In tropical regions, extensive conversion of forests and agricultural intensification are typically identified as the most prominent drivers of land-use change and biodiversity loss (Sala et al., 2000; Wright 2005). The mitigation of tropical deforestation and biodiversity protection must address livelihoods and needs of local communities (Bhagwat et al., 2008). In this sense, agroforestry is considered as a promising approach to reduce deforestation and improve rural livelihoods in the tropics (Current et al., 1995; Ashley et al., 2006). Agroforestry is a land-use management system where trees or shrubs develop around or among crops or pastureland, providing economic, social and environmental benefits (McNeely, 1995; Bhagwat et al., 2008).

Agroforestry systems are often very small in size and surrounded by open landscapes and resemble forest fragments. Species distinctiveness (presence of rare or endemic species) is frequently low, even though their species richness (total number of species) might be equal to, or higher than that of neighboring forests (O’Dea & Whittaker, 2007). Many agroforestry systems are important for protection of species and habitats outside protected areas, and agroforestry systems can be considered as refuges for biodiversity (Bhagwat et al., 2008; Manson et al., 2008; Nonato de Souza et al., 2012). These systems conserve biodiversity in remnant habitats and provide potential movement for species among these remnants (Bhagwat et al., 2008). These systems also provide environmental services such as carbon stock and sequestration (Albrecht & Kandji, 2003; Dávalos Sotelo et al., 2008; Thangata & Hildebrand, 2012), improvement of environmental quality (Tornquist et al., 1999; Geissert & Ibáñez, 2008), water harvesting, reducing water runoff, and increased recharge of aquifers, reduction of floods and droughts, among other services (Mejía et al., 2004).

Almost three quarters of the planet’s surface and 67.3% of Mexico’s surface are covered by ecosystems managed or modified by humans (Pimentel et al., 1992; McNeely, 1995; Palacio-Prieto et al., 2000; SAGARPA, 2007). Because of the dominance of these systems, their management changes can affect biodiversity conservation and ecosystem services (Tilman et al., 2002; MEA, 2005). For the montane cloud forest (MCF), despite its high strategic value for sustainable development, the key role it plays in the hydrological cycle, and being considered as reservoir of endemic biodiversity (Toledo-Aceves et al., 2011), in Mexico it is considered the most threatened terrestrial ecosystem because of land-use changes and the effects of global climate change. Currently, this ecosystem has been assigned as high priority for conservation and promotion of sustainable development (Aldrich et al., 1997; CONABIO, 2010; Toledo-Aceves et al., 2011; Calderon Aguilera et al., 2012).

Montane cloud forest (MCF) in Veracruz occupies ca. 1243.65 km² (1.73% of the total area; Ortega & Castillo, 1996; Ellis & Martinez, 2010; Castillo-Campos et al., 2011). In the central region of the state the MCF area was reduced gradually because of the expansion of the coffee cultivation. Since the late XIX century to 1960’s, MCF’s were replaced with coffee agro-ecosystems (Ruelas-Monjardín et al., 2014), and the forest fragmentation was accelerated because of the demographic pressure and territorial expansion (Williams-Linera et al., 2002), where the greatest impact on vegetation (transformation in species composition) was caused mostly by deforestation, fires, plantations establishment, and land-use conversion to pasture (Ellis & Martinez, 2010). As a result the development of several types of grown-shade coffee such as shade monoculture, simple polyculture, diverse or traditional polyculture, and rustic plantations in which the forest canopy is used as shade for coffee have taken place.

Currently, Veracruz is the second largest producer of coffee in Mexico with the 24.7% of the national coffee production, occupying an area of 1520 km², equivalent to 13.92% of total of vegetation present in the state (Olguín et al., 2011). Coffee agro-ecosystems (CAE) are developed at the lowest elevation of the MCF under similar environmental and climate conditions; therefore, coexistence and recombination (replacement) of species make them particularly complementary (Castillo-Campos et al., 2011). When coffee plantations are under shade, the system “CAE–MCF” maintains forest cover, although with less species diversity compared to the undisturbed MCF. However, because of its structure, species diversity, and environmental services provided, CAE’s are of great importance for conservation (Ellis & Martinez, 2010; Olguín et al., 2011; Toledo-Aceves et al., 2011).

