1. Seasonal and spatial availability of pasture

The Mediterranean lowlands, especially those near the sea, experience mild wet winters with abundant pasture for grazing animals and hot dry summers with relatively scarce pasture. In Greece, as elsewhere in the Mediterranean, such lowlands are often just a few days’ walk from mountain summits that offer a wealth of green pasture in summer but are under snow in winter. The traditional twice-yearly movement of herds between such seasonally complementary grazing areas widely enabled maintenance of more numerous and better fed livestock than was possible for stationary herds. This ostensibly timeless response to environmental constraints and opportunities (Braudel Reference Braudel1975; Semple Reference Semple1922; Öncel Reference Öncel2020) has provided a model for proposed vertical movements of livestock from the Neolithic onwards in Greece: for example, to summer pastures in the high Pindos (Fig. 1) by herders (perhaps resembling our Types 4 or 5) from lowland Epirus to the west (Dakaris Reference Dakaris1976, 11, fig. 3, 15–16) and by farmers (cf. our Type 3) from lowland Thessaly to the east (Jarman, Bailey and Jarman Reference Jarman, Bailey and Jarman1982, 150–1), following similar routes to the flocks of recent Sarakatsani and Vlach pastoralists, respectively. Likewise, it is widely suggested that high-altitude settlements, such as Early Iron Age Vitsa (c. 1100 m asl) in the Pindos (Vokotopoulou Reference Vokotopoulou1986, 340) and later Neolithic sites around the Lasithi basin (c. 900 m asl) on Crete (Watrous Reference Watrous1982, 10), were occupied in summer by pastoralists who were forced by snow to migrate downwards in winter (e.g. Chandezon Reference Chandezon2003, 403–4; Hammond Reference Hammond, Boardman, Edwards, Hammond and Sollberger1982, 638–9; Snodgrass Reference Snodgrass1987, 203). Such arguments of environmental necessity or opportunity are insufficient grounds, however, to infer for antiquity either long-distance seasonal mobility with livestock or pastoralist dependence primarily on herds rather than fields. On the contrary, the upland (and lowland) sites under discussion, including the relatively extreme case of Vitsa, lie close to recent, mixed agro-pastoral villages where some livestock were reared locally year-round (cf. Types 1–2; e.g. Halstead Reference Halstead1990; Forbes Reference Forbes1995, 327–8), albeit in modest numbers limited by the labour costs of providing stored fodder (Halstead Reference Halstead1998, 227–8).

Fig. 1.

Map of Greece, showing regions, topographic features, and toponyms cited in text. Key: 1 Aiginio, 2 Argilos, 3 Argos Orestiko, 4 Athens, 5 Corinth, 6 Delphi, 7 Drakolimni, 8 Ioannina, 9 Kastro-Kallithea, 10 Kefalokhori, 11 Knossos, 12 Kouphovouno, 13 Magoula Plataniotiki, 14 Makriyalos I, 15 Metsovo, 16 Nea Alos (New Halos), 17 Nea Trapezounta, 18 Palaikastro, 19 Paliambela-Kolindrou, 20 Pherae, 21 Samarina, 22 Skoteini Cave/Tharounia, 23 Sparta, 24 Thebes, 25 Vitsa.

Fig. 1 Long description

A topographic map of Greece with an inset in the top-left corner showing Greece's location within the Mediterranean. A scale bar at the bottom-left indicates 0 to 150 kilometers.

* North: In Macedonia, sites include 3 Argos Orestiko, 19 Paliambela-Kolindrou, 1 Aiginio, 14 Makriyalos I, and 17 Nea Trapezounta. To the Northeast on the coast is 2 Argilos.

* Northwest: In Epirus and the Pindos mountains, sites include 10 Kefalokhori, 21 Samarina, 7 Drakolimni, 25 Vitsa, 8 Ioannina, and 15 Metsovo. The Zagori region is labeled here.

* Central: In Thessaly, sites are 20 Pherae, 13 Magoula Plataniotiki, 16 Nea Alos, and 9 Kastro-Kallithea. Further south in Central Greece are 6 Delphi, 24 Thebes, 22 Skoteini Cave/Tharounia on Euboea, and 4 Athens near the Cithaeron range.

* South: The Peloponnese peninsula contains 5 Corinth, Arcadia, Messenia, and the Taygetos mountains with sites 23 Sparta and 12 Kouphovouno.

* Island: To the south, the island of Crete is labeled with 11 Knossos in the center near the Ida and Mesara regions, and 18 Palaikastro at the eastern tip near Lasithi.

This last point highlights the importance of livestock numbers to understanding herders’ responses to the constraints and opportunities of complementary lowland and upland pasture: not only did seasonal movement between lowland and upland generally increase potential herd size, but equally it was not usually considered worthwhile with only small numbers of livestock (e.g. Psikhogios and Papapetrou Reference Psikhogios and Papapetrou1984, 12; Nixon and Price Reference Nixon and Price2001, 405–6). Unfortunately, direct evidence of absolute numbers of livestock is normally lacking in an archaeological context (Halstead Reference Halstead2024), but the implications for herd mobility of some ancient written records of livestock numbers are discussed below, as is the equally critical issue of the availability of pasture. At this point, we simply note that no ancient settlement known to the authors is located in a part of the Greek landscape that would have enforced seasonal removal of livestock, regardless of their number.

