Tooth Wear and Feeding Ecology in Mountain Gorillas from Volcanoes National Park, Rwanda

Home , Karisoke Research Center, Tonda (orangutan), Volcanoes National Park

AMERICAN JOURNAL OF PHYSICAL ANTHROPOLOGY 00:00–00 (2015)

Tooth Wear and Feeding Ecology in Mountain Gorillas From Volcanoes National Park, Rwanda

Jordi Galbany,1* Olive Imanizabayo,2 Alejandro Romero,3 Veronica Vecellio,2 Halszka Glowacka,4 Michael R. Cranfield,5 Timothy G. Bromage,6 Antoine Mudakikwa,7 Tara S. Stoinski,2 and Shannon C. McFarlin1,8

1Center for the Advanced Study of Human Paleobiology, Department of Anthropology, The George Washington University, DC 2Dian Fossey Gorilla Fund International, Atlanta, GA 3Departamento de Biotecnologıa, Universidad de Alicante, Alicante, Spain 4Institute of Human Origins, Arizona State University, Tempe, AZ 5Mountain Gorilla Veterinary Project, University of California at Davis, CA 6Hard Tissue Research Unit, New York University College of Dentistry, NY 7Rwanda Development Board, Department of Tourism and Conservation, Kigali, Rwanda 8Division of Mammals, National Museum of Natural History, Smithsonian Institution, DC

KEY WORDS dentine exposure; aging; diet; Gorilla beringei beringei

ABSTRACT

Objectives: Ecological factors have a dramatic effect on tooth wear in primates, although it remains unclear how individual age contributes to functional crown morphology. The aim of this study is to determine how age and individual diet are related to tooth wear in wild mountain gorillas (Gorilla beringei beringei) from Volcanoes National Park, Rwanda. Material and Methods: We calculated the percent of dentine exposure (PDE) for all permanent molars (M1–M3) of known-age mountain gorillas (N 5 23), to test whether PDE varied with age using regression analysis. For each molar position, we also performed stepwise multiple linear regression to test the effects of age and percentage of time spent feeding on different food categories on PDE, for individuals subject to long-term observational studies by the Dian Fossey Gorilla Fund International’s Karisoke Research Center. Results: PDE increased significantly with age for both sexes in all molars. Moreover, a significant effect of gritty plant root consumption on PDE was found among individuals. Our results support prior reports indicating reduced tooth wear in mountain gorillas compared to western gorillas, and compared to other known-aged samples of primate taxa from forest and savanna habitats. Discussion: Our findings corroborate that mountain gorillas present very low molar wear, and support the hypothesis that age and the consumption of particular food types, namely roots, are significant determinants of tooth wear variation in mountain gorillas. Future research should characterize the mineral composition of the soil in the Virunga habitat, to test the hypothesis that the physical and abrasive properties of gritty foods such as roots influence intra- and interspecific patterns of tooth wear. Am J Phys Anthropol 000:000–000, 2015. VC 2015 Wiley Periodicals, Inc.

Primate dental morphology provides evidence of dietary adaptations, as tooth morphology is designed to Additional Supporting Information may be found in the online accommodate the mechanical properties of foods that are version of this article. processed during mastication and ingestion (Lucas, 2004; Bunn and Ungar, 2009; Guy et al., 2013). The Grant sponsor: National Science Foundation; Grant numbers: functional life of a tooth may be altered mechanically by BCS 0852866, 0964944; Grant sponsors: The Leakey Foundation; National Geographic Society’s Committee for Exploration and wear produced by attrition, or the cumulative loss of Research; the Center and Institute Facilitating Fund of The George enamel and dentine tissues resulting from contact Washington University; the 2010 Max Planck Research Award between opposing teeth, and abrasion, which results endowed by the German Federal Ministry of Education and Research from contact with hard food objects and other particles to the Max Planck Society; Alexander von Humboldt Foundation. ingested during foraging (Lucas, 2004; Guy et al., 2013; Lucas et al., 2013; Romero et al., 2012, 2013). Thus, *Correspondence to: Dr. Jordi Galbany, Center for the Advanced physical properties of food items are critical determi- Study of Human Paleobiology, Department of Anthropology, The nants of macro- and microscale enamel loss associated George Washington University, 800 22nd Street NW, Ste 6000 Washington DC 20052, USA. E-mail: [email protected] with certain dietary behaviors and ecological constraints (Galbany et al., 2011, 2014; Romero et al., 2012; Morse Received 6 May 2015; revised 2 November 2015; accepted 3 et al., 2013; Cuozzo et al., 2014). Because foraging is the November 2015 most important life requirement for primates, tooth wear can have dramatic implications for life history, as DOI: 10.1002/ajpa.22897 the effects on occlusal surface morphology associated Published online 00 Month 2015 in Wiley Online Library with wear may have consequences for individual sur- (wileyonlinelibrary.com).

