Weeds in spring cereal fields in Finland – a third survey

A survey of weeds in spring cereal fields was conducted in 16 regions of southern and central Finland in 1997–1999. Data were collected from conventional and organic farms, both of which applied their normal cropping practices. A total of 690 fields were investigated by counting and weighing the weed species from ten sample quadrats 0.1 m in size in late July – early August. Altogether 160 weed species were found, of which 134 were broad-leaved and 26 grass species. The total number of weed species ranged from 41 to 84 between regions. In organically farmed fields, the average species number was 24 and in conventionally farmed fields 16. The most frequent weed species were Viola arvensis 84%, Stellaria media 76% and Galeopsis spp. 70%. Only 18 species exceeded the frequency level of 33%. The average density of weeds was 136 m (median= 91) in sprayed conventional fields, 420 m (374) in unsprayed conventional fields and 469 m (395) in organic fields. The average airdry above-ground biomass of weeds was 163 kg ha (median=63), 605 kg ha (413) and 678 kg ha (567), respectively. Weed biomass accounted for 3% of the total biomass of the crop stand in sprayed conventional fields and for 17% in organic fields. Elymus repens, the most frequent grass species, produced the highest proportion of weed biomass.

Regular weed surveys are considered a necessary and valuable way of monitoring the responses of weed floras to changes in agricultural practices and habitats.The most frequent and abundant weed species are of interest in terms of control measures.Current research projects focus not only on agricultural features of weed occurrence but also on biodiversity in agricul-
In the 1990s, crop production in Finland underwent changes that apparently affected weed floras in fields.The most marked changes were 1) the implementation of an extensive fallow scheme in the early 1990s, which peaked in 1994 at 505 000 ha (almost 25% of the cultivated area), 2) the switch to organic farming by roughly 5000 farms, accounting now for 6% of the cultivated field area and 3) the shift from the use of phenoxy acid herbicides to sulphonylureas, which differ from the former in selectivity among weed species.
Since 1995, when Finland became a member of the European Union, the economic basis of crop production has had to contend with cuts in prices of products.To compensate for losses to farmers, the government introduced a system of area-based subsidies taking into account certain special conditions.The Agri-Environmental Support Scheme was launched with the aim, among others, of promoting the environmentally sus-tainable use of fertilizers and pesticides (Ministry of Agriculture and Forestry 1999).
A longer term change in agriculture has been the regional specialization of production.Nowadays, southern and south-western Finland are characterized by a high dominance of cereal crops in rotation whereas in central and eastern Finland the use of arable land is more diverse, many farms practising crop rotation based on grassland for silage and hay and pasture for cattle.Since this diversity of crop rotation has been found to have some effect on the species composition (Bàrberi et al. 1997) and species diversity (Doucet et al. 1999) of weed communities, we expected to find differences between regions.
The objective of the survey was to investigate the current status of weed occurrence in spring cereal fields, examining both conventionally and organically cultivated fields.This first report focuses on the botanical composition of fields.It is aimed, on the one hand, at farmers, advisory services and the chemical industry with a view to promoting specific weed control measures and, on the other, at scientists and authorities as a report on the diversity of weed flora.A more detailed analysis of the factors explaining weed occurrence and changes in weed floras over the decades will be given separately.

