We chose Soroti and Ngora districts in Teso subregion, Eastern Uganda ( Figure 1 ). The two districts share common frontiers, and are traversed by the Mbale-Soroti highway, an arterial stretch of the PAH. They were chosen because of the popularity of commercial orchards along the highway and in the villages [31] . This region belongs to the Kyoga plains agricultural zone (Teso subregion) where citrus production is highest, with over 700 hectares of land with citrus. They had a production volume of over 200,000 tonnes of citrus and approximately 67 billion Uganda shillings (US$ 28.3 million) to smallholder income in the region [32] [33] . Specifically, the Soroti Fruits Factory was commissioned in 2019 in Soroti to handle the processing of sweet oranges. Ngora district is located at coordinates 1˚29'59.99"N 33˚47'59.99"E while Soroti district lies between 1˚35'59.99"N and 33˚38'59.99"E. The vegetation in the two districts is savannah grassland with light, well-drained and sandy loam soils that favor fruit growing [34] . The rainfall regime in these two districts is bi-modal with peaks in April-May and July-August [35] . The mean annual rainfall ranges between 800 - 1000 mm while the mean annual temperature is 24˚C. The main economic activity in Soroti and Ngora districts is agriculture where small scale livestock rearing alongside cropping is practiced.

In total, 36 sweet oranges (Citrus sinensis (L.) Osbeck) and 36 mango (Mangifera indica L.) fruits were purposively sampled from orchards in Aukot and Mukura

Sub-Counties in Soroti and Ngora districts, respectively. Sampling was done between 7^th and 14^th August 2023 from orchards where the fruit trees were grown at 10 m, 100 m, and 300 m from the highway, since levels of heavy metals in soil due to road transport operations decrease with increase in distance from the road [36] . Fruit samples were washed using deionized water to remove soil particles and/or dust on their surfaces [37] .

A total of 72 soil samples were also obtained from the corresponding orchards. These distances in the increasing order were chosen basing on the fact that the concentrations of HMs in soils as a result of road transport operations decrease with the increase in distance from the road. For all the soil samples, surface litter was removed at the sampling spot. A stainless-steel spade was used to make a ‘V’ shaped cut into the soil to a depth of 25 cm. A slice of soil was obtained from the top to bottom and placed in a clean plastic bucket. Four slices of soil were obtained from each cut. The above procedure was repeated for several spots from the edges of the area under the tree canopy to ensure homogeneity. The samples were thoroughly mixed to make the composite sample and foreign materials such as gravel, pebbles, roots, and stones were removed. The composite sample was then reduced to about 0.5 kg by quartering, packed in a clean paper bag and labeled appropriately. From the field, the composite samples were taken to the laboratory where they were air dried for 1 week [37] .

Control samples of oranges (n = 3) and mango fruits (n = 3) and their corresponding soils were obtained from orchards 5 km away from the Highway i.e., in Agogomit and Awoja villages in Mukura and Aukot Sub-Counties. All the fruit and soil sample were taken on the same day.

Mango fruits were peeled using a clean stainless-steel knife to obtain the pulp and the seeds were discarded. Each sample (300 g) was oven-dried for 24 hours at 105˚C [38] . The dried samples were then ground into a powder using a porcelain pestle and mortar. The powdered samples (0.2 g) were placed in a digestion tube to which aqua regia (6 mL of 37% hydrochloric acid and 6 mL of 65% nitric acid) were added. For orange samples, fruits were peeled and the peel was discarded. From the remaining portions, juice (150 mL) was squeezed into 250 mL beakers, filtered and then kept in a refrigerator. The sample juice (4 mL) was placed into a digestion tube and then 12 mL of 65% nitric acid was added. All the mixtures were heated at 105˚C on a hot plate for 30 minutes [37] . The resultant solutions were filtered into 50 mL volumetric flasks through Whatman filter no. 42 paper and the volume topped the mark with deionized water. Soil samples were ground and then sieved through a 0.15-mm mesh. Thereafter, samples (0.2 g) were transferred into Teflon tubes, and then aqua regia was added. The tubes were transferred onto a hot plate at 105˚C for 30 minutes for the samples to digest [37] . The digestates were then filtered using Whatmann Grade 1 filter papers into a 50 mL volumetric flask. These were topped with deionized water to the mark.

