Bee diversity in secondary forests and coffee plantations in a transition between foothills and highlands in the Guatemalan Pacific Coast

Study sites

This study was conducted from March 2013 to February 2014 in SLT foothills region (Appendix 1A). SLT is located at the limit between the central highlands and costal lowlands in southeastern Guatemala, where many Kaqchikel indigenous people live. SLT is bordered by two volcanoes, Atitlán and Tolimán that range from 800 to 3,500 masl and produce a variety of microhabitats. According to the Villar classification, the primary vegetation of studied farms are within a subtropical humid forest, where broadleaf evergreen forest divides the mountain forest from the tropical humid savanna on the Pacific coast (Villar Anleu, 1998). These biotic and topographic differences give the area a dynamic ecotone with high precipitation (Villar Anleu, 1998, 2003). According to the Guatemalan National Council of Protected Areas (Consejo Nacional de Áreas Protegidas (CONAP), 2008), the department of Sololá, where SLT is located, has 35% forest cover, making it the most forest-covered department in Guatemala. This fact is associated with the high degree of community conservation but is also due to the presence of private farms that normally have their own forest reserves.

Sampling design

Sampling was conducted at three coffee farms with different management types (Table 1). The farms harvest Coffea arabica, cultivar “caturra”. The three farms have secondary forests nearby, farm 1 and 3 are separated by 0.8 km, and 4 km from farm 2 (Appendix 1B). The names of these private farms must be withheld at the request of the owners.

In each of the studied farms, three plots of 60 m² were established. Each plot was categorized as early secondary growth, late secondary growth or coffee plantation, the plots within the farm were separated by at least 0.5 km depending on the farm. Early secondary growth was characterized by early secondary forest with up to one year of development, mainly herbaceous vegetation with few bushes and an abundant incidence of light. Late secondary growth was characterized by late secondary forest with two to three years of succession, with mainly shrub vegetation and luminosity slightly restricted below the high bushes. Coffee plots were selected in patches of shaded coffee in the selected farms. Plot characteristics are presented in Table 1.

Bee sampling

Bees were sampled every month from March 2013 to February 2014 in each vegetation type plot, covering the two Guatemalan climatic seasons established as dry or summer from November to April and rainy or winter from May to October (Consejo Nacional de Áreas Protegidas (CONAP), 2008). During three days each month, five people searched for bees on flowering plants in all of the selected plots. For each flowering plant species in each plot, bee sampling was conducted for 40 min at different times between 8:00 and 12:00 p.m. The sampling schedule was done considering results of previous temperature and humidity monitoring where optimal conditions for bee activity in the area were established. Bee sampling was based on the direct search method on flowers, using net sweeping (Brosi et al., 2008; McGavin, 1997; McMullen, 1965). Bees captured from flowers were killed by freezing in individual containers. Native bee (no honey bee) specimens were mounted on insect pins labeled with field data and assigning a unique code for deposition in the “Colección de abejas nativas del Centro de Estudios Conservacionistas de la Universidad de San Carlos de Guatemala”. In contrast, honey bee specimens were sampled and recorded but not mounted. Taxonomical keys were used for bee identification to genus or species (wherever possible) (Ayala, 1990, 1999; Michener, 2007; Michener, McGinley & Danforth, 1994; McGinley, 1986; Roberts, 1972; Roubik & Hanson, 2004; Smith-Pardo, 2005; Snelling, 1974) and with the help of the bee expert Ricardo Ayala, and collaboration of Mabel Vásquez, Carmen Yurrita y María José Dardón. Collection 166 permits for conducting bee and plant sampling are 000960, 002011, 002009 and the Field 167 Research License were 007/2015 and 043/2012.

Data analysis

Seasonal richness and abundance

Over one year of study, 3,004 bee specimens, belonging to 102 species (Appendix 2) and 100 species of flowering plants with visiting bees (Appendix 3) were recorded. Bee and plant richness inside plots were significantly correlated (t₃₆ = 7.82, p = 2.8E−09, cor = 0.79) showing a close relationship between them (Fig. 1A).

The study area showed a sparse flowering season from March to October 2013 and high flowering in the early dry season from November 2013 to February 2014, closely reflecting the two climatic seasons in Guatemala. A few months after the lowest records of bee abundance (March–October 2013), the rainy season promoted vegetative growth in the secondary forest, and it was correlated with the highest values of flowering plant richness with a 21% increase (Fig. 2; Appendix 3).

