Mitigation of Cadmium Uptake in Cocoa: Ecacy of Soil Application Methods of Lime and Biochar

Although mitigation approaches have been developed to reduce Cd in cocoa beans the ecacy of the approaches have been inadequate to make them economically viable. A eld study was conducted in a cocoa farm in Biche, Trinidad using two soil amendments, lime and biochar, at recommended rates using three methods of application, soil surface application with incorporation (SA), soil injection (SI) or deep placement using an auger (AA) along with a control. The objective was to determine the application method that would be most ecacious with respect to rapidity of effect, magnitude of reduction and persistency of effect on leaf Cd. The experiment was arranged in randomized complete block design with three replications with 15 trees per replication. Phytoavailable soil Cd, soil pH, CEC and total leaf Cd concentration were monitored monthly on three guarded trees per plot over a one-year period. The results showed that both lime and biochar were effective in reducing leaf cadmium levels albeit at different levels. The ecacy of SI was signicantly better than SA in terms of rapidity of the effect on leaf Cd in comparison to the control (40% compared to 30%) as well as the effect was more persistent in SI. With biochar, again the SI was signicantly better than SA with regards to reducing leaf Cd levels in comparison to the control (35% compared to 20%) but the time taken to action and the persistency were lower compared to lime application. AA did not signicantly reduce Cd level in the leaf with lime or biochar application. determined using the dietheylene triamine penta acetic acid extraction method of and Norvell The involves extracting 10 of soil with 20 mL of reagent containing 0.005 M DTPA, a correspondingly low kg of acidic kg critical kg Zn, optimum range smaller reduction (av. 20%) that occurred 1 month after SI and lasted only for 2 months (6 MAA to 7 MAA), This was also mirrored in the changes experienced with CEC. These results suggest that SI achieved a better incorporation of biochar in the soil than SA, resulting in greater treatment effectiveness. Past studies have shown that biochar needs to be thoroughly incorporated in the soil in order for it to be effective as a technique of soil remediation (Ruysschaert et al. 2016; Guo et al. 2020). The application of biochar using the AA had no signicant effect on DEC and leaf Cd levels as was demonstrated for lime application.


Introduction
Bioaccumulation of cadmium (Cd) in the human body from consumption of Cd contaminated food over the human lifetime presents numerous health risks including damage to the kidney, liver, bones and the neurological system ( There is considerable evidence that the Cd concentration in chocolate products is a function of the level of Cd found in cocoa beans (Mounicou et al. 2002). Soil ameliorants such as lime, biochar, inorganic and organic-rich materials and soil microbial consortia have been shown to reduce the soil phytoavailable Cd and consequently the Cd concentrations in cocoa beans (Meter et al. 2020). In a eld trial, Ramtahal et al. (2019) demonstrated the effectiveness of both lime and biochar as soil ameliorants in reducing Cd uptake in cocoa but noted that their effectiveness diminished over time. This was attributed to the heavy rainfall and the sloping terrain where cocoa is grown and the inability to effectively incorporate the ameliorants into the soil.
There are many ways in which amendments can be incorporated into soils including broadcasting, soil injection and localized placement. Broadcasting otherwise known as surface application of the material can be done with or without subsequent incorporation by tillage and is considered to be typically the fastest and most economical method of application of amendments, but is prone to amendment losses due to environmental factors (Brady and Weil 2004). Ramtahal et al. (2019) employing surface application with light tilling into the top 20-cm of the top soil layer, while taking care not to damage the lateral feeder roots of the cacao tree, showed this to be an improvement over surface application without tilling (Ramtahal et al. 2018). Another recent pot study, Argüello et al. (2020) highlighted the importance of subsurface liming to improve the effectiveness of lime applications in reducing Cd uptake. Localized or spot placement involves the placement of the material in speci c locations (spots or band) or in depressions created by various tools including drills or auger within the dripline of the tree (Jain et al. 2018) and has been used successfully over the years for fertilizer application. Amendment application via soil injection typically as pressure-injected sludge into the soil has been shown to be very e cient in getting the material into the rooting zone of the plant to promote amelioration in crops (Faber et al. 2015). Although soil injection or deep placement via augered holes can be more effective as methods of application, these have not been tested for the mitigation of Cd in cocoa.
The overall objective of this study was to evaluate the effectiveness of three soil application methods, soil injection, deep placement and surface application with tilling, of two amendments, biochar and lime, in reducing the phytoavailable pool of soil Cd, its uptake and bioaccumulation in leaves under eld conditions.

