INTRODUCTION
The island of Sulawesi accounts for 45% of Indonesian cocoa (Theobroma cacao L.) production, mainly on smallholdings of 1–5 ha. After a boom in production in the early 2000s, national production of 535,000 tonnes in 2010 had declined to 420,000 tonnes by 2013 (http://www.icco.org/about-us/icco-annual-report.html), largely, as a result of reduced productivity of smallholdings in Sulawesi. This is likely to have been due to reduction of natural soil fertility, increased pest and disease impacts and inadequate cocoa management (Panlibuton and Lusby, Reference Panlibuton and Lusby2006; Ruf and Yoddang, Reference Ruf, Yoddang, Gerard and Ruf2001, Reference Ruf and Yoddang2004). The most serious pests and diseases of cocoa in Sulawesi and other regions of Indonesia are cocoa pod borer (CPB, Conopomorpha cramerella), Phytophthora pod rot (PPR, black pod, caused by Phytophthora palmivora) and vascular streak dieback (VSD, caused by Ceratobasidium theobromae). Helopeltis species and stem borers such as Zeuzera can also inflict considerable damage. Availability of labour has decreased due to emigration of young people to urban centres leading to an ageing of the cocoa farming communities. In many cases, farmers have replaced cocoa with other crops that are less labour-intensive, such as oil palm in areas with year-round rainfall, or maize or cloves where rainfall is more seasonal.
Participatory demonstrations on cocoa farms in Papua New Guinea and Indonesia have shown that the implementation of basic cultural management such as pruning and sanitation on existing farms can double the yields (Daniel et al., Reference Daniel, Konam, Saul-Maora, Kamuso, Namaliu, Vano, Wenani, N'nelau, Palinrungi and Guest2011). However, cocoa production on most soils needs an improved regime of nutrient input if it is to be sustainable. As a tree crop, regular fallow periods are not feasible for cocoa and, therefore, soil nutrients need to be replenished under the growing crop. In many areas of Sulawesi, continuous cropping with cocoa and other crops decreases the fertility previously accumulated under forest (Hartemink, Reference Hartemink2005; Ruf and Yoddang, Reference Ruf, Yoddang, Gerard and Ruf2001). Soil pH may be lower under cocoa than in fallow land (Adejuwon and Ekanade, Reference Adejuwan and Ekanade1988), reducing availability of basic cations and phosphorus, and increasing availability of toxic aluminium (Al) and manganese (Mn) cations (Hue, Reference Hue2011). Cocoa beans and harvested pods have a high content of potassium and other macronutrients and remove more K from the soil than coffee (www.fao.org/docrep/008/y7063e/y7063e.htm).
The soils in many Sulawesi cocoa smallholdings are particularly low in soil organic matter (SOM) and nitrogen. Soil C contents >3.5% and soil N contents >0.2% are recommended for cocoa production (Hardy, Reference Hardy1958; Murray, Reference Murray1967), although soil C content levels of 2% are considered adequate for cocoa production in Sulawesi. However, on two cocoa farms in different districts in Sulawesi, soil organic C contents were found to be 1.45 and 1.83%, while soil N was 0.14% at both sites (McMahon et al., Reference McMahon, Bin Purung, Lambert, Mulia, Nurlaila, Susilo, Sulistyowati, Sukamto, Israel, Saftar, Amir, Purwantara, Iswanto, Guest and Keane2015). Similarly, Agoume and Birang (Reference Agoume and Birang2009) reported a soil organic C content of 1.35% under cocoa. Higher rates of oxidation of organic matter following cultivation of tropical soils result in more rapid losses of organic C than in temperate soils. SOM is critical for normal growth and production, particularly in weathered tropical soils (Steiner et al., Reference Steiner, Teixeira, Lehmann, Nehls, De Macedo, Blum and Zech2007; Tiessen et al., Reference Tiessen, Cuevas and Chacon1994), providing a crucial reservoir of macronutrients, such as N, P and S. Many weathered tropical soils have low activity charges, due to either low clay content or clay compositions characterised by low activity (Steiner et al., Reference Steiner, Teixeira, Lehmann, Nehls, De Macedo, Blum and Zech2007), and therefore a low nutrient binding capacity. In such soils, SOM plays a critical role in nutrient availability by providing exchange sites and immobilisation by microorganisms, therefore, preventing rapid leaching and improving availability of nutrients to plants by gradual release in the rhizosphere (Burger and Jackson, Reference Burger and Jackson2003). An active and diverse soil microbial community directly and indirectly stimulates plant growth and reduces the severity of pests and disease (McMahon, Reference McMahon and Cumagun2012). SOM includes several functional pools defined by the turnover rates of C, including leaf litter (an active, labile pool of microbial biomass, soluble carbohydrates, proteins and extra-cellular enzymes), a less active pool (slow pool) consisting of particulate organic matter and a stable pool that includes various humates and fulvates (humus) (Woomer et al., Reference Woomer, Martin, Albrecht, Resck, Scharpenseel, Woomer and Swift1994).
Better nutrition linked to appropriate tree management would enable productivity to increase on existing cocoa farms, and reduce pressure to expand into forested areas. Cocoa farmers in Indonesia generally use NPK fertiliser and/or urea formulated for rice production and purchased with a government subsidy provided for food crops. Inorganic fertiliser products with more appropriate formulations for cocoa are not always available, or are too expensive for farmers. However, under current farmer practice, cocoa nutrition could be improved by organic amendment using locally available materials. In a field trial with young cocoa trees on a marginal soil in the main cocoa-growing area of South Sulawesi, we investigated the use of an organic amendment in conjunction with low cost and readily available inorganic fertilisers.
