Gliding is only present in six extant groups of mammals – interestingly, despite divergent evolutionary histories, all mammalian gliders are strictly nocturnal. Gliding mammals also seem to have relatively high rates of ultrasound use and ultraviolet-induced photoluminescence (UVP) in contrast with their close relatives. Therefore, we hypothesized that, despite diverging lineages, gliding mammals use similar modes of cryptic communication compared to their non-gliding counterparts. We developed two datasets containing the vocal range (minimum-maximum of the dominant harmonic; kHz) and UVP of 73 and 82 species, respectively; we report five novel vocal repertoires and 57 novel observations of the presence or absence of UVP. We complemented these datasets with information about body size, diel activity patterns, habitat openness, and sociality to explore possible covariates related to vocal production and UVP. We found that the maximum of the dominant harmonic was significant higher in gliding mammals when vocalizing than their non-gliding relatives. Additionally, we found that nocturnality was the only significant predictor of UVP, consistent with the previous hypothesis that luminophores primarily drive UVP in mammal fur. In contrast, however, we did not find UVP ubiquitous in nocturnal mammals, suggesting that some unknown process may contribute to variation in this trait.

Vocalizations

We developed a database beginning with a list of publications describing gliding mammal vocalizations (summarized in Table S1). The minimum requirement for each publication was describing at least one call with either a spectrographic analysis or numerical data. However, most publications described multiple call types per species or multiple species per publication (7 gliding mammals represented across 9 publications, summarized in Table S1). The databases used to search for these publications were Google Scholar, JSTOR, Web of Science, and Wiley Online Library. We used the keywords acoustics, acoustic repertoire, calls, frequency, Hz, vocalizations, and ultrasound paired with an exhaustive list of currently valid and invalid genera (the most updated nomenclature was taken from the Integrated Taxonomic Information System). Across all published calls, we took the absolute minimum and maximum frequencies (kHz) of the dominant harmonic for the final analyses (this often corresponded to the fundamental harmonic, if multiple harmonics were present [1]). For noisy calls, such as broadband calls, where the harmonics are not well defined, we estimated the minimum and maximum of the loudest parts of the call. We did not include calls produced by neonates or juveniles as there is evidence of some frequencies and calls being different in younger individuals [2,3].

To compare gliding mammals to closely related species, we systematically searched for vocalization data using the same methodology described above (Fig. 1). Flying squirrels are unique amongst the gliders as there are many extant species that occupy the same family (Sciuridae); therefore, we kept all relatives from the same subfamily (Sciurinae) and a random subset of squirrels from the other subfamilies (26 squirrels across 62 publications). Other gliders have few extant relatives and we strategically chose taxa that shared similar evolutionary histories and traits. For the scaly-tailed gliders, we selected springhares (Pedetes capensisi), the only other extant taxa of the Anomaluromorpha suborder, and a variety of small-bodied rodents (12 species across 16 publications) exhibiting a range of vocal frequencies (maximum dominant frequency range: 9.86 (Sicista subtilis [4]) - 109.8kHz (Mus musculus [5]).For colugos, the only extant members of the order Dermoptera, we selected tree shrews (Tupaia belangeri) which form a sister clade with Dermoptera [6] and similarly sized taxa from the order Primates (19 primates across 27 publications), which are the next closest sister taxa [7]. For marsupial gliders, we expanded our search to include similarly sized taxa of the order Diprotodontia as there were few records of marsupial vocalizations (5 marsupials across 6 publications). The vocalization data for two marsupial gliders (Petaurus breviceps, and P. norfolcensis) and two glider relatives (Pedetes capensis and Pseudocheirus peregrinus; Fig. 1) were not available in the literature, and we worked with co-authors and collaborators to develop novel call descriptions  for our study (methods in Article S1). We also provide vocalization data from free-ranging Petaurus australis (methods in Articles S1) to opportunistically contrast our recordings with previously reported calls in the literature [8,9]. In the literature, four species were represented by a single subspecies only: Otolemur garnettii lasiotis, Petaurista alborufus lena, Sciurus aberti kaibabensis, and Sciurus niger rufiventer.

