Evolutionary Ecology 17: 293–312, 2003.
2003 Kluwer Academic Publishers. Printed in the Netherlands. Evolutionary perspective The fast life of a dwarfed giant PASQUALE RAIA1,*, CARMELA BARBERA2 and MAURIZIO CONTE2 1 Università degli Studi del Molise, Dip. STAT, Via Mazzini 8, 86170 Isernia; 2Università Federico II di Napoli, Dip. Scienza della Terra, L.go San Marcellino 10, 80138 Napoli, Italy (*author for correspondence, e-mail: pasquale.raia@libero.it) Received 26 July 2002; accepted 10 June 2003 Co-ordinating editor: N.Chr. Stenseth Abstract. In the first half of the 1960s, a rich paleontological site was discovered at Spinagallo caves (Eastern Sicily, Southern Italy). A very abundant fossil population (at least 104 specimens) of the dwarf elephant Elephas falconeri, the smallest elephant that ever lived, was recovered. We computed the survivorship curve for this fossil population in order to investigate both the great juvenile abundance and high calf mortality which it shows. Through the analysis of E. falconeri survivorship, of some reconstructed life-history traits, and of its ecology, and taking into account the Island rule (Foster, 1964), we concluded that E. falconeri moved somewhat toward the ‘fast’ extreme of the slow-fast continuum in life-history traits in regards to its mainland ancestor E. antiquus, that is, it was somehow r-selected. In keeping with our findings, we propose a new explanation for the common occurrence of dwarfism in large mammals living on islands. We suggest the interplay of competition, resource allocation shift and feeding niche width could successfully explain this pattern. Key words: dwarfism, elephants, island rule, life history, resource allocation Introduction During the Middle to Upper Pleistocene, two distinct dwarf forms of paleoloxodontine elephants inhabited the island of Sicily (Southern Italy). The larger, Elephas mnaidriensis, is stratigraphically younger (Belluomini and Bada, 1985; Bada et al., 1991). The smaller, Elephas falconeri, was only 0.90 m tall at the withers, with this being the smallest elephant that ever lived (Roth, 1992; Lister, 1993). A rich fossil deposit of E. falconeri was found in the Spinagallo caves (Eastern Sicily) in the first half of the 1960s. A total of 2493 remains and numerous unidentifiable fragments have been recovered. The minimum number of individuals identifiable was 104, in keeping with the abundance of the assemblage-most-common bone, the left tibia (Ambrosetti, 1968). However, this estimate should be considered as a rather prudent approximation (Gilinsky and Bennington, 1994). 294 Spinagallo fossil elephants afford the very rare opportunity of studying a very large, well preserved fossil population of a dwarf elephant. This study aims to explain, through the analysis of the Spinagallo fossil population, the wide size reduction that occurred in the evolution of Elephas falconeri, in comparison with its large mainland ancestor, the straight-tusked elephant, E. antiquus. We focussed on the selective value of dwarfism, and we have shown that size reduction might arise as an adaptive response to insularism instead of being the mere outcome of calorie restriction, as is often stated in previous studies. We compared the Spinagallo population structure with some of the living African elephants, Loxodonta africana. The choice of the African elephant is justified because, even if it is not the closest knit of E. falconeri, it is probably ecologically similar to the dwarf elephant of Sicily (Caloi and Palombo, 1994), both living in open environments with a sparse tree cover. Moreover, data on African elephants are abundant and obtainable from longterm studies of stable populations (e.g. Moss, 2001). Indeed, living elephants, demographic profiles are heavily affected by poaching, meaning that poachers prefer larger, heavy-tusked males; therefore, we restrict our comparisons to a few poaching-free population studies. The Island rule Dwarfism of large mammals on islands is a rather debated issue. As a general rule, once segregated on an island, large mammals undergo a considerable size reduction, whereas small mammals greatly increase their own size. This phenomenon is known as the Island rule (Foster, 1964). As insular faunas are impoverished, if not deprived, of terrestrial predators (Sondaar, 1977), lack of predation pressure is recognized as the main explanation for the rule (Case, 1978; Lomolino, 1985). In contrast, Heaney (1978) recognized changes in body size to be mainly related to the host–island surface, whereas Case (1978) emphasized the role of behavioural features, such as territoriality. Later, different studies (Brown et al., 1993; Damuth, 1993; Kelt, 1997; Kelt and Brown, 1998) pointed out that at a certain small body mass (either 100 g or 1 kg according to different authors) mammals attain an optimal energetic tradeoff between resource provisioning and offspring production abilities. In keeping with this, they proposed that this body size limit is approached by species freed from ecological constraints such as predation or interspecific competition, circumstances that island colonizers normally face. Roy et al. (2000), however, found no support for this statement. Whatever be the explanation, apart from many remarkable examples of obedience to the rule (e.g. McFarlane et al., 1998), literature affords cases of small mammals becoming even smaller on islands, and of large mammals becoming even larger (i.e. the Kodjak bear). These 295 exceptions were thought to be dependent upon phylogeny (Case, 1978; Lomolino, 1985), meaning that closely related genera should exhibit similar body-size variation trends, whereas Lawlor (1982) proved the feeding specialization degree to justify such aberrant cases. According to Lawlor’s study, ampleness of feeding niche, together with the well known reduced island biodiversity (MacArthur and Wilson, 1967), allows generalists to cope better with the array of food items available on the colonized island, different in abundance and composition with regards to the mainland one. As argued by him, generalists should have a better survival chance than specialists on islands, a point later upheld by Ehrlich (1986). Therefore, generalists’ body size should be more amenable to record any change. An important merit of Lawlor’s paper is that it looked at species interaction, a quite overlooked level of investigation for the island rule, as later pointed out by Dayan and Simberloff (1998). Materials and methods Spinagallo cave (Eastern Sicily) elephants’ remains are now available in the University of Catania (Sicily). The bones of Elephant falconeri were laid in a single layer of red sand devoid of internal stratification and regularly overlaid by a younger stratum containing material referred to Elephas mnaidriensis and to the dwarf hippo Hippopotamus pentlandi (Ambrosetti, 1968). Ambrosetti specifically reported the assemblage as showing ‘no signs of disturbance’, exclusive of weathering due to water circulation within the embedding sands. The absence of taphonomic selection of any sort is further testified by the evenness