COVID-19 Corona Virus
South African Resource Portal
COVID-19 Corona Virus
South African Resource Portal
COVID-19 Corona Virus
South African Resource Portal
Submit an article to Indago - a peer reviewed journal
Submit an article to Indago - a peer reviewed journal
Submit an article to Indago - a peer reviewed journal

– Animal body size, and in particular body mass (live weight), determines many biological and ecological traits including food intake, metabolism, thermoregulation, generation time, longevity, growth rate, home range size, reproductive strategy, extent of sexual dimorphism and even extinction risk (e.g. Peters 1983; Damuth & MacFadden 1990). Being large comes with certain competitive advantages. First, it is associated with reduced metabolic rates and energy demands per unit mass (an elephant uses much less energy relative to its mass than a mouse does) and, second it is associated with a reduced risk of predation. Consequently, natural selection generally favours evolution towards larger mass, a phenomenon known as Cope’s Rule (Damuth & MacFadden 1990).

Climate is another factor influencing body size. In cold environments, animals tend to be larger than in warm ones. This, because their larger volume (mass) to surface area ratios mean that rates of heat production relative to heat loss are also higher, such that larger size keeps them warm. This effect, arguably the first ever macroecological pattern described by natural science, is called Bergmann’s Rule (Damuth & MacFadden 1990). Another explanation for variation in body size is the temperature-size rule; this model predicts that in warmer temperatures an animal’s development rate is accelerated (progression through the life cycle towards reproductive maturity) and faster than that of growth rates (increases in mass). Consequently, mature adult body sizes in warm climates are small. While a (negative) latitude-body size relationship has been found both within and across many species, there are known exceptions to this temperature-size rule, such as a species of grasshopper, Chorthippus brunneus, that in high temperatures attains a larger body size at maturity than in cool climates (Walters & Hassall 2006).

A third factor that can influence body size, is human predation ̶ one of the many areas in which humans directly impact the functioning of the natural world (Darimont et al. 2009). When hunting or gathering prey, humans tend to target the large, reproductive-aged adult animals since these provide the greatest quantities of meat and other products. Consequently, animals exposed to continuous and intensive ’harvesting’ by people, show rapid changes in body size as the large, mature animals are removed from the population.

Tortoises and body size/mass change
Tortoise remains are commonly found in archaeological assemblages across the globe, indicative of their role in past human diet. Although tortoises grow throughout their life, researchers examining animal bone assemblages from South African archaeological sites, have reported a reduction in average tortoise size/mass, with larger tortoises in assemblages dated to the Late Pleistocene (Middle Stone Age; ∼130,000–40,000 years ago), and smaller ones in the Holocene (Later Stone Age; ~40,000 to 2000 years ago). The researchers attributed the observed size reduction to either: (1) climate change, with drier and warmer conditions, coupled with impoverished nutrition, resulting in smaller tortoises being available in the landscape and so available as prey (e.g. Henshilwood et al. 2001); or (2) overkill, namely that tortoise size diminished as a result of excessive human predation on large reproductively mature tortoises, as a consequence of increased demand for protein due to an explosion of human population (e.g. Klein & Cruz-Uribe 1983; Steele & Klein 2005/6, 2013). Both hypotheses are plausible, so the cause of this diminution in tortoise size continues to be debated.

The Later Stone Age (LSA) levels of Wonderwerk Cave in the Northern Cape Province preserve evidence of an association of humans and the Leopard tortoise (Stigmochelys pardalis), extending back over the last ~14,000 years, a period that straddles the end of the Late Pleistocene through Holocene epochs (Holt et al. 2018, 2019). Notably, the LSA in the cave is marked by extensive climatic fluctuations – a shift from extremely moist to extremely arid conditions (Ecker et al. 2018). It is also characterized by a change in the intensity of occupation from ephemeral to intense, as well as by changes in human culture (from the Oakhurst/Kuruman lithic techno-complex to the more recent Wilton and finally the most recent period, the Ceramic LSA (Humphreys & Thackeray 1983). The Wonderwerk Cave tortoises serve as an excellent example with which to examine the relative impact of climatic versus human factors on tortoise size.

