Description of the study area

The research was carried out during the first half of 2018 in three elevational zones that grouped 13 of the 15 municipalities of the Soconusco region, located in the south of the state of Chiapas, Mexico, between coordinates 15° 19' NL and 92° 44' WL covering 4 605.4 km², which represents 6.28 % of the State area.

Climates in the region are warm and semi-warm; warm sub-humid weather predominates with summer rains, followed by warm humid weather with abundant summer rains. Therefore, according to the Köppen classification modified by ^{García (1973)}, the predominant climate is type Aw₂(w)Ig with average relative humidity of 79.4 % and average annual temperature of 26.8 °C (^{Comisión Nacional del Agua [CONAGUA], 2015}). The Soconusco region and its coastal plain has eight soil types: Litosol, Acrisol, Regosol, Solonchak, Andosol, Luvisol, Nitosol and Cambisol (^{Comisión Nacional para el Conocimiento y Uso de la Biodiversidad [CONABIO], 2015}). However, in cocoa production areas, the predominant soils are generally Andosol, Cambisol and Luvisol.

Sampling

An exploratory tour was carried out and interviews were conducted with the producers in 90 % of the cocoa plantations in each municipality of the Soconusco region, to determine the sampling areas. From this tour, the elevational ranges to be studied, the criteria for the selection of the municipalities to be examined, and the productive entities (farms) to be evaluated were defined.

The selection criteria of the municipalities were chosen based on the representativeness at the State level: 1) area planted with cocoa; 2) number of producers; 3) contribution to production (%); and 4) geographical representativeness by elevational levels (from 0 to 500 m).

The tour identified that the largest area of cocoa plantations is at elevations below 150 m, which ensured representativeness by selecting the largest number of farms at low elevations; therefore, the following elevational levels were defined: 0 to 50 m, 51 to 100 m and ≥101 m, consisting of seven, three and three farms, respectively (Table 1).

Response variables

Statistical analysis

The recorded species data were systematized in a database created with the Excel version 9.1 package, where the frequency and average of species were calculated. With the information originating from the number of trees per species and the abundance to determine the diversity indices, an analysis was made with Fisher’s least significant differences (LSD) test at P ≤ 0.05. The means were compared by multiple range contrast with the statistical package STATGRAPHICS plus version 5.1 (^{Batanero & Díaz, 2008}).

Tree composition

Table 2 shows the 35 tree species selected by producers for shade in cocoa agroforestry plantations in Soconusco, Chiapas. The species were grouped into 32 genera and 22 families. The most represented genera were Citrus (8.5 %) and Tabebuia (5.71 %), which together accounted for 25.12 % of the trees (199 in total). The most represented families were Fabaceae, Moraceae and Rutaceae with four, three and three species, respectively, followed by Apocynaceae, Bignoniaceae, Euphorbiaceae, Malvaceae, Meliaceae and Sapotaceae with two species each. These families grouped 54.1 % of the inventoried species.

A richness of 35 species was determined at all three elevations. The greatest richness was found at the elevation of ≥101 m with a total record of 21 species which meant, on average, seven species per plot, represented by the families Boraginaceae and Apocynaceae, followed by the elevation between 0 and 50 m with 26 species (on average 3.72 species per plot) represented mainly by the families Boraginaceae, Sapotaceae and Meliaceae; the lowest richness was recorded at elevations between 51 and 100 m with six species in total, averaging two species per plot represented by four families, among them Boraginaceae.

Regarding species abundance, a total of 199 trees were determined in the three elevations combined; 101 trees were recorded within the elevational range from 0 to 50 m, representing 14.42 trees per plot; 47 trees were recorded at the elevational level between 51 and 100 m, representing 15.66 trees per plot, and 51 trees were recorded at heights greater than 101 m, averaging 17 trees per plot. According to Figure 1, the laurel (Cordia alliodora [Ruiz & Pav.] Oken), pink poui (Tabebuia rosea [Bertol.] DC.) and mamey sapote (Pouteria sapota [Jacq.] H. E. Moore & Stearn.) were the most abundant species with 29, 26 and 23 trees, respectively. Of the recorded species, 13 had low abundance with only one tree; of these, eight were found mostly located at the lowest elevational level (0 to 50 m). However, the most abundant species were recorded at all three elevations.

