1 Accumulation of Cadmium in Soils, Litter and Leaves in Cacao Farms in the North Sierra 1 Nevada de Santa Marta, Colombia 2 Daniel Guarín1, *, Javier M. Martín-López2, Zamir Libohova3, Jhony Benavides-Bolaños1, Siela 3 N. Maximova4,6, Mark J. Guiltinan4,6, John Spargo4, Mayesse da Silva2, Alejandro Fernandez5, 4 and Patrick Drohan1 5 1. Department of Ecosystem Science and Management, The Pennsylvania State University, 6 University Park, PA 16802-3504 7 2. International Center for Tropical Agriculture (CIAT), Palmira, Colombia. 8 3. USDA-NRCS National Soil Survey Center, Federal Building, Lincoln, NE, United States 9 4. Department of Plant Sciences, The Pennsylvania State University, University Park, PA 16802-10 3504 11 5. Department of Earth and Environmental Sciences, University of Minnesota, Minneapolis, MN 12 55455 13 6. Huck Institute of the Life Sciences, The Pennsylvania State University, University Park, PA16802 14 * Corresponding author (+18148529683, dguarin@psu.edu). 15 16 Abstract 17 Cadmium (Cd) accumulation in Colombian cacao is a growing concern due to its potential 18 health impacts and EU regulations on Cd content in chocolate products. Furthermore, cacao 19 plays a significant role as an agricultural commodity and a tool for illegal crop replacement, 20 yet our regional understanding of Cd dynamics in cacao cultivation in the north flank of the 21 Sierra Nevada de Santa Marta is still limited. This research provides the first comprehensive 22 investigation of cadmium biogeochemistry in cacao agroecosystems by analyzing the 23 interactions between subsurface soils, topsoil, rock fragments, litter, and cacao leaf Cd 24 concentrations from 30 farms. Results reveal generally low mean total soil Cd 25 concentrations for topsoil and subsurface soils at 0.12 mg kg-1 and 0.05 mg kg-1, 26 respectively. Leaf and litter Cd concentrations are significantly higher (p < 0.05) than soil 27 mailto:dguarin@psu.edu 2 Cd, with a mean of 0.42 and 0.4 mg kg-1, respectively. Our results suggest that age 28 dependent surface-level processes such as the bioaccumulation and biocycling of Cd over 29 time through the leaves, litter, and topsoil, govern Cd in the topsoil, leading to older cultivars 30 and trees exhibiting higher Cd concentrations in leaves, litter, and soils. Subsurface soil Cd 31 is primarily driven by geogenic Cd coming from the weathering of the underlying bedrock 32 with a hypothesized contribution from pedogenic Cd being translocated to deeper soil 33 horizons from the topsoil via clay and oxide illuviation. Our research provides insights into 34 the accumulation of Cd in cacao plants and soils, which can lead to long-term preferential 35 accumulation of cadmium on soil layers and thus increase plant uptake through roots. 36 Keywords: Theobroma cacao, Cadmium, Cd biocycling, Cd bioaccumulation; Geogenic Cd, 37 Anthropogenic Cd. 38 39 1. Introduction 40 Cadmium (Cd) is one of the most toxic heavy metals present in the environment with no 41 known physiological and biological function in plants or humans (Tchounwou et al., 2012; 42 Vanderschueren et al., 2021). The consumption of Cd via ingestion of contaminated plant and 43 animal food products represents a danger to food security and global health (WHO, 2010). 44 According to the International Agency for Research on Cancer (IARC), Cd and Cd compounds 45 are ranked as Group 1 carcinogenic elements, meaning there is enough evidence to conclude 46 that they can cause cancer in humans (IARC, 2019; Romero-Estévez et al., 2019). 47 Trace amounts of Cd in soils can lead to its gradual accumulation in edible plant structures, 48 ultimately entering the human food chain (Barraza et al., 2018; Sharma et al., 1979). The cacao 49 tree (Theobroma cacao) is known to accumulate Cd in roots, leaves, pods, and beans (Barraza 50 et al., 2017; Chavez et al., 2015; Hanafi and Jomol Maria, 1998; Lewis et al., 2018; Meter et al., 51 2019; Pérez Moncada et al., 2019; Zug et al., 2019) at a higher rate than normally seen in other 52 plants (Argüello et al., 2019; Barraza et al., 2019; Gramlich et al., 2018). This has raised concerns 53 3 regarding the potential health effects of the consumption of chocolate produced with cacao beans 54 sourced from high Cd bearing crops, especially for the cacao located in Latin America and the 55 Caribbean, where soils and cacao beans tend to have higher concentrations of this element 56 compared to other cacao growing areas (Chavez et al., 2015; Meter et al., 2019; Scaccabarozzi 57 et al., 2020; Vanderschueren et al., 2021). 58 To address the health risks associated with the dietary intake of Cd through the 59 consumption of chocolate, the European Union (EU) Commission approved Cd limits in cacao 60 products (The European Commision, 2014; Vanderschueren et al., 2019). Limits were set 61 between 0.1 and 0.8 mg kg-1, depending on the cacao content of the final chocolate product. 62 Likewise, the Codex Alimentarius Commission, an international food safety organization, defined 63 the limits for Cd in chocolate, with a maximum of 0.90 mg kg-1 depending on the concentration of 64 cacao solids in the final product (Codex Alimentarius Commission, 2018; Vanderschueren et al., 65 2019). 66 In tropical Latin American countries like Colombia, Ecuador, Peru, and Brazil, the 67 derivatives of cacao beans are an important commodity used for premium quality chocolate 68 manufacturing. The sale of cacao in these countries is in many cases the main, and in some 69 instances the only, source of income for thousands of farmers (Abbott et al., 2018; Meter et al., 70 2019; Zug et al., 2019). In Colombia, cacao has been chosen as a legal alternative for coca 71 growing in areas of the country that have been historically affected by armed conflict (Abbott et 72 al., 2018; Meter et al., 2019; Zug et al., 2019). Switching to cacao cultivation in the traditional coca 73 growing territories was one of the strategies implemented by the Colombian government after the 74 signing of the peace agreement in November 2016 (Colombian Presidency Press Office, 2018), 75 making cacao a crucial crop for post-conflict social and agricultural development in Colombia. The 76 new limits of Cd in chocolate products adopted by the EU represent a concern for the future 77 continuation and success of such crop replacement programs. Therefore, identifying Cd sources, 78 interactions, and viable mitigation strategies is paramount to avoid negative impacts on farmers’ 79 4 livelihoods and the continuation and expansion of illegal crop replacement programs in the cacao 80 growing region. 81 There are two potential sources of cadmium in soils and plants. First, geogenic Cd (i.e. 82 naturally present in the environment) is naturally present in soils due to weathering processes 83 that release this element from parent materials containing heavy metals (Adriano, 1986; 84 Engbersen et al., 2019; Kabata-Pendias and Pendias, 2011). Deep marine sedimentary rocks 85 generally exhibit higher Cd concentrations because of multiple sediment provenance sources that 86 can potentially contain larger amounts of Cd bearing trace elements (He et al., 2015). Moreover, 87 sediment transport and deposition processes from potentially high Cd lithologies can lead to the 88 accumulation of Cd in alluvial soils (Capparelli et al., 2020; Chavez et al., 2015; Gramlich et al., 89 2018; He et al., 2015; Kubier et al., 2019; Rodrigues et al., 2017; Zug et al., 2019). 90 Second, anthropogenic sources of Cd in soils are attributed to a wide array of heavy metal-91 producing activities (Kabata-Pendias and Pendias, 2011). Byproducts from oil, mining, textile, and 92 agrochemical industries can explain the high content of Cd and other heavy metals in the 93 environment (Barraza et al., 2018). Agricultural practices, especially the production and use of 94 phosphate fertilizers sourced from Cd rich rocks are also identified as an anthropogenic source 95 of Cd in contaminated soils (Barraza et al., 2017; Capparelli et al., 2020; Chavez et al., 2015). 96 This study is the first of this scale in this region of Colombia and aims to expand the current 97 understanding of the dynamics of Cd accumulation in plant tissues and soils in northern Colombia. 98 Conducted on 30 farms along the north flank of the Sierra Nevada de Santa Marta (SNSM), this 99 study aims to (1) quantify the Cd levels in soils, a permanent litter layer made of rotting leaves 100 and cacao pods, soil rock fragments, and leaves from living trees; (2) evaluate the relationship 101 between different soil horizons and litter and plant Cd concentrations; (3) explore the potential 102 geogenic sources of Cd; (4) investigate the role of anthropogenic Cd and; (5) determine if high 103 Cd concentrations in plants can be attributed to a biocycling mechanism taking place by the cacao 104 tree. Overall, this study aims to contribute to a better understanding of the natural and human 105 5 factors affecting Cd accumulation in cacao crops of the SNSM, providing valuable information for 106 potential strategies for mitigation and crop selection based on local geology, soils, and 107 management practices. 108 109 2. Materials and Methods 110 2.1 Study Area 111 The study was conducted in the northern region of the SNSM, Colombia (Fig. 1). The 112 farms in this study are located at lower elevations, with a tropical climate that experiences 113 seasonal variations in temperature and precipitation. The region experiences a bimodal 114 distribution of precipitation with wet seasons during Apr-Jun and Oct-Nov, corresponding to the 115 meridional shifting of the Intertropical Convergence Zone across the tropics. The mean monthly 116 precipitation varies from a minimum of 1 mm in the dry seasons to a maximum of 1,000 mm in 117 the rainy seasons while the mean monthly temperature varies from 11.7 °C to 28 °C (Karger et 118 al., 2017). 119 The survey area is dissected by rivers and streams draining north towards the Caribbean 120 Sea (Fig. 1). Approximately a third of the study area is relatively flat and dominated by moderate 121 density agriculture. The remaining two-thirds are hilly and mountainous, extending up towards the 122 foothills of the SNSM Mountain Range. The study area is characterized by high relief, mostly in 123 the southern and central parts, and elevations ranging from 0 to 2,849 meters above sea level. 124 125 6 126 Figure 1. A. Map of South America highlighting Colombia. B. Map of Colombia with major cities and study 127 area highlighted. C. Relief map of the study area with sampling locations. Dashed black lines delineate the 128 drainage basins that make up the study area. 129 130 2.2 Regional Geology 131 The study area is located in the northern flank and alluvial flats of the Sierra Nevada de 132 Santa Marta, which is an uplifted crustal block located on the north Caribbean coast of Colombia 133 (Idárraga-García and Romero, 2010). The northwestern section of the SNSM consists primarily 134 of Paleogene plutonic and Cretaceous metamorphic lithologies from the Santa Marta geotectonic 135 province, while the northeastern portion is dominated by Jurassic, Triassic, and Paleogene 136 plutonic units, with minor occurrences of Miocene sedimentary and Paleozoic metamorphic 137 lithologies (Idárraga-García and Romero, 2010; Quandt et al., 2018; Tschanz et al., 1974). 