Crop Protection 18 (1999) 677}685
Increased frequency of phytosanitary pod removal in cacao (Theobroma cacao) increases yield economically in eastern Peru W. Soberanis, R. RmH os, E. AreH valo1, L. ZuH n8 iga1, O. Cabezas, U. Krauss* Universidad Nacional Agraria de la Selva, Tingo Marn& a, Apdo 156, Hua& nuco, Peru Received 21 June 1999; received in revised form 20 July 1999; accepted 1 September 1999
Abstract Diseases caused by Moniliophthora roreri (moniliasis), Crinipellis perniciosa (witches' broom) and Phytophthora palmivora (black pod) are the most important factors limiting cocoa production in Peru and cultural management is considered to be the only practical means of control for the smallholder. The objective of this study was to develop an epidemiologically and economically sound control recommendation based on the frequency of phytosanitary pod removal. Weekly removal of diseased pods reduced the incidence of diseases signi"cantly in comparison with fortnightly removal. In three "elds, moniliasis was decreased by 26}41%, black pod by 35}66% and witches' broom on pods by 14}57%. The cumulative e!ect was a consistent yield increase from an average of 504}660 kg ha~1 yr~1. Returns compensated for increased labour costs. Weekly pod removal was 32% more pro"table. Timecourse analyses indicated that weekly removal should be practised throughout the year under "eld conditions of eastern Peru. ( 2000 Elsevier Science Ltd. All rights reserved. Keywords: Crinipellis perniciosa; Cultural disease management; Economics; Moniliophthora roreri; Phytophthora palmivora; Theobroma cacao
1. Introduction In 1996, Peru produced 22,867 metric tons of cocoa on 32,200 ha (Fujimori et al., 1997). This production does not meet the demand of its domestic cocoa processing industry and export markets. Peru is a net importer of cocoa beans, but, since it is a net exporter of all valueadded cocoa products, the trade balance is positive. In 1996, its FOB value amounted to US$ 8.9 million (Fujimori et al., 1996) and is thus a major earner of foreign exchange. If Peru could increase productivity by an annual average of 7%, it would achieve self-su$ciency for the raw product and save almost $2 million. Fungal diseases constitute the major constraint for cocoa production in Peru (Evans et al., 1998) and other Latin American countries (Barros, 1982; EnrmH quez et al., 1982; Evans et al., 1977; HernaH ndez, 1982; Maddison
1 Present address: Instituto de Cultivos Tropicales, Jr. Tarapoto No 247, Banda del Shilcayo, Tarapoto, San MartmH n, Peru. * Corresponding address: CATIE, 7170 Turrialba, Costa Rica. Tel.: #506-556-6431; fax: #506-5526-0606. E-mail addresses:
[email protected] (W. Soberanis),
[email protected] (E. AreH valo),
[email protected] (L. ZuH n8 iga),
[email protected] (U. Krauss)
et al., 1995; Porras and SaH nchez, 1991; Trevizan, 1996; Zadocks, 1997). In Peru, the three major pathogens are Moniliophthora roreri (Cif. & Par.) Evans et al., Crinipellis perniciosa (Stahel) Singer and Phytophthora palmivora (Butl.) Butl. (Evans et al., 1998; Ram and AreH valo, 1997). However, if disease problems could be addressed more e$ciently and economically, cocoa would be one of the most lucrative cash crops. Disease control with fungicides generally has been unsatisfactory in cocoa. Because of the rapid expansion of the pod surface during the "rst two to three months of pod development, when the pod is at its most susceptible stage, protective fungicides have to be applied frequently. This is both wasteful and di$cult in tall trees (Evans et al., 1977). Field trials with systemic fungicide have also been disappointing, especially against witches' broom (Evans et al., 1977; Purdy and Schmidt, 1996) and moniliasis (Evans, 1981) on smallholdings which, in Peru, represent the typical form of cocoa cultivation. Studies in several countries indicate that fungicides can be economical for the control of Phytophthora spp. but only if disease incidence exceeds 25}50% (McGregor, 1982; Thorold, 1967). Copper-based fungicides are most e!ective and relatively inexpensive. However, by nature, they are highly soluble and adhere poorly to pods especially in
0261-2194/00/$ - see front matter ( 2000 Elsevier Science Ltd. All rights reserved. PII: S 0 2 6 1 - 2 1 9 4 ( 9 9 ) 0 0 0 7 3 - 3
678
W. Soberanis et al. / Crop Protection 18 (1999) 677}685
Fig. 1. Cocoa-growing areas in Peru.
the rainy season when they would be needed the most. Therefore, costly formulations are necessary which has rendered their use uneconomical in Peru, unless yields exceed 800 kg ha~1 yr~1 (HernaH ndez, 1991). Fungicides are seldom used in Peru and the main control method available to the smallholder is phytosanitation, i.e. the removal of diseased pods from the tree to the ground. Matlick (1998) recently stated that, although we believe that we know how to control cocoa diseases, even cultural practices are under-investigated. He recommended an evaluation of the `machete technology for smallholdersa as a short-term control measure while long-term solutions, based on biotechnology, are being developed.
