Long-term experimental plots

A long-term highbush blueberry experiment, established in 2008 on flat topography at 7.6 m above sea level at the Agriculture and Agri-Food Canada, Agassiz Research and Development Centre (Agassiz RDC) (49°14′N, 121°45′W), was selected for this study. The local climate is characterized as moderate oceanic with relatively cool and dry summers, and warm and rainy winters (British Columbia Agriculture and Food Climate Action Initiative 2015). Annual rainfall ranges from 1483 to 1689 mm with the peak of about 280 – 350 mm received in November. Average daily temperatures range between 3.4 °C in January and 18.8 °C in August (30 years average, 1981 – 2010, Agassiz Climate Daily Almanac-Agassiz CDA station). The soil is a loam textured soil of the Monroe series, classified as Dystric Brunisols (Canadian System of Soil Classification 1998) or Typic Dystroxerepts in the U.S. Soil Taxonomy (Soil Survey Staff 2010). The soil is moderately to well drained, shallow with a weakly developed profile. The 13-year experiment consists of annual ammonium sulfate (21-0-0) application at three rates (100%, 150%, and 200% of the annual recommendation for blueberry based on the BC blueberry production guide) applied either as broadcast or fertigation, for a total of seven treatments including a control (0%), arranged in a randomized complete block design with four replicates (Messiga et al. 2018). General physicochemical properties, shown in Table 1, include: pH, 4.1–5.8; electrical conductivity, 118–311 µs cm^–1; total organic carbon, 1.97%–3.74%; total N, 0.15–0.22; P_M3, 163–200 mg kg^–1; Mehlich-3 extractable Al (Al_M3), 2126–3218 mg kg^–1; and Fe (Fe_M3), 166–210 mg kg^–1.

Table 1.

General properties of soils with initial pH values of 4.1, 4.8, 5.5, and 5.8 collected at 0–30 cm soil depth in a long-term blueberry experiment receiving annual applications of nitrogen fertilizers (ammonium sulfate) at Agassiz RDC in British Columbia, Canada.

Soil sampling

Soils were selected according to their variable pH to fit into four descriptive acidity classes (Soil Survey Division Staff 1993): (i) extremely acidic (4.1), (ii) very strongly acidic (4.8), (iii) strongly acidic (5.5), and (iv) moderately acidic (5.8). We used soil samples collected from a long-term highbush blueberry experiment with a history of ammonium-based fertilizer applications that produced a range of soil pH from extremely low to near-neutral (pH = 3.8–6.5). Three treatments of the 13-year blueberry experiment and an adjacent field with identical agri-practices were therefore selected to fit these classes: (i) broadcast N (200% for pH 4.1); (ii) fertigation N (100% for pH 4.8); (iii) control (0% N for pH 5.8); and (iv) an adjacent field under rye grass (pH 5.5). For soil sampling, after discarding the layer of sawdust mulch at the soil surface approximately 3.5 kg of soil was collected down to 30 cm depth using a shovel, and then was placed in a labeled clean 10 L bucket and mixed thoroughly. The fresh soil was sieved (<2 mm) and kept at 4 °C in the laboratory. Two 80 g soil samples taken from each bucket were air dried prior to physico-chemical analyses, including pH, texture, organic matter content, cation exchange capacity (CEC), total N, total S, total C, organic P, total P, P_M3, Al_M3, Fe_M3, and other Mehlich-3 extractable cations including Ca (Ca_M3), and Mg (Mg_M3). The remaining soil was kept frozen at –20 °C until the beginning of the incubation experiment.

Incubation experiment

Eight combination treatments (four initial soils (pH 4.1, 4.8, 5.5, and 5.8) × two lime applications (with and without lime)) were used for this study. The lime treatment was selected to reach a target pH of 6.0 for soils initially at pH 4.1, 4.8, and 5.5, and a target pH of 6.5 for the soil initially at pH 5.8. The lime requirement (LR) was calculated using a modified procedure of Laboski and Peters (2012), which used the CaCO₃ equivalent ratio to moles of the H⁺ concentrations to be neutralized from initial pH to target pH. The LR to raise the initial pH to 6.0 was 9.0 g for pH 4.1, 4.0 g for pH 4.8, 0.7 g for pH 5.5, and 0.1 g for pH 5.8. The eight combination treatments, each replicated four times (32 experimental units in total), were (i) pH 4.1 without lime (pH4.1L0), (ii) pH 4.1 with lime (pH4.1L1), (iii) pH 4.8 without lime (pH4.8L0), (iv) pH 4.8 with lime (pH4.8L1), (v) pH 5.5 without lime (pH5.5L0), (vi) pH 5.5 with lime (pH5.5L1), (vii) pH 5.8 without lime (pH5.8L0), and (viii) pH 5.8 with lime (pH 5.8L1). For treatments without lime, 1200 g of soil was mixed in a clean bucket and 250 g was weighed in four Mason jars per treatment. For the treatments with lime, 1200 g of soil was placed in a clean bucket, mixed with CaCO₃ (fine powder, 99.95%–100.5% dry basis, pH 8, from Sigma–Aldrich), and 250 g was weighed in four Mason jars per treatment. The 32 Mason jars were closed with parafilm into which 10 holes were punctured with a needle to allow for gas exchange. The Mason jars were arranged in a completely randomized design in a growth chamber (Conviron Adaptis A1000, Winnipeg, Canada) at 22.5 °C for 93 days. The weight of each Mason jar at the start of the incubation was recorded to monitor the soil moisture content. Moisture content was maintained at 60% water-filled pore space to maximize microbial activity (Franzluebber et al. 1999). Each Mason jar was weighed every 3 days to check moisture content, and distilled water was added by a bottle-top dispenser if the moisture content decreased by more than 10%. After 30 and 44 days of incubation, 10 g subsamples were taken from each jar to monitor changes in soil pH. After 30 days of incubation, it was observed that the soils initially at pH 5.8 did not reach the target pH 6.0. An extra 20 mg of CaCO₃ was mixed with the soil in the jars for that treatment, which increased the soil pH to ∼6.5.

