Soil sulphur mineralization in relation to plant sulphur availability in soils under pastures

Abstract

A simultaneous comparative study was conducted in the glasshouse under identical conditions for 20 weeks using three different systems (i.e. open incubation, pots with plants and pots without plants) to evaluate soil sulphur (S) mineralization in relation to plant S availability to ryegrass plants. Soils used were obtained from camp and non-camp sites of a long-term fertilizer trial (0, 188 and 376kg superphosphate (SP)/ha/year applied since 1952) on the Lismore stony silt loam soil (Udic Ustochrept) at the Winchmore Irrigation Research Station, New Zealand. Carrier-free ³⁵SO₄-S (1.48MBq per leaching tube of 30g soil and 14.8MBq per pot of 300g soil) was added and preconditioned for two weeks in an incubator before commencing the glasshouse experiments.

Results obtained showed that rates and amounts of soil ³²S mineralization were significantly different in the three systems in all soils. In the open incubation system, a rapid mineralization of ³²S occurred during the first four weeks followed by no significant amounts of ³²S mineralization for the remainder of the 20 weeks in any of the soils studied, probably as a result of loss of labile organic S during the initial leachings and decreases in substrate availability with time. In the two pot systems, ³²S mineralization was slow initially and increased significantly with time.

Comparing the two pot systems, the effect of plants was shown only during early harvests probably due to the depletion of nutrients with increased crop removal with time in the limited amount of soil in pots at later harvests. However, ³²S mineralization was not suppressed at later harvests in the presence of plants probably due to the production of sulphohydrolases by roots and rhizosphere micro-organisms, thereby indicating the possibility of enhanced soil S mineralization in the presence of plants if soil limitations in the pots were overcome. The ³⁵S data also showed a greater dynamic role of microbial activity in the pot systems in releasing or retaining ³⁵S through soil S mineralization/immobilization at different harvests. Significantly higher soil ³²S mineralization occurred in fertilized than control treatment at some harvests in pots without plants indicating the possibility of higher soil ³²S mineralization in fertilized treatments in the presence of adequate substrate availability for microbial activity. Differences in soil ³²S mineralization between camp and non-camp soils were not significant, probably due to the lack of significant differences in initial soil ³²S fractions between these soils.

Changes in C-bonded and ester ³²SO₄-S fractions in any of the systems did not account for soil ³²S mineralization patterns probably due to a relatively small proportion of ³²S was involved in soil S cycling compared to the large amounts of these organic ³²S fractions in soils. The ³⁵S data also did not assist in identifying the mineralizable source of soil organic S in contributing to plant S uptake as many and varied transformations occurred in C-bonded and ester ³⁵SO₄-S fractions within a short period of 20 weeks.

Comparing the two pot systems, significant decreases in CaCl₂, KH₂PO₄ and KCI-40 extractable inorganic, HI -reducible and total ³²S (inorganic and labile organic S) were observed in the presence of plants. These decreases showed significant relationships to plant ³²S uptake. However, amounts of these extractable ³²S forms were largely utilised by the 8th week harvest and subsequently amounts of extractable ³²S remaining were small, indicating that a steady state of labile organic ³²S did not occur and mineralized ³²S present during these harvests might have originated from the non-labile organic S. Changes in non-labile organic S in NaHCO₃ and NaOH extractable ³²S forms showed the possibility of soil S mineralization from different soil S pools not all of which are reflected in plant S availability. In addition, the ³⁵S data showed simultaneous occurrence of mineralization and immobilization processes in different soil S pools. Thus, it is difficult to estimate plant S availability based on soil S extraction methods. A field trial was conducted as a split-plot design with cultivated (0-7Smm depth of cultivation) and uncultivated sub-plots as split-plots superimposed on main treatment plots (0, 188 and 376kg SP/ha/year) in non-camp sites in the long-term trial at the Winchmore Irrigation Research Station, New Zealand where soils were collected from the same trial for the glasshouse experiments. Carrier-free ³⁵SO₄-S was applied to all plots (323MBq/m²) and allowed to pre-condition for two weeks. Perennial ryegrass seeds were sown only in cultivated sub-plots while in uncultivated sub-plots, existing pasture (mainly a mixture of ryegrass and white clover) was allowed to regrow. The trial was under border strip irrigation. A total of five harvests (29, 66, 125, 225 and 373 days after pre-conditioning) were made so as to cover the four seasons in a year (started in October 1996 and ended in October 1997).

Results obtained showed that dry matter yields were significantly affected by fertilization and cultivation with significantly higher dry matter yields in fertilized than the control treatment although differences between the 188 and 376kg SP/ha/year treatments were not significant. Differences in dry matter yields between cultivated and uncultivated plots were observed mainly during the early harvests (due to establishment of newly sown ryegrass in cultivated plots) but not at later harvests. Comparing different harvests, seasons affected pasture growth rate in all treatment plots. Plant ³²S uptake at different harvests was dependent on dry matter yields and soil S supply, as concentrations of ³²S in rye grass and white clover tops did not show significant dilution with time. Specific activities of plant ³⁵S uptake (ryegrass and white clover tops) varied at different harvests indicating that the size of plant-available ³²S and ³⁵S pools influenced the uptake of ³²S and ³⁵S by plants in addition to the interrelationship of plant S uptake with dry matter yields.

Significantly higher ³²S mineralization was observed in fertilized compared with control treatment at later harvests after the initial immobilization of ³²S due to higher availability of inorganic ³²S at early harvests. This initial immobilization was greater in cultivated than uncultivated plots because of less plant S demand due to less growth of plants initially in cultivated plots. Plant S demand, microbial activity, amounts of labile soil organic S and turnover rates of S from plant residues, all of which are under the influence of season, affected soil S mineralization differently in different treatment plots and affected the size of the plant-available S pool at different times.

Changes in C-bonded ³²S, ester ³²SO₄-S and extractable HI-reducible ³²S forms (extracted by KH₂PO₄ and NaHCO₃) did not explain soil ³²S mineralization patterns at different harvests. Changes in the NaHCO₃ extractable S pool at different harvests indicated the turnover of soil S in different soil S pools and showed the possibility of simultaneous synthesis and decomposition of organic matter. This made it difficult to identify the mineralizable soil S pools contributing to plant S uptake. Furthermore, amounts of ³²S uptake by plants at anyone time were small (<2-4% of total soil ³²S) and did not correspond to changes in different soil extractable ³²S forms.

Since plant ³²S uptake at different harvests was related to differences in soil S supply to plants, it is more appropriate to estimate soil S mineralization by taking plant S uptake into consideration rather than S transformations in different soil S pools in the bulk of soil organic S. However, net mineralized ³²S at different harvests did not fully correspond to net plant ³²S uptake due to the accumulation of some mineralized ³²S in the soil inorganic S pool and the presence of high amounts of initial inorganic S especially in fertilized treatments. These results suggest that plant S availability could be modelled based on net changes in soil inorganic S and plant S uptake to provide a reliable measure of soil S mineralization. As seasons affected significantly soil inorganic S and plant S uptake, rates of soil S mineralization obtained under glasshouse conditions could not be used to predict plant S availability in the field. On the other hand, although net soil ³²S mineralization determined at different harvests in the field provided a more reliable measure of soil S mineralization, a greater frequency of harvests than conducted in the present field study would give a better understanding of the contributory role of soil S mineralization to plant S availability in different seasons.