Cocoa (Theobroma cacao L.) is a crop from South America and was brought to Central America in pre-Columbian times (Montemayor et al., 2002). There are an estimated 6 to 7 million cocoa farmers worldwide, and 80-90% are smallholder farmers, with production areas between 2 and 4 ha, from tropical regions of Africa, Asia, Indonesia and Central and South America (Bermúdez et al., 2022).

One of the aspects that affects cocoa cultivation commercialization is the presence of high concentrations of heavy metals (including Cd). This damages cocoa derivative quality, affecting 27% of exports (Huamaní-Yupanqui et al., 2012). Cd is generally adsorbed by roots in the soil and transported to the whole plant through the vascular system by essential cations, including Ca, Fe, and Zn (Huamaní-Yupanqui et al., 2012). High Cd concentrations in beans limit agricultural product marketing (Bravo Realpe et al., 2014).

Total Cd concentrations of geogenic origin in soil generally do not exceed 1 mg kg^-1. However, in some cases levels of up to 16.3 mg kg^-1 Cd have been found, which can be associated with weathering processes and the type of parental material, with a natural base level that would not exceed 0.5 mg kg^-1 (Kabata-Pendias, 2010).

There are many factors that influence Cd concentration in plants. Soil properties, such as soil starting material, soil organic matter (SOM), pH, carbonates, oxides and hydroxides of Fe and Mn, mineralogy of clays, salinity, anthropogenic contamination and plant genotype, affect the bioaccumulation and transfer of heavy metals to the cocoa bean (Alloway, 2013). In turn, these factors can allow heavy metal dissolution, leaching and solubility (Alloway, 2013).

The concentration of total Cd in the soil is not always a reliable indicator for predicting the concentration of Cd in the cocoa bean, as only one portion is bioavailable to the plant (Shahid et al., 2016). Alternatively, Cd and Zn in the soil present synergy when both metals have high concentrations; however, if one metal has higher concentration than the other, it is not absorbed by the plant (Huamaní-Yapanqui et al., 2012; Alloway, 2013). The total Cd concentration in the soil can be transferred to the plant if the soil pH is very acidic or the SOM is low (Rábago & Aracil, 2011).

The Cd in the plant is mobilized and transferred to the leaves and fruits, then concentrated in the grain, enabling us to verify the correlation between the bioavailable Cd in the soil and Cd in the grain (Argüello et al., 2019; Engbersen et al., 2019).

A high intake of cocoa beans at the maximum concentrations permitted by the Codex Alimentarius (FAO/WHO, 2017) in cocoa bean derivatives may promote the accumulation of Cd in the human body. This may cause serious problems in tissues and organs that are susceptible to it, such as the kidneys and liver (Casteblanco, 2018).

Furcal-Beriguete & Torres-Morales (2020) report that the European Food Safety Authority has not established a limit on Cd concentration in the unfermented or raw bean, but there are studies that show that Cd concentrations in the fermented and dried cocoa bean reach levels higher than established. The maximum limit for chocolate and other products with cocoa content between 50 and 70% is 0.6 mg kg^-1, while chocolates and products with declared cocoa content above 70% the maximum level is 0.8 mg kg^-1.

Cocoa cultivation in Panama presents high Cd concentrations in the bean, reducing its quality, price and marketing possibilities. Unlike other countries such as Peru (Huamaní-Yupanquí et al., 2012), Costa Rica (Florida, 2021) and Ecuador (Barraza et al., 2017), in Panama there are no studies on the relationships between soil Cd and its presence in cocoa beans, despite the fact that moderate concentrations of the metal are usually detected (personal communication with members of the Cooperative of multiple services of cacao Bocatoreño R.L).

It is pertinent to note that certain agronomic management such as the application of organic fertilizers, biochar, beneficial microorganisms and liming can also reduce the absorption of Cd and other heavy metals in cocoa (Dávila-Zamora et al., 2020). But cocoa cultivation areas in Panama do not maintain an adequate fertilization program.

