L'Homme a provoqué de nombreuses extinctions au cours des derniers millénaires et les espèces endémiques insulaires en ont été les victimes les plus spectaculaires [23]. Pour les rongeurs, citons les exemples fournis par les îles de la Méditerranée [26], les Caraïbes, les îles Galápagos ou l'archipel des Canaries. Les espèces endémiques insulaires actuelles sont pour la plupart en danger [16]. L'interprétation des modalités et des causes des extinctions demande l'intégration de nombreuses données relatives à la chronologie, à l'histoire du climat et des environnements et à l'installation humaine. L'obstacle majeur demeure l'inégalité dans l'abondance et la qualité de ces données, surtout pour les faunes insulaires : par exemple, les âges radio-isotopiques sont souvent en nombre insuffisant et les preuves de la coexistence de l'espèce endémique et des humains sont difficiles à réunir. L'extinction est-elle provoquée par les humains, comme souvent admis maintenant (par exemple, [12,25]) ? Mais le climat a changé au cours de l'Holocène et n'a-t-il pas aussi joué un rôle ? Pour la première fois est apporté, pour l'archipel canarien, une argumentation qui souligne mieux le rôle probable de l'Homme.

Recent human expansion entailed the extinction of many species, the island faunas supplying many examples [23]. Aboriginal island mammal faunas were poor in species compared with the continental faunas, but they hold singular species, such as dwarf elephants, hippopotamuses, and cervids, as well as giant rodents and insectivores. These insular species differed strongly from their continental relatives not only by their size but also by their body proportions [26]. Almost all of these insular mammalian species are now extinct. Most of the extinction events occurred during the last millennia and examples are well known for several islands of the Mediterranean Sea, such as Corsica, Sardinia, and Balearic Islands, the Caribbean Archipelago, the Galápagos Islands, and the Canary Islands. Today nearly all extant insular species are endangered [16]. However, the search for modalities and causes of these extinctions still remains a difficult task, which needs integrating data from various research domains, such as absolute chronology, reconstruction of environment, climate history, and the study of the stages of human settlement on the island. Indeed, humans are commonly thought to be the proximal cause of the extinction of aboriginal species (e.g., [12] and [25]). The lack of data makes explanation of the extinction often uncertain, especially radiocarbon dates and direct evidence of the co-occurrence of humans and endemic species [2]. Was the extinction natural, arising before human settlement on the island? Was there a combination of climatic change and anthropogenic actions? Or are humans the single cause of extinction? Arriving in a virgin area, they could have hunted the endemic species into extinction, or introduced competitors and fatal parasites. One of the endemic rodents of the Canary Islands, Canariomys bravoi, provides a case that illustrates the complexity of the question: for the first time the role of humans is more clearly exposed.

An endemic fauna of vertebrates (bats and birds not considered here) including reptiles (terrestrial turtles, one snake, lizards, geckos and skinks) and mammals (three rodents and one shrew) was present on the Canary Islands before human settlement. The snake, the turtles, the rodents and several species of reptiles among which are giant lizards, are now extinct [13] and [27]. Each of the endemic mammalian species of the Canary archipelago was present only on one or a few islands. The rodents, that all belonged to the sub-family Murinae, the rats and the mice of the Old World, were: Malpaisomys insularis on the oriental islands [11], Canariomys bravoi on Tenerife [7], and Canariomys tamarani on Gran Canaria [14]. The present work deals with C. bravoi, one of the giant rats. C. bravoi was considered Pliocene and Pleistocene in age [28] until radiocarbon dating provided the first chronological information: the ¹⁴C age is 12 230 ± 140 BP (reference OxA-5450 2, [20]) which – after calibration [24] – corresponds to a calendar date of 12 430–11 940 cal BC. In the present study, collagen from five bones belonging to C. bravoi, which were previously analysed for the stable carbon and nitrogen isotopic ratios have been radiocarbon dated. These specimens come from three different sites from the area of Icod, to the northwest of Tenerife (Fig. 1).

Collagen was extracted as described previously [4]. The radiocarbon age was measured using AMS (Accelerator Mass Spectrometry) at the Centre for Isotope Research, Groningen, the Netherlands. For these radiocarbon measurements, background levels are in the range 45–50 000 years.

The diet of Canariomys is shown to be purely vegetarian. Indeed, the aboriginal rat of Tenerife, as the one from Gran Canaria, belongs to the tribe of African rats Arvicanthini [10] and [14], a group that includes species with a vegetarian diet [21]. Moreover, the δ ¹³C values and their stability in Canariomys are consistent with such a vegetarian diet in an environment of C₃ plants [4]. The variability of the δ ¹⁵N values may be explained by a change in diet composition, such as varying amounts of animal proteins of terrestrial or marine origin in addition to terrestrial plants, or by a change in the isotopic composition of an unchanging food resource, that would be terrestrial plants. The stability of the δ ¹³C values strongly suggests that the observed changes in δ ¹⁵N values cannot be explained by varying amounts of marine food in the diet. Indeed, the addition of terrestrial animal proteins would lead to a coeval increase in δ ¹³C values [3], and the study of present-day black rat on Tenerife has shown that a diet mixing terrestrial and marine resources leads to covarying δ ¹³C and δ ¹⁵N values [4], which is not observed for Canariomys. The decrease of the δ ¹⁵N isotopic ratio can be explained by a change in the δ ¹⁵N values of the consumed plants, a phenomenon that can be linked to a decrease of plant growth temperatures or to a change in aridity (e.g., [1]). Changes in aridity would also affect the δ¹³C values of plants, a phenomenon that is not observed in present case. Variations in temperature are therefore the most likely explanation for the observed changes in δ ¹⁵N values.

