Naturally-occurring tooth wear, tooth fracture, and cranial injuries in large carnivores from Zambia

Study area

The Republic of Zambia occupies 752,614 km² in southcentral Africa within 15° 00′ S and 30° 00′ E. Samples were obtained from two regions, Luangwa Valley (LV) in the east and the Greater Kafue Ecosystem (GKE) in the west (Fig. 1). Both regions are comprised of national parks (NPs) and adjacent game management areas (GMAs). The LV contains four NPs totaling 15,630 km² that, combined with the adjacent GMAs, covers 62,038 km² (Astle, 1999). The GKE is an approximately 66,547 km² area comprised of Kafue NP (KNP) (22,400 km²) and adjacent GMAs (Siamudaala, Nyirenda & Saiwana, 2009). There are no fences or other barriers, and animals routinely cross between NPs and GMAs. Legal trophy hunting for lion, leopard, and spotted hyena occurs in most GMAs. Two lions killed by the wildlife authority (DNPW) as problem animals that were 10–15 km outside of LV (n = 1) and GKE (n = 1) were assigned to the geographically closest region.

Obtaining specimens

Between 2007 and 2012, one of us (PAW) examined and photographed skulls and teeth of hunted lion, leopard, and spotted hyena as part of a larger study on age estimation of trophy hunted lions (White & Belant, 2016; White et al., 2016). Skulls of hunted carnivores were made available by professional hunters and taxidermists. Skulls of two lions killed as problem animals were made available by DNPW. All osteological material was examined and photographic data collected in Zambia in partnership with DNPW (formerly Zambia Wildlife Authority, Research/Employment Permit No. #008872). A total of 118 lion (LV 60; GKE 58), 45 leopard (LV 18; GKE 27) and 13 spotted hyena (LV 6; GKE 7) skulls were evaluated in this study. Due to post-mortem damage or incomplete photographic views, not all lion skulls could be used in every analysis. Thus, samples sizes for lion varied slightly as follows: tooth wear and breakage 115 lions (LV 58; GKE 57); cranial injury 116 lions (LV 60; GKE 56).

Lion hunting in Zambia is restricted to males. Sex of sampled lions was confirmed from hides or trophy photographs. We restricted our sample of problem lions to include only males. Leopard hunting in Zambia is restricted to males, and sampled leopards were assumed to be males. However, trophy photographs or hides were not available for all sampled leopards, and the possibility cannot be excluded that some hunted individuals were females (Spong, Hellborg & Creel, 2000). Because spotted hyena are encountered less frequently, the species is taken more opportunistically and hunter selectivity is lower than for lion or leopard. Consequently, the spotted hyena sample likely consisted of both female and male specimens. For each of the three species, we assumed that hunter bias in selectivity was similar among the two regions.

Photographs

We captured digital images of each skull using a Nikon 35mm D3300 digital camera with Nikkor AF 70–300 mm f/4.5-6.3 lens (Nikon USA Inc., Melville, NY, USA). We photographed skulls with the mandible articulated from left and right lateral sides, anterior, occipital and dorsal views. Thereafter, we disarticulated the mandible and inverted the cranium to obtain a palatal view and dorsal and lateral views of the mandibles. Digital images were organized by a field ID number assigned to each specimen and stored on SD cards for later examination on a computer and display monitor.

Age estimation of samples

Because the probability of having broken a tooth or suffered an injury increases with age, it is important to control for age differences between samples from the two regions. We assigned age classes to each sampled lion, leopard, and spotted hyena based on published standards (below) that represent the best available methods of age estimation for each species. We assigned specimens to one of three age classes initially with skulls in-hand and verified age assignments later using digital images.