MATERIALS AND METHODS
Establishment of the nutrient amendment trial
The trial was established in Bone Bone sub-district, North Luwu District, South Sulawesi Province, Indonesia on land managed as a Government of Indonesia Assessment Institute for Agricultural Technology (AIAT, or BPTP, Balai Pengkajang Teknologi Pertanian) field station (2°36′21″S; 120°36′44.4E). The site was slightly elevated (112 m asl) adjacent to the coastal plain. Annual rainfall in Bone Bone was 3183 mm (2011), 2891 mm (2012), 2119 mm (2013), 2448 mm (2014) and 2311 mm (2015) (Mars, Inc., Bone Bone weather station). The site had previously grown moribund, unproductive cocoa trees under tall coconut shade. The soils at the site have low N and organic C contents and a low base saturation (BS). The yellow colouring of the clay in the lower horizon, a coarse upper horizon and a BS lower than 50% suggests that this soil belongs to a class of sandy, yellow podzols known as acrisols. Coconut palms displayed yellow fronds, and nearby cocoa showed interveinal necrosis typical of calcium deficiency, and was affected by CPB and PPR (black pod), while VSD incidence was very low.
Gliricidia sepium shade trees were planted on the site to provide even shade under the coconuts. Planting holes (60 cm3) at a 3 × 3 m2 spacing were prepared for cocoa seedlings. Five-month-old nursery-raised cocoa clones (budded seedlings) of the Malaysian clone PBC123 (Chong and Shepherd, Reference Chong, Shepherd, Pushparajah and Soon1986), known locally as Sulawesi 1, were provided by a commercial cocoa nursery in Masamba, North Luwu, and planted at the site in late December 2011. At the time of planting, mineral fertiliser, 100 g NPK (Ponska) and 150 g triple superphosphate (36%), was added to each tree in equal amounts to provide adequate and uniform nutrient conditions for establishment of all plants in the first few months after planting out. A randomised block design consisting of four blocks and treatment plots of 16 trees per plot was applied.
Soil treatments
In May 2012, after the plants were well established, eight treatments were applied to the 16-tree plots in each block in a complete factorial design of the three treatment factors (organic compost, inorganic fertiliser and dolomite) (Table 1). Compost was prepared with locally available cow manure (60%), empty oil palm bunches (15%), rice straw (10%), various leaves, including banana, grass, Gliricidia and maize (10%) and cocoa pod husks (5%). A decomposition-promoting mix of microbes (EM4) was added with water to the compost mix, ensuring that it was distributed evenly to the bottom of the compost box, thus reducing the need for mixing. Dolomite contained 18–22% MgO according to company specifications. The inorganic fertiliser was NPK (Ponska, 15% N, 15% P2O5 and 15% K2O with traces of S) and urea (see Table 1 for amounts supplied in each treatment). Two samples (each approximately 0.5 kg) were collected from different batches of the compost typically used in the trial and analysed (Table 2).
Experimental treatments in the North Luwu trial.
Soil amendments (left-hand columns) were applied twice per year to provide total quantities per tree each year as follows: 374 g NPK (Ponska) and 250 g urea (mineral fertiliser), 5 kg dolomite and 10 kg compost. Combinations (treatments E–H) were additive. The columns on the right show total quantities (g) of macronutrients (as elements) provided per tree each year in each treatment.
Macro- and micro-nutrient concentrations and other properties (mean of two samples) of the compost used in the North Luwu trial.
Organic C was determined using the Walkey and Black method, total N by the Kjeldahl method, S and P by spectrophotometric methods and other nutrients by AAS.
The soil amendments were applied twice per year, the first treatment being in May 2012. Combined treatments were additive (see Table 1). The treatments (all in solid form) were mixed into the topsoil at the periphery of the lateral root zone of each tree in a 16-tree plot (0.5–1 m from the tree base). Otherwise, all trees were managed using normal farm practice, including pruning and removal of weeds by hand within the drip zone of the cocoa trees.
Soil and leaf analyses
In 2011, baseline soil samples were collected from the trial site and adjacent land under coconuts (three samples per site). Each sample was prepared by combining four sub-samples of topsoil collected from points intermediate between trees.
Following soil amendment, leaves were collected in February 2013 from plots representing each treatment and the leaf nutrient contents were evaluated. In June 2014, soil and leaf samples were collected from treated plots and evaluated. Four soil sub-samples were collected on a rain-free day from the outer edge of the root zone (0.–1 m from the tree base) in the centre of one of the four replicate plots for each treatment. The samples were collected between 1 and 8 cm below the soil surface by first removing approximately 1 cm loose soil and other surface material from an area of 6–8 cm2 and then using a trowel to collect soil to a depth of 5–7 cm below the scraped surface, a part of the topsoil accessed by lateral roots. The four subsamples were combined and mixed, providing a composite sample of about 1 kg per treatment. The eight soil samples were stored with labels in double plastic bags (with the label placed in the outer bag), which were sealed to retain moisture. In the same plots from which soil samples were taken, 3–4 fully expanded mature leaves were collected from the middle of selected branches (from a position about 3–4 leaves below the youngest expanded leaf at the branch tip) on each of the four central trees. These were stored in paper bags.
Soil and leaf samples were analysed by two accredited soil laboratories: leaf samples in 2013 by the AIAT Soil Laboratories, Maros, South Sulawesi and soil and leaf samples in 2014 by the Indonesian Coffee and Cocoa Research Institute, Jember, East Java. To verify the soil analyses, in 2011 duplicate samples were analysed by the two laboratories, and in 2014 triplicate samples were analysed, the third sample by Gajah Mada University Soil Laboratory, Yogjakarta, Central Java.
