Transformation of added phosphorus to acid upland soils with different soil properties in Indonesia

Abstract

The transformation of added phosphorus (P) to soil and the effect of soil properties on P transformations were investigated for 15 acid upland soils with different physicochemical properties from Indonesia. Based on oxide-related factor scores (aluminum (Al) plus 1/2 iron (Fe) (by ammonium oxalate), crystalline Al and Fe oxides, cation exchange capacity, and clay content) obtained from previous principal component analyses, soils were divided into two groups, namely Group 1 for soils with positive factor scores and Group 2 for those with negative factor scores. The amounts of soil P in different fractions were determined by: (i) resin strip in bicarbonate form in 30 mL distilled water followed by extraction with 0.5 mol L⁻¹ HCl (resin-P inorganic (P_i) that is readily available to plant), (ii) 0.5 mol L⁻¹ NaHCO₃ extracting P_{i and} P organic (P_o) (P which is strongly related to P uptake by plants and microbes and bound to mineral surface or precipitated Ca-P and Mg forms), (iii) 0.1 mol L⁻¹ NaOH extracting P_i and P_o (P which is more strongly held by chemisorption to Fe and Al components of soil surface) and (iv) 1 mol L⁻¹ HCl extracting P_i (Ca-P of low solubility). The transformation of added P (300 mg P kg⁻¹) into other fractions was studied by the recovery of P fractions after 1, 7, 30, and 90 d incubation. After 90 d incubation, most of the added P was transformed into NaOH-P_i fraction for soils of Group 1, while for soils of Group 2, it was transformed into resin-P_i, NaHCO₃-P_i and NaOH-P_i fractions in comparable amounts. The equilibrium of added P transformation was reached in 30 d incubation for soils of Group 1, while for soils of Group 2 it needed a longer time. Oxide-related factor scores were positively correlated with the rate constant (k) of P transformation and the recovery of NaOH-P_i. Additionally, not only the amount of but also the type (kaolinitic) of clay were positively correlated with the k value and P accumulation into NaOH-P_i. Soils developed from andesite and volcanic ash exhibited significantly higher NaOH-P_i than soils developed from granite, volcanic sediments and sedimentary rocks. Soil properties summarized as oxides-related factor, parent material, and clay mineralogy were concluded very important in assessing P transformation and P accumulation in acid upland soils in Indonesia.

Key words:

• acid upland soils
• Indonesia
• phosphorus
• soil properties
• transformation

INTRODUCTION

Phosphorus (P) sorption experiments in acid upland soils in Indonesia showed that soils varied widely in their capacity to sorb P (Hartono et al. 2005). Principal component analyses and stepwise regression showed that oxide-related factor (aluminum [Al] plus 1/2 iron [Fe][by ammonium oxalate], crystalline Al and Fe oxides, cation exchange capacity [CEC] and clay content) were the main components actively contributing to their P sorption maxima and bonding energies in Langmuir simulation. The soils with high scores of oxide-related factor exhibited high P sorption maxima and bonding energies. Desorption experiments showed that by extracting with 0.01 mol L⁻¹ CaCl₂, the soils had low desorbability of the sorbed P (Hartono et al. 2005). One of the explanations for this result was that it might correspond to irreversible reactions of adsorbed P with soil compounds that lead to a stronger bond through rearrangement of the phosphate ions on the surface (Kuo and Lotse 1973).

In general, added fertilizer P transforms into relatively insoluble compounds of Al and Fe in acid soils (Ghani and Islam 1946). Long-term transformation of fertilizer added P under cultivation in the field in acid soils (e.g. according to field experiments) is transformed not only into the form of inorganic P of Al and Fe fractions, but also into the form of organic P fractions and these P were retained with a different bond (Oberson et al. 2001; Schmidt et al. 1996; Verma et al. 2005; Zheng et al. 2002).

Most of the upland soils in Indonesia are acidic in reaction and are deficient in available P. Acid soils account for approximately 57% of the total upland soils and are developed from different types of parent materials (Subagyo et al. 2000). Phosphorus fertilization is a key component of increasing soil productivity in these acid upland soils. The adsorption of additional P decreases as the quantity of P already adsorbed increases (Barrow 1990); thus, repeated additions have cumulative benefits. Accordingly, the best strategy to manage P in these upland soils is to apply P fertilizer in very large amounts in an initial addition or in frequent small applications, depending on the characteristics of the soils to have large and long-term residual values (Cassman et al. 1993; Kamprath 1967; Linquist et al. 1996).

