorded in one site each. In general, the forest cover was mainly due to remnants of native vegetation (38.33%), live fences (9.21%), and forest cover patches (4.29%). Isolated trees covered 0.88% of the area, vegetation fringes 0.38%, and vegetation islands 0.11% (Table 1).
Sustainability 2020, 12, x FOR PEER REVIEW 9 of 27 documented a gradient of management intensity in the varying percentage of forest cover that TAFS maintain, which ranged from 11.11% (in "Rinconada" Cuicatlán) to 89.29% (in "La Cañadita" Quiotepec) (Table 1). We also documented correspondence between the amount of vegetation cover and the capacity for biodiversity conservation in the Apancles, as shown in Figure 4. The most frequent types of agroforestry practice were live fences and remnants of native vegetation, occurring in 89% of the sampling sites, then isolated trees in 78% of the sites, forest cover patches in 44% of the sites, and finally vegetation islands and fringes, which were recorded in one site each. In general, the forest cover was mainly due to remnants of native vegetation (38.33%), live fences (9.21%), and forest cover patches (4.29%). Isolated trees covered 0.88% of the area, vegetation fringes 0.38%, and vegetation islands 0.11% (Table 1).

Capacity for Biodiversity Conservation
We documented in total 132 perennial plant species belonging to 101 genera and 39 families of Magnoliophyta (Supplementary File S1). We found significant differences only in the abundance of individuals (t = 3.414; p = 0.001; Figure 5). Total species richness ( 0 D) recorded in the sampling sites of TDF and TAFS were similar (98 and 101 species, respectively), as well as the effective number of species or communities estimated for the alpha, beta, and gamma diversity of order 1 ( 1 D) and 2 ( 2 D) ( Figure 6). However, the average alpha diversity was slightly higher in TDF ( 0 Dα = 34.67, 1 Dα = 19.92, and 2 Dα =11.12 effective species) than in the Apancles ( 0 Dα = 27.67, 1 Dα = 16.08, and 2 Dα = 9.29 effective species) ( Figure 6A, Table 2).
For beta diversity between sites ( Figure 6B), we found higher values of the effective number of communities in TAFS for order 0 ( 0 Dβ = 3.65 versus 2.83 in TDF), which indicates that in the Apancles the species turnover is mostly due to the rare species. On the other hand, the typical species in the Apancles and TDF are not being replaced ( 1 Dβ = 3.10 and 2.96 effective communities, respectively), while the turnover of dominant species in TDF sites is slightly higher ( 2 Dβ = 3.73) than in the Apancles ( 2 Dβ = 3.41).

Capacity for Biodiversity Conservation
We documented in total 132 perennial plant species belonging to 101 genera and 39 families of Magnoliophyta (Supplementary File S1). We found significant differences only in the abundance of individuals (t = 3.414; p = 0.001; Figure 5). Total species richness ( 0 D) recorded in the sampling sites of TDF and TAFS were similar (98 and 101 species, respectively), as well as the effective number of species or communities estimated for the alpha, beta, and gamma diversity of order 1 ( 1 D) and 2 ( 2 D) ( Figure 6). However, the average alpha diversity was slightly higher in TDF ( 0 D α = 34.67, 1 D α = 19.92, and 2 D α =11.12 effective species) than in the Apancles ( 0 D α = 27.67, 1 D α = 16.08, and 2 D α = 9.29 effective species) ( Figure 6A, Table 2). Table 1. Surface occupied by different types of agroforestry practices (forest cover) versus agricultural cover recorded in nine traditional agroforestry systems ("Apancles") analysed in the Cañada Oaxaqueña region. The forest cover corresponds to the area occupied by the sum of all the recorded agroforestry practices. The percentage of each type of cover is indicated in parentheses.

