New Evidence from Galeria da Cisterna (Almonda) and Gruta do Caldeirão on the Phasing of Central Portugal’s Early Neolithic

1 Introduction

The Cardial culture is the archaeological proxy for early farming in the western Mediterranean. Fifty years ago, Jean Guilaine’s review of museum collections showed that this culture had reached as far as Iberia’s Atlantic façade; the proof came from vessels or sherds bearing the diagnostic decoration that Guilaine was able to identify among the finds made at a number of Portuguese cave and open-air sites excavated in the nineteenth century or the first half of the twentieth (Guilaine & Veiga Ferreira, 1970).

One such cave site was the Galeria da Cisterna, a passage located c. 5 m above the spring of the Almonda river (Torres Novas), a tributary of the Tagus (Figures 1 and 2). Here, surface collection coupled with limited archaeological excavation had been carried out between 1937 and 1942 (Paço, Vaultier, & Zbyszewski, 1947). In 1988–1989, I explored zone AMD2 of the site, which had been spared by the previous excavations and yielded evidence of funerary usage in the Early Neolithic as well as in later pre- and proto-historic times (Zilhão, Maurício, & Souto, 1991; Zilhão, Maurício, & Souto, 1993). This work was carried out as my 1979–1988 exploration of Gruta do Caldeirão (Tomar) – a cave site located c. 40 km north–east of the Almonda spring – was coming to an end. Opening onto a side valley of the Nabão river, a sub-tributary of the Tagus, Caldeirão features a Palaeolithic sequence overlain by a thick Holocene deposit containing a record of nearly continuous funerary usage from the Early Neolithic onwards (Zilhão, 1992, 1993) (Figures 3–7).

Figure 1

Galeria da Cisterna. Excavation and stratigraphy. (a) Plan and cross-section of the passage, archaeological zones, and 1988–1989 excavation grid. (b) Stratigraphic profile along the F > G axis of zone AMD2 (elevations in cm below datum). (c) Entrance to the passage (left), overview of zone AMD2 prior to excavation (centre), and the basal fluviatile sands of layer C, exposed (in G19), or excavated (down to bedrock, in G20) (right).

Figure 2

Galeria da Cisterna. The Holocene context in zones AMD2 and AMD3. (a) Pottery and beads (the white scale bars associated with each item or group of items represent 1 cm). 1. Perforated red deer canine and bone bead imitation; 2. perforated shells (Theodoxus fluviatilis); 3. teardrop beads (Glycymeris sp.); 4 and 5. sperm-whale ivory buttons; 6. stamped pot; and 7–11. impressed ceramics (1–3, 7–11: Early Neolithic; 4 and 5. Bell Beaker; 6. Iron Age). (b) The human bone remains providing the MNI for the Cardial burials (not to scale).

Figure 3

Gruta do Caldeirão. Excavation and stratigraphy. (a) The entrance to the cave in 1980, at the beginning of the second field season. (b) Outline of cave walls at the surface of the sedimentary deposit and excavation grid (in darker shade, those in which the Neolithic was excavated). (c) Stratigraphy and post-depositional disturbance in the upper part of the sedimentary fill (top); note the large subsurface hollows in layer ABC-D (badger burrows) and a comparable feature in squares Q-R/12-13, which was filled with a mix of Neolithic and Magdalenian material and penetrated deeply into the underlying Solutrean sequence (bottom). (d) The reference profile at the end of the 1988 field season (elevations in cm below datum).

Figure 4

Gruta do Caldeirão. Excavation and stratigraphy. (a) Site formation process of the Holocene sequence inferred from stratigraphy, refitting and spatial distributions (after Zilhão, 1992, Figure 3.11). (b) Bubble plot of human osteological remains from layer ABC-D and links conjoining the broken human fibulae therein retrieved.

Figure 5

Gruta do Caldeirão. Spatial distribution of Neolithic diagnostics and associated radiocarbon-dated samples. (a) Horizontal scatter plot. (b) Vertical projection along the axis of the passage (elevations in cm below datum). NA2 = Cardial; NA1 = Epicardial; NM = Middle Neolithic.

