Keywords: masturbation, postcopulatory sexual selection, pathogen avoidance, Bayesian phylogenetics

(a) . Data assembly

We assembled the largest dataset of masturbation occurrence to date, compiled from nearly 400 sources, including 246 publications and 150 questionnaire responses and personal communications from primatologists and zoo-keepers [1,2]. We combined this dataset with a posterior distribution of 10 000 molecular phylogenies spanning the primate order, from the 10k Trees Project [23]. Our data covered 105/272 species (38.6%), 54/67 genera (80.6%) and 18/19 (sub)families (94.7%) within our phylogeny. The proportion of species from each major radiation for which we have data varies, and likely reflects a research bias towards larger-bodied primates. Thus, data were available for 19.7% of lemurs, 33.3% of lorises and galagos, 20.0% of tarsiers, 41.8% of platyrrhine monkeys, 40.4% of catarrhine monkeys and 68.2% of apes.

The data cover both sexes, with data for females covering 67.2% of genera (n_species = 49), and males covering 76.1% of genera (n_species = 83). This allowed us to run all analyses separately for females and males, to account for potential sex differences.

Both wild and captive primates are represented in our dataset, though reports of masturbation are more common for captive primates (n = 298) than wild individuals (n = 105). However, given that many species are more commonly studied in captivity, the absolute number of reports is not necessarily informative. A more appropriate measure is to compare how many studies on captive versus wild primates report masturbation, or the lack thereof. In our dataset, masturbation is reported to be present in 74.5% of studies on captive females, and 87.4% of studies on captive males, versus 35.4% of studies on wild females, and 73.3% of studies on wild males.

(b) . Phylogenetic signal and ancestral state reconstructions

Both female and male masturbation showed strong phylogenetic signal indicating that species with more recent shared ancestry have more similar trait values (electronic supplementary material, table S1).

Ancestral state reconstructions of female masturbation (electronic supplementary material, table S2;figure 1; Bayesian rjMCMC ancestral state reconstructions;n_species = 49) indicate that masturbation was present in ancestral (non-tarsier) haplorrhines (node D; mean probability = 0.76), catarrhines (node F; mean probability = 0.91), apes (node G; mean probability = 0.95) and catarrhine monkeys (node H; mean probability = 0.87). Other nodes were reconstructed with less confidence, with the model indicating that masturbation was probably present in the ancestral haplorrhine (including tarsiers; node C; mean probability = 0.65) and platyrrhine monkey (node E; mean probability = 0.63), but absent in the ancestral strepsirrhine (node B; mean probability = 0.67). The estimation of female masturbation occurrence at the root was equivocal (node A; mean probability of masturbation presence = 0.55).

Ancestral state reconstructions of male masturbation (electronic supplementary material, table S2;figure 1; Bayesian rjMCMC ancestral state reconstructions;n_species = 83) followed the same broad pattern. Masturbation was present in ancestral (non-tarsier) haplorrhines (node D; mean probability = 0.74), catarrhines (node F; mean probability = 0.92), apes (node G; mean probability = 0.95) and catarrhine monkeys (node H; mean probability = 0.97). The ancestral platyrrhine monkey node was reconstructed with less confidence, but masturbation was likely present (node E; mean probability = 0.67). Compared to females, ancestral male haplorrhines (including tarsiers; node C; mean probability of masturbation presence = 0.58), and strepsirrhines (node B; mean probability of masturbation presence = 0.49), were reconstructed with lower confidence. In keeping with females, the estimation of male masturbation occurrence at the root was equivocal (node A; mean probability of masturbation presence = 0.56).

(c) . Postcopulatory selection hypothesis

Different mating systems vary in the level of postcopulatory selection and associated behavioural and morphological phenotypes they generate [24]. For example, multi-male mating systems produce high levels of postcopulatory male–male competition, and are associated with large testes to body mass ratio [5,25] and elongated bacula [24]. We therefore employ mating system and testes mass (controlled for body size) as proxies for postcopulatory selection, using the compilation of data in Brindle & Opie [24].

