Soil morphological and chemical properties in homegardens on sandy beach ridges along the east coast of Peninsular Malaysia

ABSTRACT

The morphological and chemical properties of homegardens in the beach ridges with interspersed swales (BRIS) soils were evaluated in order to find a clue for developing sustainable agricultural management. Field survey and soil sampling were conducted at the homegardens and secondary forests in Sungai Ular Village and in the experimental farm of Malaysian Agriculture Research Institute (MARDI), Cherating Station. Chemical fertilizers including ash, charcoal waste and plant litter were applied to the homegardens in both the inland-ward area (HG-I) and in the shoreline area (HG-S). Manure was applied in HG-I and seafood waste from fish processing was buried in the soils in HG-S. High correlation was found between total carbon (T-C) and cation exchange capacity (CEC), indicating soil organic matter was the determinant factor for CEC in the very sandy BRIS soils. The levels of T-C, total nitrogen (T-N), and CEC at 0–10 cm in HG-I in parallel with increasing ages of homegardens. The soils in 0–10 cm and 20–30 cm showed higher levels of T-C and T-N with higher C/N ratios in HG-I than in HG-S and the MARDI farm. The high levels of exchangeable Ca, Mg, and K in HG-I compared with the MARDI farm, suggesting that improved CEC with increasing soil organic matter in HG-I heighten the retention of basic cations supplied as chemical fertilizer, ash, and manure. The level of available P was higher in HG-S than HG-I, attributed by the seafood waste application in the shoreline area. Thus, on the BRIS, the levels of soil organic matter and nutrients can be sustained by the homegarden management although it cannot be regarded as a closed system compared with those in the other tropical regions because chemical fertilizer was used and a certain amounts of nutrients seemed to leach down beyond plant rooting depth.

KEY WORDS:

• Entisols
• homegarden
• sandy beach ridges
• soil fertility
• soil morphology
• Spodosols

1. Introduction

Soils exceeding 90% sand content are distributed along the coastal areas all over the tropics. These soils are classified mostly as Entisols or Spodosols in the USDA classification system (Soil Survey Staff 2014). Because of their intrinsic infertility due to sandy texture, these soils are usually regarded as unsuitable for agricultural use and therefore left under forest or grassland vegetation (Thompson et al. 1996; Sommer et al. 2001; Perez et al. 2011).

Such soils can also be found along the east coast of the Malay Peninsula and are called Beach Ridges Interspersed with Swales (BRIS) soils (Nossin 1964; Paramananthan 1987). BRIS accounts for 1.23% (162,000 ha) of the total land area of Peninsular Malaysia (Zahari et al. 1982). The soils located close to the shoreline and those inland wards are reported to be classified as Typic Quartzipsamments and Typic Haplohumods, respectively, in the USDA system (Okubo et al. 2003; Roslan et al. 2010; Khairul et al. 2017). Unlike other regions, the BRIS land in the Malay Peninsula has been settled and used for agriculture, although the soil is recognized as problematic lowland soil for farming, as well as peat swamp soil and acid sulfate soil. At present, because of the lack of arable land and increasing demand for food, agricultural pressure on the BRIS land has been rising with intensive management, ranging from small-scale commercial farming by local farmers to a large-scale Cocos nucifera (coconut) plantation. Although many agronomic studies have been conducted to help increase crop production through improving fertilizer schemes (Abdul-Hamid et al. 2009; Hanafi et al. 2010; Che Lah et al. 2011), knowledge of BRIS soils is still limited to fundamental aspects such as their formation process (Okubo et al. 2003; Roslan et al. 2010). In our previous study, we investigated the influence of farming activities on the soil characteristics, comparing an experimental field of the Malaysian Agriculture Research Institute (MARDI) with the adjacent remnant forest (Khairul et al. 2017). We showed that agricultural activities even with a well-managed system in the experimental farm caused an obvious reduction in soil organic matter in the A horizon and excessive loss of nutrients from the soils.

In general, tropical homegardens have been admired as a sustainable agroforestry system, which are composed of multi-story combinations of various plants with the least input of agricultural materials and can provide daily necessities such as food, materials, and medicines (Fernandes and Nair 1986; Kumar and Nair 2004). Several studies have pointed out that soil fertility levels of tropical homegardens are sustained through efficient organic matter and nutrient cycling between diverse plants and soils and their recycling through daily input of waste such as garbage ash and manure (Torquebiau 1992; Nair 2001; Kumar and Nair 2004). However, many previous studies have reported on Ultisols, Oxisols, or other soil classes with a somewhat clayey texture (Jensen 1993a, 1993b; Funakawa et al. 2007; Pandey et al. 2007; Lattirasuvan et al. 2010; Tanaka et al. 2010, 2012). As for sandy soils used for homegardens, the high fertility of the Terra Preta in the Amazonian Basin is well known (Glaser et al. 2000; Lehmann et al. 2003). However, these soils are thought to be historical relics formed long ago after the pre-Columbian era. On the other hand, on the BRIS on the east coast of the Malay Peninsula where substantial settlement is believed to have started since the eighteenth or nineteenth century, the soils used for homegardens have not yet been investigated. During the survey in our previous study (Khairul et al. 2017), we found that homegarden soils were dark in color down to the deep layers, suggesting a preferable soil condition for plant growth. This finding may provide a clue to help develop sustainable and ecofriendly systems for agriculture in BRIS soils. Thus, the main objective of this study is to investigate and evaluate the soil morphological and chemical properties in homegardens on the BRIS in Peninsular Malaysia in order to provide fundamental information on their soil conditions.

2. Materials and methods

2.1. Study area

This study was conducted at Sungai Ular Village and the MARDI experimental farm at Cherating Station in Cherating district in February 2014 and September 2015 (Fig. 1). The average annual precipitation and temperature from 2002 to 2012 were 2645 mm and 32.1°C, respectively. The topography in the village is relatively flat, and leveling for housing had been seldom conducted. The flat topography seems to be common for the BRIS in this district (Khairul et al. 2017).

Figure 1. Map of the study sites and soil pits.

The village was divided into the west and east sides of the main road (hereafter called the inland-ward area and shoreline area, respectively) (Fig. 1), which can be characterized by the different types of residents and the related land uses. The inland-ward area is occupied mainly by farmers and to a lesser extent by households of other occupations. Many of the latter also used to be farmers. There is one aggregate of farmlands in the northwest of the village for commercial crop cultivation. The secondary forests, occasionally a vestige of abandoned residences, are scattered within the village. Meanwhile, in the shoreline area, while much of the land is used for resort hotels, restaurants, and schools, the rest is occupied mainly by fishermen, and by the managers of the restaurants and hotels to a lesser extent. The number of houses was much less than in the inland-ward area. There were no farmland and forest in this shoreline area except for a windbreak forest belt along the beach. This division of the village seemed to almost coincide with the dominant soil types found in the two areas, which are Typic Haplohumods or Typic Haplorthods at inland-ward and Typic Quartzipsamments at shoreline area, respectively.

In both the inland-ward and shoreline areas, every household possesses a homegarden of various sizes and planted with various kinds of plant. In this study, we regarded the space surrounding a house and encircled by a fence as a homegarden, except for the spaces planted with coconut trees in the shoreline area as mentioned below.

