The advance of cultivation at its northern European limit: Process or event?

Abstract

Dates for early cultivation in Finland obtained from pollen analysis and remains from archaeological sites are compared with the changes in population size derived from the summed calendar-year probability distributions. The results from these two independent proxies correlate strongly with one another indicating that population size and the advance of farming were closely linked to each other. Moreover, the results show that the adaptation and development of farming in this area was a complex process comprising several stages and with major differences between regions The most intensive expansion having occurred in and after the Iron Age. It is therefore more accurate to describe the introduction of farming into the area as a long-lasting process, rather than an event.

Keywords

archaeology pollen analysis population dynamics prehistory of Finland radiocarbon dates spread of farming

Introduction

Discussions of prehistoric farming often focus on its earliest appearance. The date of the start of farming in Finland is the subject of debate. Recent discussions have argued for dates anywhere between 5000 BC to the final millennium BC (Alenius, 2011; Alenius et al., 2013; Bläuer et al., 2013; Lahtinen and Rowley-Conwy, 2013; Mökkönen, 2010). However, it is the question of not only when farming was first practised in Finland that is of interest but also the particular processes involved with its spread throughout the country. This aspect of the phenomenon has been neglected, which reveals an implicit assumption that farming was present throughout the area when it is first observed in one location, something only previously discussed by Vuorela and Hicks (1996) – see below. Understanding the process by which farming practices advanced could contribute to the more general discussion of the spread of farming in other marginal areas throughout the Northern Hemisphere. In this paper, we will discuss when farming spread into the different parts of Finland and link this to the changes in population size starting around the time of the Bronze Age (Tallavaara et al., 2010). While the focus of this paper is on the spread of farming, it also discusses agricultural intensification within the Eastern Fennoscandia.

In most areas, farming causes a significant increase in productivity which, in turn, causes population density to increase. In Europe and North America, this increase in population size has been studied using summed calendar-year probability distributions (spds) of radiocarbon dates. In Britain, France and Denmark, population size increased rapidly after the arrival of farming (Collard et al., 2010; Shennan et al., 2013; Timpson et al., 2014). Collard et al. (2010) also showed that the earliest cereal macrofossil dates correspond with an increase in the number of radiocarbon dates on other materials which are used as a proxy for population size in their study. In Finland, estimates of population size based on radiocarbon dates covering most of Finland have recently been published (Oinonen et al., 2010; Tallavaara et al., 2010). Furthermore, a link between radiocarbon date–based population levels and agriculture has been qualitatively proposed by Tallavaara et al. (2010).

In this paper, we focus on the quantitative pollen evidence from Finland as a proxy for early farming. We consider that single grains of cereal-type pollen do not constitute reliable indicators of the presence of cultivation, for the reasons discussed by Lahtinen and Rowley-Conwy (2013). In this study, we use the earliest continuous curve of cereal-type pollen. Chronological data on the earliest observations of continuous cereal-type pollen are converted into spds that contain information on the time and magnitude of the advance of cultivation within the area. We also created an artificial reference data set of equidistant radiocarbon dates and compared the spds generated from this with the cereal pollen spds, and also with archaeological radiocarbon dates. The ‘artificial’ spd was created in order to examine the possible effects of the radiocarbon calibration process and the ‘archaeological’ and ‘pollen’ spds to analyse the links between human activity and cereal cultivation.

The spread of farming in Eastern Fennoscandia

The advance of farming based on pollen evidence has been discussed previously in two review papers (Donner, 1984; Vuorela and Hicks, 1996). Donner (1984) studied pollen results from south Finland. His conclusions were first, that there is only sporadic evidence of farming in the form of single grains of cereal-type pollen during the Bronze Age (1500 BC–500 BC) that could indicate cultivation, but that most signs of cultivation in pollen analyses can be seen during the Iron Age (500 BC–AD 1200) and second, that the introduction of rye is visible in pollen cores between AD 450 and 1000. Vuorela and Hicks (1996) estimate the spread of farming also based solely on pollen analysis. They include every single cereal-type pollen grain as evidence and conclude that farming first started in the south-west of Finland approximately 4000 ago and spread very quickly across the country. It has already been pointed out (Lahtinen and Rowley-Conwy, 2013; Simola, 1999) that this synthesis is based on the uncritical use of pollen studies and thus is very likely flawed. Furthermore, it leaves out the important question of how farming spread throughout the country. This question is highly significant to the issue of whether farming was introduced by immigrants, adopted by locals, or through a combination of both (Galeta et al., 2011). It can be assumed that if there was significant immigration into Finland, this could mean a fast introduction of farming into the most favourable areas (such as Southwest Finland) and a significant population increase.

Direct evidence for cereal cultivation in the form of macrofossils has not been collected systematically, and thus it is impossible to estimate the advance of farming practices in the area based on macrofossils. No evidence of cereal macrofossils dated to before the last millennium BC has been well supported (Lahtinen and Rowley-Conwy, 2013). Although controversial, macrofossils from Niuskala (1891–1018 cal. BC; Vuorela and Lempiäinen, 1988) and Kitulansuo (1400–1048 cal. BC; Lavento, 1998) precede this (see Lahtinen and Rowley-Conwy, 2013). It is likely that more macrofossils will be discovered in the future which will increase our knowledge on farming practices, but currently the lack of them limits interpretation. Lack of macrofossils is also a problem in the Iron Age (500 BC to 13th-century AD) and Medieval (13th-century AD to 16th-century AD) periods in many areas of Finland because most archaeological excavations are carried out in the south, creating a geographic bias in data representation. Nor can the advance of farming be estimated from tools or archaeological sites because of a lack of evidence for farming-related activities before the Medieval period (Mikkola, 2009). This could be because of recovery bias as most excavated sites are cemeteries, single objects, or hoard finds (Mikkola, 2009). This makes it very difficult to use archaeology to document migrations or carry out research on the establishment of new settlements or small-scale variations in farming practices.

It has been suggested that farming did not spread to many areas of Finland until as late as the Medieval period (in Finland, the 13–16th-century AD) or even the Early Modern period (Enbuske, 2006; Keränen, 1984; Orrman, 1991, 2003; Soininen, 1957; Taavitsainen et al., 1998; Zvelebil, 2010). This large-scale advance of farming has been seen as an independent event in the Medieval period. However, this assumption is based solely on very limited historical material. Korpela (2012) strongly objects to the theory of migration into Eastern Finland during the late Medieval period, claiming that the idea is based solely on historical documents that are very likely biased. He argues that farmers become visible in remote areas only when the Swedish or Novgorod states and churches were established and began keeping records. Thus, this assumed migration into new areas might reflect the establishment of centralised authority and record keeping, not actually a new occupation. Korpela (2012) suggests that it is very likely that the inner part of Finland was not empty, nor was wilderness used in the way that Taavitsainen et al. (1998; Luukko, 1959; Taavitsainen, 1987, 2004) have suggested. Permanent occupation could have reached remote areas, but because of the lack of a central administrative system there are no records of this. They further suggest that pollen analysis has shown evidence of cultivation in these areas during the periods of the claimed migrations. Thus, this model of medieval advancement remains hypothetical.

Dating of pollen analysis

Establishing chronological timelines for events through high-resolution dating methods is a major objective in all fields of archaeology and palaeobotany. It is particularly important when case studies are compared with one another. Only then is it possible to determine whether different events are simultaneous and thus connected to each other. In pollen studies, the most common methods used are radiocarbon, varve, or palaeomagnetic dating. However, there is always a degree of uncertainty in the results, and they should always be critically examined and discussed in light of potential variability because of the differences in materials and methods.

Varve dating is based on the calculation of annual laminated sediment layers deposited in lakes. This can be studied only in very specific sedimentation environments where lamination is not disturbed during the formation of each varve. Typically, these environments are small lakes that are topographically closed or sheltered. Varve dating is the most precise dating method available for pollen cores, and if there is little disturbance in such an environment the pollen profiles are also often very reliable. However, this method can be problematic because of missing varves, issues with the identification of individual varve layers and underestimation of sediment disturbance (Ojala et al., 2012).

Radiocarbon dating from bulk sediment has long been known to result in dates that can be too old or recent in sediment age-depth models. In Finland, the difference between varve and bulk sediment radiocarbon dates has been as much as 1500 years (Alenius et al., 2011; Tolonen, 1978a, 1978b, 1980, 1981). Improvements in calibration have narrowed this difference in some studies (see Table 1), which suggest that the offset might be smaller than was previously estimated. Finnish bedrock is not rich in carbonates, so this difference is not likely because of the effect of hard water, but is more likely because of the recycling of old organic carbon in the lake catchment area. Similar effects have been recorded in Sweden (Stanton et al., 2010; Zillén et al., 2003). Based on the above, corrections towards true ages may be worthwhile for bulk sediment radiocarbon dates.

Table 1.

