The history of early cereal cultivation in northernmost Fennoscandia as indicated by palynological research
- First Online:
- Cite this article as:
- Josefsson, T., Ramqvist, P.H. & Hörnberg, G. Veget Hist Archaeobot (2014) 23: 821. doi:10.1007/s00334-014-0446-2
- 438 Downloads
The age of the introduction of cereal cultivation in northern Europe has long been debated by researchers from many disciplines, in particular archaeology and palaeoecology. Over the past 40 years extensive palynological data have been collected concerning pre-industrial land use in northern Fennoscandia. This paper reviews palynological studies that include records of fossil cereal pollen from northernmost Sweden, Finland and Norway at latitudes north of 63°N. The geographical extent of known early cultivation sites is constantly expanding, with more than 100 records of cereal pollen pre-dating ad 1700. The oldest records of scattered cereal pollen derive from Neolithic times. Periods of continuous cultivation, indicated by cereal pollen recorded recurrently in the sediment profiles, derive from the Late Neolithic and Early Bronze Age. Collectively, the reviewed pollen records indicate that cereal cultivation was first introduced into areas close to the coast and later to the interior, and that it may have been practiced locally long before sedentary settlements based on intensive cultivation were established during medieval times. The data do not indicate a latitudinal spread of cultivation from south to north. However, methodological problems relating to pollen morphology of cereals, site characteristics and lack of connections to archaeologically excavated sites imply that the value of many early cereal pollen finds remains unclear. To increase our understanding of the context in which cereal cultivation was introduced in northernmost Fennoscandia, multidisciplinary studies integrating palaeoecology, archaeology and history are needed.
KeywordsPollen Cerealia Hordeum Agriculture Land use Scandinavia Fennoscandia
The intriguing question of when cereal cultivation was introduced into different areas of northern Europe has been explored by researchers from a range of disciplines including archaeology, anthropology and palynology (Sjøvold 1982; Zvelebil and Rowley-Conwy 1984; Berglund 1985, 1991; Johansen and Vorren 1986; Kaul and Sørensen 2012; Lahtinen and Rowley-Conwy 2013). In southern Sweden, the earliest presence of farming groups, associated with the Funnel Beaker tradition (in southern Fennoscandia ca. 4000–2800 bc), has been dated to 4100–1700 bc (Persson 1998). During the subsequent millennia, a cultivation economy associated with the Battle-Axe tradition, developed in south and central Sweden and later spread into southern Norway and southwest Finland—probably in the form of the Corded Ware tradition (Holmblad 2010). Corded Ware settlements never seem to have been established further north in Sweden, even though there are several examples of agricultural activities during the late Middle and Late Neolithic periods in this region.
Where, when and by which processes cereal cultivation first appeared in northern Fennoscandia are less well known. Throughout human history, these boreal and subarctic landscapes have been sparsely populated. Although Mesolithic hunters and gatherers affected the vegetation locally (Jensen 2004; Vorren 2005a; Hörnberg et al. 2006), the forest and mountain ecosystems in these regions have been exposed to very low levels of human influence. This is one of the reasons why these northerly parts of Europe have long been considered to be a marginal area, and consequently of limited archaeological interest (Baudou and Biörnstad 1972; Östlund and Bergman 2006). Generally, land use such as hunting, fishing and gathering natural resources can be regarded as minor disturbances in contrast to the dramatic vegetation changes resulting from the “Landnam” process (first described by Iversen 1941) when the forests were opened up substantially for grazing and cultivation (see Behre 1988; Birks et al. 1988). During recent decades, light has been shed on the introduction of cereal cultivation in northern Norway (Høeg 1972; Vorren and Nilssen 1982; Johansen and Vorren 1986; Nilssen 1988; Moe 2003) and northern Finland (Hicks 1976, 1985; Reynaud and Hjelmroos 1976, 1980; Hjelmroos 1978; Wallin and Segerström 1994), but information on early cereal cultivation in northern Sweden remains scarce (but see Huttunen and Tolonen 1972; Tolonen 1972; Engelmark 1976)—especially for areas north of latitude 65°N (Segerström 1990a, b, 1995).
This paper presents a critical review of the first signs of cultivation, as traced by pollen records, in northernmost Fennoscandia focusing on the northern parts of Sweden. The time span that we focus on extends from 5000–3000 bc with some of the earliest finds of cereal pollen in lake and peat sediments in coastal parts of northern Finland and Norway (Reynaud and Hjelmroos 1980; Vorren and Nilssen 1982), to ad 1700 when the large-scale process of farmers colonizing the interior of Sweden started (Bylund 1956; Rudberg 1957). Earlier finds of cereal/wild grass pollen types (predating the introduction of agriculture in southern Scandinavia) are not included in this overview but are presented in an Appendix (see ESM). We also present information pertaining to archaeologically excavated sites that describe some important traits of the pre-sedentary cultivation in this region. The pollen records are compared and discussed with regards to separation of cereals from wild grasses, pollen productivity and dispersion of cereals, and dating techniques and precision.