Remnants of Native
TDF was a more equitable community (evenness factor = 0.423) than Apancles (evenness factor = 0.314) ( Table 2), although both system types had similar richness and diversity.  For beta diversity between sites ( Figure 6B), we found higher values of the effective number of communities in TAFS for order 0 ( 0 D β = 3.65 versus 2.83 in TDF), which indicates that in the Apancles the species turnover is mostly due to the rare species. On the other hand, the typical species in the Apancles and TDF are not being replaced ( 1 D β = 3.10 and 2.96 effective communities, respectively), while the turnover of dominant species in TDF sites is slightly higher ( 2 D β = 3.73) than in the Apancles ( 2 D β = 3.41).

Floristic Composition
In the sampling sites, the plant families better represented were Fabaceae, Cactaceae, Euphorbiaceae, and Burseraceae, and the genera Bursera, Opuntia, Agave, and Vachellia (Table 3). We identified the species Phaulothamnus spinescens (Achatocarpaceae), which is a new record for the state of Oaxaca, a scarce species previously reported in arid zones of southwestern US, and the Mexican states of Baja California, Nayarit, Nuevo León, Puebla, Tamaulipas, and Sonora, in xerophytic scrubs at elevations of 900 to 1100 m [62]. This species was recorded only in one site of TDF and one Apancle of the community of Quiotepec at an elevation of 615 m. Table 3. Distribution of the number of genera and species (recorded in 18 sampling sites of 500 m 2 ) in the most diverse families and genera of tropical dry forests and traditional agroforestry systems of the Cañada Oaxaqueña region. The percentage with respect to the total number of genera and species is indicated in parentheses.  Table 2).

Families Genera (%)/Species (%) Genera Species (%)
The most frequent species in TDF sites were Bursera aptera (occurring in all sampled sites), Bursera submoniliformis (in 89% of the sites), Ceiba parvifolia, Pachycereus weberi, and Randia thurberi (in 78% of the sites). In contrast, 27% of species were recorded in one single site, most of them (58%; 15 species) occurred only in the TDF. In the Apancles, the most frequent species were Prosopis laevigata and Quadrella incana (occurring in 89% of the sites), as well as Escontria chiotilla, Mimosa luisana, and Pachycereus weberi (occurring in 78% of the sites). These species have multiple uses, mainly providing shade, firewood, and wood for tools and fences, as well as edible fruits, mainly cacti, which is highly valued in the region. In the Apancles, 46% of the species were recorded in one single site, 50% of them (23 species) only found in these systems.

Endemism
We recorded a high degree of endemism in the study area: 58 species (44% of all species recorded) are distributed only in Mexico. Seven species (12%) are restricted to the states of Oaxaca and Puebla, eight species (14%) are only found in the Tehuacán-Cuicatlán Valley, and one species (Agave quiotepecensis) is micro-endemic of the Cañada Oaxaqueña region, in slopes of mountains neighbouring the Sabino and Grande rivers [63]. Also relevant is the presence of Escontria, a monotypic genus endemic of Mexico, in the states of Guerrero, Michoacán, Oaxaca, and Puebla [64,65], as well as Hechtia, which is a genus quasi-endemic to Mexico, with 96% of its species restricted to the national territory [66]. TDF contained 37% (50 species) endemic species, while Apancles had 30% (39 species) of plant species endemic to Mexico (Figure 8, Supplementary File S1).
Some of the species recorded that are restricted to Mexico are within some risk category according to the Red List of Threatened Species of the International Union for Conservation of Nature (IUCN) [67] and Mexican laws (NOM-059-SEMARNAT-2010) [68] (Supplementary File S1). In addition, all species of the Cactaceae family are listed in Appendix II of the Convention on International Trade in Endangered Species of Wild Fauna and Flora (CITES) [69], which includes species that are not necessarily endangered, but whose trade must be controlled to avoid incompatible use with the survival of species.