Figure 6

Gruta do Caldeirão. The post-Cardial Neolithic contexts. (a) Distribution of Late Neolithic and Copper Age diagnostics (not to scale); note the preservation of spatial structure despite the significant post-depositional disturbance (Late Neolithic finds in the Back Chamber, Copper Age ones between Corridor and Entrance). (b) Impressed wares and body ornaments from the NA1 horizon. 1. Vessel 2; 2. Vessel 5; 3. Vessel 3; 4 and 5. Perforated Glycymeris glycymeris valves.

Figure 7

Gruta do Caldeirão. The Cardial context. (a) Artefacts from the NA2 horizon. 1. Vessel 10; 2. perforated Hynia pfeifferi; 3. perforated Theodoxus fluviatilis; 4. teardrop Glycymeris sp. beads; 5. flint trapeze; and 6. two sets of conjoined sherds from Vessel 1. (b) Spatial distributions. Only the dated human bone samples providing the MNI have been plotted. The bubble graphs represent the distribution per square of Vessel 1 sherds (by weight; total = 1,087 g) and perforated gastropod shells (by number; total = 140). The position of the undated female mandible (denoted by the letter F) is indicated. The positions of the single Vessel 10 sherd in O10 and the teardrop bead in R11 are approximate. Note the similar elevation of the dated, large size skeletal remains (corresponding to the surface of layer Eb, i.e. the cave floor atop which the bodies were laid down), and the significant vertical displacement of the smaller items. Elevations in cm below datum.

Since, only the directly dated sheep bones from the open-air locality of Lapiás de Lameiras (Sintra) have added fresh evidence on Central Portugal’s earliest farming communities (Davis, Gabriel, & Simões, 2018). Those bones are demonstrative of a Cardial-age occupation, but the associated material culture has yet to be published. Thirty years on, Cisterna and Caldeirão therefore remain the only modernly excavated sites upon which the culture-stratigraphy of western Iberia’s earliest Neolithic can rest. Here, I report on recently acquired evidence concerning both caves and conclude with implications for the understanding of the Mesolithic–Neolithic transition in this part of the world.

2 Sites

Nowadays, Cisterna functions as a rarely activated overflow spring. Most of the floor is bedrock, but shallow sediment infills existed in places. Remnant fluviatile sands of Pleistocene age containing the remains of beaver formed the base of the stratigraphic sequences in zones AMD1 and AMD2, which were capped by a dark, Holocene cave earth deposit. In zone AMD1, an eroded breccia of Late Magdalenian age containing stone tools and animal and human remains was found in the intermediate position (Trinkaus, Bailey, Davis, & Zilhão, 2011). In zone AMD2, a dense bat bone lens (layer B), denoting a prolonged hiatus in sedimentation, separated the fluviatile sands (layer C) from the archaeological deposit (layer A), which spilled onto zone AMD3 (Figure 1).

Layer A of Cisterna was completely excavated, with total water sieving of the sediment, in the course of two field seasons. Initial testing was carried out using arbitrary spit subdivisions, but the practice was abandoned, and the layer removed as a single package, once it became clear that we were dealing with a <40 cm-thick unstratified and homogenised palimpsest. Typological considerations enabled the rich artefact assemblage to be sorted out into its different components (Figure 2a): Early Neolithic (Zilhão, 2009; Zilhão & Carvalho, 2011; Table 1), Bell Beaker and Bronze Age (Martins et al., 2015; Zilhão, 2016), and Iron Age and Roman (Martins et al., 2015; Tente & Lourenço, 2016).

Magdalenian layer Eb caps the Palaeolithic sequence of Caldeirão (Figure 3). A hiatus in sedimentation followed, and so it was atop a Pleistocene cave floor that people of the Cardial culture deposed the bodies of their dead. Eventually, these funerary contexts were dispersed by processes of syn- and post-depositional disturbance. As a result, the skeletal remains and associated material culture offerings intruded the underlying deposit, where they intermingled with the lithics of the Magdalenian and with the associated faunal remains (which differed markedly in typology, taxonomic composition, and degree of fossilisation). In the Back Chamber (squares N-R/8-12 and P-R/13-15), sedimentation resumed with the accumulation of layer Ea, a sandy matrix éboulis containing a funerary context with Epicardial ceramics. The post-depositional disturbance of this context scattered it onto the Corridor (squares K-O/13-18), where elements thereof were found in layer Eb, mixed with likewise-intruded human remains and undecorated ceramics representing funerary usage of this part of the cave during the Middle Neolithic. The sequence was capped by a 1.5 m-thick organic cave earth deposit (layer ABC-D) devoid of internal stratification and containing abundant human remains, pottery, and other artefacts spanning the Late Neolithic to Early Medieval time range (Zilhão, 1992, 1993).