In lions, crown wear of the upper second premolar (P2) has been shown to increase with age (Schaller, 1972; Smuts, Anderson & Austin, 1978; White & Belant, 2016). We visually scored P2 crown wear as: (1) no wear/sharp, (2) moderate wear/rounded, or (3) heavy wear/flat, and used P2 scores in conjunction with overall tooth wear features (following Smuts, Anderson & Austin, 1978; Miller et al., 2016) in assigning each lion to one of three age classes: young adult < 5 years; mature adult ≥ 5–7 < years; and old adult ≥ 7+ years (Fig. S1). Assignment to the mature adult age class was confirmed in most cases by the partially to fully obliterated closure of the interfrontal suture after Smuts, Anderson & Austin (1978) (Fig. S2). We assigned each leopard to an age class based on wear of the upper and lower canines (C1, c1) and premolars (P3, p3,4) following the criteria of Stander (1997) with the modification that younger age classes were combined as follows: young adult ≤ 4 years; mature adult 5–6 years, and old adult 7–10 years (Fig. S3). We assigned spotted hyena to one of three age classes (young 1–3 years; mature > 3–6 years; old > 6–16 years) based on wear of the lower third premolar (p3) after Kruuk (1972) (Fig. S4). We tested for differences in age structure of sampled lion and leopard from LV vs. GKE using chi-square performed in SPSS v.26 with significance levels set at P < 0.050.

Scoring of dental condition

For the regional comparisons, we scored tooth breakage and overall wear from digital photographs alone. Scoring of tooth breakage and wear was performed subsequent to, and independent of, age estimation, and involves a more comprehensive review of wear across the entire tooth row as opposed to focusing on one or a few specific teeth (e.g. canines, premolars). Notably, PAW did the age estimation and BVV did the subsequent dental wear and fracture analyses, blind to the age estimates. As in earlier studies (Van Valkenburgh, 1988; Van Valkenburgh & Hertel, 1993; Van Valkenburgh, 2009), breakage was scored by tooth position and type, i.e., incisors, canines, premolars, and molars. A tooth was counted as broken in life (antemortem) only if there was clear evidence of subsequent wear on the tooth following breakage (Fig. S5). Teeth broken at time of death or post-mortem with apparently unworn, sharp edges were not counted as broken and were excluded from analyses.

We assigned one of five tooth wear stages to each individual: (1) little or no apparent wear observed on shear facets or blunting of cusps, (2) shear facets apparent on carnassial teeth and cusps slightly blunted on some teeth, (3) shear facets apparent on carnassial teeth and cusps blunted on most teeth, (4) carnassial teeth exhibiting strong shear facets and moderate blunting of premolar cusps, or (5) carnassial teeth exhibiting strong shear facets and/or blunted cusps, and premolars and molars with well-rounded cusps. The five categories were then reduced to three for the analysis, (1) “slight” (stage 1 only), (2) “moderate” (stages 2 and 3), and (3) “heavy” (stages 4 and 5) (Fig. S6).

We summed tooth wear stage distribution and fracture frequencies by species and location (LV or GKE) and, for lion and leopard, compared distribution of wear stage between regions using chi-square with significance set at P < 0.050. Interspecific comparisons using the entire dataset were made on (1) number of broken teeth as a percentage of the total number of teeth, (2) percentage of individuals with one or more broken teeth, and (3) percentage of tooth fracture by tooth position (incisors, canines, premolars, carnassials). For lion and leopard, the percentage of broken teeth and breakage by tooth position for each species were compared between samples from LV and GKE using chi-square with significance set at P < 0.050. Sample size of spotted hyena (n = 13) was insufficient to compare between the two regions.

We then compared the results of tooth damage in modern Zambia carnivores to historical samples for all three species obtained from various museums as described in Mann, Van Valkenburgh & Hayward (2017) using chi-square with significance set at P < 0.050. Due to methodological differences in assessing tooth wear, i.e., Zambia samples from photographs; historical samples in-hand, we restricted our comparisons of modern with historic specimens to tooth breakage. Further, it should be noted that our Zambia felid sample consists of males, whereas Mann, Van Valkenburgh & Hayward (2017) present data for the sexes combined. In addition, the historical samples span more years (several decades) than our 6-year Zambia study, and thus are not strictly comparable.

Scoring of cranial injuries

We examined skulls for evidence of past injury including healed or healing fractures or other pathologies excluding those directly associated with dental breakage, i.e., missing bone at the alveolar margin of a broken tooth. Damages that were associated with time of death were readily determined, e.g., freshly splintered bone, and excluded from analyses. We initially assessed cranial injuries with skulls in-hand and subsequently quantified damages using digital images.