2. Written records of livestock numbers or movements

For early historic Greece, herd mobility is implied by numerous treaties between lowland Classical and Hellenistic (fifth–first centuries BCE) city states, governing access to pasture in border (Georgoudi Reference Georgoudi1974) or even cross-border areas (Chaniotis Reference Chaniotis and Chaniotis1999; Chandezon Reference Chandezon and Laffont2006, 56–7). The generally modest distances and altitudes involved (Chandezon Reference Chandezon and Laffont2006) are arguably as compatible with Type 2 as with Type 3 herding, however, while the conclusion of such treaties between city states arguably precludes provision for relatively independent groups of more specialised pastoralists of Types 4 or 5.

Informative literary sources from these centuries are sparse. Dio Chrysostom’s description (Discourses 7.47) of an isolated two-family farm in the hills of first-century CE southern Euboea, with only a cow, a calf and eight she-goats, plainly recalls our Type 1 herding, while references in ‘pastoral literature’ to the movement of livestock usually refer to daily displacements between byre or pen and pasture (Chandezon Reference Chandezon and Laffont2006, 53) of our Type 2. An exception is Sophocles’ presentation of the Oedipus story (Oedipus Rex 1120–252), in which shepherds met on the summer pastures of Mount Cithaeron while minding royal flocks from lowland Thebes and Corinth. This passage vividly recalls recent Type 3 herding (normal rather than inverse transhumance) and, although set in the mythical past, the encounter on Cithaeron was arguably intended to be plausible to a fifth-century BCE audience (Georgoudi Reference Georgoudi1974, 167) and has been cited as evidence for transhumant herding in Classical Greece (e.g. Skydsgaard Reference Skydsgaard and Whittaker1988). Other texts with ostensibly better claims to historicity describe wars that seemingly began in disputes over seasonal use of pasture: for example, between Spartans and Messenians in the southern mainland over summer grazing on Mount Taygetos (Howe Reference Howe2008, 68); and by communities from the central mainland for control of winter grazing below the important shrine of Delphi (Howe Reference Howe2003). Although Howe describes such competition for pasture in terms of specialised pastoralism, in the sense of ‘agricultural systems devoted to large-scale production of animals’ (Howe Reference Howe2003, 131), the livestock in question again belonged primarily to wealthy lowland landowners who could leverage civic power to promote their interests, while the accompanying herdsmen were slaves or hired workers of low status rather than economically independent pastoralists (Hodkinson Reference Hodkinson and Whittaker1988, 55; Howe Reference Howe2003, 142). Accordingly, these textual references, apparently to transhumant herding, have more in common with our Types 2 and 3 than Types 4 and 5.

A case for long-distance herding movements and perhaps for specialised pastoralism would be strengthened by textual evidence for very large numbers of livestock, but most epigraphic records for Classical to early Roman Greece (fifth–first centuries BCE), although biased towards the interests of wealthy men, attest to only modest numbers for which significant vertical or horizontal displacement cannot be regarded as necessary. For example, the four oxen, four cows and calves, 84 sheep and lambs, and 67 goats and kids of wealthy Panaitios at fifth-century BCE Athens (Chandezon Reference Chandezon2003, 17–21) are comfortably surpassed by many recent Type-2 agro-pastoral farmers of quite modest means. The livestock of Panaitios was admittedly dwarfed by the 1000 cattle and 10,000 smaller domesticates that Jason of Pherae gathered from his Thessalian subjects for sacrifice at Delphi in 370 BCE, in an intended eye-watering display of economic power. These animals, however, drawn from a territory with exceptionally extensive arable land and wetlands providing abundant summer pasture for cattle, may have been just a fraction of the wealth on the hoof of local elite farmers (Howe Reference Howe2008, 69–71) and again more plausibly represent herding of Types 2 or 3 than Types 4 or 5.

A more promising context for such a line of reasoning is Late Bronze Age central Crete, where Linear B texts from the ‘Palace of Minos’ at Knossos record something like 80,000 (mostly male) wool sheep (Killen Reference Killen1964) and, for these flocks to be sustainable, imply the parallel existence of tens of thousands of unrecorded breeding female and replacement young sheep (Halstead Reference Halstead1999, 164–6). The biggest and most numerous recorded wool flocks were apparently concentrated in lowland locations rich in arable land, such as the Mesara plain (Bennet Reference Bennet1985, 237), and so probably exploited seasonal pasture on stubble and fallow fields rather than perennial rough graze/browse, inviting speculation of their removal to high-altitude pasture in summer (e.g. Halstead Reference Halstead, Sheridan and Bailey1981a, 204). We return to this case below in discussing stable-isotope analyses of animal teeth.

Finally, although the earliest apparent written references to the Vlachs within the borders of modern Greece date back a millennium or so (e.g. Dasoulas Reference Dasoulas, Nitsiakos, Ntinas and Siokis2024), archival evidence suggests that the expansion of Type 4 specialised, market-oriented pastoralism among Vlach inhabitants of the Pindos began perhaps as late as the sixteenth or seventeenth centuries CE (Rokou Reference Rokou2007; Dasoulas Reference Dasoulas2019).