Soil analyses

After 93 days of incubation, the soil in each Mason jar was removed and air dried for chemical analysis. Air-dried soils were analyzed for pH in distilled water (1:2 soil:solution) on a MultiLab IDS (Model 4010-3; Hendershot et al. 1993). The P_w was determined by shaking 2.0 g of air-dried soil mixed in 50 mL centrifuge tubes with 20 mL distilled water for 1 h, centrifuging the mixture at 3500 r min^–1 for 10 min, and filtering through Whatman Grade 42 filter paper (Self-Davis et al. 2000). Phosphorus in the extract was determined by the molybdenum blue colorimetric method (Murphy and Riley 1962). The P_M3, Al_M3, and Fe_M3 were determined by shaking 2.5 g of air-dried soil with a 25 mL of Mehlich-3 solution (pH 2.3) for 5 min (Mehlich 1984). The P_Ox, Al_Ox, and Fe_Ox were determined according to Ross and Wang (1993). The concentrations of P_M3, Al_M3, Fe_M3, P_Ox, Al_Ox, and Fe_Ox were assessed with an inductively coupled plasma optical emission spectrometer (ICP-OES). Total organic P was determined by the ignition method (Saunders and Williams 1955). Total P was determined by digestion using a fluxer (M4 Fluxer, Claisse) as described by Kowalenko and Babuin (2014). Phosphorus concentrations in these digests were analyzed by an ICP-OES (ICAP 7000 series, Thermo Scientific).

Calculations of degree of P saturation and P saturation index

The DPS was calculated by using the molar ratio of P_Ox, Al_Ox, and Fe_Ox (Breeuwsma and Silva 1992):

(1)

The P saturation index was calculated using two variations, PSI₁ and PSI₂, using P_M3, Al_M3, and Fe_M3, as follows (Khiari et al. 2000):

(2)

(3)

Phosphate sorption experiments

Phosphate sorption experiments were also conducted on air-dried soils (Messiga et al. 2021). In brief, solutions with a range of P concentrations were formulated in a matrix of distilled water and 100 µL of toluene (to inhibit microbial activity) for batch equilibration with the soil. Initial equilibrating solution P concentrations were 0, 10, 20, 50, 100, and 150 mg P L^–1 as KH₂PO₄. Sorption batches consisted of 2.0 g air-dried soil mixed in 50 mL centrifuge tubes with 30 mL of the equilibrating solution. The tubes were gently shaken end-over-end for 24 h at room temperature, centrifuged at 3500 r min^–1 for 10 min, and then filtered through Whatman Grade 42 filter paper. An aliquot of the supernatant solution was used to determine P in solution by ICP-OES.

Soil P fractionation

Finally, soil P fractionation was performed on the remaining air-dried soil by sequentially extracting P according to a modified method of Chang and Jackson (1958) as described by Zhang and Kovar (2009). Briefly, 0.5 g of finely ground soil (0.2 mm) was sequentially extracted by 1 mol L^–1 NH₄Cl (NH₄Cl-P), 0.5 mol L^–1 NH₄F (NH₄F-P), 0.1 mol L^–1 NaOH (NaOH-P), a mixture of 0.3 mol L^–1 sodium citrate dihydrate, Na₃C₆H₅O₇.₂H₂0 + 1 mol L^–1 NaHCO₃ for reductant-soluble P (CBD-P), and 0.25 mol L^–1 H₂SO₄ (H₂SO₄-P) at room temperature. For the purpose of precision, the inorganic P fractions are described by their extraction method (Barrow et al. 2020; Guppy 2021). Phosphorus concentrations in the extracts were analyzed by the molybdenum blue colorimetric method (Murphy and Riley 1962).

