Types

Traditionally, cocoa was classified according to trade perceptions of its physical and sensory quality and associated botanical traits, albeit imperfectly (Cheesman, Reference Cheesman1944).

• • Criollo: cultivated on a small scale because of its lack of vigour, although the area is currently expanding due to renewed interest in speciality cocoa.

• • Forastero: the bulk of the world's cocoa production is from this type.

• • Trinitaro: usually considered to be the result of crosses between the other two types and is not found in the wild. Material designated as Trinitario is grown commercially in the Caribbean and Papua New Guinea.

Recently, Motamayor et al. (Reference Motamayor, Lachenaud, de Silva e Mota, Loor, Kuhn, Brown and Schnell2008) have proposed a new classification of cocoa germplasm. Based on an analysis of over 1000 individual trees from different geographic areas within the Amazon basin, and using microsatellite markers, they identified 10 genetic groups or clusters, rather than the three listed above, namely: Maraňon, Curaray, Criollo, Iquitos, Nanay, Contamana, Amelonado, Purůs, Nacional and Guiana. The authors believe that this new classification reflects more accurately the genetic diversity available to breeders.

Traditionally, cocoa has always been propagated from seed although the potential for clones (with a plagiotropic – at an angle to the vertical – growth habit, see below) has long been recognized (Lockwood and Boamah Adomako, 2010).The breeding and adoption of new seedling cultivars has led to large increases in productivity, for example in Ghana (Edwin and Masters, Reference Edwin and Masters2005).

Stem and leaf canopy development

Leaf development occurs on the main stem (orthotropic) as well as on the fan (plagiotropic) branches. Initially about 10 leaves begin to develop and expand at the same time. This is known as a leaf flush (Greenwood and Posnette, Reference Greenwood and Posnette1950). After about 40 days a second (and third and so on) flush occurs. As this rhythmic process is largely independent of climate it is considered to be under endogenous control (Almeida and Valle, Reference Almeida and Valle2007). Leaves differ in appearance depending on the type of stem on which they arise (the petioles are about 60 mm longer on chupons, allowing the leaf to be orientated in relation to the light) (Burle, Reference Burle1961).

An individual shoot passes through alternate periods of growth and ‘dormancy’, during which leaf primordia/small leaves are developing in the apical bud (Hardwick et al., Reference Hardwick, Machado, Smith and Veltkamp1988a), and follows a similar pattern to the leaf flush cycle (Greathouse et al., Reference Greathouse, Laetsch and Phinney1971). The growth period is characterized by the expansion of leaves and elongation of the shoot. During dormancy the length of the shoot remains constant and no new leaves expand. The main stem reaches physiological maturity after one or two years from planting (at a height of 1.0–1.2 m) when the apical meristem stops growing and a jorquette of plagiotropic branches (3–5) begin to develop, and these in turn eventually branch. At the same time a second vertical stem (second chupon) appears on the main stem (first chupon). In the course of time, a second whorl of fan branches develops. This process may be repeated several times such that the canopy increases in height (up to 10 m when cultivated and even 20 m when uncultivated and under heavy shade). This is customary practice in West Africa. Cocoa may be pruned to limit its height to 3–5 m, when only the initial jorquette and subsequent branching growth are retained, further chupons are continually removed to restrict the eventual vertical growth, and fan branches may be pruned in order to develop an efficient plant structure and for cultural management.

The importance of light interception and distribution within the canopy in limiting productivity in cocoa germplasm was highlighted by a field study in Malaysia (Yapp and Hadley, Reference Yapp and Hadley1994). For cocoa with a full canopy (48 months after transplanting), planted at a conventional density (1096 trees ha⁻¹), precocity and yield were a simple function of intercepted radiation (f). In a comparison of 12 genotypes, there were large variations in the value of ‘f’ (from 0.50 to 0.82), and in the bean yield/intercepted radiation conversion efficiency (0.085–0.133 g MJ⁻¹). In contrast, for cocoa planted at high density (3333 ha⁻¹), with full canopy cover, yield was related to light distribution in the canopy and the degree of attenuation represented by the light extinction coefficient (k). Shaded leaves low in the canopy can contribute more to dry matter production when the value of ‘k’ is small (Hadley and Yapp, Reference Hadley and Yapp1993). The effect of the age of a leaf and its position within a canopy on its productivity (rates of photosynthesis and respiration) has been described by Miyaji et al. (Reference Miyaji, da Silva and Alvim1997).

Subsequently, the importance of canopy structure in cocoa was confirmed by Daymond et al. (Reference Daymond, Hadley, Machado and Ng2002a). In a comparison of 10 clones in Brazil, the leaf area index (L), for example, reached values ranging from 2.8 to 4.5 (plants spaced 3 m square) There were associated differences in the relationship between L and the fractional light interception. In addition, clones also vary in the base temperature for main stem growth ranging, for example, from 18.6 °C (AMAZ 15/15) to 20.8 °C (SPEC 54/1) (Daymond and Hadley, Reference Daymond and Hadley2004). For comparison, the minimum duration of a leaf flush cycle occurs at a daily mean air temperature of about 26 °C (Hadley et al., Reference Hadley, End, Taylor and Pettipher1994).