Endemism
We recorded a high degree of endemism in the study area: 58 species (44% of all species recorded) are distributed only in Mexico. Seven species (12%) are restricted to the states of Oaxaca and Puebla, eight species (14%) are only found in the Tehuacán-Cuicatlán Valley, and one species (Agave quiotepecensis) is micro-endemic of the Cañada Oaxaqueña region, in slopes of mountains neighbouring the Sabino and Grande rivers [63]. Also relevant is the presence of Escontria, a monotypic genus endemic of Mexico, in the states of Guerrero, Michoacán, Oaxaca, and Puebla [64,65], as well as Hechtia, which is a genus quasi-endemic to Mexico, with 96% of its species restricted to the national territory [66]. TDF contained 37% (50 species) endemic species, while Apancles had 30% (39 species) of plant species endemic to Mexico (Figure 8, Supplementary File S1).   [69], which includes species that are not necessarily endangered, but whose trade must be controlled to avoid incompatible use with the survival of species.

Reasons for Conserving the Native Vegetation
Among the native plant species, we recorded 96 (73% of the total) with at least one local use. Many species had more than one local use (i.e., edible, medicine, fodder, wood). We documented that the main motives for maintaining (through tolerance, protection, and promotion) components of TDF within agricultural plots were different benefits that can be classified as contributions: (i) material, (ii) nonmaterial, and (iii) regulating.
The material contributions included plants providing edible roots, stems, flowers, or fruits (specially Cactaceae species), some used for preparing beverages, establishing live fences, or medicines, or used as firewood, fodder, resins, poisons, and wood. Nonmaterial contributions included ornamental plants that form part of ceremonies and rituals, do not cause damage, and have a "right to live", as well as those that "cause well-being". The regulating contributions included plants that provide shade, "attract rain", maintain water, regulate the climate, are the habitat of other useful species, increase soil fertility, control pests, and protect soil against erosion ( Figure 9, Table 4, Supplementary File S1).

Reasons for Conserving the Native Vegetation
Among the native plant species, we recorded 96 (73% of the total) with at least one local use. Many species had more than one local use (i.e., edible, medicine, fodder, wood). We documented that the main motives for maintaining (through tolerance, protection, and promotion) components of TDF within agricultural plots were different benefits that can be classified as contributions: (i) material, (ii) nonmaterial, and (iii) regulating.
The material contributions included plants providing edible roots, stems, flowers, or fruits (specially Cactaceae species), some used for preparing beverages, establishing live fences, or medicines, or used as firewood, fodder, resins, poisons, and wood. Nonmaterial contributions included ornamental plants that form part of ceremonies and rituals, do not cause damage, and have a "right to live", as well as those that "cause well-being". The regulating contributions included plants that provide shade, "attract rain", maintain water, regulate the climate, are the habitat of other useful species, increase soil fertility, control pests, and protect soil against erosion ( Figure 9, Table 4, Supplementary File S1).

Capacity for Biodiversity Conservation
The findings of this study show that the TAFS analysed can conserve an important proportion of the plant species richness (68%) and abundance of individuals (53%) native to the TDF. At the same time, the Apancle systems contribute to satisfying basic human needs. These systems are sources of food, firewood, medicine, materials for construction, shade, soil fertility, hydric regulation, fodder, and inputs for ornamental and ritual uses, among others ( Figure 9).
Analysis of beta diversity showed that the high dissimilarity between sites, which can be explained by the species turnover and that we found in the Apancles (74%) and TDF (72%), has profound implications for conservation, suggesting that it is necessary to maintain several sites in order to conserve the regional diversity of natural ecosystems.  Table 4. Contributions of vegetation to the satisfaction of human needs in the Cañada Oaxaqueña region. Species mentioned in workshops and interviews are included. The contributions are ordered from the highest to lowest number of species recorded in each category.