When we revised practical manuals for cocoa production in Peru, we found that most authors failed to recommend any de"ned frequency for cultural control. Fortnightly pod removal is practised only on some well-managed farms; the majority of growers either practice it less frequently or do not practice at all. However, at least for polycyclic diseases such as moniliasis and black pod, weekly removal may be bene"cial. In Colombia, weekly removal proved to be both most e$cient and economical in comparison with monthly and twice-weekly removal (Cubillos and Aranzazu, 1979). In Costa Rica, weekly removal improved yields by more than a factor "ve in comparison with no removal (Porras et al., 1990).
W. Soberanis et al. / Crop Protection 18 (1999) 677}685
The objective of our present study is to compare weekly with fortnightly phytosanitary pod removal in pathological, as well as in economical terms, in order to arrive at a de"ned recommendation for Peruvian cocoa growers. Additionally, the results of a disease survey are presented to quantify recent trends. 2. Materials and methods 2.1. Disease survey Disease surveys were carried out at monthly intervals in 1997 and 1998. Disease incidence in the seven zones listed in Table 1, and depicted in Fig. 1, was evaluated along 10 transects using zigzag or diagonal sampling, depending on the shape of the plot. One-hundred and twenty tree samples were collected randomly in 9}42 di!erent cocoa "elds covering a minimum area of 1500 m2/"eld. Identi"cation was assisted by microscopy and/or incubation of split open pods in humid chambers for 3}4 days to allow sporulation to occur and thus be able to distinguish M. roreri from C. perniciosa, whenever necessary. Disease incidence was calculated on a present/absent basis per tree sample and year (average of twelve sampling dates). Averages per zone were weighted according to the number of "elds surveyed.
679
to July 1996. Tingo MarmH a is situated in the eastern foothills of the Andes mountains at an altitude of 666 m.a.s.l., latitude 09349@ and longitude 75357@ (Fig. 1). It receives an annual rainfall of 3000 mm, with a mean annual temperature of 243C (maximum 29.53C, minimum 19.83C) and mean relative humidity of 81% (minimum 67%, maximum 94.8%) (CeH sare, 1986; HernaH ndez, 1991; Servicio Nacional de Sanidad Agraria, 1998). The three "eld sites, A"lador I, A"lador II and Las Islas were located on the right bank of the river Huallaga in alluvial soils. The cocoa "elds had a size of approximately 0.5 ha and consisted of a mixture of eight-year-old hybrids; the spacing was 4]4 m. At the beginning of the study, all trees were pruned to a height of 3}4 m; 70% of trees were between 3.5 and 3.8 m in height. A"lador I was under sporadic (ca. 20%) mixed shade, A"lador II was under dense shade (almost 100%) of avocado (Persea americana) and Las Islas under 50% consisting of a 1 : 1 mixture of avocado and Inga edulis. Prior to the installation of the experiment, the "elds were managed `traditionallya which comprises harvesting of healthy pods and manual weed control. Diseased pods were neither removed nor were trees pruned under this system. The three "elds are situated in an area where inoculum pressure from surrounding neglected or abandoned plots is high. 2.3. Experimental design, data collection and statistical analysis
2.2. Experimental sites for xeld trials Field trials were carried out in three farmers' "elds in the Huallaga valley within a 2}4 km radius from Tingo MarmH a, Department of HuaH nuco, Peru, from April 1995
The trials were arranged in a split plot design with 100 replicate trees per location and treatment. Two-hundred
Table 1 Disease incidence (% trees infected) in cocoa (Theobroma cacao) in the river valleys of eastern Peru in 1998. Percentages represent the mean of 12 monthly measurements Valley Huallaga
Ucayali
Urubamba ApurmH mac
Pathogen
Lower Central Upper AguaytmH aSan Alejandro
Pachitea Ucayali
Upper
San FranciscoSivia
Weighted average!
Change from 1997
Crinipellis perniciosa Moniliophthora roreri Colletotrichum gloeosporioides Botryodiplodia theobromae Phytophthora palmivora and Phytophthora capsici Pellicularia koleroga Rosellinia spp Corticum salmonicolor Thielaviopsis p. Fusarium decemcellulare Ceratocystis xmbriata Armilaria sp
18.2 22.7 13.6 13.6 4.5
70.0 50.0 15.0 45.0 30.0
85.7 66.7 64.3 28.6 38.12
55.6 55.6 55.6 22.2 11.1
53.3 26.7 26.7 13.3 13.3
88.9 11.1 22.2 33.3 33.3
64.7 47.1 41.2 23.5 17.6
64.2 45.5 38.1 26.1 23.9
#0.2 #3.4 #21.7 #13.0 #10.6
0.0 0.0 0.0 0.0 0.0 0.0 0.0
10.0 20.0 0.0 0.0 0.0 0.0 0.0
28.6 26.2 11.9 9.5 9.5 7.1 4.8
0.0 11.1 0.0 0.0 0.0 0.0 0.0
13.3 6.7 0.0 0.0 0.0 0.0 0.0
22.2 0.0 0.0 0.0 0.0 0.0 0.0
23.5 11.8 0.0 5.9 0.0 0.0 0.0
16.4 14.2 3.7 3.7 3.0 2.2 1.5
#4.5 #6.0 nd" nd nd nd nd
!Weighted on a per "eld basis. "nd, no data.