Study regions, farms and fields
The present weed survey was carried out in southern and central Finland in 1997-1999.The 16 regions surveyed (Fig. 1, Table 1) are referred to in tables and figures by the numbers given in Table 1.Regions 1 and 2 were studied in 1997 as a preliminary survey.The other regions (nos.[3][4][5][6][7][8][9][10][11][12][13][14][15][16] were chosen from those studied in surveys conducted in the 1960s and 1980s.The initial intention of this survey was to visit exactly the same fields as those studied earlier but this was not entirely practicable, since spring cereals were not grown in many of the study fields in the particular study year.
The choice of study regions reflects the prevalence of spring cereal crops in rotation in different regions.Consequently, the highest number of fields was studied in south-western Finland, where spring cereals are grown in more than 50% of the arable area and annual spring-sown crops are predominant.The number of study fields decreased towards the east and north, where spring cereals account for less than 50% of the cultivated area, and many farms include grassland in their crop rotation.
The number of farms visited was 305, of which 229 engaged in conventional and 76 in organic farming.The study farms were either crop husbandry farms (174) without grassland in crop rotation or animal husbandry farms (131) with pasture in crop rotation.The fields of crop husbandry farms exceeded 70% of the studied fields in five regions (Nauvo/Korppoo, Paimio/ Tarvasjoki, Laihia, Nurmijärvi and Iitti) and those of animal husbandry farms in four regions (Jokioinen, Kitee, Nivala, Vieremä).The farm types were not evenly distributed over the study regions but reflected, as said, regional differences in farming structure.The survey region in the south-western archipelago, Nauvo/Korppoo (3), was the only one without organically cultivated spring cereal fields.The proportion of organic  Typically, a great majority of study fields were sown in mid-May (weeks no.19-21) and sprayed with herbicides in mid-June (weeks no.23-25).Weather conditions varied considerably between study years and regions (Table 2).
On each farm, one to eight spring cereal fields were examined, giving a total of 690 fields: 343 under barley (Hordeum vulgare L.), 233 under oats (Avena sativa L.), 97 under wheat (Triticum aestivum L.) and 17 in mixed cultivation.Altogether 525 study fields were cultivated conventionally and 165 organically.
To gain an overview of actual weed populations in the middle of the growing season, no instructions on crop management, e.g. on herbicide use or fertilization, were given to farmers in advance.At the sampling time, the information on cropping measures was recorded by in-terviewing the farmers.The national regulation on organic cereal production bans the application of chemical fertilizers and pesticides and sets an upper limit on the use of organic fertilizers (Ministry of Agriculture and Forestry 1999).
The average area of the 690 study fields was 4.3 ha (range 0.

Weed sampling
The present survey was carried out in 1997-1999 during a 4-week period starting in mid-July (weeks no.28/29), by which time the spring cereals had reached their heading stage and at least 1 month had elapsed since herbicide treatment.The occurrence of weeds was assessed from 10 sample quadrats randomly placed in each field.For this purpose, each field was split in a 10 × 10 cell grid in which the positions of sample quadrats were set with a random number calculator.The size of grid cells varied between fields according to the area of each field.Two sample quadrats were placed at a distance of 1-3 m from the sown field edge and the other eight more than 5 m from the edge.
Weed density was determined by counting the number of plants or shoots of grass weeds by species in a rectangular frame measuring 0.1 m 2 (25 cm × 40 cm), which was a corner area within a larger quadrat measuring 1.0 m 2 (1.0 m × 1.0 m).The larger quadrat was used for observations of the presence/absence of each species.In 1997 (Jokioinen and Lammi), only the 0.1 m 2 sample quadrats were used in the sampling.The results presented in the tables and figures derive from data collected from the 0.1 m 2 quadrats and pooled over the samples in each field.A complete list of the additional weed species found in the presence/absence observation is given in Appendix 1 as a supplement to the 42 species presented in the tables.
In four out of ten small sample quadrats, weeds and cereals were cut at the soil surface and their biomass was weighed by species after the samples had been dried in an air-flow dryer at 40°C for some days.The average biomass of a single plant was calculated by dividing the biomass of each species by the numbers of that plant in fields where the species was present, provided that data were available from at least 20 fields.

