Holocene Flora, Vegetation and Land-Use Changes on Dingle Peninsula, Ireland, as Reflected in Pollen Analytical, Archaeological and Historical Records

Abstract

Palaeoecological investigations connected with extensive pre-bog, stone walls, and field systems at Kilmore, Dingle peninsula, Ireland, are presented. The main pollen profile, KLM I, spans the last 4000 years. When the record opened, pine (Pinus sylvestris) was already a minor tree, oak (probably Quercus petraea) was the main tall-canopy tree, and birch and alder were dominant locally. Substantial farming is recorded between ca. 1530 and 600 BCE (Bronze Age) when the stone walls were likely constructed. From ca. 560 CE onwards, intensive farming was conducted for much of the time. A largely treeless landscape emerged in the late twelfth century CE. Fine-spatial reconstructions of landscape and vegetation dynamics, including the timing of blanket bog initiation, are made. Post-glacial change in the western Dingle peninsula, based on published Holocene lake profiles and drawing on the new information presented here, is discussed. Reported are (a) fossil spores of the filmy ferns Hymenophyllum tunbrigense, H. wilsonii, and Trichomanes speciosum; (b) the first fossil pollen record for Arbutus unedo (strawberry tree) in the Dingle peninsula (540 CE); and (c) the first published records for Fagopyrum fossil pollen in Ireland, indicating that buckwheat was grown at Kilmore in the late eighteenth/early nineteenth centuries.

1. Introduction

South-western Ireland is defined geographically by three prominent peninsulas, namely Dingle (Corca Dhuibhne in Irish), Iveragh, and Beara (Figure 1, inset). Each peninsula has its own particular character, but they share many features, including bedrock dominated by Devonian sandstone [1], extensive uplands including the highest mountains in Ireland (Carrauntoohil in Iveragh (1038 m) and Mount Brandon in Dingle (952 m); elevations are given relative to average sea level (asl), as indicated in recent Ordnance Survey of Ireland (OSI) maps), and a particularly rich cultural heritage, including archaeological material from most periods, beginning in the Mesolithic and extending into the early medieval period and beyond [2,3,4,5,6,7,8,9]. Several interesting biota are also present, including species with predominantly Atlantic/Lusitanian distribution patterns. These include Geomalacus maculosus (Kerry slug), Simethis mattiazzii (Killarney lily), Arbutus unedo (strawberry tree), and filmy ferns (Hymenophyllum tunbrigense, H. wilsonii, and Trichomanes speciosum, i.e., the Killarney fern) ([10,11,12]; the online version of the BSBI Atlas at https://plantatlas2020.org (accessed on 1 March 2025); and Figures S6, S7 and S9). Note: The prefix S designates figures, tables, and text that are available in the Supplementary Information (SI).

The region has what is widely regarded as some of the best examples of temperate rainforest in Ireland and Britain, namely the Quercus petraea-dominated, fern- and bryophyte-rich woodlands in Killarney National Park, the premier and oldest national park in Ireland. On limestone terrain within this park, there is a Taxus baccata-dominated, species-rich woodland that is of particular importance, given the rarity of semi-natural yew woodlands in Ireland and Europe [13,14,15]. Another important recent development is the establishment of Kerry Seas National Park/Páirc Náisiúnta na Marra, Ciarraí, in 2023, which includes important coastal and marine habitats. The terrestrial component of this national park is centred on the Dingle peninsula between Mount Brandon and the Conor Pass, where our study area lies (https://www.nationalparks.ie/kerry-seas/; accessed on 3 July 2024; see also [8,16,17]).

Our study area lies in the western Dingle peninsula in Kilmore Townland (Td.), about 3 km south of Brandon Bay (Figure 1A and Figures S9–S11). The area is sheltered by a horseshoe-shaped range of mountains that extends from Mount Brandon in the west to Sliabh a’ Fhia (Slievanea; 627 m) and Binn na Naomh (670 m) to the south (core sites KLM I and II are on a blanket-bog covered plateau near the base of these mountains (see below), and Binn os Gaoith (Beenoskee; 826 m) and Stradbally Mountain (Cnoc an tSráidbhaile; 798 m) at the north-east end of the horseshoe.

Kilmore Td. and adjoining Ballyhoneen Td., both largely blanket-bog-covered, have particularly rich archaeology. This includes a wedge tomb (megalith assignable to the Late Neolithic/Early Bronze Age), two standing stones, eight fulachta fiadh (burnt mounds), rock art (cup and circle type) on three of the wedge tomb stones (all assignable to the Bronze Age), and several enclosures, hut sites, a mound, and pre-bog stone walls [18,19] (Figure S9). In what Ó Coileáin [19] refers to as the ‘Central Complex’, pre-bog stone walls, with a total wall length of 7.35 km, extend over approximately 80 ha and define at least 30 fields (the complexes are indicated in Figure S9B). In addition, 27 archaeological sites have been recorded here, and pollen profiles KLM I and II were obtained from this complex. A second complex of pre-bog stone walls has been recorded on the eastern side of Lough Adoon, i.e., the L. Adoon East Complex [19]. The complex extends over 69 ha. Twenty-three wall lengths and several archaeological sites have been recorded, including many hut sites that suggest habitation. However, only one large field (4.5 ha) and two smaller fields, <0.5 ha, have been identified [19]. The third, smaller complex was located on flat terrain to the east of the Scorid river and immediately south of L. Adoon. This complex, referred to by Ó Coileáin [19] as the Wedge Tomb Complex, includes the Ballyhoneen wedge tomb. There are also two standing stones and four fulachta fiadh (probably Bronze Age) that further emphasise the importance of the area during the Bronze Age. Ó Coileáin [19] regards this complex as possibly a place reserved for ritual purposes. The fourth complex is located on the steeply sloping western side of L. Adoon. Ó Coileáin [19] regarded these stone walls as related to the eighteenth or nineteenth centuries CE and associated with relatively recent sheep farming.

As regards palynological investigations in south-western Ireland, the main focus has been on the Killarney area on the Iveragh peninsula and also on Beara and Mizen peninsulas, where the overall emphasis has been on Holocene vegetation dynamics, particularly as mediated by human impact in the second half of the Holocene [20]. In the case of the Dingle peninsula, the investigations by Dodson [21,22] are from the same general area as those reported here. Dodson’s investigations focused on the L. Adoon area (Loch a’ Dúin; lake of the fort), i.e., the lowlands that lie to the north and in the shelter of the Brandon/Stradbally horse-shaped mountain range. From blanket-bog covered lowlands (<150 m asl) near L. Adoon, there are two pollen profiles, namely LAH (L. Adoon Hill Bog) and LAV (L. Adoon Valley Bog). These are sited in the second and third archaeological complexes, respectively (see above; Figure S9B). Both relate to the later Holocene, the former spanning the last ca. 4000 years and extending into recent times, while the latter begins somewhat later (after 4 ka) and ends before the so-called secondary rise of Pinus, i.e., before the eighteenth century (Text S1). Both profiles showed substantial evidence of human impact, as expected, given their proximity to megalithic and other archaeological sites (Figure 1 and Figure S9). The third pollen profile, LCN, relates to nearby Lough Camclaun (Loch Cham Calláin) and derives from lake sediment that was covered by a mat of floating vegetation at the southern end of this glacial lake when cored by Dodson [21]. This site is at ∼250 m asl and abuts a steeply rising terrain (see Text S1).

The marginal sediments of a small lake (∼1 ha but once probably ≥5 ha), situated in the lowlands (∼70 m asl) west of Mount Brandon and ∼4 km from the coastline to the west, have provided a 6-m-long pollen profile (BL) that spans the Lateglacial and Holocene (excluding recent centuries) (Text S1) [23]. Barnosky referred to this site as Ballinloghig Lake (Loch Chorr Áille in OSI, DS 70) [23]. This profile provides an interesting contrast to LCN (L. Camclaun), which lies only 10 km to the east. The significance of these pollen profiles by Barnosky and Dodson with regard to local and regional vegetation history is considered in the Discussion.

From Lispole, on the lowlands east of Dingle town, OCarroll [24] presented a short pollen profile (60 cm) from organic-rich deposits sampled above a fulacht fiadh that was one of several archaeological features recorded in preparation for local road-alignment works. The profile spans ca. 350 BCE to 1150 CE, i.e., from the mid-Iron Age to the end of the early medieval period. Tall woody vegetation during this time consisted mainly of hazel, alder, and birch towards the end of the period. Records of cereal-type pollen and Plantago lanceolata suggest local farming during much of the time. In addition, AMS ¹⁴C dating of botanical remains associated with fulachta fiadh and charcoal pits provides additional evidence of trees, shrubs, and cereal cultivation. Wood and charcoal samples yielded ¹⁴C dates as follows: 3617 ± 54 BP (holly) and 870 ± 37 BP (alder), and two charcoal samples (oak) dated to the late medieval period (ca. thirteenth century), while elm and yew charcoal provided post-medieval dates (295 ± 31 BP and 421 ± 31 BP, respectively). Grains of oat and barley were ¹⁴C dated to the mid-/late medieval period, and a rye grain gave a late date (330 ± 35 BP) [24].

At the north-eastern end of the Dingle peninsula, immediately east of the medieval town of Tralee, archaeological investigations in the context of extensive roadworks have provided a wealth of evidence for human activity dating back to the early Neolithic [25,26]. The Bronze Age is particularly well represented, as is the Iron Age, which is unusual in the context of Irish archaeology [27,28]. Palaeoenvironmental investigations, consisting mainly of macrofossil, wood, and charcoal analyses, have greatly added to our knowledge of woodland composition in the area (especially plants that are poorly represented or silent in pollen records, such as blackthorn (Prunus spinosa) and bramble (Rubus fruticosus agg.)), woodland structure, trees/shrubs used for fires, and the farming economy, especially crops and associated weeds [28]. New insights into medieval (e.g., kilns for cereal drying and iron workings) and post-medieval developments (e.g., lime kilns, brick making, and land reclamation) have also been obtained from these excavations [28,29,30].

Figure 1. Maps showing the geographical details of the study region. The inset map of Ireland shows the county boundaries. The four main peninsulas of south-west Ireland, including Dingle, are labelled. The main sites referred to in the text are shown in the images from Google Earth. (A) Dingle peninsula, the main peaks (numbered) in the Mount Brandon/Stradbally mountain range, pollen profiles BL (Ballinloghig) and LCN (L. Camclaun), archaeological sites at Ferriter’s Cove (FC) and Ballyhoneen Wedge Tomb (WT), and main roads and settlements. (B) Tilted view towards the north. The main study area is shown in the image foreground. The locations of the Kilmore profiles KLM I and II, L. Adoon bog profiles LAV and LAH, and lake profile LCN are shown. (C) Detailed map centred on the main archaeological area (‘Central Complex’) in Kilmore. The locations of pollen profiles KLM I and II, surface pollen samples SS1 and SS2 (SS3 is off image), pre-bog stone walls W7 and W8, and the transect laid out along the line of W8 are shown in the image. The yellow dashed transect line indicates the transect is on uncut bog.

Figure 1. Maps showing the geographical details of the study region. The inset map of Ireland shows the county boundaries. The four main peninsulas of south-west Ireland, including Dingle, are labelled. The main sites referred to in the text are shown in the images from Google Earth. (A) Dingle peninsula, the main peaks (numbered) in the Mount Brandon/Stradbally mountain range, pollen profiles BL (Ballinloghig) and LCN (L. Camclaun), archaeological sites at Ferriter’s Cove (FC) and Ballyhoneen Wedge Tomb (WT), and main roads and settlements. (B) Tilted view towards the north. The main study area is shown in the image foreground. The locations of the Kilmore profiles KLM I and II, L. Adoon bog profiles LAV and LAH, and lake profile LCN are shown. (C) Detailed map centred on the main archaeological area (‘Central Complex’) in Kilmore. The locations of pollen profiles KLM I and II, surface pollen samples SS1 and SS2 (SS3 is off image), pre-bog stone walls W7 and W8, and the transect laid out along the line of W8 are shown in the image. The yellow dashed transect line indicates the transect is on uncut bog.

The investigations reported on in this paper aimed, in the first instance, at providing palaeoenvironmental contexts for the archaeological investigations that were being carried out at the time (the early 1990s) by Ó Coileáin [18,19]. Of particular interest to us, and also the archaeologist, was the woodland and land-use history, and dating of blanket bog initiation and spread, which provides a terminus post quem for stone-wall construction [31]. At the time of archaeological and palynological investigations, turf cutting for domestic fuel had removed peat from substantial areas within the Central Complex where KLM I and II were collected, but fortunately, considerable tracts of undisturbed peat remained and were available for investigation (Figure 1C; peat cutting has more or less ceased in the meantime; see photographs in Figures S10, S11, and S14). The KLM profiles lie close to those of Dodson [21], i.e., the peat profiles LAV and LAH, which lie within a kilometre to the east, and the lake-edge profile LCN, which is about 1.6 km to the south (Figure 1).