Based on the results of a large trial in Sabah, Malaysia, Lockwood and Pang (Reference Lockwood and Pang1996) concluded that the optimum plant density can vary between clones (the densities compared were 1096 and 3333 plants ha⁻¹). Later, Pang (Reference Pang2006) confirmed that, at the current stage of cocoa breeding, selection for adaptation to planting density was of higher priority than selection for yield efficiency (the ratio of cumulative yield over a period of time to the increment in the cross-sectional area of the trunk over the same time period).

Drought symptoms: water availability, by influencing the rate of canopy expansion, and hence light interception, can be expected to influence the productivity of young cocoa. When droughted, seedlings quickly reach a ‘point of no return’ from which they can not recover. Visible symptoms of drought include premature leaf fall (progressively younger leaves abscise), the yellowing of basal leaves, wilting, small leaves and slow trunk growth. The relief of water stress is followed by a large, synchronized flush of leaf growth (Hutcheon, Reference Hutcheon1977b), but the flushing cycle that follows is independent of plant water status (at least until the next dry season) (Hardwick et al., Reference Hardwick, Robinson and Collin1988b). Synchronized leaf growth can also be triggered by an increase in temperature (Hutcheon, Reference Hutcheon1981a).

Flowering

Flowers form in meristematic tissues located above leaf scars on the main stem and woody branches (cauliflorus). Seedling trees do not flower until after the second flush has hardened on jorquette branches, but plants budded from plagiotropic growth can flower when the second growth flush has hardened. An individual site (commonly known as a ‘cushion’, or strictly as a ‘compressed cincinnal cyme’ (Purseglove, Reference Purseglove1968), can have flowers at different stages of development whilst up to 120 000 flowers can be produced each year on a single tree. The floral meristem produces flowers over the tree's entire life span. There is marked seasonal variation in flower production (Edwards, Reference Edwards1973). Flower opening is synchronized and the anthers release their pollen early in the morning. Various small insects, including midges (Ceratopogonidae: Forcipomyia), are responsible for pollination. Unfertilized flowers abscise from the stem about 24–36 h after anthesis. Only 0.5–5% of flowers develop into mature pods.

In Ghana, Smith (Reference Smith1964) found that irrigation of young cocoa increased growth rates, advanced flowering and increased the number of flowers (in the second year after planting) but did not affect the percentage of fruit set compared with the unirrigated control treatment. In his rainfall-only treatment, flushing (leaf production) within a population was synchronized by the onset of the rains, whilst in the irrigated treatments trees responded as individuals. Sale (Reference Sale1970a) found that each time potted cocoa plants were watered following drought, vigorous flushing (and flowering) occurred, beginning about 10 days after watering. Similarly, field observations in Brazil indicated that after a period of water stress, during which flowering was inhibited, had ended with the start of the rains, synchronous flushing (and flowering) occurred (Alvim and Alvim, Reference Alvim and Alvim1977). By transferring plants from low (50–60%) or medium (70–80%) relative humidity to high humidity (90–95%), Sale (Reference Sale1970b) was able to show that flowering could also be induced by the relief from atmospheric-induced water stress. These and other factors influencing seasonal periodicity of flushing and flowering in Ghana have been described by Hutcheon (Reference Hutcheon1977b).

Fruit development

From anthesis to the maturation of the fruit takes 150 days or more. In Papua New Guinea, Bridgland (Reference Bridgland1953) reported actual total gestation periods (recorded in a progeny trial) lasting from 163 to 200 days (mean 182). These were similar durations to those found later in the Cameroon (between 167 and 200 days) (Braudeau, Reference Braudeau1969). The fruit remains attached to the tree until harvested. There are two critical points during the development of the small fruits (known as cherelles) when they may stop growing (known as ‘cherelle wilt’). The first, at about 40 days after fertilization of the flowers, coincides with the first division of the fertilized egg. The second, after about 75 days, coincides with greatly increased fat and starch metabolism and the onset of rapid growth of the fruits (McKelvie, Reference McKelvie1956). At either susceptible stage, more of the cherelles, even all of them, may wilt if the tree is flushing heavily. However, if the tree is not flushing and all the cherelles are at the same developmental stage following synchronized manual pollination, few if any wilt and if the crop is large, flushing is suppressed. If, when the pods are harvested, replacements are set by synchronized manual pollination of freshly opened flowers, they develop normally, without wilting. The trees do not flush, and they may die back.

When fruit set is not synchronized artificially, cherelle wilt is considered to be a fruit thinning mechanism, which occurs in response to a limited carbohydrate supply and competition between pods for carbohydrates, whereby younger pods are preferentially ‘wilted’. In this respect, water stress has an indirect impact on the ‘wilting’ process as carbohydrate assimilation is reduced. Whether water stress also has a direct impact on another growth process such as, for example, cell division in the cherelle is not known. Rather than being a yield-limiting factor itself, cherelle wilt therefore reflects any carbohydrate limitation to pod growth. The number of seeds per fruit is normally in the range 20–50. The rate of increase in pod dry weight is slow for the first 60 days after pollination and peaks after about 100 days before declining.

