3.2 Landscape, environment, climate

Climate Change

Much work over the last 15 or so years has led to a new paradigm in understanding when, why and how climate has changed in the Holocene, the last c 11 500 years (Chambers and Brain 2002; Mayewski et al. 2004). Very large climatic reconfigurations of the hemispheric or global atmospheric circulation occurred, perhaps with a tempo of around 1400 to 1500 calendar years, which were also abrupt, perhaps complete in decades, certainly noticeable to human populations and of sufficient scale to impact on societies (de Menocal 2001; Berglund 2003).

The climate of Scotland is determined by conditions in the North Atlantic Ocean. The record of ice-rafted sand grains in the eastern Atlantic Ocean, just off western Ireland, has assumed major significance in defining periods of hemispheric-scale rapid climate change. Sand grains originating in only a few areas around the Arctic Ocean were periodically transported much further south than normal by ‘armadas’ of icebergs. At c. 2200 BC there is a modest peak in the proportion of such ice-rafted sand grains (Bond et al. 1997; Bond et al. 2001). The North Atlantic cooled, though it is unlikely that this event had a major impact on the strength of the “gulf stream” (Oppo, McManus and Cullen 2003) and so, little impact, perhaps on marine resources. However, a substantial change in the stratification of the southern Irish Sea after c. cal 1600 BC is explained by Marret, Scourse and Austin ( 2004) as reflecting a strengthening of the “gulf stream”, leading to milder winters, possibly increased winter precipitation and reduced seasonal contrasts, which can only have benefitted agricultural productivity.   

There is no record of past temperature change for Scotland. Pollen-derived estimates of seasonal temperature differences from present (Seppa and Birks 2001; Davis et al. 2003) cannot capture phases of rapid climate change: they define the slower rhythms of the Milankovitch cycle. Davis et al. (2003) suggest for Scotland that summer temperatures were warmer by around 1ºC in this period. Winters were around 1ºC colder at c. cal 2000 BC but were close to the present by c. cal 1000 BC. Seasonal contrasts were reduced through the Bronze Age. A long way north of Scotland, in northern Fennoscandia, and so used here with caution, summers warmed between 2000 and 1000 BC by 0.5ºC. Much finer resolutions come from speleothem data. The Crag Cave record in western Ireland (McDermott et al. 2001) is thought to reflect air temperature. Here mean annual temperatures in general were colder than the Holocene average throughout the Bronze Age, as cold as the later Neolithic but warmer than the early Iron Age. Northern Fennoscandian Pinus sylvestris (Scots Pine) tree rings are correlated with July temperature (Helama et al. 2002). At centennial resolutions, colder than normal centuries were rare after c. cal 2000 BC, only the 15th century BC ranking as one of the coldest in the record.  The 12th century calBC was unusually warm.

The period 2500–2200 calBC was substantially wetter than average (Barber et al. 1994; Anderson 1998; Anderson et al. 1998; Tisdall 2000; Charman et al. 2006). Soil water tables in central and southern Scotland (data are few further north) were high at c. 2000 calBC, becoming lower by c. 1500 calBC, but rising to peak at 1400-1300 calBC (cf. Marret et al. 2004). Lower soil water tables and dryer conditions are then sustained until c. 800 calBC (Charman et al. 2006), markedly dryer after c. 1150 calBC (Swindles et al. 2010).

Periods of increased storminess should relate to the strength of North Atlantic atmospheric circulation, though also reflecting the greater availability of wind-blown sediment through relative sea level fall. Machair on the Outer Hebrides was frequently mobilised between c. 1800 and 1300 calBC (Gilbertson et al.1999). Early Bronze Age dune construction is also recorded from the early Bronze Age on the North Sea coast (Wilson et al. 2001; Orford et al. 2000). Peaks in aeolian sand transport in southern Sweden were also at c. 2200-2100 calBC, and then between c. 1050 and 850 calBC (Bjorck and Clemmensen 2004), the last at least relating to dry soils and possibly drought conditions in Northern Ireland (above: Swindles et al. 2010). Dune instability characterised the northern Irish coast after c. 1400 calBC and before c. 1200 calBC (Wilson et al. 2004).    

Landscape Change

The issue of tracing sea-level change was central to work at the Early Bronze Age Pict’s Knowe henge (Tipping, Haggart and Milburn 2007). There it was shown that the peak in sea level was of early Neolithic rather than early Bronze Age date.© RCAHMS.

It is conventionally assumed that relative sea level in northern Britain fell constantly from its highest altitude at c. 4500 calBC (Shennan and Horton 2002). Smith, Cullingford and Firth ( 2000) argued from data on the Carse of Stirling and in southern Scotland for the occurrence of a second relative sea level rise, the Blairdrummond Shoreline, which culminated within the Bronze Age, identified now at several localities on the Scottish coast (Selby et al. 2000; Smith et al. 2003) and dated to c. 1800 calBC on the west coast (Smith et al. 2007). At some localities the altitude reached by this event, the Blairdrummond Shoreline, was higher than that reached by its mid-Holocene predecessor. Recent revisions to the chronology (Smith et al. 2010) now suggest the duration of high sea level in the Blairdrummond Shoreline to have been short-lived, but nevertheless, GIS modelling of the Carse of Stirling in later prehistory by Smith et al. 2010) shows how the distribution of archaeological sites makes more sense when related to high relative sea level.

