Ethical statement
This research adhered to the Association for the Study of Animal Behaviour/Animal Behaviour Society Guidelines for the Use of Animals in Research, was approved by the University of Liverpool’s Animal Welfare Committee, and carried out under a UK Home Office Licence.
Subjects
Wild house mice (M. musculus domesticus) were captive-bred F1-F3 animals from a large outbred colony originating from five different populations in the northwest of England, UK. Animals were housed in standard rodent cages (40 × 23.5 × 12.5 cm, North Kent Plastic Cages Ltd, UK), with Corn Cob Absorb 10/14 substrate and paper wool bedding material, and ad libitum access to food (LabDiet 5002 Certified Rodent Diet, Purina Mills, St Louis, MO, USA) and water. Controlled environmental conditions were maintained at 20 to 21°C, 45% to 65% relative humidity, and a reversed 12–12 hour light cycle (lights off at 0800).
Four breeding groups were established to produce offspring for release into outdoor enclosures. Breeding groups contained two sets of three unrelated females, with each set of females housed for 14 days with an unrelated male (unrelated animals shared no parents or grandparents). This breeding plan was designed to simulate a naturalistic situation in which dominant males could sire offspring with several unrelated females resulting in discrete local populations [43]. At weaning (four weeks), offspring were given subcutaneous radio-frequency identification tags (RFID), and a small tissue sample was taken from the tip of the tail (1 to 2 mm) under general anaesthesia (halothane) for genotyping. Animals were then separated into single sex sibling groups and housed under controlled environmental conditions (see above). Social and environmental experience was thus carefully controlled with no opportunity for mating prior to release. All resulting offspring from each breeding group were released together simultaneously as founders into four separate outdoor enclosures, at age 48 to 65 days. In total, 48 male and 33 female founder mice were released to create four separate breeding populations (A: 7 males, 6 females; B: 18 males, 12 females; C: 12 males, 7 females; D: 11 males, 8 females). To investigate relationships with reproductive success, male morphological traits were measured for a sample of 24 focal males selected randomly from the four populations (A: 2/7 males, B: 11/18 males, C: 6/12 males, D: 5/7 males, see below).
Population enclosures
Outdoor enclosures (25×10 m) contained long grass as ground cover and sufficient space for all released animals to establish territories. Thirty nest boxes were provided within each enclosure and ten concrete block shelters (45 × 45 × 35 cm) were added after 12 weeks for additional shelter. Ten food and water stations, spaced evenly around the outer walls of each enclosure, provided ad libitum food (Lab Diet 5002 Certified Rodent Diet) and water. Sheet-aluzinc walls (1.3 m high with concrete foundations) prevented escape or contact between populations, and wire mesh upper walls and roof prevented predation. Mice were left to breed undisturbed for 15 to 19 weeks after release, allowing sufficient time for females to rear up to three litters to independence. Each of the four populations was then removed from the enclosures by live trapping. Sex, weight and age class (adult, sub-adult, juvenile) were recorded for all captured animals. Individually tagged ‘founder’ animals were identified and returned to the laboratory, with males housed individually under controlled environmental conditions (cages 48.5 × 11.5 × 12 cm, see above). Descendants of the founder animals were culled humanely under halothane anaesthetic and tail snips were taken for genotyping to establish parentage.
Parentage and mating assignment
Reproductive success of founder males was quantified by assigning parentage to F2 offspring (distinguished from potential F3 descendants on the basis of their weight, sex and date of capture) and is described in detail elsewhere [43]. Briefly, all F1 and F2 individuals were genotyped for a set of 24 unlinked microsatellite markers with individuals assigned to parent pairs using maximum-likelihood methods implemented in CERVUS v3.0 [44] and further validated with reference to major histocompatibility complex (MHC) and major urinary protein (MUP) genotyping [43].