The nomenclature of higher plants in this paper follows Parnell and Curtis [32]. Bryophyte nomenclature generally follows Atherton et al. [33]. Placenames mainly follow those used by the OSI and, in the case of Dingle peninsula, the OSI Discovery Series (DS) map, no. 70, 6th edition. Cognizance, however, is taken of what is in common usage locally and in Ireland generally, as well as the placenames used in the EastWest 25series maps (www.eastwestmapping.ie; accessed on 1 December 2024). In some instances, both English and Irish versions of the placenames are given (especially where both are commonly used), the former often being merely an attempt by nineteenth-century cartographers at rendering in English spelling what the Irish version sounded like, though, in the process, the placenames invariably lost their significance as descriptors of landscape, vegetation, land use, etc., that were inherent in the Irish version. Dingle peninsula is often referred to as Corca Dhuibhne (in Irish; also in English, where other spellings such as Corkaguiny are not uncommon), which is the name of the barony that includes all but the easternmost part of the peninsula.

For details of the cultural periods referred to in the text, Waddell [34] is the main reference for prehistory, while for the historical period (beginning in Ireland at ca. 400 CE), various sources are used, including Aalen et al. [17] (see also Abbreviations). The early medieval period (CE) is regarded as extending from 400–1169 and includes the early Christian (ca. 400–800) and Viking periods (ca. 800–1169). The late medieval period, also referred to as the Anglo-Norman period, extends from 1169 to ca. 1550. Occasionally, reference is made to the classical pollen zones defined by Mitchell [35] for Ireland.

2. Materials and Methods

2.1. Sampling in the Field

In October 1993, a fresh face was exposed in a recently cut bank of turf, and a peat monolith, KLM I, was removed in three contiguous sections where the peat was deepest (2.23 m thick) (Figures S12 and S13). At the sampling site (52.20576 N, 10.16296 W; ∼138 m asl), there was a small localised depression of ∼10 cm in the mineral ground so that the base of the monolith was a few centimetres below the general level of the sub-peat mineral ground. A small pine stump that had been uprooted by peat cutters was observed nearby.

Prior to collecting KLM II, the uncut bog was cored along a transect that ran south-west from a recent substantial stone-wall constructed by peat cutters to facilitate drying of cut turf to a particularly large boulder that lay where the peat met steeply rising ground and ceased (Figure 1C and Figure S14). This transect followed Wall 8 (wall descriptions and numbering after Ó Coileáin [18]). The surface along the transect was levelled with respect to the surface within the nearby large stone enclosure (see Figure 1C) using a theodolite. Peat thickness and stratigraphical features were investigated along the transect using a gouge corer. Core KLM II was collected using a Livingstone piston corer (core tube: 100 cm long, diameter: 5 cm), where the maximum thickness of peat was recorded, i.e., at 80 m from the start of the transect and about 40 m west of KLM I (co-ordinates: 52.20586 N, 10.16365 W; ∼139 m asl).

Wall 8 intersects Wall 7, which runs across it (the stone walls in these parts were exposed by peat cutters; see Figure 1C). Wall 7 runs west towards the large stone enclosure referred to above, curves, and passes by the enclosure without touching it, i.e., it avoids it and presumably post-dates it (Figure 1C). This wall also ran east of Wall 8 and curved sharply northward to disappear under shallow peat. In an area where peat had not been cut, Wall 7 was excavated, and five ¹⁴C dates were obtained by Ó Coileáin [18] (Figure 1C, Figures S11 and S15; see also Section 4).

2.2. Preparation, Identification, and Counting of Fossil Pollen Samples

Monolith KLM I was sampled between 4 and 222 cm at intervals of 2 and 4 cm in the lower and upper parts of the monolith, respectively. The basal part only of core KLM II (255–285 cm) was sampled at 2-cm intervals. Samples were 1 cm thick, with the top being used to designate depth; e.g., sample 4 cm was from 4–5 cm from the bog surface at the sampling site.

Sample preparation for pollen analysis followed standard procedures. A Lycopodium spore suspension of known concentration was added at the start to enable determination of the pollen concentration. The samples were heated with 10% KOH for 10 min and sieved to remove large and unwanted debris. The sievings were later examined for macroremains (KLM I only). The mineral-rich samples were treated with 60% HF. All samples were acetolysed for 4 min, washed with glacial acetic acid, and then washed with water. Glycerin was used to store the pollen pellets and as a mounting medium.

Pollen identification and counting were carried out using a Leitz Laborlux D microscope (Leica, Wetzlar, Germany). A magnification of ×500 was used for routine counting, and ×1250 oil immersion with phase contrast was used for identifying critical pollen types. Where pollen density was low, a magnification of ×312 was used in conjunction with higher magnifications to check for difficult-to-identify pollen types.

Pollen identification was carried out to the highest taxonomic resolution wherever possible, i.e., genus or species level; however, in some cases, this was not possible. Taxa that present a particular challenge regarding identification and/or are important from the point of view of reconstruction of the palaeoenvironment include the following: Corylus and Myrica were distinguished according to the criteria of Mohr [36]. Monolete spores (-perine) consist of fern spores (probably derived from Dryopteris and related ferns, including Thelypteris palustris) that had lost their outer perine and so could not be further distinguished. Other fern spore determinations, excluding filmy ferns, follow Moore et al. [37]. Rumex-type is consistent with Rumex sect. Acetosa in Fægri and Iversen [38] and Rumex acetosa-type in Moore et al. [37]. Rumex aquaticus-type is as defined by Fægri and Iversen [38]). Rumex-type is expected to be derived from docks that are typical of grasslands (e.g., R. acetosa), while R. aquaticus-type probably derives from a dock of wet habitat, such as R. hydrolapathum. Anagallis tenella-type (as in Moore et al. [37]) is most likely A. tenella, a common blanket-bog species. Saxifraga hirsuta-type, as described by Moore et al., potentially includes pollen of S. spathularis and S. hirsuta, both of which have been recorded in the locality [10]. Tricolporate, echinate Asteraceae (here referred to as Tubuliflorae, as distinct from Liguliflorae, e.g., Taraxacum) potentially includes pollen from several common composite daisy-like species. Only Centaurea nigra-type pollen (most likely derived from C. nigra) is plotted separately here (profile KLM I). Fabaceae (Leguminosae) potentially includes pollen from Trifolium and Vicia spp. (distinguished during counting but aggregated in the pollen plots). Ulex was distinguished and plotted separately (surface sample SS-3). Caryophyllaceae includes mainly Cerastium-type pollen (cf. Fægri and Iversen [38]).

Cereal-type pollen was defined according to the criteria described by Beug [39,40]. These pollen grains were further categorised based on their size (maximum length). Cereal-type pollen in the size range of 37 to 39 µm is, however, included in Poaceae and is regarded as most likely arising from non-cultivated grasses. Secale (rye) was distinguished from other cereal-type pollen based on the shape, size, and position of the pores, as described by Beug [39,40].

Vaccinium-type includes all ericoid pollen (tetrads) other than Calluna vulgaris, Erica tetralix, E. cinerea, Empetrum nigrum, and Arbutus unedo. A. unedo tetrads were distinguished with confidence from those of other tetrad-producing ericoid species based on large size (generally ≥50 µm), long colpi, and rather smooth surface (see Beug [39,40], Foss and Doyle [41]; Moore et al. [37]), and after comparison with modern reference samples (see below).

Non-pollen palynomorphs (NPPs) that were identified and counted include testate rhizopods such as Amphitrema flavum, Assulina, and Hyalosphenia subflava; crustaceans such as Copepoda; ascospores including Gelasinospora; and other fungal spores. Mougeotia gracillima and other Zygnemataceae taxa were also counted, as were fragments of the epidermis of E. tetralix seeds and charcoal particles ≥ 37 µm in size. Rhizopods were identified using Grospietsch [42], and other NPPs were identified using Van Geel [43]. Type numbers preceded by ‘HdV-’, in addition to the names that are in common use, are shown for NPPs in the pollen diagrams (cf. Shumilovskikh et al. [44]).

Type slides in the Palaeoenvironmental Research Unit (PRU), University of Galway, were consulted for all critical taxa, especially A. unedo and filmy fern spores. Regarding the filmy ferns, size statistics and other details derived from the study of modern materials were compiled and are presented in Table S1. These results show that H. wilsonii and H. tunbrigense can be distinguished with confidence based on size (88 µm, 74–111 µm vs. 48 µm, 36–65 µm; average sizes and size ranges are cited) and spore features, including shape, trilete marking, and size and distribution of projections on the spore surface. T. speciosum is similar to H. tunbrigense in that it is rather small (average size: 55 µm; size range, however, is large, i.e., 39–80 µm), but it can also be differentiated with confidence based on the shape of the spore, characteristics of the trilete mark, and shorter surface processes (Table S1).

High pollen counts (≥1000) were striven for, but this was not always feasible in KLM I because of low pollen concentrations. In both KLM I and II, total terrestrial pollen (TTP) was used as the pollen sum (PS). The PSs in KLM I and KLM II averaged 689 and 932, respectively, which can be expected to provide an adequate basis for percentage calculations. Bog and aquatic taxa are excluded from the PS, as well as Alnus (excluded because of exceptionally high values at the base of the profile) and Sphagnum. The percentage values for the taxa outside the PS are based on TTP and the sum of the taxa in a particular category, e.g., bog/heath. In some instances, the particular category consisted of only a single taxon, e.g., Alnus, Sphagnum, Equisetum, and E. tetralix epidermal fragments. In these instances, and in the case of NPPs, percentages are expressed relative to TTP plus the particular taxon. Concentration (grains cm⁻³) and influx (grains cm⁻² y⁻¹) values were calculated in the usual way based on the number of Lycopodium spores (exotic) counted. Regarding the surface samples, Alnus, which has only moderate values, is included in the PS, as is the normal practice.

To summarise the pollen data, individual curves were combined into composite pollen curves, including AP (arboreal pollen) and NAP (non-arboreal pollen). NAP is subdivided into NAP1 (pollen indicative mainly of grasslands) and NAP2 (pollen of cultivated plants and weeds associated with arable/disturbed habitats). Pollen assemblage zones (PAZs), based mainly on the major movements of the percentage pollen curves, were distinguished by careful inspection of the percentage plots.

2.3. Macrofossil Analyses

The material retained on the 100-µm-mesh sieves during the initial stages of pollen preparation was examined using a Wild M8 binocular microscope (Wild Heerbrugg, Heerbrugg, Switzerland) and scored for abundance on a scale of 1–4, i.e., rare, occasional, frequent, and abundant. E. tetralix seeds were identified using type slides and photographs by Grosse-Brauckmann [45]. Juncus effusus-type seeds potentially include those of J. effusus and J. conglomeratus [46]. Mosses were identified using the descriptions provided by Smith [47]. The records mainly consist of Sphagnum leaves. S. papillosum was differentiated, as were other cucullate-leaved Sphagna that probably include S. magellanicum and S. palustre (referred to as S. sect. Cymbifolia (=Sect. Sphagnum). S. austinii (=S. imbricatum subsp. austinii) was not recorded. All other Sphagnum leaves (narrow-leaved, non-cucullate-leaved specimens) were referred to Sphagnum undiff. The abundance of mineral particles and charcoal fragments (referred to as macro-charcoal, i.e., fragments ≥ 100 µm) was also noted.

2.4. ¹⁴C Dating

Six peat samples, each 1 cm thick and ∼60 g wet weight, were taken from monolith KLM I and submitted for ¹⁴C dating. ¹⁴C dates were calibrated using OxCal ver. 4.4.4 [48] and the IntCal20 calibration curve [49]. Age/depth models for profile KLM I and other profiles were produced using Clam v. 2.3.8 [50].

2.5. Determination of Ash Content and Tephra Detection

Samples were taken for ashing from depths in monolith KLM I corresponding to the pollen-analysis samples. Core KLM II was sampled continuously every centimetre from 285 cm (base) to 255 cm (top). The samples were dried at 70 °C for 24 h, cooled in a desiccator, and the dry weight was ascertained (about 3 g and 1 g for KLM I and II, respectively). The samples were then ashed at 650 °C for 4 h. After cooling, the residue (ash) was weighed, and the ash content was calculated as a percentage of sample dry weight.