3. Type and (im)permanence of occupation traces

Mobile herders are widely cited in connection with periods and regions for which archaeological evidence suggests sparse or impermanent settlement: known open-air habitation sites are variously few, small, short-lived or temporally discontinuous and, in some cases, use of caves/rock-shelters is more widely attested; or known dwellings are flimsy and apparently ephemeral; or portable material culture is scarce and impoverished. For the early first-millennium BCE ‘Dark Ages’ of southern Greece, Classical tales of intruding ‘Dorians’ from mountainous (and thus ‘pastoral’) north-west Greece (e.g. Hammond Reference Hammond1931/2, 154) encouraged linkage of a sparse archaeological record with reliance on livestock rather than crops (Snodgrass Reference Snodgrass1987, 192–8). Thin occupation levels at Early Neolithic (seventh millennium BCE) sites in Thessaly (Whittle Reference Whittle, Darvill and Thomas1996) and meagre Final Neolithic (fifth–fourth millennium BCE) archaeological records in many parts of Greece (e.g. Douzougli Reference Douzougli1998, 147–54; Cavanagh Reference Cavanagh and Renard1999; Douzougli and Zachos Reference Douzougli, Zachos, Touchais and Renard2002, 125–6) have prompted similar arguments. Reliance on herding rather than cultivation would indeed have limited potential human population and so favoured habitation sites of restricted size and perhaps number. On the other hand, insubstantial habitation remains and chronologically undiagnostic material culture reduce archaeological visibility (e.g. for the Final Neolithic: Rutter Reference Rutter, Keller and Rupp1983; Bintliff and Sarri Reference Bintliff, Sarri, Dietz, Mavridis, Tankosić and Takaoğlu2018, 254; Siamidou Reference Siamidou2026), leading Snodgrass (Reference Snodgrass1987, 184–6) to qualify Dark Age depopulation as ‘apparent’ rather than proven. Moreover, residential mobility alone does not determine the ‘permanence’ of domestic architecture: for example, elderly residents of the sedentary, mixed-farming village at Paliambela-Kolindrou in early–mid twentieth-century CE lowland Macedonia recall relatively flimsy, wattle-and-daub huts as a rapid and cheap means of housing newly married couples with scarce labour and cash (Halstead Reference Halstead2014, 297). Similarly, while deciduous teeth (Payne Reference Payne, Wilkie and Coulson1985) shed by live animals and traces of animal dung (Brochier Reference Brochier1983; Karkanas Reference Karkanas2006) confirm that some caves and rock-shelters sheltered livestock, as was common in the recent past, alternative uses in antiquity evidently included burial, other ritual activities, human habitation and storage (e.g. Sampson Reference Sampson1992; Tomkins Reference Tomkins2009). Because of their association with limestone, caves and rock-shelters are also frequent in locations marginal to modern agriculture, again favouring attribution to mobile herders. As with the high-altitude sites discussed above, however, recent local land use may suggest otherwise. For example, Skoteini cave, in a gorge at c. 450 m asl on the island of Euboea, has been linked with seasonally mobile later Neolithic herding (Sampson Reference Sampson1992, 99–101), but within living memory sheltered over winter small household herds from the nearby sedentary, mixed agro-pastoral village of Tharounia (AFN).

Turning to scarce portable material culture, seasonal displacements provide evident incentives for mobile herders to avoid large, heavy and fragile items (Rutter Reference Rutter, Wiersma and Voutsaki2017, 20), but recent Sarakatsani pastoralists carried (admittedly on pack-animals unavailable in the Neolithic) looms, bulky textiles and some ceramics (AFN). Conversely, sedentary agro-pastoral farmers may invest in organic containers rather than fine ceramics and indeed fine wares of Neolithic mixed farmers exhibit technological, formal and decorative features suggesting mimicry of wooden, basketry and leather prototypes (e.g. Childe Reference Childe1957, 60; Tomkins Reference Tomkins, Mee and Renard2007, 186–7).

In sum, apparently sparse, short-lived and flimsy habitation, predominance of cave/rock-shelter over open-air sites, and poverty of durable portable material culture are plausible, but not necessary, archaeological correlates of mobile herding and specialised pastoralism (Types 3–5), which in turn are by no means the only likely rationales for the former. Conversely, these various ambiguous proxies are also compatible with herding of Types 1 and 2.

4. Regional distribution of styles of portable material culture

While some scholars have associated mobile herders with impoverished portable material culture, others have implicated them in the long-distance dispersal of fine Middle Neolithic (sixth millennium BCE) ‘Urfirnis’ ceramics in the north-east Peloponnese (Jacobsen Reference Jacobsen and Rice1984) and, more tentatively, of Final Bronze–Early Iron Age ‘matt painted’ ceramics across Albania and northern Greece (Kilian Reference Kilian1972, 122; cf. also Heurtley Reference Heurtley1926–7, 187) and of bronze personal items across Early Iron Age northern Greece and the southern Balkans (Kilian Reference Kilian1973). It is left open whether herders would have carried these artefacts for their own use, presumably as expressions of cultural identity, or as intermediaries in regional exchange, in the manner of recent Vlach muleteers. Nor is it clear why, in the latter case, long-distance exchange should be attributed to mobile herders rather than, say, chains of marriage alliances between sedentary communities of mixed farmers. In principle, therefore, far-flung material culture styles may be as compatible with herding of Types 1–2 as of Types 3–5.

