Land use history of Village Bay, Hirta, St Kilda World Heritage Site: A palynological investigation of plaggen soils

Review of Palaeobotany and Palynology 153 (2009) 46–61

Contents lists available at ScienceDirect

Review of Palaeobotany and Palynology j o u r n a l h o m e p a g e : w w w. e l s e v i e r. c o m / l o c a t e / r e v p a l b o

Land use history of Village Bay, Hirta, St Kilda World Heritage Site: A palynological investigation of plaggen soils Margaret P. Donaldson a, Kevin J. Edwards b,⁎, Andrew A. Meharg c, Claire Deacon c, Donald A. Davidson d a

School of Geography and Geosciences, University of St. Andrews, St. Andrews KY16 9AL, UK Departments of Geography & Environment and Archaeology, University of Aberdeen, Elphinstone Road, Aberdeen AB24 3UF, UK Department of Plant and Soil Science, University of Aberdeen, Cruickshank Building, St. Machar Drive, Aberdeen AB24 3UU, UK d School of Biological and Environmental Sciences, University of Stirling, Stirling, FK9 4LA, UK b c

A R T I C L E

I N F O

Article history: Received 21 January 2008 Received in revised form 13 June 2008 Accepted 18 June 2008 Available online 1 July 2008 Keywords: pollen plaggen soils manuring St Kilda Scotland

A B S T R A C T This paper presents findings based on a palynological investigation of artificially accreting (plaggen) soils from the settlement of Village Bay, Hirta, in the St Kilda archipelago, which was perhaps the most distant and inhospitable outpost of sustained human habitation in the British Isles. The soils were developed principally through the addition of turf ash and seabird waste, although some ash may have been derived from upland peats. It is assumed that the woodland pollen signal (much lower in the soils than in an upland peat site nearby) represents off-island sources. Corylus avellana-type pollen (frequent in upland sites), along with Potentilla-type, may provide markers in the Village Bay profiles for the addition of ashed hillside turf, and possibly peat, to the plaggen soils. Cereal-type pollen is well represented through the profiles and is often strongly associated with the record for Chrysanthemum segetum (corn marigold), a frequent indicator of arable land. The Brassicaceae signal may partly reflect the cultivation of cabbages; Chelidonium majus (greater celandine) may have been grown for medicinal use. Soil mixing has rendered radiocarbon dating meaningless at this site, but the establishment of a change in cultivation regime before AD 1830 may have been identified from the patterns of pollen concentration and preservation in the profiles. © 2008 Elsevier B.V. All rights reserved.

1. Introduction The St Kilda archipelago lies 55 km west of North Uist in the Outer Hebrides and 160 km west of mainland Scotland (Fig. 1A). Possibly for millennia, the main island of Hirta, with an area less than 6.5 km2, represented the most distant and inhospitable outpost of continuous human habitation in the British Isles (Steel, 1994). The sea cliffs of Hirta, the highest in Britain at over 300 m, are just visible on a clear day from vantage points on Harris and North Uist in the Outer Hebrides, serving to emphasise the isolation and relative inaccessibility of the island group. The population of Hirta was around 180 at the time of Martin's visit in AD 1697 (Martin, 1994), although numbers had declined to 76 by 1877 due to emigration, disease and losses by drowning (Seton, 1878). By the 1920s the population was struggling to retain viability and in 1930 the remaining 36 islanders were evacuated to the mainland at their own request (Steel, 1994). The seabird colonies of St Kilda are some of the largest in the world for gannets, puffins and fulmars. The St Kildans utilised this resource, supporting themselves principally as a seabirding economy; the heavy Atlantic swell limited fishing opportunities. Eggs were taken from the

⁎ Corresponding author. Tel.: +44 1224 272346; fax: +44 1224 272331. E-mail address: [email protected] (K.J. Edwards). 0034-6667/$ – see front matter © 2008 Elsevier B.V. All rights reserved. doi:10.1016/j.revpalbo.2008.06.005

cliffs and birds were captured for meat and oil. All parts of the birds were used; eggs and meat were stored and eaten, feathers and oil were used to pay rent (Harman, 1997). Waste materials such as viscera and bones were added via midden pits to the artificially accreting (plaggen) soils. Such soils are typically amended for cultivation by the repeated addition of organic and mineral material from a variety of sources, e.g. turf, peat, burnt organic material and ash, seaweed, animal and human waste (Groenman-van Waateringe, 1992; Davidson et al., 2007). A review of agricultural land use was undertaken by Harman (1997) using information from records dating back to 1595. Sheep and cattle formed the core of pastoral activity with Seton (1878) noting the entitlement as 1200 sheep and 50 cattle. However, up to 2000 sheep were recorded on the three main islands of the group, providing meat, milk and wool, some of which was paid as a rental contribution to the steward (Harman, 1997). There were around 90 cattle at the time of Martin's visit in 1697 (Martin, 1994), but later visitors noted fewer than this (Harman, 1997). Cattle were kept inside during the winter providing a source of manure for cultivation. Both sheep and cattle were pastured at Gleann Mor (Fig. 1A) during the summer to protect the crops at Village Bay where arable activity was confined over recent centuries (Figs. 1B and 2). The remains of cultivation strips extending up- and downhill from the single row of houses on Main Street, Village Bay, can still be seen. The principal crops recorded were barley and oats (Martin, 1994; Harman, 1997);

M.P. Donaldson et al. / Review of Palaeobotany and Palynology 153 (2009) 46–61

47

Fig. 1. A: Hirta, the principal island in the St Kilda group, showing the location of Village Bay, Gleann Mor and Conachair; B: detail of the Village Bay settlement showing the location of soil profiles 2, 5, 6 and 8. M.O.D. is the Ministry of Defence military base.

48

M.P. Donaldson et al. / Review of Palaeobotany and Palynology 153 (2009) 46–61

Fig. 2. View of Village Bay, Hirta from the northwest (cf. Fig. 1), with north to the left side of the photograph. The Consumption Dyke runs from left to right, from the houses to the coast, in the right centre of the photograph.

potatoes were introduced in the second half of the 1700s. Other vegetables such as cabbage and turnip were occasionally noted. The formerly cultivated area represents the pattern of land use established when the village was rebuilt in its present location in the early 1830s (Small, 1979); the houses shown (Fig. 1B) date from 1862. Parts of the 1830s cultivation footprint were undoubtedly in use prior to that time (Fleming, 2005) although there is some uncertainty about the exact location of the earlier village. There is some evidence to suggest that the pre-1830s village stretched from the upper part of the military base (Fig. 1B) in a northwesterly direction (MacGregor, 1960) allowing the same broad area of land to have been utilised. Lazybeds, suggestive of an earlier cultivation regime, were identified on land now occupied by the western part of the Ministry of Defence (M.O.D.) base (MacGregor, 1960), but that area was not available for sampling for this work. This paper is concerned with a palynological investigation of plaggen soils from Village Bay. The work was undertaken as part of a wide-reaching study into the potential sources of pollution affecting the arable activities of an isolated seabird-dependent island community (Meharg et al., 2006; Davidson et al., 2007). The results of the chemical investigations showed the accumulation of high levels of Pb and Zn contamination in soils due to traditional manuring practices, with peat/turf ash as the primary source of contamination rather than additions from high levels of Zn in seabird bones. The bird economy of the island, allowing a high human population density to be sustained on a limited, and naturally poor, soil resource, was responsible ultimately for the greatly elevated levels of toxic metals in the soils. Previous pollen studies relating to the island of Hirta focused on areas away from Village Bay and provide little detail on past cultivation practices. Three samples from beneath Plantago sward in northwest Hirta were reported in Poore et al. (1949) who suggested that the peat, in excess of 90 cm depth, had developed from vegetation similar to that present at the time. The arboreal content (Betula, Alnus, Pinus and Corylus) of one sample totaled 8% TLP, and this, together with a ‘discovery of small wood’ (Ibid., p. 97) was interpreted as evidence of the former presence of trees on Hirta. McVean (1961) published limited data from three upland areas. Percentages of arbo-