Later (in 2004/05), KLM I was sampled continuously for tephra by taking 4-cm-thick peat slices and ashing these following similar procedures to those described above [51]. Ashing took place, however, at a lower temperature (550 °C) to minimise damage to any tephra present, but for a longer time (5 h). After ashing, the mineral matter was washed with 10% HCl to remove the surface coating from the tephra shards and sieved using a 24-µm-mesh sieve to remove small particles. Glycerol was added to the material retained on the sieve, and the suspension was plated onto slides and examined microscopically at ×250 magnification. Tephra was identified based on viscularity with the aid of a polariser. Tephra shards do not rotate plane-polarised light (PPL); therefore, they become invisible upon rotation of the polariser [52]. In contrast, quartz grains remain visible as bright entities. Similar to tephra, phytoliths, i.e., silicic plant inclusions, do not rotate PPL, but they can be distinguished from tephra by their rough edges and lack of vesicles.

One or more tephra shards were detected in several of the 4-cm-thick samples but only in two contiguous samples, i.e., 49–53 cm and 53–57 cm, which had elevated numbers of shards (see Section 3). This 8-cm interval was subsampled by taking 1-cm-thick samples and ashing following the procedures outlined above. Tephra shards were distinguished and counted in the same way as described for the initial sampling.

2.6. Pollen Analysis of Surface Samples

Surface samples, consisting mainly of Sphagnum moss, were collected from within three relevés at a depth of ∼2 cm, prepared for pollen analysis, and the pollen was counted (locations are indicated in Figure 1C). Sample preparation followed procedures similar to those used for preparing the peat samples, but an exotic was not added. Relevés were taken at the sampling sites as follows: Relevé 1 was taken in the vicinity of coring site KLM II. Molinia caerulea and sedges (Schoenus nigricans, Eriophorum angustifolium, and Rhynchospora alba) were abundant. Relevé 2, at about 95 m on the transect, was from the edge of a 1-m-wide, shallow, drainage channel on the bog surface. M. caerulea and aquatic species, especially Potamogeton polygonifolius, were abundant. Relevé 3 was from the east of the Scorid river in the immediate vicinity of a fulacht fiadh. Seven holly trees (Ilex aquifolium), five with berries, were recorded locally.

3. Results

3.1. Peat Stratigraphy, Ashing and Tephra Data

The results of the stratigraphical investigations along the bog transect are shown in Figure 2. The data relating to peat stratigraphy, ashing, and tephra investigations of monolith KLM I and core KLM II are shown in Figure 3 and Figure 4, and details are given in Tables S2–S4.

The peat body along the transect was generally 2 to almost 3 m thick (maximum thickness recorded: 290 cm) (Figure 2 and Table S2). The lower peat was usually highly decomposed and had a fine-fibrous texture. Towards the surface, the peat was more fibrous, and decomposition was lower. Occasionally, E. vaginatum remains were recorded, mainly at mid-depth. Substantial pieces of Betula wood were recorded at the 90 m point. Charcoal was not conspicuous (see detailed stratigraphy for KLM II in Table S4). The sub-peat soil was silty in texture. Stones were encountered at the base of coring points 80 m, 90 m, and 100 m.

Figure 2. Schematic representation of the peat body and underlying mineral ground based on coring and levelling along a transect where core KLM II was collected (at 80 m). The vertical exaggeration is ×10 (approximately).

Figure 2. Schematic representation of the peat body and underlying mineral ground based on coring and levelling along a transect where core KLM II was collected (at 80 m). The vertical exaggeration is ×10 (approximately).

The stratigraphical features of monolith KLM I are as follows (Figure 3, Figure 4 and Figure 5 and Table S3). The sub-peat mineral soil was silty and had a high stone content. In the peat column, changes in decomposition and texture were gradual. The basal part (192–223 cm), which contained substantial pieces of birch wood (B. pubescens), consisted of highly decomposed peat. The most pronounced feature is a distinct birch layer between 126 and 120 cm (Table S3). This layer, embedded in dark, well-decomposed peat, was traced laterally along the peat bank on either side of KLM I. A 3-cm-thick layer of dark, charcoal-enriched peat was present immediately beneath the birch layer. E. vaginatum remains were occasionally recorded but were not prominent.

Detailed stratigraphical information for core KLM II is available only for the lower part of the core (Table S4; also see Figure 6A₂). Substantial pieces of birch wood were recorded at 211 cm. Below 274 cm (to 281 cm), the peat was highly decomposed. A charcoal-rich layer was present between 278 and 280 cm. Peat with sand followed (281–284.5 cm). This rested on mineral ground consisting of light grey, silty soil (284.5–285 cm). During initial probing, 290 cm of peat was recorded in the vicinity of this coring point, which indicates a locally uneven mineral ground surface, as was the case at KLM I.

The basal peats in KLM I and II yielded elevated ash values (>20% starting at 210 cm and 283 cm in cores KLM I and II, respectively; Figure 3A,D, and Figure 6A₂). This suggests that as peat initially accumulated, it was under soligenous influence, and mineral ground was likely still exposed locally. The ashing results from KLM I, based on 1-cm- and 4-cm-thick slices, are broadly similar (Figure 3A,B). Average values of 2.1 ± 0.7% and 2.3 ± 0.8% (n = 48 and 50, respectively) were recorded (note: the lowermost highly minerogenic samples have been excluded from these statistics). Sample 153–157 cm gave 16% ash, which is regarded as an error and hence is excluded from the statistics. There was a general trend, though not pronounced, towards declining ash values until ∼60 cm (Figure 3A,B). Above this depth, the values were generally elevated, especially near the top.

Elevated tephra-shard counts were recorded at 49–53 cm and 53–57 cm (16 and 10 shards, respectively; Figure 3C). In the 1-cm thick slices, sample 53 cm (i.e., 53–54 cm) yielded the greatest concentration of shards (110 shards; samples 52 and 54 cm yielded only 6 and 8 shards, respectively; Figure 3E), which suggests a distinct tephra layer at 53 cm. Most of the shards in sample 53 cm (93 out of 110) ranged in size from 20–90 µm (Figure 3F), and four shards > 100 µm were recorded. The shards were all colourless; most shards were vesicular (>80%), about 50% had air bubbles, and 13% had only elongated channels [51].

The tephra layer is bracketed by the uppermost two ¹⁴C dates. These dates suggest that the tephra layer dates to the late thirteenth or early/mid-fourteenth centuries, i.e., at a time when there were several known Hekla eruptions [53,54]. Hekla₁₃₀₀ lasted an entire year and was a major eruption with a Volcanic Explosivity Index (VEI) of 4. However, there are only a few records of the Hekla₁₃₀₀ eruption from Ireland [53]. Hekla₁₃₄₀ was of medium size (VEI-3), and Hekla₁₂₂₂ and Hekla₁₃₈₉ were smaller and thought to be of low magnitude (VEI-2 and VEI-3, respectively). These four eruptions are hence regarded as unlikely sources of the tephra layer recorded in KLM I. Hekla 1 (1104 CE), which is well-defined at several Irish sites [53,55], is clearly too old and is therefore excluded from consideration. This leaves Öræfajökull₁₃₆₂ as the most likely candidate. Öræfajökull₁₃₆₂ was a major eruption in south-east Iceland that began in the early part of 1362 CE, lasted a few months, had a substantial impact in Iceland, and produced widely dispersed tephra that has been recorded in Greenland ice cores and as far east as Germany and Poland [53,54,56]. There are also several Öræfajökull₁₃₆₂ records from Ireland, including sites in western and south-western Ireland [52,57,58,59]. Thus, the available evidence points to Öræfajökull₁₃₆₂ as the source of the KLM I tephra layer. Accepting this, the historical date of the eruption was used to provide a fix for the chronology in this part of KLM I. As there is no evidence in the stratigraphy or pollen for rapid changes in peat accumulation rates (which would follow if the tephra and the two uppermost ¹⁴C dates were accepted as valid for the depths to which they pertain), the ¹⁴C dates were not taken into account in the construction of the age/depth curve. Using the ¹⁴C dates in preference to the tephra-derived dates results in somewhat older ages (but ≤100 y) for this part of the profile. Overall, however, the differences between accepting the tephra vs. the ¹⁴C dates are small; therefore, the approach taken appears to be justified (see also ¹⁴C dating, which follows).

Figure 3. Plots showing the results of ashing (cores KLM I and II) and tephra-shard counts (KLM I). Ash values, cores KLM I and II (1-cm-thick slices) ((A,D), respectively); ash values, core KLM I (4-cm-thick slices) (B); shard counts, 4-cm-thick slices (C); shard counts in 1-cm-thick samples, samples 49–55 cm (E); shard-size distribution in sample 53 cm (F). In (B), sample 153–157 is marked with an asterisk. The ash value (16%) is unexpectedly high, due presumably to experimental error.

Figure 3. Plots showing the results of ashing (cores KLM I and II) and tephra-shard counts (KLM I). Ash values, cores KLM I and II (1-cm-thick slices) ((A,D), respectively); ash values, core KLM I (4-cm-thick slices) (B); shard counts, 4-cm-thick slices (C); shard counts in 1-cm-thick samples, samples 49–55 cm (E); shard-size distribution in sample 53 cm (F). In (B), sample 153–157 is marked with an asterisk. The ash value (16%) is unexpectedly high, due presumably to experimental error.

3.2. ¹⁴C Dating and Construction of an Age/Depth Model

The six ¹⁴C dates relating to KLM I appear to be generally acceptable (Figure 4; Table S6). The lower four ¹⁴C dates were accepted as reliable evidence of age. The upper two dates, however, are regarded as somewhat too old, mainly on the basis of the presence of a micro-tephra layer at 53 cm that is attributed to Öræfajökull₁₃₆₂ (see above) and so are not used in the age/depth model. The uppermost levels are constrained by two dates assigned on palynological grounds. At 20 cm, a curve for Pinus recommences, which is assumed to reflect an increase in pine pollen arising from the widespread planting of pine that had already commenced by the mid-eighteenth century in Ireland [60], including Co. Kerry [61,62,63]. In the uppermost pollen spectrum at 4 cm, Pinus representation did not greatly increase, which suggests that this spectrum predates the increased afforestation of the later twentieth century. Therefore, the top of the pollen profile is unlikely to extend beyond 1950 CE, i.e., the age assigned to depth 4 cm.

Figure 4. Plot of age/depth curve for profile KLM I, as produced by Clam. The spline curve (smoothing = 0.25) is based on ¹⁴C dates (uppermost dates, in red, were excluded within Clam) and the tephra layer. The key to the stratigraphic symbols is shown in Figure 3.

Figure 4. Plot of age/depth curve for profile KLM I, as produced by Clam. The spline curve (smoothing = 0.25) is based on ¹⁴C dates (uppermost dates, in red, were excluded within Clam) and the tephra layer. The key to the stratigraphic symbols is shown in Figure 3.

Overall, the age/depth curve suggests more or less constant and low peat accumulation rates in the lower part of the profile (at/below 140 cm: average value: 0.4 mm y⁻¹), increasing rates between about 140 and 70 cm, and followed by a decline that stabilised towards the top of the profile (average 0.85 mm y⁻¹, from 52 cm to the top; this is probably artificially low as a result of drainage arising from peat cutting). Additional ¹⁴C dates would reveal greater variation in peat accumulation rates. This is hinted at by changes that suggest slow rates, including dark and more decomposed peat, very high pollen concentration and influx values in zone KLM I-1 and briefly elevated values at about 130 cm (immediately below the woody birch layer) (Figure S1).

Profile KLM II does not have independent evidence of age; therefore, a precise chronology cannot be attached. Given its location and context, it is assumed that the base of KLM II is similar in age to KLM I (4000 cal. BP) or perhaps somewhat older (based on the Pinus curve). P. lanceolata begins to increase substantially at ca. 3300 cal. BP in KLM I. A similar rise was recorded at 267 cm in KLM II (P. lanceolata attained >14% at 265 and 263 cm in mid-zone KLM II-3). As this development is probably of more than merely local significance, this feature in KLM II is assumed to be of a similar age as the rise in P. lanceolata in KLM I (3300 cal. BP). Accepting that this is broadly correct, the basal peat in KLM II has accumulated at 0.28 mm y⁻¹, which is within the same order of magnitude as that estimated for KLM I (0.4 mm y⁻¹). Assuming a steady accumulation rate in KLM II, this profile extends to ca. 2800 cal. BP. However, it may be older, as the rate of peat accumulation in the basal part most likely increased over time, as is the case in KLM I.

3.3. Pollen Analysis: General Considerations, Conventions in the Pollen Diagrams and Surface-Pollen Data

Plots of the pollen data, i.e., the surface samples and pollen profiles KLM I and II, are presented as percentage diagrams in Figure 5 and Figure 6, and as concentration and influx diagrams in Figure S1 (KLM I only). A summary of the KLM profiles and the surface-sample pollen data is available in Table S8. Pollen profiles LAH and LAV by Dodson [21,22] are presented in Figure 7. An overview of the percentage pollen data derived from the peat cores and summary interpretations in graphic form are provided in Figure 8. The relevés from the sites where surface samples were collected are listed in Table S5.

