A G R I C U LT U R A L A N D F O O D S C I E N C E Agricultural and Food Science (2022) 31: 136–148 136 https://doi.org/10.23986/afsci.115313 Site characteristics determine the duration of structure liming effects on clay soil Jens Blomquist1, Jan-Eric Englund2 and Kerstin Berglund1 1Department of Soil and Environment, Swedish University of Agricultural Sciences, Box 7014, 750 07 Uppsala, Sweden 2Department of Biosystems and Technology, Swedish University of Agricultural Sciences, Box 190, 234 22 Lomma, Sweden e-mail: jens.blomquist@slu.se Adding carbonated or non-carbonated lime to clay soils can lead to changes in aggregate stability. In Sweden, ‘struc- ture liming’ with a mixed product (normally 80–85% calcium carbonate and 15–20% calcium hydroxide) is subsidised through environmental schemes to increase aggregate stability, thereby mitigating losses of particulate phosphorus (PP). This study assessed the effects of structure liming on aggregate stability in eight clay soils in southern Swe- den, using turbidity as a proxy for aggregate stability. Turbidity in leachate from simulated rain events performed on aggregates (2–5 mm) in the laboratory was measured one and six years after application of four treatments 0, 4, 8 and 16 t ha-1 of a mixed structure liming product. The effect on turbidity was analysed for all application rates, but also as the contrast between the unlimed control and the mean of the limed treatments, to identify the gen- eral effect. A significant effect of structure liming on turbidity was found after one year. The effect decreased over time, but was still detectable after six years. However, there was a significant interaction between trial and treat- ment, indicating different reactions on different soils and suggesting that not all clay soils are suitable for structure liming if the desired objective is to lower the risk of PP losses. Clay content, initial pH and mineralogy may explain the different responses to structure liming. These findings show a need for a site-specific structure liming strategy. As a tentative recommendation, soils with a minimum clay content of approximately 25–30% and pH <7 should be preferred for structure liming. Key words: structure liming, aggregate stability, pH, turbidity, clay Introduction Adding non-carbonated lime (calcium oxide or calcium hydroxide) to clay soils leads to a series of reactions involving i) cation exchange, ii) flocculation, iii) carbonation and iv) pozzolanic reactions, as outlined by Choquette et al. (1987). Pozzolanic reactions are of a long-term nature (Firoozi et al. 2017), as new cementing products are formed when siliceous and aluminous materials react chemically with lime in the form of calcium hydroxide. Pozzolanic reactions are the main mechanism ensuring good soil-lime stabilisation, according to Choquette et al. (1987). These lime-soil reactions can result in changes in soil physical properties, as demonstrated in Swedish field trials with burnt (calcium oxide) and slaked (calcium hydroxide) lime during the 1960s and 1970s (Berglund 1977, Ledin 1981) and with ground limestone (calcium carbonate) used as reference in some later studies (Siman et al. 1984). The focus of these previous studies was normally on the agronomic aspects of liming, such as soil shrinkage, ag- gregation and penetration resistance, and they reported positive effects on the physical status of clay soils. How- ever, use of liming with non-carbonated lime in Swedish agriculture remained limited, as burnt and slaked lime products entailed high investment costs (Wiklander 1963). Interest in ‘structure liming’ with a mixed product, normally 80–85% calcium carbonate and 15–20% calcium hy- droxide, emerged in Sweden around 2010, driven by national environmental schemes under the EU Water Frame- work Directive (EC 2021) that subsidise up to 50% of the costs of liming (HaV 2021). The focus in these environ- mental schemes is mainly on the environmental aspect, as losses of particulate phosphorus (PP) have been shown to decrease when lime is mixed with clay soil (Ulén and Etana 2014). The practice is therefore relatively wide- spread, with around 65,000 hectares in Sweden being structure limed between 2010 and 2021, corresponding to approximately 2.5% of the arable land. Positive short-term effects on aggregate stability and a reduced risk of phosphorus (P) losses as a result of in- creasing application rates of structure liming products have been reported. For example, Blomquist et al. (2022) observed increased aggregate stability and an associated reduced risk of P losses about one year after struc- ture liming, while Blomquist et al. (2018) detected effects of the same magnitude 2.5 years after structure lim- ing. In contrast, Norberg et al. (2021) found no significant effect on aggregate stability one and two years after structure liming on a clay soil where effects could be expected. Thus the effect seems to depend on the inherent Received 8 March 2022 / Accepted 25 September 2022 The Scientific Agricultural Society of Finland ©This is an open access article under the CC BY 4.012 J. Blomquist et al. 137 characteristics of the soil and manifests as an interaction between soil type and lime treatment, with some soils responding to structure liming with markedly increased aggregate stability and other soils responding much less or remaining unaffected by structure liming (Gunnarsson et al. 2022). This is in line with findings by Keiblinger et al. (2016) of varying effects of liming on aggregate stability in soils with different clay content. Soil pH also plays a role, with soil structure being positively influenced on acid soils by an increase in base saturation through cation exchange and a decrease in distances between clay particles leading to aggregation (Wiklander, 1963). According to Bell (1996), pozzolanic reactions occur when the affinity of a soil for lime has been satisfied, which can be in- terpreted as a rise to 100% base saturation, with a simultaneous increase in pH. The possibility for pozzolanic re- actions to occur should therefore be greater in naturally alkaline soils or in soils where the pH has been elevated due to liming as a management practice. In this