Plant-availability of soil and fertilizer zinc in cultivated soils of Finland

203

Plant-availability of soil and fertilizer zinc in cultivated soils of Finland Markku Yli-Halla Yli-Halla, M. 1993. Plant-availability of soil and fertilizer zinc in cultivated soils of Finland.  University of Helsinki, Finland.) The Zn status of cultivated soils of Finland was investigated by chemical analyses and bioassays. The effect on ryegrass of different Zn fertilizers and Zn rates was studied in pot experiments and their effect on barley and timothy in field experiments. In an uncontaminated surface soil material of 72 mineral soils and 34 organogenic soils, total Zn (Zntot) was 10.3 -202 mg kg" 1 (median 66 mg kg" 1 ). In mineral soils, Zntot correlated positively with clay content (r = 0.81 ) and in organogenic soils negatively with organic C (r = -o.s3***). Zinc bound by organic matter and sesquioxides was sequentially extracted by 0.1 M K4P207 (Zn py ) and 0.05 M oxalate at pH 2.9 (Znox), respectively. The sum Zn py + Znox , 0 x, a measure of secondary Zn potentially available to plants, was 2 -88% of Zntot and was the lowest in clay (median 5%) and highest in peat soils (median 49%). Water-soluble and exchangeable Zn consisted of 0.3 -37% (median 3%) of Zntot, the percentage being higher in acid soils, particularly in peat soils. Zinc was also extracted by 0.5 M ammonium acetate -0,5 M acetic acid -0.02 M Na2-EDTA at pH 4.65 (ZnAc), the method used in soil testing in Finland. The quantities of ZnAc (median 2.9 mg dm" 3 , range 0.6 -29.9 mg dm" 3 ) averaged 50% and 75% of Znpy + Zn0x in mineral and organogenic soils, respectively, and correlated closely with Znpy . In soil profiles, ZnAc was with few exceptions higher in the plough layer (0 -20 cm) than in the subsoil (30 -100 cm).
In an intensive pot experiment on 107 surface soils, four crops ofryegrass took up 2 -68% (median 26%) of Znpy + Zn ox. 0 x. The plant-available Zn reserves were not exhausted even though in a few peat soils the Zn supply to grass decreased over time. Variation of Zn uptake was quite accurately explained by ZnAc but increasing pH had a negative impact on Zn uptake. Application of Zn (10 mg dm" 3 of soil as ZnSCri • 7H20) did not give rise to yield increases. In mineral soils, increase of plant Zn concentration correlated negatively with soil pH while ZnAc was of secondary importance. In those organogenic soils in which the reserves of native Zn were the most effectively utilized, plant Zn concentration also responded most strongly toapplied Zn. In two 2-year field experiments, Zn application did not increase timothy or barley yields. Zinc concentration of timothy increased from 30 mg kg" 1 to 33 and 36 mg kg" 1 when 3 or 6 kg Zn ha 1 was applied, respectively. The efficiency ofZnSCri ■ 7H20 alone did not differ from that of a fertilizer where ZnSCfi • 7H20 was granulated with gypsum. Zinc concentration of barley grains increased by foliar sprays of Na2Zn-EDTA but only a marginal response to soil-applied Zn (4.8 or 5.4 kg ha" 1 over three years) was detected in three 3-year experiments. High applications of Zn to soil (15 or 30 kg ha' 1 as ZnSCXt

INTRODUCTION
Zinc is a trace element, the average concentration of which in the earth's crust is quoted as 70 mg kg' 1 (Wedepohl 1991). There are minerals containing Zn among olivines and pyroxenes, e.g. acmiteaugite, and amphiboles, e.g. riebeckite. Also 2:1 clay minerals, mainly trioctahedral micas, contain Zn owing to isomorphic substitution of Mg or Fe ions for Zn at octahedral sites (Rankama and SAhama 1950,Lindsay 1972, Huang 1989. A substantial part of total Zn in soil occurs in clay and silt size particles (Shuman 1985), and total Zn content correlates with the content of clay or clay plus silt (Sippola 1974, Schlichting and Elgala 1975, Tjell and Hovmand 1978, Baghdady and Sippola 1983, Liang et al. 1990. Zinc is released into the soil solution from mineral structures through weathering reactions as Zn 2+ cation which is further adsorbed by various soil constituents and utilized by living organisms. The significance of Zn as a nutrient of higher plants was shown in 1926by Sommer and Lipman. Zinc is involved in several enzymatic reactions of protein and carbohydrate metabolism of plants (Marschner 1986). Zinc deficiency in crop production is extensive in calcareous soils (Sillanpää 1982), but insufficient Zn supply to dryland crops occurs also in acid soils for example in several states of the USA (Junus andCox 1987, Boswell et al. 1989),Brazil (LINS and COX 1988), Australia (Brennan and Gartrell 1990) and Zambia (Banda and Singh 1989). Zinc deficiency induced by liming has also been reported (Kowalenko et al. 1980,MACNAEiDHEetaI. 1986, Myhr 1988). In Finland, the average Zn concentration of timothy ranges from 24 to 32 mg kg' 1 (Lakanen 1969, Kähäri andNissinen 1978) and in cereal grains from 26 to 36 mg kg' 1 (Jaakkola andVogt 1978, Varo et al. 1980)although means as high as 54 mg kg' 1 have been reported in cereals (Pessi et al. 1974, Syvälahti and Korkman 1978). The Zn concentration in the crops of Finland is above the minimum physiological requirement of gramineous plants or clover, 10-20 mg kg' 1 of plant dry matter (e.g. Marschner 1986,Brennan and Gartrell 1990, Carsky and Reid 1990. Hence Zn applica-tions have not increased yields in field experiments with cereals and forage crops (JAAKKOLA and Vogt 1978, Korkman 1978, Sillanpää 1990). However, Zn concentration of crops grown in Finland is almost always below 50 mg kg' 1 , a desirable level in the fodder of ruminants (NJF 1975, Salo et al. 1990. Worldwide, annual industrial consumption of Zn ranks fifth among metals after Fe, Al, Mn and Cu (Kabata-Pendias and Pendias 1984). In Finland, 160 000 tons ofZn is manufactured annually (Tilastokeskus 1992), and 20% is consumed in the domestic markets mainly in galvanization (S. Karlman 1991, Outokumpu Oy, pers. commun.). Zinc is dispersed in the environment as emissions of metal industry and through the use of Zn-containing products. Elevated contents of Zn are found in soils of industrial areas, especially around Zn mines (Bergholm and Steen 1989), smelters (Andersson and Nilsson 1976, Elsokkary and Låg 1978, Miller and McFee 1983, along highways (DeLaune et al. 1989), under electric pylons (Al ■ Hiyaly et al. 1990) and in urban areas in general (Salomons 1984) owing to traffic and combustion of fossil fuels (Cass and Mcßae 1983). Sludge application also gives rise to elevated Zn contents of soil (Wiklander andVahtras 1977, CHRISTIE andBeattie 1989). Atmospheric deposition of anthropogenic origin is considered a major source of Zn input to the soil ofrural areas in southern Sweden and western Norway (ÖBLAD and Selin 1986, Steinness et al. 1989). The annual precipitation in southern and central Sweden and Finland is 100 -140 gZn ha' 1 (Ross 1987, Erviö et al. 1990 and an increasing accumulation ofZn in lake sediments of Finland has been observed during the last 100 years (Myllymaa andMurtoniemi 1986,Verta et al. 1989).
Zinc input into the cultivated soils ofFinland has probably increased over time, but intensified cultivation has elevated Zn uptake by the crop especially in grasslands (Rinne et al. 1974). A decrease of soil Zn concentration in northern Finland has been observed in timothy fields when the same fields were analyzed in 1974 and again 14 years later (ErviÖ et al. 1990).This has been regarded as an indication of gradual decline of plant-available Zn in intensive grassland cultivation. Also farmlevel observations indicate insufficient supply of Zn to crop or domestic animals. Grasslands on peat soils have commonly shown poor growth after 10 years of intensive cultivation, and Zn deficiency has been suggested as an explanation to this (Urvas and Soini 1984). In northern Finland, cattle have exhibited symptoms of Zn deficiency which disappeared with Zn injections. One way of contributing to a sufficient Zn supply to the cattle would be the elevation of the Zn content of forage crops and fodder cereals by Zn fertilization.
Zinc fertilization is in Finland recommended especially to fodder crops (Viljavuuspalvelu 1992). Before 1982, less than 30 000 kg ofZn (below 15 g ha 1 ) was applied annually in mineral fertilizers. The first macronutrient fertilizer (18-3-12% N-P-K) containing also 0.3% Zn was introduced in 1982 and a separate granular Zn fertilizer in 1984. Since 1982, 180 000 -420 000 kg, or 80 -210 g Zn ha' 1 has been spread annually in mineral fertilizers, more than 90% of which incorporated in macronutrient fertilizers (Kemira 1992). The principal areas of fertilizer Zn consumption have been the provinces ofVaasa, Mikkeli, Kuopio and especially the provinces of Oulu and Lappi where 500 g Zn ha' 1 , as compared to 20 -50 g ha' 1 in the southernmost provinces, has been applied annually in chemical fertilizers. Even though field experiments on Zn fertilization have been carried out in Finland, the influence of soil characteristics on the response to applied Zn has not been investigated previously. Neither has the efficiency of different commercial Zn fertilizers been compared.
The purpose of the present investigation was to examine the content and solubility of Zn in cultivated soils ofFinland and the effect of Zn fertilizers on cultivated plants. Information on the soil characteristics controlling the solubility and plant-availability of native and added Zn was sought. The study did not concentrate on soils suspected to be poor in Zn; the soil material collected represented all kinds of cultivated soils ofFinland. The emphasis was in the plough layer, but the vertical distribution of Zn was also investigated. In addition to the characterization of soil Zn by soil analyses, the availability of soil Zn was studied in a pot experiment. The effect of Zn application on the Zn content of forage was examined in pot and field experiments. Also barley, the most important fodder cereal in Finland, was included in the field experiments. The ability of soil analysis to explain Zn uptake by ryegrass and to predict the response of plant Zn concentration to Zn applications was critically studied in pot experiments. In field experiments, the efficiency of different Zn fertilizers and application methods were compared, not forgetting environmental aspects.

METHODS OF ANALYSIS
2.1 Testing the methods of soil Zn determination Soils from a material of 13 cultivated soils (Appendix 1) were mainly used for testing the methods of soil analysis. A few soils from a larger material (Appendix 2) were occasionally used.

Total Zn
In order to determine the total Zn content (Zn to t) in the soil, the solid matrix needs to be dissolved.
Hydrofluoric acid (HF) is required for complete decomposition of silicate minerals, and perchloric acid (HClOa) is a strong oxidizing agent for organic materials. Procedures with and without these hazardous chemicals were tested for the digestion of Zntot.
In the aqua regia procedure (1), a 300-mg soil sample (four replicates) was digested with 4 ml of aqua regia (AR, 1 ml of concentrated HNO3 and 3 ml of concentrated HCI). The sample was heated for 2 hours in a platinum crucible on a hot plate and allowed to react overnight. The next morning the residue was washed with warm dilute HCI into a volumetric flask. In the procedure of Lim and JACK- SON (1982) employing aqua regia and HF (2), a 300-mg soil sample (four replicates) was digested with 4 ml of AR for 2 hours at 200°C in a 100-ml volumetric flask in a sand bath. Thereafter, 5 ml of HF was added and digestion was continued for 1 hour after which 50 ml of saturated H3803 was added to dissolve the possibly precipitated metal fluorides. After cooling, the bottle was filled with deionized water. In the HNO3 -HF -H2SO4 procedure (3), a 500-mg soil sample (four replicates) was digested with 20 ml ofHNO3 in a teflon crucible on a hot plate until dry. Then, 5 ml ofH2SO4 and 15 ml of HF were added and evaporated to dryness. In order to completely remove fluoride, 5 ml of concentrated H2SO4 was added and evaporated to dryness. The residue was washed into a volumetric flask withwarm dilute HCI. From three soils, AR dissolved 55 -70% of the quantity of Zn dissolved by the two mixtures containing HF (Table 1).
In an experiment with 12 surface soils (4,11,23,30,32,35,60,61,67,69,88,105 in Appendix 2), the effect of HCIO4 addition was tested in the HNO3 -HF -H2SO4 procedure. After the digestion with HN03,3 ml ofHCIO4 and 3ml of H2SO4 were added and warmed until fumes evolved and heating was continued for 10 more minutes. Then, HF and H2SO4 were added as described above. Inclusion of the additional digestion phase into the procedure increased the average quantities of Zn extracted from 82.2 to 84.6 mg kg' 1 (+2.9%). According to the paired t-test, the difference was not statistically significant (t = 1,856 n s ), but in further digestions also HCIO4 was added in order to ensure effective oxidation of organic matter. An experiment was carried out to study possible Zn loss and contamination during the digestion procedure. Portions (500 mg) of carefully homogenized mull soil (212) were weighed into nine teflon crucibles. Next, 1) 10 ml of water, 2) 5 ml of a solution containing smgZn dm as ZnSOa • 7H20 (ZnSOq) plus 5 ml of water and 3) 10 ml of the Zn solution were pipetted into three crucibles each. The quantities of Zn added were 0, 50 and 100 mg kg' 1 of soil, respectively. The soil samples were digested according to the HNO3 -HCIO4 -HF -H2SO4 procedure as described above, and the Zn concentration in the digests was determined. No marked net loss or contamination occurred during the digestion (Table 2).

Fractionation of soil Zn
In order to characterize the chemical forms of soil Zn, it is commonly separated into fractions differing in solubility. The fractionation makes the basis for the estimation of potentially mobile Zn reserves and availability of soil Zn to plants. The sizes of the fractions are defined operationally as quantities of Zn which are extracted, often sequentially, with solutions supposed to displace Zn from the exchange complex or dissolve certain soil components resulting in a solubilization of Zn retained by them. The following fractions are commonly distinguished: (1) Zn in the soil solution, (2) exchangeable Zn, (3) adsorbed, chelated or complexed Zn, (4) Zn in secondary clay minerals and insoluble metal oxides, and (5) residual Zn bound by primary minerals (VIETS 1962).
There are no specific extractants for Zn, but various solutions are used for a simultaneous extraction of several elements, e.g. Zn, Cu, Mn, Fe, Al, Ni, Co. The sequential extraction procedures are usually combinations of single extraction methods developed earlier for specific purposes. The most frequently applied procedure in non-calcareous soils utilizes neutral salt solutions (e.g. 0.05 M CaCb, 1 M MgCb) to extract water-soluble and exchangeable Zn, pyrophosphate solutions (0.1 M K4P207 or Na4?207) for the extraction of Znbound by organic matter, and ammonium oxalateoxalic acid solutions for the dissolution of Zn bound by Fe, Al and Mn oxides (sesquioxides). This procedure was first used for the fractionation of soil Cu (McLaren and Crawford 1973) and has later been used also for the fractionation ofZn in non-calcareous soils (Elsokkary and LÅG 1978, IYENGAR et al. 1981, Bjerre and Shierup 1985, Haynes and Swift 1985, LIANG et al. 1990). H202 or NaOCI may be used instead of pyrophosphate for the dissolution of Zn bound by organic matter (Shuman 1979, Nielsen etal. 1986, Sims 1986, Singh etal. 1988. A common feature for all the fractionation procedures is that residual Zn (Znre s), remaining in the soil after removal of oxide-bound Zn and consisting mainly of Zn in the primary minerals, is dissolved with concentrated acids according to the same procedures as used in the digestion ofZntot- The same fractionation procedure is seldom used in more than one study, which complicates the comparison of results. Depending on the research objectives, different fractions are determined. In some studies a fraction of Zn supposed to be specifically adsorbed on inorganic sites has been extracted by 2.5% acetic acid (Elsokkary and LÅG 1978, Iyengar et al. 1981, Bjerre and Shierup 1985orPb(NO,3)2(Liangetal. 1990). Further,Zn bound by Mn oxide has been extracted together with Zn bound by Fe and Al oxides (Elsokkary andLÅG 1978, Shuman 1979) or separately (Sims and Patrick 1978, Iyengar et al. 1981, Miller and McFee 1983, Shuman 1985, Sims 1986, Liang et al. 1990. Zinc bound by poorly crystalline Fe and Al oxides can be extracted separately from Zn bound by crystalline oxides (Miller and McFee 1983, Shuman 1985, Sims 1986 as op-posed to extracting only one fraction, referring to Zn bound by oxide materials. In addition to differences in the extracting solutions, the same soil sample may be used throughout the procedure (Elsokkary and Låg 1978, Shuman 1979, Nielsen et al. 1986 or after the determinationof water-soluble and exchangeable Zn a new sample is weighed for the determination of the more sparingly soluble secondary fractions (Sims andPATRICK 1978, Iyengar et al. 1981).
In the present study, MgCb solution was used for the extraction of water-soluble and exchangeable  Stewart and Berger (1965) and Martens (1968). In those days, concentration of Zn was measured colorimetrically. Since the 1970'5, Zn has invariably been determined by atomic absorption spectrophotometry (AAS) where a high salt concentration in the solution analyzed may cause a high background absorption as well as crusting of the burner. Owing to low concentration of Zn in the extract, dilution as a means ofreducing the salt concentration may not be feasible. Therefore the use of less concentrated salt solutions as extractants would be desirable.
The effect of MgCl2 concentration on the extraction of Zn from eight soils (201,203,204,206,209 and 210 in Appendix 1; 10 and 73 in Appendix 2) was studied. Soil samples (10 g, two replicates) were shaken for 2 hours with 25-ml aliquots of 1 M or 0.5 M MgCh solution. The suspensions were filtered and analyzed for Zn. The means and ranges of Zn extracted (mg kg' 1 ) were as follows: The less concentrated solution extracted 65% (range 53 -74%) of that extracted with the 1 M solution. According to the paired t-test, the difference between the quantities of Zn extracted with the two solutions was highly significant (t = 14.758 ), but the results correlated closely (r = *** 0.99 ). The mean deviation of the two replicates was 5.2%, range 0.3 -22.2%.
The recovery of added Zn was studied on the extract obtained from soil 201 with 0.5 M MgCl2 at the soil-to-solution ratio of 1:5 (weight/volume, w/v). Into three 180-ml portions of the extract, obtained by combining extracts of several subsamples, 20-ml aliquots of water or solution of ZnSOa were added to produce concentrations theoretically differing by 0.05 and 0.20 mg Zn dm' 3 . The measured concentrations (four subsamples) showed accurate recovery of added Zn (Table 3).