real taxa reached 30% at two of the sites and the presence of birch– hazel scrub on the island during the ‘climatic optimum’ was inferred. Walker (1984) analysed and dated a core from Gleann Mor (Fig. 1A). Although some of the radiocarbon dates proved problematic, he discounted McVean's interpretation of birch–hazel scrub (as opposed to the possibility of small numbers of such taxa in sheltered localities), identifying instead open communities dominated by grasses, sedges, heaths and plantain sward during the last 6000 yr. In the upper levels of the core, speculatively dated to the Little Ice Age, Walker found very small amounts of cereal-type pollen which he attributed to windblown inputs originating at Village Bay as the settlement in Gleann Mor was probably abandoned by the fourteenth century. Meharg et al. (2006) published a selected taxa pollen diagram from Conachair (Fig. 1A), reproduced for information as Fig. 7. The Conachair core is dated from 5180 ± 40 BP at the base and the pollen spectra are indicative of generally open communities characterised by grasses, sedges and Potentilla-type (cf. tormentil) with Rumex acetosa in the lower levels and Calluna from zone Con-5b upwards. The arboreal pollen signal reaches values of 30% in zone Con-5, dating from 4380 ± 40 to a hiatus at 3270 ± 40 BP, which was suggested as evidence for peat-cutting prior to 1690 ± 40 BP; the woodland component was attributed to off-island sources, especially as pollen concentrations were very low. Above the hiatused section, the enhanced charcoal to pollen ratios and the appearance of cereal-type pollen were presented as evidence of domestic and arable activity on Hirta. In undertaking a study of the best known cultivated area of Hirta, the pollen and spore content of samples from four soil profiles at Village Bay was determined in order (1) to provide a context for variations in soil chemistry that are discussed in the associated papers (Meharg et al., 2006; Davidson et al., 2007), and (2) to test whether past cultivation practices could be detected within the pollen spectra. This paper focuses upon the second of these aims. This research not only augments the corpus of palynological material available from St Kilda, but presents findings that are felt to justify the pollen-analytical investigation of problematic soil materials. The potential complexity of the pollen records presented here is fully recognized; nonetheless we believe they are interpretable with caution.

M.P. Donaldson et al. / Review of Palaeobotany and Palynology 153 (2009) 46–61

2. Sites The soil profiles were excavated from within the formerly cultivated area of Village Bay at altitudes of 20–30 m on the seaward side of Main Street (Fig. 1B). While three of the profiles extended from the base of the plaggen soil to the surface, profile 8 was excavated from directly beneath the Consumption Dyke. This substantial wall, formed from field-gathered boulders and stones, runs downslope to the shore and was constructed in AD 1830, thereby giving a minimum age (pre-1830) to the upper part of profile 8.

49

40% Brassicaceae together with up to 5% Calluna vulgaris and Lactuceae. Cereal-type pollen values, including Hordeum- and Avenatypes are frequent, if rarely exceeding 2%. Above the zone boundary, in VB 2-2, the character of the pollen profile changes. The Brassicaceae decline abruptly, while Poaceae is reduced within a more varied flora that includes stronger representation from Calluna, Cerealia-types, Chrysanthemum segetum-type, Plantaginaceae and Potentilla-type. A few grains of Chelidonium majus (greater celandine) were noted near the base of this zone (Table 2). In the upper part of VB 2-2, Poaceae representation increases, achieving nearly 70% by the uppermost level.

3. Stratigraphy 5.2. Soil profile 5 (Fig. 3B) Visual inspection of the stony organo-mineral soils at Village Bay revealed stratigraphic homogeneity and the thin sections showed there to be little differentiation of patterning within the slides or between the top and bottom of profiles (Meharg et al., 2006; Davidson et al., 2007). Soil micromorphology showed the soils to have a high abundance of carbonized material and some bone fragments, particularly in the upper parts of the profiles, with evidence of biological mixing and, possibly, digging by hand. The thin sections from beneath the Consumption Dyke were similar to the other Village Bay soils but had a higher abundance of stones, bone fragments and carbonized material. 4. Palynology Initial sample preparations were at an interval of 4 cm; subsequently a small number of intervening samples were prepared from soil profiles 2, 6 and 8. Standard pretreatments (NaOH, HF and acetolysis) (Faegri and Iversen, 1989) were used to concentrate palynomorphs. A tablet containing a known number of marker grains of Lycopodium clavatum was added to each preparation so that an estimate of absolute pollen content could be made. Samples were dried using alcohols and mounted in silicone fluid of 12,500 cSt viscosity for microscope counting. A minimum counting sum of 500 total land pollen (TLP) was used for soil profiles 2 and 8; the sum for profiles 5 and 6 was at least 300 TLP. The computer programs TILIA and TILIA.GRAPH (Grimm, 1991) were used to produce selected taxa pollen diagrams for each site (Figs. 3 and 4), a composite of the summary diagrams (Fig. 5) and all other palynological figures. The placement of local pollen assemblage zones and sub-zones (Table 1, where zones are designated with VB [Village Bay] and soil site number prefixes) was informed by stratigraphically constrained cluster analyses using the program CONISS (Grimm, 1987) for taxa achieving at least 3% TLP in one sample for each site. The occurrence of species not included in the selected taxa pollen diagrams, is indicated in Table 2 by zone or sub-zone. Pollen nomenclature follows Stace (1997) and Bennett (2008). Cerealia undiff. denotes cereal pollen that cannot clearly be assigned to other cereal types. In the text the suffix ‘undiff.’ has been omitted where clarity is uncompromised. Pollen preservation status was noted by assigning each grain to one of four mutually exclusive categories — perfect, folded, broken (or split) and corroded (or thinned) (Cushing, 1967; Whittington and Hall, 2002). Summary preservation data are presented within the pollen diagrams (Figs. 3 and 4). Absolute pollen counts for selected taxa are shown in Fig. 6. Charcoal counts were completed for soil profiles 5, 6 and 8 by making an areal estimate of each piece of charcoal encountered on a representative set of traverses while noting the number of exotic marker grains. Charcoal concentration (cm2 cm− 3) and charcoal to pollen ratios (C:P) are shown in Figs. 3 and 4 at an exaggeration of ×103 and ×108 respectively.