Ó 2015 WILEY PERIODICALS, INC. 2 J. GALBANY ET AL. vival and reproduction (Logan and Sanson, 2002; King reported among great ape taxa, and inform future stud- et al., 2005). ies testing its potential later-life consequences. Numerous studies of dental ecology in the field have Today, gorillas are found in ten central African coun- demonstrated how tooth wear varies as a function of age tries, in a broad diversity of habitats ranging from [e.g., Lemur catta (Cuozzo and Sauther, 2006), Propithe- coastal lowland forests to high altitude afromontain rain cus edwardsi (King et al., 2005), Alouatta palliata (Den- forests, in which they also show significant variation in nis et al., 2004), Papio sp. (Phillips-Conroy et al., 2000; behavior and demography (Doran and McNeilage, 1998). Galbany et al., 2011), Mandrillus sphinx (Galbany et al., Although gorillas are usually classified as folivorous and 2014)]. In addition, wear-related changes in dental func- may consume tough and fibrous fallback foods, signifi- tion can significantly impact the outcome of reproductive cant differences in dietary behavior exist among popula- effort. King et al. (2005) demonstrated that sifaka tions (Doran et al., 2002; Elgart-Berry, 2004). females older than 18 years of age were characterized by Mountain gorillas are only found in two small popula- a significant loss of shearing blade length, suggesting tions at the south sector of the Virunga Volcanoes in loss of dental function. Further, advanced tooth wear in Rwanda, Uganda and Democratic Republic of Congo these older females was associated with a significant (DRC), and the Bwindi Impenetrable National Park in decline in infant survivorship, suggesting that advanced Uganda, located in the Albertine Rift. Whereas western dental wear may have consequences for a female’s abil- gorillas feed on a substantial amount of fruit, mountain ity to provision offspring in long-lived mammals such as gorillas from the Virunga Volcanoes feed mostly on non primates. These findings highlight the potential conse- reproductive plant parts, such as leaves, stems, pith, quences of dental senescence for life history. bark, roots, but very few fruits (Watts, 1998; Grueter In addition to age, tooth wear also reflects the interac- et al., 2013). The Virunga volcanoes region is character- tion between dental tissues and ecological factors, such ized by high-altitude montane forests with a dense herb as feeding behavior and habitat characteristics (Cuozzo layer and low abundance of fruit (Doran and McNeilage, et al., 2014). For example, in Amboseli baboons (Papio 2001). cynocephalus), the percent of dentine exposure on molars Moreover, mountain gorillas show wide dietary flexi- was positively correlated with the percent of time spent bility, which enables them to occupy a variety of habitats consuming gritty grass corms at the individual level within their range (Doran and McNeilage, 2001). In the (Galbany et al., 2011). Further, forest-dwelling mandrills Volcanoes National Park (VNP), the majority of their (Mandrillus sphinx) from Lek edi Park, Gabon, showed diet is composed of the leaves, stems, and pith (85.8%) of significantly higher wear rates for their age than Ambo- plant species, mainly Galium vines, thistles, celery and seli baboons, a finding that was attributed to differences nettles (Watts, 1998; Grueter et al., 2013). The Karisoke in mechanical food properties and soil composition (Gal- study area in particular is characterized by the highest bany et al., 2014). Though these species share similar- gorilla food biomass in the Virungas, and the diet of the ities in molar occlusal morphology and enamel gorillas in this area is composed of at least 54 plant spe- thickness, a high percentage of the mandrill diet year- cies (Watts, 1984; Grueter et al., 2013), with six species round is comprised of hard fruits and seeds, and the soil accounting for the 87% of the diet: vines (Galium spp.), from their habitat is comparatively rich in quartz, which thistles (Carduus nyassanus), wild celery (Peucedanum may contribute to the abrasiveness of their diet (Gal- linderi), bamboo (Yushania alpina), blackberries (Rubus bany et al., 2014). Thus, when abrasive particles in the spp.), and nettles (Laportea alatipes). Food availability diet are considered, they show a significant effect on and growth rate of vegetation does not vary seasonally, enamel loss that is independent of age (Galbany et al., with the exception of bamboo shoots (Fossey and Har- 2011, 2014). court, 1977; Watts, 1984). Mountain gorillas usually Another example of the interaction between dental tis- travel to the bamboo forests during the few months of sues and ecological factors is provided by ring-tailed the year when the fresh shoots are available. Moreover lemurs (Lemur catta) from Beza Mahafaly Special bark and roots of some plants and trees are also occa- Reserve (BMSR) in Madagascar. Cuozzo et al. (2014) sionally consumed (Watts, 1984). showed that dietary variation is the likely cause of Here, we examine age and diet as potential sources of microhabitat-related variation in tooth wear within this variance in tooth wear in a single well-studied wild popu- population, especially due to the consumption of lation of mountain gorillas (Gorilla beringei beringei) mechanically challenging foods, which are more abun- from the VNP in Rwanda. Individuals in this population dant in areas affected by anthropogenic disturbance. have been subject to continuous behavioral observations, Moreover, in great apes, examinations of individuals including observations of feeding behavior, since 1967 by from osteological collections have shown that as wear the Dian Fossey Gorilla Fund International’s Karisoke proceeds, minor differences are detected in occlusal relief Research Center (Watts, 1998; Grueter et al., 2013; Cail- topography in chimpanzees and orangutans compared to laud et al., 2014). Our specific aims are twofold. First, we more folivorous lowland gorillas (MKirera and Ungar, aim to expand on the limited available data on tooth 2003). Further, it has been shown that mountain gorilla wear in known-aged primates by examining changes in (Gorilla b. beringei) teeth exhibit less wear in both the percentage of dentine exposure (PDE) as a function of upper and lower molars than G. g. gorilla and G. g. age. Secondly, we aim to test the hypothesis that long- graueri subspecies (Cousins, 1988; Elgart, 2010), sug- term patterns of feeding behavior predict tooth wear, as gesting that mountain gorilla teeth are comparatively food choice has been shown to impact tooth wear in Afri- durable and wear may not lead to mechanical senes- can cercopithecidae species from forest and savanna cence, despite the mechanically demanding diet (Elgart, niches (Galbany et al., 2011, 2014). Moreover, the results 2010). However, to date it has not been possible to exam- of this research will have significance for efforts to esti- ine tooth wear in great apes of known chronological age. mate age at death in skeletal remains from wild moun- By characterizing how tooth wear proceeds with age, tain gorillas housed in osteological collections, as well for this may contribute to our understanding of differences unknown mountain gorillas found dead in the Virungas.