The timing of harvests varies from region to region depending on temperature and rainfall distribution (Alvim, Reference Alvim1988). For example, in the state of Bahia, Brazil (14°S), where there is no clearly defined dry season, low mean air temperatures (<23 °C) in June, July and August restrict flowering, which results in low yields seven months later. In warmer locations, harvests are mainly influenced by rainfall distribution, again with a six to seven month delay between cause (the start of the rains) and effect (mature fruit).

Genotypes differ in the sensitivity of fruit growth to changes in air temperature, which can affect time to fruit ripening, fruit losses from cherelle wilt, final pod size, bean size and lipid content (Daymond and Hadley, Reference Daymond and Hadley2008). There are also differences between cultivars in the base temperature for fruit growth (range: 7.5 °C for genotypes Amelonado and AMAZ 15/15 to 12.9 °C for SPEC 54/1).

Potential yield

The potential annual biomass (above ground) productivity of cocoa was estimated by Corley (Reference Corley1983) to be 56 t ha⁻¹. Based on an assumed harvest index of 0.20 (perhaps an over-estimate) the maximum seed yield was predicted to be 11 t ha⁻¹. This is considerably more than the best recorded yield of 4.4 t ha⁻¹ (without shade) and the best commercial yields of 1.5–2.5 t ha⁻¹ (Joe Thau-Yin Pang and G. Lockwood, personal communication) Limiting factors, in addition to water stress and pests and diseases, included shade trees (limiting photosynthesis; competition for water), excessive leaf temperatures (especially when grown without shade), low air temperatures (e.g. in Bahia, Brazil), and foliage susceptible to wind damage (benefits from shelter, but reduced leaf cooling because of less wind).

Daymond et al. (Reference Daymond, Hadley, Machado and Ng2002a; Reference Daymond, Hadley, Machado and Ng2002b) have confirmed the potential for yield improvement in cocoa by selectively breeding for efficient partitioning of biomass to the yield component. In a field experiment in Bahia, Brazil, they found, over an 18-month period, a seven-fold difference in dry bean yield between the 12 genotypes tested, ranging from the equivalent of 200 to 1400 kg ha⁻¹. During the same interval, the increase in trunk cross-sectional area ranged from 11.1 cm² to 27.6 cm², with yield efficiencies varying between 0.008 kg cm⁻² and 0.08 kg cm⁻². Of the seven clones compared, beans accounted for between 32% and 45% of the pod biomass.

Roots

According to Kummerow et al. (Reference Kummerow, Kummerow and Alvim1981), summarizing the work of others, cocoa seedlings are characterized by a fast growing tap root on which are formed a limited number of rootlets. After three or four months a first ring of lateral roots has been established. These grow vigorously always close to the soil surface (<100 mm depth) and branch at their distal parts into dense clusters of fine roots. After four to six years these laterals have spread 4–6 m from the stem forming a mat of fine roots, which extend into the decomposing litter layer. Based on detailed measurements in Brazil, Kummerow et al. (Reference Kummerow, Kummerow and Alvim1981) estimated the total length of these roots for an 11-year-old cocoa tree to be 1200 m m⁻² or the equivalent of 1.2 m² of root surface for every 1 m² soil surface. The tap root continues to grow in thickness and length reaching depths of 1.0–1.5 m depending on soil conditions. Lateral roots develop on the tap root at depth, some of which grow upwards. By combining fine root production with reasonable estimates of the dry mass of other roots, including the tap root, the total dry mass of roots was estimated to be conservatively 10 t ha⁻¹.

Similar descriptions of the root systems of cocoa have been provided by Wood (Reference Wood, Wood and Lass1985a) and Toxopeus (Reference Toxopeus, Wood and Lass1985). Both refer to the detailed work in Trinidad summarized by McCreary et al. (Reference McCreary, McDonald, Muldoon and Hardy1943), whilst Van Himme (Reference van Himme1959) has described with illustrations the results of a major study of the root system of cocoa in the Democratic Republic of Congo. The proportion by dry mass of the tap root to the total root system was found by Thong and Ng (Reference Thong and Ng1980) on an inland soil in Malaysia to be constant at 0.84 almost independent of the plant age (up to seven years). By contrast, in a similar study but this time on a coastal clay soil where a water table limited the development of the root system, the tap root represented only 0.15–0.30 of the total dry mass of the root system (Teoh et al., Reference Teoh, Chan and Chew1986).

By observing roots in a glass-sided box, filled with soil, Vogel (Reference Vogel1975), working in the Cameroon, found that roots of cocoa seedlings grew in length in a rhythmic pattern having the same period as the leaf flush. The greatest root growth rate occurred before and during each flush. Subsequently, Sleigh et al. (Reference Sleigh, Hardwick and Collin1981), using a similar method in the UK, confirmed the cyclic nature of root extension, but with maximum root development alternating with active leaf expansion. By contrast, Taylor and Hadley (Reference Taylor and Hadley1988), also in the UK, found that periods of rapid root growth coincided (within three or four days) with maximum rates of shoot growth (and vice versa). During the periods between successive leaf flushes, root growth continued but at a reduced rate. These observations were made in a controlled environment greenhouse with a nutrient film system.

