Twenty farms were selected in southern Bahia, Bra­zil (Fig. 1), located in three climatic regions: humid (H), sub-humid (S) and sub-humid to dry (SD) (Table 1), classified according to typology climatic condi­tions of the state of Bahia, according to Thornthwaite's methodology ( SEI, 2007 ).

Leaf and fruit collections in plants of the PH 16 clonal cacao genotype, recommended by the Cocoa Research Center (CEPEC/CEPLAC) because its high production and resistance to witches’broom disease ( Moniliophtora perniciosa ) ( Leite et al. , 2013 ), aged over six years, were performed in four plots of twenty farms, located under different edaphic and topogra­phic conditions (Fig. 1). Eight mature leaves per plant were collected in four plants per plot from January to February 2012, following the methodology described by Souza Júnior et al. (2012) , which consisted in col­lecting the third mature leaf from the apex of newly ripened branches in the four quadrants of the plants. The plant material was packaged separately in paper bags, according to the type of plant material, farm and region, and taken to the laboratory. The leaves were then washed with distilled water and then placed in an oven with forced air circulation at 65 °C until reaching constant mass. Finally, the samples were milled with a Willey type MA-340 mill, using 20 mesh sieves.

Four mature fruits per plant were collected, which were broken apart, separating them in husk and beans (seeds). The placenta remained attached to the husk. Seed pulp was removed with the aid of a plastic sieve, and the pulps were packed in plastic containers and frozen for further analysis. Beans and pod husk were dried in an oven at 60 °C, with forced air circulation, until reaching constant mass. Subsequently, the dried beans were separated into cotyledons and tegument, with the aid of a scalpel.

The chemical analyzes were adapted from the methodology described by EMBRAPA (2000) which uses samples of 0.20 g of natural pulp and dry matter of leaf, husk, tegument and cotyledons were digested with 4 mL of nitric acid and 3 mL of hydrogen peroxide. The digestion was performed in a digester block with initial temperature of 50 °C for 30 min and final temperature of 120 °C for 90 min. Afterwards, the samples were added into falcon tubes to 14 mL and Ni and Cu readings were measured by inductively coupled plasma atomic emission spectrometry model Varian 710-ES (Mulgrave, Austrália), and the analyses were performed in duplicate. This equipment is equipped with a radio frequency generator of 40 MHz used in the 1200-

1400 W range for the proposed study and a solid state with charge coupled device (CCD) ( Moreira, 2016 ).

Data, by climatic region, were subjected to descriptive statistics analysis and Shapiro Wilk's normality test (W) ( p <0.05), and the means were compared by the Tukey test ( p <0.05). The concentrations of Ni and Cu in the leaves and in the different parts of the fruit were submitted to Pearson's linear correlation.

In general, for all three climatic regions, the mean concentrations of Ni in the leaves and in all parts of the fruit were close to the medians (Table 2), indicating that the average would be a good reference of the central value of the sample. Larger differences between the mean and median values were observed only in the S region for leaf and cotyledons.

For all the evaluated plant materials, the concen­trations of Ni presented normal distribution in the H region, according to Shapiro Wilk's test, a fact opposite to that observed in the S region (Table 2). In the SD region, only for leaf and pulp, the concentrations of Ni presented normal distribution (Table 2). According to Cressie (1991) , data normality is not a requirement for a good set of data as long as the normal distribution curve does not have very elongated lines (high amplitude). This fact was not evidenced in the present work, since the similarities of the values of central tendency (average and median) indicate symmetrical distributions.

According to the classification proposed by Warrick & Nilsen (1980) , the coefficient of variation (CV) was high for all variables, except the pod husk (region SD), cotyledons (region H), and tegument (region S), which presented average CVs (Table 2). These results indicate a wide variability in Ni concentrations in all analyzed plant materials. This fact was corroborated by the high amplitudes (differences between the minimum and ma­ximum values) of the concentrations of Ni in the leaf and in the parts of the fruit (Table 2). Probably the high CV values and high amplitudes of Ni concentrations in the plant are due to the high geological and pedological variability of the cacao producing region of the South of Bahia and, consequently, high variability of the mineralogical, chemical, and physical attributes of the soils ( Arévalo-Hernández et al. , 2019 ). The variation of Ni concentration in the soil strongly depends on the source material ( Massoura et al. , 2006 ). According to these authors, when analyzing the concentration of Ni in soils of various parts of the world, there is a wide variation of the concentrations of this metal in the soil, which varied from 4 to 2000 mg Ni kg ^-1 soil, being the highest concentrations found in soils derived of ultramafic rocks.

