Study area

The study was conducted in cocoa AFS in different localities of the municipality of Cárdenas, Tabasco. Cárdenas is located in the Chontalpa region (17º 59’ N; 91º 32’ W) at an altitude of 2 to 17 masl. Climate is hot-humid with mean annual precipitation of 2 643 mm and a monthly mean of 355 mm; mean annual temperature is 26 ºC with a maximum of 45 ºC (^{CNA-SMN-SCDI, 2012}).

Sampling sites

Twenty experimental plots 50 x 100 m were set up on Eutric Fluvisols (FLeu) and Eutri-gleyic Fluvisols (FLeugl), which are the representative soils of the study area and of the cocoa-AFS (^{Palma et al., 2007}). In addition to the field observations, authorities of each locality were interviewed to contact cooperating producers and to select the cocoa AFS to sample.

Tree flora composition and dasometric variables

The age and area of the cocoa AFS were recorded within each sampling site. The trees were identified taxonomically and georeferenced with a GPS (Garmin model GSmap60csx^®) and assigned a number that was painted on the tree with spray paint (^{Zarco et al., 2010}). The dasometric variables recorded were: diameter at breast height (DBH_{1.3 m}, cm) measured with a diametric tape; total height (Ht, m) and commercial bole (Hc) measured with a Haga^® Pistol; basal area (BA, m²) estimated with the equation BA= 0.7854 x DBH² in which, 0.7854= constant; total and commercial volume (TV, CV, m³) estimated with the equation V= BA x ff x H, where: ff= form factor (0.70) and H= total or commercial height (^{Dauber, 1997}; ^{FAO, 2004}).

Classification of canopy based on height and DBH and potential use of trees

Tree canopy was classified following the criteria proposed by ^{Pinelo (2001)}: very low stratum (<5 m high), low (≥5 - <15 m), medium (≥15 - <25 m), and high stratum (≥25 m high). Information on DBH of trees was organized by diametric category at intervals of 10 cm (1 to 10, 10 to 20...) to determine the frequency of each class with respect to the total classes (^{Orozco and Somarriba, 2005}; ^{Zarco et al., 2010}). The potential products to be obtained from individual trees were defined in function of DBH: without use (DBH < 5 cm), firewood (≥5 - <10 cm), posts (≥10 - <15 cm), narrow boards (≥15 - <30 cm), thick boards (DBH ≥30 cm) (^{Mora and Hernández, 2007}).

Tree flora composition

In the 10 ha sampled it was found 2 856 forest trees, 67 species belonging to 58 genera and 28 families. ^{Ramírez-Meneses et al. (2013)}, sampled 6 ha in Tabasco Mexico and found 38 species, 35 genera and 24 families. ^{García (1983)}, also in Tabasco, recorded 40 species of 19 families in a survey of 72 producers, while ^{Salgado et al. (2007)} in the Soconusco region in Chiapas Mexico recorded 790 trees belonging to 23 families, 38 genera and 47 species in 7.2 ha. In our study a broader diversity of species and families were recorded than that reported by ^{Garcia (1983)} y ^{Salgado et al. (2007)}. In contrast, ^{Rosa (2003)}, in Brazil, recorded a smaller number of trees (2 514) but a larger number of species and families (293 and 52 respectively), probably because the cocoa AFS were found in cleared rainforests areas or because the sampling area was larger (15 ha). Moreover, in Nigeria ^{Do and Odebiyi (2007)} reported 487 trees belonging to 45 species and 24 families in 1.3 ha.

The average number of species per hectare was 14, ranging from 6 to 35. The most common species were E. americana and C. odorata. ^{Orozco and Somarriba (2005)}, in Bolivia, reported Swietenia macrophylla, Schyzolobium parahyba and Amburana cearensis. as more frequent species per hectare.

In this study, mean density was 286 trees ha^-1, varying from 96 to 618 trees ha^-1. The families Fabaceae and Meliaceae were more predominant. In Panama and Costa Rica, ^{Somarriba et al. (1996}; ²⁰⁰⁰⁾ recorded 278 trees ha^-1; ^{Mogollon et al. (1996)} in Venezuela reported 300 trees ha^-1, while ^{Rosa (2003)} in Brazil reported from 47 to 355 trees ha^-1. The four studies mentioned coincide with ours in tree density per hectare, and also in that the Fabaceae family was the most frequently used for shading the cocoa crop.