Ultraviolet-Induced Photoluminescence

To expand on our vocalization dataset, we assessed the UVP of pelage for 83 species. Previous literature accounted for 19 species in our dataset; we sampled an additional 64 species from the mammal collections at the Canadian Museum of Nature and the Royal Ontario Museum (one mounted specimen (Sicista subtilis), otherwise all dry-preserved pelts; specimen and museum information provided on Dryad). We sampled species from the vocalization dataset preferentially. However, we opportunistically added ten species (bold type in Table S1) to increase the sample size of luminescing species. We followed the same vocalization protocol detailed above for both opportunistic and previously published UVP species; we found vocalization data for eight additional species (four opportunistic and four from previous UVP literature).

We used a Vansky UV flashlight (395nm wavelength) to illuminate museum specimens (held 75 cm above the individual) and a Huawei P30 Pro phone (held directly beside the light) to capture any luminescence. A yellow gel filter was held in front of the camera lens to reduce the input of purple-blue light [10,11]. To minimize the additional yellow hue created by the filter, we manually color-corrected the photos in Photoshop (details in Info. S1). We took pictures of each specimen’s ventral and dorsal sides under white-light conditions, UV-light only, UV-light + filter, and UV-light + filter + correction (example provided in Fig. 2; complete photoset available on Dryad). We additionally photographed a few live Glaucomys individuals trapped in the Kawartha Highlands, Ontario, following the same protocol (animals studied under Trent University animal care protocol 27909).

In our investigation, some species expressed visible photoluminescence in white pelage or in some cases, the white ends of guard hairs. While “white” UVP has been noted in some species, this “white” coloration has been reported as a bluish-white (as seen in the striped possum (Dactylopsila trivirgata) and some marsupial gliders [13]). The underlying cause of UVP expressed as distinct colors have been linked to porphyrins (red or pink) or tryptophan metabolites (cyan, blue, lavender [13]). However, the expression of exclusively “white” coloration is not commonly reported, nor has a clear explanation been proposed for producing UVP without a dominant color. Furthermore, white human hair may emit a bluish hue similar to the pelage of minks, rabbits and goats and sheep, which have been described as being photoluminescent due to the presence of tryptophan metabolites [14]. Given that we could not photograph museum specimens in complete darkness, the available visible light may have excited white hairs that would otherwise not express UVP. Therefore, to remove the potential bias of visible light, we removed individuals that only expressed “white” photoluminescence (but model outputs for all species, including those with “white” UVP, are included in Table S2). While UVP varied dramatically in color (e.g., pink, blue/green), placement, and patterning across museum specimens and published literature, we reduced variability to absence/presence to increase the sample size in each category.

Dataset Assembly

Once we had assembled our database of vocalizing mammals with UVP records, we searched for the body mass (g), diel activity pattern (diurnal or nocturnal), social complexity, and habitat openness of the dominant habitat (open or closed) of each species. We preferentially took these data from the relevant vocalization or UVP papers, though this information was rarely provided; therefore, other resources, including articles and online databases such as Mammalian Species accounts and the Animal Diversity Web [15], were reviewed to complete our dataset. If a range was provided for the body mass, we took the mean of the given values; we took a mean of male and female body masses as we were not capturing the effect of sex on vocalization frequencies or UVP. Social variability was reduced to social or solitary living to reduce model parameters; species that exhibit dynamic social structures, where adult individuals will seasonally or cyclically shift between solitary and social living (e.g., flying squirrels engaging in social nesting during the winter only [16]), were treated as socially living.

Phylogeny

While multiple subspecies were present in the vocalization dataset, we calculated the vocalization maxima at the species level for the final dataset and analyses (Fig. 1; subspecies-specific information noted in Table S1). Only one subspecies was excluded from analyses (Peromyscus maniculatus bairdii) due to a binary variable inconsistency with the parent species – this subspecies only occurs in open habitats [17], while the parent species is most commonly found in closed habitats. In addition, Masters et al. [18] recently proposed that the Paragalago genus is a distinct clade from the Galagoides genus to which the Paragalago species had been previously assigned; we reassigned these species accordingly.