The concentration of Cu, Zn, Pb, and Cd in soil and fruit samples was established using an inductively coupled plasma-optical emission spectrometer (ICP-OES; Optima 7000 DV, Perkin Elmer Inc., Waltham, USA) used in our previous studies [15] [39] . The instrument detection limits were 0.003, 0.005, 0.005, and 0.001 ppm for Cu, Zn, Pb, and Cd respectively as described in US EPA 200.7 [40] . The specific wavelengths used for the HMs were 224.700 nm, 206.200 nm, 220.353 nm and 214.440 nm, respectively. Quantitation was performed using calibration curves constructed from working standard solutions, which were prepared from 1000 mg/L stock solutions of the individual HMs’ nitrate or chloride salt [15] .

To ensure confidence in the data obtained, all reagents and chemicals used were all of analytical grade. Other quality control measures involved analyzing blanks and spiked samples for every 10 sample runs, which gave satisfactory recoveries (98.4% - 101.2%). Analytical method precision, expressed as relative standard deviations of runs, was below 5% (3.2% - 4.7%). To avoid batch-specific errors, each sample was analyzed in triplicate under the same standard conditions.

The health risks associated with consuming the fruits, as recommended by the US EPA [41] , are categorized into cancer and non-cancer risks for both children and adults. Assessing these risks requires determining the estimated daily intake (EDI, mg/kg/day) of HMs in the edible fruit parts. The EDI can be calculated using Equation 1 [42] [43] . The variables in the equation and their corresponding values are provided in Table 1 .

$\text{EDI}=\frac{C_{HM}\times \text{IngR}\times \text{EXF}\times \text{EXD}}{W_{ab}\times T_{aet}}$ (1)

The non-cancer health risks were assessed using the hazard quotient (HQ) method (Equation (2)). Given that HMs exhibit additive toxicity, the hazard index (HI) was determined by summing the HQ values of the individual HMs (Equation (3)) [44] .

$\text{HQ}=\frac{\text{EDI}}{\text{RF}}$ (2)

$\text{HI}={\displaystyle {\sum }_i^n\text{HQ}}$ (3)

In Equation (2), RF is the oral reference dose of the specific heavy metal, with values of 0.04, 0.006, 0.14 and 0.01 mg∙kg⁻¹ for Cu, Zn, Pb and Cd, respectively [44] - [46] . Ultimately, cancer risk (CR) was evaluated based on the incremental lifetime cancer risk for ingestion of carcinogenic HMs (Pb and Cd) in the edible fruit parts (Equation (4)) [47] .

CR = EDI × CSF (4)

where CSF is the ingestion cancer slope factor = 8.5 × 10⁻³ and 5.01 × 10⁻¹ mg/kg/day for Pb and Cd [47] .

Several pollution indices are available for the evaluation of HMs pollution of soils. In this study, we adopted two indices namely the contamination factor (Equation (5)) [48] , and geoaccumulation index, an empirical metric that compares the concentration of HMs to their normalized geochemical backgrounds (Equation (6)) [49] .

Contamination factor (CF) = $\frac{C_{HM}}{C_{PIC}}$ (5)

Geoaccumulation index (I_geo) = ${\log }_2\frac{C_{HM}}{1.5C_{PIC}}$ (6)

In the foregoing equations, C_PIC represents the preindustrial concentration of the respective element in soils. These were taken to be the metal concentrations from the control sites since they were lower than those for Cu (12.5 mg/kg) and Zn (70.0 mg/kg) for the average shale [50] and 19.88 mg/kg and 0.13 mg/kg for Pb and Cd in soils from an uncontaminated site in Mbarara (Uganda) [15] . The constant of 1.5 serves as a background matrix correction factor to account for anticipated lithological variations [51] [52] .

The cumulative pollution load in the soils was evaluated based on the pollution load index method (Equation (7)) [44] .