The MARDI experimental farm extending 1 km from the shoreline to inland-ward is located to the north of the village (Fig. 1) and contains fruit tree lots such as Cocos nucifera (coconut), Manilkara zapota (ciku in Malaysia), Anacardium occidentale (cashew nut), and Averrhoa carambola (starfruit), and pasture lots with grasses such as Brachiaria humidiocola (koronivia grass). Detailed information about the farm was provided in our previous paper (Khairul et al. 2017). For the fruit trees, chemical fertilizers (200 g each of Christmas Island rock phosphate, magnesium sulfate, and ground magnesium limestone) and organic amendments (1 or 2 kg of oil-palm empty fruit bunches and palm oil mill sludge) were applied upon planting. Then, 100 to 150 g of an NPK compound fertilizer (12:12:17 + trace elements (B, Cu, Zn)) and 50 g of urea were applied under the canopy every month during the first year, followed by 200 to 250 g of the compound fertilizer every month from the second year.

2.2. Soil sampling and survey

We selected twenty-five homegardens as study sites: Twenty were located in the inland-ward area (HG-I) and five in the shoreline area (HG-S). The information on households and homegardens was collected through direct interviews of the household member who is principally responsible for homegarden management. The information includes the occupation of the owners, the age and area size (in acre) of the homegarden, the name, number and usage of planted crops, the type, amount and frequency of fertilizer application, the frequency of watering, and livestock raised in the homegarden.

To avoid the influence on soils of the diverse plants and their different management, soil sampling was conducted where the Mangifera indica (mango) and Cocos nucifera (coconut) trees were planted because they were found in every homegarden. Soils were sampled at depths of 0–10 cm and 20–30 cm about 1 m from these trees (HG-Im and HG-Sm for mango; HG-Ic and HG-Sc for coconut). In the shoreline area, the coconut trees were usually planted outside the fence of a homegarden, and therefore, we took samples from there for HG-Sc. We also collected soil samples from the spaces used for vegetable cropping (HG-Iv and HG-Sv) as well as three lots of secondary forests in the inland-ward area (SF-I). Although vegetables seemed to be an important part of homegardens, six HG-I were not planted with any vegetables for several reasons, for example, lack of time to take care them or little space for planting. Soil profiles were described at two HG-Iv as well as two SF-I (HG-Iv-1, SF-I-1, HG-Iv-2, and SF-I-2; each located adjacent to each other) and soil samples were collected based on pedogenic horizons. Meanwhile, there was no sufficient space to make soil profile in the five HG-S with avoiding to damage plants. Additional reason was no secondary forest remained in the shoreline area which can be used as control.

In the MARDI field, soils were collected from the fruit tree lots located along the two 1 km transect lines used in the previous study (Khairul et al. 2017); these lots included three coconut lots, one ciku lot, and one cashew nut lot in the inland-ward area (MA-I) and two coconut lots and one starfruit lot in the shoreline area (MA-S). The soils were collected at depths of 0–10 cm and 20–30 cm from two points, namely 1 m apart from the trees where fertilizers were applied and in between the trees (MA-I-ft and MA-S-ft and MA-I-ib and MA-S-ib).

The soil samples were air-dried and passed through a 2 mm mesh sieve for physicochemical analysis.

2.3. Analytical methods

Soil pH was determined in water and 1 M KCl in a soil to solution ratio of 1:5 using the glass electrode method (Horiba pH meter F-21). Electrical conductivity (EC) was measured using an EC meter (CM-14P; TOA). Exchange acidity was determined using a titration method with 0.01 M NaOH and the amount of exchangeable Al with 0.01 M HCl after adding 4% NaF solution. The amount of exchangeable H was calculated as the difference between exchangeable acidity and exchangeable Al. Exchangeable bases (Ca, Mg, K, and Na) were extracted with 1 M ammonium acetate at pH 7.0. The concentration of Ca, Mg, and K were determined using atomic adsorption spectrophotometry, whereas that of Na was determined using flame photometry (AA-6800; Shimadzu). The ammonium ion adsorbed in the residue was replaced by 10% NaCl and determined using the steam distillation and titration method as cation exchange capacity (CEC). The amounts of total carbon (T-C) and nitrogen (T-N) were analyzed using a CN corder (JM 1000; J-Science Lab). Available P was determined using the Bray II method (Kuo 1996) with the phosphomolybdenum blue method at wavelength 710 nm (UV-140–02; Shimadzu). For the soil profile samples, particle size distribution was determined using the pipette and sieving method: < 2 μm for clay, 2–20 μm for silt, and 20–2000 μm for sand fractions.

2.4. Data analysis

The chemical data of the soil samples were expressed on an oven-dried basis. All of the statistical analyses were performed using Excel Statistics ver. 2015 for Windows (SRI, Tokyo, Japan). The soils collected from one HG-I and one HG-S were omitted as outliers from the analysis because of extremely high values of exchangeable Ca (9.8 to 22.8 cmol_c kg⁻¹) and available P (890 to 1640 mg P kg⁻¹). Soil parameters between HG-I and HG-S were compared using Student’s or Welch’s t-test in terms of each crop. For each HG-I or HG-S, soil parameters between different crops (coconut, mango, and vegetables) were paired within a homegarden and compared using paired one-way analysis of variance (paired ANOVA), and subsequently Tukey’s multiple comparison (p < 0.05). Six HG-I where vegetable was not planted were omitted from this paired ANOVA. Soil parameters were also compared for each inland-ward area (HG-Im, SF-I, MA-I-ib, and MA-I-ft) and shoreline area (HG-Sm, MA-S-ib, and MA-S-ft) using ANOVA with Tukey’s multiple comparison (p < 0.05).

3. Results

3.1. Characteristics of the homegardens

The ages of the homegardens after establishment varied widely from 5 to 45 years with an average of 23 years for HG-I and from 30 to 50 years with an average of 42 years for HG-S. This difference seems to result from the history of the village, which was originally settled from the shoreline area by fishermen, followed by the reclamation of the inland-ward area by farmers after construction of the main road. The size of the homegardens varied from 0.25 to 4 acre for HG-I and from 0.25 to 3 acre for HG-S. Both HG-I and HG-S were enclosed with a fence. Because the inland-ward area was densely populated, bordering houses in HG-I were to denote land tenure. Meanwhile, in the shoreline area, abundant vacant land still remained as grassland, occasionally used for goat and cow grazing, and requires enclosing the homegardens to prevent damage to plants by animals.

According to the interviews for HG-I and HG-S, coconut and mango trees were planted in abundance because of the perception that these trees are suitable for the sandy soil. Many kinds of fruit trees were recorded such as Areca catechu (betel palm), Syzygium aqueum (water apple), Artocarpus heterophyllus (jackfruit), Citrus aurantifolia (key lime), and Nephelium lappaceum (rambutan). However, the abundance and frequency of these trees were low, and one or two trees were recorded within one homegarden if any. Various vegetables and tuber crops were also recorded in both HG-I and HG-S such as Psophocarpus tetragonolobus (winged bean), Vigna unguiculata (long bean), Solanum melongena (eggplant), Capsicum annuum (chili), Manihot esculenta (tapioca), and Colocasia esculenta (taro), as well as ornamental flowers such as Hibiscus sp. (rose mallow) and Allamanda cathartica (golden trumpet). Vegetables and tuber crops were planted in one or several small plots (one to several square meters). Ornamental flowers were planted mostly in flowerpots. Periodic watering was conducted for tree seedlings, vegetables, tuber crops, and flowers but not for mature fruit trees. The fruits, vegetables, and tuber crops were mostly for self-consumption or used as gifts for neighbors and relatives and were sometimes sold to a local market in the case of HG-I.