138 7 The superficial geology of the flat parts of the study area is characterized primarily by 139 alluvial plain deposits where rivers coming from the SNSM flow into the Caribbean Sea (Fig. 2). 140 Meanwhile, the geology of the higher elevation section of the watersheds is dominated by 141 metamorphosed quartz-feldspar-rich gneisses, phyllites intercalated with minor outcrops of 142 quartz-sericite schists, and granodiorites that vary to quartz-diorites and quartz-monzonites 143 (Colmenares et al., 2007). The presence of siliceous sandstones, mudstones, siltstones, Jurassic 144 rhyolites, Quaternary colluvial deposits, and limestones can be observed towards the eastern 145 basins (Gómez Tapias et al., 2015; Servicio Geológico Colombiano, 2023). 146 147 Figure 2. Map displaying the underlying lithology of the study area divided into eastern and western basins. 148 Different colors indicate distinct lithologies. Data from the Colombian Geological Survey’s geological charts 149 12, 13, 19, and 20 at a 1:100,000 scale (Servicio Geológico Colombiano, 2023) 150 151 8 2.3 Sample Collection 152 The survey was conducted between March and April 2019 on 30 cacao farms. The 153 chosen farms were already part of the Cacao for Peace program, a crop substitution initiative led 154 by USAID and USDA which considered research on Cd one of its main components. At each 155 farm, two or three zones were determined, based on landforms identified in maps and on field 156 visits. After the zones were created, one site per zone at the scale of individual trees was randomly 157 selected for litter, leaves, and soil sample collection, If no different landforms were identified, the 158 different zones were selected depending on different accompanying subsistence crops or farm 159 management plans. One soil core per site was randomly extracted around the sampled cacao 160 tree at approximately 1 meter from the main tree trunk and below the drip line. 83 soil cores were 161 collected using a Dutch auger, separated into individual soil samples based on genetic horizons, 162 and described based on US Soil Survey standards and methods (Schoeneberger et al., 2012; 163 Soil Survey Staff, 2022). Bulk density was determined by driving a steel ring into the soil, 164 extracting an undisturbed core of known, and then using the soil’s weight to calculate bulk density 165 (Blake, 2015). The total sample count consisted of 83 topsoil (0-5 cm depth), 62 leaves (a 166 homogenized sample consisting of 10 leaves per tree), and 80 litter samples, collected before soil 167 coring. 168 To investigate the potential sources of Cd, different soil depths were analyzed. Topsoil 169 Cd, the first mineral soil horizon, was measured to determine if surface processes related to farm 170 management practices such as fertilizer or pesticide use, or Cd-bearing sediments being 171 deposited in flooding or irrigation could be responsible for high Cd levels. Subsurface soil 172 horizons, which refer to mineral horizons numbered 2 and deeper (i.e., all mineral soil horizons 173 except the surface horizon), were also analyzed and averaged to test for differences between 174 surface and subsurface processes. The bottom horizon soil Cd (concentration for every soil core 175 at the deepest horizon, in contact with underlying bedrock) was examined at farms with high soil 176 Cd to determine whether in-situ weathering of bedrock was a possible source of Cd. The 177 9 distinction between topsoil and subsurface soils was informed by previous research, which 178 indicates that the primary accumulation of Cd in cacao growing soils is confined to the top 15 cm 179 of the soil profile, where the majority of the tree’s fine and active roots are located (Arévalo-Gardini 180 et al., 2017; Argüello et al., 2019; Barraza et al., 2019; Chavez et al., 2016, 2015). For every core, 181 the bottom limit was either 200 cm, refusal, or bedrock. 182 183 During sampling, additional information about the age of the cacao cultivar, use of 184 fertilizers and/or pesticides, crops mixed with cacao, yields, tree genotype, and management 185 practices were collected from farm owners, managers, and/or workers through a questionnaire. 186 Leaf and litter samples were not collected at all farms, as these samples were added to the study 187 design after soil sampling had already begun. 188 189 2.4 Sample processing and analysis 190 Litter, soil, and leaf samples were analyzed at the Laboratory of Analytical Services 191 at the Alliance of Bioversity International and International Center for Tropical Agriculture (CIAT) 192 in Colombia. The soil samples were oven dried at 60 °C and sieved through a 2-mm stainless 193 steel mesh. For Cd analysis, soil samples were ground to pass a 0.5 mm stainless steel mesh. 194 Soil pH and electrical conductivity (EC) were measured in deionized water (1:1 soil/water) using 195 a pH/conductivity potentiometer. Soil organic matter was determined using a modified Walkley-196 Black method (Walkley and Black, 1934), where soil was treated with potassium dichromate and 197 sulfuric acid to heat the samples at 120°C for 2 hours. The remaining unreacted dichromate is 198 determined via spectrophotometry at 620 nm using an automated analyzer (Skalar®). Soil 199 texture was determined using the Buoyoucos hydrometer method (Bouyoucos, 1962). Available 200 phosphorus was analyzed by the Bray II method and measured via spectrophotometry at 880 201 nm using an automated analyzer (Skalar®). When soil pH was less than 5.5, Ca, K, Na, and Mg 202 were determined using a 1 M KCl extractant and analyzed using atomic absorption spectroscopy 203 10 (AAS). When soil pH was greater than or equal to 5.5, Ca, Mg and Na were extracted using 1 M 204 ammonium acetate and analyzed by AAS. For available Fe, Mn, and Cu, soil samples are 205 digested with a mixture of dilute hydrochloric acid (1M HCl) and sulfuric acid (5M H2SO4) and 206 analyzed via AAS. Available S was determined using 0.002M Ca(H2PO4)2 at pH 4 to extract and 207 analyze by barium chloride turbidimetry. Exchangeable Al in soils was extracted with 1M KCl and 208 measured by potentiometric titration while available B was determined by mixing the soil sample 209 with hot water, filtering the extract, adding azomethine-H reagent, and measuring the intensity of 210 the resulting yellow-green color using an ultraviolet–visible spectrophotometer. 211 Total Cd and Zn in soils were determined via an aqua regia digestion (3:1 ratio of HCl: 212 HNO3) and analyzed via inductively coupled plasma - optical emission spectrometry (ICP-OES). 213 The digestion procedure began by digesting a 125 mg aliquot sample in aqua regia (1:3 HCl: 214 HNO3) for 2 hours at 125°C on an open block. Total soil Cd was measured via Inductively 215 Coupled Plasma Optical Emission Spectroscopy (ICP-OES – Thermo Scientific iCAP 7200 Duo 216 ICP-OES Analyzer). 217 For determination of Cd in plant tissues (litter and leaves), samples are air dried and 218 grounded to pass a 0.5 mm stainless steel mesh. Using a pipette filler, 5.0 mL purified 65% 219 HNO3 was added, gently mixed, loosely capped, and pre-digested for at least 16 hours. Samples 220 are then heated at 90°C for 2 hours, then at 130°C until the digestate volume is reduced to 1-2 221 mL (1-1.5 hours). After adding 2 mL more 65% HNO3 and reheating at 135°C to re-concentrate, 222 3 mL HNO3:HClO4 (2:1) was introduced, and tubes were heated at 150°C for 1 hour, then 180°C 223 for 0.5 hours. Plant tissue Cd content is then measured via ICP-OES (Thermo Scientific iCAP 224 7200 Duo ICP-OES Analyzer). The limit of quantification (LOQ) was 0.065 mg kg-1 for soils and 225 0.0495 mg kg-1 for leaf and litter samples. 226 Rock fragment samples (diameter > 2 mm) collected after the initial soil sieving were 227 analyzed by ALS Geochemistry Laboratory by method ME-MS61. This method consists of a four-228 acid digestion process at 185°C for 150 minutes using nitric and perchloric acid to oxidize the 229 11 samples, hydrofluoric acid to dissolve the silicate lattice, and hydrochloric acid to evaporate the 230 solution before measuring the concentrations of elements via inductively coupled plasma mass 231 spectrometer (ICP-MS) for a near-total recovery of most analytes with an LOQ of 0.02 mg kg-1 232 for rock Cd (ALS, 2022). Some samples of soil and rock Cd were found to be below the LOQ. To 233 use these samples in the statistical analysis we assigned a random value between zero and one 234 half of the LOQ (0 – 0.0325 mg kg-1) following (Mrozek et al., 2006). 235 236 2.5 Statistical Analysis 237 Spatial visualization was performed using ArcMap 10.8 (ESRI, 2011), and statistical 238 treatment of data was first done using the Minitab software (Minitab LLC, 2019) and R statistical 239 software (R Core Team, 2022). Descriptive statistics and a Pearson correlation analyses were 240 calculated for Cd in soils, leaves, and litter using the “pastecs” (Philippe and Ibanez, 2018) and 241 “corrplot” (Wei and Simko, 2017), respectively. Mean comparison analysis was carried out 242 through the ANOVA test using the r-package “stats” (R Core Team, 2022) followed by a Fisher 243 post-hoc test. Significant tests for mean comparison analyses were carried out using the 244 “PMCMRplus” r-package at a 0.05 significance level (Pohlert, 2021). Soil Cd profile plots were 245 drawn using the “aqp” r-package (Beaudette et al., 2013). Geological data was digitized from 246 multiple geological charts published by the Colombian Geological Survey at a scale of 1:100,000 247 (Gómez et al., 2015). 248 To explore the relationship between multiple soil physical and chemical properties and 249 total soil Cd, a stepwise regression analysis was performed. The initial explanatory variables 250 were: clay (g kg-1), silt (g kg-1), sand (g kg-1), pH, bulk density (g cm-3), soil organic matter (g kg-251 1), total Zn (mg kg-1), Ca (cmol kg-1), P (mg kg-1), Mg (cmol kg-1), K (cmol kg-1), Fe (mg kg-1), Mn 252 (mg kg-1), Cu (mg kg-1), B (mg kg-1), S (mg kg-1), Na (cmol kg-1), rock fragment Cd (mg kg-1), and 253 geology as a categorical value with igneous, metamorphic and sedimentary as the possible levels. 254 12 The variables included in the best model were selected to maximize the coefficient of 255 determination (R2) while minimizing the Bayesian Information Criterion (BIC). 