Chronological table of archaeological time periods for Scandinavia mentioned in the text
Early Bronze Age
Late Bronze Age
Pre-Roman Iron Age
Early Roman Iron Age
Late Roman Iron Age
The presented pollen data include anthropochores (cereal pollen types) and pollen types from apophytes (native species favoured by human activities) co-occurring with cereal pollen types. The pollen data, in terms of pollen-type taxonomy and nomenclature, was harmonised using the same procedure as the one used in the European pollen database (http://pollen.cerege.fr/fpd-epd/). The following principles were applied in this review. For pollen data where published criteria or keys were used to identify cereal grains we used the pollen-type names “Cerealia type” (Cerealia-t); “Avena-type” (Avena-t); “Hordeum-type” (Hordeum-t); “Triticum-type” (Triticum-t); and Secale cereale. Following the pollen-type name, the letters A (Andersen 1979), B (Beug 1961, 2004), Bt (Bennett 2004), E (Erdtman et al. 1961, 1963; Erdtman 1969), F (Fægri and Iversen 1964, 1975; Fægri et al. 1989), G (Grohne 1957), M (Moore and Webb 1978; Moore et al. 1991), P (Punt 1976–2003) or R (Reille 1992–1998) were added according to the source of the criteria/keys used. Pollen data where no published criteria or keys were used for separating cereal grains from other grasses, or where the criteria or reference to a specific key were lacking, were classified as Cerealia-t and marked with an asterisk (*)—i.e. we regarded such identifications as uncertain. Please note that Cerealia-t includes all cereals except Secale cereale, but since some grains of Secale cereale may look like Hordeum-t, this latter group can also include Secale cereale (see Beug 2004).
Physical geography and climatic conditions
The areas surrounding the Gulf of Bothnia are characterized by isostatic land uplift that occurred following the last glaciations. The uplift was most severe during the early Holocene (Påsse and Andersson 2005), and in the areas affected, the relative uplift since deglaciation has been ca. 260 m (Lindén et al. 2006). In the region around Umeå, Sweden, this uplift is still occurring quite rapidly and amounts to ca. 0.8 m per century (Ågren and Svensson 2007). In the surrounding areas, this process is slowing and at the fringes in northern Norway and Finland it has practically ceased (Påsse and Andersson 2005). Consequently, when evaluating the cultural history in the region surrounding the Gulf of Bothnia, the land uplift must be acknowledged because the coastal landscape has changed dramatically over time (see Wallin and Segerström 1994). The climate in northern Fennoscandia is mainly temperate despite its high latitude; this is due to the moderating influence of the North Atlantic Current along the northern Norwegian coast. In the western part, the climate is typically maritime and strongly influenced by Atlantic weather systems, whereas in northern Finland and northernmost Sweden the climate is more continental and largely affected by weather systems from the east. In the extreme north, the climate is typically subarctic, maritime along the Norwegian coast and more continental in north-eastern Finland.
Palynological records and archaeological context
In a broad sense, cultivation includes production of food by preparing the land to grow crops. From an archaeological point of view, two forms of cultivation that were used can be distinguished: extensive cultivation which was used by mobile groups as a minor complement to other forms of sustenance, and intensive cultivation that was performed by sedentary groups and where the cultivation outcome formed a major part of their total economy (Boserup 2003). The extensive form was a system that used simple tools, or none, and small inputs of labour relative to the land area being cultivated, whereas the intensive form required large investments, including permanent buildings such as houses, barns and fields, to be used over long time periods. Naturally, the remains of an intensive cultivation system are more readily detected through palynological and archaeological methodologies than traces of extensive cultivation (Behre 1981).
In general, the areas north of latitude 63°N in Sweden have been considered to have been largely unaffected by land use until the Medieval Period when many sedentary settlements were established near the coast and along the major rivers—indicated by both historical and archaeological records (see Rudberg 1957; Wallerström 1995). According to archaeological material, settlements with agrarian elements at latitude 61°N can be dated to times before the first millennium bc in the coastal areas on both sides of the Gulf of Bothnia (see Holm 2006; Holmblad 2010), i.e. at the same time as in coastal areas in northern Norway at latitude 68°N (Johansen 1982; Valen 2007; Arntzen and Sommerseth 2010; Jensen 2012). Clear indications of cereal cultivation were also found in coastal parts of north-central Sweden (62–63°N) from 2700 to 2500 bc (Engelmark 1978; Ramqvist 1998; Gustafsson and Spång 2007). Hence, it is evident that there has been cultivation of the land for at least 3,000 years in the regions west, east and south of the Swedish Gulf of Bothnia’s coastal areas and the interior of northern Sweden.
The coastal areas of northern Sweden and Finland along the Gulf of Bothnia
The idea that there may have been cultivation phases of cereals in northernmost Sweden before ad 1500 is not new (e.g. Sundström 1984; Segerström 1990a). Finds of cereal pollen indicating early cultivation along the coastline of the Gulf of Bothnia were presented in the 1970s. However, because of the scarcity of archaeological findings supporting the palynological studies (i.e. artefacts, monuments, structures or features that may be associated with cultivation), the implications of these pollen records remain unclear (e.g. Baudou 1977).
According to previous archaeological studies, the Örnsköldsvik area constitutes the northernmost limit of sedentary, agrarian settlements during the first millennium ad in Sweden (Fig. 1) (Ramqvist 1983, 1998). At Gene, cultivation of Hordeum sp. (barley) and Avena sp. (oats) on manured, permanent fields, and livestock farming including Bos taurus (cattle), Ovis aries (sheep), Capra aegagrus hircus (goat), Sus scrofa domesticus (pig) and Equus ferus caballus (horse) is indicated by carbonized cereal seeds and osteological analyses (Ramqvist 1983; Lindqvist and Ramqvist 1993). Unfortunately no palynological studies have been undertaken in this area. Just northeast of Örnsköldsvik, at Banafjäl, lies another settlement dated to the Pre-Roman Iron Age (Fig. 1) (Lindqvist 2000). Here, a range of non-specified pollen of Cerealia-t* and arable weeds was found in a layer between the mineral soil and peat at the bottom of a mire (dated to 195–35 bc), indicating that the land forming the base of the mire was previously used for cultivation (ESM) (Engelmark 1994).