Poaceae is the dominant taxon with representation ranging from 30–60% TLP. Heath taxa, mainly Calluna vulgaris, are also significant throughout at values from 10–25%, while other types, such as Plantago maritima and Avena-type have a constant presence at more than 2%. Strong Poaceae values at the base of VB 5-a decline slightly through the subzone while the combined heath signal, mainly Calluna, increases (Figs. 3B and 5). Plantaginaceae and Potentilla-type also rise through VB 5-a, especially in the top two spectra. Chrysanthemum segetum-type becomes continuous at N2% from the base of VB5-b while Potentilla-type displays a marked decline across the 5a-5b boundary. Calluna is more variable in VB 5-b with high values in the lower and upper parts of the subzone contributing to peaks in the heath signal (Fig. 3B). Plantaginaceae, mainly P. maritima, achieves its highest values in this subzone. Poaceae remains the dominant taxon in VB 5-b but barely exceeds 40%. In VB 5-c Poaceae achieves its highest values with a maximum of over 60%. Heaths and P. maritima decline across the upper subzone boundary, while Rumex acetosa achieves some values N2% in VB 5-c. The Pteropsida (monolete) indet. signal declines up-profile. Charcoal concentrations and C:P are variable with the highest values associated especially with the maximum pollen concentrations (Fig. 3B) which occur in VB 5-b. The percentages of Poaceae grains, and total pollen, that were found to be corroded decrease up-profile (Figs. 3B and 5). 5.3. Soil profile 6 (Fig. 4A)

5. Results

Poaceae is dominant achieving values of 50–65% TLP in the two lower subzones and declining slightly to stable values around 45% through most of VB 6-1c. Poaceae dips to its lowest value of 30% at the base of VB 6-2, thereafter increasing steadily to a maximum value of 65% at the top of VB 6-3. Apart from Poaceae, the VB 6-1 assemblage zone is characterized by the continuous presence of Cyperaceae at values of 5 to 10% and Calluna, which strengthens in VB 6-1c to values around 10%. Plantaginaceae, mainly Plantago lanceolata, are present throughout at combined values of up to 10%. Subzones 1a and 1c feature slightly enhanced values for Brassicaceae compared to VB 61b. The combined heath signal increases sharply to over 25% at the base of VB 6-2, where Poaceae is at a minimum, and remains strong in VB 6-2, declining slightly towards the upper zone boundary. Potentilla-type and Plantago maritima also achieve their highest values in this zone. Heath values decline further in VB 6-3 as Poaceae continues its recovery. Small amounts of cereal-type pollen were noted throughout Soil profile 6; Avena-type provided the strongest signal from VB 6-1c upwards where the arable weed Chrysanthemum segetum-type was also noted. The highest C:P signals were noted in VB 6-1c and particularly in VB 6-2, where they were supported by raised charcoal concentrations.

5.1. Soil profile 2 (Fig. 3A)

5.4. Soil profile 8 (Fig. 4B)

The profile is dominated throughout by Poaceae at values of 30– 65% TLP. In VB 2-1 the strong Poaceae signal is accompanied by 20–

Poaceae is dominant throughout this profile at values of 60–70% TLP in VB 8-1 and around 50% in VB 8-2. The strong Poaceae signal is

50 M.P. Donaldson et al. / Review of Palaeobotany and Palynology 153 (2009) 46–61

Fig. 3. Selected taxa pollen and spore diagrams with preservation status summaries for soil profile 2 (A) and soil profile 5 (B).

M.P. Donaldson et al. / Review of Palaeobotany and Palynology 153 (2009) 46–61

Fig. 4. Selected taxa pollen and spore diagrams with preservation status summaries for soil profile 6 (A) (×5 exaggeration curve for Hordeum-type) and soil profile 8 (B).

51

52 M.P. Donaldson et al. / Review of Palaeobotany and Palynology 153 (2009) 46–61 Fig. 5. Composite of summary diagrams from the Village Bay profiles showing the occurrence of some of the main pollen types discussed in the text, percentage corroded Poaceae (upper scale) and total pollen concentration (grains per cm3 wet sediment, lower scale).

M.P. Donaldson et al. / Review of Palaeobotany and Palynology 153 (2009) 46–61 Table 1 Local pollen assemblage zones for the Village Bay soil profiles LPAZ

Major taxa

Depth (cm)

Soil profile 2 VB 2-2 VB 2-1

Poaceae—Calluna–Chrysanthemum segetum Poaceae–Brassicaceae

46.5–3.0 72.0–46.5

Poaceae—Calluna–Plantago maritima–Avena– Potentilla–Pteropsida (monolete) indet. Poaceae—Calluna–Plantaginaceae—Avena–Potentilla– Chrysanthemum segetum Poaceae—Calluna–Plantaginaceae—Avena– Potentilla–Pteropsida (monolete) indet.

15.5–0

Soil profile 5 VB 5-c VB 5-b VB 5-a

Soil profile 6 VB 6-3 VB 6-2 VB 6-1c VB 6-1b VB 6-1a

Soil profile 8 VB 8-2 VB 8-1b VB 8-1a

Poaceae—Calluna Poaceae—Calluna–Empetrum–Plantaginaceae— Potentilla Poaceae—Calluna–Cyperaceae–Plantaginaceae— Potentilla–Rumex acetosa Poaceae–Cyperaceae—Calluna–Plantaginaceae— Rumex acetosa–Potentilla Poaceae–Cyperaceae—Calluna–Plantago lanceolata– Brassicaceae—Potentilla

Poaceae—Calluna–Ericales–Cerealia-type—Potentilla Poaceae–Lactuceae–Ericales Poaceae–Lactuceae—Calluna–Ericales–Cerealia-type

35.5–15.5 55.0–35.5

12.5–0 25.5–12.5 48.5–25.5 56.5–48.5 71.5–56.5

37.5–20.0 51.5–37.5 72.0–51.5

accompanied by heath taxa, principally Calluna and Ericales, in both zones, and by Lactuceae in VB 8-1 and Potentilla-type in VB 8-2. The combined total representation in VB 8-1a approaches 20% and is accompanied by Cerealia- or Hordeum-types at N2%. In VB 8-1b the combined heath signal is much reduced while Poaceae achieves its highest values and the cereal-types are slightly reduced. Lactuceae is a constant presence in VB 8-1 at 5 to 10%. Small amounts of Chelidonium majus were noted at most levels in VB 8-1 (Table 2). The main changes in the profile occur at the base of VB 8-2 and include sharp declines in Poaceae (to around 50%) and Lactuceae, the recovery of the heaths to around 20% and the expansion of Potentilla-type. All cereal-types are at their strongest in VB 8-2. Pollen concentration values are low throughout VB 8-1 (the lowest representation associated with a 20% peak in Pteropsida [monolete] indet. spores in the upper part of VB 8-1b), but rise markedly in VB 8.2. Charcoal concentrations increase through the profile, but C:P shows clear maxima within subzone 1b. 6. Discussion The difficulties with interpreting soil pollen spectra have been well-rehearsed (e.g. Dimbleby, 1985; Kelso, 1994; Whittington and Edwards, 1999; Long et al., 2000), but the patterning evident in the Hirta profiles persuades us that sensible inferences are possible. For ease of discussion, elements of both methodology and interpretation are presented here, although the complex nature of the findings does not readily permit of such neat partitioning. 6.1. Palynomorph preservation, biased assemblages and chronology In view of Hirta's remoteness from regional pollen sources, the principal pathways for deposition of palynomorphs in the Village Bay soils are likely to be those local to the island, although it is accepted that there will be elements of long-distance pollen transport to be considered (Section 6.3). The local processes involved would include deposition from cultivated crops and fallow weeds, windblown pollen from neighbouring plots, the surrounding vegetation of Village Bay and the rest of the island and the addition of pollen contained within organic manures and ash used to build up the soils (Sections 6.2 and 6.4).

