Materials and methods

Study area. This work was carried out in five cacao plantations in the state of Tabasco and the northern region of Chiapas, Mexico. The data was recorded in the 2011-2012 production cycle. The study area is in a hot humid climate with abundant rainfall in summer. The average annual precipitation is 2432 mm, with rainy periods that extend from June to March, and a dry period in April and May. The average annual temperature is 26 °C and Vertisol and Gleysol soils predominate (^{INEGI, 2017}). Data on the geographic location and altitude of each plantation are indicated in Table 1.

Characteristics of the plantations. The study was carried out in traditional commercial cacao plantations, made up of trinitarian-type hybrids, susceptible to BPR. The plantations were 20 to 25 years old and had a density of 784 plants ha-1 with an approximate topological arrangement of 3.5 × 3.5 m. In each plantation, an experimental plot of 180 trees was delimited by a rectangular area of 10 × 18 cocoa trees. A total of 48 trees were evaluated in the 4 × 12 central section. Plantation management consisted of the mechanical removal of basal vegetative shoots and mechanical weed control.

Fruit register. Fruits greater than 7 cm in length, including snappers, green, and ripe fruits, were recorded weekly for one year to determine the fluctuation of the flow of potential fruits to infection by P. capsici and to relate the incidence of disease.

Incidence of BPR and FPR of cocoa evaluation. In each plantation, all the fruits from 8 to 10 cm long were labeled, coming from the flows of flowering, and mooring of fruits that appeared during the evaluation year. The number of fruits was estimated through a census of 48 trees. Each fruit was observed weekly in situ to detect symptoms and signs of BPR, which consisted of dark brown circular spots with a regular border, and a thin layer of mycelium with the appearance of whitish cotton. On the other hand, to evaluate the competitive potential of the infection of the fungus that causes FPR on the incidence of BPR, the number of fruits diseased by FPR, whose symptoms were brown spot (chocolate spot) with irregular edges and abundant sporulation, was recorded weekly according to ^{Torres-de la Cruz et al. (2020}). To estimate the incidence of each disease, the number of diseased fruits was recorded among the total number of fruits evaluated per 100.

Temporal analysis of the progress of cocoa black pod rot. The temporal analysis of the BPR was performed according to ^{Torres-de la Cruz et al. (2020}). For this, graphs of the temporal progress of the incidence percentage, accumulated and not accumulated, were generated as a function of time for each plot. With the accumulated plot progress, the initial incidence (Y ₀ ), area under the disease progress curve (AUDPC), apparent infection rate (r) with the Gompertz model in its non-linear form, and final accumulated incidence (Y _f ) were estimated (^{Torres-de la Cruz et al., 2020}; ^{Campbell and Madden, 1990}). Additionally, the monomolecular and logistic epidemiological models were adjusted to analyze the possible monocyclic and polycyclic effects of the infection. The incidence values ​​were multiplied by four to scale the epidemic curve and allow adjustment to the epidemiological models since Y _f was less than 10%.

Climatic variables. Temperature and relative humidity were recorded per plantation at 2-h intervals using Hobo H8® sensors (Onset Instruments, Pocasset, MA, USA), installed 2 m above the center of the plot. Data was collected weekly. Accumulated monthly precipitation data for 2011 and 2012 were obtained from records of stations closest to the studied plantations: Poblado C-28 (Station 27078), Paraíso (Station 27034), Tulipán (Station 27051), INIFAP (Station 27095), Platanar (Station 07130).

Statistical analysis. The calculations of the apparent infection rate (r), AUDPC and the adjustment of the incidence data to epidemiological models were carried out with the NLIN, SUMMARY, and GLM procedures of SAS® (^{SAS Institute, 2004}). The determination coefficient was used as a goodness-of-fit criterion. The comparison between plantations was made with each one of the environmental variables by means of ANOVA (p = 0.5). With a weekly lag due to the average incubation period, the absolute incidence of BPR was correlated with the weekly average temperature, weekly average relative humidity, and the number of fruits. Similarly, the absolute incidence was correlated with lagged intervals of temperature and relative humidity for which the number of hours/week of temperature and relative humidity in the following intervals was calculated: temperature, a) < 19.9 °C; b) 20-26.9 °C; c) 27-29.9 °C; d) 30-34.9 °C, and e) > 35 °C; and relative humidity, a) < 59.9 %, b) 60-90 %, and c) > 90 %. The accumulated monthly incidence was correlated with the accumulated monthly precipitation, without lag, and the correlation between the final incidence of BPR and the accumulated precipitation during the months of epidemic progression (September 2011-February 2012) was analyzed. To assess the potential for competition between the causative agents of FPR and BPR, peaks of FPR with a delay of two to three weeks, during October-January, were correlated with respect to BPR peaks. Correlations were analyzed with Spearman’s coefficient (Rho) using IBM SPSS® 22.0 software.