Sorption data fitting

Sorption data for each soil pH and lime treatments were fitted using Langmuir and Freundlich isotherms. Sorption isotherms were parameterized using SAS PROC NLIN version 9 (SAS Institute Inc. 2010). The Langmuir isotherms resulted in the best goodness of fit for all soil pH and lime treatments. The Langmuir isotherm is described as follows:

(4)

(5)

Statistical analyses

To statistically compare the S_max (not k_a because of autocorrelations) parameter between not limed and limed treatments, we calculated the following F ratio:

(6)

Data of soil P analysis, DPS, PSI, and P fractionation were analyzed statistically using one-way Analysis of variance (ANOVA) with random effects as replicates and the eight treatments as fixed effects using SAS PROC MIXED (SAS Institute 2010). Data were first checked for assumptions of ANOVA using Shapiro–Wilk test and Q–Q plots for normality and for equal variance using residual plots and Bartlett's tests. Differences between least square means for treatment pairs (without lime vs. with lime at each initial pH level) were tested at significance level of P = 0.05. Nonlinear regression techniques were applied to pH, P_M3, P_w, DPS, PSI₁, and PSI₂ using NLIN procedure of SAS. The fit of the regression curves was evaluated using R² to determine how well the curve explains measured data variation. Principal component analysis (PCA) was carried out using SAS PROC PRINCOMP (SAS Institute 2010) to establish the relationships between agri-environmental indicators and chemical properties of soil and P fractions.

Liming effects on soil pH, concentrations of Mehlich-3 P, and water-extractable P, and agri-environmental indicators

As projected by the LR calculations, soil pH was increased from 4.1 to 6.0, 4.8 to 6.0, 5.5 to 6.0, and 5.8 to 6.5, respectively, after 93 days of incubation with additions of CaCO₃ (Fig. 1a). The initial soil pH value did not change after 93 days of incubation. (Fig. 1a). The concentrations of P_M3 ranged from 146 to 200 mg kg^–1 and were not significantly changed by liming (Fig. 1b). The concentrations of P_w ranged from 1.0 to 3.65 mg kg^–1 and were significantly affected by additions of lime only for soils with initial pH of 5.8 (Fig. 1c). The DPS, PSI₁, and PSI₂ were not affected by additions of lime for any of these soils, regardless of initial soil pH (Figs. 1d–1f). On average, the DPS values were 27%, 26%, 25%, and 27% for soils at initial soil pH of 4.1, 4.8, 5.5, and 5.8, respectively (Fig. 1d). On average, the PSI₁ values were 6.0%, 5.2%, 5.6%, and 6.1%, respectively (Fig. 1e). On average, the PSI₂ values were 5.8%, 5.0%, 5.4%, and 5.8%, respectively (Fig. 1f).

Fig. 1.

Effects of liming using calcium carbonate on changes in (a) soil pH, (b) Mehlich-3 P, (c) water-extractable P, (d) DPS (%), (e) PSI₁ (%), and (f) PSI₂ (%) after 93 days of incubation. Bars with different letters are statistically different at P < 0.05. [Colour online.]

Liming effects on the relationships between concentrations of water-extractable P and agri-environmental indicators

There were two patterns of the relationships between concentrations of P_w and agri-environmental indicators at the end of incubation depending on the initial soil pH. At initial soil pH ≤ 5.5, the concentrations of P_w were related to agri-environmental indicators, P_M3, DPS, PSI₁, PSI₂, PSC, and Ca_M3, by quadratic regressions (Figs. 2a–2f). The slopes of the quadratic regressions were positive for P_M3 (Fig. 2a), DPS (Fig. 2b), and PSC (Fig. 2e), but negative for PSI₁ (Fig. 2c), PSI₂ (Fig. 2d), and Ca_M3 (Fig. 2f). In addition, the R² of the regressions were 0.25 for P_M3, 0.49 for DPS, 0.22 for PSI₁, 0.23 for PSI₂, 0.23 for PSC, and 0.58 for Ca_M3. In contrast, at initial pH = 5.8, there was no trend in the distribution pattern of the concentrations of P_w with P_M3, DPS, PSI₁, PSI₂, and PSC except the decreasing trends following additions of lime and the increased soil pH (Figs. 2a–2e). However, at initial pH = 5.8, the concentrations of P_w decreased with increasing Ca_M3 following additions of lime (Fig. 2f).

Fig. 2.

Effects of liming using calcium carbonate on the relationships between water-extractable P and (a) Mehlich-3 P, (b) degree of P saturation (DPS, %), (c) P saturation index (PSI₁, (P/Al)_Mehlich-3 (%)), (d) P saturation index (PSI₂, (P/Al + Fe)_Mehlich-3 (%)), (e) P sorption capacity (PSC, mmol kg^–1), and (f) Mehlich-3 calcium (Ca, mg kg^–1) after 93 days of incubation. Red and green highlighted points are for initial and target soil pH = 5.5, respectively, which were not included in the regression. [Colour online.]

