Human-derived landscape changes on the northern Etruria coast (western Italy) between Roman times and the late Middle Ages

Roman Period (between the 3rd and the end of the 1st centuries bc)

For this historical phase, the charcoal data concern the site of Populonia, and 914 samples were analysed (Figure 2): the 337 charcoal samples dating to the 3rd century bc indicate that evergreen Quercus (40.9%) was the most common taxon at the time, together with A. unedo (22.8%) and Erica (11.3%). None of the deciduous taxa exceed 2%. Laurus nobilis (1.8%) and Abies alba (1.2%) were also found.

The first half of the 2nd century bc is represented by 155 charcoal samples: evergreen Quercus is still the dominant taxon (61.9%), while among the shrubs, the most common taxon is Rhamnus/Phillyrea (12.9%). Of the deciduous taxa, only Ostrya carpinifolia is recorded (<1%). The second half of the 2nd century bc yielded 51 samples, confirming the presence of the same taxa, with the predominance of evergreen Quercus.

The period from 100 to 50 bc yielded 176 samples. Sclerophyllous evergreens are always dominant in this period: evergreen Quercus accounts for 32.4%, while shrubs reach as high as 37.6%, with A. unedo being dominant (27.3%). In this interval of time, deciduous taxa as a whole reach 10.3%: of the latter, Ostrya/Carpinus is the best represented (5.7%), while deciduous Quercus slightly exceeds 2%. Olea europaea is below 1%. In the latter half of the 1st century bc, 195 charcoal samples were analysed. Shrub taxa broadly dominate (63.7%), with A. unedo accounting for 23.1% and Erica 18.5%, while evergreen Quercus accounts for 31.3%. Deciduous taxa are poorly represented, with only O. carpinifolia (2.6%) and deciduous Quercus (1.5%).

Late Roman Period (between the 2nd and mid-5th centuries ad)

For this period, 857 charcoal samples from the sites of Donoratico and Alberese were analysed (Figure 2). The Donoratico data fall between the Roman and Late Roman Period. The archaeological dating is not precise, with dates spanning the period between the 2nd and the end of the 3rd centuries ad. The 529 samples analysed show low values of evergreen Quercus (7.2%) and a more marked presence of evergreen sclerophyllous shrubs dominated by Erica (49%). The only deciduous trees recorded are Quercus (2.3%) and Ulmus (<1%). Pinus halepensis-pinea (1.3%) is also present. O. europaea does not exceed 5%, while Juglans regia and Vitis vinifera are both <1%. By contrast, the Alberese data span the 4th and mid-5th centuries ad. The 328 charcoal samples analysed show a completely different situation: deciduous taxa exceed 40% overall, including Cornus (19.8%) and deciduous Quercus (18.6%). Evergreen vegetation chiefly consists of Viburnum (21.6%), while evergreen Quercus and Erica are both below 2% and O. europaea <1%.

Early Middle Ages (10th century ad)

For this historical phase, charcoal data (386 samples) come from the Donoratico site (Figure 2). They indicate the strong presence of Erica (31.1%), followed by A. unedo (14%). Evergreen Quercus accounts for 5.7%. Deciduous taxa total 31.7%, with deciduous Quercus (20.7%) dominating. O. europaea is recorded (<1%).

Late Middle Ages (between the 11th and 13th centuries ad)

For this historical phase, the charcoal data concern the site of Populonia, and overall 299 samples were analysed (Figure 2). The first sequence of data concerns a settlement phase dating between the 11th and 12th centuries ad. The 248 samples analysed indicate the strong presence of evergreen Quercus (30.6%), while shrubs account for 12.1% overall, with Erica reaching 7.3%. The deciduous taxa account for 26.5%, with deciduous Quercus dominant (17.3%); other deciduous tree taxa, such as Ulmus, Fraxinus ornus and O. carpinifolia are also identified. O. europaea reaches appreciable values for the first time (9.7%), while Castanea sativa first appears in the anthracological record, slightly exceeding 1%. The 13th-century data concern sporadic frequentation of the ruins of the Roman city. The few samples analysed (51) show the presence of evergreens: evergreen Quercus, Erica and A. unedo are all found with at least 15%. The deciduous taxa, including Quercus (11.8%), reach 23.2% overall. Finally, it is worth noting that O. europaea exceeds 20% and C. sativa reaches 2%.