Location of eld study
This investigation was conducted at a cacao estate in Biche, East-Trinidad (10.41′62.96″ − 61.12′61.65″) during a one-year period from February 2018. The site consisted of 40-year-old trees of a mix of cross compatible TSH varieties grafted onto TSH 911 rootstocks. The landscape was generally at. The cocoa trees were planted at a spacing of 4 x 4 m with banana and other permanent shade trees.

Soil characteristics and amendment requirement determination
In order to determine the soil characteristics and required amendment rates for the treatment site, representative soil samples (0-15 cm) were collected, airdried, ground using a mortar and pestle, sieved to < 2mm and mixed thoroughly. Each dried sample was placed in a clear polyethylene bag, mixed well, labeled and stored until analysis. Sub-samples were taken and subjected to the following analyses: I. The particle size distribution was determined using the hydrometer method described by Gee and Or (2002).
II. Soil pH was measured in triplicate using three 10 g portions of air-dried soil mixed separately with 10 mL of deionized water, and allowed to equilibrate for 30 min. The pH of the supernatants was then measured with a calibrated pH electrode and meter (Kalra 1995) and averaged over replicates.
III. Organic matter content was measured using the method as described by Walkley and Black (1934).
IV. CEC was also determined according to Sumner and Miller (1996).
V. "Total" Cd, Fe, Mn and Zn soil metal concentrations were determined using the USEPA 3051A method (USEPA 2007). Triplicate samples each containing 0.5 g were digested using 10 mL of concentrated analytical grade 60% HNO 3 (J.T. Baker, USA) in a microwave digester (Multiwave PRO, Anton Paar, Austria). After acid digestion, samples were cooled to room temperature, diluted with 5 mL deionized water in a boiling tube and ltered through Whatman #542 lters into 25 mL volumetric asks. Each solution was made up to volume with deionized water rinses of the residues and mixed thoroughly.
VI. Phytoavailable soil Cd was determined using the dietheylene triamine penta acetic acid (DTPA) extraction method of Lindsay and Norvell (1978). The VII. The soil liming requirement was determined using the method of Eckert and Sims (1995). The soil was mixed with a modi ed triethanolamine-ammonium chloride Mehlich buffer (1:1) of pH 6.60 and the resultant pH measured (BpH). The following equation was then used to calculate the lime requirement (LR) for the soil:

Field Experimentation and treatments
The eld trial was carried out at the cocoa estate in Biche, with hydrated lime [Ca(OH) 2 ] applied at a rate of 6000 kg ha − 1 and biochar (Charcoal Green® Biochar Powder) applied at rate of 326 kg ha − 1 . Each amendment was applied with three treatments (injector, auger and surface) at the appropriate rates, along with appropriate controls (no application of amendment). The treatments was laid out in a randomised complete block design with three blocks with each plot within a block consisting of three guarded trees.