A further examination showed that the relationships among agri-environmental indicators were described by linear regressions (Figs. 3a–3d). The linear regression between PSI₁ and PSI₂ was characterized by a positive slope = 0.96 and R² = 0.99 and was not affected by liming (Fig. 3a). Similarly, the relationships between DPS and PSI₁, PSI₂, and P_M3 were not affected by liming and were described by linear regressions with positive slopes = 0.39 and R² = 0.52 for DPS vs. PSI₁ (Fig. 3b), positive slopes = 0.38 and R² = 0.53 for DPS vs. PSI₂ (Fig. 3c), and positive slopes = 20.5 and R² = 0.45 for DPS vs. P_M3 (Fig. 3d).

Fig. 3.

Effects of liming using calcium carbonate on the relationships between (a) P saturation index (PSI₁, (P/Al)_Mehlich-3, %) and PSI₂ ((P/Al + Fe)_Mehlich-3, %), and degree of P saturation (DPS, %) and (b) PSI₁, (c) PSI₂, and (d) Mehlich-3 P after 93 days of incubation. [Colour online.]

Liming effects on concentrations of aluminum, iron, and calcium

There was a decreasing trend of the concentrations of Al_M3 and Fe_M3 in soils with increasing soil pH. The addition of CaCO₃ further decreased the concentrations of Al_M3 and Fe_M3 by 24%, 11%, 2.3%, and 17% for initial soil pH = 4.1, 4.8, 5.5, and 5.8, respectively. The relationships between the concentrations of Al_M3 and Fe_M3 vs. soil pH were not affected by the addition of CaCO₃ and were described by linear regressions with negative slope = –329.61 and R² = 0.62 for Al_M3 (Fig. 4a), and negative slope = –8.06 and R² = 0.33 for Fe_M3 (Fig. 4b).

Fig. 4.

Effects of liming using calcium carbonate on the relationships between soil pH and (a) Mehlich-3 aluminum, (b) Mehlich-3 iron (Fe), (c) Mehlich-3 calcium, (d) ammonium oxalate aluminum, and (e) ammonium oxalate iron after 93 days of incubation. [Colour online.]

In contrast, there was an increasing trend of the concentrations of Ca_M3 in soils with increasing soil pH. The addition of CaCO₃ further increased the concentrations of Ca_M3 by 5.5 times, 2.4 times, 1.4 times, and 1.8 times for soils with initial pH 4.1, 4.8, 5.5, and 5.8, respectively. The relationship between the concentrations of Ca_M3 and soil pH was not affected by the addition of CaCO₃ and was described by a power regression with a positive slope = 0.48 and R² = 0.44 (Fig. 4c).

There was no specific trend in the concentrations of Al_Ox and Fe_Ox with initial soil pH as observed with Al_M3 and Fe_M3. On average, the concentrations of Al_Ox were 321, 332, 237, and 320 mmol kg^–1 for initial soil pH values of 4.1, 4.8, 5.5, and 5.8, respectively, indicating no effect of soil pH or CaCO₃ additions (Fig. 4d). The concentrations of Fe_Ox were 87, 81, 59, and 83 mmol kg^–1 for initial soil pH values of 4.1, 4.8, 5.5, and 5.8, respectively, indicating no effect of soil pH or CaCO₃ additions (Fig. 4e).

Liming effects on phosphorus sorption characteristics

The experimental data of the sorption study for all soils at the end of the 93 days of incubation were best fitted by the Langmuir equation, with R² values 0.97 and 0.99 (Fig. 5). The S_max and k_a values were affected by liming only for soils with initial pH values of 4.1 and 5.8 (Figs. 5a and 5d). For soils with initial pH 4.1, the S_max value was 1382 mg P kg^–1 without lime but significantly decreased to 1243 mg P kg^–1 after liming (Table 2; Fig. 5d). For soils with initial pH 5.8, the S_max value was 1093 mg P kg^–1 without lime but significantly increased to 1114 mg P kg^–1 after liming (Table 2; Fig. 5d).

Fig. 5.

Effects of liming using calcium carbonate on phosphorus sorption characteristics of soils after 93 days of incubation. (a) P sorption maximum; (b) binding energy of P; (c) significant probability; (d) root means square deviation. Means followed by similar letters within the row for the same initial soil pH are not statistically different using the F test at the probability level of 0.05. [Colour online.]

Table 2.

Comparison of the parameters of the Langmuir equation for the four soils at their initial pH values following additions of lime after 93 days of incubation.

For the other soils, the S_max value was 1244 mg P kg^–1 without lime and 1267 mg P kg^–1 with lime with an average S_max value of 1254 mg P kg^–1 for initial soil pH 4.8; 745 mg P kg^–1 without lime and 795 mg P kg^–1 with lime with an average S_max value of 777 mg P kg^–1 for initial soil pH 5.5; and 1093 mg P kg^–1 without lime and 1114 mg P kg^–1 with lime for initial soil pH 5.8 (Figs. 5b and 5c).

