Morphological variations of auditory bullae in otomyine rodents (Rodentia: Otomyini) in southern African biomes

Mammalian middle ear cavities differ from those of other taxa as they comprise three ossicles and in rodents, can be encapsulated by an auditory bulla. In small mammals, the middle ear cavity (bulla) was found to be enlarged in the desert‐dwelling species; however, differences in bullar size could have been due to ancestry. In this study, we sampled seven species from three genera (Myotomys, Otomys, and Parotomys) of the African murid tribe Otomyini (laminated‐toothed rats), and compared the bullar volumes and shapes between the otomyine species and within the species Myotomys unisulcatus. Photographs of museum skull specimens were taken from ventral and lateral views, and the volumes of the bullae were estimated digitally from the photographs. No sexual dimorphism in bullar volumes was found in any of the species. Corrected bullar volumes were significantly different between species and larger bullae were seen in individuals inhabiting regions with lower annual rainfall. Bullar shape (estimated using geometric morphometrics) was significantly different between the genera and the species. Parotomys have tympanic meatuses that face more anteriorly compared to both, Otomys and Myotomys. When comparing bullae within M. unisulcatus, those inhabiting regions with lower annual rainfall had significantly larger bullar volumes, but no significant difference was found in bullar shape between the regions. This study shows that otomyine rodents in more xeric habitats have different auditory structures to those inhabiting wetter habitats.

Mammals have evolved a middle ear cavity containing three ossicles, and some mammal species have evolved bones that have fused around the middle ear creating a capsule known as the auditory bulla (Anthwal et al., 2013;Manley et al., 2004;Novacek, 1977Novacek, , 1993)).
Smaller auditory bullae are expected to be suited to higher-frequency sound transmission, while larger bullae are found in mammals registering lower frequencies of a broader range (Heffner & Heffner, 1992;Heffner et al., 2001).Various reasons have been proposed as to why some mammals have evolved low-frequency hearing, and include communication over long distances (Randall, 1984(Randall, , 2010;;Randall & Lewis, 1997), prey detection (Narins et al., 1997), signaling to predators (Randall, 2001(Randall, , 2010)), and in predator avoidance (Webster, 1962;Webster & Webster, 1971, 1972).Given the fact that high-frequency sound attenuates faster in air than lower frequency sound in regions of low relative humidity (Huang et al., 2022;Kinsler et al., 2000), lowerfrequency acoustic communication would likely evolve in desertdwelling species.In arid regions, mammalian populations are generally less dense and acoustic signals would need to travel longer distances (Petter, 1953(Petter, , 1962)).In comparisons of mammals and marsupials from different families, auditory bullae were found to be enlarged in desertdwelling species (Alhajeri et al., 2015;Basso et al., 2020;Mason, 2016;Nengovhela et al., 2019;Taylor et al., 2022).However, such morphological adaptations may be due to ancestral character retention, and may be limited by the ancestral phenotype.Evolution can only change what is available, and thus an investigation of bullar size in a related group of rodents would be useful to determine whether evolution of bullar size is linked with microhabitat, irrespective of ancestry.
Rodents from the tribe Otomyini (Murinae) inhabit various biomes south of the Sahara, mainly in eastern and southern Africa (Monadjem et al., 2015;Montgelard et al., 2023).The tribe consists of three genera: Otomys, Myotomys, and Parotomys.Species from the genus Otomys F. Cuvier 1824 inhabit wetter areas, such as temperate and tropical grasslands, using the grass for protection against predators (Monadjem et al., 2015).The specific validity of the subspecies of Otomys saundersiae Roberts 1929, and whether one should be synonymized with Otomys irroratus Brants, 1827, and the other to be raised to specific status (Otomys karoensis Roberts, 1931), has been under debate (Meester et al., 1986;Taylor et al., 1993Taylor et al., , 2005)).Here, we sampled O. saundersiae specimens from both the Western Cape and Eastern Cape provinces, as well as O.
irroratus, to include any individuals from this species complex.Otomys angoniensis Wroughton, 1906 is associated with dense vegetation near water sources, and will either construct nests from reeds and grass, or will dig burrows (Bronner & Meester, 1988) vocalize using a single, high-pitched alarm call (Monadjem et al., 2015).