American Journal of Physical Anthropology TOOTH WEAR IN MOUNTAIN GORILLAS 3 MATERIAL AND METHODS TABLE 1. Known age subjects, including collection ID, name, sex, age at death, age accuracy and presence of feeding data Study population Age Feeding The mountain gorillas examined in the present study ID Name Sex Age accuracy data? are from the VNP in Rwanda, located in the south sector GP.071b Urugero Male 2.76 0 No of the Virunga volcanoes, in the Albertine Rift. This area GP.169 Akarusho Male 3.35 0 Yes extends from the northernmost point of Lake Albert to GP.125 Ihumure Male 3.72 0 Yes the southernmost point of Lake Tanganyika, forming a GP.025b Mpore Female 3.83 0 No transboundary block of three protected areas of GP.004 Dushirehamwe Male 5.48 1 No 450 km2 between Rwanda, Uganda and the DRC GP.075 Arusha Male 6.08 0 No (Weber, 1995). GP.147 Tayna Female 8.61 0 Yes All individuals included in this study (Table 1, Sup- GP.038 Mpanga Female 10.69 0 No GP.167 Icyizere Female 12.18 2 Yes porting Information Table S1) were found dead in the GP.033 Nyarusizi Male 12.71 0 No Volcanoes National Park in Rwanda, and their skeletons GP.045 Ndatwa Male 14.94 0 No recovered in a multidisciplinary collaborative effort by GP.161 Bikwi Male 18.49 4 No the Mountain Gorilla Skeletal Project (McFarlin et al., GP.148 Ntobo Female 19.79 0 Yes 2009). Age and sex are known for those gorillas that GP.026 Umurava Male 21.14 0 Yes were monitored daily by the Rwanda Development GP.069b Ziz Male 22.29 1 No GP.139 Intwali Female 24.26 0 Yes Board and Dian Fossey Gorilla Fund International’s a Karisoke Research Center (Musanze, Rwanda), where GP.150 Kuryama Male 24.84 4 No GP.153a Kubyina Female 30.10 1 No the skeletons are also housed. GP.149 Ginseng Female 31.15 0 Yes Tooth wear was measured in 58 gorillas, representing GP.124 Shinda Male 31.73 0 Yes all individuals in the collection to date that possess at GP.143 Kwiruka Female 33.71 3 Yes least one fully erupted permanent molar (M1) (Table 1, GP.121b Pablo Male 33.78 4 No Supporting Information Table S1). From these gorillas, GP.131 Tuck Female 38.31 1 Yes 19 individuals (10 males and 9 females) are associated GP.116 Puck Female 38.32 1 Yes b with chronological age known to the exact date or month GP.078 Papoose Female 43 3 Yes (615 days), one female is associated with a known year a Individuals not included in the analyses due to high tooth of birth (66 months), two female individuals are associ- wear associated with malocclusion. ated with ages known to within 61.5 years, four individ- b Individuals of probable identity. Age accuracy is given as fol- uals were first seen as adults (62–4 years, or more), and lows: 0 5 Known, exact date or within a week (64 days). three individuals disappeared from social groups before 1 5 Know the month of birth (615 days). 2 5 Know the year in their deaths and their remains later recovered at an which the birth occurred (66 months). 3 5 Individual first seen advanced stage of decomposition; ages for the latter are as pre-reproductive immature (61.5 years). 4 5 Ages for GP.121, GP.150, GP.161 are based on date last observed alive. Maximum established as date last observed alive (see Supporting ages based on date body recovered are as follows: GP.150 5 24.99 Information Table S1). Five individuals are of ‘probable years; GP.161 518.53 years. GP.121 was recovered as fully ske- ID’. These individuals died prior to the onset of the letonized remains 1 year after the date last observed alive. Mountain Gorilla Skeletal Project, which implemented a more standardized burial recording protocol. In most cases, identity was determined from historical accounts porting Information Table S1). Furthermore, two and by the Mountain Gorilla Veterinary Project based on individuals that presented pathological tooth wear due to matching of dental development status, linear body malocclusion, evidenced by dentine exposure in areas dimensions, and other dental-skeletal observations other than the cusps, were excluded from all statistical against necropsy reports from the same time period analyses reported here. Finally, all analyses were con- (2001-2006). For GP.071 and GP.025, dental emergence ducted with and without individuals of ‘probable ID’ status and skeletal dimensions are also consistent with included. Since excluding these individuals did not other similarly-aged individuals of confirmed identifica- change the outcome of analyses reported below, we report tion. In the case of GP.121 and GP.078, probable identi- results here for the total sample. ties were determined based on recorded disappearance of these individuals from their study groups and the Feeding data recovery location of postmortem remains in the forest (i.e., within their former range areas). Finally, given To examine gorilla feeding behavior as a predictor, we sample size limitations, we also report data collected used data from the Karisoke Research Center’s long- from 29 individuals of unknown age (12 males, 13 term database (Watts, 1984) to determine the percent of females, 4 unknown sex) (Supporting Information Table feeding time during which each individual consumed S1). In light of prior reports that mountain gorillas show particular food types from January 2001 through the reduced dental wear compared to western gorillas, these remainder of their lives. The proportion of time spent in individuals were measured to ensure that our sample feeding behavior was determined using 50-min focal ani- represents the range of wear states observed in this mal sampling periods, during which instantaneous sam- population. ples were collected at 10-min intervals (Grueter et al., In analyses testing the effect of age on tooth wear, we 2013). When feeding was recorded, the identity of the only included individuals that are associated with known food item was also recorded, allowing us to determine ages (61.5 years, or less; N 5 12 males and 11 females) the importance of different food types as a percentage of listed in Table 1. Individuals that were first seen as total feeding time. A total of 7,562 individual point sam- adults (62–4 years, or more), and unknown age individu- ples of feeding behavior were analyzed on 13 mountain als, were not considered in these analyses (Table 1, Sup- gorillas, including males and females (Table 1). Samples

American Journal of Physical Anthropology 4 J. GALBANY ET AL.