Liming effects on phosphorus fractions, organic P, and total P

At the end of the 93 days of incubation, the concentrations of NH₄Cl-P significantly decreased with addition of CaCO₃ for soils with initial pH 4.1 only, by 50.4% (Fig. 6a). The decreasing trends of NH₄Cl-P in soils with the other initial pH values were not significant (Fig. 6a). The concentrations of NH₄F-P significantly decreased with addition of CaCO₃ for soils with initial soil pH 4.1 only, from 59 to 29 mg kg^–1 (Fig. 6b). The concentrations of NaOH-P significantly decreased with addition of CaCO₃, from 324 to 247 mg kg^–1, for initial soil pH = 4.8 (Fig. 6c). The decreasing NaOH-P trends observed with the other initial pH with addition of CaCO₃ were not significant. The concentrations of the reductant-soluble P fraction (CBD-P) significantly increased with addition of CaCO₃, from 26 to 218 mg kg^–1, at initial soil pH = 4.1, and significantly decreased with addition of CaCO₃, from 364 to 21 mg kg^–1 and from 103 to 23 mg kg^–1, at initial soil pH 5.5 and 5.8, respectively, with no significant change for soils initially at pH 4.8 (Fig. 6d). The concentrations of the H₂SO₄-P fraction were not significantly affected by addition of CaCO₃ (Fig. 6e). The corresponding organic P fractions at different sequential extracts were not influenced by liming.

Fig. 6.

Effects of liming with calcium carbonate and 93 days of incubation on changes in molybdate-reactive P pools determined by sequential fractionation. (a) Ammonium chlorideextractable P (NH₄Cl-P), (b) ammonium fluoride-extractable P (NH₄F-P), (c) sodium hydroxide-extractable phosphorus (NaOH-P), (d) reductant-soluble P (CBD-P), and (e) acidextractable P (H₂SO₄-P). Bars with different letters are statistically different at P < 0.05. [Colour online.]

At the end of the 93 days of incubation, soil organic P and total P were not affected by additions of lime. On average over the two lime additions, organic P concentrations were 339–444, 427 mg kg^–1 (Fig. 7a), while total P concentrations were 1181–1492 mg kg^–1 (Fig. 7b).

Fig. 7.

Effects of liming using calcium carbonate and 93 days of incubation on (a) soil organic P and (b) total soil P. Bars with different letters are statistically different at P < 0.05. [Colour online.]

Relationships between P_w, Mehlich-3 P, agri-environmental indicators, inorganic P fractions, and chemical soil properties

The PCA showed that P_w was highly correlated to PSI₁, PSI₂, and Ca_M3, whereas P_M3, Al_M3, Fe_M3 inorganic P fractions and chemical soil properties were highly correlated following the addition of CaCO₃ (Fig. 8). The eight combinations of initial soil pH × addition of CaCO₃ were separated into four groups for their effects on the studied properties, including (i) pH4.1L0 and pH4.8L0, (ii) pH4.8L1, (iii) pH5.5L0 and pH5.5L1, and (iv) pH4.1L1, pH5.8L0, and pH5.8L1. Groups I and II positively affected NH₄Cl-P, Al_M3, Fe_M3, and k_a, but negatively affected P_w, PSI₁, PSI₂, and Ca_M3; Group III positively affected H₂SO₄-P and NaHCO₃-P, but negatively affected DPS, Al-P, Fe-P, PSC, S_max, and P_M3; Group IV positively affected P_w, PSI₁ and PSI₂, Ca_M3, and pH. Pearson's correlation among P_w, P_M3, agri-environmental indicators (PSI₁, PSI₂, and DPS), inorganic P fractions (NH₄Cl-P, NH₄F-P, NaOH-P, NaHCO₃-P, and H₂SO₄-P), and chemical soil properties (Al_M3, Fe_M3, Ca_M3, PSC, k_a, and S_max) were in line with the results of PCA (Table 3). The concentrations of P_w were positively correlated with Ca_M3, PSI₁, and PSI₂. Increasing soil pH was associated with decrease in Al_M3, Fe_M3, and NH₄Cl-P more strongly at initial soil pH values of 4.1 and 4.8 (Table 3). The concentrations of H₂SO₄-P were negatively correlated with k_a and S_max. Remarkably, the P_w were positively associated with PSI₁ and PSI₂ than DPS (Table 3).

Fig. 8.

Correlations between water-extractable P and Mehlich-3 P agri-environmental indicators and chemical soil properties as tested by principal component analysis (PCA): pH treatment (combinations of pH and lime), pH, TC (total C, %), TP (total phosphorus, mg kg^–1), TN (total nitrogen, %), P_w (water-extractable phosphorus, mg kg^–1), P_M3 (Mehlich-3 extractable phosphorus, mg kg^–1), DPS (degree of phosphorus saturation, %), PSC (phosphorus sorption capacity, mmol kg^–1), Al_M3 (Mehlich-3 extractable aluminum, mg kg^–1), Fe_M3 (iron-extractable aluminum, mg kg^–1), PSI₁ (phosphorus saturation index, ((P/Al)_Mehlich-3) %), PSI₂ (phosphorus saturation index, ((P/(Al + Fe))_Mehlich-3) %), Ca_M3 (Mehlich-3 extractable calcium, mg kg^–1), NH₄Cl-P (ammonium chloride-extractable phosphorus, mg kg^–1), NH₄F-P (ammonium fluoride-extractable phosphorus, mg kg^–1), NaOH-P (sodium hydroxide-extractable phosphorus, mg kg^–1), NaHCO₃-P (sodium bicarbonate-extractable phosphorus, mg kg^–1), H₂SO₄-P (acid-extractable phosphorus, mg kg^–1); S_max (phosphorus sorption maximum, mg P kg^–1); K_a (binding energy of phosphorus, L mg P^–1) after 93 days of incubation.

