Study area

The study was carried out in the Huehuetán municipality of the state of Chiapas, Mexico, which is located in the Soconusco region at 15°30'35" N and 92°24'27'' W at 35 masl. The climate is Am(w’’)ig, which corresponds to warm sub-humid with summer rains. The average annual rainfall is 2 800 mm and the average temperature is 28.5 °C. (^{Garcia, 1973}). The soils belong to the two main groups, Acrisol and Luvisol (^{INEGI, 2005}).

Sampling sites

10 random sampling units (UM) were established above the intermediate part of the Huehuetán municipality (from 80 to 540 masl), where coffee cultivation predominates (Figure 1). Each experimental site or sampling unit was named according to the name of the ejido or the cantón (Table 1). Within each one, an area of 1 000 m² (20 m x 50 m) was set aside based on what was recommended by ^{Cox (1981)} and ^{Somarriba (1999)} for tree communities in agroforestry systems.

Figure 1 Location of the sampling units. 

Tree struture analysis

The tree species found in the sampling units were taxonomically identified, with common, scientific and family names, as described by ^{Pennington and Sarukhán (2005)}.

Vertical stratification was defined by measuring the height of all the trees present in the sites, from the base of the stem to the apex. For this purpose, a Haga type altimeter was used (GmbH & Co D-90429, Germany).

The Importance Value Index (IVI), according to ^{Curtis and McIntosh (1951)}, refers to the importance of one or several species in terms of the physical structure of a community or its species composition (^{Zarco-Espinosa et al., 2010}) which reflects the percentage that each species has in the community.

The dominance, density and frequency values, as well as the value of importance per species, were obtained according to ^{Stiling (1999)}.

The basimetric area of the trees (AB) was obtained by the following formula:

Where:

n = Number of trees

DAP = Diameter at breast height (DBH)

Density is defined as the number of individuals of a species that occupy a given area and is calculated as follows:

In order to know how the species are distributed, it is necessary to calculate the frequency, by counting the sampling units where they were found:

The frequency value can be interpreted as the probability of finding a species in any randomly chosen sampling unit of the same area. The relative frequency is calculated as follows:

The Shannon-Wiener Index measures the average degree of uncertainty to predict the species to which an individual randomly selected from the sampling units belongs (^{Somarriba, 1999}).

The formula for the Shannon index is as follows:

Where:

H´ = Shannon-Wiener index

S= Number of species

Pi = Proportion of individuals of the i species

Ln = Natural logarithm

The Simpson Index measures the probability that two randomly selected individuals within a habitat belong to the same species (^{Pielou, 1969}).

The formula to calculate the Simpson index is:

Where:

S = Number of species

N = Total of present organisms (or square units)

n = Number of samples per species

Information analysis

Based on the information obtained in the field, tables were created in Microsoft Office Excel ©2010 with the number of plant species found in each sampling unit, in which the averages were calculated for analysis and discussion for each site. The Shannon-Wiener and Simpson value indices were determined using the EstimateS program (^{Colwell, 2005}), because it allows analyzing the number of plant species for a given sampling area.

The initial identification of the trees started from the common name directly in the field with the support and knowledge of the owners of the plots and the technician of the forestry consultancy. For the registration and taxonomic identification of the tree component, the work carried out by ^{Pennington and Sarukhan (2005)} and ^{Miranda (2015)} were taken as support. The taxonomic classification was made according to the APG III system.

Diversity of tree species

279 trees belonging to 23 species and 11 families of the medium evergreen forest were identified (Table 2).

The most abundant species were Tabebuia donnell smithii Rose (maderable), Inga micheliana Harms. (firewood) Cordia alliodora (Ruiz et Pavón) Oken (posts) y Cedrela odorata L. (timber). As well as fruits such as Citrus sinensis (L.) Osbeck. (Orange), Byrsonima crassifolia) (L.) Kunth (Nance) and Pouteria sapota (Jacq.) H. E .Moore & Stearn (Zapote). Cedrela odorata L, Brosimum alicastrum Sw., Bursera simaruba (L.) Sarg. and Tabebuia donnell smithii are typical of the semi-evergreen medium tropical forest (^{Villavicencio and Valdez, 2003}).

The notable presence of Inga spp. in the region has been identified in other coffee growing regions of Mexico, such as in the Sierra Norte de Puebla where ^{Martínez et al. (2007)} determined that shade in coffee plantations was composed of a single species, such as Inga spp. (chalahuites), Gliricidia sepium (Jacq.) Kunth ex Walp. (cuacuite) or Alnus acuminata Kunth (elite).

The UM with the highest number of tree species (10) was number 4 (Cantón La Estrella). After this site, UM 5 (Ejido Chamulapita), UM 6 (Cantón El Tivoli) and UM 10 (Cantón Siria) followed with 9. In UM 9 (Cantón Cairo) there were four tree species and it was the site with the lowest number of species of this type. Of the total number of species found in the UMs, the largest number of timber and fruit trees was identified in the Cantón Siria (UM 10), with a total of 56. This site is located at a lower altitude.