680
W. Soberanis et al. / Crop Protection 18 (1999) 677}685
Fig. 2. Monthly (bars) and cumulative (lines) rainfall in 1995/96 (solid "ll) and long-term average (hatched) (Servicio Nacional de Sanidad Agraria, 1998).
Fig. 4. Time course of moniliasis sporulation in three "elds near Tingo MarmH a in 1995/96.
Fig. 5. Time course of Crinipellis infection in pods in three "elds near Tingo MarmH a in 1995/96. Fig. 3. Flowering and harvest peaks in three "elds near Tingo MarmH a in 1995/96.
Fig. 6. Time course of Phytophthora spp. infection in three "elds near Tingo MarmH a in 1995/96. Fig. 8. Time course of harvest of healthy pods under weekly (solid) and fortnightly (open) disease management. Average of three "elds near Tingo MarmH a in 1995/96.
trees were selected randomly and marked in each "eld; 100 trees each were submitted to one of the two treatments: weekly versus fortnightly removal of diseased pods. Pruned, diseased material was left on the plantation #oor according to local practice. The following data were collected, either weekly or fortnightly according to the treatment: number of pods past the wilting stage ('8 cm) infected with M. roreri, C. perniciosa or Phytophthora spp. Pods infected with M. roreri were further
Fig. 7. Time course of vegetative broom formation in three "elds near Tingo MarmH a in 1995/96.
W. Soberanis et al. / Crop Protection 18 (1999) 677}685
distinguished into sporulating pods and those with earlier symptoms. Non-sporulating pods with ambiguous symptoms were split open and incubated as described under disease survey. For comparison between treatments, two weeks of evaluation of the weekly removal were pooled and compared with the value for fortnightly removal in a balanced design. Harvesting was done every fortnight according to local practice. At this time, the number of healthy, mature pods was recorded and witches' brooms (of foliage, branches and in #ower cushions) were counted and removed to avoid recording them again on subsequent evaluation dates. Mature pods which had been infected at a late stage were split open and any non-a!ected seed was pooled with seed from healthy pods for the determination of yield per tree. Yields per hectare were extrapolated by multiplying yields of 100 trees with a factor 6.25 given a planting density of 625 trees ha~1. Production costs for the traditional system were calculated according to CeH sare (1986) using up-dated prices. Returns for the traditional system were estimated from data of the preceding season (farmers' sales up to week 0 and measured yields up to week 18). For 10 randomly selected trees within each sub-plot, the number of #owers was quanti"ed every week, independent of treatment, up to a height of 3 m marked along the main stem and one jorquette branch. Data of weeks 0}66 were analysed using GENSTAT 5, Release 2.2 (Genstat 5 Committee, 1993) and the appropriate general linear model for binomial (percentage data) and Poisson (counts) error distributions. All factors (removal frequency, "eld, time), as well as their "rst order interactions, were "tted sequentially into the regression model and tested. The `r-paira procedure was employed to identify signi"cant di!erences between treatments for the same evaluation date. Climatic data were obtained from the meteorological station of the Universidad Nacional Agraria de la Selva in Tingo MarmH a, less than 1 km from each plot.