Results and discussion

Temporal behavior of cacao black pod rot. Five BPR epidemics were generated, one for each locality plot studied (Figure 1). At the Moctezuma, La Piedra, and Poblado C-28 sites, BPR occurred between 15 and 18 weeks. In Paredón and Platanar, the BPR was extended for 25 weeks. All the epidemics in general exhibited a sigmoidal curve with slight inter-sigmoidal events. The epidemics of Moctezuma, Paredón, and Platanar had a prolonged pre-exponential phase between August 30 and October 15. Atypical sigmoidal conformation BPR epidemics have also been reported for P. megakarya in Cameroon (^{Ndoumbe-Nkeng et al., 2017}). These atypical, inter-sigmoidal increases may be due to new foci of infection (Ndoumbe-Nkeng et al., 2017; ^{Ristaino, 2000}). While ^{Torres-de la Cruz et al. (2020}) attributed them to variations in the fruiting rate with intermittent infection events in the case of Moniliophthora roreri.

Figure 1 Epidemic curves of the cumulative temporal progress of cacao black pod rot incidence, caused by Phytophthora capsici, in five commercial plantations with traditional technology in Tabasco and northern Chiapas, Mexico. Productive cycle 2011-2012. 

The final incidence (Y _f ), adjusted by the number of fruits produced during the disease period, fluctuated from 6.3 to 24.8% with an average incidence of 12.5% ​​(Table 2). The lowest Y _f occurred in the Moctezuma plantation, while the highest was obtained in Platanar (Table 2). These results represent a low intensity compared to other regions in Africa with the occurrence of P. megakarya between low and high incidence (1.15 to 70%), and low and moderate (16 to 40%) depending on the annual rainfall (^{Ndoumbe-Nkeng et al., 2009}). The dependence of Phytophthora spp. precipitation for high infection rates is well documented. However, the intensity of BPR incidence may also depend on the variety of the cacao plant, sanitary management, Phytophthora species involved, and plantation cultural management (^{Akrofi, 2015}; ^{Montes-Belmont and de Los Santos, 1989}).

The intensity values ​​of the epidemics, estimated with AUDPC, fluctuated between 141.9 and 2,416.8 (Table 2). The plot with the highest AUDPC was Platanar (2,416.8) and the lowest was obtained in Moctezuma (141.9). However, the high AUDPC of the first did not correspond to the highest infection rate (r = 0.016) due to its prolonged epidemic compared to Poblado C-28 (r = 0.041) (Table 2)

The epidemic curves in all the evaluated plantations had the best goodness of fit to the Gompertz model (Table 2). According to ^{Campbell and Madden (1990}), the Gompertz model describes polycyclic epidemics, with asymmetric sigmoidal progress curves. In a polycyclic epidemic, the causative agent has the capacity to produce several cycles of infection, which allows intermittent epidemic increases. In previous work with M. roreri, polycyclic events were also shown to be favored by the permanence of infected fruits on the tree, which provide inoculum for continuous infections if there are favorable environmental conditions for the development of the disease and fruit production flows (^{Torres-de la Cruz et al., 2020}).

The characterization of BPR as a polycyclic epidemic may have implications for the establishment of management strategies aimed at reducing the secondary inoculum that delays or reduces the intensity of the epidemic rate (^{Torres-de la Cruz et al., 2020}; ^{Campbell and Madden, 1990}). In this case, the periodic removal of diseased fruits can have a negative impact on the incidence of BPR. In BPR epidemics caused by P. megakarya, it has been reported that the removal of diseased fruits can reduce the incidence by up to 30% (^{Ndoumbe-Nkeng, 2004}). The frequency of removal of inoculum sources is decisive. Weekly removals were more effective than biweekly, with a reduction of up to 66% in incidence (^{Soberanis et al., 1999}). Through numerical simulations, it was found that the optimal removal interval was 4 d (^{Nembot et al., 2017}). Additionally, the application of effective fungicides can contribute to the reduction of the primary and secondary inoculum. For example, copper sulfate and Bordeaux broth have been widely used in Mexico for the control of BPR (^{Torres-de la Cruz et al., 2019}).