By the beginning of the Bronze Age natural soil development had led to nutrient losses and podsolisation on most well-drained substrates (Davidson and Carter 2003).  Soil erosion was occurring. Climatic deterioration rather than human impact was probably causal in disrupting montane slopes and triggering soil movement across them in the early Bronze Age ( Mottershead 1978; Reid and Thomas 2006). High energy debris flows cascading down steep bedrock or talus slopes appear to cluster in time around 1800 to 1500 BC ( Ballantyne 2004). Alluvial fans in the western Highlands were constructed after c. 2000 calBC, most actively after c. 1300 calBC (Reid, Thomas and Tipping 2003). It is probable that on lower montane slopes and across the ‘flow country’ the cover of blanket peat we have today in Scotland was almost fully formed before the Bronze Age (Charman 1992; Tipping 2008). It is not an artefact of agricultural activity, though Bennett, Bunting and Fossitt (1997) argued that on Shetland, though formed prior to the Bronze Age, blanket peat spread accelerated due to agricultural change. Blanket peat could still spread across fields (Barber 1998; Carter 1998; Tipping et al. 2008) but this process was not as remorseless as once envisaged (Piggott 1972): farming communities could ‘stem the tide’. On slopes under blanket peat or under lightly grazed grassland, soil erosion seems not to have been a problem (Edwards, Hirons and Newell 1991; Tipping 1995). Within farmed landscapes, however, soil erosion from cleared hillslopes and cultivated fields may have been widespread (Edwards and Rowntree 1980; Mercer and Tipping 1994; Terry 1995; Carter 1998; Edwards and Whittington 2001) but seems not to have impacted adversely on farming routines.

(left) Blanket peat in Glen Affric (right) GIS model of blanket peat inception and spread in Glen Affric from a series of AMS 14C dates. both images (Tipping 2008) ©Richard Tipping

Some streams and rivers in northern and eastern Scotland changed their behaviour in the early Bronze Age (Macklin et al. 2005), usually constructing floodplains by depositing flood sediment, accompanied by lateral erosion and valley widening. Most are in montane and upland landscapes where climate drivers are probable (Robertson-Rintoul 1986; Tipping 2010). Lowland river systems do not display such behaviour, though they were capable of change before and after the Bronze Age, suggesting that woodland clearance and agricultural practice were not in this period causal in fluvial change. There is no evidence as yet for the scale of floodplain transformation described in central England where single channel meandering rivers emerged within the Bronze Age from natural, multi-threaded anastomosing streams as a result of intensive agriculture and concomitant changes in fluvial hydrology (Brown and Keough 1992). 

Knowledge Gaps

It is important to establish the precise altitudinal and temporal pattern of later prehistoric relative sea level change because, as Smith et al. (2010) have shown the distributions of later prehistoric archaeological sites make more sense when this is reconstructed. This is true not just for sites near the present coast but also for former estuarine regions. 

There needs to be more awareness among the archaeological community of the extents of these estuaries in the past. Archaeologists need to talk to coastal geomorphologists if opportunities for collaboration are not to be missed. For example, the work of Ellis ( 2001) on the Carse of Stirling could have contributed enormously to an understanding of sea level change had 14C dated borehole data only been surveyed to OD.

Empirical data on relative sea level change still need to be collected because computer models cannot identify features that affect specific localities. There is a need for the construction of local relative sea level changes in important early Bronze Age landscapes such as the Kilmartin Glen.

Sea level affects the geography of coastal landscapes, and so the distribution and availability of marine resources: these are rarely researched in the Scottish Bronze Age. Does one assume there was no role for these resources or has no-one looked?

Sand dunes and sheets produce a wealth of artefacts (e.g. Cowie 1996) but current understanding of their construction and chronology is very poor. Yet these are key sources in understanding key environmental stresses like storminess. The perception of hazard from storms, even more than the actual hazard, probably influenced long-distance connectivity and trade, and we need to understand this to explain the artefactual record.

People moving around was important, particularly as isotopic data show how distant were some of the movements. In this regard the navigability of rivers and estuaries is important (e.g. Strachan 2010). Whilst past behaviour and hydrological regime of rivers can be defined (Lewin, Macklin and Johnstone 2005) research has often ignored such basic attributes as channel width and depth, riparian woodland, natural log-jams and woody debris hindering movement, and the persistence of fording places.

Blanket peat spread in the uplands may have been important in the Bronze Age. Despite the sweeping statements above, there are very few localities in Scotland where the inception of blanket peat, its triggering mechanisms if any, its rate of spread and the role of farming communities in keeping it at bay has been measured. This is a central question for the understanding of later prehistory in Scotland. The time lag between acidification and blanket peat spread, and changes in the archaeological records needs muchy further research.