3. Results The results of the disease survey in 1998 are presented in Table 1. For comparison, the cocoa-growing areas of Peru are shown in Fig. 1. C. perniciosa and M. roreri were the most frequently observed pathogens. M. roreri and Phytophthora spp. incidence increased in comparison with 1997. It is worth mentioning that M. roreri was detected in the Urubamba valley for the "rst time in August 1998 (RmH os and RodrmH guez, 1998) and by the end of the year, the average annual incidence was 11.1% (Table 1). This pathogen also increased importance in other areas such as the upper Huallaga valley. In the trials, moniliasis was the most serious disease accounting for 73% of losses. Witches' broom, with 17% of losses, was next and black pod came third with 10% of
681
losses. Over the observation period, the climate followed a typical pattern with a mean temperature range of 21.2}25.93C, and average temperature minima and maxima of 19.3 and 29.63C, respectively. The temperature extremes recorded were 17.5 and 32.23C. Rainfall was 2830.6 mm yr~1. In the Huallaga valley near Tingo MarmH a, annual rainfall of 3000 mm is distributed throughout the year with a distinct rainy season from October to May (Servicio Nacional de Sanidad Agraria, 1998). August and September 1995 were somewhat drier than average, but the rainy season started on time in October with average precipitation (Fig. 2). Above average rainfall in May 1996 made up for slight de"cit from the previous year. Flowering of cocoa increased with the onset of the rainy season and peaked in week 32 (early November 1995) (Fig. 3). The peak of cocoa harvest was in week 62 (mid June 1996), 30 weeks after the peak in #owering. A small harvest peak at week 12 during the corresponding rehabilitation phase in 1995 is also visible. Pod losses due to disease generally peaked before the main harvest (Figs. 4}6). The number of sporulating pods infected with moniliasis under the fortnightly removal (Fig. 4) and also those infected with Phytophthora under both removal regimes (Fig. 6) were highest 14 weeks after the peak in #owering. Witches' broom disease of vegetative tissues (Fig. 7) and pods (Fig. 5) exhibited two peaks. The "rst, smaller peak of pod loss followed broom formation in vegetative tissue with a four-week lag period. The second, more prolonged peak, occurred 24 weeks after #owering and was also preceded by increased vegetative broom formation. No di!erence in numbers of vegetative brooms were found between treatments (data not shown). Weekly removal of diseased pods signi"cantly reduced the incidence of all three diseases (Tables 2}4). Moniliasis incidence (Table 2) was reduced by over 40% in both "elds in A"lador and by 26% in Las Islas. The number of sporulating pods was reduced by a similar magnitude (Table 2, Fig. 4). For losses due to witches' broom (Table 3), the interaction treatment]"eld was signi"cant (deviance ratio 4.09, dof"2) which means that the three "elds responded di!erently to the treatments. Whereas disease incidence in Las Islas was reduced by 57%, "elds in A"lador responded less to weekly removal. Black pod (Table 4) was reduced signi"cantly by weekly pod removal, independent of the site. As a result of improved disease control, yields in all "elds were increased signi"cantly by 26}36% with weekly removal of diseased pods (Table 5). Phytosanitary pod removal increased production costs from an estimated US$ 125 ha~1 yr~1 under traditional management to US$ 229 and US$ 293 ha~1 yr~1 for fortnightly and weekly removal frequency, respectively (Table 6). However, the increase in gross returns more than compensated for increased expenditure in labour.
682
W. Soberanis et al. / Crop Protection 18 (1999) 677}685
Net returns increased from US$ 492 ha~1 yr~1 with fortnightly to US$ 650 ha~1 yr~1 with weekly phytosanitary pod removal (Table 6). Thus, the weekly removal frequency was 32% more pro"table than the fortnightly treatment. In contrast, net returns under the traditional system are an estimated US$ 83 ha~1 yr~1. Time course analysis of losses due to individual pathogens indicated that weekly removal yielded signi"cantly fewer numbers of diseased pods during a part of the observation period only. None of the interactions with time were signi"cant. The number of pods releasing conidia of M. roreri was reduced signi"cantly during the rehabilitation period and for a 24-week period between Table 2 Moniliasis: e!ect of fortnightly and weekly removal of diseased pods on disease incidence (% pods infected) and number of sporulating pods (per plant and year) in three cocoa "elds Fortnightly
Weekly
Reduction (%)
35.0# 35.5# 42.5$
21.1! 21.0! 31.6"
40 41 26
Number of sporulating pods A"lador 1 43.5% A"lador 2 31.6# Las Islas 46.2%
26.2" 16.7! 36.5$
40 47 21
Incidence in pods (%) A"lador 1 A"lador 2 Las Islas
!}%Values followed by the same letter do not di!er at P(0.05 (comparison of six cells).
Table 3 Witches' broom: e!ect of fortnightly and weekly removal of diseased pods on disease incidence (% pods infected) in three cocoa "elds Fortnightly Incidence in pods (%) A"lador 1 A"lador 2 Las Islas
7.8# 9.8$ 7.7#
Weekly
6.5" 8.4#$ 3.3!
weeks 33 and 57 (Fig. 4). This period coincides with that of the main pod development from the peak of #owering until the onset of the harvest peak (Fig. 3). No signi"cant di!erence existed for the 24-week period preceding #owering. Somewhat curtailed levels were observed throughout harvest but beyond week 57 this failed to reach signi"cance. The time course for total losses due to moniliasis was parallel to the one for sporulating pods but no signi"cant di!erence between treatments was observed during the rehabilitation period (data not shown). The average proportion of infected pods reaching sporulation was 69.6 and 59.6% for fortnightly and weekly removal, respectively. A similar trend was found for Phytophthora: signi"cant di!erences were found for a 24 week period but slightly later, from weeks 37 to 61 (Fig. 6). The pattern of losses due to C. perniciosa was more complex. The "rst 9-week peak (Peak I, Fig. 5) and
Table 4 Black pod: e!ect of fortnightly and weekly removal of diseased pods on disease incidence (% pods infected) in three cocoa "elds
Incidence in pods (%) A"lador 1 A"lador 2 Las Islas
Fortnightly
Weekly
Reduction (%)
5.5# 7.6$ 4.1"
3.6" 4.2" 1.4!