To help resolve this debate, we studied the tortoise body size/mass throughout the LSA sequence of Wonderwerk Cave (Codron et al. 2022). We asked whether: (1) tortoises experienced a net change in their body size/mass over this period, (2) this led to a smaller (as expected) average adult size, and (3) the timing and direction of these changes can be explained by climate change or by excessive human predation.

How to estimate body mass/size of fossil animals?
Faced with remains of only bones and teeth, palaeontologists have to use indirect approaches to estimate live weights (body mass) of ancient animals. The most common method of “putting the meat back on the bones” makes use of the fact that the size dimensions of various animal body parts are closely related not only to each other, but to the organisms’ body mass. The precise nature of these size-mass relationships varies depending on the animal group and the studied body part, but in general they follow a pattern of allometric scaling, which is explained in Figures 1a and 1b. Provided we can characterize this allometric relationship mathematically – which requires analysis of size data from living animals of known body mass – we can estimate the live weight of animals based on measurements of the fossil remains (Figs. 1c & 1d). Fortunately for us, several previous studies of tortoises have reported a clear correlation between the long bone measurements and the body mass (e.g. Miller & Richard 2005; Llorente et al. 2007; Esker et al. 2019).

Figure 1. A short introduction to allometric scaling. The body size of an animal is positively related to its body mass (live weight), i.e. as the mass increases, so too does the animals’ size. However, the increase in size is less-than-proportional to the increase in mass, because mass reflects a 3D volume whereas size parameters like length and width, are measured on a ‘flat’ (two-dimensional) surface (the blue curve in (a)); in rare cases, both measures increase at the same rate, i.e. one to one, forming an ‘isometric’ relationship (the straight red line in (a)). In (b), the relationship is reversed and reflects an increase in body mass proportionally to the cube of the length, thus offering a mathematical method that allows calculation of the animal body mass from a measurement of its size (e.g. length). The accuracy of the body mass estimate depends on how strong the relationship is: in (c) individual animals (blue circles) lie close to the allometric curve (black line), and so the estimate is accurate, but in (d) other factors (differences between the sexes, ages, habitats, or many other variables) influence the body mass and so the prediction accuracy is poorer.

What to do with fossil tortoise remains
The difficulty working with fossils is that, unlike in Hollywood movies like Jurassic Park, complete skeletons are very rarely found. Instead, palaeontologists and archaeologists must work with many isolated bones, or often just bone fragments, to piece together the puzzle of the past. Another hurdle, specific to the tortoise remains recovered from the Wonderwerk Cave sequence, is that many are limb bones or other parts of the post-cranial skeleton, which cannot be measured in living tortoises without killing and dissecting them.

Figure 2. The four-step approach to the allometric reconstruction of the body mass.

To overcome the problem, we used a four-step approach (Fig. 2) in our allometric reconstructions of body mass based on the fossil remains:
i) We constructed size allometries based on measurements of the shell to 11 post-cranial bone measures, taken on modern skeletons of 152 Leopard tortoises collected from farms in the Northern Cape and Free State (Holt et al. 2021). These animals had been killed after coming into contact with electrified farm fences, an animal conservation and welfare concern we have discussed elsewhere (Holt et al. 2021).
ii) Next, we took the same five shell measurements (as previously measured on the modern skeletal sample ̶ Fig. 3) on 192 live Leopard tortoises released into Grant’s Hill, Bloemfontein, between 2016–2018. We also weighed these animals before release, thereby obtaining data on their body mass to establish their body mass-size allometric equations.

Figure 3. Measurements taken on shells of live and skeletonized Leopard tortoises to establish their body mass-size allometric equations.

iii) We then used these data to construct the allometry of shell length to body mass of these live animals, to measurements of shell and limb bones of the modern skeletal data set. In this manner, we were able to estimate body mass of the dead tortoises based on their limb bone size.
iv) Lastly, we measured over 350 fossil Leopard tortoise limb bones recovered from the Wonderwerk Cave sequence. Using the data already collected from the live tortoises and modern skeletal sample, we were able to do a back-calculation and estimate the shell dimensions, and so body mass, of each individual found in Wonderwerk Cave.