Application of treatments
For the soil injection method, the amendment requirement for each plot was calculated, mixed with water using the recommended mixing ratio and injected into the soil using a motorized backpack (Maruyama Model MS75, Fort Worth, TX, U.S.A) equipped with a soil injector (HTI-2000 Soil Injector, Minnetonka, MN, U.S.A). The soil was injected at 15 cm depth using the pressurized injection system at multiple points within the drip-zone of each cocoa tree until all the solution was emptied. With respect to the auger technique, a soil auger (AMC, SST reinforced regular, 6" long, 2 ¾" diameter, Idaho, U.S.A), was used to create ten holes around the drip-zone of each of the tree which was then lled with a slurry of each of the amendment, at the appropriate rate per tree, and covered with soil. The surface application involved the broadcasting of each amendment onto the soil surface within the drip-zone of each tree which was subsequently incorporated using a fork, by gently tilling the 0-15 cm of the upper soil horizon taking care not to damage the surface roots of the cocoa tree. For the control, no amendment was added but the method of applications was applied with water.

Data collection
The soil and trees within each plots were sampled at monthly intervals for a year. Soil and leaf samples were taken before the treatments were applied and every month thereafter. Three soil cores, 0-15 cm deep were sampled at three locations within the drip zone of each tree using a soil corer (Model A1, 12" long, 3/4" diameter, Oak eld, U.S.A), composited and subsampled for analysis. Thirty leaf samples at the Inter ush 2 stage (Greathouse et al. 1971) were obtained from throughout the tree, dried, ground and subsampled. The soil samples were analysed for phytoavailable Cd, pH, CEC and total Cd respectively while the leaf samples were analysed for total cadmium as described below.

Soil and leaf analysis for eld study
The preparation and analysis of soil samples for the determination of phytoavailable Cd, pH and CEC in the eld study were as described in Sect. 2.3. The leaf samples were carefully washed with deionized water to remove visible surface contaminants such as algae and soil. They were then spread out on paper towels to remove excess water, oven-dried in aluminum foil at 75°C until constant weight and ground to < 0.2 mm using a mortar and pestle. The dried samples were ground and placed in clear plastic bags, labeled and stored until analysis. The 'total' concentrations of Cd in the leaf triplicate samples each containing 0.5 g were digested using 10 mL of concentrated analytical grade 60% HNO 3 (J.T. Baker, USA) in a microwave digester (Multiwave PRO, Anton Paar, Austria). After acid digestion, samples were cooled to room temperature, diluted with 5 mL deionized water in a boiling tube and ltered through Whatman #542 lters into 25 mL volumetric asks. Each solution was made up to a volume with deionized water rinses of the residues and mixed thoroughly (USEPA 3051A). Total cadmium for the leaf and total and available Cd for the soil were determined by means of Flame Atomic Absorption Spectrometry (FAAS), Varian SpectrAA 880, Australia tted with a high-sensitivity hollow cathode lamp (UltrAA) to enhance Cd detection in sample solutions (EPA 7000B).

Meteorological data
The rainfall and temperature data for the study period was obtained from the nearest meteorological station.

Quality control
The water used for preparing samples and cleaning of glassware and other apparatus in this study was glass-distilled and then deionized using a water puri cation system. In order to avoid trace metal contamination, laboratory glassware and other utensils used in all analyses, were washed with a suitable detergent, soaked in an acid bath of 2 M nitric acid for at least 24 h, rinsed in distilled deionized water and dried in an oven at 50ºC. All reagents used in this study were of analytical grade.
Since no certi ed soil reference materials were available for the determination of phytoavailable Cd using the selected extractant, an Internal Quality Control Material (IQCM) was developed to act as a surrogate standard in this study. The IQCM was prepared from a bulk soil sample collected from a cacao plantation previously identi ed as having Cd-contaminated soil. The bulk sample was air-dried for 48 h, then ground and sieved through a 2 mm sieve. It was subsequently analyzed with every batch of soil samples analyzed for phytoavailable Cd, to ensure consistency of Cd levels extracted from the IQCM throughout the laboratory trial. Similarly, to monitor and control the quality of the method of determination of 'total' Cd in soils and leaves, a National Institute for Standards and Technology (NIST) Certi ed Reference Material (CRM), SRM 2709a, San Joaquin Soil and SRM 1570a, Spinach Leaves were used, respectively, and analyzed with each batch of samples for the experimental period. Recoveries of all reference materials analysed during the investigation fell within the acceptable 95-105 percentile range.

