4.7 Vegetation change and land uses in later prehistory and the historic period

Figure 22 shows the locations of pollen sites in Argyll and Bute that cover all or some of the period from c. 4000 cal BC to the present or close to it. The current data–set for this long period fragments when the times the pollen records span and their purposes are described. The sites on Mull at Gribun (Walker and Lowe 1987), Fhuaran and Torness (Walker and Lowe 1985), Loch Cill an Aonghais (Peglar 1980), Loch Lomond (Dickson et al. 1976), and Clashgour and Coire Seilich on Rannoch Moor (Bridge et al. 1990) are vegetation histories (see Section 4) rather than land use histories but the long records at Loch Cholla (Andrews et al. 1987b), Loch a'Bhogaidh (Edwards and Berridge 1994), Dubh Lochan (Stewart et al. 1984) and the four sites in and around Oban presented by Macklin et al. (2000) were intended to have human activities as their investigative focus.

Hale's pollen analyses at Glengorm on Mull (in Martlew and Ruggles 1996) were designed to throw light on inter–visibility from standing stones and rows, but the data were not dated or presented, only discussed, and the pollen record itself (Hale pers. comm.) is considered by this author not to relate in time to the monuments. The sites of Cairnbaan and Glasvaar in and near Kilmartin (Long 2002; Winterbottom and Long 2006) and at Torbhlaren (Tipping et al. 2011) were analysed to understand the landscape contexts of later prehistoric rock art sites. The first two are undated. Analyses at Dunadd (Miller and Ramsay 2001; Housley et al. 2010) commence in the later Iron Age and explore land use change around the fort in the early historic and medieval periods. Iona in the early historic period has unsurprisingly attracted much palynological interest (Bohncke in Barber 1981; Balaam in Reece 1981; Scaife and Dimbleby 1990; Tipping in McCormick 1993) but has generally disappointed. Small 'snapshots' in time were generated by Balaam, Bohncke and Tipping and although Scaife and Dimbleby (1990) argued that the sediments in Iona Loch are a complete Holocene record, this is very unlikely, and peat cutting is suggested here to have truncated the stratigraphy above the later prehistoric period. At the head of Loch Awe is a concentration of analyses that have recently specifically explored medieval and later land uses (Sansum 2004; Davies and Watson 2007).

Figure 22: Map of later prehistoric and historic period pollen records © Richard Tipping: Cairnbaan (Long 2002; Winterbottom and Long 2006); Clashgour (Bridge et al. 1990); Coire Seilich (Bridge et al. 1990); Corries, Loch Awe (Davies and Watson 2007); Dalness Chasm (Brazier et al. 1987); Dubh Lochan (Stewart et al. 1984); Dunadd (Miller and Ramsay 2001; Housley et al. 2010); Fhuaran, Mull (Walker and Lowe 1985); Glasvaar (Long 2002; Winterbottom and Long 2006); Glengorm (Hale in Martlew and Ruggles 1996); Gribun, Mull (Walker and Lowe 1987); Iona (Bohncke in Barber 1981; Balaam in Reece 1981; Scaife and Dimbleby 1990; Tipping in McCormick 1993); Loch a' Bhogaidh, Islay (Edwards and Berridge 1994); Loch a' Chrannag, Mull (Sugden 1999; Fyfe et al. 2013); Loch Cholla (Andrews et al. 1987); Loch Cill an Aonghais (Peglar in Birks 1980); Loch Lomond (Dickson et al. 1976); Lochan Taynish (Rymer 1974); Oban Davies 1997; Macklin et al. 2000); Sansum (2004): sites along Loch Awe at Cladich, Fernoch and Glen Nant; Torbhlaren (Tipping et al. 2011); Torness (Walker and Lowe 1985).

Macklin et al. (2000, 113) described "substantial deforestation" from c. 4300 cal BC at some sites, but in detail (Davies 1997) the losses are often seen to be of wetland trees and need not have been anthropogenic. 'Clearance' around Cnoc Philip at c. 2000 cal BC is from the expansion of ling heather (Calluna) and, probably, blanket peat (Davies 1997) and again need not have been anthropogenic, which perhaps challenges the suggestion of widespread synchroneity around Oban of "more settled farming practices" in the early Bronze Age (Macklin et al. 2000). The suggested correspondence between later prehistoric phases of farming with climatic change is exciting but rests on absences of cereal type pollen, which is always under–represented in pollen records.