The crops were fertilized, but the practice varied widely depending on the crops and households; common to both HG-I and HG-S, kitchen waste, household garbage, and slashed or dead tree residues were burned at a designated place, and the resulting charcoal and ash were applied. Chemical fertilizers (NPK compound) were also applied in homegardens in both areas for vegetables and tuber crops at the rate of a finger pinch per plant at one-week to 16-week intervals. Fertilizer was also applied to trees at a lesser frequency when the leaf color turned yellowish or upon the fruiting season. Meanwhile, several conspicuous differences were found in the practices between HG-I and HG-S. First, in all HG-I, urea and homemade-manure from chicken dung, or goat dung in one case, were applied at a higher frequency for vegetables. However, urea and manure were not applied in HG-S because of the lack of sources of manure. Second, the coconut trees in HG-I were actively managed, but those in HG-S were left unmanaged because of the growth stage exceeding harvestable height. Third, the households in HG-S buried seafood waste from fish processing such as fish heads, bones, and internal organs in the soils in the homegardens to avoid odors and as well as for the effects as a fertilizer. In addition, in HG-S, trash such as plastic bottles and bags, tin cans, and scrap wood was dumped in the homegardens.

3.2. Morphological and chemical properties of the soil profiles

Table 1 shows the soil profiles description of HG-Iv and the adjacent SF-I. The spodic horizon could not be recorded in all the pits because groundwater seepage caused the pit walls to collapse, preventing observation of the deeper layers. However, soil shoveled from the bottom of the soil pit contained blackish or dark brownish clods.

Table 1. Soil profiles of the homegardens and adjacent secondary forests.

Secondary forest						Homegarden
Horizon	Depth (cm)	Color	Structure^a	Root^b	Boundary^c	Horizon	Depth (cm)	Color	Structure^a	Root^b	Boundary^c
SF-I-1						HG-Iv-1 (vegetable plot)
O	10–			fine root-mat
A1	0–8	10YR2/3	3mg	4vf-m,2c	gs	Ap1	0–12	10YR2/1	2f-ms	3vf-m	cs
A2	–23/26	10YR2/3	2ms	3vf- m	gs	Ap2	–30	10YR2/1	2ms	2vf-f,1m	gs
AE	–32/34	10YR3/4	2ms	2vf-f,1c	cs	A	–57	10YR2/2	0	1vf-f	gs
EA1	–67	10YR5/8	2ms	1vf	gs	AE	–64	10YR5/3	0	1vf-f	gs
EA2	–88	2.5Y6/6	0	0	gs	E1	–80	10YR6/2	0	0	gs
E	–140+	2.5Y7/2	0	0		E2	–90+	10YR6/3	0	0
SF-I-2						HG-Iv-2 (vegetable plot)
O	3–	root mat and partially decomposed leaf litter
A1	0–8	10YR3/4	3mg	4vf-m	gs	Ap1	0–7	10YR2/1	2f-ms	3vf-f	cs
A2	–21	10YR3/3	2ms	3vf-m	cs	Ap2	–13	10YR2/2	2ms	2vf-f	cs
AE	–57	10YR4/6	2ms	2vf-m	gs	AE1	–20	10YR5/4	0	2vf-f	gs
EA1	–90	2.5Y5/4	0	1vf-m	gs	AE2	–30	10YR2/2	0	2vf-f	gs
EA2	–108	2.5Y6/3	0	0	cs	AE3	–65	10YR2/3	0	1vf	gs
E1	–137	10YR7/1	0	0	cs	AE4	–85	10YR3/3	0	0	gs
E2	–146	10YR7/2	0	0	cs	EA	–95	10YR5/2	0	0	cs
E3	–170+	2.5Y7/1	0	0		2E	–105+	10YR6/3	0	0

a) Grade: 0, massive; 1, very weak; 2, weak; 3, moderate. Size: f, fine; m, medium. Type: s, subangular blocky; g, granular.

b) Abundance: 0, no roots; 1, very few; 2, few; 3, common; 4, many. Size: vf, very fine; f, fine; m, medium; c, coarse.

c) Distinctness: c, clear; g, gradual. Topography: s, smooth.

The soil profiles of SF-I-1 were composed of the O, A, AE, EA, and E horizons. Soil color was brownish black in the A horizons and increased in value and chroma with depth to be grayish yellow at the E horizon. A moderate granular structure was observed in the A1 horizon, followed by a weak subangular blocky structure to EA1. The roots of the trees, fern undergrowth, and other herbaceous species were abundantly distributed in the A horizons, but their penetration was restricted below the EA1 horizon. All horizon boundaries were smooth in topography. The morphological properties of SF-I-2 were similar to those of SF-I-1.

The profile of HG-Iv-1 showed similar horizon stratification to that in SF-I-1, except for the absence of the O horizon and the presence of the Ap horizon. The important morphological feature of HG-Iv-1 was the soil color in each horizon, which was lower in value and/or chroma compared with SF-I-1. No clear structure was observed below the Ap horizon. The distribution and abundance of roots, mostly of crop roots, were similar to that in SF-I-1. The profile of HG-Iv-2 also showed similar morphological properties to that of SF-I-2 with darker soil colors. It is noteworthy that the value and chroma abruptly changed between the AE1 and AE2 horizons from dull yellowish brown to brownish black. At HG-Iv-1 and HG-Iv-2, charcoal was detected in the Ap and A horizons, at sizes mostly less than several millimeters.

All these soil profiles showed a sandy texture with sand content around 90% (Table 2). The levels of T-C and T-N in SF-I-1 and SF-I-2 were 27 and 1.5 g kg⁻¹ in the A1 horizon, respectively, and decreased with depth. The values in the E horizon were low at less than 0.5 g kg⁻¹ for T-C. The C/N ratio showed a similar trend to T-C and T-N. The level of CEC was low, particularly below the EA horizon. The soils were acidic with pH (H₂O) around 5 but less acidic in the E horizon. The levels of exchangeable bases were low throughout the profile, in particular in the E horizon. The levels of available P in SF-I-1 were low at less than 10 mg P kg⁻¹, whereas those in SF-I-2 were 42 and 63 mg P kg⁻¹ in the A1 and A2 horizons, respectively. Meanwhile, the soils from HG-Iv-1 and HG-Iv-2 showed higher levels of total C and N, C/N ratio, exchangeable bases, and available P than those in the corresponding horizons in SF-I-1 and SF-I-2, respectively.

Table 2. Soil physicochemical properties in the homegardens and adjacent secondary forests.