Growth
Tree height (cm) was determined at the time the treatments commenced in May 2012, and then in January 2013 and each following month until January 2014 by which time the rate of increase in height had slowed.
Productivity (yield) and bean quality
Pods were first produced in 2013, about 18 months after planting, and the first harvest was in August 2013. Annual yields were determined from all 16 trees per plot from harvests conducted twice per month between 2014 and 2015. Pods harvested in each plot were pooled and counted. Beans were extracted from harvested ripe pods, weighed and then sun-dried to determine their dry weight. Bean quality characteristics were analysed in the Mars, Inc., laboratories in Makassar, South Sulawesi. Beans were harvested and dried, and 300–400 g samples obtained from each treatment were used to determine quality parameters, including fat and shell content, proportion of waste material and bean count, as described previously (McMahon et al., Reference McMahon, Bin Purung, Lambert, Mulia, Nurlaila, Susilo, Sulistyowati, Sukamto, Israel, Saftar, Amir, Purwantara, Iswanto, Guest and Keane2015).
Flowering and incidence of pest/diseases
The incidence and severity of CPB and incidence of PPR (black pod) were determined between July 2013 and September 2015, as previously described (McMahon et al., Reference McMahon, Bin Purung, Lambert, Mulia, Nurlaila, Susilo, Sulistyowati, Sukamto, Israel, Saftar, Amir, Purwantara, Iswanto, Guest and Keane2015). Incidence of VSD was negligible in 2012 but increased slightly during the period of the trial and in 2015, trees were scored for VSD severity (light, medium or high) each month between February and September 2015. At the same time, flowering was scored as light, medium or high (McMahon et al., Reference McMahon, Bin Purung, Lambert, Mulia, Nurlaila, Susilo, Sulistyowati, Sukamto, Israel, Saftar, Amir, Purwantara, Iswanto, Guest and Keane2015), and the number of trees in each category was counted. The cumulative proportion (%) of trees with each of the flowering scores was then calculated.
Statistical analyses
Harvest yields were assessed in two ways to either exclude or include tree mortality. First, yields of dry beans per plot were adjusted to yield per tree (to account for missing trees) and the mean yield and standard error of the mean of the four replicates per treatment were obtained. The number of pods produced per tree was calculated by dividing pod harvest numbers per plot by the number of surviving trees. An estimate of yield per hectare was obtained by extrapolating per tree data to 1000 trees (the approximate number of cocoa trees per hectare in Sulawesi). Second, yield data were extrapolated directly from plot data (originally 16 trees per plot) to incorporate tree mortality and give the actual yield per hectare. The null hypothesis that yield was unaffected by the soil treatments was tested. Means and SEM for yield, pod numbers and pest/disease incidence (%) were obtained from individual plot measurements and analysed by one-way ANOVA. Pearson's correlation test was applied when appropriate. To determine whether treatments had significant effects on yield or pod numbers, data were subjected to one-way ANOVA. The proportion of pods infested in each plot was transformed to its arcsine value. The transformed data were subjected to one-way ANOVA (IBM SPSS Statistics Ver. 19) to compare treatment effects. Where the Levene test demonstrated homogeneity in the raw or transformed data, Tukey's post-hoc test was applied to separate means; otherwise, the Games-Howell test was used.
RESULTS
Tree growth and flowering
Trees supplied with compost, alone (treatment C) or in combination with mineral fertiliser (E), dolomite (G) or dolomite plus mineral fertiliser (H) grew taller than the trees in the control treatment (A) and those supplied with mineral fertiliser (B) or dolomite alone (D) (Figure 1a). The foliage of the trees treated with compost generally appeared darker green than in untreated trees or those supplied with inorganic amendments. Leaf growth on trees with no soil amendments (A) or receiving only inorganic amendments (B, D and F) was sparser and showed evidence of interveinal necrosis on leaves, typical of calcium (and/or magnesium) deficiency (Wessel, Reference Wessel, Wood and Lass1985). Such deficiency symptoms were not seen in the plots amended with compost. Furthermore, many trees died in control plots and in those receiving only inorganic amendments (Supplementary Table S1, available online at https://doi.org/10.1017/S0014479717000527), and in one plot receiving only mineral fertiliser (B) all 16 trees died, indicating a possible toxic or extreme nutrient deficiency effect. Tree deaths were greater in the mineral fertiliser treatment (B) than in all other treatments. There were no tree deaths in plots amended with compost.
Effect of soil amendments on (a) tree height (cm) in May 2012 (light bars) when soil amendment treatments were commenced and January 2014 (dark bars) and (b) yield of dry beans (determined from 2014 to 2015) for all planted trees in which no allowance was made for tree death (light bars) or for surviving trees only (dark bars). All data shown are the means of four replicates. Soil amendments were as follows: A, control; B, mineral; C, compost; D, dolomite; E, mineral/compost; F, mineral/dolomite; G, compost/dolomite; H, all amendments. Means with the same letter are not significantly different (P = 0.05).
The treatments that included compost greatly increased the proportion of trees with a high flowering score, while the mineral fertiliser and dolomite treatments (alone [B, D] or in combination [F]) had no effect (P < 0.05) (Figure S1). The most prolific flowering followed treatment with compost (C), compost plus dolomite (G) and compost plus dolomite plus mineral fertiliser (H). The addition of inorganic fertiliser reduced the benefits of compost (E), while dolomite appeared to neutralise the negative effects of mineral fertiliser (F).