5. (Palaeo)ecological indicators of anthropogenic impact on landscape

A critical issue not yet considered is the availability of pasture in antiquity. A reduction in grazing pressure over recent decades, thanks variously to lower livestock numbers and/or increased stall feeding, has been accompanied by widespread and readily visible expansion of tree cover. In the mountains of northern Greece, the resulting extension upwards of the timber-line (Fig. 2) makes clear that the present-day high summer pastures are at least partly an anthropogenic artefact of grazing by domestic livestock, an inference consistent with their poverty in endemic plant species otherwise well represented in the country’s northern highlands (Quezel Reference Quezel1967, 172) and in high-altitude pastures of southern Greece (Quezel Reference Quezel1964, 323). Trees are also expanding in many parts of the lowlands where grazing has declined. In antiquity, therefore, the availability of pasture, especially for sheep that are less suited than cattle, goats or pigs to woodland, will have depended at both high and low altitudes on anthropogenic clearance of arboreal vegetation and suppression of its recovery by livestock. In an early review of palynological evidence for the impact of farmers on the landscape of Greece, Bottema noted the difficulties of interpretation posed, on the one hand, by the altitudinal and hence climatic differences between the catchments of coring sites and, on the other, by the pre-Neolithic presence of some of the plant taxa that served as anthropogenic indicators of farmers in temperate Europe (Bottema Reference Bottema1982). He concluded that the impact of farmers first became clear from the latter part of the Bronze Age or second millennium BCE (Bottema Reference Bottema1982), not least thanks to the first consistent appearance in central and northern mainland cores of walnut ( Juglans regia) and chestnut (Castanea sativa), likely favoured for their edible nuts, and plane (Platanus orientalis), perhaps spreading due to changes in watercourses triggered by human clearance. Moreover, comparison of north Greek palynological records with present-day surface samples suggested that very extensive deforestation may not have occurred before the early modern period (Bottema Reference Bottema1974, 151, fig. 17, 166; pace Willis Reference Willis1994, 114). While the difficulty in detecting the impact of Neolithic farmers was doubtless due in part to the extraction of pollen cores from lake basins with regional catchments (Willis and Bennett Reference Willis and Bennett1994), it also suggests that early cultivation and herding were modest in scale. More recent palynological research, using a range of other proxies in addition to pollen, has detected possible traces of early farming from the Early Neolithic (seventh millennium BCE) onwards in lowland northern Greece (Ganz et al. Reference Ganz, van Vugt, Gobet, Morales-Molino, Giagkoulis, Ogi, Hächler, Schaad, Zander, Lloren, Dubois, Grosjean, Vogel, Kotsakis, Bogaard, Hafner and Tinner2024; Morales-Molino et al. Reference Morales-Molino, van Vugt, van Leeuwen, Gobet, Schwörer, Ganz, Giagkoulis, Brugger, Bogaard, Hafner, Kotsakis, Lotter and Tinner2024; van Vugt et al. Reference van Vugt, Gobet, Morales-Molino, Knetge, Lotter, Vogel, Ganz, Grosjean, Giagkoulis, Bogaard, Kotsakis, Hafner and Tinner2025) that essentially confirm the low impact and hence small scale of early cultivation and herding. A heavier impact on the landscape, suggesting herding on a scale incompatible with our Types 1 and perhaps 2, is again not seen in this region prior to the Bronze Age (third–second millennia BCE) or even historical times (first millennium BCE onwards). In semi-arid eastern Crete, human impact in the form of both cultivation (e.g. olive and cereal pollen) and grazing (coprophilous spores) is evident throughout a core at Palaikastro spanning the mid fifth to mid third millennia BCE (Final Neolithic–Early Bronze Age), with peak intensity in the early third millennium (Cañellas-Bolta et al. Reference Cañellas-Bolta, Riera-Mora, Orengo, Livarda and Knappett2018). The core was taken from a coastal wetland, however, that was probably a very attractive seasonal grazing resource, and so may reflect herding in its immediate vicinity. In the Final Neolithic, when dispersed settlements on high ground are well documented in the wider region, a clear preference for locations with access to fertile soil suggests reliance on mixed agropastoral farming compatible with small-scale herding of Types 1–2 (Kalantzopoulou Reference Kalantzopoulou2019, 180).

Fig. 2

(a) Sheep grazing at c. 1700 m asl on the summit of Profitis Ilias, north-west of Fourka village in the Pindos Mountains (August 1983); (b) their Sarakatsani shepherd described how, with a reduction in sheep numbers since the 1940s, woodland expanding up the wet west- and north-facing slopes had encroached on this area of pasture; (c) closer inspection of the woodland revealed isolated elderly conifers smothered by dense young beech trees, confirming his account.