study, aggregate stability was measured one and six years after structure liming, in order to compare the duration of the effect in the slightly longer term. Specific objectives were i) to assess changes in aggregate stabil- ity over time after structure liming; ii) to determine whether different soils react differently to structure lime; and iii) to identify critical soil variables determining the effect of structure lime on aggregate stability. Materials and methods Study area and sites Structure lime (Nordkalk Aktiv Struktur, Nordfalk Corp., Finland) was spread and incorporated in eight field tri- als at four sites (Krageholm, Lönhult, Vadensjö, Kornheddinge) in Scania County, southern Sweden, in August and September 2014. To assess the effect of structure liming on aggregate stability in soils with varying clay content, trials with low clay content (LC) and high clay content (HC) were selected at each site. All trials followed the crop rotation and management practices on the respective farm. Site coordinates and date of structure liming are shown in Table 1, together with the crops grown in the two sampling years (2015 and 2020). After spreading in 2014, the structure lime was incorporated using different till- age non-inversion implements, working with discs, tines or combinations of these, to 5–15 cm depth. At all sites, incorporation was performed in 2–3 passes on the same day as spreading or within 1–2 days. At Vadensjö, an ad- ditional fourth pass was made approximately three weeks after spreading. No inversion mouldboard ploughing was performed at any of the sites in autumn 2014. In the years from 2015 to the repeated sampling in 2020 the eight field trials carried a variety of crops such as winter wheat, spring barley, spring wheat, oats, sugar beet, winter oil seed rape and red fescue for seed produc- tion. At trial sites Krageholm and Lönhult the soil was mouldboard ploughed once in the autumn of 2018 and at trial site Vadensjö once in the autumn of 2015. Kornheddinge LC was mouldboard ploughed before and after sug- ar beet in 2016 and likewise Kornheddinge HC was ploughed before and after sugar beet in 2018, i.e. twice until the sampling in 2020. In all other years, when the trials were not mouldboard ploughed, the primary tillage in au- tumn was performed with non-inversion tillage. Experimental design and liming product The plots on each soil type were laid out in a randomised block design with three replicates, and subjected to one of four treatments with increasing application rates of structure lime. The standard application rate in Sweden is approximately 8 t ha-1 of a mixed liming product. The treatments in the trials were: an unlimed control (SL0); half Table 1. Site coordinates, dates of spreading-incorporation of structure lime in autumn 2014 and crops grown in the soil aggregate sampling years 2015 and 2020 Sites Coordinates Date lime application and incorporation Crop 2015 Crop 2020 Krageholm 55.51 °N, 13.75 °E 5 September 2014 WW WOSR Lönhult 56.19 °N, 12.71 °E 28 August 2014 WW SB Vadensjö 55.92 °N, 12.86 °E 27 August 2014 SuB SuB Kornheddinge 55.63 °N, 13.29 °E 20 and 28 September 2014 WW WW WW = winter wheat; SuB = sugar beet; SB = spring barley; WOSR = winter oilseed rape Agricultural and Food Science (2022) 31: 136–148 138 the standard application rate (4 t ha-1 structure lime; SL0.5), the standard application rate (8 t ha-1; SL1); and twice the standard application rate (16 t ha-1; SL2). The liming product used was a mixture of approximately 80–85% ground limestone and 15–20% slaked lime, with a water content of 15–25% and bulk density 0.8–1.0 t m-3. The chemical composition of the product is shown in Table 2. Clay mineralogy determination One topsoil sample (0–20 cm) per trial or site, consisting of a pooled sample from all plots on the particular trial, was examined with X-ray diffraction analysis to determine the mineralogy of the clay fraction. To assess this semi- quantitatively, the intensity (surface) of the three most pronounced clay mineralogy reflections in the X-ray curves at 14, 10 and 7 Å was measured. The percentage contribution from each of these three reflection curves was then used to calculate SmV index as: Intensity 14 Å/(Intensity 10 Å + Intensity 7 Å), as an estimate of the relationship between swelling and non-swelling clay minerals. SmV index also indicates the relative proportions of clay miner- als with high (smectite and vermiculite, intensity 14 Å) and low (illite, chlorite, kaoline, intensity 10 Å + intensity 7 Å) cation exchange capacity (CEC). A soil with a high proportion of swelling clay minerals is thus characterised by a high SmV index. Soil characteristics Initial soil chemical characteristics, together with soil organic matter content, texture and clay mineralogy (SmV index) are summarised in Table 3 for the eight field trials. The SmV index varied between 0.6 and 3.1 at the tri- als, i.e. with the relative proportion of swelling clay minerals being five-fold higher in Kornheddinge LC than in Krageholm LC/HC. Table 2. Chemical composition (dry matter basis) of the structure lime product Nordkalk Aktiv Struktur (approximately 80–85% ground limestone and 15–20% slaked lime) spread in field trials in August–September 2014. Water content 15–25% depending on storage. Particle size 0–0.5 mm with 99% passing through mesh size 0.5 mm. Source: Nordkalk Corp. Macronutrient or compound Conc. (%) Micronutrient Conc. (mg kg-1) Total Ca as CaO 51 Cd 1.8 Mg 1.0 Co 9 SiO 2 5.2 Cr 26 Al 2 O 3 2.9 Cu 45 Fe 2 O 3 1.1 Hg < 0.02 K 1.5 Ni 26 Na 2 O 0.5 Pb 58 S 1.5 Zn 260 P 0.07 V 38 Table 3. Mean (of 12 plots) soil texture, soil organic matter (SOM), nutrient status and clay mineralogy (SmV index) in topsoil (0–20 cm) in the eight field trials prior to structure liming in August–September 2014, HC/LC = high/low clay soil at each trial Trial SOMa Sandb Siltc Clayd pH P-ALe K-ALe Mg-ALe K/Mg Ca-ALe SmV % >60 µm 2–60 µm <2 µm H 2 O mg kg-1 mg kg-1 mg kg-1 quota mg kg-1 index Krageholm LC 2.4 39.5 41.4 19.1 6.2 25 82 95 0.9 1566 0.6 Krageholm HC 2.5 19.5 52.7 27.8 6.6 29 111 151 0.8 1850 0.6 Lönhult LC 5.0 43.4 28.5 28.1 7.7 129 188 146 1.4 4042 1.1 Lönhult HC 3.9 29.1 24.2 46.8 8.2 127 307 