Location and description of the study area

The research was carried out during 2013 and 2014 in the municipality of Tzucacab, Yucatán, México, between coordinates 19° 55’ 823’’ and 20° 00’ 873’’ N and 88° 57’ 818’’ and 89° 02’ 580’’ W, with 165 m of average altitude. The soil is of karstic origin, characterized by the presence of Leptosol, Cambisol and Luvisol (^{Flores and Bautista, 2005}). The climate is sub-humid warm (Aw₀) (^{García and CONABIO, 1998}). The mean annual temperature during 2015 was 27.5 °C and the accumulated annual precipitation, 1 210 mm (^{CONAGUA, 2015}). The type of vegetation is secondary medium sub-deciduous rainforest (^{Flores and Espejel, 1994}; ^{Zamora et al., 2009}). Three bovine production units (BPUs) were evaluated within a radius of 15 km. The BPUs were made up of double-purpose animals (Box Taurus x Bos indicus), in their majority cows (75-85 %). The size of the pastures ranged between 10 and 40 ha, with a resting period from grazing of between one and three months. The cattle diet was based on the consumption of grasses and secondary vegetation throughout the year. The surface and animal load of the BPUs were 130 and 0.33, 108 and 0.19, and 800 and 0.10 hectares and UA/ha/year, respectively, considering UA equivalent to one bovine of 450 kg live weight.

Based on the description of the plant succession by ^{Flores (2001)}, the forest, shrub and herb communities are distinguished. The Forest Community (FC) was characterized by the presence of trees with more than 8 m of height, diameter at breast height (DBH) higher or equal to 10 cm and 10 to 15 years of abandonment. In the Shrub Community (SC) trees below 4 m of height were found, DBH lower than 10 cm, with shrub dominance and two to ten years of abandonment; and in the Herb Community (HC), with presence of herbs, grasses, vines or lianas and two to four years of abandonment. One of the BPUs did not have HC and another one lost it during the dry season.

Description of the study

The vegetation of a medium sub-deciduous rainforest in different successional stages was characterized to understand the level of dominance of the fodder species, through the index of relative value of importance (RVI). The differences in abundance of fodder species between plant communities were analyzed. The dry matter content of the potential phytomass was estimated in the woody and herbaceous strata of the FC, SC and HC during the rainy season (October and November 2013) and dry season (April and May 2014). The chemical-nutritional contents, in situ ruminal degradability of the dry matter, and the concentration of metabolizable energy of the fodder species collected were determined at 48 h during the rainy season in the study area.

Sampling of the vegetation

A first group of variables made up of the wealth, abundance, frequency, DBH and coverage of all the plants found was evaluated, with the aim of estimating the RVI and understanding the level of dominance of the fodder species. For this purpose, in each type of plant community 75 m to the west were visited systematically (^{Bautista et al., 2004}). At the end of the visit the establishment of sampling frames began, with a distance of 10 meters between each. Four sampling frames of 16 m² (4´4 m) were placed for the woody stratum, with nested frames of 1 m² for the herbaceous stratum (^{Mueller-Dombois and Ellenberg, 1974}; ^{Bonham, 1989}). In the FC and SC, 24 nested frames were placed in total per community and in the HC only 12 were placed, due to the absence of the woody stratum and because there wasn’t always a HC in the BPUs.

The representativeness of the sample was calculated through an accumulation curve of species using the Chao 2 Mean model (Estimate S version 9.1.0, ^{Colwell 2013}), which is more adequate when there are data of presence-absence of the species (^{Álvarez et al., 2006}). The species were identified in the “Alfredo Barrera Marín” herbarium of the School of Veterinarian Medicine and Animal Husbandry of the Universidad Autónoma de Yucatán (Facultad de Medicina Veterinaria y Zootecnia de la Universidad Autónoma de Yucatán, FMVZ-UADY).

Potential phytomass

A second group of variables made up of the potential phytomass of the secondary vegetation and ruminal degradability was evaluated, and the concentration of metabolizable energy of fodder species. The potential phytomass was collected in the sampling frames of 1 m² previously established, with the objective of describing in detail the fodder potential of the plant community strata. The potential phytomass which was in reach of the animals was considered (^{Bello et al., 2001}). For their sampling, fodder, fruits, flowers and tender stems were defoliated manually within the 1 m² frames previously established and at a height of between 5 and 170 cm. The material collected was dried at 60 °C in a forced air stove during 72 h (^{AOAC, 2000}).