We scored injuries as resulting from either blunt trauma or intraspecific conflict as follows: (1) major fractures, defined as fractures that caused misalignment of one or more bones sometimes resulting in remodeling, and shallow depressions or gouges with irregular shapes were presumed to have resulted from blunt trauma sustained during prey capture and handling, such as a kick or impact of a horn boss; or (2) deep circular depressions and linear scratches, consistent with bite and claw marks, respectively, were judged to have been caused by intraspecific conflict (Fig. 2). Because some judgements were subjective, it is possible that prey-inflicted injuries vs. intraspecific conflict were erroneously assigned in some instances. Major fractures most commonly involved the zygomatic arches (Fig. 2) or nasals, while other injuries were found primarily on the premaxilla, maxilla, nasal, frontal, and parietal bones.

We recorded the number of incidents of traumas per skull, with an incident defined as occurrence of one type of pathology per animal because it could not be known if multiple injuries were inflicted during a single event. For example, a lion with one or more fractures (which could have occurred during a single predation event) was counted as one incident of blunt trauma attributable to prey-handling, and a lion with one or more bites and scratches (which could have occurred during a single lion fight) was counted as one incident of intraspecific trauma, whereas a lion with one or more fractures, one bite mark and deep scratches was counted as two incidents (fracture(s) = one blunt trauma; bite and scratches = one intraspecific trauma).

Carnivore populations

From the literature, we obtained lion population size estimates during the sampled time period of 400–750 in LV (0.0064–0.012 lions/km²) with an additional approximately 50 lion inhabiting the CO, and an estimated 250–500 lion in GKE (0.0037–0.0075 lions/km²) (IUCN, 2006). We used IUCN (2006) estimates rather than available focal studies in reporting lion densities because the former represent the larger regions inclusive of the NPs and all adjoining GMAs. Lion appeared to be both more numerous and present at higher density in LV than in GKE (IUCN, 2006).

The LV is considered as having one of the highest densities of leopard in Africa (Gregg, 2019) although few empirical data exist on Zambia’s leopard populations (Purchase, Mateke & Purchase, 2007). A study conducted from 2006–2008 in a 491 km² area of the LV reported leopard densities of 3.36 to 4.79 leopards/100 km² (Ray, 2011). Subsequent investigations (2012–2014) in a 313 km² area of the LV estimated from 5.08 to 8.50 leopards/100 km² (Rosenblatt et al., 2016). We found no current publications reporting on leopard density in GKE although there is no evidence to suggest that leopard density in GKE was as high, or higher, than in LV.

Scant data are available on the current status of spotted hyena populations in LV or GKE (Purchase, Mateke & Purchase, 2007); available reports suggest that the species occurs at relatively low density in both regions. Based on camera traps, spotted hyena appeared to be approximately 2–5 times less abundant than leopard in Ray’s (2011) LV study area. Playback censuses conducted in 2003–2004 by Carlson, Carlson & Bercovitch (2004) estimated 18–44 spotted hyenas/1,000 km² in northern Kafue National Park (NKNP) and commented that spotted hyena were shy and rarely seen. Likewise, Creel et al. (2018) reported on spotted hyena in NKNP between 2013–2016 as present but not common. This is consistent with reports of spotted hyena occurrence in the larger GKE between 2007–2012 where sightings were relatively infrequent and most often consisted of singles or pairs (White & Kim, 2018). These limited and localized results highlight the need for updated surveys on the status of spotted hyena in LV and GKE (Purchase, Mateke & Purchase, 2007).

Prey populations

From the literature, we obtained relative abundance of prey in each region. Aerial surveys conducted by DNPW in 2015 contributed to population trend data for 12 species of large mammals over the time periods 2002–2015 in LV and 2006–2015 in GKE (DNPW, 2016). Population estimates for 12 key prey species (DNPW, 2016), as well as elephant (IUCN, 2016), and hippopotamus (Chansa & Milanzi, 2010) are provided as Supplemental Information. Because all three carnivore species consume carrion, we augmented estimates of elephant population size with counts of elephant carcasses, where available (CITES Secretariat, 2016). Additionally, we referenced Creel et al. (2018) for relative abundance of prey species in NKNP from 2013–2016.

