1.1. Recovery methods

5Most of the sediment excavated between 2011 and 2016 at Klimonas was water-sieved through 5 and 3 mm meshes. Subsamples of sediments from the richest faunal units were water-sieved using a 2 mm mesh (with 10% of it sieved through 1 mm). They represent a total of 6600 litres of sediment (see chap. 24) from 34 stratigraphic units (SU), with 1 to 12 bags of 20 litres of sediments per SU. Only some of the residues were sorted under a lens or a stereomicroscope, and the sorting of sterile samples was rapidly stopped. The total volume sieved using a fine mesh and actually sorted can be estimated at around 97 bags of 20 litres each, i.e., 1940 litres of sediment.

1.2. main characteristics of the rodent bones

6We retrieved a total of 41 rodent bones or teeth (appendix 28-1), i.e., one for every 47 litres, on average. None came from modern contaminations because they all bear a dark patina, are sometimes burnt and, except on very small islands, mice never dig burrows (Niethammer and Krapp 1978). They can all be considered as PPNA specimens. Even though one specimen (lower incisor) shows clear digestion marks (dog?), the low density of rodent bones indicates that they were not accumulated on the site by predators, such as birds of prey (Andrews 1990). They probably come from commensal individuals that died on the site.

7Table 28-1 shows that all body parts are present (skull, vertebrae, anterior and posterior limbs). The lower incisors, easily identified by the excavators, were the most numerous, and indicate a minimum number of six individuals. About one third of the specimens can by identified as muridae, according to the criteria of Vigne (1995). One-third of the remains are diaphyses of unidentified rodent long bones, suggesting that this set of specimens only represents a small fraction of the original number of dead mice on the site.

Tab. 28-1 – Anatomic distribution of the 41 rodent specimens found at Klimonas in number of specimens (NISP) and minimal Anatomic Unit (MAU).

8In Sector B, five of the more than 3000 large mammal bones bear clear rodent-gnawing marks (rib, long bone, metacarpus, phalanx; fig. 28-1). These are often highly eroded due to the poor preservation of the bone surface. Here again, this very small number of marks are probably the only remains of many others destroyed by taphonomic processes. Four of these marks came from the primary fills deposited together with domestic middens in Buildings B08 and B17, just after their abandonment (SU 678-3, 6078, 6080, St 6709). Despite their scarcity, these gnawing marks clearly indicate that mice lived in the settlement and foraged the middens.

Fig. 28-1 – Fragment of a large mammal long bone showing gnaw marks from rodent incisors (KL16, SU 6737, B13).

1.3. Spatial distribution of rodent bones

9As already pointed out in chap. 24, the spatial distribution of the 41 rodent remains across the different excavation sectors is far from even. Of course, their absence in the most eroded zone (Sector F), where we did not massively wet sieve with a fine mesh because of the very low density found in the test samples, cannot be considered to signify the absence of rodent bones in this area. However, in the Central and B Sectors, the 41 rodent bones came from only 36% of the 97 sediment samples sieved with a fine mesh and then sorted. In addition, 70% of the rodent specimens came from the Communal building, versus 30% from the domestic areas of Sector B (fig. 28-2, A). This disparate distribution can be seen in the Communal building itself, where 66% of the rodent specimens came from only three of the six sieved stratigraphic units, SU 10.3, 10.4 and 10.8. However, the high rate for SU 10.3, results from the important number of sediment samples that were sieved and sorted (13 bags, i.e., c. 260 litres). Conversely, only one sediment sample was sieved with a fine mesh for SU 10.4 and 10.8. Finally, when we divide the NISP by the volume of sediment sieved for each SU (fig. 28-2, B), we observe that 89% of the total came from only three SU: 10.4, 10.8 and 10.44.

Fig. 28-2 – Distribution of the 41 rodent bones found at Klimonas among the different sectors and Stratigraphic Units (SU): A, absolute frequency; B, frequency per litre of sediment.