The conventions followed in the construction of pollen diagrams are as follows. Curves that are based on PSs that deviate from what is usually employed have names that include an asterisk to alert the reader that there is a deviation from normal practice (see Section 2). A lighter shade of brown is used for bog/heath taxa indicative of wet mire conditions and similarly for NPP taxa.

The surface sample data (three spectra) demonstrated a major contribution from local pollen producers (Figure 6B₁,B₂; Table S8). Poaceae, for instance, average 69%. This is attributable to M. caerulea, a grass that is important both locally and regionally on extensive intact and cutover bogs. The local presence of P. polygonifolius resulted in a strong representation of Eu-Potamogeton in SS2 (20%). The AP component in the surface samples was low (average: 15%, consisting mainly of Pinus and Betula). This undoubtedly derives from long-distance pollen transport, as there are few trees locally or in the wider region, apart from conifer plantations, which, at the time of sampling, were not yet extensive/mature. In SS3, Ilex achieved 5.9%, which, although high for a shrub with poor pollen dispersal, is not surprising given the presence of several mature holly specimens locally (though often female and hence not expected to produce pollen [64]). Despite the high Poaceae values that lower the percentage representation of other taxa, taxa indicative of grasslands on mineral ground are well represented. This suggests that these taxa, and especially P. lanceolata, produce substantial amounts of widely dispersed pollen.

A puzzling aspect of the surface-pollen spectra is the high cereal-type pollen values. This is especially true given that, during the later part of the twentieth century, arable farming greatly declined here, as in most parts of western Ireland ([65]; Figure 6B₁; see Section 4.3). Not only cereal-type pollen of medium size (40–45 µm) but also large pollen, including pollen ≥ 50 µm (Secale, too; a single grain is recorded in SS2), are well represented. The most likely explanation is that several decades, rather than a few years of pollen deposition, as originally thought, are represented in the surface samples.

As regards bog/heath taxa, several are under-represented and, in particular, shrubby species, including ericoids such as Calluna, E. tetralix (not recorded at all), and E. cinerea. Bog myrtle (Myrica), which was present at the three sites (and plentiful at SS3), was poorly represented. This should not be attributed to misidentification, as Corylus is also poorly represented (Myrica and Corylus pollen are difficult to separate; see Section 2). Cyperaceae are strongly represented, which is not surprising given the important role of sedges (Carex spp., Rhynchospora, and Eriophorum (bog cotton), etc., especially at SS1 and SS2). Also interesting is the strong representation of Amphitrema flavum, Assulina, and Copepoda in SS1 and SS2, particularly in SS1, where humid rather than wet conditions prevailed. In SS3, where dry conditions prevailed, these taxa (also Eu-Potamogeton) were poorly or not at all represented.

Figure 5. Percentage pollen diagram, profile KLM I. Curves with broken lines are exaggerated ×10. T = tephra layer (cf. Öræfajökull₁₃₆₂). A taxon name with an asterisk indicates that the taxon in question is outside the PS and/or a PS is used that differs from that used in neighbouring taxa (see Section 2). Rare taxa are indicated by a dot, and taxon names are abbreviated. ¹⁴C dates, dates for PAZ boundaries, a peat accumulation rate curve, macrofossil records, and stratigraphy are shown (see (A)). Intensive farming phases are shown in (B). Dark/light brown fill in curves for bog/heath taxa and NPPs (in (C)) is used to convey the idea of dry/wet conditions on the bog surface (also in Figure 8C₁).

Figure 5. Percentage pollen diagram, profile KLM I. Curves with broken lines are exaggerated ×10. T = tephra layer (cf. Öræfajökull₁₃₆₂). A taxon name with an asterisk indicates that the taxon in question is outside the PS and/or a PS is used that differs from that used in neighbouring taxa (see Section 2). Rare taxa are indicated by a dot, and taxon names are abbreviated. ¹⁴C dates, dates for PAZ boundaries, a peat accumulation rate curve, macrofossil records, and stratigraphy are shown (see (A)). Intensive farming phases are shown in (B). Dark/light brown fill in curves for bog/heath taxa and NPPs (in (C)) is used to convey the idea of dry/wet conditions on the bog surface (also in Figure 8C₁).

Figure 6. Percentage pollen diagrams, profile KLM II (A₁–A₃), and surface samples (B₁,B₂). The conventions followed are similar to those in Figure 5. In (B₁,B₂), the x-axis is magnified ×10 in the case of non-filled histograms; Poaceae is scaled ×0.5 in (B₁). A taxon name with an asterisk indicates that the taxon in question is outside the PS and/or a PS is used that differs from that used in neighbouring taxa (see Section 2).

Figure 6. Percentage pollen diagrams, profile KLM II (A₁–A₃), and surface samples (B₁,B₂). The conventions followed are similar to those in Figure 5. In (B₁,B₂), the x-axis is magnified ×10 in the case of non-filled histograms; Poaceae is scaled ×0.5 in (B₁). A taxon name with an asterisk indicates that the taxon in question is outside the PS and/or a PS is used that differs from that used in neighbouring taxa (see Section 2).

Figure 7. Percentage pollen profiles from blanket-bog peat in the L. Adoon area drawn to time scales (after [22]). The conventions followed are similar to those used in the KLM profiles. (A). Peat profile LAH from the eastern side of L. Adoon, 600 m SE of LAV. (B). Peat profile LAV from east of the Scorid river, near the Ballyhoneen wedge tomb. In (B) Monolete spores (-per.), i.e., Monolete fern spores without perine, an asterisk is added to the name to indicate that the taxon is outside the usual PS; its PS is TTP + Monolete spores (-per.) (see Section 2).

Figure 7. Percentage pollen profiles from blanket-bog peat in the L. Adoon area drawn to time scales (after [22]). The conventions followed are similar to those used in the KLM profiles. (A). Peat profile LAH from the eastern side of L. Adoon, 600 m SE of LAV. (B). Peat profile LAV from east of the Scorid river, near the Ballyhoneen wedge tomb. In (B) Monolete spores (-per.), i.e., Monolete fern spores without perine, an asterisk is added to the name to indicate that the taxon is outside the usual PS; its PS is TTP + Monolete spores (-per.) (see Section 2).

Figure 8. (A_1–3,B₁) Summary percentage peat pollen profiles from Kilmore/L. Adoon area drawn to time scales. (B₂,C_1–3) Schematic reconstruction of vegetation, land use, and local bog-surface wetness/dryness based on pollen profiles KLM I and II, LAV, and LAH. Alnus with an asterisk (in A₁,B₁) indicates this taxon is outside the usual PS; its PS is TTP + Alnus (see Section 2).

Figure 8. (A_1–3,B₁) Summary percentage peat pollen profiles from Kilmore/L. Adoon area drawn to time scales. (B₂,C_1–3) Schematic reconstruction of vegetation, land use, and local bog-surface wetness/dryness based on pollen profiles KLM I and II, LAV, and LAH. Alnus with an asterisk (in A₁,B₁) indicates this taxon is outside the usual PS; its PS is TTP + Alnus (see Section 2).

3.4. Pollen Profiles KLM I and KLM II: Woodland Dynamics and Human Impact

As regards the pollen source area for profiles KLM I and II, at the base of the profiles, it is undoubtedly mainly local. As peat accumulates and extends laterally, the source of the non-bog pollen takes on a more regional character, which, at least regarding farmland, relates to the lowlands that lie mainly to the north. Given the predominantly south-westerly and often strong winds, the vegetation of the uplands to the south and south-west, and especially that on the lower slopes near the sampling sites, can also be expected to be strongly expressed in the pollen records.

The Poaceae curve is assumed to include mainly pollen of M. caerulea, a species of wet grasslands and also blanket bog. P. lanceolata, on the other hand, is a species that is more or less confined to grassland and fallow and mineral ground generally [66], and so, in conjunction with other NAP such as Asteraceae (this includes Tubuliflorae (daisy, etc.) and Liguliflorae (dandelion, hawkweed, etc.)), provides a reliable indicator as to the origin of the Poaceae pollen. High Poaceae pollen representation, when P. lanceolata, etc., are poorly represented, can be regarded as mainly reflecting the dominance of M. caerulea at/near the sampling site, i.e., bog/heath vegetation rather than grasslands on mineral soils.

Profile KLM I, which spans from the Early Bronze Age (1900 BCE) to the near present time (mid-twentieth century), indicates that the local woodlands during most of this almost 4000-year-long period consisted mainly of birch, hazel, and alder (Figure 5; also Figure 8A₁,C₁). Profile KLM II (Figure 6 and Figure 8B₁), which began somewhat earlier, probably did not extend beyond the Bronze Age (ca. 1000 BCE) (see Section 3.2). Both KLM I and KLM II indicate the local importance of woody vegetation as peat accumulated in the study area. As KLM I opened, tree cover and human impact appeared to be broadly comparable to that at KLM II, although alder was not as dominant, and the increase in human activity adversely affected birch more than alder. Both profiles suggest that pine is rare. However, a small pine stump, presumably from the lowermost peat in the general vicinity of KLM I (see Section 2), indicates that pine trees survived as peat accumulation commenced locally.

At the base of KLM II (Zone 1 and much of Zone 2), holly and oak, together with alder and birch, made important contributions. Ilex achieved 35.5% in Zone 1 but declined in Zone 2; Quercus achieved 12.5% in Zone 2, where Alnus declined and Betula expanded. The changing contribution of trees and tall shrubs at KLM II (and KLM I) in the context of local peat initiation may be the result of a relatively short period with low levels of human impact. The reduction in farming activity was probably the facilitator, rather than the result, of peat initiation and woodland regeneration. The subsequent increase in human activity (substantial rise in P. lanceolata in both KLM I and II; Figure 5B and Figure 6A₁) is probably largely responsible for the decline in alder and other associated changes. Records for filmy ferns, and especially H. wilsonii and H. tunbrigense in profile KLM II and a record for T. speciosum, as well as a curve for H. wilsonii and several records for H. tunbrigense in the basal part of KLM I, are noteworthy (Figure 5A and Figure 6A₁).

Regarding tree cover in the Iron Age and subsequently (from about 600 BCE; starting in zone KLM I-3), the main changes are as follows (Figure 5 and Figure 8A₁,C₂). In subzone KLM I-3a (600–90 BCE), AP values are very low, but this is probably attributable more to exceptionally high Poaceae values depressing percentage AP values rather than an actual decrease in AP pollen input; this finds some support in the concentration and influx pollen diagrams (Figure S1). Low values for P. lanceolata and other NAP taxa indicative of human impact support the conclusion that Poaceae pollen is largely of local (bog) origin. Subzone 3b to the top of Zone 5 (90 BCE to 1210 CE), with the exception of Zone 4 which is of short duration (see below), is characterised by elevated AP, particularly Betula, but Quercus and Corylus are also well represented. The role of human impact and climate change on vegetation development during this period, which spans the late Iron Age to the late medieval period, is considered below (Section 3.6).

Beginning at about 1380 CE, there was another shift that involved increased Corylus representation but substantially lower Betula and Quercus values. Profile KLM I ends with the lowest AP values overall in subzone 7a (1720–1870 CE), followed by an increase, especially in Betula, as the profile ends (subzone 7b; 1870–1950 CE). These changes in AP were probably mediated by changing levels of human impact resulting from particularly large demographic shifts (see Section 4.3).

3.5. First Fossil Records for Arbutus and Filmy Ferns from Dingle Peninsula

Of particular note are the records for Arbutus (almost certainly A. unedo) in sample 122 cm, profile KLM I (1.4%; nine Arbutus tetrads recorded), which dates to 540 CE (Figure 5A; Table S8). In this part of the profile, every second centimetre was sampled, and thus, a high time-resolution record is available. As each sample integrates about 20 years of pollen production, it is likely that there was sufficient time between the sample at 124 cm and the following sample at 122 cm for Arbutus to establish locally and produce the pollen recorded here (in cultivation A. unedo starts to flower when eight years old; under natural conditions at about 25 y [67]). Given that its pollen dispersal is poor [68,69], our records indicate the local presence of the strawberry tree. The precise circumstances under which this species grew and flowered at Kilmore are of particular interest. It should be borne in mind that its range seems to be largely static in Ireland since scientific recording of flora commenced [10]. In addition, A. unedo was once much more frequent in the Kerry/West Cork region [62,70], so past presence on Dingle peninsula, without any human intervention, cannot be ruled out. Until now, however, the evidence for presence has been limited to placenames, e.g., Árd na Caithne [71,72] (also rendered Ardnacaithne; in English ‘high ground of the strawberry tree’; today usually referred to as Smerwick) near the tip of the peninsula (Figure 1A), but it is by no means certain that ‘Caithne’ in that placename translates to Arbutus [73].