Nutritional analysis

The purpose of analyzing the chemical-nutritional composition and determining the ruminal degradability and concentration of metabolizable energy was to describe the nutritional characteristics of the fodder species found in the medium sub-deciduous rainforest consumed by bovines. All the plants present in the stratum of 5 to 170 cm height were included, which in previous works had been identified as fodder species (^{Sosa et al., 2000}; ^{Ayala et al., 2006}; ^{Alonso et al., 2009}; ^{Velázquez et al., 2010}; ^{Flores and Bautista, 2012}). The collection was carried out in November 2013, considering that the dominant vegetation would be flowering (^{Flores, 2001}), thus reflecting optimal levels of nutrients for the livestock (^{Ayala et al., 2006}). Manually, 1 kg of leaves, fruits and tender stems of the species listed as fodder and present in the sampling sites was cut.

Chemical-nutritional composition of fodder species

The neutral detergent fiber (NDF) was determined through filtered bags, applying method 6 from Ankom Technology (^{Ankom Tecnology A200 and A200I}) (^{Van Soest et al., 1991}), raw protein (RP N × 6.25) with an elemental analyzer (C, N), Lecco CN-2000 series 3740 (^{Lecco, 2013}) and condensed tannins (CT), using the colorimetric method of Vanillin-Hydrochloric Acid (^{Makkar and Becker, 1993}) on 20 dominating fodder species, identified as such after calculating the RVI. The results from the CT should be approached carefully, since no liquid nitrogen was used during the sampling nor was it refrigerated later for the preservation of the samples.

Determination of ruminal degradability and concentration of metabolizable energy

After 48 hours, assays of ruminal degradability were performed in situ of the dry matter on 16 of the most dominating fodder species. The technique of nylon bag with a pore opening of 53 µm (^{Ørskov et al., 1980}) was used in two cows (430 kg of live weight) canulated in the rumen and fed with 20 kg of Pennisetum purpureum (cv Taiwan), 4 kg of Leucaena leucocephala and 1 kg of concentrate (13.5 % of PC). The metabolizable energy (ME) was estimated, using the in situ degradability of dry matter (DisDM) with the following equation (Orskov et al., 1980):

Data analysis

In order to describe the levels of dominance, the RVI of the herbaceous and woody strata of all the species (fodder and non-fodder) was estimated (^{Mueller-Dombois and Ellenberg, 1974}). The values of abundance, frequency, coverage and DBH found in the production units were integrated into a single group of data for the FC, another for the SC and a last one for the HC. The fodder species with highest dominance were selected to subject them to chemical-nutritional analysis.

Statistical analysis

To determine whether the composition of fodder species changed between the different vegetation communities studied, their similarity was estimated through the Sorensen Coefficient (^{Magurran, 1988}) using the method of paired group analysis without bias, with the arithmetic mean (UPGMA), using the Statistical Multivatiate Program (MVSP, version 3.1, 1985-2013, Kovach Computing Services) (^{Sørensen, 1948}; ^{Kovach, 2007}). The variable of absolute abundance was selected as an indicator of proportion (^{Magurran, 1988}) of fodder species inside each plant community. Variance analyses were performed of the absolute abundance of fodder species and the potential phytomass, applying factorial models of 2×2 (two plant communities for two times of the year in woody) and 3×2 (three plant communities for two times of the year in herbaceous). The SAS software was used (^{Statistical Analysis System Inc. 2000; North Caroline; USA; version 8.1.}).

Attributes of the vegetation and of the fodder species

More than one thousand individuals (1745) were found, distributed in 154 species, grouped into 49 families (fodder and non-fodder). According to the Chao 2 Mean estimator, the accumulation curve of species did not reach the asymptote (Figure 1); however, a representation of 85.5 % of the wealth expected (180 species) was obtained.

I.C.: interval of trust, S (est): species observed.

Figure 1 Accumulation curve of Chao 2 Mean species. 