List of sites from south to north in Sweden with dates of the earliest finds of cereal pollen types presented as in the original publication and with harmonised taxonomy and nomenclature. For a complete list with coordinates, elevation and archive type, see Appendix (ESM)
In the Umeå area, ca. 100 km further north, an early phase of cultivation is indicated by scattered pollen grains of Cerealia-t (B E M) and Triticum-t (B E F) and several weed pollen types (Artemisia vulgaris, Cannabis, Chenopodiaceae, Plantago major, Rumex acetosa/acetosella) dated to 900–800 bc (Tolonen 1972; Segerström 1990c) and to 700 bc (Engelmark 1976) (Table 2; Figs. 1, 2). As in the Örnsköldsvik area, recent finds of charred cereal grains at an excavated settlement site, Mariehem in Umeå, that have been identified as Hordeum (probably naked barley) and dated to 1270–1040 bc (see Viklund 2011), support previous reports of early cultivation in the Umeå area (Table 2, ESM). This settlement site was used during two periods: 1800–1700 bc and during the transition to the Pre-Roman Iron Age, i.e. the centuries around 500 bc (Forsberg 1999). No house foundations were discovered, but charred bones of goat/sheep and remains of weeds (e.g. Polygonum aviculare) indicating animal grazing at the site were recorded. At Kassjön, Umeå, permanent cereal cultivation, indicated by Hordeum-t (B E M), was practiced from ad 1200 according to Segerström (1990c), and at Prästsjön, Umeå, there is indication of permanent cultivation (Cerealia-t, B M, recorded recurrently) from ad 500 and onwards (Engelmark 1976). A bit further inland at Vindeln, recurrent finds of Secale cereale (B M) indicate cultivation from ad 850 and onwards (Table 2; Figs. 1, 2) (van der Linden et al. 2008). About 60 km north of Umeå lies Ånäset and the excavated Mårtenfäboda site (Fig. 1), a Bronze Age settlement with cleared areas and clearance cairns, fire-cracked stones, cooking pits, embankments and settlement remains from 1000 to 500 bc (Forsberg 1999). Clearance cairns are commonly found at sedentary agricultural settlements (cf. Lagerås and Bartholin 2003; Overland and Hjelle 2013). Not far from this site, near Ånäset, Wallin (1996a) found scattered occurrences of Hordeum-t (B M) in peat deposits dating back to 400 bc (Table 2; Figs. 1, 2). Another 80 km north of Umeå, close to the coastal town of Byske, lies one of the earliest Neolithic sites in northernmost Sweden that has been excavated, the Bjurselet settlement (Fig. 1). According to the archaeological excavations carried out in 1962–1968, this settlement was initially used as a temporary seal hunting station 4,000–4,500 years ago. For a couple of centuries during the second millennium bc it had typical elements of the Battle-Axe culture found further south (Knutsson 1988; Knutsson and Christiansson 1989). At this site, bones from goat/sheep, thick-butted flint adzes, the use of a southern lithic technology and Battle-Axe pottery were found. The pollen records include pollen types such as Cannabis/Humulus, Rumex acetosa/acetosella and Polygonum aviculare, indicating human influence and possibly cereal cultivation (Cerealia-t* pollen) as early as 1950 bc (Table 2; Fig. 2), and permanent settlement and cereal cultivation during the first centuries ad (Königsson 1970).
List of sites from south to north in Finland with dates of the earliest finds of cereal pollen types presented as in the original publication and with harmonised taxonomy and nomenclature. For a complete list with coordinates, elevation and archive type, see Appendix (ESM)
On the Finnish side of the Gulf of Bothnia no sites with sedentary settlements from the Late Neolithic have been excavated within the studied region. Although the Battle-Axe tradition was very strong in this region until 2300 bc, no farming societies seem to have developed here as occurred in regions further south (Vilkuna 1997; Taavitsainen et al. 1998; Holmblad 2010). Even so, the oldest palynological records from the coastal areas indicate early sporadic cultivation (Cerealia-t, B F) from 4100 to 1000 bc and onwards in the Oulu region (Reynaud and Hjelmroos 1980) (Table 3; Figs. 1, 2). North of Kemi in Tervola, Reynaud and Tobolski (1974) analysed soil samples from the lowest terrace of the Kemi River and recorded Cerealia-t* pollen from ad 650 onwards (Table 3; Figs. 1, 2). In Kilmajänkä, also situated north of Kemi, Reynaud and Hjelmroos (1980) found sporadic occurrences of Cerealia-t (B F) pollen in peat deposits dated to 300 bc (Table 3; Figs. 1, 2).
In the areas near Vasa, Wallin and Segerström (1994) studied peat profiles from small bogs and found evidence for cereal cultivation (Hordeum-t, B E F M) from 100 bc onwards. Further findings in the same area point to almost continuous land use including cereal cultivation (pollen of Cerealia-t*, Hordeum-t, Secale cereale and Triticum-t, B E F) from ad 250 to 300 (Table 3; Figs. 1, 2) (Tolonen et al. 1976; Miettinen and Vuorela 1988). In the same region, near Vörå (Pörnullbacken) lies one of the earliest Iron-Age settlements in northern Finland that has been excavated—established during the first century ad (Baudou 1991; Viklund and Gullberg 2002). At this site large quantities of charred cereal grains of Hordeum sp. indicating local cultivation were discovered during excavations in the late 1990s (Viklund and Gullberg 2002). Cerealia pollen has also been recorded in soil samples taken from the archaeological excavations of a medieval settlement of a so-called Birkarla (bailiff) family on the island of Oravaisensaari near Torneå (Fig. 1) (Koivunen 1991). These findings point to the beginning of cereal cultivation (pollen of Cerealia-t, Secale cereale, A B) around ad 1350 on this island, and continuous cereal cultivation from ad 1500 onwards (Table 3; Fig. 2) (Hjelmroos 1974, 1978). During the archaeological excavations, house foundations and iron objects from ad 1500 were found together with ceramics from ad 1400 and a bronze buckle (Hjelmroos 1978). Furthermore, on the island of Hailuoto, located just offshore of Oulu, Reynaud and Hjelmroos (1976) and Hicks (1992) found evidence for a colonization process that had already started in ad 1100 and cereal cultivation (indicated by Cerealia-t, B, and Hordeum-t and Secale cereale, F M) from at least ad 1200 (Table 3; Figs. 1, 2). The Cerealia pollen was frequently identified as Secale cereale (F M) which points to very early cultivation of rye on this island, accompanied by Centaurera cyanus—a herb that in northern Europe is closely linked with the introduction of rye cultivation (see Mikkelsen 1952; Behre 1981; Berglund 1991; Hjelle 2007; Lempiänen 2007).