Tree roots and carnivores were the main agents of disturbance in operation at Caldeirão (note, well-apparent in the stratigraphic profile illustrated in Figure 3c, the still-hollow tunnels and chambers made by the badger, which could be followed across the excavated extent of layer ABC-D of both Corridor and Back Chamber). As a consequence, elements belonging to each of the different human occupation episodes recorded in layers ABC-D, Ea, and Eb were displaced from their original locus of deposition, and this in all directions, horizontally as well as vertically, upward as well as downward (Figure 4a).

However, as illustrated by the conjoining of broken human fibulae from layer ABC-D (Figure 4b), enough of the original spatial structure survived to allow for reliable, albeit partial and limited reconstruction of Caldeirão’s original contexts of deposition, and especially so in the case of the Neolithic material (Table 1). First, based on the diagnostic ceramics retrieved in layers Eb and Ea, three archaeological horizons were defined: NM (Middle Neolithic), NA1 (Epicardial), and NA2 (Cardial). Based on the distribution of sherds that conjoined or that, based on decoration and fabric, could be unambiguously attributed to a single vessel, the spatial envelope of each horizon was then delimited. Finally, the skeletal remains of humans and domesticated animals, the stone tools, and the objects of personal ornamentation retrieved within such spatial envelopes were assigned to the corresponding horizon.

Following the same logic, it was also possible to partially reconstruct, from finds retrieved in layer ABC-D or at its disturbed interface with underlying units, two other Caldeirão find horizons: Late Neolithic and Copper Age. Note that four green rock (muscovite and variscite) beads previously assigned to horizon NA1 (Zilhão, 1992, pp. 109–110) are now considered to belong in the Late Neolithic context. This is because (a) two were found in burrowed areas of layers Ea and Eb, (b) another came from the interface between layers ABC-D and Ea, (c) layer ABC-D yielded similar finds, and (d) recent surveys of the evidence suggest that, in Iberia, variscite beads are no earlier than the fourth millennium cal BC (Odriozola et al., 2016). Such a chronology is consistent with the types of bone pins, arrowheads, and flint knives that the green rock beads were spatially associated with.

Illustration of key artefacts, together with their spatial distributions and that of the associated dating samples, is provided in Figures 5–7.

3 Dating

Table 2 lists all the radiocarbon determinations currently available for the Holocene deposits of Cisterna. The dates on beads of tooth and bone (OxA-9287, 9288) were intended to assess contemporaneity with the Cardial context from Cova de l’Or (Valencia, Spain). The hypothesis, suggested by the style of decoration seen in Cisterna’s Vessels 1 and 3 (Figure 2a, no. 10), was supported by the dating results (Zilhão, 2001). The other samples were intended at assessing the time spans represented by the different components that could be differentiated in the palimpsested assemblages from zones AMD2 and AMD3. To do so, samples were selected using (a) a Minimum Number of Individuals (MNI) approach, applied to sheep/goat astragali and human right foot first phalanges (Martins et al., 2015) and (b) spatial proximity to Cardial sherds, applied to human bones retrieved from the base of layer A in those G-row squares where the excavation had used spit subdivisions.

In all, nine of the Cisterna results fall in the time range of the Early Neolithic. Statistical comparison shows that they form two clusters whose 1σ ranges do not overlap. The earlier cluster spans the 5238–5478 cal BC interval (2σ) and includes one human bone sample, the G21-1764 radius. The later cluster spans the 5073–5358 cal BC interval and includes five human bone samples: three mandible fragments of adult individuals, a foot phalange, and a juvenile metatarsal. These results demonstrate body deposition in at least two occasions, of minimally one person in the first and of minimally four in the second. The MNI for the Cardial burials of Cisterna is therefore five, those represented by the samples illustrated in Figure 2b. Of these, we know from aDNA analysis that the later cluster includes at least one male (represented by the G21-631 mandible) and one female (represented by the F19-385 phalange) (Olalde et al., 2015).