4.7.1 Neolithic activity

Whitehouse et al. (2013) have recently argued for the beginning of the Irish Neolithic at c. 3750 cal BC. Around Coire Clachach in central Mull (Walker and Lowe 1985), Plantago lanceolata appears in the early Neolithic at c. 3850 cal BC, and some dryland trees like hazel and oak began to be lost. This pattern is described from Loch a'Bhogaidh on Islay from this date (Edwards and Berridge 1994) and from the Sorn Valley (Andrews in McCullagh 1989) from c. 3750 cal BC, though at the latter site by a single pollen grain only. At Gribun on the exposed west coast of Mull a plant community with wild grasses was established before the elm decline, further increasing from c. 2850 cal BC as the birch population fell (Walker and Lowe 1987). Losses in hazel and an expansion of grassland with P. lanceolata occurred prior to the elm decline around Iona Loch, and cereal type pollen is recorded from the elm decline (Scaife and Dimbleby 1990). Also on Iona, Bohncke (in Barber 1981) suggested Neolithic farming included cereal crops. Near Kilmartin, valley–floor woods at Torbhlaren were partly cleared from as early as c. 4300 cal BC (Tipping et al. 2011), with regeneration between c. 3200 and c. 2900 cal BC. Cereal cultivation was introduced at c. 3300 cal BC. Tipping et al. (2011) suggested that at c. 2900 cal BC oak woodland began to be conserved, or perhaps planted, and grown on for some 800 cal years until the woodland was abruptly felled at c. 2100 cal BC. This activity may have coincided with the creation of rock art on the valley floor. At Lochan Taynish, Rymer (1974) reported only limited woodland disturbance in the Neolithic, after c. 3400 cal BC. Close by at Loch Cill an Aonghais (Peglar 1980), whilst P. lanceolata appeared at the elm decline, grassland expanded only from c. 3400 cal BC. In the Sorn Valley, grassland with P. lanceolata was established around 3100 cal BC (Andrews in McCullagh 1991).

Long's (2002; Winterbottom and Long 2006) reconstructions of the local landscapes around the rock art panels at Cairnbaan and Galsvaar (Map 5) suffer from the absence of dating controls. Elm declines are recognised at both pollen sites but at the boundary between mineral soil and overlying peat, so that it is unclear whether the decline is real or is a taphonomic effect: these sites need to be re–visited and 14C dated.

4.7.2 Bronze Age activity

On Mull at Torness, P. lanceolata appears from c. 2500 cal BC and a large expansion of grasses is recorded from c. 2000 cal BC with falls in hazel. At Fhuaran a sharp fall in oak pollen proportions at c. 2000 cal BC coincides with a rise in those of wild grasses, with P. lanceolata consistently recorded after c. 1500 cal BC with a sustained decline of birch, oak and alder. At Gribun P. lanceolata was established from c. 2000 cal BC in a landscape of grass and heath (Walker and Lowe 1987). Andrews et al. (1987b) described a grazed grassland with P. lanceolata established around Loch Cholla on Colonsay from c. 2650 cal BC and around An t–Aoradh on Oronsay after c. 1700 cal BC. It is possible that arable activity is recorded from c. 1900 cal BC around Loch Cholla. Cereal type pollen is also found from c. 2000 cal BC on Islay, around Loch a'Bhogaidh (Edwards and Berridge 1994) with an expansion in grazed grassland. The Dubh Lochan record (Stewart et al. 1984) differs from those in the west of the region in showing negligible woodland clearance until the later Bronze Age at around 1100 cal BC, and little further change.

4.7.3 Iron Age activity

Andrews et al. (1987b) saw birch wood regeneration from c. 800 cal BC to c. 570 cal BC on Colonsay as indicating reduced human activities. After c. 500 cal BC, however, came further woodland clearance and the emergence of a treeless landscape. On Kintyre a loss of hazel woodland in the region is recorded from Taynish Fen by Rymer (1974) after c. 500 cal BC, more sustained after the BC–AD boundary. Losses of oak and alder woodland occurred around Loch Cill an Aonghais at this time (Peglar 1980). Around Gribun on Mull the alder and hazel woodlands collapsed at c. cal AD 150 (Walker and Lowe 1987).

4.7.4 The historic period

Grassland around Dunadd was common at the BC–AD boundary. Locally, proportions of wet woodland declined at c. cal AD 400 and those of dry woodland declined c. cal AD 650. Heathland was established locally c. cal AD 450 but in the region only after c. cal AD 1100. A peak in the proportions of birch is seen immediately following abandonment of Dunadd c. cal AD 900, though later medieval woodland clearance removed it (Miller and Ramsay 2001). The Dunadd record has its interpretative problems in that the pollen signal seems somehow muted, not what you expect around such a site, and maybe the abandoned channel lay at the time within a larger Moine Mhor, distant from fields. There are other major, perhaps more significant citadel forts than Dunadd in the region such as Dunagoil on Bute (Harding 2004, 141–4) with peat close by which would reward investigation.