They do not require tall grass for protection and retreat into their burrows when predators are present (Monadjem et al., 2015).The two species belonging to Myotomys Thomas, 1918(Myotomys unisulcatus [F. Cuvier, 1829] and Myotomys sloggetti [Thomas, 1902]) were once classified in the Otomys genus (Phukuntsi & Kearney, 2016), and unlike the Otomys and Parotomys genera, Myotomys species are not easily categorized under the same umbrella (Maree, 2002;Taylor et al., 2004).For one, M. sloggetti is a burrowdwelling species that is range-restricted to high-lying mountainous areas (Schwaibold & Pillay, 2010), whereas M. unisulcatus is found in almost every biome across South Africa and creates large nests above ground within bushes (du Plessis, 1989).Interestingly, the most recent phylogenetic assessment of the tribe clustered Parotomys and Myotomys species into a single clade (Montgelard et al., 2023), and suggested that both genera should be subsumed into Otomys.
M. sloggetti was sister to Parotomys and M. unisulcatus, which makes sense given the fact that the latter three species are the xeric-adapted species within the tribe (Montgelard et al., 2023).
We hypothesized that the volume of the auditory bullae will differ significantly between seven species within the Otomyini  having relatively larger bullae than those in wetter areas (temperate and tropical grasslands; interspecific differences).Due to their broad habitat range, we also hypothesized that bullar volume and shape will differ within M. unisulcatus, with those individuals inhabiting more xeric habitats having larger bullar volumes (intraspecific differences).

| Sampling
We sampled skulls housed in the Amatole Museum in King William's Town (South Africa) from seven species within the Otomyini tribe, aiming to obtain 20 adult females and 20 adult males.Unfortunately, this proved difficult as many of the skulls were damaged due to poor preservation or handling, and skulls were chosen for each specimen according to their condition-skulls that had broken bullae and/or nasal regions were not included in the analyses.Of the seven species, we managed to obtain and measure the following number of skulls per sex: M. unisulcatus-20 females, 16 males; M. sloggetti-13 females, 7 males; O. saundersiae-14 females, 16 males; O. angoniensis -20 females, 20 males; O. irroratus-19 females, 20 males; P. littledalei-20 females, 20 males; P. brantsii-19 females, 20 males (Figure 1; Supporting Information S1: Table 1).Individuals were chosen from localities that were toward the middle of the species' distributions (determined using distribution maps in Monadjem et al., 2015; Figure 2).Maps of the sampling distributions were created using the locality data associated with the museum specimens using the R package "mapview" (Appelhans et al., 2023), using the WGS84 (EPSG: 4326) coordinate reference system.Due to the generality of many of the localities provided in the museum collection (many GPS points were only given to two decimal points), some localities were not mapped to the precise locality of the collection of the specimens.
Southern Africa is biodiverse both in terms of fauna and flora.
Areas can be classified according to various criteria such as mean annual rainfall (MAR), vegetation cover and type, and average temperature (Anthwal et al., 2013).When classifying biomes, these three factors go hand-in-hand as they influence one another (e.g., temperature affects rainfall, which affects vegetation).Areas with lower rainfall (~184 mm per annum) are classified as xeric shrublands with minimal vegetation, open space, and minimal water (Jiang et al., 2017).Some areas are similar in terms of vegetation; however, they experience slightly different rainfall and temperaturetemperate grasslands (~438 mm per annum) and tropical grasslands (~936 mm per annum) both consist of grasses and shrubs that cover the land and can be considered to be rich in vegetation compared to xeric shrublands (Jiang et al., 2017).MAR data was used as a proxy for aridity and vegetation cover (obtained from Mucina & Rutherford, 2006).