h. Zinc hound by organic matter
The use ofpyrophosphate solution as the extractant for Zn bound by organic matter is based on the ability of pyrophosphate to solubilize humic substances (Bremner andLees 1949, Mortvedt andOsborn 1977) and on the ability of pyrophosphate anion to form soluble complexes with Zn Bar-Yosef 1982, Bar-Yosef andAsher 1983). It has been hypothesized that polyvalent cations complexed to organic matter are responsible for keeping organic matter in a flocculated and water-insoluble state. These cations can be removed by complexing with pyrophosphate anion, resulting also in the solubilization of humus (STE-VENSON 1982, p. 40). However, the mechanism responsible for the solubilization of humic substances and cations in the pyrophosphate extraction has not been fully established (BoRGGAARD 1988).
Recovery of Zn added into the pyrophosphate extracts of a mull soil (212) was studied. Soil samples were shaken with 0.1 M K4P207 (pH 10) at the soil-to-solutionratio of 1:25 (w/v) for 18 hours, and the suspensions were centrifuged. Zinc was added to the extracts as has been described earlier. Added Zn was accurately recovered (Table 3).
The commercial K4P207 chemical contained 6 mg Zn kg' 1 resulting in aZn concentration of 0.2 mg dm' 3 in the 0.1 M solution. It is possible to purify the reagent with a solvent extraction (Shuman 1979) or with a cation exchange resin (SHU-MAN 1985). However, the reagent may be used withoutpurification ifZn in the extraction solution remains completely in the liquid phase during the extraction. The influence of Zn in the pyrophosphate reagent was indirectly examined with four surface soils (Appendix 2) by studying the adsorption of added Zn to soil suspended in the 0.1 M K4P207 solution (pH 10). In the experiment, 2.5-g soil samples (four replicates) were shaken for 18 hours in the following solutions: Zinc additions to the extractant corresponded to 4.1 -9.2 mg Zn kg" 1 of soil and were recovered in the extract (Table 4), showing that added Zn was not adsorbed by the soil. The results are in agreement with those of Asher and Bar-Yosef (1982) who observed that at pH 9 Zn was not adsorbed onto a Ca-montmorillonite suspended in a 0.0096 M pyrophosphate solution containing 8.0 mg Zn dm' 3 . It can thus be concluded that Zn of the commercial chemical is not adsorbed either but only gives rise to a high background absorption. The 0.1 M K4P207 was therefore used in the extraction of Zn without purification.

c. Zinc hound by sesquioxides
Since the work of Tamm (1922), acid oxalate solutions have been widely used for the extraction of Fe and Al oxides in soil. In the dark, acid oxalate is supposed to extract only poorly crystalline oxides; in UV light, also crystalline Fe oxide is extracted. The oxalate solutions are assumed to dissolve components occluded into the Fe and Al oxides, and oxalate has therefore been used for the extraction of soil Zn. To avoid crusting of the burner of AAS, a 0.05 M oxalate solution was selected instead of more concentrated solutions commonly used in the fractionation procedures. The extraction ofZn from four soils with 0.05 M oxalate solutions was investigated at pH 2.0, 2.9, 3.3 and 4.0. The pH values were created by different ratios of oxalic acid and ammonium oxalate. Before the oxalate treatment, the samples (2.5 g, four replicates) were extracted with pyrophosphate and washed with water. The remaining samples were shaken for 4 hours with 50-ml aliquots of the four oxalate solutions, the suspensions were centrifuged and the extracts analyzed for Zn. The solution which had the lowest pH was the most efficient extractant for Zn (Table 5), probably owing to a substantial dissolution of structural Zn. There was a considerable decrease in the extractability of Zn in three soils with an elevation of pH from 2.0 to 2.9, but an additional increase in pH affected the results less markedly. A solution of pH 2.9 (0.024 M and 0.026 M in oxalic acid and ammonium oxalate, respectively) was used in the rest of the oxalate extractions.
The recovery of Zn added to the oxalate extract obtained from a mull (212) was studied. Prior to oxalate extraction the soil samples (2.5 g) were shaken with pyrophosphate and washed with water. The remaining samples were shaken in 50-ml aliquots of 0.05 M oxalate (pH 2.9) for 4 hours and the suspensions were centrifuged. Zinc was added to  (Table 3).

d. Repeated pyrophosphate extraction
The effect ofrepeated pyrophosphate extraction on the quantities of Zn dissolved sequentially with oxalate was studied with two soils (211,213). The 2.5-g soil samples (four replicates) were extracted once, twice or three times with 50-ml aliquots of pyrophosphate, washed with water and extracted once with a 50-ml aliquot of oxalate (pH 2.9). In both soils, the first pyrophosphate treatment extracted more than did the second and the third treatment (Table 6). Repeated pyrophosphate extractions tended to reduce the quantities of Zn extracted by oxalate in the mull (213), suggesting that thetwo solutions dissolved Zn to some extent from a common pool. An alternative but less likely explanation is that part of the sample was lost in the successive washings, resulting in a smaller quantity of soil remaining in the oxalate extraction. The number of pyrophosphate extractions did not have a consistent effect on the results of the oxalate extraction in the mineral soil 211. In soil 211, the quantities of Zn extracted by oxalate were substantially larger than those dissolved with the second and third extraction by pyrophosphate. In this soil there was obviously The reproducibility of the results of pyrophosphate and oxalate extractions was studied with 13 soils (201, 202, 206, 207, 208, 209, 210, 211 and 212 in Appendix 1; 10,73,90and 100 in Appendix 2). The 2.5-g soil samples (two replicates) were extracted in sequence with 50-ml aliquots of 0.1 M pyrophosphate, washed with water and extracted with 50 ml of 0.05 M oxalate (pH 2.9). Another two replicates were sequentially extracted and analyzed for Zn a few days later. The results of the first extraction were designated Py 1 and Ox 1, those of the second extraction Py 2 and Ox 2. The differences between Py 1 (5.12 mg kg ') and Py 2 (5.20 mg kg' 1 ) as well as between Ox 1 (2.15 mg kg' 1 ) and Ox 2 (2.17 mg kg' 1 ) were not statistically significant according to the paired t-test. In the 13 soils, the difference between the two means ofZnpy ranged from -0.65 to 0.38 mg kg" 1 and that of Zn 0 \ from -0.30 to 0.25 mg kg' 1 . The coefficient of variation of the four individual determinations in the 13 soils ranged 2.8 -12.4% (mean 5.3%) and 0.6 -13.4% (mean 4.5%) for ZnPy and Zn ox, 0 x, respectively.
In sequential fractionation procedures part of the sample may be lost during the numerous decantations. In order to study the importance of this source of error, 2.5-g samples (four replicates) of 13 soils were sequentially extracted by pyrophosphate for 18 hours, washed with water and extracted by oxalate for 4 hours. After the oxalate extraction, the remaining sample was washed with water, dried and ground, and a 500-mg sample (two replicates) was digested according to the HNO3 -HCIO4 -HF -H2SO4 procedure, and residual Zn (Znres) was measured. Also Zntot (two replicates) was determined. The sums of Znpy , Zn o x and Znres were compared with Zntot (Table 7). In seven soils the sum of fractions was higher(1.1 -13.6 mg kg' 1 , 2 -11%) than Zntot, while in six soils the sum of fractions was lower (0.1 -10.5 mg kg' 1 ,0.1 -10%) than Zntot-The difference between Zntot and the sum of the fractions was not statistically significant (t = 0.897 ns ', paired t-test), showing that no major loss of the sample had occurred. The sum of the fractions correlated closely with Zntot (r = 0.99 ).

Chemical and statistical analyses 2.2.1 Determination of Zn
All the extractions and digestions were performed in duplicate and were repeated when large deviations between the replicates occurred. In filtrations, Schleicher & Shiill 589 3 (Blue ribbon) filter paper was used unless otherwise mentioned. Silicon and polythene stoppers were used in capping centrifuge glass tubes and volumetric flasks because rubber stoppers were found to be sources of soluble Zn. The centrifugations were run for 10 min at 3000 G. Zinc concentrations of the MgCh extracts were measured by atomic absorption spectrophotometry (AAS) using the standard addition method. Zinc concentration of the other extracts was measured by AAS using standard solutions matched for the matrix of the extracts. a. Total Zn and chemically specific fractions ofZn in soil Total Zn. A 500-mg soil sample was digested in a teflon crucible with 20 ml of HNO3 until dry. Thereafter 3 ml of HCIO4 and 3 ml of H2SO4 were added and warmed until fumes evolved. Heating was continued for 10 minutes. Then, 5 ml ofH2SO4 and 15 ml of HF were added and evaporated to dryness. To remove the fluoride, 5 ml of concentrated H2SO4 was addedand evaporated to dryness.
Finally, 20 ml of deionized water and 10 ml of concentrated HCI solution were added, and the mixture was warmed up and washed into a 100-ml volumetric flask.
Water-soluble and exchangeable Zn was extracted by shaking 10 g of soil for 2 hours with 50 ml of 0.5 M MgCl2 in 100-mlpolythene tubes in a reciprocating shaker, centrifuged and filtered.
Extraction with o.l' M K4P207 solution (pH 10) was performed by shaking 2.5-g samples of soil in 50 ml ofpyrophosphate solution in a reciprocating shaker in centrifuge glass tubes. After 1 hour of shaking the suspensions were allowed to stand overnight (16 hours); the following morning the suspensions were shaken for 10 more minutes, centrifuged and filtered. The sample was washed by shaking with 50 ml of deionized water for 1 hour, and after centrifugation the solution was discarded.
Oxalate extraction was carried out sequentially after the pyrophosphate extraction. Oxalate solution (50 ml of 0.026 M ammonium oxalate, 0.024 M oxalic acid, pH 2.9) was added into the centrifuge glass tubes and shaken for 4 hours in daylight, centrifuged, filtered and analyzed for Zn.

h. Procedures used in soilfertility testing
Contrary to the fractionation of soil Zn, determination of Zn in soil testing does not aim at extracting chemically specific fractions. These determinations do not involve assumptions of the bonding mechanism or soil constituent to which Zn is bound but attempt to obtain information on the size of Zn reserves available to plants. The solutions should ideally extract Zn from the same soil components which supply plants with Zn, and the quantities of Zn extracted should correlate with Zn uptake by plants. Mineral acids (HCI), neutral salts (1 M MgCb) and chelating agents (EDTA, DTPA) are commonly used for the extraction of Zn in soil testing. In the present study, methods presented in the literature were applied, and studies on the optimization of the procedures were not carried out.
In the acid ammonium acetate procedure (AAAc), a solution containing 0.5 M CH3COONH4 and 0.5 M CH3COOH was made of acetic acid and NH4OH, and the pH was adjusted to 4.65 with NH4OH or acetic acid. This solution is used in soil testing in Finland to extract Ca, Mg, K and P, and this was also used as an extractant for plant-available Zn by Sillanpää and Lakanen (1966). In the present study, 20 ml of soil was shaken with 200 ml of the extractant in polythene bottles in a planar shakerfor 1 hour, filtered and analyzed for Zn.
In the AAAc-EDTA procedure , the soil is extracted with a solution containing 0.5 M CH3COONH4, 0.5 M CH3COOH and 0.02 M Na2-EDTA at pH 4.65 (Lakanen and Erviö 1971 For the DTPA procedure, developed by Lindsay and NORVELL (1978) and commonly used as an extractant for metallic micronutrients in soil testing, a solution containing 0.005 M diethylenetriaminepentaacetic acid (DTPA), 0.01 M triethanolamine (TEA) and 0.01 M CaCh at pH 7.3 was prepared. In the extraction, 10 g of mineral soil or 10ml of organogenic soil was shaken for 2 hours in polythene bottles, filtered and analyzed for Zn.

c. Zinc in plant material
The Zn content of plant material was determined at Soil Analysis Service Ltd. Plant samples were dried at 60°C and ground. Prior to analysis, approximately 1.0 g of plant material was weighed into tared glass vessels and the exact weight was recorded. The sample was dried at 105°C for 4 hours, cooled and weighed, and the dry matter content was calculated. Simultaneously, another sample (approximately 1.0 g, exact weight recorded) was weighed for dry ashing. The sample was heated in a quartz crucible at 550°C for 4 hours until a white ash was obtained. After cooling, the ash was wetted with a few drops of deionized water, 10 ml of 4 M HCI was added, and the crucible was heated for 30 minutes in a sand bath. The contents of the crucible were transferred quantitatively into a 100-ml volumetric flask which was filled with water. After the remaining solid material had settled to the bottom, a 10-ml sample was taken and the Zn concentration of the solution was determined by AAS. The consistency of the results was checked by including one standard hay sample in every set of 40 samples. In 35 determinations, the mean Zn concentration of the standard sample was 22.5 mg kg' 1 , range 19-31 mg kg' 1 , standard deviation 2.56 mg kg' 1 and coefficient of variation 11.4%.In addition, in one determination the standard gave a value of 42 mg kg' 1 , probably due to contamination at some point, resulting in repeated determinations of several samples of the set. Routinely, a duplicate determination was carried out with every eight samples.

d. Zinc in fertilizers
Fertilizer samples were dissolved with aqua regia to facilitate the determination of Zn. A sample (1.00 g) of ground fertilizer was weighed into a 250-ml beaker, and a few drops of water, 10ml of concentrated HNO3 and 30 ml of concentrated HCI were added and evaporated to dryness. After cooling, 10 ml of concentrated HCI was added and evaporated to dryness. Thereafter, 50 ml of water and 10ml of concentrated HCI were added and boiled until the precipitation was dissolved. The solution was poured into a volumetric flask, filled and filtered if necessary and analyzed for Zn.

Other analyses
The soil organic carbon was determined by a modified wet digestion method (Graham 1948). It was assumed that 80% of the carbon was oxidized in the treatment. The organic matter content was calculated by multiplying the organic C content by 1.724. The particle size composition of the mineral material in the soil samples was determined by the pipette method (Elonen 1971). The determination was made for all mineral soils (organic matter content less than 20%) and for most mull soils (organic matter content 20 -40%). Soil pH was measured in water (20 ml of soil, 50 ml of water) after 2 hours of equilibration. The bulk density of ground (<2 mm) soil was determined by weighing two 50-ml samples of soil.
Poorly crystalline Fe and A 1 oxides were extracted with 0.05 M oxalate solution at pH 3.3 (Niskanen 1989).The concentrations ofFe and A 1 in the extracts were measured by AAS. The airacetylene flame and acetylene-nitrous oxide flame were used in the determination of Fe and Al, respectively.
In order to determine the content of Zntot in textural fractions, clay (<0.002 mm) and silt plus very fine sand (0.002 -0.05 mm) fractions were separated in three soil samples. The flocculating and cementing agents were first removed with hydrogen peroxide (H202) and by a treatment with sodium dithionite (Na2S2oa) and sodium citrate using NaHCOs as a pH buffer (pH 7.3) (Olson and Ellis 1982). The actual separation of the textural fractions was carried out by an automated procedure for the gravity sedimentationdecantation technique (Rutledge et al. 1967) with the equipment ofTexas A&M University (College Station, Texas, USA).
The contents ofK, NH4 + -N and NO.V-N in fertilizers were determined after dissolution of the fertilizer (20 g) with water (1 dm 3 ). The sum of NH4 + -N and NOs'-N was taken as the concentration of total N. For the determination of total P, the fertilizer was dissolved in H2SO4 and HNO3. The pH of the fertilizers was measured in a 10% solution (w/v). The concentration of N in barley grains was measured by the near-infrared (NIR) technique. The vegetative parts were analyzed for N by the Kjeldahl method.

Statistical methods
The statistical analyses were mainly carried out according to the procedures presented by Ranta et al. (1991). The variation of the results was studied by calculating the mean deviations, MD = Xlxj-pl/N, or standard deviations, s = V| X(x, -|i) 2 ]/(« -1). In assessing the variation of replicates, the mean deviations were calculated. Otherwise, standard deviations were used.
Fractiles of 10% (Fio%) and 25% (F25%, quartiles) were formed in order to group the material in terms of various characteristics. Occasionally there were several equal results, all of them placed in the same fractile. Therefore the ultimate sizes of the fractiles may deviate from F25% and Fio%.
Means obtained from two populations were studied with the t-test. The effect of a treatment on several soils was studied by the t-test for paired measurements. Analysis of variance was used to test statistically significant differences between several means. The significance of the differences between individual means was tested by Tukey's test (HSD, P = 0.05). Means marked with the same superscript do not differ at P = 0.05.
The correlation between different characteristics of a population was studied. Owing to the skewed distribution of the results of several variables, logarithmic transformations (natural logarithms, loge) of results were commonly used. In addition to the linear correlation coefficients, Spearman rank correlation coefficients were calculated. The z-transformation test was applied to test the differences between the linear correlation coefficients.
Regression analyses were carried out in order to study the dependence between variables. Statistically not significant (P = 0.05) independent variables were excluded from the regression equation one by one and the equation was recalculated until all variables in the equation were significant. The significance of the multiple determination of the regression equations was tested by the F-test. The significance of each regression coefficient (b) was tested by the t-test. To depict the relative importance of statistically significant variables, the standard partial regression coefficients, or beta coefficients ((3), were calculated as follows (Steel and Torrie 1981): b' =b ■ si/sy, where si = standard deviation of an independent variable and s y = standard deviation of the dependent variable.

