Brain size of the lion (Panthera leo) and the tiger (P. tigris): implications for intrageneric phylogeny, intraspecific differences and the effects of captivity (2024)

Search for other works by this author on:

Article history

Abstract

INTRODUCTION

Brain size in vertebrates, especially mammals, has consistently been a provocative and revealing subject for scientific investigation and public interest. Previous research suggested the existence (and absence) of correlations between variables derived from mammalian brain size and body size, evolutionary stage, behaviour, ecology, life history, domestication, captivity and taxonomy (Jerison, 1973; Radinsky, 1978; Gittleman, 1986a, 1994; Pagel & Harvey, 1988; Kruska, 1989; Hemmer, 1990; Iwaniuk, Pellis & Whishow, 1999; Bush & Allman, 2004; Pérez-Barbería, Shultz & Dunbar, 2007). Most studies tried to identify general trends in and the biological relevance of ‘encephalization’ in a wide range of species, although each species may be represented by only a few specimens. In contrast, less attention has been paid to finer-scale investigations of encephalization within single species, or between closely related species, based on large samples. Furthermore, studies on intraspecific brain size have been conducted mainly on humans or domesticated mammals, sometimes compared with their wild ancestors (Kruska, 2005), and few have focused exclusively on wild species.

It has been suggested that detailed research on brain size in single or closely related species could contribute to increasing understanding of evolutionary biology, such as inter- and intraspecific variation, including domestication (Hemmer, 1990; Kruska, 2005). However, a major problem constraining research on intraspecific relationships between brain- and body-size parameters may be that many body-size parameters cannot be measured or estimated as accurately, or in a biologically meaningful way, as can those of brain size (Radinsky, 1978; Bininda-Emonds, 2000). There are practical difficulties in obtaining accurate individual body-size measurements, especially for large species, owing to natural fluctuations (e.g. seasonal and diurnal weight changes), measurement errors (e.g. measured in different ways or by different people) and problematic estimates (e.g. using an arbitrarily chosen population to represent a species or unassociated data for brain size and body weight) (Radinsky, 1978; Eisenberg, 1981; Bininda-Emonds, 2000). In reality, accurate and biologically meaningful measurement data are unavailable for many wild species in the field and alternative readily available parameters must be sought with which brain size can be compared. Skulls are common in museum collections, they can be measured repeatedly with a high degree of accuracy and skull size is often used in encephalization research (Schauenberg, 1971; Wayne, 1986; Yamaguchi et al., 2004).

Here, we investigate for the first time, using very large sample sizes, the intraspecific relationship between endocranial volume and skull size (greatest length of skull) of the lion (Panthera leo) and the tiger (P. tigris), with consideration of intraspecific variation and differences between wild and captive animals. We also compare their endocranial volumes with those of the closely related leopard (P. pardus) and jaguar (P. onca) to allow for a better understanding of their intrageneric phylogeny.

MATERIAL AND METHODS

Endocranial volume (cv; cm³) and greatest length of skull (gl; mm) were measured for 370 lions, 225 tigers, 32 jaguars and 42 leopards from European museum collections. A skull was classified as juvenile if the cemento-enamel junction of any permanent canine was not clearly visible above its alveolus. If these junctions were visible and yet the basioccipital–basisphenoid suture and/or frontal suture were still open, a skull was classified as subadult. If these sutures were closed, a skull was classified as adult. Only adults and subadults were analysed, owing to small sample sizes for juveniles. Based on museum labels, lions and tigers were categorized for sex, geographical origin, putative subspecies (Hemmer, 1974b; Mazák, 1981; Nowell & Jackson, 1996; Luo et al., 2004) and whether of wild or captive origin.

Greatest length of skull (from prosthion to inion) was measured using metal callipers to the nearest 0.05 mm (see Barnett et al., 2008). The cranial cavity was cleaned and filled with glass balls of c. 4 mm diameter, which were then either weighed using an electronic balance to the nearest 1 g or measured using a plastic cylinder to the nearest 1 cm³. Weights were converted into volumes using the equation: volume = 0.633 × weight + 0.939 based on linear regression coefficients (R² = 1.000, d.f. = 1, F = 24172.7, P < 0.001). The coefficient of variation for greatest skull length was 0.04 ± 0.01% (mean ± standard error, N = 5 skulls with three repeats each) and that for the cranial volume was 0.35 ± 0.08%. We are not aware of a widely accepted standard concerning accuracy, reliability and consistency of measurement and therefore follow Yamaguchi et al. (2004) in accepting a cut-off of less than 2% coefficient of variation for using measurements and volumes in subsequent analyses, which both measurements met. Data were log-transformed (ln) and regression lines were calculated. Residuals between data points and regression lines were analysed. All statistical analyses were carried out using SPSS (SPSS Inc., Chicago, IL, USA) and statistically significant differences were detected using ANOVA.