Fruit production and absolute change in BPR and FPR incidence. Fruit density, from 7 cm to ripe fruits, showed a similar temporal fluctuation in all evaluated plantations. The presence of fruits of different ages, although quantitatively variable, was maintained throughout the evaluation year. The highest production of total fruits was concentrated in the months of August to December (28 fruits tree^-1) with a productive peak in October - November (Figure 2), a period that coincided with the exponential epidemic phase in all locations-plantations (Figure 1). Statistically, the positive correlation of absolute incidence with the intensity of fruiting, with a lag of one week (Rho =0.52 - 0.74), was demonstrated (Table 3). This dependence on specific host tissue availability had already been reported in Brazil for BPR (^{Medeiros, 1967}). Similarly, ^{Torres-de la Cruz et al. (2020}) also reported a close dependence between the incidence of FPR with the density of fruits for the conditions of Mexico, which raised the hypothesis of competition between these pathogens. The absolute increase in BPR and FPR exhibited multiple incidence peaks with greater intensity and frequency in the latter mentioned (Figure 2). Clearly, FPR preceded BPR causative infection events starting between July-August vs September-October for BPR. Due to the occurrence of fruits throughout the year, the contrasting epidemic behavior between and within each disease is due to the parasitic capacity of the pathogens in relation to the local environment. Variable peaks of P. megakarya infection due to rainfall intensity were reported in Ghana (^{Dakwa, 1973}), with main peaks in August and October (^{Opoku et al., 2007}; ^{Opoku et al., 2000}). While, in Cameroon, with the same pathogen, epidemics were reported that lasted up to 23 weeks, also depending on the intensity of rain (^{Ndoumbe-Nkeng et al., 2017}). In ^Bahia, Brazil, the highest incidence of BPR (P. citrophthora, P. palmivora, and P. capsici) occurred in the coldest months of the year (June - August) (^{Oliveira and Luz, 2005}), with a maximum peak of incidence three to five months after the start of the epidemic (Medeiros et al., 1969), indicative of more intense epidemics than in this study. The epidemic with the shortest duration occurred in Moctezuma (8 weeks), the location with the lowest annual rainfall (1767 mm year^-1), while Platanar had the longest (22 weeks), coinciding with the highest rainfall (2992 mm year-1) (Figure 1). However, the epidemic intensity of BPR was lower, both in the intensity of peaks and their frequency compared to frosty pod rot, which demonstrates the greater parasitic aptitude of M. roreri for the conditions of southeastern Mexico. This was evidenced by the significant negative correlation between the absolute changes of these two diseases (Rho = -0.48 - -0.61) (Figure 2, Table 3). M. roreri is a superior fungus that could have evolved with greater infectious and environmental plasticity, conferring greater competitive aptitude on P. capcisi, at least at the fruit level, restricting the incidence of BPR to levels below 24%, especially when the FPR reaches its necrotrophic stage, generating an enzymatic and tissue environment unsuitable for oomycete infection.

Figure 2 Fluctuation of cacao fruits (greater than 7 cm - ripe fruits) and absolute incidence of cacao black pod rot (Phytophthora capsici) and cacao frosty pod rot (Moniliophthora roreri), in five traditional commercial plantations in Tabasco and northern Chiapas, Mexico. Productive cycle May 2011-May 2012. 

Correlative analysis between climatic factors and BPR. The temperature and relative humidity showed similar behavior in all the plantations, without significant differences between sites (p>0.05). However, in Platanar and Paredón, the temperature had the lowest weekly averages (25 and 24.6 °C, respectively) than the rest of the plots. On the other hand, relative humidity had the greatest inter-plot variation. However, no significant differences were found between sites (p>0.05). By averaging the temperature and relative humidity of all the sampled sites, at weekly intervals, two periods of temperatures greater than 25 °C were evidenced. The first period was presented from May 25 to October 12 (2011). The second period was recorded from March 14 to May 9 (2012). The longest period of relative humidity above 90% occurred from July to January (Figure 1).

The accumulated monthly precipitation also exhibited a similar pattern in all the sites without significant differences (p>0.05). October had the highest accumulated monthly precipitation (595 mm) with a regional average of 2596 mm yr^-1. Total annual precipitation fluctuated from 1767 to 3076 mm yr^-1, being less in Moctezuma, near the coast of the Gulf of Mexico, and greater in Platanar (Figure 1).