Fig. 1. Adult mountain gorillas processing roots from terrestrial herbaceous vegetation (THV) at Volcanoes National Park (Rwanda). The combination of age and time spent feeding on roots was significantly related to tooth-wear rates during life. Photos by Jordi Galbany. Fig. 2. Digitized images showing the percent of dentine exposure (PDE) on mandibular molars (M1–M3) with age (in years). All individuals are females. Mesial: left; buccal: down. were collected throughout the year, covering both dry and wet seasons. Nonetheless, the number of samples per individual per year, and the number of years of sampling, varied from gorilla to gorilla [number of focal sam- an individual had missing or broken teeth, the PDE for ples (average 5 581.7; range 5 158–1,228); duration of that molar position was calculated based on the avail- data collection (ranged 5 8–12 years) per individual]. We able molars (see Supporting Information Table S1). considered the percentage of time spent feeding on the most common plant parts consumed by gorillas as follows: bamboo shoots (Yushania alpina), vines (mostly Statistical analyses Galium sp.), small herbs (mostly Rumex sp.), terrestrial herbaceous vegetation – THV (mostly Carduus sp., Peu- To compare the PDE of adult known-age gorillas with cedanum linderi, Echinops hoehnelii, and Laportea ala- that of unknown-age gorillas, we performed ANOVA tipes), pith of trees, bark of trees and dead wood, tests for M1, M2, and M3; considering both sexes backberries (Rubus sp.), other fruits (Discopodium pen- together and also separately. To determine the relation- ninervium, Vernonia adolfi-frederici), leaves (Rubus sp., ship between PDE and age in this cross-sectional data Yushania alpina, Carex bequaertii), and roots (Fig. 1). set, we tested PDE as linear and quadratic functions of These foods have been previously reported to comprise age in regression analyses conducted separately for the over 90% of the mountain gorilla diet in the Karisoke M1 (N 5 23), M2 (N 5 17), and M3 (N 5 16). We used study area (see Watts, 1984; Grueter et al., 2013). analysis of covariance (ANCOVA) to test within-tooth PDE differences on individual age by sex, using arcsine Analysis of tooth wear and logarithmic transformed variables, respectively (Galbany et al., 2014). Quantification of tooth wear followed established pro- Next, we examined the effects of diet on PDE in males tocols (Uchida, 1996; Elgart-Berry, 2000; Elgart, 2010; and females. To do this, we performed a stepwise multi- Galbany et al., 2011, 2014; Morse et al., 2013). The following teeth were sampled: upper and lower first, sec- ple linear regression, in which we regressed age and all ond, and third permanent molars (M1, M2, and M3, food categories (percent of time consuming each major respectively). Occlusal digital photographs were collected food) on PDE for each molar separately. These multivari- using a Nikon D800 digital SLR camera (36.3-MP reso- ate analyses allowed us to control for multiple variables lution) with a fixed 85-mm lens. Teeth were oriented simultaneously. We performed these analyses for each based on the cemento-enamel junction plane, visible in molar using a subsample of gorillas (N 5 13) for whom skeletonized specimens, which was situated perpendicu- we had feeding data available. We also performed a lar to the optical axis of the camera lens. On each image, series of multiple regressions taking into account age the total occlusal area (TOA), which is the area of molar and each food category separately, to determine whether crown in occlusal view, and areas of dentine exposure the results were consistent. Models were compared using (DEA), were traced with ImageJ software (Abramoff Akaike’s Information Criterion corrected for small sam- et al., 2004) to calculate the Percent of Dentine Expo- ple size (AICc), as a measure of relative quality, and the sure (PDE) for each molar tooth (PDE 5 DEA/TOA*100) preferred model was identified by the minimum value of (see Fig. 2 and Galbany et al., 2011, 2014 for details). AICc (Akaike, 1974; Sugiura, 1978; Burnham and The average PDE was calculated by taking the mean Anderson, 2002), following Galbany et al. (2014). Statis- values for each tooth position, across both antimeres of tical analyses were conducted using SPSS 15.0 for the maxillary and mandibular dentitions. In cases where WindowsTM. The significance level was set at P 0.05.

American Journal of Physical Anthropology TOOTH WEAR IN MOUNTAIN GORILLAS 5

Fig. 3. Box-plot showing percent of dentine exposure (PDE) for M1 in unknown and known age adult gorillas by sex. Boxes Fig. 4. Relationship between individual age and percent of enclose 25–75% percentile values, the mean and median are dentine exposure (PDE) in mountain gorillas. The solid lines indicated with a circle and horizontal bar respectively, and represent the ﬁtted quadratic regressions for M1–M3 (P < 0.01). whiskers denote the interquartile range (IQR), mini- Crosses indicate two individuals with extremely high tooth- mum2maximum values 1.53IQR. Diamonds indicate outliers wear associated with malocclusion, not included in the (individual code number). analyses.

RESULTS provided smaller AICc values than linear regressions (Fig. 4 and Table 2). Further, a test of homogeneity of When all adult specimens from the collection were slopes (as assumed by an ANCOVA) revealed that the sex considered—including those of unknown age—the high- factor did not contribute significantly to variance for est individual mean PDE values for M1, M2, and M3 in PDE with age in M1 (F1,22 5 4.292; P 5 0.051), M2 females and male were, respectively: 53.19, 28.06, and (F1,16 5 1.209; P 5 0.288) or M3 (F1,15 5 0.309; P 5 0.586) 9.45% (GP.057, female), and 58.59, 16.33, and 5.97% molar teeth. Figure 5 summarizes mean percent of den- (GP.173, male) (see Fig. 3, Supporting Information Table tine exposure at different age intervals, including cases of S1). GP.173 was a known gorilla whose birth date is unknown-age gorillas showing extreme wear. unknown, Kwitonda, the dominant silverback in Kwi- Stepwise multiple linear regressions, which included all tonda group, managed by Rwanda Development Board the predictor variables (age and percent of time spent for tourist purposes. On the other hand GP.057 was an feeding on every food category) in one model per molar, unidentified unknown adult female, found dead before produced different results. For M1, R2 for the complete 1994. model was 0.921, with the majority of variance explained When PDE of adult known-age gorillas was compared by age (R2 5 0.882) and a smaller amount (3.9%) explained with that of unknown-age gorillas, unknown-age gorillas by percent of time spent eating roots (Table 3). For M2, showed significantly higher tooth wear for M1 (ANOVA; only age was included in the model and explained 67.5% F1,37 5 5.943, P 5 0.020) (Fig. 3), but equivalent PDE for of the variance. Finally for the M3, only the percent of M2 (ANOVA; F1,37 5 2.592, P > 0.05) and M3 (ANOVA; time consuming bamboo shoots was included in the model, 2 F1,37 5 0.931, P > 0.05). Differences in M1 tooth wear but having a negative effect (R 5 0.504) (Table 3). Finally, between the two subsamples were driven by differences percent of time spent consuming other foods did not show in PDE between known and unknown-age female gorillas any significant relationship with PDE for any molar. (ANOVA; F1,20 5 6.962, P 5 0.016), but not males (ANOVA; Multiple regressions considering each food category F1,15 5 0.972, P > 0.05). Both M2 and M3 showed no signif- separately yielded the same results for M1 and M2. icant differences in PDE between unknown and known- Only the percent of roots, in addition to age, explained age gorillas when males and females were considered sep- PDE in M1 (R2 5 0.921 for the whole model). For M2, arately (M2 females: ANOVA; F1,20 5 1.938, P > 0.05; M2 only age explained variability in PDE for all models. For males: ANOVA; F1,15 5 0.824, P > 0.05; M3 females; M3, some of the models were not significant, meaning ANOVA; F1,20 5 0.828, P > 0.05; M3 males; ANOVA; that even age did not explain variability of PDE, and F1,15 5 0.129, P > 0.05). none of the food categories explained significant variabil- In gorillas of known age, analyses of PDE for each molar ity of PDE (Table 4). (M1, M2, and M3) revealed significant linear relationships with age as predictor variable (M1: F1,21 5111.56, DISCUSION 2 2 R 5 0.842, P < 0.001; M2: F1,15 5 23.52, R 5 0.611, 2 Tooth wear and age in Virunga mountain gorillas P < 0.001; and M3: F1,14 5 14.31, R 5 0.506, P 5 0.002), and even better results when quadratic regressions were Our results clearly show that tooth wear in mountain 2 performed (M1: F2,21 5173.55, R 5 0.943, P < 0.001; M2: gorillas reflects age. This finding is in agreement with 2 F2,15 5 40.47, R 5 0.844, P < 0.001; and M3: F2,14 5 25.23, other studies on tooth wear in wild primates including R2 5 0.783, P < 0.001). In all cases, quadratic regressions strepsirrhines (King et al., 2005; Wright et al., 2008),