Experimental soils
Most laboratory studies and two pot experiments were carried out with soils of a material of 107 samples which were collected from plough layers (Ap horizons) of cultivated fields in differentparts of Finland. The samples were collected in 1987and 1989 to represent the distribution of soil classes of cultivated soils of Finland as reported by Kurki (1982). The samples were taken from rural areas at least 100 m away from roads and electric wires. Moist soil samples were air-dried at room temperature and stored in plastic bags. Part of the sample was ground to pass a 2-mm sieve and stored in a cardboard box. Mineral soils except moraines were designated according to the textural classification of Juusela and Wäre (1956). Clay refers to the fraction finer than 0.002 mm, silt to 0.002 -0.02 mm, very fine sand to 0.02 -0.06 mm and fine sand to 0.06 -0.2 mm. Mineral soils containing >30% of clay are called clay soils. Occasionally, other mineral soils are collectively called coarse mineral soils. Organogenic soils (organic matter content >20%) were divided into mull and peat soils, with organic matter contents of 20 -40 and > 40%, respectively. The soil characteristics are presented in Appendix 2 and summarized in Table 8. The vertical distribution of Zn was studied on seven soil profiles of cultivated fields (Appendix 3). Soil profiles were sampled according to visible horizon boundaries if present. Where the subsoil was apparently homogeneous, the samples were taken to represent 20-cm thick layers. Also 15 pairs of samples taken from the plough layer (A p horizon) and the respective subsoil (30 -35 cm) (Appendix 4) were investigated. The two samples of the pair were usually of the same soil class. In one case, an organogenic soil had a fine sand subsoil and one case was the contrary. there was a very high content of Zntot (420 mg kg' 1 ), while the bulk of the results ranged rather uniformly between the second highest (202 mg kg' 1 ) and the lowest value (10.3 mg kg' 1 ). The ranges ofZntot overlapped markedly in the five soil classes (Table 9). However, the average Zntot in the clay soils was higher than that of silt, loam and very fine sand soils, and excluding soil 71, the Zn tot in fine sands and moraines was still lower. The mean Zntot of the two organogenic soil classes did not differ significantly from one another or from the fine sand and moraine soils but was lower than that of the more fine-textured mineral soil classes. The mean deviation of the individual measurements of a soil sample averaged 1.8 mg kg' 1 , i.e. 2.8% of the mean, range 0 -17.6% of the mean. The mean deviation exceeded 5% in 18 soils which were poor in Zntot.
The results of Zntot were divided into quartiles (F25%) ( Table 10). The frequency of the different soil classes in each quartile demonstrates the abundance of Zntot in clay soils and the small re-serves of Zntot in the coarsest mineral soils and organogenic soils. As many as 19 out of 25 clay soils occurred in the largest F 25%, six out of the seven heavy clay soils (clay > 60%) containing more than 150 mg Zntot kg' 1 . The smallest F25% contained 10 fine sand and moraine soils, and in these soils more than half (57 -91%) of the mineral material was coarser than 0.06 mm (i.e. fine sand or coarser). However, the peat soils were poorest in Zntot. As many as 13 out of the 20 peat soils occurred in the smallest F25%.
When the results were transformed into milligrams per dnv of soil, the difference between mineral and organogenic soils became greater. The average Zntot was 31 and 14 mg dm in mull and peat soils, respectively, while the results of the mineral soils were not altered by the transformation. The present material consisted of 11 Carex peat soils; eight of them contained less than 10 mg 3 3 Zntot dm of soil. The lowest Zntot (3.2 mg dm ), occurring in a Carex peat soil of Sotkamo (97), corresponded to 6kg ha' 1 in a 20 cm deep plough layer, while clay soils commonly contained as much as 300 kg Zntot ha' 1 .
Linear correlation coefficients were calculated between Zntot (mg kg' 1 ) and some soil characteristics (Table 11). In mineral soils, excluding soil 71, Zntot correlated positively with clay content and negatively with the content of coarse mineral material (>0.06 mm). There was also a positive correlation between Zntot and Fe 0x, but the partial correlation between these two variables was not statistically significant when the effect of clay was eliminated. The dependence of Zntot on clay content in mineral soils is presented in Figure 1. Some soils (17, 21, 38, 71) contained more Zntot than the other soils of the same clay content. Inversely, the gyttja clay 12 (Vihti) and the silty clay 13 (Perniö) contained less Zn than expected. In soil 12 the organic C content (9.5%) was higher than average, resulting in a smaller quantity of mineral material and lower Zntot in the sample weighed for the analysis. The low Zntot of soil 13 may be due to the abundance of coarser materials which are usually poor in Zntot-In soil 13, 33% of the mineral material was coarser than 0.02 mm (very fine sand and coarser), while the average of these materials in clay soils was only 20%.
In organogenic soils, there was a negative correlation between Zntot and organic C. The content of mineral material is inversely reflected in the organic C content and the correlation between Zntot and organic C thus indicates the fact that mineral material was richer in Zntot than was organic matter. Figure 2 shows that all the soils which contained more than 35% organic C exhibited a very low concentration of Zntot-The results of Zntot were further studied by multiple regression analyses. In mineral soils Zntot (mg kg' 1 ) increased with increasing content ofclay and silt (%) and Al ox (mmol kg' 1 ). According to the regression analysis, an increase in organic C (C, %) was coincident with the decrease in Zntot even though there was no significant linear correlation between these two variables. The t-test of the regression coefficients and beta coefficients C|3) of the independent variables showed that clay content was by far the dominant variable (Table 12). The regression equation was as follows (n = 72): Zn,ot = 1.79 Clay + 0.66 Silt -7.90 C + 0.20 Alox + 42.14 9 *** R 2 = 0.74 In organogenic soils, the dependence ofZntot on different soil characteristics was weaker. In no -2 equation did the coefficient of determination (R ) exceed 0.30 when the results were expressed as milligrams or millimoles per kilogram of soil. However, when the regression analysis was carried out with results transformed into milligrams ormillimoies per dm of soil, a higher coefficient of 2 multiple determination (R ) was obtained. According to the following equation (n = 34), Zntot (mg  The organic C content, being relatively the more important variable (Table 12), inversely reflects the abundance of mineral material in the soil, and Alox may stand for the abundance of aluminosilicates. There was a negative correlation between organic C and Alox (r = 0.48 ), showing that the Alox decreased with decreasing mineral material, which obviously also resulted in a decrease of Zn tot.

b. Particle size fractions
Of three silty clay soils (204 and 205 of Appendix 1; 10 of Appendix 2), clay and silt plus very fine sand were separated and analyzed for Zntot-The textural fractions separated had probably lost at least part of the secondary Zn during the pretreatments with hydrogen peroxide, citrate and dithionite. Therefore, the Zntot of the soil fractions actually gives the quantity of Zn contained in primary minerals, while the results of the whole soil contained primary and secondary Zn. The clay fraction contained more Zntot than did silt plus very fine sand (Table 13). In soil 205, the clay fraction was slightly poorer in Zntot than in soil 204, while the  silt plus very fine sand of soil 10was richer in Zntot than this fraction in the other two soils. The results of these analyses confirmed that the clay content largely determines the concentration of Zntot in a mineral soil but also silt and very fine sand seem to contribute substantially to Zntot.

Fractions of soil Zn
In the fractionation of soil Zn, water-soluble and exchangeable Zn (Zn ex ) were extracted by MgCI;;. A new sample was weighed for the sequential extraction with pyrophosphate and oxalate which were assumed to dissolve Zn bound by organic matter (Zn py ) and sesquioxides (Zn" x ), respectively. Residual Zn (Zn re s) remaining in the soil after the oxalate extraction was obtained as the difference of Zntot and Znpy + Znox .

a. Water-soluble and exchangeable Zn
Water-soluble and exchangeable Zn (Zn ex ) ranged from 0.3 to 22.0 mg kg' 1 (Appendix 5). The mean Zn ex (Table 14) did not differ markedly from one mineral soil class to another, but the means in the two organogenic soil classes were considerably higher than those of the mineral soil classes. When studying the results expressed as milligrams ofZn per dm 3 of soil, the difference between the means of the five soil classes were not statistically significant.

b. Zinc bound by organic matter and sesquioxides
In 106 soils, Zn extracted with pyrophosphate (Znpy) ranged from 1.4to 53.8 mg kg" 1 . In addition, in soil 71 rich in Zntot there was also plenty of Znpy (227 mg kg' 1 ) (Appendix 5). The mean Zn py was higher in peat soils than in the other four soil classes (Table 15). However, after transforming the results to milligrams per dm of soil, statistical differences between the soil classes were nonexistent with means of 4.9 and 5.3 mg dm in the mull and peat soils, respectively.
Soil Zn extracted by oxalate (Zn ox ) sequentially after the pyrophosphate treatment ranged in 106 soils from 0.5 to 13.0 mg kg' 1 , while soil 71 contained 115 mg Znox kg' 1 (Appendix 5). The mean Znox (Table 15) was higher in clay soils than in fine sand and moraine soils or in organogenic soils. When expressing the results as milligrams ofZn per dm ofsoil, the organogenic soil classes were by far poorer in Znox than were the mineral soils with means of 1.2and 0.9 mg dm in mull and peat soils, respectively. In nine out of the 25 clay soils, Zn ox was higher than Zn py . In soils other than clay, Zn ox was lower than Znpy with only two exceptions; in organogenic soils Zn py was four to five times higher than Znox . The average mean deviation of the replicates of Znpy was 0.33 mg kg' 1 , or 5.5% of the mean and ranged from 0 to 18.9% of the mean in the 107 surface soils. The mean deviation ofZnox averaged 0.16 mg kg' 1 , or 4.8% of the mean, ranging from 0 to 25.2%. The mean deviation exceeded 10% in 15 and eight soils in the determination of Znpy and Znox , respectively.
The lowest results of Znpy (Fn%) were distributed over all mineral soil classes, while those of Znox (Fio%) occurred in fine sand and moraine soils (eight soils) as well as in organogenic soils (three soils). There was no soil in common to the smallest Fn% of Znpy and the smallest Fio% of Zn ox . The mineral soils poorest in Znox were characterized by a coarse texture and low Feox content. In the eight fine sand and moraine soils of the smallest Fio% of Znox the average concentration of Fe ox was 22.0 mmol kg' 1 which was less (t = 3.598 ) than in the rest of the fine sand and moraine soils (mean 62.3 mmol kg 1 , n = 20). The highest results (Fio%) of Zn Py occurred with three exceptions in organogenic soils, while those of Zn Gx (Fio%) were the most common in clay soils.

c. Complexed Zn
Water-soluble and exchangeable Zn extracted with MgCh was probably included also in Zn extracted by pyrophosphate (Zn py ). Therefore the results of the pyrophosphate extraction can be divided into two parts: (1) water-soluble and exchangeable Zn bound by non-specific electrostatic forces and (2) Zn presumably bound mainly by organic matter in complexed forms by covalent forces. The quantities of complexed Zn were calculated as the difference of Zn py and Zn ex (i.e. Zn py -Znex) (Table 16). This fraction was larger in peat soils than in the mineral soil classes but, again, there were no differences between the soil classes when the results were ex--3 pressed as milligrams of Zn per dm . The percentage of Znpy which was bound by covalent forces, i.e. the ratio 100 • (Znpy -Zitex)/Znpy , ranged 14 -90% (Table 16) and correlated with soil pH (r -0.65* ). The fractile (Fio%) of the smallest percentage (14 -38%) ofZn py in complexed forms consisted of six organogenic soils (pH 3.8 -5.6) and of five coarse mineral soils (pH 4.2 -5.8). The four fine sand and moraine soils of this group had only a moderate acidity (pH 4.9 -5.8) and a very low content of clay (<4%) as well as Fe ox and Al ox. Three of these soils (72,73,74) were those of the coarsest texture in the whole soil material. For comparison, the most acid clay soils (10, 20, 26) had nearly the same pH (4.6, 5.2, 5.0, respectively) but a higher percentage (52 -62%) of Znpy was not water soluble or exchangeable, indicating a difference in bonding of Zn in acid clay soils and acid fine sands.

d. Residual Zn
Residual Zn (Znre s) (Table 15) represents the fraction bound in the mineral lattices. This fraction was the largest in clay soils and decreased in mineral soils with decreasing clay content; 11 soils richest in clay were also in the largest F25% of Znres. The lowest results occurred in organogenic soils; 15 out of 20 peat soils, especially those of the highest organic C content, and five out of 14 mull soils were in the smallest F25% of Znres-The smallest P25% also contained seven fine sand and moraine soils which were very coarse in texture, containing only 1 -7% clay and 3 -13% silt. The pH of these seven soils (4.9 -6.3, mean 5.5) was lower (t = 3.295 ) than in the rest of the fine sand and moraine soils (mean 6.2). When the results of Znre s were transformed into milligrams per dm of soil, the difference between organogenic and mineral soils became greater with '1 means of 25.4 and 8.0 mg dm' for mull and peat soils, respectively, while the transformation did not have a marked influence on the results of mineral soils.

e. Relationship between Zn fractions and other soil properties
Owing to the skewed distribution of the results of secondary Znfractions (Znex , Znpy , Znpy -Ziiex and Znox), the linear correlation coefficients were calculated using the natural logarithms (loge) of the results (Table 17). In order to eliminate the effect of the skewness of the material, also the Spearman rank correlation coefficients were calculated between the various Zn fractions and soil properties. The two correlation coefficients were similar and only the linear ones are presented. Soil 71, extremely rich in Zn, was not included in the calculations.
In mineral and organogenic soils, Znex correlated negatively with soil pH. The correlation was made weaker by a few soils which in spite of a pH above 6.5 contained plenty of Zn e x-Znex correlated closely with Znpy , but it should be taken into account that Znex is actually a part ofZnpy; thecorrelation between Znex and complexed Zn (Zn py -Zn ex ) remained lower than that between Zn ex and Znpy . Organic C content seemed to correlate with Zn ex , reflecting the lower Table 17. Linear correlation coefficients between Zn fractions and some other soil properties. Tabulations were carried out using natural logarithms of concentrations of Zn, Fe and Al, expressed as mg kg-1 (Zn) or mmol kg l (Fe, Al).

Zn"
Znpy and -0.44 in mineral and organogenic soils, respectively). Soils rich in Zn py tended also to be rich in Zn ox-Znox and Znres correlated closely with Zntot in mineral soils but more weakly (z = 2.219 and 12.642 for ZnGx and Znre s, respectively) in organogenic soils. In mineral soils Znox correlated with clay, Feo x and Znres-Contrary to mineral soils, Znox did not correlate with Znres or Feox in organogenic soils but there was a significant negative correlation between Znre s and organic C.
The relationships between Zn ex and other fractions of secondary Zn were further studied by regression analyses using the natural logarithms of Zn concentrations. Both in mineral and organogenic soils Znex (mg kg' 1 ) increased with increasing Znpy (mg kg 1 ) and/or decreasing soil pH. Even though there was a statistically significant linear correlation between Znex and ZnQx in organogenic soils, ZnQ x remained insignificant in the regression analysis, owing to the positive correlation between Zn 0x and Znpy. The t-values and the beta coefficients (Table 18)  loge Znex = -0.54 pH + 0.66 loge Zn P y +1.13 R 2 = 0.84 Dependence of Znpy , Znpy -Zn ex and Zn Q x on soil characteristics was studied by regression analyses. The regression equations consisting of Zntot, organic C, pH, Fe 0 x, Alo x and in mineral soils clay content explained less than 10% of the variation of Znpy and Znpy -Znex. Instead, a considerable part of the variation ofZnox (mg dm" ) was explained by Zntot (mg dm (n = 106): loge Zriox = 0.65 loge Zntot -2.00 9 *** R =0.71 /. Distribution of soil Zn into differentfractions The distribution of soil Zn into different fractions, expressed as percentages of Zntot, was studied (Table 19). In clay and silt soils the residual fraction (Znres) commonly accounted for more than 90% of Zntot-The distribution of Zn in the mineral soil classes did not differ significantly from one class to another even though in a few coarse mineral soils a larger part of Zn was in the secondary fractions Table 18. t-Values of the regression coefficients and beta coefficients (3) of the independent variables explaining the variation of log Zn" in mineral and organogenic soils.

Mineral soils
Organogenic soils variable~t P t P pH -10.550*" -0.52 -6.826*" -0.51 (Zriex, Zripy -Zn e x, Zn 0 x) than in the rest of the mineral soils. In mull soils the percentage of Znre s was significantly smaller than in the two most finetextured mineral soil classes. In peat soils the percentages of the secondary fractions were substantially greater and Znres smaller than in the other soil classes. In the 13 soils richest in organic C (34 -50%) the sum Znpy + Zn o x corresponded to as much as 68 -88% of Zniot-In mineral soils, the ratio (Zn py + Znox)/Zntot which reflects the relative abundance of secondary fractions did not correlate with soil properties, but in organogenic soils there was a close correlation between this ratio and organic C (r= o.Bl***).
Fractions of Zn were studied separately in mineral and organogenic soils in which Zn py exceeded 10 mg kg" (mineral soils) or 20 mg kg" (organogenic soils) (Table 20). In these soils, the secondary Zn fractions ranged from 19% in clay soil 17 to 88% in peat soil 102 and were relatively larger than those in the respective soil class on average. Inquiries concerning the farming operations revealed that large quantities of farm-yard manure had been spread to the fields from where soils 53,70, 85 and 95 originated (V. Haataja, P. Luoma and J. Nieminen 1991,Kemira Oy, I. Kallioniemi 1991, Suomen Säästöpankki, Pori, pers. commun.). It is likely that the abundance of secondary Zn also in soil 17 can be attributed to the use of farm-yard manure, because the respective field was located at Viikki 0.2 km from the barn of the farm of the University of Helsinki. Soil 71 of Harjavalta had been taken about 2 km from the smelter of Outokumpu Oy. No slag or farm-yard manure had been transported to the field for several years (I. Kallioniemi 1991, Suomen Säästöpankki, Pori, pers. commun.), sug- Table 20. Soil Zn py and Zn ox as well as the ratio (%) between secondary and Zn tol , i.e. 100 • (Zn py + Zn ox )/Zn tot , in soils exceptionally rich in Zn py .
gesting that the abundance of Zn was airborne. Information of the farming of soils 74, 101 and 102 was not available.

Zinc extracted by AAAc-EDTA
In soil testing in Finland plant-available Zn is extracted by AAAc-EDTA at pH 4.65. Therefore, this method was applied also to the present surface soil material, and relationships between Zn extracted by AAAc-EDTA (ZnAc) and chemically more specific fractions of soil Zn were investigated. Soil ZnAc ranged from 0.6 to 165 mg dm' (Appendix 5). The highest result occurred for the same fine sand soil (71) which was rich in Zn according to all indices determined. The second highest result was 29.9 mg dm' and the median of the whole material was 2.9 mg dm . The means of ZnAc did not differ statistically significantly from one soil class to another (Table 21). The mean deviation of the replicates averaged 0.25 mg dm' , or4.4% of the mean, range 0 -20.8%. Soil 71 excluded, the mean deviation averaged 0.15 mg dm' 3 . ZnAc corresponded on average to 45% (range 18 -76%) and 75% (range 43 -200%) of the sum Zn py + Zn 0x (mg dm' 3 ) in mineral and organogenic soils, respectively.
The 15 soils (Fi4%) poorest in ZnAc 1.5 mg dm' 3 ) consisted oftwo clay soils, 10coarse mineral soils and three organogenic soils. The pH of the 12 mineral soilsof this group was 5.4 -6.9, and in nine of these soils, the pH was above or equal to 6.0, which was the mean pH for mineral soils. There in mineral and organogenic soils, respectively). The correlation was statistically highly *** *** significant also with Znex (r = 0.76 and 0.79 ) and Zn o x (r = 0.61 and 0.60 ). In the organogenic soils there was a significant correlation also * between ZnAc and Zntot (r = 0.42 ). Like Zn py , ZnAc did not correlate with soil pH or with the content of clay or organic matter. Despite the highly significant correlation of ZnAc both with Zn py and ZnQx, Zn py (mg dm" When the regression analyses were tabulated separately for mineral and organogenic soils, slightly different equations were obtained for the two soil groups. In organogenic soils, Zn py alone explained 89% of the variation of ZnAc (equation not shown); other soil characteristics were not significant. In the equation for mineral soils (n = 72), the contents (%) of clay and organic C were also significant variables, while Zn py (mg dm" 3 ) was relatively the most important: loge ZnAc = 0.91 loge Znpy + 0.0064 Clay The above equation shows that at a given level of Zn py, increasing clay content enhanced the extraction power of AAAc-EDTA. The negative regression coefficient of C suggests that, at a given Znpy level, increasing organic C reduced the extraction power of AAAc-EDTA in relation to that of pyrophosphate.
-1 Also the relationships between ZnAc (mg dm" ) and Zn ex (mg dm" ) were studied by multiple regression analyses. According to the beta coefficients (P) (Table 22), Znex was relatively the most important variable explaining the variation of ZnAc both in mineral and organogenic soils. The equation below shows that the relative efficiency of AAAc-EDTA to extract Zn increased with increasing soil pH. In mineral soils, also the clay content increased the efficiency of AAAc-EDTA as an extractant for soil Zn as compared to MgCla. The equations were as follows:

Vertical distribution of soil Zn
Plant roots penetrate farbelow the plough layer and therefore Zn reserves also deeper in the soil may be important in providing the plant with Zn. Therefore, the vertical distribution of Zn was studied by investigating soil profiles as well as sample pairs consisting of a sample from the plough layer and the subsoil.