of bone counting. There were, for instance, 244 carpal bones and 244 tarsal bones, 180 metatarsals and 190 metacarpals; dorsal vertebrae outnumber cervical vertebrae by a 2.66 ratio, the expected figure for the latter is between 2.71 and 3, according to a presumed number of dorsal vertebrae which is between 19 and 21 per elephant (Ambrosetti, 1968, p. 319). The only contrasting evidence is for total vertebral count of 720 specimens (with at least two hundreds of fragmentary vertebrae), but vertebral bodies ossify later at adulthood. Thus, the apparent lack of vertebrae is justified by a preservation bias. Among long bones, evenness in bone counting is evident for all but radii and fibulae, whose scarcity could be well justified by their gracile feature in Elephas falconeri (pers. obs.). The abundant remains, together with the absence of any pre-mortem fractures in the long bones that we have analysed, and the taphonomic evidences we have dealt with above lead us to exclude both accumulation by accidents (i.e. falling from a hypothetical crevice or fatal entering of the caves by inexperienced individuals) and carcasses gathering by some sort of transport. Elephants did not live in caves; they perhaps visited them looking for shelter or for 296 provisioning mineral salts of some sort, as living elephants still do. Therefore, cave visiting was neither accidental nor could any carnivore have accrued the bone material because carnivores are totally absent from the whole E. falconeri faunal unit (Bonfiglio et al., 2000). One can argue that cave accessibility could have biased the assemblage (i.e. smaller individuals might have been accessing the caves more easily than larger ones); however, it seems unlikely that the largest individuals could not enter the caves, as they amounted to at least 40 individuals. Furthermore, very young elephants are hardly over-represented because weaning forced them to stand close to adult females, whereas, all other things being equal, tiny bones have the least chance to be preserved because of their higher surface to volume ratio (Damuth, 1982). In fact, modern elephants live in family herds led by a matriarch. The group follows the matriarch’s decisions, and the youngest follow their mothers without straying away from them (McComb et al., 2001). The Spinagallo population exactly matches this structure, with adult females being accompanied by their young and only a few adult males. Additionally, the peculiar feature of a great juvenile abundance in Elephas falconeri population seems to be supported by E. falconeri findings other than that of Spinagallo elephants (Ambrosetti, 1968, p. 284). For the above reasons, we assume the fossil assemblage to represent a good approximation of the living elephant population (as already stated by Ambrosetti, 1968). Under this assumption, we computed a survivorship curve ranking the most abundant remains (the left tibia) by dimensional classes. Specimens were ascribed to each class according to maximum bone length (Kurten, 1954; Dodd and Stanton, 1990). Because 10 tibiae were not easily accessible to us as they were parts of skeletons mounted for exposition elsewhere, we refered to Ambrosetti’s published data (Ambrosetti, 1968) to make up for lack of specimens. In addition, for the sake of completeness, eight specimens were included in the analysis even if partially broken. This was made through comparison of their midshaft diameter width instead of the total bone length. We calculated a least squares regression of log transformed bone diameter vs. length (in fact, because all the eight specimens of concern are epyphises-less, we are actually referring to diaphysis length). Transversal diameter scale to the length according to the equation l ¼ 4.23 d1.082 and the slope is indistinguishable from 1 (95% confidence limits 0.918–1.247, R2 ¼ 0.923). We used this formula for adding to the analysis the eight specimens we referred to above, by using reconsitued bone length. Seven other left tibiae were too fragmentary to be ascribed to any dimensional class and were therefore excluded from the analysis. In sum, we included in our analysis 97 left tibiae. The underlying assumption is that tibia length is a good proxy for age. The choice of the tibia is a good coincidence, since in elephants tibia epiphyses fuse to diaphyses at the same time (Roth, 297 1984) thus rendering bone ranking easier. Since, for practical reasons, we did not take into account sexual dimorphism in juveniles, which is obviously less marked than for adults (Jarman, 1983), our computation makes young males older than they really were, thus underestimating the actual calf mortality. Since mammals are not ever-growing creatures, this procedure is applicable to juveniles and young adults only (i.e. it is impossible to assess adult age at death from bones beyond the moment of epiphyses fusion). This should be of no crucial importance because differences in survivorship curve among mammals are especially related to juvenile mortality (Pianka, 2000). At least one factor can negatively affect a so-computed survivorship curve: adult tibiae (those with fused epiphyses) are separated in to two normally distributed, non-overlapping groups (Fig. 1). The two adult groups (females and males) are statistically different (one tailed t-test, t ¼ )16.1; p < 0.001, see Table 1). We follow Ambrosetti (1968) in assuming that a great sexual dimorphism could explain this pattern (i.e. the two groups may probably represent similarly aged males and females). The evidence that the larger group is less numerous could indicate that many young males (assuming males to be larger than females) are absent from the fossil assemblage, possibly due to emigration, or to strongly male-biased juvenile mortality. Indeed, male mammals usually abandon their native place at sexual maturity to avoid inbreeding (Greenwood, 1980), and this is certainly true for elephants. In mammals, a slightly greater male mortality is quite common among juveniles (CluttonBrock et al., 1985; Waser, 1996; Berger and Gompper, 1999) and elephants in particular (Moss, 2001; Wittemeyer, 2001). Wittemeyer (2001) reported Samburu adult female elephants to be 2.3 times as abundant as males. adult's left tibiae dimensional classes 12 frequency 10 8 6 4 2 250-9 240-9 230-9 220-9 210-9 200-9 190-9 180-9 170-9 160-9 150-9 0 tibia length Figure 1. Distribution of left tibiae belonging to adult individuals. Specimens supposed to represent females, stippled bar. Specimens supposed to represent males, open bar. 298 Table 1. Statistics relating to left tibiae with fused epiphyses (adults, in mm) Sex n Mean Std. dev. Std. err. Female Male 27 13 174.00 230.38 7.27 14.60 1.40 4.05 Samburu and Buffalo Spring (Kenya) elephants are rarely killed by poachers (Wittemeyer, 2001). Furthermore, this figure is confirmed by the Amboseli elephants which are nearly unaffected by poaching (Moss, 2001) and is probably a quite generalized feature of elephant age profiles (e.g. Jachmann, 1980). Spinagallo adult males were half as numerous as females (13/27 see Table 1). Therefore, we did not make