Introduction

Elephant hunting is a risky business at the best of times (Cooper, 1914:63).

The current global extinction crisis has sharpened focus on prior extinction episodes and the possible role that humans may have played, particularly in the disappearance of large game as a result of human hunting. One of the most controversial of those episodes is the disappearance of megafauna, especially proboscideans, at the close of the Pleistocene (e.g., Faith et al., 2020; Grayson, 2016; Grayson & Meltzer, 2015; Koch & Barnosky, 2006; Meltzer, 2015; Sandom et al., 2014; Smith et al., 2019; Surovell et al., 2005). Although their extinction is sometimes associated with human predation, other explanations attribute their disappearance to Late Pleistocene-Holocene climate change, or a combination of natural and anthropogenic processes (e.g., Boulanger & Lyman, 2014; Guthrie, 2006; Koch & Barnosky, 2006; MacDonald et al., 2012; Meltzer, 2015). Late Pleistocene-Holocene climate change likely influenced the spatial distribution, abundances and genomic diversity of biotic communities (Cooper et al., 2015; Mondanaro et al., 2021; Nadachowski et al., 2018; Seersholm et al., 2020; Wang et al., 2021), but the role and scale of anthropogenic hunting in the disappearance of megafauna remains contentious (e.g., Eren et al., 2021; Grayson & Meltzer, 2002, 2015; Haynes, 2002; Nikolskiy et al., 2011; Shipman, 2015; Smith et al., 2019; Surovell et al., 2005). To bolster an incomplete archaeological record, scholars turn to the ethnohistoric records of elephant hunting to demonstrate the plausibility and potential productivity targeting proboscideans. These records, in concert with the assumption that large-sized animals are always the most efficient prey, has led some to conclude that preferential targeting and, in fact, over-hunting of high-ranked megafauna led to their extinction (Agam & Barkai, 2018; Ben-Dor & Barkai, 2020; Haynes, 2002; Osborn, 2016; Shipman, 2015).

A central problem with the over-hunting argument is that direct archaeological evidence for the active hunting of proboscideans (and other megafauna) remains rare (Grayson & Meltzer, 2015; but see Boëda et al., 1999, Lemorini et al., 2022; Mackie et al., 2020; Mussi & Villa, 2008; Wojtal et al., 2019). Direct evidence includes the presence of hunting weaponry embedded in animal skeletal remains, bone or lithic tool damage associated with weaponry impacts and, under some circumstances, the close physical association between tools and butchered or processed bones (e.g., Eren et al., 2021; Gaudzinski-Windheuser, 2016; Grayson & Meltzer, 2002, 2015; Villa et al., 2004; Wojtal et al., 2019). In a recent systematic evaluation of North American archaeological sites, for example, Grayson and Meltzer (2015) identified only 15 sites with direct and compelling evidence of subsistence-oriented hunting or exploitation (including scavenging) of megafauna (but see Mackie & Haas, 2021 who propose a higher number of possible kills). Most of these instances involved Mammuthus remains and some, but not all, likely represent hunting. In Europe and Eurasia direct archaeological evidence for proboscidean hunting predating 30,000 years ago is uncommon (e.g., Drucker et al., 2017; Pitulko et al., 2016; Thieme & Veil, 1985; also see Villa et al., 2004), but mammoth remains become more abundant in Gravettian and EpiGravettian Upper Paleolithic sites (see Braun & Palombo, 2012; Nikolskiy & Pitulko, 2013; Markova et al., 2010). Some contexts, such as the Pavlovian sites (25–29 ka bp), contain large quantities of mammoth bones in arrangements suggesting living spaces and features. These sites and other material evidence suggest that mammoths were important in the lifeways of some populations at certain times and places (e.g., Bocherens et al., 2015; Drucker et al., 2017). However, the taphonomic histories of some of these assemblages and origins of the mammoth bones are complicated and not unambiguous (see Bosch, 2012; Perri et al., 2015). In Africa, Faith et al., (2018) document long term declines in premodern and later modern proboscideans associated with ecological change spanning from 4.6 mya through the Late Pleistocene, suggesting that hominins played a very minor role their demise (also see Cantalapiedra et al., 2021; Faith, 2014).

Proponents of the over-hunting hypothesis argue that the paucity of prehistoric evidence is linked to preservational biases, gaps in the archaeological and paleontological records, and taphonomic factors influencing the visibility and detectability of sites (Koch & Barnosky, 2006; Meltzer, 2015: 40–41; Surovell & Grund, 2012). Martin (1973, 1984), one of the best-known proponents, argued that the speed and focus at which Pleistocene big game hunters dispatched naive megafauna mitigated against preservation and promoted the quick burial of evidence (the so-called Blitzkrieg model). The seemingly close chronology between the arrival of Pleistocene big-game hunters and demise of megafauna on continents and in regions where modern humans are relatively late migrants is also viewed as evidence supporting anthropogenic hunting and over predation of megafauna (but see Boulanger & Lyman, 2014; Meltzer, 2015). Cumulatively, the available physical evidence suggests that mammoths and other premodern proboscideans were sporadically hunted and/or scavenged and utilized for edible and non-consumptive products, such as ivory and bones, by some populations at certain places and certain times. Importantly, while instances of prehistoric proboscidean predation demonstrate that hunters dispatched these large mammals, they do not demonstrate the frequency of acquisition or over-hunting to the point of extinction.

Historic, Ethnohistoric, and Ethnographic Sources: Contextual Considerations

To bolster that incomplete and sparse archaeological record, researchers often turn to ethnohistoric records to provide insights into the plausibility, technological methods and productivity of proboscidean hunting. The richest ethnohistoric records are from Africa, dating from the late eighteenth through the early twentieth centuries, and largely pertain to savanna (Loxodonta africana) and forest (L. cyclotis) elephantsFootnote 1; a much smaller body of data exist for the third living species, Asian elephants (Elaphas maximus). These records encompass ethnographic studies and historic journals and diaries of big-game sport or trophy hunters, explorers, missionaries, colonial administrators and colonists, and a few indigenous recollections.

Most archaeologists recognize that these records are the products of specific historical technological, economic and sociopolitical contexts that make their analogical value to the past highly contestable. For example, the contexts of these records are unrepresentative of the circumstances faced by traditional hunter-gatherers (Haynes & Klimowicz, 2015; Saunders, 1992). Even so, some argue that the wide range of techniques used to acquire elephants, scale of hunting, and ivory exports reflects the ease of proboscidean procurement (e.g., Agam & Barkai, 2018; Ben-Dor & Barkai, 2020). Others draw inferences from ethnographic studies showing that large game are often preferred by hunters and on theoretical grounds should be the most highly valued prey whenever encountered (e.g., Wolfe & Broughton, 2020). Before any conclusions can be drawn from these records, however, researchers must consider the broader context(s) of historical elephant acquisition and how it influenced hunting productivity, valuation, and the scale of hunting.

Many of these records are written by outsiders and do not represent indigenous perspectives and suffer from the attendant problems associated with colonial and foreign perspectives. Many give an adventured picture of hunting and often present narrow and negative views of indigenous peoples and animals. Of these, a number of sources report information that is clearly not based on first-hand interactions or reflects limited contact with indigenous peoples (e.g., Tjader, 1910). Surprisingly, only a handful of first-hand observations of the use of traditional technologies used to acquire elephants exist (e.g., Du Chaillu, 1861; Graça, 1890; Harako, 1981; Livingstone, 1858; Usher-Wilson, 1947) and only a few indigenous voices or perspectives are heard among these sources. The only available sources recording indigenous perspectives on African proboscidean hunting are interviews or observations conducted recently, especially in the mid to late twentieth century by ethnographers, and well after laws were introduced limiting elephant hunting (Bahuchet, 1993; Duffy, 1984; Harako, 1981; Ichikawa, 2021; Lewis, 2021; Lupo, 2005; Turnbull, 1983).Footnote 2

Importantly, existing ethnohistoric records contain observations of hunting elephants within the socioeconomic contexts of a burgeoning ivory-trade, colonization and the slave trade, especially during the late eighteenth through the early twentieth centuries, and based on sport and trophy hunting beginning in the latter part of the nineteenth century. This period is characterized by regionally specific, complicated, and often catastrophic processes such as depopulation in response to the slave trade, tribal warfare, epidemics and colonial development. All these processes influenced demography, economy, technology, religion, and social and political organization that are well beyond the scope of this paper (but see Alpers, 1967; Steinhart, 1989; Walker, 2009). While the acquisition of ivory and trophies drove much of the hunting, it is important to note that in concert with other intersecting processes, elephant acquisition had transformative and far-reaching effects on indigenous economies, sociopolitical systems and ecological systems (see Galvin, 1986; Håkansson, 2004; Meredith, 2014, Spinage, 2012; Steinhart, 2000, 2001; Walker, 2009). But it is the inter-related impacts of these contexts that are central to understanding the scale of harvesting, productivity and valuation of elephants to indigenous populations.