Nomenclature and data analysis
As the present survey focused on the diversity of weed flora, all weed species found in sampling areas were assessed.Nevertheless, some genera/taxa, e.g.Galeopsis spp.and Lamium spp., had to be pooled since they could not be identified by species at the small seedling stage.
The term frequency indicates the proportion of fields where the species was found.The frequency results are classified as in Mukula et al. (1969).For each field, the total weed density and biomass were summed, and the averages, standard deviations and median values are given in the text.To ensure consistency with the manner of data presentation in previous surveys (Mukula et al. 1969, Erviö andSalonen 1987), the average weed densities and dry weights by species are presented.Median values are given for total biomass and density but not for single species since most of the median values would have been zero due to the low frequency.
The results of the three surveys should nonetheless be compared with caution as the sampling technique used in the present survey was slightly modified in terms of size and number of quadrats.Furthermore, in contrast to previous surveys, the majority, 66%, of the fields studied were sprayed with herbicides.
The similarity of species composition between regions was compared by Jaccard's simi- larity coefficient S j (Jaccard 1912) (S j = c/(A + B -c), where c = number of species common to both samples A and B, A = number of species in sample A, B = number of species in sample B).The data were pooled over all fields of each region before the analysis.The results of similarities were presented as dendrograms in which average-linkage clustering (the unweighted pairgroup method) was applied (see Krebs 1999).
The diversity of weed species was described by species richness and evenness.The number of species was used as a measure of species richness.Since the number of species depends on the sample size and since the number of sampled fields varied between regions, total species numbers between regions could not be compared.Therefore, the expected number of species E(S n ) was calculated for each region by rarefaction: where E(S n ) = expected number of species in a random sample of n individuals, S = total number of species in the entire collection, N i = number of individuals in species i, N = total number of individuals in the collection, n = sample size (number of individuals) chosen for standardization (see Heck et al. 1975, Krebs 1995).
In rarefaction, the number of species of larger samples are scaled down to the given number of individuals which permits the comparison of species numbers between samples differing in size.Since the lowest number of individuals observed was 3386, we scaled sample sizes down to 3300 individuals in the data sets of all fields.In conventional sprayed and organic fields the lowest numbers of individuals were 1519 and 1867, respectively, and we scaled the number of individuals down to 1500.In all data sets, the lowest number of individuals was in the Nurmijärvi region.The sample size of 1500 individuals was used in both data sets to permit the comparison of numbers of species between conventional sprayed and organic fields.Hill's evenness index E 2,1 where p i is the proportion of the total sample belonging to the ith species (for properties of the index, see Alatalo 1981) was used as a measure of evenness.Species richness and evenness were calculated for each field separately.Data from the ten 0.1 m 2 sample quadrats were pooled before the calculation.The evenness index was calculated by using data on the number of individuals.

Frequency of weed species
The occurrence of the 42 most frequent weed species that exceeded the overall frequency level of 5% in the small sample quadrats is presented by region (Tables 3, 8-9); the remaining species observed from the larger quadrats are listed by region (Appendix 1).
The five most common weed species were Viola arvensis, Stellaria media, Galeopsis spp., Chenopodium album and Elymus repens.These species occurred in more than 60% of the fields studied (Table 3).In contrast, most of the observed 160 weed species were fairly rare as they were found in less than 2% of the survey fields (Table 4 and Table 5).
With a frequency of 66%, Elymus repens was by far the most common grass weed (Table 3).The next most common grass species, although much lower in the ranking list, were Poa annua (frequency 12%), Phleum pratense (3%) and Alopecurus geniculatus (3%).With the exception of E. repens and P. annua, however, the grass species could not always be identified by species at seedling stage.Thus, they were only recorded and pooled to a monocot class that reached a frequency level of 26%.
The ranking list of the ten most frequent weed species in the conventional fields sprayed with herbicides was somewhat different from that in the organic fields (Table 6).Highly productive perennial weeds occurred more frequently in organic than in conventional farming; Elymus repens had a frequency of 81% in organic and 60% in conventional farming, Sonchus arvensis 54% and 31% and Cirsium arvense 34% and 15%, respectively.Similarly, typical grassland weed species such as Achillea millefolium 15% vs. 2%, Ranunculus repens 30% vs. 12% and Taraxacum officinale 30% vs. 20% were more frequent in organic than in conventional farming.
The most common weed species were found every year and in all regions, although their rank order varied between the regions.In contrast, some species, e.g.Lamium spp. in south-western Finland, Gnaphalium uliginosum in central Finland and Veronica serpyllifolia in the Vieremä region, were common only in some regions.
The analysis of species composition with similarity indices revealed two main groups of which the first consisted of two subgroups (Fig. 2), one in south-western (Jokioinen, Laitila and Paimio/Tarvasjoki) and one in southern Finland (Lammi, Iitti, Imatra/Ruokolahti and Nurmijärvi) and the Laihia region in Ostrobothnia.The second main group included regions (Tammela, Mikkeli, Laukaa/Toivakka, Vieremä, Kihniö/ Parkano, Kitee and Nivala) throughout Finland.The most striking difference in species composition between the two main groups was the absence of many species (Cerastium fontanum, Juncus bufonius, Veronica serpyllifolia, Ranunculus acris and Epilobium angustifolium) from some regions of the first group whereas in the Vol. 10 ( 2001): 347-364.
second group they were found in all regions (Table 3).The species composition in the Nauvo/ Korppoo region (no organic farming) differed most markedly from that in any of the other regions because some species, e.g.Gnaphalium uliginosum, Persicaria hydropiper and Poa annua, were not found there at all (Table 3).