Capacity for Biodiversity Conservation
The findings of this study show that the TAFS analysed can conserve an important proportion of the plant species richness (68%) and abundance of individuals (53%) native to the TDF. At the same time, the Apancle systems contribute to satisfying basic human needs. These systems are sources of food, firewood, medicine, materials for construction, shade, soil fertility, hydric regulation, fodder, and inputs for ornamental and ritual uses, among others ( Figure 9).
Analysis of beta diversity showed that the high dissimilarity between sites, which can be explained by the species turnover and that we found in the Apancles (74%) and TDF (72%), has profound implications for conservation, suggesting that it is necessary to maintain several sites in order to conserve the regional diversity of natural ecosystems.
Considering the effective numbers of species estimated for the alpha and gamma diversity of order 1 (one of the better parameters, since all species are included with a weight proportional to their abundance in the community), the capacities of plant diversity conservation of the Apancles studied are 81% and 96%, respectively. These figures suggest that Apancles can conserve the greater part of the diversity of perennial plant species of the neighbouring TDF. However, despite such remarkable conservation capacity, Apancles maintain only one half of the abundance of individuals (53%; Figure 5) and are "samples of diversity", where half of the representation of individuals has been reduced, and, thereby, part of the ecological interactions and ecosystem functions. Therefore, it is necessary to go beyond the estimation of diversity and to evaluate these effects. It has been documented that although TAFS may be similar to native vegetation in terms of species richness, the floristic composition is not consistent and may represent an excess of pioneer species that spread and establish in the disturbed areas easily [70,71]. It is important to have in mind that pioneers can have a role in creating an ecological succession in highly degraded land.
Based on this aspect, to evaluate the capacity for biodiversity conservation of the TAFS, we considered it pertinent to study which conditions best maintain the diversity inside these systems and to include more detailed approaches to robustly characterize their configuration. In this study, for instance, we documented higher dominance and lower equitability in Apancles than in TDF, even when we recorded that total species richness and diversity were similar in both systems (Table 2, Figure 6).
An outstanding aspect of the Apancles is their capacity for harbouring endemic species, since nearly 30% of species whose distribution is restricted to Mexico are maintained there. This pattern is consistent with the high degree of endemism documented in the Mexican TDF, which are the main reservoir of endemism in the neotropics with nearly 60-73% endemic species [25,36]. Hence, these are considered a priority for conservation on a global scale [31,37].

Floristic Composition
The most diverse families and genera recorded in the Cañada Oaxaqueña region (Table 3) were also reported by several studies in the neotropical TDF [72][73][74][75][76]. The review by Rzedowski and Calderón de Rzedowski [77] reported the predominance of the family Fabaceae in the Mexican TDF, as confirmed in this study. Also outstanding in the study area were the families Cactaceae, Euphorbiaceae, and Burseraceae, which is consistent with previous studies [73,78].
The genus with a higher number of species in the sampling sites was Bursera, represented by trees producing aromatic resins that are locally used in ceremonies and rituals. Some of these species have the capacity for rooting and sprouting of branches, and hence are frequently used for live fences in the Apancles. Such high diversity of Bursera has prompted some authors to characterize a type of TDF as "Cuajiotal", where photosynthetic stemless trees of this genus are dominant [79]. Oaxaca is one of the states with more species of Bursera (37), only surpassed by Guerrero (47 species) [80]. Other genera well represented in this study were Croton, Euphorbia, Lysiloma, Mimosa, Randia, and Vachellia, which have a wide distribution in TDF of the neotropics [78]. All these genera were mentioned by Rzedowski and Calderón de Rzedowski [77] among those that contain the greatest number of species and that live preferentially or exclusively in the TDF.
We found a marked differentiation in the species more abundantly recorded in the TAFS and TDF (Figure 7). In the Apancles, the most abundant were species of multi-purpose plants like Stenocereus stellatus producing edible fruit and live fences (which are relatively easy to propagate from branches, and have rapid growth and production), Prosopis laevigata and Vachellia campechiana providing wood, shade, and firewood, Quadrella incana providing fencing stakes and tool wood, and Lippia graveolens with medicinal and condimental uses. Plants more abundant in TDF (Croton alamosanus, Aeschynomene compacta, Mammillaria carnea, Echinopterys eglandulosa, and Bursera aptera) were little or not used. Peasant management increases the abundance of useful plants inside agricultural fields, as documented in several studies [5,15,40,81], through the implementation of different types of agroforestry practices.