| Data acquisition
A Canon EOS 750 D digital camera with a macro lens (Canon EFS 55−250 mm) was used to photograph the specimens on a grid paper.
Multiple photographs of the ventral and lateral views of the skulls were taken, slightly adjusting the orientation each time, to ensure that at least one usable photograph was obtained (i.e., ensuring standard orientation for all specimens).
Using the program ImageJ v.1.53k(Schneider et al., 2012), length and width of the bulla was measured digitally on the ventral F I G U R E 3 Top and middle: Visualization of linear measurements taken: L, length; W, width; H, height; SL, skull length (specimen AM5962 of P. littledalei shown).Bottom: Landmarks placed on the right auditory bullae of specimens.Landmarks 1, 6, 7, 8, 9, 11, and 17 were fixed landmarks, while landmarks 2−5 were placed equidistant between landmarks 1 and 6, and landmarks 12−16 were placed equidistant between landmarks 11 and 17. photographs (length = longest measurement of the bulla, width = from the anterior of the external auditory meatus to the anterior of the jugular foramen), and bullar height was measured digitally using the lateral photographs (height = tallest measurement of the bulla; Figure 3).Skull length was measured on the ventral photograph to account for allometry (Figure 3).Input files for the geometric morphometric analyses were created in tpsUtil64 v.1.81(Rohlf, 2015), and homologous landmarks (Figure 3) were placed on each specimen's ventral-view photograph in tpsDig2 v.2.32 (Rohlf, 2005).

| Linear morphometric analyses
The volume of the bulla was calculated using the equation provided in Schleich and Vassallo (2003): where EBV is estimated bullar volume, L is bullar length, W is bullar width, and H is bullar height.
To test for intraspecific sexual dimorphism in bullar volume, two-sample Mann−Whitney U tests (stats::wilcox.test)on bullar volumes were carried out on the males and females of each species.To understand allometric scaling within each species, a model II simple linear regression (lmodel2::lmodel2, range.x= " interval," range.y= "interval"; Legendre, 2018) using 1000 permutations of the ranged major axis method was conducted on each species' skull length against the bullar volume (both variables were log 10 -transformed).
To test for significant differences between groups (species,
To test for intraspecific sexual differences in bullar shape, Mann −Whitney U tests were conducted on the scores of the first two PC axes.Kruskal−Wallis rank sum tests (stats::kruskal.test)were used to test for significant differences between groups (species, rainfall categories), using the first two PC axes' scores as variables, separately.Dunn's post hoc tests (FSA::dunnTest) were run after each Kruskal−Wallis test to determine which category differed from which, using a Bonferroni corrected p-value.
F I G U R E 4 Scatterplot of log10-transformed bullar volume against log10-transformed skull lengths.Colors of each point and regression line corresponds to the species (key to species abbreviations as in Figure 1).
F I G U R E 5 Boxplots of otomyine bullar volumes between species (a) and rainfall categories (b).Solid bar shows the median value, box denotes the first and third quartiles, and the extent of the "whiskers" (dotted lines) show the minimum and maximum values.(c) NMDS ordination plot using morphological variables, with the species and rainfall factors fit to the ordination.NMDS, nonmetric multidimensional scaling.

| Sexual dimorphism
None of the species showed intraspecific sexual dimorphism in bullar volumes or in the scores of the first two PC axes; that is, the sexes did not different in either bullar size or shape in any of the study species (Supporting Information S1: Table 2).
Between the rainfall categories, a significant difference in the  having significantly larger bullae compared to those in more mesic areas, while there is no significant difference in bullar volume between individuals that inhabit regions with higher annual rainfall (Figure 5b; Table 2).The NMDS ordination clustered the individuals from Parotomys together (positive NMDS axis1), which was also associated with the lowest rainfall level (100−200 mm per annum; Figure 5; Table 3).