There was significant difference of the mean con­centration of Ni between the climatic regions, for the leaf and in all evaluated parts of the fruit, except for husk. The mean concentrations of Ni in leaf, cotyledons, and pulp were higher in the region SD in relation to the other regions. However, for tegument, the mean concentration of Ni in the region SD was significantly higher only in relation to the mean of the S region (Table 2). The highest values of Ni in the analyzed plant materials found in the region SD may be due to the fact that this region has less aged soils than the other two climatic regions; being these more coastal (Fig. 1). In the region SD, there is also presence of ultramafic rocks, with mineral exploitation of Ni, with the mine being considered the third largest open pit of sulfided nickel in the world ( Matta, 2016 ).

Arévalo-Gardini et al. (2017) , analyzing heavy metals in the eight major cacao growing regions in Peru, did not find significant differences between regions for Ni concentrations in leaves and in ca­cao beans. According to these authors, mean concen­trations ranged from 2.2 to 12.2 mg Ni kg ^-1 DM and from 3.5 to 9.2 mg Ni kg ^-1 DM, for leaf and beans, respectively. In turn, Bertoldi et al. (2016) studied the geographical chemical traceability, based on the analysis of 56 elements of cacao bean samples from the 23 main cacao producing countries of different continents (Africa, Asia, Central and South America). These authors observed that the average concentration of Ni in the beans, in the Central American samples, was 12.1 mg Ni kg ^-1 DM, which is significantly higher than the average values found in the other continents, which presented averages ranging from 4.9 to 6.7 mg Ni kg ^-1 DM.

Bertoldi et al. (2016) and Arévalo-Gardini et al. (2017) described in their methodological procedures that the beans were peeled and separated in tegument and cotyledons, and the Ni concentration in the tegument was approximately 15 times higher than the cotyledons (Table 2). This shows that the Ni present in cacao beans accumulates predominantly in the te­gument, which is not used in the food industry, being an industrial residue. Only the cotyledons are used for the production of liquor, chocolate, cocoa butter and its derivatives. On the other hand, the pulp that surrounds cacao beans, which can also be used in the food industry, had higher Ni concentrations. Although, the pulp was the only part of the fruit analyzed based on the natural matter, while the other vegetal materials were analyzed based on the dry matter.

In the 80 foliar samples analyzed in the present study, the concentration of Ni varied greatly from 8 to

666 mg Ni kg ^-1 DM (Table 2), with average values being much higher than those observed by Arévalo-Gardini et al. (2017) . Phytotoxic Ni concentrations vary widely among plant species and cultivars, and there are reports of phytotoxic foliar concentrations ranging from 40 to 246 mg Ni kg ^-1 DM. However, there are native plants that grow in soils naturally contaminated with Ni, which have concentrations in the leaves above

6000 mg Ni kg ^-1 DM ( Kabata-Pendias, 2011 ). However, for cotyledons, the amplitude of Ni concentrations was much lower than in the leaf, ranging from 0.5 to 19.9 mg Ni kg ^-1 DM (Table 2). These values are similar to the variations found in cacao beans by Arévalo-Gardini et al. (2017) and ratified by the concentrations of 5 to 10 mg Ni kg ^-1 DM, presented by Kabata-Pendias (2011) , as commonly found in a cocoa powder sample.

In order to compare the partition of the two metals in the analyzed plant materials based on dry matter, the leaf was adopted as the reference organ, since it is the organ-spot of plant metabolism and often used to evaluate the nutritional status of the plant ( Souza Júnior et al. , 2018 ). Considering the average Ni concentration of the three studied regions, in each plant material, the concentrations of Ni in the tegument, pod husk and cotyledons were, respectively, 60.9%, 6.3% and 4.0 % of the average concentration found in the leaf (Table 2), evidencing the dilution effect of this metal on the fruit. This indicates, in turn, the low mobility of this metal from the leaf to the fruit, and in the fruit, it was more concentrated in the tegument than in the pulp (Table 2).