Natural Woodland Loss

Across the mainland of northern Scotland north of the Great Glen, populations of pine trees that had grown on blanket peat surfaces died between c. 2200 and 1900 calBC, seen in thousands of stumps across the Highlands (Birks 1975; Dubois and Ferguson 1985, 1988; Bridge, Haggart and Lowe 1990; Gear and Huntley 1991; Daniell 1997; Huntley, Daniell and Allen 1997) and as reductions in proportions of pollen (see reviews in Tipping 1994; Bennett 1995). The presumed regional synchroneity of this event has always encouraged explanation by a climatic driving force, but new data suggest considerable diachroneity, even in northern Scotland (Tipping et al. 2008). An attempt to show that pine woodland loss was anthropogenic at a site above Golspie in north east Scotland proved this proposal to be unsubstantiated (Tipping et al. 2008). Early and middle- Bronze Age metal axe marks showed people cutting through pine stumps and branches, but the trees themselves were already long dead.   

On the Northern Isles much of the woodland, which was never dense and always tenuous, had been lost before the Bronze Age (Keatinge and Dickson 1979; Bunting 1994; Tipping 1994; Bennett et al. 1997), again perhaps principally through climatic stress. At Catta Ness in Shetland the earliest woodland reduction is dated to c. 2000 calBC (Bennett et al. 1992), possibly anthropogenic but being coincident with the pine decline the authors point out that “disentangling cause and effect at any one site is likely to be difficult” (p. 263). Birch, oak and hazel trees at altitudes above 350m OD in northern Scotland were also impacted by this climatic deterioration (Davies 1999; Tipping, Davies and Tisdall 2006), so that large expanses of the west-central highlands, Cairngorm and northern Highlands were probably changed at c. 2000 BC to open dry heath and grassland by woodland collapse.

In recent years it has been recognised that populations of oak trees growing on peat bogs also suffered abrupt, short-lived collapses, ‘dying-off’ phases, across north west Europe (Leuschner et al. 2002). Within the Bronze Age there were such phases around 2000, 1740, 1680, 1550, 1470, 1400 and 1030 calBC. These too must have been natural though cause is unknown. Oak trees regenerated following these events but a long term effect of their failure may have been more aggressive and persistent growth of adventitious hazel and birch trees.  Such successional change would mimic the pattern created in, for example, the colonisation by trees of abandoned farmland. The impacts of climate and, possibly, disease, on tree populations make it harder now to attribute anthropogenic origins for some woodland changes: there are strong competing hypotheses. The impacts of climate stress probably decreased southward and eastward, or more probably the deciduous oak-hazel woods of southern Scotland were more resistant to such stress.      

Knowledge Gaps

There is a need to disentangle natural and anthropogenic impacts on fragile woodland in northern Scotland. This should come from analyses where climatic indicators are related stratigraphically to measures of tree abundance.    There are a number of anecdotal and unexplored reports which describe pine stumps in peat bogs charred by fire. Fire can be natural in origin but can be set by people. This method of clearance would have been very effective given the flammable nature of pine. These reports suggest we should not yet dismiss human activity in causing losses of pine trees at c. 2000 BC or later in the Bronze Age.

Most palaeoecological work on the pine decline in northern Scotland has focused on tree populations that grew on climatically driven, temporarily dry peat bog surfaces. Only recently have we had pollen data to show that in some regions pine woodland was very important on well-drained mineral soils as well (Paterson 2011). This is useful to know because we can now recognise the need for them to be cleared for agriculture.

It would be good to know how much primary woodland, unaltered by people, still grew at the beginning of the early Bronze Age, and whether this was diminished at the end of the period. The ‘gut feeling’ is that very little deciduous woodland had not been altered by c. 2000 BC but that pine woodland was less altered. If correct, this might mean that there were far fewer resources within pine woodland, or just that they grew where there were few people.

Did primary woodlands comprise one dominant tree taxon or were they mixed with other trees? This is important because this affects the ease (a) with which resources (fuel and wild food) could be hunted and collected by people, (b) how easy it was for people to move through woodland and (c) what the spatial extent and scale was of dying pine and oak populations. Davies (2007) has described the delicate relationship between naturally dying pine populations and farmers but we do not know what effect, for instance, ‘great dying off’ events in oak populations had on the landscape or on people. Palynological techniques are available to understand woodland dynamics at the stand scale. These need to be used more fully.

The relationship between people and trees in later prehistory is not yet fully understood. Too often it is assumed to have been an antagonistic relationship, and terms such as ‘clearance’ and ‘impact’ are used, but these terms reflect more recent thinking, arguably divorced from nature and human need. 

Some aspects of agricultural development

Andrew Richmond (1999) argued from an extensive review of data on settlement and land use that the early Bronze Age was the first time when people committed entirely to agriculture as an economy: see also Parker Pearson (1993; 1999) and Brück (2002). Renewed focus now on the central importance of farming in the Neolithic (Rowley-Conwy 2004; Bogaard and Jones 2007; Jones and Rowley-Conwy 2007) throws into question Richmond’s interpretations, but in Argyll and on the Solway Firth there was a markedly greater interest in farming only from the earliest Bronze Age (Macklin et al. 2000; Tipping et al. 2004).

There is in general no sharply defined decline in the birch-hazel-oak woods of northern and western Scotland (Tipping 1994; Edwards and Whittington 1998; Edwards et al. 2000). It was gradual, with individual trees dying and not being replaced. Anthropogenic woodland reduction through sustained low intensity grazing pressure (Buckland and Edwards 1984) is only one of several explanations. Natural attrition through exposure (Quine 2003), blanket peat spread (Tallis 1991; Charman 1992; Tipping 2008) and soil deterioration (Wardle, Walker and Bardgett 2004) can have been implicated, with all factors probably operative. Much deciduous woodland in these regions survived through the Bronze Age (Tipping 1994; Smith 1998; Carter, Dalland and Long 2005; Bunting, Middleton and Twiddle 2007).