We found support for coevolution between masturbation presence and mating system in male, but not female, primates, providing support for the Postcopulatory Selection Hypothesis in males (electronic supplementary material, table S3;figure 2; rjMCMC models of correlated versus independent evolution;n_species = 299; male masturbation and mating system, log BF = 2.44; female masturbation and mating system, log BF = −2.40). Examination of evolutionary transitions revealed that, in male primates, shifts from masturbation absence to presence occurred in both single- and multi-male mating systems (Z = 18.8% and 0.2%, respectively; whereZ is the percentage of models where a transition did not take place, indicating the likelihood of a given transition), though more frequently in multi-male mating systems. However, masturbation was also lost frequently in single-male mating systems (Z = 0.0%), but almost never in multi-male mating systems (Z = 98.7%). This indicates that—while masturbation is a very labile trait in single-male mating systems—in multi-male mating systems, once male masturbation has evolved, it persists.

We did not find evidence for a relationship between masturbation and adult male testes mass in either female or male primates (electronic supplementary material, table S4;figure 3).

(d) . Pathogen avoidance hypothesis

Many parasites require hot, humid conditions or water, to complete their life cycles. This may explain why global parasite diversity tends to increase towards the equator, and is closely linked to climatic variables [26,27]. For this reason, we employ (i) primate geographical region (temperate versus tropical), (ii) mean annual precipitation, and (iii) environmental harshness (the degree of climatic predictability) as a proxy for pathogen load. We also invoke the more direct measure of sexually transmitted pathogen occurrence in wild populations, gleaned from the Global Primate Parasite Database (GPPD) [28], which was categorized as either present or absent for a given species.

We found very strong evidence for coevolution between masturbation and pathogen occurrence in males, but not females, providing support for the Pathogen Avoidance Hypothesis in males (electronic supplementary material, table S3;figure 4; rjMCMC models of correlated versus independent evolution;n_species = 299; male masturbation and pathogen occurrence, log BF = 11.25; female masturbation and pathogen occurrence log BF = – 1.22). Evolutionary path analyses indicate that, in male primates, shifts from masturbation absence to presence occurred regardless of whether pathogens were absent or present (Z = 0.6% and 26.7%, respectively). However, masturbation was lost at a very high rate when pathogens were absent (Z = 0.0%) but almost never when they were present (Z = 99.8%). This indicates that masturbation is a labile trait when pathogens are absent, but when pathogens are present masturbation is retained.

By contrast, our analyses with respect to environmental predictors of pathogen risk failed to find any patterns. We found no evidence in support of coevolution between masturbation and geographical region in female or male primates (electronic supplementary material, table S3). Similarly, there was no support for a relationship between masturbation and mean annual precipitation or environmental harshness, in either females or males (electronic supplementary material, table S4;figure 3).

To confirm that our results were not an artefact of potential coevolution between pathogen presence and mating system, we also checked these variables for coevolution, but did not find support for this model (electronic supplementary material, table S3).

(a) . Masturbation data

We define masturbation as the self-stimulation of the anogenital or breast region, carried out with an individual's own body parts or external tools [1,2]. This operational definition treats internal motivations as a black box, albeit without ascribing to the stance that mental causes are absent.

Data were compiled from the literature, as well as questionnaire responses. Particular care was taken to solicit information from primatologists and zoo-keepers who had extensive experience with taxa for which published literature provided little or no information about masturbation (details in [1,2]). Species with a single confirming scientific report were considered to be masturbatory. Thus, reports of masturbation presence overrode those of masturbation absence, regardless of the number of times absence was reported (i.e. if 10 people said a species did not masturbate, but one individual confirmed they did, this species would be classed as masturbatory). Conversely, a species was not simply classified as non-masturbatory if there were a lack of reports on the behaviour, because behavioural studies of many primate species are limited, and masturbation may be rare, so anecdotal incidences are unlikely to be published. Instead, we only included cases which explicitly state that the behaviour was not observed.