This paper evaluates the concentration of Cd in cocoa plants, especially in the bean, and its relationship with the available Cd and other soil properties of small producer’s fields in the province of Bocas del Toro, Panama. Our objective was to calculate Cd concentration in cocoa plants and evaluate its relationship with available Cd and other soil properties.

In Table 1, we present the limits of detection and instrumental quantification for the eight elements quantified on the atomic absorption spectrophotometer equipment, which met the requirements for precision and accuracy. By using it, the results were obtained with the necessary level of reliability. Additionally, summary measurements and 95% confidence intervals were used for soil mineral content and physical properties extracted at a depth of 30 cm from 21 evaluated plots (Table 2).

The soils showed high acidity, with a mean pH of 5.1 and a 95% confidence interval of 4.9-5.3. SOM levels were moderate, at 5.1%. There was an average concentration of 0.34 % of total nitrogen, with minimum and maximum values of 0.1 and 0.6 %, respectively. Among the sites sampled for the C: N ratio, the mean value was 9.9, with minimum levels of 4.5 and maximum levels of 26.

A value in the mean of the ECEC of 19.9 cmol₍₊₎ kg^-1 (Table 2) was observed, this level being high in Inceptisol soils. In the same way, the elements that are part of the ECEC, K, Ca and Mg, obtained high concentrations with means of 0.20, 14.1 and 5.0 cmol₍₊₎ kg^-1, respectively. The average concentrations of exchangeable Al had a moderate value of 0.7 cmol₍₊₎ kg^-1, with a low level of Al. Sat. 3.9%, being able to find in the plots, minimum and maximum concentrations of 0.6 and 13.9%, respectively.

The micronutrients in the soil presented, for the concentration of Cu, Zn, Fe and Mn extracted by Mehlich 3, values of 2.5, 3.6, 101.9 and 177.1 mg kg^-1, respectively. Soil concentrations of total Cd and bioavailable Cd (Cd_bd) yielded mean values of 0.10 and 0.02 mg kg^-1, respectively, whose 95% confidence intervals were 0.07-0.14 and 0.02-0.03 mg kg^-1, respectively.

With respect to soil texture, the largest fraction corresponded to sand particles (57.8%), followed by silt (28.9%) and clay (13.3%).

A description of the Cd concentration in cocoa leaves, fruits, and beans is presented in Table 3 as well as 95% confidence intervals. A minimum and maximum limit is also provided in Fig. 2. The average concentration of Cd in leaves (Cd_leaves) was found to be 1.04 mg kg^-1 with a minimum and maximum concentration of 0.13-4.40 mg kg^-1. The concentration of Cd in fruits (Cd_fruits) was 0.34 mg kg^-1 with a minimum concentration of 0.09 mg kg^-1 and maximum of 0.76 mg kg^-1. The concentration of Cd in cocoa beans (Cd_beans) was 0.25 mg kg^-1, with minimum and maximum levels of 0.01-0.95 mg kg^-1 respectively with lower limits of 0.15 mg kg^-1 and upper limits of 0.34 mg kg^-1.

Pearson’s correlations were calculated to evaluate the relationship between Cd_beans concentration and soil variables (Table 4). Pearson’s correlations between the concentration variables of Cd_bd, Cd_leaves and Cd_fruits were also calculated (Table 5).

Cd_beans (Table 4) was positively correlated with total Cd and Cd_bd (r = 0.66 and r = 0.50 respectively; p<0.05). A positive correlation was also found with the percentage of slime and the content of Fe (r = 0.47 in both; p<0.05). The content of organic matter and Zn maintained a negative correlation with Cd in the cocoa bean with values of (r = -0.50; p<0.05) and (r = -0.38; p<0.10) respectively.

There was no significant correlation between Cd_leaves and C_bd (p=0.1169, Table 5). The concentration of Cd_leaves presented a positive correlation with Cd_beans (r = 0.84, p<0.0001) and with Cd_fruits (r = 0.57, p = 0.0075). In turn, the Cd_fruits correlate positively with the Cd_beans (r = 0.64, p=0.0016).