F I G U R E 6 Boxplots of the scores of the first two principal component (PC) axes for the Otomyini species, categorized according to the rainfall categories that the individuals inhabit (left boxplots) and into the separate species (right boxplots).Solid bar shows the median value, box denotes the first and third quartiles, and the extent of the "whiskers" (dotted lines) show the minimum and maximum values.Central panel shows the wireframe shape differences for the positive and negative ends of the PC axes (gray, dotted line, mean configuration; black, solid line, axis extreme configuration).
We investigated whether species in the Otomyini tribe showed interspecific and intraspecific differences in auditory bullar volume and shape.It was found that those otomyine species inhabiting regions with lower annual rainfall exhibited auditory bullae with larger volumes.This matched what was found by Nengovhela et al. (2019) and we build on that study with the additional finding that this trend of inflated bullae in more arid regions is also found within a single, widely-distributed species.
Parotomys species have bullae that are more voluminous and that have more anteriorly facing meatuses.Commonly known as "whistling rats," Parotomys species have unique communication methods among the otomyines in that they whistle loudly when danger is perceived (Hall, 2015;Kingdon, 2015;Monadjem et al., 2015), and larger bullae could aid in the perception of these calls.Lower frequencies travel further distances than higher frequencies, which is beneficial to those rodents living in open spaces, such as xeric shrublands (Walsberg, 2000), and for Parotomys, the ability to register these lower frequencies may have resulted in the evolution of larger bullae (García-Navas & Blumstein, 2016).
These small mammals cannot hide amongst dense grass in these dry, vegetation-sparse regions, and thus communicating to their burrowmates of potential dangers is crucial in this environment (Walsberg, 2000).Specific tests for auditory ability in detecting particular frequencies have not been done for the otomyine rodents; however, studies of other rodents have indicated that bullar hypertrophy is linked with the detection of low frequency sounds (Manoussaki et al., 2008;Nengovhela et al., 2019;Vater & Kössl, 2011;West, 1985).
Another difference found between Parotomys and other otoymines is the direction of the external meatuses, with Parotomys exhibiting more forward-facing external openings to the middle ear.
Sound localization in mammals appears to driven by a combination of three cues: binaural time-difference, binaural intensity-difference, and monaural pinna cues (Heffner & Heffner, 2016).There is a correlation between functional head size and the ability to hear lowfrequency sounds in mammals (Heffner & Heffner, 2016).This was explained to be due to the smaller head size reducing the maximum binaural time-difference, increasing the need to use the binaural intensity-difference cue to localize sound (Masterton et al., 1969).
Whether the more forward-facing meatuses, combined with the inflated auditory bullae, produce a better (or worse) ability for sound localization in Parotomys remains to be seen, but it is interesting that this combination is found in arid-adapted otoymines, but not in the mesic-adapted species.
M. unisulcatus individuals that inhabit regions with lower annual rainfall exhibited larger bullar volumes compared to their conspecifics from other more mesic regions.M. unisulcatus is variable in body coloration, body size, and dietary intakes, and differs in nesting habits across their range (De Graaff, 1981), with those in the western (more arid) region being larger and lighter in color, whilst those inhabiting the eastern (more mesic) regions are smaller, darker-colored individuals (Edwards et al., 2011).Larger absolute bullar volumes of the individuals in regions with lower annual rainfall (western regions of southern Africa) may be a consequence of their larger overall size.
In mammals, correlations of body size and temperature (Bergmann's rule-larger body size in colder regions; Ashton et al., 2000;Meiri & Dayan, 2003), and limb dimensions and temperature (Allen's ruleanimals adapted to cold climates have shorter and thicker limbs and bodily appendages) have been found.Smaller mammals, however, do not follow Bergmann's rule as strongly (Ashton et al., 2000), and in some species, larger bodied conspecifics are found in warmer We can conclude that there is a link between auditory structures and aridity in the southern African species of the Otomyini tribe.