(b) . Postcopulatory selection data

Primate species were classified according to whether their mating system and anatomy were likely to be associated with high or low postcopulatory competition. In the first case, species with single-male mating systems (monogamy, polygyny) were considered as having low postcopulatory competition, while species with multi-male mating systems (polygynandry, polyandry) were considered to have high postcopulatory competition, following Brindle & Opie [24]. Species with intraspecific variability that routinely spans both sides of the binary grouping were discarded from the analyses. In the second case, species with larger adult male testes mass (while controlling for body size) were identified as experiencing higher postcopulatory selection [25]. Mating system (single-male versus multi-male;n_species = 181) and testes mass (n_species = 75), were therefore employed as proxies for postcopulatory selection pressure. These data were taken from Brindle & Opie [24].

(c) . Pathogen load data and environmental proxies

Two different methods were adopted to evaluate the risk of infection from sexually transmitted pathogens: a direct approach, in which we assessed the frequency with which species were recorded to carry sexually transmitted pathogens; and an indirect approach, in which we assessed environmental correlates of pathogen abundance. Data on environmental measures were available for a wide sample of primate species [39], while the sample for sexually transmitted pathogens was less extensive [28].

In the first approach, a measure of sexually transmitted pathogen load in wild populations was gleaned from the Global Mammal Parasite Database v. 2.0 (GMPD; note that here ‘parasite' also refers to viral, bacterial, and fungal pathogens [40]). Data were filtered to include only sexually transmitted pathogens, and only those primate species included in the 10k Trees GenBank phylogenies. Data in which primates were only labelled to genus were also removed. While it is unlikely that pathogens are truly absent across a species, for the purposes of our analyses, we marked them as such if all reported pathogen screens were negative in at least 10 individuals (see electronic supplementary material for further discussion).

In the second approach, given the limited availability of data on pathogen load, we assessed the environmental correlates of pathogen infection risk. In the light of previous studies (e.g. [26,27]), we identified three potential predictors. These comprised (i) geographical region (temperate versus tropical;n_species covered = 7 and 117, respectively), (ii) mean annual precipitation (n_species covered = 147) and (iii) environmental harshness (n_species covered = 147) as proxies for pathogen load. These data were collected from Boteroet al. [39], who measure precipitation as the mean annual precipitation averaged across a species' geographical range, and environmental harshness as the amount of exposure to drier, less productive environments, with colder, and more variable annual temperatures (with higher scores indicating harsher environments). Only data for primate species included in the 10k Trees GenBank phylogenies were collated from the database. Analyses of the environmental correlates were restricted to only include species for which data were available for all variables within a model.

(d) . Phylogenies

Comparative analyses must incorporate phylogenetic information to control for the confounding effects of shared ancestry between species, given that cross-taxa data are not statistically independent [41]. We therefore conducted phylogenetic analyses across a posterior distribution of 10 000 molecular phylogenies following GenBank taxonomy, from the 10k Trees Project v. 3.0 [23]. Where a single tree was required, we generated a maximum clade credibility tree from our distribution of 10 000 molecular ultrametric phylogenies, using TreeAnnotator (part of the BEAST software package [42]).

(e) . Phylogenetic signal

The phylogenetic signal of female and male masturbation was calculated using the ‘phylo.d' function within the R package ‘caper' [43,44].

(f) . Ancestral state reconstructions

We used a Bayesian MCMC framework to reconstruct the evolutionary history of masturbation across the primate order for females and males, using the ‘MultiState' function in BayesTraits v. 3.0 [45]. The analyses were carried out at the species level, with missing data included as such, so that nodes could take one of two states (masturbation present versus masturbation absent). A reversible-jump hyperprior approach was employed, seeding from an exponential distribution with a range of 0.00–0.05. Hyperpriors allow the details of the prior distribution to be estimated from the data, rather than ana priori assumption, which is useful where investigators have little or no information about the mean and variance of rate coefficients [46]. Each model was run three times, and the results were checked for equivalence. We then selected the model with the median log marginal likelihood (following [47]). Models were run for 5 000 000 iterations, with a burn in of 50 000 iterations. Ancestral states of interest were reconstructed via the ‘Add MRCA’ function in BayesTraits v. 3.0 [45]. The nodes reconstructed were the most recent common ancestors of (A) all primates, (B) strepsirrhines, (C) haplorrhines, (D) haplorrhines excluding tarsiers, (E) platyrrhine monkeys, (F) catarrhines, (G) apes and (H) catarrhine monkeys. Using this approach, the probability of a given state at any one node is generated. We consider probabilities > 0.70 as providing high confidence of a given state, > 0.60 as low confidence and < 0.60 as equivocal (following [47]).