The incidence of BPR was positively correlated with moderate values ​​with temperature periods lower than 19.9 °C (Rho =0.43 - 0.63) and with temperature periods in the range of 20 - 26.9 °C (Rho = 0.38 - 0.81) (Table 4). Temperatures in the range of 27 to 29.9 °C showed a negative correlation (Rho= -0.53 - -0.80) (Table 4). Two sporangia germination mechanisms of P. palmivora have been demonstrated due to the temperature. Direct germination with temperatures close to 25 °C, rupture of the sporangium, release, and germination of zoospores at 15 - 20 °C (^{Erwin and Ribeiro, 1996}). P. capcisi, it is possible that the predominant free humidity in October, associated with the increase in rainfall, and with temperatures close to 25 °C activate the direct germination of sporangia constituting the primary inoculum, while temperatures below this threshold, in humid conditions it can be responsible for the rupture of sporangia to produce secondary inoculum.

In this work, although the relative humidity, in general (Rho = 0.47 - 0.60) and the higher humidity 90% was moderately correlated with the incidence (Rho = 0.41- 0.77) (Table 3 and 4), in agreement with other works (^{Akrofi 2015}; ^{Deberdt et al., 2008}; ^{Oliveira and Luz, 2005}; ^{Dakwa, 1973}), this factor may actually be an estimator of saturation moisture on plant tissues, which is determinant for the motor function of flagellate zoospores. Consequently, as expected, the accumulated monthly precipitation and the precipitation associated with the period with BPR incidence (September 2011-February 2012) had a moderate correlation with absolute changes in disease (Rho = 0.34 - 0.76, and Rho = 0.68, respectively) (Table 3 and 4). In Africa, P. megakarya caused 70.3% BPR incidence in sites with rainfall greater than 2200 mm, and 1.15% in those with less than 800 mm (^{Ndoumbe-Nkeng et al., 2009}).

In all plantations, BPR had its exponential epidemic phase in October, coinciding with the highest monthly precipitation (452-780 mm) (Figure 1), and with the first period of average temperature above 25 °C. However, the longer duration of the epidemic process in Platanar, associated with the highest rainfall (2992 mm year^-1), suggests that the temperature is less restrictive for the infection. The free, superficial humidity on the tissues, more than the relative humidity, is determinant for the infection of the fungus, which is congruent with its reproductive physiology and dissemination. This coincides with several works that associate precipitation with the occurrence of the disease (^{Ndoumbe-Nkeng et al., 2009}; ^{Dakwa, 1973}; ^{Medeiros, 1967}).

The correlative effects of absolute BPR incidence with temperature and humidity were significant with a weekly lag between climatic induction and symptom onset. This statistical association suggests the existence of an average incubation period of one week for P. capsici in Trinitario cacao for Mexican conditions. In P. megakarya, the BPR incubation period was estimated to be six days in Amelonado cacao trees (^{Ndoumbe-Nkeng et al., 2009}; ^{Adebayo et al., 1981}).

Conclusions

The analysis of the epidemic structure of black pot rot (BPR) in commercial Trinitario cacao plantations with traditional technology in six locations in southeastern Mexico showed that this disease has a short and low-intensity epidemic process restricted to the September-February period. The Y _f ranged from 6 to 24%, with an average incidence of 12.4%. The highest incidence occurred in Platanar, Pichucalco, Chiapas, and the lowest in Moctezuma, Paraíso, Tabasco. The exponential epidemic phase occurred in October, coinciding with the highest intensity of rainfall (452-780 mm), so this factor could be decisive. All epidemics fit the Gompertz model (r = 0.059 - 0.123, R² 0.97-0.99). The primary inoculum was favored by precipitation, relative humidity greater than 90%, and temperatures from 20 to 26.9 °C. Temperatures below 20 °C also contributed to the polycyclic progress of the disease, possibly due to the secondary inoculum produced by indirect germination of sporangia. Despite the availability of susceptible tissue (i.e., fruits) throughout the production process, P. capcisi had less parasitic aptitude than M. roreri, which induced earlier epidemics, with greater duration and intensity. The characterization of BPR as a polycyclic epidemic justifies management strategies aimed at reducing the secondary inoculum to restrict the epidemic rate. Etiological studies are recommended to determine the status of the causative agent(s) of BPR.