The differences in soil properties related to oxide-related factor and clay mineralogy in these upland soils may have different P distributions to the added P. Information on P transformation of added water-soluble P into various P pools and the effect of the soil properties to the P transformation process in these acid upland soils have not been well evaluated. Understanding the transformation of added P, for example, in one cropping season (90 days), in these soils is very important to identify appropriate P management strategies. The objectives of the present study were to examine: (1) the transformation process of added P to soils, (2) the differences in the dominant P fraction among soil types and how these fractions are determined by soil properties.

MATERIALS AND METHODS

Soil samples

Surface horizons from 15 acid upland soils from Sumatra, Java and Kalimantan were used in this study. Soil samples were air-dried and crushed to pass through a 2-mm mesh sieve. Selected chemical and physical properties that can affect P dynamics were analyzed by Hartono et al. (2005). Principal component analysis demonstrated that three principal components in Indonesian acid upland soils influenced P sorption, namely oxide-related factor (Al + 1/2 Fe [by ammonium oxalate], crystalline Al and Fe oxides, CEC and clay content) (PC₁), acidity plus 1.4 nm mineral-related factor (exchangeable Al and 1.4 nm minerals) (PC₂), and organic carbon (C)-related factor (organic C and organically bound Fe) (PC₃) (Hartono et al. 2005). Parent material of the soil samples and their chemical and physical properties are presented in Table 1. In Table 1, soil samples were divided into two groups according to the factor scores of PC₁ of surface soils. Group 1 denotes soils with factor scores of PC₁ more than zero and Group 2 denotes soils with factor scores of PC₁ less than zero. The dominant silicate clay minerals in the soils of Group 1 were kaolinite (>90%), except for soils from Gajrug, which were dominated by smectite (95%), and soils from PD1 were kaolinite (71%) and Al-interlayered vermiculite–chlorite intergrades (29%). The dominant silicate clay minerals in the soils of Group 2 were kaolinite for soils from PD2, Kota Bumi and Rimbo Bujang (>95%) and kaolinite and vermiculite for soils from Sebuluh, Kota Bangun, SM-BP and SM 1, and kaolinite and smectite for soils from Teluk Dalam. The percentage of kaolinite in soils from Kota Bangun, SM-BP, SM1 and Teluk Dalam was more than 75%, while in soils from Sebuluh it was approximately 50% (Hartono et al. 2005). All of the Java soils belonged to Group 1, whereas the Kalimantan soils belonged to Group 2.

Incubation

Duplicate 0.5-g samples of each soil were treated with 300 mg P kg⁻¹ as a solution of KH₂PO₄, mixed thoroughly, and incubated for 1, 7, 30 and 90 days at 25°C and 80% of field capacity, which corresponds to 40–65% in gravimetric moisture content (w/w). Controls without the addition of P were also included for each soil.

The P addition rate, 300 mg P kg⁻¹, was chosen based on the lowest P sorption maxima obtained in the previous study. Ninety days was chosen as the end of incubation to evaluate P transformation in one cropping season.

Phosphorus fractionation

Soil incubation was followed by P fractionation. The protocol of P fractionation according to Tiessen and Moir (1993) is summarized in Fig. 1. This P fractionation method was selected because it is considered to have a wide range of P fractions of inorganic and organic P, from the most available fraction to the most unavailable fraction to plants. The modified Hedley method, which is similar but includes CHCl₃ treatment before NaHCO₃ extraction to lyse microbial cells, was

Table 1 Parent material and some chemical and physical properties of the soils used