For the final species dataset (n = 93), we pruned 1000 node-dated completed trees from the mammalian supertree on VertLife, an online database used to produce pruned random distribution trees of vertebrate species [19]. The nexus outputs were compiled into a consensus tree using the phytools [20] package in R [21] (Fig. 1). Petaurus notatus is a recently described species (previously incorporated within P. breviceps), and therefore, it was the only species not available on Vertlife; we incorporated this species into the final consensus tree by splitting the P. breviceps lineage at a divergence time of 1Ma [22].

Analyses

We built phylogenetic generalized least square (PGLS) models to account for variation in the vocal range that could be explained by phylogenetic relatedness. PGLS models estimate phylogenetic relatedness as lambda (λ), which varies between 0 (no phylogenetic trace) and 1 (absolute Brownian motion) [23,24]. Full models were built for each frequency limit (β₀ + body mass (β_Mass) + diel activity pattern (β_Diel) + sociality (β_Sociality) + habitat openness (β_Open) + UVP (β_UVP)) using the caper [25] package in R [21]. We reported the regression coefficient estimates (x̄ ± SE) to evaluate significance and effect size (F-statistic, P-value, and adjusted R²).

We also built a phylogenetic generalized linear mixed (PGLM) model for binary data using the ape [26] package in R to assess the presence of UVP. This binary PGLM model accounted for variation in UVP while dealing with the bimodal distribution that violates other tests [27]. The same independent variables were used (β₀ + β_Mass + β_Diel + β_Sociality + β_Open); β_Mass was standardized to have a mean of 0 and variance of 1, while the categorical variables were reconstructed into dummy variables (2 categories = 0, 1) for the PGLM model. We standardized the variables to improve the interpretation of regression coefficients as they more accurately represent the effect size of the independent variables (Ives & Garland, 2010). The PGLM model represents the phylogenetic signal (s²) as the scalar magnitude of the phylogenetic variance across all species comparisons (estimated from the phylogenetic variance-covariance matrix [27]).