53

Taphonomic concerns are important for interpretations and they have been considered systematically by applying the nine tests of Bunting and Tipping (2000) to highlight possible occurrences of postdepositional biasing. Two test results raised questions about the integrity of the Village Bay profiles. The abundance of corroded pollen (Figs. 3–5) frequently exceeds the recommended cut-off (35% TLP) for severely deteriorated pollen, described by Bunting and Tipping (2000) as ‘degraded and/or amorphous’. However, the corroded category used in the current work includes grains that are slightly thinned and pitted. The abundance of severely deteriorated grains is unknown, but it is certainly less than the percentage of corroded grains (Figs. 3–5). In addition, the abundance of resistant taxa, those that remain recognizable even in a degraded state, was close to or exceeded the 6% level recommended (Ibid.) at three of the Hirta sites (2, 6 and 8) where Lactuceae and Brassicaceae were significant in the profiles. These two pollen types have strong associations with agriculture and as it is known that the site was used for cultivation, then arguably the quantity of these two resistant types is not sufficient cause for denying the interpretability of the data. Poaceae is the largest contributor to the pollen profiles of Village Bay and percentages of corroded Poaceae have been included in the summary diagrams (Fig. 5) as an indicator of conditions detrimental to the preservation of pollen, in the soils themselves and at points along the pathway to incorporation in the soils. Such conditions would include frequent exposure of soils to wetting and drying through soil mixing and cultivation, which would enhance microbial activity. Variations in total fossil concentration are shown in the pollen and summary diagrams (Figs. 3–5), while absolute data for selected taxa are shown in Fig. 6. Such changes may indicate differences in pollen production and/or preservation through the profile, but can also result from soil compaction, disturbance, or enrichment with pollen-rich material from elsewhere. Given that normal pedogenesis is not being dealt with here (e.g. as in podzol development), but rather plaggen formation, then palynomorph deterioration through time and/or down-profile, as reflected in attenuations in concentrations and increased palynomorph damage and spore representation, would not necessarily be expected. To varying degrees, such patterns are seen in profiles 2, 5 and 8. This may be coincidental, or it may signify that once emplaced, plaggens can become podzolised or can mimic such soil palynological processes. To some extent, all the profiles show an inverse relationship between total fossil concentration values and the percentages of corroded Poaceae (Fig. 5), suggesting that when there is plenty of pollen in the soil, much of it is in good condition, and vice versa. This relationship is quite clear throughout profiles 2 and 8 and only slightly equivocal in parts of profiles 5 and 6. The inverse relationship in profiles 2 and 8 is particularly striking at around 40 cm below ground level, where a sharp reduction in high values for corroded Poaceae is mirrored by a marked rise in the total fossil concentration. This is highly suggestive of the addition of composts derived from pollen-rich and previously undisturbed organic materials such as peat and turf (Meharg et al., 2006) to the upper part of these profiles. Shallow composting was recorded by Mackenzie (1911; reported in Harman, 1997, p. 200) during his stay on Hirta from 1830–43, and, amongst other practices, prompted agricultural improvements. The highest absolute quantities of pollen are found in profile 6 where, in contrast to the other sites, values are high in the lower levels. This feature may reflect conditions particular to that site; profile 6 has significant Cyperaceae representation in zone VB 6-1, indicative of poorly drained land where pollen preservation ought to be good. One interesting feature of profiles 2, 5 and 6 is that values for total pollen concentration and percentage corroded Poaceae change direction (on the scales used here, they approach and criss-cross each other over most of the upper parts of the profiles). This correspondence may result from the sites being regularly manured and cultivated in a differing and probably more systematic manner than

54

M.P. Donaldson et al. / Review of Palaeobotany and Palynology 153 (2009) 46–61

Table 2 List of taxa that have been omitted from one or more of the selected taxa pollen diagrams for the Village Bay soil profiles, by zone and sub-zone

(continued on next page)

M.P. Donaldson et al. / Review of Palaeobotany and Palynology 153 (2009) 46–61

55

Table 2 (continued)

+ indicates presence at one or more levels at b2%TLP; ++ indicates presence at one or more levels at N2%TLP. Shading indicates inclusion in the relevant pollen diagrams.

formerly. In profile 8 the graphs suggest that such a change had yet to be reached fully at the top of the profile, but this process could have been terminated by the building of the Consumption Dyke. Low percentages of arboreal pollen are found in the uppermost zone at Conachair (after cal AD 250–430; Fig. 7) and tree and shrub pollen is very low in all soil profiles from Village Bay. Even had macrofossils been found within the mixed soils of the profiles, their radiocarbon dating, let alone that of the soils themselves, would have been questionable (Matthews, 1985; Walker, 2005), making dating and profile correlations difficult. The Consumption Dyke provides a latest date (AD 1830) for the top of profile 8. Although total Pb was observed to decrease in the upper portions of profiles 2, 6 and 8, this cannot be attributed to atmospheric inputs from industrial sources as it is evident well below the 1830 surface of site 8 (see Meharg et al. (2006) for further discussion). 6.2. Microscopic charcoal The high quality of pasture in the archipelago is well attested (cf. Fleming 2005). There is no historical evidence known to us of heathland burning (muirburn) to promote browse on St Kilda, and the smallness of the islands may have made it unlikely in earlier (especially prehistoric) times as inferred for elsewhere in the Outer Hebrides (Edwards et al., 1995; Edwards, 1996). Varying amounts of charcoal were found throughout the three profiles that were assessed (Figs. 3B and 4), but there is no clear

association between the charcoal curves and the records for particular pollen taxa (e.g. Calluna and other heaths which are readily seen elsewhere as being suggestive of natural or intentional burning — cf. Bunting, 1996; Edwards, 1996; Edwards et al., 1995, 2000). Within profiles 5 and 6 the graphs for charcoal concentration and charcoal: pollen behave in a similar manner. This is supportive of the view that the bulk of the charcoal and pollen accumulating in the plaggen soils of these profiles was derived from the same source, i.e. burnt composts and ash (Meharg et al., 2006) and that the microscopic charcoal is largely domestic in origin. In profile 8, however, while charcoal concentration and C:P show similar trends in the lower part of the diagram (Fig. 4B), there is a considerable divergence in behaviour from 50 cm upwards. In midprofile the very high C:P may in part be explained by the exceptionally low total pollen concentrations. The charcoal concentration remains significant, indicating a source that did not include substantial amounts of well-preserved pollen. At least two possibilities for this pattern suggest themselves. Firstly, the pollen may have been destroyed as a result of a high temperature peat burn, thus enhancing the values for C:P (pollen can survive burning if temperatures are not excessive [cf. Carrión et al., 2000; Bunting et al., 2001; Ghosha et al., 2006]). A second potential candidate, that would not require burning to produce a charcoal signal, is the inner roofing material that was removed from the houses in most years. The incorporation of this smoke-impregnated material, which may have been derived from heaths (Emery, 1996) or barley straw, would have added charcoal to

M.P. Donaldson et al. / Review of Palaeobotany and Palynology 153 (2009) 46–61

Fig. 6. Concentration data for selected pollen taxa.

