The Irish Burnt Mound Database

As part of this project a database was constructed of Irish sites reported as ‘burnt mounds’ or fulachtaí fiadh (Fig. 1). The database was compiled from the Irish Excavations Database (http://www.excavations.ie) as it was at the end of 2015, using the search terms ‘burnt mounds’, ‘fulachtaí fiadh’, and ‘burnt spread’ in the site type field. This returned 465, 775, and 200 hits respectively. In each case Townland, excavation licence number, grid reference, publications(s) arising, excavator, excavation.ie reference number, were recorded along with the abstracted material from the site description including dimensions of the mound, dimensions and construction material of the trough, and dates where available. The distribution map was created through the use of similar search terms through the Irish National Monuments website (http://www.archaeology.ie) which were exported as a shape file which was then overlain on the base map.

Charcoal, plant macrofossils and pollen analyses

Charcoal from Cahiracon included oak, with some birch (Betula), hazel, ash, alder, Maloideae sp. (apple family), Prunus spinosa (blackthorn), holly and Salix sp. (willow). Radiocarbon dating of charcoal from the burnt mound dates the site to the Late Bronze Age (Table 1). Two monoliths (M7 & M5) were taken for environmental analysis; M7 from the west-facing edge of the baulk directly above the trough which would relate to post-trough use activities and M5 from the north end of the same excavation. The pollen diagram from M7 (Fig. 4) which has been divided into five pollen zones (CC7P1-CC7P5) is dominated by mixed deciduous woodland of alder, hazel, and oak. However, there are dramatic changes with the lower zones dominated by oak which is replaced after a partial clearance by oak, alder and hazel. The basal zone (CC7P1) indicates a large clearing or partially open landscape with significant heathland or bog. The rise in hazel and alder in zone CC7P2 indicates the regeneration of woodland over, or close to, the site continued into zone CC7P3. The fall in oak early in zone CC7P4 represents a second clearance phase and late in zone CC7P4 a second regeneration of woodland occurs over the site or area which on this occasion is dominated by oak rather than hazel and alder. The remarkably high oak values (80% TLP) strongly suggest its presence on site. Lastly, in zone CC7P5 there is another phase of woodland clearance with some expansion of heathland and cereal cultivation appears restricted to this last clearance phase. From the stratigraphy the pollen and spore zone CC7P1 is correlated with the abandonment of the trough and the later zones with archaeological phases 5–7 which include the use of the hearth and mound site to the north, east, and west of the trough and the eventual abandonment of the entire site. Monolith M5 provides a sequence from pre- to post-fulacht activity as it spanned two natural units, a lower peat and an upper peat with in-between sediment from the edge of the burnt mound. The pollen diagram, although allocated a single zone, shows a muted clearance signal similar to that in CC7P1 but with cereal type pollen (Supplementary Fig. S1). Given the high levels of oak and alder it suggests that the site was very close to or, more likely within, oak woodland, with some alder trees nearby. During the period covered by the diagrams there is a partial clearance of woodland associated with burning on site corresponding with the construction of the mound. After the burning oak regenerates, but there are also species characteristic of disturbed woodland such as holly and birch. The episode is also associated with arable cultivation of oats or wheat which ceases during the regeneration phase.

Fig. 4 Cahiracon monolith 7 pollen diagram

From Inchagreenoge charcoal identifications are dominated by alder, with lesser amounts of ash, hazel, birch, blackthorn, Prunus avium/padus (wild/bird cherry), Maloideae, elm, Taxus (yew), and willow. The monolith section (0.58×0.10 m) was taken through the stratigraphy of the southern baulk of the site which comprised a base of natural boulders, a peat unit, the edge of the burnt mound, another peat unit, a thin clay, and finally a soil A horizon. In addition, a large section of the trough (C28, fill C35) was cut in situ and shipped to the laboratory (Fig. 5). The pollen diagram from the core profile (Supplementary Fig. S2) has a basal pollen zone (ISCP1) dominated by Pinus (pine) but this almost completely disappears at the level of the fulacht fiadh sediments. The pine is difficult to interpret due to its highly efficient dispersal of pollen, however, at approximately 30% TLP it is high enough to regard as of local or extra-local origin. The high common Polypodium (polypody) and other Pteropsida (ferns) but low Pteridium (bracken) also indicate that the site was at the edge of or close to a woodland. In zone ISC2 the spectra is dominated by grass, herbs and sedges indicative of grazed pasture or meadow. Oats (Avena) or wheat also appear at 32cm which corresponds with the fulacht fiadh sediments (context 21). The simplest interpretation of the data is that pine woodland was felled and/or burnt prior to the construction of the fulacht fiadh and the site or land nearby used for both cereal cultivation and pastoral activity. The Holocene decline in pine is highly diachronous in Ireland being earlier (c. 2400–2100 cal BC) in northern Ireland (Hall et al. Reference Hall, Pilcher and McCormac1994) but later in central and western Ireland where it can be as late as c. 600BP (Molloy & O’Connell Reference Molloy and O’Connell2007). Pine is also known to survive in refugia with locally favourable conditions long after its regional decline due to a combination of competition driven by climate change and human activity (Brown Reference Brown1988). Inchagreenoge is Early Bronze Age in date and this relatively early date helps explain the existence of pine woodland which must have been growing on the slopes surrounding the bog in which the two fulachtaí fiadh were constructed. Animal bones from the fill of this trough included large domestic ungulates (cattle).