Frequency of tree species by plantation age

The age of cocoa AFS sampled varied from 6 to 35 years-old Table 1). In this entire range of ages, Mote (E. americana), 28.4%, Spanish cedar (C. odorata), 20.1%, and Cocoite (G. sepium) 8.7% were outstanding. This agree with ^{Cordova et al. (2001)} who reported 25% E. americana in Cárdenas, Tabasco, but contrasts with this same study for G. sepium (75% of the shade trees). ^{Rosa (2003)} in Brazil indicated that the most frequent species used to shade cocoa were Schefflera morototoni (Aubl.) Maguire, and Artocarpus heterophyllus Lam., with 8 and 7%, respectively. In Nigeria 10 species accounted for 76% of the trees recorded: Elaeis guineensis Jacq., Cola nitida (Vent.) Schott et Endl., C. sinensis, Mangifera indica, Anacardium occidentale L., Psidium guajava L., Persea americana Mill., Ricinodendron heudelotii Muell. Arg., Citrus reticulata L. and Cocos nucifera L. (Oke and Odebiyi, 2007). In Talamanca, Costa Rica, C. alliodora, Citrus spp., C. nucifera, Inga spp., and C. odorata were outstanding (^{Somarriba et al., 2003}), but in Bolivia the outstanding species were S. macrophylla, S. parahyba, A. cearensis, Centrolobium ochroxylum and C. odorata (^{Orozco and Somarriba, 2005}). These studies, compared with ours, recorded similar species but different density, which is reasonable since the species used for shade in the cocoa AFS vary not only among countries but also among regions of the same country.

Table 1 Frequency of species according to the number of trees recorded in cocoa AFS of different ages, Cárdenas, Tabasco, Mexico. 

In 6-, 20- and 25-year-old cocoa AFS, the most frequent tree species used to shade the cocoa AFS was C. odorata, with a frequency of 78.0, 41.9 and 15.1%, respectively. In 15- and 20-year-old plantations G. sepium was the most frequent with 35.1 and 27.1%. Erythrina americana was the most frequent in the 18-, 25- and 30-year old cocoa AFS with frequencies of 46, 45.4 and 56.6%, respectively. Tabebuia rosea was most frequent in plantations 18- and 35-years-old with 15.2 and 15.8%. Diphysa robinioides was the most frequent in 27- and 33-year-old plantations accounting for 47.8 and 38.5%. In 30-, 33- and 35-year-old plantations, C. arborescens was most frequent with 17.5, 12.8 and 16.9%, respectively (Table 1).

Species with greater basal area by age of cocoa AFS

The 20-year-old cocoa AFS had the smallest BA and the 25-year-old AFS the largest: 12.2 and 22.7 m² ha^-1, respectively. The three forest species with the largest BA in the sampled AFS (6 to 25 years old) were E. americana (6 m² ha^-1), E. poeppigiana (3.8 m² ha^-1) and C. odorata (1.6 m² ha^-1), while each of the remaining 64 species had a BA of ≤1 m² ha^-1 (Figure 1).

Figure 1 Species with the largest basal areas (m² ha^-1) in cocoa AFS of different ages in Cárdenas, Tabasco, Mexico. 

Cedrela odorata was the outstanding species in BA in both 6- and 20 year-old cocoa AFS (15.7 and 2.2 m² ha^-1, respectively). Guazuma ulmifolia was outstanding in 6- and 18-year old cocoa AFS (0.6 ad 1.9 m² ha^-1). In those cocoa AFS aged 15, 25 and 30 years, it was E. poeppigiana (7.8, 9.1 and 1.9 m² ha^-1). Gliricidia sepium had the largest BA in 15, 20 and 35-year-old AFS (4.3, 2.7 and 3.5 m² ha^-1), E. americana in AFS aged 15, 18, 25 and 30 years (3.3, 8.3, 8.2 and 14 m² ha^-1), M. indica in 18-, 30- and 33-year-old AFS (1.4, 0.9 and 2.6 m² ha^-1); S. saman in 25- and 33-year-old AFS (1.7 and 7 m² ha^-1). The 27-, 33- and 35-year-old ASF shared D. robinioides as the outstanding species in BA (3.5, 2.5 and 1 m² ha^-1) (Figure 1).

Basal area (BA) of all the registered trees was 185 m², with a mean of 18.5 m² ha^-1 and a range of 8.3 to 34.6 m² ha^-1. Ramirez-Meneses et al. (2009) in Cárdenas, Tabasco, reported a mean of 48.2 m² ha^-1, highlighting the species S. saman (12 m² ha^-1), D. robinoides (7.8 m² ha^-1) and G. ulmifolia (5.6 m² ha^-1). These values are higher than those found in our study since the species are of larger diameter and higher frequency (S. saman).^{Somarriba et al. (1996)}, in Panamá, reported a mean of 11 m² ha^-1; the species C. alliodora (12 m² ha^-1), T. ivorensis (11 m² ha^-1) and T. rosea (10 m² ha^-1) had the largest BA. ^{Concha et al. (2007)} in Lima, Peru, reported average BA of 5.71 m² ha^-1 and the species with the largest BA were Inga sp, Citrus nobilis and Piptadenia favia. ^{Somarriba and Domínguez (1994)}, in Costa Rica, reported an average BA of 4.1 m² ha^-1; the outstanding species were T. ivorensis (52 m² ha^-1), T. rosea (4.5m² ha^-1) and C. alliodora (2.8 m² ha^-1). ^{FHIA (2007)}, in Honduras, reported an average BA of 4.8 m² ha^-1. The four cited authors recorded smaller BA than that of our results because the trees were young and had small diameters.