The calibration of late 1970s radiocarbon samples has been redone using Oxcal version 4.2 with calibration curve IntCal13 (Bronk-Ramsey, 2009; Reimer et al., 2013). The table shows the difference between the calibrated probability distribution and varve ages in Lake Ahvenainen (Tolonen, 1978a, 1978b) and Lake Lampelonjärvi (Tolonen, 1980). Ages are in BC(−)/AD(+) and – for ¹⁴C data – given as mean values and their standard deviations. Varve ages have been given as 1-year precision for Lake Ahvenainen (AHV) or as (age_MIN + age_MAX)/2, where ageMIN and ageMAX are the minimum and maximum ages provided by Tolonen (1980) for the varves of the Lake Lampelonjärvi (LAM). Uncertainties are given as ((age_MIN − age_MAX)/2.

Site	Code	Radiocarbon age (BP)	Calendar age (mean) (cal. yr BC/AD)	Varve age (cal. yr BC/AD)	Difference (cal. yr)
AHV	Su-690	800 ± 100	1105 to 1285 (1195)	1473	−278
AHV	Su-691	970 ± 100	960 to 1170 (1065)	1168	−103
AHV	Su-692	1450 ± 100	475 to 675 (575)	833	−258
AHV	Su-693	2010 ± 100	165 BC to AD 125 (−35)	546	−581
AHV	Su-722	2080 ± 100	250 BC to AD 10 (−120)	37	−157
AHV	Su-723	2300 ± 100	−550 to −230 (−390)	−304	−86
AHV	Su-724	2770 ± 100	−1075 to −845 (−960)	−915	−45
AHV	Su-695	3440 ± 100	−1885 to −1635 (−1760)	−1464	−296
AHV	Su-698	3290 ± 100	−1705 to −1465 (−1585)	−1634	49
AHV	Su-696	3550 ± 100	−2034 to −1765 (−1900)	−2106	206
AHV	Su-701	4040 ± 100	−2755 to −2445 (−2600)	−2987	387
AHV	Su-702	4450 ± 100	−3295 to −3005 (−3150)	−3630	480
AHV	Su-700	4820 ± 100	−3710 to −3470 (−3590)	−4200	610
LAM	Su-581	1450 ± 50	550 to 640 (595)	1515 ± 45	−920
LAM	Su-582	1840 ± 130	20 to 330 (175)	1170 ± 20	−994
LAM	Su-583	2310 ± 90	−560 to −250 (−405)	1070 ± 30	−1476
LAM	Su-576	1830 ± 130	30 to 340 (185)	580 ± 25	−395
LAM	Su-577	2100 ± 150	−340 BC to AD 40 (−145)	365 ± 25	−510
LAM	Su-578	2330 ± 70	−565 to −295 (−430)	180 ± 40	−610
LAM	Su-579	3580 ± 150	−2160 to −1750 (−1955)	−1880 ± 80	−75

Radiocarbon dating from bog peat samples has been shown to be more reliable because the organic material has grown in-situ and, unlike in lakes, there is less significant carbon recycling (Goslar et al., 2005). Studies show that pure sphagnum (peat moss) samples are very suitable for carbon dating and one study even suggests that there is no reservoir effect when only sphagnum leaves are used in dating samples (Blaauw et al., 2004).

Furthermore, the calibration process of radiocarbon dates may induce magnified peaks when summing the individual calendar-year probability distributions into one common distribution (Brown, 2015; Michczyńska et al., 2007; Williams, 2012), as is done in this study. This is largely because of the steep slopes of the calibration curve that concentrate the calendar-year probability into these periods of time. Regardless of these slopes, summarised distributions can still contain valid information, and the peaks are therefore not totally spurious. Furthermore, calibration curve plateaux cause an opposite effect by spreading the calendar-year probability distribution across broader periods of time. The location and magnitude of these effects can be assessed by comparing them to the spd of the series of equidistant, artificial radiocarbon dates throughout the study period. In addition, it has been suggested that for the Holocene these effects could be smoothed by applying a 500-year running average of the summed probability distributions (Brown, 2015). The two opposite effects of peaking and broadening thus to some extent cancel each other out.

Palaeomagnetic dating is based on measuring the palaeomagnetic secular variation (PSV) observed from several different proxies (Holcomb, 1986; Lund, 1996). This variation is seen in the sediment’s magnetic orientation (natural remanent magnetisation), and when compared with known changes in the Earth’s magnetic field, it can be used for dating sediment layers (Creer, 1982; Mackereth, 1971). In other words, it is a relative tool that is calibrated with other dating methods, such as varves or radiocarbon from other environments in close proximity. The advantage of this method is that it can be used to date lake sediments which would otherwise provide misleading radiocarbon dates because of carbon recycling, because it is not dependent on dating in a single sedimentation environment (Mackereth, 1971). Several studies suggest that it is highly precise (error 50–100 years) in lakes in Finland (Haltia-Hovi et al., 2010; Kotilainen et al., 2010).

In summary, each dating method has particular issues and different levels of uncertainty that cannot always be estimated, and it is important to understand that accuracy to within less than 100 years is simply impossible. The margin of error is likely to be significantly greater. In addition, the role of the radiocarbon calibration process should be considered. Despite this, we expect to observe summed and averaged calendar-year probability distributions displaying trends of cultivational development within the study area. Moreover, by combining several published studies, we try to minimise the possibility of interpreting or over-emphasising any one case study.

Materials and methods

The data cover published and unpublished institutional reports of pollen studies conducted in Finland and the surrounding areas. The total number of studies used in this work is 112. This data set provides the best current approximation for the spread of farming within the country. The data are presented in Table 2.

Table 2.

The radiocarbon dates from pollen analysis used in this study. The results of calibrated radiocarbon dates are given as 95.4% confidence ranges.