(g) . Correlated evolution

Two binary traits (such as masturbation presence versus absence, and single-male versus multi-male mating systems) can produce four different combinations of observable states within a species across a phylogeny. If traits have coevolved, the rate of change between two states should depend on the background state of the other [46]. If the traits followed an independent pattern of evolution, the rate of change of one state should be independent of the other. By combining information on ancestral states with posterior rate distributions (i.e. transition rates), it is possible to discern the probable coevolutionary pathway two traits have taken [46]. We tested for coevolution between (i) masturbation and mating system and (ii) masturbation and pathogen presence, in both female and male primates. In addition, we checked for coevolution between mating system and pathogen presence, to exclude the possibility that the independent variables were influencing one another.

To do this, we employed reversible-jump Markov chain Monte Carlo (rjMCMC) techniques in BayesTraits v. 3.0 [46] to simultaneously test for coevolution between binary variables and estimate transition rates between states. This method fits continuous-time Markov models to discrete traits, allowing them to transition to another state at any time. The rjMCMC visits each model in proportion to its posterior probability, searching for the best-fitting model to describe the coevolution of two traits across a phylogeny. We compared the fit of a dependent model of evolution, in which transition rates were free to vary, switching ‘on' and ‘off', to that of an independent model, where transition rates between states were constrained to be equal. Transition rates were estimated by taking the mean posterior rate estimate across all posterior models for which a given transition was active. We also calculated the proportion of iterations in which each parameter was estimated to be zero (i.e. when a transition was ‘switched off') within the rjMCMC model.

To ensure these estimates were accurate and stable, with little run-to-run variation, each chain was run three times. If the chains converged well, the model with the median log marginal likelihood was chosen (following [47]). Markov chains were run for 5 000 000 iterations, after a burn-in of 500 000 samples. We employed a reversible-jump hyperprior approach, seeding from an exponential distribution of 0–0.5. Model fit was established using Log Bayes Factors (BFs) = 2(m_D–m_I). Natural log marginal likelihoods of the dependent (m_D) and independent (m_I) models were estimated using a stepping-stone sampler set with 100 stones, each run for 10 000 iterations, and the default parameters ofα = 0.4 andβ = 1. BFs were interpreted following Kass & Raftery [48]: 0–2, minimal support; 2–6, positive support; 6–10, strong support; greater than 10, very strong support.

(h) . Phylogenetic logistic regressions

We built Bayesian phylogenetic binary logistic regression models using thebrms package in R [44,49]. We modelled masturbation presence as a binary response (present versus absent) separately for two different environmental fixed effects (mean annual precipitation and environmental harshness), and one postcopulatory fixed effect (testes mass). All analyses were run for female and male primates separately, resulting in four environmental models and two postcopulatory models. We accounted for phylogenetic relationships in all models by including a variance-covariance matrix of phylogenetic distance as a random effect. We included adult male body mass as a fixed effect in our postcopulatory models to control for the effects of body size on testes mass, rather than using residuals from a separate regression between the two variables as data, which can lead to biased parameter estimates [50].

We ran four chains for 10 000 iterations, with a burn-in of 5000 iterations using default priors. Delta was adapted to 0.9999 and maximum tree depth set to 12. Model performance was evaluated by visually assessing trace plots as well as examining R-hat values (with those <1.01 considered to have mixed well) and effective sample sizes (Bulk ESS should be greater than 100 times the number of chains, so in our case greater than 400). We considered statistical differences to be substantial if the 95% Bayesian credible intervals did not overlap zero, and moderate if 80% Bayesian credible intervals did not overlap zero.