American Journal of Physical Anthropology 6 J. GALBANY ET AL.

TABLE 2. Linear and quadratic regressions for predicted percent of dentine exposure (PDE) in molars (M1-M3) with age Model Tooth N R2 FPAICc Equation Linear M1 23 0.842 111.56 <0.001 68.82 PDE 5 0.303*age-2.122 M2 17 0.611 23.52 <0.001 50.60 PDE 5 0.194*age-2.530 M3 16 0.506 14.31 0.002 15.86 PDE 5 0.082*age-0.999 Quadratic M1 23 0.943 173.55 <0.001 48.22 PDE 5 0.008*age220.016*age M2 17 0.844 40.47 <0.001 33.79 PDE 5 0.006*age220.086*age M3 16 0.783 25.23 <0.001 15.85 PDE 5 0.002*age220.024*age AICc: Corrected Akaike Information Criterion. Signiﬁcant differences P 0.05 are shown in bold.

Fig. 5. Dentine exposure model at different age intervals in mountain gorillas from Volcanoes National Park. Both upper teeth (left) and lower teeth (right) are shown. Drawing modified from Swindler (2002). platyrrhines (Dennis et al., 2004) and cercopithecines the known-aged sample, our findings are consistent with (Phillips-Conroy et al., 2000; Galbany et al., 2011, 2014). previous analyses of tooth wear in African great apes, Mountain gorillas show a significant positive relation- which concluded that mountain gorillas, despite their ship between PDE and age, and for all molar types the thinner enamel (Swindler, 2002), show the lowest relationship was quadratic, as found in previous studies amounts of wear on occlusal surfaces in M1 and M2 on cercopithecine species (Galbany et al., 2014). Differ- among gorillas, and also lower tooth wear rates than ent degrees of wear observed among teeth likely reflect chimpanzees (Pan t. troglodytes and P.t. schweinfurthii) earlier eruption and functionality for M1, followed by (Elgart, 2010). M2 and M3 (Smith et al., 1994). However, none of the However, when gorillas of unknown age were individuals in the known-age sample present a very included, maximum individual mean PDE values for M1 high degree of tooth wear. In fact, the highest PDE and M2 were higher (58.59 and 28.06%, respectively). (averaged across right and left maxillary and mandibu- Thus, PDE was significantly greater in the unknown-age lar molars) found in M1 was 11.99%, and 9.91% for M2. sample compared to known-age gorillas among females, Even when cases of pathological tooth wear were consid- but not males. One possible explanation for this differ- ered here, the highest PDE in M1 was just 17.88%. For ence is that the known-aged sample is biased towards

American Journal of Physical Anthropology TOOTH WEAR IN MOUNTAIN GORILLAS 7

TABLE 3. Step-wise multiple regression analyses for predicting TABLE 4. Independent multiple regression analyses for predict- molar wear (PDE) from foraging ecology ing molar wear (PDE) in M1-M3, from age and each food category M1 (N 5 13) Model R2 AICc P (age) P (food) Step no. Variable R2 Coefﬁcient P M1 (N 5 13) 1 Age 0.882 0.415 <0.001 Bamboo shoots 0.894 17.72 <0.001 0.310 2 Roots 0.921 0.304 0.050 Vines 0.891 18.05 <0.001 0.373 Constant 26.084 0.008 Small herbs 0.887 18.49 <0.001 0.495 Regression equation: THV 0.913 15.09 <0.001 0.085 PDE(M1) 5 0.405*(Age) 1 0.291*(Roots) – 5.661 Pith 0.887 18.51 <0.001 0.501 F 5 58.21, P < 0.001 Bark 0.882 19.13 <0.001 0.961 Blackberry fruits 0.901 16.79 <0.001 0.191 Other fruits 0.882 19.12 <0.001 0.949 M2 (N 5 13) Leaves 0.886 18.68 <0.001 0.566 Step no. Variable R2 Coefﬁcient P Roots 0.921 13.89 <0.001 0.050 M2 (N 5 11) 1 Age 0.675 0.220 0.002 Bamboo shoots 0.769 14.42 0.002 0.109 Constant 22.898 0.085 Vines 0.675 18.16 0.006 0.999 Regression equation: Small herbs 0.768 14.44 0.007 0.111 PDE(M2) 5 0.220*(Age) – 2.898 F 5 18.70, P 5 0.002 THV 0.683 17.88 0.007 0.661 Pith 0.706 17.05 0.002 0.383 Bark 0.764 14.65 0.001 0.121 M3 (N 5 12) Blackberry fruits 0.676 18.13 0.004 0.882