Fig. 5 Inchagreenoge trough C28 plan showing stake-holes and location of the large in situ bulk sample

Charcoal data from Cragbrien (C21) indicates that hazel, alder, blackthorn, oak and wild/bird cherry were growing in the vicinity. The high-resolution pollen sequence (Fig. 6) is directly correlated with the base of the mound, and so start of its accumulation at 84 cm. The basal zone (CBP1) indicates wet alder-dominated woodland with both pine and mixed oak/hazel woodland. That the woodland was not completely closed and had gaps or clearings is indicated by the high Hedera (ivy) and the wide range of herb types present many of which are light demanding. The fall in alder pre-dates construction of the burnt mound, but by how much is unknown. The reduction in pine and oak are later and immediately pre-date the burnt mound level and an interesting increase/spike in willow (zone CBP2a). This could be due to these trees having been left by the streamside (25 m to the east) or possibly by willow branches with catkins (male) being brought onto the site. By the level of the mound in zone CBP2b the local area is open with scattered trees and the herbs are dominated by pastoral indicators. The apparent rise in pine in zone CBP2a is almost certainly an artefact of the opening up of the local canopy allowing far more regional and long distance pollen to reach the ground surface. The Hordeum (barley) type record is difficult to interpret as it could be Glyceria (reed grass) which is typical of such wet locations, however, the correspondence with oat/wheat type and Secale (rye) in zone CBP3 and early zone CBP4 suggests that the clearing was used for arable cultivation immediately after the deposition of the mound sediments. The construction of the mound also seems to be associated with a high grass to sedge ratio and a peak in Plantago lanceolata (ribwort plantain) suggesting grazing. Zone CBP3 largely post-dates the burnt mound levels and covers a reversal in the Poaceae/Cyperaceae (grass/sedge) ratio which suggests either/or a reduction in grazing pressure or an increase in local surface wetness. The final zone CB4 shows a peak in bracken suggesting a reduction in grazing pressure. The simplest interpretation of the vegetation sequence is: clearance of an area of alder carr and oak woodland close to the river at a time when there was still a largely wooded landscape. The burnt mound is then created in an open environment characterised by rough pasture and some heathland of bracken and heather. After the mound has gone out of use and is being buried by alluvial sediment, grazing pressure decreases and cereal cultivation increases only to be abandoned at the end of the sequence. This suggests that the construction of the burnt mound is not unreasonably associated with a pastoral economy but with some cereal cultivation in the vicinity.

Fig. 6 Craigbrien pollen diagram

The pollen analysis and plant macrofossils from Killescragh and Caraun More (A024/22 and A024/23, Hatton unpublished, Supplementary Data Table S1) reveal that the earliest peats analysed from site A024/22 (F22 & F8) indicate a local landscape dominated by open wet woodland comprising alder, birch, hazel, ash, bird cherry and yew. Pine and elm were also growing in the vicinity and may have occupied the well-drained slopes of the nearby esker ridges. The plant macrofossil assemblages from the earliest deposits analysed indicate the presence of open fen woodland with a diverse wetland understorey, which is likely to have occupied the low-lying area between the River Craughwell and the esker. The expansion of yew is thought to be associated with reductions in intensive woodland clearance in the Neolithic (O’Connell & Molloy Reference O’Connell and Molloy2001) and Iron Age although it is regarded as over-represented in pollen diagrams (Newman et al. Reference Newman, O’Connell, Dillon and Molloy2007). The possibility of immediate post-burnt mound anthropogenic activity in the area is also suggested by the presence of a single charred grain of barley and charcoal fragments of hazel, ash and elm, however, this context (F8), although above and close to the burnt mound (Killescragh 2), is not in direct stratigraphic association. Plant remains from the peat horizon underlying a wooden trackway (F21) suggest the continued presence of wet woodland, although some of the arboreal remains may relate to the trackway construction. Charcoal from the burnt mound deposit suggests that hazel and ash may have been used for fuel for fires associated with the feature. Plant macrofossils from the overlying peat deposit reflected open fen vegetation with little woodland cover. Samples from below a possible trough base at A024/23 indicate the presence of local wet woodland comprising alder, birch, hazel, and yew. Birch and hazel charcoal from the peat above the possible trough base may derive from fuel associated with the burnt mound.