Types of Agroforestry Practices Implemented in the Apancles
Peasants of the Cañada Oaxaqueña maintain native varieties of maize, beans, and squashes, some "improved" commercial varieties or hybrid cultivars, and exotic fruit trees coexisting with components of neighbouring TDF through traditional and modern practices, similarly described by Durand [82] for the rural areas of Mexico. This real situation in the Tehuacán-Cuicatlán Valley, as in other regions of Mexico, should discourage oversimplified conclusions about the supposition of functional relationships between knowledge and management, and that traditional societies are in all cases ecologically sustainable [82].
Some types of agroforestry practices are more passive than others, and in some cases it is questionable if they are genuine agroforestry practices; for instance, when a remnant of native vegetation or a vegetation island may be maintained in an agricultural field simply because people do not have machines or sufficient labour to remove the vegetation. In contrast, live fences, one of the most frequent practices and with high capacity for biodiversity conservation [83], actively contribute to maintaining and recovering elements of TDF because they do not interfere with agricultural practices, and are areas continually enriched as new components replace others [5,14,15,46]. These facts make it necessary to evaluate more specifically which type of agroforestry practices are more efficient for conserving biodiversity and contributing to satisfy people's needs. However, we could see that peasants actively design their plots, where they make management decisions planned and with clearly defined purposes.
In this study, we identified a type of agroforestry practice scarcely described before, i.e., forest cover patches, which are composed of native and exotic species that constitute orchards with high economic importance in the region. This practice contributes to satisfying human needs more than conserving native biodiversity, but through these patches people obtain monetary incomes and maintain other socio-ecological functions.
The forest cover maintained in the Apancles (45.98%) is also remarkable, as it is much higher than that recorded in other ecosystems in the region (on average 25%) [84,85]. We found that the communities with higher vegetation cover (Quiotepec and Dominguillo) conserve higher richness and diversity of TDF species, whereas in Cuicatlán, where intensive agriculture predominates, forest cover, species richness, and diversity are all low in Apancles (Figure 4). In consequence, increasing forest cover in TAFS would increase their capacity for biodiversity conservation.

Implications of the Apancles for Conservation of Biocultural Diversity
Our study reveals that Apancles are important for biodiversity conservation and satisfying human needs. The main reasons motivating the maintenance of native vegetation inside Apancles are the material contributions (as edibles plants, firewood, formation of live fences, medicines, construction wood, fodder, and tool wood), then the regulating contributions (mainly shade, soil fertility, and protection against erosion), and with a relatively lower weight, the nonmaterial contributions (as plants for ceremonies, rituals, and ornaments). These results are similar to those reported by other studies in the region [15,18,46].
Currently, the need to establish horizontal communication between scientists and people with traditional ecological knowledge is recognized in order to effectively attend to the complex multidimensional environmental and social challenges affecting local people [86][87][88][89][90]. Science possesses a dynamic and effective agenda for producing new knowledge and local people have gathered knowledge and tested experience for millennia [9,91,92]. For this reason, the challenges would be more effectively addressed by the combination of views, methods, and perspectives of both approaches.
It is also necessary to integrate some antagonist perspectives in science about the trade-offs associated with biodiversity conservation and food security of the human population [93,94]. The debate has been around the strategy of land-sparing, which proposes intensifying industrialized agricultural production and locating areas for conservation in different places [95], whereas land-sharing argues that primary productive activities can be compatible with biodiversity conservation [1,96,97], as this study showed. Recent reviews of the topic [98] suggest that none of the strategies are sufficient for finding one single solution to find a balance between producing and conserving, because of the high complexity of socio-ecological systems and their multiple contexts. Instead, they appear more effective in constructing local management alternatives, considering the specific contexts, the local needs, motivations, knowledge, techniques, and customs. In all cases it is important to have in mind that the main aims are that the management of productive systems should be effective and rentable, without risk to human well-being and biodiversity conservation [98]. We consider that the complementarity of both strategies and other multiple options should be analysed contexts, considering local peoples' views and those of other sectors interacting in local contexts.
Summarizing, it is relevant that many peasants maintain strategies to take advantage of the components of nature that allow them to ensure their permanence [1][2][3][4]. At the same time, the new challenges of a rapidly changing world would be more effectively solved through dialogs between local people, scientists, and other actors (nongovernmental organizations and government, among others), offering accompaniment from a committed science. TAFS, with their advantages and limitations, offer viable opportunities to find solutions to the purposes of satisfying human needs and biodiversity conservation.