256 257 3. Results 258 3.1 Cadmium in soils 259 The analysis of 390 genetic soil horizons in the study area revealed generally low total soil 260 Cd concentrations (less than 1 mg kg-1), except for topsoil samples from one farm (Fig. 3). 261 Moreover, the occurrence of soil samples with Cd concentrations higher than 0.8 mg kg-1 was 262 limited to two out of 30 farms, specifically farms 26 and 90, in the western watersheds of the study 263 area (Fig. 3 and 6). The mean total Cd for the topsoil samples across the study area was 0.12 (± 264 0.01) while the mean Cd concentration for subsurface horizons (second horizon and deeper) was 265 0.05 (± 0.01) mg kg-1. The maximum Cd concentration for topsoil samples analyzed was 1.11 mg 266 kg-1 at farm 90 while the maximum Cd content for subsurface horizons was 0.51 mg kg-1. 267 13 268 Figure 3. Graphic representation of Cd content based on genetic soil horizons for all farms (F = Farm, P 269 = Soil pit) and soil cores (S = Soil sampling site per farm). 270 14 3.2 Cd in leaves and litter 271 In this study, leaf samples taken from live trees (n = 62) exhibited a mean Cd concentration 272 of 0.43 (± 0.09) mg kg-1, ranging from 0.06 mg kg-1 to 4.04 mg kg-1, the latter measured at site one 273 in farm 90. The second-highest leaf Cd concentration of 3.29 mg kg-1 was observed at site 2 in 274 farm 90 while, the third-highest value, outside of farm 90, was 1.51 mg kg-1 at farm 26, 3 km away 275 from farm 90. The mean leaf Cd concentration was found to be significantly higher (3 to 4 times) 276 than Cd concentrations in both topsoil and mean subsurface soil horizons (Fig. 4). 277 Litter samples (n = 80) consisting of fallen leaves collected under the drip line of the 278 sampled tree showed a mean concentration of 0.41 (± 0.05) mg kg-1, ranging from a minimum of 279 0.03 mg kg-1 to a maximum of 2.68 mg kg-1, recorded at site one in farm 90. Consistent with the 280 trend observed in leaf Cd concentrations, the second-highest litter Cd values were also found at 281 farm 90, with a concentration of 2.39 mg kg-1, significantly higher than the third-largest litter Cd 282 value of 1.39 mg kg-1 found in farm 27. The concentration of Cd in litter samples was significantly 283 higher than the Cd levels in topsoil and subsurface soil horizons but showed no significant 284 difference against leaf Cd concentrations (Fig. 4). 285 286 Figure 4. Boxplots of mean and distribution of Cd for leaves, litter, topsoil, and subsurface soils. Different 287 letters indicate statistically significant differences (p < 0.05) between the means of each group measured 288 via ANOVA Fisher mean comparisons. The boxplots show the interquartile range and median, and the 289 whiskers represent the standard error. 290 15 291 3.3 Relationships between plant, litter, and soil Cd. 292 Our findings show that leaf and litter Cd concentrations are significantly higher than topsoil 293 or subsurface soil Cd, while there is no significant difference between topsoil and subsurface soil 294 Cd concentrations, nor between leaf and litter Cd concentrations (Fig. 4). The Cd concentrations 295 in leaves exhibited significant linear correlation with total soil Cd for every horizon measured 296 across all farms, regardless of varying horizons, thicknesses, and depths (Fig. 5). The correlation 297 between leaf Cd and total soil Cd in the deepest horizon (closest to bedrock) was stronger (r2 = 298 0.81, p < 0.05) than the correlation between leaf Cd and total soil Cd in the first horizon (r2 = 0.66, 299 p < 0.05). 300 301 302 Figure 5 Correlations between total soil Cd per soil horizon (Hz) and leaf Cd concentrations across all 303 sites. * = Significant at an alpha of 0.01. 304 16 305 At the individual farm scale, our findings indicate that leaf Cd concentrations were 306 significantly higher (p < 0.05) than soil Cd concentrations in all farms except farm 26, measured 307 via ANOVA Fisher mean comparisons (Fig. 6). Moreover, litter Cd concentrations were significantly 308 higher than topsoil Cd in all farms, except farms 25 and 26, where no significant difference was 309 observed. Litter Cd concentrations were significantly higher than subsurface total soil Cd for all 310 farms analyzed. In the eastern watersheds, both leaf and litter Cd concentrations are within a 311 similar range, which is four to five times higher than both topsoil and subsurface soil. In the 312 western basins, in most farms, except for farms 26 and 90, both topsoil and subsurface soil Cd 313 concentrations were under 0.5 mg kg-1. 314 17 315 Figure 6 Top panel: Eastern watersheds and Cd concentration in leaf, litter, topsoil, and subsurface soil (left to right) for each farm. Bottom panel: 316 Western watersheds and Cd concentration in leaf, litter, topsoil, and subsurface soil (left to right) for each farm. F = Farm. P = Soil pit. * Farms where 317 leaf samples were not collected. ** Farms where leaf and litter samples were not collected. 318 18 3.4 Cadmium and geology 319 The results of this study show that the Cd content in soils overlying sedimentary rocks (n 320 = 28) was significantly higher than those overlying older igneous rocks (n = 28) for every genetic 321 soil horizon except for the topsoil (Fig. 7). In soils overlying sedimentary rocks (Quaternary alluvial 322 deposits and Neogene shallow marine units) the mean total Cd concentration in the topsoil was 323 0.13 mg kg-1, which was not significantly higher than topsoil samples overlying metamorphic rocks 324 (Cretaceous gneisses and schists) with a mean total Cd value of 0.12 mg kg-1 or igneous rocks (n 325 = 27; Paleogene and Triassic plutonic intrusions) with a mean total Cd concentration of 0.04 mg 326 kg-1. However, for all subsurface horizons, soils overlying sedimentary rocks had significantly 327 higher Cd levels than soils overlying igneous rocks, but not metamorphic rocks (Fig. 7). 328 329 Figure 7. 95% Confidence interval around the mean plot for Cadmium concentration in leaves, litter, surface, 330 and bottom soil horizons for all farms sampled (n = 30) grouped by underlying bedrock type. Different letters 331 indicate statistically significant differences (p < 0.05) between the means of each group measured via 332 ANOVA Fisher mean comparisons. Note how the mean total soil Cd is significantly higher in soils overlying 333 sedimentary lithologies than those overlying igneous and metamorphic rocks below the first horizon. 334 19 The comparison of Cd concentrations in soil and rock fragments at the same horizon and 335 farm revealed that rock fragment Cd concentrations were relatively low and within a similar range 336 as soil Cd concentrations (Fig. 8). At all farms, except 3, 26, and 39, rock Cd concentrations are 337 higher than total Cd concentrations measured in soils from the same depth. The mean Cd 338 concentration for rock fragments was 0.07 (±0.02) mg kg-1 for igneous rocks, 0.07 (±0.02) mg kg-339 1 for metamorphic rocks, and 0.1 (±0.04) mg kg-1 for sedimentary rocks. Classifying rock fragments 340 by lithology, we observed a mean total Cd concentration of 0.13 (±0.05) mg kg-1 for quaternary 341 alluvial units, 0.07 (±0.02) mg kg-1 for gneiss, 0.04 (±0.02) mg kg-1 for sandstone and 0.03 (±0.00) 342 mg kg-1 for granodiorite. 343 344 Geology N Mean Cd (±SE of mean) Lithology N Mean Cd (±SE of mean) [mg kg-1] [mg kg-1] Sedimentary 12 0.1 (±0.03) Qt Alluvial (Sed) 7 0.13 (±0.06) Igneous 3 0.03 (±0.00) Gneiss (Met) 20 0.07 (±0.02) Metamorphic 20 0.07 (±0.02) Sandstone (Sed) 5 0.04 (±0.02) Granodiorite (Ign) 3 0.03 (±0.00) Table 1. Near-total mean Cd concentration and standard error for the mean for rock fragments taken after 345 soil sieving by rock type (left) and lithology (right). Sed = Sedimentary, Met = Metamorphic, Ign = Igneous. 346 Cadmium concentration was not significantly different (p >0.05) between rock types. 347 348 20 349 Figure 8. Cd concentration of soil and rocks from the same horizon and depth in the study area. 350 351 3.5 Relationships of soil Cd with rock fragment Cd and other soil properties 352 A correlation analysis for every soil horizon in which rock fragments were tested reveals 353 a significant positive relationship between soil Cd and rock Cd. Additionally, rock fragment Cd 354 showed a significant correlation with litter Cd, while no significant correlation was observed with 355 leaf Cd. Total Soil Cd shows a significant linear relationship with total soil Zn and Bray II available 356 soil P. Meanwhile, soil Cu concentrations are significantly correlated with leaf and litter Cd and 357 not with soil Cd (Table 2). 358 359 Litter Cd 0.88* Soil Cd 0.77* 0.71* Total Soil Zn 0.55* 0.60* 0.51* Soil P 0.70* 0.70* 0.58* 0.30 Soil Cu 0.44* 0.54* 0.15 0.23 0.43* 21 Rock Cd 0.29 0.46* 0.60* 0.16 0.31 0.04 Leaf Cd Litter Cd Soil Cd Total Soil Zn Soil P Soil Cu Table 2 Pearson correlation matrix (r) for rock fragment Cd, litter Cd, leaf Cd, and soil elemental 360 chemistry. Bold values and * = significant at al alpha of 0.05. 361 362 Differentiating the correlation analysis by topsoil and subsurface soils allows for observing 363 different processes that may be controlling soil Cd at different depths. When looking at the 364 differences between topsoil and subsurface soil elemental and texture relationships, results show 365 that soil Cd exhibited no significant correlation with sand, clay, and silt content in surface soils. In 366 contrast, the relationship was stronger between soil Cd and silt and sand content at the 367 subsurface horizons. Notably, the correlation between clay content and soil Cd remains not 368 significant at both topsoil and subsurface horizons. Likewise, no significant correlations between 369 soil Cd and pH at both topsoil and subsurface layers were found. 370 371 372 22 Figure 9. Correlation matrix testing multiple soil variables (Clay, silt, and sand percentages, pH, SOM, total 373 Zn and available P, K, Ca, Mg, Fe, Mn, Cu, B, and S) and total soil Cd. Left panel = Topsoil (Horizon 1). 374 Right panel = Subsurface soils (Horizon 2 and deeper). The upper right of each matrix displays the Pearson 375 correlation value (r-value). The lower left of each matrix displays the p-value, dominant slope, color codes, 376 and asterisks for significance: p-values (0.001, 0.01, 0.05, 0.1) <=> symbols (“***”, “**”, “*”, “ ”). 377 378 A stepwise regression analysis revealed that different factors are associated with Cd 379 distribution in the soil profile (Table 3). In the topsoil, the model with the highest predictive power 380 was obtained when considering pH, Bray II available P (mg kg-1), available Fe (mg kg-1), available 381 Cu (mg kg-1), Sand (%), available S (mg kg-1), and total Zn (mg kg-1). Meanwhile, for subsurface 382 horizons, the model with the highest predictive power was obtained when considering Bray II 383 available P (mg kg-1), pH, available Cu (mg kg-1), available Mn (mg kg-1), available K (mg kg-1), 384 Clay (%), bulk density (g cm-3), soil organic matter (g kg-1), and geology as a categorical variable 385 (igneous, metamorphic, or sedimentary). 