Total Zn
The seven profiles differed greatly in the content of total Zn (Zntot) (Table 23). In profiles PI, P 3 and P 6, where the entire profile or a part of it consisted of clay soil, the peak Zntot exceeded 150 mg kg 1 .
In turn, in profile P 4 dominated by fine sand as well as in the Carex peat profiles P 5 and P 7, all layers contained less than 50 mg Zntot kg' 1 , most layers even less than 10mg kg" 1 . The Zntot of organogenic layers was dependent on the Zntot of the mineral subsoil. This was demonstrated in profile P 7 which had a Carex peat topsoil very poor in Zntot and a fine sand subsoil which also was poor in Zntot. On the contrary, in profile P 6 the soil below the organogenic surface horizons was heavy clay rich in Zntot, and also the surface horizons were richer in Zntot as compared with the other two organogenic soil profiles.
The content of Zntot showed marked changes in relation to the depth within each soil profile. The highest Zntot occurred either in the plough layer or m the deepest horizon sampled. The lowest /.nun occurred between these horizons, with the exception of profile PI (silty clay) and P 3 (silt) where it was in the two or three uppermost horizons sampled. In the other five profiles (P2, P 4, P 5, P 6, P 7) Zntot was markedly higher in the plough layer than in the next layer below. In all profiles but one (P5) was the content of Zntot in the bottom of the profile at least as high as that in the plough layer. Mineral soil had obviously been mixed in the plough layer of Carex peat profile P 5, increasing the Zntot in the plough layer of this profile. An evidence of the external source of the mineral matter was the presence ofclay (9%). Moreover, the uppermost layer contained only 9% organic C, while in the layer below there was as much as 52% of organic C.
The Zntot of the mineral soil profiles PI, P 2 and P 3 correlated closely with the clay content (r = 0.89** -o.99***); in profile P 4, Zntot correlated (r = 0.81 ) with very fine sand which was the finest textural fraction present in abundance. In the organogenic profiles P 5 and P 7, Zntot correlated ** negatively with organic C, r = -0.92 and r = -0.77 ns ', respectively.
former soil surface currently covered by organic materials. In the three mineral soil profiles (PI, P 2 and P 3) dominatedby clay or silt there was a tendency that deeper in the soil profile ZnAc first decreased, increasing again in the deepest horizons. In profile PI, ZnAc was highest in the deepest layer sampled. On the contrary, in the coarse mineral soil profile (P4) and in the two Carex peat profiles there was no increase of ZnAc in the deeper layers.
In 14 out of the 15 plough layer and subsoil sample pairs, the subsoil was poorer in ZnAc than was the respective plough layer (Table 24, details in Appendix 4). According to the t-test for paired measurements, the difference in ZnAc between the plough layer and subsoil was highly significant (t = 4.804***, n = 15). The only exception was the sample pair from Forssa which consisted ofa mull plough layer and a subsoil of heavy clay which was richer in ZnAc-In the other sample pairs, ZnAc in the subsoil averaged 39% of that in the plough layer. Only in heavy clay subsoils did ZnAc exceed 2.0 mg dm' 3 .

Zinc extracted by AAAc-EDTA
Within a soil profile, the sampled layers differed markedly from one another in terms of Zn extracted by AAAc-EDTA (ZnAc , Table 23). In all profiles except P 5 and P 6 was the content ofZnAc higher in the plough layer than in the next few underlying ones. In the Carex peat profile P 5, all the horizons were equally poor in ZnAc-In profile P 6, consisting of an organogenic surface horizon and a heavy clay subsoil, the maximum ZnAc was measured in the upper part of the mineral subsoil which is the 3.4 Extractability of Zn added to soil The ability of common extractants to dissolve Zn from soil was studied on four surface soils (18,34,78,104). Deionized water or ZnSOa solution was added to weighed portions of air-dry soil (three replicates, 200 ml each), and the samples were incubated at an approximate field capacity for two weeks after which the samples were air-dried and ground with a porcelain mortar and a pestle. The soil samples of the first incubation (Zn addition 3 9.06 mg dm ) were extracted (two replicates) with the following solutions: 1.0.5 MMgCla 2. 0.5 M ammonium acetate -0.5 M acetic acid at pH 4.65 (AAAc) 3.0.5 M ammonium acetate -0.5 M acetic acid -0.02 M Na2-EDTA at pH 4.65 (AAAc-EDTA) 4. DTPA -TEA -CaCh at pH 7.3 (DTPA) The unbuffered MgCl2 solution extracted exchangeable Zn (Zn e \) at soil pH, AAAc at a constant pH of4.65. The results of AAAc-EDTA demonstrate the effect of the addition of a chelating agent. The use of DTPA allows comparison to be made with an extractant containing the other widely used chelating agent.
The primary purpose of the second incubation experiment (Zn addition 9.26 mg dm ) was to study the extractability of added Zn in sequential extraction by 0.1 M K4P207 and 0.05 M oxalate. In both experiments, the results of extractions for which the samples were weighed (MgCb, DTPA, pyrophosphate and oxalate) were transformed to milligrams per dm 3 of soil by multiplying the results by the bulk density determined for the incubated soil samples.
As far as native Zn is concerned, MgCb was the weakest extractant in clay (18) which was the least acid (pH 6.2) of the four soils, while in the three more acid soils MgCl2 and AAAc extracted equal quantities of native Zn (Table 25). In the clay soil, the higher efficiency ofAAAc was probably due to its pH (4.65) which is 1.5 units below the soil pH. The acidity may have caused dissolution of Zn reserves which would not be exchangeable at native soil pH. From the other three soils AAAc probably extracted only the exchangeable Zn fraction.
Addition of EDTA to the AAAc solution markedly enhanced the extraction of native Zn. In clay (18) and organogenic soils (78, 104) the quantities dissolved were at least tripled, and in the highly acid loam (34) they increased by 50%. DTPA was a weaker extractant than AAAc-EDTA in the three soils except the peat (104) where DTPA and AAAc-EDTA extracted native Zn with an equal efficiency. The sequence of efficiency in mineral soils was thus: In organogenic soils the sequence was as follows: MgCli = AAAc < DTPA < AAAc-EDTA Table 25. Soil Zn (mg dm 3 ) extracted by MgCl 2 (Zn ex ), AAAc (Zn AAAc ), AAAc-EDTA (Zn Ac ) and DTPA (Zn DTPA ) from soil samples incubated with or without added Zn. The percentage of added Zn which was extracted appears in parentheses.
Zn + = Zn added (9.06 mg dnr 3 of soil), Zn-= native Zn. 1 In each soil, the results of the samples incubated with and without added Zn were tested separately for each extraction. 1 In each soil, the results of the samples incubated with and without added Zn were tested separately for each extraction.
Adsorption of added Zn into the non-exchangeable form was observed especially in the clay soil (18) where as much as 88% of added Zn was not extracted with MgCb. This tendency was weaker in the loam (34) in which more than half of added Zn remained exchangeable. In mineral soils added Zn was more efficiently extracted by AAAc than by MgCb but in the organogenic soils these two solutions extracted added Zn equally effectively. Despite the low pH of the organogenic soils, a considerable part of added Zn was adsorbed into forms not dissolved by MgCla or AAAc. This may be due to the formationof organic complexes, and the extraction of this fraction seemed to require chelating agents (EDTA and DTPA).
AAAc-EDTA dissolved 81 -92% of added Zn, being thus the strongest extractant for added Zn in the two mineral soils and in the mull. In peat, AAAc-EDTA and DTPA were equal in efficiency. Despite its high pH, DTPA was a strong extractant for native and added Zn in organogenic soils, but in mineral soils DTPA was relatively less effective. As for added Zn, the results of AAAc-EDTA extraction were less affected by soil characteristics than those of the other extractions.
Pyrophosphate extraction removed the bulk of added Zn from soil, the rest being dissolved by oxalate (Table 26). The residue of added Zn in the soil after pyrophosphate extraction seemed to be highest in the clay soil and lowest in the two organogenic soils. The recovery of added Zn was not affected by soil pH. The apparent recovery ofadded Zn seemed even to exceed 100% in all soils. The recovery was calculated as the difference between the sums ofZn py + ZnGx extracted from soils incubated with and without added Zn. The confidence limits at the 95% level were 0.2 -0.3 mg dm' 3 for Znpy and 0.1 -0.2 mg dm' 3 for Zn ox* The present results were within these limits.

Total Zn
The Zntot in the surface soil material was higher than the results of spectrographic determination of Zntot published earlier in Finland (Vuorinen 1958, Erviö and Virri 1965, Sippola 1974). For example, in the study by Sippola (1974) (Tjell and Hovmand 1978), clay, loam and sandy soils of Germany (Baghdady andSippola 1983, Lichtfuss andAndresen 1983) and clay loam soils of various parts of Canada (Bishop and MacEachern 1973, Nielsen etal. 1986, Liang et al. 1990, are similar to or slightly smaller than those in the respective textural classes of the present material. Clay and clay loam soils of warmer regions, e.g. Virginia and Georgia, USA (IYENGAR et al. 1981, Shuman 1985 and the Nile delta, Egypt (Baghdady and Sippola 1983), have also had similar or slightly lower Zntot than occurs in texturally corresponding soils ofFinland. However, in unpolluted soils of other countries, Zntot seldom exceeds 150 mg kg' 1 , a value commonly found in clay soils of Finland. In coarse sandy soils of warmer climates, i.e. Georgia, USA (Shuman 1985), and Australia (Brennan and Gartreli. 1990), Zntot is commonly below 10 mg kg' 1 which is less than in any of the coarse mineral soils of the present study.
A close correlation between clay content and Zntot is likely to be a consequence of the differences in soil mineralogy in the textural fractions (Sippola 1974). The analyses of clay and silt plus very fine sand of three soils directly showed that the coarser fraction was poorer in Zntot than was clay. However, silt can occasionally be as rich in Zntot as is clay (Andersson 1979, Armour et al. 1990.This may be the explanation why the unpolluted soils of Kuhmoinen (soil 21) and Hollola (soil 38) had a higher Zntot than the other soils of similar clay contents.
The negative correlation between Zntot and or-ganic C in organogenic soils reflects the origin of Zntot in mineral material. The studies on the vertical distribution of Zn emphasize that Zntot in organogenic soils was dependent on Zntot of the mineral soil below the organogenic layers. Organogenic soils (profile P 6 from Jokioinenand the sample pair from Forssa) on a clayey subsoil were rich in Zntot-On the contrary, the Carex peat profile P 7 (Muhos), most probably also profile P 5 (Sotkamo), had developed on coarse mineral soils poor in Zntot with the consequence that also the organogenic layers were poor in Zntot-In mull and peat soils, the mean Zntot was 48.7 and 35.1 mg kg' 1 , respectively, whilea higher mean of 65.5 mg kg' 1 was measured in 55 cultivated Histosols of Canada (Levesque and Mathur 1986). Nine out of the 20 peat soils of the present study contained less Zn than was the minimum (28.5 mg kg' 1 ) in the peat soils of Canada. This comparison confirms that soils extremely poor in Zntot prevail among cultivated peat soils of Finland.
Moreover, the results of URVAS et al. (1992) suggest that the peat soils of this study did not even contain the poorest ones occurring in Finland.
However, Zntot in cultivated peat soils of Finland commonly exceeds that in virgin peatlands of Germany where a Zntot of 5 -50 mg kg' 1 (mean 20 mg kg' 1 ) has been measured (Teicher et al. 1987).

Fractions of soil Zn
The decreasing trend of Zn e x with increasing soil pH agrees with the results of several studies (Sims andPATRiCK 1978, Nielsen etal. 1986, Sims 1986, Palko and Yli-Halla 1990,reflecting the corresponding increase of specific adsorption of Zn. In mineral soils, Zn ex was of the same level as in other studies on acid mineral soils (IYENGAR et al. 1981, Shuman 1985, Nielsen et al. 1986. Even in very acid clay soils Zn tended to be bound by covalent forces to a larger extent than in the coarsest mineral soils of similar pH. This may reflect the abundance of sites capable of specific adsorption ofZn in clay soils, which were richer in organic matter and Fe oxide, important components adsorbing Zn in acid soil (Mcßride and Blasiak 1979,Brummer et al. 1983, Tiller et al. 1984, Pulford 1986. In organogenic soils, the high percentage of Zntot in exchangeable form was probably attributable to the strong acidity in these soils. In a few peat soils, more than 30% of Zntot was exchangeable, which suggests that the small native reserves of Zn in these soils may even be susceptible to leaching. A few observations corroborate that sequential pyrophosphate and oxalate extractions dissolved Zn by and large from different soil components. First, Zn py (mg dm ) was approximately equal in all soil classes while Zn ox decreased in the mineral soils with decreasing clay content. Second, unlike Znpy, Zn 0x correlated significantly with Zntot. The fraction of Zn P y has been assumed to consist of Zn bound by organic matter. In the present study Zn py or complexed Zn (Zn py -Zn e x) did not correlate with organic C, but the dark color of the pyrophosphate extracts and the appearance of the soil after the extraction suggest effective removal of organic matter, with the consequence that also most Zn bound by organic matter was probably extracted. Pyrophosphate may also dissolve Zn from other sources. Oxides of Fe, Al and Mn are major adsorbents for Zn added to soil (e.g. Mullins et al. 1982, Mcßride 1989. In an incubation experiment of the present study the bulk of Zn added to soil was extracted with pyrophosphate and only a minor part was recovered as Znox which was supposed to represent sesquioxide-bound Zn. This observation indirectly suggests that at least some Zn from sesquioxides is extracted with pyrophosphate. The hypothesis is supported by an observation that pyrophosphate dissolves Fe from sesquioxides (BaS-COMB 1968).
There was some evidence that in mineral soils the occurrence of ZnD x may indeed be connected to Fe oxides. There was a correlation between Zn 0 x and Fe 0x, and the lowest Zno x exclusively occurred in the mineral soils poorest in Feo x. Several organogenic soils were also rich in Fe0x, but contrary to the mineral soils this was not connected to the abundance of Znox-Moreover, in organogenic soils Zn o x did not correlate with Znres or organic C which reflect the quantity of mineral material in the soil. Therefore, the source of Zn 0x in organogenic soils requires further research.
The mean percentage ofZn ox was low compared to the results of comparable sequential extraction studies from Georgia and Virginia, USA (Shuman 1979, Iyengar et al. 1981, and from the Nile delta, Egypt (Elsokkary 1979). In those soils, scarce in organic matterand rich in crystalline Fe oxide, Zn ox was the major fraction of secondary Zn, amounting to 25% ofZntot. Inversely, in the temperate soils of Canada (Nielsen et al. 1986, Liang et al. 1990 the Zn 0x fraction was of the same magnitude (below 5% of Zntot) as in the present material. Oxalate is not selective for poorly crystalline oxides but over time also crystalline oxides are dissolved (Borggaard 1979(Borggaard , 1992. In the present sequential extraction, poorly crystalline oxides and Zn bound by them may have been removed already in the pyrophosphate treatment and only the Zn bound to more crystalline oxides may have remained in the successive oxalate extraction. The scarcity of Znox in temperate soils may thus be attributed to the small quantity of crystalline oxides, owing to the young age and high content of organic matter which retards crystallization. The hypothesis presented above may be valid especially in organogenic soils most of which were poor in Zn Qx. Pyrophosphate and oxalate extractions completely removed Zn which had recently been added to the soil. The strong extraction power of these solutions was also shown by Payne et al. (1988) who observed that Zn added to a Rhodic Paleustult (560 kg Zn ha" 1 during 17 years) was recovered as Zn P y and Zn 0x, but there was no accumulation in the residual fraction. It seems therefore justified to regard the sum Znpy + Zn ox as the quantity of secondary Zn.
In most studies from other countries, the percentage of Znres in mineral soils has been lower than that in the present mineral soils where approximately 90% of soil Zn occurred as Znre s. For example, in soils of the British Columbia and Saskatchewan, Canada (Nielsen et al. 1986, Liang et al. 1990, Znres amounted to 71% and 83% (ranges 46 -93% and 69 -91%, respectively). Results of that level have also been obtained in soils of Delaware and Georgia, USA (Iyengar et al. 1981, SHU-MAN 1985, while in alluvial soils of Egypt, Znres was only 45% of Zntot (range 39 -61%) (Elsokkary 1979). The higher percentage of Znres in the present study reflects the young age of soils of Finland. The result also shows that the cultivated soils of Finland are not polluted with Zn because in polluted soils Znres represents a smaller fraction of Zntot (LÅG and Elsokkary 1978, Räsänen andHämäläinen 1991). The high percentage of secondary Zn in the organogenic soils reflects the scarcity of primary minerals in these soils.
Application ofmanures, sludges and Zn-containing mineral fertilizers as well as atmospheric deposition of Zn can result in excessive accumulation of secondary Zn in surface soil Kershaw 1989, Driel andSmilde 1990). Owing to mineral additives of fodder, manures in Finland commonly contain more than 200 mg Zn kg" 1 in the dry matter (Kemppainen 1989).The present results showed that also in Finland there are soils where abundant use of manure has resulted in elevated reserves of secondary Zn. The high concentration of Zn in soil 71 of Harjavalta was probably due to atmospheric deposition of Zn from the local metal industry which for example in 1988 emitted 100 000 kg of Zn, approximately 17% of all industrial Zn emissions of the country (Aunela and Larjava 1990). Even though this single soil sample is not necessarily representative, this observation warrants concern for excessive Zn content of soil in the vicinity of metal industry. However, the Zn concentration of this sample was much lower than the extreme values exceeding 1000 mg kg' 1 in the secondary fractions in soils heavily polluted by Zn (Elsokkary and Låg 1978, Folkeson and Andersson-Bringmark 1988, DeLaune et al. 1989, Jordao and Nickless 1989. Also in Finland, highly elevated concentrations of Zn have earlier been reported in lake shore sediments and surface soils of industrial areas (Räsänen and Hämäläinen 1991).

