ultimate drivers of forced extra-pair copulations in birds lacking penises: jackdaws as a case-study

Forced copulation is common, presumably because it can increase male reproductive success. Forced extra-pair copulation (FEPC) occurs in birds, even though most species lack a penis and are widely thought to require female cooperation for fertilisation. How FEPC persists, despite a presumed lack of siring success and likely non-negligible costs to the male, is unknown. Using the jackdaw (Corvus monedula) as a case study, we use SNPs to quantify extra-pair paternity rate through FEPC and evaluate explanations for the persistence of FEPC in species without a penis. We then collate evidence for FEPC across penis-lacking birds. Combining genetic and behavioural analyses, our study suggests that the most likely explanations for the maintenance of FEPC in jackdaws are that it provides a selective advantage to males, or that it is a relic. Our literature review shows that across birds lacking a penis, FEPC is taxonomically widespread, yet little is known about its evolution. Broader implementation of the approach used here, combining both genetic and behavioural data, may shed light on why this widespread sexual behaviour persists. Additional work is necessary to understand whether a penis is needed for paternity through forced copulation, and to quantify the costs of FEPC.

Methods
Behavioural data

We fitted cameras with microphones (380TVL CMOS camera, Handykam, Redruth, U.K.) inside jackdaw nest-boxes during the 2014, 2015, 2018 and 2019 breeding seasons. Cameras were set to record starting from between 6AM and 11AM BST (n = 575 recordings). This captures the period when our previous observations, and other studies (Gill et al. 2020), indicate the majority of sexual behaviour occurs. In addition, a small subset set to record in the afternoon (n = 8 recordings; 13:30PM - 14:30PM BST). Videos were recorded at three different stages of the breeding season: the nest building stage, the incubation stage, and the nestling stage, with one to three videos per pair per stage (total number of pairs = 130). The nest building stage was filmed from when building had begun and a nest ‘cup’ was becoming apparent (Hahn et al., 2021) until the nest was complete; the incubation stage was filmed 2 – 10 days after the female’s fertility window until hatching, and the nestling stage was filmed at 4 – 10 (nestling 1) and 17 – 23 (nestling 2) days following the first egg hatching. Note that videos were filmed for various research purposes, and in-line with these protocols, no videos were filmed during the female’s fertile period (5 days pre-clutch initiation to the penultimate lay date; (Gill et al., 2020; Henderson et al., 2000)).

Sampling, DNA Extraction and Sequencing

For jackdaw chicks that survived until 25 days old, blood samples for DNA extraction were collected at ringing. In 2018 and 2019, nest-boxes with chicks <25 days old were monitored closely so that deceased chicks could be collected before removal from the nest by parents. Deceased chicks were frozen at -20℃ as soon after death as possible, and tissue samples from the liver were collected for DNA extraction. DNA was extracted from blood using Thermo Scientific GeneJET Whole Blood Genomic DNA Purification, and from liver using QIAGEN DNeasy Blood & Tissue Kit, both following manufacturer’s protocols. See Supplementary Materials for further details of DNA extraction and quality control.

Following DNA extraction, samples were selected based on fully sampled family units (i.e. broods where both social parents had been sampled), site (sample number was proportional to site size), and quality of DNA extraction. Five duplicate samples (from the same DNA extraction) were included in the sequencing pool in order to estimate approximate sequencing error rate. We also included known social half-siblings (known through re-pairing of the social parent) so we could examine whether our final analysis had the power to detect half-sib relationships, which is key to estimating extra-pair paternity rates. A female with no known offspring or parents was also included, because she had been observed for several years to associate with a pair who were included (along with their offspring) in the sample selection. We also had video evidence of this female laying an egg in that pair’s nest-box (see Methods: behavioural data).

Our final sample for sequencing consisted of 188 individuals (plus five duplicates) across two sites. 149 individuals were included from Sites Y and Z (treated as one site due to gene flow; see Methods: Bioinformatic Pipeline), comprising 113 (plus two duplicate) offspring from 47 broods and 22 sibships. Two parent samples were also duplicates. From Site X, 43 individuals were included, which comprised of 33 (plus one duplicate) offspring from 13 broods and six sibships. Quality control of DNA extracted from liver samples revealed a moderate level of degradation in many samples. Therefore, only six samples from deceased chicks were included in the analysis; all other samples were extracted from blood.

Samples were sent to Exeter Sequencing Services for library preparation and ddRAD sequencing. See Supplementary Materials for detailed library prep and sequencing methods.

Bioinformatic pipeline

Sequence data underwent strict quality filtering, presented in detail in the Supplementary Materials. Following quality filtering, reads were run through process radtags in STACKS v2.54 (Rochette et al., 2019). Reads were then aligned to the jackdaw reference genome (GenBank accession JABDSK000000000; Weissensteiner et al., 2020) using GSNAP/GMAP v2020-10-27 (Wu & Nacu, 2010) specifying a maximum of ten mismatches (-m 10), an indel penalty (-i 2) and turning off terminal alignments (–min-coverage = 0.95). Only reads that aligned uniquely were retained (-n 1, --quiet-if-excessive) (Paris et al., 2017).

Following alignment, we retained samples with >1 million aligned reads. These were run through the ref_map module of STACKS, specifying a minor allele frequency of 0.15 (--min-maf 0.15), retaining one random SNP per read (--write-random-snp) and retaining only SNPs present in >50% of samples (-r 0.5). Finally, we used PLINK v1.9 (Purcell et al., 2007) to discard loci in linkage disequilibrium. To do this, we considered all individuals without sampled parents as ‘founders’ for Site Y + Site Z, two sites with substantial overlap and assumed gene flow (now referred to as Site YZ), and Site X, a separate site with little to no observed overlap with Site YZ. We tested for linkage disequilibrium using --indep, evaluating 50 SNP windows, five SNPs at a time with a variance inflation factor cut-off of 2 (Levine et al., 2019), and filtered out SNPs in linkage disequilbrium from our final datasets.