At Coonagh West, pollen was extracted from six contexts from the excavation respectively top, middle, and base of the trough fill, and below the trough stones, deeper below the trough stones, and under a basal timber (Table 1, Supplementary Data Fig. S3). These were all counted to an unusually high count (>600 grains) in order to improve statistical inferences. The concentration of pollen and spores was high (20,000–100,000 grains ml) and very few showed signs of degradation or corrosion. The site has therefore almost certainly not suffered from differential preservation and the counts can be taken as an accurate reflection of the pollen and spore influx to the site. All six samples are dominated by trees, but this is where the similarity ends. It is apparent that the three levels beneath the stones and timber (CWG1, CWH1, CWF5) are all very similar to the point of having almost exactly the same percentage values of major types such total trees (77.7%, 77.5% & 77.9% TLP) and the same frequency order of trees. It is reasonable to suggest that these three samples came from an alder-oak-hazel dominated woodland surrounded by mixed woodland with a significant component of pine. The other samples have significant differences. Sample CWA1 is dominated by alder annd oak with high pine, hazel, elm, and ash. As with all the samples elm is surprisingly high (>4% TLP) and suggests that it was a significant component of the regional woodland (see later discussion). Birch and willow, although present, do not appear to be major components of the woodland, however, the Salicaceae, due to their dioecious biology, are generally under- and rather erratically represented in pollen diagrams even from hazel-willow carrs. A clearing is also indicated by high holly, which only flowers in sunlight and most prolifically at the edges of woodland, and a range of herbaceous pollen types including grasses and sedges but also Lactuceae (dandelions family), Chenopodiaceae (goosefoot family), ribwort plantain, Rumex acetosa (common sorrel) and bracken spores. A single grain of cereal type pollen (oat/wheat type) was also recorded. The closed nature and close proximity of the surrounding woodland is also indicated by the high levels of polypody and the occurrence of Chelidonium (celandine). Sample CWB1 is also dominated by alder and oak with high hazel, pine, and ash. The extremely high alder value (1910 grains counted) and the presence of clusters of alder grains clearly indicates that anthers had blown into the trough from either overhanging or adjacent trees. An unusual occurrence is Malus sylvestris (crab apple) at 1.1% TLP-A along with even higher holly than in sample CWA1 at 5.4 %TLP-A. The non-tree component is significantly higher at 27.9% TLP-A with unusually high levels of common sorrel at 10.4 %TLP-A, Filipendula (meadowsweet) at 5.1 %TLP-A and Ranunculus (buttercups) at 2.5% TLP-A. Cereals (oat/wheat type) are also high at 1.2 &TLP-A and one grain of Linum bienne type (pale flax) was recorded. Sample CWC1 is also dominated by alder and oak with high hazel, pine, and ash. Alder is still very high (865 grains counted) but no clusters were observed and both the herbaceous total at 19% TLP-A and unusually high holly (nearly 6% TLP-A) indicate a clearing or forest gap. The high herbs are again buttercup family, ribwort plantain, common sorrel and meadowsweet. Cereals are present, Cannabis type (hemp, 1 grain), pale flax type and Stellaria holostea (greater stitchwort). A similar gap or clearing was recorded at Leahy which also had unusually high Lonicera (honeysuckle; Brown et al. Reference Brown, Hatton and Davis2004). The remarkably high concentration of alder pollen in the lower levels of the Coonagh West trough suggests deliberate collection and deposition especially when we consider that it was possible that the trough was not cleaned out after its final use (Reilly Reference Reilly2010). Alder, common sorrel, and meadowsweet were commonly used in dyeing in Ireland.

Plant macrofossil analysis conducted on columns 1 and 3 from Coonagh West all produced a similar plant assemblage (Supplementary Table S2; Durham Archaeological Services 2007) and this can be compared with the pollen data from samples CWA1, CWB1, and CWC1 (Table 2). This is meaningful in this case because of the undoubtedly localised source area for most of the pollen and spores. There is a strong correspondence amongst both the common elements (e.g. alder, oak, etc) but also some rarities such as greater stitchwort. It reinforces the inference of a small damp grass-dominated clearing within wet/damp alder-hazel-oak woodland within the regional woodland with pine, elm, and ash. The composition of the local woodland matches unusually well the main wood types used in the construction of the trough with alder (51%), oak (23%), hazel (10%), Maloideae (8%), ash (5%), and willow (3%). In particular it is rare to be able to match less common species, in this case crab apple, which is also a potentially important prehistoric food resource. The slight mismatch with willow is probably due to the un-representivity of pollen influx value as mentioned earlier. The interesting difference is a lack of any elm or pine in the macrofossils, supporting the contention that they were growing in the region but not on the flood plain or in the immediate vicinity of the site.

The pollen sequence from Jigginstown is unusually complex given its limited depth (Fig. 7). The basal zone, JG1, almost certainly represents the end of the Lateglacial and very beginning of the Holocene and the succeeding zone, JG2, covers a period during which the Boreal woodland of the Early–Midle Mesolithic is established as is confirmed by radiocarbon dates. This woodland becomes richer after the appearance of hazel in zone JG4. There are indications of disturbance, both natural and later human, to this open and probably fragmented, woodland. Pollen zone JG5 covers a period when a clearing is created in the pine-hazel dominated woodland, perhaps with the aid of fire and this is initially used for grazing. Subsequently arable cultivation occurs in the close vicinity of the site. The definite presence of Linum catharticum (purging flax) in three consecutive levels at the base of the zone JG6 strongly suggests that it is being grown or processed near the site. It can be used as a yellow dye-plant but is also well known for its medicinal properties as it is anthelmintic (expels parasites from the gut), diuretic, emetic, and purgative. Macrofossil remains have been found at a number of Bronze Age sites in the British Isles including Knights Farm (Berkshire), Runnymede (Surrey), Stackpole Warren (Pembrokeshire), Wilsford Shaft (Wiltshire), and from several crannogs including Oakbank Crannog, Loch Tay, Tayside (Tomlinson & Hall Reference Tomlinson and Hall1996). The unusually high representation of Centaurea nigra (black knapweed) throughout zone JG6 (at over 5% TLP) could be due to the close proximity of a rough, damp knapweed-infested meadow or, alternatively, it may have been being encouraged/cultivated or brought onto site due to its well-known use in fabric dyeing (it produces a yellow dye). This is an unusual pollen and spore diagram, as it appears that oak dominated temperate woodland never developed around this site and the activities associated with the fulachtaí fiadh occurred in a pine and hazel dominated landscape. The evidence of purging flax and the anomalous levels of black knapweed suggest that one function, or at least usage, of the fulachtaí fiadh may have been related to the manufacture and dyeing of textiles.