353 0.9 10542 1.1 Vadensjö LC 4.0 46.3 32.8 20.9 7.2 130 97 121 0.8 3100 2.9 Vadensjö HC 3.3 44.1 29.5 26.4 8.2 89 148 229 0.7 11425 2.9 Kornheddinge LC 2.8 53.8 25.4 20.8 7.4 48 99 87 1.1 3042 3.1 Kornheddinge HC 2.6 45.8 25.8 28.5 7.8 114 203 218 1.1 7050 1.9 asoil organic matter, measured as loss of ignition corrected for clay content; bwet sieving; ccalculated as difference between the sand and clay fractions; dhydrometer sedimentation; eextraction with 0.1 M ammonium lactate + 0.4 M acetic acid, pH 3.75 (Egnér et al. 1960) J. Blomquist et al. 139 Soil classification according to the FAO World Reference Base for Soil Resources (WRB) was not performed at the experimental sites, but has been done previously for neighbouring Swedish long-term soil fertility trials (Kirch- mann et al. 1996, Carlgren and Mattsson 2001). According to the results, the Krageholm, Lönhult and Kornhed- dinge soils can be classified as Haplic Phaeozems and Vadensjö as a Eutric Cambisol. Data on pH and clay content for the eight field trials soils are depicted in Figure 1, to show the range in properties expected to be decisive for soil aggregate stability. Follow-up analyses of soil chemical variables in all plots were performed in both 2015 and 2020. Aggregate stability, sampling and measurements Soil sampling for aggregate stability measurements was performed in one year (2015) and six years (2020) af- ter liming. Sampling at Vadensjö in 2015 was performed in spring, shortly after drilling of the sugar beet crop, and sampling in 2020 was performed in summer in the growing sugar beet crop, as autumn sampling was con- sidered unsuitable. At all other sites, both samplings were performed in autumn after harvest of a cereal or oil- seed crop and shallow tillage to approximately 6–8 cm. Loose soil from the tilled depth was collected and sieved into three different size classes (with mean aggregate diameter >5 mm, 2–5 mm and <2 mm, respectively). The fraction of aggregates with diameter 2–5 mm was stored dry and aerated, and later transported to the labora- tory at the Swedish University of Agricultural Sciences (SLU) in Uppsala. At the laboratory, the aggregates were air-dried again to equalise differences in water content and then subjected to two simulated rainfall events 24 hours apart, with a rainfall intensity of 32–39 mm per hour in each event. Each rainfall event lasted for one hour and the irrigation boom moved back and forth continuously without stopping at the ends during irrigation, so the aggregates were subjected to simulated rain for five minutes in the one-hour period. The leachate was collected after each rainfall event for analysis of turbidity and electrical conductivity (EC). Only results for the samples col- lected after the second rainfall event, denoted Turbidity A2 and EC A2, respectively, are reported in this paper as they have shown less variation in earlier studies. The muddy leachate samples were shaken for 10 minutes and then allowed to sediment for 4.5 hours. After sedimentation, a supernatant sample was taken at 5.6 cm depth and turbidity measured with a turbidimeter (Hach TL 2360 Turbidimeter, Hach, Loveland, Co.). The turbidity val- ues obtained provided an estimate of the clay concentration in the leachate (Etana et al. 2009) and were thus a proxy for aggregate stability and risk of particulate P (PP) losses (Puustinen et al. 2005, Ulén et al. 2012). Electri- cal conductivity in soil increases after liming, so the EC values in leachate were used as an indicator of presence of soluble calcium in the soil solution. Fig. 1. Average initial pH H2O and clay content in the eight field trials at spreading and incorporation of structure lime in August–September 2014. Kra=Krageholm, Lön=Lönhult, Vad=Vadensjö, Kor=Kornheddinge, LC = lower clay content, HC = higher clay content. Actual clay content is given in Table 3. Agricultural and Food Science (2022) 31: 136–148 140 Statistical analyses The data obtained for pH, ammonium lactate (AL)-extracted calcium, turbidity and EC (Turbidity A2 and EC A2) were subjected to analysis of variance (ANOVA) (General Linear Model (GLM) in Minitab 18, Minitab Inc.). The mean values obtained were expressed in absolute and relative terms (unlimed control = 100). In the GLM, trial, treatment and the interaction between trial and treatment were used as fixed factors, and block nested in trial as a random factor in a hierarchical model. Pairwise comparisons of treatment averages were made according to Tukey’s test. Unless otherwise stated, the significance level used in the calculations was p < 0.05. To test the overall effect of liming versus the unlimed control, the control values were compared against the mean of the three structure lime treatments using a contrast. For the comparisons, SAS software (SAS 9.4) was used. Results pH and calcium changes Soil pH and calcium content (Ca-AL) in the different treatments before liming (2014), one year after liming (2015) and six years after liming (2000) are shown in Figure 2. One year after liming, pH was significantly higher (p < 0.001) in all liming treatments compared with the unlimed control, with an increase in treatment SL0.5, SL1 and SL2 of 0.2, 0.4 and 0.6 pH units, respectively (Fig. 2a). How- ever, further analysis revealed a significant interaction between trial and treatment (p = 0.007) (Fig. 3). 