Species diversity
Altogether 188 weed species were found in the large (1.0 m 2 ) sampling quadrats and 160 in the small (0.1 m 2 ) quadrats.The total number of observed species, S OBS , in regions ranged from 41 to 84 (Table 7).In three regions (Nauvo/Korppoo, Laihia and Nurmijärvi), the number of observed species was below 50 and in two regions (Tammela and Mikkeli) it exceeded 80.The same regions had the lowest and the highest expected number of species, E(S n ), calculated by rarefaction analysis (Table 7).No clear regional trend in the variation in the total number of species was observed.For example, the regions with the highest species number included regions in both eastern (Kitee, Mikkeli and Imatra/Ruokolahti) and south-western (Tammela and Paimio/Tarvasjoki) Finland.
In 11 out of 15 regions, the total number of observed species, S OBS , was higher in organically farmed than in sprayed conventional fields, whereas the expected number of species, E(S n ), was higher in organically farmed fields in nine regions (Table 7).However, even though the number of species might be almost the same, the species composition could be very different.For example, in the Imatra/Ruokolahti region the difference in the number of observed species, S OBS , was only two but the proportion of co-occurring species was as low as 54.9% (Table 7).

Salonen, J. et al. Weeds in spring cereals in Finland
The average number of weed species per field was 18.In organically farmed fields, the average species number was 24 and in conventionally farmed fields 16.However, in conventional fields the difference between sprayed and unsprayed fields was ten species (15 vs. 25 species).The difference in the average number of weed species between the poorest and the richest regions was 11 in both sprayed conventional and organic fields (Fig. 3).The number of species was, however, higher in organic fields (variation in averages 14-25.1)than in sprayed conventional fields (6.6-17.6).The greater the number of species in sprayed conventional fields, the lower were the values of the evenness index (i.e.some species dominated the weed community); a similar trend was not observed in organic fields.
The three most abundant weed species accounted for 32-53% of the total density in each region.There were considerable differences in the abundance of weeds between the regions (Table 8), reflecting both annually changing factors such as weed control measures and weather conditions and more permanent factors such as soil conditions and farming practices.Some less frequent species, e.g.Juncus bufonius, Persicaria hydropiper and Veronica serpyllifolia, were of importance in some regions.
Although the average biomass production of weeds varied between the regions (Table 9), the annual averages pooled over the regions were at roughly the same level: 227 kg ha -1 in 1997, 251 kg ha -1 in 1998 and 244 kg ha -1 in 1999.
In all regions, the three most abundant weed species together accounted for more than 50% of the average total biomass production in the region.The same species were quite often the most productive (e.g.Elymus repens, Chenopodium album) in different regions.Perennial  Vol. 10 ( 2001): 347-364.
weeds such as Cirsium arvense, Sonchus arvensis and Tussilago farfara also ranked high, since one individual perennial plant alone may produce more biomass than 100 seedlings of annual broad-leaved weeds.
The average biomasses of individual weed plants and their rank order differed in sprayed conventional and organic cropping as indicated in the list of the 15 most productive species (Table 10).Eleven of these were the same species in both cropping types.
Broad-leaved species accounted for 43% of the average total biomass production in sprayed conventional fields and for 72% in organic fields.Elymus repens was the most efficient biomass producer, as it accounted for 26% of the total weed biomass production pooled over all organic fields and for as much as 50% in sprayed conventional fields.
In all, the proportion of weed biomass relative to total vegetative biomass (crop + weeds) was fairly low (mean = 3.0%) in sprayed conventional fields, somewhat higher (mean = 12.6%) in untreated conventional fields and highest (mean = 17.1%) in organic fields.

