The present study was carried out between February 2014 and September 2016 in a 40-year old cacao plantation located in the municipality of Cardenas in the Experimental Field of the Colegio de Postgraduados, Campus Tabasco (17˚ 59' 23.98'' N and 93˚ 37' 10.24'' W), at 24 masl. Tabasco contributes 64% of the total cacao production in Mexico, and Cardenas is the second largest municipality in the state in terms of cacao production [25] .

The climate of the study area is hot-humid, with abundant rain falling in summer [26] . The average annual temperature is 26˚C, with a monthly average maximum of 30.3˚C in May and an average minimum of 20˚C in December and January. The absolute maximum and minimum temperatures reach 40˚C and 10˚C. The average annual rainfall is 2643 mm, with a monthly maximum average of 335 mm in September and a monthly average minimum of 10 mm in April [27] . The annual average relative humidity is 83%, with a maximum of 86% in January and February and a minimum of 77% in May. The highest wind speed (30 km h⁻¹) was recorded in November and December, while the lowest (20 km h⁻¹) was recorded in June [26] .

The chemical analysis of the soil in the plantation showed the following results: pH 5.61; EC, 53.40 µS cm⁻¹; organic matter, 3.25%; total, N 0.14%; Olsen P, 12.96 mg/kg; K, 0.36 Cmol/kg; Ca, 9.13 Cmol/kg; Mg, 2.19 Cmol/kg; Na, 0.08 Cmol/kg; CIC, 15.64 Cmol/kg; Fe, 192.16 mg/kg; Cu, 5.31 mg/kg; Zn, 2.49 mg/kg; Mn, 15.00 mg/kg; clay, 42.20%; silt, 16.24% and sand, 41.56%. The soil is classified as a texturally clayey and silty soil.

The management of the plantation consisted of two weeding during the year. The density of the plantation was 625 trees per hectare, which is the typical plan tation density in the region. The shade trees associated with the plantation under study were mahogany (Swietenia macrophylla King), Spanish cedar (Cedrela odorata L.), beechwood (Gmelina arborea Roxb. Ex Sm.), teak (Tectona grandis L.F.), chipilcohite (Diphysa robinioides Benth.) and jinicuil (Inga jinicuil Schltdl. & Cham. Ex G. Don), which are typical of cacao agroforestry systems in Cárdenas, Tabasco, Mexico [8] . Such trees are established without any arrangement. The plantation was surrounded by other plantations of coconut (Cocos nucifera L.), bananas (Musa paradisiaca L.), cacao and an orchard with different tropical fruits, such as caimito (Chrysophyllum cainito L.), mamey sapote (Pouteria sapota Jacq. H.E. Moore & Stearn), jinicuil (I. jinicuil), soursop (Annona muricata L.) and Malay apple (Syzygium malaccense (L.) Merr. & L. M. Perry).

The present study evaluated the effect of sanitary pruning of cacao and shade trees, and the application of fertilizer with K and NPK. Before the experiment was established, we measured the intensity of sunlight that passed through the shade tree canopy to the cacao tree canopy to ensure the correct pruning shade trees and adjust the light intensity of the plantation to 60%. The instrument used for such measurements was a LI-190SA QUANTUM SENSOR from LI-COR Bioscience^TM (Nebraska, U.S.A.). The pruning of cacao trees consisted of eliminating crossed, dry or damaged branches and removing diseased or damaged fruits and chilillos. The fertilizer was applied in a band at a distance of 1.5 m and a depth of 5 - 10 cm around the cacao trees. The fertilizer sources were urea (46% N), triple calcium superphosphate (46% P₂O₅) and potassium chloride (20% K₂Cl). The dose of K applied was 300 g per tree, and that of NPK was 600 g per tree.

The experimental design was split-plot in a factorial arrangement of treatments 2² × 3. The large plot was shade trees pruning and small plot was cacao trees pruning. The evaluated factors were: pruning of shade trees (with and without pruning), cacao pruning (with and without pruning) and fertilization (without fertilizer, with K, with NPK). There were a total of 12 treatments with three repetitions and four trees per repetition ( Figure 1 ).

The variables number of healthy, diseased and damaged fruits, number of healthy and diseased chilillos, and number of flowers per tree were assessed every 15 days. Healthy fruits were those without the presence of moniliasis and/or black pod (Phytophthora spp.) and ready for harvest. Diseased fruits were those with presence of moniliasis and/or black pod. Damaged fruits were those attacked by squirrels (Sciurus aureogaster F. Cuvier) or birds (Melanerpes aurifrons Wagler). Healthy chilillos were cacao fruits up to two months old without the typical hump caused by the presence of M. roreri. Diseased chilillos were those with the aforementioned hump. The number of flowers was assessed from the base of the stem up to a height of 1.5 m.