Nickel is a relatively mobile element in the phloem and can be readily transferred from sources (mature leaves) to preferential metabolic drains such as leaves and young fruits ( Page & Feller, 2005 ; Page et al. , 2006 ). When it reaches the draining organs, Ni can be exported from the tissues of the phloem to the apoplast again, where it is absorbed by neighboring cells ( Deng et al. , 2017 ). The translocation of Ni via phloem is bidirectional, including downward and upward motions. Deng et al. (2016) verified that 89% of Ni are exported from mature leaves to young leaves of Noccaea caerulescens , while only 11% are exported from mature leaves to roots. This suggests that upward movement is the predominant direction for translocation in the phloem, and that young leaves and reproductive organs are the primary preferential metabolic drains of Ni via phloem ( Deng et al. , 2017 ). Moreover, Estrade et al. (2015) also observed that the fractionation of Ni isotopes between leaves and flowers occurs in the initial stages of growth of Alyssum murale , resulting from the net transfer of Ni to the leaves. According to these authors, in the stage of full bloom, the Ni isotopic compositions between leaves and flowers are leveled, indicating that the great redistribution of phloem occurs at this stage.

Studies with radioactive ⁶³ Ni, applied via foliar in different plant species, demonstrated that Ni can be rapidly transferred to young leaves and young seeds ( Fismes et al. , 2005 ; Riesen & Feller, 2005 ). This fact was also observed in Ni hyperaccumulating plants, which accumulate this metal in leaves, flowers and seeds at high concentrations ( Zhang et al. , 2014 ; Groeber et al. , 2015 ). As the reproductive organs are the main translocation drains of the phloem, this evidence indicates that hyperaccumulating plants can carry substantial amounts of Ni via phloem. The high concentration of Ni in the phloem sap was documented in several Ni hyperaccumulating plant species growing in tropical ultramafic soils, such as Euphorbia helenae (3.1%) ( Reeves et al. , 1996 ) and Phyllanthus balgooyi (16.9%) ( van der Ent & Mulligan, 2015 ; Mesjasz-Przybylowicz et al. , 2016 ). However, in the present work, young leaves of cacao are probably the preferential metabolic drains for Ni from mature leaves, to the detriment of young fruits, due to the low mobilization of Ni from mature leaves to cacao fruits. Although there are no new branches during the fruiting of cacao trees, pruning, used to eliminate dry brooms, caused by the fungus Moniliophthora perniciosa , interferes with the phenology of the plant, breaking the apical dominance of the branches and stimulating the appearance of new leaves concomitant with the fruiting.

Positive and significant correlations of Ni concen­trations among all the studied plant materials were observed (Table 3). This demonstrates interdependence between the accumulation of Ni in the leaves and in the different parts of the fruit. It also suggests that the leaf, which is already analyzed for plant nutritional diagnosis, can also be used as an indicator of the accumulation of Ni in parts of the fruit used for food production, such as cotyledons and pulp.

The upward movement of Ni to the aerial part follows the flow of the xylem, which final destinations are mature leaves, due to the high transpiration in these tissues, while young leaves receive a small proportion of Ni via xylem ( Deng et al. , 2017 ). The Ni transported via xylem is in the form of free hydrated cations, and when it reaches the smaller veins and fills the apoplastic spaces in the leaves, it is loaded in the foliar symplast or remains in the apoplast ( Deng et al. , 2017 ). In leaves, Ní is preferentially distributed in epidermal cells, at least in the active tissues of the symplast ( Tappero et al. , 2007 ). In leaf symplast, Ni is rapidly transferred to vacuoles, particularly in epidermal cells, where it is chelated mainly by carboxylic acids ( Deng et al. , 2017 ). The mesophilic paralisade cells become an important compartment when there is increase in Ni concentration in the leaves ( Broadhurst et al. , 2004 ).

For all the analyzed plant materials, the values of central tendency measurements of Cu concentrations presented similar behavior to those of Ni, that is, similar mean and median values. Almost all Cu data presented a non-normal distribution, except for pod husk, in the three regions, and for cotyledons and leaf in the region H (Table 4). Regardless of the analyzed plant material, CV for Cu were lower than those observed for Ni (Tables 3 and 4), indicating a lower variability of that. According to the classification presented by Warrick & Nilsen (1980) , the great majority of CV for Cu fits as average.