Location	Soil order^†	Parent material	Depth (cm)	pH H2O (1:1.5)	Organic C (g kg^−1)	Clay (%)	Pyrophosphate		Oxalate		DCB		Base Saturation (%)	Exch. Ca (cmolc kg^−1)	Exch. Mg (cmolc kg^−1)	Exch. Al (cmolc kg^−1)	CEC (cmolc kg^−1)	Bray 1-P (mg P kg^−1)
							Alp (g kg^−1)	Fep (g kg^−1)	Alo (g kg^−1)	Feo (g kg^−1)	Ald (g kg^−1)	Fed (g kg^−1)
Group I
Darmaga	Inceptisols	Andesite	0–13	4.37	22.2	70.0	1.89	2.45	3.24	3.64	7.46	45.1	22.6	3.19	0.63	4.21	18.1	2.91
Gajrug	Ultisols	Volcanic ash-sedimentary rock mixture	0–14	4.24	25.2	65.2	4.97	4.15	6.38	9.11	8.67	29.2	15.0	4.29	1.04	23.1	40.3	5.65
Tegal Garu	Ultisols	Volcanic ash	0–15	4.34	15.6	65.5	0.80	0.34	2.49	18.2	5.75	48.4	8.88	0.36	0.62	2.70	13.8	2.91
Karang Pandan	Alfisols	Volcanic ash	0–10	4.77	12.9	67.4	0.42	nd	3.07	4.70	6.31	50.3	45.0	4.01	1.55	nd	13.1	3.30
Pringsurat	Alfisols	Volcanic ash	0–9	4.56	11.0	88.6	0.39	nd	3.48	4.67	7.24	57.6	43.8	4.06	1.16	nd	13.3	0.90
PD1	Ultisols	Andesite	0–14	5.61	53.9	55.5	1.41	1.93	5.65	20.7	13.2	53.9	62.5	12.7	0.28	nd	21.1	1.19
Sitiung	Ultisols	Volcanic ash-sedimentary rock mixture	0–14	4.42	19.4	66.3	1.27	0.62	4.60	5.08	8.26	37.9	7.27	0.76	0.05	5.02	17.0	4.73
Group II
PD 2	Ultisols	Granite	0–27	4.64	18.2	23.7	0.39	0.10	0.87	4.09	2.04	9.94	37.7	2.28	0.63	nd	8.83	7.18
Kota Bumi	Alfisols	Volcanic sediments	0–20	4.34	17.0	36.7	0.52	0.11	1.78	6.85	3.98	17.4	36.1	2.04	0.44	nd	7.57	2.24
Rimbo	Ultisols	Volcanic sediments	0–4	4.04	20.9	46.2	0.40	0.13	2.81	11.9	5.90	28.3	7.90	0.19	0.36	1.64	9.31	7.15
Bujang
Sebuluh	Ultisols	Sedimentary rock	0–9	4.12	10.8	28.7	0.50	0.40	1.47	2.50	2.89	20.0	30.6	1.75	1.74	3.01	12.1	9.22
Kota Bangun	Ultisols	Sedimentary rock	0–4	4.51	13.6	14.8	0.39	0.66	0.59	1.41	1.05	5.82	18.3	0.33	0.54	1.21	5.74	12.3
SM-BP	Ultisols	Sedimentary rock	0–11	4.60	17.3	16.1	0.25	1.10	0.64	5.03	1.26	12.2	19.7	0.44	0.56	1.22	6.70	10.3
SM 1	Ultisols	Sedimentary rock	0–10	4.66	18.1	31.3	0.51	0.57	1.26	3.61	2.60	19.0	49.0	3.54	1.80	0.83	11.7	6.86
Teluk Dalam	Ultisols	Sedimentary rock	0–9	4.37	9.50	29.9	0.47	0.71	0.97	2.97	2.16	17.8	43.6	2.37	0.98	1.62	8.83	48.0
Hartono et al. (2005). ^† Soil Survey Staff (1998). Group I, soils with factor scores of oxide-related factor (PC1) greater than zero; Group II, soils with factor scores of oxide-related factor (PC1) less than zero; nd, non-detectable.

Figure 1  Flow chart of the phosphorus (P) fractionation into various inorganic and organic P fractions.

not applied in the present study because a possible readsorption of lysed P on soil minerals would bring additional difficulties to the interpretation of experimental results. Phosphorus in solution was determined using the procedure of Murphy and Riley (1962). Characterization of sequential soil P extraction was explained as follows: (1) resin-P_i was interpreted as P that is readily available to plants, (2) NaHCO₃-P_i, -P_o were interpreted as P that are strongly related to P uptake by plants and microbes and bound to the mineral surface (Mattingly 1975; Tiessen and Moir 1993) or precipitated Ca-P and Mg-P forms (Olsen and Sommers 1982), (3) NaOH-P_i, -P_o were interpreted as P that are more strongly held by chemisorption to Fe and Al components of the soil surface, (4) HCl-P_i was interpreted as Ca-P of low solubility. Residual P was determined by subtracting from total P the sum of resin-P_i, NaHCO₃-P_i, -P_o, NaOH-P_i, -P_o, and HCl-P_i (Schmidt et al. 1996) for the original soils with no P addition. Residual P was interpreted as occluded P and recalcitrant organic forms (Dobermann et al. 2002; Tiessen and Moir 1993).

Phosphorus transformation evaluation

The transformation of added P into other fractions was evaluated by the recovery of P fractions after each period of incubation. The values were obtained by subtracting the control from the recovery of P fractions. Statistical analyses (correlation analysis and anova followed by a Tukey's test) were applied using SYSTAT 8.0 (SPSS Inc. 1998).