2 Other fruits 0.731 16.07 0.003 0.231 Step no. Variable R Coefficient P Leaves 0.796 13.06 0.001 0.062 1 Bamboo SH 0.504 20.283 0.021 Roots 0.734 15.98 0.002 0.222 Constant 4.692 0.004 M3 (N 5 10) Regression equation: Bamboo shoots 0.546 4.42 0.445 0.250 PDE(M3) 520.283*(Bamboo SH) 1 4.692 F 5 8.121, P 5 0.021 Vines 0.456 6.24 0.055 0.714 Small herbs 0.636 2.21 0.084 0.096 Significant differences P 0.05 are shown in bold. THV 0.447 6.40 0.072 0.854 Pith 0.453 6.28 0.047 0.743 Bark 0.644 4.53 0.027 0.088 individuals of younger ages. Despite the long history of Blackberry fruits 0.454 6.28 0.048 0.736 observation at Karisoke (since 1967), a number of habitu- Other fruits 0.511 5.16 0.053 0.360 ated adults represented in our sample were first observed Leaves 0.526 4.85 0.077 0.307 as mature individuals and are thus of unknown chrono- Roots 0.478 5.83 0.039 0.524 logical age. Life expectancy also differs between the sexes. Significant differences at P 0.05 are shown in bold. AICc: Cor- For example, the oldest known age male ever observed at rected Akaike Information Criterion. Karisoke, Canstbee (still living), was born in November 1978 (37 years old by November 2015), while the oldest consumption of particular foods, namely roots, show a known age female, Papoose (disappeared from the study significant relationship with the degree of tooth wear group and presumed dead in June 2007; GP.078 in the that individuals experience for their age. current study) reached 43 years old (61.5 years of age Age-related tooth wear variability between and within accuracy). A second possibility is that these cases of species can be related to differences in feeding ecology and increased wear may be attributed to differences in rang- tooth–food item interactions (Lucas, 2004). Teeth can be ing patterns, which for some groups have changed consid- abraded by particles harder than enamel; these particles erably since the start of observational research at may be intrinsic to plants, as in the case of phytoliths, or Karisoke. The total range area of Karisoke study groups exogenous gritty particles from the soil (Romero et al., in Volcanoes National Park has experienced a 75% 2012, 2013; Lucas et al., 2014). Research on the mecha- increase since 2007 (Caillaud et al., 2014). Further, nisms behind abrasion processes at a nanometric scale changes in gorilla food availability and altitudinal distri- shows that quartz dust is a rigid abrasive capable of frac- bution within the study area have also been documented turing and removing enamel pieces, while phytoliths suffer (Grueter et al., 2013). Though, similar data for range deformation during their contact with enamel, causing areas of unhabituated gorillas, and for those study groups small grooves on the enamel surface, but not dramatic tis- monitored for tourism, are not available. As the mountain sue loss (Lucas et al., 2013, 2014). While some contend that gorilla skeletal collection grows over time, the influence of phytoliths may have a role in tooth wear (Rabenold and these factors on variation in PDE warrants further study. Pearson, 2014; Xia et al., 2015), it seems clear that exoge- Tooth wear and diet-related factors in mountain nous gritty contaminants on plant matter are more likely gorillas to cause increased tooth wear (Sanson et al., 2007; Romero et al., 2012; Lucas et al., 2013, 2014; Galbany et al., 2014). The current study tests whether individual differences We found that the percent of time consuming roots in dietary composition can account for variation in tooth (5.46% on average, range 0.81–17.68%) by mountain goril- wear in Virunga mountain gorillas. It should be noted las has a significant positive effect on tooth wear for the that the study is limited in that the available diet data M1. This food item, consumed during the whole year only cover a short time frame, beginning in 2001. How- (Watts, 1984), may have low potential to generate wear by ever, despite having low dental wear overall compared to itself. For example Kniphofia thomsonii root consumed by other great apes, we found that differences in the mountain gorillas in the Ugandan sector of the forest, but

