Multicore Study of Upper Holocene Mire Development in West-Frisia, Northern Netherlands: Ecological and Archaeological Aspects

: We studied twelve late Holocene organic deposits in West-Frisia, The Netherlands. Pollen, spores, non-pollen palynomorphs, mosses, other botanical macrofossils and insect remains were recorded for reconstructions of changing environmental conditions. Eastern West-Frisia was a cultivated landscape during the Bronze Age, but it became a freshwater wetland in the Late Bronze Age. In most of our sites, radiocarbon dates show that time transgressive inundation of soils preceded the climate shift at 850 cal BC for several centuries. We suggest that solar forcing of climate change may have delivered the ﬁnal push to the inundation and depopulation of West-Frisia, which had already commenced several centuries before, due to sealevel rise. We did not ﬁnd evidence for signiﬁcant Bronze Age tree growth in West-Frisia before the inundations. Vegetation successions in the new wetlands developed from shallow mineral-rich freshwater to rich-fen vegetation. Subsequently poor fen vegetation with birch and pine developed, and the natural succession led to ombrotrophic raised bog vegetation. Complete successions from shallow, mineral-rich lakes to raised bog lasted between 1000 and 1500 calendar years. We hypothesize that medieval drainage and reclamation became possible only when the mires of West-Frisia had reached the raised bog stage. Reclamation of raised bogs by medieval farmers (drainage, eutrophication, peat digging) caused compaction, oxidation and loss of the upper part of the peat deposit. Seeds of salt-tolerant and salt-demanding plant species indicate that the medieval sites were inundated during storm surges with brackish or salt water, which triggered the farmers to build artiﬁcial mounds and, later, dikes. Under mounds and dikes, peat deposits remained protected against further decay. With our data we deliver a long-term perspective on contemporary ecosystem dynamics of freshwater wetlands, relevant for nature conservation and future climate change. , Moina and Simocephalus , are the recorded faunal taxa, while Dinoﬂagellata and Foraminifera were probably reworked from the underlying marine sediments that were deposited before the Bronze Age habitation. The ascospores of the coprophilous fungi Podospora , Cercophora , Sporormiella and Apiosordaria verruculosa indicate feces as a substrate.


Introduction
Conservation and management of landscapes for people and nature is important for societal and ecosystem health [1]. Information on past environments, human impacts and ecosystems can provide valuable insights to inspire and guide conservation and management policies [2][3][4]. However, landscape scale information from palaeoecological systems is often lacking because the spatial density of suitable study sites is too low. The landscape in West-Frisia (Noord-Holland, The Netherlands) sits on top of a thick sequence of Holocene marine, fluvial and peat deposits [5] and has a long history of human habitation [6][7][8]. The similarity between West-Frisia and other wetland regions in North-West Europe means that findings are likely to provide insights into how climate and humans have changed landscapes and vegetation that are informative for developing management and conservation strategies. In this study we present extensive new landscape scale palaeoecological data from West-Frisia that shed new light on the nature of the past vegetation and the role of past climate change in shaping the landscape.

The Abundance of Trees in the West-Frisian Landscape during the Bronze Age
Eastern West-Frisia was inhabited by Bronze Age farmers (between ca. 1700 and 800 BC) and is now one of the best preserved and most intensively investigated Bronze Age landscapes in northwest Europe [6][7][8], however, the vegetation of the area during the Bronze Age is still a matter of discussion. The established paradigm for the vegetation suggests that the landscape of West-Frisia through the Bronze Age consisted mostly of grassland and arable fields on sandy ridges. In this model, upland trees are thought to have been virtually absent [9,10], with wetland trees like Salix and Betula only in depressions [11]. This 'treeless landscape' paradigm has been challenged, based on new interpretations of pollen records, wood finds and bones of mammals, to suggests that forests made up a significant part of the Bronze Age landscape in West-Frisia; with Alnus carr in wet depressions, Fraxinus-Ulmus forests at drier sites and grasslands in the relatively high and dry areas [6,12]. Furthermore, it has been suggested that, during the Bronze Age, lakes, marshy grasslands and dry arable fields alternated with patches of woodland, shrubs and even with riparian Fraxinus and Quercus forest [7]. In this study we try to find arguments pro or contra the established paradigm and the alternative ideas of van Amerongen [6,12] and van Zijverden [7]. See research question 1.

The Role of Climate Change in Driving Human and Vegetation Dynamics in West-Frisia
During the mid-Holocene (ca. 5000 BC to ca. 1500 BC) the region of West-Frisia formed part of the Bergen tidal basin. Behind an inlet near the present village of Bergen a multi-channel river system deposited sediment into tidal flats. Around 2100 BC the eastern part of the Bergen tidal basin area became disconnected from the western part [7]. The disconnection was due to a slow down in sea-level rise that allowed beach barriers and dunes to develop along the Dutch coast [5]. During the Bronze Age habitation (ca. 1700 to 800 BC) the inlet near Bergen closed around 1570 BC, increasing the area of land available to people.
Archaeological evidence indicates that towards the end of the Bronze Age (ca. 800 BC) houses were more frequently situated on creek ridges, and possibly on small artificial mounds, suggesting rising groundwater tables [8,13]. This interpretation is supported by a concomitant rise in the water table indicated by the formation of shallow freshwater lakes on top of soils that had been exposed during the Bronze Age, and the appearance of new wetlands [14,15]. This landscape change in West-Frisia is coincident with a climate event detected around the world that has become known as the 2.8 ka event [16]. A decline of solar activity around 850 cal BC (Subboreal/Subatlantic transition in raised bogs) has been suggested to have triggered an abrupt climate shift, to cooler and wetter conditions [14,15]. Around this time agricultural communities abandoned the area of West-Frisia likely due to a declining ability to produce food in an increasingly wet landscape. Evidence for this comes from the site of Enkhuizen-dijk, a shallow freshwater lake that terrestrialized and developed into a Quaternary 2020, 3, 12 3 of 31 peat-forming mire, and finally to an ombrotrophic raised bog [14,15,17]. In this study we examine landscape scale palaeoecological evidence to assess if a rapid rise in groundwater occurred at a large scale across West-Frisia (research question 2).