References

[1] Newar SL, Bowman J. 2020. Think Before They Squeak: Vocalizations of the Squirrel Family. Frontiers in Ecology and Evolution 8:1–15. DOI:10.3389/fevo.2020.00193.
[2] Nikol’skii AA. 2007. A Comparative Analysis of The Alarm Call Frequency in Different Age Rodent Groups (in Russian). Zoologicheskii Zhurnal 86:499–504.
[3] Schneiderová I, Schnitzerová P, Uhlíková J, Brandl P, Zouhar J, Matějů J. 2015. Differences in alarm calls of juvenile and adult European ground squirrels (Spermophilus citellus): findings on permanently marked animals from a semi-natural enclosure. Zoo Biology 34:503–512.
[4] Volodin IA, Panyutina AA, Abramov A V., Ilchenko OG, Volodina E V. 2019. Ultrasonic bouts of a blind climbing rodent (Typhlomys chapensis): acoustic analysis. Bioacoustics 28:575–591. DOI: 10.1080/09524622.2018.1509374.
[5] Hoffmann F, Musolf K, Penn DJ. 2012. Ultrasonic courtship vocalizations in wild house mice: spectrographic analyses. Journal of Ethology 30:173–180.
[6] Nie, W, Fu B, O’Brien PCM, Wang J, Su W, Tanomtong A, Volobouev V, Ferguson-Smith MA, Yang F. 2008. Flying lemurs - the ‘flying tree shrews’? Molecular cytogenetic evidence for a Scandentia-Deromptera sister clade. BMC Biology 6:18. DOI: 10.1186/1741-7007-6-18.
[7] Beard KC. 1993. Phylogenetic systematics of the Primatomorpha, with special reference to Dermoptera. In: Szalay FS, Novecaek MJ, McKenna MC eds. Mammal Phylogeny: Placentals. Verlag: Springer, 129–150. DOI: 10.1007/978-1-4613-9246-0_10.
[8] Kavanagh RP, Rohan-Jones WG. 1982. Calling behaviour of the yellow-bellied glider, Petaurus australis Shaw (Marsupialia: Petauridae). Australian Mammalogy 59:95–111. DOI: 10.1071/AM82011.
[9] Whisson DA, McKinnon F, Lefoe M, Rendall AR. 2021. Passive acoustic monitoring for detecting the yellow-bellied glider, a highly vocal arboreal marsupial. PLOS ONE 16:e0252092. DOI: 10.1371/journal.pone.0252092.
[10]Kohler AM, Olson ER, Martin JG, Anich PS. 2019 Ultraviolet fluorescence discovered in New World flying squirrels (Glaucomys). Journal of Mammalogy 100:21–30. (doi:10.1093/jmammal/gyy177)
[11] Nummert G, Ritson K, Nemvalts K. 2023 Photoluminescence in the Garden dormouse (Eliomys quercinus). Zoology 157. (doi:10.1016/j.zool.2023.126075)
[12] Reinhold L. 2021 Mammals with fluorescent fur: Observations from the Wet Tropics. North Queensland Naturalist 51:1–8.
[13] Reinhold LM, Rymer TL, Helgen KM, Wilson DT. 2023. Photoluminescence in mammal fur: 111 years of research. Journal of Mammalogy. DOI: 10.1093/jmammal/gyad027.
[14] Millington KR. 2020 Anomalous fluorescence of white hair compared to other unpigmented keratin fibres. International Journal of Cosmetic Science 42:289–301. (doi:10.1111/ics.12614)
[15] Myers P, Espinosa R, Parr C, Jones T, Hammond G, Dewey T. 2023. The Animal Diversity Web (online). Available at https://animaldiversity.org (accessed 18 September 2023).
[16] Garroway CJ, Bowman J, Wilson PJ. 2013 Complex social structure of southern flying squirrels is related to spatial proximity but not kinship. Behavioural Ecology and Sociobiology 67:113–122. (doi:10.1007/s00265-012-1431-3)
[17] Wecker SC. 1963 The Role of Early Experience in Habitat Selection by the Prairie Deer Mouse, Peromyscus maniculatus bairdi. Ecological Monographs 33:307–325.
[18] Masters JC, Génin F, Couette S, Groves CP, Nash SD, Delpero M, Pozzi L. 2017 A new genus for the eastern dwarf galagos (Primates: Galagidae). Zoologial Journal of the Linnean Society 181:229–241.
[19] Upham N, Esselstyn J, Jetz W. 2019 Inferring the mammal tree: species-level sets of phylogenies for questions in ecology, evolution, and conservation. PLOS Biology 17:e3000494.
[20] Revell LJ. 2012 phytools: an R package for phylogenetic comparative biology (and other things). Methods of Ecology and Evolution 3:217–223. DOI: 10.1111/j.2041-210X.2011.00169.x.
[21] R Core Team. 2022 R: A Language and Environment for Statistical Computing. 
[22] Cremona T, Baker AM, Cooper SJB, Montague-Drake R, Stobo-Wilson AM, Carthew SM. 2021 Integrative taxonomic investigation of Petaurus breviceps (Marsupialia: Petauridae) reveals three distinct species. Zoological Journal of the Linnean Society 191:503–527.
[23] Freckleton RP, Harvey PH, Pagel M. 2002 Phylogenetic analysis and comparative data: A test and review of evidence. American Naturalist 160:712–726. DOI: 10.1086/343873.
[24] Martin K, Tucker MA, Rogers TL. 2016 Does size matter? Examining the drivers of mammalian vocalizations. Evolution 71:249–260. DOI: 10.1111/evo.13128.
[25] Orme D, Freckleton R, Thomas G, Petzoldt T, Fritz S, Isaac N, Pease W. 2018 caper: Comparative Analyses of Phylogenetics and Evolution in R. 
[26] Paradis E, Schliep K. 2019 ape 5.0: an environment for modern phylogenetics and evolutionary analyses in R. Bioinformatics 35:526–528. DOI: 10.1093/bioinformatics/bty633.
[27] Ives AR, Garland T. 2010 Phylogenetic Logistic Regression for Binary Dependent Variables. Systematic Biology 59:9–26. DOI:10.1093/sysbio/syp074.

R is required to open UltrasoundandUltraviolet_Script.R and UltrasoundandUltraviolet_Trees.nex 
Microsoft Excel can be used to view UltrasoundandUltraviolet_Data.csv and UltrasoundandUltraviolet_MuseumSpecimens.csv 
Any PDF viewer can be used to view UltrasoundandUltraviolet_UVP.pdf