Elephant ivory (or “white gold”) was an established tradeable commodity (Alpers, 1992) that became part of regional economic interactions by at least 5000 years ago, if not earlier (de Flamingh et al., 2021; Schuhmacher et al., 2009; Valera et al., 2015; Ylvisaker, 1982). However, the demand and trade of ivory significantly intensified from the eighteenth through the nineteenth centuries as measured by exports to North America, Europe and Asia (Alpers, 1967; Von Oppen, 1993; Spinage, 2012; Thorbahn, 1979) with estimates suggesting the scale of harvesting was unprecedented. In some locations at certain times it resulted in the localized reduction in numbers and/or extirpation of elephants (Feinberg & Johnson, 1982; Spinage, 1973, 2012: Fig. 16.6; Walker, 2009). For instance, a tenfold increase in the price of ivory between the late eighteenth and early ninenteenth centuries greatly amplified the hunting efforts of indigenous and foreign hunters in East Africa and eventually forced excursions inland as elephant populations closest to the coast declined (Coutu, 2015). Similarly, a 300% increase in the price of ivory after 1830 CE increased hunting in south central and southeastern Africa (Von Oppen, 1993:63). In some places other trade commodities, especially slaves, were intertwined with ivory expeditions enhancing the profits of long-distance treks (Sundstrüm, 1974:70, 80, 88–91; Walker, 2009:119). In response to the demand for ivory and other products, expeditions sometimes involving hundreds of people and spanning one to three months were not uncommon, especially in the late nineteenth century (e.g., Beachey, 1967).

Against this backdrop, the profits generated from ivory and later trophy hunting gave rise to an industry with large numbers of indigenous people working as professional hunters, middlemen, commercial vendors, porters, guides, trackers and assistants to foreign hunters. The number of indigenous Africans engaged in elephant hunting during the eighteenth and nineteenth centuries remains unknown, but the quest for profits gave rise to many so-called “full-time elephant hunters” among peoples who were not hunter-gatherers and were formerly engaged in other activities (see Steinhart, 1989, 2001; Walker, 2009:66). The Vili, for instance, who were frequently cited as traditional elephant hunters, were originally food producers and were among the first to engage in ivory trade with Portuguese and others by the late sixteenth century (Walker, 2009:66). Traditional hunter-gatherers were also engaged in elephant hunting and the ivory trade, but considerably less information is known about their involvement (but see Lupo, 2016). For some, the increased demand for ivory and other trade goods had an especially pivotal influence on existing mutualistic or patron/client arrangements with neighboring food producers. Among the Waata, Okiek and Mbuti, existing power disparities with neighboring food producers were amplified by the ivory trade and influx of European trade items. Early ethnographers note that neighboring food producers historically requested that hunter-gatherers target elephants to protect their plantations from marauding animals (Putnam, 1948; Schebesta & Schütze, 1954). But with increased market demand, hunters who had traditionally supplied meat through trade arrangements were compelled to hunt elephants and surrender much of the ivory and meat to local elites who reaped the profits (e.g., Lalouel, 1950; Richards, 1995).

Importantly these records shed light on how context, circumstances and processes influenced the scale of harvesting, the range of techniques used, and the productivity and valuation of elephant products. It is clear, for example, that economic rewards generated by the high market demand for ivory from the eighteenth and nineteenth centuries incentivized indigenous populations and greatly increased the frequency of proboscidean hunting to a level well beyond subsistence needs (Alpers, 1992). That means researchers cannot assume the amount of ivory exported from Africa and Asia or the frequency of mentions of elephant hunting and elephant hunting technology in the ethnohistoric and ethnographic records reflects efficiency or productivity of capture (e.g., Agam & Barkai, 2018; Ben-Dor & Bakrai, 2020). Despite the unprecedented economic and political circumstances that strongly influenced elephant hunting during this period (from which most of the ethnohistoric derive), the records can still provide a useful framework for understanding some of the constraints associated with the use of traditional hunting technologies. Here we investigate prey handling costs (i.e., costs associated with pursuit, processing and capture after encounter) and risks of failure associated with traditional technologies used to hunt proboscideans. These data can provide a systematic method for evaluating the productivity of hunting proboscideans that extends beyond valuations based solely on the body-size and amounts of different tissues associated with the animal. When used within a comparative analysis, these data can provide invaluable insights on the general productivity of different prey and hunting methods.

Traditional Elephant Hunting Techniques

At historic contact a variety of hunting techniques were used to acquire elephants (Supplementary Appendix 1). Some of these technologies were of limited distribution because of specific requirements (i.e., the use of horses or camels and swords), while others, such as certain snares and traps, were not widely used. This summary focuses on the most widely used or mentioned technologies. Most can only be considered “traditional” in the sense they likely predate the widespread use of firearms. Almost all of the records pertaining to indigenous elephant acquisiti encompass metal (usually iron) hunting technology (e.g., spears, arrows, etc.) which was widespread throughout Africa before the eighteenth century. Turnbull (1983:20) and others suggested that metal technology was not required to kill elephants and that poisoned sharpened wooden spears were formerly used by some populations (also Harrison, 1905; also see Hubback, 1905:113–114).Footnote 3 However, it is possible that metal projectiles provided a more effective delivery system for poisons than wood or stone and metal knives were likely more efficient than stone tools for butchering carcasses and reduced processing time (e.g., Mathieu & Meyer, 1997).

Nontraditional firearms were first introduced as trade items by 1635 CE and became widespread shortly after 1850 CE (Spinage, 2012; von Oppen, 1993:166). Although greatly desired and highly valued because they were considered superior to traditional hunting technologies (Cuthbert, 1924), firearms never completely replaced the use of other hunting technologies, such as spears, which often served as a back-up technology. Some groups, such as forest foragers, continued to use traditional technologies and had little access to firearms prior to the twentieth century due to sociopolitical constraints.

Projectiles

Projectile technologies used at historic contact include poisoned and unpoisoned spears, harpoons, arrows and darts. The technology used to dispatch elephants, especially spears, was specialized equipment designed to penetrate the skin and muscles of the animal. In most but not all instances, the use of projectiles usually required more than one person; participants often numbered in the dozens, especially in open grassland or wooded savannas. According to Thompson (1827:364), large parties of spearmen attacked and, “In this manner they will sometimes carry on their attack for the whole day and before he falls he is often pierced by more than a thousand assagais (spears). Not infrequently he escapes them…”. In one of the few detailed accounts of a spear hunt, Livingstone (1858:620, but also see Du Chaillu, 1899:72) describes how 70 to 80 spearman each threw one or more spears into an elephant. Livingstone eventually shot the animal 12 times, but it nonetheless escaped alive and was never found, even after an extensive search covering a 12-mile radius spanning about two days.

Coordinated attacks by multiple spear or bowmen inflicted large numbers of wounds that resulted in death by exsanguination or septicemia. In some areas, thrusting and throwing spears were considered more effective than arrows at delivering a fatal impact with large prey (Chaboo et al., 2018). In the forested habitats of central and western Africa, hunts relied on one or more stealthy hunters using a handheld projectile and targeting a vulnerable body location such as the abdomen or anus. The wounds could prove fatal but required 1 to 2 days (or more) of tracking before the animal actually died (Turnbull, 1965). Projectiles used in concert with poisons were more efficacious than unpoisoned projectiles, but still often required many hours or days of tracking before the animal became weakened and/or died. The main advantage in the use of poison was that it weakened the animal and reduced long pursuits. Even so, the efficacy of poisons varied as a function of the toxic ingredients and concentration, animal-size, season, freshness, location of the wound, and amount injected by the weaponry. For animals as large as elephants, it could take 24 h to 3 days or more for the animal to die from the poison on the projectiles (Cuthbert, 1924:145; but see Neuwinger, 1996). Holman (1967:37) working among the Kamba and other groups in East Africa reported that a well-placed shot could cause an elephant to collapse within 200 m where it would suffer and “thrash” for hours. But among East African Okiek and Waata, who were famed for using some of the strongest African hunting poisons available (Acokanthera sp; Neuwinger, 1996), elephants still usually took 2–3 days to die after being hit, and it was uncommon for an animal to die within 24 h (Woodhouse, 1913:21). But even those populations who had access to the appropriate poison did not always use it on elephants. The East African Hadza, for example, hunt animals as large as giraffes with poisoned projectiles but do not hunt elephants (Woodburn, 1968). According to the Hadza, their poison is not strong enough to kill elephants, although they will consume the meat if the animal is dispatched by someone else (O’Connell et al., 1988).