The landscape of the Roman Period

This part of ancient Etruria became Roman in the 3rd century bc; the main economic activity distinguishing the area throughout the previous Etruscan phase was that of mining and metal-working which began in the 8th century bc, peaked under Romanisation and then momentarily stopped in the 1st century ad. Several archaeologists have suggested that this period of intense mining activity caused major deforestation throughout the coastal area (Corretti, 2009; Hughes, 2005). Previous spot archaeobotanical studies suggested that cutting activities during the Etruscan Period heavily altered the natural state of the pre-existing woodlands (Sadori et al., 2010b).

The samples from the settlement of Populonia date to the 3rd, 2nd and 1st centuries bc, while those from the hilltop of Donoratico are dated between the 2nd and the end of the 3rd centuries ad. Between the 3rd and 1st centuries bc, the anthracological record shows the dominant evergreen Quercus, which in this area essentially consists in Q. ilex, with several sclerophyllous macchia shrubs; deciduous taxa are very poorly represented (Figure 3). From the 2nd century ad to the end of the 3rd century ad, evergreen macchia shrubs were abundant (c. 84%), while Q. ilex stood at c. 8%. Deciduous taxa compared with the previous phase are even less represented (Figure 3).

Overall, during these 600 years, Q. ilex forest initially seems to dominate the landscape. The forest then seems to contract appreciably, indicating a change to more open macchia formations in which Erica is dominant. Interestingly, the pollen sequence of Lake Accesa showed the same dynamic (Figure 3). Ericaceae start to increase at c. 2000 cal. yr BP (mid-1st century bc) and culminate at c. 1750 cal. yr BP (early 3th century ad), while Q. ilex declined. The expansion of the macchia might have been triggered by a widespread climatic event; indeed, the period in question is characterised by dry conditions as shown by low lake levels both at Lake Accesa and Lake Fucino, recorded also on a wider scale in the central Mediterranean and Anatolian lakes, corresponding to the warm bond interval at c. 2700–1800 cal. yr BP (Dermody et al., 2012).

Between the 3rd century bc and 3rd century ad, evergreen shrublands increased in several regions: from c. 2100 cal. yr BP (mid-2nd century bc), pollen data indicate the spread of macchia in south-eastern Italy, including the Gargano Peninsula (Caroli and Caldara, 2006) and Salento (Di Rita and Magri, 2009), as well as in central Sicily (Sadori et al., 2013) and in western and south-eastern coastal Sicily (Calò et al., 2012; Noti et al., 2009; Tinner et al., 2009). On the other hand, pollen data also indicate for the same period a substantial stability in the forest cover, as in the case of the Versilia Plain on the Tyrrhenian coast 100 km further north of our study area (Bellini et al., 2009; Colombaroli et al., 2007); a widespread deciduous oak forest was present along the lower Tyrrhenian coast in Latium (Goiran et al., 2014; Sadori et al., 2014) and around Naples (Allevato et al., 2010; Russo Ermolli et al., 2014). Also outside Italy, some areas like the southern Balkan Peninsula on the coasts of Acarnania (Jahns, 2004) and Attiki (Kouli, 2012) were unaffected by any forest decline. The evidence of different dynamics in a uniform climatic phase could suggest that changes in the vegetation cover may have been induced by non-climatic factors.

Closer analysis of the historical context may be able to supply a different interpretation. Archaeological evidence concerning the Romanisation of coastal northern Etruria indicates a period of intense human activity (Cambi and Botarelli, 2004; Cambi and Acconcia, 2011; Dallai, 2000), thanks to the number of archaeological sites discovered (Figure 3): from the 2nd century bc onwards, there emerges a complex network of settlements consisting of villas and farms which increased threefold compared with the previous century, reaching its maximum expansion between the 1st century bc and 1st century ad (Botarelli, 2004; Dallai, 2003b; Vaccaro, 2008). In our opinion, our data essentially reflect growing human impact along the northern Etrurian coast: this major presence in the area coincides with the minimum frequency of forest fires testified by microcharcoal in the record at Lake Accesa throughout the whole study period (Figure 3). This may be ascribed to greater area control and land management whereby accumulation of dead biomass is prevented, thereby reducing the fire risk (Tinner et al., 1998).