Number	Name	Beginning	End	Median	Dating method (if not radiocarbon)/source	Y (N)	X (E)	Reference
1	Aholampi	−1957	−1613	−1782	B	61.895	25.217	Koivula et al. (1994)
2	Ahvenainen	−354	230	−28	B	61.031	25.119	Tolonen (1978a)
3	Ahvenistonjärvi	−382	−222	−302	Varves	60.995	24.417	Grönlund and Simola (2009)
4	Alasenjärvi	1275	1632	1403	B	61.012	25.740	Vuorela (1978)
5	Antinlampi	1034	1215	1125	B	62.316	25.982	Vuorela et al. (1993)
6	Arminjärvi	428	880	653	B	61.024	24.345	Vuorela (1975)
7	Asilammi	−482	60	−229	B	61.563	25.148	Tolonen (1990)
8	Bonästräsk	667	890	772	B	60.073	23.356	Tolonen et al. (1976)
9	Gördorna	1297	1406	1348	P	60.176	22.683	Alenius (2008)
10	Hampträsk	654	968	788	B	60.291	25.256	Sarmaja-Korjonen (1992)
11	Hannusjärvi	775	969	881	B	60.090	24.411	Alenius (2011)
12	Hattulanselkä	127	326	205	Terrestrial macrofossil (P)	61.045	24.410	Grönlund and Simola (2009)
13	Haukkasuo	73	642	363	P	60.844	26.936	Tolonen and Ruuhijärvi (1976)
14	Heinälampi	1456	1556	1506	Varves	63.124	27.654	Grönlund et al. (1992)
15	Iidesjärvi	−749	53	−274	B	61.484	23.812	Alhonen (1981)
16	Isorahka	679	952	813	P	60.900	21.633	Tolonen et al. (1976)
17	Isoskärret	−804	−232	−556	P	60.155	22.750	Asplund and Vuorela (1989)
18	Jaatilanjärvi	237	562	399	B	61.879	25.186	Vuorela (1994)
19	Joutjärvi	777	1207	1010	B	60.977	25.700	Vuorela (1978)
20	Kaakotinlampi	536	1024	769	B	61.414	25.868	Vuorela (1981)
21	Kaartinlammensuo	987	1276	1128	P	60.741	24.195	Vuorela (1993)
22	Kangerjoki	1471	0	1673	P	66.119	28.99	Hicks (1976)
23	Kantala	172	428	330	P	61.164	24.042	Tolonen (1978b)
24	Karvalampi	1219	1393	1286	B	63.338	25.625	Taavitsainen et al. (2007)
25	Katajajarvi	−1148	−751	−924	B	61.168	26.846	Alenius et al. (2009)
26	Katajärvi 2	1047	1268	1195	B	61.168	26.846	Alenius et al. (2009)
27	Katamossen	711	1025	898	P	63.183	22.300	Wallin and Segeström (1994)
28	Katinhännänsuo	85	646	385	B	60.390	24.467	Vuorela (1972, 1975)
29	Kattilanlahti	1238	1322	1280	Varves	61.400	27.17	Simola et al. (1988)
30	Ketohaka	−171	527	167	B	60.420	23.140	Tolonen (1985)
31	Kirjavanlampi	393	836	613	M, B	61.695	30.76	Alenius et al. (2004)
32	Kirkkojärvi	641	1181	883	B	60.683	21.646	Vuorela (1975)
33	Kirkkolampi	901	1116	1005	B	61.785	30.001	Alenius and Laakso (2006)
34	Kirkkosaari	1446	1798	1567	P	60.857	24.491	Vuorela (1993)
35	Kirvesjärvi	1850	1870	1860	Varves	62.200	26.57	Simola et al. (1988)
36	KissalammiA	−752	−252	−502	Varves	61.256	24.355	Tolonen (1981)
37	KissalammiB	−1812	−1192	−1502	Varves	61.256	24.355	Tolonen (1981)
38	Kitulansuo	687	992	853	P	61.506	27.365	Saastamoinen (1999)
39	Konnunsuo	1219	1450	1346	P	61.039	28.460	Tolonen and Ruuhijärvi (1976)
40	Könttärinlahti	644	1039	844	B	62.284	24.709	Vuorela (1994)
41	Kuittijärvi	100	300	200	Varves	65.192	30.321	Alenius et al. (2011)
42	Kuivajärvi	−915	−795	−842	B	60.782	23.858	Vuorela and Kankainen (1992)
43	Kunnonniemensuo	780	1279	1088	P	62.100	30.21	Tolonen (1984)
44	Kynnarträsken	901	1036	1002	B	60.095	24.166	Alenius (2011)
45	Labböleträsk	428	990	723	B	60.090	22.735	Alenius (2008)
46	Lappträsket	892	1152	1004	B	60.048	23.656	Tolonen et al. (1976)
47	Lehijärvi	81	681	434	B	61.052	24.3003	Vuorela (1975)
48	Lemusuo	1029	1394	1223	P	60.198	23.223	Vuorela (1985)
49	Levisträsk	260	529	388	B	59.987	23.266	Alenius (2011)
50	Likolampi	1220	1480	1350	Varves	60.894	27.597	Tomminen (2006)
51	Linnajärvi	−381	73	−136	B	61.632	25.189	Tolonen (1990)
52	Linnasuo	427	603	521	P	61.633	25.196	Tolonen (1990)
53	Lintunemossen	994	1275	1135	B	63.124	22.184	Tolonen et al. (1976)
54	Loimaansuo	−2035	−1416	−1714	B	61.136	22.673	Vuorela (1975)
55	Lojärvi	637	1148	845	B	60.209	24.510	Tolonen et al. (1976)
56	Lovojävi	−1049	−176	−618	B	61.077	25.033	Huttunen and Tolonen (1977)
57	Löytysenlampi	−91	127	19	B	60.975	28.155	Tomminen (2006)
58	Majenemossen	1045	1415	1267	P	63.133	22.3	Miettinen and Vuorela (1988)
59	Mäyrälampi, Keuruu	379	795	589	B	62.353	24.618	Koivula et al. (1994)
60	Mäyrälampi, Hankasalmi	−378	−2	−184	B	62.342	26.234	Koivula et al. (1994)
61	Mossen	1216	1625	1363	P	60.165	22.695	Vuorela (1990)
62	Mustikkalampi	1046	1153	1100	Varves	61.430	28.2	Simola et al. (1988)
63	Myllypuro	−748	−198	−396	P	61.556	21.979	Aalto et al. (1980)
64	Nälköönsuo	720	1297	1073	P	60.297	24.196	Tolonen and Ruuhijärvi (1976)
65	Nautajärvi	1394	1454	1424	Varves	61.805	24.679	Ojala and Alenius (2005)
66	Nerderskogen	1280	1432	1352		63.267	22.283	Wallin and Segeström (1994)
67	Niemispääbog	716	1260	1008		61.900	22.633	Tolonen et al. (1976)
68	Niikkalanlampi	1185	1290	1250	B	61.371	30.907	Vuorela et al. (2001b)
69	Orijärvi	730	930	830	Palaeomagnetic	61.400	27.14	Alenius et al. (2008)
70	Pakarinlampi	1363	1437	1400	Varves	62.428	28.644	Huttunen and Simola (1986)
71	Pärkönsuo	664	1119	881	Varves	60.850	21.662	Tolonen et al. (1976)
72	Parusuo	214	682	480	P	61.042	21.617	Vuorela (1991)
73	Pegrema	1326	1649	1500	B	62.367	34.678	Vuorela et al. (2001b)
74	Petarträsk	256	530	381	B	59.967	23.900	Alenius (2011)
75	Pieni summanen	385	770	575	B	62.674	25.327	Taavitsainen et al. (2007)
76	Pienikuuppalampi	5	400	201	B	61.170	29.55	Miettinen et al. (2002)
77	Piilosuo	143	605	402	P	60.783	24.650	Tolonen and Ruuhijärvi (1976)
78	Pitkälampi	1420	1480	1450	Varves	62.256	30.462	Grönlund and Asikainen (1992)
79	Pohjanlampi	55	401	224	B	62.221	25.780	Taavitsainen et al. (2007)
80	Pytärälampi	1278	1362	1320	Varves	62.299	28.591	Huttunen and Simola (1986)
81	Puutienlampi	563	637	600	Varves	62.071	28.899	Simola et al. (1985)
82	Rukatunturi	1026	1392	1214	P	66.166	29.1507	Hicks (1976)
83	Ryönänsuo	1528	0	1798	P	60.438	24.177	Vuorela (1993)
84	Säkinlampi	430	659	588	B	62.475	26.497	Taavitsainen et al. (2007)
85	Santamäensuo	125	680	447	B	60.420	23.150	Tolonen (1985)
86	Siikasuo	1216	1634	1372	P	61.300	22.065	Vuorela (1991)
87	Sistuslampi	782	1225	993	Modelling	62.704	25.222	Koivula et al. (1994)
88	Skiitanlahti	425	665	577	B	61.391	30.918	Vuorela et al. (2001b)
89	Söderbyträsket	1022	1155	1096	B	60.055	22.432	Alenius (2008)
90	Sotkulampi	1520	1580	1550	Varves	61.470	27.51	Simola et al. (1988)
91	Storträsk	625	764	663	B	60.093	23.236	Alenius (2011)
92	Suurijärvi, Kerimäki	1151	1248	1200	Varves	61.828	29.060	Simola et al. (1985)
93	Suurijärvi	580	740	660	Varves	61.834	29.057	Grönlund (1991)
94	Syrjälä Mire	356	552	466	P	61.213	28.103	Vuorela (1995)
95	Syrjälänsuo	346	559	465	P	61.172	28.052	Vuorela and Kankainen (1993)
96	Taruslampi	422	950	677	B	61.486	25.625	Vuorela (1982)
97	Tervalampi	520	680	600	Varves	61.683	29.389	Simola et al. (1985)
98	Tjärnen	900	1030	997	B	60.047	23.225	Alenius (2011)
99	Träskmyre	416	655	551	P	63.15	21.583	Wallin and Segeström (1994)
100	Tullerinsuo	233	623	433	P	61.339	21.948	Vuorela (1991)
101	Työtjärvi	−769	−204	−469	B	60.996	25.464	Donner et al. (1978)
102	Työtjärvi	−769	−204	−469	B	60.996	25.464	Donner et al. (1978)
103	Vähä-Pitkusta	1301	1631	1434	Terrestrial macrofossil (P)	60.488	23.649	Hakala et al. (2004)
104	Väiskänsuo	−767	−97	−383	Modelling	60.917	21.700	Tolonen et al. (1976)
105	Vasikkasuo	1668	1947	1820	P	64.678	27.857	Vuorela and Kankainen (1991)
106	Vikperä	975	1245	1099		63.233	22.217	Wallin and Segeström (1994)
107	Vitsjön	1020	1445	1265	B	59.963	23.315	Tolonen and Tolonen (1988)
108	Vitsjön bog	1267	1617	1379	P	59.965	23.310	Tolonen and Tolonen (1988)
109	Vohtenkellarinsuo	94	649	399	P	60.409	22.678	Vuorela (1983)
110	Vuojärvi	670	1119	892	B	62.414	25.935	Vuorela et al. (1993)
111	Vuorijärvi	1576	1624	1600	Varves	62.170	27.400	Simola et al. (1988)
112	Ylimysneva	1485	0	1725	P	62.142	22.87	Huttunen (1990)

B: bulk sediment; p: peat; M: modelled.

The data consist of radiocarbon dates of pollen studies where a continuous cereal-type pollen curve has been observed. The date of the layer where the first indication of cereal-type pollen was visible is entered into the database. In several pollen records, cereal-type pollen shows a hiatus. In these cases, both available radiocarbon dates are included. If the layer with the early signal was not dated, it is interpolated with the Bayesian data analysis tool with Oxcal (version 4.2; Bronk-Ramsey, 2009). In summary, sites were excluded if (1) the data needed extrapolation (the layer was not dated itself, and it did not lie in between radiocarbon dated layers), (2) the error margins could not be estimated and (3) the dating was considered considerably too old and erroneous by the original authors. None of the sites was excluded on the basis that the event itself did not fit in with the wider picture of the beginning of the cultivation in Finland.