Much of the oak-dominant woodland of central and southern Scotland also survived through the Bronze Age. The density of Bronze Age archaeological remains in the Cheviot uplands led Burgess (1984, 1985, 1990, 1995) to see the hills as treeless, highly populated and economically vigorous (cf. Gates 1983). Pollen diagrams show abundant trees still remained (Tipping 1997a; 2010) with farmsteads and field systems isolated within woodland, supporting a low population. The archaeological record probably overestimates the numbers of people creating it, and palaeoecologists probably overestimate the numbers of trees there were. Some, perhaps a minority, of the woodland was probably still primary, still unaffected by agricultural activities, but its survival in some instances need not represent woodland unaffected by people. Resource protection from livestock, conservation and management may have existed also (Tipping et al. 2008). Fuel was needed: Dickson and Dickson 2000, 70-75) show the wide range of species used for fuel in upper Clydesdale. Monuments used tall and straight oak trees. Some trees may have been sacred (Darwin 1994; Newman et al. 2007, the latter in an Iron Age context).

Within deciduous woods that suffered few natural setbacks, anthropogenic woodland clearance is almost ubiquitously described as ‘small’ and ‘temporary’, and in contrast to the slow attrition of woodland in the north and west, woodland clearance phases in central and southern Scotland were often short-lived, episodic, and always followed by woodland regeneration (Tipping 1994). Judith Turner (1965, 1970, 1975) tried to understand the spatial extent of woodland clearance from an integrated network of pollen analyses, ‘three-dimensional’ pollen diagrams.  By small, Turner (1975, 95) suggested that Bronze Age clearings were ‘up to a few hundred metres in diameter at the very most’. Re-examination of her data in light of a better understanding of pollen transport might suggest that clearings were of several square kilometres, far larger and perhaps related to ground shared by several rather than single farmsteads. Tipping (2000) calibrated Turner’s original 14C based chronology to conclude that the duration of each clearance event may have been c. 150 years, again longer than the duration expected of individual roundhouses (Halliday 2007).

This impermanence discerned in the palaeoecological record is at the scale of the farming unit (Gerritsen 1999). There are difficulties in measuring settlement shift archaeologically. At Lairg in northern Scotland it was nigh  impossible to demonstrate that two roundhouses in a cluster were occupied contemporaneously despite a comprehensive dating strategy (McCullagh and Tipping 1998): this sequence might now benefit from Bayesian approaches to dating. Gerritsen’s (1999) description of Iron Age ‘wandering settlements’ in Denmark, though from a later period and with buildings very different to Scottish roundhouses, does fit very well with the patterns of shifting land use described in pollen records from the Scottish uplands. Later prehistoric roundhouses may have been occupied for only a few decades before needing to be rebuilt (Barber and Crone 2001; Halliday 2007; Pope 2008) (but see Ralston and Ashmore (2007) for a critique of this idea) but the spatial scale of apparent settlement shift is bigger than implied only by the need to rebuild. ‘Wandering’ seems to indicate a freedom of movement unconstrained by rigid systems of land tenure: the organic growth of small patches of cultivation shown for the Danish earlier Iron Age by Sorensen (2007) would also fit well into Scottish Bronze Age landscapes. But why settlements ‘wandered’ is unclear. Slash and burn cultivation (Rowley-Conwy 1981; Huntley 2007) is an inappropriate explanatory model borrowed from environments entirely different to ours and from soils more vulnerable to nutrient loss than ours. Besides, the routine application of domestic waste to restore soil nutrients is known from Bronze Age agricultural systems (Bakels 1997; Guttmann, Simpson and Davidson 2005). Long term fallow is another possibility (Bevan 2007) with historical parallels in western Scotland (Dodgshon 1998) though not persisting over several human generations.  It is, perhaps, to be borne in mind that disease or vermin might well become endemic in an old house leading to rational and/or superstitious desertion.

On cleared ground field systems were established. The emergence of cairnfields in northern Britain is most often seen as a process of gradual organic growth (Quartermaine 2002), linked to the beginnings of land tenure (Johnston 2002). Cairnfields have been seen as foci of sustained upland agriculture (Jobey 1968; Davies and Turner 1979; Hoaen and Loney 2007). Cayless ( 2000) found, however, from pollen analyses within and largely reflecting the use of the cairnfield at Stanshiel Rig in upper Annandale (RCAHMS 1997 ) that agricultural activity was intense but only brief, a century or so in the earliest Bronze Age, after which there was little further  agricultural activity. Cairnfields need not have represented major or permanent transformations of the landscape. Other field systems have proved difficult to understand archaeologically. At An Sithean on Islay what was thought to have been a single field system was shown to be more fragmentary (Barber and Brown 1984). It is unclear how extensive was the small coaxial field system 14C dated to the Later Bronze Age at Tulloch Wood near Forres (Carter 1993). Cord rig (Topping 1989) probably has a long rather than a short chronology (Carter 1994; Tipping 2010). Away from Scotland, what was seen as a unitary landscape typified by coaxial fields on Dartmoor (Fleming 1988) is now being deconstructed, appearing far less coherent (Johnston 2005; Fyfe et al. 2008). Scotland has as yet nothing like the fully integrated agrarian landscapes proposed for southern England (Yates 2007).