386 387 Soil Horizons Topsoil (HZ = 1) R2 BIC Subsurface (HZ ≥ 2) R2 BIC pH, P, Fe, Cu, Sand (%), S, total Zn 0.55 -55.55 P, pH, Cu, Mn, K, Clay (%), BD, SOM, Geology 0.80 -195.92 P, Fe, Cu, Sand (%), S, total Zn 0.53 -57.86 pH, Cu, Mn, K, Clay (%), BD, SOM, Geology 0.79 -198.30 Fe, Cu, Sand (%), S, total Zn 0.52 -60.47 Cu, Mn, K, Clay (%), BD, SOM, Geology 0.78 -199.10 Cu, Sand (%), S, total Zn 0.51 -62.56 Mn, K, Clay (%), BD, SOM, Geology 0.77 -199.34 Sand (%), S, total Zn 0.49 -64.60 K, Clay (%), BD, SOM, Geology 0.75 -199.82 S, total Zn 0.47 -65.98 Clay (%), BD, SOM, Geology 0.73 -198.45 total Zn 0.24 -46.44 BD, SOM, Geology 0.70 -195.71 SOM, Geology 0.66 -193.68 Geology 0.60 -186.24 Table 3. Stepwise regression model for topsoil and subsurface soil horizons. BIC = Bayesian information 388 criterion. A smaller BIC indicates a better fitting model. 389 23 390 3.6 Variability between farms 391 The range of Cd concentrations in leaves, litter, topsoil horizons, and subsurface soil 392 showed no significant differences at most farms (p < 0.05; Fig. 10). However, farms 90 and 26 393 were the exception, with significantly higher topsoil total Cd (0.65 and 0.51 mg kg-1, respectively). 394 Surrounding farms, 100 to 200 meters away, are in the same lithological unit (gneiss bedrock) but 395 still show significantly lower Cd concentrations (Fig. 11). 396 397 398 Figure 10. Boxplots of Cd concentration in leaves (top left panel), litter (bottom left panel), topsoil (top right 399 panel), and subsurface horizons (bottom right panel) for all farms sampled (n = 30). Different letters in the 400 same panel indicate statistically significant differences (p < 0.05) between the means of each group 401 measured via ANOVA Fisher mean comparisons. Note the difference in the range values for the Cd levels 402 for each panel. 403 404 24 405 The cluster of farms surrounding Farm 90 and Farm 26, showed significantly lower Cd 406 concentrations in leaves, soils, and litter (Fig. 11). In live leaves taken from the trees, values at 407 farm 90 (2.63 mg kg-1) are significantly higher than those at farms 38 (0.29 mg kg-1), 89 (0.13 mg 408 kg-1), 40 (0.35 mg kg-1), and 91 (0.21 mg kg-1), the latter being no further than 1 km and overlying 409 the same bedrock lithology. For litter Cd levels, farm 90 showed significantly higher values (2.06 410 mg kg-1) than all farms in the study area. Both farms 90 and 26 had significantly higher topsoil Cd 411 levels than neighboring farms (0.65 and 0.51 mg kg-1, respectively). In addition, Cd levels for the 412 bottom soil horizon, in contact with the underlying lithology at farm 90 was significantly higher 413 (0.21 mg kg-1) than the neighboring farms, regardless of underlying lithologies or watersheds. 414 415 25 416 Figure 11. A) Farm 90 underlying lithology, stream network, and surrounding farms. B) Farm 90 altitude, 417 stream network, and surrounding farms. C) Boxplots for leaves (top left), litter (bottom left), topsoil (top 418 right), and bottom soil horizons (bottom right) for farm 90 and surrounding farms. Different letters indicate 419 statistically significant differences (p < 0.05) between the means of each group measured via ANOVA Fisher 420 mean comparisons. Note the difference in the range values for the Cd levels for each panel. 421 422 26 4. Discussion 423 4.1 Cadmium in soils of the SNSM 424 In comparison to other cacao-growing regions of Colombia, our study area showed lower 425 ranges of plant and soil Cd. In the Antioquia department, Gil et al. (2022) reported soil total Cd 426 concentrations that ranged from 1.22 to 2.03 mg kg-1, while for litter and beans concentrations 427 ranged from 0.43 to 3.34 mg kg-1 and 0.62 and 1.44 mg kg-1, respectively. Similarly, Rodríguez-428 Albarrcín et al. (2019) reported higher values of Cd from cacao producing areas of central 429 Colombia, with mean Cd concentrations of 0.63, 0.66, 0.65, and 0.61 mg kg-1 in soils (0-30 cm), 430 litter, beans, and leaves, respectively. Bravo et al. (2018) reported even higher values of soil Cd 431 concentrations of 3.74, 3.50, 2.76, and 1.16 mg kg-1 Cd in the Boyacá department, the towns of 432 San Vicente de Chucurí, El Carmen de Chucurí, and Arauca, respectively. However, they found 433 lower values in farms in Saravena (1.16 mg kg-1), Tame (0.74 mg kg-1), Arauquita (0.66 mg kg-1), 434 Pauna (0.73 mg kg-1), and Mapiripí (0.50 mg kg-1) (Bravo et al., 2018). Hence, despite some 435 outlier farms, the farms in the north flank of the SNSM exhibit lower total Cd levels compared to 436 other cacao growing areas of Colombia. 437 438 4.2 Relationships between soil Cadmium and other soil parameters 439 Contrary to findings from previous studies in Colombia and Latin America (Argüello et al., 440 2019; Gil et al., 2022; Vanderschueren et al., 2021), we found no significant correlations between 441 soil Cd and pH and only a weak relationship with soil organic matter content. These results are 442 likely due to the already small amounts and variations in soil Cd, which may diminish any effect 443 of soil texture, pH, or soil organic matter on soil Cd concentrations. 444 Heavy metal levels in soils, including Cd, generally increase with increasing clay 445 concentration and finer textures (Chavez et al., 2015; Gramlich et al., 2017; He et al., 2015). 446 However, we found no significant correlations between clay percentages and soil Cd for both 447 topsoil and subsurface soil horizons. Silt and sand percentages and total soil Cd had only a weak 448 27 relationship at subsurface horizons (r = 0.3, p < 0.001; r = -0.26, p < 0.001, respectively), while 449 topsoil Cd shows no relationship with either sand or silt percentages (Fig. 9). The significant 450 relationship between subsurface soil Cd and sand content was negative, but positive for silt 451 content. This suggests that soil texture determines total soil Cd for the subsurface but not in the 452 topsoil. 453 Previous research showed that sand had a negative relationship with Cd retention in soils, 454 while no significant correlation was found with silt content (Hooda and Alloway, 1998). The soils 455 in the study area have low levels of organic matter (0.9–1.29%; 95% confidence interval around 456 the median) and are highly weathered and sandy (45.07–49.17%; 95% confidence interval around 457 the median for sand content; 17.19-19.23%; 95% confidence interval around the median for clay 458 content). These texture characteristics suggest there are limited clay-sized particles and organic 459 matter, which typically provide the major cation exchange capacity in soils. However, we found 460 that total soil Cd correlates with silt content, which suggests that in these soils, the silt fraction 461 likely serves as the main exchange site for Cd. The limited exchange capacity sites might also 462 explain why total soil Cd concentrations remain low in the study area as any external Cd entering 463 the system has restricted binding sites to remain in the soil and potentially transfer to crops. 464 Stepwise regression results indicate that the processes controlling Cd distribution in 465 subsurface horizons are primarily driven by geogenic Cd interactions with regional lithology and 466 other soil properties (Table 3). However, for topsoil Cd, the factors that explain the variation in 467 soil Cd are different and their predictive power is weaker. This could imply that soil Cd in the 468 topsoil is governed more by management practices or the bioaccumulation of Cd via cycling of 469 Cd through the leaves, litter, and topsoil. Meanwhile, Cd concentrations for subsurface horizons 470 are mediated more by pedogenic processes such as eluviation, clay illuviation, and bedrock 471 weathering. 472 In the subsurface horizons, we found a significant positive correlation between total soil 473 Cd and soil Cu (Fig. 9). Cd, Zn, and Cu have similar mobility and vertical migration patterns (Gibb 474 28 and Cartwright, 1982) and Cu can be found as a trace metal in rocks where Cd is present in 475 quantifiable amounts, such as black shales (Dumoulin et al., 2011) and sulfide bearing ores 476 (Tabelin et al., 2018). 477 478 4.2.1 The Role of Zinc 479 We found a significant positive correlation between total soil, leaf, and litter Cd and total 480 soil Zn concentrations, which suggests that synergism between Zn and Cd in plant tissues and 481 soils contributes to the accumulation of Cd in cacao plants (Arévalo-Gardini et al., 2017) (Table 482 2). This observation is consistent with the results of Rodríguez-Albarrcín et al. (2019), who 483 reported a significant and strong positive correlation (r = 0.70, p < 0.05) between soil Zn and Cd 484 in cacao beans in central Colombia. 485 Since Zn and Cd share ionic similarities, both elements could be taken simultaneously by 486 plant exchange sites where Zn has a structural or metabolic function (Arévalo-Gardini et al., 2017; 487 Tang et al., 2014). While Zinc and Cadmium are considered elemental analogs (Kabata-Pendias 488 and Pendias, 2011), a Zn deficiency can trigger Cd uptake by plants (Brown et al., 1995; Chaney, 489 2010; Vanderschueren et al., 2021). However, in a review article, Vanderschueren et al., (2021) 490 did not identify total soil Zn as a significant factor to predict bean Cd concentrations in cacao. The 491 authors found that Cd accumulation in cacao producing areas of Latin America was still 492 considerable, even when a Zn deficiency was not present. 493 The simultaneous adsorption of Zn and Cd suggested by our findings can be explained 494 by the common uptake and transport pathways shared by both elements. Heavy metal transit 495 from soil to plant tissue follows a simplified pattern consisting of (i) metal transport from soil 496 solution to plant root, (ii) transport across the root and loading into the xylem, and (iii) transport of 497 the element to the leaf tissue via the xylem where it is then stored in leaf epidermal and mesophyll 498 cells (Papoyan et al., 2007). 499 29 The transporters involved in these metal translocation processes, such as members of the 500 ZIP transporter protein family (Han et al., 2006) like the ZNT1 protein (Pence et al., 2000), or the 501 HMA2 and HMA4 ATPases (Nouet et al., 2015), can absorb and transport these elements 502 simultaneously, given the ionic similarities between Cd, Zn, and Cu (Arévalo-Gardini et al., 2017; 503 Papoyan et al., 2007). This explanation supports the positive correlation between these elements 504 in our study. Furthermore, none of these elements exhibit high values in soils, which could negate 505 the competition effect described by other studies (Brus et al., 2005; Castro et al., 2015; Gramlich 506 et al., 2017; Grant et al., 1998). Hence, our results suggest that when there is not a Zn deficiency 507 or surplus, the plant absorbs ionically similar metals such as Cu, Zn, and Cd simultaneously. 