Abstract

Drying is a key technology for the preservation and processing of citrus fruits; however, its effects on bioactive compounds, antioxidant properties, and quality attributes vary considerably by method. This study aims to quantitatively evaluate the effects of different drying methods on the bioactive compounds, antioxidant activities, and colorimetric parameters of citrus matrices using a multilevel meta-analysis conducted in accordance with PRISMA guidelines. A total of 23 studies comprising 413 effect estimates were included. Effect sizes were calculated using multilevel random-effects models estimated by restricted maximum likelihood (REML) and combined with a robust variance estimate. The results show significant decreases in vitamin C content (34%), total phenolic compounds (29.4%), and total flavonoids (37.8%) after drying compared with fresh or control samples. Freeze-drying generally exhibits the best retention of bioactive compounds, while solar and hot air drying result in the greatest losses. Conversely, some variables, notably ABTS activity and narirutin content, show apparent increases after drying (>100% retention), which can be attributed to improved extractability or concentration effects rather than a true increase in compounds. Colorimetric parameters are relatively less affected by drying processes. High heterogeneity (I² > 98%) is observed for most variables studied, reflecting the diversity of citrus matrices, drying conditions, and analytical methods used, thereby justifying subgroup analyses. This meta-analysis therefore provides quantitative data to guide the selection of appropriate drying technologies to preserve the functional quality of citrus-derived products.

1. Introduction

Citrus fruits represent a significant share of global fruit production and are widely processed into juices and derivative products. Approximately one-third of total production is destined for industrial processing, generating substantial quantities of co-products ¹. Peels and pulps, which comprise 50 to 65% of the fruit's weight, are a rich source of bioactive compounds, including polyphenols, flavonoids, and vitamin C, and are recognized for their nutraceutical properties ¹. Long considered waste, these co-products are now attracting increasing interest as valuable functional ingredients in the food industry ². Recent studies have also highlighted the potential of citrus fruits and their co-products in developing functional foods and nutraceutical ingredients for sustainable food applications ^{3, 4}. Dehydration is a key step in the management and valorization of these matrices. Drying reduces water activity, extends shelf life, and limits microbial growth, while also facilitating the extraction of phenolic compounds ^{1, 5}. However, drying conditions strongly influence the stability and recovery of bioactive compounds. High temperatures generally promote the degradation of heat-sensitive compounds and decrease antioxidant activity, while moderate conditions can improve polyphenol extractability by disrupting cell walls ¹. Thus, the effect of drying results from a complex balance between degradation and the release of bioactive compounds. Conventional techniques, such as hot-air or solar drying, are widely used for their simplicity and low cost, but they often lead to significant losses of vitamins, polyphenols, and organoleptic qualities ¹. Conversely, freeze-drying better preserves nutrients, color, and aroma thanks to low temperatures and reduced pressure, though it incurs a high energy cost ². In parallel, several emerging technologies, including microwave, infrared, and vacuum drying, have been developed to simultaneously improve product quality and process efficiency. These technologies modify heat and mass transfer as well as the structure of plant matrices, directly influencing the accessibility of bioactive compounds ¹. Recent studies have also shown that innovative drying technologies can improve flavonoid retention and antioxidant capacity, depending on thermal parameters and plant matrix characteristics ⁴. Despite these advances, comparing the effects of different technologies remains difficult due to the wide variability in operating conditions, pretreatments, and matrices studied. Furthermore, literature remains fragmented, with studies often focusing on a single technology or a limited number of compounds. This methodological heterogeneity limits the identification of optimal drying strategies for preserving the bioactive compounds of citrus fruits ¹. To date, quantitative data that simultaneously integrate the effects of different drying technologies on the bioactive compounds, antioxidant properties, and colorimetric attributes of citrus matrices remain limited. In this context, the present study aims to conduct a multilevel quantitative meta-analysis to evaluate the effect of the main drying methods on the retention of bioactive compounds, antioxidant activities, and colorimetric parameters of citrus matrices. This approach also aims to identify the main sources of heterogeneity associated with drying technologies, operating conditions, and matrix characteristics, thereby providing scientific data to guide the selection of drying strategies for the preservation and valorization of citrus-derived products.