Traps and Snares

Traps included subterranean pit traps (or deadfalls), spear/harpoon traps, and various trigger traps and snares. Importantly, many of these devices targeted an array of prey and were not necessarily designed specifically to kill proboscideans. For example, harpoon or spear drop traps were quite common throughout the continent, but were particularly efficacious on hippopotami (Hippopotamus amphibius), and less useful for elephants (Neumann, 1898:80). Subterranean deadfalls were placed in strategic locations near water sources, along trails, and within artificially constructed or natural corridors where they trapped a variety of animals (Hall, 1977). In the historic record, these traps were most often used to protect crops and food stores from marauding elephants (Putnam, 1948; Takeda, 1996; Torday & Joyce, 1910; Turnbull, 1965). When pit traps were used for hunting, animals were actively driven or baited towards an array of deadfall pits within a confined area (e.g., Baldwin, 1863:153; Hall, 1977; Jacolliot, 1888:157; Sanderson, 1907:75; Shaw & Van Armelo, 1972). While there are records of elephants being caught in pit traps, several sources note that proboscideans quickly became trap wary and were able to detect and avoid pits (Selous, 1899:15; Shaw & Van Armelo, 1972; Weeks, 1909:124). This is echoed by indigenous voices. Turnbull (1962: 238), for instance, describes how Mbuti hunters laughed at the pit traps and spear drop traps used by neighboring food producers to procure elephants:

They both laughed at the strange hunting methods of the villagers. The digging of a trap was a long and arduous task, as was the setting up of spear fall. The latter method was particularly difficult. A stout spear blade is hafted into a heavy tree trunk which then has to be hauled high above the ground… But animals soon come to recognize any kind of trap and simply avoid it. We passed several, and sure enough, we could see the tracks going right up to the edge, then carefully circling around in safety.

These observations match recent studies that show that proboscideans have an enlarged hippocampus which is associated with superlative long-term spatio-temporal memory skills (Hart et al., 2008; Polansky et al., 2015). Further, unless these drop traps were closely monitored or caused the immediate demise of the animal, elephants were known for their ability to dig their way out and escape (Jacolliot, 1888:157). Stow (1905:89–90) reported that elephants were known for assisting companions out of pit traps, especially when the animal was young.

The spear drop trap consisted of a suspended weighted spear or harpoon set over an animal path or another strategic location that was triggered when a cord was tripped by the animal (Lagercrantz, 1934). To kill extremely large game and to deliver a fatal blow, the animal had to be precisely positioned under a heavy or weighted blade. According to Bell (1923:2) some of these weighed up to 181 kg and had to be carried by several men. Many historical observers comment on the ineffectualness of this device for killing elephants because it had to hit the body very precisely otherwise the device simply wounded the animal (Bell, 1923:2; Lalouel, 1950:199; Neumann, 1898:80; Sikes, 1971:306; Hubback, 1905:113–114; Regnault, 1911; but see Barns, 1925:276). Lalouel (1950:199) commented that many different aspects had to come together to deliver a killing blow, and according to Hubback (1905:113–114) “…in actual practice it generally misses the vertebrae and gives a nasty wound on the back…if the wound is not serious enough it doesn’t cause death.” Variations of traps involved the use of manned platforms or perches where armed hunters waited until elephants were driven or baited towards them. This required hunters to wait hours (usually at night) until elephants passed beneath the trap. Takeda reports (1996:22) the manned platform technique was not used very often because, “The method requires long waits on the platform in the trees and is physically taxing….” Other kinds of traps, such as foot traps, were not widely used and did not immediately kill the animal. Similarly, snares were less commonly used because it was difficult to find material strong enough to hold or immobilize an elephant. Most snares were designed to entrap a specific body-part, such as the trunk or foot, and were largely used to deflect elephants from agricultural stores or crops. Sikes (1992:222), for instance, describes the use of several different snares, all of which wounded but did not kill the animal.

Collaborative Drives

Collaborative drives are defined here as hunts involving organized groups of people with the objective of killing more than one animal by driving, chasing or baiting the animals into a natural or artificial trap. Some involved the use of fire and were seasonal affairs that could only be conducted in the dry season (Spinage, 2012:276–292; Turnbull, 1972:25). Cooperative drives involving large numbers of people (> 100), including the use of fire, often required extensive planning as described by Turnbull (1972:24–25). Some involved the use of fairly elaborate constructed features such as rope and log enclosures (Kingsley, 2004) in concert with natural impediments such as thick vines or landscape features (Du Chaillu, 1861:114–115; Hall, 1977; Holub, 1881:241–242; Kingsley, 2004). For instance, Mary Kingsley (2004) describes a cooperative elephant hunt among the Fan where a rope and tree enclosure was constructed on a landscape and enhanced by the use of bait laced with sedatives (also see Bruel, 1910:118). The hunt spanned several days and nights during which the elephants were driven into the enclosure and people invested substantial effort monitoring the trapped herd. According to Kingsley (2004) “the crowd of men and women spend their days round the enclosure, ready to turn back any elephant who may attempt to break out…Their nights they spend in little bough shelters by the enclosure, watching more vigilantly than by day.” Some of these drives were highly organized, specialized events conducted within the contexts of institutionalized leadership or were exhibitions that were anticipated and planned for months (Du Chaillu, 1861; Holub, 1881:241–242). Where available, hunters riding horses and camels could very effectively coordinate hunts that dispatched large numbers of elephants and even entire herds (Baker, 1883). Horses and camels were particular effective because these animals could easily outrun elephants (Douglas-Hamilton, 1987). For hunter-gatherers, the lack of being able to assemble enough people to coordinate an attack was a constraint (see for example Lee, 1979:234), but Maguire (1928:136) and von Höhnel (1894:13) both describe Okiek collaborative efforts.

Reconstructing the Productivity of Ethnohistoric Proboscidean Hunting Techniques

To evaluate the productivity of different traditional elephant hunting techniques, we use the rationale from the Prey Choice Model, which is well described in the literature (Lupo, 2007; Morin et al., 2022; Smith, 1991; Stephens & Krebs, 1986). Data derived from ethnohistoric records are used to quantify the handling costs, success rates, and the post-encounter return rates for hunting elephants using different traditional technologies. Note that costs of searching for or finding prey are not considered here as these represent a separate cost. Post-encounter return rates are based on energetic returns per unit of handling time (kcal/hr) after the prey is encountered. These values are important because they are often used to rank available resources and predict dietary choice by foragers in the prey choice model (Lupo, 2007). In general, foragers are expected to always selectively pursue the highest-ranking prey whenever it is encountered, and lower ranked prey are incorporated into the diet as a function of the availability of higher ranked prey. One limitation for applying this model to the zooarchaeological record is that determining the handling costs, particularly with traditional technologies, remains a challenge. To accommodate this difficulty, zooarchaeologists routinely assume that body size equates to prey rank. Following this logic, proboscideans, the largest-sized terrestrial prey, are assumed to be the highest ranked prey resource and would be preferentially pursued whenever available to hunters. However, Broughton (1994: Fig. 1C) long ago hypothesized declining post-encounter return rates with some of the largest-sized prey. Furthermore, recent and subsequent empirical research clearly demonstrate that prey body-size does not always correlate to prey rank in a systematic fashion (see Bird et al., 2009, 2012; Grayson et al., 2021; Hill et al., 1987; Lupo, 2007; Lupo & Schmitt, 2016; Morin et al., 2022; Schmitt et al., 2018; Smith, 1991; Winterhalder, 1981).