Figure 1 Distribution pattern of species abundance in cocoa agroforestry systems in Soconusco, Chiapas.  

Regarding the IVI, the families Boraginaceae, Fabaceae, Sapotaceae and Meliaceae were among the five most important, which were represented by species located in at least two of the heights studied. The family Boraginaceae was represented by high IVI species at the three elevational levels, which suggests that they are successful taxa with species capable of developing in different environments and heights (Figure 2). In contrast, the family Fabaceae was not represented by any higher IVI species, but it did have species with a lower IVI, such as the zope (Schizolobium parahyba [Vell.] S. F. Blake) and chalum (Inga micheliana Harms), at two elevational levels.

Figure 2 Most important families and species at three elevational levels in Soconusco, Chiapas. IVI: importance value index. 

Unlike the analysis carried out in the families, no important species were recorded developing at the same time at all three height levels, only in two of them (Figure 2). However, some species were found at all three heights, but with lesser or greater importance in one or the other; such is the case of laurel, primavera, mamey sapote, Spanish cedar (Cedrela odorata L.) and avocado (Persea americana Mill.).

Diversity

Table 3 shows that the diversity of species and families, estimated by means of the Shannon index, was significantly (P ≤ 0.05) different among elevational levels; the highest diversity values were found at heights of 51 to 100 m and greater than 101 m. On the other hand, in the diversity estimated through the Simpson index, no significant (P > 0.05) differences were found among the three elevations.

Similarity

According to Table 4, the Jaccard and Sørensen qualitative indices indicate that the greatest similarity was found between tree compositions from 0 to 50 m and ≥101 m; of the total number of species recorded, 12 were common in the three elevational ranges. However, according to Morisita-Horn's quantitative method, the similarity between heights was greater in AFSs located at elevational levels greater than 51 m.

Stored carbon

The land used by the cocoa AFSs, in the three elevational ranges studied in Soconusco, stores high levels of carbon with species that accumulate between 3.11 Mg·ha^-1 and 205 Mg·ha^-1. Based on the information in Table 5, species growing at low and high elevations stored 39.5 % and 36 % of the total, respectively, while at medium elevation (51-100 m) only 24.5 % is stored.

Tree composition

The greatest abundance per plot was found in the elevational range ≥101 m, mainly due to the number of species recorded in the municipalities of Tuxtla Chico and Acapetagua, with dominance of trumpet tree (Cecropia obtusifolia Bertol.), chapón (Stemmadenia donnell-smithii [Rose] Woodson) and laurel. The tree composition of the three elevational ranges presented typical species of the agroecosystems where cocoa is cultivated, the record of which was similar to that reported in this same region by ^{Salgado et al. (2007)} and ^{Roa-Romero et al. (2009)}.

The family Fabaceae, as one of the most represented, grouped species of the cocoa AFSs, as in the case of chalum (I. micheliana) in the municipalities of Tuxtla Chico, Acacoyagua and Suchiate. This species is commonly found as a shade tree in perennial crops, due to the architecture of its umbrella-shaped crown that allows the homogeneous entrance of light in the plantations. As a service tree it provides nitrogen due to its fixing capacity, and fertility through the pruning residues that are used in the form of dead cover (^{Barrance et al., 2013}).

When examining the regularity of species abundance values when placed in decreasing order, it was observed that the distribution pattern corresponded to the mathematical models explained by ^{Magurran (1989)}, ^{Margalef (1995)} and ^{Krebs (1999)}, whose curve allowed interpreting the abundance information. This pattern began with a progression similar to a geometric series where few species (13) are dominant and practically rare; that is, a constant proportionality between abundances and species was assumed, so that the series was observed as a straight line in a logarithmic scale. Subsequently, an adjustment to a series with normal logarithmic distribution was observed, where there was a small number of abundant species and a large proportion of species with low abundance, which determined that the curve that it describes has the shape of an inverted jota. This distribution expresses with greater clarity the relationship of trees by species and where species with intermediate abundance were recorded, with respect to the total, and which became more frequent. Finally, the distribution pattern curve showed the conditions of the broken stick model in which species are equally abundant and can be organized into abundance classes.