2. Methodology

A literature search was conducted in accordance with PRISMA recommendations, using the Web of Science, Scopus, and PubMed as bibliographic databases, with complementary searches conducted through ScienceDirect, SpringerLink, and Wiley Online Library. The search was initially conducted between September 2024 and February 2025 and updated in July 2026. Grey literature was explored via Google Scholar, and the references of selected articles were manually reviewed. References were managed using Zotero, and study selection was performed independently by two reviewers.

Experimental studies published between 2010 and 2026 on citrus matrices subjected to various drying processes were included. The variables studied included bioactive compounds, antioxidant activities, and colorimetric parameters. Studies without usable quantitative data, as well as reviews and conference proceedings, were excluded.

After removing duplicates, titles, abstracts, and full texts were independently reviewed by two reviewers. Discrepancies were resolved by consensus. The selection process is presented in the PRISMA diagram.

The extracted information concerned study characteristics, drying conditions, response variables, and statistical parameters (mean, standard deviation, and sample size). Additional information related to the evaluated citrus matrices and drying methods was also recorded when available. The extracted data were used for qualitative description of the selected studies and quantitative synthesis. Multiple comparisons from the same study were retained to maximize the available information while accounting for dependence among observations in the statistical analyses.

The risk of bias was assessed using a framework adapted for experimental studies in food science, inspired by the Joanna Briggs Institute and Cochrane approaches. The results were summarized in heat maps and domain-level distributions.

All statistical analyses were performed using RStudio. Effect sizes were calculated as log response ratios (logRR) or standardized mean differences (SMDs), depending on the outcome variable and measurement scale. Because multiple effect sizes could be extracted from the same study, multilevel random-effects models were fitted using restricted maximum likelihood (REML), with effect sizes nested within studies. Cluster-robust variance estimation with CR2 correction was applied to account for within-study dependence. Heterogeneity was assessed using Cochran’s Q test, I² statistics, and variance components from the multilevel models. Subgroup analyses and meta-regression models were conducted to explore potential sources of heterogeneity related to drying methods and study characteristics. Sensitivity analyses were performed using a leave-one-study-out approach, sequentially removing each study's effect sizes to assess the robustness of the pooled estimates. Potential small-study effects were assessed using funnel plots and Egger’s regression test when at least 10 independent studies were available. Funnel plot asymmetry was interpreted cautiously as an indication of potential publication bias. The certainty of evidence was qualitatively evaluated using an approach adapted from GRADE principles.

3. Results

The study selection followed PRISMA guidelines and is summarized in Figure 1. A total of 949 articles were identified. After removing duplicates and screening titles and abstracts, 624 articles were excluded. Thus, 325 articles were retained for further analysis. Of these, an additional 176 documents were excluded, including books, literature reviews, and meta-analyses. The remaining 149 articles underwent full-text review. Following this step, 126 studies were excluded for failing to meet the predefined eligibility criteria. These exclusions mainly concerned studies involving citrus-based formulations, composite or mixed matrices containing citrus ingredients rather than direct citrus matrices subjected to controlled drying treatments, as well as studies lacking sufficient methodological details or quantitative data required for meta-analysis. Finally, 23 studies were included in the qualitative and quantitative synthesis.

PRISMA flowchart describing the study selection process for the systematic review. A total of 949 records were initially identified. After screening and eligibility assessment, 23 studies were included in the final analysis. Reasons for exclusion at each stage are indicated in the diagram.

Table 1 presents the distribution of the variables studied, the number of corresponding studies, and the number of associated independent comparisons, distinguishing between solid and liquid matrices. A total of 413 independent comparisons were extracted from the 23 included studies presented in Table 2. Total phenolic compounds (TPC) constitute the most documented variable, with 108 comparisons from 8 studies. Colorimetric parameters (L*, a*, b*, ΔE) are reported in 6 studies, totaling 61 comparisons. Antioxidant activity measured by DPPH is described in 6 studies (48 comparisons), while FRAP and ABTS tests are reported in 5 studies (39 comparisons) and 1 study (6 comparisons), respectively. Vitamin C is analyzed across 6 studies, yielding 47 comparisons. Regarding specific flavonoids, hesperidin was reported in two studies (22 comparisons), naringin in three studies (24 comparisons), narirutin in one study (16 comparisons), and neohesperidin in two studies (8 comparisons). Total flavonoids (TFC) were described in 4 studies, totaling 34 comparisons. Most comparisons involved solid matrices, including whole fruits, citrus peels, and pomace. Liquid matrices were underrepresented and were used only for DPPH antioxidant activity analysis.