508 509 4.3 Cadmium and Geology 510 Our results agree with previous studies that have found that sedimentary rocks, especially 511 organic-rich lithologies such as dark shales and siltstone, tend to have higher concentrations of 512 total Cd compared to igneous and metamorphic rocks (Birke et al., 2017; He et al., 2004; Kabata-513 Pendias and Mukherjee, 2007; Kabata-Pendias and Pendias, 2011; McLaughlin and Singh, 1999; 514 Thornton, 1981; Traina, 1999). 515 Hydrothermal alteration can also be a source of natural Cd contamination in rocks. This 516 alteration occurs when there is extreme heat from magmatic intrusions, a source of water, and an 517 extensive fault network that brings water in contact with local lithology to form superheated and 518 solute rich solutions (Tabelin et al., 2018). This can lead to metals like Cd, Zn, and Cu being 519 exchanged with Ca in the lattice of the minerals of the host rock (Mar and Okazaki, 2012). This 520 process has been shown to be responsible for Cd accumulation in phosphate rocks from which 521 phosphate fertilizers are sourced, becoming a geogenic source of Cd when present in the area 522 or an anthropogenic source of Cd when phosphate fertilizers sourced from contaminated rocks 523 are applied (Mar and Okazaki, 2012; Nathan et al., 1997, 1996; Nziguheba and Smolders, 2008; 524 Schipper et al., 2011). 525 30 Analysis of the geology reveals that igneous and metamorphic rocks present in the study 526 area exhibit localized hydrothermal alteration of the host rocks, potentially leading to increased 527 concentrations of Cu, Zn, and Cd. These alterations are supported by the presence of chalcopyrite 528 and pyrite in the western sections of the gneiss unit that is found in the upper parts of the 529 watersheds of the study area reported by Colmenares et al. (2007). Likewise, disseminated 530 sulfide ore deposits with concentrations of pyrite and chalcopyrite have also been found on the 531 south flanks of the SNSM, as well as in the Rhyolite units found in the eastern part of the study 532 area (Colmenares et al., 2007). However, the low Cd concentrations in rock fragments and 533 subsurface soils suggest that this may only be a minor source of Cd in the cacao farms of the 534 study area, compared to other sources. 535 Overall, we observed significant differences in Cd concentrations in subsoil samples 536 overlying metamorphic, igneous, or sedimentary rocks. The geological history of plutonic 537 intrusions, host rock lithologies, and faulting suggests the potential for geogenic Cd via 538 hydrothermal alteration. However, no significant differences were found in the Cd concentration 539 for topsoil and leaves. Likewise, Cd concentrations remain low in rock fragments and deep soils 540 in contact with the bedrock. Our results show no significant differences in Cd levels in leaves and 541 topsoil samples between bedrock lithology units, suggesting that any influence of parent material 542 on soil Cd concentrations is muted by a stronger influence of surface processes. This indicates 543 that the anthropogenic influence and the bioaccumulation of Cd over time may have weakened 544 the geogenic signature of Cd levels in the farms of this study. 545 546 4.4 Variability within farms 547 Excluding Farm 90 and Farm 26, which displayed high Cd concentrations, we observed 548 no significant differences between farms in the vicinity of the 2 farms with comparatively high Cd 549 contents (Fig. 11). Soil organic matter was not significantly different between farms surrounding 550 Farm 90, indicating this soil property could not account for the observed differences. The mean 551 31 topsoil pH at Farm 90 was 5.4, which was not significantly different from that of surrounding farms 552 such as Farm 91, 40, and 89, which had mean pH values of 5.6, 5.6, and 5.7, respectively. 553 However, Farm 90 exhibited a significantly higher total soil Zn than surrounding farms, which 554 suggests that the mechanisms responsible for the accumulation of Cd at significantly higher rates 555 in soils and plants at farm 90, could also be responsible for observed differences in soil Zn. 556 The management history at Farm 90 and the surrounding farms was studied to assess the 557 potential contribution of certain management practices. At Farm 90, cacao trees were planted 558 alongside mandarin trees (Citrus reticulata), orange trees (Citrus sp.), and Gliricidia sepium as 559 shade trees. The farm owner reported no use of synthetic fertilizers that could have potentially 560 contributed to high levels of Cd. The owner also reported that farm 90 has been under cacao 561 production “for generations” (Farm owner; personal communication), which is longer than the 562 average cacao producing farm in the area. For comparison, the approximate age reported by farm 563 owners or managers of the cacao cultivars of surrounding farms is 13 years (Farm 89), 9 years 564 (Farm 40), 9 to 12 years (Farm 39), 9 to 12 years (Farm 38), and 15 years (Farm 26). These 565 findings suggest that bioaccumulation of Cd at the surface may have occurred over decadal 566 scales at Farm 90, which may explain why the Cd levels were higher in litter, leaves, and topsoil 567 compared to nearby farms. Likewise, higher subsurface soil Cd at farm 90 may be explained by 568 the age of the cacao cultivar, as Cd from the topsoil and the litter layer slowly migrates downwards 569 in the soil profile by clay and oxide illuviation processes (Quezada-Hinojosa et al., 2015). 570 571 4.5 Cadmium biocycling by the cacao tree 572 Cd and other heavy metals are transported to leaves and pods via the xylem, where it is 573 stored in leaf epidermal and mesophyll cells (Papoyan et al., 2007). Research indicates that Cd 574 in nibs and beans originates from phloem redistribution from stems, leaves, or branches, which 575 hold the largest fraction of total plant Cd (Blommaert et al., 2022; Vanderschueren et al., 2023). 576 Findings from Blommaert et al. (2022) show that the main pathway for Cd accumulation in beans 577 32 is not xylem transport from the roots, but rather a phloem-mediated transport from Cd pools in the 578 bark. This means Cd is taken from the soil via the roots and transported to the branches and bark 579 where it is then transferred to leaves and pods. This implies that leaf Cd displays the maximum 580 possible value for bean Cd. Therefore, this study used Cd concentration in leaves as a proxy for 581 bean Cd concentration, used for chocolate production to which the new EU regulations apply. 582 Plant and litter Cd concentrations two to three times higher than topsoil Cd, plus a strong 583 significant positive correlation between the concentration of Cd in leaves and litter with topsoil 584 samples, supports the hypothesis of active Cd biocycling and bioaccumulation processes 585 occurring in cacao cultivars of our study area (Fig. 12). The warm and humid tropical environment 586 of our study area promotes high rates of litter decomposition which facilitates Cd cycling from 587 fallen leaves and pod husks, into the topsoil and absorbed again by plant surface roots (Gramlich 588 et al., 2018). As a result, after every harvest cycle and leaf fall, Cd accumulates on upper soil 589 surfaces, leading to increased plant uptake over the long term. This phenomenon has been 590 observed by other researchers who reported the enhanced Cd phytoavailability resulting from the 591 recycling of highly available Cd in the litter layer (Barraza et al., 2019; Gramlich et al., 2018; 592 Imseng et al., 2019; Meter et al., 2019; Rodríguez Albarrcín et al., 2019). 593 The process of bioaccumulation (i.e. plant concentrations that exceed soil concentrations) 594 can occur in farms that have been under cacao cultivation for longer periods of time. 595 Vanderschueren et al (2021) derived from a meta-analysis of multiple articles reporting Cd fluxes 596 from the litter layer that the annual Cd flux flowing from the litter layer to the topsoil is close to 10 597 g Cd ha-1 y-1, much higher than what would be derived from fertilizers or geogenic sources. In 598 Bolivia, the enrichment of the topsoil with Cd required approximately 50 years of litter 599 decomposition (Gramlich et al., 2017; Vanderschueren et al., 2021). In Peru, the use of fertilizers 600 with trace levels of Cd plus an added biocycling mechanism was found to increase topsoil Cd 601 levels and was suggested as the main source of Cd contamination in the northern part of the 602 33 country (Guarín et al., 2023). In the Magdalena basin in the Antioquia department of Colombia, a 603 study showed highly significant correlations (p < 0.01) between Cd in cacao beans and soil litter 604 Cd content (r = 0.91), the altitude of the farms (r = -0.82) and the age of the plant (r = -0.62), 605 further supporting the idea that the age of the cultivar contributes to the bioaccumulation process 606 taking place at the surface (Gil et al. 2022). Given the relationship between approximate cultivar 607 age and plant and soil Cd concentrations, the results of this study suggest that bioaccumulation 608 of Cd over time may be the leading process behind high Cd levels in cacao plant tissues. 609 610 Figure 12. Correlation between the topsoil Cd concentrations and the Cd levels 611 seen in plant leaves and plant litter overlying the soil profile. Litter is shown in red, and leaves are shown 612 in black (p < 0.05). Note the difference between scales for Cd in leaves (black - left) and litter (red - right). 613 34 The results of this study suggest that geogenic Cd seen in some farms of the study area 614 may kickstart and amplify the bioaccumulation effect evidenced in cacao cultivars. Cadmium 615 concentrations for the bottom horizon, far from the effects of surface processes, were significantly 616 higher at Farm 90 compared to surrounding farms, even though farms nearby overlie the same 617 lithologic unit, and occur on the same topographic position (backslope). Likewise, the rock 618 fragments from Farms 26 and 90 exhibit a comparatively higher Cd content than those from 619 neighboring farms, suggesting that these farms possess a larger initial pool of geogenic Cd. With 620 time, the process of Cd biocycling amplifies the pre-existing elevated concentrations of soil and 621 rock Cd, resulting in a more pronounced accumulation of Cd in the plants at these farms, 622 compared to surrounding farms, which have a smaller potential pool of geogenic Cd. However, 623 further research on the mineralogy of soils and rocks of this area is needed to explore a potential 624 localized hydrothermal alteration of the gneisses underlying these specific farms or other possible 625 soil formation processes that could explain this anomaly. Additionally, further exploration into the 626 weathering stages of underlying bedrock and rock fragments needs to be researched to determine 627 how available is this pool of geogenic Cd to cacao plants of the study area. 628 629 4.7 Recommendations 630 The findings of this study are promising for the continuation and expansion of the voluntary 631 crop replacement programs in the north SNSM of Colombia. The low levels of Cd in both plants 632 and soils in most farms sampled in this study are encouraging for farmers in the region who rely 633 on the sale of cacao to profitable foreign markets like the European Union. 