AAAc-EDTA extractions
In unpublished material of Soil Analysis Service Ltd., the means of ZnAc were 5.0, 6.1 and 4.9 mg dm"' 1 in the 1262,10240and 11701 analyses carried out in 1986, 1987 and 1988, respectively (Sippola and Tares 1978) were of the same magnitude as those of the present material. On the contrary. Sillanpää (1982) and Urvas et al. (1992) presented results on 90 mineral soils and 112 organogenic soils in which ZnAc was con-• 2 siderably smaller (mean 2.7 and 2.1 mg dm , respectively). Because the present results correspond to those of the large materials of Soil Analysis Service Ltd. and Sippola and Tares (1978), it is justified to conclude that the observations made on the material can be extended to the cultivated soils of Finland in general. The extractant AAAc-EDTA consists of three components: (1) acetic acid, (2) ammonium acetate and (3) Na2-EDTA. The chemical nature of ZnAc can be assessed by studying the fractions which might be extracted by each component alone. It is likely that the aqueous solution extracts water-soluble Zn, and Zn i+ bound by electrostatic forces is exchanged by NH4 + cations of the solution.AAAc-EDTA extracted more Zn than did MgCh, the difference being especially pronounced at the high pH range of the experimental soils. In several studies, 2.5% CH3COOH has been used in the extraction of Zn specifically adsorbed on the surfaces of sesquioxides (Elsokkary and Låg 1978, Iyengar et al. 1981, Payne et al. 1988). In AAAc-EDTA, the concentration ofacetic acid is 3% and therefore it is likely that specifically adsorbed Zn is extracted also by AAAc-EDTA. This assumption is supported by the observation that increasing clay content enhanced the extraction power of AAAc-EDTA in relation to MgCb in mineral soils. In clay soils the number of sites available for specific adsorption of Zn is probably higher than in coarse mineral soils, owing to the higher contents of Fe ox and Al ox-EDTA has been added to the AAAc solution in order to enhance the extraction of metallic trace elements (Lakanen and Erviö 1971). However, the Zn-EDTA complex is most stable at pH 6.5 (Lindsay 1972), and pH 4.65 should theoretically be far from ideal in order to facilitate effective extraction of Zn by EDTA. Yet, the extraction ex-periment (Section 3.4) showed that EDTA increased the extractability of both native and added Zn especially from organogenic soils as compared to the extraction by AAAc alone. Moreover, the extraction power of AAAc-EDTA was at least equal to that of DTPA where the pH 7.3 corresponds to the pH of maximum stability of the Zn-DTPA complex (Lindsay 1972). EDTA is an effective extractant for organically bound Cu (Rashid 1974, Stevenson 1982). It has also been observed that the addition of EDTA in the Mehlich 2 extractant (CH3COOH -NH4CI -NH 4F -HCI) enhanced the extractability of soil Zn especially when the content of organic matter increased (Mehlich 1984). It may thus be assumed that also AAAc-EDTA extracts Zn bound by organic matter even at pH 4.65. However, organic matter is more effectively extracted by pyrophosphate than by EDTA (McLaren and Crawford 1973). This observation seems to apply also to organically bound Zn because in mineral soils the extraction power of AAAc-EDTA decreased in relation to pyrophosphate with increasing content of organic C. The close correlation between ZnAc and Znpy suggested that AAAc-EDTA extracted Zn mainly from the same sources as did pyrophosphate. It may thus be concluded that, in addition to water-soluble and exchangeable Zn, the AAAc-EDTA extracts Zn specifically adsorbed by sesquioxides and organic matter. Goldschmidt (1937) observed in Germany and Hibbard (1940) in California, USA higher concentrations of Zn extracted from surface soils as compared to the subsoil. They both suggested independently that this may be due to bioaccumulation as a result of uplift of Zn from deeper layers by plant roots. When plant residues are decayed, Zn from plant tissue is mineralized and retained in the surface soil. The hypothesis of bioaccumulation is also corroborated by the observations on the present mineral soil profiles and earlier in a more extensive material ofpeat soils of Finland (Urvas 1986). In the present fine-textured mineral soil profiles, the minimum ZnAc occurred in the upper part of subsoil, and a higher concentration of ZnAc was measured deeper in the profile. Zinc uptake by plant roots may have depleted the reserves of ZnAc in the upper part of subsoil but not in the deeper layers because roots of herbaceous plants do not penetrate to a considerable extent deeper than 1 m (Dwyer et al. 1988). This observation also suggests that bioaccumulation of Zn is an important factor contributing to ZnAc in the plough layer. Absence ofcorrelation between Znre s and ZnAc also suggests that a considerable part ofsecondary Zn in the surface soil may not have been released from the primary minerals residing in the plough layer but may have been brought there from external sources, e.g. by bioaccumulation and atmospheric deposition. At the beginning of the experiment, the fertilizer solutions were mixed in the soil. For the subsequent crops they were pipetted onto the surface of the soil. To prevent a decrease of pH in the course of the trial due to N fertilization and root exudates, 250 mg of Ca(OH) 2 (6.8 meq dm' 3 ) was mixed into each pot. The seeds (250 mg per pot) were covered with 150 ml of quartz sand washed with 3 M HCI. The pots were watered with deionized water once or twice a day. The first crop was cut 29 days after sowing, and the average growing period of the successive crops was 24 days. The yields were dried at 60°C, weighed and analyzed for Zn. At the end of the experiment, the pH of the soil in the pot was determined. In spite of the lime application, soil pH decreased during the experiment (Table 27).

Dry matter yields
The mean deviation of the total dry matter yield between the two replicates was 0.30 g per pot, or 3.1% of the average yield. The average total dry matter production (sum of four crops) in clay, silt loam and very fine sand soils was slightly higher than that of the organogenic soils (Table 28). Dry matter yields did not correlate with any of the indices of secondary Zn of soil. Dry matter production in mineral soils was positively correlated with Zntot (r = 0.48** ) and soil pH (r = o.34* **). The correlation between Zntot and the yield reflects the trend that fine sand and moraine soils, poorer in Zntot than the clay soils, tended to produce smaller yields than did the more fine-textured soils. In organogenic soils, dry matter yield did not correlate significantly with the soil properties measured.

Zinc concentration and uptake
The mean Zn concentration of the grass within each yield (Table 29) did not differ markedly from one soil class to another (HSD values not presented). 1 Means in each column were tested separately. Each soil class, particularly fine sand and moraine, contained a few soils (especially soil 71) which produced grass very rich in Zn as compared to the bulk of the material. There was a tendency of Zn concentration to be the lowest in the second crop in which the largest quantity of dry matter was pro-duced. In peat soils, the mean Zn content in the fourth crop was lower than that in the first one, while in the mineral soils Zn concentration in the last two crops was at least as high as in the first one. Uptake of Zn (Table 30) was calculated by multiplying the dry matter yield with Zn concentration of the respective grass sample. Because the roots were not weighed and analyzed, Zn uptake represents the quantity of Zn transported to the aboveground parts of ryegrass. In clay soils, Zn uptake remained constant from crop to crop, whereas in silt, loam and very fine sand soils, the maximum Zn uptake occurred in the third crop. Instead, in the fine sand and moraine soils, Zn uptake increased towards the end of the experiment, despite decreasing dry matter production. Because both Zn concentration and uptake increased, the Zn supply to plants can be assumed to increase over time in these soils. In organogenic soils, Zn uptake by the fourth crop was smaller than that by the three earlier ones. In peat soils, both concentration and uptake were smaller in the fourth crop, which may reflect a decreased Zn supply to the plants. Within a crop, the only statistically significant difference in Zn uptake between the five soil classes occurred in the fourth crop in which Zn uptake from clay soils was greater than that from the peat soils (HSD values not presented).
The cumulative Zn uptake in the four yields ranged from 620 to 6190 pig dm of soil, mean 1470 pig dm' . The mean deviation in Zn uptake between the replicates was 52 pig dm' , corresponding to 3.7% of the mean. The differences in Zn uptake were much greater within each soil class than between the classes which did not differ significantly from one another. Correlation between plant Zn concentration and Zn uptake was very close in each of the four crops (r = 0.87 -0.96 ) but negligible between the cumulative dry matter yield and Zn uptake (r = 0.08 ns ' and r = 0.26 ns ' in mineral and organogenic soils, respectively). Thus, the quantity of Zn taken up was by and large determined by the Zn concentration of the plant material.

Dependence ofZn uptake on soil properties
The correlation coefficients between a few soil characteristics and the cumulative Zn uptake in the four crops are presented in Table 31. In the calculations, natural logarithms (log e ) of soil Zn concentrations (mg dm' 3 ) were used. The results of soil 71 were not included in the calculations. In organo- with Zn ex and Znpy -Zn ex , and more closely with Zn py than with Zn py -Znex . The partial correlation coefficients between Zn uptake and Znpy -Znex, when eliminating the effect of Zn ex, were highly significant (r = 0.56 and 0.65 in mineral and organogenic soils, respectively), suggesting that also complexed Zn contributed to plant-available Zn. The correlation coefficient with Zn o x was also significant, but the partial correlation coefficient between Znox and Zn uptake, when the effect of Znpy was eliminated, was not statistically significant. The relationship between ZnAc and Zn uptake in mineral and organogenic soils is presented in Figures 3 and 4, respectively.
In the multiple regression analysis, ZnAc (mg dm' ), soil pH and organic C (%) explained 82% of a the variation of the cumulative Zn uptake (pg dm' ofsoil) in mineral soils. In an equation consisting of Znpy (mg dm' 3 ), Znox (mg dm' 3 ) and pH, the coefficient of multiple determination was slightly lower. At a given level of Zn py , Zn ox and ZnAc, Zn uptake decreased with increasing soil pH. According to the beta coefficients (Table 32), ZnAc and Zn py were by far the most important soil characteristics determining the magnitude of Zn uptake by the grass. The two equations for mineral soils (n = 72) were as follows: Zn uptake = 593 loge Zn Ac -188pH -40.8 C + 2150 R 2 = 0.82 Zn uptake = 439 log e Znpy + 222 loge Zn ox -202 pH The dependence ofZn uptake of each crop on the characteristics of mineral soils was also studied. Within each crop, ZnAc and Znpy (loge of theresults expressed as mg dm' ) were the most important soil characteristics explaining the variation of plant Zn uptake (|4g dm 3 of soil) (Table 33). Soil pH was significant in the first two crops but lost significance towards the end of the experiment. Also the pH measured at the end of the experiment was used as an independent variable, but it did not prove statistically significant even in the last two crops. On the contrary, Zn Gx was significant in the last crop. Increasing clay content (%) promoted Zn uptake, but increasing organic C (%) had a negative impact. When Znpy was divided into two components (Znex and Znpy -Ztiex) they were both significant (beta coefficients not shown), and the equations had nearly the same coefficients of multiple determination as those containing Znpy . In multiple regression analyses of the results of organogenic soils, the cumulative Zn uptake (ug 3 3 dm "of soil) was explained only by Zn Ac (mg dm ') or Zn py (mg dm'     range 0.6 -40.3%). It was greater in the peat soils as compared to the other soils (Table 34). In organogenic soils there was a correlation (r = 0.80 ) between organic C and utilization ofZntot, which is explained by the inverse relationship between organic C and Zntot-In five out of 20 peat soils Zn uptake exceeded 20% of Zntot-In mineral soils, decreasing pH promoted the utilization ofZntot (r = 0.42 ), and clay content correlated negatively with the relative utilization of Zntot (r = -0.51 ). This is because clay content correlated with Zntot, but Zn in the fine-textured mineral soils was to a higher extent in the residual fraction unavailable to plants.
In the short term, secondary Zn fractions serve as the reserve of plant-available Zn in soil. Therefore, rather than utilization of Zntot, it is more appropri-ate to investigate the use of secondary Zn reserves. Relative utilization of secondary Zn in soil is affected by the size of the reserves as well as by their availability. For example, a given Zn uptake by the crop corresponds to a stronger relative utilization of Zn in a soil which has small reserves as compared to another soil which contains more Zn in a plantavailable form. Strong relative utilization of soil Zn may reflect the tendency of those reserves to be exhausted by plant uptake. Zinc uptake by ryegrass corresponded on average to 109% of Zriex and ranged 15 -535%. This result suggests that in addition to the water-solubleand exchangeable fraction, also less soluble Zn must have been taken up in several soils. Therefore the emphasis of studies on the relative utilization of soil Zn was in the Znpy and Zn ox fractions.
The relative utilization of secondary Zn was calculated as the ratio of Zn uptake (me: dm' 3 ) to the sum Znpy + Znox (mg dm" 3 ), i.e. 100 •Zn uptake/(Zn py + Zn ox ). Because the uptake of Zn in a pot experiment was used in the calculation of this index, the validity of the numerical values obtained is limited to this particular experiment. Zinc uptake amounted to 27.3% of Znpy + Zn0x, the range from 1.8% in soil 71 to 68.8% in peat soil 97 (Table 34). The secondary Zn fractions were utilized on average slightly more effectively in peat soils than in mineral soils, but the differences were not statistically significant. In 11 soils (two coarse mineral soils, two mull soils, seven peat soils) poor in Zn the relative utilization of secondary Zn reserves exceeded 40%.
Because correlation and regression analyses did not describe accurately the relationships between different soil characteristics and therelative utilization of secondary Zn, the quartiles of this index were investigated. The quartiles were designated I -IV according to increasing relative utilization of secondary Zn (Table 35). Coarse mineral soils were evenly distributed but in several clay soils Zn reserves tended to be poorly utilized, while in several organogenic soils strong relative utilization was observed.
The quartiles of relative utilization of secondary Zn were compared to those of Zn uptake. There were soils of all possible combinations of these two dimensions (Table 36). Large reserves of soil Zn were in most soils connected to a weak relative utilization, but small reserves were not necessarily effectively utilized by the grass. The four extreme groups of soils, shaded in the comers of Table 36, were studied in more detail. These soils were supposed to possess outstanding characteristics connected to given patterns of utilization of soil Zn. The rest of the soils were supposed to have the same characteristics less illustratively. The four extreme groups of soils were as follows: Group 1: combination of high Zn uptake (1720 -3 ... 6190 |ig dm ) (quartile I) and weak relative utihzation of secondary Zn (quartile I). The group consisted of 12 soils of large Zn reserves. All the seven soils containing more than 10 mg ZnAc dm occurred in this group. Soil pH (5.3 -7.2) was of minor importance in these soils with excessive reserves of plant-available Zn. Group 2: high Zn uptake (quartile I) and strong relative utilization (39%) of secondary Zn (quartile IV). The group contained only one very fine sand (40) and one peat soil (99). These soils were acid (pH 4.2 and 5.5) and the ZnAc (2.9 and 3.5 mg dm 3 ) was around the median of the material. These soils are examples of rapid reduction of Zn reserves which were at least average in size. Owing to a small number of soils, this group was improperly portrayed.
Group 3\ strong relative utilization of soil Zn (quartile IV) and low Zn uptake by plants (quartile IV). This group consisted of one gyttja clay (soil 12), three fine sands (50, 51 and 66) and four or-  (Table 37). These soils, except soil 50, belonged to the smallest quartile of ZnAc-The four organogenic soils were even within the Fio% of strongest relative utilization of secondary Zn as well as in the smallest Fio% of Zn uptake. Zinc uptake in these soils was thus limited by the small reserves. Group 4: low Zn uptake (quartile IV) connected with weak relative utilization of soil Zn (quartile I). In the four mineral soils of this group (Table 38) soil pH was distinctly higher, and Zntot, Zn P y + Zn 0x, and ZnAc were higher than in group 3, but ZnAc was below the median of the material (2.9 mg dm "). Zinc uptake by the grass was thus limited by relatively small reserves of Zn which, owing to a rather high pH, were poorly available.
In groups 3 (low uptake, strong relative utilization of secondary Zn ) and 1 (high uptake, weak relative utilization) the soils had the most distinctive characteristics. Also some characteristics of the combination of weak relative utilization and low Zn uptake (group 4) could be defined. Groups 3 and 4 exhibited two different combinations of characteristics resulting in a limited Zn supply to plants, Group 3 contained soils with small Zn reserves which, owing to the strong acidity, had a high availability to plants, resulting in a strong relative utilization. In turn, the soils of group 4 had larger Zn reserves which, owing to a higher pH, showed a poorer availability resulting in a weak relative utilization of these reserves. These conclusions are supported by the differentpatterns of Zn concentrations of grass grown in the organogenic soils of group 3 and in the mineral soils of group 4 ( Figure  5). In the organogenic soils of group 3, the Zn concentration of grass decreased from crop to crop, suggesting declining reserves ofplant-available Zn in soil. In group 4, the Zn concentration of plants increased during the experiment, suggesting increased availability, probably owing to acidification of the soil in the course of the experiment.

Effect of Zn application on plant Zn concentration 4.2.1 Experimental
A pot experiment was carried out in order to study the relationships between soil characteristics and the effect of Zn application on Zn concentration of ryegrass. The experiment was carried out in the same way as the one presented earlier (Section 4.1) with the exception that the pots were not limed. Zinc (10 mg dm" 3 ofsoil) was applied as a solution of ZnSOa. One crop of Italian ryegrass was grown for 30 days in polythene boxes containing 0.2 dm of soil. There were two pots of each soil to which Zn was applied; two pots were grown without added Zn. Of the 107 surface soils (Appendix 2), 101 soils were available for this trial. Those not available (15,27,29,51,88,107) were of average characteristics.

Dry matter yields and plant Zn concentrations
The average dry matter production was 3.23 g per pot; the highest yields were nearly double the lowest ones (Table 39). The mean deviation of dry matter yield between the two replicates was 0.10 g per pot, or 3.1% of the average yield. The yields grown without added Zn correlated closely (r = * + + 0.88 ) with the yields fertilized with Zn. According to the paired t-test, Zn application did not affect the size of the yield in any of the soil classes. The yields were not increased by Zn application even in soils producing grass with the lowest Zn concentration. The mean deviation ofZn concentration between the two replicates was 1.6 mg kg' 1 and 1.0 mg kg' 1 in pots grown with and without applied Zn, corresponding to 4.5% of the mean in both treatments. The average Zn concentration of grass grown in mineral soils without applied Zn (19.5 mg kg' 1 ) was Fig. 5. Zinc concentration of four ryegrass crops grown (a) in very acid peat soils (group 3) where a strong relative utilization of secondary soil Zn was measured and (b) in slightly acid and neutral mineral soils (group 4) where a weaker relative utilization of secondary soil Zn occurred. lower (t = 4.072***) than that in the corresponding organogenic soils (27.1 mg kg" 1 ). In all soil classes, Zn application elevated Zn concentration of ryegrass significantly (t-values of the paired ttest not presented). The Zn concentration of the grass fertilized with Zn was lower (t = 3.276 )in mineral soils (34.0 mg kg" 1 ) than in organogenic soils (42.1 mg kg" 1 ). Zinc uptake correlated closely with the Zn concentration of the grass (r = 0.84 and r = 0.93 in pots grown with and without Zn application, respectively), but there was no correlation between plant Zn concentration and dry matter yield.