RESULTS

Greatest lengths of skull (mm), cranial volumes (cm³) and Schauenberg's indices (Schauenberg, 1969: greatest length of skull/cranial volume) for all four species of big cat are summarized in Table 1. The relationship between greatest lengths of skull and cranial volumes is shown in Figure 1. The interspecific regression line for all four species was ln(cv) = 0.943 [ln(gl)] + 0.100 (R² = 0.544, d.f. = 1, F = 789.78, P < 0.001), with a residual mean (M) of 0 and residual standard deviation (SD) of 0.130, whilst a better-fitted line was obtained with larger F-value and reduced residual SD, if tigers were excluded {ln(cv) = 0.978 [ln(gl)] – 0.173: R² = 0.793, d.f. = 1, F = 1670.03, P < 0.001, residual M = 0, residual SD = 0.086}. Residual analysis showed a significant difference in the distribution of data points around the regression line amongst the four species (d.f. = 3, F = 398.10, P < 0.001: Fig. 2). If tigers were excluded, the F value was substantially reduced, although the difference was still statistically significant (d.f. = 2, F = 16.56, P < 0.001). The intraspecific regression line for leopards was ln(cv) = 0.832 [ln(gl)] + 0.547 (R² = 0.646, d.f. = 1, F = 77.74, P < 0.001) and for jaguars was ln(cv) = 0.680 [ln(gl)] + 1.442 (R² = 0.451, d.f. = 1, F = 26.51, P < 0.001).

The intraspecific regression line for lions was ln(cv) = 0.667 [ln(gl)] + 1.637 (R² = 0.437, d.f. = 1, F = 280.91, P < 0.001). Residual analyses revealed no significant differences between sexes [male (N = 201), female (N = 159): d.f. = 1, F = 2.34, P = 0.127], ages [adult (N = 311), subadult (N = 49): d.f. = 1, F = 0.30, P = 0.586] or sex and age combined [adult male (N = 168), adult female (N = 143), subadult male (N = 33), subadult female (N = 16): d.f. = 1, F = 1.67, P = 0.197]. Therefore, sex and age classes were combined for the following analyses. There was a significant difference in residuals between captive and wild individuals [captive (N = 39 of which 30 were of known origin), wild (N = 306): d.f. = 1, F = 38.09, P < 0.001: Fig. 3], as well as between eight commonly recognized subspecies [melanochaita (N = 5), krugeri (N = 18), bleyenberghi (N = 31), nubica (N = 236), azandica (N = 7), senegalensis (N = 23), leo (N = 8), persica (N = 16): d.f. = 7, F = 16.74, P < 0.001: Fig. 4]. However, owing to the small overall sample size (N = 39), the average residual of captive animals may have been lowered by the eight Asian lions, which had the lowest average residual (Fig. 4). Additionally, the eight Barbary lions (leo) were all captive, potentially causing sampling bias. When only known-origin sub-Saharan lions were included in the analysis, the difference between captive and wild individuals was still statistically significant [captive (N = 11), wild (N = 296): d.f. = 1, F = 19.68, P < 0.001]. When only wild individuals were included in the analysis, there was still a statistically significant difference between subspecies [melanochaita (N = 2), krugeri (N = 14), bleyenberghi (N = 29), nubica (N = 228), azandica (N = 7), senegalensis (N = 16), persica (N = 7): d.f. = 6, F = 17.50, P < 0.001]. However, there was no statistically significant difference between wild sub-Saharan subspecies (i.e. excluding persica and leo: d.f. = 5, F = 0.94, P = 0.456). Actual values for all adult lions are summarized in Table 2. Whilst captive animals had significantly greater Schauenberg indices (i.e. smaller brains relative to skull size) than those of wild sub-Saharan lions (male d.f. = 1 F = 15.44 P < 0.001, female d.f. = 1 F = 7.88 P = 0.006), Asian lions showed the opposite trend (see Table 2). Residuals from pooled data for both sexes were significantly greater (i.e. larger brain size relative to skull size) for captive vs. wild Asian lions [wild (N = 7), captive (N = 8): d.f. = 1, F = 5.35, P = 0.038].

