Climate and Climate Change
The Iron Age is taken here to mean the period between c. 800 BC and c. AD 500, the latter date a median estimate given the diachroneity of this boundary across Scotland. Climate reconstructions which reflect the very long-term, Milankovitch-driven millenial relationship between the Earth and the Sun (Davis et al. 2003) suggest summer temperatures in north west Europe, including Scotland, to have been slightly warmer than today, and winter temperatures not dissimilar to today. It is the more abrupt, centennial scale climatic fluctuations superimposed on these trends that had at the very least, the potential to impact on human livelihood (deMenocal 2001; Berglund 2003; Turney et al. 2005; Charman 2010). The summary here is a description and synthesis of palaeoclimatic data only. Inferred human responses to Iron Age climate change are considered later.
Bond et al.‘s (1997) record of sand grains in marine sediment transported in “armadas” of icebergs to the latitude of western Ireland, centred on c. 800BC, is a graphic though poorly resolved description of the hemispheric, probably global scale of this rapid climate change (Mayewski et al. 2004; Chambers et al. 2007). Oppo et al. (2003) report cold ice-bearing surface ocean water off western Ireland between c. 1100 and c. 400 BC, the only time this occurred in the last c. 5000 years, because the “gulf stream” was weakened. Marine resources would almost certainly have collapsed.
Measures of storminess will have been related to the strength of the North Atlantic jetstream. Wilson et al.‘s (2004) synthesis identified the period c. 1100-450BC as one of widespread sand blow, as do Bjorck and Clemmensen (2004) in Denmark, but increased storminess is also recognised in several case studies after c. 500BC (Wilson et al. 2001; Wilson 2002; de Jong et al. 2009), and in the Outer Hebrides Gilbertson et al. (1999) found that only the centuries after AD200 were as affected.
Temporal detail comes from more closely dated terrestrial records. Speleothem data are annually resolvable but complex in the climatic variables they describe. McDermott et al.‘s (2001) record from western Ireland is regarded as describing annual temperature more than annual precipitation. If this is correct, it bears little relation to changing ocean conditions. Rising temperatures from c. 1200BC, then stable and with limited variability around 800BC, are followed by increased variability but falling temperatures to c. 425BC. Oscillations were then extreme until c. 200BC, after which there were highly variable but falling temperatures to c. AD400. A different way to understand such changes is provided by Swindles et al.‘s (2010) records in Antrim from peat-based measures of drought (summer water deficit), with three dry phases, c. 1150 to c. 800BC, c. 320BC to c. AD150 and c. AD250 to c. AD470.
Speleothem records in Inchnadamph are interpreted to depict annual precipitation more than temperature (Proctor et al. 2002). Declining precipitation between c. 900 and c. 700BC fits well with warm and dry indications in Irish sequences. Much higher precipitation is seen from c. 700BC to c. 300BC. Drought in northern Ireland c. 320BC to c. AD150 is matched in dry conditions in northern Scotland, persisting beyond c. AD500, but lower temperatures would have reduced the risk of drought. Temperature and precipitation are more difficult to separate in the peat-based effective precipitation records across northern Britain synthesised by Charman et al. (2006) but there is considerable agreement with other records. The most direct interpretation of these is in how wet bog surfaces were, and these will relate to the wetness of mineral soils. The period c. 900 to c. 750BC was the driest in the later Holocene record. The abrupt shift at c. 750BC to very much wetter bog surfaces is astonishing, and until c. 40BC they remained very wet. Although rapidly dryer over some 50 years to c. 400BC, bog surfaces were still not dry, and did not become so until after c. AD200.
The case for a dramatic climate change, from warm and dry to cool and wet, in the LBA or perhaps EIA, is supported by the pollen core evidence from several sites in the Forth Valley (see Davies 2006 for a discussion). Ellis (2000b) interprets the evidence from the Forth Valley as representing gradual climatic deterioration. The available dates correlate well with those from peat bog recurrence surfaces from across north-west Europe, which have been dated to c. 500BC (Bell and Walker 1992, 72). Renewed glacial activity in Europe has also been reported for the mid first millennium BC (Bell and Walker 1992, 72). Human responses to early Iron Age rapid climate change.
It is important to be chronologically precise in this discussion. This section will not consider social instability and apparent upland abandonment in the later Bronze Age, prior to c. 800BC. It will focus on the period at and after c. 800BC. This is probably not just ‘splitting hairs‘. Highly resolved climate proxies indicate the exceptional rapidity of this event, and interpretations of climatic and social change generated before this was understood , which assume a gradual, centuries-long slide from c. 1200BC into final collapse in the early Iron Age have probably conflated what may have been two distinct phases of climatic instability in late prehistory.