Age structure of sampled carnivores

When considering samples from the Luangwa Valley (LV) and Greater Kafue Ecosystem (GKE) combined, the majority of lion (81%) and leopard (73%) were mature adults, i.e., ≥5 years old, with 11% of lions and 13% leopards having estimated ages of ≥7 years, i.e., old adults (Table 1) (Data S1). There was no significant difference in the age distribution between regions either for lion (X² = 1.701, df = 2, P = 0.427) or leopard (X² = 2.639, df = 2, P = 0.267). The 13 spotted hyena specimens included young (31%), mature (23%), and old (46%) individuals (Table 1) (Data S1).

Tooth wear and fracture frequency between species

The distribution of individuals among tooth wear categories was similar between lions and leopards, with lions showing a tendency towards having more individuals in the heavy wear class (Fig. 3A) (Data S2) despite the two species having similar age distributions (Table 1). However, lions showed twice the tooth fracture rate of leopards on a per tooth basis (4% vs. 2%; X² = 8.150, df = 2, P = 0.002) (Table 2) (Table S1) (Data S2). Although lions also exceeded leopards in the percentage of individuals with one or more broken teeth (53% vs. 44%, Table 2), this difference was not significant (X² = 0.957, df = 2, P = 0.211). Across the tooth row, lion had significantly more broken incisors and premolars than leopard (P < 0.030) but the two species did not differ greatly in their rates of canine or carnassial tooth fracture (Fig. 3B) (Table S1) (Data S2).

As expected based on their fondness for consuming bone, spotted hyena had more tooth wear and higher tooth fracture rates than lion or leopard. Among the hyenas sampled, there were none that exhibited only slight tooth wear, and the majority were placed in the heavy wear category (Fig. 3A) (Data S2). On a per tooth basis, spotted hyena had a fracture rate of 12%, which is three times that observed for lion (4%, X² = 50.117, df = 2, P < 0.001) and six times that observed for leopard (2%, X² = 71.223, df = 2, P < 0.001) (Table 2) (Table S1). On a per individual basis, 77% of the spotted hyena had at least one broken tooth, whereas that proportion was 53% for lion and 44% for leopard (Table 2) (Data S2). The higher tooth fracture observed in spotted hyena was distributed across the tooth row, with canine teeth and cheek teeth being more likely to be broken than incisors (Fig. 3B).

Tooth fracture frequency within species

Sample sizes of lion and leopard were sufficient to explore differences in tooth fracture frequency between populations in LV and GKE. Among the lions, the differences between the two regions were striking with LV lion exhibiting significantly higher fracture frequencies than their GKE counterparts, both in percentage of broken teeth (5% vs. 2.6%, X² = 13.213, df = 2, P < 0.001) and percentage of individuals with one of more broken teeth (62.1% vs. 43.9%; X² = 3.827, df = 2, P = 0.038) (Table 2) (Data S2). Notably, the higher fracture frequency of LV lions was apparent across the entire tooth row, from incisors to carnassials (Fig. 3B) (Table S1). In association with their higher tooth fracture rates, LV lions also had more rapid rates of tooth wear than GKE lions, as evidenced by a smaller proportion of individuals in the slight wear class along with a larger proportion of individuals in the heavy wear class (Fig. 3C; X² = 6.623, df = 2, P = 0.036), despite the fact that the age distributions of lions did not differ between LV and GKE (Table 1). Unlike the lions, the leopards from the two regions did not differ significantly in fracture frequency on either a per tooth (X² = 0.161, df = 2, P = 0.424) or per individual basis (X² = 0.375, df = 2, P = 0.381) (Fig. 3C) (Table S1) (Data S2).