7.4a 7.4a 7.0a 7.4a 7.6b 7.2ab 7.5a 7.8c 7.3b 7.5a 8.0d 7.6c 6.0 6.5 7.0 7.5 8.0 8.5 2014 2015 2020 pH SL0 SL0.5 SL1 SL2 a a aa a a a ab aa b a 0 1000 2000 3000 4000 5000 6000 7000 8000 9000 2014 2015 2020 Ca (m g kg -1 so il) SL0 SL0.5 SL1 SL2 Fig. 2. (a) Soil pH and (b) ammonium lactate-extracted calcium (Ca-AL) content before structure liming (2014), one year after liming (2015) and six years after liming (2020) in the unlimed control (SL0) and in treatments receiving 4, 8 and 16 t ha-1 structure lime (SL0.5, SL1 and SL2, respectively). Mean for all eight field trials. Different letters on bars indicate significant differences between treatments within the year. a) b) J. Blomquist et al. 141 The soil at the different experimental trials showed different degrees of pH increase in the liming treatments (Fig. 3). The soils at the two Krageholm trials, both with initial pH of around 6.5 (see Fig. 1), reacted to increasing applica- tion rates with consistent pH increases, while the Lönhult HC and Vadensjö HC soils, with initial pH values above 8, did not show any further pH increase in the lime treatments. Six years after liming, significant differences in pH were detected only in treatments SL1 and SL2 (Fig. 2a) com- pared with the unlimed control (p < 0.001). There was no significant interaction between trial and treatment (p = 0.18) at that time. The initial calcium content (Ca-AL) varied widely between the treatments (Fig. 2b), but the differences were not statistically significant. In the first year after liming, only the double liming rate treatment (SL2) differed signifi- cantly (p = 0.001) from the unlimed control and there was no significant interaction between trial and treatment (p = 0.70). Six years after liming, the ANOVA results showed significant differences (p = 0.035), but the pairwise comparisons showed no significant differences in soil calcium content between the treatments. There was no sig- nificant interaction between trial and lime treatment (p = 0.52). Aggregate stability Aggregate stability (Turbidity A2) of 2–5 mm aggregates from field trials sampled one and six years after liming are shown in Figure 4a, while the corresponding values of electrical conductivity in leachate (EC A2) are shown in Figure 4b. One year after structure liming, there were statistically significant differences in turbidity between the field trials (p < 0.001) and between the treatments (p = 0.001). In treatments SL1 and SL2, turbidity decreased significantly, by 33% and 34%, respectively, while in treatment SL0.5 there was a non-significant decrease of 22% (Fig. 4a). There were no significant differences between the limed treatments (SL0.5–SL2). However, there was a significant inter- action between treatment and trial (p = 0.049), indicating that the eight field trials reacted differently to structure liming. Electrical conductivity (EC A2) also differed significantly between trials (p < 0.001) and between treatments (p < 0.001) (Fig. 4b), but with no significant interaction between site and treatment (p = 0.23) (Fig. 4). In treat- ments SL1 and SL2, EC A2 increased significantly, by 18 and 29%, whereas treatment SL0.5 (10% increase) was not significantly different from the unlimed control (SL0). Six years after structure liming, the significant differences in turbidity between trials persisted (p < 0.001), but there was no longer a significant difference due to liming treatment (p = 0.102) (Fig. 4a). However, there was still a significant interaction (p = 0.003) between trial and treatment (Fig. 4a). For electric conductivity, no significant differences were found between the treatments (p = 0.55) (Fig. 4b). 6.0 6.5 7.0 7.5 8.0 8.5 SL0 SL0.5 SL1 SL2 pH Krageholm LC Krageholm HC Lönhult LC Lönhult HC Vadensjö LC Vadensjö HC Kornheddinge LC Kornheddinge HC Fig. 3. Interaction plot for mean soil pH measured in 2015, one year after structure liming, in the unlimed control (SL0) and in treatments receiving 4, 8 and 16 t ha-1 structure lime (SL0.5, SL1 and SL2, respectively) at the different experimental trials. HC/LC = high/low clay. Agricultural and Food Science (2022) 31: 136–148 142 Due to the significant interactions on turbidity, it was not possible to draw general conclusions on aggregate sta- bility from the average for the eight field trials in the first and sixth year after liming. To achieve higher resolu- tion, the data were analysed for the individual sites and sampling years. The full results for the different sites in the sampling years 2015 and 2020 (see Table S1 in Supplementary information) revealed an interaction between site and treatment for both turbidity and electric conductivity. To further clarify these differences in response to structure liming in general compared with the unlimed control, the average values for the three limed treatments (SL0.5, SL1 and SL2) were compared against the control values for all trials and for each site separately (Fig. 5). For the eight field trials combined, there was a significant effect of liming in 2015 (p < 0.001) with a 29% decrease in turbidity in the limed treatments, and also in 2020, with a 10% decrease in turbidity persisting in the limed treat- ments (p = 0.032). However, there was considerable variation between the sites, which was evident as an inter- action between trial and treatment in both 2015 (p = 0.049) and 2020 (p = 0.003). In terms of the individual sites, structure liming only significantly affected the turbidity at Krageholm and Lönhult at the first measurements (in 2015). The soils at the two trials at Krageholm reacted with a 35% reduction in turbidity in that year and the limed treatments showed a significant difference from the control (p = 0.013) (Fig. 5). In 2020, there was still a remain- ing significant difference compared with the control (p = 0.0018) at that site. At the two trials at Lönhult, there was a 61% reduction in turbidity in 2015 (p = 0.018), but the effect of structure liming was no longer significant in 2020 (p = 0.65) (Fig. 5). The Vadensjö and Kornheddinge sites showed no signs of aggregate stability changes, either when grouped together or analysed individually, in either of the sampling years (Fig. 5). Fig. 4. (a) Relative turbidity and (b) relative electrical conductivity in leachate collected after a second simulated rainfall event on 2–5 mm aggregates of tilled soil collected one year (2015) and six years (2020) after structure liming at a rate of 4, 8 and 16 t ha-1 (SL0.5, SL1 and SL2, respectively), calculated as a mean for all trials based on statistical analyses of original data (absolute values) and expressed relative to values obtained for the unlimed control (SL0 = 100). Different letters on bars indicate significant differences between treatments (p < 0.05). 