any correction for greater female abundance. In addition to survivorship analysis, we made extensive use of allometric formulae to predict many life-history variables of E. falconeri. The allometric approach is widely accepted by the majority of paleobiologists (e.g. Damuth and MacFadden, 1990) and will not be discussed further here. Results Despite a probable underestimation of juvenile abundance (see below), juvenile specimens (specimens with non-fused epiphyses) constitute 58.8% of the whole population (57 out of 97 specimens). From the data in Ambrosetti (1968), who used a slightly less restrictive criterion for the inclusion of specimens in the analysis, we calculated specimens with detached epiphyses to be 56.7% (59/ 104) of the total for the left tibia, 57.0% for the right tibia (53/93), 67.0% for the right femur (63/94), 62.9% for the left femur (44/70), 62.9% for the right humerus (44/70), 67.1% for the left humerus (47/70), 64.1% for the right radius (25/39), 51.9% for the left humerus (25/52), 61.1% for the right ulna (41/67) and 52.3% for the left ulna (34/65). We report these data to demonstrate that the high abundance of immature specimens is not an artifact of the choice of the left tibia. We used a v2 comparison for proportions (Zar, 1984) to test whether those proportions differed among each other, and we found that they did not, v20:05;9 ¼ 8:024, 0.75 > p > 0.50. Indeed our figure of 58.8% is one of the smallest. The most outstanding feature of E. falconeri survivorship curve is the very high calf mortality. Thirty out of 97 individuals (30.9%) belong to the smallest three classes; beyond the third size class, mortality takes place at a much slower pace. For instance only 8.2% (8/97) of individuals belong to the fourth, fifth or sixth size class. Whitehouse and Hall-Martin (2000) reported yearling’s mortality of 6.2% for Addo Park (South Africa) elephants and quoted some 10% as the mean value for African elephants. Moss (2001) calculated male mortality during the first ten years of life to be 25%, whereas that of females is set to 299 survivorship curve 100 90 % of survivors 80 70 60 50 40 30 20 10 adults 170-179 160-169 150-159 140-149 130-139 120-129 110-119 90-99 100-109 80-89 70-79 60-69 50-59 0 dimensional classes Figure 2. Survivorship curve computed for Spinagallo’s Elephas falconeri population. 16%. Even if neither of those studies can be directly compared to our study, the suggestion of an exceedingly high calf mortality in the Spinagallo population remains obvious. Our survivorship curve resembles a type II curve (see Fig. 2, cf. Pianka 2000). A type 1 curve is typical of long-living mammals (Pianka, 2000), and not surprisingly, it was obtained from a fossil assemblage of the woolly mammoth Mammuthus primigenius (Kurten, 1954; Guthrie, 2001). High calf mortality is a quite unexpected pattern for animals, such as elephants, recognized to be addicted K-strategists. Indeed, because of their size, adult elephants are nearly predation-free, whereas juveniles are not. Modern elephant calves could experience a high mortality risk due to large predators (Ruggiero, 1991). E. falconeri adult size was closer to that of very young modern elephants, yet Spinagallo elephants had no predators to fear, and no large predator existed in the whole mammalian fauna to which E. falconeri belonged (Bonfiglio, 2000). Discussion Extant elephants are long-living. The African elephant Loxodonta africana reaches sexual maturity only when it is 11–15 years old (Calef, 1988; Moss, 300 2001; Whitehouse, 2002), it has an extremely long pregnancy duration, gives birth to precocial offspring, and commonly has only one calf at a time. Roth (1992) estimated Elephas falconeri to weigh around 100 kgs, some 150 times less than a 15 ton Elephas antiquus. According to the former estimation and to the allometric relationship between physiological time and body mass which is t ¼ M0.25 (Charnov, 1991; Krokonis and Scheffer, 1992; Calder, 1996), we calculated that E. falconeri was sexually mature when 3–4 years old and that it had a pregnancy duration of some 189 days, life span of around 26 years and fasting endurance of some 71 days (allometric formulae taken from Calder, 1996, pregnancy period and fasting endurance taken from Roth, 1992). It is evident that Elephas falconeri lived somewhat ‘faster’ than extant elephants and, almost certainly, than its forebear as well. The questions we are dealing with are, ‘Why were the elephants of Spinagallo dwarfed?’ and ‘Why had they such high calf mortality?’ Roth (1990) furnished a possible explanation for the extreme dwarfism of E. falconeri. She proposed that the reduction in body size was a response to overcrowding and at least two consecutive events of overcrowding had to occur. In a subsequent paper, Roth (1992) provided a general view of dwarfism in fossil elephants. She attributed this phenomenon to the species’ population dynamic in ecologically restricted environments such as islands. However, as she noted on her own, it still remains hard to explain why Sicily housed the most reduced form of elephant being, by far, both the greatest and the nearest to the mainland of the Mediterranean islands. Sicily’s current surface area was established since the Middle Pleistocene (Bonfiglio et al., 1996), and E. falconeri remains are in fact present in the extreme East (Spinagallo) and the extreme West (Luparello, Palermo) boundaries of the island (Caloi et al., 1993). In fact, overcrowding appears more probable in smaller islands, where the population density should reach earlier any limiting value, and on islands farther from the mainland, since the arrival of the continental faunas, certainly easier in Sicily, always dismiss island endemic faunas (Palombo, 1996). Therefore, we think that dwarfism in E. falconeri deserves an alternative explanation. The most striking evolutionary change for mammals living on islands is body size change. Apart from phylogeny (Stearns, 1983), body size accounts for the best part of variance in lifehistory traits (Western, 1979; Stearns, 1983; Promislow and Harvey, 1990; Calder, 1996), maybe through physiological constraints (Calder, 1996; Geffen et al., 1996). Thus, body size has been recognized as a prime factor in life strategy regulations in mammals (Calder, 1996). Nevertheless, it is well known that many important variations in life-history traits do exist. These variations depend on species’ ecology, which affect life-history traits through several, strongly covarying factors, such as mortality schedule (Harvey and Zammuto, 1985; Pomislow and Harvey, 1990; Charnov, 1991, 2001; Purvis and Harvey, 1997; Stearns et al., 2000), environmental situation (Stearns and Koella, 1986; 301 Tuljiapurkar, 1990; Brommer, 2000), population density (Guo et al., 1991; Mueller et al., 1991; Moorcroft et al., 1996; Sinervo et al., 2000) and even social organization (Geffen et al., 1996; Keller and Genoud, 1997). One can conceivably think that body size is enough to arrange mammal lives along the socalled slow–fast continuum in their life-history traits, while species’ ecology could