10Mice died in greater numbers in these places. These three stratigraphic units are closely connected with the packed earth floors of the three main successive reconstruction phases of the Communal building (chap. 5):

• SU 10.44 represents the thin packed earth floor of the primary building, located near the entrance (D34 square meter). SU 10.46, which provided the complete hemimandible, is very similar to SU 10.44; only 2 m² of this floor was visible because it was either eroded or not excavated beneath the earth benches. It is therefore impossible to surmise about the horizontal density distribution of the rodent bones in this floor;
• SU 10.8 is a vast and complex stratigraphic unit deriving from the dismantling of the walls and benches of the first building. It was used both as a backfill for building the second Communal building on the primary one and, in the south half, as the soil of the second Communal building, replacing the packed floor SU 10.6, which appeared to be poor in rodent bones. Here again, it is impossible to say more about the distribution of rodent bones in this vast SU because all the sampled sediment comes from the same square metre, F35, located near the entrance. It is only possible to observe that mouse remains were more numerous in the southern part (SU 10.8) than in the northern section of the floor in this second building;
• SU 10.4 corresponds to the extension in the northwest area (ZY36-ZZ37) of the packed floor of the third reconstruction phase of the Communal building (Building 3). Again, it is impossible to determine whether these rodent bones are representative of a high density throughout the floor of this third building, or whether they represent a local concentration, as the other lateral developments of this floor (SU 10.2 and 10.5) were not sampled for small vertebrates. Only one sample was taken for the whole floor, in SU 10.4 (square metre ZZ36, against the earth wall).

11Even though we cannot emit any hypotheses regarding the horizontal distribution of rodent bones in the Communal building, it is clear that their main concentrations on the site were in the three main successive reconstruction phases of this special building, and in different spots for each phase. murid rodents were present from the earliest to the latest settlement phases of the Communal building, which also correspond to the total duration of the whole village (chap. 10). Their close association with the (packed) soils of these buildings (fig. 28-3) indicates that these bones did not come from building earth material (where they would probably have been totally destroyed during the construction process). mice most likely lived and died in the building. This suggests that, like some other PPNA communal buildings on the continent (Stordeur and Abbès 2002, Stordeur 2015), the Communal buildings of Klimonas played a major role in the storage of cultivation products. At the same time, this strengthens our aforementioned suggestion that these murids were highly commensal.

Fig. 28-3 – Location of the three sediment samples (mouse outline) that yielded the most important densities of rodent bones on a west-east profile in the Communal building fill. The red, green and blue thick lines delimitate the surface of the floor of the three successive semi-embedded earth buildings built in the large circular St 10 excavation.

1.4. First conclusions

12The 41 rodent bones uncovered at Klimonas represent at least six individuals. However, these are probably just a few eroded remains of the many mice that lived in the village as commensal rodents, as evidenced by gnawing marks on larger mammal bones (fig. 28-1). The mice died in situ and cannot be associated with any accumulation agent such as the hawk or small carnivores. Interestingly, the concentration of these bones in three clusters corresponds to the packed circulation floors of the three successive reconstruction phases of the Communal building. This confirms that many commensal mice were attracted to the village by grain storage and suggests that crops were mostly stored in the Communal building, as was also the case in some of the communal buildings of the continental PPNA, such as Jerf el-Ahmar.

2.1. method and material

13The different mediterranean mice of the genus mus are twin species (Gerasimov et al. 1990). Given the absence of ancient DNA or of complete skulls from the archaeological deposits, their identification to species and subspecies taxonomic level can be achieved through a geometric morphometric (GMM) approach (Cucchi et al. 2002, 2012). This is especially true when distinguishing the Cypriot endemic mouse from the commensal grey mouse (Cucchi et al. 2002, 2006). We identified the species present in the village using a 2D GMM analysis of the first lower molar (M/1) occlusal view following Cucchi et al. (2020).

14Only one first mandibular molar was suitable for GMM (fig. 28-4). In order to obtain valid results, we decided to compare it to an extensive dataset of potential sympatric wild and commensal subspecies from the mediterranean Basin, both modern (N = 513) and archaeological (N = 716, extending from the Natufian to the Late Bronze Age; Cucchi et al. 2020). The fossils m. cypriacus and m. m. domesticus from Cyprus, namely from mylouthkhia, Khirokitia and Cape Andreas (Cucchi et al. 2002, 2006), are included in this dataset.

Fig. 28-4 – Occlusal view of the right first and second lower molars of the mouse hemimandible collected from the basal packed earth floor of the Communal building (SU 10.46).