The intraspecific regression line for tigers was ln(cv) = 0.667 [ln(gl)] + 1.821 (R² = 0.503, d.f. = 1, F = 227.32, P < 0.001). Residual analyses revealed no significant differences between sexes [male (N = 126), female (N = 99): d.f. = 1, F = 0.07, P = 0.790], ages [adult (N = 193), subadult (N = 32): d.f. = 1, F = 2.61, P = 0.107] or sex and age combined [adult male (N = 103), adult female (N = 90), subadult male (N = 23), subadult female (N = 9): d.f. = 1, F = 0.001, P = 0.982]. Therefore, sex and age classes were combined for the following analyses.

In contrast to lions, there was no statistically significant difference (P = 0.05) in residuals between captive and wild tigers, although captives clearly showed lower mean residuals [captive (N = 34, all known origin), wild (N = 169): d.f. = 1, F = 3.31, P = 0.070: Fig. 3]. There was a significant difference in residuals between nine putative tiger subspecies [altaica (N = 10), virgata (N = 7), amoyensis (N = 10), corbetti (N = 15), tigris (N = 69), ‘jacksoni’ (N = 6), sumatrae (N = 44), sondaica (N = 47), balica (N = 7): d.f. = 8, F = 3.45, P = 0.001: Fig. 5]. When only wild individuals were analysed, there was still a significant difference in the residuals between putative subspecies [altaica (N = 2), virgata (N = 3), amoyensis (N = 9), corbetti (N = 6), tigris (N = 64), ‘jacksoni’ (N = 5), sumatrae (N = 31), sondaica (N = 38), balica (N = 7): d.f. = 8, F = 2.65, P = 0.01]. However, there was no significant difference between the six continental subspecies (wild only: d.f. = 5, F = 0.08, P = 0.996), but there was between the three Sunda Islands subspecies (wild only: d.f. = 2, F = 11.16, P < 0.000). Actual values for adult tigers from the Malay Peninsula and Sunda Islands are summarized in Table 3.

DISCUSSION

Brain size in the genusPanthera

The results clearly demonstrate that the tiger has a larger cranial volume for its skull size than do the lion, the jaguar and the leopard. As a result, a small tiger skull can be easily distinguished from either a large jaguar or leopard skull based on its cranial volume (see Fig. 1). Our results are broadly consistent with those of Eisenberg (1981) (except for the jaguar), which are largely based on previous literature. The encephalization quotient (the ratio between observed and expected brain weights for a defined body weight) of the tiger (1.35) is greater than those of the leopard (0.85) and the lion (0.57–0.83), but that of the jaguar (1.36) is greater than expected compared with our results, which are based on skull size instead of body weight (Eisenberg, 1981). Although we could not measure actual brain size (e.g. brain weight), endocranial volume is commonly used as an acceptable estimator of brain size for many mammals, including carnivorans (Gittleman, 1986a). Also, Röhrs & Ebinger (2001) suggest that closely related species (e.g. congeneric) tend to show similar ratios between brain and brain-case size. Therefore, we consider that the results reflect differences in actual brain size of the four closely related big cat species. We are not aware of any behavioural, ecological or sociobiological characteristics that distinguish the tiger from the other three Panthera species for which we have data – in fact, the lion might be regarded as unique, especially with regard to sociality (Nowell & Jackson, 1996; Sunquist & Sunquist, 2002). Encephalization and sociality in carnivores have been discussed previously (Hemmer, 1978a, b; Gittleman, 1986a, 1994; Pérez-Barbería et al., 2007). Although Hemmer (1978a) suggested that social species had proportionally larger brains, this was refuted by Gittleman (1986a) based on a larger sample size. Our results also do not support Hemmer's (1978a) suggestion concerning the four big cat species, which probably shared a common ancestor c. 3.7 million years ago (Johnson et al., 2006).