A chronological framework for the Caldeirão sequence was initially produced 35 years ago, in the early days of accelerator mass spectrometry dating (Zilhão, 1992, 1993) (Table 3). That dating project’s rationale was to assess the age of the Early Neolithic occupation of the site and, in the process, the integrity of the taphonomy-based assemblies of skeletal remains, stone tools, and ornaments assigned to the ceramics-defined NM, NA1, and NA2 find horizons (Figures 6b and 7a). The reconstructions passed the test, as the samples yielded results consistent with the Cardial for horizon NA2 (OxA-1033, 1034, 1035), the Epicardial for horizon NA1 (OxA-1036, 1037; TO-350), and the Middle Neolithic for horizon NM (TO-349). Additionally, sample TO-351 showed that two Visigoth iron buckles (one of armour, another of bridle) found in layer ABC-D of squares P10 and P11 related to funerary usage of the place in Early Medieval times. It could also be concluded that the statistically earlier conventional charcoal result previously obtained for layer Eb (6870 ± 210 BP; ICEN-296) represented a natural component of the deposit unrelated to the site’s Early Neolithic occupation.

Table 3 also includes new results contributed by ongoing paleonutrition and paleogenetics projects (Carvalho, 2018; Linscott, personal communication; Allentoft, personal communication). The sampling targets were as follows: (a) the four adult male left tibiae from horizon NA2; (b) cranial fragments from layers ABC-D and Eb bearing a preserved petrous; and (c) small elements of the skeleton (teeth, phalanges, metacarpals, and metatarsals) judged to be in situ within the Magdalenian and Solutrean levels (to make sure they were not Neolithic intrusions), or, conversely, found in layer ABC-D, but suspected, due to their degree of fossilisation, of having been upwardly displaced from the Magdalenian or the Solutrean (a few reworked lithic diagnostics of which having been retrieved alongside; Matias, Aubry, & Zilhão, 2019).

The new Caldeirão results add four bone remains previously thought to be in situ Palaeolithic finds to the tally of horizons NA2 (three) and NM (one). They also raise to three the MNI of layer ABC-D’s Early Medieval (Visigoth or Islamic) burials. More to our point here, they significantly improve the temporal resolution of the site’s Cardial phase. Collectively considered, the overall results allow partition of NA2 into two phases whose chronological boundaries coincide with those defined by the Cisterna set (Figure 8): c. 5250–5500 cal BC, for which, given what we know of the chronology of the Early Neolithic in southern France and Spain (e.g. Bernabeu et al., 2011; Manen et al., 2019), the “Classic Cardial” label is appropriate; and c. 5000–5250 cal BC, for which a “Late Cardial” label will be used here.

Figure 8

Calibrated radiocarbon ages and periodisation of the Early Neolithic in Central Portugal. Calibration used the IntCal20 Northern Hemisphere curve in Calib 8.1 for Windows (Reimer et al., 2020; Stuiver & Reimer, 1993). The directly dated human bones based on which the funerary activity of horizon NA2 times could be phased are labelled accordingly. The upper chronological boundary of Mesolithic burial activity at Muge is taken from Peyroteo-Stjerna (2020).

Statistical comparison further shows that the four individuals represented by the dated tibiae entered Caldeirão on at least three separate occasions and that the site’s two dated temporals, even though representing opposite sides of the braincase, are not from the same person. The radiocarbon age of the right temporal from R11 (6415 ± 36 BP, UBA-40088) is consistent with either the first or the second occasion, but its spatial proximity to the tibia from the same square (6444 ± 33 BP; Wk-38580) suggests the first. The radiocarbon age of the left temporal from Q12 (6237 ± 35 BP, UBA-40087) is consistent with the age of the tibia from P12 (6260 ± 33 BP, Wk-38579), which represents the third occasion. The other NA2 dates are consistent with such a periodisation into three phases of burial activity: from more recent to earliest, NA2-A, NA2-B, and NA2-C (Table 3; Figure 8).