Yield and bean quality
All treatments that included compost yielded significantly (P < 0.01) more dry beans than the control or treatments with mineral fertiliser and dolomite (alone or in combination), whether dead trees were excluded from the extrapolation or not (Figure 1b). Yields in the plots treated with a combination of compost with mineral fertiliser, dolomite or mineral fertiliser and dolomite (treatments E, G or H) were not significantly greater than those treated with compost alone (treatment C), although there was evidence of an increase in yield when compost was combined with other nutrient sources.
The relatively high bean count (number of beans per 100 g dry beans) generally found in the PBC123 clone (McMahon et al., Reference McMahon, Bin Purung, Lambert, Mulia, Nurlaila, Susilo, Sulistyowati, Sukamto, Israel, Saftar, Amir, Purwantara, Iswanto, Guest and Keane2015) is evident in Table S2. However, compost treatments C and H appeared to decrease the proportion of flat beans and the bean count, indicating that these treatments caused an increase in average bean size. The proportion of waste and shell content was also lowered by compost amendment alone (C). Fat content was unaffected by the soil amendments.
Soil properties and leaf nutrient contents
Lateral roots of cocoa, responsible for the bulk of nutrient and water uptake, occur in the soil to a depth of 20 cm (Hartemink, Reference Hartemink2005; Toxopeus, Reference Toxopeus, Wood and Lass1985) and analysis of properties of soil collected to this depth is generally recommended. However, in this study, composite soil samples were collected from within the top 15 cm of soil critical for cocoa nutrition, where 80% of cocoa lateral roots are concentrated (de Geus, Reference de Geus1973), providing data for comparison between treatments. Baseline soil properties determined on the trial site in 2011, prior to establishing the trial, and on adjacent land under tall coconut indicated that the soils of both sites were marginal, with low C (<1.7%), N (<0.18%) and pH (4.7) (Table 3). Average clay content was low at both sites (13%), while sand content was relatively high (>49%). Exchangeable Al was detected at up to 2.28 cmol kg−1 in the unamended soil at both the trial site and under adjacent coconuts (Tables 3 and 4). The soil data from the controls (Table 4) confirm the baseline data in Table 3, particularly the critically low organic C and N contents and low soil pH. Although the applied compost had a relatively high pH (approx. 7.5), soil pH readings in the compost-amended soil were no higher than in control soils indicating that the amounts of compost applied were not enough to affect soil pH. Soil pH was slightly higher (>5.0) in the plots amended with dolomite (Table 4). Cation exchange capacity was low (<12) at the trial site, and Mg was critically low. The high sand content and low cation exchange capacity are indicative of a heavily leached soil, which is likely given the high annual rainfall in North Luwu (see Materials and methods). Concentrations of soil P and K were adequate (Tables 3 and 4). Soil data from the trial site (Table 4) and leaf contents (Table 5) indicate that the levels of available soil S were adequate. Although data on the adequate levels of soil S for cocoa production are not available, leaf S contents (Table 5) were considerably higher than those reported in cocoa previously (Fahmy, Reference Fahmy1977; Nelson et al., Reference Nelson, Webb, Berthelsen, Curry, Yinil and Fidelis2011), suggesting that the supply of this element in the soil was not limiting. Soil concentrations of Fe and Mn were adequate or high (Table 4).
Soil nutrient contents and other soil properties in topsoil (0–5 cm depth) at the North Luwu trial site and adjacent land under coconuts in 2011 (prior to establishing the trial).
Soil analyses were conducted by the Indonesian Coffee and Cocoa Research Institute.
*After Fahmy (Reference Fahmy1977).
‡Ammonium acetate (pH 4.8) extraction.
†CEC determined at the AIAT Soil Laboratories, Maros.
Effect of soil amendments on soil nutrient and Al concentrations, cation exchange capacity (CEC), base saturation (BS) and pH in the trial at Bone Bone 2 years after treatments A–H were first applied.
Soil treatments were applied twice per year beginning in May 2012. Soil properties were determined in the soil laboratory of the Indonesian Coffee and Cocoa Research Institute, East Java (June 2014). Soil treatments: A, control; B, mineral; C, compost; D, dolomite; E, mineral/compost; F; mineral/dolomite; G, compost/dolomite; H, all amendments. Data are based on one measurement from pooled soil samples.
*pH determined by AIAT Soil Laboratories, Maros.
†Walkey and Black method.
‡Kjeldahl method.
§25% HCl extraction.
¶Bray-I method.
**Ammonium acetate (pH 4.8) extraction.
††Ammonium acetate (pH 7) extraction.
‡‡KCl (1 N) extraction.
§§Salt extraction.
¶¶HCl (0.1 N) extraction.
Exch., exchangeable; cmol kg−1 = mg 100 g−1.
Leaf nutrient contents of macro- and micro-nutrients (in italics) in cocoa trees in North Luwu in February 2013 (I) and June 2014 (II).
Soil amendments were first applied in May 2012: A, control; B, mineral; C, compost; D, dolomite; E, mineral/compost; F; mineral/dolomite; G, compost/dolomite; H, all amendments. Right-hand columns: estimates of normal leaf nutrient concentrations according to Murray (Reference Murray1967) and Fahmy (Reference Fahmy1977).
*Adapted by Nelson et al. (Reference Nelson, Webb, Berthelsen, Curry, Yinil and Fidelis2011).
nd, not determined.
The high average Ca concentration of the applied compost (Table 2) may partly account for an increased soil BS, as well as leaf Ca content, two years after organic amendment commenced, but adding dolomite alone had no such effect (Table 4). Amendments had little effect on soil N, which was presumably rapidly taken up by the plants, or otherwise leached and/or volatilised. Uptake rates of N were apparently unaffected by the soil amendments since leaf concentrations were similar in all treatments, regardless of vigour (Table 5). Soil concentration of S, which also occurs mainly in anionic form, was apparently not influenced by the soil treatments.