There is considerable pollen-analytical evidence for upland cereal cultivation from the early Bronze Age (Tipping 2002). The crop was almost entirely of barley: warm summers, particularly after c. 1600 BC may have encouraged this.  There is little evidence for a specialisation emerging between lowland mixed or arable farming and upland livestock production (Tipping 2002). The impression from palaeoecological data is that Bronze Age farmers were doing much the same thing everywhere in a landscape without core areas and consequently with no margins.

At Lairg (McCullagh and Tipping 1998), however, soils were cultivated around the roundhouses but it could not be shown that cereals which were stored had been grown locally. Trade with specialist arable farms is possible. Preston, Pearman and Hall (2004) identified several arable weeds that appear in Britain for the first time in the later Bronze Age, which they tentatively attribute to increasing connectivity through trade, but specialist arable farms may have allowed more rapid spread through creating larger expanses of ploughed ground. In this model, upland areas were settled by people committed to predominantly pastoral activities. Pastoralism has been suggested to have been a Bronze Age specialism at Newton Mearns in the Clyde Valley (Toolis 2005), and in this and earlier discussion (Halliday 1985; Cowley 1998) a transhumant economy has been invoked. This remains speculative. The ubiquity of the phrase ‘mixed farming’ to describe the agriculture of these communities might be correct or might be a rather glib euphemism meaning little. 

One of the dominant features to the Bronze Age of much of Scotland is the manner in which human land-use attempts to counter, but could often exacerbate, the trajectory of declining soil condition. A prominent footprint left by such behaviour are farming methods which concentrate soil nutrients (either by transport of soil or by transport of animal dung) into smaller managed units. These may equate with fields, as recognised in other areas and in later periods; the establishment and maintenance of fields may have generated acute changes in perceptions of land tenure, inheritance and social hierarchy.

Knowledge Gaps

There is a need to question again the role of agriculture in the Bronze Age, and in particular how it developed from that in the later Neolithic.

There is still a mismatch between the landscapes imagined from archaeological data and those imagined by palaeoecologists. This is a particular issue in the Bronze Age as settlements become easier to identify and excavate. In part this is because archaeologists are describing evidence for settlement and palaeoecologists describe land uses, which are inevitably more extensive. Both disciplines are moving closer together: landscape archaeology has directed attention to recording at spatial scales that the palaeoecologist is happiest describing and the palaeoecologistd are learning to adjust their reconstructions to smaller, human scales. But new issues emerge. The recent suggestion that houses had rather short lifespans has substantial implications for reconstructions of settlement density, although this idea makes more likely the interpretations from palynology of rather empty uplands in the Bronze Age.

How can the suggestion that core agricultural areas did not exist in the Bronze Age be tested? Re-evaluation of past work on, e.g charred plant remains, might identify more sophisticated techniques (van der Veen 1992) employed in some regions and not others. What does the archaeological settlement record reveal: are there contrasts in the size or complexity of settlements? 

It is difficult for palaeoecologists to recognise woodland conservation and management though it is likely that such techniques were practiced in later prehistory. Experimental approaches to recognising coppicing are ongoing but at present we continue to assume that woodland destruction was the only intention of farming communities.

The apparent mobility of Bronze Age farming communities at the temporal and spatial scales described by Judith Turner 50 years ago is not understood. Again, short-duration settlement might hold one key, although houses can be rebuilt on the same site. There seem to be few environmental reasons for this behaviour (above). The apparent impetus to move around every few generations needs to be explored from landscape archaeological approaches coupled with Bayesian or wiggle matched dating strategies from high temporal resolution, local scale pollen analyses.

If mobility is poorly understood, then sedentism is not clear either.  How do we demonstrate sedentism?

Scotland seems not to have seen the large scale co-axial field systems recorded, for example, in the Thames valley. Why not? This cannot be because of site destruction: the Thames Valley is heavily disturbed yet still the field systems are found. But it might be through the lack of very large excavation projects in Scotland (see Catling 2011). Was their absence also through environmental constraints? The Lothians and Angus were surely sufficiently productive to lead to this scale of organisation. Might differences in land use relate to differences in social structure?

There is a need to develop techniques to recognise agricultural specialisms. Can the balance between arable and pasture be defined? Charred plant analyses rarely identify pasture (van der Veen 1992) and pollen records are biased against crop growing. Can one begin to identify what choices in agricultural techniques were available at any one time? Is it possible to define why some techniques were known about but not chosen? When were innovative practices discovered? What motivated the innovation at that time? Are there techniques to define, for instance, the season a site was occupied, because only this can unambiguously demonstrate transhumance. There is a need to integrate the study of faunal remains and animal husbandry technique with evidence from soils, pollen and charred plant remains.