Fig. 7 Jigginstown pollen diagram

At Ballygawley, a total of 2530 macroscopic charcoal identifications were carried out from features and deposits across all periods of site use. Alder and hazel were identified as the most abundant timbers used through each period with other arboreal taxa more popular as fuel wood in varying periods, eg, oak was utilised mainly in the Neolithic, while holly was resourced in the Late Bronze Age–Early Iron Age. Other tree and shrub types used for fuel include: wild cherry, Maloideae, willow, blackthorn, and ash. Ring-curvature shows fragments mainly had strong and moderately curved growth rings suggesting that branch wood was the main timber size used as wood fuel (Wheeler et al. Reference Wheeler, Timpany, Mighall and Scottin press). Fragments with weakly curved growth rings were present in low numbers which, together with the worked (waterlogged) wood from the trough linings, shows that large-sized timbers (eg, trunk wood) were also used on occasion. A total of 37 bulk samples were analysed for macroscopic plant remains from archaeological features and palaeochannel deposits. The overall assemblage from Ballygawley changes little chronologically, consisting largely of wet/damp loving plants indicative of fen conditions that are typical of stream side assemblages including: Viola palustris (marsh violet), Potentilla palustris (marsh cinequefoil), Potamogeton sp. (pond weed), and Persicaria hydropiper (water pepper), together with a range of sedge species. Evidence of the local woodland is limited with only Sambucus nigra (elder) fruits present, suggesting the area around the mound activity was cleared. The presence of pasture land near to the site is also indicated from the presence of a number of meadowland plants such as Persicaria maculosa (redshank), buttercups, Galeopsis tetrahit (hemp-nettle), Chenopodiaceae, and Stellaria sp. (chick weed). There is some evidence for cultivation during the prehistoric period with the presence of charred cereal grain of naked barley recovered from a pit (9539) associated with burnt mound (9009). The grains have been dated to the Middle Bronze Age (1610–1410 cal BC; SUERC-20608; 3215±35BP and cereal-type pollen is also present in the upper layers of Monolith 2 (LPAZ S1II4).

The pollen sequence from Ballygawley monolith 2 has been divided into four local pollen assemblage zones (LPAZ S1II1–4) (Fig. 8). The pollen diagram shows local alder carr with willow and birch, formed in the valley bottom, river banks, and flood plain terraces while dryland woodland of oak, hazel, and elm was present on the drier valley slopes and surrounds. The herbaceous pollen assemblage shows the field layer consisted mainly of grasses and damp grassland taxa such as sedges, Peucedanum type (probably wild angelica), and Urtica sp. (nettles). Within LPAZ S1ii1 and at the beginning of LPAZ S1ii3, regular fluctuations in the pollen of oak, alder, and hazel appear as a zig-zag pattern during and following the end of activity at burnt mound 9031. It is suggested this zig-zag pattern reflects some form of woodland management (eg, coppicing cycles) of the local woodland in association with the sampled burnt mound and one 40 m to the south-west (9034), dated to 2460–2200 cal BC (SUERC-20634; 3850±35BP). This potential pattern of managed woodland exploitation has been replicated at another burnt mound site at Roughan, Co. Tyrone. This zig-zag pattern seen in the arboreal and shrub pollen is absent within some levels from the burnt mound in LPAZ S1II2, and after the mound material has been deposited (upper LPAZ S1II3 and LPAZ S1II4); this is likely to be an effect of poor preservation through the mound material and counting at increased intervals in the upper levels. Poor pollen preservation might also explain the decrease in total arboreal pollen from 103 cm to 95 cm although a phase of woodland clearance cannot be discounted. Any impact on woodland cover is short-lived as arboreal pollen percentages recover to close to their pre-burnt mound values during the early stages of LPAZ S1ii3. Peaks in microscopic charcoal of all sizes (<21–>50 µm) are seen during the levels of the burnt mound activity reflecting the local burning taking place. Non-pollen palynomorph and spore data from levels in the lower and upper parts of burnt mound material show peaks in the presence of Gloeotrichia (HdV-146) implying increased levels of eutrophism at the site, which could be the result of water used in the troughs for cooking or other purposes, or from stagnant water lying close by. Sphagnum was probably used to line the troughs and its spores peak in the burnt mound material. Peaks in coprophilous fungi Cerocophora-type (HdV-112) and Sordaria-type (HdV-55A) are also recorded during and following the burnt mound activity suggesting the presence of animals around the site.

Fig. 8 Pollen and spore diagram from Monolith 2 at Ballygawley

Beetles (Coleoptera)

Coleoptera fragments were high enough for meaningful analysis throughout the Cahiracon monolith M7 (Supplementary Data Table S3). The basal sample (0.3–0.5 m) was the most diverse of three with the assemblage comprising 102 individuals from 45 taxa. A number of taxa characteristic of swampy or waterside environments are present (15%). These include the scirtid Cyphon sp., the pselaphid Reichenbachia juncorum, and the cryptophagid Paramecosoma melanocephalum. The water beetles present (Hydraena riparia, Anacaena globulus) strongly suggest a slow-flowing water body and the most abundant ecological grouping in this sample is indicative of accumulations of foul, rotting organic matter (c. 20% of all individuals present). These include a wide variety of staphylinids such as Micropeplus fulvus, Tachinus laticollis, Philonthus varians, and Rugilus rufipes. In addition, four dung-specific taxa are present, including two species of Aphodius (A. contaminatus and A. sphacelatus) with a preference for the dung of cattle or horse (Koch Reference Koch1989), numerous fragments of Geotrupes (at least one individual), and the staphylinid (Anotylus sculpturatus) which may also inhabit carrion. Two woodland edge taxa seen in the other samples are again present; Stenichnus collaris and Othius punctulatus. In addition three genuine woodland taxa are recorded; the relatively non-specific weevil Strophosoma melanogrammum, the weevil Curculio pyrrhoceras, which is oligophagus on oak, its larvae living in the galls of the oak gall wasp Cynips quercusfolii (Koch Reference Koch1992), and the cossonine weevil Mesites tardii. This is the largest species of cossonine weevil in the British Isles and has a preference for firm dead wood in coastal locations (Warren & Key Reference Warren and Key1991), so it may result naturally from the proximity of the site to the Shannon Estuary or from the import of driftwood for fuel. M. tardyi seems surprisingly common in Irish prehistoric samples compared with the UK where it is almost entirely driftwood related. Why this might be is not clear. In addition the presence of the weevil Anthonomus rubi is characteristic of woody members of the Rosaceae, probably Rubus fruticosus agg. (blackberry) of which a number of seeds were evident in sorting.