Fig. 2 Long description

The series consists of three vertically stacked photographs.

Panel a, at the top, shows a close-up view of a flock of white sheep grazing on a grassy, uneven mountain summit. A shepherd dog stands among them. In the background, hazy blue mountain peaks are visible under a clear sky.

Panel b, in the middle, provides a wide-angle view of the landscape. A large, dense forest of dark green trees covers the upper slopes of the mountain. Below the tree line, a vast area of light green and brown pasture slopes downward. A small, white cluster of sheep is visible in the distance on the lower left portion of the pasture.

Panel c, at the bottom, is a vertical-oriented close-up of the forest interior. A tall, thick-trunked conifer with dark, textured bark and sparse upper needles is surrounded and partially obscured by the bright green, dense foliage of younger beech trees, illustrating the encroachment of deciduous woodland into older coniferous stands.

A recently published core from the small high-altitude lake of Drakolimni in the Pindos Mountains sheds the first direct light on likely anthropogenic impact around the timberline, in this case at c. 2100 m asl (Hersberger et al. Reference Hersberger, Ganz, Kiener, Gobet, van Vugt, Giagkoulis, Breu, Heiri, Vogel, Zahajská, Bogaard, Hafner, Kotsakis and Tinner2025). From the base of the core at c. 4250 BCE, coprophilous fungi indicate nearby grazing by herbivores, but these may well have been wild chamois (Rupicapra rupicapra) or deer (Cervus elaphus or Capreolus capreolus), especially given the attraction of the lake as a source of water. From the second millennium BCE (later Bronze Age), an apparent lowering of the timberline and peaks in microscopic charcoal and dung-related fungal spores suggest increased grazing and burning (presumably for pasture) close to the coring site, while fluctuations in the timberline and more intense traces of human impact are seen from 600 BCE to the present. The authors of this study interpret the later Bronze Age grazing indications in terms of seasonal transhumance, leaving open whether this was in the context of specialised pastoralism or mixed agro-pastoral farming (Hersberger et al. Reference Hersberger, Ganz, Kiener, Gobet, van Vugt, Giagkoulis, Breu, Heiri, Vogel, Zahajská, Bogaard, Hafner, Kotsakis and Tinner2025, 15). Some 20 km to the south in the Zagori area, known later Bronze Age burials at c. 900–1100 m asl lay claim (like the modern agropastoral villages) to small islands of cultivable land in a landscape predominantly better suited to rough pasture (Halstead Reference Halstead1990; Vasileiou Reference Vasileiou2019–20; Reference Vasileiou and Karantzali2021). Ten km to the east near Samarina and at around 1500 m asl, well within the upper limits of cereal cultivation in living memory (Halstead Reference Halstead1990, 64), Bronze Age traces of human activity include ¹⁴C-dated pits apparently from the burning of tree stumps (Biagi et al. Reference Biagi, Starnini, Efstratiou, Nisbet, Hughes and Woodward2023), more plausibly linked to clearance for cultivation (cf. Halstead Reference Halstead2018, 45–6) than for extension of pasture. Given the likely presence nearby of communities practising cultivation, Bronze Age seasonal grazing at Drakolimni is most parsimoniously related to herding of Type 3 or even Type 2.

6. Skeletal remains of domestic animals as evidence of their mobility and diet

The skeletal remains of domestic animals have, until very recently, played a minor role in arguments over mobile herding in ancient Greece. Bones of domestic ruminants (cattle and, especially, sheep and goats) have been cited in favour of seasonal mobility (e.g. Jarman, Bailey and Jarman Reference Jarman, Bailey and Jarman1982, 150–2; Vokotopoulou Reference Vokotopoulou1986, 340), but recent management of these species has frequently been stationary. Dental evidence for age at death has also been used to estimate the season(s) in which livestock were present at individual sites and so to question wholesale seasonal mobility (e.g. Becker Reference Becker1999; Halstead Reference Halstead, Bailey, Whittle and Cummings2005). Such estimates, however, assume early stock-rearers did not manipulate the seasonality of breeding and, while apparently the case for the north Balkan Neolithic (Balasse et al. Reference Balasse, Renault-Fabregon, Gandois, Fiorillo, Gorczyk, Bacvarov and Ivanova2020), this has been questioned for the west Mediterranean Neolithic (Tornero et al. Reference Tornero, Balasse, Bréhard, Carrère, Fiorillo, Guilaine, Vigne and Manen2020), and no evidence is yet available for Greece.

The different livestock species have complementary feeding preferences, with sheep best suited to open grassland. On this basis, the tendency for sheep to dominate faunal assemblages in a more or less wooded earlier Neolithic landscape has been interpreted in terms of the penning of livestock primarily on land cleared for crops, with the more balanced composition of later Neolithic assemblages perhaps indicating expansion of herding onto surrounding woodland, scrub and wetland to which goats, cattle and pigs are variously better suited (Halstead Reference Halstead, Hodder, Isaac and Hammond1981b, 322–4). Of course, early farmers may not have tied livestock to their preferred habitats, but this argument has subsequently received independent support from stable-isotope data.