RESULTS

Original P distribution

Original P distribution is presented in Table 2. The values of resin-P_i and NaHCO₃-P_i and -P_o, which were considered to be biologically available P (labile P) of the two groups, were much lower than those of residual P, NaOH-P_i and -P_o. Most P in these soils was accumulated in residual P, which was relatively insoluble. Phosphorus was also significantly accumulated as organic P in the form of labile organic P (NaHCO₃-P_o) and

Table 2 Original phosphorus (P) distribution and total P of the soil samples

Location	Depth (cm)	Resin-Pi (mg P kg^−1)	NaHCO3-P		NaOH-P		HCl-Pi (mg P kg^−1)	Total-P (mg P kg^−1)	Residual-P (mg P kg^−1)
			NaHCO3-Pi (mg P kg^−1)	NaHCO3-Po (mg P kg^−1)	NaOH-Pi (mg P kg^−1)	NaOH-Po (mg P kg^−1)
Group I
Darmaga	0–13	12	3	22	46	79	2	469	305
Jasinga	0–14	9	3	33	41	75	1	375	213
Tegal Garu	0–15	6	26	44	43	73	5	446	249
Karang Pandan	0–10	7	8	34	98	52	4	609	406
Pringsurat	0–9	6	9	25	89	49	5	535	352
PD1	0–14	6	9	11	38	57	nd	346	225
Sitiung	0–14	5	3	24	55	75	2	519	355
Group II
PD 2	0–27	9	10	25	28	69	9	394	244
Kota Bumi	0–20	6	4	15	21	47	4	238	141
Rimbo Bujang	0–4	9	5	6	20	30	nd	228	158
Sebuluh	0–9	10	4	23	28	56	1	346	224
Kota Bangun	0–4	9	11	13	13	32	1	189	110
SM-BP	0–11	9	13	11	20	31	1	208	123
SM 1	0–10	7	10	25	37	71	2	378	226
Teluk Dalam	0–9	19	34	22	90	144	11	516	196
Group I, soils with factor scores of oxide-related factor (PC1) greater than zero; Group II, soils with factor scores of oxide-related factor (PC1) less than zero; nd, non-detectable.

Table 3 Range of the total recovery of the added phosphorus

	Incubation time (days)
	1	7	30	90
Range (%)	91–100	83–93	83–100	78–100
Mean (%)	97	89	94	90
SD (%)	3	3	6	6
SD, standard deviation.

more strongly held by chemisorption to Fe and Al components of the soil surface (NaOH-P_o). HCl-P_i (Ca-P) was much lower than resin-P_i and NaHCO₃-P_i and -P_o. On average, the P fraction in the original soils of the two groups followed the order: residual P > NaOH-P_o > NaOH-P_i > NaHCO₃-P_o > resin-P_i ≈ NaHCO₃-P_i.

A lack of resin-P_i and NaHCO₃-P_i in the original soils indicated low P availability, and suggested that P fertilizer application in these upland soils was low and infrequent. The low application in these upland soils was also confirmed by the relatively low amount of NaOH-P_i, which ranged from 38 to 98 mg P kg⁻¹ in Group 1 and from 13 to 90 mg P kg⁻¹ in Group 2. For comparison, the amount of NaOH-P_i in acid upland soil from Matalom, Philippines, was 133 mg P kg⁻¹ when 466 kg P ha⁻¹ was applied (Dobermann et al. 2002).

Kinetics of P in different fractions during the incubation

The range of total recovery of added P after 1, 7, 30 and 90 days incubation is presented in Table 3. The recovery of added P was close to 100% in all periods of incubation, and only a small portion may be fixed in the residual P.

Resin-P_i

Recovery of added P in the form of resin-P_i is presented in Fig. 2a for soils with factor scores of PC₁ more than zero (Group 1) and Fig. 2b for soils with factor scores of PC₁ less than zero (Group 2). After 1 day incubation, added P was distributed in resin-P_i ranging from 128 to 213 mg P kg⁻¹ in Group 1 and from 190 to 220 mg P kg⁻¹ in Group 2. After 7 days incubation, the recovery ranged from 67 to 167 mg P kg⁻¹ in Group 1 and from 94 to 174 mg P kg⁻¹ in Group 2. After 30 days incubation, the recovery of resin-P_i ranged from 60 to 144 mg P kg⁻¹ in Group 1 and from 65 to 137 mg P kg⁻¹ in Group 2. After 90 days incubation, the recovery of resin-P_i ranged from 43 to 110 mg P kg⁻¹ in Group 1 and 54 to 111 mg P kg⁻¹ in Group 2.

Soils of Group 1 exhibited lower resin-P_i compared with soils of Group 2 after 1 day incubation. After 7, 30 and 90 days incubation, both groups exhibited similarly low amounts of resin-P_i compared with the initial P addition (300 mg P kg⁻¹). These results suggested that

Figure 2  Kinetics of the recovery of phosphorus (P) fractions in soils with factor scores of PC1 greater than zero and soils with factor scores of PC1 less than zero.

most resin-P_i was transformed to other fractions. From 1 to 7 days incubation, recovery of resin-P_i decreased significantly, but the difference was not significant from 7 to 30 days or from 30 to 90 days incubation. This suggests that in all the soils studied, after 7 days incubation the added P transformed slowly to other fractions.