A corollary of the Caldeirão dating pattern is that the skeletal remains of Cardial people found therein were laid down over an extended period, ruling out interpretations of the site as a communal burial ground. Rather, the pattern suggests spaced-out funerary usage, leading to intermittent accumulation, over the generations, of the bodies of a few selected individuals, one at a time.

4 Burial

The MNI for the skeletal remains in horizon NA2 of Caldeirão is five: the four adult male left tibiae that were dated plus the mandible and dentition of a young adult female (Jackes & Lubell, 1992). The assemblage also includes the lower left deciduous incisor of a 7-month-old infant (inventory number P12sc662). This tooth, however, will not be considered here. It came from the very base of layer Eb and bears the matrix of red silty sands characteristic of the immediately underlying Solutrean layer (Fa); it is therefore quite likely that it represents either décapage error or upward displacement (note, in Figure 3c, the large burrow in adjacent square Q12, which penetrates >40 cm into layers Fa and Fb).

In the monographic publication of the Neolithic of Caldeirão, I argued that, notwithstanding the disturbance, the layout of the burial space could be gleaned from the distribution of the archaeological remains forming the NA2 horizon (Zilhão, 1992, pp. 74–77, Figure 4.14). Based on the interpretation of the artefacts as grave goods and the assumption that the clusters of skull bones denoted the original emplacement of the bodies, the following associations were proposed: of Vessel 1 (Figure 7a, no. 6) with the female, which would have been laid down against the north wall in squares O/11-12; of the teardrop Glycymeris shell beads (Figure 7a, no. 4) and the trapeze (Figure 7a, no. 5) (and perhaps three other microliths as well) with the male laid down against the south wall in squares Q-R/12-13; and of the concentration of over 100 perforated shells of Theodoxus fluviatilis and Hinia pfeifferi (Figure 7a, nos 2 and 3) with another male laid down in adjacent square R11. The new dating results are broadly consistent with this model, as is the sexing via aDNA of some of the dated samples: the O12-57 metacarpal, the P12-251 phalanx, and the two temporals, Q12-186 and R11-8, all of which are of male individuals (Table 3; Figure 7b). For instance, as the two dated temporals are of different radiocarbon ages, the hypothesis stands unrefuted that the two clusters of cranial material found against the south wall of the Back Chamber represent at least two different individuals laid down in different spots and at different times.

The dated adult male tibia from R11, which defines burial phase NA2-C, is one of the south wall individuals. This square also yielded a diagnostically male left innominate, and the dated male temporal from the same square is likely to belong to the same individual. In this case, the radiocarbon results strengthen the interpretation that the close spatial association of the remains with the Theodoxus and Hinia beads from that square means that the latter are this person’s grave goods.

The more recent temporal from the Q-R/12 corner represents a second south wall male – perhaps the same individual whose tibia was found against the opposite wall, is of statistically identical radiocarbon age, and defines burial phase NA2-A. This tibia comes from square P12 and a diagnostically male pelvis was found alongside, broken in four fragments retrieved from O12 and P12. A metatarsal from square N15, c. 4 m away (Figure 5), is of the same radiocarbon age (6242 ± 35 BP, OxA-35146). These four bones may well belong to the same skeleton. If so, burial phase NA2-A would have been affected by more widespread scattering than the preceding ones, implying that assigning grave goods on the basis of spatial proximity is in this case more problematic. For instance, even though not refuted, the hypothesis is weakened that the teardrop beads and geometric microliths from Q-R/12-13 are associated with this male (alternatively, these items could relate to the second adult male from burial phase NA2-C, defined by a tibia from Q12 because, with available evidence, we cannot exclude that cranial remains of this second adult male are included in the adjacent Q-R/12-13 cluster). On the other hand, if the remains of the NA2-A individual are widely scattered in the area around the angle between Corridor and Back Chamber, then the possibility exists that the three thick undecorated body sherds that define Vessel 15, all retrieved in that area (square P14), belong in his set of grave goods.