Amendment with compost, compost plus dolomite or compost plus dolomite plus mineral fertiliser increased soil C content slightly (Table 4), but levels remained well below estimated adequate levels recommended for cocoa (Table 3). Soil P and K concentrations, determined either by Bray-1 or acid extraction, were increased by treatment with mineral fertiliser. However, tissue concentrations of P or K were maintained within the normal range in these trees (Table 5) and were mainly unaffected by the soil treatments. Leaf concentrations of K only became subnormal in the larger trees, presumably due to the low exchangeable K available in the soil, which became more limiting in the larger and higher-yielding trees. The growth-limited trees in the control and inorganic treatments apparently took up and redistributed sufficient quantities of K, which is highly mobile within plants, to maintain normal leaf concentrations. Similarly, P leaf concentrations were only slightly below normal levels in the larger trees; these trees also had higher yields, and, therefore, demand for P apparently exceeded its rate of uptake from the soil. Reduced availability of P is likely to have been an effect of low soil pH, since the P soil concentrations determined by the Bray-1 method were adequate (Tables 3 and 4).
Micronutrient tissue concentrations were consistent with those determined in soil and cocoa leaf tissue in cocoa growing areas of PNG by Nelson et al. (Reference Nelson, Webb, Berthelsen, Curry, Yinil and Fidelis2011). Leaf concentrations of Fe and Mn were high in all of the soil treatments (Table 5). However, soil and leaf concentrations of Fe and Mn were not affected greatly by soil treatments (Tables 3 and 5) and no correlation was found between soil and leaf concentrations (not shown). Previous assessments of normal leaf concentrations of Fe in cocoa vary: while Fahmy (Reference Fahmy1977) reported 50 mg kg−1 to be sufficient for normal growth, other reports suggest leaf concentrations up to 175 mg kg−1 are required (Nelson et al., Reference Nelson, Webb, Berthelsen, Curry, Yinil and Fidelis2011).
These analyses indicate that the soil amendments did not have a large impact on some critical soil properties, particularly N and C content. However, based on the criteria of Murray (Reference Murray1967) and Fahmy (Reference Fahmy1977), leaf concentrations of Mg were deficient in all of the treatments, while Ca leaf concentrations were deficient in the unamended control and mineral fertiliser treatment. Concentrations of Mg in leaf samples were extremely low in the control and mineral fertiliser treatments. Adding dolomite or compost apparently alleviated the factor(s) suppressing uptake of Ca and Mg, resulting in higher leaf concentrations of these nutrients (Table 5). Nevertheless, leaf contents of Ca and Mg in trees treated with compost alone (1.2 and 0.3%, respectively) were higher than those in the dolomite-treated trees (0.8 and 0.2%, respectively): this might partly account for the higher growth and productivity of the compost-treated, compared to the dolomite-treated trees. All treatments that included mineral fertiliser had elevated exchangeable Al levels (2.69–3.32 cmol kg−1), while those treated with compost, dolomite or compost plus dolomite had levels (2.20–2.29 cmol kg−1) similar to those in the unamended control (2.28 cmol kg−1) (Table 4).
Nutrient ratios
A low leaf N/P ratio in the control trees was only slightly increased by amendment with compost (treatment C), suggesting N uptake rate was not the most limiting factor at this site. All treatments with dolomite increased the N/P ratio. In all treatments, the soil exchangeable (Ca+Mg)/K ratios were much lower than the ratio of 25 or more recommended by Hardy (Reference Hardy1958), although treatment with compost (C) or dolomite and compost (G) greatly increased the ratio. The leaf Ca/K ratio was low in the control and inorganic fertiliser treatments but was apparently increased by amendment with compost or compost plus dolomite (Table 6). In the control soil (treatment A), Ca/K and Ca/Mg ratios were 4.3 and 3.1, respectively, compared to 7.4 and 4.5 in the compost-amended soil (treatment C) and 1.7 and 1.5 in the mineral fertiliser amended soil (treatment B). In the latter treatment (B), the extremely low ratio of Ca to other exchangeable cations is likely to have had a suppressive effect on Ca uptake by plants. Supplying dolomite (treatment D) alone did not affect soil ratios of exchangeable cations but resulted in higher leaf Ca/K and Mg/K ratios compared to the control treatment.
Ratios of exchangeable soil cations and leaf macronutrients (determined in June 2014) in the soil amendment trial at North Luwu.
Supplying dolomite with compost (treatment G) appeared to have a synergistic effect, as both soil and leaf Ca/K and Mg/K ratios increased greatly under this treatment; the lower Ca/Mg ratio in this treatment might be due to the high Mg content of dolomite. It is likely that Mg was limiting to growth and yield, since low leaf Mg/K ratios (observed in control plants) were lowered even further as a result of mineral fertiliser supply, whether supplied alone (treatment B) or in combination with dolomite (treatment F), while the leaf Mg/K ratio was increased by compost supply (Table 6). For the eight soil treatments, productivity (mean dry bean yield) was highly correlated with leaf Mg concentration (Pearson's r = 0.90**) and Mg/K ratio (Pearson's r = 0.94**), but not with leaf Ca concentration (r = 0.40) or Ca/K ratio (r = 0.41). Since the trees with higher Mg/K ratios were also the highest yielding and healthiest trees in the trial, with no obvious deficiency symptoms, it appears that organic amendment resulted in selectively enhanced Mg uptake by the cocoa plants.