There could be many reasons for this difference, such as different . Parotomys Thomas, 1918 (whistling rats) are found in the less densely vegetated xeric shrublands and construct extensive burrow systems, and is predominantly distinguished from Otomys by their larger auditory bullae, shorter nasal bones, and the fact that both Parotomys species F I G U R E 1 Photographs (in greyscale and high contrast) of the ventral side of a representative skull from each study species.Specimens shown: Myotomys sloggetti-AM25311; Myotomys unisulcatus-AM31084; Otomys angoniensis-AM8570; Otomys irroratus-AM29343; Otomys saundersiae-AM30777; P. littledalei-AM5968; Parotomys brantsii-AM4537.Cladogram of the relationships shown to the left (from Montgelard et al., 2023), and photographs are surrounded by boxes colored to reflect the mean annual rainfall levels in the regions where they occur.
bullar volume was evident (χ 2 [5, N = 236] = 125.76,p < .0001),with those inhabiting lower-rainfall regions (100−200 mm per annum) T A B L E 1 Dunn's post hoc test results showing difference levels in auditory bullar volume size and shape between sample species.
Figure 5; Table 3).The individuals from different rainfall regions differed significantly in bullar shape (PC1 [53%]: χ 2 [5, N = 236] = 150.21,p < .0001;PC2 [15%]: χ 2 [5, N = 236] = 21.87,p = .001;Table2), with individuals from regions with the lowest annual rainfall (100−200 mm per annum) having more anterior-facing tympanic meatuses (landmarks 7 and 8) and those from regions with higher annual rainfall levels having larger and more posterior-facing meatuses (PC1 scores; Figure6; Supporting Information S1: Figure1), climates.Temporal changes in skull size within otomyine species showed either no change over time (P.brantsii) or a decrease in size with a warming climate (M.unisulcatus, O. auratus, and O. angoniensis;Nengovhela et al., 2015Nengovhela et al., , 2020)).While we did find a significantly negative correlation with date collected and bullar volume (r = −0.25,t 34 = 468.27,p < .0001,calculated in Microsoft Excel), and date collected and head length (r = −0.16,t 34 = 556.31,p < .0001), in our M. unisulcatus data set-in other words, bullar volume and head length reduced with time (and a warming climate)-there was no seeming pattern of date collected and locality collected (i.e., the specimens collected in the more arid regions were collected between 1925 and 1987).So, in this species, larger bullar volumes in more arid regions appear to be due to larger body sizes, though the underlying reason for the differential body sizes in different biomes is still not clear.
foraging and defense methods required for different environments, though it appears that reception of lower-frequency vocalizations may be a driving factor.Investigating other tribes or families would be beneficial to confirm this theory and would be interesting to see if the same trends are seen in larger rodents.Also, to verify whether the Parotomys species and arid-dwelling M. unisulcatus do indeed F I G U R E 7 Boxplots of (a) bullar volumes, (b) first principal component (PC) axis scores, and (c) second PC axis scores among Myotomys unisulcatus individuals inhabiting areas with different annual rainfall levels.Solid bar shows the median value, box denotes the first and third quartiles, and the extent of the "whiskers" (dotted lines) show the minimum and maximum values.(d) Scatterplot of the scores of the first two PC axes investigating the bullar shape differences within M. unisulcatus.Wireframes to the right and below represent the shape differences for the extreme positive and negative ends of each axis (gray line, mean configuration; black line, axis extreme configuration).(e) Model II linear regression of log 10 -transformed skull lengths and bullar volumes of M. unisulcatus.
T A B L E 2 Dunn's post hoc test results showing difference levels in auditory bulla volume and shape of sample otomyine species between regions classified by annual rainfall (mm per annum).Dunn's post hoc test results showing difference levels in auditory bulla volume and shape between Myotomys unisulcatus inhabiting sites with differing mean annual rainfall levels.
T A B L E 3Note: Z-values shown in the pairwise comparison tables (below diagonal), and those results that were significant (Bonferroni adjusted p-values) are shaded in gray.