Climate change may have been involved in the spread of A. unedo, given that there is evidence of climate fluctuations at this time [74,75]. Immediately preceding the Arbutus records in KLM I, Betula achieved exceptionally high pollen representation, which coincided with the local expansion of birch on the bog surface (Figure 5A and Figure 8A₁,C₁). This development was preceded by a dry phase indicated by strong Calluna representation and a sharp decline in bog taxa, such as Cyperaceae and Narthecium, which are indicative of wet conditions (Figure 5C and Figure 8A₁,C₁). The dry phase probably triggered colonisation of the bog surface by birch (for the phenomenon of birch colonising mainly shallow blanket-bog peat, see [76]; for recent examples of a birch on Irish bogs, see [77]). Although the evidence for climate change is rather strong, we favour introduction by people as the preferred explanation, especially given the static range of Arbutus in recent times (see above). Our fossil records coincide with an upsurge in human activity that is a feature of the early medieval (Christian) period in Ireland generally, and specifically on Dingle peninsula [6,27,78], including Kilmore. In profile KLM I, the Arbutus records coincide with the start of exceptionally high values for cereal-type pollen that signal the beginning of a short period (it lasted not more than about three human generations) centred on the seventh century CE (PAZ KLM I-4; farming Phase II; Figure 5 and Figure 8C₁). During this period, arable farming assumed particular importance locally. Woody vegetation declined, presumably due to sustained human impact, which likely caused the local extinction of A. unedo. For recent further discussion of A. unedo, and specifically native versus archaeophyte status in Ireland and its biogeographic affinities, see [20,72,79].

The filmy fern record in KLM I is noteworthy. H. wilsonii has a continuous curve at the base of the profile (KLM I-1, 2a and the lower part of 2b; ca. 1900–1250 BCE; Figure 5A). It is also well represented at the top of KLM I-3b and in the lower spectra of KLM I-5a (early medieval period). In the upper part of the profile, there are occasional records. H. tunbrigense is not as strongly represented. It was recorded in only 1.5% of the spectra, mainly as single spores in the basal spectra. This contrasts with H. wilsonii, which was recorded in 34.3% of the spectra and attained a maximum of 6%, which suggests that it was locally common. Although poorly represented, H. tunbrigense was almost certainly present locally, especially in the early part of the record (prior to 1500 BCE). The only record of T. speciosum was a single spore in the basal spectrum. Local presence is possible, indeed likely, given its recent distribution in the Cork/Kerry region (Figure S6C,D)—but long-distance spore transport is also a possibility.

3.6. Farming Phases and First Fossil Record for Buckwheat Cultivation in Ireland

As regards human impact and farming, particularly intensive phases are identified (Figure 5 and Figure 8).

Phase i. This corresponds mainly to zone KLM I-2 and spans at least 700 years, ending at ca. 600 BCE. This involved both pastoral and arable farming (Figure 5B and Figure 8C₁). In zone KLM I-1, cereal-type pollen is also well represented, so arable activity seems to have featured throughout the mid-and late Bronze Age. The impact on local woodland cover was substantial (high Pteridium values indicate increased openness), with alder and birch adversely affected. The filmy fern record is interrupted at the transition KLM I-2a/2b, which also points to a decline in woodland in the general vicinity of the coring location.

Phase ii. The main activity during this short-lived phase (≤100 y; zone KLM I-4) is centred on 600 CE. It involved mainly arable farming (cereal growing). Maximum cereal-type pollen values for the profile as a whole are recorded. P. lanceolata is scarce, so the peak in Poaceae is probably best attributed to M. caerulea. The sharp decline in AP (mainly Betula) and the large increase in Poaceae suggest that woodlands were strongly and adversely affected. The first records of Secale pollen relate to this phase. However, only occasional grains of Secale have been recorded in the profile. Rye was not an important crop; indeed, it may have had only an accidental presence in other cereal crops (most likely barley, and perhaps oats and some wheat [24]; see Section 1).

Phase iii. This phase (ca. 750–1200 CE) broadly corresponds to KLM I-5. In the early years, there was little cereal cultivation. As farming intensified (cf. high P. lanceolata; also much cereal-type pollen indicative of a mixed farming economy), woody vegetation was impacted, including oak, as the phase ended (upper part of subzone 5b). At this point, P. lanceolata registered 30%, and Rumex-type also peaked. This indicates that typical grassland species, and especially ribwort plantain, were present in abundance but not locally, as presumably, the bog was, by now, extensive (at the sampling site, peat had achieved a thickness of almost 1.5 m, and peat accumulation was well initiated where Wall 7 was excavated).

Phase iv. This phase (ca. 1300–1700 CE) broadly corresponds with KLM I-6b and involved both pastoral and arable farming. Mid-fourteenth-century developments, such as the Black Death, adverse climate, and political upheavals (cf. [80]), are not obvious from the data. Pastoral farming was most important (cf. P. lanceolata) towards the end of the period, when oak (remnants of oak woodlands had probably survived locally up until then) was severely affected. Hazel, however, flourished, which suggested that farming, though important, was not so intensive in the region as to result in complete clearance of woody vegetation. A farming economy that relied on semi-open wood pastures rather than completely open pastures, as in recent times, is also a distinct possibility (cf. [81,82]).

Phase v. This phase corresponds mainly with KLM I-7a, where P. lanceolata averages 13.6%. This is the most intensive farming phase and spans ca. 1820 to 1950 CE. Potatoes, which do not appear in the pollen record [83], rather than cereals, were the main crops at this time. A detailed consideration of developments during this period is provided in Section 4.3.

Of particular interest are the fossil pollen records for Fagopyrum (presumed to be F. esculentum, i.e., buckwheat, the species that is usually sown as a crop in Europe; Figure S8B). Pollen was recorded in two spectra (depths 20 and 16 cm). The age/depth model suggests that these spectra date to ca. 1750 and 1800 CE, respectively.

Buckwheat has been a common crop in much of central Europe since the twelfth century CE [84]. In northern Germany, it was often grown in tandem with rye as the main crop in areas with low-fertility soils. Not surprisingly, many of the pollen diagrams reproduced in Feeser et al. [85] have a substantial Fagopyrum curve that frequently begins in the late medieval period, peaks in the seventeenth and eighteenth centuries, and continues into the early twentieth century [86,87]. On extensive bogs in the north German lowlands, buckwheat cultivation became associated with Moorbrandkultur, i.e., bog drainage, followed by firing of the surface vegetation and sowing buckwheat in the ash that resulted from the firing [88]. As regards Britain and Ireland, Pescott and Akeroyd in BSBI plantatlas2020.org (accessed on 12 March 2025) suggest that buckwheat was “a significant grain crop in Britain and parts of Ireland until the 19th century”. However, Greig [89] is of the opinion that although fossil Fagopyrum pollen has been recorded at British sites, buckwheat was never a popular food crop in Britain. Data presented by Dark [90] also convey the impression that it was an infrequent crop in Britain during the medieval period. However, the Oxford English Dictionary (OED) has entries for buckwheat dating back to the sixteenth century [91]. As far as Ireland is concerned, there seems to be no published pollen evidence to date, although there is further palynological evidence for buckwheat cultivation from the study area (see Section 4.1.1) that supports the idea that it was at least trialled as a crop in Kilmore before the Great Famine.

Kelly [92], in his comprehensive account of early Irish farming that is based mainly on Old Irish law texts, regards it as unlikely that ‘rúadán’ (=the red one) refers to buckwheat, given that in Europe, buckwheat is a food-plant that post-dates the Irish law texts. Furthermore, Kelly ([92], p. 225) is of the opinion that it “has never been a regular crop in Britain or Ireland”. Lucas [93] does not mention it in his account of Irish food, which also relies on early historical sources, and it is not referenced in a recently published tome on the history of Irish food [94].

From the above literature, it seems that buckwheat has never been cultivated widely in Ireland. However, more intensive searches—not made any easier on account of the various names by which it has been referred to; e.g., bockwheat, beechwheat, and French wheat [95]—indicate that it was rather widely known and used as a foodstuff. For example, Robert Payne (reprint edited by A. Smith [96], p. 10), advising on the feeding of rabbits (connies) in 1589, recommends to prospective English settlers in Ireland “you may fatte them with graines mixed with oates, brane or French wheate [i.e., buckwheat]”. Many decades later, with reference to the Confederate Wars in Ireland, military stores, including “french wheat, rye, and wheat” etc., were “certified as delivered at Cork, Dublin and Carrickfergus” in September 1642 [97], p. 31. However, Professor Wade, according to the Dictionary of Irish Biography (DIB), “a noted Irish botanist”, states in his 1814 memoir to the Bog Commissioners ([98], p. 24) that buckwheat “Is a crop not at all known in Ireland; but by good farmers in England is considered as very valuable for feeding horses, fattening hogs, and keeping poultry”. On the other hand, Professor Mackey, in his landmark Flora Hibernica ([99], p. 224), gives the following entry for Fagopyrum/Buckwheat: “Cultivated ground, but introduced by cultivation, being often sown as food for pheasants and other poultry”. Given the contradictory statements by Wade and Mackey, it appears that although buckwheat as a crop was known in early nineteenth-century Ireland, it was not widely cultivated or appreciated as a food for human consumption (floristic records from Ireland are few and relate mainly to recent years; Figure S8A).

Mitchell ([100] p. 172; also in earlier and later editions of Reading the Irish Landscape) refers to an unnamed German traveller to Ireland in 1828 who recorded that buckwheat was frequently seen as a crop along with oats and potatoes. Based on this and Old Irish law texts, Mitchell assumed that buckwheat was grown as a crop in nineteenth-century Ireland.

We have identified the German traveller referred to by Mitchell [100] as the well-known German landscape designer Hermann von Pückler-Muskau, who travelled widely in Ireland between August and December 1828, and shortly afterwards published an extensive account in Briefe eines Verstorbenen in four volumes. Volume 1 contains Letter 31, in which Pückler-Muskau described his stay at Bermingham House, near Tuam, Co. Galway, in September 1828 [101]. He was particularly struck by the deep bogs (raised bogs) that were reclaimed and extensively farmed (“gedeiht die Bruchwirthschaft”) and on which buckwheat—he refers to it as “Haidekorn”, i.e., an old version of the more common German word ‘Heidekorn’, literally ‘heath corn’ [101], p. 218—and potatoes and oats were grown. These accounts—in the original German and also in translation—were very popular at the time and, indeed, subsequently. Recently, they have been republished in English in a large-format, illustrated, and annotated volume by Parshall [102], while the letters relating to Ireland have been translated in part and published with commentaries by Bourke [103]. In the latter publication, the passage relating to buckwheat in Letter 31 was omitted.

As regards buckwheat at Kilmore, there is the possibility that it was introduced to Dingle by the Palatines from the Palatinate/Pfalz region in south-western Germany. These farmers were settled in Co. Limerick in the early eighteenth century and, later that century, moved into north Kerry [104,105,106]. However, there is no mention of buckwheat in the accounts of these industrious German farmers. Arthur Young, the ‘improving’ English landlord who travelled extensively in Ireland in the 1770s, visited Newcastle West, Co. Limerick, to report specifically on the Palatine farmers. He does not mention buckwheat, although, in his accounts, he frequently reports on arable farming, including the increasing popularity of the potato as a crop [106]. Interestingly, however, he refers specifically to the Maharees, as “famous for their corn products… famous for the best wheat in Kerry. All under the plough… they have known two crops of barley gained from the same land in one year and the second better than the first” ([106], vol. II, pp. 127, 128). The area referred to is on the north coast of the Dingle peninsula, about 10 km north-east of Kilmore (Figure 1A). This serves to emphasise the major changes in land use that have taken place during the last two centuries, as a largely arable-based economy has been almost completely replaced by pastoral-based farming. It also provides context for the exceptionally high cereal-type pollen representation in KLM I, which is difficult to envisage from a modern perspective.

4. Discussion

4.1. Palaeoecological Records from Kilmore in the Context of Earlier Published Records

The geographically closest and most pertinent pollen profiles relating to the Kilmore area are those by Dodson [21,22] (LAV, LAH, and CLN) and Barnosky [23] (BL) (see Section 1; site locations in Figure 1A,B). To facilitate comparison with the KLM profiles, these profiles were replotted on time scales (Figure 7, Figure 8 and Figure 9). Details of the profiles and the procedures followed in replotting the profiles, including age/depth models, are available in Text S1 and Figures S2–S5. The two peat profiles, LAH and LAV, from the L. Adoon area, are first considered and compared with the nearby KLM profiles from Kilmore.