The sample stratigraphically above (0.2–0.3 m) contained a moderately diverse fauna (62 individuals from 37 taxa) relative to sample size. Again, the principal component of the assemblage indicates a swampy or stream-side locality as is once again indicated by the inclusion of members of the Pselaphidae and Carabidae of the genus Pterostichus. The water beetle fauna present is characteristic of slow-moving water bodies (Anacaena globulus, Helophorus brevipalpis, Hydraena riparia). The staphylinid Olophrum consimile is of particular interest in that it has traditionally been considered an upland species in the British Isles (eg, Joy Reference Joy1932; Tottenham Reference Tottenham1954). However, a number of continental European studies suggest it is often also present at the edges of lakes, ponds, bogs, and streams with luxuriant vegetation (eg Böcher Reference Böcher1995; Koch Reference Koch1989). Several taxa present are characteristic of woodland or woodland edge environments including the staphylinid Othius punctulatus, the carabid Bembidion harpaloides (which mainly lives under bark of logs and dead trees; Luff Reference Luff1998), and the cantharid Rhagonycha femoralis. The only obligate woodland species is once again the non-specific, deciduous tree dwelling Strophosoma melanogrammum. Four taxa characteristic of rotting vegetable matter are present; the hydrophilids Megasternum boletophagum and Cercyon sp. and the staphylinids Philonthus sp. and Tachyporus hypnorum. Dung beetles of the genera Geotrupes and Aphodius are present, the latter represented by A. fimetarius which may be found in dung and more rarely in compost and stable manure heaps and rotting vegetation, especially Brassicaceae (Koch Reference Koch1989). A single phytophagous species is present; a weevil of the genus Gymnetron, the majority of which favour plantains as a host plant. The uppermost sample (0–0.1 m) also contained a moderate to highly diverse fauna (97 individuals from 42 taxa), strongly dominated by taxa of swampy or stream-side localities (eg, the pselaphids Bryaxis bulbifer, Brachygluta haematica, Reichenbachia juncorum, and the carabids Pterostichus anthracinus, P. gracilis, and P. vernalis), comprising approximately 30% of all taxa present. The limited water beetle community indicates a slow flowing water body (Hydraena riparia and Anacaena globulus both prefer such environments) with vegetation probably including Scirpus lacustris (common club-rush), upon which the larvae of Limnobaris t-album feed (Hoffman Reference Hoffmann1954). Some of the taxa present suggest local woodland or the presence of woodland litter, such as the scydmaenid Stenichnus collaris, which is usually found in damp woodland or woodland margins and the staphylinid Lordithon exoletus which lives on agaric fungi in similar environments (Koch Reference Koch1989). Two taxa are strongly associated with woodland; the elaterid Prosternum tesselatum which has a strong preference for oak and pine (Laibner Reference Laibner2000), both of which are common within the pollen diagram at this time, and the weevil Strophosoma melanogrammum which lives upon a wide range of deciduous tree taxa. Several taxa indicate the presence of foul, rotting organic matter (eg, the staphylinids Gyrohypnus punctulatus, Proteinus ovalis, and Tachinus laticollis, in addition to members of the Hydrophilidae). This may have been in the form of dung, as fragments of two Aphodius dung beetles were present (too degraded to identify to species). Also present are two carabids characteristic of open localities which exhibit a preference for cultivated ground; Bembidion obtusum and Trechus quadristriatus. This is in concurrence with the pollen diagram which records both barley- and oat/wheat type within this zone (CC7P5).

All three samples have a broadly similar character in that they imply a swampy area in proximity to both a slow moving water body and an area of woodland (Fig. 9). More specifically the upper two samples both include some taxa of open cultivated ground which, in conjunction with the pollen evidence, strongly suggests arable cultivation. This element is absent from the lower sample, which has a more extensive dung and refuse fauna, possibly suggesting a switch from a pastoral system with the presence of cattle or horses in the vicinity to a mixed system. Also, marginally more woodland taxa are present towards the base of the column, which is in keeping with the pollen evidence of local woodland. Also in agreement is that this woodland included oak as is indicated by Curculio pyrrhoceras. This beetle is associated with oak galls which have a long use of history in the cloth dyeing industry. It is possible that the cossonine weevil Mesites tardii would have arrived on the site as a result of human activity, as its preferred habitat is driftwood which, given the apparently regular burning of such sites, may well have been an exploitable resource.

Fig. 9 Sample by sample ecological comparison for site Cahiracon 3/37/7(M7), Killesraigh. and Coonagh West site A005 2021 and (b) Ecological categories are: WS (Slow water), WR (Running water), T (woodland), SI (Silvicolous), REF (Refuse/rotting vegetable matter), PD (Pasture/dung), M (Meadow), LATH (Lathridiidae – mould beetles), GR (Grassland), AM (Aquatic/marsh) and AD (Arable/disturbed)

The Coleoptera from the Inchagreenoge monolith was sparse but the sample from 0.57–0.42 m included taxa of slow waters (Agabus bipustulatus and Helophorus brevipalpis), emergent vegetation (Donacia sp.), and shaded locales (Olophrum sp.). The leg of an Aphodius dung beetle is present. In the sample from 0–0.2 m depth one Dytiscid head (Agabus sp.) suggested slow water and there were carabids of open ground (Pterostichus sp.) and hygrophilous staphylinids (Stenus sp.). This is in agreement with the pollen and spore diagram but due to the low concentration and poor state of preservation prohibited further inference.