Discussion
Species diversity in cereal fields was fairly high, a total of 160 weed species, actually 188 species, being found in the surveyed fields.The classification of species into broad-leaved/grass and annual/perennial types showed a similar distribution over the frequency classes.Characteris-tic of weed infestation in spring cereal fields is the occurrence of broad-leaved annuals and Elymus repens.
Only 42 weed species/taxa exceeded the overall frequency level of 5%.Furthermore, although the number of weed species observed was fairly high, the number of dominant weed species was quite low, as only 18 species exceeded the frequency level of 33% and more than 60% of the species remained below the 2% frequency level.In terms of successful cereal production, shifts in the rank order of the most frequent and dominant species should have greater implications for the planning of weed management than have actual changes in the diversity of weed communities, as suggested also by Légère and Derksen (2000).
The present finding of 160 ( 188) weed species along with the earlier finding of 304 species (Mukula et al. 1969) shows that spring cereal fields support a much more diverse weed flora than which is important in terms of crop protection.Due to the difference in sample sizes, however, the findings of these two surveys cannot be directly compared.Further, we may expect that some weed species detected previously are now extinct or extremely rare in Finnish arable fields.Such a decline in arable flora has been observed in other countries, too (Albrecht 1995, Andreasen et al. 1996, Sutcliffe and Kay 2000).
Regional specialization of agricultural practices has taken place in many countries -including Finland -during the last decades.The study regions in southern and south-western Finland can be regarded as this country's main cereal production area.In other study regions, the use of arable land is more diverse as many farms base their crop rotation on grassland for silage and hay and pasture for cattle.These two areas proved to have a different weed species composition and species diversity.Regions in central and eastern Finland had, in general, higher average weed species numbers than did regions in southern and south-western Finland.Species absent from southern and south-western Finland were those typical of Finnish grasslands (Raa- Vol . 10 (2001): 347-364. tikainen andRaatikainen 1975).Therefore, regional specialization in agricultural production can be regarded as one reason for differences in the composition of weed communities between these two areas.
Another factor affecting the species composition in different regions (probably in interaction with cropping practices) was soil properties, organic soils being predominant in some regions of central and eastern Finland (e.g.Kitee and Nivala).Soil properties have been found to be an important factor explaining the species composition of weed communities (Andreasen et al. 1991).Salonen (1993) and Erviö et al. (1994) analysed the data of the second weed survey and found that Chenopodium album, Lamium spp.and Fallopia convolvulus were more abundant in clay than in coarse mineral or organic soils.Furthermore, Galeopsis spp.and Polygonum lapathifolium favoured organic soils and Poa annua and Lapsana communis coarse mineral soils.
It should also be kept in mind that in both cereal and animal husbandry production systems the intensity of farming has increased tremendously and in many ways.The most striking changes during the last four decades have been the greater use of herbicides, increased nitrogen fertilization and improved tillage and sowing practices with combined seed and fertilizer drill methods (Elonen 1983, Mukula andRantanen 1987).It is thus unlikely that any single factor alone can adequately explain the changes in weed flora (Haas andStreibig 1982, Salonen 1993).
Another factor which should be taken into account when comparing regions is that the surveys were conducted in different years and under different weather conditions.Both 1997 and 1999 were fairly dry years, whereas 1998 was very wet.Surprisingly, however, the annual averages of weed biomass pooled over the regions were close to each other.Moreover, some regions were studied during the latter part of the onemonth survey period; this obviously had some effect on the biomass results but not so much on the frequency and density records.Therefore, it was appropriate to use a qualitative similarity index, which treats all species equally, irrespective of their abundance, in the comparison of species composition between regions although the rare species are assigned the same importance as the dominants.