35 45 66
!}$Values followed by the same letter do not di!er at P(0.05 (comparison of six cells).
Table 5 E!ect of fortnightly and weekly removal of diseased pods on yield (kg ha~1 yr~1) in three cocoa "elds
Reduction (%)
17 14 57
!}$Values followed by the same letter do not di!er at P(0.05 (comparison of six cells).
Yield (kg ha~1 yr~1)
Fortnightly
Weekly
Increase (%)
A"lador 1 A"lador 2 Las Islas
597.9# 349.3! 565.8#
791.7$ 474.3" 712.6$
32 36 26
!}$Values followed by the same letter do not di!er at P(0.05 (comparison of six cells).
Table 6 Production costs and net returns under traditional, fortnightly and weekly phytosanitary pod removal (US$ yr~1) given a farm-gate price for dry cocoa beans of US$ 1.43 kg~1
Traditional (no removal)! Fortnightly removal of diseased pods Weekly removal of diseased pods
Production cost ($ ha~1 yr~1)
Gross returns ($ ha ~1 yr~1)
Net returns ($ ha ~1 yr~1)
Bene"t:cost ratio
125 229 293
208 721 943
83 492 650
1.7 3.1 3.2
!Production costs for the traditional system were calculated according to CeH sare (1986) using up-dated prices. Returns of 145.5 kg ha~1 yr~1 for the traditional system were derived from data of the previous season (up to week 18).
W. Soberanis et al. / Crop Protection 18 (1999) 677}685
the last 9 weeks going into harvest (Peak IIb), failed to reach signi"cance. During the 14-week period from week 43 to 57 (Peak IIa), the di!erence was signi"cant. Overall, higher numbers of healthy pods were harvested for the majority of the year under weekly disease control (Fig. 8). The only consistent exception was the 16-week period following #owering (for comparison see Fig. 3).
4. Discussion Moniliasis, which appeared and spread in Peru during the 1990s (Evans et al., 1998), is the most damaging cocoa disease in Peru and losses of 100% have been recorded (Servicio Nacional de Sanidad Agraria, 1998). This disease only a!ects fruit; young pods are most susceptible. Although symptom development depends on the age of the pod at infection, the result of infection is almost invariable pod loss unless the pods are attacked close to maturity (Evans, 1981; Evans et al., 1977). The same applies to C. perniciosa but to a lesser extent to Phytophthora. Thus, in Peru, C. perniciosa, M. roreri and Phytophthora spp. are the most important pathogens: being very common (Table 1) and causing direct pod loss. The trials were conducted in a typical year. Under these conditions which are typical for South-American cocoa-growing regions (Alvim, 1981; HernaH ndez, 1982; Maddison et al., 1995), a typical production cycle was observed. Yield losses coincided with the patterns reported in the literature (Evans, 1981; Evans et al., 1977; HernaH ndez, 1982, 1991; Porras and SaH nchez, 1991; Purdy and Schmidt, 1996; Rudgard et al., 1993; Thorold, 1967). Present recommendations on cultural disease control are still poorly de"ned. In Peru, the maximum removal frequency is fortnightly. However, there is support for the hypothesis that even fortnightly removal is insu$cient (Cubillos and Aranzazu, 1979; Porras et al., 1990). HernaH ndez (1991) recommended fortnightly pod removal if losses due to moniliasis are less than 15%/yr, otherwise weekly. Porras and SaH nchez (1991) generally recommend at least weekly pod removal for both, M. roreri and Phytophthora spp. For Phytophthora spp., removal as frequently as every day or every other day has been suggested but this proved to be uneconomical. No di!erences were found between weekly and fortnightly removal but monthly removal was worst (Thorold, 1967). Concrete recommendations for the control of witches' broom are even more limited. In their review, Purdy and Schmidt (1996) state that despite a multitude of studies, critical epidemiological parameters, including latent infection period, are inadequately de"ned and highly variable. Known inoculum sources are still poorly correlated with disease occurrence. They concluded that meaningful control recommendations have to be designed at a regional or even a plantation level. In general terms, they agreed with the current recommendation for Peru (HernaH ndez,
683
1991) of the complete removal of all diseased tissue at the end of the dry season followed by a second round 4}5 months later. However, HernaH ndez (1991) suggested leaving pruned material on the plantation #oor, as was done in our study according to local practice, whereas Purdy and Schmidt (1996) assert that its destruction is essential. Whether or not destruction is economically bene"cial depends on the likelihood of basidiocarp formation on the ground which is in#uenced by decomposition rates and thus local conditions (Rudgard et al., 1993). Studies on phytosanitary pruning against witches' broom concentrate on the complete removal of infected tissue: fruits and all vegetative brooms which is a very labour-intensive operation (Rudgard et al., 1993). Pod infection is least a!ected by this type of pruning (Evans and Bastos, 1981). Ever since Evans and SoloH rzano (1982) suggested that basidiospores of C. perniciosa can travel at least 50}70 km, it is unlikely that phytosanitary pruning of this kind will be e!ective unless carried out in a synchronised manner over a large area. However, this is impractical on smallholdings in Peru which are typical