Pollution load index (PLI) = (CF_Cu × CF_Zn × CF_Pb × CF_Cd)^1/4 (7)

The potential ecological risk index (RI) was employed to assess any harmful effects that could arise due to HMs pollution of the soils (Equations (8) and (9)) [48] .

Ecological risk coefficient (E_i) = T_i × CF (8)

$\text{RI}={\displaystyle {\sum }_{i=1}^{n=4}{E}_i}$ (9)

The toxic response factor (T_i) for a heavy metal indicates its toxic and ecological sensitivity levels, and its values for Cu = Pb = 5, Zn = 1 and Cd = 30 [48] . The criteria for classification of the different ecological and pollution indices are given in Table 2 .

Continued

Analytical data were inspected, and their normality was determined using Shapiro-Wilk Test. Subsequently, One-Way Analysis of Variance was used to determine statistically significant differences in the mean concentrations of HMs in fruits and soils. Tukey's post hoc test was applied for multiple comparisons where statistical differences were detected at p < 0.05. The mean concentrations of the HMs were compared with the acceptable limits given by the World Health Organization and/or Food and Agricultural Organization of the United Nations (WHO/FAO) [53] [54] . Pearson’s correlation analysis was performed to assess the relationships between HMs concentrations in fruits and soil samples. The analyses as well as data visualizations proceeded in Origin Pro 2025a (Origin Lab Corporation, Northampton, MA).

The concentration of Cu in mangoes ranged from 1.49 ± 0.26 mg/kg to 1.81 ± 0.37 mg/kg. On the other hand, orange juice samples had Cu between 0.05 ± 0.01 mg/kg and 0.10 ± 0.02 mg/kg ( Table 3 ). Similar to the observation with Cu, the concentrations of Zn varied between 2.41 ± 2.18 mg/kg to 8.11 ± 1.2 mg/kg in the fruit samples. In mangoes, the concentration of Zn ranged from 1.81 ± 0.37 mg/kg to 5.76 ± 1.30 mg/kg. In the orange juice samples, Zn was from 0.04 ± 0.03 mg/kg to 0.100 ± 0.010 mg/kg. Cadmium was below the limit of detection (<0.001 mg/kg) in all the fruit samples, while the concentration of Pb varied from 0.01 ± 0.00 mg/kg to 0.06 ± 0.03 mg/kg. In all the fruit pulps, mean total metal concentrations followed the chemical sequence Zn > Cu > Pb > Cd.

The concentration of HMs in the soil samples ranged from <0.001 - 0.44 ± 0.21 mg/kg for Cd to 20.15 ± 1.52 - 40.03 ± 7.30 mg/kg for Zn ( Figure 2 ). Soil samples from orchards where oranges are cultivated had comparatively higher concentrations of all the HMs. The concentration of Zn was highest at 10 m for both mango

Note: Values are presented as mean ± standard deviation of replicates (n = 3). Means carrying different letters in a column are statistically different as per Tukey test (p < 0.05).

(40.03 ± 7.30 mg/kg) and orange (35.28 ± 4.07 mg/kg) orchards, decreasing with distance from the road (p < 0.05). Background values at 5 km showed lower Zn levels (21.11 ± 3.34 mg/kg for mango and 19.00 ± 1.30 mg/kg for orange orchards). For Cu, a similar trend was observed where its concentrations were high near the road at 10 m (30.50 ± 3.62 mg/kg in mango orchards and 30.85 ± 5.97 mg/kg in orange orchards) and decreased steadily with distance. Only Zn and Cd in mango orchard soils within 10 m radius were statistically different (p < 0.05).

Lead concentrations were highest at 10 m (21.00 ± 3.13 mg/kg in mango orchards and 19.87 ± 2.71 mg/kg in orange orchards) and dropped significantly (P < 0.05) at 100 m and 300 m. Background Pb concentrations were relatively low (8.88 ± 0.10 mg/kg and 9.28 ± 0.40 mg/kg for mango and orange orchards, respectively), emphasizing road-associated Pb contamination. Lastly, Cd was detectable only at 10 m in both mango and orange orchards, with concentrations of 0.44 ± 0.21 mg/kg (mango) and 0.35 ± 0.19 mg/kg (orange). Cd was below detection limits (<0.001 mg/kg) beyond 100 m and at the control distance (5 km). In all the samples, mean metal concentrations decreased in the order Zn > Cu > Pb > Cd.