• 1.Brindle M.2022. Autosexual behaviour in primates: form, phylogeny and function. Doctoral thesis, University College London, London, UK. [Google Scholar]
• 2.Sommer V, Thomsen R, Brindle M. 2022. Masturbation in primates. In The Cambridge handbook of evolutionary perspectives on sexual psychology (ed. Shackelford T). Cambridge, UK: Cambridge University Press. [Google Scholar]
• 3.Dixson AF. 2012. Primate sexuality: comparative studies of the prosimians, monkeys, apes, and human beings. Oxford, UK: Oxford University Press. [Google Scholar]
• 4.Roth L, Briken P, Fuss J. 2022. Masturbation in the animal kingdom. J. Sex Res.59, 1-13. ( 10.1080/00224499.2022.2044446) [DOI] [PubMed] [Google Scholar]
• 5.Dixson AF, Anderson MJ. 2004. Sexual behavior, reproductive physiology and sperm competition in male mammals. Physiol. Behav.83, 361-371. ( 10.1016/j.physbeh.2004.08.022) [DOI] [PubMed] [Google Scholar]
• 6.Brindle M. 2018. Postcopulatory selection. In Encyclopedia of animal cognition and behavior (eds Vonk J, Shackelford T), pp. 1-4. Cham, Switzerland: Springer International Publishing. [Google Scholar]
• 7.Wikelski M, Baurle S. 1996.Proc. R. Soc. B263, 439–444. ( 10.1098/rspb.1996.0066) [DOI]
• 8.Thomsen R, Soltis J. 2004. Male masturbation in free-ranging Japanese macaques. Int. J. Primatol.25, 1033-1041. ( 10.1023/B:IJOP.0000043350.75897.89) [DOI] [Google Scholar]
• 9.Pound N, Javed MH, Ruberto C, Shaikh MA, Del Valle AP. 2002. Duration of sexual arousal predicts semen parameters for masturbatory ejaculates. Physiol. Behav.76, 685-689. ( 10.1016/S0031-9384(02)00803-X) [DOI] [PubMed] [Google Scholar]
• 10.Baker RR, Bellis MA. 1993. Human sperm competition: ejaculate adjustment by males and the function of masturbation. Anim. Behav.46, 861-885. ( 10.1006/anbe.1993.1271) [DOI] [Google Scholar]
• 11.Dubuc C, Coyne SP, Maestripieri D. 2013. Effect of mating activity and dominance rank on male masturbation among free-ranging male rhesus macaques. Ethology119, 1006-1013. [DOI] [PMC free article] [PubMed] [Google Scholar]
• 12.Meston CM. 2000. The psychophysiological assessment of female sexual function. J. Sex Educ. Therapy25, 6-16. ( 10.1080/01614576.2000.11074323) [DOI] [Google Scholar]
• 13.Suarez SS, Pacey AA. 2006. Sperm transport in the female reproductive tract. Hum. Reprod. Update12, 23-37. ( 10.1093/humupd/dmi047) [DOI] [PubMed] [Google Scholar]
• 14.Puts DA, Dawood K. 2006. The evolution of female orgasm: adaptation or byproduct?Twin Res. Hum. Genet.9, 467-472. ( 10.1375/twin.9.3.467) [DOI] [PubMed] [Google Scholar]
• 15.Johnson M, Everitt BJ. 1988. Essential reproduction. Oxford, UK: Blackwell Scientific Publications. [Google Scholar]
• 16.Mortimer D. 1989. Sperm transport in the human female reproductive tract. Oxf. Rev. Reprod. Biol.5, 30-61. [Google Scholar]
• 17.Waterman JM. 2010. The adaptive function of masturbation in a promiscuous African ground squirrel. PLoS ONE5, e13060. ( 10.1371/journal.pone.0013060) [DOI] [PMC free article] [PubMed] [Google Scholar]
• 18.Dixson AF, Mundy NI. 1994. Sexual behavior, sexual swelling, and penile evolution in chimpanzees (Pan troglodytes). Arch. Sex. Behav.23, 267-280. ( 10.1007/BF01541563) [DOI] [PubMed] [Google Scholar]
• 19.Nunn CL, Altizer S. 2004. Sexual selection, behaviour and sexually transmitted diseases. In Sexual selection in primates: new and comparative perspectives (eds Kappeler PM, Van Schaik CP), pp. 117-130. Cambridge, UK: Cambridge University Press. [Google Scholar]