The drying methods identified in the included studies were grouped into different technological categories, namely freeze-drying (FD), hot air drying (HAD), infrared drying (IR), microwave drying (MW), vacuum drying (VD), solar or shade drying (SUN), spray drying (SD), and hybrid processes (HYBRID) (Supplementary Table S1). For each category, several names and technological variants were identified in the literature. This terminological diversity necessitated harmonizing the methods into standardized analytical categories to allow comparison within the meta-analysis.

Table 3 presents a descriptive summary of the operating parameters for the identified drying methods, including the number of observations, temperatures, and treatment times. Median temperatures range from 52.5°C for freeze-drying to 150°C for spray drying, while the other processes fall between 22.5°C and 80°C. Drying times also exhibit significant variability, ranging from 35 minutes for microwave drying to 4320 minutes for solar drying.

The risk-of-bias assessment revealed a predominance of low-to-moderate levels across most studies. However, some limitations were identified, particularly regarding reproducibility and the description of experimental conditions. Variability between methodological domains was also observed, suggesting heterogeneous quality among the included studies. The detailed results of this assessment are presented in Figure 2, which shows the distribution of bias levels across studies and domains.

The overall results of the multilevel meta-analysis are presented in Table 4. Overall, the drying processes resulted in significant changes in several bioactive compounds and functional properties. Vitamin C showed a significant decrease (logRR = -0.416; 95% CI [-0.608; -0.223]), corresponding to a mean retention of 66%. Similarly, total polyphenols (TPC) and total flavonoids (TFC) showed decreases (logRR = -0.334 and -0.475), with mean retentions of 71.6% and 62.2%, respectively. Regarding antioxidant activities, ABTS activity showed a significant increase (logRR = 0.968), corresponding to a mean retention of 263%. In contrast, DPPH and FRAP activities did not show significant changes. Specific flavonoids showed variable results depending on the compounds studied. Hesperidin (logRR = -0.947), naringin (logRR = -0.077), and neohesperidin (logRR = -0.785) showed no significant changes, while narirutin showed a significant increase (logRR = 0.769), corresponding to a mean retention of 215.7%. The colorimetric parameters L*, a*, and b* showed no significant changes. The overall color change (ΔE), expressed as SMD, also showed no significant effect. High heterogeneity was observed across most of the variables studied, with I² values exceeding 98%.

High heterogeneity was observed for almost all variables studied, with I² values exceeding 98%. The variable ΔE had an I² value of 59.4%. Cochran's tests (Q) were significant for all analyses (p < 0.001). Detailed Q and I² values are presented in Table 4.

•Regarding bioactive compounds and antioxidant activities

The effects of the different drying methods are illustrated in Figure 3. For vitamin C, the observed retention rates were 81.4% for freeze-drying (FD), 70.1% for vacuum drying (VD), 61.0% for hot air drying (HAD), and 55.0% for solar drying (SUN). For total polyphenols (TPC), the observed retention rates were 99.6% for FD, 75.4% for VD, 73.7% for MW, 65.6% for HAD, and 58.6% for SUN. For total flavonoids (TFC), the retention rates were 98.7% for FD, 93.2% for VD, 81.5% for HAD, 62.3% for SUN, and 28.7% for MW. Specific flavonoids also showed variations depending on the drying method. For hesperidin, the observed retention rates were 59.6% for FD, 43.3% for HAD, 67.8% for MW, and 17.0% for OD. For naringin, the observed values were 100.6% for FD, 85.2% for HAD, and 98.2% for MW. For narirutin, the observed retention rates were 256.7% for FD, 182.0% for HAD, and 215.2% for MW. Neohesperidin showed retention rates of 78.1% for FD, 31.7% for HAD, and 32.1% for OD. Antioxidant activities also varied depending on the drying method. For ABTS, the observed retention rates were 448.2% for FD and 201.0% for HAD. For DPPH, the observed values were 119.9% for FD, 102.5% for VD, and 99.7% for HAD. Regarding FRAP, the observed retention rates were 133.7% for FD, 113.7% for VD, and 89.6% for HAD. Detailed results for all variables, including specific flavonoids and colorimetric parameters, are provided inTable S2.