Offspring were assigned to litters for each founder female based on sex-specific growth curves. Methods are described in detail elsewhere [43]. Briefly, for each female, we plotted offspring weight against capture date and used the sex-specific growth curves from captive F2 animals together with paternity to assign offspring to separate litters. We took a conservative approach and only assigned offspring from the same sire to separate copulations if they were very unlikely to come from the same litter (based on their body mass and a maximum litter size of 9 for wild house mice). For focal males that did not sire offspring in the outdoor enclosures, subsequent pairings with females from the same population confirmed normal fertility. This was achieved on recapture by pairing each focal male with a (non-sibling) female in MB1 cages (45 × 28 × 13 cm) under standard laboratory conditions (as described above), and monitoring pairs for litter production.
Measurement of baculum and other male reproductive traits
To investigate relationships between male genitalic traits and reproductive success, focal male mice (see above) were culled humanely under halothane anaesthetic. Bacula were prepared post-mortem following dissection and storage of penises at −20°C. On defrosting, bacula were thoroughly cleaned of surrounding tissue using a combination of dissection under a microscope at 20× magnification and repeated soaking in 0.05 g ml-1 KOH (24 hours × 2), and stored in 70% ethanol [17]. The baculum in house mice consists of a large proximal bone (representing the main base and shaft of the baculum) and a much smaller distal bone at its tip (Figure 1, see also [45]). In this study the small distal bone was removed during the cleaning process and measurements were made of the large proximal bone (Figure 1). This bone is relatively simple in form. Consistent with previous functional analyses of the baculum [10-18], we have therefore focused our analysis on measurements of baculum length and width - two major, independent axes of morphological variation [see Additional file 1: Tables S1 and S2]. A digital image of each bone was obtained using a flatbed scanner (CanoScanLiDE 30, Canon Inc.) at a resolution of 1,200 dpi, using a solid black background to create contrast during scanning. Images were processed using Image J software (version 1.38×, [46]) with measurements obtained digitally using the Measure, Threshold and Analyse Particles tools. Following inversion, images were converted to 32-bit and rotated if necessary to align the image on a vertical axis before setting the scale. For each baculum, we recorded maximum length (baculum length), maximum width of the base (baculum base width), and width of the shaft at its narrowest point (baculum shaft width) (Figure 1). The latter two measurements are positively correlated, whereas length and width varied independently in our dataset and so we analysed these aspects of baculum variation separately [see Additional file 1: Table S2]. Linear measurements were made to the nearest 0.01 mm (length and base width) or 0.001 mm (shaft width). Sample sizes for the different measurements vary because some bacula were damaged during processing.
Also measured at dissection were testes mass, seminal vesicles mass, and preputial glands mass (each paired, to the nearest 0.001 g). Mean body mass was calculated as an average of pre-release, post-capture and post-mortem body mass (to the nearest 0.01 g).
Statistical analysis
To investigate morphological traits influencing the total number of offspring produced by the subject males, we used generalized linear mixed models (GLMMs) with a logarithm link function and Poisson distribution, fitted using the Laplace approximation to restricted maximum likelihood estimation (lmer procedure in the lme4 R package [47]). To investigate traits influencing the average number of offspring sired per litter we used linear mixed effect models (LMEs), fitted by maximum likelihood using the lme4 package in R. Significance was assessed by comparison of the model with and without the variable of interest included, using likelihood ratio tests. For both approaches (GLMMs, LMEs), population and focal male sibling group were included as random effects, to control for relatedness of subjects and shared environmental and social conditions within the four outdoor enclosures. All morphological traits were log transformed prior to analysis to ensure normality.
In the first stage of analysis to investigate morphological traits influencing male reproductive success, we constructed separate models for each trait of interest (baculum length, shaft width and base width, testis mass, seminal vesicle mass, preputial gland mass), with body mass as a covariate. Traits with significant influence were then combined for each measure of male reproductive success to fit the best model in each case, with body mass again retained as a covariate. Results of analyses for individual traits and combined models are provided in Additional file 1: Tables S5 and S6. Preliminary analyses revealed no significant influence of variation in sex ratio of founder animals on measures of male reproductive success; hence sex ratio was not included in further analyses. Results of analyses with population as a fixed effect are provided in Additional file 1: Tables S7 and S8.