• 20.Tan M, Jones G, Zhu G, Ye J, Hong T, Zhou S, Zhang S, Zhang L. 2009. Fellatio by fruit bats prolongs copulation time. PLoS ONE4, e7595. ( 10.1371/journal.pone.0007595) [DOI] [PMC free article] [PubMed] [Google Scholar]
• 21.Donovan B. 2000. The repertoire of human efforts to avoid sexually transmissible diseases: past and present. Part 2: strategies used during or after sex. Sex. Transm. Infect.76, 88-93. ( 10.1136/sti.76.2.88) [DOI] [PMC free article] [PubMed] [Google Scholar]
• 22.Hart BL, Korinek E, Brennan P. 1987. Postcopulatory genital grooming in male rats: prevention of sexually transmitted infections. Physiol. Behav.41, 321-325. ( 10.1016/0031-9384(87)90395-7) [DOI] [PubMed] [Google Scholar]
• 23.Arnold C, Matthews LJ, Nunn CL. 2010. The 10kTrees website: a new online resource for primate phylogeny. Evol. Anthropol. Issues News and Rev.19, 114-118. ( 10.1002/evan.20251) [DOI] [Google Scholar]
• 24.Brindle M, Opie C. 2016. Postcopulatory sexual selection influences baculum evolution in primates and carnivores. Proc. R. Soc. B283, 20161736. ( 10.1098/rspb.2016.1736) [DOI] [PMC free article] [PubMed] [Google Scholar]
• 25.Harcourt AH, Harvey PH, Larson SG, Short RV. 1981. Testis weight, body weight and breeding system in primates. Nature293, 55-57. ( 10.1038/293055a0) [DOI] [PubMed] [Google Scholar]
• 26.Nunn CL, Altizer SM, Sechrest W, Cunningham AA. 2005. Latitudinal gradients of parasite species richness in primates. Divers. Distrib.11, 249-256. ( 10.1111/j.1366-9516.2005.00160.x) [DOI] [Google Scholar]
• 27.Guernier V, Hochberg ME, Guégan JF, Harvey P. 2004. Ecology drives the worldwide distribution of human diseases. PLoS Biol.2, e141. ( 10.1371/journal.pbio.0020141) [DOI] [PMC free article] [PubMed] [Google Scholar]
• 28.Nunn CL, Altizer SM. 2005. The global mammal parasite database: an online resource for infectious disease records in wild primates. Evol. Anthropol. Issues News Rev.14, 1-2. ( 10.1002/evan.20041) [DOI] [Google Scholar]
• 29.Dittrich L. 1968. Beobachtungen zur Masturbation bei Säugetieren im Zoologischen Garten. Zoologischer Anzeiger180, 299-312. [Google Scholar]
• 30.Thomsen R, Soltis J, Teltscher C.. 2003. Sperm competition and the function of male masturbation in non-human primates. In Sexual selection and reproductive competition in primates: new perspectives and directions, Special Topics in Primatology, vol. 3 (ed. Jones CB), pp. 436-453. n.l.: American Society of Primatologists. [Google Scholar]
• 31.Edström AM, Malm J, Frohm B, Martellini JA, Giwercman A, Mörgelin M, Cole AM, Sørensen OE. 2008. The major bactericidal activity of human seminal plasma is zinc-dependent and derived from fragmentation of the semenogelins. J. Immunol.181, 3413-3421. ( 10.4049/jimmunol.181.5.3413) [DOI] [PMC free article] [PubMed] [Google Scholar]
• 32.Schulze M, Jakop U, Schröter F, Herrmann C, Leiding C, Müller K, Jung M, Czirják GÁ. 2020. Antibacterial defense in bull and boar semen: a putative link to the microbiome and reproductive strategy?Theriogenology157, 335-340. ( 10.1016/j.theriogenology.2020.07.033) [DOI] [PubMed] [Google Scholar]
• 33.Levin RJ. 2003. The ins and outs of vaginal lubrication. Sexual Relationship Therapy18, 509-513. ( 10.1080/14681990310001609859) [DOI] [Google Scholar]