•Regarding colorimetric parameters

The colorimetric parameters L*, a*, b*, and ΔE did not show significant variation across the drying methods (Figure 4). For ΔE specifically, the estimates, expressed as standardized mean differences (SMDs), ranged from 80.98 to 113.02 depending on the method (Table S2), with confidence intervals that systematically included zero, confirming the absence of a significant effect.

The results of the overall tests of the effects of the drying methods revealed significant effects of the drying methods for several variables, including vitamin C (QM = 63.53; p < 0.001), total polyphenols (TPC) (QM = 15.87; p = 0.026), ABTS activity (QM = 141.27; p < 0.001), FRAP activity (QM = 12.17; p = 0.007), narirutin (QM = 43.85; p < 0.001), neohesperidin (QM = 8.27; p = 0.041), as well as the colorimetric parameters L* (QM = 58.52; p < 0.001), a* (QM = 14.80; p = 0.011) and b* (QM = 14.80; p = 0.011). 13.83; p = 0.017). No significant effect was observed for DPPH, hesperidin, naringin, and ΔE. Detailed results are presented in Table S3.

Meta-regressions revealed significant associations between temperature and the colorimetric parameters L* (estimate = -0.0015; 95% CI [-0.003; -0.000]; p = 0.021) and a* (estimate = 0.0198; 95% CI [0.011; 0.029]; p < 0.001). Drying time was also significantly associated with a* (estimate = 0.0030; 95% CI [0.002; 0.004]; p < 0.001). In contrast, no significant association was observed between total polyphenols (TPC) and temperature (estimate = -0.0014; 95% CI [-0.005, 0.002]; p = 0.443) or duration (estimate = 0.0000; 95% CI [-0.000, 0.000]; p = 0.253), nor between b* and temperature (estimate = -0.0001; 95% CI [-0.003, 0.002]; p = 0.965). The results are provided in Supplementary Table S4.

Leave-one-study-out sensitivity analysis was performed to assess the stability of the overall estimates. Excluding each study in turn did not significantly change the effect sizes for the variables studied. The overall estimates remained generally stable across all analyses. Summary and detailed results of the sensitivity analyses are presented in Supplementary Table S5. The corresponding influence plots are shown in Figures S1, S2, and S3 for bioactive compounds and flavonoids, antioxidant activities, and colorimetric parameters, respectively.

Publication bias was assessed using funnel plots and Egger's test when sufficient studies were available. Funnel plots show variable distributions depending on the parameters studied. Visual asymmetries are observed for some variables. Funnel plots corresponding to bioactive compounds, specific flavonoids, and antioxidant activities are shown in Figure S4, while those for colorimetric parameters are shown in Figure S5. Detailed results of Egger's tests are presented in Supplementary Table S6.

4. Discussion

The results of this meta-analysis show that drying processes generally reduce the levels of the main bioactive compounds in citrus matrices. Vitamin C appears to be one of the most sensitive compounds, with an average retention of 66% and a significant decrease (logRR = -0.416). Total polyphenols (TPC) and total flavonoids (TFC) also show significant decreases, with average retentions of 71.6% and 62.2%, respectively. These results confirm the high sensitivity of vitamin C to thermal and oxidative treatments, as reported by Santos and Silva ²⁸. The decrease in phenolic and flavonoid compounds is also consistent with the observations of Kumar et al. ¹⁵. However, the differences observed among the variables suggest varying levels of stability that depend on the chemical structure, cellular localization, and thermal sensitivity of the bioactive compounds.