Inter-metal correlation analysis showed that in both soils and mango fruits, there were weak to very strong positive correlations between HM pairs ( Figure 3(a) ). However, except for Cu in mangoes that had a negative correlation with Zn (r = −0.086, p > 0.05) and Pb (r = −0.030, p > 0.05) in the mango pulps, and Cu in the soils (r = −0.054, p > 0.05). On the other hand, Pearson’s correlation analysis showed that in soils and oranges, there were very strong positive correlations between metal pairs ( Figure 3(b) ). However, some associations, especially for Cu and Pb in oranges were not statistically significant (p < 0.05).

The EDI calculated for mangoes ranged from 1.74 × 10⁻⁵ mg/kg/day for Pb (at 5000 m) in children to 6.67 × 10⁻² mg/kg/day for Zn at 10 m in adults ( Table 4 ). For oranges, the EDI were 2.89 × 10⁻⁶ mg/kg/day for Pb (at 5000 m) in children to 1.16 × 10⁻³ mg/kg/day for Zn at 10 m in adults. The hazard quotients were 1.24 × 10⁻⁴ for Pb (at 5000 m) in children to 1.11 × 10¹ for Zn at 10 m in adults for mango pulps, and 2.07 × 10⁻⁵ to 1.93 × 10⁻¹ in orange juices, respectively. The corresponding hazard indices were from 1.08 × 10⁻¹ at 100 m in children to 5.24 × 10⁻¹ at 300 m in adults for mango pulps, and 1.45 × 10⁻² in adults (at 5000 m) to 7.23 × 10⁻³ at 10 m in children for orange juices.

Note: **Calculated in mg/kg/day.

For cancer risks, the estimation was only done due to Pb ingestion since Cd was not detectable in the fruits ( Table 5 ). In mangoes, the cancer risks ranged from 1.476 × 10⁻⁷ in children consuming mangoes grown within 5000 m radius of the highway to 44.27 × 10⁻⁷ for adults consuming mangoes grown within 10 m radius from the highway. For oranges, the cancer risks ranged from 0.246 × 10⁻⁷ for fruits grown 5000 m away from the highway (consumed by children) to 5.9 × 10⁻⁷ for fruits grown at 10 m away from the highway that are consumed by adults.

The potential effects of highway traffic on orchard soil pollution were assessed. The CF, I_geo and pollution load indices were calculated. The CF showed that the values lied in the classification of 1 ≤ CF ≤ 3 and 3 ≤ CF < 6 for Cu, Zn, Pb and Cd in mango orchards within 10 m radius. Beyond this distance, the CF were generally about 1 or less. In orange orchards, the same trend was observed and Cd was the major contributor to pollution ( Table 6 ). For the PLI, the values were from 0.99 to 1.81, with the highest values for soils taken from within 10 m radius of the orchard soils. For index of geoaccumulation (I_geo) in soils from orchards within 10 m radius, values were generally less than 1 for both mango and orange orchard soils. These were thus classified under class 1 (0 < I_geo < 1) for Cu, Pb and Zn ( Table 2 ; Table 6 ). For this index, the HMs showed an order: Cu > Pb > Zn > Cd.

Continued

The ecological risk coefficients were from 0.95 to 11.85 and 1.19 (for Zn) to 13.50 for Cu at 10 m ( Table 6 ). In order of severity, the risk coefficients could be arranged as Pb > Cu > Zn for mango orchard soils, and Cu > Pb > Zn for orange orchard soils. The potential ecological risk index ranged from 11.15 to 26.06 for soils from all orchards.