The coleoptera from the earliest peats from Killescragh and Caraun More (A024/22) are suggestive of a predominantly wet, but open environment with occasional trees (Fig. 9). One sample in particular demonstrates the presence of woody Rosaceae and possibly elm. The predominant vegetation type appears to have been aquatic grasses, with the field chafer, Phyllopertha horticola present in each of the three samples analysed from this phase. The presence of large herbivores is indicated in some samples, but the evidence does not suggest these were abundant or that pastoral agriculture represented a factor in the landscape at this time. Of the three samples analysed contemporary with the burnt mound, two indicate a more or less open flood plain landscape, the phytophages indicating plant taxa capable of rapid regeneration (e.g. rosebay willowherb: Chamerion angustifolium) and disturbed conditions (eg, ribwort plantain). Grazing indicators are sparse and the overall impression is of a landscape not dissimilar to that indicated in samples prior to mound construction, perhaps wetter with the elaterid Prosternon tesselatum living at distance on higher ground. However, the third sample analysed from this phase, which underlay a trackway remnant, provided a wealth of information regarding local and extra-local woodland at the site. In addition to the suite of aquatic and marshy taxa present in the majority of the other samples, it included five woodland taxa no longer found in Ireland, including the first record here of elm bark beetle, Scolytus scolytus, two species of Cerylon, and the chrysomelid Oomorphus concolor, indicative of ivy. It is probable that many of these woodland indicators arrived with the construction materials for the trackway itself, but they provide a picture of a diverse primary local woodland with oak, ash, willow, elm and Prunus/Crataegus. The presence of this diverse woodland assemblage has obvious implications for the availability of fuel for the mound itself; despite being situated in a largely cleared local landscape, diverse secondary woodland was still present in close enough proximity to allow its harvest for trackway building. The post-mound assemblages comprise what appears to be a more or less cleared flood plain environment, not unlike that observed in the pre-mound state but with some indication of local woodland or occasional tree presence. The moderate diversity of dung taxa and the presence of indications of ribwort plantain suggest a disturbed and grazed environment, typical of many later period lowland flood plains. Two assemblages recovered from this phase include a high proportion of fast water taxa and possible allochthonous taxa (ie, dryland species within a predominantly wetland assemblage) and may represent increased severity of floods.

Only one sample (Feature 27; Sample 43) from Caraun More produced insect remains. The sample yielded a sparse assemblage of 24 poorly preserved individuals from 14 taxa. This was dominated by the throscid Trixagus dermestoides, present as a total of eight individuals. T. dermestoides is primarily a woodland taxa, often found in clearings and leafy debris. The material processed, while devoid of leafy remains, clearly incorporated a large proportion of fine woody debris, reminiscent of wood shavings. The assemblage also included a few aquatic taxa of slow or stagnant waters, two dung beetles (one Aphodius sp. and one Geotrupes sp.), and a chrysomelid of the genus Longitarsus. Interestingly the assemblage also included a small potentially synanthropic element. A single Ptinus sp. was recovered, many species of which are strongly synanthropic, alongside two individuals of the Lathridius minutus group. L. minutus, while typically synanthropic, is not an obligate synanthrope and may be recovered from woodland debris where it lives upon fungal hyphae. As such, its presence in this assemblage appears to relate to the decomposing wood ‘shavings’ which were home to Trixagus dermestoides. It is conceivable that this debris was created during working of wood for trough construction.

Two pre-mound samples from the palaeochannel fill stratigraphically below a trough at Coonagh West were analysed and found to be extremely rich beetle assemblages. Numerically these were dominated by taxa of slow or running waters, with taxa of fast-flowing waters conspicuously absent (Fig. 9). Refuse taxa were well represented, with Anotylus rugosus the most common species of this ecotype. A range of the taxa represented are typical of a waterside location and dung taxa are present in frequencies suggesting moderate presence of large herbivores. Numerically these are dominated by Aphodius sphacelatus/prodromus type, but the presence of A. depressus is suggestive of a shaded, possibly woodland environment. In contrast, A. merdarius is predominantly a taxon of open environments with a preference for dung of cattle, though infrequently found in woodland margins; this suggests the local presence of woodland, possibly providing considerable shade in places. Of particular interest is the diverse woodland assemblage preserved within these samples. These include a range of silvicolous taxa exemplified by the carabids Nebria brevicollis and Agonum assimile. This latter is a nocturnal woodland floor predator, resting during the day under loose bark (Stork et al. Reference Stork, Hammond, Russell and Hadwen2001). The most common woodland taxon present is the large cossonine weevil Mesites tardyi. Other woodland taxa included two species of bark beetles (Scolytidae), Scolytus mali and S. rugulosus, neither of which are currently recorded in Ireland, although S. mali has been recorded from early medieval Dublin (cf. Whitehouse Reference Whitehouse2006); both of these taxa are typical of woody plants of the Rosaceae, earning S. mali the English name of the orchard bark beetle. Other woodland components indicated include oak (the weevil Curculio pyrrhocerras lives upon oak galls) and ash (of which the Scolytid Leperisinus varius is typical). Several taxa suggest an abundance of local dead wood, including the Eucnemid Melasis buprestoides and the Cucujid Pediacus dermestoides, which is characteristic of freshly cut or broken tree stumps but rare in Ireland (Alexander Reference Alexander1994). Several taxa present suggest the local presence of grassland or pasture. These include the Elaterid Adrastus pallens, usually found on stream and river meadows, in pastures and at woodland margins, in addition to many individuals of the Carabid Trechus quadristriatus, typical of open, rather dry country with short vegetation, including agricultural land. Non-woodland phytophagous taxa are relatively sparse, but include the Nitulid Brachypterus glaber, characteristic of stinging nettle and hence increased nitrogen input, the Chrysomelid Hydrothassa marginella, found on a range of plants by fresh water (including Caltha palustris (marsh marigold) and buttercups) and the weevil Thryogenes festucae, found in sedges in waterside locations. In summary, the assemblages while diverse (particularly as regards the woodland element), are not unlike many modern flood plain assemblages, indicating an area of mixed oak woodland, possibly secondary in nature (suggested by the presence of ash), adjacent to a largely cleared flood plain environment. This may have been agricultural pasture (implied by the presence of nettles and some dung taxa), possibly with the presence of some cattle.