2.2. Results and conclusion

15When we plot together the dental shape of the archaeological and modern mus of Southwest Asia and Europe, the m/1 of Klimonas clusters within the m. m. domesticus group (fig. 28-5). This is also confirmed by machine learning prediction (k-NN, appendix 28-2). This is the only identifiable specimen from Klimonas that can be attributed without any doubt to m. m. domesticus. It is also the earliest evidence of a house mouse stowaway transported onto an island, 300 to 500 years before the earliest known introduction of the commensal house mouse to Cyprus (Cucchi et al. 2002).

Fig. 28-5 – Projection of the first two axes of the canonical variate analysis (CVA) computed on the dental shape variation in modern wild mice and mus musculus subspecies (dots with colours) and archaeological mus sp. (grey dots and acronym of the site). Each dot represents the CVA mean scores for each modern and archaeological samples. For more information about the archaeological dataset see Cucchi et al. (2020).

3.1. mainland-Island metapopulation concept and bio-proxy of connectivity

16The ecological concept of metapopulations considers that a cluster of populations from the same spatially and temporally separated species are still interconnected through dispersal. The “mainland-island” metapopulation model (MacArthur and Wilson 1967) suggests that mainland populations act as a source for smaller insular populations, the sustainability of which is mostly dependent on migration flow. Following this “mainland-island” model, we used the phenotypic relationship between the PPN m. m. domesticus populations from Cyprus, and three potential mainland departure regions (Anatolia, Northern Euphrates and Southern Levant) for maritime routes towards Cyprus from the surrounding continent (Peltenburg et al. 2000), as proxies to elucidate the connectivity between Cyprus and the mainland during the PPN. This approach is relevant if we take into consideration (1) the current genetic differentiation of domesticus populations in three mitochondrial haplogroups (Turkey, northern Near East and southern Near East) in this region (Bonhomme et al. 2011) and (2) the phylogenetic signals encapsulated in the shape of mammal molars (Caumul and Polly 2005, Cucchi et al. 2017, Hulme-Beaman et al. 2018), supporting the use of dental shape as a phylogenetic proxy when aDNA preservation is missing. One could argue that our PPN mainland dataset does not sample PPN m. m. domesticus populations from the coasts where maritime routes should have originated. Instead, we have assumed that the house mice translocated to Cyprus were part of an economic package that colonized Cyprus after crossing long distances via the leapfrog dispersal model (Anthony 1997). In this latter model, source populations come from densely populated settlements located inland. Conversely, the dispersal pathway would have followed the known routes of more mobile PPN communities, which have left no trace between the putative area of origin and Cyprus (Peltenburg et al. 2001).

3.2. Shifting connectivity between Cyprus and the continental cultural sphere during the PPN

17Molar shape diversity in the PPN m. m. domesticus metapopulations (fig. 28-6, A) from the PPNA of Klimonas and Early and Late PPNB mylouthkia samples (1A and 2B, respectively) are closely related but significantly different (MANOVA: Pillai < 0.0001). The divergence of the Aceramic Neolithic Cape Andreas, and especially the Khirokitia samples of the 7^th millennium cal BC, confirms a lack of homogeneity among these metapopulations throughout time. By dividing the continental Natufian and PPN domesticus samples into three geographic morphogroups (Anatolia, Northern Levant and Southern Levant) and investigating their input into the diversity of the PPN domesticus of Cyprus (see method in Cucchi et al. 2020), we found that these continental sources contributed differently over time to the phylogeographic signal of the Cypriot PPN m. m. domesticus (fig. 28-6, B).

Fig. 28-6 – Diversity and mainland connectivity of the m. m. domesticus PPN metapopulations of Cyprus. A, canonical variate analysis showing dental shape differentiation among the m. m. domesticus PPN metapopulations of Cyprus. B, continental signal in the dental shape variation of the domesticus metapopulations from PPN Cyprus. Pie charts showing the greatest phenotypic relationship with (1) the Anatolian (blue), (2) the Northern Euphrates (yellow) and (3) the Southern Levant (red) PPN morphotypes according to the k-nearest neighbours (k-NN) machine learning algorithms.