No grounds exist to question the association between Vessel 1 and the O11 female mandible found against the north wall in close association with most of the vessel’s sherds (Figure 7b). The mandible is undated, but the male tibia defining burial phase NA2-B was found nearby (in two conjoined fragments from O11 and P11), and so were the male metcarpal and the male phalanx that intruded the underlying Solutrean deposit, all three of which are of indistinguishable radiocarbon age (Table 3; Figure 8). If this NA2-B individual also had accompanying grave goods, two association scenarios are conceivable: (a) given where the tibia was found, it could represent a male deposed alongside the O11 female, in which case Vessel 1 could be a shared offering; (b) both his cranium and grave goods lie in the unexcavated parts of layer Eb (rows 8–9 and squares P-R/10). If the former, we would be dealing with a double burial, perhaps that of a couple, and an exception to the one-at-a-time rule inferred from the dating evidence. If the latter, Vessel 10, represented in the NA2 assemblage by a single fragment from adjacent square O10 – a decorated handle with a fabric similar to Vessel 1 (Barnett, 1992; Figure 7a, no. 1) – could, like the tibia, stand for the spilling onto the excavated area of a largely unexplored burial context preserved against the unexcavated bottom end of the Back Chamber. This latter hypothesis is supported by the fact that the fourth dated human bone that is definitional of phase NA2-B – the O9sc97 metacarpal (Table 3) comes from square O9.

5 Phasing

The shell ornaments from horizon NA2 of Caldeirão (Figure 7a, nos 2–4) are associated with individuals from the earliest burial phase, NA2-C. Statistically identical ages were obtained for the directly dated tooth and bone pendants from Cisterna (Figure 2, no. 1), whose large assemblage of shell ornaments (Figure 2, nos 2 and 3) – over 200 whole, perforated shells of small gastropods (mostly Theodoxus fluviatilis), and over 40 teardrop beads (mostly cut from valves of Glycymeris sp., but also of Unio sp. and even of Sepia officinalis cuttlebone) (Zilhão, 2009) – is typological akin to those from horizon NA2 of Caldeirão. Bringing together the evidence from the two sites, we can conclude that shell beads, pierced red deer canines, and bone beads imitating their shape are characteristic of the Cardial from the onset.

No ceramic vessels could be directly associated with Caldeirão’s NA2-C individuals. Their bodies may have been laid out with none or, as is possibly the case with the adult male from NA2-B, such ceramics may await discovery in the site’s unexcavated area. However, the type of decoration seen in Cisterna’s Vessels 1 and 2 – dense, comb, or shell impressions extensively covering the body from rim to base, associated with large handles, and sometimes defining anthropomorphic motifs (Figure 2a, no. 10) – is characteristic of the very start of the Classic Cardial of the regional sequence of Valencia (Spain) (Bernabeu et al., 2011). Based on correlation with that sequence, I have suggested that Vessel 1 of Cisterna stood for a phase of the Portuguese Cardial preceding that represented by Vessel 1 of Caldeirão (Zilhão, 2009), which is decorated in a style characterised by shell impressions forming bands below the rim and garlands between prehension knobs (Figure 7a, no. 6). Put another way, my hypothesis was that, in Bernabeu et al. (2011) terms, Vessel 1 of Cisterna and Vessel 1 of Caldeirão belonged in the Portuguese equivalents of Phase 3 and Phase 4 of Valencia’s Classic Cardial, respectively.

The Valencia correlation is consistent with assignment of Vessel 1 of Caldeirão to this site’s burial phase NA2-B, and with assignment of Vessel 1 of Cisterna to the material culture package represented by the shell and bone beads that define both sites’ earliest moments of funerary activity – at Caldeirão, burial phase NA2-C. These inferences find additional support in the spatial patterning displayed by the human bone dates for horizon NA2 of Caldeirão, which suggests that depositions become more recent as one moves clockwise from square R11 along the walls of the Back Chamber. The pattern is apparent in Figure 7b and further strengthened by consideration of the smaller elements of the skeleton whose age is statistically the same as the NA2-B tibia’s, all of which come from north wall grid units (O9, O12, P12; Table 3). It is unlikely, therefore, that Vessel 1 and the female it presumably is associated with belong in the earliest burial phase, NA2-C (whose dated remains are all found along the north wall), or in the latest, NA2-A (whose dated remains are all found farther outward).

