1. Introduction
To gain a greater understanding of trophic interactions within the environment it is critical to have an in-depth understanding of species’ diet (Rytkönen et al., 2019). Our understanding of biodiversity and ecosystem functions is significantly increased by investigating spatial and temporal variation in trophic interactions within food webs (Rytkönen et al., 2019; Yu et al., 2012). Estimating diet composition can provide knowledge crucial for understanding the structure of ecological communities and the flow of energy and nutrients (Kartzinel et al., 2015; Nielsen, Clare, Hayden, Brett, & Kratina, 2017). Furthermore, characterising consumers’ resources can provide information about niche specialisation at both inter- and intra-specific scales (Kratina, LeCraw, Ingram, & Anholt, 2012). The accurate identification of components within an omnivorous diet is however, still considered challenging (da Silva et al., 2019; De Barba et al., 2014; Tercel, Symondson, & Cuff, 2021). Due to the costly, laborious and taxonomically demanding nature of exploring omnivorous diet, studies attempting to elucidate all dietary aspects are rare (Pompanon et al., 2012; Tercel et al., 2021).
Morphology-based methods of diet analysis can be time consuming and biased towards identification of distinguishable and intact undigested or semi-digested dietary items (Pompanon et al., 2012; Symondson, 2002). Additionally, morphology-based methodologies often record ingested taxa at a coarse taxonomic resolution, missing subtle differences in the taxa consumed and resulting in a lack of ability to make fine scale inferences relating to species’ ecology (Mata et al., 2016). The use of molecular techniques such as high-throughput sequencing (HTS) in conjunction with DNA barcoding, referred to as “metabarcoding” is being frequently utilised to assess the diet of a range of organisms (Cuff et al., 2021; Evens et al., 2020; Kartzinel & Pringle, 2020; Thompson & Newmaster, 2014; Zalewski, Szymura, Kowalczyk, & Brzeziński, 2021). This technique requires minimal a prioriknowledge on the dietary composition of the target organism (Alberdi et al., 2017; Valentini et al., 2009) and a wide range of ingested taxa can be identified to fine taxonomic levels (King, Read, Traugott, & Symondson, 2008). For the application of metabarcoding to study bird diet, faecal samples are highly suitable, as they contain residual dietary DNA and can be collected easily to study species which may otherwise have been difficult to locate or directly observe (Pompanon et al., 2012; Taberlet, Bonin, Zinger, & Coissac, 2018).
Within metabarcoding studies, detection of plant species have traditionally used sections of plant genes rbcL and matK , which, when used in conjunction, have the power to provide up to 75% species-level discrimination (de Vere et al., 2012). Limitations on amplicon length in HTS (maximum of 2 x 300 base pair reads on an Illumina Miseq), as well as primers designed to amplify short barcodes in order to detect DNA in degraded samples (Ando et al., 2013; Dunn et al., 2018; Pompanon et al., 2012) has resulted in these gene regions providing reduced taxonomic resolution in analysis of faecal samples (Pompanon et al., 2012). The Internal Transcribed Spacer 2 (ITS2) nuclear gene has been proposed as a suitable barcode for herbivorous dietary analysis (Moorhouse-Gann et al., 2018). Universal primers targeting the ITS2 region have been developed, producing amplicons between 187-380 base pairs (Dunn et al., 2018; Moorhouse-Gann et al., 2018). This has enabled the most variable region within the gene to be targeted, with the amplicon length suitable for use within DNA metabarcoding studies of the Turtle dove (Streptopelia turtur , Linnaeus), Pink pigeon (Nesoenas mayeri , Prévost) and Telfair’s skink (Leiolopisma telfairii , Desjardin) (Dunn et al., 2018; Moorhouse-Gann, 2017; Moorhouse-Gann et al., 2018). In a diet which contains a wide range of invertebrate taxa, DNA barcodes from the COI mitochondrial gene region have become the standard and are used in many species-level identification studies (Kress, García-Robledo, Uriarte, & Erickson, 2015). This is due to the extensive taxonomic coverage and depth within the Canadian, European, UK and USA taxonomic COI reference sequence databases (Porter & Hajibabaei, 2018). Such large databases reduce the possibility of false taxonomic assignment and improve higher taxonomic resolution (Andújar, Arribas, Yu, Vogler, & Emerson, 2018; Porter & Hajibabaei, 2018; Somervuo et al., 2017).