Both the average and median weed density and weed biomass were fairly high, particularly in organic production and in the unsprayed fields of conventional production.In the early 1980s, the average weed density in unsprayed sample quadrats was 170 plants m -2 (median 124) and the average weed biomass 320 kg per hectare (median 183) (Salonen 1993).Furthermore, the biomass production of weeds in the 1960s averaged 1000 kg per hectare (Mukula 1974).
The use of herbicides started to increase in cereal production in Finland in the early 1960s, the peak volume (kg active ingredients) being reached in the early 1980s (Hynninen and Blomqvist 1997).Since the 1990s, sulphonylureas have replaced phenoxyacid herbicides to such an extent that the current sprayed area of spring cereals is about the same for both herbicide types.Herbicide treatment in cereal production is still a common practice on conventional farms (Londesborough et al. 2000).Most of the survey farms used the lowest recommended application rate.The main reason for annual interruptions to herbicide application was the unsuitability of weather conditions, either drought or an excess of rain.
Viola arvensis, Fumaria officinalis and Galium spurium represent species that were common in cereal-dominated production areas in southern Finland but cannot be successfully controlled with sulphonylureas.In general, the intensive manipulation of arable land in favour of crops has also favoured species such as Elymus repens and Viola arvensis, which can adapt to the strong competition in cereal stands.
In organic farming Chenopodium album, Stellaria media, Galeopsis spp.and Viola arvensis were found in more than 90% of fields studied.The majority of Finnish organic farms converted from conventional to organic cropping after 1995, when Finland became a member of the European Union.Therefore, the botanical composition of fields is still in the process of changing.However, in terms of species diversity, even the short period of organic farming has already been beneficial.Not only did the organic fields have a higher number of weed species but the sprayed conventional and organic fields differed in the most frequent species.Use of herbicides has been found to reduce the number of species in weed communities (Hyvönen and Salonen 2002).
In general, farmers should consider some kind of weed control during the growing season, as the weed infestation on organic farms was fairly high and hardly any direct control measures were carried out.Particularly, perennial weed species such as Elymus repens, Sonchus arvensis and Cirsium arvense may threaten the future of organic cereal production if their control is not given due consideration in crop rotation.Of interest is that these perennial species were not particularly common in a recent Swedish survey of organic farms (Rydberg and Milberg 2000).
Taking into consideration the frequency and biomass production of weed species, Elymus repens is by far the most harmful weed species for spring cereals in Finland.Glyphosate is one of the best selling herbicides in this country (Londesborough et al. 2000) but the conditions for spraying are often unsuitable and its efficacy on E. repens is seldom optimal as it is mostly applied on stubble after harvest in September.
In conclusion, economy was the driving force in the decision-making of crop production in the 1990s.Indirectly, this is reflected in weed floras now that organic farming has become a common practice and the inputs for cereal production are judiciously considered in conventional farming, too.The decrease in cropping intensity gives weeds an opportunity to make a comeback.

A
G R I C U L T U R A L A N D F O O D S C I E N C E I N F I N L A N D

Table 1 .
Number of fields surveyed by region, production type and cereal species.

Table 2 .
Weather conditions in survey regions.

Table 4 .
Distribution of annual and perennial weed species into frequency classes.

Table 5 .
Distribution of broad-leaved and grass weed species into frequency classes.

Table 6 .
Frequency of ten most common weed species in two cropping systems.

Table 7 .
Observed (S OBS ) and rarefied (E(S n ) with SD) number of species by region and production type, and proportion of co-occurring species between sprayed conventional and organic fields.

Table 10 .
The 15 most productive weed species (air-dry biomass, g/plant) in two cropping types.