Bee - Pollinator Monitoring Project , Tanzania

The project is hosted at CAWM, Mweka Tanzania and is being implemented in collaboration with local and international partner institutions. Local institutions include Sokoine University of Agriculture (SUA), Tanzania Wildlife Research Institute (TAWIRI), Tanzania Commission for Science and Technology (COSTECH), Ministry of Agriculture - Tanzania, Tropical Pesticide Research Institute (TPRI) and National Museum of Tanzania (NMT). Partner institutions from outside Tanzania include the University of Würzburg (Germany), Agricultural Research Council ARC (South Africa), Royal Belgian Institute of Natural Sciences RBINS (Belgium), and the University of Kansas (USA).

Goals: The project's main goal was to determine the current distribution and status of bee pollinators in Tanzania. Other project objectives were:

• To strengthen the capacity of Tanzanians in the aspects of biodiversity informatics; plant-bee interactions; DNA-based and morphological identification techniques; and collection management;

• To develop and implement a standardized bee pollinator monitoring programme;

• To share data on bee species, abundance and their interactions with plants via dedicated databases, such as Global Biodiversity Information Facility (GBIF), Tanzania Biodiversity Information Facility (TanBIF) and African Pollinator Initiative (API);

• To disseminate results to the scientific community through peer-reviewed publications and conference presentations; and

• To raise awareness of the general public on the importance of bee pollinators through various media.

The project is financed by the JRS Biodiversity Foundation, USA.

The study was carried out in a set of study sites established in agricultural (transformed), grazing (degraded) and natural savannah (conserved) lands to represent different land-use categories as presented in Table 1. Agriculture intensity was measured, based on magnitude of land use intensification, i.e. moderately intensive agriculture habitat was mainly characterized by smallholder farms with field sizes of less than 1 ha with mixed crops, such as maize, beans and sunflower. It is also characterized by moderate use of agricultural machines and agrochemicals, while intensive agriculture was characterized by monoculture farms. In this habitat, there is a high use of heavy agricultural machines and agricultural inputs (i.e. pesticides and chemical fertilizers). Nonetheless, grazing intensity was measured, based on the visual inspection of on-site signs of obvious grazing like shortened tufts of grass, presence or absence of livestock footprints and by calculating the distance between study sites to bomas (livestock enclosures and living grounds of families holding large herds of livestock) using remote sensing and GIS techniques. Study sites with signs of intensive grazing activity were very near to bomas (average distance 0.09 ± 0.05 (SD) km while study sites with moderate livestock grazing intensity were at a far distance to bomas (i.e. average distance 25.3 ± 27.6 km (SD).

Data were collected in 40 study sites distributed along savannah, grazing and agriculture gradients in the three regions. A paired patch study design (i.e. sampling plots were positioned in two contrasting habitats within each study site) was used to minimize spatial autocorrelation. In each study site, two 50 x 50 m sampling plots were positioned and spaced at least 150 m apart. The coordinates of the plots were recorded at the mid-point between the paired plots. Bee data collection involved a combination of standardized pan trapping and random walk methods. These techniques have successfully been used for sampling bee species in northern Tanzania (Classen et al. 2015, Classen et al. 2017, Classen et al. 2020) and in other parts of the world (e.g. Noyes 1989, Stephen and Rao 2007, Westphal et al. 2008, Yi et al. 2012, Spafford and Lortie 2013). In each plot, four clusters of UV-Reflecting pan traps (each with yellow, white and blue) were installed and left in the field to collect bees for 48 hours. Two of the clusters were installed using a 120 cm pole to increase the chances of collecting bees foraging on shrubs and the other two were installed using a 35 cm pole to capture bees foraging on herbaceous plants. In each of three quota water-filled pan traps, a drop of scentless colourless liquid soap was added to break the surface tension and prevent bees from escaping. The total sampling effort for this technique summed to 1,152 hours per site. For the standardized random walk, two researchers actively collected bees for two hours within each sampling plot using sweep nets. This method summed to a sampling effort of four man-hours per study site.

Controlling data: For each of the study sites, we recorded the habitat type, GPS coordinates and elevation (metres above sea level, m a.s.l.). The coordinates and elevation of localities were derived from a hand-held Garmin GPS (Model: GPSMAP64s; resolution ± 3 m; Garmin Ltd, Taiwan). In addition, for each study site, information on weather parameters (temperature and precipitation) and forage resources were recorded. The specimens collected were preserved in 70% ethanol before being mounted and identified by afro-tropical bee taxonomists (Alain Pauly and Connal Eardley). Bees were identified following the nomenclatural system of Michener 2007 "The Bees of the World, Second Edition" with the exception of the family Halictidae that followed Pauly 1990 and Pauly 1999. Both Michener (2007) and Halictidae taxonomic publications contain keys, diagnosis and descriptions of bees. The reference collections for identified bee species are available at the CAWM, Mweka.

The study was conducted in the northern part of Tanzania i.e. Kilimanjaro, Arusha and Manyara regions (Fig. 1). The study regions are located between latitude 3°30’ S and 4°45’ S and longitude 4°30' E and 5°45’ E. The study regions have two rainy seasons: a long rainy season from March to May and a short rainy season in November and December. Average annual rainfall ranges geographically between 1300 mm and 2400 mm. Annual mean maximum temperature (hottest season) is 25.4°C between July and September and minimum temperature (cold season) is 12.8°C between May and June.

Figure 1.  

Map of the study area showing study sites. (a) Location of Tanzania (pale yellow background) on the map of Africa; (b) Location of the study area (pale yellow background) in Tanzania; (c) Enlarged map of the study area showing sampling sites (grey dots) in northern Tanzania i.e Kilimanjaro, Arusha and Manyara regions.

3°30’ S and 4°45’ S Latitiude and ; 4°30' E and 5°45’ E Longitude.

This data paper describes a total of 953 occurrences for bee species representing four families, 20 genera and 45 species (Table 2), amongst 20,507 species that have been described worldwide (Ascher and Pickering 2020). Seven families of bee species (Andrenidae, Halictidae, Apidae, Melittidae, Colletidae, Megachilidae and Stenotridae) are currently recognized globally (Michener 2007), though only four (Andrenidae, Apidae, Halictidae and Megachilidae) have been recorded in this study. In this sample, seven species (Apis mellifera (Linnaeus, 1758), Macrogalea candida (Smith, 1879), Lasioglossum bowkeri (Cockerell, 1920), L. rubritarse (Cockerell, 1937), L. transvaalense (Cameron & Cockerell, 1937), Seladonia foana (Vachal, 1899) and S. hotoni (Vachal, 1903) are reported to occur across all land-use types: agricultural (transformation), grazing (degradation) and natural savannah (conservation), whereas other species are found in a subset of land-use types (Table 3).

The Halictidae was richest in species, with 27 species, followed by Apidae with 16 species. Two families (Andrenidae and Megachilidae) were represented by single species: Andrena notophila (Cockerell, 1933) and Lithurgus pullatus (Vachal, 1903), respectively (Table 3). Greater numbers of records from Kilimanjaro (511 occurrences), compared to Arusha (410 occurrences) and Manyara (32 occurrences) is attributed to more sample plots in the region and not fewer bee species in Arusha or Manyara regions.

These data can be freely used, provided their source is cited.