Possible human responses to climate change in the early Bronze Age

Between c. 2000 and 1500 BC there was a major and sustained colonisation of the uplands throughout Britain, into previously unoccupied or scarcely visited montane landscapes (Burgess 1980; 1984; Fleming 1988; Parker Pearson 1993; Ashmore 1996; 2001; RCAHMS 1997 ; Cowley 1998; Tipping 2002, 2010). This has usually been interpreted as an expansion of the population through population growth. ‘Expansion’ is a difficult term, because this implies the filling-up of a core area with people and the colonization of marginal ones: people were pushed uphill. There are, however, no data, palaeoecological or archaeological, from which to suggest increasing population pressures at this time (Tipping 2002). There is little reason to think lowland soils could be exhausted of nutrients given the evidence for manuring (Bakels 1997). There was no climatic ‘golden age’ at the beginning of the Bronze Age: instead there was a major climatic deterioration. A dislike for determinist arguments has led to lack of awareness of climate change, though the tide is beginning to turn again (Haselgrove and Pope 2007). There is however a need for caution (Tipping 2002; Coombes and Barber 2005): the scale of hemispheric climate change is stunning but climate change need not automatically have led to societal disruption. 

Burgess (1984; 1992; 1995) imagined settlers either pushed into the uplands because lowland soils became nutrient-poor or being drawn into the uplands through climatic amelioration, but these arguments can easily become circular (Bradley 2002). They are unlikely to have been correct. Lowland soils have always supported the highest populations, and there is no evidence that this was not true in the early Bronze Age. The second is very unlikely on palaeoclimatic grounds. Colonisation occurred not because climatic amelioration encouraged it but in spite of large scale climatic deterioration (above: Tipping and Tisdall 2004). Davies ( 2007) has elegantly theorised how the demise of upland pine dominant woods in northern Scotland at around 2000 to 1800 BC created landscapes with diverse soils and resources and the ‘right conditions’ for farmers to be drawn to the uplands. To paraphrase Baillie (1998): ‘bad for trees – good for humans’. What those ‘right conditions’ were remains to be defined. Davies (2007) argued that pastoral specialists were advantaged. Economic drivers may have promised a benefit that outweighed ecological and climatic risks (Walsh 2005). But it is not clear if such specialisms existed (above), and Davies (2007) identifies the lack of grazing in peat- and heath-dominated uplands, the short growing season for grasses and the absence of meadow and hay as the major limitations on economic growth. Dependence on trade may have had very high risks. An insular, subsistence economy, carefully constructed and honed, and purposefully simple, with no requirement to trade, may have characterised pioneer communities high in the hills. The simplicity of this tried and tested system may have made it most resilient to environmental stresses because there was little that could go wrong (Tipping 2005).

Davies’ (2007) model does not in all its characteristics apply to regions of deciduous woodland in central and southern Scotland. A different driving force is required to explain the same colonisation of the uplands. This is not yet understood. Population growth is one driver, but is poorly understood because lowland areas are those most impacted by destruction of the archaeological landscape (Haselgrove 2002). The best measure of population density in lowland areas will probably come from securely dated pollen data but these are to date too few to synthesise.   

Knowledge Gaps

There is a need to explain why the uplands began to be colonised in the early Bronze Age. Given that the early Bronze Age was characterised by increased waterlogging, probably a direct increase in precipitation, do we look to the explanation for late Bronze Age migration developed by van Geel: are the two periods analogous? The lowlands are the sources of most Bronze Age archaeological finds but not settlement (Cowie and Shepherd 2003). Did lowland settlement cease? Environmental evidence needs to be combined with a better understanding of society, in terms of kinship and cosmology.

There is a tendancy to think of core areas for agricultural productivity, and therefore margins. Climatically, except in the early Bronze Age, around 1500 calBC and at the boundary with the Iron Age, there were probably fewer climatic contrasts between lowland and upland and between west and east, particularly after c. 1500 calBC as seasons became more similar. Perhaps one should think of a rural economy without core areas, in which upland farmers did not recognise themselves as being disadvantaged.    

Possible human responses to climate change in the late Bronze Age

Arguments have been presented over many years that climatic deterioration impacted on large areas of the British uplands towards the end of the Bronze Age, forcing their abandonment by farmers (for example Piggott 1972; Burgess 1984, 1985, 1989; Barber 1998; Robinson and Dickson 1998; Amesbury et al. 2008). Interpretations of abandonment from northern British landscapes focused on the downward depression of agricultural limits and so the abandonment of upland landscapes, influenced by Parry’s (1975, 1978) work on ‘little ice age’ impacts in south east Scotland. Soil acidification through nutrient loss, and the spread of blanket peat are also postulated (Robinson and Dickson 1998).

Challenges to these interpretations have been made on archaeological grounds (Gates 1983; Young and Simmonds 1995, 1999; Young 2000; Bevan 2007 ), arguing for continuity of settlement across the Bronze-Iron Age transition. Cessation of anthropogenic activity is, of course, very hard to define (Caseldine 1999). Palynological evidence in the uplands of Scotland was synthesised by Tipping (2002), who suggested that there was little evidence for retreat. Dark ( 2006) reached the same conclusion from a larger data-set throughout Britain. Recent interpretations have, however, suggested that withdrawal from the uplands in the late Bronze Age can be recognised. This was characterised not by abandonment but by the re-organisation of farming practice, and in particular the decision to focus crop-growing at lower altitudes (Davies 1999; Tipping et al. 2008; Tipping 2010). This period may have seen the first differentiation of upland and lowland farming practices and the first time that some form of agrarian specialism emerged (Halliday 1993).