Recent use of stable carbon (δ ¹³C), nitrogen (δ ¹⁵N) and oxygen (δ ¹⁸O) isotope values of domestic animal bones and teeth, as proxies for their average long-term diet (δ ¹³C and δ ¹⁵N of bone collagen) and for seasonal variation in their diet and the temperature of ingested water (δ ¹³C and δ ¹⁸O of incremental tooth enamel samples), marks a significant advance on hitherto limited appeals to faunal evidence. First, δ ¹³C and δ ¹⁵N values of domestic animal bones and crop grains from earlier Neolithic (seventh–sixth millennia BCE) Knossos on Crete and perhaps Kouphovouno in the Peloponnese suggest close integration of livestock with intensively cultivated (high δ ¹⁵N) garden plots around the settlement (Halstead and Isaakidou Reference Halstead, Isaakidou, Gron, Sørensen and Rowley-Conwy2020, 89–90; Diffey and Bogaard Reference Diffey, Bogaard, Cavanagh, Mee and Renard2025; Vaiglova et al. Reference Vaiglova, Gardeisen, Buckley, Cavanagh, Renard, Lee-Thorp and Bogaard2020) and thus the predominance of Type 1 or 2 herding with minimal or modest mobility. Secondly, high δ¹³C values in cattle bone collagen and incremental tooth enamel samples from Late Neolithic (sixth millennium BCE) Makriyalos I in Macedonia reveal that some of these large domesticates seasonally (and others perhaps year-round) grazed saline rough pasture on the coast, at least a few kilometres from the site (Vaiglova et al. Reference Vaiglova, Halstead, Pappa, Triantaphyllou, Valamoti, Evans, Fraser, Karkanas, Kay, Lee-Thorp and Bogaard2018, 16, fig. 11), indicating that Type 2 herding beyond the cultivated landscape is also of great antiquity. Thirdly, incremental tooth enamel samples exhibiting marked oscillations between low (cool/wet) and high (hot/dry) δ ¹⁸O values, such as are expected of livestock present year-round at the same altitude (Pederzani and Britton Reference Pederzani and Britton2019; also Chazin, Gordon and Knudson Reference Chazin, Gordon and Knudson2019), are seen in sheep and cattle at Late Neolithic Makriyalos I (Vaiglova et al. Reference Vaiglova, Halstead, Pappa, Triantaphyllou, Valamoti, Evans, Fraser, Karkanas, Kay, Lee-Thorp and Bogaard2018, 15–16, figs 10–11), in goats and most sheep at Archaic–Classical (seventh–fourth centuries BCE) Argilos in Macedonia (Alagich et al. Reference Alagich, Gkotsinas, Perreault, Bonias, Dotsika, Adler and Smith2025) and Middle–Late Neolithic (sixth–fifth millennia BCE) Kouphovouno (Vaiglova et al. Reference Vaiglova, Gardeisen, Buckley, Cavanagh, Renard, Lee-Thorp and Bogaard2020, 12, table 6, 13, fig. 9), in goats at Late Bronze Age (second millennium BCE) Knossos (Isaakidou et al. Reference Isaakidou, Styring, Halstead, Nitsch, Stroud, Roux, Lee-Thorp and Bogaard2019, 51, fig. 8), and in sheep at Classical–Hellenistic (fifth–third centuries BCE) Kastro Kallitheas (Bishop et al. Reference Bishop, Garvie-Lok, Haagsma, MacKinnon and Karapanou2020), Pherae, Nea Alos (New Halos) and Magoula Plataniotiki in south-east Thessaly (Filioglou et al. Reference Filioglou, Valenzuela-Lamas, Patterson, Pena, Presslee, Timsic, Delgado Huertas, Prummel and Çakirlar2024). Conversely, for sheep at Late Bronze Age Knossos, dampened δ ¹⁸O curves are consistent with displacement in summer to cool, high-altitude pasture. Tooth enamel δ ¹³C values approximately track the δ ¹⁸O curve in most of the preceding cases, consistent with more marked seasonal variation in diet among stationary than vertically mobile animals, but a few exceptions may reflect consumption of stored fodder out of season (δ ¹³C) or of water from seasonally mixed sources such as lakes or springs (δ ¹⁸O) (e.g. Isaakidou et al. Reference Isaakidou, Styring, Halstead, Nitsch, Stroud, Roux, Lee-Thorp and Bogaard2019, 50–2; Pederzani and Britton Reference Pederzani and Britton2019, 89). Incremental δ ¹⁸O data from Greece are as yet sparse for secure generalisation, but long-distance vertical movement of livestock is arguably documented, if at all, only for sheep from Late Bronze Age Knossos. On the other hand, incremental analysis of strontium isotope ratios (⁸⁷Sr/⁸⁶Sr) in teeth, partly reflecting the geological setting in which animals were reared, suggests that a few domesticates consumed at Late Neolithic Makriyalos I, Late Bronze Age Knossos and the aforementioned Classical–Hellenistic sites in south-east Thessaly spent at least part of their early life non-locally. The geology of Greece is complex and bio-available strontium (the critical variable) barely mapped (e.g. Isaakidou et al. Reference Isaakidou, Styring, Halstead, Nitsch, Stroud, Roux, Lee-Thorp and Bogaard2019; Frank et al. Reference Frank, Frei, Moutafi, Voutsaki, Orgeolet, Kristiansen and Frei2021; Vaiglova et al. Reference Vaiglova, Halstead, Pappa, Triantaphyllou, Valamoti, Evans, Fraser, Karkanas, Kay, Lee-Thorp and Bogaard2018; Whelton et al. Reference Whelton, Lewis, Halstead, Isaakidou, Triantaphyllou, Tzevelekidi, Kotsakis and Evershed2018), however, so ⁸⁷Sr/⁸⁶Sr data have as yet shed limited light on the horizontal distances traversed by livestock, and at least some of the documented displacements may have involved, for example, a one-off change of ownership rather than seasonal movement between pastures.