The coast of northern Norway
List of sites from south to north in Norway with dates of the earliest finds of cereal pollen types presented as in the original publication and with harmonised taxonomy and nomenclature. For a complete list with coordinates, elevation and archive type, see Appendix (ESM)
The interior areas of northern Fennoscandia
According to previous research, early temporary cultivation and subsequent development of an agricultural economy in northern Fennoscandia first occurred in the coastal areas (see Johansen and Vorren 1986; Segerström 1990a; Myrdal et al. 1998). It is also in such areas that the majority of the palynological investigations included in this review were carried out. In fact, the geographical extent of pollen studies performed in the interior of northern Sweden, Finland and especially Norway, is limited (Fig. 1). Nevertheless, in northern Sweden ca. 150 km inland from Umeå, finds of Cerealia-t (B E F M) from lake sediments indicate cereal cultivation as early as ad 1000 (Cerealia-t recorded continuously from this date) in Åsele (Segerström 1982), and from the same date in the Hotagen area (sporadic occurrences of Secale cereale, Bt) close to the Norwegian border (Giesecke 2005) (Table 2; Figs. 1, 2). Close to Näsåker and to Hoting, in peat deposits dated to ad 1000-1500, von Stedingk and Baudou (2006) recorded single and scattered cereal pollen of Hordeum-t (M).
Further north, about 100 km inland from Piteå lies Kåtaselet and the settlement of Lappviken, which was used during both the Neolithic and Medieval periods. Here, Robertsson and Miller (1985) found Cerealia-t* pollen together with pollen of Epilobium, Caryophyllaceae, Chenopodiaceae, Artemisia, Asteraceae, Urtica and Rumex-types in soil samples taken from a Late Medieval house foundation (Table 2; Figs. 1, 2). Unfortunately, no dates are available for this pollen record. Another 100 km inland from Kåtaselet finds of cereal pollen in the Arjeplog area indicate early cultivation in sub-montane forest areas. At the Munka settlement, just north of the Arctic Circle and ca. 80 km north of Arjeplog, Josefsson et al. (2009) found scattered Cerealia-t (M) in peat deposits dated to ad 1600 (Table 2; Figs. 1, 2). About 100 km west of Arjeplog, in areas where cultivation is generally believed not to have occurred during historic or prehistoric times, scattered Hordeum-t (M) pollen was recorded in peat deposits that dated from ad 1350 in Gieddeålge (Staland et al. 2011) and ad 1700 in Hiednikvalta (Karlsson et al. 2009) (Table 2; Figs. 1, 2).
In the northernmost parts of Sweden, there are several pollen studies that indicate cereal cultivation prior to ad 1700. In a study of past vegetation changes and land-use history along the Lule River valley, Segerström (1990b) collected data from two lakes with varved sediments: Edefors, and the previously mentioned Heden near Boden. The sediments from the inland locality Edefors included Cerealia-t (A B M) pollen dated to ad 750 (Table 2; Fig. 1). Furthermore, finds of Hordeum-t were dated to ad 1200 and of Secale cereale to ad 1450. Continuous finds of Cerealia-t were recorded from ad 1650 onwards. Earlier indications of land use in this area include finds of pollen types from Artemisia vulgaris, Caryophyllaceae, Asteraceae and Rumex acetosa/acetosella from ad 500 (Segerström 1990b).
West of Porjus, at two forest sites situated ca 200 km inland from the coast, Hörnberg et al. (1999) and DeLuca et al. (2013) found scattered Hordeum-t and Secale cereale pollen grains (M) in peat sediments dated to ad 1400 and onwards (Table 2; Figs. 1, 2; ESM). These findings are believed to indicate small local areas of cultivation created in forests managed with fire by people from a nearby village, where cultivation of Hordeum is described in tax documents from the latter part of the 18th century. In the Torneträsk area (Abisko), Jankovská and Kociánová (2010) studied past vegetation and palsa development and found scattered Cerealia-t (F M P R) pollen in the uppermost 65 cm of a palsa possibly dating to 600 bc (Table 2; Figs. 1, 2). Hörnberg et al. (2012) studied the vegetation development within a nature reserve about 100 km north of Torneå (at Ahmarova, northwest of Övertorneå), and found single Hordeum-t and Secale cereale pollen grains (M) dated to ad 1600 indicating cultivation at this site (Table 2; Figs. 1, 2).
In the interior of northern Finland there are several areas with interesting records of pollen types indicative of early cereal cultivation: Kuopio, Pihtipudas, Nurmes, Vaala, Kuusamo and Övertorneå. In Siilinjärvi near Kuopio, Grönlund et al. (1992) studied varved lake sediments close to the Pöljä Middle Neolithic settlement site (Fig. 1) which has been dated to 2900–1800 bc (Pälsi 1929; Carpelan 1979). Grönlund et al. (1992) recorded Cerealia-t* pollen during two short periods, the Bronze Age (before 665 bc, i.e. the oldest varve year) and ad 150–200 (Table 3; Figs. 1, 2). Both time periods were also shown to be characterized by a decrease of Picea pollen, while Betula and non-arboreal pollen increased, which was interpreted as evidence of large impact by humans. Cerealia-t* pollen were recorded almost continuously from ad 1650 and onwards. However, there is no evidence of a sedentary Iron Age settlement in this area (cf. Pihlman 1995) although the excavated Luukonsaari site, which has given its name to a ceramic style dating to the Pre-Roman Iron Age, lies close to Kuopio (Meinander 1969).
