What influences prey sequence recovery?
There are many factors that can influence the recovery of prey sequences and thus the success of the brute force approach. The length of the target sequence is important, with small amplicons being favored due to a longer half-life and hence an increased detectability (Kamenova et al., 2018). Whether the extracts are based upon recently ingested or heavily digested materials (e.g. feces) will also have an impact on recovery (Kamenova et al., 2018). For small invertebrates, whole-body extraction is us usually the only feasible alternative. With whole-body samples, we find it probable that the consumer DNA has the greatest impact on prey sequence recovery as it competes for amplification and detection during PCR and sequencing. The severity of the overabundance problem is difficult to predict in advance, however, given that different consumers have variable genome sizes, cell numbers and target gene copy-numbers. Indeed, the ratio of the prey to consumer sequence may also vary from one season to the next, between different sexes, feeding or life stages. The exact ratios acquired here may thus not be representative for studies of other invertebrates although the study design is otherwise identical. Nonetheless, we show that the problem of consumer DNA overabundance may be overcome by sequencing deeper.
With a great overabundance of putative consumer DNA (98%), our results underline the problem of dietary samples from small animals. In fact, over 3.8 billion of the sequence reads belonged to the top three abundant zOTUs, recruiting 1.6, 1.4 and 0.8 billion reads, respectively. Although we could not resolve the taxonomy of these zOTUs beyond Maxillopoda, we are confident that these represent the three sampled copepod species just based on the abundance of their read counts. The first sequence variant dominated samples of Oithona similis , the second dominated samples from Microcalanus spp., and the third dominated samples from Microsetella norvegica . Other abundant maxillopod zOTUs with similar distribution patterns were also identified. These may be artifacts of the dominant sequences, real variants of the gene sequenced, or remnants of copepod prey. As we are unable to distinguish between these alternatives, we conservatively chose to discard all Maxillopoda sequence reads from our analyses.
It is however not just consumer DNA that is problematic as contaminants may further dilute samples, making finding enough prey data more difficult. In the datasets analyzed in this study, we observed more reads from the four extraction negatives sequenced for the pilot dataset (10.8 mill. reads) than in the combined twenty extraction negatives from the full dataset (9.3 mill. reads). This holds also at the level of real samples, given that the pilot dataset contains a higher contaminant percentage (1.6%) than the full dataset (1.1%). If we somehow “worked cleaner” during preparation of the second sequencing run is not known, but we hypothesize that increased contamination could explain some of the observed gap in prey recovery (0.4% in pilot versus 1.2% in full dataset), since also contaminant sequences may compete for amplification and sequencing. A lot of non-prey sequences would inevitably lead to a lower output of prey, and the more contaminated samples are, the greater the problem. Future dietary studies may arguably increase the yield of prey sequences by paying attention to lab routines and making them as clean as possible.
Other possible influences on prey sequence recovery could be technical – for instance that deeper sequencing leads to a greater number of rare prey zOTUs surpassing a set cutoff-value (we denoised each sample individually with a four reads cutoff), or related to the biology of the consumer. Several Arctic copepod species may for example enter diapause prior to the polar night. A dormant state ensues, where energy spent on motility, reproduction and feeding is drastically lowered (Conover, 1988). Because internal wax ester storages are used for energy, feeding activities are no longer required. Consequently, some copepod species may be expected to contain little or no prey DNA during certain periods of the year. The copepods studied here are not expected to enter diapause, but may to some extent reduce metabolic rates during winter (Conover & Huntley, 1991).