In this work, we compared tree diversity and richness among one grown-shade coffee plantations (CAE) and two sites of montane cloud forests (MCF), one preserved (MCF1) and other perturbed (MCF2), in the coffee region of Coatepec-Xico, Veracruz, Mexico. We also analyzed the importance of coffee plantations as a refuge of tree species, holding a potential role for conservation.

The most representative/abundant species were: i) MCF1: Parathesis melanosticta and Hedyosmum mexicanum, ii) MCF2: Beilschmiedia mexicana, Clethra macrophylla and Carpinus tropicalis, and iii) CAE: Citrus spp. and Inga vera (Table 1). For MCF1 the 14 most abundant species accounted only 15.26 %; whereas for MCF2 and CAE the 14 most abundant species accounted 70 and 77.5% respectively (Table 1). For MCF2 we found several species that evidenced the perturbation degree: Citrus spp., Heliocarpus donnell-smithii, Lippia myriocephala, Myrsine coriácea, Solanum schlechtendalianum and Trema micrantha.

2721 individuals from 154 tree species were registered in MCF1 and MCF2; 116 species in MCF1 and 38 in MCF2. In CAE we registered 64 tree species with 2947 individuals (Table 1). The highest and lowest species diversity corresponded to MCF1 and MCF2 respectively; this was reflected in the Margalef (DMg) and the Menhinick (DMn) indices (Table 2). The Shannon-Wiener (H) index also indicated that MCF1 had the greatest diversity of species, and the Simpson dominance (λ) index showed that CAE was the least diverse site, whereas MCF1 was the most diverse (Table 2).

Regarding floristic similarity, we found that MCF1 and MCF2 shared 15 species, with Alchornea latifolia, Psychotria sp. and C. macrophylla as the most represented (higher number of individuals; Table 3). The greatest similarity was found between sites MCF2 and CAE, sharing 16 species, of which the most frequent species were C. tropicalis and Citrus spp. (Table 3). This similarity between MCF2 and CAE was also observed in the cluster analysis (Figure 2). The sites sharing less number of species (12) were MCF1 and CAE (Table 3). The Jaccard (IJ) and the Sørensen (IS) similarity coefficients, and the Morisita-Horn (IM-H) and the Sørensen (Iscuant) indices also confirmed that sites with less similarity were MCF1 and CAE, and those having the highest similarity were MCF2 and CAE. Values obtained to calculate these indices were higher for both qualitative and quantitative data (Table 4).

We confirmed similarity/dissimilarity among sites using different qualitative and quantitative methods. Measuring species’ relative abundance and similarity among sites allowed us to identify those species whose low representation make them more sensitive or vulnerable to environmental perturbations: B. mexicana, Miconia glaberrima and I. punctata. It is important to note that the most similar sites were MCF2 and CAE, but only with 16 species, i.e. the 20% of species present in MCF2 and CAE as a whole. Low similarity among sites might be due to the low number of common species between pairs of comparisons, finding a high percentage of species exclusive of each site, which contributes to biodiversity conservation at regional scale. These findings highlight the CAE importance for conservation and the high tree biodiversity in the region, especially considering that the sites are not far apart geographically (Figure 1).

Another important finding is the high dissimilarity between MCF1 and MCF2, where the low number of species in MCF2 shows the shocking biodiversity loss in perturbed areas. Although ecosystems such as MCF2, are subject to influences determined by other species (e.g. predators, competitors, invaders), and temporal and spatial variations of environmental conditions, such as nutrient availability, temperature and precipitation (Chapin et al., 2000; Bellemare et al., 2002), human activities and perturbations have a great impact on them. Human perturbations can decrease local diversity or richness, as it was seen in MCF2; however, for CAE this is not necessarily true. Human perturbations can widely change floristic composition of ecosystems, but agroforestry can help to mitigate the impacts of land-use change and preserve local biodiversity.