Importantly, search costs, or the costs associated with looking for animals, are normally considered as a separate cost spread across all targeted resources within the diet. Even so, empirical evidence suggests the opportunity costs associated with the tradeoffs between search and handling are not always mutually exclusive, especially when hunters pursue and track prey over great distances (Hill et al., 1987). While search costs are not included in the pursuit costs discussed here, it is clear from descriptions that elephant hunting required focused attention on a single prey even before the animal was encountered. After fresh evidence of an elephant(s) was discovered, hunters immediately followed-up even when an animal had not been seen or encountered. For example, Bahuchet (1985:241–245)Footnote 4 comments on the necessary dedication of hunters after evidence of elephants were discovered, noting the hunters often ate and slept very little during long pursuits.

To derive quantitative measures from the available records, only sources based on first-hand observations or where direct interactions with indigenous peoples are indicated were used. Whenever possible, quantitative data was confirmed by at least two independent sources. Observations and data pertaining to indigenous African hunter-gatherers such as the Okiek (formerly Ndorobo), !Kung (several sub-groups) and forest foragers (Mbuti, Aka, Baka) were prioritized, but these sparse sources were supplemented by cases derived from African food producers, many of whom also hunted elephants.

Most ethnohistoric records are incomplete and do not list information about elephant encounter rates, pursuit costs or success rates (as measured by the number of animals hit with a weapon against the number of acquired carcasses). In most instances, the missing parameters are estimated based on assumptions about walking speed, length of hunting bout, and number of hunters using existing records. To evaluate the efficiency of hunting African elephants with different technologies, we adopted methods recently outlined by Morin et al. (2022) for creating a systematic method for calculating post-encounter return rates (Rprey).

Elephant Hunting Success Rates

Unfortunately, there are very few data on the failure or success rates of different traditional hunting techniques. While it is clear from the records that hunters were highly skilled and motivated, it is also clear that proboscideans were formidable prey and possessed physical and cognitive attributes that made them difficult and dangerous to pursue. Further, although there are many different ways that hunts and subsequent pursuits can be unsuccessful (see Lupo & Schmitt, 2016), the sparsity of data precludes quantification of all of these scenarios. The ability to find or encounter an animal is linked to prey density and distribution, but given their large-size and the markers they leave, the inability to find an elephant might have been a less common form of failure after imminent signs were encountered. For example, several sources document hunters easily following trails or paths habitually used by elephants, monitoring the vegetation for signs of saliva, examining the ground for fresh dung or tracks and listening for the sounds of elephants eating (e.g., Foa, 1899; Haynes, 2006, 2012). Given existing observations, it is likely most failures were not a result of failure to find prey, but rather due to the inability to inflict a lethal wound or make subsequent attacks, and/or loss of animal after it has been hit and failure to follow and find the carcass.

A general statement about African big-game hunting by a nineteenth century European sport professional hunter using firearms concerning success rates is telling. Foa (1899:106) stated that a good hunter can only count on getting one out of every two animals he pursued, and at a minimum he was likely to wound and/or miss three or four animals for every one that is killed. Livingstone (1858) also made a revealing comparison between the kill rates of indigenous spear-hunters and nonindigenous hunters. According to him (1858:84), “bushmen” hunting parties with spears chased or harassed the animal(s) until it became tired and then dispatched it, but only obtained < 1 elephant per hunter. In comparison, European hunting parties, often mounted on horses and armed with firearms and trained dogs, usually dispatched 20 elephants per man.

Of the handful of existing observations of indigenous elephant hunting efforts, inability to make a lethal kill is the dominant form of failure. This is exemplified by Harako (1981) who reported one observation of Mbuti spear hunters targeting elephants. The hunting party made two attempts and both ended in failure because the weaponry missed and did not penetrate the skin (also see Lupo & Schmitt, 2016). Working among the Valley Bisa, Marks (1976:Table 9, 211–212) reported data on failure rates of elephants that were shot or wounded by Bisa hunters using muzzle-loaders. The success rate for muzzle-loaders as measured by the number of animals killed divided by the number targeted (wounded) is between 0.09-0.25. By way of comparison, much higher success rates for killing elephants were reported by rangers using high-powered (0.47-0.58 caliber) rifles and shotguns (Marks, 1976). More recently, Ichikawa (2021) reports the success rate of Mbuti elephant hunters between 1974 and 1975. According to his 15 months of observations, one of six attempted hunts succeeded in dispatching an elephant (0.16 success rate). He estimates that most bands only got one elephant per six-month interval because they are not hunted very often and, although large quantities of meat result from these activities, “it is rather a rare occasion, and not reliable for daily subsistence” (Ichikawa, 2021:461).

Anecdotal accounts report some information on the efficacy of other traditional technologies such as pit and harpoon or spear drop traps. Weeks (1909:124) suggested that weighted spears from drop traps penetrated the animal’s body four times out of six, but often resulted in nonfatal injuries because the spears often failed to penetrate vital organs (see also Johnston, 1902:609; Junker, 1890; Hubback, 1905:113–114). Junker (1890) found that two of the three elephants he shot had been previously wounded by spear drop traps.Footnote 5 Weeks (1909:124) reported that, “It is seldom anything is caught in these pit-traps”, an opinion echoed by Selous (1899:15) who stated that, “I have never known of any but young animals secured in this manner’ (but see Baldwin, 1863:271–272; Barrow, 1801:209; Sikes, 1971).

Other qualitative sources are sparse but revealing. Hunting failure is also indicated by the incidence of observed predation wounds on animals (cf. Strauss & Packer, 2013). Here the data encompass accounts made by hunters who found pre-existing, but nonlethal, wounds on their kills or observed animals with trauma. Predation wounds include healed or festering wounds from weaponry, and actual ammunition and weapons, such as spears, embedded in elephants from previous near-death encounters. Predation wounds from weaponry were not an uncommon occurrence and are frequently mentioned by historic hunters (see especially Cuthbert, 1924:83, 145; Foa, 1899; Letcher, 1911; Livingstone, 1858; Neumann, 1898:111; Marks, 1976:172; Thompson, 1827; Sikes, 1992:222). Marks (1976:172), for example, cites a control officer who reported in 1953 that every elephant encountered in the Luangwa Valley (Zambia) had wounds from previous weaponry attacks. Probably the most precise data were reported by Foa (1899) who logged and detailed 21 elephant kills between 1894 and 1897. Of those kills, 86% of the carcasses yielded “native” bullets from previous nonlethal attempts/assaults, including one animal with an astounding 30 bullets from a previous hunting attempt(s). Assuming each attack consisted of three bullets, the average number of shots it took Fao to kill an elephant, it is likely that some of these animals withstood several separate attacks by several different shooters.Footnote 6

Yet another qualitative source of evidence is reports by ivory workers of healed injuries from spears and ammunition embedded in tusks (Combe, 1801; Kunz, 1916:222–223; Miles, 2003). It is impossible to estimate the frequency of these kinds of wounds among the general animal population, but Combe (1801) stated that the discovery of such damage in ivory was common. In a review of historical nonlethal tusk injuries, Miles (2003) notes that tusks were not only damaged during unsuccessful attacks with weaponry but were sometimes broken off when the animals escaped from pit traps.

Obviously, the frequency of these qualitative phenomena cannot be converted into useful population level estimates of failure for each type of hunting technology. However, what these data demonstrate is that evidence of failure, even with firearms, was not an uncommon occurrence. Ideally, to account for failed attempts using all the traditional techniques used to target elephants would require detailed information which is not available. In this analysis, we assume the success rate for all traditional technologies was at least as good as that involving early firearms—around 20%. Given that firearm use demonstrably increases the success rate of hunting different kinds of prey (see Morin et al., 2022), it likely over-estimates the success rate of some traditional technologies. Cooperative drives may arguably yield higher success rates. However, from the sparse data that are available, it is clear these hunts were not always successful. One of the few first-hand participant observations is described by Usher-Wilson (1947) who witnessed a collaborative two-day fire drive involving some 2,000 participants enclosing an eight-mile area. This effort captured many different animals, but the small herd of elephants encountered during the hunt broke through the fire line and escaped.