		SF-I-1						HG-Iv-1 (vegetable plot)
Horizon		A1	A2	AE	EA1	EA2	E	Ap1	Ap2	A	AE	E1	E2
Depth	cm	0 − 8	−23/26	−32/34	−67	−88	−140+	0 − 12	−30	−57	−64	−80	−90+
pH (H2O)		5.05	5.18	5.30	5.26	5.10	5.89	5.08	4.58	4.85	5.47	5.46	5.50
pH (KCl)		4.25	4.46	4.76	4.69	4.71	5.07	4.17	4.36	4.56	4.73	4.86	4.61
EC	mS m^−1	2.5	1.4	1.2	1.1	0.6	0.6	10.3	8.3	5.4	2.1	1.9	2.1
T-C	g kg^−1	27.3	23.0	12.2	2.5	1.9	0.3	39.0	39.2	26.0	6.7	5.8	6.2
T-N	g kg^−1	1.52	1.24	0.52	0.16	0.13	0.09	2.3	2.1	0.9	0.9	0.9	0.9
C/N ratio		18.0	18.5	23.4	15.8	14.8	4.0	31.4	32.8	31.0	10.1	9.6	9.2
CEC	cmolc kg^−1	6.40	4.65	2.62	1.10	1.03	0.10	5.58	5.47	4.42	0.23	0.28	3.25
Exch. Ca	cmolc kg^−1	0.07	0.02	nd	0.01	0.01	0.01	0.57	0.15	1.51	0.12	0.13	0.37
Exch. Mg	cmolc kg^−1	0.11	0.03	0.01	0.01	0.02	0.02	0.03	0.02	0.02	0.02	0.02	0.03
Exch. K	cmolc kg^−1	0.04	0.02	0.01	0.01	0.01	0.00	0.22	0.08	0.07	0.03	0.04	0.09
Exch. Na	cmolc kg^−1	0.03	0.02	0.02	0.03	0.01	0.01	0.08	0.02	0.07	0.01	0.02	0.05
Exch. Al	cmolc kg^−1	0.80	0.19	0.06	0.03	0.00	0.00	1.41	1.51	0.97	0.23	0.18	0.05
Exch. H	cmolc kg^−1	0.44	0.50	0.33	0.28	0.24	0.19	0.62	0.23	0.06	0.12	0.10	0.10
Avail. P	mg P kg^−1	9	6	5	3	7	5	147	62	12	69	66	47
Clay	%	6.2	6.7	6.8	5.5	6.5	2.0	7.7	6.1	9.9	3.4	2.5	3.0
Silt	%	1.6	1.9	1.7	1.2	0.2	0.5	1.6	7.7	2.3	16.1	10.8	23.1
Sand	%	92.2	91.4	91.5	93.3	93.3	97.5	90.7	86.2	87.8	80.5	86.7	74.0
		SF-I-2						HG-Iv-2 (vegetable plot)
Horizon		A1	A2	AE	EA1	EA2	E1	E2	E3	Ap1	Ap2	AE1	AE2	AE3	AE4	EA	E
Depth	cm	0 − 8	−21	−57	−90	−108	−137	−146	−170+	0 − 7	−13	−20	−30	−65	−85	−95	−105+
pH (H2O)		4.66	4.91	4.97	4.99	4.97	5.52	5.49	5.4	6.83	6.32	6.98	6.43	5.89	5.69	5.75	5.98
pH (KCl)		4.17	4.40	4.78	4.74	4.72	5.01	4.92	4.88	6.07	5.89	5.61	5.71	5.25	4.95	5.19	5.09
EC	mS m^−1	3.1	2.0	1.5	0.9	1.0	0.6	0.6	0.6	9.7	6.0	2.7	3.3	4.3	3.5	3.1	2.3
T-C	g kg^−1	27.2	20.2	15.4	2.6	1.5	0.3	0.5	0.5	56.5	34.2	13.4	22.6	19.0	11.2	5.2	0.8
T-N	g kg^−1	1.48	0.90	0.50	0.14	0.13	0.07	0.10	0.10	3.2	2.3	0.8	1.1	0.7	0.5	0.4	0.2
C/N ratio		18.3	22.5	30.9	19.0	11.5	3.5	4.7	5.21	42.9	28.1	13.4	21.2	21.2	14.8	8.7	1.6
CEC	cmolc kg^−1	7.66	5.12	4.18	1.20	0.80	0.08	0.30	0.30	8.37	5.81	2.45	4.32	4.07	2.10	0.89	0.18
Exch. Ca	cmolc kg^−1	0.04	0.02	0.01	0.00	0.01	nd	0.01	0.01	3.00	2.70	2.50	2.56	2.29	1.17	0.68	0.10
Exch. Mg	cmolc kg^−1	0.04	0.01	0.06	0.03	0.02	nd	0.02	0.01	0.03	0.04	0.01	0.02	0.07	0.04	0.03	0.06
Exch. K	cmolc kg^−1	0.03	0.01	0.01	nd	nd	nd	0.00	nd	0.18	0.05	0.02	0.02	0.03	0.03	0.03	0.01
Exch. Na	cmolc kg^−1	0.03	0.02	0.02	0.01	0.01	0.01	0.01	0.01	0.25	0.24	0.15	0.18	0.13	0.07	0.05	0.01
Exch. Al	cmolc kg^−1	1.24	0.56	0.15	0.12	0.06	nd	nd	0.03	0.03	0.08	0.10	nd	0.03	0.18	0.05	nd
Exch. H	cmolc kg^−1	0.07	0.12	0.16	0.15	0.14	0.10	0.12	0.11	0.05	0.10	0.05	0.10	0.10	0.17	0.13	0.10
Avail. P	mg P kg^−1	42	63	8	6	6	3	9	16	401	253	343	166	66	32	39	27
Clay	%	5.6	5.1	10.2	8.0	7.9	6.4	6.3	9.7	7.7	4	3.9	3.8	6.2	9.7	6.4	5.1
Silt	%	1.5	2.1	3.1	0.5	0.5	0.1	0.5	0.6	0.1	9	0.1	5.7	0.1	0.1	0.1	0.1
Sand	%	92.9	92.8	86.7	91.5	91.6	93.5	93.2	89.7	92.2	87.0	96.0	90.5	93.7	90.2	93.5	94.8

nd: not detected

Exch., exchangeable; Avail., available.

3.3. Soil chemical properties in the inland-ward and shoreline areas

Table 3 compares the soil chemical properties of HG-I and HG-S. Irrespective of the type of crop, the levels of T-C, T-N, C/N ratio, and CEC were higher at HG-I than at HG-S, in which the trend was more obvious at depth 20–30 cm than 0–10 cm based on the ratio of HG-I to HG-S. No significant difference was found in exchangeable Ca and Mg at 0–10 cm and Ca at 20–30 cm. The level of exchangeable K at both depths differed significantly between HG-I and HG-S, except for the surface soils in vegetable plots and the subsurface soils in mango plots. The ratios of exchangeable Ca, Mg, and K at HG-I to those of HG-S were small compared with those of T-C, T-N, and CEC. On the other hand, available P was conspicuously higher at HG-Ic than at HG-Sc. However, the opposite trends were found for mango and vegetables despite a significant difference only at depth 0–10 cm between HG-Im and HG-Sm.

Table 3. Comparison of soil chemical properties at depths 0–10 cm and 20–30 cm between homegardens in the inland-ward and shoreline areas.