Figure 4. Location of the farms where the tortoises killed by electric fences were collected, Grant’s Hill where the live tortoises were released, and Wonderwerk Cave where the Later Stone Age tortoises were excavated.

Insights from tortoise allometry
Our four-step allometric approach was a success! Results from the analysis of the modern skeletal sample of Leopard tortoises, showed that the 11 post-cranial skeletal elements we had measured, reliably predicted tortoise body mass (>90% of the variation; Fig. 5). Shell size was an even better predictor of body mass, with all five shell dimensions explaining around 99% of the variance in mass! Moreover, the allometric relationships between the different bones and the shell were not influenced by size and shape differences between male and female tortoises, nor did they vary depending on whether the bone was from the left or right side of the skeleton. This is important, because it means the same equations could be applied to the entire Wonderwerk Cave fossil assemblage, for which the identification of the animal’s sex from individual bones, is impossible.

Figure 5. Reconstructing the fossil tortoise body mass. We started by establishing the allometric relationship between the shaft width of the humerus and shell length, based on data from tortoise skeletons (a). Next, we identified the allometric relationship between the shell length and the body mass, based on measurements of live animals (b). We then measured the widths of fossil humeri and used the equation from (a) to estimate the shell length of the fossil animal (red segment in (c)). Finally, we took the estimated shell length and, based on the equation derived from (b), and back-calculated to estimate the live body mass of the fossil animal (red segment in (d)). Note that all fossil animals recovered from Wonderwerk Cave lie on the smaller side of the size range of the species.

In total, our study provided 999 allometric equations that future researchers can use to study the size and mass relationships in Leopard tortoises, and at the same time creating opportunities to ask new questions about their evolution. For example, the ulna (elbow bone) was a notable under-achiever: its dimensions account for less than 50% of the variation in the body mass. Such weak allometric relationships imply that the ulna perhaps less of a role in weight-bearing than some of the others, and so is adapted to perform some other yet-to-be-discovered function.

Evolution of tortoise body mass at Wonderwerk Cave
Armed with this set of reliable allometric equations, we were able to make confident estimates of the live body masses of the fossil tortoises from Wonderwerk Cave. Interestingly, we found that their average body mass was quite low, ranging between 280 and 550 g (Fig. 5d), though this varied a bit between the different skeletal element used for estimation. By contrast, the average body mass of modern Leopard tortoises in South Africa is between 10 and 20 kg (Boycott & Bourquin 2000). The Later Stone Age people that occupied Wonderwerk Cave clearly did not have a preference for, or access to, very large individuals!

Taking our analysis a step further, we subdivided the fossil sample into seven units based on their age within the cave sequence, with Layer 4d the oldest (dated to ~11,800–10,400 years before present) and Layers 3a/2b the most recent (dated to ~500–2,300 years before present; Table 1). These units are correlated with known changes in climate in the past, as well as shifts in human cave occupation intensity and the type of stone tool industries found in the cave i.e. tools that share characteristics (Ecker et al. 2018; House et al. 2022; Fig. 5).
Table 1. Wonderwerk Cave Later Stone Age archaeological levels, dates and past climate reconstructions (based on Ecker et al. 2018: Table1; House et al).
Note that the dates for each layer do not overlap indicating breaks in human cave occupation. An exception is Layer 4a/4LH that partly overlap with Layer 4b. This, as layer 4LH is found only in a specific part of the cave.
Later Stone Age levels are coloured to show tortoise size/mass trends- smaller in 4c & 4d and larger from 4b upwards.