Burgess (1989) and Baillie (1989) saw this within the later Bronze Age, at c. 1250 BC. They related it to volcanic activity (see also Grattan, Gilbertson and Charman 1999). That model has not fared well (Buckland, Dugmore and Edwards 1997). The period 1200 to 800 calBC undoubtedly includes within it a phase of global abrupt climate change (Mayewski et al. 2004; Chambers et al. 2007) but Mayewski et al. (2004) ‘fudge’ the chronology by identifying the entire period 1200-800 calBC as one of ‘rapid’ climate change. Closer dating of climate change in this period is difficult because it lies within a major radiocarbon plateau but van Geel and colleagues in The Netherlands used wiggle matched 14C analyses (van Geel and Mook 1989) to identify a short period around 850 BC as critical. This boundary is significantly different to that of c. 1250 BC that Burgess (1989) and Baillie (1989) envisaged as the crisis. Analyses of changing water tables in Dutch raised mosses have suggested that the greatest impact on people was likely to have been in lowland landscapes (van Geel et al. 1996), not in the uplands. van Geel et al. (1996, 1998) have argued that climatic deterioration led to elevated water tables and population movement away from established farmland and out onto salt marshes.  More recently van Geel et al.(2004) have explored the links in this period between climatic deterioration and population movements in central Europe. van Geel and Berglund (2000) also argued that this climatic stress led after c. 500 BC to the restructuring of society and its revitalisation across north west Europe.

Knowledge Gaps

Was there a ‘retreat from the margins’ at the end of the Bronze Age? How could this be measured? Archaeological evidence for settlement abandonment and closure seem to have occurred towards the end of the Bronze Age (Barber 1998; McCullagh and Tipping (1998) but the difficulties in demonstrating these from absences of evidence requires no explanation here. Turning this issue around and looking for positives it should be possible to demonstrate that lowland areas were farmed more intensively or extensively. This was what Tipping et al. (2008) tried to show from pollen analyses, with equivocal success. Can we show this archaeologically? What competing hypotheses are there for this apparent abandonment?  These need to be explicitly tested. Is climate stress the most economical explanation? Pollen analyses can suggest that the land did not become derelict: in this sense nothing was abandoned and it becomes important in future to use more precise terminology. Even where crop-growing may have ceased (more absences of evidence) the hills were turned to livestock production. The difference between settlement and land use is critical here. Which would have been more important to people at the time?

There has been considerable confusion over when the ‘retreat from the margins’ occurred. This issue needs to be clarified.

Colin Burgess used in his interpretations what he saw as reductions in the amount of metalwork being circulated in the Ewart Park phase as evidence for populations under climatic stress. This needs to be re-evaluated. Understanding of late Bronze Age metalwork has changed. Is the Ewart Park phase still dated to the period Burgess thought it was? Is metalwork production/circulation a valid indicator of environmental stress? Plunkett (2009), for instance has related metalwork production to the vigour of Bronze Age farming in Ireland, but has suggested that metalwork production was a measure of socio-economic or political factors rather than failures in subsistence.  

Can van Geel’s model (van Geel and Berglund 2000; van Geel et al. 2004)be applied to Scotland?

Human geography

The nature and extent of Bronze Age human settlement in Scotland was affected by two abiding constraints on settlement: the need to avoid in-breeding in domesticated plant and animal communities and human communities; and seasonal threats to survival. The spread of human settlement to every ecological niche in Scotland seems to imply that both constraints were overcome. It is likely however, that both constraints continually re-asserted themselves causing settlement to retract and domesticate and human populations to crash.

As the heterodox ecologies changed through time and the range of ecologies on offer shrank, it is likely that human populations developed specialist knowledge to cope with local variations and so one might presume social, technological, economic and political differences to become sharply defined as the Bronze Age progresses: highland and lowland habitat boundaries reiterated in human geography boundaries.

What has become clear is the very long-term attachments to landscape and place during the Scottish Bronze Age, with continuity of settlement landscapes far more prevalent than previously supposed (Pope forthcoming). Whether the nature of this settlement represents settlement sedentism, however, is a topic for further research (cf. Halliday 2007). At some long-term sites – such as Green Knowe or Kintore – the impression gained thus far is that occupation might instead be considered ‘episodic’ and bound up with generational household shifts. Gaining an understanding of associated systems of land use in now paramount.

Some of the skills of building, especially the skills of initiating and managing large projects were no longer practiced, or were directed at less monumental communal structures (such as field systems, constructed terracing, enclosure construction). In contrast, the highly developed construction skills for domestic accommodation visible in some areas (e.g. Orkney, NE Scotland) had become commonplace. The most radical change can be seen in the development of skills  and capacity for changing landscape both at the local, probably intentional, level (excavation of stone and minerals) and at the much larger, and perhaps unintentional level including (loss of habitat, landform change, and water course changes).

By the end of the 2nd millennium BC, a large proportion of the landscape within the Highland zone, essentially anthropogenic in its foundation and organisation, had changed from land suitable for arable to land only useable as pasture. Within this degraded landscape, settlement and landuse strategies shifted from an expansionist approach, exploiting change, to one which expressed itself as resistance to change. The dominant impression is one of resilience and survival. Contemporary shifts in strategy probably occurred on lowland better quality soils but later erosive landuse has made any evidence relating to these areas very faint and uncertain.