Roots of cocoa appear to grow in a rhythmic pattern but its timing in relation to a leaf flush is uncertain (possibly this is a result of the different methods of root measurement used). There are practical implications in terms of the sensitivity of cocoa to water stress at different stages in the flush cycle, or the best time to transplant seedlings to the field, and subjects worthy of further research.

In a remarkably detailed experiment, Moser et al. (Reference Moser, Leuschner, Hertel, Hölscher, Köhler, Leitner, Michalzik, Prihastanti, Tjitrosemito and Schwendenmann2010) simulated an El Niňo drought in Central Sulawesi, Indonesia (1.55°S, 120.02°E; alt. 585 m) by reducing rain reaching the soil by 70–80% (with plastic roofs placed beneath the cocoa canopy) for 13 months in a six-year-old cocoa plantation. The cocoa was shaded by Gliricidia sepium. All aspects of dry matter production were monitored, including roots, relative to the unprotected control area. Excavations before the drought revealed a superficial distribution of roots in the profile with >83% of the fine roots (<2 mm diameter) and >86% of coarse roots (2–50 mm), both by dry mass, being found in the top 0.40 m. There were significant quantities of fine roots to a depth of 1.0 m whilst the deepest were found at 2.0 m. Coarse roots extended to 1.5 m. Surprisingly, there were no differences in root distribution or mass between the control and the sheltered cocoa at the end of the drought 13 months later. A few roots extended laterally up to >1.0 m, although coarse and large roots (2–150 mm) were concentrated within 0.5 m of the stem. The roots of Gliricidia penetrated much deeper than those of cocoa with considerable numbers of fine roots present at 2.5 m depth.

Little work appears to have been done on genotypic variation in root:shoot ratios.

Summary: crop development

1. 1. Cocoa has a dimorphic growth habit.

2. 2. Flushes of leaf and shoot growth, synchronized by the commencement of the rains after a dry season (or an increase in temperature), alternate with periods of ‘dormancy’.

3. 3. Light interception by the leaf canopy, and light distribution within the canopy, are important determinants of yield.

4. 4. Visible symptoms of drought include premature leaf fall (progressively younger leaves abscise), the yellowing of basal leaves, wilting, small leaves and slow trunk growth.

5. 5. Flowering is inhibited during periods of water stress (and by low temperatures) but the start of the rains results in synchronous flowering.

6. 6. Water stress during fruit development results in small pods and may be an indirect contributory factor in cherelle wilt.

7. 7. Roots grow in a rhythmic pattern but there is conflicting evidence of its timing in relation to shoot growth flushes.

8. 8. Although cocoa is considered to be shallow rooting, roots can extend to depths of 1.5–2.0 m. A large proportion of the roots (80%) are located in the top 0.2–0.4 m.

9. 9. Roots can extend laterally considerable distances (>5 m), but the majority are within 0.5 m of the stem.

10. 10. Only a few of these processes have been quantified in practically useful ways.

Stomata

Stomatal size and frequency as well as the structure and distribution of leaf waxes of two seedling cocoa varieties (Catongo and Catongo/SIAL) were described by Gomes and Kozlowski (Reference Gomes and Kozlowski1988). In a greenhouse study in Wisconsin, USA, the average density of stomata, which are found only on the abaxial (lower) surface of the leaf, was about 700 mm⁻². Both leaf surfaces were covered with heavy deposits of amorphous wax, except for near the stomatal pores. By comparison, in Malaysia, stomatal densities in mixed hybrid seedlings averaged 820 mm⁻² in irrigated seedlings but 1110 mm⁻² in unirrigated plants (increase in density associated with smaller leaves) (Huan et al., Reference Huan, Yee and Wood1986). In a comparison of eight contrasting genotypes in a greenhouse environment in the UK, Daymond et al. (Reference Daymond, Tricker and Hadley2009) recorded stomatal densities ranging from 788 to 1081 mm⁻² (mean 960). In contrast, the densities recorded by Balasimha et al. (Reference Balasimha, Subramonan and Chenchu Subbaiah1985) in a comparison of 40 accessions in South India averaged only about 100 mm⁻² (range 80–122). Such a low density compared with other research suggests an error in measurement or in reporting.

In pioneering research, Alvim (Reference Alvim1958) used the infiltration technique to monitor the degree of stomatal opening in cocoa, and showed, with cuttings and young plants, that the method could be used as a practical indicator of water stress in cocoa. In a field study with five-year-old plants, which were not suffering from water stress, stomatal opening increased as the light intensity increased with maximum opening in strong sunlight.

Using a diffusion porometer in Ghana, Hutcheon (Reference Hutcheon1977a) found that stomatal conductance was controlled mainly by the leaf water status. Partial stomatal opening at low light intensities is a feature of shade-tolerant species like cocoa, as is the capacity of the stomata to respond quickly to changing light intensity. The stomata of well-watered field-grown cocoa plants generally remained open in full sunlight.