NaHCO₃-P_i

The recovery of added P in the NaHCO₃-P_i fraction is presented in Fig. 2c for soils of Group 1 and Fig. 2d for soils of Group 2. After 1 day incubation, added P was distributed in NaHCO₃-P_i from 17 to 33 mg P kg⁻¹ in Group 1 and from 26 to 48 mg P kg⁻¹ in Group 2. After 7 days incubation, the recovery ranged from 10 to 36 mg P kg⁻¹ in Group 1 and from 34 to 92 mg P kg⁻¹ in Group 2. After 30 days incubation, the recovery ranged from 9 to 34 mg P kg⁻¹ in Group 1 and from 51 to 106 mg P kg⁻¹ in Group 2. After 90 days incubation, the recovery ranged from 22 to 35 mg P kg⁻¹ in Group 1 and from 56 to 106 mg P kg⁻¹ in Group 2.

The distribution of the added P to this fraction was higher in Group 2. There was not any significant increase in Group 1 during the incubation period, suggesting that most added P was not transformed into this fraction after 30 and 90 days incubation. In contrast, soils of Group 2 showed a significant increase in this fraction after 30 and 90 days incubation, suggesting that significant amounts of added P were transformed into this fraction.

NaHCO₃-P_o

The recovery of added P in the NaHCO₃-P_o fraction is presented in Fig. 2e for soils of Group 1 and Fig. 2f for soils of Group 2. The two groups showed that the added P was not distributed into this fraction.

NaOH-P_i

The recovery of added P in the NaOH-P_i fraction is presented in Fig. 2g for soils of Group 1 and Fig. 2h for soils of Group 2. After 1 day incubation, NaOH-P_i ranged from 65 to 120 mg P kg⁻¹ in Group 1 and from 28 to 57 mg P kg⁻¹ in Group 2. After 7 days incubation, the recovery ranged from 79 to 144 mg P kg⁻¹ in Group 1 and from 44 to 89 mg P kg⁻¹ in Group 2. After 30 days incubation, the recovery ranged from 115 to 163 mg P kg⁻¹ in Group 1 and from 49 to 103 mg P kg⁻¹ in Group 2. After 90 days incubation, the recovery ranged from 126 to 155 mg P kg⁻¹ in Group 1 and from 77 to 116 mg P kg⁻¹ in Group 2.

The soils of Group 1 exhibited a larger NaOH-P_i fraction than those of Group 2 in each period of incubation. After 30 and 90 days incubation, in the former soils, approximately 40 to 50% of added P was transformed into this fraction. The equilibrium of P transformation was reached at 30 days in soils of Group 1 because an increase in this fraction was not observed after 30 days incubation. For soils of Group 2, the amount of NaOH-P_i was comparable with that of NaHCO₃-P_i and the former fraction tended to increase during incubation.

NaOH-P_o

The recovery of added P in the NaOH-P_o fraction is presented in Fig. 2i for soils of Group 1 and Fig. 2j for soils of Group 2. Fig. 2i,j shows that there is no regular pattern concerning the distribution of added P to this fraction. Most soils of Group 1 and some soils of Group 2 exhibited some amounts of the NaOH-P_o fraction after 30 and 90 days incubation, and the former group exhibited higher amounts of the NaOH-P_o fraction.

HCl-P_i

The recovery of added P in HCl-P_i is presented in Fig. 2k for soils of Group 1 and Fig. 2l for soils of Group 2. The two groups of soils showed that the added P was not distributed in this fraction. This is because the soils were acid and contained negligible amounts of free CaCO₃ to form Ca-P, which is extracted by 1 mol L⁻¹ HCl.

DISCUSSION

Decreasing trend in resin-P_i during the incubation in relation to soil properties

Soils of Group 1 with high clay content exhibited lower resin-P_i in the early stage of incubation (1 day), although this difference disappeared later, and also exhibited lower NaHCO₃-P_i fraction than soils of Group 2, with lower clay content, in each period of incubation. This supports the results of Kamprath and Watson (1980), in which finer-textured soils normally have larger P-buffering capacities and lower extractable P contents than coarser-textured soils. The difference in P transformation process between soils of Group 1 and Group 2 suggested that the form and distribution of added P were related to the soil characteristics summarized as oxide-related factor (PC₁), that is, Al + 1/2 Fe (by ammonium oxalate), crystalline Al and Fe oxides, CEC and clay content.