American Journal of Physical Anthropology 8 J. GALBANY ET AL. not by mountain gorillas in the Karisoke study area, was in this population (Romero et al., 2012; Lucas et al., shown to have a low toughness value (0.93 kJ m22; Elgart- 2013) and between mountain gorillas and other great Berry, 2004). Nonetheless, roots may represent a source of ape taxa (e.g., Elgart, 2010). gritty extrinsic abrasive particles in the diet. To obtain roots, gorillas pull the plant out of the soil with both hands, Conclusions and future directions especially Carduus nyassanus, and then peel off or remove Age-related patterns of tooth wear differ among pri- the outermost layer using their incisors, to access to the mate species. The significance of our findings in compar- inner part of the root (see Fig. 1. JG and VV field observa- ative context are difficult to interpret, as only Old World tions). Sometimes, mountain gorillas also rub the roots monkeys are represented among known-aged samples with their hands, apparently to remove soil. However, all studied to date and, otherwise, occlusal morphology of of these procedures do not fully eliminate the soil from the cercopithecoid and hominoid molars differs substantially roots, and so during the process of root consumption, gorillas are directly exposed to grit particles that may be even in their unworn state (Swindler, 2002). Nonethe- ingested with the root. less, we note that mountain gorillas examined here show In this context, our finding that the percentage of time reduced tooth wear with age as reflected in PDE com- spent feeding on bamboo shoots has a negative effect on pared to more frugivorous primates such as mandrills, tooth wear on the M3 may reflect a bias in the feeding the latter of which appear to exhibit more substantial data, as bamboo shoots are consumed only during rainy wear-related changes in molar occlusal surfaces evi- seasons (Watts, 1984). However, we suggest that this denced by a flatter occlusal morphology and higher PDE finding may reflect the negative correlation between the (Galbany et al., 2014). Our findings are also in agree- percent of time consuming bamboo shoots and roots ment with previous studies that also found that moun- (R2 5 0.452, P 5 0.004). Thus, gorillas that consumed less tain gorillas present very low tooth wear, compared to bamboo shoots ate more THV roots, and this may contrib- other great ape taxa (Elgart, 2010). ute to more wear on their M3s. A previous study of moun- The present study also contributes to an understand- tain gorilla foods from the Virunga massif determined that ing of the relationship between tooth wear and primate the fracture toughness of bamboo shoots is very low, in fact diets. Our results lend support that the consumption of much lower than those found in other mountain gorilla particular foods, namely roots, has a significant effect on foods such as bark, leaves, pith or stems (Elgart-Berry, variability in age-related patterns of molar wear 2004). This implies a low or null direct abrasion of enamel, between individuals within the Virunga population. As and lower attrition between tooth-to-tooth contact due to root consumption had a positive effect on wear in our shorter chewing cycles (Morse et al., 2013). The most fre- study, future research to characterize mineral composi- quently consumed foods were shown in the latter study to tion and particle size in soils from the Virungas (and be low to moderately tough on average (1.92 kJ m22), while from other gorilla habitats) are necessary to examine bamboo shoots eaten by Virunga gorillas showed lower the role of exogenous grit in explaining wear differences toughness values (0.2 kJ m22). These low toughness values among gorilla populations. Also, future studies that cou- may indicate that processing of vegetation by itself has less ple dental topographic (Glowacka et al., submitted), potential to wear down the enamel. occlusal fingerprint (Benazzi et al., 2011), and microwear Our finding that increased root consumption is associ- analyses with more detailed examinations of variability ated with increased tooth wear among mountain gorillas is in food material properties would significantly advance consistent with previous studies. For example, tooth wear our understanding of the mechanical demands of gorilla associated with consumption of gritty food has also been diets, the potential of tooth wear to discriminate individ- documented in baboons (Papio cynocephalus), where those uals with specific food preferences, and the consequences individuals who consumed more underground storage of tooth wear for processing specific foods. organs (USOs) showed a higher PDE with age (Galbany et al., 2011). Findings on ring-tailed lemurs (Lemur catta) ACKNOWLEDGMENTS also suggest the importance of microhabitat and gritty diets for explaining tooth wear differences within a single The authors gratefully acknowledge the Rwandan Gov- primate population (Cuozzo et al., 2014). Thus, extrinsic ernment and Volcanoes National Park authorities for their grit present in the soil may also be an important contribu- support of the Mountain Gorilla Skeletal Project (MGSP). tor to tooth wear differences among species living in differ- The MGSP is indebted to the core research partners, field ent habitats. In turn, it was previously shown that baboons assistants and other staff of the Rwanda Development from Amboseli (Kenya) are characterized by significantly Board’s Department of Tourism and Conservation, Dian lower molar PDE with age compared to mandrills from Fossey Gorilla Fund International’s Karisoke Research Lek edi Park in Gabon (Galbany et al., 2014). The percent Center, Mountain Gorilla Veterinary Project—Gorilla Doc- of quartz in the soil was also significantly lower in the tors, and the many other contributing students and Amboseli baboon habitat (1.5%), whereas mandrills from researchers from Rwandan and US institutions for their Lek edi Park inhabit an ecosystem with a high presence of continuing efforts to preserve mountain gorilla skeletal mineral quartz (8%) (Galbany et al., 2014). This remains and generate associated long-term data as a increased exposure to quartz, in addition to a higher bite- resource for scientific study. Without their tireless efforts force demanding diet in mandrills, likely results in faster and commitment to the gorillas, these studies would not be enamel cracking and early dentine exposure (Lee et al., possible. DFGFI also gratefully acknowledges the public 2011; Galbany et al., 2014). and private agencies, foundations, and individuals that Although root consumption explained only 3.9% of the have provided support for the Karisoke Research Center. variation in M1 PDE among mountain gorillas examined Finally, the authors are also grateful to Keely Arbenz in the current study, we suggest that the presence of Smith for her contributions to MGSP research, and Nuria grit warrants further attention as a possible factor con- Garriga, Winnie Eckardt, and anonymous reviewers who tributing to variation in tooth wear among individuals provided helpful comments.