7. Socioeconomic preconditions of specialised pastoralism

In diametric opposition to the belief that timeless characteristics of Mediterranean climate and topography imposed or strongly encouraged vertical pastoral mobility are arguments that specialised pastoralism depended on three historically contingent socioeconomic preconditions: a lack of obstructions to free passage by inter-polity borders; the organisation of lowland winter pasture in large blocks subject to common patterns of cropping/fallowing; and the existence of markets in which pastoral produce could be exchanged for agricultural staples.

The difficulties that inter-polity borders pose for long-distance movement of herds were highlighted in the last two centuries by the fragmentation of the Ottoman Empire into much smaller nation states (e.g. Wace and Thompson Reference Wace and Thompson1914, 11–12; Karamanes Reference Karamanes2011, 150). As noted above, written evidence for Classical–Hellenistic treaties regulating cross-border grazing is compatible with fairly short-distance pastoral mobility of Types 2 or 3, while polities sufficiently extensive and cohesive to facilitate mobility of Types 4 or 5 first appeared in Greece, with the possible exceptions of Final Palatial Knossos in central Crete (Bennet Reference Bennet, Duhoux and Davies2011, 148–51) and of the eighth–fourth-century BCE southern Peloponnese following the Spartan conquest of Messenia (Howe Reference Howe2008), from the fourth century BCE onwards.

The importance of lowland land-tenure and land-use (regular fallowing) regimes that made winter pasture available in large blocks was likewise highlighted over the last two centuries by the breaking up of monastic and secular agricultural estates, especially in the central and northern mainland (e.g. Karavidas Reference Karavidas1931; Vergopoulos Reference Vergopoulos1975; Nitsiakos Reference Nitsiakos1986–7). Lowland land use of this type may have existed in southern and central Greece, patchily in the context of extensive cultivation by Late Bronze Age Mycenaean palaces (Killen Reference Killen1994–5; Reference Killen1998; Halstead Reference Halstead1999) and perhaps more widely from the later Hellenistic–early Roman period when widespread decline of rural ‘farmsteads’ in regional surface surveys (Snodgrass Reference Snodgrass, Murray and Price1990) is partly matched by the emergence of large rural estates (Alcock Reference Alcock1989, 107–8 and fig. 5; Stewart Reference Stewart, Rizakis and Lepenioti2010, 220–1).

As to markets for the exchange of pastoral for agricultural produce, there is a variety of anecdotal literary evidence for the sale of wool, cheese and meat in the market of Classical Athens (e.g. Hodkinson Reference Hodkinson1992; Chandezon Reference Chandezon, Wilkins and Nadeau2015), at a time when the city’s population was unusually large with a spending power subsidised by the state (Hodkinson Reference Hodkinson1992, 56). Even this market, however, was apparently provisioned by herds much smaller than those of recent Type 4–5 pastoralists, while other Greek cities of this era were substantially smaller (e.g. Bintliff Reference Bintliff, Alston and van Nijf2008), and the apparently greater size of some Hellenistic and Roman cities (e.g. Alcock Reference Alcock1993) is also debated (Reinders and Prummel Reference Reinders and Prummel1998; Bintliff Reference Bintliff1996). The extent to which pastoralists are necessarily dependent on markets (Bates and Lees Reference Bates and Lees1977) has been disputed in cases where they are integral members of the same social group as the farmers from whom they obtain agricultural staples, as suggested for the pre-Inca Andean highlands (Parsons, Hastings and Matos M. Reference Parsons, Hastings and Matos1997) and early urban Mesopotamia (Porter Reference Porter2012). The same can be argued for the named ‘herders’ in Mycenaean palatial records of wool flocks over which the central authority had at least some use rights (Killen Reference Killen1993; Halstead Reference Halstead, Voutsaki and Killen2001). This was not the basis, however, on which recent Type 4 or 5 (or, for the most part, Type 3) herders in Greece acquired essential plant staples. Admittedly, pastoralists in Greece sometimes procured grain directly from the producer, rather than in the market, particularly so Vlach (Type 4) herders who used their horses on the threshing floors of lowland farmers in return for a fraction of the threshed and winnowed crop, but exchange in urban markets provided access to a much wider pool of cultivators and so was less vulnerable to local harvest failures.