Leaf water status

In a review of early research on the water relations of cocoa, Fordham (Reference Fordham1972) emphasised the importance of recording the internal water status of a plant if the results of research on crop water use were to have application at sites away from where the research was undertaken. Attempts to relate rainfall amount and distribution to yields in Ghana, Nigeria and Trinidad had, for example, merely emphasized that the relationships are not simple and have little generic value. Similarly, attempts to quantify the role of irrigation in crop productivity in Ghana and Trinidad had failed to establish causal relationships as they did not always take into account either the influence of pore size distribution on soil water availability, or the impact of the aerial environment (e.g. effects of solar radiation/shade, humidity and wind) on evaporative demand at the crop surface. Since preliminary results had been encouraging, Fordham (Reference Fordham1972) advocated the use of the pressure chamber technique for determining the water status of field-grown cocoa plants, a method subsequently confirmed, in a comparative test, as being suitable for cocoa by Yegappan and Mainstone (Reference Yegappan and Mainstone1981) in Malaysia.

In Ghana, Hutcheon (Reference Hutcheon1977b) described the sequential changes in leaf water potential (and its components) as water stress increased, as recorded in a diverse range of experiments. Initially in a wet soil, values were high early in the morning, declining to about −1.2 MPa by the middle of the day before recovering in the afternoon. As the soil dried midday values declined, reaching lows of −1.5 MPa when midday stomatal closure restricted further water loss. As the soil dried out, recovery from midday closure became weaker until leaf water potentials remained low even at night. There was no evidence of diurnal changes in osmotic potential but some evidence of considerable seasonal changes (down to −2.4 MPa).

In Ecuador, by contrast, low evaporation rates, due to dull, cool conditions, at the start of the long (six months) dry season prevented the midday leaf water potential, measured in the upper canopy of unirrigated, shaded 12-year old trees, from declining below that recorded in irrigated trees (about −0.8 MPa) until there had been 3–4 months without rain. By the end of the dry season, the leaf water potential had reached −1.6 MPa. There was no similar reduction in stomatal conductance which remained relatively constant throughout the dry season. Osmotic adjustment may have occurred. Similar measurements made in semi-arid north-east Brazil, where irrigation is also necessary and where evaporation rates are higher than in Ecuador, indicated how drought stress occurred earlier in the dry season (leaf water potential reached −1.4 MPa within two months) with a large differential in stomatal conductance between irrigated and unirrigated trees (Orchard, Reference Orchard1985).

Gas exchange

In Ghana, Hutcheon (Reference Hutcheon1977a) found a close relationship between photosynthesis rate (measured with the ¹⁴CO₂ method on a wide range of cocoa genotypes and growing conditions) and stomatal conductance, indicating that stomata are an important controlling factor in photosynthesis. Stomata are sensitive to the dryness of the air. For example, in a greenhouse experiment in the UK (under low light conditions), stomatal conductance fell (cv. West African Amelonado) when the saturation deficit of the air exceeded about 1.0 kPa, and continued to decline over the range tested (up to 3.0 kPa). Rates of photosynthesis (measured with an infra red gas analyser) also declined over the same range of values. In contrast, transpiration at first increased with the saturation deficit but, at values above 1.0 kPa, remained constant (Raja Harun and Hardwick, Reference Raja Harun and Hardwick1988). In another greenhouse experiment, this time in Indiana, USA, Joly (Reference Joly1988) judged that net photosynthesis of seedlings declined once the leaf water potential fell below about −0.8 to –1.0 MPa. The relationship (with pooled data) between net photosynthesis and stomatal conductance was again (Hutcheon, Reference Hutcheon1977a) best described by an exponential asymptotic equation. Assimilation rates varied linearly with transpiration. The three cultivars tested (EET399 and EET400, half siblings of Ecuadorian Oriente origin, and one Trinitario type, UF613) differed in their instantaneous water-use efficiencies, that is net photosynthesis divided by transpiration (Joly and Hahn, Reference Joly and Hahn1989).

By contrast, in a greenhouse/controlled environment study in Maryland, USA, three different genotypes (CCN 51, LCT EEN 37/A and VB 1117 representing different types) responded in similar ways to changes in ambient conditions (light intensity, carbon dioxide concentration and dryness of the air). As an example, the short-term effects of increasing the saturation deficit of the air (from 0.9 to 2.2 kPa) were to reduce the rate of net photosynthesis slightly, to increase transpiration but to have minimal effect on stomatal conductance. Cocoa was considered to be unusually ineffective in limiting transpiration in dry air compared with other rain forest tree species (Baligar et al., Reference Baligar, Bunce, Machado and Elson2008).

The sensitivity of the stomata of cocoa to the saturation deficit of the air was demonstrated further in Colombia by Hernandez et al. (Reference Hernandez, Cock and El-Sharkawy1989). Stomatal conductance declined rapidly in shaded plants (cv. ICA4xIMC67) as the saturation deficit was increased from 0.5 to 3.5 kPa, whilst apparent carbon dioxide uptake was reduced at values above about 2.0 kPa. Transpiration at first increased as the saturation deficit was increased from 0.5 to 1.0 kPa, but then declined, reaching very low values at 4.0 kPa. The implications of these observations to the sustainable management of cocoa under field conditions were considered, in particular the value of shade (water use efficiencies are likely to be higher under shade, although in practice this will depend on such things as the amount of light intercepted by the shade canopy, the leaf area index of the cocoa canopy and the ambient light conditions) and irrigation (responses will be limited during periods of hot, dry air).