One of the few studies reporting correlations between life history parameters and brain size suggests that, amongst orang-utan species (Pongo pygmaeus and P. abelii) and subspecies (P. pygmaeus sspp.), a shorter interbirth interval correlates with smaller brain size (Taylor & van Schaik, 2007). Therefore, Gittleman's (1986b) general study of carnivorans gives the lion's interbirth interval as 25 months, the leopard's as 24 months and the tiger's as 32 months, which might explain differences in brain size. However, Gittleman's (1986b) data need careful re-evaluation. For example, Nowell & Jackson (1996) gave a range of 11–25 months (N = 38) for the lion's interbirth interval, an average 15 months for the leopard's and 20–24 months (N = 7) for the tiger's. There appears to be no convincing evidence that the tiger has a longer interbirth interval than the lion's.

There is a popular notion that tigers are ‘bigger’ than lions (e.g. Sunquist & Sunquist, 2002). Hemmer (1974a) suggested that the tiger has a relatively smaller head (skull length) for its body size (head-and-body length) than either the lion or the leopard, both of which possess similar head-to-body ratios. Therefore, the tiger's relatively bigger brain size may reflect its bigger body compared with that of the lion, which has a bigger skull relative to its body size. However, careful re-evaluation of original field data and relatively well-documented hunting records does not support this popular notion. The modern wild-living tiger has an estimated average body weight (i.e. excluding stomach contents) of c. 160 kg for adult males and c. 115 kg for adult females, whilst modern wild-living lion weigh c. 175 kg (males) and c. 120 kg (females), where ‘average’ is the mean body weight of commonly recognized putative subspecies (Yamaguchi, 2005a, b; Kitchener & Yamaguchi, in press). Therefore, we conclude that the tiger has a relatively bigger brain than the lion's (by c. 16%), given their very similar average body sizes.

However, the brain size to skull size ratios of the lion, the jaguar and the leopard are similar. These three species are phylogenetically closer to each other than to the tiger, sharing a common ancestor c. 2.9 million years ago (Johnson et al., 2006), and our results may reflect this phylogenetic relationship more than any functional explanation. Additionally, the general shape of the brain is slightly different amongst extant felids, such that relatively larger brains have a more globose shape (Radinsky, 1975). The tiger's skull is more globose than the lion's (Haltenorth, 1937; Poco*ck, 1939) and this may relate to brain shape, which in turn results in differences in relative brain size. However, the main question still remains – why did the tiger evolve a relatively bigger brain (or why did the other species evolve smaller brains) after the tiger's ancestor split from the ancestor to the other three species? Alternatively, can such a difference be explained by intrageneric variation or merely by chance? Answers to these questions might be found by analysing similar data from their fossil relatives and smaller members of the Pantherinae (e.g. snow leopard, P. uncia, and clouded leopards, Neofelis spp.), as well as a comparative anatomical study of which parts of the brain contribute to overall differences in brain size between lions and tigers.

Intraspecific variation

There is significant variation in brain size between commonly recognized putative subspecies of the lion and the tiger. This is comparable with that described for orang-utans (Taylor & van Schaik, 2007). The Asian lion has a relatively small brain compared with those of sub-Saharan lions, whose brain sizes are similar. Although the Barbary lion also had a relatively small brain, all the skulls in this study are from captives and so it is not clear whether this was characteristic of wild Barbary lions. However, mtDNA shows that the Barbary lion is phylogenetically closer to the Asian lion than to sub-Saharan lions (Barnett et al., 2006a, b). Therefore, the relatively small brain sizes of these two subspecies may reflect this close phylogenetic relationship.

There are interesting differences in brain size between putative tiger subspecies from the Malay Peninsula and the Sunda Islands. Tiger populations in this region have been connected to each other intermittently, owing to sea-level changes associated with glacial–interglacial oscillations, and the effects of the gigantic Toba eruption c. 75 000 years ago (Kitchener & Dugmore, 2000; Kitchener & Yamaguchi, in press). This area has tigers with both the relatively biggest (Balinese and Javan tigers) and smallest (Sumatran and Malayan tigers) brains, compared with those of other continental subspecies, whose brains are similar to each other's (Table 3, Fig. 5). Differences between the three adjacent island subspecies are notable, because, based on skull morphology, Mazák & Groves (2006) suggested that Balinese and Javan tigers are closer to each other than to the Sumatran tiger. Our results appear to be consistent with those of Mazák & Groves (2006) suggesting tentatively that past geological events resulted in two distinct tiger lineages in the Malay Archipelago, with a possible boundary between Sumatra and Java. However, further evidence is required; for example, ancient biomolecular research on Balinese and Javan tigers, in order to ascertain whether brain size differences amongst Sunda Islands tigers reflect their intraspecific phylogeny.