Statistical Analysis
Data were analyzed using ANOVA through the General Linear Model (GLM) routine. Tukey's test was used to detect differences between means at probability level P < 0.05. Analysis was done using the software Number Cruncher Statistical System (NCSS 2007, LLC, USA). Normality of data and variance homoscedasticity were tested prior to carrying out ANOVA. For the eld experiment, the initial level of Cd in the soil were used as co-variates in ANOVA to eliminate within eld differences affecting the results.

Agro-ecology of the experimental site
The experiment was conducted in cocoa farm situated in the lowlands in Biche, Trinidad and Tobago, with a topography described as at to slightly undulating. Its topsoil was predominantly clay (65%) as demonstrated by the soil textural analysis, with relatively low organic matter content ( Table 1). The soil was acidic in nature with a pH of 4.76 and had a correspondingly low CEC of 5.84 cmol kg − 1 , typical of acidic soils. The total soil Cd concentration of the study area averaged 1.47 mg kg − 1 which is above the critical limit (0.43 mg kg − 1 ) for Cd in agricultural soils (USEPA 2002). The concentrations of the other metals Zn, Mn and Fe were within the optimum range for agricultural soils. Total Fe (mg kg − 1 ) 52891 The rainfall pattern (Fig. 1) was typical of the humid tropics with a dry period characterised by relatively low rainfall from December to May followed by a period of heavy rains during June to November (wet season). Consequently, the rst four months and the last three months of the study period experienced on average 60 ± 13 mm and 19 ± 10 mm rainfall respectively compared to a monthly average of 232 ± 35 mm during June-November, 2018. The monthly average temperature uctuated between 25.5 to 27.5 o C during the study period and dipped below 26.5 o C between the months of November to March.

Effect of application methods of lime and biochar on soil pH
Lime application either by soil injection (SI) or surface application (SA) signi cantly increased (P < 0.05) soil pH from an average of 4.7 to pH > 7.0 within a month of application, while the auger method of application (AA) did not signi cantly (P > 0.05) increase soil pH compared to the control (Fig. 2a). Further, lime application by SI maintained the pH above 7 for a period of 7 months after which it showed a modest decline but was above pH 6.5 through the study period. In contrast, surface application showed a steady decline in pH, three months after application (MAA), but the pH was maintained at around pH 6 throughout the study period. Overall, soil injection of lime was signi cantly better in maintaining a higher soil pH than SA, particularly after 3 months of application. In the case of biochar, neither the application method, auger, injector nor surface, signi cantly (P < 0.05) altered pH levels compared to the control for the duration of the study (Fig. 2b).

Effect of application methods of lime and biochar on soil phytoavailable Cd
The DTPA-extractable soil Cd (DEC) calculated as proportional change in relation to the control (PCDEC) where the value of 1 is the reference and values above or below represents an increase or decrease in phytoavailable Cd respectively. As expected the PCDEC showed a signi cant (P < 0.05) decline with lime application for SI and SA; while that for AA was not signi cantly different (P > 0.05) from 1 throughout the study period (Fig. 3a). There was also a signi cant (P < 0.05) PCDEC x month interaction. The effect of SA of lime on PCDEC was not signi cantly different (P > 0.05) from the control until 3 MAA following which there was a signi cant decline (average 23%; P < 0.05) which persisted up to 6 MAA. Following 6 months PCDEC for SA increased and was not signi cantly different (P > 0.05) from 1. In contrast, SI of lime resulted in a signi cant reduction (P < 0.05) in PCDEC compared to the control from 2 MAA (av. 11% reduction below 1) with a generally consistent decline until 5 MAA at which time it was 51% below 1. After 5 MAA, there was a gradual increase in PCDEC for SI, however, it remained signi cantly lower (P < 0.05) than the control until the end of the study period.
The PCDEC for SI for biochar treatment showed a steady and signi cant decline (P < 0.05) (av. 11%) below 1 from 3 MAA until 6 MAA then uctuated. A decline was observed with SA only after 4 MAA and uctuated following month 5 and remaining mostly not signi cant (P < 0.05) from 1. With respect to the AA, the PCDEC uctuated and remained at or above 1 and was not signi cantly different (P > 0.05) throughout the duration of the study.