There was not significant difference in Cu concentrations between the climatic regions in any of the fruit parts. However, the mean concentration of Cu in the leaves of the region H was significantly higher than the means of the other two regions (Table 4). This fact can be attributed to the greater use of copper fungicides in that region for the control of brown rot and witch broom diseases ( Veloso & Santana, 2000 ). However, there are reports of Cu deficiency in cacao in the region SD ( Souza Júnior et al. , 1999 ). Considering the appropriate leaf range of 10 to 20 mg Cu kg ^-1 DM, for cacao trees grown in the South of Bahia ( Souza Júnior et al. , 2018 ), of the 80 samples analyzed in the three regions, 60% of the plants would be deficient in Cu, 48.3% with adequate concentrations and only 3.7% with excessive concentrations. In addition, the highest frequency of deficiency occurred in the regions SD and S, which had 83.3% and 58.3% of plants with Cu deficiency.

On average, for the three regions together, Cu con­centrations in the pod husk, tegument and, cotyledons were 1334%, 1685%, and 16%, respectively, higher than the Cu leaf concentration (Table 4). It evidences the effect of concentration of this metal on the fruit, which indicates that Cu would have good mobility in phloem for cacao trees. However, it would focus on fruit residues (husk and tegument) and not on parts frequently used in the food industry, such as cotyledons and pulp (Table 4). Arévalo-Gardini et al. (2017) also observed, in eight cacao producing regions in Peru, Cu concentrations in cacao leaves lower than those found in cacao beans (tegument + cotyledons), ranging from 7.2 to 10.3 mg Cu kg ^-1 DM and 18.8 to 30.4 mg Cu kg ^-1 DM, respectively. On the other hand, different from that observed for Ni (Table 3), no significant correlations were observed for Cu concentrations among the different evaluated plant materials.

In plants, Cu can be transported in the form of free Cu (I) or Cu (II), however, it is usually transported as Cu complexes ( Yruela, 2009 ). There is strong evi­dence that Cu, once in the xylem sap, is transported in Cu (II) form by specific metal chelators ( Printz et al. , 2016 ). Possible ligands candidates are small molecules, including carboxylates of organic acid such as citrate and malate, amino acids [nicotianamine (NA), histidine (His) and cysteine (Cys)], high affinity Fe (III) chelating compounds, and NA derivatives ca­lled phytosiderophores, such as muginoic and 2-deoxi­­muginic acids, as well as peptides and proteins [metallothioneins (MTs)] ( Alvarez-Fernandez et al. , 2014 ). At pH of the xylem sap, NA and His are the major Cu ligands, suggesting that these amino acids are the major ligands during the long-distance transport of Cu in the xylem ( Pich & Scholz, 1996 ).

MTs are Cys-rich proteins capable of coordinating the Cu (I), Zn (II) and Cd (II) ions by their thiol groups ( Wan et al. , 2013 ). These proteins are involved in the redistribution of Cu from senescent leaves to draining organs such as young leaves or seeds in develop­ment ( Benatti et al. , 2014 ). NA is ubiquitous in higher plants and present in all tissues, and is involved in metal transport ( Clemens et al. , 2013 ). Histidine has three metal binding sites (carboxylate, α-amino, and imidazole groups). The last group forms strong bonds and strong complex, especially with Ni and Cu ( Alvarez-Fernandez et al. , 2014 ). There is evidence that His is involved in the long-distance transport of metals in plants, mainly Ni in the xylem of hyperaccumulating species of the genus Alyssum ( Deng et al. , 2017 ). His (in the μM-mM range) and Ni concentrations in the xylem sap are significantly and linearly correlated with various Ni hyperaccumulating Alyssum , such as A. lesbiacum , in response to the increase of concentrations of metals in the growth media ( Krämer et al. , 1996 ).

In general, leaf, tegument, and fruit pulp of clonal cacao genotype PH 16 were the plant materials that accumulated more Ni. The largest accumulation of Cu was in the tegument (cacao beans/seeds coating) and in the pod husk, whereas in the cotyledons there was little accumulation of these metals. There was high variability of Ni and Cu concentrations in all the analyzed plant materials. The concentrations of Ni were influenced by the climatic region, a fact not observed for the Cu, except for its foliar concentration. There was interdependence between Ni accumulation in the leaves and in the different parts of the fruit, a fact not observed for Cu. The low Ni and Cu accumulation in cacao beans cotyledons, raw material for the manufacture of chocolate and other food products, has shown that these metallic elements do not affect consumers' health.