The wealth of fodder species was 54, grouped into 21 families, in contrast with the eight fodder species found in the municipality of Tzucacab, Yucatán (^{Zamora et al., 2009}) and 35 fodder species for bovines in the Yucatán Peninsula (^{Flores and Bautista, 2012}), indicating a high wealth of fodder species in the sites studied. The fodder wealth was numerically higher in the rainy season, except in the SC (Table 1), suggesting the presence of species with highest resistance to drought in this type of community. The Fabaceae family was the most represented (Table 2) and the most dominant (Table 3), in agreement with ^{Flores and Bautista (2012)}, ^{Zamora et al. (2008)} and ^{Gutiérrez et al. (2012}). This is explained because it is one of the most important families in the world flora and in the tropics (^{Flores, 2001}), which is why it has been considered an important fodder in animal production (^{NAS, 1979}).

In terms of the type of fodder species and their dominance in relation to the type of vegetation or time of abandonment (Table 3), similar species to those found by ^{Flores (2001)} were observed in the henequen zone of Yucatán, with ranges of RVI relatively similar in Hyptis suaveloens (RVI 10.19 %), Piscidia piscipula (RVI 79.74 %) and Guazuma ulmifolia (RVI 25.02 %) in sites with two, five and fifteen years of abandonment, respectively. The woody fodder species had the tendency of occupying the first places of relative importance (RVI) and the herbaceous fodders lower places, except in the Herbaceous Community, where the herbaceous fodder species in the medium sub-deciduous secondary rainforest dominated (Table 3). The dominance of perennial fodder species could favor the permanence of potential food in the critical time of the year, especially species such as Acacia pennatula, Celosia virgata, Guazuma ulmifolia, Hybanthus yucatanensis and Piscidia piscipula (^{CICY, 2010}), while in the HC the Poaceas dominated primarily, because it is a young stage of succession with low tolerance to shade and resistant to grazing (^{Gleen-Lewin et al., 1992}; ^{Villegas et al., 2001}). On the other hand, the similarity between the Forest and Shrub communities was 0.54 in rain season and 0.47 in dry season; in the Herb Community in rain and dry seasons it was 0.36, and in Herb Community with the Forest and Shrub communities it was 0.27 (Figure 2), which indicated that the type of species changed between plant communities, offering a great variety of species.

CH1: Herbaceous community in rain season; CH2: Herbaceous community in dry season. CA1: Shrub community in rain season; CA2: Shrub community in dry season. CF1: Forest community in rain season; CF2: Forest community in dry season.

Figure 2 Dendrogram of similarity from the Sorensen Coefficient of the species considered as fodder in the sub-deciduous medium rainforest. 

On the other hand, interaction was found from the effect of season and community in the abundance of woody fodder species (p=0.01), with it being higher in the rain season of the Forest Community (p<0.0001) and in the dry season in the Shrub Community (p=0.0019) (Table 4). In terms of the abundance in the Forest Community, it was favored in woody species in the rain season and the Shrub Community was less affected in the dry season, making the SC a food reservoir in the critical drought season. The high variance in the abundance of herb species in the FC was explained by the low tolerance of this stratum to the shade (^{Gleen-Lewin et al., 1992}), which is why different light conditions created by the woody stratum can vary within the same site, originating constant differences in their abundance. In this sense it is necessary to understand the selectivity and preferences of the livestock when selecting their diet to understand how the nutritional wealth of the secondary vegetation is exploited.

Potential phytomass

The potential phytomass of the secondary vegetation was 3272 kg DM/ha in the rainy season and 1454 kg DM/ha in the dry season throughout all the strata and communities, which agrees with the amounts of fodder available reported by ^{CICC (2009)} for the Yucatán Peninsula, with the animal load used in the study sites being consistent with the capacity of animal load indicated for the region (^{CICC, 2009}). In the woody fodder stratum there was higher potential phytomass in the Forest Community in the rain season (2030±741 kg DM/ha) explained by the impact of the season (p<0.0001). In the herbaceous stratum a higher phytomass potential was found in the Herb Community (3977±2299 kg DM/ha in the rain season and 2451±3336 kg DM/ha in the dry season) explained by the effect of the community (p=0.011). The high variance in the potential phytomass is explained by the presence of rock outcrops where no vegetation was found, characteristic of Leptosols (^{Flores and Bautista, 2005}). The potential phytomass contribution of the HC, due to the presence of herbs and grasses, could be a key element to complement the source of food for the cattle in the secondary subtropical rainforest, since grasses have been traditionally associated to livestock production because they are an important source of dry material, with regrowth capacity and resistance to grazing that may have good nutritional quality (^{Villegas et al., 2001}). The medium sub-deciduous rainforest is characterized by a defoliation of between 50 and 75 % (^{Flores and Espejel, 1994}), indicating that the presence of potential phytomass in the dry season could mean a strategic dietary resource for the survival of the cattle, in addition to the fallen leaves and husks (^{Pettit et al., 2011}). Although traditional livestock production in the south of Yucatán is based on feeding with rainfed grasses (^{Osorio and Marfil, 1999}), possibly the presence of areas in successional young stages (with grasses), intermediate (with woody), and mature (with developed trees), could be linked to grazing practices in the rainy and dry seasons. Since there are no reports available in the study sites of potential phytomass, the results obtained are very important to achieve a better exploitation of the fodder resource in these production systems.

