Step 1: Obtaining the pollen sample
Pollen can be obtained from a range of environmental samples. Most often it is sampled from insects, either directly from the body (Chang et al., 2018; Pornon et al., 2016; Suchan, Talavera, Saez, Ronikier, & Vila, 2019), or through provisions in nests (Gresty et al., 2018; McFrederick & Rehan, 2016), honey (de Vere et al., 2017; Jones, Brennan, et al., 2021), or honeycomb (Tanaka, Nozaki, Nakadai, Shiwa, & Shimizu-Kadota, 2020). Alternatively, pollen may be sampled from the air (Brennan et al., 2019; Leontidou et al., 2018). How, where, and when a sample is collected affects the number and diversity of target species sampled and the plant richness and diversity captured. For example, when sampling directly from a pollinator’s body, researchers should consider the influence of body size and the pilosity of insects on the number and diversity of pollen retained (Cullen et al., 2021). Likewise, capturing foraging insects by hand-netting may be limited by surveyor bias. Some sampling methods are specific to particular taxonomic groups (e.g., honey sampling or pollen trapping; Judd, Huntzinger, Ramirez, & Strange, 2020). The temporal scale of a pollen sample should also be considered. Pollen from an individual represents a single foraging trip, whilst pollen trapping usually entails pooling samples so multiple trips from multiple bees can be sampled over a short period. Pollen provisions in solitary bee nests represent multiple trips by one individual. The temporal scale of airborne samplers is dependent on study design. The sampling period defines the knowledge which can be gained (e.g., sampling pollen from social bees with a long flight period during one month of the year does not give a global picture of resource use). The sampling period should also consider floral phenology, as pollen will only be present when the plants are in flower.
The quantity of pollen required for the selected sequencing method, and the quantity of pollen available per sample will determine whether samples need to be pooled. Pooling of samples should be done with caution, as it could decrease the accuracy of diet estimations relative to the separate analysis of samples, especially for detecting species that occur at relatively low abundance (Mata et al., 2019). The number of pollen grains present in a sample will vary according to the sample type. For pollen collected directly from the bodies of pollinators, this will vary according to pollinator species. Estimates show that Hymenoptera can carry relatively high numbers of pollen grains (Apis mellifera and Bombus lucorum carry 2100-2500 pollen grains; Bombus pascuorum , ~800 grains), Diptera carry fewer (Syrphidae ~ 500 grains; Empididae ~80 grains), and Lepidoptera carry fewer again (~20 grains), while the few Coleoptera observed carried up to 200 grains (Pornon et al., 2016). The number of samples collected, whether they are pooled, the number of sampling events across time and space, and the replication level within each of these samples will depend on the questions being addressed. Once the pollen samples are obtained, they seem to be relatively resilient and easily storable over extended periods under different conditions without introducing biases (e.g., frozen, room temperature, silica gel dried; Quaresma et al., 2021).
Control samples should be used to validate the methodology for each study system. These include negative controls, ideally including field-collected and laboratory blanks, to assess for contaminating DNA from external sources, and positive controls or mock communities to demonstrate that the sequencing method is effective, and if relevant, quantitative, for the species in the study system. Contamination should also be minimised through good laboratory practice. Extra precautions can be taken for specific sample types, such as using separate collection materials (i.e., nets, vials) to avoid cross-contamination between specimens during collection (Pornon et al., 2016). For some sample types, contamination will be unavoidable. For example, some types of nest pollen from solitary bees may be contaminated by other plant DNA during nest construction (Gresty et al., 2018; Alexander Keller et al., 2015; McFrederick & Rehan, 2016). Finally, analysis of negative controls will allow for detection and removal of contaminating sequences during the bioinformatics step (e.g., Davis, Proctor, Holmes, Relman, & Callahan, 2018).