For each of the variables evaluated, an analysis of variance (ANOVA) was performed to test statistical differences between the factors studied: pruning of shade trees, pruning of cacao trees, fertilization, and the interactions. The ANOVA was performed on the data at 6, 12, 18 and 24 months after applying the treatments (months). The significance of the interactions and the factors evaluated was recorded at p ≤ 0.05. For significant interactions, a least squares mean test was performed to determine statistically different interactions (p ≤ 0.05). For non-significant interactions, a multiple comparison of means of the factors was carried out using the Tukey test (p ≤ 0.05). The assumption of normality of the ANOVA errors was verified using the Shapiro-Wilk test (p > 0.05); while the assumption of homoscedasticity was verified using the graphs of the predicted values vs. the residuals, whose points presented a random distribution in the plane, which meant homogeneity of variances in the groups. The information was processed using the Proc Mixed Procedure and the slice instruction of the SAS software version 9.4 [28] .

The ANOVA results indicated a significant effect (F = 2.85, p > F = 0.060) of the interaction among the three evaluated factors on the number of healthy fruits at 24 months ( Table 1 ). The best interaction in terms of its effect on the number of healthy fruits was shade tree pruning, no cacao pruning and K fertilization ( Figure 2 ). These results contrast with those obtained by other authors, such as [12] and [29] , who reported a lack of response of cacao trees to fertilization treatments in agroforestry systems. This discrepancy likely stems from competition, primarily among cacao trees, given the higher plant density in the agroforestry systems reported by those authors (1242 plants) compared to our study (625 plants). Furthermore, the age of the cacao trees and the fertilizer dosage and formulation may also contribute.

MAAT, Months after applying the treatments. LP, Large plot (shade trees pruning). SP, Small plot (cocoa pruning). FERT, Fertilization.

At 24 months, the number of healthy fruits was also significantly affected by shade tree pruning, the interaction between cacao and shade tree pruning, and the interaction between cacao pruning and fertilization ( Table 1 ). The mean comparison test indicated that shade tree pruning significantly increased this variable, and that the effects of NPK fertilization and the control treatment without fertilizer were statistically equals ( Table 2 ). This can be attributed to the fact that nutrient content in the soil was enough for the number of fruits that can be produced by cacao trees.

The interaction among the three factors did not have a significant effect on the number of diseased fruits at any of the evaluation dates ( Table 1 ). The factors that had the greatest effect on this variable were cacao pruning at 6 months, shade tree pruning and fertilization at 12 months and shade tree pruning at 18 and 24 months. The greatest effect of the interaction between cacao and shade tree pruning was observed at 6, 18 and 24 months. With the exception of the first evaluation date, shade tree pruning significantly increased the number of diseased fruits on all evaluation dates.

Cacao pruning increased the number of diseased fruits at 6 months, decreased it at 12 and made no significant difference at 18 and 24 months ( Table 2 ). This can be attributed to the fact that cacao pruning is usually an annual practice, although this was not done in the plantation under study. As indicated [30] the damage to fruits due to disease is a direct function of the age of the cacao plantation. The plantation under study was 40 years-old, and the age of maximum production is 28 years [31] . However, the trees responded to fertilization.

The ANOVA results showed a significant effect (F = 4.32, p > F = 0.0148) of the interaction among the three evaluated factors on the number of damaged fruits at 18 months. The main effect of fertilization on this variable was also recorded at this time ( Table 1 ). The mean comparisons indicated that K fertilization significantly decreased the number of damaged fruits ( Table 2 ). Independent of fertilization, the combination of cacao pruning and shade tree pruning decreased the number of damaged fruits ( Figure 3 ). The best interactions to reduce the number of damaged fruits were K fertilization, no shade tree pruning and cacao pruning as well as shade tree pruning, cacao pruning and no fertilization. The above two interaction groups contrasted with the interaction among NPK fertilization, cacao pruning and no shade tree pruning, which had the highest number of damaged fruits ( Figure 3 ).

The damage to cacao fruits caused by squirrels and birds was similar to that by moniliasis [1] . Pruning can thus be used to reduce the effect of the above problems by decreasing the number of rest locations for squirrels and birds, improving the aeration and regulating the humidity in the plantation. Another way to reduce the damage by these animals is to avoid cultivating shade trees that attract the animals [29] , which was not the case in the plantation under study and surrounding plantations.

MAAT, Months after applying the treatments. Variation factors: STP, Shade tree pruning [1) no pruning, 2) pruning]. CP, Cacao pruning [1) no pruning, 2) pruning]. FERT, Fertilization [1) no fertilizer, 2) K, 3) NPK]. NHF, Number of healthy fruits; DF, Diseased fruits; Df, Damaged fruits; HCh, Healthy chilillos; DCh, Diseased chilillos; NFl, Number of flowers. Values with the same letters by date, variation factor and variable are statistically equal (Tukey, p ≤ 0.05).