In Lake Karvalampi, in the parish of Pihtipudas, Vuorela (1997, 2007) found early indications of cultivation (scattered pollen from Cerealia-t, B) dated to 2300 bc (Table 3; Figs. 1, 2), and continuous cultivation (Cerealia-t, Secale cereale and Cannabaceae pollen) from ad 1300. Near Nurmes, Augustsson et al. (2013) recorded a single Hordeum-t (B P) pollen in sediments from Lake Lehmilampi dated to 3000 bc (Table 3; Figs. 1, 2). They also recorded Hordeum-t pollen continuously from 1500 to 200 bc. In the Vaala region (Lake Ahmasjärvi, Lake Nimisjärvi and Sotkasuo), Reynaud and Hjelmroos (1980) recorded Cerealia-t (B F) pollen together with pollen types of Artemisia and Cannabis/Humulus from 5500 to 3450 bc (Table 3; Figs. 1, 2). In Ahmasjärvi, Cerealia-t (B F) (and later Secale cereale, B F) was recorded almost continuously from 2700 bc. From the Kuusamo district, close to the Arctic Circle, cereal cultivation is indicated by scattered finds of Cerealia-t* pollen from ad 1300 in the sediments of Lake Paanajärvi (Bondestam 1999), and from ad 1700 in peat deposits at Kangerjoki (Hicks 1976) (Table 3; Figs. 1, 2). In the Övertorneå area, on the Finnish side, the oldest records of Cerealia-t (B F) are dated to 3550 bc–ad 1100 (Hjelmroos 1978; Tikkanen 1978; Reynaud and Hjelmroos 1980) (Table 3; Figs. 1, 2).
Few palynological studies have been performed in the interior of Norway. However, in Skaitidalen east of Røkland, Høeg (1972) recorded scattered Hordeum-t (B E F G) pollen in deposits dated to 3350 bc and recurrent finds during the first millennium bc (Table 4; Figs. 1, 2). Other pollen recorded during this time period included Rumex acetosa-type, Urtica, Chenopodiaceae, Plantago major, Plantago lanceolata and Spergularia-type that together with Hordeum-t pollen indicate continuous farming and cereal cultivation. In Oaggejokka near Karasjok, Høeg (2000) found Hordeum-t* pollen together with Rumex and Artemisia pollen types (not specified), Chenopodiaceae, Plantago major and P. lanceolata indicating farming including cereal cultivation at 1350 bc (Table 4; Figs. 1, 2).
Discussion and methodological considerations
Over the past 40 years a variety of palynological information has been collected concerning pre-historical land use in northern Fennoscandia. Pollen types identified as Cerealia-t or Hordeum-t have commonly been recorded in deposits dating from times for which it is generally assumed that cereal cultivation was not yet introduced. The palynological data included in this review indicate that cultivation of cereals may have been practiced locally over several thousand years—in coastal areas as well as in the interior—long before the establishment of sedentary settlements based on intensive cultivation during the Medieval Period. There are many factors influencing the results of pollen analyses that make comparisons between these different studies/sites problematic. Some of the more important are: the land-use intensity; sampling techniques and the size of the sampling site; the kind of biological archive used (soil, peat or lake sediment); the resolution of the pollen analysis; the number and precision of radiocarbon dates; the pollen productivity and dispersion of cereals; and separation of cereals from wild grasses on the basis of pollen morphology—each of them with several inherent problems and challenges, some of which are highlighted and discussed below.
Identification of cereal pollen
Interpretations of pre-historic agrarian land use are often based on the presence or absence of putative cereal pollen, but separating the pollen of cereals from wild grasses characterized by large pollen grains is problematic and has been the subject of many previous studies (e.g. Beug 1961, 2004; Leroi-Gourhan 1969; Andersen 1979; Köhler and Lange 1979; Vorren 1986; Odgaard 1994; Hannon and Bradshaw 2000; Edwards et al. 2005; Tweddle et al. 2005; Joly et al. 2007). The principal criteria for separating cultivated from wild species of Poaceae are grain shape and size, pore diameter, pore annulus breadth and height, and exine sculpture (Beug 1961, 2004; Andersen 1979; Köhler and Lange 1979). Generally, grass pollen with grain and annulus diameter >37.0 and >8.0 μm, respectively, have been identified as characterizing Cerealia pollen (Beug 1961; Andersen 1979), and in a more recent key by Beug (2004) pore diameter and annulus width should exceed 2.7 μm, and annulus thickness should be larger than 2.0 μm. However, acetolysis, hydrofluoric acid (HF) treatment and mounting medium (silicon oil or glycerine) all affect pollen size (see Moore et al. 1991 and references therein). Joly et al. (2007) performed careful measurements of 35 different species of both wild and cultivated grasses mounted in glycerine and found that a biometric threshold of two values of diameter (47 μm for the grain and 11 μm for the annulus) provides a more precise tool for differentiating these groups. However, the pollen grains of several genera of wild grasses also fulfil the criteria (large sizes) of Joly et al. (2007), which imply that the characteristics of the exine sculpture (if visible) constitute an important tool in the identification. Using and presenting appropriate criteria and references is thereby crucial for identifying early cereal cultivation and has been stressed previously for example by Behre (2007) and Lahtinen and Rowley-Conwy (2013). Preferably, several methods should be applied. In several studies included in this review, the methods, criteria or the references used for identifying cereal pollen were not presented (here classified Cerealia-t*, Tables 2, 3; ESM). Such records of cereal pollen types are thus questionable and unfortunately of limited value in the discussion of the earliest cereal cultivation.