Elephant Pursuit Costs Using Different Hunting Technologies

Pursuit times are inclusive of the time it takes to pursue the animal after it has been hit. Where applicable, it also includes the time expended constructing features specific to kinds of hunting technology such as platforms, pit traps, or enclosures. Pursuit costs per animal are calculated by multiplying the number of participants by the time spent pursuing the animal. Most reported elephant hunts involved more than one person, except for hunts executed by forest foragers (e.g., Mbuti, Aka).Footnote 7 Unless explicitly stated by the source, it is assumed here that at least two people are involved in each pursuit. Cooperative drives are defined as hunting episodes that involved the coordinated activity of 10 or more participants. Some of these events had hundreds of participants and, while it is not clear if everyone was equally invested in the hunt, sources agree that elephant hunts often required large numbers of assaults, assailants and weaponry. Morin et al. (2022) argued that with large cooperative groups involving men, women and children, many of the participants are not directly involved in the acquisition phase of the hunt. Because large numbers of participants inflate pursuit costs, they advocate for discounting the number of participants to accommodate those that are not directly involved in the acquisition of the prey. However, because elephants often require many assaults and most of the observations reported here describe the activities of spearman, group size is not discounted.

Table 1 shows the calculated pursuit times in minutes for different common traditional technologies derived from the ethnohistoric records. Here, pursuit time reflects the amount of time spent pursuing the animal after it has been encountered multiplied by the number of participants—note these values are not corrected for hunting success. These values are illustrated in Fig. 1 and show that firearms have a much lower mean pursuit time in comparison to all other traditional techniques. The difference in mean pursuit times for firearms compared to spears and poisoned projectiles, the two Indigenous techniques with the largest samples, is statistically significant (firearms-spears, t = 2.786, df = 26, p = 0.011; firearms -poisoned projectiles, t = 2.831, df = 17, p = 0.012). This corresponds to the general expectation that early firearms increased the efficiency of hunting elephants and further demonstrates the necessity of considering technological differences when evaluating prey productivity. Interestingly, there is no statistically significant difference in mean pursuit times between poisoned projectiles and spears for elephants (t = 1.054, df = 25, p = 0.302). This is surprising given that the expectation would be that poisoned projectiles would shorten pursuit times. One possible explanation is that while poisons do not necessarily shorten pursuit times, they do reduce the distance over which a dying animal traveled (i.e., 15 km versus 50 km search radius). If this is so, then pursuits are still long but the area over which hunters might have to search for the carcass or wounded animal is smaller than if poisons were not used. Alternatively, the small number of observations may preclude detecting statistical differences between these two technologies. Unfortunately, there are too few data points for traps and drives to make statistically meaningful comparisons.

Table 1 Pursuit time (in minutes) of African elephant acquisition constructed from ethnohistoric records
Fig. 1
figure 1

Comparison of elephant raw pursuit times (uncorrected for success) across different traditional technologies and firearms

For additional context on the cost of pursuing proboscideans with traditional hunting technologies, Fig. 2 compares the mean pursuit times (uncorrected for the risk of failure) of elephants to an array of different large game African species (> 100 kg in live weight) hunted with poisoned projectiles (Lupo & Schmitt, 2016). Poisoned projectiles are considered here because data for alternative technologies such as traps etc., are not available or do not apply to some species. These data demonstrate the well-known phenomenon that with poisoned projectiles prey body-weight influences pursuit times; larger-bodied animals require more poison and more time before they die. In comparison to other big-game, giraffes and elephants have long pursuit times with poisoned projectiles. Although giraffes have long pursuit times, these results might be driven by the giraffe sample which is derived from the !Kung, among whom they were rarely hunted, and includes one particularly long hunt spanning eight days (Marshall (1976:14). Furthermore, in comparison to elephants, giraffe are fleet sprinters but they cannot sustain speed over long distances (Dagg, 2014). To escape predation, they run short distances and hide in thick vegetation and finding them after they are wounded may not be as difficult in comparison to elephants.

Fig. 2
figure 2

Comparison of raw pursuit times (uncorrected for hunting success) for poisoned projectiles across different taxa. Most of the data were derived from Lupo and Schmitt (2016: Appendix ACite ESM.) and augmented as listed: Zebra (Equus spp.), Tomita (1966:162), Kohl-Larsen (1958), J. F. O’Connell (unpublished field notes); Alcelphines (wildebeest [Connochaetes taurinus] and hartebeest [Alcelaphus buselaphus]), J. F. O’Connell (unpublished field notes), Marshall (1976:137), Neuwinger (1996); gemsbok (Oryx gazella), Neuwinger (1996); kudu (Tragelaphus strepsiceros), Neuwinger (1996), Lebzelter and Neuse (1934:60); eland (Taurotragus oryx), Silberbauer (1965:57), Kohl-Larsen (1958); giraffe (Giraffa spp.) Blurton Jones and Konner (1976:341)

Figure 3 compares kg/min per estimated pursuit times corrected for hunting success of different hunting elephant technologies (spears, firearms, traps, poison projectiles and cooperative drives) and the same.array of African prey captured with spears and poisoned projectiles. The values shown here reflect the average kg/min acquired during a single hunting bout. The data show the increased efficiency in elephanthunting associated with firearms which have greatly elevated kg per min of pursuit time. Of the traditional proboscidean hunting techniques investigated here, mass drives yield the lowest kg/min. Although successful cooperative drives can yield large number of carcasses, the number of participants, elevates pursuit costs and yields some of the lowest kg/per min of pursuit time. Traps yield the highest kg/min followed by spears and poisoned projectiles. Even so, other prey such as kudu, buffalo, zebra, bush duiker, steenbok yield higher kg/min of pursuit time.

Fig. 3
figure 3

Comparison of kg/per minute (min) of pursuit (corrected for hunting success) among an array of African animals including bush duiker (Sylvicapra grimmia), African buffalo (Syncerus caffer), impala (Aepyceros melampus), steenbuck (Raphicerus campestris), hartebeest, wildebeest, gemsbok, eland, kudu and giraffe

Figure 3 compares kg/min per estimated pursuit times corrected for hunting success of different hunting elephant technologies (spears, firearms, traps, poison projectiles and cooperative drives) and the same.array of African prey captured with spears and poisoned projectiles. The values shown here reflect the average kg/min acquired during a single hunting bout. The data show the increased efficiency in elephant hunting associated with firearms which have greatly elevated kg per min of pursuit time. Of the traditional proboscidean hunting techniques investigated here, mass drives yield the lowest kg/min. Although successful cooperative drives can yield large number of carcasses, the number of participants, elevates pursuit costs and yields some of the lowest kg/per min of pursuit time. Traps yield the highest kg/min followed by spears and poisoned projectiles. Even so, other prey such as kudu, buffalo, zebra, bush duiker, steenbok yield higher kg/min of pursuit time.

As also shown in Fig. 3, efficiency as measured by kg per unit of pursuit time with the introduction of firearms is increased in comparison to all traditional technologies (as reported by Morin et al., 2022). Even so, muzzle-loaders, the earliest widely used firearm, were notoriously inefficient at killing large game because they propelled low-velocity projectiles that were wildly inaccurate, and assaults on elephants required multiple attacks because the projectiles lacked penetration power (Stigand & Lyell, 1906). Larger caliber breech-loading rifles were not widely used until the mid-nineteenth century, but these weapons still required multiple repeated assaults to kill elephants. It was not until after 1895 with the introduction of smokeless powder and large caliber, high velocity rifles (so-called elephant guns) that firearms became highly efficient weapons at killing elephants with fewer shots (Fadala, 2006). The data used in this analysis does not include any pursuits with elephant guns, but by most accounts these weapons were pivotal in the reduction of elephant populations (Fadala, 2006).

The data presented here show that regardless of the traditional hunting method used, the acquisition of elephants was costly and usually associated with long pursuits. Although poisoned projectiles might have facilitated dispatching these large animals, comparisons between this technology and unpoisoned spears do not show a decline in mean pursuit effort. The influence of long pursuits is especially notable in comparisons between other large-sized fauna and elephants where smaller-bodied but more easily dispatched animals yield higher kg/per unit of pursuit time.

Elephant Post-encounter Return Rates

Pursuit costs (multiplied by success rates) are but one component of prey handling costs. Handling costs also include processing costs. As shown in several previous analyses, processing costs are predicted by prey body-mass (Lupo & Schmitt, 2016; Morin et al., 2022). Although pursuit costs comprise the largest proportion of the handling costs for elephants, processing costs are important. Of all the prey considered here, elephants have the highest processing costs (see Lupo & Schmitt, 2016). Even with metal tools, the carcasses can take a long time to break down and always involved large numbers of people. Processing costs and edible carcass fractions were derived from Morin et al.’s (2022:32) regression formulae. To find Rprey (post-encounter return rates) for different traditional technologies, values were input into the automated excel calculator provided by Morin et al., (2022:SOM3). The Rprey values reported here differ from those previously reported by Lupo and Schmitt (2016) because here the values reflect different hunting technologies rather than a composite of all technologies combined. The values shown in Fig. 4 also reflect the use of a new methodology as outlined by Morin et al. (2022).