The relationship between the concentration of bioavailable Cd (Cd_bd) and the physicochemical characteristics of the soil could be modeled with five variables according to the backward selection method (Table 6a ), where the adjusted R² of the multiple regression model was 0.54. The resulting model was: Cd_bd = 0.16 + 5.4 × 10^-4 Silt (%) - 8.9 × 10^-4 Clay (%) - 0.03 pH + 1.9 × 10^-3 Zn + 1.2 × 10^-4 Fe.

The concentration of Cd_bd is mostly influenced by soil pH, due to the high predictive coefficient (Cp) of Mallows (24.53); the lower the soil pH there is more Cd_bd. The percentage content of clay follows in predictive value to the pH; the lower the percentage of clay, the higher the concentration of Cd_bd. The opposite occurs with the percentage of silt, the higher the percentage more Cd_bd. Zn and Fe complete the set of significant variables to explain Cd_bd; in this case the higher the Fe and Zn, the higher the content of Cd_bd.

The relationship between cadmium in leaves (Cd_leaves) and the physicochemical characteristics of the soil could be modeled with five variables according to the backward method (Table 6b ). The adjusted R² of the model was 0.47.

The estimated multiple regression model was: Cd_leaves = - 0.26 + 0.50 Al + 0.02 Silt (%) + 0.04 Clay (%) - 0.15 SOM (%) + 22.37 Cd_bd, interpreting that the variables that better predict according to Mallows’ Cp were exchangeable Al (11.74) and silt content (9.40), while Cd_bd had a positive relationship with Cd_leaves.

The relationship between Cd_fruits and the physicochemical characteristics of the soil could be modeled with fiur variables according to the backward method (Table 6c). The adjusted R² of the model was 0.44. The resulting model for Cd in fruits was: Cd_fruits = 0.36 + 11.40 Cd_bd - 0.09 Al -0.47 K - 0.03 Zn.

According to Mallows’ Cp, the variables that better predict Cd concentration in fruits were Cd_bd (15.75) and Zn (8.97). Except for Cd_bd, which had a positive relationship with Cd_fruits, the other variables affected it negatively.

The relationship between the concentration of Cd_beans and the physicochemical characteristics of the soil was modeled with four variables according to the backward method (Table 6d). The adjusted R² of the model was 0.68. The resulting model was: Cd_beans= - 0.64 + 20.44 Cd_bd - 0.03 Zn + 0.03 Clay (%) + 0.04 Mg.

As in the case of Cd_fruits, the variable of greater intervention for the concentration of Cd_beans was the Cd_bd with a Cp of Mallows of 35.23 and with a positive association, as well as Mg and Clay. The only variable negatively affecting the amount of Cd_beans was the Zn content.

In soil, the concentration of heavy metals (e.g., Cd) is closely linked to the content of SOM, which forms organometallic complexes, immobilizing their absorption by plants (Arévalo-Gardini et al., 2016). Researchers such as Cameselle & Gouveia (2019) mention that bioavailable Cd in soil can vary due to interactions with other metals. In turn, the concentration of Cd in the soil depends on the depth at which the sample is taken (Goebes et al., 2019).

The average pH value found in this study is considered very acidic (Snoeck et al., 2016). Low pH values allow for the highest solubility of toxic elements such as Al (0.7 cmol₍₊₎ kg^-1 on average in this study) and Cd. This is significant in cocoa cultivation since the absorption processes of the Cd concentration are favored when the pH is very acid. In the same way, when the pH in the soil increases, Cd is adsorbed by colloids, blocking its bioavailability (Liu et al., 2020).

This could be verified in the regression model (Table 6a) where the pH of the soil was presented with a p-value < 0.05 and a high prediction value, indicating that is the variable that most interferes in the concentration of Cd_bd; the lower the pH, the more Cd_bd. Several studies affirm that the bioavailability of Cd is closely related to pH values (Arévalo-Gardini et al., 2016; Florida, 2021).

Generally, the average SOM content is greater than 3% (Table 2) which is adequate for cocoa cultivation (Snoeck et al., 2016) while helping the nutrition of the crop and immobilization of bioavailable Cd in the soil (Fernández-Nieto & Betancourt-González, 2018). The SOM at the levels found could have contributed to the concentration of total N in the soil maintaining high values (SEMARNAT, 2002), which is an important element in the inorganic nutrition of plants and improving product quality.