Because of prolate grain shape, large size and the sub-polar position of the pore (see Köhler and Lange 1979) pollen of Secale cereale is the only cereal pollen type that can be confidently identified to species level. Interestingly, in the coastal areas of both northern Finland and Sweden there are early records of Secale cereale pollen occurring recurrently from ad 300 and 850 and onwards (Tables 2, 3; Fig. 2), indicating rye cultivation from the Late Roman Iron Age and the Migration Period, i.e. concurrently with areas in southern Scandinavia (see Behre 1992). Also in interior Sweden (Näsåker, Hotagen, Edefors and Porjus) there are scattered occurrences of Secalecereale from the period ad 950–1400. Pollen from all other cereals can only be identified to the genera or group of genera/group of species level, and there are often species of wild grasses included in the pollen type.
The palynological records cited in this review show that large pollen grains of grasses are often divided and grouped into two types, Cerealia-t and Hordeum-t (Tables 2, 4; ESM), where both groups are “problematic” and include species of both wild and cultivated grasses. In northern Fennoscandia, distinguishing cereal pollen from pollen of Leymus arenarius (a large wild grass growing on sandy soils mainly along the shoreline of the Gulf of Bothnia), Elytrigia repens (a common wild grass on nutrient-rich soils) and Glyceria fluitans (a wetland wild grass with large pollen, mainly found in southern and central Fennoscandia and with only a few sporadic occurrences along the shoreline of the Gulf of Bothnia) is especially difficult. For this reason, single occurrences of Cerealia-t or Hordeum-t pollen found in sediments many thousands of years old by for example, Høeg (1972), Reynaud and Hjelmroos (1980), Segerström (1990b), Jensen (2004), Hörnberg et al. (2006), Vuorela (2007) and Augustsson et al. (2013) are questionable (Tables 2–4; ESM). Most certainly, the recorded Cerealia-t or Hordeum-t pollen predating the introduction of cereal cultivation in southern Scandinavia (see Berglund 1991; Behre 2007) and north-eastern Europe (see Poska et al. 2004) is in fact pollen from some of these wild grasses (Tables 2–4; ESM).
At some locations, cereal pollen identified as Triticum-t has been recorded. In northern Sweden (Anundsjö) and Norway (Lofoten) (Tables 2, 4), sediments with Triticum-t pollen recurring simultaneously with Hordeum-t pollen and pollen of apophytes such as Cannabis/Humulus and Urtica has been dated to 2650–2200 bc. These findings suggest that Triticum was cultivated sporadically during the Middle and Late Holocene, and contribute to the ongoing discussion regarding early cereal cultivation, not only of Hordeum but also of Triticum, in northernmost Fennoscandia (Johansen 1979; Zvelebil and Rowley-Conwy 1984 with comments in Norw. J. Rev. 1984–1985; Berglund 1985; Johansen and Vorren 1986; Taavitsainen et al. 1998; Lahtinen and Rowley-Conwy 2013). Data sets from northern Fennoscandia where Avena-t pollen constitutes the earliest recorded cereal pollen type are rare (see ESM), except for a few places in northern Norway (e.g. Moe 1991; Nilssen 1988). The Avena–t pollen is in general recorded in more recent times (cf. Jensen 2012), possibly due to a late introduction of A. sativa, but also because this pollen may have been included in the Triticum-type (see Andersen 1979; Moore et al. 1991). The wild grass, A. fatua, has a very restricted distribution in Fennoscandia (Hultén 1971) and in northernmost Sweden (Stenberg 2010).
The bio-statistical approach presented by Joly et al. (2007) represents one approach to separating pollen of cereals from wild grasses. However, an absolute distinction between these pollen types is probably impossible since the size ranges of cereal and wild grass pollen overlap (cf. Beug 2004; Behre 2007). One way to increase the prospect of successful identification of fossil cereal pollen is by establishing a regional reference pollen collection including both native, regional wild grasses and local “traditional” cereal crops (with careful measurements of grain, annulus and pore diameter and identification of specific exine characteristics) and for which all samples (reference and study sites) are handled and chemically treated in exactly the same way. Furthermore, because the traditional interpretation of pollen taxa and their relation to specific forms of land use (including cereal cultivation and animal farming) is based on land use in southern environments that are different from those in northern Fennoscandia (cf. Behre 1981; Bos and Janssen 1996) alternative sets of indicators should be used, as proposed, by Hicks (1985, 1988), Vorren (1986) and Josefsson et al. (2009). Field and micro-morphological analyses of relict cultural soils, as performed by Simpson et al. (1998), and their fossil pollen contents (Segerström 1991) together with the use of non-pollen palynomorphs to identify animal husbandry (van Geel 2001; Gauthier et al. 2010) represent additional methodologies that could be applied.
Source area of cereal pollen grains
Much previous palynological research on identifying early cultivation has been interpreted in relation to regional vegetation history, focusing on reductions in arboreal pollen percentages (indicating forest clearance), increases in grasses and herbs (signifying increases in open land) and the appearance of pollen from cereals and associated weed species (see Behre 1981, 1988; Birks et al. 1988). However, several studies included in this review point to a sequence of agricultural development in this northern region that was different from the dramatic Landnam phase found in southern Fennoscandia and further south (see Huttunen and Tolonen 1972; Engelmark 1976, 1978; Reynaud and Hjelmroos 1976).