4.1.1. Profile LAH (L. Adoon Hill Bog) (Figure 7A)

The pollen profile LAH is from near the north-eastern shore of L. Adoon, 620 m south-east of core LAV (see below), in an area rich in archaeology that relates mainly to later prehistory (Bronze Age and Iron Age) and also many pre-bog stone walls that have not been dated [19] (Figure 1B and Figure S9B). In LAH, there are only occasional Pinus records until near the top of the profile (zone LAH-6), where Pinus reaches 5.7% (ca. 1850 CE or possibly younger). Betula dominated at the base of the profile (LAH-1; average: 70%). At the bottom of zone LAH-2, it fell to 43% and then averaged 33% over the rest of the zone. There were corresponding increases in Poaceae and bog/heath taxa, which suggests that local birch woodlands were largely replaced by bog/heath vegetation. Apart from a decline in Quercus and Alnus at the LAH-2a/2b boundary (there seems to have been local and regional woodland clearance at about 760 BCE, i.e., near the end of the Bronze Age, these AP taxa remained steady at about 6–10%. In LAH-2a, cereal growing was important locally, which is indicated by high cereal-type representation and exceptionally (indeed impossibly) high Plantago major/media values (averages 1.0% and 2.4%, respectively). The importance of grasslands is difficult to evaluate, as much of the Poaceae pollen may arise from bog/heath communities. However, the low P. lanceolata values suggest that bog/heath may be the main source of Poaceae pollen.

Zone LAH-3, which spanned most of the first millennium up to ca. 860 CE, shows an intensification of farming that starts gradually (first two centuries CE) and, based on the NAP taxa (especially P. lanceolata, Rumex type, and Asteraceae (includes Liguliflorae and Tubuliflorae pollen; these have comparable representation)), progressively increases. Substantial woodland clearance, probably at local and regional levels, did not occur until zone LAH-4, at the base of which P. lanceolata had an exceptional peak of 52% (ca. 900 CE), which suggests ribwort-plantain-rich vegetation close to the sampling site. Given that the Poaceae values are relatively low (28%), while indicators of arable farming are well represented (cf. cereal-type and P. major/media), the high P. lanceolata values are probably indicative of local arable activity rather than a grassland sward (cf. [66]). Micro-charcoal also attains the highest representation in zone LAH-4, which again supports the idea of much human activity locally and probably regionally.

Zone LAH-5 (1140–1680 CE) reflects a landscape that is more or less treeless, presumably due to further intensification of farming (exceptionally high values for both NAP1 and NAP2 taxa). Considerable arable and pastoral farming is taking place close to, or at least in, the general vicinity of the site. Cereal-type pollen, recorded in all spectra and averaging 1.5%, indicates substantial cultivation of cereals. Flax may have also been cultivated, at least within the region. A single Linum pollen grain was recorded at 75 cm, i.e., ca. 1260 CE. This probably arises from flax (L. usitatissimum), although the non-cultivated L. bienne as the pollen source cannot be excluded. Cannabis-type pollen was recorded in this zone (45 cm) and in earlier zones, which suggests hemp cultivation, although hop (Humulus lupulus), which produces a Cannabis-type pollen, cannot be excluded as the pollen source [107]. It is expected that hemp (C. sativa) was grown as a source of fibre rather than as an oil. Hemp was particularly important for the mercantile and navy ships in these times, and so its cultivation was actively encouraged, especially in the eighteenth and early nineteenth centuries ([108]; as regards south-western Ireland, see [109]).

Of particular interest is the single record of Fagopyrum pollen at 55 cm (ca. 1430 CE). The cultivation of buckwheat (F. esculentum; see also below) as a crop in central Europe had begun by this time [87], but it is doubtful whether it had reached Ireland or Britain. The isolated record in LAH-5 is therefore regarded as the result of long-distance pollen dispersal (see de Klerk et al. [110], who highlight records of Fagopyrum pollen from periods prior to widespread cultivation in Europe). Notably, there are peaks in Pediastrum (22%) and Cyperaceae (49%) at the top of LAH-4 and considerable values for Potamogeton at the top and bottom of LAH-5. This indicates that the local bog surface was wet and probably waterlogged for most of the time.

Zone LAH-6 spans approximately 1700 to 1850 CE. The increase in Pinus at the base of the zone is assumed to reflect the beginning of pine plantations in the wider region. There are no records for cereal-type pollen, probably reflecting the increasing importance of potato cultivation (silent in the pollen record; see Section 3.6), especially on marginal lands at/close to the site. Of particular interest are the further records for Fagopyrum (recorded in the five spectra of Zone 6 and averaging 2.7%), which undoubtedly signals local cultivation of buckwheat. This supports the evidence for local buckwheat cultivation, as provided by profile KLM I (see Section 3.4).

Subzone 6b (uppermost two spectra) is distinct in that P. lanceolata declines from an average of 18.3% in subzone 6a (three spectra) to 3%, while Poaceae increases substantially and Pteridium peaks in the uppermost spectrum. These changes probably reflect the decline/abandonment of farming, especially in marginal areas, which followed the Great Famine (1845–1852 CE) in many parts of western Ireland, including Co. Kerry [17,65].

4.1.2. Profile LAV (L. Adoon Valley Bog) (Figure 7B)

This blanket-bog profile is from about 500 m east of KLM I and 600 m north of LAH (Figure 1B and Figure S9B). Apart from the lowermost spectra (zone LAV-1), where Pinus approached 2%, there are only occasional Pinus records. Woody remains are plentiful in the lower part of LAV, so it is not surprising that AP values are high (average 60% in zones LAV-1 to 4), with the main contributors being Alnus at the base, Betula (especially LAV-3 and 4), and Salix (LAV-3, upper part). These trees/shrubs were probably present at or near the sampling site, i.e., growing on mineral soils and peat. AP values are distinctly lower in the upper part of both LAV and LAH, reflecting the completely open, more or less treeless landscape that characterised much of medieval and subsequent times, i.e., from about 1000 CE onwards. This feature is not as pronounced in KLM I, which suggests that complete openness may not have taken place until the early eighteenth century. P. lanceolata is well represented in profile LAV, apart from zones 1 and 3, where P. lanceolata averages only 0.7% compared with 7% in the other zones. This suggests that pastoral farming has been emphasised over long periods. In contrast, occasional cereal-type pollen suggests little or no arable farming in the vicinity of the site.

The records (six) of large ericoid tetrads in sample LAV 40 cm are noteworthy (Figure 7B). The sources of these pollen grains could be A. unedo and Rhododendron ponticum, both of which produce exceptionally large tetrads. If the age ascribed to the sample is correct (ca. 1420 CE), then R. ponticum is unlikely given that the shrub was first introduced to Britain in 1763, and probably several decades later to Killarney where, by the early nineteenth century, it was of importance in at least some parts of the Killarney woods [63,111]. Attributing the records to A. unedo is not justified based on the size of the tetrad alone. Furthermore, the A. unedo records in profile KLM I are much older (ca. 540 CE) (Figure 5). Given the demise of woody vegetation and relatively intensive farming, the local survival of Arbutus over the long interval between these records is unlikely. Therefore, the source of these large ericoid tetrads remains unresolved.

4.1.3. General Comparison of Peat Profiles from Kilmore and L. Adoon

The three peat-derived profiles from Kilmore (KLM I) and L. Adoon areas (LAV and LAH) enable fine-spatial reconstruction of vegetation and land-use change over approximately the last four millennia, to which the time-limited information provided by profile KLM II can be added.

To better facilitate the comparison of these four blanket-bog-derived pollen profiles, summary profiles are plotted on a time scale (Figure 8A_1–3,B₁). To further simplify the data and facilitate the reconstruction of past vegetation and land-use cover, as well as local bog wetness at each site, curves are also shown that aim to give a general impression of vegetation and land-use changes, as well as local bog-surface wetness at the four sites (Figure 8B₂,C_1–3). Curves showing the contribution of tall woody vegetation are presented under two categories: (a) regional woodland (mainly oak), which is based largely on Quercus values, and (b) local tall shrub/woodland, which corresponds mainly to AP excluding Quercus (oak). Birch (Betula) is the largest contributor to this latter category, but alder (Alnus) may also be important; therefore, the estimate of its contribution is shown by a curve within the composite AP curve. Willow (Salix) makes a substantial contribution to parts of profile LAV, as indicated schematically on that profile. Hazel (Corylus) is of some importance only in profile KLM I and so it is shown in Figure 8A₁ (see composite curves on the left-hand side of pollen profile KLM I). The estimation of pastoral farming is based primarily on P. lanceolata rather than Poaceae values, the latter being regarded as largely indicative of grasses growing locally on bogs and other wet habitats. Cultivation of cereals, buckwheat, and hemp relies on relevant curves, that is, cereal-type, Fagopyrum, and Cannabis-type pollen. Potato cultivation relies on historical and statistical data. The reconstruction of local bog-surface wetness/dryness is based mainly on Cyperaceae, Narthecium, and Calluna as indicators of wet and dry conditions, respectively. In the case of profile KLM I, for which NPP data are available, Assulina, Amphitrema flavum, and Copepoda are regarded as indicators of local wet conditions [43]. The wet/dry signal is not always clear-cut, in large measure probably due to the small-scale variation in wetness/dryness that is frequent on blanket bog surfaces [112].

The main points arising from the comparison of the four blanket bog pollen profiles, as summarised in Figure 8, are as follows. Blanket bog initiation at the four sites was broadly contemporaneous, having started at about 2000 BCE. Peat accumulation rates in the three profiles for which there is independent evidence of age (KLM I, LAH, and LAV) are also broadly similar (∼0.6 mm y⁻¹).

In the early years (until ca. 1500 BCE), the main tall canopy species was oak, presumably Q. petraea, which is the dominant species of Atlantic oak woodlands in the west Cork/Kerry region [113]. Birch (B. pubescens) is important and dominates locally at all sites. However, alder (A. glutinosa) is important at KLM I, KLM II (it dominates here), and LAV. As peat accumulation is initiated, human impact, as indicated by NAP (especially P. lanceolata), is largely imperceptible, which suggests that peat initiation is not directly connected with human activity; land abandonment is a more likely cause (cf. [114]). Other factors that were probably also at play include climate (lower temperature/increased precipitation) and pedogenesis involving natural and/or human-induced soil acidification.

Between about 1500 and 600 BCE, i.e., the mid-Bronze Age to the early Iron Age, varying but generally high levels of pastoral farming were recorded at KLM I (referred to as farming Phase i; Figure 5B and Figure 8A₁,C₁) and also at LAV (Figure 7B and Figure 8C₂). At KLM I, cereal growing appears to have been important early in the period (1500–1350 BCE). At KLM II, the situation is similar, but this profile is short and probably does not extend to the end of the interval under consideration (Figure 8B_1,2). Pastoral activity is also recorded at LAH, but it is much weaker (Figure 7A and Figure 8C₃). The LAH profile at this time may merely reflect farming activity at a distance, e.g., near KLM I and LAV. During this time, peat surfaces at all sites supported heathy/blanket bog vegetation, with much ling (Calluna) and sedges (Cyperaceae). Overall, peat surfaces appeared to be dry based on the abundance of Calluna pollen.

From about 600 BCE to 500 CE at KLM I, pastoral farming was greatly reduced (Figure 8A₁,C₁). Arable farming persists but at a reduced level. During the interval, starting shortly before the BCE/CE transition and continuing to about 500 CE, woody vegetation expanded, especially oak and birch (corresponding to the LIAL). It is at the end of this interval that Arbutus pollen is recorded, coinciding with the beginning of a sharp increase in arable farming (top of KLM I-3b; ca. 540 CE) and substantial woodland clearance, at least at a local level (KLM I-4; ca. 560–630 CE). In LAV, the record during this time is comparable (zone LAV-3), but scrub vegetation is much more important (as well as birch, willow is important, the high pollen representation suggesting that willow grew on the rather dry bog surface). A clearance feature corresponding to that in KLM I-4 was not recorded, so this appears to be of local significance only. In profile LAH (cf. LAH-2b), the contribution of woody species (especially oak and birch) remains steady until about the BCE/CE transition, when AP values decline. At this point, Poaceae expanded, and cereal-type pollen was recorded, but there was very little P. lanceolata. Overall, there are different vegetation dynamics at LAH compared with KLM I and LAV, and human activity at LAH seems to be lower during this time.

From about 750 CE onwards, pastoral farming became important once again at KLM I. The increasing trend continued until about 1200 CE. Although birch declines in importance, the role of oak is maintained or indeed increases until the end of the period (ca. 1100 CE) when exceptionally high P. lanceolata values indicate ribwort-plantain-rich grasslands in the vicinity of KLM I. Where precisely these grasslands were located is not easily envisaged. The ¹⁴C dates from the pre-bog stone-wall excavation (Figure 9; Table S7) suggest that peat initiation began at this location as early as the end of the second century CE, i.e., in the late Iron Age, perhaps after the abandonment of the stone-wall field system. Therefore, it seems that stone-wall construction took place no later than the mid/late Iron Age, although earlier construction (Bronze Age) is more likely. This may have been connected with the rather intensive farming recorded at the transition KLM I-2a/b, i.e., ca. 1300 BCE (intensive farming was also recorded in KLM II and LAV).