On the other hand, the ECEC (19.9 cmol₍₊₎ kg^-1) is considered high (Cabalceta & Molina, 2006), while the exchangeable K is within the critical (optimal) level and the exchangeable cations Ca and Mg are sufficient in a soil classified as Inceptisol (Cabalceta & Molina, 2006). The microelements (Cu, Zn, Fe and Mn) presented high concentrations exceeding the critical levels described for each of them by Calbacete & Molina (2006).

The Zn content in the soil can interfere with the adsorption and absorption of Cd due to its chemical similarity (Sikora & Wolt, 1986), since they are competing elements and therefore the Zn content negatively affects the uptake of bioavailable Cd by the plant. It could be verified in this work (Table 6a), where Mallows’ Cp demonstrated that Zn is among the variables that predict the bioavailability of soil Cd (Huamaní-Yapanqui et al., 2012; Alloway, 2013).

The plots evaluated showed sandy loam, loam and sandy clay loam textures (Soil Survey Staff, 2014). According to the granulometry found of sand, silt and clay (Table 2) the study did not present textures with a high percentage of clay, but abundant in sand, similar to those found in studies carried out in Ecuador by Moreno et al. (2018) representing a very unfavorable factor because there may be a high mobilization of Cd moving by leaching to underlying surfaces at the depth of 30 cm (Kruszewski et al., 2018).

A high level of tolerance was observed for Cd_leaves (Table 3), according to Kabata-Pendias (2010) and Alloway (2013), who recommend a concentration between 0.05 and 0.5 mg kg^-1. According to Barraza et al. (2017) and Arguello et al. (2019), the Cd concentration values in leaves can be used as an approximate diagnosis, when the cocoa crop does not present its fruits. This may mean that there is a 3:1 ratio (cocoa leaves: beans).

A statistical correlation was found between the total Cd in the soil and the Cd_grains (Table 4). Ramtahal et al. (2016) reported similar results on cocoa plantations in Trinidad and Tobago. The results indicate that we can predict the amount of Cd in fruits and grains based on the amount of bioavailable Cd.

Cd_leaves did not show a significant correlation with Cd_bd in the soil, but when combined with other variables, it was significant (Table 6b). This may be because other metals, such as Al, can interact with the adsorption of this metal by the plant (Cameselle & Gouveia, 2019). Other authors, such as Huaraca-Fernández et al. (2020) refer to the fact that Cd_bd in soil is associated with ions (e.g., carboxyls), and these are related to SOM, which by increasing its contents decreases the concentration of bioavailable Cd and affects Cd_leaves (Table 6b).

It is worth mentioning that according to the European Union there are still no tolerance levels for Cd in cocoa beans (Barraza et al., 2017; Furcal-Beriguete & Torres-Morales, 2020) so the Cd level is taken as a reference for chocolate with total dry matter content ≥ 50% of the Codex Alimentarius. This should not exceed 0.80 mg kg^-1 (FAO/WHO, 2017). According to the European Union in its regulation No. 488 (EC, 2014) this level may also apply to cocoa fruit. In this study the Cd_grain was well below this limit, with an average of 0.25 mg kg^-1.

Similarly, it could be found in Table 2 that the concentration of total and bioavailable Cd in the soil could be related to the average concentration of Cd found in the leaves, fruits and cocoa beans (Table 3), since The Cd transfer coefficient of the soil and plant is 1.0-10.0 (Alloway, 2013).

The concentrations of bioavailable cadmium are within the levels of precision and accuracy of the analytical equipment used, for the 21 Inceptisol soil plots extracted at a depth of 30 cm and are relatively low despite the fact that the average pH level is very acid.

The content of soil organic matter presented medium levels, favoring the non-solubility of interchangeable Al and bioavailable Cd.

The models developed for bioavailable Cd, Cd in cocoa leaves, fruits and beans can be used to correct some chemical and biological properties of the soil that can be managed at the farm level, including liming to increase pH and adding organic material to increase the SOM content.