Dates were calibrated with Oxcal (version 4.2, Bronk-Ramsey, 2009) using calibration curve Intcal13 (Reimer et al., 2013). Calibration was performed even if a calibrated date was already published, in order to make the data set more comparable and accurate, especially in the case of older studies. The resulting individual calendar-year probability distributions (spd) were then summed to a single distribution. Normally distributed calendar-year probability distribution of varve and palaeomagnetic dates was also summed to yield summed spds for each dating method used for comparison. Most of the varve studies used did not provide an error estimation. However, in Finland, errors are typically 1–3% (Ojala et al., 2012). We adapted a 3% margin of error for all varve dates to reflect this.

The resulting total sum distribution was compared with the spd of the archaeological radiocarbon data from southern Finland discussed by Tallavaara et al. (2010) and Oinonen et al. (2010). This contains 465 radiocarbon dates ranging from 9480 to 800 BP. As described by Oinonen et al. (2010), many of these dates have been published in the Radiocarbon Dates List – documents (I–VI), and recently as part of an online database (Junno et al., 2016). This data set has been complemented by unpublished dates released to us by archaeologists and/or their institutes. Unfortunately, permissions to use these latter dates cover only the sum distributions, thus preventing us from publishing individual dates.

To investigate the possible effects of the radiocarbon calibration process, we established a set of artificial and equidistant radiocarbon dates through the study period of 3000 BC–AD 1950 (that is approximately 4300–0 BP) with 40 radiocarbon years between them and with ±40 radiocarbon-year standard deviations. This totals 124 artificial and equidistant dates to be compared with the full data set of 112. Because of the assumed recycling of old carbon in bulk dates, we have adopted a factor to correct these based on the recalibrated radiocarbon dates and varve ages from Lake Ahvenainen (Tolonen, 1978b) and Lake Lampelonjärvi (Tolonen, 1980). This average factor is ΔT = 250 years (see Table 1), meaning the bulk ages are, on average, 250 years too old compared with the varve. The large standard deviation, averaging 510 years, indicates that an even larger correction factor might be necessary.

The spds of artificial, archaeological and pollen data sets in the range 2500 BC–AD 1200 have been explored using principal component analysis (PCA) to identify independent factors responsible for the distributions. The resulting spds were also smoothed by 500-year running averages to suppress the calibration effects in the radiocarbon data. As a comparison, the smoothing was also performed on the results of Tallavaara et al. (2010). To discuss the results with consistency, we have defined threshold values T₁₀ and T_max which are times when the calendar-year probability reaches 10% of a peak maximum and maximum, respectively. These are used to provide quantitative estimates for the advance of cultivation.

Results

As already mentioned, the accuracy of each dating method varies, possibly introducing systematic biases into distributions. In the following, we assume that the varve and palaeomagnetic data from the pollen analysis provide the most accurate (not necessarily most precise) distributions free of systematic effects. The averaged pollen analysis spds of the different dating methods were compared with the averaged spds of this varve and palaeomagnetic data. The 250-year shift of the bulk sediment spd, observed in recalibrated dates from (Tolonen, 1978b, 1980), improves the correlation between the distributions drastically from r = 0.36 (bulk vs varve and palaeomagnetic spds) to r = 0.68 (shifted bulk vs varve and palaeomagnetic spds). This adds confidence to our correction procedure. It is noteworthy that the correlation coefficient (shifted bulk vs varve and palaeomagnetic spds) increases with larger shifts, until the shift reaches ΔT = 700 years, indicating that the correction could be even larger. Even though we do not have solid evidence for applying a similar correction procedure for peat dates, the same test was carried out between averaged peat spd and varve and palaeomagnetic spd. In this case, the best correlation was obtained when peat spd was not shifted at all (r = 0.54). We interpret this as an indication that the peat dates do not require correction. The total spd summing all the pollen data was made based on these considerations.

The cumulative geographical distribution of the mean radiocarbon dates of the pollen data for the different periods is presented in Figure 1. The sum distributions of artificial, archaeological and pollen data are presented in Figure 2 together with breakdowns of the latter into different dating methods. The calibration process itself introduces a few magnified peaks into the summed probability distribution (Figure 2a) at around 800 BC, 400 BC, AD 600–700, AD 1000 and AD 1300–1400 (a double peak), and this should be taken into account when interpreting the results. All the calibration-induced peaks can possibly be identified from the bulk sediment spd and also from the archaeological data. Particularly, the effect is pronounced at 800 BC, AD 600–700, AD 1000 and AD 1300–1400.

Figure 1.

Map of the cumulative distribution of median ages of radiocarbon dates obtained from pollen cores (Table 2).

Figure 2.

Summed calendar-year probability distributions assessed in the work. All the distributions are given with 5-year resolution: (a) artificial equidistant dates, (b) archaeological dates based on Oinonen et al. (2010) and Tallavaara et al. (2010), (c) bulk sediment dates from the pollen data, (d) peat dates from the pollen data, (e) varve and palaeomagnetic dates from the pollen data, (f) total sum distribution of (c)–(e) and (g) total sum distribution of (c)–(e) with bulk dates corrected by +250 years. Three intensification periods discussed in this paper are marked with Roman numbers.

PCA of the three – total corrected for pollen – spds reveals the following (see Figures 3a and b and 4a and d):

There is a very strong correlation between archaeological and pollen data with r = 0.87.

There is a weak correlation between the artificial and archaeological or pollen data with r = 0.34–0.35.

The first principal component (PC1) explains 71% of the total variance of the data and the shape of the PC1 scores resemble the archaeological or pollen spds.

The second principal component (PC2) explains 25% of the variance and the shape of the PC2 scores resemble that of the artificial spd.

Figure 3.

Results of the principal component analysis: (a) biplot of principal component scores (grey) of PC1 and PC2, and eigenvectors of the original data sets. Sharp peaks of the spds are displayed in the top-right corner of the PC space with calendar years shown for peak identification. (b) PC1 and PC2 scores as a function of calendar years. Percentages of variability of the data the component explains are also provided.

Figure 4.

Averaged calendar-year probability distributions (500-year running average, Williams, 2012) and similarly averaged PC1 scores: (a) artificial equidistant dates, (b) archaeological dates based on Oinonen et al. (2010) and Tallavaara et al. (2010), (c) total sum distribution of pollen data with bulk dates corrected by +250 years and (d) PC1 scores. Note that (a), (b) and (d) are restricted to AD 1200 because of time scale of (b).

The peaks found are suppressed by using 500-year running average (Figure 4a–c). For comparison, similarly averaged PC1 scores are also plotted in Figure 4d. The PC1 score plot shows the same major features as the averaged pollen spd. Quantitative correlation analysis between the distributions (range 2500 BC–AD 1200) confirms the visual observations: both non-averaged (r = 0.87) and averaged (r = 0.98) archaeological and pollen spds correlate extremely strongly.

The corrected total pollen spd shows three different phases of agricultural advance. The first modest advance is during the second millennium BC (T_{10_1} ~ 1900 BC, T_{max_1} ~ 1550 BC), coinciding with the Kiukainen culture and the Early Bronze Age. A minimum is reached at around 1000 BC, after which the second advance can be seen in the first millennium BC, during the Late Bronze Age (T_{10_2} ~ 950 BC, T_{max_2} ~ 500 BC). The third broad phase of advance is in the first millennium AD during the Iron Age and lasts until the Middle Ages in Eastern Fennoscandia (T_{10_3} ~ AD 100, T_{max_3} ~ AD 1300). This last maximum is also clearly seen in all the subsets of non-averaged data (see Figure 2c–e): in the old organic carbon-corrected bulk spd at ~AD 1250, in the peat spd at ~AD 1350 and in the varve and palaeomagnetic spd at ~AD 1400.

Discussion

The quantity of radiocarbon dates can allow the estimated reconstruction of the changes in population sizes (Timpson et al., 2014). Within the study area, the population history has been assessed by recording radiocarbon dates obtained from Finnish archaeological finds (Oinonen et al., 2010; Tallavaara and Seppä, 2012; Tallavaara et al., 2010). These population size estimates are based on radiocarbon dates from archaeological materials and are thus totally independent of our pollen-based analysis or radiocarbon dates obtained from the pollen cores. This method has been criticised because of the underlying assumption that population size and productivity of consumables correlate (Contreras and Meadows, 2014); also, radiocarbon calibration-induced effects can bias interpretations of spds if not taken into account (Brown, 2015).

However, regardless of whether the change in calendar-year probability distributions of radiocarbon dates is caused by population increase (i.e. increase in productivity) or because of a cultural change resulting in greater archaeological visibility, this distribution correlates extremely strongly (R = 0.98) with the spd of the first continuous appearance of cereal-type pollen. The strong correlation between the observed cereal-type pollen and archaeological spds and, furthermore, the observed strong PC1 displaying a similar signal to the above spds suggests that there is a common factor behind the two distributions. It seems that the PC2 can be linked to calibration effects: the similarity of the PC2 score data to the artificial spd of equidistant dates is striking. PC1 must therefore be related to something else. We suggest that this component represents the human influence behind the distributions. Furthermore, it is equally clear that the earliest strong signals of cereal pollen found in the sediments are because of humans cultivating these lands. Therefore, our suggestion that human influence lies behind the PC1 is plausible. More confidence concerning this connection comes from the similarity of pollen spd and PC1 scores (Figure 4): all the essential features in the two plots are the same. Although we tentatively suggest that PC1 records human influence, in the following we base our conservative discussions mainly on the 500-year averaged total spd instead of the similar PC1 scores.

