Skip to main content
Dryad logo

Heterogeneous agroecosystems support high diversity and abundance of trap nesting bees and wasps amongst tropical crops

Citation

Coutinho, Jeferson; Coca, Catalina; Boscolo, Danilo; Viana, Blandina (2020), Heterogeneous agroecosystems support high diversity and abundance of trap nesting bees and wasps amongst tropical crops, Dryad, Dataset, https://doi.org/10.5061/dryad.d51c5b001

Abstract

Land-use intensification for agricultural purposes modifies the structure of natural environments in various ways and at different spatial scales. These modifications can affect ecological processes and the community structure of multi-environment users such as solitary bees and wasps. Understanding the role of distinct habitat descriptors in promoting such changes is one of the major challenges of empirical studies. In this study, we use a multi-scale approach to evaluate how landscape compositional and configurational heterogeneity, vegetation structural complexity, and the proportion of agricultural landscape composition affect communities of bees and wasps that nest in pre-existing cavities in remnants of native vegetation bordering agroecosystems. We selected 25 sampling points along a gradient of amount of surrounding agriculture and landscape diversity within natural physiognomies located in Chapada Diamantina, Bahia, Brazil. Through model selection using Akaike's information criterion, we verified the complementary roles of landscape heterogeneity and local vegetation in structuring these hymenopteran communities. Abundance in the groups showed different tendencies depending on the descriptors employed, pointing to the importance of evaluating within-group specificity. Furthermore, bees and wasps presented differential responses to landscape composition, but they did not differ in relation to configurational complexity. In more heterogeneous landscapes or sites with more complex local vegetation, the proportion of agriculture had a positive influence on the response evaluated. Efficient management of agricultural landscapes therefore requires increased landscape heterogeneity and conservation or restoration of native vegetation remnants at the local scale.

Methods

This study was conducted in Chapada Diamantina, Bahia, Brazil, in the irrigated agricultural development region located in the municipalities of Mucugê and Ibicoara, (41°28’40”S, 13°09’10”W).  

To assess the influence of landscape structure on associated biological variables, a land-use map was drawn based on the supervised classification of satellite images. These images, dated September 14th, 2011, were obtained from the LANDSAT 5 Thematic Mapper (TM) satellite. The spatial resolution of the images was 30 m. The images underwent geometric correction, georeferencing, atmospheric correction (DOS2) and radiometric calibration before the classification procedure was applied (Moreira et al., 2015). A total of 13 classes were employed, including nine classes of vegetation and four complementary classes. The vegetation classes used in the study were based on those proposed by Veloso et al., (1991) for the classification of Brazilian vegetation. It was necessary to adapt certain classes to the context of this study, which resulted in the following classes: disturbed vegetation (recently abandoned areas occupied by ruderal vegetation); grassy-woody savannah; parkland savannah; wooded savannah; forested savannah; semi-deciduous forest; rocky park savannah; rocky wooded savannah; rocky steppe savannah; anthropic areas (primarily including agricultural lands, with minor contributions from roads, buildings and bare soils of anthropogenic origin); clouds; water; and shadows (Figure 1C). The classification was performed using the maximum likelihood classification algorithm (Novo, 2008; Weng 2011) available in the application ENVI 4.7 ITT 2009.

Based on this map, we chose 25 sampling units and applied a focal-site, multiscale study design. Each sampling unit was located in the center of a series of concentric circles with radius varying from 200 to 3000 m (Figure 1D). These scales were used because they are compatible with the home ranges of most hymenopterans (Zurbuchen et al., 2010a). We distributed these sampling units throughout the agricultural development region and calculated the proportion of agricultural use (AGRI), the landscape diversity index (SDI, which is a relative measure of patch diversity and is equal to zero when there is only one patch in the landscape and increases as the number of patch types or the proportional distribution of patch types increases), and the landscape configuration (configurational complexity), using the area-weighted mean shape index (AWMSI, which is the average shape of the landscape elements weighted by the area occupied by each element and increases with increasing patch shape irregularity). These landscape metrics were calculated of all 15 the different spatial scales (200 to 3000 m), using a distance of 200 m between each circumference, which allowed us to evaluate the influence of the composition and landscape configuration on the community of solitaries bees and wasps (more details about these metrics are provided in McGarigal et al., 2012). Landscapes with more-irregular patches, presenting greater interspersion between classes, can facilitate the flow of organisms between these classes (Moreira et al., 2015), and is justifies the choice of AWMSI as a configuration metric. The proportion of crop land is a landscape composition metric that has been demonstrated to be related to negative effects on several biological groups and ecosystem services (Garibaldi et al., 2011; Benjamin et al., 2014;; Flores et al. 2019), even though crops can supply complementary resources for other biological groups, such as some bee species (Klein et al., 2007). It is relevant to evaluate this variable orthogonally to landscape diversity, thereby disassociating the effects of each descriptor on the abundance and richness of wasps and solitary bees as well as the rate of parasitism, what was guaranteed during sample site selection. These metrics were calculated in the Patch Analyst module ArcGIS 9.3 ESRI 2008 (Rempel et al., 2012) and the Maximum Likelihood Classification algorithm, which is available in the software ENVI 4.7 ITT 2009. Linear regressions were performed to detect the best fit scale for each response variable for each descriptor of the surrounding landscape (Table S2) (Steffan-Dewenter 2002; Steffan-Dewenter et al., 2002). For these regressions, we employed the lm function in R 3.4.4 (R Development Core Team 2018).