for the majority of Latin American cocoagrowing areas. Thus, our objective was to recommend a control measure which individual smallholders could implement under high disease pressure from external sources. Table 6 and Fig. 3 suggests that both removal frequencies improved yield by approximately a factor of four but season to season variation probably played a part too. This is a conservative estimate because rehabilitation had begun 12 weeks earlier. No direct measurements were available because traditionally managed sub-plots which we had included as controls in our trials were rehabilitated by all three growers before the 1996 harvest peak. However, data collection from these plots were discontinued because the onset and diligence of rehabilitation varied between farms and rendered a meaningful comparison impossible. For moniliasis and black pod, weekly removal of diseased pods reduced disease losses signi"cantly compared with fortnightly removal, independent of the "eld. Additionally, the number of sporulating, Moniliophthora-infected pods, the source of inoculum of the pathogen, was lowered consistently and immediately after rehabilitation began. Thus, weekly removal increased yields and reduced the risk of dissemination of inoculum which may have long-term bene"ts which were not assessed here. M. roreri has a latent period of two to three months during which the pods appear healthy but internal symptom are well advanced. After the appearance of external symptoms, M. roreri produces vast numbers of conidia within one week (Evans et al., 1977, Evans, 1981). This rapid time-course has been corroborated in Peru (AreH valo, 1992; HernaH ndez, 1991). Conidia are the only known infective propagules of M. roreri and any physical agitation (rain, wind, interference by animals and man) is
684
W. Soberanis et al. / Crop Protection 18 (1999) 677}685
su$cient to dislodge them (Evans, 1981). Pods remaining suspended in the canopy can release viable conidia for up to nine months (Evans et al., 1977). In this manner, mature pods can release seven billion spores (44]106 cm~2) which could be detected at 1 km from nearest cocoa "eld, whereas sporulation comes to a halt within one to two months if the pods are on the ground (Evans, 1981). HernaH ndez (1991) states that man is an important dissemination agent, if pods are removed when already sporulating, and he recommended the use of bags to cover pods and to leave them on the soil surface covered by leaf litter. This recommendation, however, has not been adopted by farmers although it appears obvious that pods should be removed before they sporulate. Since sporulation commences within one week of lesion development and external symptoms prior to that can be inconspicuous, it seems most logical to remove pods at weekly intervals for this disease. For witches' broom control, the degree of short-term response di!ered between "elds. This is not surprising given the long latent phase of several months of this pathogen. However, likely medium to long term bene"ts at least in some "elds seems to justify weekly removal especially where other important diseases are controlled at the same time as was the case in Tingo MarmH a. We do not envisage that weekly pod removal will substitute overall phytosanitary pruning of vegetative brooms. The latter was not investigated here but the two peaks of vegetative broom formation observed in Fig. 7, corroborate the recommendation of HernaH ndez (1991) and Rudgard et al. (1993). Instead, our recommendation should be seen as an additional means of reducing losses due to witches' broom particularly suited to smallholders in high-inoculum areas. M. roreri and C. perniciosa cannot be distinguished during early infection by "eld symptoms alone (Evans et al., 1977) and, for practical reasons, Ram and AreH valo (1997) recommended an integrated approach to all major diseases. The increase in yield between 26 and 36% represents the cumulative bene"t between weekly and fortnightly removal in such an integrated approach. However, signi"cantly more healthy pods in the weekly treatment were already measured during the "rst weeks of rehabilitation (Fig. 8) before a treatment e!ect would be expected given the long latent periods of M. roreri and especially C. perniciosa. In contrast, a rapid response of black pod is not surprising. This may have been due to the tendency of both moniliasis (data not shown) and black pod incidence (Fig. 6) to be somewhat more pronounced with the fortnightly removal although neither was signi"cant. If weekly pod removal were implemented with similar success on 21.4% of the area under cocoa, it would provide self-su$ciency of cocoa for Peru. The additional cost of weekly pod pruning of US$ 64 ha ~1 yr~1 was outweighed by increased returns. Thus, on-farm, an extra US$ 158 ha ~1 yr~1 could be earned. The "ndings of this study were convincing not