Excavation of the bottom end of the Back Chamber is planned and will provide a test of these conclusions. Pending the outcome, whether the differentiation of burial phases NA2-C and NA2-B of Caldeirão has wider significance and translates into a chronological subdivision of Central Portugal’s Classic Cardial must remain an open issue. With current evidence, the parsimonious interpretation, which Figure 8 retains, is that the two styles – Cisterna’s “extensive comb impressions,” and Caldeirão’s “bands and garlands of shell impressions” – were broadly coeval in the region.

The timespan of the Epicardial as defined by horizon NA1 of Caldeirão is unrepresented among Cisterna’s 18 dates on animal and human bone samples. The reasoning above associates the dated human radius from Cisterna with Vessels 1 and 2, and with the site’s shell and bone bead assemblage. The age of the other four individuals represented by the dated remains illustrated in Figure 2b is consistent with assignment to either the Classic Cardial or the Late Cardial. These must therefore be the chronological phases to which belong a significant proportion (if not all) of the site’s other 38 Early Neolithic decorated vessels. Vessel 28 (Figure 2, no. 9), featuring a narrow band of large, irregular cardial impressions disposed in two rows immediately below the rim would be consistent with the Late Cardial, while Vessel 5 (Figure 2, no. 11) is relatively close to Vessel 1 of Caldeirão and, stylistically, amenable to inclusion in the Classic Cardial. Vessel 8 (Figure 2, no. 7) is the best candidate to represent a possible Epicardial phase that radiocarbon dating failed to detect; in Table 1, however, a dozen of other vessels decorated with a range of non-Cardial impressions, incisions, and grooves are conservatively assigned to that phase as well.

6 Conclusion

Based on Cisterna and Caldeirão, the Cardial people of Central Portugal used caves for funerary purposes. Bodies were laid down one at a time, over several generations, reflecting that only a few selected individuals received this type of treatment. The associated grave goods include personal ornaments, geometric microliths, and ceramic vessels. The direct dating of sexed and aged human remains allows for the differentiation of three burial phases within the NA2 find horizon of Caldeirão. Combining the evidence from both sites, the hypothesis can be advanced, but not demonstrated, that the earliest two burial phases are associated with Classic Cardial impressed wares and the latest with Late Cardial ones.

The existence of a pre-Cardial phase of Ligurian-style affinities has been suggested for Valencia; impressed groove wares would be the index fossil of this phase (Bernabeu, Molina, Esquembre, Ramón, & Boronat, 2009). However, as shown by Martins et al. (2015), the associated radiocarbon dates are statistically indistinguishable from the Classic Cardial’s, and so such wares might be diagnostic of the latter’s earliest stages rather than of a separate cultural tradition. The evidence from Cisterna is consistent with Martins et al.’s interpretation, as the impressed groove technique is represented therein (by Vessel 4, counted as Epicardial in Table 1; Figure 2a, no. 8) but none of the dates obtained for the site fall in the time range of the Ligurian Impressa.

In Central Portugal, the Impressa time range is a period during which only Mesolithic hunter–gatherers are documented in the archaeological record. They accumulated the well-known shell-middens of the inner estuary of the Tagus, near Muge, 45 km downstream from the Almonda spring (as the crow flies), and of the Sado, farther south. Based on the available evidence, I have for long argued that these hunter–gatherer communities had persisted for a few centuries alongside the first farming communities (Zilhão, 1992, 1993, 2000, 2001, 2003). A recent study (Peyroteo-Stjerna, 2020) has added detailed, high-quality radiocarbon evidence that fully corroborates such a coexistence pattern, which, as illustrated in Figure 8, may have lasted for the entire duration of Central Portugal’s Classic Cardial – i.e. for up to two and a half centuries (c. 5500–5250 cal BC). This overlap highlights the potential of the region for testing models of last hunter–gatherer/early farmer interaction, which the ongoing stable isotope and aDNA analysis of the directly dated Cardial individuals from Cisterna and Caldeirão are expected to shed much light on.