Despite uncertainty regarding the date from which open borders, consolidated fallow blocks and markets were each sufficiently developed in Greece, it is unlikely that all three of these suggested sociopolitical preconditions for Type 4–5 specialised pastoralism were met before the Hellenistic or Roman periods at the earliest. The requirements of open borders and consolidated fallow grazing were probably met in the territory of Final Palatial Knossos on Crete, if not also in those of Mycenaean polities on the southern mainland, but the ‘herders’ who might have taken advantage of these opportunities were not independent pastoralists.

8. Interim synthesis: non-local mobile herding in ancient Greece

Previous claims of non-local mobile herding in ancient Greece have largely rested either on remote and ambiguous proxies (sparse habitation traces; impoverished portable material culture; long-range stylistic similarities) or on highly circumstantial and ahistorical arguments of environmental necessity/opportunity, in both cases inspired by recent mobile herding (Types 3–5), while ignoring alternative, more stationary (and, seemingly, less interesting) forms of animal husbandry (Types 1–2). Critical evaluation of these various claims has suggested that many are compatible with herding of our Types 1–2, that is with scales and forms of animal husbandry that have probably been a continuous feature of the Greek landscape since the arrival of domestic animals and for which terms such as ‘pastoralism’, ‘specialised’ or ‘mobile’ would be misleading. Even the most persuasive indications of non-local livestock movements (δ¹³C hints of saltmarsh grazing at Late Neolithic Makriyalos I; coprophilous fungi and microcharcoal at Drakolimni in the high Pindos; dampened seasonal δ¹⁸O curves in sheep teeth at Late Bronze Age Knossos on Crete, suggesting upland grazing; and treaties regulating grazing between Classical–Hellenistic cities) are more-or-less compatible with herding of Type 2 rather than Types 3–5.

Of course, absence of evidence is not evidence of absence, especially given the lowland, urban, sedentary and agricultural biases of ancient written sources and the limitations of available archaeological proxies (Forbes Reference Forbes1995). There are good grounds for optimism, however, that a combination of archaeological and palaeoecological investigation, inspired not least by pioneering work in mountain regions of the west Mediterranean (e.g. Ejarque et al. Reference Ejarque, Miras, Riera, Palet and Orengo2010; Orengo et al. Reference Orengo, Palet, Ejarque, Miras and Riera2014; Moreno et al. Reference Moreno, Cevasco, Pescini, Gabellieri, Álvarez, Gomez-Mediavilla and López-Estébanez2019; van Vugt et al. Reference van Vugt, Garcés-Pastor, Gobet, Brechbühl, Knetge, Lammers, Stengele, Alsos, Tinner and Schwörer2022; Walsh et al. Reference Walsh, Burri, Knockaert, Magniez, Dulias, Pike, Orton, Giguet‑Covex and Mocci2023), can significantly advance understanding of the history of high-altitude land use – including mobile herding – in Greece. First, applications of both pedestrian and remote survey methods (Orengo and Knappett Reference Orengo and Knappett2018; Kalantzopoulou Reference Kalantzopoulou2019; Papadatos and Kalantzopoulou Reference Papadatos, Kalantzopoulou, Driessen and van Wijngaarden2022; Moudopoulos-Athanasiou Reference Moudopoulos-Athanasiou2022; Biagi et al. Reference Biagi, Starnini, Efstratiou, Nisbet, Hughes and Woodward2023; Apostolou et al. Reference Apostolou, Garcia-Molsosa, Venieri, Georgiadis, Mayoral and Orengo2024) at high altitudes and in rocky terrain are revealing widespread traces of past human activity, the location or morphology (e.g. Chang and Tourtellotte Reference Chang and Tourtellotte1993; Orengo and Knappett Reference Orengo and Knappett2018, 484–5; Kalantzopoulou Reference Kalantzopoulou2019, 180) of which, especially if informed by ethnographic/oral-historical accounts of pre-mechanised land use, may favour association with cultivation or herding. Secondly, macroscopic and stable-isotope analysis of on-site plant and faunal remains may determine whether cultivation was intensive or extensive and herding stationary or seasonally mobile, while off-site palaeoecological data may clarify the scale of land use. Granted favourable preservation conditions, therefore, as well as effective data recovery and analysis, it is realistic to anticipate increasingly clear distinction between herding of Types 1, 2 and 3–5 and thus an understanding of the history of medium- to long-distance seasonal mobility with livestock and also of its impact on the development of some characteristic present-day plant communities. These methodological tools will not distinguish, however, between the seasonal camps (as opposed to permanent agropastoral settlements) of Type 3 herders and the seasonal sites of their Type 4–5 counterparts.

The antiquity of specialised pastoralist communities of Types 4–5 is of interest, not only as a significant element of the cultural heritage of the Vlachs and Sarakatsani, but also more generally for understanding the historical conditions that gave rise to these very distinctive ways of life. Given the extent to which the Vlachs and Sarakatsani, and to a less striking degree Type 3 groups such as the Kopatsaraioi and Valtetsiotes, maintained distinctive collective identities over considerable distances of both space and time, one way forward may be to revisit the relationship between specialised pastoralism and material culture in the broadest sense, including livestock and animal products as well as more conventional craft items.