The study was carried out in the Ferlo, more specifically in the communes of Doundodji, Dodji and Labgar ( Figure 1 ), in the department of Linguère, Louga region (Senegal).

These communes have similar characteristics to Ferlo, which is a silvopastoral area located in northern Senegal. It is semi-desert [15] and belongs to the Sahelian bioclimatic zone [16] . Indeed, the Ferlo is located between latitudes 16˚15 and 14˚30 North and longitudes 12˚50 and 16˚ West, with an area of about 70,000 km². It is the largest eco-geographical region in Senegal. From an administrative point of view, this agro-ecological region covers part of the regions of Saint-Louis and Louga and the entire Matam region [5] .

The Ferlo is a region in the Sahelian zone, with a dry season lasting nine months and a rainy season lasting three months. It is part of the semi-arid Sahelian domain and experiences variability in precipitation, with an average annual rainfall ranging from 300 mm in the north to 500 mm in the south [15] - [19] . The climate is characterized by low relative humidity of 35% per year and strong evaporation from 1800 to 2200 mm·year⁻¹ [6] [8] [20] . The flora in the Ferlo is ordered according to rainfall gradient, with a zonal arrangement of phytogeographical domains interrupted by a few azonal groups at the Ferlo valley level. The main types of flora are shrubby pseudo-steppes and shrub to treed savannahs, which are deciduous, open, and often spiny, with Acacia raddiana and Balanites aegyptiaca predominating [7] [18] .

The Ferlo region is home to spiny trees, rich woody flora, and a transition zone between silvo-pastoral zones and groundnut basins. The region has developed into silvopastoral reserves and classified forests, with higher frequency of B. aegyptiaca. The herbaceous mat includes species like Aristida mutabilis, Schoenefeldia gracilis, and Cenchrus biflorus [17] . Two types of soil are distinguished in Ferlo: sandy dune systems with neutral to weakly acidic subarid reddish-brown soils and tropical ferruginous soils with a sandy to sandy-clay texture. Soils evolve according to latitude, from subarid brown to tropical ferruginous, with sandy-clay soils with clay accumulation in deep horizons. The area is characterized by flat, monotonous relief, often found in dune plateaus called “seno” in Fulani [17] [18] .

The study was carried out in the communes of Dodji, Labgar and Doundodji. These sites were chosen in a participatory manner but also on the basis of the important statistics available to the Linguère forestry service in terms of the availability of the resource.

Due to poor fruit production during the sampling season, nuts were collected from animal enclosures, where livestock such as goats had ingested and excreted intact seeds. Collection was carried out with the support of local residents and forest agents. Only nuts with intact morphology (no cracks, breaks, or insect damage) were retained. In each site, 10 kg of seeds were collected and transported in labeled bags to the Zootechnical Research Center (CRZ) in Dahra for further analysis.

At the laboratory, the nuts were cleaned and labeled. Individual nut weight was measured using a precision balance (Mettler Toledo), and length and diameter were recorded with a digital caliper ( Table 1 ). Based on these measurements, a multivariate analysis was used to classify the nuts into three size categories: small, medium, and large ( Figure 2 ). From each site, 750 nuts per category were randomly selected, totaling 2250 nuts.

Almond samples were categorized based on size and site, and biochemical analyses were conducted to determine dry matter, moisture, organic matter, fat, and crude fibers content.

To determine the organic matter, 2.5 g of ground material was weighed in a crucible previously dried and tared and then put in the oven at 550˚C for 3 hours. At a temperature of 100˚C, the crucibles are transferred to a desiccant for 30 minutes, and then weighed with the calcinations residue.

For crude fibers, the test portion is 1 g of ground material. Two successive hot extractions are carried out, with 150 ml of H₂SO₄, then 150 ml of KOH, the reagents being boiling when they are added to the sample. Each attack lasts 30 minutes from the time the boil resumes. The heating is first set to maximum and decreased as soon as the boil returns. After each attack, the product is rinsed with boiling distilled water. After the two hot extractions, rinse 3 times with acetone using a wash bottle. After rinsing, the bags were dried in the crucibles in a ventilated oven at 105˚C for 16 hours. The bags were taken out of the oven, cooled in a desiccant containing blue silica gel for 1/2 h. The contents were poured into bags in the crucibles and then weighed without letting them regain moisture. They were then charred for 3 hours at 550˚C.