displace them from those values (i.e. physiological time, diet specialization, etc.) predicted on the sole basis of their body size. Maximization of individual fitness should be viewed as the prime cause of those variations. Interestingly, Promislow and Harvey (1990) demonstrated mortality schedule to strongly affect reproductive strategies. Under high mortality rates, females gain interest in giving birth to larger litters with lighter offsprings, whereas in more predictable environments (i.e. under lower mortality rate), heavier offsprings are favoured because of their greater intraspecific competition abilities. Many laboratory studies on fruitflies do confirm these predictions (e.g. Gasser et al., 2000). It is worth noting here that Promislow and Harvey distinguished between extrinsic and intrinsic mortality, that is, between mortality due to predation or environmental conditions (extrinsic) aside from mortality mainly related to reproductive efforts (intrinsic, sensu Promislow and Harvey, 1990). The Spinagallo elephants lived under conditions of very low extrinsic mortality. Their related fauna comprised neither competitors nor land predators (Ambrosetti, 1968; Alcover and McMinn, 1994; Bonfiglio et al., 2000); thus, extrinsic factors seem insufficient to account for both high calf mortality and overall juvenile abundance. We will return to this important point later on. Although many authors suggested that insular faunas live under resource deprivation (Lawlor, 1982; Lomolino, 1985; Lister, 1993; but see Case, 1978) and that resource deprivation could in turn justify dwarfism of large mammals through stunting, we feel this explanation to be unlikely. As Dayan and Simberloff pointed out (1998), there are many contradictory statements about this point: ‘…high population densities (for island dwelling mammals) have been envisioned as promoting decreased body size, but also as leading to a K-selection regime that generates increased body size…’ They later state in the same section that ‘Some researchers have argued that islands will have enhanced competition because of a decreased resource base, while others have argued that islands will have decreased competition because of the absence of some mainland competitors…’ Isotopic evidences (Koch et al., 1998; Zazzo et al., 2000) show how, despite nearly unchanged morphological features in their dentition, both extinct elephants and mammoths could switch from browsing to grazing or mixed 302 feeding as well, thus testifying that elephantids were capable of some food adaptability in the past as well as in the present. Because of their large body sizes and of their hind-gut-fermenting digestive strategy (Demment and Van Soest, 1985; Janis, 1989), elephants can thrive on even very low quality food and accept a wide variety of food items. Perhaps elephants are the most effective feeding generalists. As Lawlor (1982) demonstrated, dietary generalists fare better than dietary specialists as island colonizers, therefore, bearing in mind that dwarfism is a rapid process, we argue that elephants were unlikely to suffer from starvation once they colonized islands. Elephants probably colonized islands with only a few dispersers capable of metabolizing almost any edible item at their disposal. As their population grew, their body size shrank thus diminishing per-capita total energetic demand, thus preventing resources to be depleted. Their disproportionate diversity as Pleistocene fossil inhabitants of islands might indeed be justified by those facts. It is worth noting that the same species, E. antiquus, reached Sicily twice within 2–300 kya during Pleistocene, showing two very different degrees of dwarfism, the highest leading to E. falconeri, which had no competitors in the accompanying fauna. If resource deprivation is the cause of dwarfism, how is it possible that food shortage led to a greater dwarfism for E. falconeri than for E. mnaidriensis (the younger dwarf elephant of Sicily), as food shortage was even more severe for the latter, due to the presence of a diverse (Bonfiglio et al., 2000) competing herbivore fauna? Instead, we think the optimal body size hypothesis (Brown et al., 1993; Damuth, 1993; Kelt, 1997; Kelt and Brown, 1998) to be relevant here. As a lonesome colonizer, E. falconeri tried to approach the optimal size limit, while E. mnaidriensis paid its due to incumbent species (smaller herbivores). Incumbency of niche spaces plays a great role in evolution (Rosenweig and McCord, 1991) and probably explains the apparent failure of optimal body size hypothesis in a mollusc’s evolution (Roy et al., 2000). Additionally, calorie restriction, as well as low food quality, is expected to delay sexual maturity and to decrease reproductive efforts together with a shift toward a greater investment in somatic maintenance (Holliday, 1989; Shanley and Kirkwood, 2000; Rollo, 2002). Despite the high calf mortality, immature individuals make up the majority (58.8%) of the population of Spinagallo elephants; therefore, reproductive efforts were certainly high in E. falconeri. In fact, as well as resource deprivation delays maturity, abundant and/or high quality food supply causes faster growth and earlier reproduction, allegedly at reduced body size (Stearns and Koella, 1986; Brommer, 2000) – the resulting shift in growth curve trajectory is heritable (Roff, 2000). According to Seydack and Bigalke (1992) and Seydack et al. (2000), elephants living in open habitats feed on higher quality resources, which favours a quicker development and enhances reproductive rate. Elephas falconeri is 303 morphologically adapted to an open environment, similar to that of the extant Loxodonta africana (Caloi and Palombo, 1994; Palombo, 1996). Moreover, Addo National Park (South Africa) elephants referred to in Seydack et al.’s papers show a rather rare feature in adult females lacking tusks. In keeping with the rate of increase of Addo elephants, the highest ever recorded for elephants (Calef, 1988; Whitehouse and Hall-Martin, 2000), Seydack et al. (2000) correctly interpreted that this feature is due to augmented resource allocation to reproductive efforts. It is certainly remarkable that female Spinagallo elephants lack tusks as well (Ambrosetti, 1968). One could argue that females of Elephas maximus also lack tusks, yet we do not know any Asian elephant population whose females have tusks. Dealing with Addo park elephants, Seydack et al. (2000) reported on a ‘variant’ phenotype, and tusk absence seems to be correlated with high reproduction rates in this typical ‘source population’. With regard to tusks, Spinagallo elephants were certainly ‘variant’, Elephas antiquus females bearing tusks. Tusklessness has no widely accepted explanation. It is an inherited, non-dominant trait (Jachmann et al., 1995; Abe, 1996; Whitehouse, 2000). Whitehouse (2002) proved that the opinion of tusklessness as borne out from intense poaching (Jachmann et al., 1995; Abe, 1996) is no more than a truism. South African elephants were heavily poached, but Addo Park females were entirely tuskless, whereas another South African