100a 100a 78ab 87a 67b 94a 66b 91a 0 20 40 60 80 100 120 2015 2020 Re l. tu rb id ity A 2 SL0 SL0.5 SL1 SL2 100a 100a 110ab 94a 118bc 105a 129c 100a 0 20 40 60 80 100 120 140 2015 2020 Re l. el . c on du ct iv ity A 2 SL0 SL0.5 SL1 SL2 a) b) J. Blomquist et al. 143 The diminishing treatment effect over time, together with the obvious site-specific nature of the response to liming, demanded even higher resolution in the analysis, so the LC and HC trials at Krageholm and Lönhult were analysed separately (Fig. 6). Structure liming (average of treatments SL0.5–SL2) had no significant effect on turbidity in leachate from Krageholm LC soil aggregates in 2015 (p = 0.76) or 2020 (p = 0.44). In contrast, for Krageholm HC soil the turbidity decreased significantly (by 46%) in 2015 (p = 0.012) and still showed a significant decrease (by 25%) in 2020 (p = 0.003). The trials at Lönhult displayed significant (p = 0.026 at Lönhult LC) and close to significant (p = 0.062 at Lönhult HC) reductions in turbidity in 2015, but no significant effects in 2020 (p = 0.47 at Lönhult LC, p = 0.91 at Lönhult HC). See Table S2 in Supplementary information for full results. In a previous study examining the effect on turbidity approximately one year after structure liming at the eight trials used in the present study together with another 22 field sites, we found that structure liming had the most pronounced short-term effect on soils with high clay content and low initial pH (Blomquist et al. 2022). Therefore, Fig. 5. Relative turbidity as a mean of the three structure liming treatments SL0.5–SL2 (applying 4, 8 and 16 t ha-1 structure lime, respectively), compared with the value in the unlimed control (SL0), where SL0 = 100. Results from the second rainfall simulation on 2–5 mm aggregates of tilled soil collected one year (2015) and six years (2020) after structure liming. Results for all trials (All) and for the individual sites: Krageholm (Kra), Lönhult (Lön), Vadensjö (Vad) and Kornheddinge (Kor). Values in bold italics indicate a significant difference compared with the unlimed control. 71 90 65 78 39 104 96 112 91 95 0 20 40 60 80 100 120 All 2015 All 2020 Kra 2015 Kra 2020 Lön 2015 Lön 2020 Vad 2015 Vad 2020 Kor 2015 Kor 2020 Re l. tu rb id ity A 2 Fig. 6. Relative turbidity as a mean of the three structure liming treatments SL0.5– SL2 (applying 4, 8 and 16 t ha-1 structure lime, respectively), compared with the unlimed control (SL0), where SL0 = 100. Results from the second rainfall simulation on 2–5 mm aggregates of tilled soil collected one year (2015) and six years (2020) after structure liming in trials with low clay content (LC) and high clay content (HC) at Krageholm (Kra) and Lönhult (Lön). Values in bold italics indicate a significant difference compared with the unlimed control. 94 87 54 75 59 109 32 101 0 20 40 60 80 100 120 Kra LC 2015 Kra LC 2020 Kra HC 2015 Kra HC 2020 Lön LC 2015 Lön LC 2020 Lön HC 2015 Lön HC 2020 Re l. tu rb id ity A 2 Agricultural and Food Science (2022) 31: 136–148 144 further analysis was performed on the eight trials in the present study divided into two groups, with low clay con- tent (LC) and high clay content (HC) (Table 4), where low and high refer to the relative clay content at each site (see Table 3). In the LC group, consisting of the trials with the lowest clay content at each site, the clay content varied from 19 to 28 % (group mean 22%). For this group, there was a significant effect on turbidity in 2015 (p = 0.046), but no sig- nificant differences between the treatments in pairwise comparisons, while there was no significant effect of treat- ment in 2020 (p = 0.259). Likewise, the LC group showed a significant treatment effect on electrical conductivity in 2015 (p = 0.008), but no remaining effect in 2020 (p = 0.652). The clay content in soils in the HC group, consisting of the trials with the highest clay content at each site, varied between 26 and 47 % (group mean 32%). For the HC group, there were significant treatment effects on turbidity in both 2015 (p = 0.012) and 2020 (p = 0.015). In 2015, treatments SL1 and SL2 both gave significant reductions, of 35 and 39% respectively compared with the control, whereas in 2020 a significant decrease (of 20%) remained only in SL2. However, in 2020 there was a significant interaction (p = 0.010) between trial and treatment. For this group, there was a significant effect of structure liming on electrical conductivity (p = 0.040) in 2015 and also a significant interaction (p = 0.012) between trial and treatment, but no remaining effect on electrical conductivity at the sampling in 2020 (p = 0.549). Discussion This analysis of soil aggregates from eight trials showed clear positive effects of structure lime on soil pH and aggregate stability one and six years after application. Declining pH effect The comparison between 2015 and 2020 showed that changes in both pH and Ca-AL as a result of structure lim- ing diminished over time. In 2015, there was a significant pH effect in all limed treatments in comparison with the unlimed control, but in 2020 there was only a statistically significant pH effect in treatments SL1 and SL2. This is in line with Blomquist et al. (2018), who reported a time-dependent rise and fall in pH over three years as a result of structure liming. Similarly, Norberg et al. (2021) found an initial rise in soil pH in the first year after structure liming, followed by a decrease in the second year. Table 4. Turbidity A2 and electrical conductivity A2 in 2015 and 2020 for the four trials with relative low clay content (LC) and the four with high clay content (HC). Significant differences are indicated by different letters (p < 0.05). Treatment Turbidity A2 Electrical conductivity A2 Year 2015 2015 2020 2020 2015 2015 2020 2020 Low clay content (LC) Abs. Rel. Abs. Rel. Abs. Rel. Abs. Rel. SL0 4.21a 100 54.8a 100 66a 100 122a 100 SL0.5 4.20a 100 51.0a 93 74ab 113 118a 97 SL1 3.09a 73 54.1a 99 92ab 140 123a 101 SL2 3.27a 78 62.4a 114 100b 153 132a 108 p Trial 0.000 0.051 0.000 0.000 p Treatm 0.046 0.259 0.008 0.652 p Trial*Treatm 0.524 0.467 0.845 0.963 High clay content (HC) SL0 8.78a 100 121a 100 117a 100 200a 100 SL0.5 5.96ab 68 102ab 84 127ab 109 185a 93 SL1 5.67b 65 111ab 92 124ab 106 215a 107 SL2 5.35b 61 96.5b 80 136b 116 192a 96 p Trial 0.002 0.000 0.000 0.000 p Treatm 0.012 0.015 0.040 0.549 p Trial*Treatm 0.111 0.010 0.012 0.558 J. Blomquist et al. 145 The initial pH increases in 2015 also differed between the soils in the different trials, resulting in