Models developed in The Netherlands have stressed impacts on lowland rather than upland areas by precipitation increases at c. 800BC (van Geel et al. 1996, 1998) in which elevated water tables in soils drove populations away from established farmland and onto more marginal areas like salt marshes. Barber‘s (1998) argument for Arran comes closest to this model, though with upland soil water-logging and the blanket spread of peat leading to abandonment, but most case studies in northern Britain consider later Bronze Age abandonment: there are currently (Tipping 2002) no archaeological data in northern Britain other than on Arran that relate settlement change directly to the climatic excursion at and after c. 800BC.
In many respects current knowledge of Iron Age climate is very refined. This precision in reconstruction needs now to be related to agro-economic models to predict anticipated agrarian responses to climatic stress. For instance, Swindles et al.‘s (2010) data on drought should have had deleterious impacts at specific periods in forcing the wilting of shallow-rooted grass pasture. Can this be seen? How might this be identified in palaeoecological analyses? How might this impact have affected pastoral economies? Some of this thinking is being done. Van Geel et al. (2004) have explored the links between increased soil water tables and population movements in the nomadic Scythian culture in central Europe. Van Geel and Berglund (2000) argued that climatic stress led directly after c. 650BC to substantial population increases in northwest Europe: crisis at c. 850 cal BC was followed by the restructuring of society and its revitalisation.
As throughout much of northern Britain, there is evidence for extensive forest clearance in the latter half of the first millennium BC in the Lowlands. With the exception of those from Rae Loch, all of the radiocarbon dates suggest that this process was underway before the Roman army had even set foot in Britain, as has also been recognised in northern England (Tipping 1997). What is also clear is that these clearances took place at different times in different places and on different scales, just as they did in northern England (cf. Dumayne-Peaty 1998a). The overall impression is of mixed and fluctuating landuse in the Iron Age, with deforestation happening well before the Roman invasion in many places and woodland regeneration occurring in most areas in the post-Roman period (cf. Dumayne 1993a; b; 1994; Dumayne and Barber 1994; Dumayne-Peaty 1998a; b; 1999). As in northern England, the data for lowland Scotland suggest a marked intensification of agriculture from c.350BC onwards, leading to dramatic deforestation (Tipping 1997). Arable and pastoral aspects of the landscape can be recognised, but the relative proportions of these cannot be deduced from the data gathered thus far.
This evidence refutes van der Veen‘s (1992, 153) assertion that the Scottish landscape was not cleared until the Roman period. The evidence from southern and eastern Scotland adds weight to Hanson‘s (1996) argument, that extensive deforestation was well underway over much of northern Britain by the late pre-Roman Iron Age. Indeed, Hanson‘s argument that this gradual process had more to do with the expansion of settlement and agricultural activity than the specific timber requirements of the Roman army, is convincing. Evidence accumulated over the last two decades provides little support for Whittington and Edwards‘ (1993, 20) contention, derived from the evidence at Black Loch and the Aberdeenshire lochs of Braerroddoch and Davan that the dramatic changes in landuse, which took place in the first few centuries AD, were caused by the devastation wrought by the Roman army. It is only fair to note, however, that well-dated modern and archaeologically-useful pollen diagrams are still a rarity in many areas (Tipping 2005).
Sea Level Change
The Main Postglacial Shoreline, dated to 5800-6850 14C years BP, was thought to have been the highest Holocene raised shoreline in Scotland (Smith et al. 2000, 489). However, work by Smith et al. (2000) on isostatic land uplift during the Holocene indicates that there was also a later period of high relative sea level (the Blairdrummond Shoreline) in the Forth Valley and elsewhere, pace Ellis (2000a, 247 and 254; 2000b; Ellis et al. 2002) and Reid (1993, 3). Tipping and Tisdall (2005) have reviewed aspects for sea-level change for the Antonine Wall zone, and the Beauly Firth has seen a detailed study, but in many areas the sequence of land uplift is poorly known. In some areas this is of considerable significance, such as the southern littoral of the Moray Firth.
Richard Tipping (1994) has argued that the sampling strategy and temporal resolution of pollen diagrams needs to be improved. He also provides a useful cautionary note when he points out that the actual extent of farmed land cannot yet be determined from pollen data (Tipping 1994, 33-35). A far greater density of securely dated pollen profiles is required before anything but the most generalised picture of landscape development over the later prehistoric period can be given; large parts of the country have no reliable cores. In SE Scotland, the relatively small number of lochs in the area means that potential pollen core sites are limited, and the Forth Valley mosses may hold the most potential for elucidating these issues. Raised mosses and valley peat bogs still survive in Cardross Moss, Gartrenich Moss, Flanders Moss West, Flanders Moss East and Ochtertyre Moss (Soil Survey of Scotland 1982) and these probably present the most potential for enhancing understanding of the later prehistoric environment in the eastern lowlands. Research as part of the Angus Field School (Dunwell and Strachan 2007; Strachan et al. 2003; McGill 2003) indicated that in some heavily-impacted areas, suitable sites simply do not survive. Research into sea level change is moving fast and further inter-disciplinary research would do much to elucidate understanding of how Iron Age people experienced the landscape.