population, that of Knysna has a low incidence of tusklessness (Whitehouse, 2002). Whitehouse (2000) was critical about the resource-quality hypothesis of Sydack and Bigalke (1992) and Seydack et al. (2000) and proposed that tusklessness is related to a non-selective genetic drift for Addo elephants. Addo Park elephants were fenced in 1931. At that time, four out of eight females present were tuskless. If tusk presence gives to some females a selective advantage over the others (Raubenheimer, 2000), it is hard to believe that tusklessness is ‘non-selective’ for Addo elephants, 98% of females currently being tuskless. Indeed, Whitehouse (2002) does not reject the hypothesis that the great reproductive effort in Addo elephants is linked to tusklessness, but emphasizes the role of genetics for this trait. For our purpose, the important finding is that increased reproductive investment and tusklessness are correlated, irrespective of the causal relationship. Comparison with Addo elephants, helps us in suggesting that a rather favourable environment, without both interspecific competition and predation, and with high quality, at least seasonally (see below), abundant food supply allowed E. falconeri to enhance its own reproductive efforts. This conclusion is crucial and deserves further explanations. Indeed, there are many evidences that total energy intake in mammals shares among growth, repair and reproduction (Holliday, 1989; Shanley and Kirkwood, 2000; Chicon, 2001; Roff, 2001) with the observation that reproduction rate invariably 304 enhances when growth ceases or slows down. We argue that a considerable energy saving in growth (i.e. to grow up to 100 kg is definitely cheaper than growing up to 15 tons) allowed Elephas falconeri to enhance its reproduction rate. With the high abundance of immature individuals, our estimate of 58.8% should be an underestimating figure because the tibia is the first long bone to fuse (thus growth takes place beyond tibia epiphyses fusion) and tusklessness in females to stand as proofs. Without predators to escape from or other large herbivores to outcompete, the advantage of a large body size vanished, and the balance of competition efforts shifted from interspecific to intraspecific competition selecting for smaller individuals able to reproduce earlier and at higher rates. At least two observations strongly bolster this hypothesis. The first, anticipated reproduction could be proven by the fact that adult Spinagallo elephants were paedomorphic (Ambrosetti, 1968; Caloi and Palombo, 1994; Palombo, 2001). The second is that according to the mean method (Plavacan, 1994), sexual dimorphism in linear measurements was 1.32 for E. falconeri, a quite high level (cf. data in Jarman, 1983). It is readily argued that under high levels of competition for females, sexual dimorphism should enhance. We have suggested above that the high calf mortality in E. falconeri could be related to reproductive factors (i.e. smaller calves are less resistant than larger ones). Paedomorphosis indicates that E. falconeri grew fast, and fast-growing organisms have higher juvenile mortality, mainly because of physiological factors (Mangel and Stamps, 2001). Under this consideration, the puzzling feature of the high calf mortality under very low potential for extrinsic-based mortality is resolved. An ‘intrinsic’ explanation could now be invoked for Spinagallo elephants. Thus, in comparison with its mainland forebear, the straight-tusked elephant E. antiquus (Caloi et al., 1993), E. falconeri clearly adopted a completely different life strategy. Pianka (1970, 2000) suggested that species evolve according to two different strategies, r- and K-selection. He underlined that r-strategists have a reduced body size, higher reproductive rates, earlier sexual maturity, high juvenile mortality, scarce competitive abilities and a wide morphological variability (Pianka, 1970, 2000). Since we believe that the r- and K-selection model applies here, in the remainder of this paragraph, we will analyse how, if ever, predictions of this model we listed above are verified in the case of this dwarf elephant species. It is important to underline that we are not saying E. falconeri was r-selected, we are indeed arguing that it shifted toward the ‘fast’ continuum in mammalian life-history traits. To begin, we have seen above that E. falconeri was paedomorphic, and Ambrosetti (1968) and Palombo (1996) also remarked it was morphologically highly variable. 305 One of the important causes of r-selection is a highly seasonal and unpredictable environment. Palombo (1996) suggested that the habitat of E. falconeri was more open than that of E. antiquus, thus somehow resembling that of the modern African elephant, Loxodonta africana. Pollen data (Bertoldi et al., 1989; Suc et al., 1995) upheld this conclusion. At the time of E. falconeri, Sicily had an environment dominated by grasses, with dispersed, deciduous trees, and severe summer droughts. This environment should have been highly seasonal as regards to resource availability. Whatever be the actual extent of seasonality, resources for E. falconeri were annually inconstant, not scarce. In order to cope with seasonality, modern elephants migrate, looking for water and/or food (Laursen and Bekoff, 1978; Shoshani and Eisenberg, 1982), but an animal confined to an island could not migrate. Besides, large living herbivores are fine-tuned to the changes occurring seasonally in their environment: They know when to move and in which direction. The same can hardly hold true for E. falconeri. We argue that for the latter, another type of environmental unpredictability could exist, just based on the fact that E. falconeri, as well as any other colonizer, was not well synchronized with its own environment, simply because it was not a long standing species on its island. We suggest that the scarce knowledge of its environment played a role in the evolution (at least at the onset) of dwarfism; it is worth mentioning again that only a few thousand years occur for a species to become dwarfed on an island. Under the r-selection strategy, reproductive season should be tuned with seasonality. If Spinagallo elephants reproduced all the year round, as modern elephants do (Laursen and Bekoff, 1978; Shoshani and Eisenberg, 1982), we expect a continuous distribution if one ranks individuals by size (or age, i.e. since no gaps occur in the reproductive period, no gap should occur in size distribution). We report a frequency histogram of left tibiae length for E. falconeri, computed including only bones with detached epiphyses (i.e. only immature individuals) (Fig. 3). Present day elephants reproduce all the year round (Laursen and Bekoff, 1978; Shoshani and Eisenberg, 1982; Moss, 2001) even if wet seasons are clearly preferred for parturition (Moss, 2001). If reproductive events are not gathered, a continuous distribution of immature individuals, from calves to young adults, is expected in such a histogram. If E. falconeri cows give birth synchronically, only one reproductive season per year is expected, in keeping with some 189 