Figure 9. Diagrammatic cross-section of the excavated pre-bog stone wall (Wall 7), including ¹⁴C dates (for details of the ¹⁴C dates see Table S7). The sketch is after Ó Coileáin [19].

Figure 9. Diagrammatic cross-section of the excavated pre-bog stone wall (Wall 7), including ¹⁴C dates (for details of the ¹⁴C dates see Table S7). The sketch is after Ó Coileáin [19].

Regarding vegetation and land use, there was a step change in the upper part of all three profiles. This is characterised by a substantial decline in AP, including Quercus, and greatly elevated NAP1 and NAP2. These changes signal the creation of the open, largely treeless landscape of today in the context of substantial pastoral and arable farming practices. In KLM I, this is referred to as farming Phase v, which spans ca. 1720–1950 CE. It corresponds to zone KLM I-7, which is subdivided into 7a and 7b, with the boundary at ca. Beginning at ca. 1870 CE, the post-Great Famine population decline, bad weather, and general economic malaise resulted in rural decline and a switch away from arable farming (cereals and potatoes) in favour of cattle and sheep [65,115].

The correlation of the upper parts of profiles KLM I, LAV, and LAH, which is based largely on palynology, is not straightforward (Figure 8). However, the differences in the pollen assemblages appear to reflect small-scale and, hence, real differences in vegetation and land use rather than being an artefact of a weak chronology. The top of LAH (zone 6) has features similar to KLM I and is assumed to span more or less the same period, although probably not extending as far into recent times (the twentieth century may not be reached). Woody vegetation is scarce from ca. 900 CE (base of LAH-4), and particularly after 1100 CE (base of LAH-5). The LAV record is broadly similar in that pronounced woodland demise (birch rather than oak) started at 800 CE (base of LAV-5), but in this profile, intensive farming accompanied by birch clearance began earlier (base of LAV-4; ca. 400 CE). The overall evidence points to the late first millennium CE rather than later periods, such as after the Elizabethan plantation of Munster in the late sixteenth century [116], as a critical juncture in the creation of a landscape largely devoid of tall woody vegetation. In Ireland, the available historical and palynological evidence suggests considerable variation in the timing of final woodland clearances, with dates extending from the mid-/later first millennium CE to the early eighteenth century ([83,114,117,118,119]; see also Section 4.2.2).

The records of Fagopyrum pollen are particularly noteworthy. Fagopyrum was recorded in two spectra in KLM I-7a (dated to 1750 and 1800 CE, respectively) and in LAH, where it averaged 2.7% over five consecutive spectra at the top of the profile (LAH-6; 1700–1850 CE). Fagopyrum is not recorded in LAV, which is not surprising given that this profile, according to its age/depth model, ceased in the mid-eighteenth century. There can be no doubt that buckwheat was cultivated in this part of the Dingle peninsula in the eighteenth and probably into the nineteenth centuries. To the best of our knowledge, these are the only fossil records of buckwheat cultivation in Ireland to date.

The evidence for the wetness of the bog surface at the coring sites is also of interest in terms of possible climate change. Substantially increased bog-surface wetness was recorded in LAV, starting at about 900 CE and continuing until almost 1500 CE (Figure 8C₂). The record at LAH is rather similar (Figure 8C₃). At KLM I, wet conditions also began at about 900 CE but were only weakly expressed during the second millennium CE (Figure 8C₁). Given the high levels of human impact, the hydrology at this site may be affected by human activity, which is likely to have counteracted any tendencies towards wet bog surfaces mediated by precipitation/evaporation and/or temperature changes.

4.2. Long-Term Holocene Environmental Change

4.2.1. Pollen Profile LCN from Lough Camclaun [21]

Profile LCN, from the southern end of L. Camclaun (at 235 m asl; ∼1.6 km to the south of KLM I), is regarded as reflecting mainly the Coumanara uplands, which include Binn na Naomh, which rises to 670 m asl at 560 m to the south-west of the site (Figure 1A,B). The profile spans most of the Holocene and is regarded as extending into the twentieth century (Figure 10A and Figure S4). There are five ¹⁴C dates available, but these have large errors (≥110 y), and four of the dates relate to the lower part of the core. The age/depth model is based on a smooth spline curve (Figure S4) that uses three of the available ¹⁴C dates and 0 BP assigned to the top of the profile (this is based on the Pinus curve, which fails to rise substantially as might be expected if the profile extended into the later twentieth century). While a more tightly constrained age/depth model is desirable, the curve is regarded as acceptable for the present purposes.

The selected percentage curves are plotted on a time scale in Figure 10A. Zonation is similar to that by Dodson [21], but here, subzones are used, so the divisions are finer. At the base of the profile (zone LCN-1), Betula is dominant, and the Corylus curve has been initiated. This is followed by the rise of Pinus, then Quercus, and later Ulmus, which seldom exceeded 2% in the profile. The sequence in the spread of tall trees, however, is not clear, which contrasts with the BL profile in which hazel dominates from an early stage, then oak and elm, and finally pine (see below), which may be regarded as more typical for south-western Ireland [20].

During the first half of the Holocene (zones LCN-2 to 4; ca. 10.4–5.4 ka), the local woodlands were dominated by pine and oak, and hazel was important until ca. zones LCN-3/4 transition (ca. 7.4 ka), after which alder began to expand. Pinus exhibits elevated values at ca. 8.4 ka, but this is probably as much an artefact of percentage calculations (monolete spores derived from ferns decline substantially at about this time) as the effect of a climate oscillation such as the 8.2 ka event [120]. Elevated Pinus values (highest in the profile) at and shortly after 6 ka are often seen in Irish pollen diagrams and probably reflect pine growing on expanding blanket bogs and, ultimately, climate change [20,121]. The classical Elm Decline follows (zones LCN-4/5 transition); however, as usual, where Ulmus values are low, the Elm Decline is poorly expressed.

The upper part of the profile (zones LCN-5 to 7) is characterised by a steep and inexorable decline in AP that starts in subzone LCN-5b and is largely complete at the base of zone LCN-6 (AP declines from 37% to 28% over the interval 4.2–3.8 ka, approximately). At about 3.8 ka (LCN-5/6 transition), there is a distinct shift in the limnic environment, probably more dystrophic as a result of the input of humic acids from the blanket bog, which is well initiated by this time, as reflected by distinct declines in Isoetes and Pediastrum.

Elevated P. lanceolata, Poaceae, and Pteridium values point to a substantial opening up of the landscape, presumably as a result of increased farming in the region relating to an upsurge in activity during Late Neolithic/Early Bronze Age times. The changes in the Pinus and Quercus curves are particularly noteworthy. At its highest, i.e., near the top of zone LCN-4, Pinus is at 39%; in LCN-5a, it averages 12% and finally drops to ∼3% near the base of LCN-6. A small peak of 18% in LCN-5a dates to ca. 4.8 ka, probably reflecting the expansion of pine onto bog surfaces, i.e., the so-called ‘pine flush’ [114]. Expansion of bog first clearly manifests itself in LCN-5a (3450–2400 BCE; increase in Calluna and Sphagnum), during which P. lanceolata is consistently recorded. Human impact, as well as climate change (wetter and cooler, resulting in bogs too wet to support pine), were undoubtedly factors that favoured the expansion of bog and heath, starting in the Chalcolithic and continuing apace during the Bronze and Iron Ages.

Interestingly, while Pinus values declined, Quercus and Betula expanded, which suggests a distinct shift in woodland composition in favour of oak and birch at the expense of pine. By about 2.4 ka, pine was probably extinct locally and regionally, and other woody species, including oak, were scarce. Quercus values recovered somewhat in the interval 1.6–0.4 ka, i.e., in medieval times, which suggests a decline in human impact in this relatively remote upland region. This is supported by rather low P. lanceolata and cereal-type values; however, it should be noted that Poaceae (probably arising mainly from bogs), Calluna, and Cyperaceae are strongly represented (and included in the PS) so that the percentage values of all other taxa are artificially depressed.

Figure 10. Percentage Holocene lake pollen profiles from Dingle peninsula plotted on a times cale. The conventions followed are similar to those used in the KLM profiles. (A) Profile LCN from L. Camclaun is after Dodson [21]. (B) Profile BL from Ballinloghig is after Barnosky [23]. In profile BL, bog taxa are excluded from the PS because of their dominance towards the top of the profile. Their dominance and also high Sphagnum values (shown schematically) are indicative of infilling and bog development at the coring location.

Figure 10. Percentage Holocene lake pollen profiles from Dingle peninsula plotted on a times cale. The conventions followed are similar to those used in the KLM profiles. (A) Profile LCN from L. Camclaun is after Dodson [21]. (B) Profile BL from Ballinloghig is after Barnosky [23]. In profile BL, bog taxa are excluded from the PS because of their dominance towards the top of the profile. Their dominance and also high Sphagnum values (shown schematically) are indicative of infilling and bog development at the coring location.

Regarding human impact, the evidence for early farming impact (Neolithic) is weak. In the mid/later Neolithic at ca. 5 ka, P. lanceolata values increased for the first time. Pediastrum values also increased substantially, which suggests increased mineral input to the lake, but charcoal values were low, indicating that fire was unimportant. There are consistently elevated charcoal values coinciding with or following the Elm Decline (ca. 5.4 ka) and also earlier (at ca. 7.4 ka (Boreal/Atlantic transition)), though it is unlikely that these latter are the result of human activity, given that population levels were probably low. In the Late Holocene, charcoal values are elevated between ca. 2.8 and 2.4 ka, i.e., in the late Bronze Age and early Iron Age; however, given the low P. lanceolata values, human activity may not be responsible.

The first major farming impact (pastoral) is registered at ca. 4 ka, which coincides with the Chalcolithic/Early Bronze Age transition. Cereal-type pollen is first consistently recorded (but at low values) in the early Iron Age (ca. 2.7 ka). The high P. major/media values (especially in subzones 5b and 6a; Chalcolithic/Early Bronze Age) are unexpected as this taxon of disturbed/trodden ground is usually associated with arable farming. The significance of this finding is unclear. In early medieval times (ca. 1.6–1.2 ka i.e., 350–750 CE), Quercus values are somewhat elevated (approaching 2%), but P. lanceolata and NAP2 taxa (indicative of arable farming) are well represented. This points to a farming economy, perhaps wood pasture, in which oak, the main surviving tall-canopy tree but now rather scarce, was selectively favoured.

PAZs LCN-6 and 7 correspond to, on palynological grounds (cf. especially Pinus), the time span represented in the peat profiles from the Kilmore and L. Adoon areas (see above). Human impact first manifested itself clearly in LCN-5b (substantial rise in P. lanceolata). At about 1800 BCE (LCN-6), i.e., at the same time or slightly later than the start of peat accumulation at KLM I and II, LAH, and LAV, AP values decline sharply, and NAP1 increases. These changes signal woodland clearance and the expansion of bog and grassland as a result of human impact, and probably a wetter and cooler climate. Interestingly, there is no hint in CLN of the important role that birch and alder continued to play elsewhere (sites KLM I, LAH, and LAV; all < 2 km apart). This serves to confirm that LCN reflects vegetation and land use at a broad regional scale (cf. [122]).

Changes at the transition LCN-6a/b (400 BCE) are rather small but significant in that they reflect the regional extinction of pine and greatly reduced oak and other tall woody vegetation. After this, changes continue to be small, although oak recovers somewhat, probably largely because of the regional character of the profile, smoothing of the curves due to mixing in the lake-sediment column, and the fact that the main land-use changes are taking place mostly at lower elevations on the relatively fertile lands to the north of LCN.

4.2.2. Holocene Pollen Profile from Ballinloghig (BL) [23]

The lake-sediment profile BL is from about 10 km west of the area of primary interest (Kilmore/L. Adoon), but the setting is very different in that the lake lies in the lowlands to the west of the Brandon mountain range and is hence exposed to the persistently strong Atlantic south-westerlies (the Atlantic is <5 km to the west) (Figure 1B). However, the site is largely surrounded by a moraine and is therefore sheltered from the prevailing winds. The 6.4-m-long core, which is from the western margin of the lake, includes more than a metre of Lateglacial sediment that is not considered here (see [23]).

Profile BL spans most of the Holocene (sediment from the top 100 cm was not analysed; the record probably stops prior to the end of the medieval period) (Figure 10B; Text S1). For most of the early and mid-Holocene (zone BL-2; ca. 11–7 ka; chronological control is weak here and in the profile as a whole; see Figure S5), Corylus dominates (at about 50–68%; mainly at ∼60%), Quercus is at ∼10% (after ca. 10 ka), Pinus has low values (often < 5%), and Betula has moderate values (generally 10%). Ulmus seldom exceeds 4%, and Fraxinus and Taxus are only sporadically recorded.