Table 3.

Pearson's correlation matrix.

Low soil pH and relationship between concentrations of P_M3, P_w, and agri-environmental indicators

The concentrations of P_M3 across all soils (Fig. 1b) fall into the “very high” P class (above 150 mg kg^–1) and represented agricultural soils typically found in the BC Fraser Valley (Kowalenko et al. 2007; Reid and Schneider 2019). According to the literature and local recommendations, these high P_M3 concentrations are associated with increased risk of environmental impact due to P losses with runoff (Kowalenko et al. 2007; Reid and Schneider 2019). A value of DPS > 25% (Fig. 1d) is considered the critical limit above which P runoff is enhanced in acidic soils (van der Zee et al. 1987), and all the studied soils had DPS values above this critical limit. However, only soils with initial pH 5.8 had concentrations of P_w of 3.8 mg kg^–1 (Fig. 1c), which is close to critical concentration values set at 4.0 mg kg^–1 across different agro-ecosystems in BC (van Bochove et al. 2012) and 3.7 mg kg^–1 in silage corn and blueberry plantings (Messiga et al. 2021) in the Fraser Valley. The soils with initial pH 5.8 have not received any N fertilizers for the past 10 years (Messiga et al. 2018). The other soils with initial pH values of 4.1, 4.8, and 5.5 had concentrations of P_w below the critical level (Fig. 1c), which contrasted with their P_M3 and DPS values, indicating significant pH effects on the solubility of P and assessment of the risk of P losses with runoff. The soils with initial soil pH values of 4.1, 4.8, and 5.5 received ammonium sulfate at different rates, which resulted in the decreased soil pH and reduced changes in P_w concentration patterns compared with soils with initial pH 5.8 (Messiga et al. 2018). The significant decline with liming of P_w at pH 5.8 could be explained by the dissociation of P ions leading to speciation shifts. At pH 6.5, phosphate will occur as HPO₄^2– more than as H₂PO₄^–, which will increase phosphate sorption (Chang and Overby 1986; Han 2020). The mismatch between concentrations of P_M3, DPS, and concentrations of P_w as influenced by soil pH indicates the need to improve our understanding of the functioning of these indicators to assess the risk of P losses with runoff across the range of cropping systems and management practices found in the Fraser Valley.

The low PSI₁ and PSI₂ (Figs. 1e and 1f), which also contrasted with high concentrations of P_M3 (150–200 mg kg^–1) and DPS > 25% (Figs. 1b and 1d), were in line with low P_w concentrations associated with soils at initial pH ≤ 5.5 (Fig. 1c). The discrepancy between the indicators of risk provided by DPS and PSI₁ and PSI₂ as they relate to P_w and P_M3 of soils with initial pH ≤ 5.5 could be explained by a combination of factors inherent to the extraction solutions and the chemistry of P, Al, Fe, and Ca, and to some extent, phytates driven by soil pH. We found that the concentrations of Al_M3 in soils with initial pH values of 4.1 and 4.8 were higher than soils with initial pH values of 5.5 and 5.8 (Table 2; Figs. 4a and 4b). On average, the concentration of Al_M3 was 3056 mg kg^–1 in soils with initial pH values of 4.1 and 4.8 compared with 2594 mg kg^–1 in soils with initial pH values of 5.5 and 5.8 (Table 2; Figs. 4a and 4b). The concentrations of Al_M3 with initial soil pH ≤ 5.5 followed the same trend as P_M3 (Fig. 1b). At these low initial pH levels, Al³⁺ cations present in the primary minerals are easily expelled in the soil solution where they form secondary minerals of Al oxides and oxyhydroxides. These secondary Al oxides and oxyhydroxides further react and precipitate with phosphate ions (Penn and Camberato 2019). In addition, the low initial pH prevailing in the soil solution increases the amount of pH-dependent charges at the edges of Al oxides and oxyhydroxides and other broken bonds of silicate minerals onto which the fixation of phosphate ions is enhanced (Penn and Camberato 2019). It is therefore possible that at low initial pH, the combination of high concentrations of Al oxyhydroxides provides large surface area for adsorption sites and large pH-dependent charges enhance the retention of phosphate ions in the soil solution, thus reducing the concentration of P_w. Furthermore, the opposite trends between the concentrations of P_w and P_M3 and Al_M3 indicate that phosphate ions retained at the surfaces of Al oxides and oxyhydroxides are not soluble in water and therefore are not accounted for by the colorimetric blue methods used in their assessment (Murphy and Riley 1962; Self-Davis et al. 2000; Messiga et al. 2021).