Chemical-nutritional composition, ruminal degradability and metabolizable energy concentration of the fodder species

The chemical-nutritional composition of the dominant fodder species analyzed varied between 35 and 77 % in their NDF content; 5 to 27 % of RP; non-detectable, at 12.75 % of CT; from 49 to 94.75o% of DisDM at 48 h; and from 7.57 to 12.43, MJ of ME, with Senna uniflora standing out, with 94.66 % in DisDM at 48 h (Table 5). The ranges of nutritional values found here agreed with those found in other studies (^{Ayala et al., 2006}; ^{Palma et al., 2011}), especially in contents of raw protein above 12 % in woody species and percentages of DisDM around 70 %. The high content of NDF in Poaceas (over 69 %) is characteristic of tropical grasses, except Lasciasis ruscifolia (Table 5), whose perennial characteristic and high dominance could make it strategic during the dry season, while the concentrations of nutrients and secondary compounds was quite diverse, which in turn defined digestibilities and metabolizable energy concentrations in a broad range of variation. This has also been found in other studies of diverse vegetation in communities subject to grazing (^{Velázquez et al., 2010}; ^{Flores y Bautista, 2012}).

Based on the dominance of fodder species reflected in the RVI (Table 3) and the values of chemical-nutritional composition obtained (Table 5), a trend was observed in the values of the species of the FC that suggests a significant offer of energy, due to the lower content of NDF, higher DisDM and concentration of metabolizable energy, which implies a higher nutritional contribution; although this plant community offers a lower amount of potential phytomass, it could be complementary to the contribution of dry matter of the herbaceous community. It is important to point out that for the exploitation of tree fodder species, they should be available for the animal, or else, be destined to fodder for cutting. On the other hand, the species with low concentrations of NDF, high DisDM and ME concentration could promote a higher voluntary consumption (^{Dulphy and Demarquilly, 1994}) and a faster passage rate (^{Ku et al., 1999}), which could represent a possible source of energy of fast ruminal availability that allows maximizing the microbial protein synthesis in the rumen, as well as increasing the contribution of microbial nitrogen to the small intestine (^{Sauvant and Van Milgen, 1995}), while the presence of secondary metabolites in the plants evaluated could induce mitigation in the production of enteric methane when modifying the microbial populations in the rumen and, as consequence, modifying the products of fermentation (acetic, propionic and butyric acid) of rumen carbohydrates (^{Gerber et al., 2013}), in addition to the nematicide effect, and which would contribute to decreasing the load of intestinal parasites (^{Hoste et al., 2006}; ^{Mueller, 2006}). The fodder species found, clearly differentiated by type of community, may represent an offer of food for the bovine livestock with different nutritional characteristics that could be complemented throughout the year for a better distribution of the availability of fodder.

In this manner, the nutritional potential of the secondary vegetation described before constitutes a potential of use for grazing in this type of plant communities. Avoiding felling to induce the monocrop of improved pasturelands, when promoting this grazing in the natural vegetation, in addition to providing complementary fodder in critical times, allows its fodder species’ structure and diversity to foster a higher connectivity between fragments of primary vegetation; the very presence of secondary vegetation optimizes environmental services such as nitrogen fixing, phosphorus solubility, and improves the biological activity of the soil and offers habitat for wild fauna (^{Murgueitio et al., 2008}; ²⁰¹¹).