Late Medieval Degredation of the West-Frisian Landscape
Archaeological evidence from the Iron Age and early Medieval period (ca. 800 BC to ca. AD 1000) from the wetlands of West-Frisia is almost absent, apart from settlements along the borders of Lake Wervershoof, an early Iron Age settlement site near Opperdoes [6,7], and the early medieval trade centre Medemblik [18]. About 1000 years ago, people started large-scale reclamation of the wetland areas in the western Netherlands by digging drainage ditches in the peat deposits.
Today most of the peat deposits in West-Frisia dating from the Iron Age have disappeared due to oxidation and peat digging [19]. However, some peat deposits have been found to have been preserved under subsequent medieval accumulation layers, e.g., artificial mounds [20] and dikes [17]. Access to the Iron Age peats for palaeoecological investigation has been possible during the last ten years thanks to archaeological excavations targeting the medieval deposits. It is the palaeoecological analysis of these peats that is presented here. Specifically, organic deposits from the following twelve sites are presented ( Figure 1 and Table 1): Beets (BTS), Berkhout (BKH), Bovenkarspel (BVK), Enkhuizen-LangeTuinstraat (ELT), Enkhuizen-Paktuinen (EPK), Enkhuizen-Raamstraat (ERS), Enkhuizen-Vijzeltuin (EVT), Enkhuizen-Westeinde (EWE), Grootebroek-Zesstedenweg (GBZ), Oostwoud (OWD), Venhuizen-Elbaweg (VEL) and Venhuizen-Spuiter (VSP). Furthermore, the new sites are compared with two previously published palaeoecological studies of West-Frisian peat deposits, at Klokkeweel [11] and Enkhuizen-dijk [17]. examine landscape scale palaeoecological evidence to assess if a rapid rise in groundwater occurred at a large scale across West-Frisia (research question 2).

Late Medieval Degredation of the West-Frisian Landscape
Archaeological evidence from the Iron Age and early Medieval period (ca. 800 BC to ca. AD 1000) from the wetlands of West-Frisia is almost absent, apart from settlements along the borders of Lake Wervershoof, an early Iron Age settlement site near Opperdoes [6,7], and the early medieval trade centre Medemblik [18]. About 1000 years ago, people started large-scale reclamation of the wetland areas in the western Netherlands by digging drainage ditches in the peat deposits.

Research Questions
In this palaeoecological study we focused on four research questions and assess the suitability of the landscape scale palaeoecological data from West-Frisia for informing conservation and management practice today: (1) Is there evidence for upland forest during the late Bronze Age habitation phase in West-Frisia as suggested by van Amerongen [6,12] and van Zijverden [7]?
(2) Is there evidence for abrupt groundwater rise (start of formation of shallow lakes and mires) in eastern West-Frisia around 850 cal BC [14]?
(3) Did medieval people only drain wetlands that had developed to ombrotrophic raised bogs, because successful drainage with man-made ditches was possible in raised areas only?
(4) Are the reconstructed vegetation successions relevant for present day and future management of wetland nature and landscapes [2][3][4]?

Materials and Methods
During archaeological excavations of medieval sites by the West-Frisian Archaeological Survey, fresh vertical cross sections were made, and metal boxes were used to sample the organic deposits ( Figure 2). In most cases the samples include the transition from sandy clay (the former Bronze Age, soil surface layer), through the natural accumulation of organic material (gyttja and subsequently peat), into the layer representing the start of medieval, man-made accumulation.

Research Questions
In this palaeoecological study we focused on four research questions and assess the suitability of the landscape scale palaeoecological data from West-Frisia for informing conservation and management practice today: (1) Is there evidence for upland forest during the late Bronze Age habitation phase in West-Frisia as suggested by van Amerongen [6,12] and van Zijverden [7]?
(2) Is there evidence for abrupt groundwater rise (start of formation of shallow lakes and mires) in eastern West-Frisia around 850 cal BC [14]?
(3) Did medieval people only drain wetlands that had developed to ombrotrophic raised bogs, because successful drainage with man-made ditches was possible in raised areas only?
(4) Are the reconstructed vegetation successions relevant for present day and future management of wetland nature and landscapes [2][3][4]?