The ANOVA results showed a significant effect (F = 3.05, p > F = 0.05) of the interaction among the three evaluated factors on the number of healthy chilillos at 18 months. The main effect of cacao pruning and fertilization on this variable was also recorded at that time ( Table 1 ). The mean comparisons did not show statistically significant differences between the cacao pruning treatments or between the shade tree pruning treatments. Significant differences were observed between fertilization treatments. Fertilization with NPK significantly increased the number of healthy chilillos ( Table 2 ). The best interaction to increase the number of healthy chilillos was NPK fertilization, no cacao pruning and no shade tree pruning ( Figure 4 ).

The ANOVA results showed a significant effect of the interaction among the three evaluated factors on the number of diseased chilillos at 18 months (F = 9.51, p > F = 0.0001) and at 24 months (F = 2.89, p > F = 0.0584). On those dates, the main effect of cacao pruning on the number of diseased chilillos was also observed ( Table 1 ) and it led to a significant decrease in this variable ( Table 2 ). The best interactions to decrease the number of diseased chilillos were shade tree pruning, cacao pruning and NPK or K fertilization at 18 months ( Figure 5 ). At 24 months, the best interact tree pruning and NPK or K fertilization ( Figure 5 ).

The absence of an effect of the evaluated treatments on the number of healthy and diseased chilillos during the first year can be attributed to the 9.3 months that the moniliasis epidemic lasts in cacao plantations [4] ). Thus, cacao fruits remain infected throughout the entire cacao production cycle. In other words, eliminating possible inocula (diseased chilillos) in a plantation has a positive effect one year after the activity has ceased.

The reasons why K fertilization reduced diseased chilillos could be several. Potassium deficiency can increase the accumulation of soluble sugars and amino acids, which are nutrients for pathogens. An adequate supply of potassium reduces this accumulation, decreasing the plant's susceptibility to disease [32] . Potassium strengthens the plant’s structural and biochemical defenses, improving its ability to resist pathogen attack. Also, K contributes to the synthesis of lignin and cellulose, essential components of cell walls. Strong cell walls act as a physical barrier, making it difficult for pathogens to penetrate [32] [33] .

The ANOVA results showed a significant effect of the interaction among the three evaluated factors on the number of flowers at 6 months (F = 2.6, p > F = 0.0768), 12 months (F = 2.96, p > F = 0.0544) and at 18 months (F = 7.83, p > F = 0.0006). An effect of the interaction between fertilization and cacao or fertilization and shade tree pruning on the number of flowers was also observed at 6 months ( Table 1 ). The best three-way interaction to increase the number of flowers was cacao pruning, no shade tree pruning and no fertilization. This interaction significantly increased this variable compared to all other interactions ( Figure 6 ). This finding can be attributed to the fact that the nutrient content of the soil, which was determined according to the analysis performed here and in Aikpokpodion [34] , was sufficient to meet the needs of the flowering process. Another explanation is that 60% of incident light on cacao trees was enough to stimulate the flowering process, as indicated by Tezara et al. [35] .

At 12 months, the ANOVA results showed a significant effect of pruning, their interactions, and the interaction between fertilization and each of the pruning

treatments on the number of flowers ( Table 1 ). Shade tree pruning significantly decreased the number of flowers per tree, while cacao pruning and NPK fertilization increased it ( Table 2 ). The best interactions to increase this variable were cacao pruning without shade tree pruning, NPK fertilization without shade tree pruning, and cacao pruning with NPK fertilization (data not tabulated). The best interaction among the three factors to increase the number of flowers was NPK fertilization, cacao pruning and no shade tree pruning ( Figure 6 ). In contrast to [36] , who associated the increase in flowering with increased lighting, the interaction of cacao and shade tree pruning in the present study decreased the number of flowers. Such a result could be attributed to the depletion of plant reserves or photosynthate sources. Additionally, it may be due to the increased stress experienced by the cacao trees because of the increase in incident light. This explanation is consistent with [37] and [38] , who indicated that cacao trees exhibit optimal physiological behavior under low radiation conditions.

At 18 months, the ANOVA results showed a significant effect of the interaction between tree pruning, fertilization and each of the pruning treatments on the number of flowers ( Table 1 ). Shade tree pruning, cacao pruning and K fertilization significantly increased this variable ( Table 2 ). The best interaction between two factors to increase the number of flowers was shade tree pruning and NPK fertilization (data not tabulated). At this time, the best interactions among the three evaluated factors to increase it were the combination of K fertilization, cacao pruning and no shade tree pruning and the combination of NPK fertilization, cacao pruning and shade tree pruning ( Figure 6 ).

Shade tree pruning is economically costly and dangerous [38] ; in addition, the greater amount of incident light on cacao trees increases the requirement of nutrients for the flowering process. Therefore, we would advise an initial pruning, repeating the pruning after one year and fertilizing with NPK at the same time, and then fertilizing with K at 18 months after the first pruning. This advice takes into account that fertilization is not a common practice in cacao cultivation [39] - [41] ; however, when applied correctly and combined with pruning and disease management, it can significantly increase the yield of cacao trees, as reported by Ortiz et al. [4] .