Botanical and palynological data lack in the Canary Islands; it is therefore necessary to search for palaeoclimatological facts on a wider area. On one side, a palynological analysis at Lake Tigalmamine (1650 m asl in the Middle Atlas) in Morocco [6] shows a strong shift in environmental conditions between the Middle and Late Holocene. This climatic shift is characterised by a lowering of temperatures, especially of winter temperatures. On the other side, a core from ODP drill 568C located 20°45′N and 18°35′W off the African coast at the level of Cap Blanc (Mauritania) shows a strong increase in aeolian dust during the Holocene since ca 6000–5000 BP [8] and [9]. The dust originates from the Sahara, which has become the desert area that we know today. These data indicate a strong climatic change and likely stronger differences in temperatures between northwestern and tropical Africa. The time correlation between the change noticed in Morocco and the one in tropical Africa supports a global determinism, as acknowledged by many studies [17] and [18].

Did this climatic event drive to the extinction of Canariomys bravoi on Tenerife? Nothing supports such a hypothesis as the giant rat survived two or three additional millennia. However, the youngest ¹⁴C date allows seeing humans as the likely responsible of the autochthonous rat extinction. The δ ¹⁵N values demonstrate that in any case Canariomys recorded a change in its environment. It is nevertheless difficult to hypothesise either a general decrease in temperature or a climatically forced lowering of the vegetation zone. The Cueva del Viento locality has yielded several specimens of C. bravoi, whose skeletons were not dispersed, many bones being in connexion. It is therefore likely that animals entered the cave but were unable to leave and subsequently died. It is not possible to determine whether these animals were coming from the lower or upper part of the slope and that may be important for interpreting the δ ¹⁵N values in the context of this island with a very strong relief. Indeed, the elevation of the site is ca 650 m asl and the steep slope makes area of lower and higher elevations within close distance to the site.

The most recent radiocarbon date indicates that humans and C. bravoi have much likely lived at the same time on Tenerife. Ages in relation with aboriginal settlements refer to the last millennium BC and first millennium AD. For a time, it was accepted that no absolute dates for archaeological sites in the Canary Islands were older than 2490 BP, in the site of Barrondo Hondo in Tenerife [22]. A more recent compilation of ¹⁴C ages does not change the conclusion for Tenerife, with radiocarbon ages extending from 2770 to 1390 BP [19]. The co-occurrence of humans and the giant rat on Tenerife is still not demonstrated by archaeological facts, contrary to C. tamarani on Gran Canaria [5] and [14] and Malpaisomys insularis on the eastern islands [11].

The dates reported in the present work are the first direct evidence for chronological overlap between the extinct endemic rat from Tenerife and human occupation. As the date of human settlement is still not definitely established, it is possible that humans settled first in the more suitable low-altitude part of the island and therefore pushed Canariomys upwards. Afterwards, hunting and competition with domestic animals and commensal species or their parasites may have driven Canariomys to extinction.

As mentioned in the introduction, endemic insular species of rodents still exist. Among them are several Murinae, as the giant rats Crateromys and Phloeomys in the Philippines Islands, and Tokudaia and Diplothrix on a few islands of the Okinawa Archipelago [21], the former rodent being of the size of a large field mouse and the latter of the size of a large rat. The area of the Philippines Islands is much larger that any of the Canary Islands and may explain the survival of these large endemic species [15]. The two Japanese species, which occupy rather small islands, the largest one being of the size of Gran Canaria, are highly endangered.

The present study shows that Canariomys bravoi recorded a climatic shift that occurred at the beginning of the Late Holocene. It is extremely likely that the giant rat was present when the first humans arrived on Tenerife, in agreement with observations already made on Gran Canaria and western Canary Islands, where remains of endemic rodents are collected in archaeological deposits [5], [11] and [14]. Humans still remain the main cause of species extinctions. Consequently, we argue that humans played a role in the extinction of this endemic rodent. Since the exact date of human settlement on the island is still unknown, the shift of δ ¹⁵N values may indicate that Canariomys bravoi either recorded a climatic trend or that it was first forced by humans to occupy more elevated areas, where δ ¹⁵N values are lower than in the lowland. Humans may have hunted the endemic rodent, but they have certainly brought it on the verge of extinction because of competition with domesticated and commensal animals, both being able to have brought diseases with them.