In the early Holocene, there was an anomalous pollen spectrum with no Quercus or Pinus and only 1.1% Corylus. The age/depth model suggests that the spectrum dates to ca. 10.6 ka, so it is very unlikely that the 8.2 ka climatic oscillation is represented. This view is supported by the behaviour of the pollen curves, especially Pinus, which is not typical of the 8.2 ka event [120].

The dominance of hazel and the relatively minor role played by pine contrast sharply with the record from L. Camclaun. A variety of factors are probably at play, especially the more montane character at L. Camclaun, differences in bedrock and drift geology giving more fertile soils at Ballinloghig, and local climatic conditions including greater exposure to Atlantic gales at Ballinloghig and less seasonal temperature variation on account of its distinctly maritime setting.

In the upper part of the profile (zones BL-3 to 5; 7 ka onwards), considerable changes are recorded. In the interval 7–6 ka (PAZ BL-3), Pinus achieved >30%, while Corylus drops to ∼30%, from which it does not recover. A slow but steady rise in Alnus to ∼12% also occurs during BL-3. Records for P. lanceolata (see exaggerated curve for this taxon in Figure 10B), as well as an increase in Poaceae, suggest a disturbance. Research over recent decades has shown that P. lanceolata records are a feature of the late Atlantic period (pre-Elm Decline, i.e., around 4 ka and earlier) in Irish Holocene pollen profiles, especially in profiles from western Ireland [20,123]. The adjustments in the contributions of pine, hazel, and alder, as well as the indications of the opening up of the woodland cover (increases in NAP and Pteridium) during BL-3, are probably the result of climate oscillations rather than human impact (Mesolithic presence at Ferriter’s Cove, <10 km to the west, relates to the end of the Mesolithic, i.e., ca. 6.5 ka [9,124]).

The classical Elm Decline (here dating to ca. 5.8 ka), at the base of BL-4, is poorly defined, as is usual where Ulmus values are low. There is no evidence of early Neolithic Landnam, as is frequently recorded in Irish pollen diagrams [76]. Major changes, however, are recorded in BL-4b (4.4–4 ka), including a decline in Pinus to very low values. A rise in Poaceae at the end of BL-4a, as well as increases in Cyperaceae and Calluna, continue into BL-4b. This probably reflects lake infilling and wider landscape changes that involved the expansion of bog and heath. Given that there is little evidence for human impact up to this point, it is assumed that natural processes involving increasing soil acidification and increased wetness are the main drivers of these changes.

Zone BL-5 (4–1.2 ka) record deforestation and human impact. At the beginning of Zone 5, the first substantial increase in human impact is registered (cf. increase in P. lanceolata also Poaceae and Pteridium, and decline in Corylus). Pinus values are so low that local/regional presence of pine (P. sylvestris) cannot be assumed. However, pine probably persisted in the locality/region until the top of subzone 5a (ca. 3 ka). Final clearances followed (subzone 5b/c boundary; ca. 2.6 ka) that gave rise to the largely treeless present-day landscape. Given the weak chronology, especially in this part of the profile, the ages suggested by the age/depth model (Figure S5) for this part of the profile should, at best, be regarded as rough approximations. The overall pattern, however, is quite similar to that at L. Camclaun, but the treeless character of the landscape may have come about earlier at Ballinloghig, i.e., at about the Bronze Age/Iron Age transition (BL-5b/5c transition).

4.3. Historical/Statistical Records with Reference to Recent Farming and Comparison with the Pollen Evidence

Apart from historical records, including accounts by travellers [125] (see also Section 3.6), statistics relating to agriculture and population, systematically collected by central authorities since 1841, are an important source of information on population changes and the rural economy in Ireland. The collection of detailed records from all parts of the island commenced in 1847 [126,127,128]. Statistics were collected annually for the first 70 years and subsequently with less regularity and sometimes with considerable interruptions. Annual reports have also been published, providing a detailed breakdown of the statistics and general commentary, as well as useful overviews (e.g., [115,127]).

Rather than attempting to assess the detailed statistics that are available, the data compiled by the Central Statistics Office (CSO) for the publication ‘Farming since the Famine’ [127], and specifically those data relating to Co. Kerry, are now summarised. Statistics from smaller administrative units (e.g., barony and townland) are clearly more desirable for making comparisons with the pollen data, but these statistics have yet to be compiled; therefore, the data at the county level (Co. Kerry) are used and plotted in Figure 11. They are regarded as broadly indicative of the main trends in farming and population applicable in the study area, but it should be noted that the Dingle peninsula suffered greater population decline during the Great Famine and for decades subsequently compared with Co. Kerry as a whole [129].

Regarding the population, the main trends are as follows. The early decades of the nineteenth century leading to the Great Famine saw a rapid expansion of the population that was largely dependent on the potato as a food source. The maximum population at the base of plot F in Figure 11 relates to 1841; the population probably peaked a few years later before the severe famine conditions that commenced in 1845. After that, the demographic trends can be summarised in four phases. Phase α (1845–1860): a period of rapid decline connected with and following the Great Famine. This was caused by death from hunger and disease, and outwards migration. Phase β (1860–1880): the population stabilised for about two decades from the adverse effects of the Great Famine. Phase γ (1880–1970): a steady decline in population, mainly due to outwards migration. Phase δ (1970 and later): a gradual increase in the population, attributable to economic recovery connected with joining the European Economic Community (EEC; subsequently the EU). The population increase was largely confined to the main towns, such as Tralee and Killarney, while the rural population, especially those directly involved in farming, continued to decline [65].

The main trends in the rural economy are as follows: After the disaster of the Great Famine (which began in 1845, was the most severe in 1846 and 1847, but continued in Dingle for several further years [129]), agriculture stabilised over the course of about three decades (Phase 1, ca. 1853–1890), during which arable farming declined, while pasture (including meadow) increased and peaked at ca. 1876. Thus, the already dominant role of pastoral farming relative to arable farming was further emphasised (Figure 11C–E).

During Phase 2 (1914–1920) and especially Phase 3 (1940–1945), acreage under cereals increased due to factors associated with World War I and II (Figure 11C). Despite the government’s emphasis on wheat growing during and immediately after World War II, oats, the traditional crop of western Ireland [92,105], continued to be the main cereal. At the end of the 1940s, arable farming, which was already in decline, rapidly fell to below 2% of the land area (Figure 11E), and horse and pony populations declined (Figure 11A) in response to changes in the farming economy and increased use of cars and tractors. At/before Ireland joined the EU in 1973 (beginning of Phase 4), a substantial increase in the bovine herd occurred. In the 1980s, there was a spectacular increase in sheep numbers (more than doubling; Figure 11B) as a result of incentives provided by the EU Common Agricultural Policy (CAP).

Only profiles KLM I and LAH extend into the period represented by the population and agricultural statistics but at a much lower temporal resolution. The intensive land use that followed the population explosion at the end of the eighteenth and early nineteenth centuries is well reflected in the substantial increase in P. lanceolata and NAP1 generally. Arable farming also increased with potatoes achieving great importance (Solanum tuberosum is silent in the pollen record; potato cultivation is shown in Figure 8C₁,C₃ based on historical records). The subsequent decline in farming and some resurgence of woodlands during the late nineteenth and early twentieth centuries are expressed at the top of both KLM I and LAH. Surface-pollen spectra derived from moss polsters clearly show a secondary rise in Pinus (reflecting pine plantations) and the planting of spruce (Picea) (Figure 6B₁). Fagopyrum pollen records in KLM I and LAH (Figure 5B, Figure 7A and Figure 8) indicate the cultivation of buckwheat in the Kilmore/L. Adoon area during the late eighteenth and possibly early nineteenth centuries. Whether it was cultivated as early as the fifteenth century is doubtful, given the scant evidence (a single pollen grain in LAH; Figure 7A and Figure 8A₃).

5. Conclusions

Detailed palynological investigations in Kilmore Townland, Dingle peninsula, at a site with extensive pre-bog stone walls and other archaeological features, facilitated the reconstruction of vegetation history and land use during the last 4000 years. As the records opened (profiles KLM I and II), tall-canopy trees were rare and consisted mainly of oak. However, other trees and tall shrubs (mainly alder and birch) are important locally.

The main profile, KLM I, presents a dynamic picture of arboreal (AP) and non-arboreal pollen (NAP). The main factor determining the direction and magnitude of changes in woody vegetation and biodiversity is farming activity. This activity is reflected in the varying but generally high values for P. lanceolata (ribwort plantain) and cereal-type pollen.

Five distinct phases of active farming have been recognised (Figure 5B and Figure 8C₁). Phase i spans 1300–600 BCE (mid/late Bronze Age, extending possibly into the Iron Age). This is followed by a long lull that includes a period of reduced activity, referred to as the Late Iron Age Lull, which spans ca. 90 BCE–550 CE. Phase ii is of short duration and is characterised mainly by arable farming, centred on ca. 600 CE and lasting for about a century. Phase iii is a period (ca. 750–1200 CE) of ever-increasing farming, mainly pastoral-based, but with cereal cultivation gaining importance towards the end. Farming is most intense at ca. 1200 CE, which coincided with early Norman times, i.e., the early years of the late medieval period. Phase iv bridges the late medieval and early modern periods (ca. 1300–1700 CE). Intensive pastoral and arable farming further reduces the already scarce woody vegetation, including oak. Phase v (ca. 1720 onwards), the most intensive farming period (especially the early part), coincides with the population explosion of the late eighteenth/early nineteenth century that was interrupted by the Great Famine (1845–1852), a catastrophic development that initiated a decline in population and farming that persisted well into the twentieth century.

Excavation by Ó Coileáin [18] of a pre-bog stone wall at Kilmore indicates that peat initiation at the excavation site began in the early first millennium CE (Figure 9). Palaeoecological investigations (profiles KLM I and KLM II; also LAH and LAV by Dodson [21,22]), however, show that peat initiation elsewhere in the study area was much earlier, i.e., at about the beginning of the Bronze Age or shortly thereafter (ca. 4 ka). As regards the extensive stone-wall network, the available evidence suggests construction in prehistory, i.e., prior to ca. 400 CE. As far as could be ascertained, wall construction predates peat initiation. Wall construction during farming phase i, and early in that phase (cf. KLM I-2a; KLM II-3; ca. 3.8 ka (Early Bronze Age)), is suggested as a distinct possibility based on the intensity of farming activity at that time, as revealed by the KLM profiles and profile LAV.

Additional noteworthy features of profile KLM I include a distinct tephra layer (most likely attributable to the Öræfajökull eruption of 1362 CE), and pollen records for the strawberry tree (Arbutus; 540 CE) and buckwheat (Fagopyrum; late eighteenth/early nineteenth centuries CE). The Arbutus records are the first independently dated Irish records from outside the main present-day distribution of the strawberry tree in Ireland, i.e., Killarney, the south-west Iveragh peninsula, and the Beara peninsula. Fagopyrum, which has seldom been regarded as a crop or food source in Ireland, is shown to have a long history on the island (from the late 1500s onwards), and its cultivation is demonstrated palynologically for the first time (fossil pollen was recorded by Dodson, but the records were not included in his publications [21,22]).

Records from the KLM profiles for filmy fern spores provide new insights into the history of this interesting Eu-Atlantic component of the European flora (see also [20,109]). Criteria are provided for distinguishing the spores of the three species recorded in Ireland, i.e., Hymenophyllum tunbrigense, H. wilsonii, and Trichomanes speciosum.

A lake-sediment pollen profile, LCN from L. Camclaun, by Dodson [21] provides information on Holocene woodland dynamics in the upland environment near the main study area. During most of the early Holocene, pine and oak were dominant. In the mid-Holocene (5.6 ka), pine began an inexorable decline, with the so-called ‘pine decline’ recorded at ca. 4 ka. After this, pine and woody vegetation generally had a greatly reduced role in the landscape. Pine likely became extinct in the region shortly before 2 ka.

A lake-sediment pollen profile, BL from Ballinloghig, by Barnosky [23] from the western lowlands abutting Mount Brandon suggests rather different woodland dynamics in the Atlantic-exposed end of the Dingle peninsula. Oak and hazel are much more important during the early Holocene. Pine, however, expands at the expense of hazel at about 7 ka, when alder also begins to expand. As at L. Camclaun, a distinct ‘pine decline’ is recorded at about 4 ka. Woody vegetation (especially oak and birch) seems to have persisted in greater quantities for a longer time at this far-western site (to ca. 2.6 ka). Occasional Taxus pollen is recorded in BL (until ca. 4 ka), but there are no records for Taxus in LCN or in the peat profiles from the Kilmore/L. Adoon areas. This suggests that yew never achieved any importance in these parts of Dingle peninsula, which contrasts with Killarney [68,130] and also with the Beara and Mizen peninsulas [20], where yew achieved moderate to substantial importance, especially in the mid-Holocene.