Six of the 10 species with the highest IVI, at one or more of the elevational levels studied in Soconusco, were found at the elevational level of 51 to 100 m; this is due to the abundance, frequency and dominance of their basal diameter. Mamey sapote was among the species with the highest importance level at the three elevational levels, coinciding with the records of ^{Roa-Romero et al. (2009)}. This is due to the fact that this species is a fruit tree of great acceptance for its contributions to the economic income of the producer. Commonly, the tree is part of the tree structure of coffee and cocoa plantations, established or naturally regenerated with or without agroforestry design. The location of the mamey sapote is random, as recorded in the plots of Tapachula, Huehuetán and Acapetagua. On the other hand, the mango (Mangifera indica L.), with greater presence in the height from 0 to 50 m, was one of the species with a lower importance value among the 10 most important species; however, this fruit tree is also of great importance to the producers because of the income it provides in light of the main crop’s low prices. Mango was found in cocoa plantations in four municipalities, a situation similar to that recorded by ^{Ramírez et al. (2013)} in a 30-year-old cocoa AFS in Tabasco.

Among the timber species that were commonly recorded, at different heights and with a high importance index, is the laurel. Its popularity lies in the value of its wood for local use; it can be combined with annual and perennial crops in an AFS, and it is also used as a medicinal plant (^{Barrance et al., 2003}). The presence of laurel in cocoa AFSs provides an alternative source of income for the producer. ^{Sánchez, Pérez-Flores, Obrador, Sol, and Ruiz-Rosado (2016)} also identified this and a large number of other timber species that can potentially be sustainable in cocoa AFSs.

Diversity and similarity

It was assumed that there is low diversity of species and families in the Soconusco region, considering that, in general, Shannon index values range from 1.5 to 3.5 (^{Zak & Willig, 2004}) and that values greater than 3 mean high diversity. The lowest values were calculated at levels between 51 and 100 m, being able to influence the Tapachula municipality plot, whose diversity was null, since it had only one species (mamey sapote) with eight trees. However, according to the calculated Simpson index and considering that the range of values goes from 0 (low diversity) to 1 (high diversity), the cocoa AFSs presented medium diversity with a tendency to high diversity, since the values are higher than 0.50; the highest indices were found at heights greater than 100 m, which is related to their greater richness of species. In this context, it can be deduced that diversity is influenced by a greater height range (>100 m), since the sampled area was the same as in the elevational range from 51 to 100 m, where there was less species diversity and richness. These results may be due to the fact that at heights greater than 100 m there are still many remains of the original forest, whose species diversity has been maintained over time despite the application of shade management techniques for the development of the crop. It also influenced the existence of plantations in municipalities that have greater equity (relationship between richness and relative abundance of each species) and diversity, such as Tuxtla Chico. In this municipality, exploratory tours showed that producers have maintained an adequate diversity of species as shade, with the aim of obtaining additional income, and that they are part of traditional management; these results are similar to those obtained by ^{Roa-Romero et al. (2009)}. The Simpson index is a measure of the probability that two trees taken randomly belong to the same species and is strongly related to the most abundant species in the sample, which are less sensitive to species richness. Therefore, the closer the index is to unity, the greater the probability of dominance of a species (^{Polo, 2008}).

In the Soconusco region, the low diversity in cocoa AFSs at heights between 51 and 100 m is due to a lower equity between abundance and richness, where an individual-species ratio of 5.7 was calculated; this is not the case at heights greater than 100 m, where a tree composition equivalent to 1.52 trees per species was recorded, followed by the composition recorded at the lowest height studied (0 to 50 m), which was 2.51. Therefore, there is more equity in the cocoa AFSs that are located at heights greater than 100 m, because the ratio between richness and relative abundance of each species is greater.

With respect to the importance of species diversity as shade, the results indicated high variability of illumination in cocoa plantations at all elevational levels, given primarily by the number and type of species per plot, the canopy size of each species and poor management of tree diversity. According to ^{Silva et al. (2013)}, if this diversity is properly managed, important benefits for the crop can be achieved.

On the other hand, from the qualitative point of view, the similar diversity in tree composition between the lowest and highest height studied is due to the presence of a large number of shared species and to greater diversity. However, from the quantitative point of view, the results of greater similarity between the highest heights may be related to a smaller difference between the number of recorded trees, in addition to greater similarity of the edaphoclimatic conditions with rainfall between 1 600 and 4 000 mm·year^-1, an average monthly temperature between 22 and 24 °C and Andosol and Acrisol soil types.