Data were subjected to analysis of variance (ANOVA), followed by post-hoc comparison using the Least Significant Difference (LSD) test at a significance level of p < 0.05. Statistical analyses were performed using Statistix 8.1 software.

The analysis of variance (ANOVA) revealed that most biochemical parameters—dry matter, moisture, fat, and organic matter—did not significantly vary across provenance or size categories. However, a significant interaction effect between provenance and nut size was observed for crude fiber content (p = 0.0364), indicating that fibre levels were not uniform across the different size-provenance combinations. This interaction suggests the influence of local environmental conditions and possibly post-harvest handling practices on fiber accumulation ( Table 2 ).

OM: organic matter, CB: crude fibre, H: moisture, MSA: analytical dry matter, fat: fat.

The biochemical characteristics of the samples analyzed depend on the origin ( Table 3 ). The dry matter is relatively similar for medium and small sizes, averaging around 95%. On the other hand, for large calibers, Labgar and Dodji have the highest values, 94.99% and 94.91% respectively, while Doundodji is at 91.01%. As for humidity, it has similarities for medium and small sizes between all sites ( Table 4 ), but shows disparities at the level of large sizes, where Doundodji records the highest value at 8.99%, compared to Labgar (5.01%) and Dodji (5.03%).

Dry matter content was high across all samples, with an average above 94%, confirming the dry nature of B. aegyptiaca kernels. However, large nuts from Doundodji showed a notably lower DM (91.01%) compared to Labgar and Dodji (both near 95%). These differences might stem from variable soil conditions, local microclimates, or the fact that nuts in Doundodji were collected from animal enclosures, potentially leading to moisture uptake from fecal matter or environmental humidity.

Although not statistically significant, higher moisture content was observed in large nuts from Doundodji (8.99%) almost double that of other provenances. This discrepancy might reflect inadequate post-harvest drying or ambient humidity during collection, corroborating findings from [21] and [22] that emphasize the importance of harvest and storage conditions on seed moisture and storability. The moisture values in this study remain lower than those reported for other indigenous fruits such as Tamarindus indica (14% - 18%) or Adansonia digitata (11% - 24%), supporting the classification of B. aegyptiaca as a dry-seeded species.

Fat content ranged from 41.40% to 45.60%, with Doundodji’s small-sized nuts displaying the lowest values. These results are slightly lower than the 52% reported by Ndiaye [21] , suggesting either varietal differences or possible degradation of lipids due to storage or environmental conditions. Nevertheless, the high and consistent fat content in medium and large sizes supports B. aegyptiaca’s potential as an oilseed crop. The literature suggests that lipid biosynthesis in dryland species is strongly influenced by genotype and water availability during kernel development.

Crude fiber content showed significant variation only in interaction between provenance and size. The highest values were found in small and medium nuts from Doundodji (18.32% and 16.42%), indicating that fiber accumulation might be driven by environmental stressors such as soil compaction, grazing pressure, or nutrient deficiency, as shown in studies on similar Sahelian species [23] . These findings align with the hypothesis that stress can increase lignin and structural carbohydrate content in plant tissues.

Organic matter levels remained above 96% across all samples, suggesting a generally high nutritional density. The minor variations among sites and sizes point to a relatively uniform biochemical profile, possibly due to shared genetic material or similar agronomic practices, as described by Garcia et al. [24] .

Correlation analysis ( Table 5 ) confirmed a strong negative correlation between fat and crude fibre (r = –0.84, p < 0.01), implying that as lipid content increases, structural carbohydrate levels decline. This inverse relationship suggests a physiological trade-off between energy storage (lipid accumulation) and structural reinforcement (fiber), which is typical in seeds adapting to arid environments. These results are consistent with those from dryland crops like Ziziphus mauritiana, where high-fat varieties also showed lower structural carbohydrate content [25] .

PCA was conducted to reduce dimensionality and assess the clustering of samples based on biochemical traits. The first two components accounted for 94.34% of the variance (48.35% and 45.99% respectively). The PCA plot ( Figure 3 ) highlights a clear separation of Doundodji samples, particularly those of small and medium size, along the fiber-fat axis, supporting the statistical findings of size-provenance interactions.