		Coconut				Mango				Vegetable
		HG-Ic (n = 19)	HG-Sc (n = 4)		Ratio^a	HG-Im (n = 19)	HG-Sm (n = 4)		Ratio^a	HG-Iv (n = 13)	HG-Sv (n = 4)		Ratio^a
0–10 cm
pH (H2O)		6.02 ± 0.56	5.92 ± 0.51		1.0	5.99 ± 0.52	6.28 ± 1.55		1.00	6.13 ± 0.67	5.63 ± 1.44		1.1
EC	mS m^−1	5.4 ± 3.0	1.9 ± 0.5	**	2.9	4.6 ± 3.0	8.3 ± 4.7	*	0.5	8.1 ± 6.0	6.2 ± 3.6		1.3
T-C	g kg^−1	30.4 ± 16.5	6.4 ± 1.1	**	4.7	23.2 ± 11.7	8.4 ± 2.5	**	2.8	29.1 ± 11.6	12.4 ± 4.7	*	2.3
T-N	g kg^−1	1.7 ± 1.0	0.5 ± 0.1	**	3.5	1.1 ± 0.6	0.6 ± 0.2	**	1.9	1.5 ± 0.9	1.0 ± 0.4		1.5
C/N ratio		18.2 ± 1.9	13.3 ± 0.8	**	1.4	24.5 ± 14.4	14.4 ± 2.0	**	1.7	21.6 ± 7.0	12.8 ± 1.2	**	1.7
CEC	cmolc kg^−1	8.10 ± 2.95	1.89 ± 1.17	**	4.3	6.96 ± 3.72	2.69 ± 0.76	**	2.6	8.14 ± 3.78	4.25 ± 1.88		1.9
Exch. Ca	cmolc kg^−1	1.44 ± 1.56	0.97 ± 1.34		1.5	1.71 ± 1.77	2.31 ± 1.65		0.7	2.28 ± 3.68	2.85 ± 4.54		0.8
Exch. Mg	cmolc kg^−1	0.25 ± 0.32	0.09 ± 0.06		2.8	0.15 ± 0.14	0.16 ± 0.04		0.9	0.16 ± 0.12	0.08 ± 0.03		1.9
Exch. K	cmolc kg^−1	0.06 ± 0.04	0.03 ± 0.01	*	1.7	0.04 ± 0.02	0.07 ± 0.03	*	0.6	0.05 ± 0.02	0.05 ± 0.02		1.0
Avail. P	mg P kg^−1	128 ± 171	15 ± 10	*	8.6	68 ± 60	150 ± 35	*	0.5	156 ± 120	278 ± 177		0.6
20–30 cm
pH (H2O)		5.83 ± 0.75	5.95 ± 0.53		1.0	5.79 ± 0.42	6.33 ± 1.67		0.9	5.98 ± 0.54	5.90 ± 1.33		1.0
EC	mS m^−1	3.04 ± 1.64	0.94 ± 0.26	**	3.2	2.45 ± 1.34	3.25 ± 4.47		0.8	3.23 ± 1.38	2.16 ± 2.46		1.5
T-C	g kg^−1	21.5 ± 11.0	2.3 ± 1.2	**	9.4	21.5 ± 10.3	1.8 ± 1.5	**	11.9	23.2 ± 12.7	2.8 ± 3.1	**	8.2
T-N	g kg^−1	0.9 ± 0.5	0.2 ± 0.1	**	3.8	1.0 ± 0.6	0.2 ± 0.1	**	4.8	1.1 ± 0.6	0.3 ± 0.2	*	3.8
C/N ratio		24.5 ± 10.0	9.0 ± 2.2	**	2.7	23.7 ± 7.4	8.1 ± 2.7	**	3.0	21.6 ± 3.9	8.3 ± 3.1	**	2.6
CEC	cmolc kg^−1	6.16 ± 4.14	1.09 ± 0.74	**	5.7	6.19 ± 3.37	0.96 ± 0.80	**	6.4	7.05 ± 3.57	1.19 ± 0.88	**	5.9
Exch. Ca	cmolc kg^−1	0.91 ± 1.47	0.38 ± 0.67		2.4	0.84 ± 1.16	0.31 ± 0.17		2.7	0.87 ± 1.72	0.36 ± 0.36		2.4
Exch. Mg	cmolc kg^−1	0.08 ± 0.02	0.09 ± 0.01	*	0.9	0.07 ± 0.07	0.03 ± 0.02	*	2.2	0.04 ± 0.04	0.02 ± 0.01	*	2.5
Exch. K	cmolc kg^−1	0.03 ± 0.02	0.01 ± 0.01	*	2.0	0.02 ± 0.01	0.01 ± 0.01		1.6	0.03 ± 0.01	0.01 ± 0.01	*	1.8
Avail. P	mg P kg^−1	56 ± 93	5 ± 7	*	10.3	30 ± 35	46 ± 55		0.7	61 ± 59	124 ± 212		0.5

HG-I, homegardens in the inland area; HG-S, homegardens in the shoreline area. Average value ± standard deviation.

^a The ratio of HG-I to HG-S. Exch., exchangeable; Avail., available.

* and ** indicate a significant difference at p < 0.05 and < 0.01, respectively, between HG-I and HG-S using the Student’s or Welch’s t-test.

3.4. Soil chemical properties in terms of planted crops

Tables 4 and 5 compare soil chemical properties at depths 0–10 cm and 20–30 cm in terms of different crops in HG-I and HG-S. In the inland-ward area, there were no clear differences in soil parameters between HG-Ic, HG-Im, and HG-Iv, except for available P at depth 0–10 cm. The latter was significantly higher in HG-Iv than in HG-Im (Table 4). Meanwhile, in the shoreline area, the levels of available P and CEC at depth 0–10 cm were significantly higher in HG-Sv than in HG-Sc. In addition, many of the soil parameters, in particular at 0–10 cm, tended to be high in HG-Sm and HG-Sv compared with HG-Sc (Table 5).

Table 4. Comparison of soil chemical properties under different crops of the homegardens in the inland-ward area.

		HG-Ic (n = 13)		HG-Im (n = 13)		HG-Iv (n = 13)
0–10 cm
pH (H2O)		6.02 ± 0.64		5.92 ± 0.58		6.13 ± 0.67
EC	mS m^−1	6.0 ± 3.4		4.9 ± 3.4		8.1 ± 6.0
T-C	g kg^−1	31.7 ± 18.6		22.7 ± 13.7		29.1 ± 11.6
T-N	g kg^−1	1.7 ± 1.2		1.0 ± 0.7		1.5 ± 0.9
C/N ratio		19.0 ± 1.6		27.5 ± 16.7		21.6 ± 7.0
CEC	cmolc kg^−1	7.62 ± 2.49		6.73 ± 4.34		8.14 ± 3.80
Exch. Ca	cmolc kg^−1	1.19 ± 1.45		1.26 ± 1.53		2.28 ± 3.68
Exch. Mg	cmolc kg^−1	0.16 ± 0.20		0.12 ± 0.13		0.16 ± 0.12
Exch. K	cmolc kg^−1	0.05 ± 0.03		0.04 ± 0.03		0.05 ± 0.02
Avail. P	mg P kg^−1	76 ± 62	ab	65 ± 65	a	156 ± 120
20–30 cm
pH (H2O)		5.80 ± 0.87		5.84 ± 0.46		5.98 ± 0.54
EC	mS m^−1	3.2 ± 1.8		2.6 ± 1.4		3.2 ± 1.4
T-C	g kg^−1	22.0 ± 10.4		24.8 ± 10.2		23.2 ± 12.7
T-N	g kg^−1	0.9 ± 0.4		1.1 ± 0.6		1.1 ± 0.6
C/N ratio		26.7 ± 11.5		24.6 ± 8.4		21.6 ± 3.9
CEC	cmolc kg^−1	6.18 ± 4.26		6.83 ± 3.82		7.05 ± 3.57
Exch. Ca	cmolc kg^−1	0.84 ± 1.61		0.93 ± 1.40		0.87 ± 1.72
Exch. Mg	cmolc kg^−1	0.06 ± 0.06		0.08 ± 0.08		0.04 ± 0.04
Exch. K	cmolc kg^−1	0.03 ± 0.02		0.02 ± 0.01		0.03 ± 0.01
Avail. P	mg P kg^−1	30 ± 26		35 ± 41		61 ± 59

HG-I, homegardens in the inland area. c: coconut; m: mango; v: vegetables.

Average ± standard deviation.

Exch., exchangeable; Avail., available.

Different letters indicate a significant difference among crops using Tukey’s multiple comparison (p < 0.05).

Table 5. Comparison of soil chemical properties under different crops of the homegardens in the shoreline area.