Influence of soil characteristics on the response to applied Zn
The average increase of Zn concentration caused by Zn application did not differ statistically from one soil class to another (Table 40), and the utilization of added Zn by the grass was low in all soil classes. The increase (mg kg" 1 ) in Zn concentration of grass correlated slightly negatively with the dry matter yield (r = 0.44 ), showing that the Zn concentration of a smaller yield was more strongly increased than that of a larger one. In soils where grass of low Zn concentration was produced without added Zn, a high response to applied Zn was not necessarily measured, and even in soils producing grass of high Zn concentration a large response to applied Zn was occasionally observed. This resulted in a nonexistent correlation (r = -0.06 n s ) between Zn concentration of grass grown without applied Zn and the increase of Zn concentration.
It can be seen in Figure 6 that in soils fertilized with Zn, soil pH had a more marked impact on plant Zn concentration as compared to soils not fertilized with Zn. In neutral and slightly acid soils the effect of added Zn on plant Zn concentration was much smaller than in the more acid soils. The relationship between soil properties and response of plant Zn concentration to applied Zn in mineral soils was studied in more detail by multiple regression analyses. The results of soils 71 and 73 were not included in the tabulation, because soil 71 had an excessive ZnAc concentration and in soil 73 the increase of plant Zn concentration (+53 mg kg' 1 ) was much higher than in any other soil. The effect of added Zn on plant Zn concentration (y, mg kg' 1 ) decreased upon increasing soil pH which, according to the beta coefficients ( In organogenic soils the relationship between soil pH and plant Zn concentration was inconsistent. As a matter of fact, the grass of the lowest Zn concentration was produced in the two most acid organogenic soils (100, pH 3.8; 105, pH 4.1), which is a striking difference from mineral soils where acidity enhanced Zn uptake by the grass. In the two Fig. 6. Dependence of Zn concentration of ryegrass on soil pH in mineral soils fertilized with 10 mg Zn dm 3 (Zn+, symbol O) and those not fertilized with Zn (Zn., symbol □). grass grown in soils where native Zn py was easily available and strongly utilized responded strongly to added Zn and vice versa. The grass grown in three soils (77, 87 and 101) responded to added Zn markedly less than would be expected on the basis of the relative utilization of Znpy . In soils 77 and 101, the deviationbetween the replicates was large, possibly contributing to the inconsistent results; soil 87 was rich in clay (69%), possibly reducing the availability of added Zn. Omission of the three soils increased the coefficient of determination (R ) from 0.48 to 0.68.
peat soils 100 and 105, the Zn concentrations of the grass not fertilized with Zn were as low as 11 and 9 mg kg' 1 , respectively, and no higher than 21 mg kg' 1 in grass fertilized with Zn. In organogenic soils the increase (mg kg' 1 ) of plant Zn concentration was not explained by other soil properties, either. Instead, the response was dependent on the relative utilization (%) ofnative Zn py in the pot experiment (100 • Zn-uptake/Znpy ) (Figure 7). In other words,

Response of ryegrass to applied Zn in soils poor in ZnAc
The need for Zn fertilization is in Finland currently assessed on the basis of soil ZnAc-It is assumed that the need for Zn fertilization of a soil is the greater the less ZnAc is extracted from the soil. It might further be presumed that the greatest responses in plant Zn concentration would occur in soils poorest in ZnAc-According to the previous regression analyses, a decrease in ZnAc indeed enhanced the response of plant Zn concentration to added Zn, but the regression equations did not accurately explain the variation of the response. Therefore the increases of Zn concentration were divided into quartiles (F25%) which were compared with the quartiles of soil ZnAc-Because, according to the present recommendations, Zn fertilization is most commonly applied to soils poorest in ZnAc, the effect of Zn application on plant Zn concentration grown on these soils was studied with particular care.
There were 19 mineral soils in the smallest quartile of ZnAc (ZnAc < 1.85 mg dm' 3 . Table 42). Only in three of these mineral soils was there a strong response (the largest F 25%, quartile I) of plant Zn concentration to applied Zn. In the bulk of these mineral soils the increase of plant Zn concentration was smaller, placing them in the second or third quartile of response, but in none of the mineral soils concerned was the increase ofZn concentration of the grass very small (quartile IV). Within the group of these 19 mineral soils, the increase of plant Zn concentration did not correlate with soil ZnAc, but Fig. 7. Relationship) between relative utilization (%) of Zn extracted by pyrophosphate (Zn py ) and increase of Zn concentration of ryegrass upon application of 10 mg Zn dm 3 in 34 organogenic soils. Table 42. Soil Zn Ac and pH in mineral and organogenic soils poorest in Zn Ac (Zn Ac < 1.85 mg dnr 3 ). The soils are divided into groups according to quartiles (F 25 . ;.) of the increase of Zn concentration of ryegrass when 10 mg Zn dm-3 was added to the respective soil. the observed response was to some extent explained by soil pH. The three soils responding strongly to added Zn (44,52, 66) had apH equal to or below 6.0, while in most of the soils in which a smaller response was observed the pH was commonly above 6.0 which was the average of the mineral surface soils. It should be pointed out that in the whole material (n = 101) the greatest increases (30 -53 mg kg' 1 ) in plant Zn concentration occurred in very acid (pH 4.2 -5.4) mineral soils (10,40,52,73) containing 1.8 -3.8 mg ZnAc dm' 3 , and only one of these (52) was among the soils poorest in ZnAc (within the smallest F 25%). This observation further confirms that low ZnAc does not guarantee strong response of plant Zn concentration to applied Zn.

Increase of Zn
There were six organogenic soils in the smallest F25% of ZnAc, all of which were very acid (Table   42). In five of those, a strong response (quartile I) of plant Zn concentration to applied Zn was measured. In all the five organogenic soils, secondary Zn reserves were quite effectively (the largest Fio%) utilized. Thus, in organogenic soils a very low concentration of ZnAc resulted in a large increase in plant Zn concentration as a consequence of Zn application. The only exception was the extremely acid soil 105 (pH 4.1) in which the grass responded less strongly. In this soil, the pH was as low as 3.8 at the end of the experiment.

Experimental
The effect of liming and different rates of applied Zn on the yield and Zn concentration of ryegrass was investigated on clay, fine sand, Carex peat and Sphagnum peat (Appendix 6) in a pot experiment. The Sphagnum peat was commercial light-colored peat; the other experimental soils were taken from cultivated fields. The pots (four replicates) were •3 -3 filled with 7.5 dm of Sphagnum peat or 7 dm' of the other three soils. Lime was mixed into the assigned pots at the beginning of the experiment three days before the actual fertilization. In clay, fine sand and Carex peat, one rate ofCa(OH)2 (clay and fine sand: 7.0 g per pot, Carex peat: 14 g per pot) was applied to elevate soil pH by one unit. The Sphagnum peat was originally extremely acid (CaCb-pH 2.8) and the grass did not grow without liming. Therefore two rates of Ca(OH)2 (16 and 23 g per pot) were applied to Sphagnum peat and no unlimed pots were grown.
Zinc was applied in a ZIISO4 solution to clay, fine sand and Carex peat at the rates of 0, 2.7, 10, 20 and 40 mg Zn dm' 3 and to Sphagnum peat at the rates of 0,5, 10, 20, 50 and 100 mg Zn dm' 3 . The excessive Zn rates were applied in order to get information on Zn tolerance of ryegrass. At the beginning of the experiment, the following quantities (mg dm' 3 of soil) ofother nutrients were mixed in the entire soil of the pot as analytical grade chemicals: Ca was added to the soil at different rates depending on the liming. The macronutrients were mixed in the soil as solids and the micronutrients as solutions. The quantity of seed sown was 2.5 g per pot. The pots were grown in a greenhouse and watered daily with deionized water. In clay, fine sand and Carex peat, two crops were grown without reseeding. For the second crop, N, P and K solutions (NH4NO3, KCI and KH2PO4) were pipetted onto the surface of the soil in four portions at intervals of four to five days, resulting in a total of 200, 50 and 200 mg of N, P and K per dm 3 of soil, respectively. For Sphagnum peat, the experiment continued for two years. Three crops were grown in the first summer. During the winter the pots were stored outdoors covered with plastic foil, and in the second spring fertilized and reseeded. Then, two more crops were grown in pots to which 5, 20, 50 or 100 mg Zn dm' 3 had been applied. At two rates ofZn application (0 and 10mg Zn dm' 3 ) five crops were grown in the second year. In clay, fine sand and Carex peat, each crop was analyzed for Zn. In Sphagnum peat, a few crops were pooled in the second year in order to reduce the number of analyses. At the end of the experiment, soil samples of each pot were analyzed for ZnAc and pH.
During the experiment, atmospheric deposition of Zn was monitored in the greenhouse. Six Petri dishes (area 60.8 cm ) were kept open in the greenhouse compartment where the pots were grown and six capped Petri dishes served as controls. After two months the Petri dishes were washed with 25 ml of 1 M HCI and the extracts were analyzed for Zn.

Dry matter yields and plant Zn concentrations
The total dry matter yields, not presented in detail, were 33.6,27.3 and 28.3 g per pot in clay, fine sand and Carex peat, respectively, and they were not affected either by Zn application or by liming. In Sphagnum peat the grass tended to be paler green at the highest Zn rate (100 mg dm' 3 ) throughout the experiment as compared to the other pots. There seemed to be a slight yield decrease towards the higher rates of Zn application (Table 43), which may be an indication of excess of Zn. The dry matter production in Sphagnum peat at the higher liming rate was 5% higher than that at the lower rate 1 Results of the two liming rates were tested separately.

244
(F = 47.083***). The additional three crops grown O in pots ofoor 10 mg Zn dm increased the total dry matter yield to an average of 181 g per pot, with no differences between the treatments. Application of Zn elevated the Zn concentration of grass substantially in all soils (Figure 8). In Sphagnum peat Zn addition increased the Zn concentration of grass most effectively. For example, Zn application of 10 mg dm elevated the Zn concentration of grass by 65 mg kg" 1 in the first crop in and Zn concentration (mg kg' 1 ) of grass (Table 44).
Despite the slight curvilinearity of the response in clay soil and in Carex and Sphagnum peat, the linear regression equations were used. The equa-tions of Sphagnum peat were calculated using the results of the first and second crop and Zn additions up to 50 mg dm" 3 . All possible pairs of the regression coefficients were tested separately for each crop and liming rate (Ranta et al. 1991, p. 395). According to the regression coefficients, the effect of Zn application was the greatest in Sphagnum peat, followed by fine sand. The effect was smaller in clay and still smaller in Carex peat, although the difference between these two soils was not significant in the second crop in the limed soils. Liming lowered the efficiency of Zn application in clay (P = 0.001) and especially in fine sand (P = 0.001) but had a relatively smaller but significant effect (P = 0.05) in the two peat soils. The efficiency of added Zn decreased in clay and unlimed fine sand from the first crop to the second one, while in Sphagnum peat the efficiency increased. In Carex peat and limed fine sand the efficiency was unchanged from the first to the second crop.
Despite substantial increases in Zn concentration of grass, the utilization (%) ofadded Zn in clay, fine sand and Carex peat remained low, ranging in the unlimed pots from 0.4 to 1.7%and in the limed pots from 0.0 to 1.4%. In the first two crops grown in Sphagnum peat, the utilization of added Zn ranged from 3.1% to 6.1%. In the total of five crops, the utilization ranged from 22 to 8%, decreasing steadily from the lowest to the highest rate of added Zn. In the pots grown for three more crops, the utilization of added Zn rose to 34.5% at the rate of 10mg Zn dm' without any indication of decreased Zn availability over time. Assuming that the same deposition had fallen over the grassed pots (area 363 cm ), a fallout of 49 jig per pot can be expected over 2 months. The average total uptake of Zn from the control pots of clay, fine sand and Carex peat ranged from 900 pg per pot in clay to 1500 pg per pot in fine sand. The deposited Zn thus ranged from 3 to 6% of the measured Zn uptake in the control pots. The deposition corresponded to 78 g ha" 1 over 12 months.

Soil analyses at the end of the experiment
Zinc addition to soil substantially elevated the quantities of ZnAc determined at the end of the experiment (Table 45). The samples taken from clay soils to which oorlomg Zn dm had been applied were lost before the analysis. In all clay samples available, the final ZnAc was lower than the amount of applied Zn plus the initial ZnAc, the difference being much greater than the Zn uptake by the grass. Inversely, in the unlimed fine sand the ultimate ZnAc did not differ from the sum of Zn application plus the initial ZnAc-This difference between clay and fine sand may suggest retention of Zn in the clay soil into forms not extractable by AAAc-EDTA, while the availability of added Zn was maintained in the sand. In the two peat soils, on the contrary, the quantities of ZnAc determined at the end of the experiment exceeded the sum of applied Zn and original ZnAc-This anomaly may be attributed to increasing bulk density of the organogenic soil during the experiment, owing to compaction and oxidation of organic matter.
Determination of soil pH at the end of the experiment showed that there was a significant (t-value .763 , respectively), while in Carex and Sphagnumpeat the difference between the two lim-ing rates was not statistically significant (P = 0.308 and P = 0.279, respectively).

Discussion
The adequacy of soil Zn supply to plant can be assessed by measuring the dependence of plant growth on the Zn status of soil (Lins and Cox 1988, Armour et al. 1990, Brennan and Gartrell 1990. The critical level of Zn in soil has been determinedas the quantity of extractable Zn above which no yield increase is obtained by Zn application. The present pot experiments did not allow to determine the critical level of ZnAc, because the dry matter production was not dependent on the status of soil Zn, and the yields were not increased by Zn application even though a few soils in the material were rated poor in ZnAc (<l.O mg dm ). The adequate size of the native Zn reserves was also displayed by the fact that in spite of the high growth intensity in the pot experiment, the average Zn concentration of the ryegrass grown in 107 soils was of the same magnitude as the average of fieldgrown timothy in Finland (30.8 + 13.2 mg kg" 1 by LAKANEN 1969, 32.0± 8.5 mg kg' 1 by Kähäri and Nissinen 1978). Zinc needs to enter the solution phase as Zn 2+ cation in order to be taken up by the plant. It is the fraction extracted with MgCb that contains watersoluble and exchangeable Zn readily available to plants. However, the four crops of grass were able to take up more Zn than was originally extracted with MgCL from several of the 107 surface soils especially in the higher pH range. It is feasible that during the experiment there was a flux of Zn from the more strongly bound fractions into the soluble one. This hypothesis is supported by the results of Bakhsh et al. (1990) and Torres-Martin and Gallardo-Lara (1991) who measured an increase in Zn extracted with 0.01 M CaCL or 1 M MgCb during the plant growth with a consequent decrease in the fraction bound by Fe oxide. As compared to the bulk of the soil, increased concentration of Zn in the rhizosphere of maize, wheat and barley has also been measured by Linehan et al. (1985) and Merckx et al. (1986). The increase was attributed to the influence of the lower pH and the abundance of chelating root exudates in the rhizosphere, both mobilizing Zn from reserves less soluble than Zn ex . Some authors have reported a very close correlation between soil Znex and Zn concentration of plants grown in pot experiments (Davies 1985, Nielsen et al. 1987 and in the field (Merkel andKöster 1977, Sauerbeck andStyperek 1985). Neutral salt solutions (NaNOs, CaCh) have been proposed for extractants of Zn especially in polluted soils rich in Zn (Hani and Gupta 1985,Sauerbeck and Styperek 1985, Houba et al. 1990). These solutions may not, however, be the most suitable in unpolluted soils where plants also utilize Zn reserves which are less soluble than Zn ex . In the present study, Zn extracted by MgCh did correlate significantly with Zn uptake by ryegrass, but in mineral soils AAAc-EDTA which dissolved also complexed Zn yielded results which correlated even more closely with Zn uptake by the grass. It was also observed by Nielsen et al. (1987) that in non-contaminated soils poor in Zn, only ZnoTPA, but not Znex , correlated with Zn uptake by plants.
Soil ZnAc and Znpy , determinedat the beginning of the pot experiment, were relatively the most important soil characteristics to explain the variation of the uptake of soil Zn throughout the four crops. AAAc-EDTA and pyrophosphate thus extracted Zn from the same fractions which serve as the source of Zn for plants. Contrary to Znex underestimating Zn supply in soils ofhigh pH, ZnAc and Zn P y overestimated the Zn supplying power in the same mineral soils. It is possible that Zn unavailable to plants may be extracted by AAAc-EDTA buffered to pH 4.65 especially from soils which are less acid than the extractant. This assumption is supported by the results of JAHIRUDDIN et al. (1986) obtained in acid soils of Scotland. There, 1 M ammonium acetate, adjusted to different pH values, dissolved more Zn at pH 4.8 than did the solution adjusted to soil pH above 4.8, and the additional quantity of Zn dissolved at pH 4.8 was the greater the higher was the soil pH. The present study showed that also pyrophosphate at pH 10 was less sensitive than ryegrass to decreased Zn solubility induced by increasing pH. Corresponding observa-tions have been reported earlier (Haq andMiller 1972, Hornburg andBrummer 1991) when extracting soil Zn with various EDTA and DTPA containing solutions.
The above findings support a pH-dependent interpretation of ZnAc, in principle suggested by Sillanpää (1982). This means that at a low pH a given result of ZnAc corresponds to a larger Zn supply to plants than the same result in a soil of higher pH, and the more acid soil gets a higher rating in soil testing. A highly pH-dependent interpretation of ZnoTPA in soils of Australia has also been proposed (Brennan and Gartrell 1990). However, when extracting soil Zn with chelating agents (EDTA and DTPA) the need ofa pH correction is not as crucial as it is when mineral acids are used as extractants for Zn. For example, in the study of JUNUS and Cox (1987) the results of Mehlich 3 extraction explained the variation of Zn uptake by soybean and com from soils of North Carolina, USA, only when soil pH was taken into account.
In extremely intensive pot experiments also ZnQ x was a source of plant-available Zn in mineral soils even though Zno x has been considered virtually unavailable to plants (Iyengar et al. 1981, Payne et al. 1988).The increase of plant Zn concentration towards the end of the experiment (Section 4.1) in the mineral soils may reflect the increased solubility of Zno x over time. The decrease of soil pH during the experiment can have caused partial dissolution of the Zn o x fraction, which has been reported to respond to soil acidification (Haynes and Swift 1985). In turn, in organogenic soils the decrease of plant Zn concentration during the trial may reflect the scarce Zn ox reserves in these soils.
In several previous studies (Wear 1956, Leyden and Toth 1960. Aasen 1981, Sims 1986, JUNUS and Cox 1987,Boswell et al. 1989 liming of an acid soil has been shown to reduce the fertilizer efficiency of added Zn. The same finding as well as an observation that ZnAc decreased upon liming were made in this study in a pot experiment with clay and fine sand (Section 4.3). However, a more noteworthy observation of pH and fertilizer efficiency of added Zn was made in the larger material of mineral surface soils (Section 4.2) where soil pH proved a relatively more important characteristic than native ZnAc in determining the effect of added Zn on plant Zn concentration. According to several studies, the equilibrium Zn concentration in soil solution or in the liquid phase ofa soil suspension decreases and Zn adsorption increases with increasing pH of an acid soil (Mcßride and Blasiak 1979, Brummer et al. 1983, Jeffery and Uren 1983,Tiller et al. 1984, Pulford 1986, Msaky and Calvet 1990. A decrease of Zn concentration in the soluble phase seems to indicate an increased specific adsorption (Tiller et al. 1984, Shuman 1986, Sims 1986 which evidently results in a decreased tendency for desorption and lower plant-availability of Zn (El Kherbawy and Sanders 1984). Thus, soil pH by and large determines which part of fertilizer Zn remains plant-available and which is converted into unavailable forms.
Unlike mineral soils, Znpy or ZnAc explained alone the variation of the uptake ofZn from organogenic soils by ryegrass without a significant contribution of other soil properties. The insignificance of other variables may partly be causedby the small number of organogenic soils and a narrower pH range, which may not have allowed the influence of soil pH to be observed as clearly as in mineral soils. It is also possible that the influence of soil pH on Zn solubility in organogenic soils is different from that in mineral soils. This conclusion can be drawn on the basis of the results of a pot experiment (Section 4.3) where liming highly affected the response of ryegrass to applied Zn in the two mineral soils but had only a minor influence in the two peat soils despite a significant increase in soil pH also in these soils. A similar observation was made by Woltz et al. (1953) in New Jersey, USA, where an elevation of the pH from 6.0 to 7.0 decreased the Zn uptake by red clover in mineral soils but did not do so in a peat soil. Also Jeffery and Uren (1983) observed that mixing peat in an acid mineral soil reduced the decrease of Zn concentration in soil solution when the pH was elevated. Mcßride and Blasiak (1979) attributed the pH-dependent increase of Zn adsorption in acid mineral soils to the reactions of Zn with Fe, Al and Mn oxides and considered the reactions with organic matter to be ofminor importance. If this is true and the fraction of Zn bound by organic matter plays a more important role as a source of soluble Zn in organogenic soils, soil pH may in organogenic soils indeed have a less marked effect on Zn solubility than in mineral soils.
In organogenic soils, the AAAc-EDTA method was able to recognize soils which, according to the pot experiment, had the smallest reserves of plantavailable Zn. In those peat soils native Zn reserves were most strongly utilized, and also added Zn remained available to plants, effectively elevating plant Zn concentration. However, the soil characteristics controlling the relative utilization of Zn in organogenic soils remained ambiguous. Moreover, in a very acid peat soil (100, pH 3.8) Zn concentra-tion and uptake of ryegrass were low in spite of abundant ZnAc reserves in the soil. In this soil and in another highly acid peat (105, pH 4.1), also the response to applied Zn was small even though at this pH the solubility of added Zn was inevitably high. An increase in Al 3+ concentration in a nutrient solution culture has been reported to reduce the concentration of Zn in ryegrass (Rengel and Robinson 1989). It has also been observed that timothy grown in acid sulphate soils rich in exchangeable A 1 had a low Zn concentration in spite ofa rather high ZnAc in therespective soils (Palko 1986). The adverse effects of A 1 may thus have caused the low Zn uptake of soil and fertilizer Zn in the two extremely acid peat soils.