Two samples investigated from a trough fill at Coonagh West (C395 & C366) were dominated by taxa of slow-moving waters or waterside locations. A few woodland taxa are present, including a single woodworm (Anobium punctatum) and the weevils M. tardyi and Strophosoma melanogrammum. The latter common, feeding on the leaves of a wide range of tree species (Harde Reference Harde1984). Phytophages are restricted to a single individual of H. marginella and the weevil Apion hydrolapathi, the latter of which is oligophagous on Rumex spp. (docks). This small assemblage possessed a fairly generic waterside character, though the presence of some dead wood and possible local woodland is implied. The reduced woodland component of the fauna is in keeping with it being of potentially later origin. While elements of this assemblage may originate from the trough itself (ie, the dead wood component), the fauna probably derive in part from post-use infilling resulting from overbank sedimentation.

Two samples from the post-mound phase of the Coonagh West site yielded worthwhile assemblages. In contrast to previous samples discussed, the aquatic fauna represents a relatively small fraction of this which, coupled with indifferent preservation, suggests a potentially drier depositional environment. However, several waterside taxa are present and a number of woodland taxa, once again including M. tardyi and Leperisunus varius. These are accompanied by Cerylon histeroides and Scolytus rugulosus, primarily a taxon of woody Rosaceae. Several taxa also suggest the presence of local open ground, including the field chafer, Phyllopertha horticola and dung beetle Aphodius prodromus/sphacelatus. Despite comprising an appreciably smaller assemblage, the overall impression is of an environment not unlike that of the pre-mound conditions; a wooded area (including ash and woody Rosaceae) adjacent to a partially cleared flood plain. This suggests that, following mound use, either some woodland remained or that this deposit represents a significantly later accumulation following growth of secondary woodland. The construction and use of the trough/mounds appears to be marked by a change in the beetle fauna that is indicative of high woodland component being replaced by more grassland and refuse species (Fig. 10). This is a more pronounced picture of the changes to the forest than is given by the pollen and plant macrofossils. This is probably due to the bias towards very local changes that is typical of beetle assemblages, whereas the extra-local and regional component of the pollen remains a major part of that spectra. No carrion beetles were noted and dung beetles were not high enough to support any concentration higher than some grazing, such as the corralling of stock at the site. It is clear from these case studies that synanthropic taxa are conspicuous by their absence and statistical analyses reveals that the burnt mound samples cluster well together but show little overlap with a variety of modern and ancient samples (Fig. 10). Woodland taxa are at times extremely common and diverse, reflecting the likely presence of remnants of primary woodland at the time of monument construction, becoming more fragmentary over the course of mound use. However, canopy taxa are sparse, suggesting that many of these woodland taxa may be arriving with fuel rather than directly from overhanging trees. Dung beetles are, in comparison to late Holocene flood plain assemblages, poorly represented and indicate low stocking levels (or indeed the complete absence of pastoral agriculture) in the regions in which these sites were constructed and used.

Fig. 10 a. DCA analysis sub-dividing the woodland component; b. Minimum Variance Cluster Analysis: burnt mounds and other sites (including modern sites)

Thirty-seven samples from Ballygawley were analysed for insect remains, again taken from archaeological features such as wicker trough lining and palaeochannel deposits. The overall assemblage is similar to that of the macroscopic plant remains including many stream side indicators, such as Colymbetes fuscus and Hydroporus palustris together with taxa indicative of the presence of damp/wet plants including Sitona hispidulus and Leiosoma deflexum (Tetlow & Davis Reference Tetlow and Davis2009). The occurrence of insects, such as Geotrupes sp., Aphodius sp., and Megasternum obscurum, associated with decaying matter (probably decaying vegetation) and sometimes dung, is of interest and suggests that animals were present around the site (ibid.). The potential presence of domesticates is also shown in the Bronze Age faunal bone assemblage with cattle, some pig, and occasional sheep/goat recovered. Bones (200+) were largely recovered from the palaeochannel fills, eight showing evidence of butchery. Tools related to butchery were also recovered from the palaeochannel fills including two bone points for hanging skins, 12 lithic scrapers, and two stone knives.