18The sole PPNA specimen from Klimonas is related to the Central Anatolian PPNB morphotypes, as defined by the numerous Catalhöyük specimens (N = 105), though 3000 years later than Klimonas. This result is equivocal, since a single specimen cannot be considered as an overall representative of a population, because we only have one location in Central Anatolia and because we do not know what the phenotypic variation of the Central Anatolian mouse populations was at the time of the Klimonas occupation. On the other hand, the pattern of continental influence on the Early and middle PPNB mylouthkia specimens shows a main input from the Northern Levant, while the Khirokitia and Cape Andreas specimens (Khirokitia culture, contemporaneous with the South Levant PPNC) indicate a clear Southern Levantine prevalence. These results suggest that maritime connections between Cyprus and the different continental PPN cultural spheres changed over time: initially an Anatolian connection during the PPNA, then a North Levantine connection during the PPNB colonisation event(s) (8500–7500 cal BC), followed by a widening to the Southern Levantine connection during the PPNC/early Pottery Neolithic after 7000 cal BC.

3.3. PPN of Cyprus: shifting continental connectivity

19In Cyprus, the presence of the endemic mouse mus cypriacus in Epipaleolithic Akrotiri-Aetokremnos, and the occurrence of m. m. domesticus in the PPNA village of Klimonas, both on the southern coast, suggest that the spread of the house mouse occurred during the PPNA colonisation of the island. The hypothesis that the house mouse could have taken advantage of regular introductions of cultivated cereals from the continent is supported by the presence of macro remains of spikelet base awns and glums of emmer wheat (Triticum dicoccoides), and maybe einkorn (T. monococcum aegilopoides), introduced from the continent, as well as chaff in building material (chap. 23).

20Following its PPNA introduction, m. m. domesticus is found in the Early, middle and Late PPN and aceramic contexts from the western and southern coast up to the north-eastern tip of the island, implying that stowaway house mice found suitable ecological conditions and settled as large commensal populations in the 9^th–7^th millennia villages of Cyprus. However, the phenotypic differentiation amongst early Neolithic m. m. domesticus samples of Cyprus suggests that these populations were structured as metapopulations, in agreement with commensal rodent biology (Bonhomme et al. 2011, Combs et al. 2017).

21The phenotypic relationships between the PPN m. m. domesticus metapopulations and the different continental morphogroups that could have acted as potential dispersal sources highlight the extension, or even shifting, of the PPN connectivity of Cyprus towards the south over time. It appears, therefore, that PPN Cyprus was perhaps not a maritime hub like modern Cyprus but an island with changing continental connectivity during the course of the PPN. These phases of colonisation from different continental sources, also observed in the modern genetic diversity of m. m. domesticus populations (Bonhomme et al. 2011, Garcia-Rodriguez et al. 2018), are in agreement with the other archaeozoological and archaeological data (Vigne et al. 2019), providing us with an indirect insight into Cypriot colonisation history and the trajectory of its cultural evolution during the Neolithic.

22The PPNA sample from Klimonas relates to a Central Anatolian morphogroup, indicating that the PPNA hunter-farmers who settled in Cyprus could have originated from Anatolia, following the shortest crossing distance between the Cilician coast and Cyprus (Vigne et al. 2014), relying on seafarers’ knowledge (Peltenburg et al. 2001, Finlayson 2004, Watkins 2004). A hypothetical Anatolian origin of the Cypriot PPNA is not in agreement with a PPNA origin from the middle Euphrates, which is better supported by material culture studies (Briois and Guilaine 2014, Briois and Astruc 2017, Vigne et al. 2017), or the taxonomic evidence of the dispersal of cultivated crops (Colledge et al. 2004). However, since the PPNA of Northern coastal Syria or the Adana Plain in Anatolia have yet to be documented, this PPNA Anatolian connection remains hypothetically plausible.

23Early and middle PPNB samples from Kissonerga-Mylouthkia show that connections shifted towards the Northern Levant and lasted throughout the PPNB. Such connections are in agreement with parallels between PPNB sites in the North Levant and Cyprus (Guilaine et al. 2000, Peltenburg et al. 2001). But by the middle of the 7^th millennium BC, the mice from Khirokitia and Cape Andreas-Kastros suggest either a new connection or a shift, this time with the Southern Levant, coinciding with the broad change observed in the lithic industry sequence from Cyprus during this period. Our results suggest that Cyprus remained connected with continental cultural spheres but adjusted to local resources and produced practical responses to the demands and pressures of small-sized agricultural communities (Peltenburg et al. 2001). This change in lithic industry could have been a deliberate expression of local identity (Vigne et al. 2019), despite contacts in a broader cultural setting (Finlayson 2004). The penetration of new cultural ingredients from the Southern Levant could have contributed, by syncretism (Vigne 1997), to the idiosyncratic nature of the Late PPN culture of Cyprus.