56

M.P. Donaldson et al. / Review of Palaeobotany and Palynology 153 (2009) 46–61 Fig. 7. Conachair peat core, selected taxa pollen diagram (×5 exaggeration curves for Lactuceae and Brassicaceae), based on Fig. 4 in Meharg et al. (2006). The 14C calibrations were carried out within the program Calib. 5.0.2html (Calib, 2008) with 2σ age ranges rounded to the nearest 10 years.

57

58

M.P. Donaldson et al. / Review of Palaeobotany and Palynology 153 (2009) 46–61

the compost stream, while any pollen it had contained formerly would have had ample time to deteriorate through oxidation. In the upper part of profile 8, zone VB 8-2, the divergence in the charcoal signals is reversed (Fig. 4B). Charcoal concentration is variable in this zone with an overall upward trend, while C:P is considerably reduced, mainly in response to the increased values for total pollen concentration. One possible explanation for this is that the charcoal in VB 8-2 comes from more than one source. Such charcoal streams could include the continuation of composting with smokeimpregnated roofing material, with the addition of ash derived from turf obtained from the upper slopes of Hirta. The use of such material is supported by the enhanced Potentilla signal in VB 8-2 which is discussed below. The addition of ash is also reflected in the elevated values of Pb and Zn in the old cultivated soils (Meharg et al., 2006). 6.3. The woodland signal The total woodland component (trees and shrubs) in each of the profiles is minimal, with values only occasionally exceeding 5% TLP. These low values are understandable in terms of off-island sources given that there are no significant woodland taxa on the island, even in areas inaccessible to sheep. The low-growing dwarf willow (Salix repens) has been found on some of the upper slopes (Gwynne et al., 1974; Revised Nomination, 2003, p.52) but, apart from one appearance at N2% in the lower part of profile 2, Salix pollen is present only sporadically. Despite low values, the woodland signal is quite variable; the majority of the small peaks include Corylus avellana-type at N2% TLP (Fig. 5) and are associated with high values for heaths. Myrica gale, the pollen of which may be mistaken for that of C. avellana (Edwards, 1981), has not been recorded in the island's vegetation (Gwynne et al., 1974; Preston et al., 2002), so it is reasonable to argue that most, if not all, of the woodland signal is derived from off-island sources (cf. Fossitt, 1994; Brayshay et al., 2000; Edwards et al., 2000). Despite this assumption, woodland pollen was retained in the pollen sum because the low values barely impact upon the curves of the other taxa; indeed, to exclude them would be to pre-judge their significance and would limit comparability with the Conachair profile (Figs. 1A and 7). The Conachair core was obtained at an altitude of over 300 m approximately 1 km north of Village Bay. The profile includes a hiatus, dated between 3270 ± 40 (1660–1450 cal BC) and 1690 ± 40 BP (cal AD 250–430), that has been interpreted as possible evidence of peat cutting (Meharg et al., 2006). Such activity suggests a pathway whereby peat, cut, dried and used as fuel, could have entered the waste stream as ash for manuring. The missing Conachair peat, however, pre-dates historical records for Hirta. Those records that do exist indicate that turf (i.e. sods of grass and heather), not peat, was used for fuel during the last few centuries of occupation (Macgregor, 1960; Martin, 1994). Martin, visiting in 1697, reported that ‘The chief ingredient in their composts is ashes of turf mixed with straw; with these they mix their urine' (Martin, 1994, p. 416). It may be the case, of course, that peat was used as fuel during earlier phases of occupation, and may have been extracted from other sites in more recent times. The Conachair pollen spectra include percentages of woodland taxa at around 15% TLP in the post-1690 BP (cal AD 250–430), levels, but the highest values of up to more than 50% (between the calibrated dates of 3260–2900 and 1660–1450 cal BC) are associated with low pollen concentrations (Fig. 7). Corylus avellana-type was the principal contributor to the woodland record throughout, apart from in the upper 5 cm where Pinus sylvestris became dominant. This marker for the expansion of coniferous plantations on the mainland over the last 150 yr was not detected in the Village Bay profiles. High values for woodland taxa have been noted previously from a site (McVean, 1961) close to Conachair that revealed pollen spectra similar to those reported here. Pollen concentration data were not available at that time and the interpretation placed on the findings, that woodland or scrub had been nearby, could be challenged in the

light of current work. It is difficult to imagine woodland growing on Hirta during the Holocene. If trees had been able to establish themselves, it is highly unlikely that they would have thrived sufficiently to produce pollen on the very exposed upper slopes of the island, such as around Conachair. It is far more realistic to suggest that a woodland signal detected anywhere on Hirta is derived directly or indirectly from off-island sources. At Conachair the woodland signal may be linked to the altitude of the site and its interception of long-distance air-streams washed by frequent rainfall, as well as to low pollen productivity from the exposed upland vegetation. The presence of a significant woodland pollen signal at Conachair impacts on the interpretation of the Village Bay profiles, because it shows the potential of the higher areas of Hirta to accumulate a regionally biased pollen signal. The islanders' habit of collecting turfs for fuel from the hillsides around Village Bay provides a more likely pathway for the introduction of a woodland-enhanced pollen signal to the plaggen soils. Turfs do not burn as hotly as peat (Macgregor, 1960) and the combustion of some of the organic content would concentrate palynomorphs already trapped within the more minerogenic parts of the turf. The peaks in the Corylus avellana-type signal from the cultivated sites occur almost exclusively with high heath values and probably provide a marker for the addition of waste derived from ericaceous turfs to the soils. A further connection is provided by Potentilla-type, the signal for which is second only to Poaceae throughout the Conachair profile, occasionally even exceeding it (see below). 6.4. The heath signal The combined heath signal contributes up to 25% TLP to the Village Bay profiles and is the most prolific of the sub-dominant taxa. The high variability of the heath spectra in profiles 2, 6 and 8 contrasts with the relatively steady heath signal in profile 5 and raises questions about how the heath pollen came to be incorporated into the soils of the cultivated area. Had the heath input been derived solely from airborne deposition, it might be argued that its registration would have been less variable between and within the pollen profiles, although changes in the non-heath flora would have influenced heath taxa pollen percentages. Total heath pollen concentrations (Fig. 6) show more variability than the percentage profiles indicate, and support many of the stepped changes in the summary diagrams (Fig. 5). If, as is suspected, the heath pollen is derived substantially from manuring, then the variability of the signal is less surprising. It is known that heather and grass turf was cut from the hillsides around the village for use as fuel (MacGregor, 1960). The three profiles that are likely to have been cultivated most recently, profiles 2, 5 and 6, display a steady decline in the heath signal over the upper 20 to 25 cm, suggestive of a gradual reduction in the input of heath pollen to the soils over the period represented. The period covered by this decline is unknown and likely to be different for each profile. What is more certain is that the upper few centimetres of profiles 2, 5 and 6 will include the last few years of occupation prior to the evacuation of Hirta. During that period of the island's history, cultivation, and by implication manuring, would have been reduced, probably both in spatial extent and frequency, as the population declined and interaction with and dependence on the mainland increased. The extent of heath vegetation on Hirta has increased since evacuation (Gwynne et al., 1974; Revised Nomination, 2003, p. 52), a change predicted by Petch (1933) who suggested that the lack of turfcutting and the removal of the Blackface sheep would allow heather to thrive more widely. The Soay sheep that remain on Hirta appear to be uninterested in the heathlands (Gwynne et al., 1974; Clutton-Brock and Pemberton 2004). There is no evidence for the expansion of heath taxa in the upper parts of profiles 2 and 6, and only slight support in profile 5 where the uppermost sample was obtained just 1.5 cm below the ground surface (although evidence of the most recent vegetation