In a field study in India, Balasimha et al. (Reference Balasimha, Daniel and Bhat1991) also found that photosynthesis declined as the saturation deficit of the air increased above about 2 kPa. Drought tolerant accessions maintained higher leaf water potentials during dry months than drought susceptible ones.

In Malaysia, container-grown plants were subjected to water stress of different durations. Leaf water potentials fell to minimum values of about −3.0 MPa in droughted plants with a suggestion that the critical value below which stomatal conductance and photosynthesis both declined was about −1.5 MPa. There was an indication that one genotype (KKM25) could withstand water stress better than the other two (KKM4, KKM5) tested (Razi et al., Reference Razi, Halim, Kamariah and Noh1992).

A similar study was undertaken in a semi-arid region of Venezuela (08°31′N 71°71′W; alt. 1100 m) this time with field-grown, shaded, four-year-old plants (cv. Guasare, Criollo type) irrigated every 3, 12 or 25 days. Minimum leaf water potentials were in the range −1.4 to −1.7 MPa, and there was evidence of osmotic adjustment, but only in the 3- and 12-day treatments. In severely stressed plants, daily photosynthesis was reduced by 25% and transpiration by 39% compared with the 3-day treatment, which implies an increase in water-use efficiency (Rada et al., Reference Rada, Jaimez, Garcia-Núňez, Azócar and Ramirez2005). In a greenhouse experiment in the UK, significant differences in the instantaneous water-use efficiencies were recorded in the eight genotypes compared, ranging from 3.1 (IMC47) to 4.2 mmol mol⁻¹ water (ICS1) (Daymond et al., Reference Daymond, Tricker and Hadley2009).

Shading, as well as water, also influences gas exchange processes. In a nursery experiment in Ghana, young clonal plants were shaded with netting at three different levels (32.5, 55 and 76% shade). The plants were kept well watered. During the two rainy seasons, stomatal conductance, photosynthesis and transpiration all declined as the level of shade increased, but during the dry season the situation was reversed (Acheampong et al. Reference Acheampong, Daymond and Hadley2009). This difference between the seasons was explained by the dryness of the air. Saturation deficits averaged about 1.4 kPa in the rains and 2.7 kPa in the dry season.

That wind can physically damage cocoa was confirmed in a wind tunnel experiment in Wisconsin, USA. Of particular interest though was the observation that an increase in wind speed from 1.5–3.0 m s⁻¹ to 6.0 m s⁻¹ increased stomatal conductance, but reduced transpiration. This was apparently the result of a reduction in the leaf:air saturation deficit associated with cooling of the leaves by the wind (Gomes and Kozlowski, Reference Gomes and Kozlowski1989).

Drought tolerance

In South India, where there is an extended dry season, Balasimha et al. (Reference Balasimha, Rajagopal, Daniel, Nair and Bhagavan1988) developed a procedure for determining the relative drought tolerance of cocoa accessions. Based on an assessment of a number of plant parameters, including effective stomatal control of transpiration when under water stress, five accessions (NC23, NC29, NC31, NC39 and NC42) were identified as being drought tolerant. Later, Balasimha et al. (Reference Balasimha, Anil Kumar, Viraktamath and Ananda1999) evaluated the drought tolerance of the progeny of crosses between four high yielding and three drought-tolerant pollen parents (NC23/43, NC29/66, and NC42/94). Using the rapid excised leaf screening method (based on changes in leaf water potential of an excised leaf in 90 minutes (Balasimha and Daniel, Reference Balasimha and Daniel1988)), and follow-up field measurements during the dry season over a five-year period, they identified two hybrids (1–21 × NC42/94; 1–29 × NC23/43) exhibiting drought tolerance characteristics on the basis of their higher leaf water potentials and stomatal resistances during dry weather.

In a comparison of eight clones (all grafted on to local Comum variety rootstocks) in a greenhouse experiment in South Bahia, Brazil, three were identified as being drought tolerant on the basis of the degree of osmotic adjustment recorded, but only when dehydration was imposed rapidly (Almeida et al., Reference Almeida, Brito, Aguilar and Valle2001). The osmotic adjustment was associated with the accumulation of potassium and phosphorus ions in the leaf. These three clones (SPA5, SIAL70 and TSH516) were recommended for growing on drought prone shallow soils in the region. By contrast, Premachandra and Joly (Reference Premachandra and Joly1994) could find no evidence of osmotic adjustment in cocoa seedlings subjected to water stress over a 22-day period in a glasshouse experiment in Indiana, USA.

According to Bae et al. (Reference Bae, Kim, Kim, Sicher, Lary, Strem, Natarajan and Bailey2008), it may be possible to enhance drought tolerance in cocoa by altering polyamine levels, which are associated with the response of plants to drought, either by selection or by genetic manipulation. Although transgenic techniques in cocoa are at present discouraged because of consumer concerns, the genes involved in polyamine biosynthesis have been identified.