In the reviewed palynological records, scattered occurrences of pollen identified as Cerealia-t or Hordeum-t are found far back in time. As discussed by Segerström (1990a), the interpretative value of such pollen merits discussion, and if it in fact is believed to represent early cereal cultivation one must ask what source area it reflects as well as what level of intensity of cultivation (i.e. duration and size of the cultivated area) it represents. Defining the pollen source area is critical (Jacobson and Bradshaw 1981; Seppä and Bennett 2003), and calculations of source areas of cereal pollen for large lake basins in southern Sweden, based on regional pollen productivity estimates and fall speed of pollen, suggest that ca. 50 and 70 % of the cereal pollen comes from an area with a maximum distance of 3 and 7 km, respectively (Hellman et al. 2009). These results indicate that cereal pollen recorded in sediments from large lakes will mainly come from relatively small areas even in open and highly cultivated landscapes. Moreover, in northernmost Fennoscandia, where most regions have been densely forested in the past, the source area of cereal pollen is restricted—especially if the sampling site is a small lake or mire or a closed canopy hollow (Bradshaw 1988).
The cereal pollen types Hordeum-t, Triticum-t and Avena-t do not produce large amounts of pollen as they are to some extent self-pollinated, and they disperse over short distances in forested environments because canopy filtering prevents pollen transport by wind and because the horizontal wind-drift is low (Vuorela 1973, 1975; Hicks 1988). Certainly, some cereal pollen grains may have been dispersed during handling of the crop (Vuorela 1975; Jensen 2012). It has also been demonstrated that most of the pollen from entomophilous cereals (e.g. Hordeum vulgare) in cultivated fields is dispersed short distances into surrounding forests in comparison to wind pollinated species such as Secale cereale that are more widespread (Vuorela 1973, 1985; Segerström 1991). Hence, small patches of cultivation used for a limited time in a forested landscape will rarely result in a strong signal of cereal pollen in palynological analyses. Still, it is not unusual that old records of cereal pollen types found at sites across northern Fennoscandia are explained by long distance dispersal (Segerström 1990b; Lahtinen and Rowley-Conwy 2013).
Although long distance dispersal of cereal pollen types cannot be excluded in any pollen analysis (see Behre 2007), this explanation should not disqualify alternative interpretations, including small-scale cultivation, even though a sampling site seems inappropriate for cultivation today. Naturally, recurrent recordings of carefully identified cereal pollen types have a higher interpretive value—especially if they occur in association with apophytes (as those proposed by Hicks 1985, 1988; Vorren 1986 and Josefsson et al. 2009), and/or are recorded close to archaeological sites with identified cereal macrofossils. The old findings of Cerealia-t, Hordeum-t or Triticum-t pollen grains recorded more or less continuously, for example in Anundsjö and Umeå (Sweden) and Vörå, Vasa and Hailuoto, Oulu (Finland) provide such examples (Fig. 2). At archaeological sites it is also important to pay extra attention to cereal macrofossils since hunter-gatherers may have used wild grasses like Elymus arenarius as food (see Guðmundsson 1996).
Some of the oldest findings of cereal pollen types included in this review are from northern Finland, i.e. Pihtipudas, Vaala, Nurmes, Oulu and Övertorneå (Fig. 1). At some of the sites studied by Reynaud and Hjelmroos (1980), the suggested dates indicating very early cereal cultivation were based on single or scattered occurrences of Cerealia-t—long before periods with recurrently occurring cereal pollen. At other sites in northern Finland, however, Cerealia-t was recorded recurrently during periods 4000–200 bc (see Tikkanen 1978; Reynaud and Hjelmroos 1980; Augustsson et al. 2013). These findings are based on old dating techniques (but not Augustsson et al. 2013) and lack archaeological context and therefore remain uncertain. In the interior of northern Sweden and Norway there is also a lack of archaeologically excavated sites with agricultural elements from the Iron Age and further back in time. Consequently, the findings of cereal pollen types occurring recurrently at sites near Vaala and Oulu in Finland and near Åsele (from ad 950 and onwards) in Sweden and Røkland (first millennium bc) in Norway merit further studies in the interior parts of Fennoscandia (Fig. 1).
Precision of dating
The principal dating techniques applied by palynologists are radiocarbon dating and varve dating (i.e. counting annual layers in lake sediments). Radiocarbon dating includes two main methods: the more time-consuming conventional method that detects the 14C activity in large samples, and the fast modern accelerator mass spectrometry (AMS) method that detects the proportion of 14C atoms and only requires small samples (Bowman 1990). The half-life of 14C was originally determined to 5,568 ± 30 years (Libby 1955), but was later set to 5,730 ± 40 years (Godwin 1962). To avoid confusion there is a recommended convention to use the Libby half-life (Bowman 1990). The robustness of the radiocarbon dating depends on a variety of factors that influence the radiocarbon content in the dated material (e.g. temporal variation in 14C production, radiocarbon reservoir effects and contamination) that may in turn affect the estimated age (see Weiner 2012; Jull et al. 2013). These factors, or the underlying assumptions for the dating techniques, however, will not be discussed in this review. Our main focus concerning dating has been to assess the reliability of the age estimations of recorded cereal pollen in different studies. In general, older studies including conventional radiocarbon dating are less accurate in comparison to more modern AMS ones mainly because the dated sample was larger and often included many levels in the stratigraphy. We have treated many old 14C dates as uncertain. Although such dates do not precisely pinpoint a specific time event, such as the first appearance of cereal pollen, they do provide a rough estimate of the age of such findings.
In contrast, varved sediments, formed in lakes with a minimum of bioturbation, in general give accurate calendar ages when counted (Renberg 1981; Segerström 1990b, c). However, problems in the development of varved sediments may occur due to, for example, adverse weather conditions, sub-laminations, large flood events or by deviations in annual deposition pattern (Lowe and Walker 1997). Although there are a number of factors that may affect dating based on varve-counting, we have considered these dates as certain.