Materials and Methods
During archaeological excavations of medieval sites by the West-Frisian Archaeological Survey, fresh vertical cross sections were made, and metal boxes were used to sample the organic deposits ( Figure 2). In most cases the samples include the transition from sandy clay (the former Bronze Age, soil surface layer), through the natural accumulation of organic material (gyttja and subsequently peat), into the layer representing the start of medieval, man-made accumulation. Sub-samples for microfossil analysis (ca. 0.5 cc each) were taken for all sites and prepared following Faegri and Iversen [32]. The microfossils were embedded in glycerol gelatin and sealed in with paraffin wax on microscopic slides. Pollen was identified with the keys and illustrations of Moore et al. [33], Beug [34] and the pollen reference collection of IBED, University of Amsterdam. In addition to pollen grains other microfossils with a characteristic morphology (non-pollen palynomorphs or NPP) were also recorded in the pollen slides. Identification and ecological interpretation of NPP followed van Geel [35], Pals et al. [11], van Geel and Aptroot [36], and van Geel et al. [37][38][39]. See also Appendix A for additional information on the ecological indicator value of NPP.
Pollen of herbaceous swamp and aquatic taxa, fern spores and NPP are normally excluded from the sum for percentage calculations, but in the present project we focused on the strictly local vegetation successions and therefore we included, in addition to the pollen total, the fern spores and algae in the sum for percentage calculations. Diagrams were produced using the TILIA program Sub-samples for microfossil analysis (ca. 0.5 cc each) were taken for all sites and prepared following Faegri and Iversen [32]. The microfossils were embedded in glycerol gelatin and sealed in with paraffin wax on microscopic slides. Pollen was identified with the keys and illustrations of Moore et al. [33], Beug [34] and the pollen reference collection of IBED, University of Amsterdam. In addition to pollen grains other microfossils with a characteristic morphology (non-pollen palynomorphs or NPP) were also recorded in the pollen slides. Identification and ecological interpretation of NPP followed van Geel [35], Pals et al. [11], van Geel and Aptroot [36], and van Geel et al. [37][38][39]. See also Appendix A for additional information on the ecological indicator value of NPP.
Pollen of herbaceous swamp and aquatic taxa, fern spores and NPP are normally excluded from the sum for percentage calculations, but in the present project we focused on the strictly local vegetation successions and therefore we included, in addition to the pollen total, the fern spores and algae in the sum for percentage calculations. Diagrams were produced using the TILIA program [40]. The percentage curves in the diagrams are exaggerated ten times in light gray. Macrofossil data are shown as bars, representing counted numbers or volume-percentages as indicated below the x-axis.
Sub-samples for macrofossil analysis were taken from all sites in 1 cm thick slices from the same levels as the pollen samples. Each macrofossil sample had a volume of ca. 15 cc. The samples were treated according to Mauquoy and van Geel [41]. Fruits, seeds, other plant remains, and zoological remains were hand-picked from the residue using a stereomicroscope and the material was stored in glycerol. Most fruits and seeds were identified using the catalogue of Cappers et al. [42] and the reference collections of the University of Amsterdam and the Cultural Heritage Agency (RCE, Amersfoort). Juncus seeds and Poaceae fruits were identified with the keys and illustrations of Körber-Grohne [43,44]. In some cases, identifications of seeds or fruits were to "type-level". Table 2 shows the possible species for those types. For identification of vegetative plant remains the reference collection of the University of Amsterdam and illustrations by Mauquoy and van Geel [41] were used. Bryophytes were identified using BLWG verspreidingsatlas [45], Landwehr [46], Nyholm [47], Siebel and During [48], Smith [49] and Touw and Rubers [50]. An exceptional case was the identification of Tomentypnum nitens. Fossil material observed in the present study did not show rhizoids on stems and leaves. Such rhizoids are an important characteristic feature of the species. However, in rare occasions T. nitens plants occur with decreasing frequency of tomentum, and plants can even be almost without rhizoids [47,51]. Macrofossils like fruits and seeds were counted, but for vegetative remains, like mosses, volume percentages in the macrofossil samples were estimated. Some of the recorded seeds, other plant remains, and some zoological remains are illustrated in Figures 3 and 4. Insect remains were studied from just two sites (EVT and EWE) due to logistical and financial constraints. The Coleoptera were identified using reference specimens from the collection of the Naturalis Biodiversity Center in Leiden. The nomenclature follows Vorst [52]. The remains of other insect groups were also identified. Each "insect" sample had a volume of ca. 15 cc and the minimum number of individuals per taxon in any one sample was mostly one, sometimes two or three. The preservation of much of the insect material was excellent. Examples are the small Sphaerocerid fly in a late stage of adult development found inside its puparium ( Figure 4H), a parasitoid wasp developing inside a fly puparium ( Figure 4G) and a tiny parasitoid wasp Anteris ( Figure 4I) found inside an arthropod egg in which it developed. We have to await further contemporary research on the parasite and its hosts before this and some other finds can be used in palaeoecological reconstructions.
Material for radiocarbon dating was sub-sampled from above-ground material (such as seeds, leaves and bud scales). The choice for the levels to be dated was mainly based on the wish to date at least the oldest and the youngest undisturbed organic levels. There were limitations in the choices because of the availability, or absence, of above-ground materials. Twenty-nine radiocarbon dates were obtained from eleven sites (Table 3). Radiocarbon dates were calibrated using OxCAL 2019-06-09 (IntCal13; [53,54]). Graphical illustrations of the calibration results are available online as Electronic Supplementary Material (ESM Document 1). For calculation of the duration of vegetation successions the choice of the age in calendar years was made based on the highest probability.

Results and Interpretation
For practical reasons we only illustrate here the diagram of the site Enkhuizen-Vijzeltuin ( Figure 5). For all sites the complete diagrams are available online as ESM Documents 2 and 3. Zone boundaries for each site were determined based on our ecological knowledge of the recorded taxa and on major vegetation changes. Here we present the vegetation successions at the twelve West-Frisian sites in alphabetical order. In addition to the ecological reconstructions, attention is given to chronological aspects based on radiocarbon dates, and to the duration of pre-Medieval hiatuses.

Beets (BTS, Diagram in ESM2)
During the excavation only undisturbed peat (50-9 cm depth) and the transition to the start of a human impacted peat layer were sampled. Deeper peat was not cored and therefore was not available for this study. Based on a radiocarbon date at 7 + 8 cm depth (ca. AD 300; Table 3) and the rise of Carpinus pollen we conclude that peat growth stopped in late Roman time (ca. AD 300), but it may well be that younger peat disappeared because of digging for fuel and/or by oxidation.
The lower 25 cm of the studied peat core was formed in relatively dry conditions (Sphagnum sect. Acutifolia, Aulacomnium palustre, Calluna vulgaris and its parasitic fungus Meliola ellisii, and HdV-10), while the upper part represents more wet local conditions (Sphagnum cuspidatum, Rhynchospora alba).

Berkhout (BKH, Diagram in ESM2)
The vegetation succession is not essentially different from the sites in, and near Enkhuizen, but the three radiocarbon dates (Table 3) point to relatively early starting peat growth (ca. 2550 cal BC) at the Berkhout site. The succession from the start of zone BKH-1/2 (mineral-rich shallow freshwater) to the ombrotrophic zone BKH-4 took about 1100-1300 years.
Within zone BKH-3 we distinguish five sub-zones, based on the alternation of the bryophyte species Calliergon giganteum and Tomentypnum nitens/Paludella squarrosa: Mineral-rich fen with C. giganteum.

BKH-3E (10-4 cm Depth)
Mineral-rich fen with HdV-123 and C. giganteum. Some temporary disturbance (flooding with water, enriched in minerals) is evidenced by the presence of Mougeotia, Spirogyra, Zygnema-type and HdV-128A in the sample at 5 cm depth.