Life history

It is well known that domesticated animals have smaller brains than those of their wild ancestors, for example, c. 20–30% in carnivorans (Hemmer, 1990; Kruska, 2005). It has also been suggested that the small brains of domesticated animals are genetically determined – for example, when wolves (Canis lupus) and poodles (C. familiaris) were hybridized experimentally, the resultant offspring had intermediate brain sizes – and therefore are the result of artificial selection (Kruska, 2005). Our results show that captive lions and tigers tend to have smaller brains (c. 3.5–10.5% actual volume without standardizing for the skull size – see Table 1) than those of wild ones. Although we do not know for how many generations these animals have been bred in captivity, we assume that animals with known geographical origins are probably wild-caught, although they may have been reared in captivity, which could have affected their skeletal development. If so, most captive animals in this study are wild-caught and statistically significant decreases in relative brain size occurred within animals' lifetimes. Hollister (1917) observed that four wild-caught, captive-reared East African lions had smaller brains (by c. 20% actual volume) than six wild lions of similar ages from similar locations. Therefore, a substantial reduction in brain size may occur in captive-reared wild animals without genetic change.

Our results also suggest an unexpected contrast between Asian and sub-Saharan lions, which may relate to a possible effect of captivity on their brain sizes. Captive Asian lions have larger brains than those of wild ones, whilst sub-Saharan lions show the opposite tendency. This observation may be the result of the small sample size for the Asian lion, even although we measured most known Asian lion skulls in European museum collections. It is suggested that food availability may affect brain size amongst orang-utan populations; for example, female orang-utans living in resource-poor environments appear to have the smallest brains (Taylor & van Schaik, 2007). However, the estimated population density of the only remaining Asian lion population in the Gir Forest is c. 7/100 km², which is comparable with the upper range of those in sub-Saharan Africa (Nowell & Jackson, 1996), thereby precluding the possibility that Asian lions are living in a poor environment. However, we speculate that wild Asian lions are living in an environment where (unidentified) factors reduce brain size and they are (partially) released from these factors in captivity, in spite of other (again unidentified) factors in captivity reducing substantially the brain sizes of sub-Saharan lions. Examining skulls of animals from different putative subspecies, which are reared in very similar captive conditions, would provide useful information.

Acknowledgements

We thank the European Union SYNTHESYS programme for allowing NY to visit Musée Royale de l'Afrique Centrale, Tervuren, Belgium, Muséum National d'Histoire Naturelle, Paris, France and Museum für Naturkunde der Humboldt-Universität, Berlin, Germany. We also thank Paula Jenkins, Daphne Hills, Louise Tomsett at the Natural History Museum, London, Malcolm Harman at the Powell-Cotton Museum, Kent, Malgosia Nowak-Kemp at the Natural History Museum, University of Oxford, Oxford, UK, Robert Asher, Peter Giere, Irene Thomas, Wolf-Dieter Heinrich at the Museum für Naturkunde der Humboldt-Universität, Berlin, Katrin Krohmann, Thomas Martin at the Forschungsinstitut und Naturmuseum Senckenberg, Frankfurt, Doris Mörike at the Staatliches Museum für Naturkunde in Stuttgart, Stuttgart, Germany, Vincent Nijman, Adri Rol at the Zoölogisch Museum, University of Amsterdam, Chris Smeenk, Hein van Grouw at the Nationaal Natuurhistorisch Museum, Leiden, the Netherlands, Wim Wendelen at the Royal Museum for Central Africa, Tervuren, Belgium, Jacques Cuisin, Francis Renoud, at the Muséum National d'Histoire Naturelle, Paris, Marie-Dominique Wandhammer, Virginia Rakotondrahaja at the Musée Zoologique, Strasbourg, France, for their kind support for the access to their collections.

References

© 2009 The Linnean Society of London

Issue Section:

Original Articles

Email alerts

Citing articles via

More from Oxford Academic