With respect to the soil from Gajrug of Group 1, it exhibited exceptionally high resin-P_i among soils of Group 1 and was similar to some of the soils from Group 2 (Fig. 2a) in all periods of incubation, although the clay contents of this soil were relatively high (65%) (Table 1). The soil from Gajrug had the highest factor scores of PC₂ as well PC₁ (Hartono et al. 2005), which means that this soil contained high amounts of exchangeable Al and the silicate clay mineral of this soil was dominated by 1.4 nm mineral, namely smectite. In addition, it also had a significant amount of Al and Fe oxides. This suggests that exchangeable Al in smectite was not reactive with H₂PO⁻ ₄. Exchangeable Al and smectite positively affected labile resin-P_i. Exchangeable Al and 1.4 nm minerals were reported to decrease P sorption maximum (Hartono et al. 2005). Smectite was also reported to have a lower P adsorbing surface than kaolinite (Sollins 1991), suggesting that clay mineralogy affects the P transformation process as shown in the case of the Gajrug soil.

Table 4 First-order resin-Pi in equilibrium (c) and the kinetic rate constant for phosphorus (P) transformation (k)

Location	Depth (cm)	c (mg P kg^−1)	k (d^−1)	r^2
Group I
Darmaga	0–13	104	0.842	0.936
Gajrug	0–14	140	0.748	0.926
Tegal Garu	0–15	104	0.977	0.952
Karang Pandan	0–10	95	1.209	0.959
Pringsurat	0–9	75	1.394	0.981
PD1	0–14	57	1.226	0.993
Sitiung	0–14	107	1.054	0.943
Group II
PD 2	0–27	106	0.759	0.962
Kota Bumi	0–20	118	0.914	0.902
Rimbo Bujang	0–4	119	0.874	0.927
Sebuluh	0–9	138	0.736	0.900
Kota Bangun	0–4	110	0.191	0.918
SM-BP	0–11	69	0.608	0.983
SM 1	0–10	88	0.687	0.977
Teluk Dalam	0–9	136	0.705	0.917
Group I, soils with factor scores of oxide-related factor (PC1) greater than zero; Group II, soils with factor scores of oxide-related factor (PC1) less than zero.

The trend of decreasing recovery of resin-P_i as a function of incubation time is more or less uniform for all soils and is well described by the first-order kinetic equation P_t = c + P_de^−kt , where P_t is the amount of resin-P_i at time t, c is the amount of resin-P_i in equilibrium, P_d is the amount of decayed-resin-P_i, k is the rate constant and t is the reaction time in days. The resin-P_i in equilibrium (c) and the first-order kinetic rate constant (k) for P transformation is presented in Table 4. The values of c and k varied among soils. Most soils of Group 1 had higher k values than the soils of Group 2. Unlike the k values, c values show no consistent pattern with regard to soil group.

The k values (Table 4) suggested that the soils of Group 1 were faster than the soils of Group 2 in the added P transformation process. Again, soil from Gajrug exhibited the lowest k value among the Group 1 soils and the highest c value in these soils, suggesting the effect of exchangeable Al and 1.4 nm mineral (smectite).

The correlation between factor scores of PC₁, PC₂, PC₃ and recovery of P fractions after 90 days incubation and c and k values is presented in Table 5. The factor score of PC₁ is positively significantly correlated with the k value (P < 0.05). This suggests that the transformation process in soils of Group 1 was faster than in soils in Group 2 because the P adsorbing surfaces in soils of Group 1 were much larger than those in Group 2 (Griffin & Jurinak 1974; Kuo & Lotse 1973), which is reflected by higher oxide-related factor scores.

Table 5 Simple correlation coefficient (Pearson r) between factor scores of PC1, PC2, PC3 and the recovery of P fractions after 90 days of incubation and resin-Pi in equilibrium (c) and rate constant (k)

	c	k	Resin-Pi	NaHCO3-Pi	NaOH-Pi
Factor score of PC1	−0.25	0.59*	−0.30	−0.84**	0.82**
Factor score of PC2	0.68**	−0.49	0.66**	−0.00	−0.28
Factor score of PC3	0.24	0.08	−0.24	−0.06	0.13
*P < 0.05;
**P < 0.01 with Bonferroni probabilities.

In contrast, PC₂ was positively significantly correlated with the c value and resin-P_i (P < 0.01), suggesting that exchangeable Al and 1.4 nm minerals increased resin-P_i. This result confirmed that soil acidity, which is associated with exchangeable Al, and the content of 1.4 nm minerals, of which smectite and vermiculite are common, should be included in models used to estimate P sorption maxima in Indonesian acid upland soils.