Bovenkarspel (BVK, Diagram in ESM2)
No radiocarbon dates are available for the Bovenkarspel core, but based on the Fagus pollen curve we conclude that the organic sequence was formed after 800 cal BC [35].

Enkhuizen-Lange Tuinstraat (ELT, Diagram in ESM2)
A radiocarbon date at 65 cm depth (Table 3) indicates the local start of aquatic organic accumulation around 1000 cal BC. The radiocarbon date at 29 cm depth shows the age of the top of undisturbed mineral-poor Betula-peat (second to third century AD). The duration of the recorded vegetation succession was ca. 1250 years.

Zone ELT-2 (63-51 cm Depth)
This assemblage zone represents an Equisetum phase, with Carex acutiformis, C. disticha, Conium maculatum, Lycopus europaeus, Pedicularis palustris, Sparganium and Utricularia spec. The zone is further characterized by the appearance of Carex rostrata and Menyanthes trifoliata. This assemblage is indicative for mineral-rich fens. Spores of Closterium cf. rostratum, Mougeotia and Zygnema-type point to temporary presence of shallow stagnant freshwater.

Zone ELT-4 (27-1 cm Depth)
This is a mixed assemblage zone of highly decomposed macrofossils. Indicators for originally ombrotrophic conditions are Sphagnum (spores), Tilletia sphagni (parasite on Sphagnum) and Polytrichum. Tomentypnum nitens at the base of the ELT-4 indicates contact with mineral-rich water. There are various indicators for inundations. The presence of Dinoflagellata, Foraminifera, Juncus gerardii, Sperularia media/salina, Triglochin maritima and the snail Hydrobia ulvae point to inundations with brackish or salt water. The presence of Botryococcus, Pediastrum, Spirogyra, HdV-128A and HdV-128B, Eleocharis palustris, Lemna spec., Ranunculus subsp. Batrachium and Moina point to inundation with fresh water, possibly due to dissolution of brackish water with rain.

Enkhuizen-Paktuinen (EPK, Diagram in ESM2)
A radiocarbon date at 62 + 63 cm depth (Table 3) points to an age of ca. 1050-950 cal BC for the start of the aquatic phase (zone EPK-1). The date at 18 + 19 cm depth shows that mineral-poor Betula-peat was formed until the first century AD. The duration of the recorded vegetation succession was ca. 1100 years. Evidence for later peat growth (if any) is absent because of oxidation, decomposition and mixing with medieval plant remains (Zone EPK-5). Equisetum is temporarily dominant. Algae are still present. Semi-aquatic taxa are Carex disticha, C. elongata, C. pseudocyperus, C. riparia, C. rostrata, Lycopus europaeus, Lythrum salicaria, Oenanthe aquatica and Ranunculus sceleratus. The cladocera Daphnia and Simocephalus were recorded. A peak of Neurospora ascospores indicates local fire.

Zone EPK-4 (30-23 cm Depth)
Thelypteris shows relatively low percentages, while Betula spec. and Menyanthes trifoliata continue to be important taxa. Calliergon giganteum indicates relatively mineral-rich conditions, and point to seepage of calcium and iron-rich groundwater.

Enkhuizen-Raamstraat (ERS, Diagram in ESM2)
Five sampled levels in the lower part of the sequence (zones ERS-1 and ERS-2) were too poor in microfossils to show percentages. Black dots in the diagram only indicate presence of taxa. Three radiocarbon dates (90-92, 84-86 and 68-70 cm levels; Table 3) indicate that inundation with shallow freshwater started around 1300 cal BC and lasted until about 1000 cal BC. Subsequent undisturbed peat growth occurred during ca. 1200 years, until the second century AD (radiocarbon date 16-17 cm; ombrotrophic raised bog peat). Any later peat formation is not traceable because of human disturbance (drainage, oxidation and possibly peat digging).

Enkhuizen-Vijzeltuin (EVT)
The radiocarbon date at 70 + 69 cm level (Table 3) Figure 4L) is a newly recorded globose "macrofossil". Its wall is covered with small mineral particles. It was also observed (but not counted) in the gyttja phase of other West-Frisian sites.
In the lower half of this zone the macroscopic parts of aquatic invertebrates are common, but in the upper half of EVT-1 these organisms diminish or disappear, indicating terrestrialization that is also reflected by a decline of Ranunculus subg. Batrachium and Characeae, the appearance of Typha latifolia and T. angustifolia and the occurrence of charcoal and fungal spores of the fire indicator Neurospora.
The find of a subfossil egg capsule of the cockroach Ectobius ( Figure 4J) in the sample at 21 cm depth is noteworthy. These only accidentally house-dwelling cockroaches currently do not occur in the lower parts of the Netherlands. The recent distribution of Ectobius spp. in the Netherlands is restricted to the Pleistocene and coastal sand dune landscapes, now the only areas with low nutrition levels and open vegetation types. At the site Vijzeltuin (EVT) Ectobius lived on the raised bog, in the open pine forest, which it prefers (pine seeds and other pine macrofossils present in the same sample).

Enkhuizen-Westeinde (EWE, Diagram in ESM3)
The radiocarbon dates at 37 and 25 + 26 cm depth (Table 3) point to inundation of the site around 900-850 cal BC. The date at 13 cm depth corresponds to the Hallstatt plateau in the calibration curve [58] and therefore it represents a very long period (ca. 350 years of uncertainty between ca. 740 and 410 cal BC) on the calendar time axis. The period of the last ca. 2500 years is not represented as undisturbed peat.
The insects Ochthebius spec, Coelostoma orbiculare, Tanysphyrus lemnae, Hydroporus and Trichoptera larvae also point to shallow water and Donacia semicuprea lives on Glyceria and Sparganium. Several ascospore types of coprophilous fungi in EWE-1 indicate a nearby presence of feces as a substrate: Apiosordaria verruculosa, Bombardioidea-type, Cercophora, Podospora, Sordaria and Sporormiella. Neurospora spores (HdV-55B/C) indicate nearby fires, and Riccia spores indicate moist arable land [17]. During the excavation parcellation ditches were observed below the peat deposit. The cysts of Dinoflagellata may be reworked (bioturbation) from the clayey, silty subsoil that was deposited before the Bronze Age habitation.