Strategies and Perspectives for Public Policies
Based on the consideration of factors that put the TAFS identified by Moreno-Calles et al. [99] at risk, we have delineated some proposals for strengthening these systems, which, if implemented could support participatory processes with local people: (i) to promote increasing the forest components inside agricultural fields; it would be desirable to increase the forest cover with multipurpose species native to the region, which would favour regional biological conservation and provide benefits for the peasants; (ii) to value and rescue local views, knowledge, and techniques sustaining biocultural diversity; (iii) to enhance programs directed toward favouring the existence and improvement of TAFS from the perspectives of academia, governments, and civil society organizations, among others; (iv) to encourage the involvement of young people in agroforestry management; and (v) to communicate alternative strategies of conservation promoted by the Mexican government, such as payment for environmental services, promoting areas for voluntary conservation, and unities for the conservation of wildlife. However, to include TAFS in public policies at the regional scale, it is necessary to enhance a strategy of communication among different sectors of the area, particularly with those decision-makers and political actors involved in environmental legislation.
Similarly to other protected areas of Mexico with great extents of TDF (i.e., Sierra de Huautla Biosphere Reserve, in the state of Morelos, or Chamela-Cuixmala Biosphere Reserve, in Jalisco), it is also crucial to encourage spaces for social participation where local people make public their opinions and perspectives, and to have access to decision-making and mechanisms for compensating the cost of conservation [100].

Conclusions
Biodiversity loss has been shown to threaten the maintenance of human well-being [101]. Moreover, the maintenance of socio-ecological systems is highly dependent on biodiversity [102]. In this context, although conservation supposes contraposition with the satisfaction of social needs, it is possible and necessary to reconstruct and enhance systems where the production and maintenance of ecosystem integrity are possible [100]. TAFS represent an outstanding opportunity to maintain socio-ecological systems for the long term but they need improvements, which scientific and participatory research may identify, to increase their capacities for both purposes.
We recognize that TAFS associated with TDF in the study area harbour an important richness, diversity, and endemism of plant components of natural ecosystems. The Apancles are an expression of traditional and contemporaneous management with features of sustainability and respectful coexistence between societies and ecosystems, so they may be viable options for constructing sustainable agricultural systems. Therefore, these systems should be studied more deeply, revalued in terms of their role in caring for nature, improved in their capacities for conserving and producing, and explored as regional strategies for biodiversity and biocultural diversity conservation at local, regional, national, and global scales. Funding: The first author acknowledges CONACYT for their support with a Ph.D. scholarship, the PAEP program for the economic support received through the IIES-UNAM. To the DGAPA-UNAM for supporting the PAPIIT IN206217 and IN206520 projects "Domesticación y manejo in situ de recursos genéticos en el Nuevo Mundo: Mesoamérica, la región andina, amazónica y el nordeste de Brasil" and "Agricultura y Agroforestería social y familiar en contextos de cambios locales y globales", respectively. This study is part of the thematic network of Agroforestry Systems of Mexico. In addition, the authors thank financial support from CONACYT, research project A1-S-14306.