FERTILIZERS AS ZINC SOURCES IN POT AND FIELD EXPERIMENTS
Qualitative relationship between soil and plant Zn can be studied in laboratory and pot experiments, but these studies do not tell the quantitative effect of applied Zn on the crop grown in the field where Zn fertilization is actually practiced. Moreover, in the pot experiments reported earlier in this study only ZnSOa was used as the source of Zn. In practice, the quantities of solid ZnSOa corresponding to Zn fertilizer recommendations are too small to allow uniform spreading, and therefore fertilizers of lower Zn concentration have been developed in which ZnSOa is incorporated in different carriers, facilitating uniform application with common machinery. In addition to the application to the soil, Zn fertilization of the crop can be carried out by foliar sprays. In order to study the effect of different sources, rates and application methods ofZn on the yield and Zn concentration of the crop, field experiments were carried out with timothy and barley at the Kotkaniemi Experimental Farm of Kemira Oy and in the neighboring fields in Vihti in southern Finland (60°22' N, 24°22' E). To complement the field experiments, a pot trial was conducted on the effect of different fertilizers on the Zn concentration ofryegrass.

Fertilizers
The fertilizers consisted of commercial fertilizers as well as products specifically manufactured for the experiments (Table 46). The commercial Zn fertilizer 'Sinkkilannos' consists of ZnSOa mixed with gypsum, and the dry mixture is granulated. In addition to straight Zn fertilizers, there were four NPK fertilizers to which Zn was added as ZnSOa.
Addition of Zn to the commercial NPK fertilizer 'Vähäfosforinen Y-lannos' (18-3-12, N-P-K) is carried out by mixing ZnSOa in the fertilizer slurry resulting in the distributionof Zn in the entire granule. The coated NPK fertilizers containing Zn were made by coating grains of a commercial NPK fertilizers (17-6-12 or 25-4-4, % N-P-K) with ZnSOa in the pilot hall of Kemira Oy Espoo Research Centre. Based on the results of Ellis et al. (1965), it was hypothesized that Zn might be more available when it is applied onto the granule as opposed to mixing in the slurry before granulation. After the field experiments were set up, it turned out that the fertilizer 18-3-12 contained 0.26% of Zn instead of 0.30% on which the experimental design was based. Therefore, Zn application to the plots receiving this fertilizer remained in reality slightly lower than intended.

Experiments with ryegrass and timothy a. Comparison ofZnfertilizers in a pot experiment
Agronomic efficiency ofa few Zn fertilizers used in the field experiments was tested in a pot experiment with ryegrass in clay, fine sand and Carex peat (Appendix 6). The clay and fine sand were taken from the sites were the field experiments with timothy were carried out. Zinc sulphate mixed in the soil served as reference fertilizer. The other treatments resembled the alternatives available in practical farming of forage crops. The commercial fertilizer 'Sinkkilannos' was mixed in the soil or was given as a topdressing,and the two Zn-containing NPK fertilizers were surface-applied (Table  47). The two Zn rates were 2.7 (2.3 in NPK 18-3-12) and 10 mg Zn dm' 3 of soil.
3 There was 7 dm of soil per pot and four replicates. At the beginning of the experiment, the same were pipetted onto the surface of all pots in four portions at a few days intervals. The pots were grown in a greenhouse and watered daily with deionized water. The grass yields were dried at 60°C and analyzed for Zn.
h. Application of Zn fertilizers to timothy in the field The objective of the 2-year experiments, performed in clay and fine sand soils (Appendix 6), was to study whether Zn concentration of timothy was equally affected by a single application of straight Zn fertilizers and applications of Zn incorporated in NPK fertilizers spread for each yield. The swards were sown in 1990 and harvested in 1991 and 1992. During the experiment, all plots, except the control, received 3 or 6 kg Zn ha" 1 (5.2 kg ha" 1 in NPK 18-3-12) as a single dose or as multiple smaller applications. Foliar application of Zn to the growing sward was not made because Zn added that way can be adsorbed on the foliage without taking part in the reactions of the plant, which may cause irrelevant results of plant analysis. The different Zn applications, presented in detail in Appendix 7, were as follows: 1) Single application of a straight Zn fertilizer (ZnSOaor 'Sinkkilannos') at sowing (1990).
2) Single application of a straight Zn fertilizer ('Sinkkilannos') onto the sward in spring of the first year of cropping (1991).
3) Applications of Zn-containing NPK fertilizers (NPK 18-3-12, Coated NPK II or Coated NPK III) onto the sward in spring and after cutting the first crop in both years of harvest (1991 and 1992).
When setting up the experiments, the fields were harrowed twice. Then, the granular Zn fertilizer 'Sinkkilannos' was broadcast and ZnSOa, dissolved in water, was sprayed on the appropriate plots. The NPK fertilization was applied with a fertilizer drill and timothy seeds were sown with a seed drill, simultaneously mixing 'Sinkkilannos' and ZnSOa in the soil. The next spring, granular 'Sinkkilannos' was top-dressed on the assigned plots. In both experimental years, all plots received N, P and K fertilization as a top dressing at the rates of 90, 32 and 64 kg ha" 1 , respectively. As mentioned earlier, two of the NPK fertilizers contained added Zn. After the first cut in mid-June, when a few heads of timothy were emerging, NPK fertilization was applied again at the same rates and the swards were harvested for the second time later during the growing season. Owing to a drought in 1992, the experiment in clay soil was irrigated on June 30 (28 mm of water) and the latter cut of both experiments took place as late as the end of September. In addition to the actual Zn fertilization, all plots received 19 g Zn ha" 1 in the NPK fertilizer when the experiments were established. During the years of harvest, the plots received a maximum of 130 g Zn ha" 1 as impurities of NPK fertilization. This quantity corresponds to 5% of the smaller application of 3kg Zn ha" 1 and can be considered negligible.
The stand was cut with a forage harvester. The yield of each plot was collected in a glass fibre box attached to the harvester. The plant samples, to be analyzed for Zn, N and moisture, were taken just before the actual harvesting. The samples, composed oftwo subsamples of 0.25 m from each plot, were collected by hand using stainless steel scissors and weighed together with the harvested yield of the respective plot. Soil samples were taken from each plot and analyzed for pH and ZnAc in the beginning and at the end of the experiment.

Field experiments with barley a. Comparison ofZnfertilizers
The objective of the three 3-year field experiments carried out in 1990 through 1992 on clay, fine sand and mull soils (Appendix 6) was to establish whether Zn concentration ofbarley can be elevated with moderate Zn applications given in different fertilizers and whether a single application at the beginning of the experiments differs in efficiency from smaller annual applications. During the experiment, all plots, except the control, received 5.4 kg Zn ha" 1 (4.8 kg ha" 1 in NPK 18-3-12) either as a single application at the beginning or as three annual applications of 1.8 kg Zn ha" 1 (4 x 1.6 kg ha" 1 in NPK 18-3-12). The treatments, presented in detail in Appendix 8, were as follows: 1) Single application of a straight Zn fertilizer (ZnSO4 or 'Sinkkilannos') in the first spring of the trials. 2) Annual doses of Zn incorporated in granular NPK fertilizers  or Coated NPK I).
3) Annual doses of Na2Zn-EDTA sprayed each spring onto the soil. 4) Annual doses of Na2Zn-EDTA sprayed on the foliage.
Each spring the soil was harrowed twice. After harrowing, the granular Zn fertilizer ('Sinkkilannos') was broadcast (in 1990 only), and ZnSOa (in 1990 only) and Na2Zn-EDTA (annually) were sprayed on the appropriate plots as water solutions. These fertilizers were mixed in the soil with a combined seed and fertilizer drill in connection with sowing and application of the NPK fertilizers. The granular NPK fertilizers, including added Zn in two fertilizers, were applied by the placement method. At the Feekes 5 growth stage (Large 1954), when the plants had usually reached the height of 10 -15 cm, water solution of Na2Zn-EDTA was sprayed on the assigned plots. In 1992, the experiment on clay soil was irrigated with 13 mm of water on July 1. In addition to the actual Zn fertilization, the experiments on clay and fine sand received 37 gZn ha' 1 as impurities of other fertilizers during three years, except the plot fertilized with the NPK 18-3-12 and Coated NPK I. In mull soil, the corresponding quantity was 120 g Zn ha' 1 , owing to the larger Zn concentration of the NPK fertilizers used in that experiment.
The grain yield, harvested with an experimental harvester (Hege 125), was weighed and analyzed for moisture and Zn. Also the straw was analyzed for Zn and moisture. The straw yield was weighed in the last two experimental years.

h. Application of different Zn rates
The objective of the two 2-year field experiments with barley was to investigate the effect of high Zn rates (15 and 30 kg ha' 1 ) on the Zn concentration of barley at different growth stages. The experiments, presented in detail in Appendix 9, were carried out in the same fine sand and clay soil (Appendix 6) in which the different Zn fertilizers were experimented. After harrowing in the first spring (1991) of the experiment, powdery ZnSOa was applied by hand at the rate of 0, 15 or 30 kg Zn ha" 1 and mixed in the soil by harrowing for one more time. The NPK fertilizer was applied with the combined seed and fertilizer drill to supply N, P and K at the rates of 110, 22 and 44 kg ha' 1 , respectively. In the second spring (1992), no Zn was added and the results of the second experimental year thus reflect the residual effect of the Zn fertilization. The quantity of Zn as impurity in the NPK fertilizer during the two years was 34 g ha' 1 . In 1992, the experiment on clay soil was irrigated with 20 mm of water on July 1.
The plant samples were collected four times during the growing season from each plot by cutting the plants each time from two areas of 0.25 m . The stand was sampled at the beginning of tillering (Feekes 2), at the end of tillering (Feekes 5), at the end offlowering (Feekes 10.5) and at maturity. The number of days from sowing to these growth stages was 38 -39 days, 50 -52 days, 69 -70 days and 104 days, respectively, in 1991 and 26 -28 days, 40 -43 days, 61 days and 91-96 days, respectively, in 1992. The plant samples were dried and weighed. The grains were removed from the ears of mature plants by hand and the remnants of the ears were combined with the straw. The plant samples were analyzed for Zn; the grains and straw were weighed and analyzed separately. At the end of the experiments, a composite soil sample of each plot was analyzed for ZnAc.

Weather
Precipitation and temperature were measured at the Kotkaniemi Experimental Farm. The three growing seasons in which the field experiments were conducted differed strongly in weather of the early summer. In 1990 and1992,May and June were dry and warm (Table 48). In 1992, the drought continued until the end of July, resulting in a shortage of water especially in the experiments on the clay  (Table 49), but only in fine sand also the granular 'Sinkkilannos', mixed in the soil, had a significant effect. In the second crop, the effect of the fertilizers was different; granular 'Sinkkilannos' elevated plant Zn concentration relatively more efficiently than in the first crop, probably owing to a longer time available for the dissolution of the granules. In the second crop, the surface-applied 'Sinkkilannos' elevated plant Zn concentra-lion in all soils at least as effectively as ZnSOa mixed in the soil. Actually, surface-applied 'Sinkkilannos' was the only fertilizer which elevated plant Zn concentration significantly at the lower application rate in the second crop in fine sand and Carex peat. The only significant effect of the two Zn-containing NPK fertilizers on plant Zn concentration was observed in the second crop in clay with the coated NPK I. At the higher application rate (10 mg dm' 3 ) the increase of plant Zn concentration was substantially higher. In fine sand, 'Sinkkilannos' mixed in the soil was equal to ZnSC>4 in efficiency in both crops, but in the first crop in Carex peat and in the second crop in clay, ZnSOa was more efficient.
Also at the higher Zn rate top-dressing of 'Sinkkilannos' was a less effective way of application than was mixing in the soil in the first crop in fine sand and clay soils. In the second crop the contrary was observed: topdressing of 'Sinkkilannos' was the most effective way of Zn application in all soils. The efficiency of the surface-application of 'Sinkkilannos' in the second yield was also emphasized when comparing the results obtained at the two Zn rates. Surface-applied 'Sinkkilannos' at the lower rate elevated Zn concentration of ryegrass as much as did the higher rate of 'Sinkkilannos' mixed in the soil. Despite substantial increases in Zn concentration of ryegrass, the utilization of added Zn remained at 1 -2% in all soils.