We have thus gathered together the existing chronological evidence relevant to the advance of cultivation in Finland and have produced a more comprehensive pattern than has been hitherto observed. We now discuss the pattern in chronological order relative to the other available evidence.

Our data selection procedure does not allow the inclusion of the suspected find of buckwheat pollen (Alenius et al., 2013) dated to before 5000 BC as continuous pollen evidence for this species has not been discovered from this period. There is therefore no calendar-year probability allotted to this very early period of time. We remain interested in this find and hope that future studies will clarify its status.

As mentioned, farming is not the only possible explanation for an increase in population size. However, farming can be one of the causes, because it can increase the quantity of food extracted from an area. As discussed by Tallavaara et al. (2010), the first large peak in the quantity of spd of archaeological radiocarbon dates around 4000 BC, but none of the pollen studies supports such a large-scale increase in cultivation at this time. Only a few single cereal-type pollen grains are known from this early period. For the reasons discussed above and by Lahtinen and Rowley-Conwy (2013), we consider such single grains to be unreliable indicators of cultivation in the absence of other line of evidence. It is notable that later peaks in archaeological radiocarbon dates are matched by increases in the frequency of cereal-type pollen. It is therefore most unlikely that the peak in archaeological radiocarbon dates around 4000 BC indicates the introduction of farming practices. This population increase coincides with several proxies indicating a broad climatic optimum: it has been mentioned that this led to a natural increase in productivity leading to a higher carrying capacity for hunter-gatherer populations (Tallavaara and Seppä, 2012). In eastern Finland, it may have been locally supported by landscape changes after the Vuoksi breakthrough occurred (Oinonen et al., 2014). The Vuoksi river was created because the post-glacial isostatic rebound made the previous outflow of the lake Saimaa physically impossible. The existence of this new channel decreased the water table of the lake Saimaa and reviled approximately 5000 km² new fertile land at around 4000 BC (Oinonen et al., 2014).

The earliest evidence for farming has been discovered through lipid residue analysis. This suggests that the Corded Ware (CW) culture practised animal husbandry (Cramp et al., 2014). Animal husbandry does not necessarily involve cultivation, which explains the absence of indications of cultivation from our pollen data, and why it is not visible in our population size estimate either. Moreover, there is a lack of any archaeological evidence for cultivation by CW people (Nordqvist and Häkälä, 2014), although Siiriäinen (1980) argued that their lack of hunting-related tools could indicate a subsistence based on something other than hunting. The first, modest indication for cultivation in the pollen spd dates to Kiukainen Culture (mean approximately 1500 BC). Signs of cultivation from this period are seen in three pollen studies, two with bulk sediment dates – Aholammi (1957–1613 BC, Koivula et al., 1994) and Loimaansuo (2035–1716 BC, Vuorela, 1975) – and one with a varve date – Kissalammi (c. 1500 BC, Tolonen, 1981). Even taking into account our most moderate temporal shift (ΔT = 250 years) in bulk sediment dates, these dates form the first increase in calendar-year probability within our plots, covering the period of the Kiukainen Culture to the Bronze Age. Preliminary results have also suggested that starch grains of barley have been detected on a pottery sherd of Kiukainen although the sherd itself dates from the Bronze Age (Juhola et al., 2014). A radiocarbon date obtained from a burnt sheep/goat bone also dates to this period (2200–1950 BC; Bläuer and Kantanen, 2013). Animal husbandry is also evidenced in the Bronze Age by a cow maxillary molar from Nakkila, south-west Finland (1427–1271 cal. BC, Bläuer et al., 2013). Thus, some independent evidence – although still very scarce – is starting to appear for both cultivation and animal husbandry from the Kiukainen Culture and the Bronze Age – in accordance with our spd plots. There are still somewhat speculative, but reasonable, challenges to this to be considered. If the temporal shift (ΔT) of the bulk sediment dates is larger than 250 years (as our correlation analyses may indicate), two of the three available dates would fall into the Bronze Age, thus making the early intensification a later development. Furthermore, as already discussed by Lahtinen and Rowley-Conwy (2013), the pollen of wild and cultivated species may be misidentified. Juhola et al. (2014) have already suggested that future investigations on the reliability of the starch identification method are also needed before these results can be unanimously accepted. Therefore, the intriguing question of the early development of cultivation needs much future work.

Cultivation seems to advance again during the Late Bronze Age, from c. 900 BC onwards, and the averaged signal shows a maximum at around 500 BC. The peak in the raw data concentrates within more the narrow period of 700–200 BC. Evidence of milk lipids has been discovered from Bronze Age vessels (undated; Cramp et al., 2014), and direct dates have been obtained from two barley grains from southern Ostrobothnia (1000–844 BC; 831–552 BC; Holmblad, 2010), thus agreeing with our observation. Evidence for cultivation in this early period also remains limited, but this could all be indicative of minor Late Bronze Age farming practised on a local scale, as seen at similar latitudes in northern Sweden (Viklund, 2011). This second advance coincides with the strong increase in the archaeological finds illustrated by Tallavaara et al. (2010) (Figure 2). In addition to the overall significant correlation between the spds, this agreement adds more confidence to the suggestion that cultivation contributed to the assumed population increase.

The third advance of cultivation started from c. AD 100 onwards, during Iron Age. This is mainly dominated by radiocarbon dates derived from peat samples and is supported by bulk sediment, varve and palaeomagnetic dates after AD 300. A local maximum is reached at around AD 700, after which an ultimate maximum is eventually reached at AD 1300 (T_{max_3}), during the Medieval period. The peak at approximately AD 500–700 followed by a decline is observed at the non-averaged data (Figure 2g) and in the PC1 scores (Figure 3b). Although it was not possible in this study to estimate which varieties or species of cereals produced these signals of cultivation, we suspect that the distribution may be connected to the introduction of rye. During the Iron Age, a change in macrofossil evidence in southern Sweden indicates an intensification of rye cultivation (Grabowski, 2011). As already mentioned, Donner (1984) also claims that rye spread to Finland between the 5th and 11th centuries AD. This could partly explain the rise in the quantity of pollen studies indicating farming during the first millennium AD. This rise in frequency is not necessarily an indication just of the advance of farming, as Donner (1984) has already argued: rye produces more pollen than barley and therefore has greater visibility in the pollen record. However, because of the simultaneous increase in population size, it would be reasonable to suggest that farming may have spread into new areas. The start of rye cultivation could partially explain the population increase because rye is more difficult to cultivate than barley for climatic reasons, but it provides larger harvests in the northern latitudes (Solantie, 1988). Another possibility is that this population increase was connected to a growth in trade evidenced by an increase in imported objects from AD 700 onwards (Huurre, 1979; Uino, 1999). Increased trade could have provided needed food supplements during harvest failures, therefore reducing pressure upon farmers.

As mentioned, a decrease in non-averaged pollen spd and PC1 scores occurs at c. AD 700. This is visible in varve, bulk and peat dated cores (Figure 2). It is in agreement with the observation of a possible genetic bottleneck (Sundell et al., 2010). This would suggest a decline in the occupation of new areas by farming communities. This same decline in the number of farms at c. AD 600 is recorded in Upland in southern Sweden (Lagerås, 2013) and Estonia (Tvauri, 2014), possibly because of the climatic anomaly at AD 536 (Gräslund, 2007; Tvauri, 2014). Furthermore, a tree ring summer temperature reconstruction from northern Sweden suggests that summers were cooler between AD 600 and 700 (Briffa et al., 1992). This could indicate a difficult time for farming-based economies in northern latitudes. Another possibility is that the northern border of farming was reached, and instead of occupying new areas, farming was just intensified within the areas it already occupied, which would not be visible in this study. However, this seems a less likely scenario because the same phenomenon has been reported in neighbouring countries.

The archaeological spd does not include radiocarbon data from the historical period (c. AD 1200 onwards). Thus, comparison of pollen spd and archaeological spd is not possible for the historical period. Moreover, in light of pollen analyses, it seems likely that the occupation of new areas was a long-lasting practice among farmers, which continued even until the 16th-century AD in Finland. Furthermore, these results suggest that there was no separate Medieval spread of farming, but rather that the third advance of farming, which started during the Iron Age, simply continued during this period. After the most intensive stage of this expansion up to the 16th-century AD, historical sources suggest that Finns continued to spread, but to other parts of the Swedish state, into Sweden’s forested uplands, into Lapland, and to the new world in North America.

The geographical distribution of radiocarbon dates from the pollen studies is shown in Figure 1. The maps show no clear pattern of advance, but rather intensification taking place in various areas simultaneously. This is in agreement with Mannermaa and Deckwirth (2010), who have suggested, based on osteological evidence, that even during the Iron Age subsistence was based on both farming and hunter-gathering. Thus, it is likely that the advance of farming took place in widely separated areas, leaving much ‘non-farmed’ land in between, in which other subsistence activities could have been practised. This could also indicate high mobility among people, the adoption of farming from neighbouring areas by the local population, or both.