Plant richness was similar between seasons (F_{1, 1} = 2.53, p = 0.37), as well as total bee abundance and richness (F_{1, 1} = 40.6, p = 0.09 and F_{(1, 1)} = 66.1, p = 0.07, respectively). However, bee groups responded in different ways, native bees and social bees showed higher abundances during the dry season (F_{(1, 1)} = 195.4, p = 0.05, F_{(1, 1)} = 118.5, p = 0.05, respectively) while honey bee and stingless bee abundance showed a similar seasonal trend but was no statistically significant (F_{1, 1} = 19.1, p = 0.1; F_{1, 1} = 72.9, p = 0.07, respectivelysee Table 2). There was an interesting finding in February 2014 (Fig. 2) when bee richness increased by 57% (from 28 to 53 species), presenting the highest value in the study, but out of phase with respect to the peak of local flowering plants (December 2013), coinciding with the beginning of drought, with the biggest drop of flowering (14%). In the following month when the coffee plants were flowering (March 2014), the greatest number of bees was recorded (437 records), but the native bee species began to decrease (10%).

Interestingly, honey bees and stingless bees were significantly and positively correlated (r = 0.62, t₃₆ = 4.78, p = 2.95E−05, Fig. 1B), as were the honey bees and native bees (r = 0.61, t₃₆ = 4.61, p = 4.84E−05, Fig. 1C). Highlighting that for the three groups of bees, the availability of floral resources is important for their activity.

Bees in coffee plantations and the surrounding secondary forest

The farms presented different bee richness and abundance. As shown in Fig. 3, farm 1 had the highest number of bees followed by farm 2 and farm 3. Farms 1 and 2 presented the highest bee abundance in plots with secondary forest, while in farm 3 the coffee plantations presented the highest abundance.

Bee richness mainly comprises five families (Fig. 3). Most of the bees are Apidae (84.8%), followed by Halictidae (8.9%), Andrenidae (2.6%), Megachilidae (2.3%) and Colletidae (1.5%). Species from the five bee families were registered in the three farms. On farm 3, 91% of captures belonged to Apidae, while farms 1 and 2 had a higher representation of Colletidae and Andrenidae. Vegetation type (early secondary forests, late secondary forests and coffee plantations) showed a significant effect on total bee abundance (F_{2, 8} = 7.92, p = 0.01), bee richness (F_{2, 8} = 6.38, p = 0.02), native bees (F_{2, 8} = 5.33, p = 0.03), stingless bees (F_{2, 8} = 8.41, p = 0.01), social bees (F_{2, 8} = 9.01, p = 0.008), honey bees (F_{2, 8} = 6.2, p = 0.02), and Apidae and Megachilidae families (F_{2, 8} = 10.01, p = 0.007, F_{2, 8} = 19.01, p = 0.0009, respectively), particularly between early growth and coffee plantation (Fig. 4).

The diversity estimators Chao1, ACE, ICE (Fig. 5) calculated per farm and vegetation type did not present any significant difference (p > 0.05). The Chao2 estimator showed significant differences among farms (F_{2, 2} = 7.903, p = 0.048). The Shannon-Wiener diversity index (H) did not present significant differences among farms. In farms 1 and 2, the richness estimators gave higher values to plots with early secondary forest, while the highest estimate value on farm 3 was for the coffee plantation. The species accumulation curves showed stabilized curves that started to flatten down in some plots (Fig. 6). Coffee plantations registered the lowest values in bee diversity, and accordingly, the richness estimators predicted the lowest number of species (Figs. 5 and 7). In farms 1 and 2, the coffee plantations registered most of the bee records during the coffee flowering season (March 2014), otherwise, neither bee activity nor early growth vegetation were recorded because coffee farmers clean adventive vegetation from coffee plantations, particularly in farm 2. On farm 3, the coffee was rarely cleaned in this manner, which contributed to the establishment of early secondary forest vegetation.

An important finding was the presence of a new species of bee genus Rhathymus Lepeletier & Serville, 1828, Rhathymus atitlanicus, described in Ayala, Hinojosa-Díaz & Armas-Quiñónez (2019). Those bees were only found in secondary forest of farm 1 and farm 3 (Appendix 2).

Finally, the cluster analysis of bee diversity per vegetation type (Fig. 8) presents a significant pattern that groups the study plots into two significant aggregations. The strongest aggregation with 97% confidence is composed by the early secondary forests from farm 1 and farm 3 and the coffee plots of farm 3 (the farm that allows secondary forest plants into the coffee plantation). The other aggregation with 97% confidence, is composed by the coffee and late secondary forests of all three farms. Within this group there are significant differences among the plots of farm 1, where coffee is harvested following traditional practices, and on the other side of the same group, is the late secondary forest linked with the coffee of farm 2. Cluster analysis shows that the early secondary forest in farm 2 is 77% different to the rest of the plots.