Zone EWE-2 (26-20 cm Depth)
The taxa at the start of this zone (Cicuta virosa, Epilobium palustre, Lycopus europaeus, Lythrum salicaria, Ranunculus flammula, Salix and Solanum dulcamara) still point to nutrient and mineral-rich local moist conditions. The beetles Hydraena, Noterus crassicornis and Hydroporus point to the presence of water, Gyrinus aeratus to sufficient surface water, Limnobaris dolorosa to the presence of Cyperaceae and Plateumaris bracata to Phragmites australis.
An important role in the terrestrialization of the site is played by Equisetum (most probably E. fluviatile). A declining nutrient level is indicated by Menyanthes trifoliata, and the occurrence of Carex pseudocyperus, C. riparia, Pedicularis palustris and Potentilla palustris, which point to mesotrophic mineral-rich fens. HdV-598 is a newly recorded microfossil that occurs in this zone. It is still unidentified and has a characteristic form and wall sculpture ( Figure 4K).

Zone EWE-3 (20-13 cm Depth)
The vegetation succession is taken over by Thelyptris palustris (monolete psilate fern spores and sporangia). Galium-type (probably G. palustris), Menyanthes, Rumex acetosa-type, Silene flos-cuculi and finally Betula spec. seeds, coinciding with a maximum in Betula pollen, all fit in the picture of a further decline of mineral richness. Betulapion simile feeds on Betula, while Sialis, Cyphon and Laccobius indicate the nearby presence of open water.

Zone EWE-4 (13-3 cm Depth)
The transition from EWE-3 to EWE-4 shows a decline of fern spores, and a sudden increase of Sphagnum, its parasite Tilletia sphagni, and Ericales, together reflecting ombrotrophic bog conditions. However, there is no undisturbed bog deposit present at the site, because the samples show a mix of ombrotrophic taxa, Pediastrum and HdV-128A (both occur in relatively nutrient-rich freshwater) and Foraminifera, the marine mollusk Littorina and Triglochin (salt or brackish conditions; probably as a consequence of marine inundations). Secale and some other taxa (e.g., Asteraceae, Brassicaceae, Chenopodiaceae) reflect nearby human impact on former raised bog deposits. Some ascospores of the dung-inhabiting Sordaria and Podospora were recorded.
Nearby human impact is reflected by several pollen taxa: Secale, increased Brassicaceae, Caryophyllaceae, Chenopodiaceae, Persicaria maculosa-type and Linum usitatissimum. We cannot exclude that GBZ-5 already represents the basal layer (12th/13th century AD) of the man-made mound.

Oostwoud (OWD, Diagram in ESM3)
The radiocarbon dates at 68 and 61 cm level (Table 3) show that the start of the deposition of gyttja and subsequent peat growth at the Oostwoud site occurred around 1650 cal BC. The date at 29 cm shows that the peat deposit is undisturbed only until around 750 to 550 cal BC (Hallstatt plateau in the radiocarbon calibration curve [58]).

Zone OWD-1 (86-66 cm Depth)
From 86 to 66 cm depth some macrofossils were recorded in the sandy clay, but the concentration of microfossils was too low, and the preservation was too bad for reliable quantification. Seeds of Juncus effusus-type, J. articulatus-type, J. bufonius, Carex riparia, Chenopodium polyspermum, Potentilla spec., Rumex spec. and Solanum nigrum were preserved in the sandy substrate. In the upper part of OWD-1 Carex otrubae/vulpina, Hypericum tetrapterum, Mentha aquatica/arvensis and Urtica dioica occur.

Zone OWD-3 (60-51 cm Depth)
Thelypteris palustris becomes a dominant species and it is growing together with Lotus, Lythrum salicaria, Menyanthes trifoliata, Pedicularis, Potentilla-type, Carex rostrata and C. acuta-type, which point to mineral-rich fen. The presence of Mougeotia spores shows that temporary shallow stagnant freshwater still prevailed, probably in springtime when the spores were formed.

Venhuizen-Elbaweg (VEL) and Venhuizen-Spuiter (VSP)
During excavations at the sites Elbaweg and Spuiter, profiles of the "Westfriese Omringdijk" (Late Medieval dike for protection against the sea) were made and a compacted, ca. 15 thick peat layer, below the oldest, medieval layer of the dike was sampled at both sites, including the layers below and above the peat layer. The sampled sediments below the peat layer consisted of sandy clay, and therefore the microfossil contents have a predominantly allochthonous, marine origin, mixed with pollen of upland taxa. The peat layers (zones VEL-2 and VSP-2) were formed by local peat-forming plants. The sampled sediments deposited on top of the peat had been collected and transported to the sampling site as sods by late Medieval people when constructing the first dike. That dike-material was also built up of sandy clay, but without an orderly stratigraphy. At the site Elbaweg a second, ca. 3 cm thick horizontal organic layer, situated about 35 cm above the peat layer, showed a stagnation phase in the history of dike building (see below, assemblage zone VEL-4).

Zone VEL-2 (74-52 cm Depth)
A peak of Salix pollen at the base of the peat deposit represents the first local carr vegetation, followed by high percentages of Alnus pollen. Willow and alder represent local wet, nutrient-rich conditions. Oenanthe aquatica, Urtica dioica, Lemna spec. (seeds and pollen) and the aquatic invertebrates Lophopus crystallinus and Simocephalus were also recorded in the peat layer. The upper few centimeters of the peat layer show a peak of Betula that indicates a decline of available nutrients, due to isolation from the mineral soil.

Zone VEL-4 (19-16 cm Depth)
The thin organic layer of VEL-4 shows a peak in Poaceae (together with fruits of Phragmites australis), Plantago major and a considerable number of spores of coprophilous fungi: Sordaria, Sporormiella, Cercophora, Apiosordaria verruculosa and Podospora. We suppose that VEL-4 represents a stagnation phase in dike accumulation, while the large number of spores of coprophilous fungi shows that the first, low dike was grazed by herbivores.