day long pregnancy period (Roth, 1992), and some gaps in age profiles should occur. Theoretically, continuous distributions may be tested by means of a v2 comparison with a Poisson distribution (Zar, 1984), yet this test requires the probability of occurrence of individuals to be the same for any given age (or size) class, which is clearly not the case because of mortality and preferential reproduction in wet seasons. In addition, a non-random distribution might result from intervening gaps due to occasional reproduction suppression in times of severe droughts (Haynes, 1985; 306 10 distribution of juvenile's tibiae frequency 8 6 4 2 177.5 167.5 157.5 147.5 137.5 127.5 117.5 107.5 97.5 87.5 77.5 67.5 57.5 47.5 0 tibia diaphyses length Figure 3. Frequency histogram of left tibiae without epiphyses (juvenile individuals). Dudley et al., 2001). As for droughts, Haynes (1985) and Dudley et al. (2001) reported that droughts commonly affect young Hwange Park (Kalahari region) elephants because they are unable to reach the water table, which could be 2 m in depth. Haynes found gaps in his population structure to be highly correlated with intense drought periods. Adult Spinagallo elephants were some 0.9 m at the withers, and even as adults they were probably unable to reach the water table, let alone the fact that they might not have been as clever as a fully grown adult African elephant to dig such deep holes. This analysis is plagued from its reluctance to statistical testing and even if some gaps occur in the distribution, this fact should not be taken as being conclusive. That of r- and K-selection is just a general model, a useful oversimplification to explain evolutionary trends. However, even if the r- and K-selection model is still a rather debated one, we feel that demography and morphological features of E. falconeri show an intriguing adherence to this model. The r- and Kselection model was already invoked to explain the Island rule for rodents (Tamarin, 1978). A conclusion similar to that of Tamarin’s was proposed by Adler and Levins (1994), but those authors were critical about the r- and Kselection model itself. Nevertheless, strong experimental evidences on fruitflies (Guo et al., 1991; Mueller et al., 1991) and lizards (Sinervo et al., 2000) seem to prove that the r and K selection model does apply. As for E. falconeri, we suggest that in the few thousand years of evolution dwarfism requires to occur (Lister, 1989; Vartanyan et al., 1993), the very slow rate of population increase typical for elephants (maximum 7%, according to Calef, 1988) was unsuited to match new opportunities given by low initial densities, relaxed intra- and inter-specific competition and lack of potential 307 predators. Through mechanisms such as reaction norms (Stearns and Koella, 1986; Adler and Levins, 1994; Brommer, 2000), elephants probably were selected to give birth progressively earlier and at lower body weights, shifting resource allocation from maintenance to reproduction. Positive selection for highly reproductive individuals under very low extrinsic mortality further favoured this trend. Our treatment of evolutionary aspects of dwarfism is in rather good agreement with a similar prediction coming from Demetrius’ evolutionary entropy (Demetrius, 2000). As argued by him, large mammals far below their carrying capacities are likely to become dwarfed as a result of maximization of individual fitness, a prediction quite similar to that which we state here. Summary and conclusions The goal of this study was to investigate dwarfism in the extinct Sicilian elephant Elephas falconeri. This was made through the analysis of the survivorship curve for a very abundant fossil population of E. falconeri recovered from the Spinagallo caves (Sicily). Many studies placed emphasis on resource limitation as a key factor for dwarfism of large mammals living on islands. Stunting could certainly arise from calorie restriction (Shanley and Kirchwood, 2000), but reproduction is thought to be reduced to a considerable extent when this is the case. Indeed, the study of some morphological and ecological reconstructed traits of E. falconeri, such as the absence of tusks in females, the paedomorphosis and the high morphological variability, lead us to think of the hypothesis of resource limitation as being inapplicable. Instead, the particular ecology of this species delineates dwarfism like an evolutionary response directed to maximize individual fitness in a condition of inconstant, but seasonally abundant, resource supply, whereas no competition or predation evidence is there. Comparison of E. falconeri with living Addo elephants gives a strong support for our hypothesis. Addo’s are not dwarfed; nevertheless, their heyday started only a few decades ago (in 1931, Whitehouse and Hall-Martin, 2000) and not on an island, and dwarfism requires more time to occur (Lister, 1989; Vartanyan et al., 1993). Further, they compete with an ungulate-rich fauna; hence, mediumsized herbivore niche may be pre-filled, thus hampering dwarfism (was the same true for Elephas manaidriensis?). In addition, we argue that the occurrence, in the Pleistocene record of the Mediterranean islands, of some very different evolutionary responses (in terms of body size change), in different ecological conditions, even for the same species (here the case of the two dwarf elephant species of Sicily is perhaps emblematic) should be re-evaluated in the light of interspecific competition. To be less vague, we propose that community 308 assembly rules could play a role in explaining the island rule, with the case of Elephas falconeri standing as a special, one-species assembly case. To conclude, we suggest the evolution of the dwarfism in E. falconeri roughly occurred through the following steps: (i) Colonization of the Island by a founder population greatly limited in number. (ii) Positive selection toward those individuals capable of shifting resource allocation from somatic maintenance to reproduction, with the attainment of a ‘dwarf ’ body size at reproduction for paedomorphic individuals. This was possible given the absence of both predators and potential competitors and the demonstrated heritability of growth curve changes (Roff, 2000). (iii) Progressive fine-tuning of the reproductive period in order to cope with a seasonal food supply, whereas migration was prevented by geographical barriers. Acknowledgements Rossana Sanfilippo kindly housed two of us during our stay in Catania. Francesco Sciuto made the material on E. falconeri in his care readily available to us. David McFarlane kindly gave us some bibliographic information at the very beginning of the manuscript preparation. Anna Whitehouse and Eric Raubenheimer kindly gave us a copy of some of their papers. Adrian M. Lister, Anna Loy, Maria Rita Palombo, Louise V. Roth, Marco Signore, N. Chr Stenseth and one anonymous reviewer read the manuscript and provided us with meaningful suggestions and important advice that greatly improved the manuscript. References Abe, E. (1996) Tusklessness among the Queen Elizabeth National Park elephants, Uganda. Pachyderm 22, 46–47. Adler, G.H. and Levins, R. (1994) The island syndrome in rodent populations. Quart. Rev. Biol. 69(4), 473–490. Alcover, J.A. and McMinn, M. (1994) Bioscience 44, 12–19. Ambrosetti, P. (1968) The Pleistocene dwarf elephants of Spinagallo (Siracusa, South-Eastern Sicily). Geol. Rom. 7, 277–398. Bada, J.L., Belluomini, G., Bonfiglio, L., Branca, M., Bugio, E. and Delitala, L. (1991) Isoleucine epimerization ages of Quaternari mammals of Sicily. Il Quaternario 4(1a), 5–11. Belluomini, G. and Bada, J.L. (1985) Isoleucene epimerization ages of the dwarf elephants of Sicily. Geology 13, 451–452. Berger, J. and Gompper, M.E. (1999) Sex ratios in extant ungulates: products of contemporary predation or past life histories? J. Mammal. 