• 34.Nunn CL. 2003. Behavioural defences against sexually transmitted diseases in primates. Anim. Behav.66, 37-48. ( 10.1006/anbe.2003.2130) [DOI] [Google Scholar]
• 35.Nunn C L. 2011. The comparative approach in evolutionary anthropology and biology. Chicago, IL: University of Chicago Press. [Google Scholar]
• 36.Gewirtzman A , Bobrick L, Conner K, Tyring SK. 2011. Epidemiology of sexually transmitted infections. In Sexually transmitted infections and sexually transmitted diseases (eds Gross C, Tyring S K), pp. 13-34. New York, NY: Springer. [Google Scholar]
• 37.Patterson JE, Ruckstuhl KE. 2013. Parasite infection and host group size: a meta-analytical review. Parasitology140, 803-813. ( 10.1017/S0031182012002259) [DOI] [PMC free article] [PubMed] [Google Scholar]
• 38.Rowe N, Myers M. 2017. All the world's primates. Charlestown, RI: Pogonias Press. [Google Scholar]
• 39.Botero CA, Dor R, McCain CM, Safran RJ. 2014. Environmental harshness is positively correlated with intraspecific divergence in mammals and birds. Mol. Ecol.23, 259-268. ( 10.1111/mec.12572) [DOI] [PubMed] [Google Scholar]
• 40.Stephens P R, et al. 2017. Global Mammal Parasite Database version 2.0. Ecology98, 1476-1476. [DOI] [PubMed] [Google Scholar]
• 41.Felsenstein J. 1985. Phylogenies and the comparative method. Am. Nat.125, 1-15. ( 10.1086/284325) [DOI] [Google Scholar]
• 42.Drummond AJ, Rambaut A. 2007. BEAST: Bayesian evolutionary analysis by sampling trees. BMC Evol. Biol.7, 1-8. ( 10.1186/1471-2148-7-214) [DOI] [PMC free article] [PubMed] [Google Scholar]
• 43.Orme D, Freckleton R, Thomas G, Petzoldt T, Fritz S, Isaac N, Pearse W. 2018. Caper: comparative analyses of phylogenetics and evolution in R.R package version 1.0.1. Seehttps://CRAN.R-project.org/package=caper.
• 44.R Core Team. 2020. R: a language and environment for statistical computing. Vienna, Austria: R Foundation for Statistical Computing. (https://www.R-project.org/) [Google Scholar]
• 45.Meade A, Pagel M. 2017. BayesTraits: a computer package for analyses of trait evolution. Version 3. Seehttp://www.evolution.reading.ac.uk/BayesTraitsV3/BayesTraitsV3.html.
• 46.Pagel M, Meade A. 2006. Bayesian analysis of correlated evolution of discrete characters by reversible-jump Markov chain Monte Carlo. Am. Nat.167, 808-825. ( 10.1086/503444) [DOI] [PubMed] [Google Scholar]
• 47.Jordan FM, Gray RD, Greenhill SJ, Mace R. 2009. Matrilocal residence is ancestral in Austronesian societies. Proc. R. Soc. B276, 1957-1964. ( 10.1098/rspb.2009.0088) [DOI] [PMC free article] [PubMed] [Google Scholar]
• 48.Kass RE, Raftery AE. 1995. Bayes factors. J. Amer. Stat. Assoc.90, 773-795. ( 10.1080/01621459.1995.10476572) [DOI] [Google Scholar]
• 49.Bürkner PC. 2017. brms: an R package for Bayesian multilevel models using Stan. J. Stat. Softw.80, 1-28. ( 10.18637/jss.v080.i01) [DOI] [Google Scholar]
• 50.Freckleton RP. 2002. On the misuse of residuals in ecology: regression of residuals vs. multiple regression. J. Anim. Ecol.71, 542-545. ( 10.1046/j.1365-2656.2002.00618.x) [DOI] [Google Scholar]
• 51.Brindle M, Ferguson-Gow H, Williamson J, Thomsen R, Sommer V. 2023. The evolution of masturbation is associated with postcopulatory selection and pathogen avoidance in primates. Figshare. ( 10.6084/m9.figshare.c.6653124) [DOI] [PMC free article] [PubMed]

Data Availability Statement

The data and code are provided in the electronic supplementary material [51].