The diversity and similarity results showed that the agroecosystems where cocoa AFSs are developed show signs of low levels of disturbance, despite anthropogenic intervention in these natural environments; the abundance, richness and diversity of species at the three height levels have directly influenced the benefits they bring to producers, regardless of the income earned from the main crop (cocoa). In this sense, producers in the Soconusco region have a diversity of species for timber, energy, medicinal and food uses that make possible the existence of cocoa plantations with economic, ecological and social potential. These similarity measures have already been reported in other cocoa agroforestry environments by ^{Hervé and Vidal (2008)} and ^{Zapfack et al. (2002)} in Cameroon, who indicate that the type of management is a factor that determines a lower percentage of similarity, a situation inherent to each cocoa region.

According to the results, the establishment of fruit and timber species, based on the interests of the producer for the economic benefits they provide, influenced the tree composition, diversity and similarity of cocoa agroecosystems by elevational levels. This was related to the region’s traditional culture and the intentional selection of producers in search of greater services in conjunction with the main crop. On the other hand, it also influenced the composition of the natural forest since many species originating from the primary forest have developed with a certain dominance and acceptance on the part of the producers. Also, as a result of the climate and soil variation with respect to height, the diversity of the vegetation has changed. In ascending order, we observed medium-height tropical forest, high-height tropical forest, and oak, mountain cloud and pine forests, which are natural vegetation of the Sierra Madre.

Carbon storage

A direct relationship was determined between aerial biomass carbon storage and species abundance. The greatest amount of carbon stored and estimated in the biomass of the species was found at the lowest elevational level (0 to 50 m), due to the greater number of recorded trees per total area, including very abundant species such as pink poui, mamey sapote and Spanish cedar. At the same elevational level (0 to 50 m), dominant species with total DBH of 473 m²·ha^-1 were recorded, while at an elevation of 51 to 100 m the lowest abundance and species with lower DBH were recorded, which may have influenced the results. ^{Cerda, Espin, and Cifuentes (2013)} determined that carbon storage capacity correlates with the basal area of timber and fruit trees. This means that tree thickness and size is as important as tree abundance and species richness.

In the present research, carbon storage was greater than that determined by ^{Segura and Andrade (2008)} in a cocoa AFS in Talamanca, Costa Rica, where they recorded 122 ± 24 Mg C·ha^-1; the difference may be due to the tree composition of each territory. In this context, ^{Somarriba et al. (2008)} stated that an average total carbon level between 80 and 120 Mg·ha^-1 would be the most appropriate in order not to harm cocoa production; quantities above this value are considered high levels. ^{Ortiz, Riascos, and Somarriba (2008)} determined that cocoa-laurel systems fixed between 43 and 62 Mg C·ha^-1 in 25 years and that in one year they fixed between 1.7 and 2.5 Mg C·ha^-1, which corresponds to low storage levels. These same authors refer to research carried out by CATIE (Tropical Agricultural Research and Teaching Center) in Talamanca in 2005, where they report carbon accumulation rates in cocoa plantations, over a 25-year period, of between 2.1 and 2.8 Mg C·ha^-1·year^-1, whose densities ranged between 100 and 150 trees·ha^-1. In the present research, the high volumes of carbon accumulated in the biomass of the trees correspond to densities between 127 and 163 trees·ha^-1.

The capacity of agroforestry ecosystems to store carbon in the form of aerial biomass varies according to the age of the trees, and the diameter and height of the shade components. ^{Andrade and Ibrahim (2003)} state that AFSs can fix and store between 12 and 228 Mg·ha^-1, including soil organic carbon, which represents between 20 and 46 % of the carbon sequestered in primary forests.

From the knowledge of tree composition and species diversity by elevational levels in the cocoa AFSs of Soconusco, producers and decision-makers will be able to evaluate shade management alternatives in plantations, with the aim of designing efficient systems in each edaphoclimatic and physiographic condition. These alternatives may consider species of importance to producers according to their traditions and socio-economic needs, based on the uses of the diverse forest potential in the region. Likewise, strategies with an ecological and agronomic approach can be adopted, with the aim of preserving the diversity of existing species and applying integrated management technology that benefits both cocoa cultivation and forest species, in order to improve the sustainability of these agroecosystems.