Zone VEL-5 (16-5 cm Depth)
The spectra of microfossils and macro-remains have much in common with zone VEL-3.

Venhuizen-Spuiter (VSP, Diagram in ESM3)
The radiocarbon date at the base of the peat layer (Table 3) shows that peat growth started, like at the site Elbaweg, between ca. 750 and 450 cal BC. Plant remains from a small filled up depression (ditch) that disturbed the peat layer of zone VSP-2, were dated around 500 cal AD.

Zone VSP-1 (43-30 cm Depth)
The sandy clay sediment of this zone is characterized by microfossils indicating marine conditions (relatively high percentages of Chenopodiaceae and Dinoflagellata) and freshwater algae (Pediastrum, HdV-128A and HdV-128B). This zone represents the peat layer. Alnus shows relatively high percentages, indicating that the peat was formed by an alder carr, with local Apiaceae, Potamogeton/Triglochin and ferns.

Forest in West-Frisia during Late Bronze Age? (Research Question 1)
Van Amerongen [6,12] challenged the paradigm of an almost treeless landscape in West-Frisia during the Bronze Age. She suggested that there may have been a considerable forest cover: Alnus carrs in wetlands, Fraxinus-Ulmus forests at drier sites and grasslands in the relatively high and dry areas. In addition, van Zijverden [7] supposed that lakes, marshy grasslands and dry arable fields alternated with patches of woodland, shrubs and even with riparian Fraxinus and Quercus forest.
If forest would have been present during the late Bronze Age, then one could expect evidence for drowned forest at our sites. Local forest would have been reflected by high tree pollen percentages and we would expect clumps of tree pollen [59] that we did not observe. We also would expect arboreal fruits/seeds, and tree roots in the upper soil just below the shallow lake sediment that we found at most of the sites. We did not find such evidence. In other words: our samples representing the Late Bronze Age do not point towards a forested landscape, apart from willow and birch carr around the freshwater lake Klokkeweel [11].
We suppose that people with their domesticated herbivores will have hampered the immigration of tree species because of grazing pressure starting after the marine phase at the moment that the area became habitable for plants, for people and for domesticated and wild herbivores. We have no evidence for local tree growth on dry soils before the recorded rise of the groundwater and therefore we do not accept the hypothesis of van Amerongen [6,12] and van Zijverden [7] about forested areas in West-Frisia during the Bronze Age.

Rise of Groundwater and Climate Change (Research Question 2)
Van Amerongen [6] and van Zijverden [7] argued against the hypothesis that abrupt climate change and the consequent sudden rise of the groundwater table caused the depopulation at the end of the Bronze Age in eastern West-Frisia [14,15]. The hypothesis was mainly based on the radiocarbon dated site Enkhuizen-dijk and on the youngest archaeological Bronze Age dates of West-Frisia, and a link was made with the transition in raised bogs, elsewhere in The Netherlands, from "Older Sphagnum peat" to "Younger Sphagnum peat" at the start of a steep rise of ∆ 14 C around 850 cal BC (2.8 event). Martin-Puertas et al. [60] and van Geel et al. [16] found strong additional evidence for a link between a steep rise of ∆ 14 C starting around 850 cal BC and abrupt climate change to cooler and wetter conditions. The present study shows that inundation of most of our sites in eastern West-Frisia (ELT, EPK, ERS, EVT, GBZ; all in the Enkhuizen area) already started centuries before 850 cal BC, i.e., between 1300 and 1000 cal BC ( Figure 6). A gradually rising sealevel apparently caused a rise in ground water [61], and the amount of shallow pools to increase. This hydrological change finally made eastern West-Frisia inhabitable around 800 cal BC. A short-term climate-related rise in ground water at 850 cal BC may have occurred; however, it seems that this was super imposed on a longer-term rise in ground water indicated by the process of paludification that had already been going on for several centuries. The hypothesis suggesting that solar forcing caused abrupt climate change around cal 850 BC is therefore not clearly tested with these data, because the impact of this event may, or may not, have been overridden by ongoing sealevel rise. However, we observe that by cal 850 BC all study sites contain sediments indicative of shallow lake environments, suggesting that by this time the transition to wetlands across West-Frisia may have been given final impetus by the climatic shift. The first freshwater inundations of the sites Berkhout and Oostwoud are much earlier (respectively ca. 2550 cal BC and ca. 1650 cal BC) but those sites are relatively far away from the sites in and near Enkhuizen.
was super imposed on a longer-term rise in ground water indicated by the process of paludification that had already been going on for several centuries. The hypothesis suggesting that solar forcing caused abrupt climate change around cal 850 BC is therefore not clearly tested with these data, because the impact of this event may, or may not, have been overridden by ongoing sealevel rise. However, we observe that by cal 850 BC all study sites contain sediments indicative of shallow lake environments, suggesting that by this time the transition to wetlands across West-Frisia may have been given final impetus by the climatic shift. The first freshwater inundations of the sites Berkhout and Oostwoud are much earlier (respectively ca. 2550 cal BC and ca. 1650 cal BC) but those sites are relatively far away from the sites in and near Enkhuizen.

Medieval Start of Peat Land Reclamation and Settlements (Research Question 3)
Based on archaeological evidence in the Netherlands, Heidinga [62] supposed that the trigger for medieval reclamation of wetlands in the western Netherlands was a dry 10th century AD, and therefore population density pressure in the areas of sandy soils. Wetlands were available for reclamation, but we emphasize that not all wetlands on top of peat deposits could be sufficiently drained to allow reclamation. The suitability for artificial drainage and reclamation was determined by the water table position and the vegetation type. In wetland vegetation successions we often see a trend starting with plant species that require contact with the mineral subsoil that moves towards species that prefer mineral-poor conditions. During the vegetation succession the peat deposit increased in thickness and the roots lost contact with the mineral subsoil and the peat-forming vegetation became more and more dependent on rainwater.
There are strong indications that in most of our studied cores there is a hiatus between the medieval anthropogenic accumulation layer and the underlying natural peat deposit. At site EVT the age difference of ca. 1000 calendar years between the two upper radiocarbon dates illustrates this. Some upper peat will have disappeared because of oxidation, and even peat digging may have played a role. In spite of this complication, at ten of our sites we found that the recorded natural vegetation succession continued and probably ended with a stable phase with Sphagnum, Betula and three sites even with Pinus in the local vegetation. In acid raised bogs dominated by Sphagnum, the