634 Our findings suggest that the biocycling of Cd between the litter layer, topsoil, and plant 635 tissues may be responsible for high Cd concentrations in cacao plants. In theory, removing the 636 permanent litter layer could be implemented as a mitigation strategy, but the practicality of 637 removing leaf litter to mitigate cadmium contamination in cacao cultivars is currently questionable. 638 The benefits provided by this organic layer, including increased soil organic matter, nutrient 639 35 availability, and erosion control make it unlikely that smallholder family and indigenous farmers 640 will adopt such practice (Meter et al., 2019). Furthermore, the additional labor required would 641 likely disproportionally fall on women, who are often responsible for a greater share of physical 642 labor on cacao farms (Abbott et al., 2018). Therefore, a more feasible approach to Cd mitigation 643 would involve implementing strategies such as phytoremediation (Casteblanco, 2018; He et al., 644 2015; Liu et al., 2018), liming to raise soil pH and immobilize soil and litter Cd (Hamid et al., 2019; 645 Ramtahal et al., 2018; Vanderschueren et al., 2021), biochar to raise soil pH and immobilize soil 646 and litter Cd (Ramtahal et al., 2019, 2018), and exploring alternatives to potentially Cd bearing 647 fertilizers (Chen et al., 2007). Additionally, other more labor intensive management practices for 648 Cd mitigation would include replacing old and declining cultivars with high Cd contents with 649 younger cacao trees from genetic varieties that favor less Cd in the final product (Engbersen et 650 al., 2019; Lewis et al., 2018; McLaughlin et al., 2021; Vanderschueren et al., 2021). Lewis et al. 651 (2018) found that grafting high yielding scions onto low Cd accumulating stock plants can be a 652 short-term strategy to mitigate Cd bioaccumulation and possibly increase yield. 653 Cooperatives can play a crucial role in assisting farmers in farms with high Cd by diluting 654 their product with cacao beans coming from low Cd farms (Vanderschueren et al., 2021). This 655 strategy could mean that cacao production is still viable for these farmers and continued harvests 656 can be commercially feasible while Cd mitigation measures are implemented. However, selling to 657 the lower-paying bulk cacao market should be a last resort for farmers as premium origin and fine 658 flavor cacao generates higher income from higher selling prices. 659 660 5. Conclusions 661 Our study examined the concentration of cadmium in soil and plants in the cacao growing 662 region north of the Sierra Nevada de Santa Marta, Colombia. Our results revealed generally low 663 total soil Cd concentrations, with only a small number of samples exceeding recommended limits. 664 The mean topsoil and subsurface Cd concentration in the study region was lower than in other 665 36 cacao growing areas in Colombia and Latin America. Our analyses also found that the Cd 666 concentration in soils overlying sedimentary rocks was significantly higher than those overlying 667 older metamorphic rocks for every genetic soil horizon, except for the topsoil. 668 Subsurface soil and rock fragment Cd and elemental concentrations indicate that the 669 processes controlling Cd distribution in the subsurface horizons seem to be primarily driven by 670 geogenic Cd, determined by the weathering of regional lithology, and its interaction with other soil 671 properties such as soil texture, P, Zn, and Cu concentrations with a hypothesized contribution of 672 pedogenic Cd in older cultivars in which downwards Cd translocation may happen via clay and 673 oxides illuviation. 674 This study underscores the role of a persistent litter layer, made of fallen leaves and 675 harvest residues, in driving the biocycling and bioaccumulation of Cadmium within plant tissues 676 and cacao products. This process involves the leaching of bioavailable Cd from the litter layer into 677 the topsoil, uptake by shallow active roots, elevated Cd concentrations in leaves (which contribute 678 to the litter layer upon falling), and notably, the influence of cultivar age, amplifying Cd levels with 679 each leaf fall and harvest cycle. Our recommendations for mitigation strategies emphasize an 680 understanding of this observed phenomenon in Latin America and this study, while also 681 considering the social implications of proposed interventions. 682 Finally, our findings are promising for the continuation and expansion of the voluntary 683 crop replacement programs and to farmers in the region who rely on the sale of cacao to profitable 684 foreign markets like the European Union, as Cd levels in both plants and soils in most farms 685 sampled remain low. 686 687 Acknowledgments 688 This work was supported by the USDA National Institute of Food and Agriculture and 689 Hatch Appropriations under Project PEN04569 and accession number 1003147 and Project 690 PEN04879 and Accession #7005892 granted to The Pennsylvania State University College of 691 37 Agricultural Sciences, the Huck Institutes of the Life Sciences, and the Penn State Endowed 692 Program in Molecular Biology of Cacao. The authors acknowledge Fedecacao for the field 693 support, farmers for providing access to their farms, CIAT for the project implementation, field 694 data collection, and providing laboratory analyses and the database, and USDA-FAS and USAID 695 for funding through the Cacao for Peace Program. 696 697 Declaration of interests 698 The authors declare that they have no known competing financial interests or personal 699 relationships that could have appeared to influence the work reported in this paper. 700 701 Declaration of Generative AI and AI-assisted Technologies in the writing process 702 During the preparation of this work, the authors used Open AI’s ChatGPT to check for grammar 703 and spelling mistakes. After using this tool/service, the authors reviewed and edited the content 704 as needed and take full responsibility for the content of the publication. 705 706 References 707 Abbott, P.C., Benjamin, T.J., Burniske, G.R., Croft, M.M., Fenton, M.C., Kelly, C.R., Lundy, 708 M.M., Camayo, F.R., Wilcox Jr., M.D., 2018. An Analysis of the Supply Chain of Cacao in 709 Colombia. https://doi.org/https://doi.org/10.13140/RG.2.2.19395.04645 710 Adriano, D.C., 1986. Trace elements in the terrestrial environment., Trace elements in the 711 terrestrial environment. https://doi.org/10.2134/jeq1987.00472425001600010030x 712 Aikpokpodion, P.E., Lajide, L., Aiyesanmi, A.F., 2012. Metal fractionation in soils collected from 713 selected cocoa plantations in Ogun state, Nigeria. World Appl. Sci. J. 714 https://doi.org/10.5829/idosi.wasj.2012.20.05.2769 715 ALS, 2022. Four Acid Digest & Advanced ICP-MS Technology [WWW Document]. URL 716 38 https://www.alsglobal.com/en/geochemistry/generative-exploration/four-acid-digestions 717 Arévalo-Gardini, E., Arévalo-Hernández, C.O., Baligar, V.C., He, Z.L., 2017. Heavy metal 718 accumulation in leaves and beans of cacao (Theobroma cacao L.) in major cacao growing 719 regions in Peru. Sci. Total Environ. 605–606, 792–800. 720 https://doi.org/10.1016/j.scitotenv.2017.06.122 721 Argüello, D., Chavez, E., Lauryssen, F., Vanderschueren, R., Smolders, E., Montalvo, D., 2019. 722 Soil properties and agronomic factors affecting cadmium concentrations in cacao beans: A 723 nationwide survey in Ecuador. Sci. Total Environ. 649, 120–127. 724 https://doi.org/10.1016/j.scitotenv.2018.08.292 725 Barraza, F., Maurice, L., Uzu, G., Becerra, S., López, F., Ochoa-Herrera, V., Ruales, J., 726 Schreck, E., 2018. Distribution, contents and health risk assessment of metal(loid)s in 727 small-scale farms in the Ecuadorian Amazon: An insight into impacts of oil activities. Sci. 728 Total Environ. 622–623, 106–120. https://doi.org/10.1016/J.SCITOTENV.2017.11.246 729 Barraza, F., Moore, R.E.T., Rehkämper, M., Schreck, E., Lefeuvre, G., Kreissig, K., Coles, B.J., 730 Maurice, L., 2019. Cadmium isotope fractionation in the soil-cacao systems of Ecuador: A 731 pilot field study. RSC Adv. https://doi.org/10.1039/c9ra05516a 732 Barraza, F., Schreck, E., Lévêque, T., Uzu, G., López, F., Ruales, J., Prunier, J., Marquet, A., 733 Maurice, L., 2017. Cadmium bioaccumulation and gastric bioaccessibility in cacao: A field 734 study in areas impacted by oil activities in Ecuador. Environ. Pollut. 229, 950–963. 735 https://doi.org/10.1016/j.envpol.2017.07.080 736 Beaudette, D.E., Roudier, P., O’Geen, A.T., 2013. Algorithms for quantitative pedology: A toolkit 737 for soil scientists. Comput. Geosci. 52, 258–268. 738 https://doi.org/10.1016/j.cageo.2012.10.020 739 Birke, M., Reimann, C., Rauch, U., Ladenberger, A., Demetriades, A., Jähne-Klingberg, F., 740 Oorts, K., Gosar, M., Dinelli, E., Halamić, J., 2017. GEMAS: Cadmium distribution and its 741 sources in agricultural and grazing land soil of Europe — Original data versus clr-742 39 transformed data. J. Geochemical Explor. 173, 13–30. 743 https://doi.org/10.1016/j.gexplo.2016.11.007 744 Blake, G.R., 2015. Bulk Density, in: Methods of Soil Analysis: Part 1 Physical and Mineralogical 745 Properties, Including Statistics of Measurement and Sampling, 9.1. American Society of 746 Agronomy, Inc., pp. 374–390. https://doi.org/10.2134/agronmonogr9.1.c30 747 Blommaert, H., Aucour, A.M., Wiggenhauser, M., Moens, C., Telouk, P., Campillo, S., 748 Beauchêne, J., Landrot, G., Testemale, D., Pin, S., Lewis, C., Umaharan, P., Smolders, E., 749 Sarret, G., 2022. From soil to cacao bean: Unravelling the pathways of cadmium 750 translocation in a high Cd accumulating cultivar of Theobroma cacao L. Front. Plant Sci. 751 13, 1–19. https://doi.org/10.3389/fpls.2022.1055912 752 Bouyoucos, G.J., 1962. Hydrometer Method Improved for Making Particle Size Analyses of 753 Soils 1 . Agron. J. https://doi.org/10.2134/agronj1962.00021962005400050028x 754 Bravo, D., Pardo-Díaz, S., Benavides-Erazo, J., Rengifo-Estrada, G., Braissant, O., Leon-755 Moreno, C., 2018. Cadmium and cadmium-tolerant soil bacteria in cacao crops from 756 northeastern Colombia. J. Appl. Microbiol. 124, 1175–1194. 757 https://doi.org/10.1111/ijlh.12426 758 Brown, S.L., Chaney, R.L., Angle, J.S., Baker, A.J.M., 1995. Zinc and Cadmium Uptake by 759 Hyperaccumulator Thlaspi caerulescens and Metal Tolerant Silene vulgaris Grown on 760 Sludge-Amended Soils. Environ. Sci. Technol. 29, 1581–1585. 