M.P. Donaldson et al. / Review of Palaeobotany and Palynology 153 (2009) 46–61

changes could have been missed). The lack of a substantial increase in heaths in the upper parts of the most recently cultivated profiles gives some support to the view that the airborne component of the heath input was minimal during the period of cultivation and that the bulk of heath pollen found in these profiles was introduced through manuring. The stepped or sharp changes in the heath records are most noticeable in profiles 2, 6 and 8, and may represent episodes of manuring with heath-rich or heath-poor material that failed to incorporate all of the previously utilised soil, leading to partial preservation of the lower levels (Groenman-van Waateringe, 1992). Potentilla was noted as a minor component of the heath-rich vegetation identified by Gwynne et al. (1974) (Fig. 1A). In profiles 2, 6 and 8, raised values for Potentilla pollen (Figs. 3A and 4) occur with high values for Calluna and offer support for the introduction to the soils of organic material rich in heather pollen, such as turfs from the slopes of Village Bay and from upland areas beyond this (Section 6.3). This pathway for the addition of Calluna pollen to the soils is supported by historical sources cited by Macgregor (1960), which indicated that turf from the nearby slopes were used as fuel in preference to peat from the hilltops because less effort was required for collection. 6.5. Select observations on taxa and land use The Village Bay profiles are dominated throughout by Poaceae even during phases when crops such as cereals, or possibly brassicas, may have been growing on-site. The dominance of Poaceae in the pollen record is not surprising as grassland floras dominate the current vegetation of Hirta (Fig. 1A) (Gwynne et al., 1974). Evidence from Conachair (Fig. 7) (Meharg et al., 2006) and Gleann Mor (Walker, 1984; authors, unpublished) indicates that grassland floras have been dominant for much of the last 5000–6000 years on the island. After Poaceae, the heath representation is the strongest in all profiles (see Section 6.4). This also is not unexpected given that ‘Dry’ and ‘Wet Calluna Heath’ floras (Gwynne et al., 1974) form a substantial part of the present island vegetation, encircling the formerly cultivated area on higher slopes (Fig. 1A). Soil profile 6 is alone in having a phase where Cyperaceae is continuous at values N2%, although the strong sedge representation may be more attributable to environmental conditions than to land use. Of the four sites examined, soil profile 6 is the lowest in altitude and furthest from the houses along Main Street. Both factors may have contributed to the strength of Cyperaceae, a taxon which usually indicates poorly drained ground. Such a site would have required more effort to improve and cultivate, potentially for less return, than ground higher upslope and closer to the houses. Significant sections of each profile appear to be well-mixed in that several herbaceous components, such as Cerealia-types, Cyperaceae, Brassicaceae, Lactuceae and Potentilla-type contribute continuously within a narrow range of values (Figs. 3 and 4). The most marked changes in the spectra can be seen in soil profile 2. Here the strong Brassicaceae signal declines abruptly at the zone boundary, above which cereals, Chrysanthemum segetum-type and other cereal weeds, Plantaginaceae and Potentilla-type, become more significant. Soil profiles 6 and 8 show similar patterns; the changes are less striking than for profile 2 but nonetheless suggest phases in which different sources of pollen had a stronger or weaker influence. Cereal pollen is rarely transported far from its point of origin so it is likely that cereals were being grown on-site, or nearby, throughout the period covered by the soil profiles. The presence of ‘Cereal-type’ in Fig. 5 is supported by the concentration values shown in Fig. 6. There is also a marked association between Potentilla-type and the cereal signal in all the profiles and this could be related to manuring (see Section 6.4). Brassicaceae pollen cannot be identified to species level using light microscopy, so it would be unwise to make claims about the cultivation of cabbages or other brassica crops from the pollen record alone. Indeed, the fact that cabbages are harvested before they flower might seem to reduce the likelihood of the Brassicaceae pollen

59

deriving from cabbages. There is, however, ample historical evidence for the cultivation of cabbages (cf. kale) in Village Bay (Seton, 1878; Nimlin, 1979), at least as far back as 1758 (Table 1 in Harman, 1997), including a report from 1890 describing cabbages ‘which are very wild’ (Ibid., 1997, p. 203). Brassica sp., regarded as introduced plants by Turrill (1927) were absent in 1948 (Poore et al., 1949). Given that cabbages are known to have been grown on the island, it seems likely that the Brassicaceae signal is derived largely from crops that were allowed to go to seed or from manuring with waste material derived from cabbage enclosures (MacGregor, 1960). The scant presence of any Brassicaceae pollen in the upper levels of profiles 2, 5 and 6 suggests that opportunistic weeds were not the source of the Brassicaceae signals in the lower parts of the sequences. Another potential contributor to the Brassicaceae pollen record is Cochlearia officinalis (common scurvy-grass; Preston et al., 2002), a plant of coastal and mountainous habitats (Stace, 1997). It was not noted in the vegetation surveys of Gwynne et al. (1974) although one plant was found in 1948 on Plantago sward by Poore et al. (1949). Boyd (1979) reports it as being a component of Hirta's flora in 1961–62. This plant was known in the Scottish islands as a remedy against scurvy (Martin, 1994, p. 378; Fleming, 2005, p.85), although its benefits may not always have been recognized by all St Kildans given the report of incipient scurvy (MacNeill, 1866, pp. 8–9) cited in Harman (1997, p. 261). It is possible that Cochlearia was gathered for use as compost at some sites. In addition to the cereal-type pollen, there are two other indicators of habitation and cultivation which are considered to be worthy of comment. Chrysanthemum segetum pollen (corn marigold) was found at all sites and displays an association with cereal pollen (cf. Whittington and Edwards, 1995). C. segetum is known as a weed of arable land (Clapham et al., 1987) and was described as a common weed of the oats on Hirta by Turrill (1927). It was present in 1931 but absent in 1948 (Poore et al., 1949) and 1961–62 (Boyd, 1979). Burnt and waterlogged macrofossils have been identified in various occupation phases on Hirta (Huntley, 1996) confirming that the plant grew on the island. The link with cereal pollen is particularly strong in the upper zone of profile 2, where C. segetum values peak at 20% TLP, but it is also significant in the upper sections of profiles 5 and 6. In profile 8, which was truncated in 1830, the link is less clear. Cereal cultivation since 1830, represented by profiles 2, 5 and 6, may have involved imported seed contaminated with C. segetum. Chelidonium majus (Table 2) was noted at all the cultivated sites but only appeared regularly in the lower part of profile 8. Greater celandine had a broad role as a medicinal plant (Dickson and Dickson, 2000; Purple Sage, 2008), being used to treat eye complaints, and may have been cultivated for this purpose. C. majus pollen has previously been found at An Lag (Huntley, 2008; Kilda, 2008), situated on the slope to the north of the village. The plant's current range in Scotland is limited mainly to the south and east, although it is known at four sites on the west coast close to the Inner Hebridean islands of Skye and Mull (National Biodiversity Network, 2008). There is the possibility that C. majus was imported to St Kilda for the purpose of treating various disorders; this medicinal role may have been particularly important when contact with the mainland was fairly limited. As contact improved, more modern remedies may have been sought. A nurse visited the island regularly from the 1890s onwards (Emery, 1996) and undoubtedly influenced treatments for various conditions. 7. Conclusions This palynological study of four Village Bay sites is supportive of previous work (Meharg et al., 2006) that hillside and upland turf ash was added to the waste stream used to create the plaggen soils. It is possible that ash from peat collected in the Conachair area may also have been used, but the dates for the hiatus at that site do not support its use during the last 1500 years. There may be other sites in the