		HG-Sc (n=4)		HG-Sm (n=4)		HG-Sv (n=4)
0–10 cm
pH (H2O)		5.92 ± 0.51		6.28 ± 1.55		5.63 ± 1.44
EC	mS m^−1	1.9 ± 0.5		8.3 ± 4.7		6.2 ± 3.6
T-C	g kg^−1	6.4 ± 1.1		8.4 ± 2.5		12.4 ± 4.7
T-N	g kg^−1	0.5 ± 0.1		0.6 ± 0.2		1.0 ± 0.4
C/N ratio		13.3 ± 0.8		14.4 ± 2.0		12.8 ± 1.2
CEC	cmolc kg^−1	1.89 ± 1.17	a	2.69 ± 0.76	ab	4.25 ± 1.88	b
Exch. Ca	cmolc kg^−1	0.97 ± 1.34		2.31 ± 1.65		2.85 ± 4.54
Exch. Mg	cmolc kg^−1	0.09 ± 0.06		0.16 ± 0.04		0.08 ± 0.03
Exch. K	cmolc kg^−1	0.03 ± 0.01		0.07 ± 0.03		0.05 ± 0.02
Avail. P	mg P kg^−1	15 ±10	a	150 ± 35	ab	278 ± 177	b
20–30 cm
pH (H2O)		5.95 ± 0.53		6.33 ± 1.67		5.90 ± 1.33
EC	mS m^−1	0.9 ± 0.3		3.3 ± 4.5		2.2 ± 2.5
T-C	g kg^−1	2.3 ± 1.2		1.8 ± 1.5		2.8 ± 3.1
T-N	g kg^−1	0.2 ± 0.1		0.2 ± 0.1		0.3 ± 0.2
C/N ratio		9.0 ± 2.2		8.1 ± 2.7		8.3 ± 3.1
CEC	cmolc kg^−1	1.09 ± 0.74		0.96 ± 0.80		1.19 ± 0.88
Exch. Ca	cmolc kg^−1	0.38 ± 0.67		0.31 ± 0.17		0.36 ± 0.36
Exch. Mg	cmolc kg^−1	0.02 ±0 .01		0.03 ± 0.02		0.02 ±0 .01
Exch. K	cmolc kg^−1	0.01 ± 0.01		0.02 ± 0.01		0.01 ± 0.01
Avail. P	mg P kg^−1	5 ± 7		46 ± 55		124 ± 212

HG-S, homegardens in the shoreline area. c: coconut; m: mango; v: vegetables.

Average ± standard deviation.

Exch., exchangeable; Avail., available.

Different letters indicate a significant difference among crops using Tukey’s multiple comparison (p < 0.05).

For HG-Im, a significant correlation (p < 0.05) was found between the levels of T-C, T-N, and CEC at depth 0–10 cm and the ages of the homegardens (r = 0.65, 0.52, and 0.60, respectively). Figure 2 depicts such a correlation for T-C. Similar trends were also found in HG-Ic (r = 0.20, 0.20, and 0.40) and HG-Iv (r = 0.54, 0.41, and 0.53), although that only for T-C in HG-Iv was significant (r = 0.54). The content of available P at 0–10 cm of HG-Im and HG-Iv also correlated with homegarden ages (Fig. 3), whereas no correlations were found in exchangeable bases (data not shown). For HG-S, no clear tendencies were found in soil organic matter with homegarden ages.

Figure 2. Relationship between T-C and homegarden age in HG-Im at depth 0–10 cm. n = 19. **, p < 0.01.

Figure 3. Relationship between available P and homegarden age. Closed circle: HG-Iv, open circle: HG-Im. n = 13 and 19, respectively. *, p < 0.05 and **, p < 0.01.

3.5. Soil chemical properties of the homegardens, secondary forests, and the MARDI experimental farm

Table 6 compares the soil chemical properties at depths 0–10 cm and 20–30 cm between homegardens, secondary forests, and the MARDI farm. In the inland-ward area, the soils at depth 0–10 cm in HG-Im and SF-I were significantly higher in T-C, T-N, and CEC than those in MA-I-ib and MA-I-ft. The C/N ratio and the levels of available P tended to be high at HG-Im, compared with SF-I, MA-I-ib, and MA-I-ft. The soils at depth 20–30 cm showed similar trends to those at 0–10 cm though significant differences for the C/N ratio and exchangeable K were also found. Meanwhile, in the shoreline area, the differences in soil properties between HG-Sm, MA-S-ib, and MA-S-ft were not clear, compared with those in the inland-ward area.

Table 6. Comparison of the soil chemical properties between the homegardens, secondary forests, and the MARDI experimental farm.

Inland
		HG-Im (n = 19)			SF-I (n = 3)		MA-I-ib (n = 5)		MA-I-ft (n = 5)
0–10 cm
pH (H2O)		5.99 ± 0.52	b		4.91 ± 0.15	A	5.14 ± 0.31	a	5.14 ± 0.43	a
pH(KCl)		4.88 ± 0.59	b		4.19 ± 0.13	ab	3.97 ± 0.50	a	4.00 ± 0.57	a
EC	mS m^−1	4.58 ± 2.95			3.45 ± 2.05		1.66 ± 0.32		2.40 ± 0.70
T-C	g kg^−1	23.2 ± 11.7	b		30.4 ± 13.3	B	5.4 ± 3.1	a	7.3 ± 1.3	a
T-N	g kg^−1	1.1 ± 0.6	ab		1.7 ± 0.9	B	0.4 ± 0.1	a	0.5 ± 0.1	a
C/N ratio		24.5 ± 14.4			18.9 ± 1.7		12.5 ± 3.1		14.0 ± 2.4
CEC	cmolc kg^−1	6.96 ± 3.72	b		9.47 ± 2.44	B	1.58 ± 0.88	a	1.96 ± 0.17	a
Exch. Ca	cmolc kg^−1	1.71 ± 1.77			0.08 ± 0.09		0.34 ± 0.27		0.71 ± 0.61
Exch. Mg	cmolc kg^−1	0.15 ± 0.14			0.02 ± 0.01		0.13 ± 0.08		0.18 ± 0.09
Exch. K	cmolc kg^−1	0.04 ± 0.02			0.04 ± 0.03		0.02 ± 0.02		0.03 ± 0.03
Avail. P	mg P kg^−1	68 ± 60			23 ± 26		53 ± 70		27 ± 39
20–30cm
pH (H2O)		5.79 ± 0.42	b		5.12 ± 0.13	A	5.57 ± 0.19	ab	5.38 ± 0.37	ab
pH(KCl)		4.64 ± 0.42	b		4.59 ± 0.06	ab	4.15 ± 0.22	a	4.01 ± 0.22	a
EC	mS m^−1	2.45 ± 1.34			1.59 ± 0.24		0.90 ± 0.40		1.56 ± 1.17
T-C	g kg^−1	21.5 ± 10.3	b		19.3 ± 2.8	B	1.6 ± 0.6	a	2.1 ± 1.2	a
T-N	g kg^−1	1.0 ± 0.6	b		0.8 ± 0.0	ab	0.2 ± 0.1	a	0.2 ± 0.1	a
C/N ratio		23.7 ± 7.4	b		25.9 ± 4.2	B	9.8 ± 4.6	a	11.1 ± 3.0	a
CEC	cmolc kg^−1	6.19 ± 3.37	b		6.31 ± 2.06	B	1.14 ± 0.48	a	1.27 ± 0.38	a
Exch. Ca	cmolc kg^−1	0.84 ± 1.16			0.01 ± 0.00		0.09 ± 0.12		0.07 ± 0.07
Exch. Mg	cmolc kg^−1	0.07 ± 0.07			0.01 ± 0.00		0.03 ± 0.03		0.04 ± 0.05
Exch. K	cmolc kg^−1	0.02 ± 0.01	b		0.01 ± 0.00	ab	0.00 ± 0.01	a	0.00 ± 0.01	a
Avail. P	mg P kg^−1	30 ± 35			8 ± 3		24 ± 45		39 ± 44
Shoreline
		HG-Sm (n = 4)			MA-S-ib (n = 3)		MA-S-ft (n = 3)
0–10 cm
pH (H2O)		6.28 ± 1.55			5.43 ± 0.20		5.81 ± 0.30
pH(KCl)		5.56 ± 1.60			4.47 ± 0.19		5.21 ± 0.33
EC	mS m^−1	8.34 ± 4.65			1.47 ± 0.34		3.50 ± 1.26
T-C	g kg^−1	8.4 ± 2.5			4.4 ± 0.7		8.5 ± 5.6
T-N	g kg^−1	0.6 ± 0.2			0.4 ± 0.1		0.7 ± 0.5
C/N ratio		14.4 ± 2.0	b		10.9 ± 0.6	A	11.6 ± 0.7	ab
CEC	cmolc kg^−1	2.69 ± 0.76			1.51 ± 0.07		3.14 ± 1.57
Exch. Ca	cmolc kg^−1	2.31 ± 1.65			0.67 ± 0.26		1.83 ± 0.85
Exch. Mg	cmolc kg^−1	0.16 ± 0.04	ab		0.15 ± 0.05	A	0.29 ± 0.09	b
Exch. K	cmolc kg^−1	0.07 ± 0.03	b		0.01 ± 0.02	a	0.07 ± 0.04	b
Avail. P	mg P kg^−1	150 ± 35			46 ± 24		103 ± 88
20–30 cm
pH (H2O)		6.33 ± 1.67			5.40 ± 0.10		5.63 ± 0.55
pH(KCl)		5.20 ± 1.18			4.47 ± 0.06		4.57 ± 0.18
EC	mS m^−1	3.25 ± 4.47			1.07 ± 0.62		1.03 ± 0.08
T-C	g kg^−1	1.8 ± 1.5			1.6 ± 0.4		0.8 ± 0.5
T-N	g kg^−1	0.2 ± 0.1			0.2 ± 0.0		0.1 ± 0.1
C/N ratio		8.1 ± 2.7			7.9 ± 2.8		5.8 ± 1.9
CEC	cmolc kg^−1	0.96 ± 0.80			1.03 ± 0.03		1.18 ± 0.15
Exch. Ca	cmolc kg^−1	0.31 ± 0.17			0.22 ± 0.09		0.21 ± 0.12
Exch. Mg	cmolc kg^−1	0.03 ± 0.02	a		0.10 ± 0.03	b	0.09 ± 0.03	b
Exch. K	cmolc kg^−1	0.02 ± 0.01			0.00 ± 0.01		0.02 ± 0.02
Avail. P	mg P kg^−1	46 ± 55			14 ± 8		51 ± 17