Structural heterogeneity of vegetation at the local scale

In addition to the structural complexity of the landscape, the structural complexity of vegetation at the local scale may be crucial for the establishment of nests by bees and solitary wasps, which are central-place foragers and highly mobile arthropods (Shackelford et al., 2013), as well as their parasitoids (Kendall and Ward 2016). Therefore, we measured the complexity of vegetation in the immediate surroundings of the selected sampling units.

To calculate the indices describing the local vegetation structure, we established a plot of 20 x 10 m in the immediate surroundings of the traps used to capture wasps and bees, extending 50 m north of the sampling unit. We measured the circumference at ground level (CGL), the number of branches and the height of all plants with a CGL greater than or equal to 3 cm. These values were used to calculate the volume of woody vegetation (m3 ha^-1; hereafter, vegetation volume), which is an estimation of the quantity of wood available, and the total number of branches in the plot, which is an estimate of the structural complexity of vegetation (Table S3). These indices were estimated with Fitopac 2.1.2.85 software. Additionally, plant richness and abundance, calculated inside the plot were used as indicators of plant diversity for each sampling unit. These variables have historically been important predictors of the taxonomic and functional composition of wasps (Kendall and Ward et al., 2016) and trap-nesting bees (Loyola and Martins, 2008). To evaluate the effects of local vegetation on the wasps richness and abundance, as well as kleptoparasitism rates, we adopted a remote sensing technique to measure plant richness, vegetation structure and productivity using the two-band enhanced vegetation index (EVI2). This index was calculated from the physical reflectance values of bands corresponding to the red and near-infrared wavelengths from LANDSAT 5 satellite images taken on 06/02/2001 (wet season), which were atmospherically corrected and geometrically and radiometrically calibrated (Jiang et al. 2008). We calculated the mean vegetation index (MEVI) at multiple scales, with circular buffers ranging between 25 and 150 m with a 25 m progression from the centre of the sampling unit (Moreira et al. 2015). Moreira et al. 2015 found that the EVI2 was a good surrogate measure of the number of branches and plant richness for savannah physiognomies in the studied region. For bees, we adopted plants richness and abundance as habitat local predictors measured in the field.

Sampling of solitary wasps and bees and their natural enemies

Wasps and bees were sampled with trap nests (Krombein, 1967, as modified by Taki et al., 2008). We conducted field inspections every 20 days for a period of 12 months (i.e., a total of 17 samples). Boxes were attached to wooden stakes at 1.2 m above the ground, spaced 10 m from one another (Figure S1A). Each nest box was constructed from a 2-L milk carton (9.5 cm × 9.5 cm × 16.5 cm) with a polystyrene insert (Figure S1B). Each box contained 36 paper tubes of four different sizes (3, 5, 7, or 9 mm internal diameter × 15 cm depth) and was covered with burlap and plastic tiles to provide shade and protect the nests from overheating (Taki et al., 2008). Collection of the contents of plugged tubes (indicating occupied nests) was conducted every three weeks. Occupied tubes were replaced with new empty tubes of the same size. There were always available tubes of all sizes, so that nesting cavities were not a limiting factor. The nested tubes were transported to the laboratory and checked every day until adult emergence. We recorded the number of individuals, parasites and brood cells for each tube. All parasites were kleptoparasites, which means that they take prey or food stored by wasps and bees. In the laboratory, the traps were maintained at an average temperature of 24 °C. After emergence, the adults were collected, labelled and subsequently identified to the lowest taxonomic level possible.

We used the collected data to determine bee and wasp species richness and bee and wasp brood cell numbers, where the last parameter was an indirect measurement of abundance and reproductive success (Steffan-Dewenter, 2002). The number of brood cells is more accurate indication of the reproductive investment made by females. A high number of brood cells can be explained either by a high number of females that colonized the trap nest or by a high reproduction per colonizing female (Holzschuh et al. 2010).  We classified wasps into two functional groups, lepidopteran larva predators (family Vespidae) and spider predators (Crabronidae) to detect the influence of landscape metrics and habitat descriptors on the richness and abundance of these groups, considering that they can exhibit different responses to these drivers (Holzschuh, 2009). For the functional groups of wasps, we measured abundance as the number of emerged adults.  We use the total brood cell number for this group because it was difficult to determine the species identities of nests. For the bees, it was easy to differentiate the brood cells for the species found. Parasitism rates for bees and wasps were calculated separately as the total number of parasitized nests in all sampling units / total nests per sampling unit. The specimens were deposited in the Natural History Museum of the Federal University of Bahia, in the Goeldi Museum of Pará and in the Museum of Natural Sciences of the Federal University of Paraná.

Funding

Conselho Nacional de Desenvolvimento Científico e Tecnológico, Award: 556050/2009‐6

Global Environment Facility