a significant inter- action between treatment and trial whereby the lower the initial pH value in the soil, the greater the increase in pH as a result of structure liming (Fig. 3). A similar interaction was found by Siman et al. (1984), although they con- centrated on the difference between trials and did not perform statistical analyses on interactions in the grouped data. The interaction observed in the present study could be due to the fact that calcium carbonate dissolves less readily in alkaline soils than in acidic soils. For soil Ca-AL content there was a less pronounced reaction to structure liming, with a significant increase at the highest liming rate (treatment SL2) compared with the control in 2015, but no remaining statistically significant differences between treatments in 2020. This fits well with results reported by Olsson et al. (2019), who detected significant increases in soil Ca-AL content with increasing application rate of ground limestone (calcium carbon- ate) from 0 to 32 t ha-1. Decreasing aggregate stability The significantly improved aggregate stability observed in 2015, approximately one year after structure liming, had decreased by the time of the repeated sampling five years later (Fig. 4). Thus, on average for the eight field trials the initial effect was transient. However, there were great differences in reaction to liming between the soils (Figs. 5 and 6), suggesting that it is inappropriate to use average values of the effects of structure liming in general advice and recommendations to farmers. The great variation in aggregate stability observed following structure liming means that it can be a blunt tool when used to reduce the risk of phosphorus losses from clay soils at na- tional level. In fact, the variation seen indicates a need for a site-specific application strategy and for soil charac- teristics to be taken into consideration. In a previous study, we observed this type of interaction between liming treatment and trial in the short term (Blomquist et al. 2022) and suggested that the interaction derives from soil properties such as clay content and soil organic matter, clay mineralogy, initial pH and management factors such as tillage before and after structure liming. Neighbouring trials with opposing results This inherent soil properties perspective was applied to the two sites where an effect on aggregate stability was actually detected (Krageholm and Lönhult). When the Krageholm HC and LC trials were analysed together, the lim- ing treatment gave a significant treatment effect compared with the unlimed control in both 2015 and 2020 (Fig. 5), i.e. structure liming had an initial and a lingering effect. However, further analysis revealed that the effect on ag- gregate stability at Krageholm derived only from the trial with higher clay content (Krageholm HC, clay content 28%), which showed an increase in aggregate stability in both sampling years, while the soil at Krageholm LC trial (clay content 19%) did not show any improvement in aggregate stability in either year. Despite the difference in clay content, these two field soils are situated close to each other (within 200 m). The contrasting behaviour of the two soils at the Krageholm site shows the importance of clay content in determining the duration of structure liming effects. Similarly, Keiblinger et al. (2016) found that the effect of quicklime (CaO) on aggregate stability was related to the clay content, and attributed this to the flocculating effect of applying calcium ions to a higher sur- face area (clay content) and also to greater availability of exchange sites (cation exchange capacity, CEC). Clay is crucial The importance of clay content is emphasised further by the results in Table 4. The group of LC soils (mean clay content 22%) showed significant treatment effects on aggregate stability (turbidity) initially, but transient effects over time. In the HC group (mean clay content 32%), the effects of structure liming were greater and significant in both sampling years. Electrical conductivity, used here as an indicator of presence of soluble calcium in the soil solution, showed significant increases in 2015 and non-significant effects in 2020 for both the LC and HC groups. Considering these differences between the groups in effect on turbidity but not EC, it can be deduced that more calcium ions were consumed in aggregate stability reactions in the soils in the HC group. The initial pH was lower and soil organic matter content was higher in the LC group than in the HC group. Based on suggestions by Blomquist et al. (2022) that structure liming may be more effective in stabilising aggregates on soils with low pH and high SOM, stronger responses could have been expected for the LC group, which was not the case (Table 4). Thus, it appears that clay content, rather than pH and organic matter, was the conclusive soil property for the differences between the LC and HC groups in this study. Agricultural and Food Science (2022) 31: 136–148 146 Initial pH also has an effect The Lönhult site produced some intriguing results regarding aggregate stability. Lönhult HC and LC trials analysed together showed a significant treatment effect compared with the unlimed control in 2015, but this effect had vanished by 2020 (Fig. 5). Both trials showed the same pattern in 2015 (Fig. 6), although the effect for Lönhult HC was only close to statistical significance (p = 0.062). The clay content was sufficiently high in both cases (28% and 47% in LC and HC soil, respectively) for a liming response to be detected, as discussed above using the clay con- tent in Krageholm HC soil (clay content 28%) as the yardstick. The transient effect can possibly be explained by the high initial pH in both Lönhult soils (pH 7.7 in Lönhult LC, pH 8.2 in Lönhult HC) compared to mean pH of 6.4 in Swedish arable soils (Kirchmann et al. 2020). On comparing aggregate stability of three clay soils after addition of quicklime, Keiblinger et al. (2016) found the lowest increase for the soil with the highest initial pH, which they attributed to presence of calcium carbonate in the initial soil. We did not measure soil calcium carbonate content in this study, but the high pH