Effect of application methods of lime and biochar on total leaf Cd
For both lime and biochar amendments ( Fig. 4a and 4b), leaf Cd levels for the control remained unchanged (5 mg kg − 1 ) until 3 MAA (not signi cant P > 0.05), following which there was a gradual but signi cant (P < 0.05) increase in Cd levels up to 7 MAA (av. 14%), which then remained quite consistent for the remaining 5 months. Both the effects of 'month', 'application method', 'amendment' and their interactions had a signi cant (P < 0.05) in uence on total leaf Cd.
In the case of lime treatment by auger (AA), the leaf concentration of Cd was not signi cantly different (P < 0.05) from the control for the duration of the experiment, except at month 7, when it was signi cantly (P < 0.05) lower than the control (Fig. 4a). In contrast, for SI and SA of lime, there was a declining trend for leaf Cd concentration up to 6 MAA, following which it showed an increase until the end of the study. At 3 MAA, leaf concentration of Cd of SI was signi cantly lower than that for the control and persisted at that level (average 40% below control) up to 6 MAA. Although there was a signi cant (P < 0.05) increase in leaf concentration of Cd following 6 MAA, the SI treatment remained signi cantly (P < 0.05) below that of the control until month 10. This rate of decrease in leaf Cd concentration for SA of lime was more gradual and was signi cantly lower than that of the control at 3 MAA (11% lower than control), dropping to an av. of 30% at 6 MAA and remained signi cantly lower up to 10MAA. Overall, while lime application by SI and SA were both effective in reducing leaf Cd concentration, the SI was effective as indicated by a (a) faster action of lime in reducing leaf Cd levels (b) achieving a higher magnitude of reduction and (c) the effectiveness lasting for a longer period than SA. Lime application by AA was not shown to be effective in this study.
With respect to biochar application, leaf Cd levels for AA were not signi cantly different (P < 0.05) over time, compared to the control (Fig. 4b). For SA, leaf Cd concentrations stayed relatively constant until 5 MAA following which there was on average a signi cant reduction in Cd concentration (av. 20%; P < 0.05). This reduction was short-lived with leaf Cd levels returning to previous levels 6 MAA for the rest of the study period. SI resulted in a more rapid and signi cant (P < 0.05) decline reaching 35% reduction compared to the initial level, 5 MAA. A subsequent loss in effectiveness was observed as leaf levels gradually increased over the remaining months, but remained signi cantly (P < 0.05) lower than the control. There was no signi cant relationship between biochar application and pH for all methods (Fig. 6a).
As expected, there was a signi cant correlation (Fig. 5a) between DTPA-extractable Cd (DEC) and soil pH for the lime treatment (SI treatment: r = − 0.879; P < 0.05; SA treatment: r = -0.578; P < 0.05). The correlation for AA was weak (r = 0.188) and not signi cant (P > 0.05). Consequently, as pH of the soil increased in SI and SA treatments, DEC decreased. There was also a signi cant correlation between DEC and leaf Cd (SI treatment: r = 0.67; P < 0.05; SA treatment: r = 0.48; P < 0.05) (Fig. 5b and 5c) for SI and SA treatments indicated by a strong correlation between DEC and Leaf Cd and between pH and leaf Cd. In contrast, biochar application (SI and SA) as expected did not have a signi cant correlation (P > 0.05) with soil pH but had signi cant effect on reducing DEC and consequently signi cantly (P < 0.05) reducing leaf Cd as indicated by the positive correlation (SI: r = 0.708, P < 0.05; SA: r = 0.645, P < 0.05) (Fig. 6b). As expected for biochar by AA, there was no signi cant correlation (P > 0.05) between pH and DEC nor DEC and leaf Cd.
Although no signi cant relationship (P > 0.05) were found between CEC and other measured soil properties, there was a general increasing trend up to 4 MAA followed by a gradual decline when the soil was limed by the injector and surface methods (Fig. 7).