As illustrated in Fig. 6, the average body mass of tortoises was smallest in the oldest parts of the sequence pre-dating 6900 years ago (Layers 4d–c), and thereafter increased, remaining almost constant from ~6900 years ago (Layers 4b) through to the uppermost layers (Layer 2b) some 500 years ago. Thus, the average body mass of tortoises did not change at the transition from the Kuruman/Oakhurst to Wilton stone tool industry or again between the Wilton and Ceramic LSA. Additionally, body mass did not follow the expected trend of cave occupation intensity, which serves as a proxy for population size or frequency of cave occupation. If human predation was the key factor influencing body size/mass, then we would expect to find the smallest tortoises in with levels experiencing the most intense cave occupation. However, at Wonderwerk Cave the opposite trend is found. The lowest average tortoise body mass occurs in levels 4d and 4c when cave occupation was scarcest, while the largest average body mass occurs in Layer 4b onwards, periods that experienced more intense cave occupation (Rhodes et al. 2022). Thus, we have no clear evidence that human activity had any adverse impact on the evolution of tortoise size over the occupation time of Wonderwerk Cave layers.

Climate change, on the other hand, probably played a more important role. The earliest part of the Wonderwerk Cave sequence is characterized by warm, semi-arid to arid conditions. Thus, the low average body mass of tortoises recovered from this period is consistent with the predictions of Bergmann’s Rule. The average body mass increased in the middle parts of the sequence (layers 4b and 4a/4aLH), characterized by warm, but moist and humid conditions, perhaps reflecting the greater nutritional value of the food that was available to the tortoises.
While, from level 3b upwards climatic conditions appear to have changed and become warmer and drier than in 4b and 4a/4LH, yet the average tortoise body mass did not change over this period. It might be that the change in climate at this point in time was too small to have a significant impact, but more research is needed to solve this conundrum.

Figure 6. Evolution of the Wonderwerk Cave tortoise body mass through time. Our results indicate a clear increase in average size from the oldest parts of the sequence (Layers 4c and 4d; purple), but no further change through the rest of Holocene (green). The black horizontal bars are medians, and each circle in the graph depicts an individual fossil.

The Later Stone Age Leopard tortoise assemblages of the Wonderwerk Cave sequence documents an approximately 12,000-year-long association with humans. Using information derived from hundreds of measurements of modern Leopard tortoises in the region, we were able to reconstruct changes in the tortoise body mass pattern over this period. Our study revealed an evolutionary shift to increased body size/mass as recently as the Middle Holocene, ~6900 before present.

The influence of climate on the body size/mass of many vertebrate species is not limited to the past. With global warming we are witnessing size change in numerous species, from small rodents to large sized mammals (e.g. Yom-Tov & Geffen 2018; Nengovhela et al. 2020). Notably, global warming may adversely affect reptiles (including tortoises) specifically incubation temperature (Booth 2006), and so impacts hatchlings in terms of their size (their growth rate), sex, shape, colour, behaviour etc. It may also affect the distribution range of reptile species. Consequently, trends observed in the fossil record, such as at Wonderwerk Cave, may improve our understanding of the outcome of climate change on present-day biodiversity.

We would like to acknowledge the Palaeontological Scientific Trust (PAST) for their financial support towards SH’s PhD research, on which this paper is based.


Booth, D.T. 2006. Influence of incubation temperature on hatchling phenotype in reptiles. Physiological and Biochemical Zoology, 79(2), 274-281.

Boycott, R.C. & Bourquin, O. 2000. The Southern African tortoise book. Hilton, South Africa: O. Bourquin.

Codron, D., Holt, S., Wilson, B. & Horwitz, L.K. 2022. Skeletal allometries in the leopard tortoise (Stigmochelys pardalis): predicting chelonian body size and mass distributions in archaeozoological assemblages. Quaternary International, 614, 59-72.

Darimont, C.T., Carlson, S.M., Kinnison, M.T., Paquet, P.C., Reimchen, T.E. & Wilmers, C.C. 2009. Human predators outpace other agents of trait change in the wild. Proceedings of the National Academy of Sciences, 106(3), 952-954.

Damuth, J. & MacFadden, B.J. 1990. Body size in mammalian paleobiology. Cambridge, UK: Cambridge University Press.