Fig. 4
figure 4

Comparison of Rprey for elephants taken with different technologies to other African species. Note that here pursuit costs are corrected for hunting success

While elephants can yield very large quantities of meat and fat and have a high edible fraction in comparison to most of the animals considered here, including giraffes (Fig. 4), they are not the highest ranked prey in the array examined here irrespective of the traditional technology used in acquisition. Again, firearms show greatly elevated efficiency in comparison to traditional technologies. But as shown in Fig. 4, with traditional projectile technologies Rprey is higher for animals that are smaller in body-size but that have shorter pursuit times and greater chances for success. These results suggest some smaller prey would be more efficient choices than proboscideans if energetic efficiency or return rate is the over-arching goal. Further, when one considers the considerable labor force and specialized tools required to pursue proboscideans, they were probably rarely pursued on encounter. Descriptions suggest that hunters needed to prepare for these kinds of hunts and in most instances were armed with specialized equipment (also see Supplementary Appendix 1). In comparison, other kinds of prey such as bushbuck, steenbok, zebra and wildebeest are animals that do not require large labor forces or specialized equipment.

Hunting Risky and Dangerous Prey for Prestige

According to ethnohistoric records, the procurement of elephants was and continues to be uniformly viewed as a dangerous activity associated with a high potential for physical harm (Bruel, 1910; Bryden, 1903; Graça, 1890:427; Lewis, 2021; Thompson, 1827:364, Fig. 4). All the sources cited here agree that elephant hunting is/was an inherently dangerous activity. Most of the danger involved the hunter being killed or injured by the elephant. But because proboscideans often required multiple assaults, participants were often accidentally injured or killed by other hunters. Close proximity hunting with projectiles, especially spears, not only exposed the hunter to dangerous encounters with an enraged animal, but large numbers of hunters throwing or shooting projectiles at the same target could be lethal for the participants. In the chaos caused by panic among elephants attempting to flee or protect themselves, many people were killed or wounded. This is well-illustrated by Bell’s (1923:166; also see Cooper, 1914:183) observation of a coordinated attack with 800 spear-men where, “In the consequent commotion casualties among the spear-men are frequent…” In a review of historical records from central Africa, von Oppen (1993:166) concluded that, with or without firearms, elephants were hard to kill with indigenous methods and the hunts involved high danger. He cites Joaquim Rodrigues Graça, a nineteenth century Portuguese explorer in west-central Africa, who reported that elephants existed there in large numbers (e.g., “like herds of cattle”) because there were few hunters willing to pursue them due to the high danger of this activity. Graça goes on to graphically describe how elephants killed hunters using firearms; his description is paralleled by several other observers of this time period and earlier (Graça, 1890:427; Burchell, 1822:301; Lichtenstein, 1812:226; Selous, 1907:355–356; Thompson, 1827:454–455). Historical European hunters and sportsmen commonly reported fatalities among fellow hunters, guides, trackers, horses and dogs resulting from elephant encounters. Foa (1899), for example, reported that over a seven-year period he personally knew four indigenous hunters killed by elephants (three on the spot and one from injuries sustained in a hunt) and an additional 3 men who were seriously wounded (see Lewis, 2021 for more recent accounts).

Coordinated mass drives could be especially lethal to the participants. According to Kingsley (2004), “One elephant hunt I chanced upon at the final stage had taken two months’ preparation…There were eight elephants killed that day, but three burst through everything, sending energetic spectators flying, and squashing two men and a baby …” Importantly, indigenous hunter-gatherers concur with these views and voiced the inherent danger of this activity in different conversations with ethnographers (Duffy, 1984; Bahuchet, 1993; Harako, 1981:544–545; Ichikawa, 2021; Lewis, 2021; Lupo, 2005). Duffy (1984:143) summarizes, “Even before elephant hunting was declared illegal in the Ituri, the actual killing of one of these beasts by a spear-wielding Mbuti was a relatively rare event. Not every band included a hunter willing or skillful enough to pit himself against the large animal in the forest, and most Mbuti knew of a friend or relative who had died horribly when attempting to do so” (see also Harako, 1981:545). An interview with one of the last living Aka elephant hunters conducted by one of us also supports this view (Lupo, 2005). The elderly Aka hunter who had killed some 55 elephants in his life-time using a rifle or spear described the hunts like “a war” and had known several hunters killed by elephants.

The high profits generated from hunting elephants, the inherent danger in the activity, and the skills required, meant that relatively few (often specialists) pursued this activity, but those who did benefited significantly (Table 2). Among food producers, highly skilled elephant hunters earned status through the wealth and the display of material items they acquired. In these groups, elephant hunters and other kinds of specialists formed guilds which often had stringent entry requirements, including a fee. Where these guilds existed among food producers, elephant hunters had the highest social status (see Marks, 1976). However, hunter-gatherers do not earn status through the acquisition or display of wealth, but through productive activities, such as hunting (e.g., Godoy et al., 2007; Gurven, 2004; Gurven & von Rueden, 2006; Hawkes et al., 2014; Smith et al., 2010). Among hunter-gatherers the high danger associated with hunting dangerous and high-risk prey such as elephants often conferred prestige on the individuals who pursued them (see Lupo & Schmitt, 2016). Prestige does not confer power or authority and is defined as status that is freely given by excelling in certain skills and knowledge, and that often yields benefits to individuals such as deference or influence (after Henrich & Gil-White, 2001; Smith et al., 2016). Among forest foragers, tûmas were widely recognized for their hunting prowess, knowledge of magic and charms, and were highly respected by neighboring food producers (Bahuchet, 1985:241–244; Bruel, 1910:117; Harako, 1981; Lewis, 2021; Regnault, 1911:287–288; Turnbull, 1965:205-b). Among forest foragers, tûmas often had apprentices who trained for years before they reached the level of specialist (see.

Table 2 Proportion (number) of elephant and/or high risk big-game hunters

Harako, 1981). Only a few individuals reached this level of specialization, but these individuals often had reputations that extended over many villages (e.g., Harako, 1981). Respect for these people among both communities was so high that tûma were often used as intermediaries for settling disputes between neighboring foragers and food producers. As described by Harako (1976:89) forager villages with tûmas in residence received extended social and economic benefits from neighboring food producers, including participating in elephant meat feasts that followed such hunts. Consequently, others often benefited by the presence of specialists in their midst.

Reframing Pleistocene Proboscidean Hunting

The existing ethnohistoric data are admittedly sparse but offer a glimpse of how productivity and risk differ among traditional technologies, and underscore how proboscidean characteristics influence the efficacy of different technologies. Importantly, these data show that irrespective of the traditional hunting technology used to acquire proboscideans, prior to the widespread use of firearms, these animals were considerably more expensive to procure than other available choices. Although the traditional techniques discussed here can dispatch elephants, historical sources report these hunting methods were less effective than firearms and did not significantly impact elephant populations (see Bryden, 1903). These data call into question the use of other proxies, such as the common mention of elephant hunting, amount of ivory exported and wide range techniques used in the ethnohistoric record that seemingly point to the ease of successful capture of proboscideans (e.g., Agam & Barkai, 2018). But the global ivory industry of the eighteenth and nineteenth centuries is central to understanding the scale of proboscidean procurement and frequency and range of techniques utilized. The motivation of this procurement was not driven by subsistence (Haynes & Klimowicz, 2015) and was clearly linked to the acquisition of ivory; while other elephant products (meat, fat, hair, skin) had secondary uses (Sikes, 1971:310). Indigenous elephant hunters were not driven by a desire for meat, but they were highly motivated by ivory profits that greatly exceeded any gains they might have made from meat acquisition (see Sundström, 1974:80, 88, 90, 91 for trade values). The unprecedented scale of harvesting, especially between 1840–1914, does not reflect the success rates of individual hunters, but the output of a commercial industry involving large numbers of people, many of whom were intensively committed to the procurement of ivory on a full-time basis – and had the weaponry suitable to the task (e.g., Milner-Gulland & Beddington, 1993; Parker, 1979; Spinage, 1973). Clearly, to fully understand the nature of the scale and frequency of historically documented hunting requires a deep dive into the complexities and contexts of elephant acquisition during that period.