Comparison with historical samples

Tooth fracture rates on a per tooth basis in male lion and leopard from Zambia were largely comparable to those observed in historical samples of both sexes of these species that originated from different parts of Africa as published in Mann, Van Valkenburgh & Hayward (2017) (Table 2). There were significant differences between the historical and Zambia samples in a few cases. Among lion, the proportion of individuals with at least one broken tooth was significantly greater in the Zambia lion (53% vs. 33.8%; X² = 4.322, df = 2, P = 0.025) as was the proportion of teeth fractured in LV lion (5% vs. 3.5%; X² = 6.717, df = 2, P = 0.007) but not GKE lion. The per tooth fracture rate for Zambia leopard was significantly less than that recorded by Mann, Van Valkenburgh & Hayward (2017) (2.1% vs. 3.7%; X² = 8.641, df = 2, P = 0.003). The most striking difference is in the spotted hyena. Our small sample (13) collected over 6 years in Zambia broke their teeth much more frequently on both a per tooth basis (12% vs. 4%; X² = 44.816, df = 2, P = 0) and per individual basis (77% vs. 51%; X² = 694.932, df = 2, P = 0) than the spotted hyena sample in Mann, Van Valkenburgh & Hayward (2017) that included 179 skulls spanning decades and multiple countries.

Cranial injuries

Over half (68/116; 59%) of all our sampled lions had at least one cranial injury ascribed to natural causes, and 16 individual lions (14%) showed evidence of both blunt trauma and intraspecific conflict (Data S3). LV lion had significantly higher rates of blunt trauma injuries attributable to prey-handling compared to lion in GKE (52% vs. 30%) (X² = 5.422, df = 2, P = 0.019) (Table 3). Among the 31 LV lions with blunt trauma injuries, 15 consisted of major healed or healing fractures (Fig. 2) as opposed to only five lions with major injuries in GKE. In contrast, the incidence of injuries ascribed to intraspecific conflict was similar for lion from LV and GKE (32% vs. 30%) (Table 3).

Among leopard, the number of incidents attributable to blunt trauma and intraspecific conflict combined were similar between regions (LV 72%; GKE 78%) (Table 3) (Data S3). No major fractures were seen in leopards. Due to small sample sizes and the number of incidents that could not be assigned to specific categories, differences in the frequency of blunt trauma injuries related to prey-handling vs. intraspecific conflict in LV vs. GKE leopards were not conclusive (Table 3).

Of the 13 spotted hyena skulls examined, four exhibited signs of significant natural trauma, including one with multiple penetrating bite wounds across its rostrum which may have been inflicted by a lion or a hyena (Fig. 2). Because the likely causes of the natural injuries in spotted hyena (prey-handling vs. intraspecific conflict vs. lion) could not be determined to a reasonable degree, all were classified as non-specific (Table 3) (Data S3).

Prey diversity, abundance and trends

Prey species diversity was similar in both regions with a few exceptions; a small population of endemic Thornicroft’s giraffe occurs only in LV, and lechwe Kobus leche occur only in GKE. Relative abundance and density of prey varied between regions and among species (DNPW, 2016) (Data S4). Buffalo (DNPW, 2016), elephant (IUCN, 2016), and hippopotamus (Chansa & Milanzi, 2010) occurred in much higher abundance and density in LV compared with GKE. Data on elephant carcasses were available only from a specified survey area in LV where from 4 to 49 carcasses per year (mean of 19 carcasses/year) were reported from 2002–2012 (CITES Secretariat, 2016) (Data S5). Hippopotamus surveys conducted between 2005–2008 found that LV contained the largest concentration in Zambia, with the Luangwa river accounting for 62% (25,000) of the country’s overall hippopotamus population compared to 10% (4,000) in the Kafue river in GKE (Chansa & Milanzi, 2010).

Higher numbers of several species of small to medium-sized antelopes (e.g., lechwe, puku K. vardonii, waterbuck K. ellipsiprymnus, sable Hippotragus niger) were recorded in GKE (Data S4). Impala Aepyceros melampus were the most abundant antelope by far in both systems, although density of puku and sable were also high in GKE. Lechwe populations in GKE were highly localized in the northwest corner of the ecosystem (DNPW, 2016). A slightly different picture of prey composition in Kafue is given by Creel et al. (2018) who reported the species of highest densities in northern Kafue National Park (NKNP) as puku, impala, and warthog (Phacochaerus aethiopicus). Aerial surveys indicated that for the majority of key wildlife species in GKE, wildlife numbers that had generally declined from the mid to late-1990s increased over the period 2006–2012. Similarly, over the period 2002–2012 in the LV, numbers of most wildlife species were reported to be stable or increasing (DNPW, 2016) (Data S4).