Ecker, M., Brink, J., Horwitz, L.K., Scott, L. & Lee-Thorp, J.A. 2018. A 12,000 year Holocene record of changes in herbivore niche separation and palaeoclimate (Wonderwerk Cave, South Africa). Quaternary Science Reviews, 180, 132-144.

Esker, D., Forman, S.L. & Butler, D.K. 2019. Reconstructing the mass and thermal ecology of North American Pleistocene tortoises. Palaeobiology, 45, 363-377.

Henshilwood, C.S., Sealy, J.C., Yates, R., Cruz-Uribe, K., Goldberg, P., Grine, F.E., Klein, R.G., Poggenpoel, C., Van Niekerk, K. & Watts, I. 2001. Blombos Cave, Southern Cape, South Africa: Preliminary Report on the 1992-1999 Excavations of the Middle Stone Age Levels. Journal of Archaeological Science, 28, 421-448.

Holt, S., Codron, D. & Horwitz, L.K. 2018. Bone mineral density in the leopard tortoise: Implications for inter-taxon variation and bone survivorship in an archaeozoological assemblage. Quaternary International, 495, 64-78.

Holt, S., Horwitz, L.K., Hoffman, J. & Codron, D. 2019. Structural density of the leopard tortoise (Stigmochelys pardalis) shell and its implications for taphonomic research. Journal of Archaeological Science: Reports, 26 Art. 101819.

Holt, S., Horwitz, L.K., Wilson, B. & Codron, D. 2021. Leopard tortoise Stigmochelys pardalis (Bell, 1928) mortality caused by electrified fences in central South Africa and its impact on tortoise demography. African Journal of Herpetology, 70(1), 32-52.

House, A., Bamford, M. & Chikumbirike, J. 2022. Charcoal from Holocene deposits at Wonderwerk Cave, South Africa: A source of palaeoclimate information. Quaternary International, 614, 73-83.

Humphreys, A.J.B. & Thackeray, A.I. 1983. Ghaap and Gariep. Cape Town: South African Archaeological Society.

Klein, R.G. & Cruz-Uribe, K. 1983. Stone Age population numbers and average tortoise size at Byneskranskop Cave 1 and Die Kelders Cave 1, Southern Cape Province, South Africa. The South African Archaeological Bulletin, 38(137), 26-30.

Llorente, G.A., Ruiz, A., Casinos, A., Barandalla, I. & Viladiu, C. 2007. Long bone allometry in tortoises and turtles. In: Wynekon, J., Godfrey, M.H. & Bels, V. (Eds), Biology of turtles: from structures to strategies of life. CRC Press, pp. 85-95.

Miller, K. & Richard, G.F. 2005. Influence of body size on shell mass in the Ornate Box Turtle, Terrapene ornata. Journal of Herpetology, 39, 158-161.

Nengovhela, A., Denys, C. & Taylor, P.J. 2020. Life history and habitat do not mediate temporal changes in body size due to climate warming in rodents. PeerJ, 8, Art. e9792.1.

Peters, R.H. 1983. The ecological implications of body size. Cambridge, UK: Cambridge University Press.

Rhodes, S., Goldberg, P., Ecker, M., Horwitz, L.K., Boaretto, E. & Chazan, M. 2022. Exploring the Later Stone Age at a micro-scale: New high-resolution excavations at Wonderwerk Cave. Quaternary International, 614, 126-145.

Steele, T.E. & Klein, R.G. 2005/6. Mollusk and tortoise size as proxies for Stone Age population density in South Africa: implications for the evolution of human cultural capacity. Munibe, 57, 221-237.

Walters, R.J. & Hassall, M. 2006. The Temperature-Size Rule in Ectotherms: May a general explanation exist after all? The American Naturalist, 167(4), 510-523.
Yom‐Tov, Y. & Geffen, E. 2011. Recent spatial and temporal changes in body size of terrestrial vertebrates: probable causes and pitfalls. Biological Reviews, 86(2), 531-541.

Comments are closed.