in both Lönhult soils point to very high base saturation. In such alkaline conditions, the solubility of structure lime consisting of approximately 80% calcium carbonate would be low and dissolution would be retarded. Therefore, the loss over time of the initial increase in aggregate stability at Lönhult can be attributed to initially high pH. In contrast, the initially much lower pH at Krageholm (6.2–6.6) most likely allowed continuous release of calcium ions over time, resulting in gradually increased base saturation and a significant effect on aggregate stability in the sixth year after structure liming. It is known that an increase in base saturation on acidic soils leads to better soil structure (Wiklander 1963). Cation exchange short-lived The cation exchange mechanism, proposed as the reason for increased aggregate stability at Krageholm, is rela- tively short-lived and needs continuous release of calcium ions to counteract losses through leaching down the soil profile. Pozzolanic reactions, on the other hand, are of a more long-term nature (Firoozi et al. 2017). A prereq- uisite for pozzolanic reactions to occur is that the affinity (i.e. base saturation) for calcium is satisfied (Bell 1996), as otherwise the calcium ions are consumed in cation exchange, leaving fewer available for pozzolanic reactions. Soil affinity for calcium was satisfied initially at the Lönhult site and the 15–20% of calcium hydroxide in the struc- ture lime possibly resulted in pozzolanic reactions from the start, as a significant increase in aggregate stability was observed in 2015. However, an alkaline environment with elevated pH of >10.5 (Akula and Little 2020) or ap- proximately 12 (Beetham 2015) is essential for the occurrence of modifying and stabilising pozzolanic reactions. Whether such alkaline conditions prevailed in the soil at any of the sites at application of structure lime or directly afterwards was not measured in this study. Discriminating mineralogy If pozzolanic reactions actually took place at Lönhult, the lack of persistent aggregate stability increase in 2020 is surprising and the initial aggregate stability increase in 2015 is even more confusing. It is possible that pozzolanic reactions occurred, but that the amount of non-carbonated lime in the structure lime product (15–20% calcium hydroxide) was too low to give a sufficient and long-lasting effect detectable in aggregate stability measurements using turbidity as a proxy. The initial pH at the Vadensjö and Kornheddinge sites was also >7.5, so if pozzolanic reactions occurred at Lönhult they should also have occurred at those two sites. However, no corresponding aggregate stability increase was observed at Vadensjö and Kornheddinge, possibly because the higher clay con- tent (average 37%, acting as the pozzolan) at Lönhult was the discriminating variable compared with Vadensjö and Kornheddinge (average clay content 24% and 25%, respectively). Another discriminating soil property was clay mineralogy, with a much lower SmV index at Lönhult (1.1) and higher values, indicating more expansive and swelling clay minerals, at Vadensjö (2.9) and Kornheddinge (1.9–3.1) (Table 3). Higher SmV index also indicates higher CEC, requiring more calcium ions to raise the base saturation level. The SmV index value at Lönhult may have been sufficiently low to saturate the affinity for calcium ions, leaving surplus ions for pozzolanic reactions. It has been suggested that differences in mineralogy between soils can influence the effectiveness of stabilisation processes (Kassim and Chern 2004). The mechanisms involved in aggregate stabilisation can be examined in detail using e.g. X-ray diffraction or differ- ential thermogravimetric analysis (Akula and Little 2020). Such an approach was beyond the scope of this study and therefore further work is needed to confirm the mechanisms involved in the observed increase and decrease in aggregate stability following structure liming of field soils. J. Blomquist et al. 147 Conclusions This study compared the effects of applying structure lime (approximately 80–85% ground limestone and 15–20% slaked lime) to clay soils on aggregate stability, using laboratory-measured turbidity as a proxy, one and six years after liming. As a mean for the three structure lime doses tested (4, 8 and 16 t ha-1), there were significant increas- es in aggregate stability compared with the unlimed control after one year and also after six years, although the effect declined over time. However, in both years there was a significant interaction between trial and treatment, indicating differences in response to structure liming on different soils. Thus, a site-specific application strategy is required, if the desired objective is to increase aggregate stability in order to mitigate phosphorus losses. Soil clay content and initial pH seemed to be decisive for a more persistent effect of liming on aggregate stability. Based on the results, we propose that soils with a minimum clay content of approximately 25% and pH <7 should be preferred for structure liming with the mixed product used in this study. The results also indicated that cation ex- change was the main mechanism involved in aggregate stabilisation with the structure lime product used, since stabilisation by pozzolanic reactions and carbonation would require products with higher solubility and the abil- ity to create very high pH (above 10–12). Acknowledgements The County Administration Board (CAB) in Skåne provided financial support for this project (Lst nr 583-23006- 2020). We gratefully acknowledge Jens Ratcovich, then working at CAB Skåne, who farsightedly initiated this fol- low-up evaluation. We also thank the farmers and farm managers at the field sites for letting us access their land for sampling. Finally, we want to express our gratitude to Fredrik Hansson and colleagues at HS Skåne Odlarservice and Lars Wadmark, then working at Nordkalk Corp., for initiating the structure liming field trials. References Akula, P. & Little, D.N. 2020. Analytical tests to evaluate pozzolanic reaction in lime stabilized soils. MethodsX 7: 100928. https:// doi.org/10.1016/j.mex.2020.100928 Beetham, P. 2015. Enhancing the understanding of lime stabilisation processes. Thesis, Loughborough University. 