However, ethnohistoric and ethnographic information was and remains an important source of data often used as frameworks by archaeologists to understand the distant past (e.g., Churchill, 1993; Haynes & Klimowicz, 2015; Holliday, 1998; Lupo & Schmitt, 2002; Villa & Lenoir, 2009). Yet contextual circumstances invite a reconsideration of how existing data might inform and be applied to the archaeological record of proboscidean procurement. Although it is impossible to “factor out” historical contexts in archaeological inquiry, it is possible to identify how known and recorded circumstances influence specific behaviors of interest and the inferences drawn from them. Here we have shown how ethnohistoric circumstances influenced the scale, frequency and mention of elephant hunting in existing records. We have also used those same records to identify characteristics associated with proboscidean acquisition that likely transcend time and space, and hypothesize how some of these might be manifested in the material record.

Prehistoric Proboscideans as High-Risk Prey?

The acquisition of modern elephants is constrained, in part, by predator avoidance strategies and physical characteristics which make them difficult to capture and some of these features were likely similarly expressed in Pleistocene proboscideans. In addition to their size, thick skin and overall strength, all extant proboscideans possess cognitive, auditory, acoustic and olfactory characteristics which assist them in avoiding predation and make them particularly difficult to capture with certain types of hunting technology. Their superlative sense of smell, in concert with exceptional long-term spatio-temporal memory skills, gives them a remarkable ability to detect and avoid traps. Exceptional hearing allows them to quickly learn to identify specific sounds such as the voices of different individuals and natural predators and adjust their defensive reactions accordingly (McComb et al., 2003, 2014; Thuppil & Coss, 2013). Their unique range of vocalization involve a wide range of sounds including infrasonic vibrations which allow them to communicate with each other over distances of several kilometers and through dense vegetation (Langbauer et al., 1991; Payne, 1998).

Despite these superlative predator defense mechanisms, elephants are vulnerable under some circumstances. Disruption in their social-organization through the loss of older animals (males or females) can have significant consequences for the survivorship of these herds (Breuer et al., 2016; Slotow & van Dyk, 2001). Elephants require large amounts of food and water and utilize large territories, but under environmental circumstances where resources are limited their vulnerability to predation increases. For example, exceptional dry seasons and drought can amplify predation rates when elephants become tethered to spatially restricted resources (Loveridge et al., 2006; Power & Shem Compion, 2009).

With the possible the exception of insular dwarf species (e.g., Mammuthus lamarmorai, Palaeoloxodon spp.), extinct Pleistocene proboscideans such as straight-tusked elephants (Paleoloxodon antiquus), Mammoths (Mammuthus primigenus, Mammuthus columbi) and Mastodons (Mammut americanum) shared many of the same physical features displayed by elephants (e.g., large-body size, large brain, expanded ears and nose), and likely had some of the same cognitive, auditory, olfactory and acoustical abilities. Direct evidence is sparse, but similarities between modern elephants and Pleistocene mammoths and straight-tusked elephants in long-term spatio-temporal abilities are reflected by the remnants of ancient trails habitually used by proboscideans (Haynes, 1993; Helm et al., 2022; Neto de Carvalho et al., 2021; Retallack et al., 2018).

Even so, estimating the costs and risks of hunting failure associated with extinct proboscideans is challenging and clearly rests on the range of hunting techniques and technologies available to hunters (Lupo & Schmitt, 2016). Some may argue that data reported here are not applicable because significant ecological differences existed between ethnohistoric contexts and those of the Late Pleistocene-Early Holocene. While there is an ecological mismatch between the data presented here and that of the late Pleistocene-Holocene, the factors that raise costs of acquisition and make proboscideans high risk prey transcend time and space. The fact that modern elephants are difficult to capture due to their strength, intellect and other characteristics is not linked to any particular ecological zone or time period. Further, the existing data spans the continent of Africa and ecological zones from tropical jungles to wooded savannas, yet it reveals a consistent pattern. The consistent theme of all of the records discussed here is the high costs associated with proboscidean capture under different ecological circumstances (i.e., closed vegetation versus open savanna). This suggests the difficulty of proboscidean capture is not necessarily linked to any particular ecological biome.

Probably the most common hunting technology used by Late Pleistocene hunters consisted of projectiles made from stone and later from a wider variety of materials (e.g., wood, ivory, bone, antler). In some locations poisons may have been used but direct evidence is quite limited (e.g., Langley et al., 2016). Evidence of cooperative pedestrian drives to acquire many large-sized animals, especially bovids, exists from the Middle Paleolithic onwards in Europe (e.g., Brugal et al., 1998; Gaudzinski & Roebroeks, 2000) and in North America mass drives of bison appear some 13,000 years ago (Speth, 2017). However, at present there is no evidence that fire was used on a landscape level to drive prey or manage ecological resources (Daniau et al., 2010). Furthermore, there is little evidence that cooperative drives were used to acquire multiple mammoths or other proboscideans. The often-cited mass mammoth capture site at La Cotte de St. Brelade (Jersey), has been recently re-evaluated and appears to reflect utilization and selective transport of individual animals acquired through encounter hunting or scavenging rather than cooperative drive (Scott et al., 2014; Smith, 2015). Similarly, while various kinds of traps might have been used to acquire fur bearing animals (e.g., Holliday, 1998), there is no evidence that deadfall or spear drop traps as described in the ethnohistoric record were employed. However, it is clear that cooperation among multiple hunters would have been required to dispatch proboscideans. As described here, with few exceptions the ethnographic record describes elephant hunting that involved more than one person. In the archaeological record, cooperative hunting might be inferred based on prey-size and aggression, but actual evidence demonstrating cooperative efforts among different hunters to dispatch proboscideans in the prehistoric record is challenging (e.g., Gaudzinski-Windheuser et al., 2023; then see Starkovich, 2023).

Assuming that lithic projectiles were the most commonly used technology to acquire large-sized animals in the Late Pleistocene, the ability to make lethal strikes and repeated attacks was likely a significant factor determining hunting success or failure. In the ethnohistoric record, metal-tipped projectiles (with or without poison) were capable of inflicting lethal damage especially if delivered through a large number of strikes. Experiments show that lithic projectiles and tools are capable of piercing the tough skin (e.g., Frison, 1989) and removing the edible tissues of modern elephants (Gingerich & Stanford, 2018; Starkovich et al., 2020). But as reviewed by Eren et al. (2021), the results of various penetration experiments do not necessarily reflect realistic hunting circumstances involving moving targets. Further, although woolly and Columbia mammoths had skin that overlapped in thickness with that of extent elephants (approx. 3–10 cm), woolly mammoths were covered with some of the thickest hair known among living arctic animals and this likely impeded penetration (see Boeskorov et al., 2016). In their experiments Eren et al. (2021) found that lithic projectile penetration ranged between 14.8- 22.9 cm but was not be deep enough to pierce the ribcage-protected thorax and vital organs. Further, they found a low incidence of impact scars indicative of penetration on Clovis points associated with mammoth bones from sites in North America (but see Kilby et al., 2022 for counterpoints). The association of barely damaged Clovis points and mammoth skeletal remains at some sites could reflect wounding evidence from one or more nonlethal attacks-similar to those associated with modern elephant populations (see Eren et al., 2021; Haynes, 1982). These results suggests that target penetration for some prehistoric proboscideans may have been far more difficult than experimental evidence implies, and the risk of failing to make a lethal hit was fairly high and possibly comparable to that experienced by modern hunters.

The significance of potential high hunting failure rates to full-time hunter-gatherers would have far-reaching consequences. In the ethnohistoric record, the risks of hunting failure were offset by the commercial gain associated with the ivory market. Given the value of ivory and industry surrounding its acquisition, one might expect a range of techniques to be used even if that technology only rarely succeeded. Further, the high costs of pursuit demonstrated in the historic record likely represent extraordinary efforts to procure a very high value target and are unrepresentative of populations solely driven by subsistence. In the historic record, hunters searched for carcasses for days to find high value ivory. When the full-time pursuit of these animals was driven by a profit motive, failed hunts were costly but only in monetary terms; the participants did not depend on the outcome for nutritional support. Clearly different circumstances apply to full-time subsistence hunters. The opportunity costs (costs associated with not pursuing other available resources) would have been excessive for subsistence hunters who could not afford to spend days searching for a carcass. But the characteristics that make elephants costly to hunt (physical, cognitive, olfactory) would still incur a high risk of hunting failure (i.e., failure to make an immediate kill). Under these circumstances, one might predict an even higher hunting failure rate than those reported here for subsistence hunters who could not afford to follow a wounded animal for days until it succumbed. For subsistence hunters a focus only on high-risk prey with a low success rate would be very costly and difficult to sustain for long periods of time. Rare windfalls do not support families with meat and fat on a regular basis (Hawkes, 1991). Even among contemporary hunter-gatherers, the full-time pursuit of high-risk prey targets, such as elephants, represented an unreliable source of nutrition for supporting subsistence efforts (Ichikawa, 2021:461).