761 https://doi.org/10.1021/es00006a022 762 Brus, D.J., De Gruijter, J.J., Römkens, P.F.A.M., 2005. Probabilistic quality standards for heavy 763 metals in soil derived from quality standards in crops. Geoderma 128, 301–311. 764 https://doi.org/10.1016/j.geoderma.2005.04.014 765 Capparelli, M.V., Moulatlet, G.M., Abessa, D.M. de S., Lucas-Solis, O., Rosero, B., Galarza, E., 766 Tuba, D., Carpintero, N., Ochoa-Herrera, V., Cipriani-Avila, I., 2020. An integrative 767 approach to identify the impacts of multiple metal contamination sources on the Eastern 768 40 Andean foothills of the Ecuadorian Amazonia. Sci. Total Environ. 709, 136088. 769 https://doi.org/10.1016/J.SCITOTENV.2019.136088 770 Caridad-Cancela, R., Paz-González, A., De Abreu, C.A., 2005. Total and extractable nickel and 771 cadmium contents in natural soils. Commun. Soil Sci. Plant Anal. 36, 241–252. 772 https://doi.org/10.1081/CSS-200043057 773 Casteblanco, J.A., 2018. Heavy metals remediation with potential application in cocoa 774 cultivation. Granja 27, 21–35. https://doi.org/10.17163/lgr.n27.2018.02 775 Castro, A. V., de Almeida, A.-A.F., Pirovani, C.P., Reis, G.S.M., Almeida, N.M., Mangabeira, 776 P.A.O., 2015. Morphological, biochemical, molecular and ultrastructural changes induced 777 by Cd toxicity in seedlings of Theobroma cacao L. Ecotoxicol. Environ. Saf. 115, 174–186. 778 https://doi.org/10.1016/J.ECOENV.2015.02.003 779 Chaney, R.L., 2010. Cadmium and Zinc. Trace Elem. Soils 409–439. 780 https://doi.org/10.1002/9781444319477.ch17 781 Chavez, E., He, Z.L., Stoffella, P.J., Mylavarapu, R.S., Li, Y.C., Baligar, V.C., 2016. Chemical 782 speciation of cadmium: An approach to evaluate plant-available cadmium in Ecuadorian 783 soils under cacao production. Chemosphere 150, 57–62. 784 https://doi.org/10.1016/j.chemosphere.2016.02.013 785 Chavez, E., He, Z.L., Stoffella, P.J., Mylavarapu, R.S., Li, Y.C., Moyano, B., Baligar, V.C., 2015. 786 Concentration of cadmium in cacao beans and its relationship with soil cadmium in 787 southern Ecuador. Sci. Total Environ. 533, 205–214. 788 https://doi.org/10.1016/j.scitotenv.2015.06.106 789 Chen, S., Sun, L., Sun, T., Chao, L., Guo, G., 2007. Interaction between cadmium, lead and 790 potassium fertilizer (K2 SO4) in a soil-plant system. Environ. Geochem. Health 29, 435–791 446. https://doi.org/10.1007/s10653-007-9088-y 792 Clemens, S., Aarts, M.G.M., Thomine, S., Verbruggen, N., 2013. Plant science: The key to 793 preventing slow cadmium poisoning. Trends Plant Sci. 18, 92–99. 794 41 https://doi.org/10.1016/j.tplants.2012.08.003 795 Codex Alimentarius Commission, 2018. Report of the 12th Session of the Codex Committee on 796 Contaminants in Foods. https://www.fao.org/fao-who-797 codexalimentarius/meetings/detail/en/?meeting=CCCF&session=12 (accessed 1 October 798 2023). 799 Colmenares, F., Mesa, A.M., Roncancio, J., Arciniegas, E., Pedraza, P., Cardona, A., Romero, 800 A.J., Silva, C.A., Alvarado, S., Romero, O.A., Vargas, A.F., 2007. Geología de las planchas 801 11, 12, 13, 14, 18, 19, 20, 21, 25, 26,27, 33, 34 y 40. Proyecto: “Evolución geohistórica de 802 la Sierra Nevada de Santa Marta,” CONTRATO DE PRESTACIÓN DE SERVICIOS No.: 803 PS 025-06. Bogotá, D.C. 804 Colombian Presidency Press Office, 2018. Sustitución de cultivos: llegó la hora del cacao. 805 http://www.posconflicto.gov.co/sala-prensa/ noticias/2018/Paginas/20180418-sustitucion-806 de-cultivos-llego-la-hora-delcacao. (accessed 1 October 2023) 807 Dubois, J., Okopnik, F., Benitez, N., Vedy, J., 1997. Origin and spatial variability of cadmium in 808 some soils of the Swiss Jura. Enregist. Sci. 1–8. 809 Dumoulin, J.A., Slack, J.F., Whalen, M.T., Harris, A.G., 2011. Depositional setting and 810 geochemistry of phosphorites and metalliferous black shales in the carboniferous-permian 811 Lisburne group, northern Alaska. US Geol. Surv. Prof. Pap. 1–30. 812 Engbersen, N., Gramlich, A., Lopez, M., Schwarz, G., Hattendorf, B., Gutierrez, O., Schulin, R., 813 2019. Cadmium accumulation and allocation in different cacao cultivars. Sci. Total Environ. 814 https://doi.org/10.1016/j.scitotenv.2019.05.001 815 ESRI, 2011. ArcMap 10.1. Environmental Systems Resource Institute, Redlands, California. 816 Gibb, J.P., Cartwright, K., 1982. Retention of zinc, cadmium, copper, and lead by geologic 817 materials., Cooperative Groundwater Report - Illinois State Geological Survey. 818 Gil, J.P., López-Zuleta, S., Quiroga-Mateus, R.Y., Benavides-Erazo, J., Chaali, N., Bravo, D., 819 2022. Cadmium distribution in soils, soil litter and cacao beans: a case study from 820 42 Colombia. Int. J. Environ. Sci. Technol. 19, 2455–2476. https://doi.org/10.1007/s13762-821 021-03299-x 822 Gómez, J., Montes, N.E., Nivia, Á., Diederix, H., Compilers, 2015. Geological Map of Colombia 823 2015. Scale 1:1 000 000. 824 Gómez Tapias, J., Nivia, Á., Montes R, N., Almanza M, M., Alcárcel G, F., Madrid M, C., 2015. 825 Notas explicativas: Mapa Geológico de Colombia, in: Gómez, J. & Almanza, M.F. 826 (Editores) (Ed.), Compilando La Geología de Colombia: Una Visión a 2015. Servicio 827 Geológico Colombiano, Bogotá, pp. 9–33. 828 Gramlich, A., Tandy, S., Andres, C., Chincheros Paniagua, J., Armengot, L., Schneider, M., 829 Schulin, R., 2017. Cadmium uptake by cocoa trees in agroforestry and monoculture 830 systems under conventional and organic management. Sci. Total Environ. 580, 677–686. 831 https://doi.org/10.1016/j.scitotenv.2016.12.014 832 Gramlich, A., Tandy, S., Gauggel, C., López, M., Perla, D., Gonzalez, V., Schulin, R., 2018. Soil 833 cadmium uptake by cocoa in Honduras. Sci. Total Environ. 612, 370–378. 834 https://doi.org/10.1016/J.SCITOTENV.2017.08.145 835 Grant, C.A., Buckley, W.T., Bailey, L.D., Selles, F., 1998. Cadmium accumulation in crops. Can. 836 J. Plant Sci. 78, 1–17. https://doi.org/10.4141/P96-100 837 Hamid, Y., Tang, L., Sohail, M.I., Cao, X., Hussain, B., Aziz, M.Z., Usman, M., He, Z. li, Yang, 838 X., 2019. An explanation of soil amendments to reduce cadmium phytoavailability and 839 transfer to food chain. Sci. Total Environ. 660, 80–96. 840 https://doi.org/10.1016/j.scitotenv.2018.12.419 841 Han, F., Shan, X., Zhang, S., Wen, B., Owens, G., 2006. Enhanced cadmium accumulation in 842 maize roots - The impact of organic acids. Plant Soil 289, 355–368. 843 https://doi.org/10.1007/s11104-006-9145-9 844 Hanafi, M.M., Jomol Maria, G., 1998. Cadmium and zinc in acid tropical soils: III. Response of 845 cocoa seedlings in a greenhouse experiment, in: Communications in Soil Science and 846 43 Plant Analysis. https://doi.org/10.1080/00103629809370084 847 He, S., He, Z., Yang, X., Stoffella, P.J., Baligar, V.C., 2015. Soil Biogeochemistry, Plant 848 Physiology, and Phytoremediation of Cadmium-Contaminated Soils. Adv. Agron. 134, 135–849 225. https://doi.org/10.1016/bs.agron.2015.06.005 850 He, Z.L., Hai Ping Xu, Zhu, Y.M., Yang, X.E., Chen, G.C., 2004. Adsorption-desorption 851 characteristics of lead in variable charge soils. J. Environ. Sci. Heal. - Part A 852 Toxic/Hazardous Subst. Environ. Eng. 39, 1949–1967. https://doi.org/10.1081/ESE-853 120039367 854 Hooda, P.S., Alloway, B.J., 1998. Cadmium and lead sorption behaviour of selected English and 855 Indian soils. Geoderma 84, 121–134. https://doi.org/10.1016/S0016-7061(97)00124-9 856 IARC, 2019. Agents classified by the IARC monographs. Vols. 1 – 123, IARC Monographs. 857 Idárraga-García, J., Romero, J., 2010. Neotectonic study of the Santa Marta Fault System, 858 Western foothills of the Sierra Nevada de Santa Marta, Colombia. J. South Am. Earth Sci. 859 29, 849–860. https://doi.org/10.1016/j.jsames.2009.11.004 860 Imseng, M., Wiggenhauser, M., Keller, A., Müller, M., Rehkämper, M., Murphy, K., Kreissig, K., 861 Frossard, E., Wilcke, W., Bigalke, M., 2019. Towards an understanding of the Cd isotope 862 fractionation during transfer from the soil to the cereal grain. Environ. Pollut. 244, 834–844. 863 https://doi.org/10.1016/J.ENVPOL.2018.09.149 864 Kabata-Pendias, A., Mukherjee, A.B., 2007. Part II Biogeochemistry of Trace Elements, 865 Springer. 866 Kabata-Pendias, A., Pendias, H., 2011. Trace Elements In Soils and Plants, CRC Press, Inc. 867 Boca Raton. 868 Karger, D.N., Conrad, O., Böhner, J., Kawohl, T., Kreft, H., Soria-Auza, R.W., Zimmermann, 869 N.E., Linder, H.P., Kessler, M., 2017. Climatologies at high resolution for the earth’s land 870 surface areas. Sci. Data 4, 1–20. https://doi.org/10.1038/sdata.2017.122 871 Kirkham, M.B., 2006. Cadmium in plants on polluted soils: Effects of soil factors, 872 44 hyperaccumulation, and amendments. Geoderma 137, 19–32. 873 https://doi.org/10.1016/j.geoderma.2006.08.024 874 Kubier, A., Wilkin, R.T., Pichler, T., 2019. Cadmium in soils and groundwater: A review. Appl. 875 Geochemistry 108, 104388. https://doi.org/10.1016/J.APGEOCHEM.2019.104388 876 Lewis, C., Lennon, A.M., Eudoxie, G., Umaharan, P., 2018. Genetic variation in bioaccumulation 877 and partitioning of cadmium in Theobroma cacao L. Sci. Total Environ. 640–641, 696–703. 878 https://doi.org/10.1016/J.SCITOTENV.2018.05.365 879 Liu, L., Li, W., Song, W., Guo, M., 2018. Remediation techniques for heavy metal-contaminated 880 soils: Principles and applicability. Sci. Total Environ. 633, 206–219. 881 https://doi.org/10.1016/j.scitotenv.2018.03.161 882 Mar, S.S., Okazaki, M., 2012. Investigation of Cd contents in several phosphate rocks used for 883 the production of fertilizer. Microchem. J. 104, 17–21. 884 https://doi.org/10.1016/j.microc.2012.03.020 885 McLaughlin, M.J., Singh, B.R., 1999. Cadmium in Soils and Plants, in: Cadmium in Soils and 886 Plants. Springer Netherlands, Dordrecht, pp. 1–9. https://doi.org/10.1007/978-94-011-887 4473-5_1 888 McLaughlin, M.J., Smolders, E., Zhao, F.J., Grant, C., Montalvo, D., 2021. Managing cadmium 889 in agricultural systems, 1st ed, Advances in Agronomy. Elsevier Inc. 890 https://doi.org/10.1016/bs.agron.2020.10.004 891 Meter, A., Atkinson, R.J. and Laliberte, B., 2019. Cadmium in cacao from Latin America and the 892 Caribbean. A review of research and potential mitigation solutions. Bioversity International, 893 Rome. https://scioteca.caf.com/handle/123456789/1506 (accessed 1 October 2023) 894 Minitab LLC, 2019. Minitab Statistical Software, Version 17. State College, PA. Available at: 895 www.minitab.com (accessed 6 September 2019). 896 Mrozek, S.A., Buck, B.J., Drohan, P.J., Brock, A.L., 2006. Decorative landscaping rock as a 897 source for heavy metal contamination, Las Vegas, Nevada. Soil Sediment Contam. 15, 898 45 471–480. https://doi.org/10.1080/15320380600847716 899 Nathan, Y., Benalioulhaj, N., Prévôt, L., Lucas, J., 1996. The geochemistry of cadmium in the 900 phosphate-rich and organic-rich sediments of the Oulad-abdoun and Timahdit basins 901 (Morocco). J. African Earth Sci. 22, 17–27. https://doi.org/10.1016/0899-5362(95)00124-7 902 Nathan, Y., Soudry, D., Levy, Y., Shitrit, D., Dorfman, E., 1997. Geochemistry of cadmium in the 903 negev phosphorites. Chem. Geol. 142, 87–107. https://doi.org/10.1016/S0009-904 2541(97)00078-8 905 Nouet, C., Charlier, J.B., Carnol, M., Bosman, B., Farnir, F., Motte, P., Hanikenne, M., 2015. 