The rate at which Neolithic farming spread across Europe as a whole has been estimated at no more than about 1 km a year, although regional variations are significant (Bocquet-Appel et al., 2012; Galeta et al., 2011; Pinhasi et al., 2005). For example, the LBK culture spread over Central Europe at a rate of approximately 6 km/yr (Dolukhanov et al., 2005). This would mean that even with a much slower rate, Finland (1200 km in length) could have been settled by farmers more quickly than we see in the data. It is not possible to draw conclusions based on migrations, or at least not using such studies alone, but this study suggests a very slow expansion process over several thousand of years. This was therefore different to the Central European neolithisation. Moreover, this process was not an even one: our study suggests at least three separate periods of advance within it.

Summary

We have collected the existing chronological information on the advance of cultivation at its northern limits in Finland and compared the resulting spds to the available archaeological data. Through applying PCA to spds, it has been possible to identify calibration effects and to explore the underlying factors within the distributions.

Two independent proxies, the spds of the archaeological and the pollen dates, correlate extremely strongly (R = 0.98). We suggest that this is because of a common underlying cause: the advance of cultivation. This correlation leads us to argue that the anomalous appearance of a few single cereal-type pollen grains at around 4000 BC is not because of the appearance of cultivation at this time.

The advance of farming involved several phases, of which three major ones (1500 BC, 900–500 BC and AD 100 onward) were observed. Consequently, the advancement of farming should be seen as a process rather than an event.

Finally, our summed calendar-year pollen studies suggest a possible decline in the spread of farming at c. AD 700. This would coincide with the possible genetic bottleneck suggested by Sundell et al. (2010), and with similar phenomena observed in neighbouring countries.

Footnotes

Acknowledgements

The authors would like to thank the two anonymous reviewers for the constructive comments which improved this article.

Funding

This study was funded by the Jenny and Antti Wihuri foundation.

References

Aalto

Taavitsainen

J-P

Vuorela

(1980) Paleobotanical investigation at the site of a Sledge Runner Find, Dated to about 4900 B.P., in Noormarkku, SW Finland. Suomen Museo 87: 41–65.

Alenius

(2008) Appendix 2: The palaeoecological study of three mires on the island Kemiönsaari, SW Finland. In: Asplund

(ed.) Kymittæ: Sites, Centrality and Long-Term Settlement Change in the Kemiönsaari Region in SW Finland. pp. 565–584. Available at: http://www.doria.fi/bitstream/handle/10024/42058/B312.pdf?sequence=1.

Alenius

(2011) From forest to a farmland: Palaeoenvironmental reconstruction of the colonization of Western Uusimaa. Iskos 19: 87–116.

Alenius

Laakso

(2006) Palaeoecology and archaeology of the village of Uukuniemi, Eastern Finland. Acta Borealia 23(2): 145–165.

Alenius

Mikkola

Ojala

(2008) History of agriculture in Mikkeli Orijärvi, eastern Finland as reflected by palynological and archaeological data. Vegetation History and Archaeobotany 17: 171–183.

Alenius

Mökkönen

Lahelma

(2013) Early farming in the Northern Boreal Zone: Reassessing the history of land use in Southeastern Finland through high-resolution pollen analysis. Geoarchaeology 28: 1–24.

Alenius

Grönlund

Simola

. (2004) Land-use history of Riekkalansaari Island in the northern archipelago of Lake Ladoga, Karelian Republic, Russia. Vegetation History and Archaeobotany 13(1): 23–31.

Alenius

Lavento

Saarnisto

(2009) Pollen-analytical results from Lake Katajajarvi: aspects of the history of settlement in the Finnish inland regions. Acta Borealia 26(2): 136–155.

Alenius

Saarnisto

Taavitsainen

. (2011) Records of land use and medieval settlement history in Vuonninen, Dvina Karelia, Northwest Russia. Geoarchaeology 26(1): 142–163.

10.

Alhonen

(1981) Stratigraphical studies on lake Iidesjärvi sediments part I: Environmental changes and paleolimnological development. Bulletin of the Geological Society of Finland 52–53: 97–107.

11.

Asplund

Vuorela

(1989) Settlement studies in Kemiö: Archaeological problems and palynological evidence. Fennoscandia Archaeologica VI: 67–79.

12.

Blaauw

van der Plicht

van Geel

(2004) Radiocarbon dating of bulk peat samples from raised bogs: Non-existence of a previously reported ‘reservoir effect’? Quaternary Science Reviews 23(14–15): 1537–1542.

13.

Bläuer

Kantanen

(2013) Transition from hunting to animal husbandry in Southern, Western and Eastern Finland: New dated osteological evidence. Journal of Archaeological Science 40(4): 1646–1666.

14.

Bläuer

Korkeakoski-Väisänen

Arppe

. (2013) Bronze Age cattle teeth and cremations from a monumental burial cairn in Selkäkangas, Finland: New radiocarbon dates and isotopic analysis. Estonian Journal of Archaeology 17: 3–23.

15.

Bocquet-Appel

Naji

Vander

. (2012) Understanding the rates of expansion of the farming system in Europe. Journal of Archaeological Science 39: 531–546.

16.

Briffa

Jones

Bartholin

(1992) Fennoscandian summers from AD 500: Temperature changes on short and long timescales. Climate Dynamics 7: 111–119.

17.

Bronk-Ramsey

(2009) Bayesian analysis of radiocarbon dates. Radiocarbon 51(1): 337–360.

18.

Brown

(2015) Through a filter, darkly: Population size estimation, systematic error, and random error in radiocarbon-supported demographic temporal frequency analyses. Journal of Archaeological Science 53: 133–147.

19.

Collard

Edinborough

Shennan

. (2010) Radiocarbon evidence indicates that migrants introduced farming to Britain. Journal of Archaeological Science 37: 866–870.

20.

Contreras

Meadows

(2014) Summed radiocarbon calibrations as a population proxy: A critical evaluation using a realistic simulation approach. Journal of Archaeological Science 52: 591–608.

21.

Cramp

LJE

Evershed

Lavento

. (2014) Neolithic dairy farming at the extreme of agriculture in northern Europe. Proceedings of the Royal Society B: Biological Sciences B 281: 20140819.

22.

Creer

(1982) Lake sediments as recorders of geomagnetic field variations – Applications to dating post-Glacial sediments. Hydrobiologia 92: 587–596.

23.

Dolukhanov

Shukurov

Gronenborn

. (2005) The chronology of Neolithic dispersal in Central and Eastern Europe. Journal of Archaeological Science 32(10): 1441–1458.

24.

Donner

(1984) Some comments on the pollen-analytical records of cereals and their dating in Southern Finland. Fennoscandia archaeologica I: 13–17.

25.

Donner

Alhonen

Eronen

. (1978) Biostratigraphy and radiocarbon dating of the Holocene lake sediments of Työtjärvi and the peats in the adjoining bog Varrassuo west of Lahti in southern. Annales Botanici Fennici 15: 258–280.

26.

Enbuske

(2006) Erämaiden asutus – pyyntikulttuurin elinkeinot ja olosuhteet. In: Häggman

Aalto

(eds) Suomalaisen arjen historiaa – Savupirttien Suomi. Helsinki: Welling Göös.

27.

Galeta

Sládek

Sosna

. (2011) Modeling neolithic dispersal in central Europe: Demographic implications. American Journal of Physical Anthropology 146: 104–115.

28.

Goslar

van der Knaap

Sheila

Maja

. (2005) Radiocarbon dating of modern peat profiles: Pre- and post-bomb. Radiocarbon 47(1): 115–134.

29.

Grabowski

(2011) Changes in cereal cultivation during the Iron Age in southern Sweden: A compilation and interpretation of the archaeobotanical material. Vegetation History and Archaeobotany 20(5): 479–494.

30.

Gräslund

(2007) Fimbulvintern, Ragnarök och klimatkrisen år 536–537 e.Kr. Saga och Sed 93–123.

31.

Grönlund

(1991) Sediment characteristics in relation to cultivation history in two varved lake sediments from East Finland. Hydrobiologia 214: 137–142.

32.

Grönlund

Asikainen

(1992) Reflections of slash-and-burn cultivation cycles in a varved sediment of lake Pitkälampi (North Karelia, Finland). Laborativ Arkeologi 6: 43–48.

33.

Grönlund

Simola

(2009) Paleoekologiaa Hämeen linnan liepeillä – Luonnontiede avaa ikkunoita menneisyyteen. Arx Tavastica 13: 32–57.

34.

Grönlund

Kivinen

Simola

(1992) Pollen-analytical evidence for Bronze-Age cultivation in Eastern Finland. Laborativ Arkeologi 6: 37–42.

35.

Hakala

Sarmaja-Korjonen

Miettinen

(2004) The origin and evolution of Lake Vähä-Pitkusta, SW Finland – A multi-proxy study of a meromictic lake. Hydrobiologia 527: 85–97.