60

M.P. Donaldson et al. / Review of Palaeobotany and Palynology 153 (2009) 46–61

locality that show evidence of more recent peat cutting. Given that hillside turf is known to have been used historically as fuel on Hirta and that scars from turf-cutting remain visible on slopes around Village Bay, it is likely that turf ash provided the bulk of the ashed material added to the plaggen soils. The presence within the Conachair spectra of significant amounts of regionally-derived woodland pollen has provided a potential marker for the addition of hillside and upland turf (and peat) ash to the profiles of Village Bay. Such a pathway could be further investigated by the analysis of modern samples from Hirta and by assessing the pollen content of ashed and partially burnt turf and peat from the island. This palynological study has also revealed discrete phases when land use appears to have been relatively invariant, but this could to some extent be due to soil mixing. Evidence for cereal cultivation is seen throughout the soil profiles and is particularly marked in the upper levels at all the sites. Palynological evidence for the cultivation of other crops is unclear. The Brassicaceae signal may in part reflect the cultivation of cabbages on site or the addition of cabbage waste to the soils. Chelidonium majus may have been grown for medicinal use. The relationship between total pollen concentration and corroded grass pollen may reflect specific changes in soil management. Soil pollen profiles and plaggen soils in particular can be difficult to interpret. While those from Hirta present their own problems, they have also been shown to be both interpretable and capable of revealing fresh insights into the land use history of an iconic location. Acknowledgements The Leverhulme Trust is thanked for the financial support which made this research possible. The National Trust for Scotland, and especially Susan Bain, provided invaluable field and logistical assistance. Particular thanks are due to Richard Tipping and two anonymous referees for many useful comments and suggestions on an earlier draft of the paper. References Bennett, K.D., 2008. Catalogue of pollen types. http://www.chrono.qub.ac.uk/pollen/pcintro.html (last accessed June 2008). Boyd, J.M., 1979. Natural history. In: Small, A. (Ed.), A St Kilda Handbook. National Trust for Scotland, Edinburgh and University of Dundee, Department of Geography, Occasional Paper No. 5, Dundee, pp. 20–35. Brayshay, B.A., Gilbertson, D.D., Kent, M., Edwards, K.J., Wathern, P., Weaver, R.E., 2000. Surface pollen-vegetation relationships on the Atlantic seaboard: South Uist, Scotland. Journal of Biogeography 27, 359–378. Bunting, M.J., 1996. The development of heathland in Orkney, Scotland: pollen records from Loch of Knitchen (Rousay) and Loch of Torness (Hoy). The Holocene 6, 193–212. Bunting, M.J., Tipping, R., 2000. Sorting dross from data: possible indicators of postdepositional assemblage biasing in archaeological palynology. In: Bailey, G., Charles, R., Winder, N. (Eds.), Human Ecodynamics. Oxbow Books, Oxford, pp. 63–69. Bunting, M.J., Tipping, R., Downes, J., 2001. “Anthropogenic” pollen assemblages from a Bronze Age Cemetery at Linga Fiold, West Mainland, Orkney. Journal of Archaeological Science 28, 487–500. Calib, 2008. Radiocarbon calibration program: version Calib 5.0.2html. http://calib.qub. ac.uk/calib/ (last accessed June 2008). Carrión, J.S., Scott, L., Huffman, T., Dreyer, C., 2000. Pollen analysis of Iron Age cow dung in southern Africa. Vegetation History and Archaeobotany 9, 239–249. Clapham, A.R., Tutin, T.G., Moore, D.M., 1987. Flora of the British Isles, 3rd edition. Cambridge University Press, Cambridge. Soay Sheep. In: Clutton-Brock, T.H., Pemberton, J.M. (Eds.), Dynamics and Selection in an Island Population. Cambridge University Press, Cambridge. Cushing, E.J., 1967. Evidence for differential pollen preservation in late Quaternary sediments in Minnesota. Review of Palaeobotany and Palynology 4, 87–101. Davidson, D.A., Wilson, C.A., Meharg, A.A., Deacon, C., Edwards, K.J., 2007. The legacy of past manuring practices on soil contamination in remote rural areas. Environment International 33, 78–83. Dickson, C., Dickson, J.H., 2000. Plants & People in Ancient Scotland. Tempus Publishing Ltd., Stroud. Dimbleby, G.W., 1985. The palynology of archaeological sites. John Wiley, Chichester. Edwards, K.J., 1981. The separation of Corylus and Myrica pollen in modern and fossil samples. Pollen et Spores 23, 205–218.