Transformation of P into different fractions

The two groups of soils showed a different direction in added P transformation after 7, 30 and 90 days incubation when their resin-P_i were similar. Fig. 2c,d,g,h and a correlation analysis (Table 5) suggested that soils in Group 1 accumulated most of the remaining added P as NaOH-P_i fraction and only a small amount as NaHCO₃-P_i fraction. In contrast, the soils of Group 2 accumulated P not only as NaOH-P_i fraction, but also as NaHCO₃-P_i in significant amounts. With respect to NaHCO₃-P_i, there were two possibilities regarding where this P fraction was derived from: (1) from the surface of minerals (Mattingly 1975; Tiessen and Moir 1993), (2) from the precipitated Ca-P and Mg forms (Olsen and Sommers 1982) because of reactions with dissolved Ca and Mg. The possibility of precipitated Ca-P and Mg-P forms was calculated by using the amount of added P (300 mg P kg⁻¹), water content during incubation, soil pH (1:1.5) and the exchangeable Ca and Mg of the soils after zero (0) days incubation. From the calculation of both cations to precipitate with H₂PO⁻ ₄, only Ca-P precipitation could possibly occur in both Group soils (Lindsay 1979).

However, because there was not any increase in NaHCO₃-P_i observed during the period of incubation in Group 1 soils, it is concluded that the abundant oxides surface retards the transformation of the added P into NaHCO₃-P_i extractable forms. Thus, for soils of Group 1, P accumulation in NaOH-P_i results from the high P adsorbing surfaces provided by soil properties related to oxide-related factor and kaolinite. The majority of the remaining added P was chemisorbed by Al and Fe oxides and the broken edges of Al layers of kaolinite. The broken edges of Al layers of kaolinite are variable charge surfaces containing – Al(OH) groups with P sorption properties similar to those of Al and Fe oxides surfaces (Parfitt 1978), although chemisorption by kaolinite was less than that by Al and Fe oxides (Sollins 1991). The appearance of NaHCO₃-P_i fraction in each stage of incubation showed that at a given amount of P chemisorbed by Al and Fe oxides (NaOH-P_i) also made equilibrium with the P to sorbed on the surfaces of minerals (NaHCO₃-P_i) besides with resin-P_i.

In soil from Gajrug (Group 1), in which the predominant silicate clay mineral was smectite, although the P transformation process was relatively slow and resin-P_i was high compared with other Group 1 soils, the remaining added P was not accumulated in NaHCO₃-P_i (Fig. 2c) but rather in NaOH-P_i (Fig. 2g). This was because of a high contribution of oxide-related factor components (i.e. Al + 1/2 Fe [by ammonium oxalate], crystalline Al and Fe oxides, CEC and clay content) (Table 1).

The NaOH-P_i fraction was considered to be P that was chemisorbed on amorphous Al and Fe minerals (Dobermann et al. 2002; Tiessen et al. 1984; Zheng et al. 2003). In the soils examined in the present study, factor scores of PC₁ were strongly correlated with the NaOH-P_i fraction (Table 5), suggesting that P in these upland soils was chemisorbed on both amorphous and crystalline Al and Fe oxide minerals.

Significant P accumulation in the NaHCO₃-P_i in the soils of Group 2 was considered to be enhanced primarily by low P fixing surfaces that contribute to the increase in NaOH-P_i. It was P adsorbed on the surface of minerals and/or precipitated Ca-P forms.

Organic P in the form of NaHCO₃-P_o was not detected in all periods of incubation and only small amounts of NaOH-P_o were detected, particularly after 30 and 90 days incubation in both groups of soils. In this experiment, only inorganic P fertilizer (KH₂PO₄) was applied. The insignificant effect of added inorganic P fertilizer on NaHCO₃-P_o and NaOH-P_o contents was previously observed by other investigators in their short-term experiments (Beauchemin and Simard 2000; Hedley et al. 1982; Ivarsson 1990; Zheng et al. 2003). However, the small occurrence of NaOH-P_o in some soils of Group 1 and Group 2, for example, in the 30 and 90 day incubations, was supposed to result from the incorporation of added P into microbial biomass. The P_o from dead microorganisms was chemisorbed by Al and Fe oxides and/or the broken edge of Al layers of kaolinite and its amount was higher in soils with high clay and Al and Fe oxide contents (soils of Group 1) as reported by O’Halloran (1993) and Beauchemin and Simard (2000). There was not any added P distributed in NaHCO₃-P_o in both groups, suggesting that NaHCO₃-P_o acted as a transitory pool rather than as a sink in P transformation.

This experiment was conducted under the condition that neither additional input of substrates for microbial activities nor influence of plant growth was expected. It is likely that P transformation in the organic fractions will be more dynamic if plants were included in the system because of P uptake, root exudates and more variable microbial activity. However, as it has been reported that added inorganic P fertilizer had little impact on organic P fractions in short-term applications when plant growth was included (Beauchemin and Simard 2000; Hedley et al. 1982; Ivarsson 1990; Zheng et al. 2003); thus, the results of this experiment are still predictable of field conditions, particularly in one cropping season (90 days).