36.

Haltia-Hovi

Nowaczyk

Saarinen

(2010) Holocene palaeomagnetic secular variation recorded in multiple lake sediment cores from eastern Finland. Geophysical Journal International 180(2): 609–622.

37.

Hicks

(1976) Pollen analysis and archaeology in Kuusamo, north-east Finland, an area of marginal human interference. Transactions of the Institute of British Geographers 1(3): 362–384.

38.

Holcomb

(1986) Dating recent Hawaiian lava flows using paleomagnetic secular variation. Geological Society of America Bulletin 7: 829–839.

39.

Holmblad

(2010) Coastal Communities on the Move: House and Polity Interactions in Southern Ostrobothnia 1500 BC–AD 1: Archaeology and Environment, vol. 26. Umeå: Department of Historical, Philosophical and Religious Studies.

40.

Huttunen

(1990) Vegetation and palaeoecology of a bog complex in S Finland. Aquilo. Series Botanica 28: 27–37.

41.

Huttunen

Simola

(1986) Heinäveden asutushitoria luonnonarkistojen valossa. In: Hämynen

(ed.) Heinäveden historia I. Heinävesi: Municipality of Heinävesi, pp. 39–49.

42.

Huttunen

Tolonen

(1977) Human influence in the history of Lake Lovojärvi, S. Finland. Finskt Museum 1975: 68–105.

43.

Huurre

(1979) 9000 vuotta Suomen esihistoriaa. Helsinki: Otava.

44.

Juhola

Etu-Sihvola

äreoja

T N

. (2014) Starch analysis reveals starchy foods and food processing from Finnish archaeological artefacts. Fennoscandia Archaeologica XXXI: 79–100.

45.

Junno

Uusitalo

Oinonen

(2016) Radiocarbon dates of Helsinki University – ¹⁴CARHU. Available at: www.oasisnorth.org/carhu (Article in preparation).

46.

Keränen

(1984) Kainuun asuttaminen. Studia Historica Jyväskyläensia 28: 1–282.

47.

Koivula

Raatikainen

Kankainen

. (1994) ihmisen vaikutus keskisuomessa siitepölytutkimuksen valossa. Jyväskylän Yliopiston Suomen Historian Julkaisuja 21: 1–93.

48.

Korpela

(2012) Migratory Lapps and the population explosion of Eastern Finns: The early modern colonization of Eastern Finland reconsidered. In: Damm

Saarikivi

(eds) Networks, interaction and emerging identities in Fennoscandia and beyond, Tromsø, 13–16 October 2009. Helsinki: Société Finno-Ougrienne, pp. 241–261.

49.

Kotilainen

Saarinen

Winterhalter

(2010) High-resolution paleomagnetic dating of sediments deposited in the central Baltic Sea during the last 3000 years. Marine Geology 166: 51–64.

50.

Lagerås

(2013) Medieval colonisation and abandonment in the south Sweden Upland: A review of settlement and land use dynamics inferred from the pollen record. Archaeologia Baltica 20: 77–90.

51.

Lahtinen

Rowley-Conwy

(2013) Early farming in Finland: Was there cultivation before the Iron Age (500 BC)? European Journal of Archaeology 4: 660–684.

52.

Lavento

(1998) Sisämaan vanhemman metallikauden väestö tutkimusongelmana. Muinaistutkija 4: 44–55.

53.

Lund

(1996) A comparison of Holocene paleomagnetic secular variation. Journal of Geophysical 101(B4): 8007–8024.

54.

Luukko

(1959) Erämark. Kulturhistorisk lexicon för nordisk medeltid IV: 39–45.

55.

Mackereth

FJH

(1971) On the variation in direction of the horizontal component of remanent magnetisation in lake sediments. Earth and Planetary Science Letters 12: 332–338.

56.

Mannermaa

Deckwirth

(2010) Suomen varhainen karjatalous: tutkimuksen nykytila ja ongelmat. In: Hirvilammi

(ed.) Varhainen viljely Suomessa. Loimaa: Sarka - The Finnish Museum of Agriculture, pp. 54–81.

57.

Michczyńska

Michczyński

Pazdur

(2007) Frequency distribution of radiocarbon dates as a tool for reconstructing environmental changes. Radiocarbon 49(2): 799–806.

58.

Miettinen

Vuorela

(1998) Archaeological and palynological studies of the agricultural history of Vörå and Malax, Southern Ostrobothnia. Fennoscandia Archaeologica V: 47–68.

59.

Miettinen

Grönlund

Simola

. (2002) Palaeolimnology of Lake Piene-Kuuppalanlampi (Kurkijoki, Karelian Republic, Russia): Isolation history, lake ecosystem development and long-term agricultural. Journal of Paleolimnology 27(1): 29–44.

60.

Mikkola

(2009) Kaskesta peltoon – rautakautisen maanviljelyn tutkimushistoria. Arkeologipäivät 2008: 38–44.

61.

Mökkönen

(2010) Kivikautinen maanviljely Suomessa. Suomen Museo 2010: 5–35.

62.

Nordqvist

Häkälä

(2014) Distribution of Corded Ware in the areas north of the Gulf of Finland – An update. Estonian Journal of Archaeology 18: 3–29.

63.

Oinonen

Pesonen

Tallavaara

(2010) Archaeological radiocarbon dates for studying the population history in eastern Fennoscandia. Radiocarbon 52: 393–407.

64.

Oinonen

Pesonen

Alenius

. (2014) Event reconstruction through Bayesian chronology: Massive mid-Holocene lake-burst triggered large-scale ecological and cultural change. Holocene 24(11): 1418–1427.

65.

Ojala

AEK

Alenius

(2005) 10000 years of interannual sedimentation recorded in the Lake Nautajärvi (Finland) clastic–organic varves. Palaeogeography, Palaeoclimatology, Palaeoecology 219(3–4): 285–302.

66.

Ojala

AEK

Francus

Zolitschka

. (2012) Characteristics of sedimentary varve chronologies – A review. Quaternary Science Reviews 43: 45–60.

67.

Orrman

(1991) Maantieteelliset tekijät ja kiinteän asutuksen leviäminen Suomessa ja eräissä Ruotsin osissa esihistoriallisen ajan lopulla ja keskiajalla. In: Kyösti

Koivumen

Mäkivuoti

. (eds) Studia historica septentrionalia 21: Suomen varhaishistoria: Tornion kongressi 14.-16.6.1991. Oulu: The Historical Association of Northern Finland, pp. 365–373.

68.

Orrman

(2003) The Cambridge History of Scandinavia (ed Helle

). Cambridge: Cambridge University Press.

69.

Pinhasi

Fort

Ammerman

(2005) Tracing the origin and spread of agriculture in Europe. PLoS Biology 3(12): E410.

70.

Reimer

Bard

Bayliss

. (2013) IntCal13 and Marine13 radiocarbon age calibration curves 0–50,000 years cal BP. Radiocarbon 55(4): 1869–1887.

71.

Saastamoinen

(1999) investigating prehistoric economy with pollen analysis – The Early Metal Age dwelling site at Kitulansuo in Ristiina, East Finland. In: Huurre

(ed.). Dig it all: Papers dedocated to Ari Siiriäinen. Helsinki: Suomen muinaismuistoyhdistys, pp. 131–148.

72.

Sarmaja-Korjonen

(1992) Fine-interval pollen and charcoal analyses as tracers of early clearance periods in S Finland. Annales Botanici Fennici 146: 1–75.

73.

Shennan

Downey

Timpson

. (2013) Regional population collapse followed initial agriculture booms in mid-Holocene Europe. Nature Communications 4: 2486.

74.

Siiriäinen

(1980) On the cultural ecology of the Finnish stone Age. Suomen Museo 1980: 4–40.

75.

Simola

(1999) Siitepölyanalyysin mahdollisuudet ja rajat. In: Fogelberg

(ed.) Pohjolan poluilla – Suomalaiset nykytutkimuksen mukaan. Helsinki: Finnish Society of Science and Letters, pp. 155–158.

76.

Simola

Grönlund

Uimonen-Simola

(1988) Etelä-Savon asutushistorian paleoekologinen tutkimus. Joensuun Yliopisto. Karjalan Tutkimuslaitoksen Julkaisuja 84: 1–55.

77.

Simola

Grönlund

Huttunen

. (1985) Pollen analytical evidence for Iron Age origin of cup-stones in the Kerimäki Area. Iskos 5: 527–531.

78.

Soininen

(1957) Erämaiden asuttaminen. In: Hämeen historia: 2,1. Noin vuodesta 1540 vuoteen 1721. Karisto. Hämeenlinna, pp. 5–59.

79.

Solantie

(1988) Climatic conditions for the cultivation of rye with reference to the history of settlement in Finland. Fennoscandia archaeologica V 3–20.

80.

Stanton

Snowball

Zillén

. (2010) Validating a Swedish varve chronology using radiocarbon, palaeomagnetic secular variation, lead pollution history and statistical correlation. Quaternary Geochronology 5(6): 611–624.

81.