Discussion
The application of lime and biochar to Cd contaminated soils have been shown to be effective in mitigating Cd uptake into the cacao tree and into the beans On the other hand, although SA was able to raise the pH to a level not signi cantly different from 7 within 1 MAA, the levels started declining 3 MAA but remained at a level above 6 throughout the study period. Consequently, both DEC and leaf Cd concentrations started declining 2 MAA and attained their lowest levels 6 MAA which were 70 % and 75% of the control, respectively, compared to 50% and 60% for SI. Following 6 MAA both DEC and Leaf Cd levels started rising for SA and the leaf Cd levels were not signi cantly different from the control by 8 MAA. Hence the magnitude of reduction and duration of effectiveness of SA was lower than that of SI. This could be attributed to greater losses due to surface run-off in SA compared to SI. Lime and its ability to effectively increase the pH of the soil is diminished with increasing rainfall, acidity of the rain, soil and site management factors (Goulding and Blake 1998).
In contrast, the AA method was not able to effect a change in soil pH in comparison to the control and consequently was unable to signi cantly reduce DEC levels in the soil, which resulted in leaf Cd levels not signi cantly different from that of the control throughout the study period. That lime using the auger method was unable to effectively elevate soil pH indicates that lime placed in auger created soil depressions along the drip line of the tree was unable to migrate laterally to effect soil pH in the clayey soils of that location. Clayey soils are less permeable and are mostly characterized with low in ltration (Aswathanarayana 2001). It is also plausible that the water accumulating in the depressions facilitated the leaching of the lime below the root zone making it ineffective. Other studies have shown that most of the functional roots of cocoa remain within the top 20 cm within the enriched top soil layer (Hartemink 2005;Toxopeus 2008).
Soil injection helps to prevent treatment loss by placing it below the zone subject to erosion (WPHA 2018 Overall the application of the amendments using the SI technique was much more e cient than SA in reducing Cd accumulation in cacao leaves. Neither lime nor biochar application by AA was shown to be effective. Further, the study demonstrated the effectiveness of lime over biochar as a soil amendment in reducing Cd uptake in cacao. Overall, lime application by SI had (a) a faster action in reducing leaf Cd levels within 1 MAA compared to biochar that took up to 3 MAA; (b) achieved a higher magnitude of reduction than biochar (40% compared to 35%) and (c) resulted in the effectiveness lasting longer than biochar (4 months compared to 1 month), before leaf Cd levels started to increase. The study also indicated that lime application even with SI required repeat applications, yearly, as was reported by Ramtahal  The effect of application methods (auger, injector and surface) of lime (a) and biochar (b) on soil pH in a eld study of Theobroma cacao L Figure 3 The effect of application methods (auger, injector and surface) of lime (a) and biochar (b) on DTPA-extractable Cd, expressed as a proportion of the control, in a eld study of Theobroma cacao L Figure 4 The effect of application methods (auger, injector and surface) of lime (a) and biochar (b) on total leaf Cd, in a eld study of Theobroma cacao L Figure 5 The relationships between DTPA-extractable Cd and pH (a), total leaf Cd and DTPA-extractable Cd (b) and total leaf Cd and pH (c) for different application methods of lime in a eld study of Theobroma cacao L Figure 6 The relationships between DTPA-extractable Cd and pH (a), total leaf Cd and DTPA-extractable Cd (b) and total leaf Cd and pH (c) for different application methods of biochar in a eld study of Theobroma cacao L Figure 7 The effect of application methods (auger, injector and surface) of lime (a) and biochar (b) on soil cation exchange capacity (CEC), in a eld study of Theobroma cacao L