Effect of parent material on NaHCO₃-P_i and NaOH-P_i fractions

The effect of parent material on NaHCO₃-P_i and NaOH-P_i fractions is presented in Table 6. Soils from sedimentary rocks exhibited significantly (Tukey's test, P < 0.05) higher NaHCO₃-P_i than soils developed from andesite, volcanic ash, volcanic sediment and volcanic ash–sedimentary rock mixture, but not from soils developed from granite.

Table 6 Effect of parent material on NaHCO3-Pi and NaOH-Pi fractions after 90 days of incubation

Parent material	NaHCO3-Pi (mg P kg^−1)	SD (mg P kg^−1)	NaOH-Pi (mg P kg^−1)	SD (mg P kg^−1)	N
Andesite	35cd	1.58	141a	5.66	4
Volcanic ash	39cd	2.97	139a	13.0	6
Volcanic ash – sedimentary rock mixture	28d	3.40	129b	4.36	4
Volcanic sediments	62bc	1.87	106bc	2.65	4
Sedimentary rock	89a	21	95c	15.1	10
Granite	76ab	2.12	102c	2.12	2
Means followed by the same letter within a column of NaHCO3-Pi are not significantly different (Tukey's test, P < 0.05). Means followed by the same letter within a column of NaOH-Pi are not significantly different (Tukey's test, P < 0.01).

In contrast, soils developed from andesite and volcanic ash exhibited significantly (Tukey's test, P < 0.01) higher NaOH-P_i than soils developed from granite, volcanic sediments, volcanic ash–sedimentary rock mixture and sedimentary rocks. Standard deviations of NaHCO₃-P_i and NaOH-P_i were relatively low on all parent materials.

Differences in parent materials could be used to assess the transformation and accumulation of P applied in these upland soils. Soils that were developed from volcanic ash and andesite accumulated the most applied P into NaOH-P_i because of relatively high soil properties related to oxide-related factor. In contrast, in soils from sedimentary rocks, granite and volcanic sediments because of relatively low soil properties related to oxide-related factor, applied P was accumulated not only into NaOH-P_i fraction but also into NaHCO₃-P_i fraction with significant amounts. Most of the volcanic soils were situated in Java and, in part, in Sumatra, while most Kalimantan soils were developed from sedimentary rocks. These results can be used to determine P transformation and accumulation when fertilizer P is applied in acid upland soils in Java, Sumatra and Kalimantan according to their parent materials.

This study suggested that the application of P fertilizer in very large amounts in these upland soils could be done in the kaolinitic soils with high oxide-related factor component and developed from andesite and volcanic ash parent materials. This is because most of the added P was accumulated in the NaOH-P_i fraction. This fraction acted as a sink for added P and long-term residual values could be expected (Beck and Shanchez 1994; Guo et al. 2000; Zheng et al. 2003).

The heavy application of P fertilizer in soils developed from sedimentary rocks with low oxide-related factor components and smectite as the predominant clay mineral is not recommended because of a relatively slow P transformation process and significant accumulation in labile P fractions (resin-P_i and NaHCO₃-P_i). Although P will be more available for plants, there will be the risk of P loss and the transfer from soils to waterbodies. Frequent and small amounts of P application are recommended.

Conclusions

Parent materials, oxide-related factor contributed by oxalate-soluble Al and Fe, crystalline Al and Fe oxides, CEC and clay content were very important in assessing P transformation and P accumulation in acid upland soils in Indonesia. As for clay, the type of silicate clay minerals should be considered because P in smectic soil, which is associated with exchangeable Al, was more extractable and slower with respect to the P transformation process compared with kaolinitic soils. The P transformation process in these upland soils was very fast, especially for kaolinitic soils with high amounts of oxide-related factor components. The equilibrium was reached in 30 days and most of the P added was accumulated into the NaOH-P_i fraction. The low P adsorbing surfaces enhanced the P transformed into NaHCO₃-P_i in soils with low amounts of oxide-related factor. Soils developed from andesite and volcanic ash accumulated the added P into NaOH-P_i fraction significantly more than soils developed from granite, volcanic sediments and sedimentary rocks. In contrast, soils from sedimentary rocks, granite and volcanic sediments, because of relatively low oxide-related factor components, accumulated the added P not only into NaOH-P_i fraction but also into NaHCO₃-P_i fraction in significant amounts. The latter fraction was bonded on the surface of minerals or precipitated Ca-P forms for soils with high amounts of calcium. This finding is of significance for further reviews in programming P fertilization in Indonesian acid upland soils.

ACKNOWLEDGMENTS

We thank the farmers in Java, Sumatra and East Kalimantan for their assistance in the collection of the soil samples. We also thank Mr Tetsuhiro Watanabe for his assistance in the field, Dr Hitoshi Shinjo for his assistance in the statistical analysis and Mr M. Turner for his assistance in the preparation of this manuscript.