Sundell

Heger

Kammonen

. (2010) Modelling a neolithic population bottleneck in Finland: A Genetic Simulation. Fennoscandia Archaeologica XXVII: 3–20.

82.

Taavitsainen

(1987) Wide-range hunting and swidden cultivation as prerequisites of the Iron Age Colonization in Finland. Suomen Antropologi 4:213–233.

83.

Taavitsainen

(2004) Culture clach or compromise? The notion of boundary in the utilization of wilderness areas. In The European frontier – Clashes and compromises in the Middle Age. In: Staecker

(ed.) The European frontier: Clashes and compromises in the Middle Ages; international symposium of the Culture Clash or Compromise (CCC) project and the Department of Archaeology, Lund, 13–15 October 2000 Lund: Lund University, Department of Archaeology and Ancient History, pp. 45–55.

84.

Taavitsainen

Simola

Grönlund

(1998) Cultivation history beyond the periphery: Early agriculture in the North European Boreal forest. Journal of World Achaeology 13: 155–196.

85.

Taavitsainen

Vilkuna

Henry

(2007) Suojoki at Keuruu – A Mid 14th Century Site of the Wilderness Culture in the Light of Settlement Historical Process in Central Finland. Helsinki: The Finnish Academy of Science and Letters.

86.

Tallavaara

Seppä

(2012) Did the mid-Holocene environmental changes cause the boom and bust of hunter-gatherer population size in eastern Fennoscandia? The Holocene 22(2): 215–225.

87.

Tallavaara

Pesonen

Oinonen

(2010) Prehistoric population history in eastern Fennoscandia. Journal of Archaeological Science 37: 251–260.

88.

Timpson

Colledge

Crema

. (2014) Reconstructing regional population fluctuations in the European Neolithic using radiocarbon dates: A new case-study using an improved method. Journal of Archaeological Science 52: 549–557.

89.

Tolonen

(1980) Comparison between radiocarbon and varve dating in. Boreas 9: 11–19.

90.

Tolonen

(1984) Interpretation of changes in the ash content of ombrotrophic peat layers. Bulletin of the Geological Society of Finland 56(1–2): 207–219.

91.

Tolonen

Ruuhijärvi

(1976) Standard pollen diagrams from the Salpauselkä region of Southern Finland. Annales Botanici Fennici 13: 155–196.

92.

Tolonen

Siiriäinen

Hirviluoto

(1976) Iron Age cultivation in SW Finland. Finskt Museum 83: 5–66.

93.

Tolonen

(1976) On Prehistoric Cultivation in Sääksmäki, S. Finland. Finskt Museum 83: 67–84.

94.

Tolonen

(1978a) Paleoecology of annually laminated sediments in Lake Ahvenainen, S. Finland. I. Pollen and charcoal analyses and their relation to human impact. Annales Botanici Fennici 15: 177–208.

95.

Tolonen

(1978b) Paleolimnology of annually laminated sediments in Lake Ahvenainen S. Finland. II. Comparison dating methods. Annales Botanici Fennici 15: 209–222.

96.

Tolonen

(1981) An absolute and relative pollen analytical study on prehistoric agriculture in South Finland. Annales Botanici Fennici 18: 213–220.

97.

Tolonen

(1985) Palaeoecological reconstruction of vegetation in a prehistoric settlement area, Salo, SW Finland. Annales Botanici Fennici 22: 101–116.

98.

Tolonen

(1990) Pollen analythical evidence of ancient human action in the Hilford area of Kuhmoinen, central Finland. In: Taavitsainen

J-P

(ed.) Ancient Hillforts of Finland. Helsinki: Suomen Muinaismuistoyhdistys, pp. 247–269.

99.

Tolonen

(1988) Synchronous pollen changes and traditional land use in south Finland, studied from three adjacent sites: A lake, a bog and a forest soil. In: Lang

Schlüchter

(eds) Lake, Mire and River Environments. Rotterdam: Brookfield, pp. 83–97.

100.

Tomminen

(2006) Paleoekologisia tutkimuksia: Luumäen Niemenkylä ja Lappenrannan Kauskila. Suomen Museo 2005: 139–155.

101.

Tvauri

(2014) The impact of the climate catastrophe of 536–537 ad in Estonia and neighbouring areas. Estonian Journal of Archaeology 18(1): 30–56.

102.

Uino

(1999) Muinais-Karjala arkeologian kannalta. In: Fogelberg

(ed.) Pohjan poluilla – Suomalaiset nykytutkimuksen mukaan. Helsinki: Finnish Academy of Science and Letters, pp. 338–350.

103.

Viklund

(2011) Early farming at Umeå in Västerbotten. Charred cereal grains dated to the Bronze Age. Fornvännen 106: 238–242.

104.

Vuorela

(1972) Human influence on the vegetation of Katinhäntä bog, Vihti, S. Finland. Acta Botanica Fennica 98: 1–21.

105.

Vuorela

(1975) Pollen analysis as a means of tracing settlement history in SW-Finland. Acta Botanica Fennica 104: 1–48.

106.

Vuorela

(1978) Local settlement history of Lahti area as shown by pollen analysis. Bulletin of the Geological Society of Finland 50: 45–57.

107.

Vuorela

(1981) The vegetational and settlement history in Sysmä, central south Finland, interpreted on the basis of two pollen diagrams. Bulletin of the Geological Society of Finland 51–53: 47–61.

108.

Vuorela

(1982) Siitepolyanalyysi arkeologisessa tutkimuksessa. Lahden museo- ja taidelautakunta, Tutkimuksia 19: 42–62.

109.

Vuorela

(1983) Vohtenkellarinsuo, a bog in Paimio, SW Finland with a cultural origin. Bulletin of the Geological Society of Finland 55: 57–66.

110.

Vuorela

(1985) On the vegetation and agricultural history of Perniö, SW Finland. Annales Botanici Fennici 22: 117–127.

111.

Vuorela

(1990) Pollenanalytiska studier. In: Zillacius

(ed.) Finska Skaren: studier I Abolansk kulturhistoria. Foreningen Konstsamfundets publikationsserie 7. Helsinki: Konstsamfundet, pp. 115–132.

112.

Vuorela

(1991) Lounais-suomen varhaismetallikautinen asutus ja viljely siitepölyanalyysin valossa. Karhunhammas 13: 2–23.

113.

Vuorela

(1993) The past and present nature of the Rasi area. Helsinki Papers in Archaeology 5: 11–39.

114.

Vuorela

(1994) Keski-Suomen asutushistoriaa Keuruun ja Jämsän järvikerrostumien arkistoissa. Espoo: Geological Survey of Finland.

115.

Vuorela

(1995) Pollen evidence of Stone Age and Early Metal Age settlement in Taipalsaari, Southern Karelia, Eastern Finland. Fennoscandia Archaeologica XII: 207–214.

116.

Vuorela

Hicks

(1996) Human impact on the natural landscape in Finland: A review of the pollen evidence. PACT 50: 245–257.

117.

Vuorela

Kankainen

(1991) Siitepölyanalyyttinen tutkimus asutuksen vaikutuksista puolangan kotilan kylässä. Espoo: Geological Survey of Finland.

118.

Vuorela

Kankainen

(1992) Tammelan asutushistoria Kuiväjärven kerrostumien paleoekologisen tutkimuksen valossa. Espoo: Geological Survey of Finland.

119.

Vuorela

Kankainen

(1993) Luonnon ja kulttuurimaiseman kehitys Taipalsaaressa. Espoo: Geological Survay of Finland.

120.

Vuorela

Lempiäinen

(1988) Archaeobotany of the oldest cereal grain in Finland. Annales Botanici Fennici 25: 33–45.

121.

Vuorela

Lempiäinen

Saarnisto

(2001a) Land use pollen record from the Island of Valamo, Russian Karelia. Annales Botanici Fennici 38: 139–165.

122.

Vuorela

Saarnisto

Lempiäinen

. (2001b) Stone Age to recent land-use history at Pegrama, northern Lake Onega, Russian Karelia. Vegetation History and Archaeobotany 10: 121–138.

123.

Vuorela

Uutela

Saarnisto

. (1993) Vuojärven ja Antinlammen kerrostumat Laukkaan asutus- ja luonnonhistorian arkistoina. Espoo: Geological Survey of Finlando.

124.

Wallin

Segerström

(1994) Natural resources and agriculture during the Iron Age in Ostrobotnia, western Finland, investigated by pollen analysis. Vegetation History and Archaeobotany 3(3): 89–105.

125.

Williams

(2012) The use of summed radiocarbon probability distributions in archaeology: A review of methods. Journal of Archaeological Science 39: 578–589.

126.

Zillén

Snowball

Sandgren

. (2003) Occurrence of varved lake sediment sequences in Värmland, west central Sweden: Lake characteristics, varve chronology and AMS radiocarbon dating. Boreas 32(4): 612–626.

127.

Zvelebil

(2010) Innovating hunter-gatherers – Mesolithic in the Baltic. In: Bailey

Spikins

(eds) Mesolithic Europe. Cambridge: Cambridge University Press, pp. 18–59.