HG-I and HG-S, homegardens in the inland and shoreline areas, respectively; MA-I and MA-S, the MARDI experimental field in the inland and shoreline areas, respectively. m: mango; ib: in-between trees; ft: points 1 m apart from trees with fertilizer application. Average ± standard deviation.

Exch., exchangeable; Avail., available.

Different letters indicate a significant difference among crops using Tukey’s multiple comparison (p < 0.05).

4. Discussion

4.1. Characteristics of the soils in the study sites

The soil profiles in SF-I-1 and SF-I-2 were composed of the O, A, and E horizons and their transitional horizons. The soils were acidic and very sandy in texture with low levels of exchangeable bases and available P, except relatively high levels of available P in the A1 and A2 horizons in SF-I-2. The latter might be due to the remaining effects of previous land use, although we could not obtain information due to obscure ownership. The soil profiles observed in the secondary forests were similar to those in the remnant forests, which were located adjacent to the MARDI experimental farm (Khairul et al. 2017). Although the spodic horizon could not be found, blackish or dark brownish soil clods were found under the groundwater table, suggesting the presence of the spodic horizon in these profiles. Assuming the spodic horizon underlies the E horizon, the soils are classified as Typic Haplohumods or Typic Haplorthods in the USDA classification systems (Soil Survey Staff 2014), which corresponds to the Jambu series of Spodosols in the Malaysian system (Paramananthan 1987). On the other hand, in the HG-S, soil profiles could not be described due to the space limitation. However, based on the observation of soil pits upon sampling, the soils in HG-S seemed to be similar to Typic Quartzipsamments as reported in our previous study (Khairul et al. 2017).

The soil profiles in HG-Iv-1 and HG-Iv-2 had similar horizon stratification to those in the corresponding secondary forests, except for the absence of the O horizon and the presence of the Ap horizon due to cultivation. These soils could be classified as the same soil class as those in the secondary forests. The important features of soils in these homegardens were darker soil color in terms of value and/or chroma and higher levels of T-C and T-N content throughout the profiles compared with those in the corresponding secondary forests. This result indicates the augmentation of organic matter pools of soils, as discussed below. It should be noted that the C/N ratio in the Ap horizon exceeded 30 and was rather high compared with the reported values ranging from 10 to 20 for homegardens in other regions (for example, Funakawa et al. 2007; Lattirasuvan et al. 2010). Taking into consideration a high C/N ratio in charcoal (Cantrell et al. 2012), organic matter from charcoal may contribute significantly to soil organic matter pools in the BRIS homegardens.

High positive correlations were found between T-C and CEC in the 0–10 cm and 20–30 cm soils (Fig. 4). Correlations were similarly found when data were analyzed in terms of each crop or the samples collected from the soil profiles (data not shown). In contrast, correlations were not found between the clay content and CEC values. Therefore, soil organic matter could be the determinant factor for CEC of the soils in the BRIS homegardens. The amount of exchangeable bases was much lower than the CEC values, indicating that development of negative charges of soil organic matter would be limited under acidic soil conditions. These findings coincided with those from the MARDI field and remnant forests in our previous study (Khairul et al. 2017).

Figure 4. Relationship between T-C and CEC of HG-I and HG-S at two different depths. Left: 0–10 cm, right: 20–30 cm. n = 69 for each depth. **, p < 0.01.

4.2. Influence of the age and size of homegardens on the soil chemical properties

Terra Preta is well known as being very sandy but fertile anthropogenic soil and was reported to contain soil organic matter up to 15% due to continuous charcoal and black carbon application (Glaser et al. 2000). However, this soil is believed to be a relic of human activities since the pre-Columbian era. Meanwhile, the maximum age of HG-I after establishment was 45 years old. The levels of T-C, T-N, and CEC at depth 0–10 cm significantly increased in HG-Im (Fig. 2) and tended to increase in HG-Ic and HG-Iv with homegarden age. In addition, HG-Iv-1 and HG-Iv-2 showed higher soil organic matter than the corresponding SF-I-1 and SF-I-2 as mentioned above. These results indicated that routine application of organic matter such as manure and charcoal contributes to increasing organic matter in the surface soils, which agrees with previous reports (Torquebiau 1992; Jensen 1993a, 1993b). It is also suggested that the buildup of soil organic matter occurs in a relatively short duration of several decades. For HG-S, no clear tendencies were found in soil organic matter with homegarden ages.