Multi-element analyses

The aim of multi-element analysis was to detect any anomalous enhancement of trace elements that might indicate any human activity such as tanning, food processing, or preserving, or any industrial activities, all of which can be reflected by trace element accumulations. The bulk sample of the Inchagreenoge trough (C28) was dissected and six samples of wood and three of sediment taken, washed (if wood) and then digested in nitric acid before dilution and analysis by ICP-MS. This use of ICP-MS is based upon the comparison of the sediment as control samples with wood from various parts of the inside of the trough (Fig. 11). The control samples come from below, inside, and outside the trough and should represent the geological background of the site. The principle is that wood values less than the control represent dilution, as would be expected, whilst values above the control samples represent accumulation or enhancement due to an external addition at some point in the past. As might be expected many elements are higher in the control samples, especially the underlying sediment. These include; strontium (Sr), molybdenum (Mo), cadmium (Cd), antimony (Sb), barium (Ba) and uranium (U). These are mostly elements that are derived from the geology, probably from the Carboniferous Limestone which underlies the site. However, there are a small number of elements with higher concentrations in the wood than in the control samples. These include sodium (Na), titanium (Ti), chromium (Cr) and arsenic (As). With the exception of sodium, which could be of meteoric (rainfall) and/or groundwater origin, these elements are typically associated with either tanning or the dyeing of textiles. Multi-element analysis was also undertaken on 21 samples from the trough at Coonagh West using a 3D grid of samples (Supplementary Data Fig. S4). Five of the more common metals lie in general within the normal range found in UK soils, however, the higher values of lead (Pb) and zinc (Zn) lie well above this range (Table S4). Started in 2000 the Irish Soil Database (Fay et al. Reference Fay, McGrath, Zhang, Carrigg, O’Flaherty, Kramers, Carton and Grennan2007) now provides comparative values of heavy metal concentration from across Ireland. As can be seen from Table 3 the values for five of the more common metals lie at the upper end, or above in general within the normal range found in UK soils exceeding the median values for Irish soils and for some (Cr, Cu, Pb) exceeding the highest natural values found. However, both at Inchagreenoge and Coonagh West the values of Pb and Zn also lie well above the range of values from ancient agricultural field systems in Scotland (Wilson et al. Reference Wilson, Davidson and Cresser2008) and Zn and Cu lie above the levels found in sewage effluent sludge within the Shannon catchment (Reid et al. Reference Reid, Brougham, Fogarty and Roche2009). A full analysis of all the 44 elements analysed will only be possible when more background soil and geological data becomes available for this region of Ireland, but at Coonagh West and Inchagreenoge the anomalously high values of Pb and Zn are most easily explained by the importation of soil from an area or areas with intrusive igneous rocks which have undergone mineralisation, or some form of bio-accumulation related to human activities.

Fig. 11 ICP-MS multi-element analysis of Inchagreenoge (3/45/1) trough (C28). E1–E6 Trough fill, E9W & E12 Wood and other samples control/sediment

An initial analysis of high outlier concentrations from the 3D sampling suggests that some samples are enriched relative to others. The atypical samples appear to be CWA6, CW1C, and CW2C which are low. What is noticeable about these is that they are all derived from the basal levels of the trough fill (level C) and could be expected to be susceptible to groundwater leaching. These are all significantly lower in Cr, than all the other samples. However, the samples above (H, G, D, & B) have higher concentrations of Co, Zn, Pb, Mo, and in one case Cu. This suggests that there is some depth-related structure to the data, particularly in Co, Zn, and Mo, mean concentrations of which are respectively higher than at level F (below the timbers). The atypical sample here is CW6 A and, given that this is a near-surface (top of trough fill) sample, it probably reflects modern pollution and iron mobility.

Heavy metals occur naturally in zones of mineralisation (Zn, Mo, Cu, Pb) and Ti, As, and Cr occur in titano-magnetites, arseno-pyrites, and serpentinites. Serpentine bearing rocks are relatively common in Ireland, occurring in the Dalradian Southern Highlands Group in Tyrone, north-west Mayo and in north Connemara, and in the Appin Group in the Connemara green marble (Daly Reference Daly2001) as well as in association with Tertiary igneous activity in north-west and Northern Ireland (Preston Reference Preston2001). Cr is known to be a particularly effective mordant which has been used inadvertently as part of the use of soil as a mordant with natural dyes. Heavy metals are particularly suitable for the dyeing of wool, after thorough washing as possible, and with acid dyes such as that derived from oak galls or leaves which contain tannic acid (Liese Reference Liese2004). The Shannon catchment is the largest catchment in Ireland draining over 12,000 km² and both Inchagreenoge and Coonagh West are located on the flood plain at its outlet. The geology of the catchment is dominated by Carboniferous (Dinantian) limestone with small areas of Ordovician rocks and, at its downstream end, Devonian and Silurian sandstones and siltsones (Holland Reference Holland2001). The whole basin is covered by glacial deposits varying from till to glacial sands and gravels. The catchment includes very little of the highly mineralised areas of Ireland such as the south-west peninsula or the Tertiary igneous provinces which are high in Cu, nickel (Ni), Zn, and other metals and for most heavy metals the area has relatively low values by comparison with the rest of Ireland (Fay et al. Reference Fay, McGrath, Zhang, Carrigg, O’Flaherty, Kramers, Carton and Grennan2007). The regions nearest to the lower Shannon with higher heavy metal values include the Galway/Burren area to the north and the Galtee Mountains to the south (ibid.). Given this size, and its mixed sedimentary lithology, outlet sediments would be expected to be relatively low background levels in trace elements and heavy metals. This may be a coincidence, caused by the concentration of these elements in fine alluvial sediments washed into the troughs or it could represent some importation of a foreign soil which had particular properties valued by Bronze Age people. The alternative, which may be more plausible, is the concentration of these elements in peat ash or urine which was collected in urine pots as a mordant in the medieval period particularly used for dyeing using weld (Priest-Dorman Reference Priest-Dorman2002). The most obvious use would be for either tanning and cleaning animal skins or for the washing and processing of wool with the soil, ash and/or urine being used as a mordant. Use of troughs for a combination of mordant and vat dyeing (as is known from Europe; Joosten et al. Reference Joosten, Maarten, van Bommel, Hofmann-de Keijzer and Hans Reschreiter2006) is consistent with the low-level of vegetation disturbance, the accumulation of some organic materials, some agriculture, and the multi-element analyses to date.