Between the 2nd and 3rd centuries ad, the charcoal data indicate a forest landscape around Donoratico in which macchia vegetation prevails, while Q. ilex drops to fairly low values (Figure 3). Both these data coincide with what is indicated in the pollen diagram (peak of Ericaceae associated with a minimum of Q. ilex-type; Figure 3). Deciduous Quercus and deciduous trees generally reach the minimum in the anthracological diagram, while they appear more stable in the Lake Accesa pollen diagram which may be affected by the fact that the lake is further inland and higher above sea level. According to the census of the archaeological sites, in the above two centuries, there is a decline in the number of settlements, with fewer than half the number active in the previous century (Figure 3). However, for this period, the archaeologists discuss of a new form of land management in Etruria, called latifundium, which concentrates extensive landed property in the hands of a few landowners (Cambi, 1993; Citter, 1996; Vaccaro, 2008) who manage from their farmsteads extensive production based on cereals and livestock (Carandini, 1985; Regoli, 2002). Thus, the great development of macchia and the contraction of tree species can well be interpreted as the effect of overgrazing. The microscopic charcoal accumulation rate in the sediment of Lake Accesa indicates an increase in fires from the mid-1st century ad, which could be interpreted as evidence for the use of fire related to pasturage. It is worth noting that, on the basis of these data, it seems evident that during the time interval between the 2nd and 3rd centuries ad, there was greater impact on the forest cover, even compared with the Etruscan Period in which the area’s economy was based on mining and metal-working (Botarelli, 2004; Cambi and Acconcia, 2011; Cambi and Botarelli, 2004).

For the final part of the Roman Period, between the 4th and mid-5th centuries ad, the anthracological record comes from the river port of Alberese and shows deciduous Quercus as the dominant tree (c. 27%) with other deciduous taxa such as Cornus (Figure 3). The percentage of macchia shrubs halved compared with previous phase; of these, only Viburnum reached appreciable values, while Q. ilex continued to be scarce. The importance of deciduous vegetation which reached c. 60% overall is in agreement once again with the vegetation picture shown by the pollen sequences of Lake Accesa (Figure 3), which indicates after c. 1750 cal. yr BP (early 3rd century ad) the increase in deciduous Quercus until c. 1500 cal. yr BP (mid-5th century ad) when it reached its maximum percentage in the time span considered.

The expansion of deciduous species is also recorded in Etruria, 100 km further north, at Lake Massaciuccoli (Colombaroli et al., 2007) between the 3rd and 5th centuries ad. The deciduous forest cover also increased along more southern sites on Tyrrhenian Sea, such as in Latium (Sadori et al., 2010a) and central Sicily (Sadori et al., 2013). On Adriatic Sea, a similar situation has been recorded in the Salento Peninsula (Di Rita and Magri, 2009) and along the South Dalmatian coast (Jahns and Van Den Bogaard, 1998). The expansion of deciduous vegetation to the detriment of evergreens in the Mediterranean during the Holocene has often been explained climatically with humid peaks in the cooler climatic phases (Colombaroli et al., 2009). However, in this case, the spread of deciduous vegetation falls within a dry phase detected in various European and Mediterranean regions (Magny et al., 2013; Peyron et al., 2013), which corresponds locally to the lowest level at Lake Accesa in the last 2500 years (Magny et al., 2007). Therefore, a causal relationship with climate being behind the expansion of deciduous species detected by our charcoal data at the end of the Roman Age should be excluded. A possible cause is suggested by the trend in the number of settlements which continued to fall between the 4th and mid-5th centuries ad (Figure 3) because of progressive depopulation (Vaccaro, 2008). This abandonment could have triggered the forest cover change.

In the Mediterranean, the record of successional processes in progress in the abandoned fields during the second half of the previous century indicates the key role played by deciduous oaks (Quézel and Médail, 2003). The phenomenon of recent rural depopulation has led to the development of deciduous oaks in formerly cultivated lands. This process has been described in the northern areas of the Mediterranean basin, especially in Provence (Barbero et al., 1990; Tatoni et al., 2004), but this is also happening at lower latitudes such as in south-eastern Sicily (Di Pasquale et al., 2004). Deciduous oaks have in fact a greater competitive potential than evergreen oaks in the processes of spatial occupation in abandoned fields (Barbero et al., 1990; Di Pasquale and Garfì, 1998). The deciduous expansion phase which characterises the Late Roman Age may thus have been caused in Etruria, as in other regions of the Mediterranean, by rural depopulation at the end of the Roman Empire and, more generally, by the abandonment of areas hitherto cultivated. The appreciable presence of Cornus in the anthracological diagram is in this sense important because it is an endozoochorous species which characterises the intermediate phases of forest succession in abandoned areas (Quézel and Médail, 2003). This hypothesis may be useful to account for other periods as well, varying in settlement intensity and human impact from c. 5000 cal. yr BP.

Finally, it should be pointed out that the increase in frequency of fires recorded at Lake Accesa also shows in this case a negative relationship with the number of settlements; this could be interpreted as the effect of the increase in potentially combustible biomass caused by a lack of farm management practices being applied to agro-forestry spaces (Tinner et al., 1998).