Incidence and severity of pest/diseases
The average incidence of CPB in ripe pods determined between July 2013 and September 2015 was high, varying from 35 to 65% (Figure 2a). Preliminary data collected on CPB incidence indicates that a significantly lower rate of infestation occurred in the trees treated with mineral fertiliser, probably due to a sparse canopy in these trees. CPB incidence in the trees supplied only with compost (treatment C) was not different from control. Furthermore, incidence of severely, moderately or lightly infested pods was similar between treatments. PPR incidence for the two-year cumulative harvest varied from approximately 10 to 16% in ripe pods (Figure 2b); however, no significant differences between treatments were detected. VSD incidence was low with no severe VSD recorded (Figure S2). At the time of commencement of the trial, VSD was virtually absent but had increased two years later. This may be due to the increased population and growth of the cocoa, enabling VSD infections to become established.
Incidence of low, moderate and severe (a) CPB and (b) PPR in pods harvested from cocoa trees in the soil amendment trial, North Luwu. The treatments included control (no amendment) (A), mineral fertiliser (B), compost (C) and dolomite (D) and combinations of these: mineral/compost (E), mineral/dolomite (F), compost/dolomite (G) and all three (H).The data shown are cumulative means (and SEM) of four replicates for harvests (conducted twice per month) from July 2013 to September 2015. For total CPB incidence, the same letter indicates that no significant difference (P = 0.05) was detected between the treatments, while total PPR incidence did not significantly differ between treatments.
DISCUSSION
This trial has shown that compost amendment of a marginal soil in North Luwu greatly benefited cocoa production. Organic amendment increased cocoa tree survival rate, growth rate and flowering, and resulted in more than a 5-fold higher yield compared with unamended controls and treatments with only inorganic amendments (mineral fertiliser and/or dolomite). In addition, compost treatment improved bean quality and increased bean size. Treatments that included compost gave estimated yields of about 1000 kg ha−1 annum−1 consistent with the yields previously reported for cocoa clone PBC123 in Sulawesi (McMahon et al., Reference McMahon, Bin Purung, Lambert, Mulia, Nurlaila, Susilo, Sulistyowati, Sukamto, Israel, Saftar, Amir, Purwantara, Iswanto, Guest and Keane2015).
Soil N and exchangeable concentrations of the basic cations, Ca2+ and Mg2+, were in the critically low range at the trial site and the data indicate that compost had a substantial role in the provision of these nutrients, either by direct decomposition of the compost, and/or by influencing their availability through critical roles of SOM. SOM can be partitioned into different functional pools. The rapid growth response to the compost treatment in this trial suggests that compost stimulated processes in the labile organic C pool (Woomer et al., Reference Woomer, Martin, Albrecht, Resck, Scharpenseel, Woomer and Swift1994), resulting in increased availability of nutrients, particularly macronutrients. The low C:N ratio (7) of the manure-based compost used in this experiment, and the low C:N ratio (10) in the soil at the study site, would have promoted rapid decomposition and release of nutrients from the added compost. Soil microbial populations may be enhanced by the addition of organic matter (Burger and Jackson, Reference Burger and Jackson2003) and microbial activity plays a crucial role in nutrient availability. Preliminary tests conducted at the North Luwu trial site showed that soils treated with compost had higher microbial activity. Response to the treatments was also likely to have been affected by the relatively young age of the cocoa trees (less than 1 year at the time the treatments were applied). Recommended fertiliser amendments differ between actively growing and mature cocoa trees (Wessel, Reference Wessel, Wood and Lass1985).
Supplying mineral fertiliser alone resulted in growth rates and productivity no higher than untreated trees, as well as marked interveinal necrosis typical of Ca deficiency symptoms in leaves. These symptoms and the low soil exchangeable and leaf concentrations of Ca and Mg and low (Ca+Mg)/K ratios (Tables 4–6), would suggest that the poor growth and yield of these trees was a result of low availability and/or uptake of Ca and Mg, and that this deficiency was exacerbated by applications of the mineral fertiliser used in the experiment. This demonstrates that the rice formulation of NPK (15/15/15) and urea, used widely by farmers and applied in the trial, is inappropriate for cocoa, particularly on such a marginal soil. Compost application, in contrast, alleviated this deficiency.
The soil treatments had less effect on the leaf nutrient contents of macronutrients other than Ca and Mg, suggesting that they did not affect their uptake rates to the same degree. A high proportion of soil N, particularly in weathered tropical soils, occurs in organic matter (whether in living form or as organic molecules; Steiner et al., Reference Steiner, Teixeira, Lehmann, Nehls, De Macedo, Blum and Zech2007) and it is apparent that the compost amendments enhanced availability of N, enabling greater growth and pod yield. Similarly, overall uptake of other macronutrients was increased by organic amendment, resulting in larger and more productive trees, but such an increase did not occur following inorganic amendment, despite the increased supply of macronutrients. Nevertheless, the supply of K was inadequate in the NPK fertiliser (15/15/15) applied in this trial. A composite 12/12/17/2 NPKMg fertiliser is recommended to cocoa growers in Malaysia (Noordiana et al., Reference Noordiana, Syed Omar, Shamshuddin and Nik Aziz2007) and a NPKMg fertiliser with the same formulation is applied as part of a yearly package (along with urea and kieserite) to plantations in East New Britain, Papua New Guinea (Nelson et al., Reference Nelson, Webb, Berthelsen, Curry, Yinil and Fidelis2011). A higher proportion of potash (compared to N and phosphate) in composite fertilisers is also recommended by FAO to cocoa growers in the Ivory Coast: for sandy, coastal soils, a composite 13/10/15 NPK fertiliser is suitable, while on inland, granitic soils, 12/15/18 NPK is advocated (http://www.fao.org/docrep/006/ad220e/AD220E04.htm). However, in the experiment in North Luwu, the similar (or lower) leaf concentrations of K and P, and higher concentrations of Ca and Mg, in the trees provided with compost, compared to the control and trees treated with inorganic amendments, suggest that uptake of Ca and Mg was promoted by compost treatment to a greater degree than K and P (Table 5). Since N and K are mobile within plants they are readily redistributed from older to young leaves (Marschner, Reference Marschner1995). Generally, P becomes less available as soil pH decreases and the low soil pH at the trial site might have reduced P availability. However, the slight increases in pH due to treatment with dolomite did not result in higher P leaf concentrations (Table 5). As was the case for other macronutrients, a greater amount of total P occurred in the trees treated with compost, but this was due to their larger size and greater productivity, rather than increased tissue concentrations. In contrast, tissue concentrations of Ca and Mg were clearly depressed in the control and trees treated with inorganic fertilisers, and these were corrected by supplying compost (Table 5).