Our results also indicate that Mehlich-3 solution extracts more Al and P at low pH compared with high pH (Figs. 4a and 4b). This is explained by high Al on the cation exchange complex at low pH and the increased Ca onto negatively charged reacting surfaces at high pH that facilitate phosphate adsorption (Barrow 2017). At low pH, the acidity of Mehlich-3 solution promotes the dissolution of Al oxides and oxyhydroxides such as gibbsite, thus increasing the concentration of Al_M3 (Penn et al. 2018). The affinity of F^– with Al further increases the concentration of Al_M3 by complexing additional Al oxides and oxyhydroxides resulting into the formation of AlF3 that is detected upon analysis of the Mehlich-3 extract by ICP (Mehlich 1984). These increased concentrations of Al_M3 associated with low soil pH and Mehlich-3 solution decrease the values of PSI₁ and PSI₂ (Figs. 1e and 1f) relative to P_M3 and DPS values (Figs. 1b and 1d;Figs. 3b–3d).

The concentrations of Al_Ox did not vary with initial soil pH (Fig. 4d) compared with Al_M3, indicating a more stable reaction of the ammonium oxalate extracting solution on Al and Fe over the range of acidic pH (Van der Zee et al. 1987). The oxalate extraction dissolves amorphous Al and Fe at the surfaces of which P is retained and not the crystalline Al and Fe (McKeague and Day 1966). These amorphous Al and Fe are made up mainly of Al and Fe hydrous oxides originating from pedogenesis and weathering processes (McKeague and Day 1966). Our results therefore confirm that at low soil pH as those observed in our study, Mehlich-3 extraction tends to overestimate the concentrations of Al_M3, which may result into significant decrease in PSI values.

The literature also shows that Al and Fe bound phytates could be dissolved by Mehlich-3 extracting solution particularly at low-pH conditions (Penn et al. 2018). However, we did not find significant changes in total organic P or total P among soils with varying initial soil pH to support the role of phytates in the studied soils (Figs. 7a and 7b). In addition, we did not observe any changes in P sorption characteristics of the soils, which may be ascribed to compounds other than Ca added with lime (Figs. 5a and 5d). At the pH conditions prevailing in our soils, the acidity of Mehlich-3 solution had little effect on Fe because Fe oxides and oxyhydroxide minerals such as goethite and hematite remain partly insoluble. This is also well demonstrated by the 1:1 regression between PSI₁ and PSI₂ (Fig. 3a) and similar relations between DPS vs. PSI₁ and DPS vs. PSI₂ (Figs. 3b and 3c). Similar observations were made on a range of soils in the Fraser Valley with Fe having minor effect on PSI (Messiga et al. 2021). The minor role of Fe could also be associated with the pedogenesis of the soil as Dystric Brunisols (Canadian System of Soil Classification 1998) or Typic Dystroxerepts (Soil Survey Staff 2010), which are moderately developed with the absence of appreciable quantities of Fe sesquioxides within the solum (Smith et al. 2011). The high concentrations of Al_M3 and P_M3 at low initial soil pH are therefore driven by the solubility of Al oxides and oxyhydroxides and further enhanced by the acidity of Mehlich-3 extracting solution (Tran et al. 1990).

Increasing soil pH and changes in agri-environmental indicators

The concentrations of Al_M3 significantly reduced with addition of CaCO₃ (Fig. 4a). Liming can increase OH^– ions that increase precipitation of Al(OH)₃ (Haynes 1982; Penn and Camberto 2019). The Ca derived from CaCO₃ added to the soil decreased Al_M3 concentrations, which is consistent with the neutralizing power of CO₃^2– towards the acidic component (HNO₃ and CH₃COOH) of Mehlich-3 extraction (Zhang and Kovar 2009). In addition, the added Ca further reacted with fluoride (F^–) used in the Mehlich-3 solution, which likely fixed some phosphates from the solution leading to decreased P_M3 concentrations in limed soils (Penn et al. 2018). It is interesting to observe that the addition of CaCO₃ with its neutralizing power on the acidic component in Mehlich-3 extraction, which decreased Al_M3, did not affect the concentrations of Al_Ox (Figs. 4a, 4b, 4d, and 4e). This result further supports the more specific reaction of the ammonium oxalate extracting solution on Al and Fe over the range of acidic pH (Van der Zee et al. 1987). One objective of this study was to understand how in soils with initial low pH, the release of phosphate ions previously retained onto Al oxides and oxyhydroxides surfaces into the soil solution would change upon increase in soil pH induced by addition of lime. We had expected that decreased Al_M3 upon liming would result in high P_M3 and P_w concentrations and thus increased PSI values and risk of P losses by runoff. Our results showed that PSI values did not significantly change with increased soil pH from low to near-neutral following applications of CaCO₃. Therefore, the decreased concentrations of Al_M3 were not translated into increased PSI values and then potential risk of P losses (Figs. 1e and 1f). The lack of PSI change observed in our study could be primarily due to re-association of P with Ca added with CaCO₃ (Fig. 4c). The increase in the concentration of Ca_M3 with addition of CaCO₃ was described by a power regression indicating a more than proportional increase of Ca_M3 with increasing soil pH (Fig. 4c). The P sorption data showed that the effect of Ca was influenced by the initial soil pH and thus the LR or amount of CaCO₃ needed to reach the target pH. We found that the addition of CaCO₃ affected the sorption characteristics of soils at initial pH values of 4.1 and 5.8 (Figs. 5a and 5d) but not at initial pH values of 4.8 and 5.5 (Figs. 5b and 5c). A comparison of the parameters of the Langmuir equations describing the sorption data of the soils at the respective initial pH upon addition of CaCO₃ showed that S_max and k_a values were only influenced at initial pH values of 4.1 and 5.8 (Figs. 5a and 5d).