Bae et al. (Reference Bae, Sicher, Kim, Kim, Strem, Melnick and Bailey2009) have also identified another possible way of enhancing drought tolerance in cocoa. They found that colonization of young cocoa seedlings by the endophytic fungus Trichoderma hamatum (isolate DIS 219b) promoted growth and delayed the onset of drought symptoms through changes in gene expression. The primary direct effect of colonization was to promote root growth (dry weight and fresh weight) regardless of the plant water status. This delayed drought-induced changes in stomatal conductance, net photosynthesis and green fluorescence emissions as well as wilting.

Summary: plant water relations

1. 1. Stomata are found on the abaxial surface of the leaf at densities of 700–1100 mm⁻².

2. 2. Stomata open in low light intensities and remain fully open in full sunlight on well-watered plants.

3. 3. Leaf water potentials in well-watered plants can decline to −1.2 MPa at midday, before recovering.

4. 4. As the soil dries, minimum values of leaf water potential can reach −3.0 MPa in droughted plants.

5. 5. Partial stomatal closure begins at a leaf water potential of about −1.5 MPa.

6. 6. Stomatal conductance is sensitive to dry air, declining as the saturation deficit increases from about 1.0 up to 3.5 kPa.

7. 7. Net photosynthesis declines over a similar range of values.

8. 8. Transpiration initially increases but when the saturation deficit exceeds about 1.0 kPa it begins to decline (or remains constant), reaching very low values at 4 kPa.

9. 9. Cultivars differ in their instantaneous water-use efficiencies.

10. 10. Leaf water potential and stomatal conductance measurements have been used to identify drought resistant cultivars.

11. 11. There is conflicting evidence about the contribution of osmotic adjustment to drought tolerance.

Summary: crop water requirements

1. 1. Little has been reported on the water use of cocoa.

2. 2. Field data (based on the sap flow method) suggest ETc rates of <2 mm d⁻¹ which, for a crop with a complete canopy, would appear to be low.

3. 3. The corresponding Kc value is only 0.3, much less than the theoretical value for a cocoa crop or any crop with complete ground cover.

Yield forecasting

A number of attempts have been made to relate cocoa yields to rainfall, for example in Trinidad (Dunlop, Reference Dunlop1925), Papua New Guinea (Bridgland, Reference Bridgland1953), Ghana (Ali, Reference Ali1969; Maidment, Reference Maidment1928; Skidmore, Reference Skidmore1929) and Nigeria (Toxopeus and Wessel, Reference Toxopeus and Wessel1970), also reviewed by Fordham (Reference Fordham1972). But, as the role of rainfall in crop growth is not a simple one, these attempts have had limited success. Recently, a physiological growth and production model for cocoa has been developed and described by Zuidema et al. (Reference Zuidema, Leffelaar, Gerritsma, Mommer and Anten2005). Known as SUCROS-Cocoa, it is based on the SUCROS-family of models with parameter values taken from the literature. It simulates biomass production and bean yield for different situations and locations. Lack of good independent data has limited the validation of the model. Regression analysis showed that over 70% of the variation in simulated bean yield could be explained by a combination of annual solar radiation and rainfall in the two driest months. Yield losses due to drought of up to 50% depending on location and soil type were predicted. Heavy shading (>60%) reduced yields by more than 30%. Opportunities to improve the model were identified.

Responses to irrigation

In view of the apparent sensitivity of cocoa to drought, there have been surprisingly few irrigation experiments, but perhaps less surprising given the limited likelihood of commercial scale irrigation. Murray (Reference Murray1961) reported the results of one such experiment in Trinidad, but problems in applying and monitoring the amount of water applied in the dry season (acknowledged by the author) meant that the results obtained over five years (no consistent yield advantage from irrigation) had little value. Low atmospheric humidity in the dry season was suggested as a possible limiting factor. An irrigation experiment in Ghana was also compromised by poor design (recognized as a weakness) that meant the data could not be analysed statistically. Yield increases of 12, 17 and 40% were obtained from mature Amelonado trees by keeping the soil close to field capacity in the three years 1960–1962. These increases were less than anticipated at the time due, it was again thought, to ‘dry air’ constraints. No absolute yields were presented (Hutcheon et al., Reference Hutcheon, Smith and Asomaning1973).

Another irrigation experiment was undertaken in Malawi (16°31′S 35°10′E; alt. 52 m) by Lee (Reference Lee1975). The depth of water applied was based on three evaporation pan factors (0.6, 0.8 and 1.00 × Epan (Kenya type)), with two irrigation intervals based on calculated soil water deficits (52 or 104 mm). Detailed records were kept for two years (1971/72 and 1972/73) when the average annual rainfall was 710 mm and the depths of water applied over the long dry season were 920–1650 mm. Yields of wet beans from a 4–5 year-old Amazon cultivar (T76) were similar for all three water applications, averaging about 2200 kg ha⁻¹. More frequent irrigation out-yielded less frequent irrigation by 285 kg ha⁻¹ (2318 and 2032 kg ha⁻¹), although plot to plot variability was large (CV >30%). Hot dry winds damaged the crop towards the end of the dry season. The alluvial soil had a water holding capacity of 230 mm m⁻¹. The seasonality of harvest was not affected by any of the treatments.