Edwards, K.J., 1996. A Mesolithic of the Western and Northern Isles of Scotland? Evidence from pollen and charcoal. In: Pollard, T., Morrison, A. (Eds.), The Early Prehistory of Scotland. Edinburgh University Press, Edinburgh, pp. 23–38. Edwards, K.J., Mulder, Y., Lomax, T.A., Whittington, G., Hirons, K.R., 2000. Humanenvironment interactions in prehistoric landscapes: the example of the Outer Hebrides. In: Hooke, D. (Ed.), Landscape, the richest historical record. Society for Landscape Studies Supplementary Series, vol. 1, pp. 13–32. Edwards, K.J., Whittington, G., Hirons, K.R., 1995. The relationship between fire and long-term wet heath development in South Uist, Outer Hebrides, Scotland. In: Thompson, D.B.A., Hestor, A.J., Usher, M.B. (Eds.), Heaths and Moorlands: Cultural Landscapes. HMSO, Edinburgh, pp. 240–248. Emery, N., 1996. The Archaeology and Ethnology of St. Kilda No. 1; Archaeological Excavations on Hirta 1986–1990. National Trust for Scotland, HMSO, Edinburgh. Faegri, K., Iversen, J., 1989. In: Faegri, K., Kaland, P.E., Krzywinski, K. (Eds.), Textbook of Pollen Analysis, 4th edition. John Wiley & Sons, Chichester. Fleming, A., 2005. St Kilda and the Wider World: Tales of an Iconic Island. Windgather Press, Macclesfield. Fossitt, J.A., 1994. Modern pollen rain in the northwest of the British Isles. The Holocene 4, 365–376. Ghosha, R., D'Rozario, A., Bera, S., 2006. Can palynomorphs occur in burnt ancient potsherds? An experimental proof. Journal of Archaeological Science 33, 1445–1451. Grimm, E.C., 1987. CONISS: a FORTRAN 77 program for stratigraphically constrained cluster analysis by the method of incremental sum of squares. Computers and Geosciences 13, 13–35. Grimm, E.C., 1991. TILIA and TILIA. GRAPH. Illinois State Museum, Springfield, Illinois. Groenman-van Waateringe, W., 1992. Palynology and archaeology: the history of a plaggen soil from the Veluwe, The Netherlands. Review of Palaeobotany and Palynology 73, 87–98. Gwynne, D., Milner, C., Hornung, M., 1974. The Vegetation and Soils of Hirta. In: Jewell, P.A., Milner, C., Morton-Boyd, J. (Eds.), Island Survivors: The Ecology of the Soay Sheep of St. Kilda. Athlone Press, London, pp. 36–87. Harman, M., 1997. An Isle called Hirte. Maclean Press, Isle of Skye. Huntley, J.P., 1996. Plant remains (sections 3:4:1, 4:4:1, 5:5:1). In Emery, N., The Archaeology and Ethnology of St. Kilda No. 1: Archaeological Excavations on Hirta 1986–1990. National Trust for Scotland, HMSO, Edinburgh, pp. 92–95, 134–137, 169–172. Huntley, J.P. 2008. http://www.dur.ac.uk/j.p.huntley/research_sites/kilda.html (last accessed June 2008). Kelso, G.K., 1994. Pollen percolation rates in Euroamerican-era cultural deposits in the Northeastern United States. Journal of Archaeological Science 21, 481–488. Kilda, 2008. http://www.kilda.org.uk/arch-investigations.htm (last accessed June 2008). Long, D.J., Tipping, R., Carter, S., Davidson, D.A., Boag, B., Tyler, A., 2000. The replication of pollen stratigraphies in soil pollen profiles: a test. In: Harley, M.M., Morton, C.M., Blackmore, S. (Eds.), Pollen and Spores: Morphology and Biology. Royal Botanic Gardens, Kew, pp. 481–497. Macgregor, D.R., 1960. The Island of St. Kilda: a survey of its character and occupance. Scottish Studies 4, 1–48. MacNeill, M., 1886. Report of Malcolm MacNeill, inspecting officer of the board of supervision, on the alleged destitution in the Island of St Kilda, in October 1885. Presented to both Houses of Parliament by Command of her Majesty, March 1886. Parliamentary Paper, vol. 57. London. McVean, D.N., 1961. Flora and vegetation of the islands of St. Kilda and North Rona in 1958. Journal of Ecology 49, 39–54. Martin, M., 1994. A Description of the Western Islands of Scotland circa 1695 by Martin Martin, Gent, Including A Voyage to St. Kilda by the Same Author and A Description of the Western Isles of Scotland by Sir Donald Monro. Birlinn, Edinburgh. Matthews, J.A., 1985. Radiocarbon dating of surface and buried soils: principles, problems and prospects. In: Richards, K.S., Arnett, R.R., Ellis, S. (Eds.), Geomorphology and Soils. Allen & Unwin, London, pp. 271–288. Meharg, A.A., Deacon, C., Edwards, K.J., Donaldson, M.P., Davidson, D.A., Spring, C., Scrimgeour, C.M., Feldmann, J., Rabb, A., 2006. Ancient manuring practices pollute arable soils at the St Kilda World Heritage Site, Scottish North Atlantic. Chemosphere 64, 1818–1828. National Biodiversity Network, 2008. Interactive map of Greater Celandine (Chelidonium majus). http://www.searchnbn.net/ (last accessed June 2008). Nimlin, J., 1979. Historical St. Kilda & Exploring Hirta. In: Small, A. (Ed.), A St. Kilda Handbook. National Trust for Scotland, Edinburgh and University of Dundee. Department of Geography, Occasional Paper, vol. 5. Dundee, pp. 76–89. Petch, C.P., 1933. The vegetation of St. Kilda. Journal of Ecology 21, 92–100. Poore, M.E.D., Robertson, V.C., Godwin, H., 1949. The Vegetation of St. Kilda in 1948. Journal of Ecology 37, 82–99. Preston, C.D., Pearman, D.A., Dines, T.D., 2002. New Atlas of the British and Irish Flora. Oxford University Press, Oxford. Purple Sage, 2008. Greater Celandine. http://www.purplesage.org.uk/profiles/ greatercelandine.htm (last accessed June 2008). Revised Nomination, 2003. Revised Nomination of St Kilda for inclusion in the World Heritage Site List. Crown Copyright, Edinburgh. Seton, G., 1878. St. Kilda Past and Present. Blackwood, Edinburgh and London. A St. Kilda Handbook. In: Small, A. (Ed.), National Trust for Scotland, Edinburgh and University of Dundee. Department of Geography, Occasional Paper, vol. 5. Dundee. Stace, C., 1997. New Flora of the British Isles, 2nd edition. Cambridge University Press, Cambridge. Steel, T., 1994. The Life and Death of St. Kilda. Harper Collins, London.

M.P. Donaldson et al. / Review of Palaeobotany and Palynology 153 (2009) 46–61 Turrill, W.B., 1927. The flora of St. Kilda. Report of the Botanical Society and Exchange Club of the British Isles 8, 428–444. Walker, M.J.C., 1984. A pollen diagram from St Kilda, Outer Hebrides, Scotland. New Phytologist 97, 99–113. Walker, M.J.C., 2005. Quaternary dating methods. Wiley, Chichester. Whittington, G.W., Edwards, K.J., 1995. A Scottish Broad: historical, stratigraphic and numerical studies associated with polleniferous deposits at Kilconquhar Loch. In:

61

Butlin, R.A., Roberts, N. (Eds.), Ecological Relations in Historical Times. Blackwell, London, pp. 68–87. Whittington, G.W., Edwards, K.J., 1999. Landscape scale soil pollen analysis. Journal of Quaternary Science 14, 595–604 (Quaternary Proceedings 7). Whittington, G.W., Hall, A.M., 2002. The Tolsta Interstadial, Scotland: correlation with D–O cycles GI-8 to GI-5? Quaternary Science Reviews 21, 901–915.