The landscape of the Middle Ages

The hiatus between the 6th and 10th centuries ad is because of the absence of samples: Populonia shows no regular human frequentation from the end of the 1st century bc to the 11th century ad; the hilltop of Donoratico was abandoned between the mid-4th and 7th centuries ad and sufficient material for analysis is dated from the 9th century ad; the site of Alberese after the 5th century ad was abandoned and covered with alluvial deposit.

At the end of the Early Medieval Period, between the 9th and 10th centuries ad, in the vegetation at Donoratico, there was a considerable presence of macchia, while deciduous Quercus was the dominant arboreal taxon to the detriment of Q. ilex (Figure 3). The landscape around Donoratico continued to be dominated by deciduous oaks which characterised the previous centuries and which still remain the dominant trees.

As regards Late Medieval times, the Populonia data show a progressive reduction in the forest component and a shift towards a more open macchia environment (Figure 3). The stability of the forest which may be inferred from the Accesa data suggests a change in the vegetation on a local scale (Figure 3). The most interesting information in this last phase concerns olive (O. europaea) and chestnut (C. sativa).

Olive charcoal data are scarce from the Roman Period to the end of the Early Middle Ages and increase significantly only from the Late Middle Ages onwards (11th and 12th centuries ad), doubling their frequencies in the 13th century ad (Figure 3). Pollen data from Lake Accesa are yet again consistent with our findings and show a percentage constantly below 0.5% until 1000 cal. yr BP (mid-10th century ad), when they increase moderately to a maximum of 5% (Figure 3). On the whole, these data indicate that, before the Late Middle Ages, olive-growing did not belong to the landscape of northern coastal Etruria, but it constituted only small and scattered plots probably located close to farmsteads. This evidence conflicts with the traditional idea of archaeologists and historians who point to widespread olive cultivation in Etruria between the 3th century bc and 1st century ad during the Roman Republic and Early Empire (Barbieri, 2010; Brun, 2011; Carandini, 1985). On a broader spatial scale, the Italian pollen data seem to confirm a scant presence of olive in the Roman Period in many areas which are today given over to olive-growing, both in central Italy (Bellini et al., 2009; Colombaroli et al., 2007; Mercuri et al., 2002), southern Italy (Caroli and Caldara 2006; Di Rita and Magri, 2009; Russo Ermolli and Di Pasquale, 2002) and Sicily (Calò et al., 2012; Noti et al., 2009; Sadori et al., 2013; Tinner et al., 2009). Medieval historians also partly confirm our findings: for central Italy, in particular, the documents demonstrate that only from the end of the 14th century ad was there a major, definitive development of olive farming (Cortonesi, 2005). It has to be stressed that olive farming in central Italy became established precisely in the centuries of the ‘Little Ice Age’: this would confute the hypothesis posited by several researchers who tentatively explained the scant presence of olive farming in the Roman Period with the recurrence of cold climatic phases (Di Rita and Magri, 2009).

Chestnut shows a very similar trend to that of olive: it appears in the 11th century ad in very low percentages and again closely matches the data from Lake Accesa which show that chestnut begins to spread a little before, starting from c. 1000 cal. yr BP (mid-10th century ad). Both proxies yield an extremely weak signal, even if we know that in the immediate neighbourhood on the Metalliferous Hills, less than 100 km from our study area, chestnut was present at least from the 7th century ad, where it was essentially used for timber (Buonincontri et al., 2014; Di Pasquale et al., 2008). The use of timber has been clearly inferred by means of charcoal data in central and southern Italy at least until the Early Middle Ages (Allevato et al., 2012a; Di Pasquale et al., 2010).

Various pollen sequences up and down peninsular Italy document, from c. 1000 cal. yr BP onwards, a great expansion of chestnut (Branch and Marini, 2013; Mercuri et al., 2013; Noti et al., 2009; Sadori et al., 2013; Tinner et al., 1999; Vescovi et al., 2010), contradicting the theory of the spread of this plant between the Late Roman Age and the Early Middle Ages (Conedera et al., 2004; Rottoli, 2014). The extremely low value yielded by charcoal analysis in northern Etruria and the slight rise in chestnut pollen from Lake Accesa would constitute evidence, in this case, of a landscape in expansion, and probably dates the time during which chestnut becomes a food crop ‘globally’ (Quirós Castillo, 1998). As in the case of olive, the emergence of such landscapes is thus Late Medieval and may well be attributed to the set-up of a new economic organisation with the establishment of the feudal economic system (Wickham, 2005) and later to the establishment of the political system of the Communi (Cortonesi, 2005; Wickham, 1988), which led to the need to satisfy a rising demand for food and luxury products (Bosi et al., 2009).