The present findings indicated varying concentrations of Cu, Zn, Pb and Cd in mangoes and oranges obtained along Mbale-Soroti highway. The levels of all the metals detected in fruits were below the regulatory international limit of 40, 1.5, 0.50 and 0.02 mg/kg, respectively [53] , which suggested minimal health risks associated with the consumption of the fruits. The distribution pattern of metals observed could show the influence of both environmental and plant-specific factors. The significant differences of Cu in the fruit samples suggest uniform deposition or minimal soil-to-fruit transference of the metal although Cu is essential in the plant metabolism, there could have been a strong physiological regulation in Cu uptake and translocation across the plant tissues [55] , leading to the observed variability. Relatedly, significant differences in Zn levels between mango pulp and orange juice (p < 0.05) suggested possible varying species-specific translocation dynamics or external deposition rates between the fruits. The distribution of zinc tends to vary with the fruit skin permeability, physiology, juice content, and the fruit surface area for dust adherence and the subsequent penetration into edible fruit tissues [56] .

Compared to previous studies, the levels of Cu, Zn, Pb, and Cd reported in mangoes and oranges from Uganda as per the current study were generally lower ( Table 7 ). Few studies conducted in Ethiopia [57] , Nepal [58] and India [59] reported even lower HM levels although the samples were taken from orchards near highways. In contrast, higher levels win HMs were observed in fruits grown along highways in Turkey [60] , where there are high traffic density along the highways, industrial emissions could contribute to the elevated dust load and hence metal contamination.

Factors such as antagonism, soil pH, organic matter, nature of the plant, and the chemical form could suggest why cadmium was not detectable in fruits [61] . Cadmium uptake can also be competitively inhibited by the presence of Zn since they are chemically similar, a mechanism of antagonism reported in plant systems [61] . Other factors like root barrier mechanisms, restricted xylem intake could further hinder the translocation of Cd.

Note: – means Not reported. BDL = Below detection limit; *These values were converted to mg/kg to allow for comparison.

Uganda does not have set permissible limits for HMs in agricultural soils. However, the WHO target values for unpolluted soils are 36, 50, 85 and 0.8 mg/kg for Cu, Zn, Pb and Cd respectively [54] . In the current study, HM levels in the orchard soil were all below the WHO threshold target values, as well as the control site, and the global average levels in the upper continental crust (38.9, 64.0, 27.0 and 0.40 mg/kg for Cu, Zn, Pb and Cd, respectively) [73] [74] . These findings show that there is minimal degree of contamination of the orchard soils along the Mbale-Soroti Highway. This observation suggests that vehicular activities could not be the sole contributing factor of HMs in such areas. Some of the HMs detected, such as Cu and Zn in orchard soils might be accumulating in such soils due to agricultural inputs as the possible sources. The application of fertilizers and fungicides containing the HM salts could have enriched the soils with these metals [75] . The presence of these metals in soil reflects their persistence in soils, accumulation due to the agricultural activities, and limited atmospheric deposition from the traffic load. The low levels of Pb and Cd could be ascribed to low emissions from limited leaded petrol traffic on the highway and the low industrial activity in the study area.

In comparison with related studies in other countries, HM levels reported in the current study were quite lower than those reported elsewhere. For example, Egbenda et al. [65] quantified mean Cu, Zn and Pb in mango orchard soils as 40.85, 11.10 and 2.565 mg/kg. Similarly, da Silva et al. [75] reported Cu (2.26 - 40.10 mg/kg), Zn (0.20 - 102.87 mg/kg) and Pb (undetected - 9.29 mg/kg) in mango orchard soils of Pernambuco, Brazil. Cheng et al. [70] found Pb (3.58 - 213 mg/kg) and Cd (0.05 - 12.9 mg/kg) in orange orchard soils of China. Another study reported the respective mean concentration of 90.5, 159, 21.8, and 0.94 mg/kg for Cu, Zn, Pb and Cd in orange orchard soils from Korea [76] . In Perlis, Malaysia, mango orchard soils contained Pb (0.32 - 0.42 mg/kg) and Cd (0.26 - 0.98 mg/kg) [77] . However, Bi et al. [68] previously reported exceptionally low levels of Pb (0.0008 - 0.063 mg/kg) and Cd (0.000019 - 0.00021 mg/kg) in mango orchard soils from the relatively pristine area of Hainan Island, China.