Effect of Zn fertilizers on timothy in the field
All the four timothy yields (Table 50) obtained from the fine sand field during the two growing seasons were ofnormal size. Owing to poor growth, only two of the four blocks could be harvested in clay soil in the first experimental year and the first yields remained small also in the harvested blocks. Later, the yields obtained from the clay soil were of the same magnitude as those from the fine sand.
The results concerning the clay soil represent the yield of those two blocks harvested in both years. Dry matter yields did not respond to Zn applications. The results of Zn concentration and uptake were tested with the analysis of variance using Zn fertilization and the crop (1/1991, 11/1991, 1/1992, 11/1992) as the two criteria of classification. The results of the blocks served as replicates. The Fvalues in Table 51 suggest that in both experiments the systematic differences between the Zn concentrations of the four crops were at least as significant as those caused by the different fertilization treatments.
In clay soil, Zn concentration of timothy varied in a rather narrow range in the four crops of each treatment (Table 51). In fine sand, the ranges were slightly wider mainly because the Zn concentrations of crop 11/1992 were substantially higher than those of the other crops. In neither experiment did the lower Zn rate (3 kg ha' 1 ) elevate the mean Zn concentration of timothy significantly. At the higher Zn rate (6 kg ha' 1 ) 'Sinkkilannos', applied by mixing in the soil or by top-dressing, elevated Zn concentration of timothy significantly in both soils. In addition, Zn concentration of the grass fertilized with ZnSOa (6 kg Zn ha' 1 ) or with the coated NPK 111 (6 kg Zn ha' 1 ) differed significantly from the control in clay soil. 'Sinkkilannos' mixed in the soil seemed to be slightly more effective at both levels than ZnSOa alone, but the difference between these two fertilizers was not significant.
A decrease in the effectiveness of Zn application during the experiments was not detected. As a matter of fact, the greatest increases in Zn concentra- The growth of barley was ample in 1990 and 1991, but in 1992 the yields were reduced in clay by the drought (Table 52). In none of the experiments were grain or straw yields affected by Zn fertilization. Zinc concentration of grain and straw (Table 53) was the highest in mull which had the lowest pH and highest content of ZnAc-Inversely, the lowest concentrations occurred in barley grown in fine sand which had the highest pH and was poorest in ZnAc-Zinc concentration of grain was, with few exceptions, at least twice the Zn concentration of the straw. Foliar application of Na2Zn-EDTA elevated Zn concentration of grain and straw significantly in all experiments. The increase in Zn concentration of grain was 4.3, 3.7 and 4.5 mg kg' 1 in clay, mull and fine sand, respectively. In mull and fine sand, none of the soil-applied Zn fertilizers affected Zn concentration of grain or straw. On the contrary, in clay all soil-applied Zn fertilizers, except Na2Zn-EDTA mixed in the soil, elevated Zn concentration of the grain and the two Zn-containing NPK fertilizers increased Zn concentration of the straw. No systematic difference could be detected in the effect of Zn fertilizers applied in the soil annually and that applied only at the beginning of the experiment. It should be pointed out that the Zn concentration of the crop did not correlate with respectively). In the three years, Zn concentration of grain of the control plots varied within a narrow range (3 mg kg' 1 in clay and mull, 1 mg kg' 1 in fine sand). In clay there was substantial annual variation in Zn concentration of grain in the plots fertilized with Zn. In the first experimental year, Zn concentration was below 30 mg kg' 1 in all treatments, while in the two following years the average concentrations of the different treatments ranged from 32 to 38 mg kg' 1 . The annual variation of straw Zn concentration was by far the greatest in mull where the means were 27.4, I l.6and 14.2mg kg' 1 in 1990, 1991 and 1992, respectively, in plots other than those of Na2Zn-EDTA application. In clay and fine sand the annual variation in Zn concentration of straw was less marked. The mean annual Zn uptake by barley grains was 89, 212 and 94 g ha' 1 in the control plots of clay, mull and fine sand, respectively. In clay, all Zn fertilizers slightly increased Zn uptake but only the Coated NPKI gave rise to a significant increase (24 g ha' 1 , +27%). In fine sand, foliar application of NaiZn-EDTA elevated Zn uptake significantly (by 23 g ha' 1 , +24%); utilization of foliar-applied Zn in grain was 1.3%, 0.6% and 0.8% in clay, mull and fine sand, respectively. Mean annual Zn uptake by the straw in the control plots was 15, 42 and 17 g ha' 1 in clay, mull and fine sand, respectively. Foliar application of NaiZn-EDTA increased the quantity of Zn harvested in the straw by 12,21 and 18 g ha' 1 in clay, mull and fine sand, respectively, corresponding to 0.7 -1.1% of the foliar-applied Zn.
5.3.2 Plant Zn concentration as affected by different Zn rates Zinc application of 15 or 30 kg Zn ha' 1 did not affect the dry matter yields at any growth stage, and only the means of the dry matter produced at different growth stages are presented (Table 54). There was a decreasing trend in Zn concentration of vegetative plant material in the course of the growing season (Table 55). In 1991, Zn concentrations in the samples taken at the Feekes 2 and 5 growth stages were higher than in the vegetative parts of the plants at later growth stages in both soils. In 1992, the samples taken at Feekes 2 growth stage had a significantly higher Zn concentration than the vegetative parts of the later growth stages. The higher Zn concentration in grain, as compared to that of straw, suggests an effective translocation of Zn from the vegetativeparts. The phenomenon was pronounced in fine sand where Zn concentration of the straw was extremely low in 1992. Even though ZnAc was of the same magnitude in both soils, Zn concentration of barley at any growth stage was higher in clay soil than in fine sand which had a higher pH. In clay, Zn application elevated Zn concentration of barley only in the first experimental year (F = 12.503 ) and showed no residual effect in the second year. In fine sand where Zn application of 5.4 kg ha' 1 to the soil (see Section 5.3.1) did not affect Zn concentration ofbarley, the higher rates in the present experiment had a significant effect in the first year (F = 19.820 ) and there was an increase in plant Zn concentration (F = 4.107 ) 1 The grain and straw samples of 1991 from the experiment on the clay soil were destroyed by fire. 1 Zinc concentration and uptake were tested separately in each soil, each year and growth stage.
2 In 1991, Zn uptake in the clay soils refers to the sampling at Feekes 10.5 growth stage. also in the second year. In neither of the soils did Zn concentration of the plants at Zn rate 15 kg ha" 1 differfrom that at Zn rate 30 kg ha" 1 . The utilization of applied Zn was extremely low, 0.2 -0.3%. When the plant Zn concentrations were compared separately at each sampling (Table 55), the differences between the Zn rates were not always statistically significant even in the first year, owing to large variation in the Zn concentration of plants fertilized with Zn. The influence of Zn application on Zn concentration of barley was greater in the early growth stages than later in the growing season (Feekes 10.5), reflecting the accumulation of dry matter and suggesting that Zn was taken up at the early part of the growing season. Soil analyses at the end of the experiments showed that the effect of the application of high rates of ZnSCH on soil ZnAc was rather small and statistically insignificant both in clay (F = 1,554 n s ) and in fine sand (F = 2.298 n s  (1986,1992) on Zn fertilization of timothy show that the efficiency of fertilizer Zn can be much higher in strongly acid peat soil, and even a small application of Zn (0.55 kg ha' 1 ) may elevate plant Zn concentration significantly.
The low or absent response ofZn concentration of barley grains to 4.8 or 5.4 kg Zn ha' 1 applied to soil is in agreement with other Finnish field experiments where small rates (1.75 kg ha' 1 ) of Zn have been given (Jaakkola andVogt 1978, Syvälahti andKorkman 1978). A higher Zn application (15 and 30 kg ha' 1 ) elevated Zn concentration of barley significantly in fine sand, but even then Zn concentration of grain remained around 20 mg kg' 1 , reflecting the poor availability of Zn in neutral soil. However, the response was of the same level as has been observed in the neutral and slightly acid mineral soils of Norway and Canada where 50 and 20 kg Zn ha' 1 , respectively, were applied to barley (Myhr 1988, Gupta 1989. The drought in 1992 may partly explain the small residual effect of the high Zn rates (15 and 30 kg ha' 1 ) on barley. In the first year of the experiment (1991) there was plenty of rain in May and June, and the roots of barley were probably active in the plough layer, resulting in the observed response to applied Zn. Owingto the drought in the second year (1992) the plough layer was dry and plant roots were able to take up nutrients from that part of the soil less effectively. In 1992, the roots probably grew to a greater extent into the deeper soil layers where they were no more in contact with the applied Zn. This hypothesis is supported by the findings made in Canada by Dwyer et al. (1988) according to which the maximum rooting depth of barley and the quantity of roots in the deeper soil layers increase when there is shortage of water in the surface soil. The hypothesis does not, however, explain why there was some response by barley to soil-applied Zn also in 1992 in clay soil in the other experiment where the different Zn fertilizers were tested (Section 5.3.1).
Granulation and spot-placement commonly decrease the agronomic efficiency ofZnSOa added to neutral and calcareous soils (Brown and Krantz 1966, Allen and Terman 1967, Mortvedt and Giordano 1969. However, in the present study, carried out in acid soils, the granulated 'Sinkkilannos' mixed into the soil elevated plant Zn concentration at the same rate as did ZnSOa. In the granulated product, ZnSOa is incorporated in gypsum, and Zn cations seem to be readily released from the matrix into the soil solution. Also the top-dressed 'Sinkkilannos' increased Zn concentration of grass in the field and pot experiments at least as effectively as did the fertilizer mixed in the soil. Availability of Zn in granular fertilizers added onto the soil surface requires that Zn be dissolved from the granule and further to move into the soil and to get into contact with active plant roots. The rains in the early summer of 1991 right after broadcasting the fertilizers or the daily watering in the pot experiment probably resulted in an effective disintegration of the granules of'Sinkkilannos' and enhanced the penetration of Zn into the root zone. In a shortterm pot experiment, top-dressing retarded the fertilizer effect of 'Sinkkilannos', but in the field there was evidently enough time for the surface-applied granules of 'Sinkkilannos' to dissolve before the first harvest, and consequently there was no difference between the application methods. The negligible agronomic efficiency of Zn contained in NPK fertilizers cannot solely be attributed to the granular form of the fertilizers because the granular 'Sinkkilannos' did elevate plant Zn concentration in the very same experiments. The low efficiency is rather due to the chemical reactions occurring in the fertilizer between Zn and the other components. During the manufacturing process of the present NPK fertilizers the acid orthophosphate slurry is ammoniated (Kivioja 1987), resulting in an elevation of pH. The accompanying decrease in the water-solubility and plant-availability of Zn added to the fertilizer as ZnSOa (Mortvedt 1968, Mortvedt andGiordano 1969 b) is due to the precipitation of insoluble Zn compounds (e.g. Zn phosphates, Zn hydroxides) in the fertilizer grain , Allen and Terman 1967, Mortvedt and Giordano 1969. According to Mortvedt (1968), above pH 5 the availability of Zn in ammoniated orthophosphate fertilizers is less than 20% of what is observed when ZnSCH is applied separately or incorporated in unammoniated (pH 3) orthophosphate fertilizer. The pH of the present NPK fertilizers ranged between 5.0 and 5.4, suggesting a low water-solubility ofZn in the fertilizer. The present field and pot experiments showed that the sparingly soluble Zn compounds are not necessarily dissolved during short-term experiments even in acid soils, resulting in an inconsistent fertilizer effect. Because the NPK fertilizers coated with ZnSO4 had an equally low availability of Zn, it is likely that ionic activities also in the vicinity of the fertilizer granule exceed the solubility products of sparingly soluble Zn compounds.
The current results disagree with those of Sillanpää (1990) who applied Zn-containing NPK fertilizers to barley in ten field experiments in Finland. In those experiments, high Zn rate (11.6 kg ha' 1 ) and low soil pH (CaCb-pH 4.2 -5.5) probably facilitated the mean increase of 8 mg kg' 1 in grain Zn concentration. Moreover, in the fertilizers of Sillanpää, part of Zn (2 kg ha' 1 ) had been added as Zn-EDTA. According to Mortvedt and Giordano (1969 a), the plant-availability ofZn added as Zn-EDTA in macronutrient fertilizers is not reduced as much as that ofZnSO4.
Increased Zn concentration of barley straw by foliar sprays of Na2Zn-EDTA can at least partly be caused by the adsorption offoliar-applied Zn on the surfaces ofplant leaves. Therefore the increased Zn concentration of straw by this treatment must not be considered an indication of high fertilizer efficiency. Because NazZn-EDTA was sprayed at an early growth stage before there was any shoot or ear in the crop, the increase of grain Zn concentration can be attributed to the introduction of applied Zn into the physiological reactions of the plant. However, soil pH strongly dominated the grain Zn concentration also in this treatment. In some studies application of Zn in a chelated form to neutral or calcareous soils has been at least twice as effective as application of ZnSOa (Mortvedt and Giordano 1969 a, Boawn 1973, Hergert et al. 1984. The difference between the two sources has not been significant in acid soils (Hergert et al. 1984) and not always even in neutral soils (SCHNAPpinger et al. 1972). Also in the present study, the soil-applied Na2Zn-EDTA failed to increase plant Zn concentration, which shows that the observed effect of foliar application of NazZn-EDTA can be attributed primarily to the application method rather than to the chelated form ofZn. This conclusion is supported indirectly by the results ofPATER-SON et al. (1991) who elevated Zn concentration of barley grains with foliar sprays of ZnSOa, while the application of ZnSO4 to the soil was without effect.
Further, it needs to be pointed out that foliar application of Na2Zn-EDTA was the only treatment increasing Zn concentration of barley grain in fine sand where insufficient Zn supply to barley may have been approached.
6 GENERAL DISCUSSION AND CONCLUSIONS in spite of generally sufficient Zn reserves for plant growth the present material contained a few soils poor in Zn. Even though worldwide zinc deficiency is commonly connected to calcareous soils, low plant-availability of Zn may occur also in slightly acid and neutral mineral soils of Finland. Moreover, the soil testing method applied in Finland (AAAc-EDTA, pH 4.65) seems to overestimate the Zn supply to plants in these soils. However, the soils with the scarcest reserves were among Carex peat and Ligno Carexpeat soils, which were poor in Zn also according to international comparisons.
Owing to the small number of organogenic soils in the present material, it is not possible to draw conclusions on the geographic occurrence of Zn deficiency in Finland even though cultivated peat soils are the most common in the northern parts of the country. Peat soils constitute less than 10% of the cultivated area of Finland (Kurki 1982), but they are not all poor in Zn. Only 1.5% of samples analyzed for ZnAc in soil testing in 1986 -1988 contained ZnAc less than 1.0 mg dm' (unpublished data of Soil Analysis Service Ltd.) and were rated poor in Zn according to the current interpretation (Viljavuuspalvelu 1992). The present results as well as those of Urvas (1985,1990) suggest that yield response to applied Zn is hardly observed in short-term experiments even in these soils. Consequently, a ZnAc concentrationbelow 1.0mg dm" in the soil does not necessarily indicate insufficient Zn supply to the crop. In accordance with the results of soil analyses and pot experiments, yield increases owing to Zn fertilization have not been detected in the field in Finland. Therefore, Zn fertilization in the great majority of cultivated soils of Finland is justified by the possible elevation of Zn concentration of the crop. It seems feasible that in the poorest peat soils Zn reserves can be exhausted over time in intensive grassland cultivation. In the studies of Sillanpää and Rinne (1975) the quantity of Zn harvested in three cuttings ofsilage grass amounted to 280 g ha" 1 at the annual N fertilizer level of 300 kg ha' 1 . At this rate, the uptake ofZn in 10years amounts to 2.8 kg 1 3 ha , or 1.4 mg dm in a 20-cm layer. This corres-ponds to the reserves ofsecondary Zn in the poorest peat soils. Low pH of most peat soils further contributes to the high availability and effective utilization of soil Zn. The above calculation supports the recent finding by Erviö et al. (1990) of the decline of ZnAc in the cultivated soils of northern Finland. Soil characteristics strongly affect the response of plant Zn concentration to Zn fertilization. In strongly acid soils Zn application elevates Zn concentration of grass, also facilitating the maintenance of sufficient supply ofZn to plants in peat soils under intensive grassland cultivation. But even high rates of Zn to slightly acid and neutral soils elevate the Zn content of the crop less effectively even if the soils were poor in ZnAc-In those soils, foliar sprays increase plant Zn concentration more effectively. In order to avoid applications of Zn to soil with no fertilizer effect, both soil pH and ZnAc need to be taken into consideration when Zn fertilizer recommendations are given. Owing to the inconsistent effect of Zn-containing NPK fertilizers on Zn concentration ofcrop, the use of separate Zn fertilizers should be preferred.
The reserves of secondary Zn (10 -20 kg ha" 1 in a 20-cm deep plough layer) were of the same magnitude as Zn fertilizer recommendations (5 -20 kg Zn ha' 1 , Viljavuuspalvelu 1992). The utilization of added Zn is commonly far below 5% and therefore the recommended application substantially increases the reserves of secondary Zn in soil. The low utilization is caused by the strong adsorption of Zn in the soil and not by the reluctance of the plants to take up Zn. This conclusion can be drawn on the basis of the results of pot experiments where high Zn concentration in the grass occurred when Zn was added to an unhumified Sphagnum peat and to strongly acid coarse mineral soils of obviously low Zn adsorption capacities.
The large variations of dry matter yield of timothy and barley in the field and ryegrass in the pot experiment were not reflected as a negative correlation between the size of the yield and the Zn concentration. This suggests that Zn uptake by the plants is probably not limited by the capacity factor (quantity ofplant-available Zn in soil) but rather by intensity (Zn concentration in soil solution). The adsorbed Zn fraction is much larger than the dissolved one, resulting in a strong buffering of soil Zn concentration (Elgawhary et al. 1970). As Zn is taken up by plant roots from the soil solution, the decrease of concentration is readily replenished from the adsorbed fraction, provided the soil is not poor in Zn. It can therefore be concluded that plant uptake does not markedly reduce the Zn concentration of soil solution. Consequently, the Zn concentration of plant tissue grown in a given soil can be the same regardless of the size of the yield.
The average Zn concentration of cereal crops grown in Finland is at the same level as in other countries of temperate climates. In Norway, in Prince Edward Island, Canada and in southwestern Sweden, mean Zn concentrations ofbarley and oats have ranged between 28 and 48 mg Zn kg' 1 (FROSLIE et al. 1983, Winter and Gupta 1987, Eriksson et al. 1990. Also timothy grown in Finland has on average at least the same Zn concentration as has been reported elsewhere (METSON et al. 1979, Winter and Gupta 1983,BoiLAetal. 1985. Owing to the great variation in Zn concentrations of timothy and barley (Jaakkola andVogt 1978, Kähäri andNissinen 1978), it is likely that in areas of poor soil Zn, locally produced fodder may contain much less Zn than is the national average.
The level of dietary Zn (50 mg kg' 1 ) recommended for cattle in the Nordic countries (NJF 1975, Salo et al. 1990 appears to be high as compared to recommendations given elsewhere. In the USA, a concentration of 40 mg kg' 1 is recommended for dairy cattle (NRC 1978) and 20 -30 mg kg' 1 for beef cattle (NRC 1976). In New Zealand, a recommendation of 15 -25 mg kg' 1 for grazing livestock is given (Towers and Grace 1983). In a compilation prepared in England (ARC 1980) 30 mg kg' 1 was regarded as the sufficient level in experimental conditions, but it was also pointed out that in studies made in the field higher concentrations have occasionally been of advantage. The average Zn concentration of timothy and barley occurring in Finland would thus be considered sufficient for cattle in most countries, and the evidence of a general need of a higher Zn level in fodder is not conclusive. Zinc concentration of timothy and barley at least in mineral soils does not reach the level recommended in Finland (50 mg kg' 1 ) without excessive Zn fertilization. Because high Zn rates result in an undue accumulation of Zn in the soil, it seems needless to aim at Zn concentrations beyond the current average level in the crop by increased Zn fertilization. The recommended Zn level in the diet should still be reached by direct supplementation into the fodder.
Besides soil Zn status, pH and Zn fertilization, also other factors affect the actual Zn concentration in fodder of domestic animals. Clover and other dicotyledons have a higher Zn concentration than gramineous fodder crops (Reay and Marsh 1976, Yläranta and Sillanpää 1984, McLaren et al. 1991. Nitrogen fertilization has also been shown to elevate plant Zn concentration, probably owing to the decrease of pH (Boawn et al. 1960. For example, in field experiments by Rinne at al. (1974) and Sillanpää and Rinne (1975) Zn concentration in the grass increased from 30 mg kg' 1 to 39 mg kg' 1 when the N fertilization was increased from nil to 600 kg ha' 1 . On the other hand, the experiments of ETTALA and KOSSILA (1979,1980) showed that on average 34% ofZn in silage grass was lost during the ensiling. These examples propose that the choice of crop as well as different agricultural practices other than Zn fertilization can affect the Zn content of the fodder at least as much as was commonly observed to be the effect of Zn application in the current field experiments with timothy. Also excessive Zn concentration in plants need to be considered. The highest concentration in ryegrass grown without addedZn, occurring in the soil of Harjavalta, exceeded 100 mg kg' 1 and was of the same magnitude as was reported in grass grown in a Zn-contaminated harbor dredge in the Netherlands (Smilde et al. 1982). High concentrations of Zn can thus occur locally in the neighborhood of industry also in Finland. The high concentration of the grass grown in the soil of Harjavalta also shows that Zn accumulated in the soil probably as a result of atmospheric deposition was plant-available. This conclusion is corroborated by findings showing that the bulk of Zn in the deposition both in urban (Gatz and Chu 1984) and rural (Lindberg and Harriss 1981) environments is water-soluble. Various plants may exhibit symptoms of Zn toxicity when the Zn concentration of the plant exceeds 120 -220 mg kg" 1 (Beckett andDavis 1977, Sauerbeck 1982) but according to the present study, as much as 500 -700 mg Zn kg" 1 was tolerated by ryegrass with only slight adverse effects. This result agrees with those by Gerzabeck and Schaffer (1989) according to whom the toxicity limit in ryegrass was higher than 400 mg kg" 1 . Tolerance to large doses of Zn by domestic animals appears to be even greater (Miller et al. 1965, Ott et al. 1966). According to Ott et al. (1966), toxicity symptoms occurred only when the dietary Zn concentration exceeded 900 mg kg" 1 .
The highest Zn fertilization rate currently recommended in Finland is 20 kg ha" 1 (Viljavuuspalvelu 1992), corresponding to 10mg dm"' in a 20-cm thick plough layer. The present pot experiments and field experiments by Urvas (1992) showed that at least in strongly acid coarse mineral soils and peat soils Zn concentration of grass can be elevated beyond the recommended level (50 mg kg" 1 ) by application of maximum recommended Zn doses. However, in unpolluted cultivated soils Zn concentration of grass fertilized at that Zn rate is likely to remain below 100 mg kg" 1 . It seems thus evident that aZn concentration toxic to plants or animals cannot be reached when field crops are fertilized with the recommended rates of Zn.