The low exchangeable concentrations of basic cations could be attributed in part to the marked soil acidity at the trial site, indicated by the low soil pH and relatively high concentrations of exchangeable Al (Tables 3 and 4). Possible explanations for the poor performance of cocoa trees treated with only mineral fertilisers in the North Luwu trial are the following: first, an increase in concentration of acidic cations in the soil, particularly soluble forms of Fe, Mn or Al, resulted in a toxic effect, perhaps by inhibiting uptake of Ca and Mg, and second, the ratio of exchangeable (Ca+Mg)/K was decreased by amendment with NPK, suppressing uptake of Ca and/or Mg. Taking the first of these alternatives into consideration, the treatments had little effect on soil or tissue Fe and Mn concentrations, but mineral fertiliser applications increased soil concentrations of exchangeable Al (Table 4), possibly to toxic levels for the young cocoa trees. In many cultivated tropical soils, exchangeable, monomeric ions of Al are toxic to plants, resulting in losses in agricultural production (Baligar and Fageria, Reference Baligar and Fageria2005; Marschner, Reference Marschner1995). Studies by Baligar and Fageria (Reference Baligar and Fageria2005) in South America showed that shoot growth of cocoa was reduced by Al saturation of the soil CEC exceeding 19%, which is considerably lower than the approximately 30% Al saturation of the CEC in control soils in North Luwu (Table 4). In a further study in Malaysia, the A-horizon of an acid sulphate soil containing 2.3–4.3% exchangeable Al caused reduced growth of cocoa seedlings (Shamshuddin et al., Reference Shamshuddin, Muhrizal, Fauziah and Husni2004). Cocoa is reported to be more sensitive to elevated soil Al than other tree crops such as oil palm (Shamshuddin et al., Reference Shamshuddin, Muhrizal, Fauziah and Husni2004). Availability of basic cations in the soil is strongly affected by exchangeable Al. Applying mineral fertiliser to cocoa trees in the North Luwu trial may have exacerbated this effect: exchangeable soil Al was 3.3 cmol kg−1 in soils supplied with mineral fertiliser, 31% higher than in control soils and Ca and Mg tissue concentrations were critically low. With increased soil acidity, toxic cations, especially Al that is bound strongly by negative exchange sites, displace Ca and Mg from exchange sites in the soil (Marschner, Reference Marschner1995; Rengel and Robinson, Reference Rengel and Robinson1989). However, in North Luwu the concentration of exchangeable Al was as high in the compost-treated soils as in control soils.
Organic matter has previously been shown to play an important role in reducing Al toxicity (Hue, Reference Hue2011). An important effect of compost may be to increase soil pH, thereby decreasing availability of toxic acidic cations, including monomeric Al species. Addition of organic matter has been reported to increase soil pH by enhancing processes of ammonification, binding of acidic cations and decarboxylation (Hue, Reference Hue2011). However, while compost amendment in the North Luwu trial had a distinct effect on growth and productivity, this was not associated with an increase in soil pH. Possibly, mineralisation processes (that increase soil pH) were counter-balanced by rapid nitrification and leaching (Burger and Jackson, Reference Burger and Jackson2003; Marschner, Reference Marschner1995; Steiner et al., Reference Steiner, Teixeira, Lehmann, Nehls, De Macedo, Blum and Zech2007), and/or inherent soil acidity, which buffered any pH increases. Liming is known to supply Ca and/or Mg to plants and to decrease concentrations of monomeric, toxic species of Al by increasing soil pH (Adams, Reference Adams and Adams1984). However, in this trial, 5 tonnes ha−1 annum−1 dolomite supplied alone did not significantly increase growth or productivity of the cocoa trees; perhaps higher amounts were needed, although these would be impractical.
As mentioned already, the concentration of exchangeable Al was not lower in the soils treated with compost than the control soils. Since soil pH was not increased by compost application in the trial, the supplied organic matter possibly reduced Al toxicity by directly binding and sequestering toxic ions. Additionally, stimulation of microbial activity in the soil by addition of compost might have been involved in Al detoxification, therefore allowing increased uptake of Ca and Mg. Humic acids have a high number of COO− and O− sites that can link to Al cations. Suthipradit et al. (Reference Suthipradit, Edwards and Asher1990) demonstrated fulvic acid forms complexes with Al in the soil. Organic matter in the labile pool may form complexes with Al, thus reducing its toxic effects (Marschner, Reference Marschner1995, p. 614). Al is not readily transported to shoots (Bartlett and Riego, Reference Bartlett and Riego1972) and organic pools most probably prevent uptake of Al, concomitantly facilitating Ca and Mg uptake, rather than promoting a mechanism of detoxification within the plant. However, Shamshuddin et al. (Reference Shamshuddin, Muhrizal, Fauziah and Husni2004) found that a negative correlation occurred between shoot dry weight of cocoa seedlings and tissue concentrations of Al.