Total timber and commercial volumes (TV, CV)

There is a large quantity of timber in cocoa AFS that can and should be used in a sustainable way. A TV of 1 923.8 m³ in logs was found in the 20 sampled sites. The average TV was 192.4 m³ ha^-1. It ranged 70.4 to 619.86 m³ ha^-1. Ten species accounted for 87.4% of the TV; the outstanding species were E. poeppigiana 33.5% (64.4 m³ ha^-1), E. americana 20.9% (40.3 m³ ha^-1) and C. odorata 8.1% (15.5 m³ ha^-1) (Figure 2). ^{Somarriba et al. (1996)}, in Panama, reported the species C. alliodora (90 m³ ha^-1), T. ivorensis (81 m³ ha^-1) and T. rosea (46 m³ ha^-1) as having larger volumes than those of our study, as they were established preferentially for shade. The values similar to those reported by ^{FHIA (2007)} in Honduras for the species C. megalantha 118 m³ ha^-1, T. donnell 33.9 m³ ha^-1, C. arborea 33.6 m³ ha^-1 and V. gaumeri 32.6 m³ ha^-1. ^{Somarriba and DomÍnguez (1994)} in Costa Rica reported T. ivorensis with 35 m³ ha^-1, C. alliodora with 21 m³ ha^-1 and T. rosea with 19 m³ ha^-1 TV.

Figure 2 Species with the highest timber volumes (m³ ha^-1) in cocoa AFS in Cárdenas, Tabasco, Mexico. 

Commercial timber volume (clean bole) recorded in the 20 sampled sites was 526.29 m³ log, an average of 52.6 m³ ha^-1, ranging from 21.9 to 146.7 m³ ha^-1. Ten species accounted for 82.9% of the total VC; of these, E. poeppigiana 27.4% (14.4 m³ ha^-1), E. americana 18.7% (9.9 m³ ha^-1) and C. odorata 11.9% (6.1 m³ ha^-1) were outstanding (Figure 3). Calero (2008), in Costa Rica reported to C. alliodora as the species with the largest VC (31 m³ ha^-1); this is explained by its preferential establishment for shade. Cedrela odorata had a smaller VC, which was also smaller than that recorded in our study.

Figure 3 Species with the largest commercial volumes (m³ha^-1) in the cocoa AFS, in Cárdenas, Tabasco,Mexico. 

Strata of canopy by height

Mean Ht was 10.1 m varying from 2 to 35.5 m; 5.8% of the trees were placed in the stratum of very low canopy (height <5 m). These were new-planted timber species such as C. odorata, T. rosea and C. arborescens. Most of the trees (84.2% of the total) were concentrated in the low canopy (≥5 - <15 m) and 1% in the high canopy (≥25 m) (Figure 4). In this last category were grouped species such as Eritrina (E. poeppigiana), Samán (S. saman) and Chestnut (A. altilis). ^{Ramírez-Meneses et al. (2013)} in Cárdenas, Tabasco, also reported some trees of 36 m, mainly of the species S. saman, G. ulmifolia and C. arborescens, evidencing that cocoa AFS contains tree species of similar height to those found in the tropical rainforests. ^{Somarriba et al. (2003)} reported up to 30 m for the upper canopy of cocoa AFS in Talamanca, ^{C. R. Somarriba et al. (1996)} in Panama and the ^{FHIA (2007)} reported average heights of 17 and 13 m, respectively, which are higher than those recorded in our study, probably because the timber species recorded are not pruned.

Figure 4 Trees distribution by height (m) in cocoa AFS in Cárdenas, Tabasco, Mexico. 

Classification of trees by diameter at breast height (DBH_1.3m)

DBH values varied from 1 to 146 cm, with a mean of 23 cm. Of the 2 856 trees registered, 91% had a DBH of 1 to 40 cm and among these, 53% had 10 to 30 cm DBH (Figure 5). ^{Orozco and Somarriba (2005)} stated that in Bolivia 45% of the trees in cocoa AFS had DBH of 10 to 20 cm, while ^{Ramirez-Meneses et al. (2013)} reported a maximum DBH of 137 cm for some species in Cárdenas, Tabasco. ^{Somarriba et al. (1996)} in Panamá and ^{FHIA (2007)} in Honduras reported average DBH of 25 and 28 cm, respectively. Both authors reported larger DBH than those found in our study, probably because they averaged only three timber species, while we averaged all the timber species found in the cocoa AFS.

Figure 5 Categories of trees by diameter at breast height (DBH_1.3m) of the cocoa AFS in Cárdenas, Tabasco, Mexico. 

Potential use of trees in function of diameter at breast height (DBH_{1.3 m})

According to the DBH, the main timber uses of the registered trees were narrow and thick boards, 39 and 27.4%, respectively, and 6.9% of the trees were recorded without any use (Figure 6) because they were reforestation species established in areas without shade. Use varies with species, age, diversity and culture, among other factors. ^{Somarriba et al. (2003)}, in Talamanca, Costa Rica, registered 34% of the trees with use for thick boards, and ^{Orozco and Somarriba (2005)} in Bolivia reported 15% with this use.

Figure 6 Potential timber use by diameter at breast height (DBH_1.3m) of trees from the cocoa AFS in Cárdenas, Tabasco, Mexico.