American Journal of Physical Anthropology TOOTH WEAR IN MOUNTAIN GORILLAS 9 LITERATURE CITED Lee JW, Constantino P, Lucas P, Lawn B. 2011. Fracture in teeth—a diagnostic for inferring tooth function and diet. Biol Abramoff MD, Magelhaes PJ, Ram SJ. 2004. Image Processing Rev 86:959–974. with Image J. Biophoton Int 11:36–42. Logan M, Sanson GD. 2002. The effect of tooth wear on the Akaike H. 1974. A new look at the statistical model identifica- feeding behaviour of free-ranging koalas (Phascolarctos ciner- tion. IEEE Trans Automat Control 19:716–723. eus, Goldfuss). J Zool Lon 256:63–69. Benazzi S, Kullmer O, Grosse IR, Weber GW. 2011. Using occlusal Lucas PW. 2004. Dental functional morphology: how teeth wear information and finite element analysis to investigate work. Cambridge: Cambridge University Press. 355p. stress distributions in human molars. J Anat 219:259–272. Lucas PW, Omar R, Al-Fadhalah K, Almusallam AS, Henry AG, Bunn JM, Ungar PS. 2009. Dental topography and diets of four Michael S, Thai LA, Watzke J, Strait DS, Atkins AG. 2013. Old World Monkeys species. Am J Primatol 71:1–12. Mechanisms and causes of wear in tooth enamel: implications Burnham K, Anderson D. 2002. Model selection and multi- for hominin diets. J R Soc Interface 10:20120923. model inference: a practical information-theoretic approach, Lucas PW, van Casteren A, Al-Fadhalah K, Almusallam AS, 2nd ed. New York: Springer. 488 p. Henry AG, Michael S, Watzke J, Reed DA, Diekwisch TGH, Caillaud D, Ndagijimana F, Giarrusso AJ, Vecellio V, Stoinski Strait DS, Atkins AG. 2014. The role of dust, grit and phyto- TS. 2014. Mountain gorilla ranging patterns: influence of liths in tooth wear. Ann Zool Fennici 51:143–152. group size and group dynamics. Am J Primatol 76:730–746. McFarlin SC, Bromage TG, Lilly AA, Cranfield MR, Nawrocki Cousins D. 1988. Why is the gorilla such a large primate? Int SP, Eriksen A, Hunt D, Ndacyayisenga A, Kanimba Misago Zoo News 35:9–16. C, Mudakikwa A. 2009. Recovery and preservation of a moun- Cuozzo FP, Head BR, Sauther ML, Ungar PS, O’Mara MT. 2014. tain gorilla skeletal resource in Rwanda. Am J Phys Anthro- Sources of tooth wear variation early in life among known-aged pol Suppl 48:187–188. wild ring-tailed lemurs (Lemur catta)attheBeza Mahafaly M’Kirera F, Ungar PS. 2003. Occlusal relief changes with molar Special Reserve, Madagascar. Am J Primatol 76:1037–1048. wear in Pan troglodytes troglodytes and Gorilla gorilla Cuozzo FP, Sauther ML. 2006. Severe wear and tooth loss in gorilla. Am J Primatol 60:31–41. wild ring-tailed lemurs (Lemur catta): a function of feeding Morse PE, Daegling DJ, McGraw WS, Pampush JD. 2013. Den- ecology, dental structure, and individual life history. J Hum tal wear among Cercopithecid monkeys of the Ta€ı Forest, Evol 51:490–505. Cote^ d’Ivoire. Am J Phys Anthropol 150:655–665. Dennis JC, Ungar PS, Teaford MF, Glander KE. 2004. Dental Phillips-Conroy JE, Bergman T, Jolly CJ. 2000. Quantitative topography and molar wear in Alouatta palliata from Costa assessment of occlusal wear and age estimation in Ethiopian Rica. Am J Phys Anthropol 125:152–161. and Tanzanian baboons. In: Whitehead PF, Jolly CJ, editors. Doran D, McNeilage A. 1998. Gorilla ecology and behaviour. Old World monkeys. Cambridge: Cambridge University Press. Evol Anthropol 6:120–131. p 321–340. Doran DM, McNeilage A. 2001. Subspecific variation in gorilla Rabenold D, Pearson OM. 2014. Scratching the surface: a cri- behavior: the influence of ecological and social factors. In: tique of Lucas et al. (2013)’s conclusion that phytoliths do not Robbins MM, Sicotte P, Stewart KJ, editors. Mountain goril- abrade enamel. J Hum Evol 74:130–133. las: three decades of research at Karisoke. Cambridge: Cam- Romero A, Galbany J, De Juan J, Perez-P erez A. 2012. Short bridge University Press. p 123–149. and long-term in vivo human buccal dental-microwear turn- Doran D, McNeilage A, Greer D, Bocian C, Mehlman P, Shah over. Am J Phys Anthropol 148:467–472. N. 2002. Western lowland gorilla diet and resource availabil- Romero A, Ramırez-Rozzi FV, De Juan J, Perez-P erez A. 2013. ity: new evidence, cross-site comparisons, and reflections on Diet-related buccal dental microwear patterns in Central indirect sampling methods. Am J Primatol 58:91–116. African pygmy foragers and Bantu-speaking farmer and pas- Elgart-Berry A. 2000. The relationship of fracture toughness of toralist populations. PLoS One 8:e84804. plants to the morphology and durability of Gorilla (Gorilla Sanson GD, Kerr SA, Gross KA. 2007. Do silica phytoliths gorilla) mandibles and teeth. Ph.D. 277p. really wear mammalian teeth? J Archaeol Sci 34:526–531. Elgart-Berry A. 2004. Fracture toughness of mountain gorilla Smith B, Crummett TL, Brandt KL. 1994. Ages of eruption of (Gorilla gorilla beringei) food plants. Am J Primatol 62:275–285. primate teeth: a compendium for aging individuals and com- Elgart AA. 2010. Dental wear, wear rate, and dental disease in paring life histories. Yrbk Phys Anthropol 37:177–231. the African apes. Am J Primatol 72:481–491. Sugiura N. 1978. Further analysis of the data by Akaike’s infor- Fossey D, Harcourt AH. 1977. Feeding ecology of free ranging mation criterion and the finite corrections. Commun Stat mountain gorillas (Gorilla gorilla beringei). In: Clutton-Brock Theory Methods 7:13–26. TH, editor. Primate ecology: studies of feeding and ranging Swindler DR. 2002. Primate dentition. An introduction to the behavior in lemurs, monkeys and apes. London: Academic teeth of non-human primates. Cambridge: Cambridge Univer- Press. p 539–556. sity Press. Galbany J, Altmann J, Perez-P erez A, Alberts SC. 2011. Age Uchida A. 1996. Craniodental variation among the great apes. and individual foraging behavior predict tooth wear in Ambo- Cambridge: Harvard University Press. 184p. seli baboons. Am J Phys Anthropol 144:51–59. Watts DP. 1984. Composition and variability of mounain gorilla Galbany J, Romero A, Mayo-Aleson M, Itsoma F, Gamarra B, diets in the Central Virungas. Am J Primatol 7:323–356. Perez-P erez A, Willaume E, Kappeler PM, Charpentier MJE. Watts DP. 1998. Long-term habitat use by mountain gorillas 2014. Age-related tooth wear differs between forest and (Gorilla gorilla beringei). 1. Consistency, variation, and home savanna primates. PLoS One 9:e94938. range size and stability. Int J Primatol 19:651–680. Grueter CC, Ndamiyabo F, Plumptre AJ, Abavandimwe D, Weber W. 1995. Monitoring awareness and attitude in conserva- Mundry R, Fawcett KA, Robbins MM. 2013. Long-term tem- tion education: the mountain gorilla project in Rwanda. In: poral and spatial dynamics of food availability for endangered Jacobsen SK, editor. Conserving wildlife: international educa- mountain gorillas in Volcanoes National Park, Rwanda. Am J tion and communication approaches. New York: Columbia Primatol 75:267–280. University Press. p 28–48. Guy F, Gouvard F, Boistel R, Euriat A, Lazzari V. 2013. Pro- Wright P, King SJ, Baden A, Jernvall J. 2008. Aging in wild spective in (Primate) dental analysis through tooth 3D topo- female lemurs: sustained fertility with increased infant mor- graphical quantification. PloS One 8:e66142. tality. Interdiscip Top Gerontol 36:17–28. King SJ, Arrigo-Nelson SJ, Pochron ST, Semprebon GM, Xia J, Zheng J, Huang D, Tian ZR, Chen L, Zhou Z, Ungar PS, Godfrey LR, Wright PC, Jernvall J. 2005. Dental senescence Qian L. 2015. New model to explain tooth wear with implica- in a long-lived primate links infant survival to rainfall. Proc tions for microwear formation and diet reconstruction. Proc Natl Acad Sci USA 102:16579–16583. Natl Acad Sci USA 112:10669–10672.

American Journal of Physical Anthropology