Bohncke (in Barber 1981) presented pollen data from Iona Fosse, 14C dated to around cal AD600. These have always perplexed in their interpretation of ash and oak woodland around the monastery when the ditch began to infill, not found by either Balaam (in Reece 1981) and Tipping (in McCormick 1993), who both sampled for pollen analyses a peat buried beneath the bank of the inner vallum of the monastery, pre–dating monastic colonisation by 300–500 years. It is simplest to assume that the tree pollen in Bohncke's record pre–dated any early historic period. Even so, ash–oak woodland would be unusual on the west coast and Bohncke's pollen spectra might be distorted by taphonomic processes. Balaam's and Tipping's samples came from different parts of the same bank, it is thought, and in detail they do not correlate (Tipping in McCormick 1993) but broadly it can be suggested that an open woodland of, probably, hazel was lost as Calluna heath expanded, and this was then succeeded by grazed grassland, but no record of other agricultural activities could be recognised. Ditch I stood next to a hedge of elder (Bohncke in Barber 1981). It is doubtful if new sediment stratigraphies will be found on Iona. Kingarth on Bute (Laing, Laing and Longley 1998) is almost surrounded by peat bogs. Lismore in the early historic period is yet to be understood fully. And what human activities might we expect around Eigg's monastic settlement with its links to the ascetic Céli Dé movement, the "desert isle in the ocean" (Fraser 2009, 370).

The transformation of the diverse landscapes around Oban from wooded to largely treeless was a product of clearance in a couple of hundred years after c. cal AD 850 (Macklin et al. 2000). Though the pollen data for the eastern Loch Lomond oakwoods (Stewart et al. 1984) reveal no evidence for intense woodland management in the last several centuries identified by Lindsay (1975) from the documentary record, Sansum (2004) profitably explored both in establishing the history over the last millennium of semi–natural western oakwoods at sites at the head of Loch Awe (Map 6), emphasising 18th and 19th century management of oak and hazel for commercial ventures. In the same area, Davies and Watson (2007) used both data sources to explore over the last c. 400 years the management of hill pastures and upland woods. Both studies employed what are known as 'small hollows', peat in basins small enough you can almost hop over, which describe the close details of a taskscape.

These brief descriptions of the results of many pollen analyses in the region might be seen as poor reflections of the interpretations but in most instances they are often not: they are what we have. It is not unsurprising in research strategies to see more data demanded but this simple observation is true for Argyll and Bute. The historic period in Argyll and Bute has been poorly served until very recently (see Section 10.2). In part this is because the establishment of organised agrarian landscapes in the last c. 1500 years has led to an apparent lack of landscape change. There is perhaps little for the pollen analyst to be interested in even though the demonstration of prolonged agricultural stability is often very significant to the archaeologist.

Pollen records like those on Mull and Rannoch Moor are from locations far from population centres. It is difficult, for example, to extract information on human activities from the sites on Rannoch Moor analysed by Bridge et al. (1990), partly because only partial pollen diagrams showing only major pollen taxa were presented, but also because the landscape was probably devoid of human occupation for long periods. It is useful to understand land uses at the margins of occupation, although we tend to interpret these as extensions of core activities, like mixed farming, simply expanding onto more intractable soils and climates. We are not good at identifying in the pollen record distinctive locale–specific activities such as shieling that may explain why people were in such seemingly marginal settings.

We have not investigated critical landscapes like Kilmartin (Sheridan 2012) or, in any depth, Achnacree (Carter and Dalland 2005). There is no clear understanding of the landscape settings of later prehistoric 'ritual' sites like Ruggles' corpus on Mull (Ruggles, Martlew and Hinge 1991; Ruggles and Martlew 1992; Martlew and Ruggles 1996) despite flawed attempts (Hale in Martlew and Ruggles 1996). Proving that trees were not blocking the view between a monument and what was to be looked at will be difficult but is feasible: intervisibility is a fundamental but un–tested assumption in archaeology (Fisher et al. 1997; Cummings and Whittle 2003; Gibson 2004). Later prehistoric 'ritual' landscapes, if they existed, must have been managed or monuments would have been lost within regenerating woodland. Perhaps they were inside woodland: we should not think that the dynamic early Bronze Age agrarian landscape at Torbhlaren (Tipping et al. 2011) tells us anything about Kilmartin: Torbhlaren is still marginal to this.

Only in recent decades have we come to ask questions of direct archaeological significance, working where the archaeologist is working, as at Glengorm in Mull, on Iona, however frustratingly short of information these data are, and at Dunadd and Torbhlaren. The idea at Dunadd was excellent: what impact did the precisely–dated rise to regional dominance of the fort have on local livelihoods? We need many more data in this critical period for Argyll. We can barely put together a coherent narrative for later periods. Did the transfer of power away from Dunadd have significance for the regional economy? What is the meaning of the regeneration of birch trees at Dunadd c. cal AD 900?