Prestige Hunting in the Pleistocene

If hunting Pleistocene proboscideans carried a high risk of failure, then a re-examination of the circumstances under which these prey might be targeted is warranted. In the ethnohistoric record, ivory profits drove hunting frequency. But among historic and recent hunter-gatherers, hunting high-risk prey was an activity that conferred prestige to a handful of individuals that undertook the risk. While the introduction of firearms and a rising global demand for ivory increased the benefits associated with this activity, these circumstances did not create prestige-seeking behavior.

Prestige-seeking has deep roots in our evolutionary past and builds off dominance hierarchies that characterize primates (e.g., Plourde, 2008; O’Connell et al., 2002). Empirical evidence shows that prestige-seeking through hunting dangerous and/or large-sized, high-risk prey and other activities is a common means for garnering prestige among egalitarian hunter-gatherers (Table 3; e.g., Altman, 1987; Bliege Bird & Bird, 2008; Clastres, 1998:279; Coxe, 1804; Elliott, 1886:153; Hawkes & Bleige Bird, 2002; Hooper, 1853; Lantis, 1938; Sandell & Sandell, 1996; Stearman, 1989; Willoughby, 1907:78–79). Certain kinds of prestige-seeking activities require high skill levels that signal to others underlying characteristics, qualities or intentions (Hawkes & Bleige Bird, 2002; Bliege Bird & Smith, 2005; Quinn, 2019; Zahavi, 1977). Recent research shows that among hunter-gatherers, young men are guided by a prestige-bias when learning important subsistence skills, especially hunting (see Chudek et al., 2012; Hewlett, 2016). For some, the prestige-bias in skill acquisition is a marker of modern human cognition and a pathway in social transmission that underwrote flexible and rapid dissemination of information with high fidelity (Garfield et al., 2016) and the adoption of new technologies and skills required to expand into novel areas in the Late Pleistocene (Amick, 2017; Gaffney et al., 2021).

Table 3 Cross-cultural examples of prestige hunting

Importantly, the manifestation of prestige-seeking varies and is malleable depending on ecological and especially sociopolitical circumstances, opportunities and constraints (see Buela, 2020; Quinn, 2019). Among so-called nonegalitarian hunter-gatherers exemplified in portions of Upper Paleolithic Europe (45,000–11,000 CE), the prehistoric and ethnographic record of the Pacific Northwest coast (Smith & Codding, 2021)Footnote 8 and other locations (see Kelly, 2007), prestige-seeking is embedded within an organization based on dominance and competition associated with institutional inequalities, including hereditary hunting specialists (e.g., Davies, 2020; Kelly, 2007; Lantis, 1938; Moreau, 2020; Prentiss et al., 2012; Smith & Codding, 2021).

In these societies, prestige is often associated with power and may be manifested through ritualized material displays, competition, intergroup conflict and differences in social status manifested by differences in wealth and property ownership. Nonegalitarian hunter-gatherers tend to be associated with specific ecological conditions such as defendable clumped resource patches (Smith & Codding, 2021), but additional factors and processes are implicated (see Moreau, 2020).

In contrast, egalitarian hunter-gatherers have few recognized social positions and prestige does not confer power, dominance or support wealth and material inequalities. Empirical evidence presented here and elsewhere shows that prestige hunting is practiced by only a handful of individuals who possess the skills, sacred knowledge or magic to dispatch the animal, and have access to a supportive network of apprentices or specialized hunting technology. These positions are not hereditary, but family members may also act as assistants. Certain ecological circumstances may be associated with prestige-hunting among egalitarian hunter-gatherers, such as heterogenous environments where high-value prey are rare and high risk favors the amplification of prestige-seeking or risky strategies among hunter-gatherers (Codding et al., 2011).

Prestige Hunting in the Zooarchaeological Record

Although prestige-seeking activities are widely documented among hunter-gatherers, evidence of these activities in the archeological record is challenging to identify, and the most direct sources of information—zooarchaeological assemblages—are often overlooked (e.g., Bird & O’Connell, 2006; Codding & Jones, 2007; Quinn, 2019). Part of the problem stems from the deeply ingrained assumption that large game are always the highest valued resource based solely on body-size and the amount of edible product associated with the carcass. But clearly this is not always the case (Bird et al., 2009; Grayson et al., 2021; Hill et al., 1987; Lupo & Schmitt, 2016; Madsen & Schmitt, 1998; Morin et al., 2022; Schmitt et al., 2018; Smith, 1991; Winterhalder, 1981). Under this assumption, researchers expect that large-bodied game will be targeted and low abundances of these purportedly high ranked prey are usually attributed to over-hunting or climate change. To complicate matters, costly prey associated with prestige-hunting are often also (but not always) large-sized animals (Lupo & Schmitt, 2016; Lupo et al., 2020; Morin et al., 2020).

Nevertheless, the zooarchaeological record offers an important and compelling source of evidence for prestige hunting. Recast within the contexts of prestige-hunting, the low zooarchaeological abundances of certain high-risk prey reflects low predation rates rather than scarcity or decreased encounter rates due to over-hunting or climate change. Importantly, the manifestation of prestige hunting will differ according to hunter-gatherer sociopolitical organization (e.g., nonegalitarian versus egalitarian) and these differences will be measurable in faunal spatial distributions, abundances and part representation. Among nonegalitarian hunter-gatherers, prestige-hunting involving dangerous and high-risk prey may be manifested through feasting, conspicuous, visual ritualized displays of trophies, including the remains of prestige animals (Table 4). Because hunting specialists tend to be hereditary positions, one might expect consistent but low levels of predation involving target prey over time. In these contexts, different material markers of nonegalitarian sociopolitical organization will also be apparent (Moreau, 2020). The high status and display value associated with prestige prey would encourage the collection and scavenging of remains from animals that had died naturally to augment trophies from hunted animals, which may be the case of Late Pleistocene proboscideans in some Pavlovian sites (Germonpré et al., 2020; Perri et al., 2015; but see, Demay et al., 2019; Haynes & Klimowicz, 2015; Svoboda et al., 2011; Wilczyński et al., 2015). In contrast, prestige hunting among egalitarian hunter-gatherers will involve generous meat sharing but not conspicuous trophy displays. While trophy-taking may occur, conspicuous ritualized displays are not expected because leveling-mechanisms mitigate against excessive aggrandizement.Footnote 9 Because these positions are skill-based and not hereditary, low-levels of discontinuous predation are possible when specialists die or leave.

Table 4 Differences in prestige hunting among hunter-gatherers with different sociopolitical organization

Conclusions

The ethnohistoric record can provide some valuable insights into the productivity and risks associated with traditional hunting technologies. But the usefulness of these records is tied to the abilities of the researcher to recognize how historical contexts and processes influenced relevant records and questions of interest, especially when novel technologies, economic opportunities and sociopolitical processes are at play. As discussed here, the ivory industry of the eighteenth and nineteenth centuries greatly increased the scale and frequency of hunting, numbers of people involved in the activity, and the valuation of elephant products. While sparse, most of the existing data on traditional elephant hunting techniques suggest that prior to the introduction of firearms elephant hunting was relatively inefficient relative to other available resources. Only a few individuals possessed the skills required to dispatch proboscideans and this activity was often associated with prestige-seeking. While the scale and frequency of prehistoric proboscidean hunting remains unclear, existing evidence suggests these dangerous, large-sized animals were infrequently targeted by some populations in some places at certain times. Archeologists will undoubtedly continue to discover the remains of prehistoric proboscideans and some of these will likely be the remnants of hunting events. But given the current state of knowledge, the question before researchers now is not whether or not prehistoric hunters could exploit high value and high-risk resources, but rather when and under what circumstances can we expect hunters to over-exploit that resource to the point of extinction? A second larger question to ask is, what are the circumstances under which the acquisition of high-risk prey might become valued to the point where they are part of and reflect underlying sociopolitical complexity?