Medieval Start of Peat Land Reclamation and Settlements (Research Question 3)
Based on archaeological evidence in the Netherlands, Heidinga [62] supposed that the trigger for medieval reclamation of wetlands in the western Netherlands was a dry 10th century AD, and therefore population density pressure in the areas of sandy soils. Wetlands were available for reclamation, but we emphasize that not all wetlands on top of peat deposits could be sufficiently drained to allow reclamation. The suitability for artificial drainage and reclamation was determined by the water table position and the vegetation type. In wetland vegetation successions we often see a trend starting with plant species that require contact with the mineral subsoil that moves towards species that prefer mineral-poor conditions. During the vegetation succession the peat deposit increased in thickness and the roots lost contact with the mineral subsoil and the peat-forming vegetation became more and more dependent on rainwater.
There are strong indications that in most of our studied cores there is a hiatus between the medieval anthropogenic accumulation layer and the underlying natural peat deposit. At site EVT the age difference of ca. 1000 calendar years between the two upper radiocarbon dates illustrates this. Some upper peat will have disappeared because of oxidation, and even peat digging may have played a role. In spite of this complication, at ten of our sites we found that the recorded natural vegetation succession continued and probably ended with a stable phase with Sphagnum, Betula and three sites even with Pinus in the local vegetation. In acid raised bogs dominated by Sphagnum, the net organic production and therefore peat accumulation increased, and the mires became raised bogs where the Sphagna, with their extremely high-water storage capacity [63], caused groundwater tables to rise above the regional water tables in mineral soils. Therefore, the raised bogs were relatively easy to drain by people, as after drainage initially their surface will have been still above mean sea level and also above water tables in adjacent mineral soils. We hypothesize that drainage and reclamation became possible only when the mires had reached the raised bog stage. Drainage of the preceding fen vegetation would have failed because after drainage the peaty soil surface would have been too wet to be turned into arable land or meadows. In other words: during a long period after the Bronze Age the local conditions were too wet for people to settle. The late Medieval farming communities could reclaim raised bogs (still without windmills and without dikes) with drainage ditches. An argument against our hypothesis (medieval settlement on raised bogs only) would be that oligotrophic conditions would hamper the growth of crop plants and also domesticated large herbivores would not thrive (grazing plants with low nutrient quality) in a newly reclaimed ombrotrophic bog and that would make life for farmers rather complicated. Natural drainage, compaction of peat and temporary inundations may have occurred and in that way mineral-poor peat deposits became fertilized with clay. In other words: we cannot exclude that water streams after storm surges triggered extra natural drainage (new gullies in the raised bogs), thus bringing minerals and nutrients for plants and therefore facilitating the start of human habitation. Natural drainage, erosion and oxidation (more available nutrients) may well have occurred before the arrival of the farmer communities. The hiatuses between undisturbed upper peat layers and the first anthropogenic accumulation layers make it difficult to find detailed answers to this question.

Human Impact and Increasing Marine Influence in West-Frisia
Indicators for salt marsh and marine conditions, such as Juncus gerardii and Foraminifera in the upper peat layers, just below the medieval accumulation layers were probably mixed by bioturbation and reworked by human action. The presence of salt marsh indicators together with freshwater indicators at our West-Frisian sites suggests that the raised bogs were temporarily flooded with brackish or salt water. This happened during storm surges [64,65] when many peat deposits in the northern Netherlands were eroded and the freshwater basin of Almere increased in size and gradually developed to the saltwater basin Zuiderzee [5]. Van Popta et al. [66,67] examined the dynamic cultural landscape of the north-eastern Zuiderzee region, ca. 23-50 km east of Enkhuizen, in the period AD 1100-1400. The cultivation of land and marine erosion of peatlands (floods and rising sea level) were considered as the main factors that caused the transformation of large parts of the landscape in the central Netherlands, from peatlands with freshwater basins into the tidal lagoon Zuiderzee. The southwestern part of the present province of Frisia (Friesland) became disconnected from West-Frisia. Reclamation and related compaction and oxidation of peat turned against the farmers, as it caused rapid erosion during storm surges. In a few centuries, large peatlands in the north-eastern Zuiderzee region [66] and in the area between West-Frisia and Frisia disappeared as a consequence of the eroding power of the Zuiderzee [5]. Residents of the coastal zones were struck by the many floods, as cultivated land was washed away, complete settlements were inundated and eroded, and thousands of people drowned [66].
Van Smeerdijk et al. [68] studied the botanical composition of samples from three chronologically successive medieval artificial mounds (terp) at Diemen, near Amsterdam, and reconstructed the increasing role of salt water in late medieval time. The authors recorded freshwater conditions for the oldest terp, followed by brackish conditions for a terp that had been built in the last part of the 12th century AD, while a 14th century terp even consisted of marine clay sods. Our West-Frisian sites show evidence for inundations from the sea before the terps were built. The Diemen sites, however, were located much more to the south (less and later marine influence) and the palaeoecological registration of the gradual salinization of the landscape at Diemen was possible due to different successive ages of the three terp sites.

Forest and Toponyms
At present, there is no natural forest in West-Frisia, but there is toponymic evidence for late medieval presence of forest and brook forest when villages were founded on the peat deposits. Names of several existing villages point to forest (-hout or -woud), or carr woodland (-broek): Berkhout, Schellinkhout, Aartswoud, Hoogwoud, Nibbixwoud, Oostwoud, Midwoud, Westwoud, Oudeboxwoude, Scharwoude, Hensbroek, Spanbroek, Benningbroek, Broek op Langedijk, Grootebroek. We suppose that artificial drainage of the raised bogs will even have facilitated germination and expanded growth of birch and pine, and the founders of villages will have experienced the West-Frisian landscape, or parts of the landscape, as forest.