Indeed, the S_max and k_a values decreased for initial pH 4.1 and increased for initial pH 5.8 upon addition of CaCO₃. The low S_max and K_a values obtained with addition of CaCO₃ at initial pH 4.1 shows that the reduction of Al sorption sites was not compensated by the activity of added Ca. Finally, we would also argue that the decreased concentrations of P_w after changing the initial soil pH from 4.1 to 6 are due to the amount of Ca added to the soil with CaCO₃ to reach the target pH (Fig. 4c). Indeed, less Ca was added to the soil with initial soil pH = 4.8 and 5.5 to reach the target soil pH = 6. Our results are in line with previous works on Mehlich-3 extractions and the declining Al_M3 concentrations reported with increased soil pH (Penn et al. 2018; Messiga et al. 2021). Another important aspect of our study is the high concentration of P_w at initial soil pH 5.8 despite similar total organic and total P across the range of soils (Figs. 1a, 7a, and 7b). This is an indication of the inhibition activity of Al oxides and oxyhydroxides at high pH (Penn et al. 2018).

Increased soil pH and phosphorus fractionation

The use of NH₄Cl solution (Chang and Jackson 1958) could explain the contrast between NH₄Cl-P and concentrations of P_w with increasing initial soil pH (Figs. 1c and 6a). Gianello and Amorim (2015) showed that NH₄Cl solution performs well at extracting Al and other cations in acidic soils indicating the contribution of some amount of P associated with Al (Figs. 6a and 6b). The decrease in the concentrations of NH₄Cl-P and NH₄F-P with additions of CaCO₃, particularly for soils with initial pH 4.1 (Fig. 5a), is consistent with the trend of P_w concentrations (Fig. 1c). The strong adsorption of F^– on soil colloids at pH > 5.5 explains the decrease in NH₄F-P with increasing pH, which would otherwise displace adsorbed P ions into the solution (Chang and Overby 1986). It is worth noting that the decrease in NH₄Cl-P and NaOH-P following addition of CaCO₃ for soils with initial pH 4.1 was translated into increased concentrations of NaHCO₃-P (Fig. 6d). We also found similar trends among the other P fractions and CBD-P in soils with initial pH values of 5.5 and 5.8 (Fig. 6e), although the change was not significant. These results are consistent with the role of Ca in sorption characteristics of the soil upon addition of CaCO₃, particularly S_max and k_a values for soils at initial pH values of 4.1 and 5.8 (Figs. 5a and 5d). It is therefore possible that the CBD-P and H₂SO₄-P represent the sink of P accumulation from other fractions upon liming. The behavior of NaOH-P showed less significant variations with addition of CaCO₃, which is consistent with the low solubility of Fe at the initial pH under consideration. This result is consistent with the small variation of Fe_M3 and Fe_Ox with soil pH (Figs. 4b and 4e).

Relationship between P and agri-environmental indicators, chemical soil properties, and P fractions

From the PCA results, soil pH was inversely related to Al_M3, Fe_M3, and NH₄Cl-P (Fig. 8). The decrease in Al_M3 and Fe_M3 with increasing soil pH has several consequences, including an increase in the concentration of P_w and PSI₁ and PSI₂. However, the close relationship between P_w and Ca_M3 indicates that phosphate ions released by Al and Fe oxides and oxyhydroxides with increased soil pH were later re-associated with Ca. In the conditions of our study, it is difficult to state whether CBD-P and H₂SO₄-P represent transient fractions of higher solubility that would buffer the solution P following uptake by plants or transport with runoff or leaching. To the best of our knowledge, the solubility of H₂SO₄-P at target pH selected for this study is high compared with that of NH₄F-P and NaOH-P (Barrow 2017). It is well known that the availability of P to plants in acidic soils increases with liming of soil (Penn and Camberato 2019). The present study shows that co-precipitation of phosphate anions with Ca²⁺ can temporarily buffer this pH-induced increase in P availability. It would be interesting to understand how other liming materials such as dolomitic lime or MgCO₃ with less affinity with phosphate would affect the concentration of P_w and PSI values upon changes in soil pH.