In the Terra Preta, Glaser et al. (2000) found that black carbon was particularly enriched at depth 30–40 cm, probably due to the perturbation and coverage process by earthworm and termite activities. In our study, few earthworms and termites and their channels and nests were observed during the field survey except for the surface soils. Taking into consideration the smooth horizon boundaries, another possible reason for the accumulation of soil organic matter down to the deep layers in HG-Iv (Table 1) is the downward movement of organic matter applied to the soil surface through large pores in the sandy soils resulting from water permeation. Furthermore, the sudden decrease in value and chroma and increase in T-C observed at the AE2 to AE4 compared with the overlying AE1 horizon in HG-Iv-2 (Tables 1 and 2) might suggest that the organic matter moved down, resulting in secondary formation of these horizons after the start of the land use as homegardens.

In addition to T-C, T-N, and CEC, the content of available P at 0–10 cm of HG-Im and HG-Iv also correlated with homegarden ages (Fig. 3), whereas no correlations were found in exchangeable bases (data not shown). This finding could be ascribed to the cumulative effects of P fertilizer application in the surface soils and low solubility of P.

On the other hand, Saha et al. (2009) in India reported that smaller homegardens with higher tree densities showed higher soil carbon stock, suggesting that the amount of leaf litter supply per unit area is an important factor in the buildup of soil organic matter. However, our study could not find any relationship between soil parameters and the size of homegardens.

Large variations in T-C, T-N, CEC, and available P were found as shown in Tables 4 and 5, compared with relatively small standard deviations in the MARDI farm soils (Table 6). The reason for the variations in these properties could be ascribed mainly to the homegarden ages as mentioned above. Meanwhile, large variations in exchangeable bases seem to result mainly from different fertilizing practices among households, including the type, amount, and frequency of fertilizer application. Despite such large variations in soil parameters, the following can be proposed based on the obtained data and information.

4.3. Influence of different locations of homegardens on the soil properties

The different occupations of farmers and fishermen in the inland-ward and shoreline areas, respectively, seem to result in several differences in management practices of homegardens. The most important difference in terms of fertilizing soils is the use of manure in the inland-ward area and fish waste in the shoreline area. In the inland-ward area, the households raise poultry, produce manure from the dung, and apply it to their homegardens, or otherwise purchase manure from their neighbors. The households in the shoreline area do not rely on manure but apply waste from fish processing for fertilization.

The properties of the soil organic matter (T-C, T-N, and CEC) in HG-I were superior to those in HG-S, irrespective of crop type (Table 3), which might be ascribable to the difference in management practices: Manure application could lead to a higher level of soil organic matter in HG-I than HG-S. Meanwhile, the influence of seafood waste application in the shoreline area was reflected in a higher level of available P in HG-Sm and HG-Sv than HG-Im and HG-Iv (Table 3). The opposite trend for HG-Sc and HG-Ic could be explained by the location of the HG-Sc outside the homegarden without the application of fish waste.

The soils of one HG-I and one HG-S showed extremely high values of exchangeable Ca (9.8 to 22.8 cmol_c kg⁻¹) and available P (890 to 1640 mg P kg⁻¹), so we omitted the data from the statistical analysis. Unique management practices were employed in these homegardens. The households in HG-I have about 10 goats and produce vast amounts of manure from the dung in the hut, which is frequently applied as manure to the homegardens including the fruit trees. On the other hand, the households in HG-S manage a relatively large-scale fish processing factory and apply the fish waste directly to the base of the coconut and mango trees because of their belief that the waste helps the trees to grow well.

4.4. Influence of the different crops in the homegardens on the soil properties

In the inland-ward area, the levels of T-C, T-N, and CEC were similar between different crops (Table 3 and 4). Taking into consideration the higher frequencies of manure application to vegetables in HG-Iv, it could be suggested that organic matter supplied from leaf litter from the mango trees at HG-Im and the dense coconut root systems at HG-Ic might compensate for less input of manure to soils to sustain the level of soil organic matter. Meanwhile, the level of available P was significantly higher in HG-Iv than HG-Im, and similar tendency was also found in exchangeable Ca. These results could be ascribed to frequent application of fertilizer and manure for vegetables. In Indonesia, Funakawa et al. (2007) compared soil properties between homegardens, which contained predominantly tree species and annual crops, both without chemical fertilizer application, and found that the former was higher in soil organic matter because of the supply of litterfall, whereas the latter was higher in available P because of manure application.

Meanwhile, in the shoreline area, HG-Sc showed relatively low fertility levels compared with HG-Sm and HG-Sv. The reason for this might be the location of HG-Sc outside the fences, in which the land was left almost bare with some undergrowing grass as well as little application of chemical fertilizer. It has been reported that insufficient cover or removal of vegetation cover for the ground surface of sandy Spodosols causes an increase in temperature and markedly accelerates the decomposition of soil organic matter (Smethurst and Nambiar 1990; Kendawang et al. 2004; Khairul et al. 2017).

4.5. Soil fertility of the homegardens on the BRIS

In the inland-ward area, the levels of T-C, T-N, and CEC in HG-Im were similar to those in SF-I and superior to those in MA-I-ib and MA-I-ft, where standardized management was carried out (Table 6). Furthermore, comparison of the soil profiles between homegardens and the adjacent secondary forests in the inland-ward area revealed higher levels of soil organic matter in the former (Table 2). Meanwhile, the levels of T-C, T-N, and CEC in HG-Sm were similar to those in MA-S-ib and MA-S-ft. The primary difference in management practices of soil organic matter was the combined application of charcoal and manure in HG-I, suggesting the importance of that combined application for sustaining soil organic matter in the use of the BRIS for homegardens. Many previous studies have reported favorable soil conditions in homegardens in terms of soil organic matter compared with crop fields or grassland (Gajaseni and Gajaseni 1999; Funakawa et al. 2007; Lattirasuvan et al. 2010; Tanaka et al. 2010, 2012). However, because the combined effects of charcoal and manure application have not been clearly verified in these studies and our study, further investigation is required on the different sources of organic matter in the BRIS soils.

Meanwhile, the levels of exchangeable Ca, Mg, and K in HG-I tended to be higher than, and the levels in HG-S were similar to those in the MARDI farm (Table 6), suggesting that improved CEC with increasing soil organic matter in HG-I could enhance the retention of basic cations supplied as chemical fertilizer, ash, and manure.

Many studies have reported that nutrients derived only from manure application can sustain soil nutrient levels in traditional tropical homegardens without chemical fertilizer input (Torquebiau 1992; Jensen 1993a, 1993b; Gajaseni and Gajaseni 1999; Kumar and Nair 2004), and the homegardens are regarded as sustainable agricultural systems in terms of closed nutrient dynamics similar to tropical forests (Kumar and Nair 2004). In contrast, our study revealed that application of chemical fertilizer in addition to the input of manure, ash, and charcoal contributes in maintaining the sandy BRIS soil fertility. Relatively high levels of nutrients, in particular available P were detected even below the AE in HG-Iv-1 and AE3 in HG-Iv-2 (Table 2), where no plant root was observed (Table 1). Though the content of available P in HG-Iv-2 was gradually decreased toward the lower horizon, the contents were relatively high compared with lower horizon of SF-I-2. It is suggested certain amounts of nutrients moved down beyond plant rooting depth. In this sense, the BRIS homegardens could not be regarded as a closed system compared with those in the other tropical regions. More detailed studies are required to clarify organic matter and nutrient budgets in order to elucidate the mechanisms of homegarden managements which allow to sustain soil organic matter and the concomitant fertility levels under the BRIS condition.

Acknowledgments

This research was financially supported by a Sasakawa Scientific Research Grant (No. 26-531). The first author is staff of Faculty of Agriculture, Universiti Putra Malaysia.

Additional information

Funding

This research was financially supported by a Sasakawa Scientific Research Grant (No. 26-531).