only for participating farmers who eliminated the `traditionala control treatment but also for a group of promoter farmers in the Tocache area of the Upper Huallaga Valley (Fig. 1) which faces problems similar to those of Tingo MarmH a. Preliminary results in 1997/98 indicate that weekly removal of diseased pods is not only an acceptable and implementable recommendation but also that the incidence of moniliasis was reduced to an average of 20% of trees on smallholdings which adopted this technology. From the time-course of individual diseases, it is obvious that weekly removal is bene"cial from #owering onwards up to the onset of the harvest peak, and possibly extending throughout it. During a 16-week period (weeks 30}46), which started with the increase of #owering, the number of healthy pods harvested did not di!er between treatments (Fig. 8). Is it, then, necessary to extend weekly pod removal throughout the whole year? Pods harvested during the period weeks 30}46 would have been initiated approximately 30 weeks earlier, i.e. in weeks 0}16. Of this period, weeks 0}12 fall into the previous season's pre-harvest phase when weekly removal should be practised. During the remaining four weeks in question (weeks 13}16), only two removal operations would be saved by reverting temporarily to fortnightly removal. The saving in terms of labour seems negligible in comparison with the risk that infected pods remaining in the canopy could sporulate and act as a carry-over inoculum source between seasons. Although the di!erence failed to reach statistical signi"cance, Figs. 4}6 suggest that this is possible. Therefore, we recommend weekly removal of diseased pods throughout the year. This is also a clearer message to farmers and may prevent them from making discretional cut-backs of the sanitation frequency. This risk exists mostly between the increase of #owering (week 30) and week 46 when the grower does not experience increased yield levels which could motivate diligent husbandry. From Figs. 4}6, however, it is clear that this is a crucial period for disease control. Whether the recommendation of weekly removal throughout the year holds in areas with di!erent disease distributions and/or a shorter rainy season remains to be investigated.
Acknowledgements This research was jointly funded by the United States Department of Agriculture}Agricultural Research Service (USDA-ARS), and Fondo Contravalor Peru Canada and managed by CABI Bioscience, the Centro de Investigacio& n y Desarrollo en la Alto Huallaga (CIDH) and the Convenio UNAS-ARS-NAS. A starting grant from the Organisation of the American States (CICAD/OAS) was received. The disease survey was assisted by the Asociacio& n Agrn& cola de Avanzada, Aucayacu, Cooperativa Agraria Cacaotera ACOPAGRO, JuanjumH , Cooperativa
W. Soberanis et al. / Crop Protection 18 (1999) 677}685
Agraria Tocache, Ltda., Tocache and the Cooperativa Agroindustrial Naranjillo Ltda., Tingo MarmH a, and the following "eld o$cers: ED steban Altamirano, David AreH valo, Je!erson CaH rdenas, Julio Flores, TomaH s Malgarejo, Alejandro PimenteH l, Manuel Pin8 a, Tulio Pisco, Walter Rategui and RoH mulo RodrmH guez. Walter Panduro assisted in the maintenance of trials and Lucinda Vela in editing the original thesis. References Alvim, P. de T., 1981. Recent studies on the environmental physiology of cocoa. In: Proceedings of the Seventh International Cocoa Research Conference, Douala, Cameroon, 5}12 November, 1979, pp. 85}89. AreH valo, E., 1992. Estudio de la Moniliasis del Cacao Causada por Moniliophthora roreri (Cif. & Par.) Evans et al. en la Selva Norte del Peru. M.Sc. Thesis, Universidad Nacional Agraria de la Molina, Lima, Peru. Barros, O., 1982. Avanzes en la represioH n de la moniliasis del cacao. In: Proceedings of the Eighth International Cocoa Research Conference, Cartagena, Colombia, 18}23 October, 1981, pp. 401}405. CeH sare, O., 1986. Manejo del Cultivo del Cacao. FOPEX, Lima, Peru, pp. 59. Cubillos, G., Aranzazu, F., 1979. ComparacioH n de tres frecuencias de remocioH n de frutos enfermos en al control de Monilia roreri Cif. & Par. El Cacaotero Colombiano 8, 27}34. EnrmH quez, G.A., Brenes, O., Delgado, J., 1982. Desarrollo e impacto de la moniliasis del cacao en Costa Rica. In: Proceedings of the Eighth International Cocoa Research Conference, Cartagena, Colombia, 18}23 October, 1981, pp. 375}380. Evans, H.C., 1981. Pod rot of cocoa caused by Moniliophthora roreri (Monilia) roreri. Phytopathological Papers, Vol. 24. Commonwealth Mycological Institute, Kew, UK, pp. 44. Evans, H.C., Bastos, C.N., 1981. Preliminary results of research on witches' broom disease of cocoa (Crinipellis perniciosa) in the Amazonian region of Brazil. In: Proceedings of the Seventh International Cocoa Research Conference, Douala, Cameroon, 5}12 November, 1979, pp. 255}256. Evans, H.C., SoloH rzano, G.R., 1982. Witches' broom disease: wrong experiment right result. In: Proceedings of the Eighth International Cocoa Research Conference, Cartagena, Colombia, 18}23 October, 1981, pp. 415}418. Evans, H.C., Edwards, D.F., RodrmH guez, M., 1977. Research on cocoa diseases in Ecuador: past and present. PANS 23, 68}80. Evans, H.C., Krauss, U., RmH os, R., Zecevich, T., AreH valo, E., 1998. Cocoa in Peru. Cocoa Growers' Bull. 52, 7}22. Fujimori, A., Mun8 ante, S., Masuda, M., Maza y Silupu, S., 1996. Comercio Exterior Agrario 1993}1996 (A Junio). Ministerio de Agricultura, O"cina de InformacioH n Agraria, Lima, Peru. Fujimori, A., Mun8 ante, S., Masuda, M., Maza y Silupu, S., 1997. ProduccioH n Agricola de Principales Cultivos, 1996. Ministerio de Agricultura, O"cina de InformacioH n Agraria, Lima, Peru.