The effect of flowering seasonality on bees

The bees recorded over the whole year in secondary forest and coffee fields could give an insight into the synchronization that exists between bees and phenology of flowering plants, due to the periods of highest bee abundance matching with flowering periods and the correlation between these parameters throughout the study. This correlation supports previous findings (Brosi, 2009; Banks et al., 2014; De Marco, Monteiro & Coelho, 2004; Donald, 2004; Taki et al., 2013) about the great importance of secondary forests in the provision of resources to bees. This seasonality can have some important consequences for bee diversity but also for coffee production, since the highest abundances of native bees were observed during secondary forest flowering (March 2014), close to the coffee flowering that varies according to the first rains. Early flowering of coffee could coincide with secondary forest flowering, which would cause native bees to interact with the high honey bee abundance possibly giving rise to a saturated environment of pollinators for the coffee and secondary forest. However, the asynchrony of pollinators and flowering periods in coffee fields must be treated with care and taken into account in crop management as a priority for improved of coffee production (Boreux et al., 2013).

When investigating the relationship between honey bee and native bee abundance, we found a positive correlation between them, as other studies in Mexican coffee plantations have found (Badano & Vergara, 2011). This fact suggests that honey bees may not yet have saturated this ecosystem, and that, according to Banks et al. (2013), the contribution of pollinators from nearby forests to the coffee plantations is still high. This notion can be supported by the fact that honey bees only maintain high populations during coffee flowering, otherwise, their cultured populations are kept to a minimum in the area. However, the abundance of the groups bees and the total of bees are showing differences between vegetation types probably due to the nearby farm vegetation, especially surrounding secondary vegetation and primary forest were the native bees usually nest.

Secondary forests as coffee pollinator enhancers

The data of bee abundance, richness and diversity supports the importance of secondary forests for bee pollinators (Arnan et al., 2011; Carvalheiro et al., 2012; Banks et al., 2013; Brosi et al., 2008; Boreux et al., 2013), not only to maintain bee diversity, but also to improve coffee production in the Pacific Coast Foothills of Guatemala. The significant results in the nested anovas remarks how the vegetation interact with bees, they are most abundant in early secondary forests(Table 2). Meanwhile, the presence of stingless bees in the secondary forests and also in the flowering coffee plants, suggests that these bees, which depend exclusively on the forest for nesting, also depend on secondary forests for the acquisition of essential resources (Winfree, 2010).

Secondary forests showed that they can maintain and provide resources for the native bees during periods when the coffee is not flowering. The farmers that use conventional agriculture need to place greater emphasis on preserving secondary forests, rather than only pristine forests (Blanque, Ludwing & Cunningham, 2006). Pollination tests are required in order to compare the pollination efficiency of native bees and honey bees in these plantations. In this way, it would be possible to more definitively establish the importance of preserving native bees and the places where they obtain resources, such as the secondary forest, for coffee production (Winfree, 2010).

Effect of farm management

We found important differences in bee diversity between farms. Farm 1 shows the highest diversity values, suggesting that, despite the lack of high technology, traditional knowledge remains effective in preserving native bee diversity, especially for the stingless bees. The conventional farm with high-intensive management (farm 2) keeps the coffee plantations cleared of early secondary forest but maintains surrounding secondary vegetation on the farm; therefore, the presence of Tephrosia spp. could function as a provision plant for native bee species. A positive interaction between farm 2 and bee family abundance can suggest the importance of nearby forest, since the farm 2 manage their own forest reserve, providing the bees with a complex landscape that may enhance bee resource acquisition, as suggested previously by different authors (Carrié et al., 2017; Winfree et al., 2009).

Also, it is evident that the conventional farming with low-intensity management and a disturbed surrounding forest (farm 3) has the lowest bee diversity and, that it is the early secondary forest left in the coffee fields for a short time period that functions as resource provider for the surrounding bee diversity. This could explain why, in the cluster analysis (Fig. 8), the coffee of farm 3 was grouped together with the early secondary forest of farm 1 and 2.

None of the studied farms use honey bee breeding to pollinate coffee at the time of this study. Some neighboring farms, however, do have managed bee hives and it is possible that the honey bees recorded during the study came from those neighboring farms. The farm with the closest neighboring honey bee hives was farm 3, which showed the lowest bee abundance, richness and diversity values as well as the highest honey bee abundance. On the other hand, this farm (3) also shows a poor surrounding forest management providing few resources for native bees, contrary of farm 1 that promotes this mixed-landscape (Carrié et al., 2017).

Another factor to consider that may affect bee diversity is the use of chemicals in the farms: the quantity of insecticides used for pest control inside (such as that used to control the Mediterranean fruit fly), but also those used outside farms in neighboring crops cultivated near coffee (Brittain et al., 2010; Schmitt et al., 2009; Schüepp, Rittiner & Entling, 2012).