This study on the biochemical characterization of Balanites aegyptiaca kernels provided insights into the variability of key biochemical traits across three geographic provenances. While the analysis of variance showed no statistically significant main effects of provenance or size on most parameters, a significant interaction between these two factors was observed for crude fiber content (p < 0.05). This interaction suggests that the effect of nut size on fiber content depends on the origin, and vice versa.

Dry matter content remained relatively stable in medium and small-sized nuts, with average values around 95%, indicating consistency in water-extractable content across these categories. However, in large-sized nuts, a marked difference was noted: Labgar and Dodji recorded the highest values (approximately 95%), whereas Doundodji presented a substantially lower value (91.01%). This variation may be attributed to genetic differences or localized agronomic practices, as supported by [24] , who documented agromorphological diversity among B. aegyptiaca individuals from different provenances.

Regarding moisture content, although not statistically significant, large-sized nuts from Doundodji had markedly higher moisture (8.99%) compared to Labgar (5.01%) and Dodji (5.03%). This discrepancy likely stems from post-harvest conditions, particularly the collection of nuts from animal enclosures, which can expose them to ambient humidity and biological degradation. These findings are consistent with studies indicating that storage environment and handling practices significantly affect moisture retention and, ultimately, the shelf life of agricultural products. The moisture values obtained in this study differ substantially from those reported by [25] , which ranged from 14.6% to 24.1%. These lower values reinforce the classification of B. aegyptiaca as a dry-seeded fruit species, as suggested by [26] . The fruit’s quality index, estimated at 2.6, places it in the semi-dry category according to [27] . Together, these indicators support its high storage potential under ambient conditions, with reduced risk of microbial spoilage [26] [27] .

Fat content showed relative uniformity across large and medium sizes at all sites, with values between 44% and 45%. However, for small-sized nuts, Dodji (44.59%) and Labgar (44.42%) had higher fat levels than Doundodji (41.43%). In contrast, [21] reported significantly higher values (up to 52%), suggesting potential regional variability or differences in extraction protocols. The uniformity observed in our results may reflect the predominance of genetic control over fat synthesis, with environmental factors playing a secondary role. Furthermore, the data reveal an inverse relationship between fat and crude fibre content kernels with higher fat content tend to have lower fiber levels a relationship further explored in the correlation analysis.

Although crude fibre content did not vary significantly with size or provenance alone (CV = 13%), the significant interaction indicates complex dynamics. Doundodji samples, particularly small-sized nuts, showed the highest fiber levels (18.32%). These results are lower than those reported by [5] in Téssékéré, despite both areas being ecologically similar. Differences in collection, storage, and processing methods may explain these discrepancies. It is also worth noting that fiber accumulation can be influenced by physiological stress, nutrient availability, and maturity stage, factors that may vary among sites.

Moreover, the nutritional value of B. aegyptiaca pulp has been contextualized by [5] , who compared its sugar content with other Sudano-Sahelian fruit species. According to their findings, B. aegyptiaca pulp has a sugar content comparable to Ziziphus mauritiana (37.6%) [28] , but lower than Phoenix dactylifera (69.9%) [29] , Detarium microcarpum (64.47%) [30] , and Maerua pseudopetalosa (67.72%), yet higher than Adansonia digitata (7.2% - 11.8%) and Tamarindus indica (14.4% - 18.6%) [25] . This suggests that Balanites fruits possess a favorable nutritional profile and may be suitable for broader dietary use by local populations.

The findings of [31] confirm that crude fiber content can vary according to growing conditions and cultivation practices, which aligns with our observation of variability among the sites studied. Such variation underscores the importance of localized agronomic management and the need for standardizing post-harvest practices if biochemical quality is to be preserved or improved.

Finally, organic matter content was remarkably consistent across all sizes and origins, with values generally exceeding 96%. This consistency suggests biochemical stability and supports the hypothesis that well-managed environmental and agricultural conditions promote uniform quality. The work of [32] has shown that stability in organic matter is often associated with controlled cultivation systems and minimal external disturbances, which may be the case in the sampled regions.