By the end of the 2nd millennium BC, a large proportion of anthropogenic landscape within the Highland zone had changes from land suitable for arable to land only useable as pasture. Within this degrading landscape, settlement and landuse strategies shifted from an expansionist approach, exploiting change, to one which expressed itself as resistance to change. The dominant approach was resilience and survival. Similar dated shifts in strategy probably occurred on lowland better quality soils but later erosive landuse has made the signal very faint and uncertain.

Knowledge Gaps

How communities developed and communicated techniques of land husbandry and management is not understood. Furthermore it is not known if the model of expanding core areas of Neolithic farming and the  consequent alteration of the Neolithic wildscape is well enough understood to furnish a reliable description  of the base line from which Bronze Age landuse developed.

It is equally unclear whether the model for Bronze Age arable landscapes of  initial expansion in the earlier centuries of  the 2nd millennium calBC followed by contraction at the end of that millennium and in the early centuries of the next was a single event spanning the 2nd millennium or is a post facto amalgamation of many – perhaps single generation events occuring within that period.  Currently developing techniques for the discernment of chronological subtlety in the palaeoenvironmental record should be able to assist with this difficulty.

Neither is the role of land tenure and inheritance in the evolution of the landscape understood.  It would seem probable that the local disposition of ‘fields and fences’ was implemented at family level but were there political agencies operating at a larger land unit (e.g. estate, territory or state level) within which tenure from generation to generation influenced the landscape.  Very little is known about how settlements spread, took in and converted land and how that process worked within the constraints of seasonal survival and the avoidance of in-breeding. Did some populations live in isolation for many generations; did others compete, perhaps aggressively, to overcome such constraints? Careful inspection of areas where large tracts of prehistoric landscape survive, working out horizontal as well as vertical stratigraphical relationships, might well throw light on these important issues.

Understanding these land use systems associated with settlements through analysis which combines an understanding of architecture, landscape, and material culture assemblages, alongside an understanding of settlement temporality, and the relatedness of settlements across the landscape, must be a priority.

If such economic territories existed, were there means of exchanging surpluses (e.g. processed grain or butchered meat) across their boundaries? Could such a systems be recognized in the archaeological record?  It is possible that refinements of isotopic analysis of animal or, indeed vegetational, remains may be able to tackle these problems in the future.

The final part of the Bronze Age presents the most extensive archaeological record of landscape abandonment from any period since the start of the Holocene. Little is known, however, about the processes or speed of abandonment – for which more evidence may be expected from Palaeoenvironmental studies. Population statistics from before, during or after these abandonments are terra incognita and it is difficult, currently, to suggest proxies that would yield any reliable information in a circumstance where burials are virtually unknown. Without these data, and without the means of acquiring them, our understanding is seriously impaired. Current research is woefully ignorant of social relations within any Bronze Age context. For example, one may suspect that slaves existed but how power was accumulated, exerted and inherited to allow this is not yet understood.

The answers to such questions might be available to careful excavation but for much of Scotland there is a potential mismatch between the upstanding appearance of many Bronze Age settlement remains and the condition of their stratified deposits. The survival of such remains is often coincident to acid soils too wet and/or acidic for modern arable farming and these preserving conditions tend to be very hostile towards subtle archaeological evidence that leaves residues of information that are not easily recovered by current excavation techniques. The use of nano-techniques for the recovery of microscopic residues ,even at the molecular levell is absolutely essential in the struggle to gain as full information as possible from such sites.  

It is fair to say that very little is known about how settlements spread, took in and converted land and how that process worked within the constraints of seasonal survival and avoidance of in-breeding. Did some populations live in isolation for many generations; did others compete, perhaps aggressively, to overcome such constraints?

The answers to such questions might be available to careful excavation but for much of Scotland there is probably a mismatch between the upstanding appearance of many Bronze Age settlement remains and their condition. Survival of such remains is often coincident to acid soils too wet and/or acidic for modern arable farming. Those preserving conditions tend to bleach out the subtle archaeological evidence leaving residues of information that are not easily recovered by current excavation techniques. So grappling with human geography means grappling with more tailored techniques of acquiring the information.

See also the ScARF Case Study: Impact of the Environment, Orkney


As mentioned in this section, the excavations in Arisaig in the West Highlands are available to download as SAIR 15: Early Land-use and Landscape Development in Arisaig.

"Re-alignment of a 6km section of the A830 road in Arisaig provided an opportunity to investigate the archaeology of this poorly understood area of the West Highlands. A combination of archaeological and palaeoenvironmental techniques were used to investigate the road corridor. Archaeological survey, followed up by selected excavations, identified a previously unrecorded Bronze Age kerb cairn and two areas of shieling huts. Investigation of the shielings obtained evidence for repeated reuse of sites and reconstruction of structures through the medieval and post-medieval periods. In both cases, Bronze Age deposits were also recorded at the base of the medieval sequence. Analysis of a long peat core from a basin close to one of the shielings revealed a history of continuous but gradual decline in woodland, starting in about 3200 BC and continuing to the present day. Collation of archaeological and palaeoenvironmental data from the present project and previous investigations in the area have allowed the creation of a tentative model of landscape evolution for Arisaig."