232 p. Bell, F.G. 1996. Lime stabilization of clay minerals and soils. Engineering Geology 42: 223–237. https://doi.org/10.1016/0013- 7952(96)00028-2 Berglund, G. 1977. Mikroaggregatanalysen som testmetod vid strukturkalkning. Uppsala: Lantbrukshögskolan. Institutionen för markvetenskap, Avdelningen för lantbrukets hydroteknik. https://pub.epsilon.slu.se/5616/. Stenciltryck. Blomquist, J., Englund, J.-E. & Berglund, K. 2022. Soil characteristics and tillage can predict the effect of ‘structure lime’on soil ag- gregate stability. Soil Research 60: 373–384. https://doi.org/10.1071/SR21022 Blomquist, J., Simonsson, M., Etana, A. & Berglund, K. 2018. Structure liming enhances aggregate stability and gives varying crop responses on clayey soils. Acta Agriculturae Scandinavica, Section B-Soil & Plant Science 68: 311–322. https://doi.org/10.1080/ 09064710.2017.1400096 Carlgren, K. & Mattsson, L. 2001. Swedish soil fertility experiments. Acta Agriculturae Scandinavica, Section B-Plant Soil Science 51: 49–76. https://doi.org/10.1080/090647101753483787 Choquette, M., Bérubé, M.-A. & Locat, J. 1987. Mineralogical and microtextural changes associated with lime stabilization of ma- rine clays from eastern Canada. Applied Clay Science 2: 215–232. https://doi.org/10.1016/0169-1317(87)90032-9 Egnér, H., Riehm, H. & Domingo, W. 1960. Untersuchungen über die chemische Bodenanalyse als Grundlage für die Beurteilung des Nährstoffzustandes der Böden. II. Chemische Extraktionsmethoden zur Phosphor-und Kaliumbestimmung. Kungliga Lant- brukshögskolans Annaler 26: 199–215. Etana, A., Rydberg, T. & Arvidsson, J. 2009. Readily dispersible clay and particle transport in five Swedish soils under long-term shallow tillage and mouldboard ploughing. Soil and Tillage Research 106: 79–84. https://doi.org/10.1016/j.still.2009.09.016 EC 2021. The EU Water Framework Directive - integrated river basin management for Europe. https://ec.europa.eu/environment/ water/water-framework/index_en.html [Online]. Accessed 21 September 2021. Firoozi, A.A., Olgun, C.G., Firoozi, A.A. & Baghini, M.S. 2017. Fundamentals of soil stabilization. International Journal of Geo-En- gineering 8: 1–16. https://doi.org/10.1186/s40703-017-0064-9 Gunnarsson, A., Blomquist, J., Persson, L., Olsson, Å., Hamnér, K. & Berglund, K. 2022. Liming alkaline clay soils: effects on soil structure, nutrients, barley growth and yield. Acta Agriculturae Scandinavica, Section B — Soil & Plant Science 72: 803–817. https://doi.org/10.1080/09064710.2022.2089590 HAV 2021. Havs- och vattenmyndigheten. LOVA. https://www.havochvatten.se/anslag-bidrag-och-utlysningar/havs--och-vatten- miljoanslaget/lova.html. Kassim, K.A. & Chern, K.K. 2004. Lime stabilized Malaysian cohesive soils. Malaysian Journal of Civil Engineering 16: 13–23. Agricultural and Food Science (2022) 31: 136–148 148 Keiblinger, K.M., Bauer, L.M., Deltedesco, E., Holawe, F., Unterfrauner, H., Zehetner, F. & Peticzka, R. 2016. Quicklime applica- tion instantly increases soil aggregate stability. International Agrophysics 30: 123–128. https://doi.org/10.1515/intag-2015-0068 Kirchmann, H., Börjesson, G., Bolinder, M.A., Kätterer, T. & Djodjic, F. 2020. Soil properties currently limiting crop yields in Swed- ish agriculture-An analysis of 90 yield survey districts and 10 long-term field experiments. European Journal of Agronomy 120: 126–132.https://doi.org/10.1016/j.eja.2020.126132 Kirchmann, H., Snäil, S. & Eriksson, J. 1996. Properties and Classification of Soils of the Swedish Long-Term Fertility Experi- ments: III. Sites at Västraby and S. Ugglarp. Acta Agriculturae Scandinavica B-Plant Soil Sciences 46: 86–97. https://doi. org/10.1080/09064719609413120 Ledin, S. 1981. Physical and Micromorphological Studies of the Effects of Lime on a Clay Soil. PhD thesis. Sveriges lantbruksuni- versitet (Swedish University of Agricultural Sciences). Norberg, L., Aronsson, H., Blomberg, M. & Ulén, B. 2021. Strukturkalkning för minskat fosforläckage - En fältstudie på mellanlera i Halland. Uppsala: Sveriges Lantbruksuniversitet. Ekohydrologi 170. (in Swedish). Olsson, Å., Persson, L. & Olsson, S. 2019. Influence of soil characteristics on yield response to lime in sugar beet. Geoderma 337: 1208–1217. https://doi.org/10.1016/j.geoderma.2018.11.020 Puustinen, M., Koskiaho, J. & Peltonen, K. 2005. Influence of cultivation methods on suspended solids and phosphorus concen- trations in surface runoff on clayey sloped fields in boreal climate. Agriculture, Ecosystems & Environment 105: 565–579. https:// doi.org/10.1016/j.agee.2004.08.005 Siman, G., Berglund, K. & Eriksson, L. 1984. Effect of large lime quantities on soil structure, nutrient balance and crop yields. A. Effect of lime on soil properties, crop yields and nutrient concentrations in yield products. Rapport 148, Avdelningen för Växt- näringslära, Institutionen for Markvetenskap. Uppsala: Sveriges Lantbruksuniversitet. (in Swedish). Ulén, B., Alex, G., Kreuger, J., Svanbäck, A. & Etana, A. 2012. Particulate-facilitated leaching of glyphosate and phosphorus from a marine clay soil via tile drains. Acta Agriculturae Scandinavica, Section B-Soil & Plant Science 62: 241–251. https://doi.org/10. 1080/09064710.2012.697572 Ulén, B. & Etana, A. 2014. Phosphorus leaching from clay soils can be counteracted by structure liming. Acta Agriculturae Scan- dinavica, Section B-Soil & Plant Science 64: 425–433. https://doi.org/10.1080/09064710.2014.920043 Wiklander, L. 1963. Kalkens betydelse för markens struktur. Särtryck ur Svenskt jordbruk och skogsbruk 1913–1962. (in Swedish). Site characteristics determine the duration of structure limingeffects on clay soil Introduction Materials and methods Study area and sites Experimental design and liming product Clay mineralogy determination Soil characteristics Aggregate stability, sampling and measurements Statistical analyses Results pH and calcium changes Aggregate stability Discussion Declining pH effect Decreasing aggregate stability Neighbouring trials with opposing results Clay is crucial Initial pH also has an effect Cation exchange short-lived Discriminating mineralogy Conclusions Acknowledgements References