Relevance for Management of Wetlands (Research Question 4)
Palaeoecological data can provide a valuable long-term perspective on contemporary ecosystem dynamics [2,3,69] but are not fully utilized in conservation management [70]. The present study is of interest for nature management of wetlands. Knowledge about the duration of the successive vegetation phases from shallow, mineral-rich freshwater to ombrotrophic Sphagnum bogs is useful, although duration depends on the local hydrological situation and climate [71]. In most of our West-Frisian cores, development of the vegetation followed the normal succession patterns in fresh water. Vegetation development started with shallow open water and aquatic species such as Lemna, followed by a rich-fen stage with semi-terrestrial species such as Equisetum, Menyanthes trifoliata, Carex spp., Mentha aquatica and Thelypteris palustris. In general, these vegetation patterns are also found today [4]. However, bryophytes such as Tomentypnum nitens, Meesia triquetra and Paludella squarrosa, which were present in several of our sequences, do not occur in the Netherlands anymore [45]. They are still common in fens in Scandinavia, but have strongly decreased in other European countries. At some of our sites, the seepage of calcium and iron-rich groundwater probably played a role in the rich-fen stage, especially the ones with mosses such as C. giganteum, T. nitens and P. squarrosa [55,56]. When the peat layer became thicker, this mineral-rich groundwater did not reach the surface layer anymore.
At most of our sites the rich-fen stage was followed by Sphagnum-dominated poor fen, bog forest with Betula and Pinus and raised bog with Calluna vulgaris and Erica tetralix, following present day analogs [4]. More minerotrophic carr with Alnus glutinosa were only present at the VEL and VSP sites. Development of Sphagnum-dominated bogs means that peat growth could keep pace with rising mean sea water tables, and the raised bog vegetation even continued to grow above the regional groundwater tables outside the bog area.
Terrestrialization stages of mire vegetation are important to northwest European nature conservancy because of high biodiversity, particularly of mineral-rich fens. Based on macrofossils and microfossil analysis Faber et al. [4] reconstructed vegetation successions in two former turbaries in the Netherlands. For both sites, the establishment of mineral-rich fens took about 60 years, and that vegetation type only remained for about 30 years before it was followed by Sphagna and other species of ombrotrophic conditions. On a landscape scale, succession from open water to raised bog may take more time [71]. Based on radiocarbon dates the duration of the West-Frisian vegetation successions from shallow freshwater to mineral-poor bogs was between 1000 and 1500 calendar years. This is comparable with succession rates in peatlands in northern Canada [71], but rather long compared with late Holocene Sphagnum peat found directly on podzolic soils in the eastern Netherlands and Germany [63]. Fen to bog transitions may be influenced by internal factors such as local hydrology, and external factors related to climate change. Although thickness of the fen peat differed between our sites from 23 to 74 cm, the succession patterns and periods were remarkably similar. This suggests that the general rise of the (sea) water level was more important than local hydrology. The relatively long duration of the wetland succession phases in West-Frisia compared to paludification of podzolic soils in the eastern part of the Netherlands is linked to the initial situation. Podzolic soils have formed in sandy parent material, which has low buffer capacity to acidification. The clayey soils of western Frisia are however highly buffered and nutrient-rich, and even additionally enriched in nutrients by people and their domesticated animals during the Bronze Age.
Another example of knowledge obtained with palaeoecological methods is the answer to the question about extinction, or survival, of Pinus sylvestris in northwest Europe in the late Holocene [70,72]. See also Jamrichová et al. [73] for pine in Central Europe. We observed macro-remains of Pinus sylvestris (needles, seeds, periderm, roots, combined with relatively high pollen percentages; Figure 3K-T) in the early raised bog zones ERS-4, EVT-3 and GBZ-4. These pine records from West-Frisia date about 2000 years ago. Pals et al. [11] in their study of the nearby Klokkeweel site (Figure 1) dated a local wetland pine phase between ca. 900 and 800 cal BC and Brinkkemper [74] mentioned pine cones from roughly the same period in peat deposits in the southwestern part of the Netherlands. We conclude that pine was still present in the Netherlands during the late Holocene, like in northwest Germany [63], at least the presence of pine is established in mineral-poor bogs.
The succession patterns in the West-Frisian bogs are also important for the future, and potential effects of increased sea levels with global climate change. Inundation of the Bronze Age cultural landscape began around 1000 BC, when sea levels were approximately 2 m lower than today. After inundation, the wetlands of our study needed 1000-1500 years to develop from open ponds into raised bog, and the wetland surface continued to rise with the water table. However, with global climate change, sea level may rise to 3 m above the current level within the coming century. For the development of raised bogs, which may grow up with the water table, this is way too fast.

Conclusions
Van Amerongen [6,12] and van Zijverden [7] challenged the model of an almost treeless landscape in West-Frisia during the Bronze Age settlement period. If forest had been present during the late Bronze Age, then we would have found evidence for submerged forest at our sites. Our samples representing Late Bronze Age lack such evidence and therefore we do not accept the hypothesis of van Amerongen and van Zijverden about forested areas in West-Frisia during the Bronze Age. We suppose that people with their domesticated herbivores will have hampered the immigration of tree species because of grazing pressure when the area became accessible.
We reconstructed the start of freshwater inundation and successive peat growth in eastern West-Frisia and we found that, at the majority of our sites, inundations already started several hundred years before 850 cal BC. We suggest that solar forcing of climate change may have delivered the final push to the inundation and depopulation of West-Frisia which had already commenced several centuries before, due to sealevel rise.
We hypothesize that medieval drainage and reclamation became possible only when the mires of West-Frisia had reached the raised bog stage.
Our study is of interest for nature management of wetlands. The duration (between 1000 and 1500 calendar years) of the successive vegetation phases, from shallow, mineral-rich freshwater to ombrotrophic Sphagnum bogs may be used as a reference for newly created wetlands under eutrophic conditions. Non-pollen palynomorphs, especially spores of fungi and Zygnemataceae were valuable indicators for the reconstruction of changing local environmental conditions.