We found that vegetation in CAE included a wide variability of species, and richness increased probably for a species recombination with the MCF surrounding CAE (Villavicencio-Enriquez & Valdez-Hernández, 2003). We observed evidence of this recombination finding species similarities between CAE and the perturbed MFC2. High diversity might be due to a species shift with the MFC nearby (Williams-Linera, 2002). Also, the highest floristic similarity between MCF2 and CAE indicates that CAE is also a perturbed ecosystem. In CAE, the lower diversity in comparison with MFC1 is probably caused by the dominance of some species, partially Citrus spp. and I. vera.

The floristic composition in CAE is the result of the system’s function directed to coffee cultivation. We found in CAE that 33.36% of the individuals were I. vera, which are promoted by farmers. Here, it is clear that diversity is influenced by local management, and not only by topography, precipitation or temperature. Trees provide numerous benefits such as building materials, food and firewood, generate family income, promote ecological conditions for wildlife habitats and ecological balance, and also protect against soil erosion (Salam et al., 2000). It has been shown that farmers in agroforestry systems select and eliminate certain tree species according to their preferences and beliefs (Salam et al., 2000; Russell & Franzel, 2004), and also to morphological characteristics (Schroth, 1995); therefore, species composition is conformed by ecological and biophysical variables, and management as well. Also, farmers are paid to modify their farming practice to provide environmental benefits (Salam et al., 2000; Kleijn & Sutherland, 2003). This management provides economic profit and income for local farmers, but it also contributes to improve social levels through the production of important goods, including export crops, fruits, raw material and firewood. Agroforestry systems success lies in the ecological productive capacity over the long term and also in economic benefits (Michon & de Foresta, 1995).

Maintaining biological diversity is essential for productive agriculture, and ecologically sustainable agriculture is in turn essential for maintaining biological diversity (Pimentel et al., 1992). This maintenance by CAE is reflected in the high number of different species compared to MCF2 and MCF1, which shows CAE’s conservation potential, in spite of the presence of exotic (e.g. Citrus spp.) and secondary tree species (Table 1) that would be indicators of disturbance. Also, the high proportion of species registered in CAE can support evidence of the services that can provide this system, although we did not evaluate environmental services. Preservation of this agro-ecosystem might represent a possible solution to minimize local biodiversity loss and improve conservation in the central region of Veracruz, especially because the coffee cultivation is more beneficial to the environment than pasture and monocrops such as sugarcane (Esperón-Rodríguez et al., 2016), because coffee conserves tree cover and allows connectivity between open landscapes and forest fragments.

CAE’s in Veracruz have a potential role as reservoirs of biodiversity maintaining the forest cover; their conservation as refuges must be considered a priority especially in areas where deforestation and land-use change are increasing. Conservation plans should be addressed to maintain local connection and species recombination between CAE and preserved forests. Knowing the local biodiversity can help local farmers to make better management decision, introducing agroforestry systems with consideration of the markets and products, and also the potential productivity gains and food crops. It must be noticed that although similar, CAE’s are not substitute for natural forests; therefore, surrounding forest play an important role in conservation, especially for species that cannot thrive in human modified landscapes. Local management must prioritize the biodiversity preservation and conservation.

But, what causes this biodiversity in Veracruz? We registered high biodiversity in our study sites despite the size of the sampled area (1500 m²). When we compared our results with previous studies from Veracruz, Mexico and South America (Table 5) we found a high biodiversity in Veracruz. Although differences may be due to several factors, precipitation is a factor that caught our attention because rainfall is a highly varying parameter in Veracruz (Barradas et al., 2010), and previous studies have shown the importance of water for the species development in the MCF of Veracruz (Esperón-Rodríguez & Barradas, 2015). We hypothesize that the high biodiversity found in our study might be because of a relatively low precipitation compared to other regions.

Regardless what is causing this high biodiversity, our results indicate that the central region of Veracruz is an important refuge for species, where CAE’s

plays an important role in the conservation of biodiversity.