80(4), 1084–1113. 309 Bertoldi, R., Rio, D. and Thunell, R. (1989) Pliocene–Pleistocene vegetational and climatic evolution of the South-Central Mediterranean. Palaeog. Palaeocl. Palaeoe. 72, 263–275. Bonfiglio, L., Di Geronimo, S., Insacco, G. and Marra, A.C. (1996) Riv. Ital. Pal. Strat. 102(3), 375–384. Bonfiglio, L., Marra, A.C. and Masini, F. (2000) The contribution of Quaternary vertebrates to paleoenvironmental and paleoclimatological reconstructions in Sicily. In M.B. Hart, (ed.) Climates: Past and Present. Geological Society, London. Special publications, Vol 181, pp. 171–184. Brommer, J.E. (2000) The evolution of fitness in life-history theory. Biol. Rev. 75, 377–404. Brown, J.H., Marquet, P.A. and Taper, M.L. (1993) Evolution of body size: consequences of an energetic definition of fitness. Am. Nat. 142, 573–584. Calder, W. (1996) Size, Function and Life History. Dover Publications, Olineole, New York, USA. Calef, G.W. (1988) Maximum rate of increase in the African elephant. Afr. J. Ecol. 26, 323–327. Caloi, L. and Palombo, M.R. (1994) Functional aspects and ecological implications in Pleistocene endemic herbivores of Mediterranean islands. Hist. Biol. 8, 151–172. Caloi, L., Kotsakis, T., Palombo, M.R. and Petronio, C. (1993) The Pleistocene dwarf elephants of mediterranean islands. In J. Shoshani and P. Tassy, (eds) The Proboscidea: Evolution and Palaeoecology of Elephants and Their Relative. Oxford Science Publications, Oxford, UK. Case, T.J. (1978) A general explanation for insular body size trends in terrestrial vertebrates. Ecology 59, 1–18. Charnov, E.L. (1991) Evolution of life history variations among female mammals. Proc. Nat. Acad. Sci. USA 88, 1134–1137. Charnov, E.L. (2001) Evolution of mammal life histories. Evol. Ecol. Res. 3, 521–535. Chicon, M. (2001) Diversity of age-specific reproductive rates may result from ageing and optimal resource allocation. J. Evol. Biol. 14, 180–185. Clutton-Brock, T.H., Albon, S.D. and Guinness, F.E. (1985) Parental investment and sex differences in juvenile mortality in birds and mammals. Nature 313, 331–133. Damuth, J. (1982) Analysis of the preservation of community structure in assemblage of fossil mammals. Paleobiology 8, 434–446. Damuth, J. (1993) Cope’s rule, the island rule and the scaling of mammalian population density. Nature 365, 748–750. Damuth, J.D. and MacFadden, B.J. (eds) (1990) Body Size in Mammalian Paleobiology. Cambridge University Press, 397pp. Dayan, T. and Simberloff, D. (1998) Size patterns among competitors: ecological character displacement and character release in mammals, with special reference to island populations. Mammal Rev. 28(3), 99–124. Demetrius, L. (2000) Directionality theory and the evolution of body size. Proc. Roy. Soc. Lond. B 267, 2385–2391. Demment, M.W. and Van Soest, P.J. (1985) A nutritional explanation for body-size patterns of ruminant and nonruminant herbivores. Am. Nat. 125, 641–672. Dodd, J.R. and Stanton, R.J. (1990) Paleoecology, Concepts and Applications. John Wiley and Sons, New York, USA, 502pp. Dudley, J.P., Criag, G.C., Gibson, D.St.C., Haynes, G. and Klimowicz, J. (2001) Drought mortality of bush elephants in Hwange National Park, Zimbabwe. Afr. J. Ecol. 39, 187–194. Ehrlich, P.R. (1986) Which animal will invade? In H.A. Mooney and J.A. Drake, (eds) Ecology of Biological Invasions of North American and Hawaii. Springer Verlag, New York, pp. 79–95. Foster, J.B. (1964) The evolution of mammals on island. Nature 202, 234–235. Gasser, M., Kaiser, M., Berrigan, D. and Stearns, S.C. (2000) Life-history correlates of evolution under high and low adult mortality. Evolution 54(4), 1260–1272. Geffen, E., Gomper, M., Gittleman, J.L., Hang-Kwang Luh, MacDonald, D.W. and Wayne, R.K. (1996) Size, life-history traits, and social organization in the canidae: a reevaluation. Am. Nat. 147(1), 140–160. Gilinsky, N.L. and Bennington, J.B. (1994) Estimating number of whole individuals from collections of body parts: a taphonomic limitation of the paleontological record. Paleobiology 20(2), 245–258. 310 Greenwood, P.J. (1980) Mating systems, philopatry, and dispersal in birds and mammals. Anim. Behav. 28, 1140–1162. Guo, P., Mueller, L.D. and Ayala, F.J. (1991) Evolution of behaviour by density-dependent natural selection. Proc. Nat. Acad. Sci. 88, 10905–10906. Guthrie, R.D. (2001) Reconstruction of woolly mammoth life history. Proc. 1st International Congress The World of Elephants Rome, 16–20 October 2001. Harvey, P.H. and Zammuto, R.M. (1985) Patterns of mortality and age at first reproduction in natural populations of mammals. Nature 315, 319–320. Heaney, L.R. (1978) Island area and body size of insular mammals: evidence from tri-colored squirrel (Callosciurus prevosti) of SouthEast Asia. Evolution 32, 29. Holliday, R. (1989) Food, reproduction and longevity: is the extended lifespan of calorie.restricted animal an evolutionary adaptation? Bioessays 10, 125–127. Haynes, G. (1985) Age profiles in elephants and mammoth bone assemblage. Quat. Res. 24, 333– 345. Jachmann, H. (1980) Population dynamics of the elephants in the Kasungu National Park, Malawi. Neth. J. Zool. 30, 622–634. Jachmann, H., Berry, P.S.M. and Imae, H. (1995) Tusklessness in African elephants: a future trend. Afr. J. Ecol. 33, 230–235. Janis, C. (1989) A climatic explanation for patterns of diversity in ungulate mammals. Paleontology 32, 463–481. Jarman, P. (1983) Mating system and sexual dimorphism in large, terrestrial mammalian herbivore. Biol. Rev. 58, 485–520. Keller, L. and Genoud, M. (1997) Extraordinary lifespans in ants a test of evolutionary theories of ageing. Nature 389, 958–960. Kelt, D. (1997) Assembly of local communities: consequences of an optimal body size for the organization of competitively structured communities. Biol. J. Linn. Soc. 69, 15–37. Kelt, D.A. and Brown, J.H. (1998) Diversification of body sizes: patterns and processes in the assembly of terrestrial mammal faunas. In McKinney and J.A. Drake (eds) Biodiversity Dynamics, Turnover of Populations, Taxa and Communities. Columbia UP, New York, USA, pp. 109–132. Koch, P.L., Hoppe, K.A. and Webb, D.S. (1998) The isotopic ecology of late Pleistocene mammals in North America Part 1. Florida. Chem. Geol. 152, 119–138. Krukonis, G. and Schaffer, W.M. (1991) Population cycles in mammals and birds: does periodicity scale with body size? J Theoretical Biology 148, 469–493. Kurten, B. (1954) Population dynamics-a new method in Paleontology. J. Paleont. 