Despite birds being one of the most studied animal classes, few studies have used molecular techniques to improve understanding of their trophic ecology (Alonso et al., 2014). In comparison with studies on mammals, in particular bats, the application of faecal metabarcoding within passerines studies is limited. However, this is an evolving field, with research being undertaken on an expanding number of passerine species (da Silva, Mata, Lopes, Lopes, & Beja, 2020; Shutt et al., 2020). For woodland birds that forage high in the canopy and are often hidden from view, obtaining accurate identification of dietary items through observation alone is particularly challenging, and many studies simply use foraging location to infer food availability and individual fitness (Charman, Smith, Dodd, Gruar, & Dillon, 2012; Mackenzie, Hinsley, & Harrison, 2014). Within temperate environments, passerine birds often demonstrate a broad dietary range, incorporating a wide range of taxa (Cholewa & Wesołowski, 2011; Shutt et al., 2020), however, dietary variability within generalist woodland species is relatively poorly understood. It has been proposed that the diet of generalists could vary spatiotemporally, based upon resource availability and preferences of the consumer. This variation within the diet may result in geographical patterns in population density, breeding productivity, and local adaptation to resource use (Shutt et al., 2020).
The Hawfinch (Coccothraustes coccothraustes , Linnaeus) is one of many bird species closely associated with woodland habitats to have shown major declines in the UK over a period of a few decades (Kirby et al., 2018). Hawfinch breed across the Palearctic, with Britain its western range limit (Kirby et al., 2015). While little is known about Hawfinch ecology, studies show Hawfinches have declined across Britain, with the species now too rare to have regular status assessments by national annual monitoring schemes (Kirby et al., 2015). Instead, population change is inferred from distribution data compiled from bird atlas surveys (Balmer et al., 2013). These atlas data indicate a 76% reduction in the number of 10km squares occupied between 1968 and 2011 (Kirby et al., 2015, 2018) and are further evidenced by Langstonet al. (2002), who estimated a 40% population decline between the mid 1980’s to the late 1990’s. Hawfinch dietary studies are limited, with all previous information obtained through personal observations. Hawfinch are thought to be dietary specialists due to morphological adaptations (a large, powerful bill) allowing them to feed on large tree seeds such as cherry (Prunus sp., Linnaeus), hornbeam (Carpinus betulus , Linnaeus), beech (Fagus sylvatica , Linnaeus) and Wych elm (Ulmus glabra , Hudson) (Mountford, 1957). During the breeding season (typically from April to June), Hawfinch diet has been observed to include sycamore (Acer pseudoplatanus , Linnaeus), hawthorn (Crataegus monogyna , Jacquin), blackthorn (Prunus spinosa , Linnaeus), wild service tree (Sorbus torminalis , Crantz), dogwood (Cornus alba , L’Héritier), larch (Larix decidua , Miller) and oak (Quercus sp., Linnaeus) (Bijlsma, 1998; Bryant, 2011; Mountford, 1957; Tomiałojć, 2012; von Haartman, 1978). Hawfinch incorporate invertebrates into their diet, particularly during the breeding season to provide sufficient nutrition for egg production, consuming Lepidoptera, Coleoptera, Hemiptera, Annelida, Gastropoda and Araneae (Mountford 1957).
In this study we used multi-marker DNA metabarcoding to construct a high-resolution diet profile of a woodland passerine, aiming to elucidate the taxonomic composition and richness, as well as showing that HTS is a powerful approach for analysing species interactions. We hypothesised that Hawfinch diet would show variation across: (a) study sites and (b) sexes. This study also discusses the suitability of multi-marker metabarcoding approaches for exploring omnivory.