On the other hand, in UM 8 (Cantón El Caucho), which is also at a low altitude with respect to the other eight sites, the least amount of timber and fruit trees was counted (13). This suggests the producer's preference for some species.

The difference in the number of species in the UMs may also be related to the age of the coffee plantations. The oldest plots registered an increase in the richness of trees in uses in relation to the youngest coffee plantations. In those older than 40 years, such as in UM 4 (Cantón La Estrella), there was an increase in the diversity of tree species. In addition to age, insulation can play a role; it is a community far from rural areas where tropical forests remains prevail.

^{Salvador-Morales et al. (2019)} cite similar results, when studying agroforestry systems with Theobroma cacao L.; they found greater diversity and tree structure in T. cacao plantations over 40 years compared to younger plantations.

Of all the species found, most belong to the Fabaceae and Bignoniaceae families. These species are mainly used for firewood and timber.

Vertical stratification

The dominant height of the vegetation in the UMs was an average of 20 m, but specimens above 25 m were observed. Within the tree component, three strata were recognized. The upper stratum was considered to be trees with a height of 18 to 27 m, and in it, 19 individuals were identified, with dominance of Tabebuia donnell smithii (31.58 %). Aspidosperma megalocarpon Mull. Arg. trees represented 21.05 % and those of Inga laurina (Sw.) Willd. 15.79 % of the total number of trees.

In the middle stratum, an average height of 14 m was recorded in 167 individuals. The most common trees were Tabebuia donnell smithii (22.75 %), Inga micheliana (22.15 %), Cordia alliodora (17.36 %), Cedrela odorata (14.97 %) and Citrus sinensis (10.17 %). Finally, in the lower stratum with height less than 9 m, 93 trees were counted, 26.88 % of which were Tabebuia donnell smithii, 15.05 % to Inga micheliana, 9.67 % to Persea americana Mill and 7.52 % to Byrsonima crassifolia (L .) Kunth (Figure 2).

Figure 2 Vertical stratification of the tree component of the Coffea arabica L. agroforestry system in Soconusco, Chiapas. 

The timber or fruit trees in the coffee production system exhibited greater height. This characteristic can be related to the producer's interest in subjecting them to cultural practices of pruning and thinning, in order to favor the entry of light to the coffee plant, and consequently, reduce or limit the growth of these shade trees. In these cases, in addition, species that are not important for the producer are eliminated and a decrease in diversity can be induced.

Shannon-Wiener Index

The Shannon-Wiener index measures the average degree of uncertainty in predicting the species to which an individual taken at random from the UM belongs. The average value obtained in the coffee UMs of Soconusco was 1.2, which can be considered low in richness, when compared to those of other coffee regions in Mexico, such as the case of the Vicente Guerrero community (^{Magdaleno et al., 2005}); thus, values from 3.55 to 3.89 were calculated. Other researchers obtained values of 3.5 in the Sierra de Atoyac, Veracruz (^{García-Mayoral et al., 2015}). In the coffee region of Coatepec, ^{Gómez-Martínez et al. (2018)} registered an index of 8.3 in the traditional system; 2.94 in Tuxtla Chico, state of Chiapas and 2.71 in Tapachula, Chiapas (^{Roa-Romero et al., 2009}). ^{Zapata (2019)}, in a study carried out in agroforestry systems with coffee in three municipalities of Cundinamarca, Colombia, found that the tree species of greatest ecological importance in the AFS were Citrus sinensis, Calliandra pittieri Standl., Inga edulis Mart. and Cordia alliodora, and that it is related to the preferences of producers for these species due to favorable biophysical, environmental and/or economic interactions, by associating them with coffeé. In this study it was possible to appreciate that the producers have changed from the traditional shade species for timber tree species and fruit species and even introduced species, which at some point will bring them economic benefits.

Current results suggest increasing agronomic management in agroforestry systems associated with C. arabica through greater control of native vegetation with greater thinning to increase solar radiation for physiological purposes. Thus, it seeks to achieve increases in the yield of the coffee crop, in addition to reducing the effects of coffee rust; concomitantly, generate the possibility of increasing the density of coffee plants, and in other cases, that of some tree species of particular interest, which leads to a reduction in tree diversity.

Simpson Index

Simpson's index considers the dominance of the most representative species, that is, as the index increases, diversity decreases (^{Pielou, 1969}). The predominant species in the coffee plantations are Inga micheliana and spring Tabebuia donnell smithii. The index value obtained in this study was 2.6. These species are widely accepted by producers. Inga micheliana (Chalum) for firewood and Tabebuia donnell smithii (Primavera) for the high commercial value of its wood.