28, 286–292. Laursen, L. and Bekoff, M. (1978) Loxodonta africana. Mammal. Species 92, 1–8. Lawlor, T.E. (1982) The evolution of body size in Mammals: evidence from insular population in Mexico. Am. Nat. 119(1), 54–72. Lister, A.M. (1989) Rapid dwarfing of red deer on Jersey in the last interglacial. Nature 342, 539– 542. Lister, A.M. (1993) Mammoths in miniature. Nature 362, 288. Lomolino, M.V. (1985) Body size of mammals on islands: the island rule reexamined. Am. Nat. 125, 310–316. MacArthur, R.H. and Wilson, E.O. (1967) The Theory of Island Biogeography. Princeton University Press, Princeton (NJ), USA. Mangel, M. and Stamps, J. (2001) Trade-offs between growth and mortality and the manteinance of individual variation in growth. Evol. Ecol. Res. 3, 583–593. McComb, K., Moss, C., Durant, S., Baker, L. and Sayialel, S. (2001) Matriarchs as repositories of social knowledge in African elephants. Science 292, 491–494. McFarlane, D.A., MacPhee, R.D.E. and Ford, D.C. (1998) Body size variability and a sangamonian extintion model for Amblyrhiza, a west indian megafaunal rodent. Quat. Res. 50, 80–89. Moorcroft, P.R., Albon, S.D., Pemberton, J.M., Stevenson, I.R., Mueller, L.D., Guo, P. and Ayala, F.J. (1991) Density-dependent natural selection and trade-offs in life history traits. Science 253, 433–435. 311 Moss, C.J. (2001) The demography of an African elephant (Loxodonta africana) population in Amboseli, Kenya. J. Zool. Lond. 255, 145–156. Mueller, L.D., Guo, P. and Ayala, F.J. (1991) Density-dependent natural selection and trade-offs in life history traits. Science 253, 433–435. Palombo, M.R. (1996) Large Pleistocene mammals of the mediterranean islands. Vie et Milieu 46, 365–374. Palombo, M.R. (2001) Paedomorphic features and allometric growth in the skull of Elephas falconeri from Spinagallo (Middle Pleistocene, Sicily) Proc. 1st International Congress The World of Elephants, Rome, 16–20 October 2001. Pianka, E. (1970) On r and k selection. Am. Nat. 104, 592–597. Pianka, E. (2000) Evolutionaty Ecology. Addison Wesley Inc., San Francisco, USA, 511pp. Promislow, D.E.L. and Harvey, P.H. (1990) Living fast and dying young: a comparative analysis of life history variation among mammals. J. Zool. Lond. 220, 417–437. Purvis, A. and Harvey, P.H. (1997) The right size for a mammal. Nature 386, 332–333. Raubenheimer, E.J. (2000) Development of the tusk and tusklessness in African elephant (Loxodota africana). Koedoe 43(2), 57–64. Roff, D.A. (2000) Trade-offs between growth and reproduction: an analysis of the quantitative genetic evidence. J. Evol. Biol. 13, 434–445. Rollo, C.D. (2002) Growth negatively impacts the life span of mammals. Evol. Develop. 4, 55– 61. Rosenzweig, M.L. and McCord, R.D. (1991) Incumbent replacement: evidence for long-term evolutionary progress. Paleobiology 17(3), 202–213. Roth, V.L. (1984) How elephants grow: heterochrony and the calibtration of developmental stages in some living and fossil species. J. Vert. Paleo. 4(1), 126–145. Roth, V.L. (1990) Island dwarf elephants: a case study in body mass estimation and ecological inference. In J. Damuth and J.B. MacFadden, (eds) Body size in Mammalian Paleobiology. Cambridge University Press, New York, USA, pp. 51–179. Roth, V.L. (1992) Inferences from allometry and fossils: dwarfing of elephants on islands. In D. Futuyma and J. Antonovics (eds) Oxford Surveys in Evolutionary Biology, Vol. 8 Oxford University Press, New York, USA. Roy, K., Jablonski, D. and Martien, K. (2000) Invariant size-frequency distribution along a latitudinal gradient in marine bivalves. Proc. Nat. Acad. Sci. 97, 13150–13155. Ruggiero, R.G. (1991) Opportunistic predation on elephant calves. Afr. J. Ecol. 29(1), 86–89. Seydack, A.H.V. and Bigalke, R.C. (1992) Nutritional and life history tactics in the bush pig (Potamochoerus porcus): development of an interactive model. Oecologia 90, 102–112. Seydack, A.H.V., Vermeulen, C. and Huisamen, J. (2000) Habitat quality and the decline of an african elephant population: implications for conservation. South. Afr. J. Wild Res. 30(1), 34– 43. Shanley, D.P. and Kirchwood, T.B. (2000) Calorie restriction and aging: a life history analysis. Evolution 54(3), 740–750. Shoshani, J. and Einsenberg, J.F. (1982) Elephas maximus Mammal. Species 182, 1–8. Sinervo, B., Doughty, P., Huey, R.B. and Zamudio, K. (1992) Allometric engineering: a causal analysis of natural selection on offspring size. Science 258, 1927–1930. Sinervo, B., Svensson, E. and Comendant, T. (2000) Density cycles and an offspring quantity and quality game driven by natural selection. Nature 606 (4069), 85–88. Sondaar, P.Y. (1977) Insularity and its effect on mammalian evolution. In P.C. Goody, M.K. Hecht and B.M. Hecht (eds) Major Patterns in Vertebrate Evolution. Plenum Press, New York, USA, pp. 671–707. Stearns, S.C. (1983) The influence of size and phylogeny on patterns of covariation among lifehistory traits in the mammals. Oikos 41, 173–187. Stearns, S.C., Ackermann, M., Doebeli, M. and Kaiser, M. (2000) Experimental evolution of aging, growth, and reproduction in fruitflies. Proc. Nat. Acad. Sci. 97, 3309–3313. Stearns, S.C. and Koella, J.C. (1986) The evolution of phenotypic plasticity in life-history traits: predictions of reaction norms for age and size at maturity. Evolution 40(5), 893–913. 312 Suc, J.P., Bertini, A., Combourieau-Nebout, N., Diniz, F., Leroy, S., Russo-Ermolli, E., Zheng, Z., Bessais, E. and Ferrier, J. (1995) Structure of West Mediterranean vegetation and climate since 5.3 ma. Acta. Zool. Cracov. 38, 3–16. Tamarin, R.H. (1978) Dispersal, population regulation, and k-selection for field mice. Am. Nat. 112, 554–555. Tuljiapurkar, S. (1990) Delayed reproduction and fitness in variable environments. Proc. Nat. Acad. Sci. 87, 139–143. Vartanyan, S.L., Garutt, V.E. and Sher, A.V. (1993) Holocene dwarf mammoths from Wrangel islands in the Siberian artic. Nature 362, 337–340. Waser, P.M. (1996) Patterns and consequences of dispersal in Gregarious carnivores. In J.L. Gittleman (ed) Carnivore, Behavior, Ecology and Evolution (2). Cornell University Press, Ithaca and London, pp. 267–295. Western, D. (1979) Size, life history, and ecology in mammals. Afr. J. Ecol. 17, 185–204. Whitehouse, A.M. (2002) Tusklessness in the elephant population of the Addo Elephant National Park, South Africa. J. Zool. Lond. 257, 249–254. Whitehouse, A.W. and Hall-Martin, A.J. (2000) Elephants in Addo Elephant National Park, South Africa: reconstruction of the population’s history. Oryx 34, 46–55. Wittemeyer, G. (2001) The elephant population of samburu and Buffalo springs National Reserves, Kenya. Afr. J. Ecol. 39, 357–365. Zar, J.J. (1984) Biostatistical Analysis. Prentice Hall, Inglewood Cliffs, New Jersey, USA, 718pp. Zazzo, A., Bocherenes, H., Brunet, M., Beauvilain, A., Billiou, D., Mackaye, H.T., Vignaud, P. and Mariotti, A. (2000) Herbivore paleodiet and paleoenvironmental changes in Chad during the Pliocene using stable isotope ratios of tooth enamel carbonate. Paleobiology 26(2), 294–309.