906 Functional analysis of the three HMA4 copies of the metal hyperaccumulator Arabidopsis 907 halleri. J. Exp. Bot. 66, 5783–5795. https://doi.org/10.1093/jxb/erv280 908 Nziguheba, G., Smolders, E., 2008. Inputs of trace elements in agricultural soils via phosphate 909 fertilizers in European countries. Sci. Total Environ. 390, 53–57. 910 https://doi.org/10.1016/j.scitotenv.2007.09.031 911 Papoyan, A., Piñeros, M., Kochian, L. V., 2007. Plant Cd2+ and Zn2+ status effects on root and 912 shoot heavy metal accumulation in Thlaspi caerulescens. New Phytol. 175, 51–58. 913 https://doi.org/10.1111/j.1469-8137.2007.02073.x 914 Pence, N.S., Larsen, P.B., Ebbs, S.D., Letham, D.L.D., Lasat, M.M., Garvin, D.F., Eide, D., 915 Kochian, L. V., 2000. The molecular physiology of heavy metal transport in the Zn/Cd 916 hyperaccumulator Thlaspi caerulescens. Proc. Natl. Acad. Sci. U. S. A. 97, 4956–4960. 917 https://doi.org/10.1073/pnas.97.9.4956 918 Pérez Moncada, A.U., Gómez Ramírez, M., Ordoñez Serralde, D.P., Peñaranda Rolón, M.A., 919 Wilches Ortiz, A.W., Ramírez, L., Rengifo Estrada, A.G., 2019. Arbuscular mycorrhizal 920 fungi (AMF) as a strategy to reduce the absorption of cadmium in cocoa (Theobroma 921 cacao) plants. Terra Latinoam. https://doi.org/10.28940/tl.v37i2.479 922 Philippe, A., Ibanez, F., 2018. pastecs: Package for Analysis of Space-Time Ecological Series. 923 CRAN. R-Package. https://CRAN.R-project.org/package=pastecs. 924 46 Pohlert, T., 2021. PMCMRplus: Calculate Pairwise Multiple Comparisons of Mean Rank Sums 925 Extended. CRAN. R-Package . https//CRAN.R-project.org/package=PMCMRplus. 926 Quandt, D., Trumbull, R.B., Altenberger, U., Cardona, A., Romer, R.L., Bayona, G., Ducea, M., 927 Valencia, V., Vásquez, M., Cortes, E., Guzman, G., 2018. The geochemistry and 928 geochronology of Early Jurassic igneous rocks from the Sierra Nevada de Santa Marta, 929 NW Colombia, and tectono-magmatic implications. J. South Am. Earth Sci. 930 https://doi.org/10.1016/j.jsames.2018.06.019 931 Quezada-Hinojosa, R.P., Föllmi, K.B., Verrecchia, E., Adatte, T., Matera, V., 2015. Speciation 932 and multivariable analyses of geogenic cadmium in soils at Le Gurnigel, Swiss Jura 933 Mountains. Catena 125, 10–32. https://doi.org/10.1016/j.catena.2014.10.003 934 Quezada-Hinojosa, R.P., Matera, V., Adatte, T., Rambeau, C., Föllmi, K.B., 2009. Cadmium 935 distribution in soils covering Jurassic oolitic limestone with high Cd contents in the Swiss 936 Jura. Geoderma 150, 287–301. https://doi.org/10.1016/j.geoderma.2009.02.013 937 R Core Team, 2022. R: A Language and Environment for Statistical Computing. R Foundation 938 for Statistical Computing, Vienna, Austria. Available at: https://www.R-project.org 939 (accessed 19 January 2022). 940 Ramtahal, G., Chang Yen, I., Hamid, A., Bekele, I., Bekele, F., Maharaj, K., Harrynanan, L., 941 2018. The Effect of Liming on the Availability of Cadmium in Soils and Its Uptake in Cacao 942 (Theobroma cacao L.) In Trinidad & Tobago. Commun. Soil Sci. Plant Anal. 49, 2456–943 2464. https://doi.org/10.1080/00103624.2018.1510955 944 Ramtahal, G., Umaharan, P., Hanuman, A., Davis, C., Ali, L., 2019. The effectiveness of soil 945 amendments, biochar and lime, in mitigating cadmium bioaccumulation in Theobroma 946 cacao L. Sci. Total Environ. 693, 133563. 947 https://doi.org/10.1016/J.SCITOTENV.2019.07.369 948 Ramtahal, G., Yen, I.C., Bekele, I., Bekele, F., Wilson, L., Maharaj, K., Harrynanan, L., 2016. 949 Relationships between Cadmium in Tissues of Cacao Trees and Soils in Plantations of 950 47 Trinidad and Tobago. Food Nutr. Sci. 07, 37–43. https://doi.org/10.4236/fns.2016.71005 951 Rodrigues, S.K., Abessa, D.M.S., Rodrigues, A.P. de C., Soares-Gomes, A., Freitas, C.B., 952 Santelli, R.E., Freire, A.S., Machado, W., 2017. Sediment quality in a metal-contaminated 953 tropical bay assessed with a multiple lines of evidence approach. Environ. Pollut. 228, 954 265–276. https://doi.org/10.1016/J.ENVPOL.2017.05.045 955 Rodríguez Albarrcín, H.S., Darghan Contreras, A.E., Henao, M.C., 2019. Spatial regression 956 modeling of soils with high cadmium content in a cocoa producing area of Central 957 Colombia. Geoderma Reg. 16, e00214. https://doi.org/10.1016/J.GEODRS.2019.E00214 958 Romero-Estévez, D., Yánez-Jácome, G.S., Simbaña-Farinango, K., Navarrete, H., 2019. 959 Content and the relationship between cadmium, nickel, and lead concentrations in 960 Ecuadorian cocoa beans from nine provinces. Food Control 106, 106750. 961 https://doi.org/10.1016/J.FOODCONT.2019.106750 962 Salmanzadeh, M., Schipper, L.A., Balks, M.R., Hartland, A., Mudge, P.L., Littler, R., 2017. The 963 effect of irrigation on cadmium, uranium, and phosphorus contents in agricultural soils. 964 Agric. Ecosyst. Environ. 247, 84–90. https://doi.org/10.1016/j.agee.2017.06.028 965 Satarug, S., Vesey, D.A., Gobe, G.C., 2017. Current health risk assessment practice for dietary 966 cadmium: Data from different countries. Food Chem. Toxicol. 106, 430–445. 967 https://doi.org/10.1016/j.fct.2017.06.013 968 Scaccabarozzi, D., Castillo, L., Aromatisi, A., Milne, L., Castillo, A.B., Muñoz-Rojas, M., 2020. 969 Soil, site, and management factors affecting cadmium concentrations in cacao-growing 970 soils. Agronomy 10, 1–15. https://doi.org/10.3390/agronomy10060806 971 Schipper, L.A., Sparling, G.P., Fisk, L.M., Dodd, M.B., Power, I.L., Littler, R.A., 2011. Rates of 972 accumulation of cadmium and uranium in a New Zealand hill farm soil as a result of long-973 term use of phosphate fertilizer. Agric. Ecosyst. Environ. 144, 95–101. 974 https://doi.org/10.1016/j.agee.2011.08.002 975 Schoeneberger, P.J., Wysocki, D.A., Benham, E.C., Soil Survey Staff., 2012. Field book for 976 48 describing and sampling soils, version 3.0. National Resources Conservation Service, 977 National Soil Survey Center, Lincoln, NE. 978 Servicio Geológico Colombiano, 2023. Estado de la Cartografía Geológica. Escala 1:100000 979 [WWW Document]. URL 980 https://srvags.sgc.gov.co/Flexviewer/Estado_Cartografia_Geologica/ 981 Sharma, R.P., Street, J.C., Verma, M.P., Shupe, J.L., 1979. Cadmium uptake from feed and its 982 distribution to food products of livestock. Environ. Health Perspect. Vol. 28, 59–66. 983 https://doi.org/10.2307/3428905 984 Soil Survey Staff, 2022. Keys to Soil Taxonomy, 13th ed. USDA-Natural Resources 985 Conservation Service. 986 Tabelin, C.B., Igarashi, T., Villacorte-Tabelin, M., Park, I., Opiso, E.M., Ito, M., Hiroyoshi, N., 987 2018. Arsenic, selenium, boron, lead, cadmium, copper, and zinc in naturally contaminated 988 rocks: A review of their sources, modes of enrichment, mechanisms of release, and 989 mitigation strategies. Sci. Total Environ. 645, 1522–1553. 990 https://doi.org/10.1016/j.scitotenv.2018.07.103 991 Tang, L., Qiu, R., Tang, Y., Wang, S., 2014. Cadmium-zinc exchange and their binary 992 relationship in the structure of Zn-related proteins: A mini review. Metallomics 6, 1313–993 1323. https://doi.org/10.1039/c4mt00080c 994 Taylor, M.D., 1997. Accumulation of cadmium derived from fertilisers in New Zealand soils. Sci. 995 Total Environ. 208, 123–126. https://doi.org/10.1016/S0048-9697(97)00273-8 996 Tchounwou, P.B., Yedjou, C.G., Patlolla, A.K., Sutton, D.J., 2012. Heavy Metal Toxicity and the 997 Environment. pp. 133–164. https://doi.org/10.1007/978-3-7643-8340-4_6 998 The European Commision, 2014. Comission Regulation (EU) No 488/ 2014 of 12 May 2014 999 Amending Regulation (EC) No 1881/2006 as regards maximum levels of cadmium in 1000 foodstuff. Official Journal of the European Union. L138, 75–79. 1001 http://data.europa.eu/eli/reg/2014/488/oj. 1002 49 Thornton, I., 1981. Geochemical Aspects of the Distribution and Forms of Heavy Metals in Soils. 1003 Eff. Heavy Met. Pollut. Plants 1–33. https://doi.org/10.1007/978-94-009-8099-0_1 1004 Traina, S.J., 1999. The Environmental Chemistry of Cadmium: An Overview, in: Cadmium in 1005 Soils and Plants. Springer Netherlands, Dordrecht, pp. 11–36. https://doi.org/10.1007/978-1006 94-011-4473-5 1007 Tschanz, C.M., Marvin, R.F., Cruz B., J., Mehnert, H.H., Cebula, G.T., 1974. Geological Society 1008 of America Bulletin Geologic Evolution of the Sierra Nevada de Santa Marta , Northeastern 1009 Colombia. Geol. Soc. Am. Bull. 85, 273–284. https://doi.org/10.1130/0016-1010 7606(1974)85<273 1011 United States Environmental Protection Agency (U.S. EPA), 2002. Supplemental Guidance for 1012 Developing Soil Screening Levels for Superfund Sites 1–187. 1013 Vanderschueren, R., Argüello, D., Blommaert, H., Montalvo, D., Barraza, F., Maurice, L., 1014 Schreck, E., Schulin, R., Lewis, C., Vazquez, J.L., Umaharan, P., Chavez, E., Sarret, G., 1015 Smolders, E., 2021. Mitigating the level of cadmium in cacao products: Reviewing the 1016 transfer of cadmium from soil to chocolate bar. Sci. Total Environ. 781, 146779. 1017 https://doi.org/10.1016/j.scitotenv.2021.146779 1018 Vanderschueren, R., De Mesmaeker, V., Mounicou, S., Isaure, M.-P., Doelsch, E., Montalvo, D., 1019 Delcour, J.A., Chavez, E., Smolders, E., 2020. The impact of fermentation on the 1020 distribution of cadmium in cacao beans. Food Res. Int. 127, 108743. 1021 https://doi.org/10.1016/J.FOODRES.2019.108743 1022 Vanderschueren, R., Montalvo, D., De Ketelaere, B., Delcour, J.A., Smolders, E., 2019. The 1023 elemental composition of chocolates is related to cacao content and origin: A multi-element 1024 fingerprinting analysis of single origin chocolates. J. Food Compos. Anal. 83, 103277. 1025 https://doi.org/10.1016/J.JFCA.2019.103277 1026 Vanderschueren, R., Wantiez, L., Blommaert, H., Flores, J., Chavez, E., Smolders, E., 2023. 1027 Revealing the pathways of cadmium uptake and translocation in cacao trees (Theobroma 1028 50 cacao L.): A 108Cd pulse-chase experiment. Sci. Total Environ. 869, 161816. 1029 https://doi.org/10.1016/j.scitotenv.2023.161816 1030 Walkley, A., Black, I.A., 1934. An examination of the Degtjareff method for determining soil 1031 organic matter, and a proposed modification of the Chromic Acid Titration method. Soil Sci. 1032 37, 29–38. https://doi.org/10.1097/00010694-193401000-00003 1033 Wei, T., Simko, V., 2017. corrplot: Visualization of a Correlation Matrix. Available 1034 Https://github.com/taiyun/corrplot. 1035 WHO, 2010. Exposure to cadmium: A major public health concern. In: Preventing Disease 1036 Through Healthy Environments. http://www.who.int/ipcs/features/cadmium.pdf (accessed 1 1037 October 2023). 1038 Zug, K.L.M., Huamaní Yupanqui, H.A., Meyberg, F., Cierjacks, J.S., Cierjacks, A., 2019. 1039 Cadmium Accumulation in Peruvian Cacao (Theobroma cacao L.) and Opportunities for 1040 Mitigation. Water. Air. Soil Pollut. 230. https://doi.org/10.1007/s11270-019-4109-x 1041 1042