Integrating palaeoecology, archaeology and history to reveal land-use history
The early presence of cereals in palynological records from different parts of northern Fennoscandia indicates a process involving both coastal and interior communities of incorporating cultivation and animal farming successively into their economic basis (that primarily relied on hunting and fishing, gathering of natural resources and reindeer herding, for example) prior to the establishment of sedentary settlements (Bergman 2010). How cereal cultivation was introduced into this region, i.e. through contact between local communities and farmers from areas further south (Hjelmroos 1978; Ramqvist 1983; Liedgren 1988; Taavitsainen et al. 1998; Vorren 2002; Ramqvist 2012), or by immigrant farmers settling in these northerly areas (Baudou 1989), or both, remains unclear.
One major challenge when studying Neolithic, Bronze Age and Iron Age cultivation in these northerly areas is to understand the spatial and temporal characters of cereal cultivation undertaken by the mobile hunter–gatherer groups. For example, the extent to which modification of local economies (to incorporate cultivation) was accepted by these people (see Engelmark 1976) is uncertain, as is the way that cultivation was implemented over time—whether continuously or through reintroductions several generations apart. This implies that meaningful interpretation of the level of intensity of early cereal cultivation and related land-use practices is impossible unless the palynological information is linked to human settlements in the region and to the land-use practices of people living in those settlements (Binford 1980; Cronon 1983), i.e. studies from small influx basins (Bradshaw 1988; Calcote 1998) should be located close to dwelling sites that have been verified and excavated by archaeological means (see Berglund 1985; Josefsson 2009; Hjelle et al. 2012; Kaul and Sørensen 2012). The early finds of cereal pollen in areas used by people of different “cultural identities” (for example the staðir farms of north and central Norway, see Moe 1991; Vorren et al. 1990; the Karelian settlements on the Finnish–Russian border, see Bondestam 1999; and the nomadic Sami settlement sites in northern Sweden, see Josefsson et al. 2009)—all with different sets of economic strategies and settlement/movement patterns—suggest that a broad spectrum of methods and alternative interpretations of palynological and archaeological findings should be considered.
A comprehensive understanding of early cultivation in northernmost Fennoscandia is hampered by a lack of large, interdisciplinary projects (however, see Jensen and Elverland 2009; Kaul and Sørensen 2012) that could facilitate knowledge transfer among research groups and disciplines, but that could also develop joint empirical studies (cf. Berglund 1991; Taavitsainen et al. 2007). New, intriguing objectives for such an approach would be to use archival records to examine how and at what scale groups that traditionally have not been considered to be farmers may have practiced cultivation during pre-historic times and relate this information to agricultural development in other parts of Fennoscandia, for example by using petroglyphs. Furthermore, in the areas surrounding the Gulf of Bothnia, diachronic analyses of place names may enable interpretations of temporal and spatial patterns of land use along the river valleys.
According to this review of palynological records, there are no clear patterns of a latitudinal spread of cultivation from south to north as suggested, for example, by Wallin (1996b). However, the palynological and archaeological data do indicate that cereal cultivation was first introduced in areas close to the coastlines of the Gulf of Bothnia and the Norwegian/Barents Sea and later to the interior of Sweden and Finland, lending support to previous discussions by Johansen and Vorren (1986) and Myrdal et al. (1998). The earliest and most reliable indications of cereal cultivation during the Late Neolithic and Early Bronze Age along the coastlines of the Gulf of Bothnia (i.e. Anundsjö, Umeå, Vasa and Oulu/Hailuoto) are based on recurrent findings of cereal pollen (where the authors have used well-defined criteria and references to distinguish cereal pollen types) at sites close to archaeologically excavated sites with agricultural elements. Following the same principles, there are no reliable indications of early cereal cultivation in the interior parts of northern Fennoscandia, but the strongest early indications are from Late Neolithic/Early Bronze Age (Vaala and Nurmes, Finland), Late Bronze Age (Røkland, Norway) and the Viking and Medieval Periods (Åsele and Edefors, Sweden).
The palynological records do not, however, provide knowledge of the processes or context of the early adoption of cereal cultivation, nor how it spread through northern Fennoscandia (for example regarding the role of immigration versus diffusion, see Johansen 1979; Zvelebil and Rowley-Conwy 1984; Armit and Finlayson 1992). These are key issues necessary for a full understanding of subsistence and economic strategies from the Neolithic to the Iron Age in northernmost Europe, which need to be addressed through an interdisciplinary approach, including palaeoecological, historical and archaeological methodologies (see Berglund 1985; Josefsson 2009; Sköld et al. 2010; Rick and Lockwood 2013). Integrating these methodologies can provide new sets of data that may result in a re-evaluation of our present view regarding where and when early cultivation in northernmost Fennoscandia was initiated. It is even possible that new archaeological finds of old small-scale cultivations in new environments may change our perspective on how to interpret stray finds of cereal pollen types, just as we have recently re-evaluated our view of the immigration of tree species (e.g. Picea and Larix) at the end of the last glaciation in northern Fennoscandia (cf. Kullman 2002; Segerström and von Stedingk 2003; Bergman et al. 2005; Paus 2010; Carcaillet et al. 2012).
We wish to thank our fellow colleagues within the research program “Recalling the past” for valuable discussions, and two anonymous reviewers for very useful and much appreciated comments on earlier versions of the manuscript. The English has been corrected by Sees-Editing, UK. This study was financially supported by the Bank of Sweden Tercentenary Fund (grant no. M11-0361:1) and the Centre for Environmental Research (CMF) in Umeå.