As suggested above, an alternative explanation to Al toxicity for the poor performance of trees treated with mineral fertiliser alone is that NPK/urea supply had an adverse effect on nutrient ratios. This is supported by the low exchangeable (Ca+Mg)/K ratio in soils treated with NPK/urea, compared to the other treatments (Table 6). In addition, treatment with NPK/urea alone resulted in low tissue Mg/K and Ca/K ratios (Table 6). Suppressed uptake of Ca and Mg might have resulted from increases in the soil exchangeable K concentration due to mineral fertiliser amendment (Table 4). The treatments resulting in limited growth and production were associated with very low leaf Mg/K ratios, while the highest yielding trees had raised Mg/K ratios (Figure 1; Table 6), suggesting that the poor growth and yield of these trees might be a result of excess inorganic K supply (supplied as KCl in the composite NPK fertiliser used in the trial). Mg is particularly susceptible to displacement from binding sites by K and other cations (Kurvitis and Kirkby, Reference Kurvitis and Kirkby1980). Organic amendment not only provided additional Mg (Table 2), but apparently resulted in facilitation of Mg uptake. This is suggested by the much higher tissue Mg/K ratio in the trees treated with compost plus dolomite (treatment G) compared to those treated with dolomite alone (treatment D) (Table 6). Apparently, availability of Mg supplied with dolomite was enhanced by compost. Tipping the balance of basic cations in the soil beyond a critical level by adding mineral fertiliser alone could have resulted in severe restrictions in uptake of Ca and Mg, and consequently reduced growth and occasional tree death. Therefore, in addition to a role played by soluble Al ions in reducing uptake of Ca and Mg, decreased nutrient ratios might have been a key factor in the poor growth and death of mineral fertiliser-treated trees. Further study would be needed to establish which of the two alternative mechanisms suggested here, enhanced Al inhibition or altered nutrient ratios in the soil and plants, were primarily responsible for the negative effect of the inorganic treatments applied in the trial. However, the results demonstrate that applying local NPK products and/or urea, formulated for rice production, despite their ready availability to farmers, is unsuitable for cocoa production in some soils, particularly if exchangeable cations are marginal. Inorganic fertilisers commonly used by cocoa farmers in Indonesia do not include Mg in their formulations. In contrast, composite fertilisers that include Mg, such as NPKMg (12/12/17/2), are reported to be suitable for cocoa grown in soils with a low availability of exchangeable cations, such as oxisols in Malaysia (Noordiana et al., Reference Noordiana, Syed Omar, Shamshuddin and Nik Aziz2007). Lime could be applied to provide an additional source of Ca, as well as raising the soil pH.
In this experiment, compost was provided at the high rate of 10 kg tree−1 annum−1 to examine the effect of added SOM on cocoa tree growth and productivity. However, such rates of application are unrealistic for most cocoa farmers, and it will be useful to investigate the efficiency of lower amounts of compost or compost applied in inter-row trenches rather than broadcast under the tree. Since the quantities of compost used in the trial are not practical for most cocoa smallholdings, this raises the possibility of supplying nutrients to cocoa trees by combining inorganic fertiliser and compost, although the trial showed no clear benefit in mixing organic and inorganic amendments, as reported previously (Van Lauwe et al., Reference Van Lauwe, Aihou, Aman, Iwuafor, Tossah, Diels, Sanginga, Lyasse, Merckx and Deckers2001). Compost could also be supplied as a strategy to improve soil health and the uptake of nutrients supplied with inorganic fertiliser. Further field studies are needed to quantify the effect of organic applications on nutrient uptake from inorganic sources and also to develop more appropriate formulations of mineral fertilisers for cocoa, since the NPK formulation used here was originally developed for rice.
Preliminary data indicate no significant effect of nutrient amendment on the incidence of PPR (black pod) or VSD, but CPB incidence was significantly lower in the mineral fertiliser treatment. The sparse foliage of these trees was perhaps an unsuitable habitat for the resting adult moths, which may have preferred the leafier compost-treated trees. An alternative explanation is that physiological factors influenced pest non-preference.
CONCLUSION
The results of this trial confirm the crucial role that SOM can play in nutrition of cocoa in marginal tropical soils. The higher Ca and Mg content, CEC and BS in the compost-treated soils and the higher tissue concentrations of Ca and Mg, taken together with substantially increased growth and yield, suggest that availability of these nutrients was limiting to growth and productivity of the cocoa trees. Interveinal necrosis observed in leaves of control and mineral fertiliser-treated trees was indicative of Ca deficiency, while low Mg/K ratios suggest added K may have exacerbated limited Mg availability. Despite increasing the supply of Ca and Mg with dolomite, dolomite-treated trees were also smaller and had lower yields than compost-treated trees. Clearly, organic matter played a role in promoting uptake of these nutrients, resulting in up to five times higher yields and highly improved survival of trees in treatments with compost. Further studies are required to determine appropriate formulations of inorganic fertilisers for use in conjunction with organic amendments on cocoa.
Acknowledgements
This research was conducted under the Australian Centre for International Agricultural Research (ACIAR) project HORT/2010/011 and the support of the Australian Government is gratefully acknowledged. We would like to thank Richard Markham (ACIAR) for his guidance of the project and acknowledge valuable comments on this research provided by Dr John Bako Boan of the Indonesian Coffee and Cocoa Research Institute and the reviewers in the development of the revised manuscript.
SUPPLEMENTARY MATERIAL
To view supplementary material for this article, please visit https://doi.org/10.1017/S0014479717000527