In Côte d'Ivoire, Jadin and Jacquemart (Reference Jadin and Jacquemart1978) compared two methods of irrigation (sprinklers and drip) with an unirrigated control treatment on the development and yield of young cocoa over a two-year period. Sprinkler irrigation was applied when the measured soil water deficit reached 20 mm, whilst drip irrigation was applied when the soil water tension reached 20 kPa at 0.20 m depth. As a result in part of these differences in scheduling irrigation, considerably more water was applied with the sprinklers in the dry season (535 mm) than with drip (224 mm). Irrigation, particularly drip, speeded up the rate of development, increased the number of flowers, and increased yields but did not affect the periodicity of the growth cycle.

The results of an unreplicated field scale irrigation trial in Peninsular Malaysia were later reported by Huan et al. (Reference Huan, Yee and Wood1986). Supplementary irrigation (drip) was applied daily to a 0.5 ha block of mixed hybrid seedlings on a coastal estate (marine clay) after a dry period (no rain for two weeks), except on days after there had been 5 mm or more rain, or it was actually raining. A similar 0.5 ha block acted as the unirrigated control. The trial lasted nearly three years (1981–1983). Annual dry bean yields were increased by irrigation from 1500 to 2400 kg ha⁻¹ (+60%) in 1982 and from 1150 to 1450 kg ha⁻¹ (+28%) in 1983. This followed an increase in pod number (averaging +39%) and in bean weight (+7%). The quantities of water applied were not specified. These results can only be considered indicative of responses to irrigation at this site, and have limited generic value.

In the simulated El Niňo drought experiment reported by Moser et al. (Reference Moser, Leuschner, Hertel, Hölscher, Köhler, Leitner, Michalzik, Prihastanti, Tjitrosemito and Schwendenmann2010) in Indonesia (described above under Roots and Crop water requirements) there were no significant differences in cocoa leaf, stem and branch wood, or fine root biomass production (above and below ground) between the control treatment and the one in which rain through-fall was reduced by 70–80%, even though the soil profile dried to permanent wilting point during the year. By contrast, there was a reduction in dry bean production over the year as a whole from 740 ± 180 to 670 ± 30 kg ha⁻¹ (both low yields when compared with the best commercial yields), with the later harvests more affected than the early ones. Possible causes of the limited response to drought in terms of net biomass production were proposed. These included active osmotic adjustment in the roots (measured), high atmospheric humidity in both treatments, and drought mitigation through shading by Gliricidia.

In contrast to the trials described above, the principal practical objective of an experiment summarized by Hutcheon (Reference Hutcheon1981b) was to find out if irrigation would induce Amelonado trees to flower throughout the dry season in order to produce pollen for use in manual pollination of a seed orchard in Ghana. When the cherelles were continuously removed, irrigated trees produced 30% more flowers than those that were unirrigated, although the flowering patterns were the same. There was no benefit from using over-tree sprinklers (to reduce internal water stress by raising the humidity) rather than micro-sprinklers under the trees.

Drought mitigation

Repeated mulching (with fresh, moist plantain pseudostems) improved the establishment of cocoa seedlings during the extended six-month dry season in Ecuador (Orchard and Saltos, Reference Orchard and Saltos1988). Mulched plants maintained stomatal opening at similar levels to irrigated plants whilst, at the end of the first year of establishment, there were substantial increases in the dry weight of shoots and roots, and in leaf area over and above those recorded in the irrigated treatments. The benefits of applying a mulch during the first three to four years after establishing cocoa seedlings in the field can be large. But, the cost of growing, cutting, transporting and spreading the mulch can be prohibitive unless it is available on the spot as, for example, when leguminous shrubs such as Flemingia macrophylla are used as temporary shade (Wood, Reference Wood, Wood and Lass1985c). The role of shade trees in drought mitigation (or enhancement) is complex and variable depending on many variables and is worthy of a separate review.

Specific advice to growers on irrigation practices is even harder to find but the Central Plantation Crops Research Institute, Kerala, South India gives this (edited) advice to cocoa producers in some areas of southern India, where long periods of dry weather lasting three to six months can occur (CPCRI, 2010):

‘Irrigate at weekly intervals during the summer [presumably refers to a sole crop]. When cocoa is grown as a mixed crop with arecanut, irrigate once a week during November-December, once every six days during January-March and once in four to five days during April-May with 175 l water tree^{− 1}. Maximum yields are obtained when cocoa is drip irrigated with 20 l day⁻¹ tree⁻¹.’

Assuming a planting density of 1600 trees ha⁻¹ (2.5 m × 2.5 m) these figures equate to rates of water use equivalent to 5.6–7.0 mm d⁻¹ or at 1100 trees ha⁻¹ (3 m × 3 m) 3.9–4.8 mm d⁻¹, and 2.2 or 3.2 mm d⁻¹ for drip irrigation. No estimates of the yield benefits are given or of the total quantity of water to be applied over a season. Presumably this is not known.

Summary: water productivity

1. 1. There is a paucity of reliable published data quantifying the yield and other benefits that could result from well-managed irrigation of cocoa in different locations.

2. 2. It remains to be seen under what conditions irrigation (and/or drought mitigation, including mulching of young trees) is financially worthwhile.

3. 3. Dry air can be expected to limit responses to irrigation.

4. 4. It is not possible to specify yield response functions to water with the limited information available.

5. 5. Shade trees add a further complication.