These marked differences in the HM levels reflect the importance of local factors shaping soil HM contamination. The geographical location, physicochemical properties and nature of anthropogenic activities may enrich the soil contamination with HMs [78] . For instance, pH, organic matter, and clay content strongly influence the retention, mobility, and bioavailability of HMs. Acidic soils increase the solubility of Cd and Pb, while organic matter can immobilize Cu and Zn via complexation. Environmental factors such as rainfall via surface run-off and wind, alters the extent of atmospheric deposition, while anthropogenic activities such as pesticide use, vehicular density enrich soils with HMs.

Therefore, the observed pattern of HMs in the orchard soils along the Mbale-Soroti highway shows that these soils are relatively uncontaminated. The Cu and Zn levels reflect agricultural inputs, while the low Pb and Cd levels could be due to limited atmospheric deposition. The comparatively lower levels than those reported elsewhere are indicative of combined effects of soil chemistry, traffic intensity, and land-use practices.

Correlation analysis gave an insight into how HMs can be transferred between soils and fruits. In mango fruits, Cu metal showed either very weak positive correlations or negative correlations with Cu in the soils and other metals in both matrices. The observed weak associations proposed distinct origins and limited soil-to-fruit transfer of Cu, probably due to the strong binding to the soil organic matter, clay soil particles, which in turn restrict mobility and hence plant uptake. The significant positive correlations between metal pairs suggest a common or combined source, probably agricultural inputs like Cu- and Zn-containing fungicides and fertilizers, or atmospheric deposition due to vehicular emissions [79] . Additionally, strong correlations of metals may also indicate similar retention mechanisms, like co-precipitation with carbonates, sorption to oxides, comparable geochemical behaviours in solid matrices. It also indicates that the metals entered into the environment through anthropogenic contributions [80] . When the same metal exhibits a positive soil-fruit relationship, it indicates accumulation from the soil into the edible tissues as a result of active uptake and translocation [6] [71] [81] , and the absence of such a pattern shows limited bioavailability in soils.

The EDI calculated values of Cu, Zn, Pb, and Cd for both mangoes and oranges, all were lower than the oral reference doses established by international agencies. Subsequently, the hazard quotients and hazard indices for both adults and children did not surpass 1. This observation suggested a negligible risk of non-carcinogenic effects expected in individuals due to regular consumption of mango and orange pulps from the orchards. On the other hand, the incremental cancer risk values were only calculated for Pb as the sole driver of carcinogenic health risks. The risk values calculated ranged from 0.246 × 10⁻⁷ to 44.27 × 10⁻⁷. These values were within the acceptable risk range (10⁻⁴ to 10⁻⁶) as per the US EPA, and the unacceptable value for the risks should be surpassing 10⁻⁴. Therefore, for the current study, the cancer risk values confirmed that there is no significant carcinogenic risk for both adults and children due to Pb exposure through ingestion the fruits.

The pollution indices to substantiate the contamination status of orchard soils along the Mbale-Soroti Highway showed that the contamination factor (CF) values were within the “low contamination” category as per Hakanson’s classification [48] , while the pollution load index (PLI < 1) denoted unpolluted soils [6] observed. The geo-accumulation index (I_geo) values were consistent with “unpolluted to low contamination” [49] soils. Relatedly, the ecological risk coefficients (E_i < 40 and RI < 95) showed low contamination indicative of low ecological risks [48] . These findings altogether suggested that there is low pollution of soils as well as ecological risks to the biodiversity in the orchard soils, which is similar to a previous observation in orange orchard soils reported by Xue et al. [82] . The findings thus demonstrated that orchard soils in the study area pose low ecological and health risks. The fruit contamination could be influenced by atmospheric deposition and agricultural practices than by direct soil-to-fruit transfer. This emphasizes the need to consider both environmental inputs and plant uptake mechanisms when assessing food safety near highways.