685
Genstat 5 Committee of the Statistics Department of Rothamsted Experimental Station, 1993. Genstat 5, Release 3, Reference Manual. Oxford University Press, New York. HernaH ndez, F.A., 1982. Algunos aspectos de la biologmH a de Crinipellis perniciosa (Stahel) Singer en la regioH n de Uraba (Colombia). In: Proceedings of the Eighth International Cocoa Research Conference, Cartagena, Colombia, 18}23 October, 1981, pp. 323}328. HernaH ndez, T.A., 1991. Cacao * Sistemas de ProduccioH n en la Amazonia Peruana. Programa de PromocioH n Agroindustrial y Desarrollo Rural Alternativo, Proyecto d PromocioH n Agroindustrial AD/ PER/86/459, UNFDAC } UNDP/OSP, Tingo MarmH a, Peru, pp. 70. Maddison, A.C., MacmH a, G., Moreira, C., Arias, R., Neira, R., 1995. Cocoa production in Ecuador in relation to dry-season escape from pod rot by Crinipellis perniciosa and Moniliophthora roreri. Plant Pathol. 44, 982}998. Matlick, B.K., 1998. Machete technology? What small cocoa farmers need! Paper presented at the First International Workshop on Sustainable Cocoa Growing, Smithsonian Migratory Bird Center, Panama City, Republic of Panama, 30 March}02 April, 1998, http://www.si.edu/smbc/matlick.htm. McGregor, A., 1982. Evaluation of "ve systemic fungicides for control of black pod of cocoa. In: Proceedings of the Eighth International Cocoa Research Conference, Cartagena, Colombia, 18}23 October, 1981, pp. 295}299. Porras, V.H., SaH nchez, J.A., 1991. Enfermedades del Cacao. FundacioH n Honduren8 a de InvestigacioH n AgrmH cola (FHIA), Honduras. Porras, V.H., Cruz, C.A., Galindo, J.J., 1990. Manejo integrado de la mazorca negra y la moniliasis del cacao en el troH pico huH medo bajo de Costa Rica. Turrialba 40, 238}245. Purdy, L.H., Schmidt, R.A., 1996. Status of cocoa witches' broom: biology, epidemiology, and management. Ann. Rev. Phytopathol. 34, 573}594. Ram, A., AreH valo, E., 1997. Manejo Integrado para el Control de la Moniliasis del Cacao en el PeruH . Pilot Project: AsesormH a e InvestigacioH n para el Desarrollo Integral Andino-AmazoH nico (AIDIA)GTZ, Lima, Peru. RmH os, R., RodrmH guez, C., 1998. La moniliasis [Moniliophthora roreri (Cif. & Par.) Evans et al.] del cacao en la cuenca del alto Urubamba: primera ocurrencia e impacto econoH mico esperado. Paper presented at the XV Congreso Peruano de FitopatologmH a, Pucallpa, Peru, 27 September}02 October, 1998. Rudgard, S.A., Maddison, A.C., Andebrhan, T., 1993. Summary report of the International Witches' Broom Project (IWBP). Cocoa Growers' Bull. 46 (Special Issue), pp. 32. Servicio Nacional de Sanidad Agraria, 1998. Programa de Control Fitosanitario de la Moniliasis del Cacao. CoordinacioH n SENASA HuaH nuco, Tingo MarmH a, Peru. Thorold, C.A., 1967. Black-pod disease of Theobroma cacao. Rev. Appl. Mycol. 46, 225}237. Trevizan, S.D.P., 1996. Mudanc7 as no sul da BahmH a associadas a` vassoura-de-bruxa do cacau. Paper presented at the 12th International Proceedings of Cocoa Research Conference, Salvador, BahmH a, Brazil, 16}26 November, 1996, in press. Zadocks, J.C., 1997. Disease Resistance Testing in Cocoa. A review prepared on behalf of FAO/INGENIC. Wageningen. The Netherlands, pp. 15.
