Breeding ground food availability is critical to the survival and productivity of adult birds. The common cuckoo Cuculus canorus is a brood-parasitic Afro-Palearctic migrant bird exhibiting long-term (breeding) population declines in many European countries. Variation in population trend between regions and habitats suggests breeding ground drivers such as adult food supply. However, cuckoo diet has not been studied in detail since before the most significant population declines in Europe began in the mid-1980s. 20th century studies of cuckoo diet largely comprised field observations likely to carry bias towards larger prey taxa. Here we demonstrate the potential value of 1) using high-throughput DNA sequencing of invertebrate prey in faeces to determine cuckoo diet with minimal bias towards large prey taxa, and 2) using crowd-sourced digital photographs from across Britain to identify lepidopteran cuckoo prey taxa during recent years post-decline (2005-2016). DNA analysis found a high frequency of Lepidoptera, including moths of family Lasiocampidae, prominent within the past literature, but also grasshoppers (Orthoptera) and flies (Diptera) that may be overlooked by field observation methodologies. The range of larval lepidopteran prey identified from photographs largely agreed with those previously documented, with potential signs of reduced diversity, and identities of key adult prey taxa were supported by molecular results. Notably, many identified cuckoo prey taxa have shown severe declines due to agricultural intensification, suggesting this has driven spatial patterns of cuckoo loss. Landscape-scale, lowland rewilding interventions provide opportunities to understand the scale of reversal of previous agricultural intensification that may be necessary to restore prey populations sufficiently to permit recolonization by cuckoos.

Study site and faecal sample collection

We collected faeces from common cuckoo adults in Dartmoor National Park, UK (Fig. 1) between April and June 2017 (and one bird ringed in May 2016). 52 samples were obtained by collecting fresh faeces dropped by birds using natural perches (trees, shrubs and rocks). Seven further samples were collected through licensed mist-netting. Within Dartmoor, Warren House (50.61194°N; 3.87028°W) and Burrator (50.52333°N; 4.01528°W) were selected as valleys with scattered trees, with abundant perches and short-grazed grass beneath, facilitating detection of cuckoos and their faeces. Sites were walked from 06:00 am when weather conditions were not wet and windy. Once located, each cuckoo was watched until defecation took place. Faeces were searched for within a 3 m radius of the estimated location of defecation, and collected within 20 minutes, allowing time to watch for further defecation events or to note the bird’s new location if it moved. All visibly fresh faeces found were collected in separate 8 ml vials containing 90% ethanol and refrigerated at 5°C within 12 h. For data analysis, we pooled all faecal samples collected from the same 3 x 3 m area at the same time, into groups referred to as ‘sampling events’.

Mist-netting was carried out at Holne Moor, Dartmoor (50.52717°N; 3.86433°W). from 04:00 am using a triangular array of three 18 m mist-nets (30 x 30 mm mesh) (Ecotone, Sopot, Poland) with a decoy female cuckoo at the centre and a sound lure of male and female cuckoo calls, at a BTO guideline volume (Blackburn et al. 2006). Each trapped bird was placed for ten minutes in a veterinary carry case with a clean, laminated card floor. All faecal material deposited was collected using the same protocols as above.

DNA extraction and sequencing

DNA was extracted using a CTAB-based protocol. DNA sequencing of a 290 base-pair (bp) region of the mitochondrial COI DNA was used to confirm bird species (Hebert et al. 2004, Kerr et al. 2009) from the faecal samples, unless collected by mist-netting, together with a 157 bp region to identify prey taxa using arthropod COI primers (Zeale et al. 2011). Libraries for a total of 48 samples confirmed as originating from cuckoo by sequencing or collection during mist-netting, plus controls, were prepared in triplicate and 300 paired-end sequenced using the Illumina MiSeq v3 SBS reagents (Illumina Corp, USA). Details of the methods used to extract the DNA, confirm the species, and prepare, sequence and identify taxa are provided in supplementary materials.

Sequencing data analysis

DNA sequence reads from Illumina MiSeq runs were de-multiplexed, and trimmed to remove primer sequence, duplicates and low quality regions (Ewing & Green 1998, Trevelline et al. 2016). Dada2 (Callahan et al. 2016) was used to identify amplicon sequence variants (ASVs). The sequence and total number of reads of each ASV (n = 268) in each replicate PCR from faecal samples were exported into a summarising table. The mean number of unique ASVs read per sample was 14.49 (range 3-38, supplementary materials Table A1). For taxonomic identification, ASVs were entered as search queries in the Barcode of Life Database (BOLD) version 4 (Ratnasingham & Hebert 2007) Species Level Barcode Records. We noted that Lepidoptera and other species previously given as cuckoo prey in the literature (e.g. Wyllie 1981) have accessioned sequences in BOLD. Sequences from samples were considered to match a database sequence if similarity exceeded 98% (following Clare et al. (2011) and King et al. (2015)). The species with the highest percentage similarity was taken to be the prey species of origin (following King et al. (2015)). 160 ASVs matched eukaryote taxa, accounting for 1,864,169 reads on the MiSeq platform. Sequences matching prokaryotic taxa were excluded. 108 read sequences of appropriate length (>150 bp) for which no match could be found through BOLD, were recorded as the aggregate result 'No Match'. UK records for each top-matching species were searched online using Google. Species with no previous records of UK occurrence were highlighted (following King et al. (2015)). Sequences relating to these species were assumed to originate from a UK-occurring species within an encompassing higher taxon and were included in calculations of frequency of taxa on this basis.

Crowd-sourced digital photography

The search terms “cuckoo” and “Cuculus” were entered in the search tools of Flickr (http://www.flickr.com), Birdforum Gallery (http://www.birdforum.net/gallery/ search.php), Birdguides Gallery (http://www.birdguides.com/ gallery/), Rare Bird Alert gallery (http://www.rarebirdalert.co.uk/RealData/gallery. asp) and Google Images (http://images.google.com). Images had to meet the following criteria: i) contained a full-grown, nutritionally independent common cuckoo handling an identifiable and wild-caught prey item and ii) identified the country and month in which the image was recorded. Plumage was used to classify each bird as juvenile or adult. Prey items in images were identified to the most specific possible taxonomic level, with all identifications made by one person (BH) with relevant taxonomic expertise (Henwood & Sterling 2020). The geographic distribution of the images extended throughout Britain, from 58 locations comprising both inland and coastal sites, though images in Scotland were more clustered to coastal and island locations and few images were available from Wales. The appearance, date, location and accompanying information were used to identify images of the same bird and prey item (referred to in subsequent text as ‘prey handling events’). Photographs of 44 prey handling events in adult cuckoos, and 24 in juveniles, met the criteria for inclusion in analysis.

NCBI Sequence Read Archive under BioProject number PRJNA606798