Subgroup analyses reveal significant differences between drying technologies. Freeze-drying (FD) exhibits the highest retention rates for vitamin C (81.4%), TPCs (99.6%), and TFCs (98.7%), confirming its effectiveness in preserving heat-sensitive compounds through low temperatures and limited oxidative processes ¹⁵. Conversely, solar drying (SUN) and hot air drying (HAD) result in the greatest losses, likely due to prolonged exposure to heat and oxygen ²⁹. Microwave drying (MW) shows more contrasting effects, with good retention of TPCs (73.7%) but a significant decrease in TFCs (28.7%), suggesting variable effects depending on the nature of the compounds ³⁰. Despite the good performance of freeze-drying, its energy cost limits its industrial application. In this context, hybrid processes appear as promising alternatives, offering an interesting compromise between preservation of bioactive compounds, energy efficiency, and industrial applicability for the sustainable valorization of citrus matrices and co-products.

Specific flavonoids exhibit contrasting responses depending on the compound and drying method. Hesperidin shows relatively low retention under HAD (43.3%) and OD (17.0%), while higher values are observed under FD (59.6%) and MW (67.8%). Naringin exhibits retention close to 100%, while narirutin reaches very high values under FD (256.7%) and MW (215.2%). These retentions exceeding 100% likely reflect apparent relative increases in the measured concentrations after drying. Several mechanisms can explain these observations, including the disruption of cellular structures, which improves the extractability of phenolic compounds ³¹, as well as certain chemical transformations induced by heat treatment ¹⁶. These results are consistent with previous studies reporting an apparent improvement in some flavonoids after drying ¹⁵. Conversely, neohesperidin exhibits lower retention, suggesting greater instability under prolonged heat treatment.

Antioxidant activities vary depending on the analytical test. ABTS activity shows the most significant increases, with retentions reaching 448.2% under FD and 201.0% under HAD, while DPPH and FRAP show more moderate variations. These differences may be related to the specific chemical mechanisms assessed by each method. The observed increases have already been reported in the literature and may be associated with improved extractability of phenolic compounds or the formation of new antioxidant compounds resulting from Maillard reactions ^{16, 29}. Chen et al. specifically point out that certain drying conditions can increase apparent antioxidant activity by releasing bound compounds or transforming complex molecules ²⁹. The coexistence of decreases in bioactive compounds and increases in certain antioxidant activities illustrates the complexity of the transformations induced by drying processes.

The colorimetric parameters L*, a*, and b* did not show significant variation across the overall analyses, although some differences were observed depending on the drying method. Meta-regression analyses, in particular, showed significant associations between temperature and the parameters L* and a*. Regarding ΔE, the observed estimates ranged from 80.98 to 113.02 across drying methods, with no overall significant effect. These results are generally consistent with several studies reporting that some biochemical changes induced by drying do not necessarily translate into significant visual changes ³². The variability observed between studies may be related to differences in matrices, operating conditions, and instrumental methods used for color evaluation.

One of the main findings of this study is the very high heterogeneity observed across most variables, with I² values exceeding 98%. This heterogeneity is confirmed by significant Q tests across all analyses. This variability can be explained by several factors, including the diversity of the matrices studied, differences in operating conditions, applied pretreatments, and the heterogeneity of analytical methods used between studies ²⁹. Meta-regression analyses show that certain parameters, particularly temperature, significantly influence some colorimetric variables. However, these factors only explain a limited portion of the observed heterogeneity.

This study has several important strengths. To our knowledge, it is one of the first multilevel meta-analyses dedicated to the effects of drying processes on the bioactive compounds and functional properties of citrus matrices. The use of multilevel models with REML estimation and robust CR2 variance accounted for dependence among comparisons within the same study. The simultaneous integration of bioactive compounds, specific flavonoids, antioxidant activities, and colorimetric parameters is also a significant advantage. However, several limitations must be highlighted. The number of studies remains limited for certain variables, particularly specific flavonoids and some antioxidant activities. Furthermore, the high heterogeneity observed limits the generalizability of the overall estimates. Finally, methodological differences across studies complicate direct comparisons of results. However, the use of cluster-robust (CR2) inference yielded reliable estimates of uncertainty while accounting for dependence among effect sizes from the same studies (Supplementary Table S7). The assessment of the certainty of evidence revealed variable levels of confidence across the evaluated parameters, mainly due to high heterogeneity and the limited number of available studies for certain outcomes. The detailed results of this evaluation are presented in Supplementary Table S8.

The results obtained highlight the importance of selecting drying technologies to preserve bioactive compounds in citrus fruits. Freeze-drying appears to be the most effective method overall, although some alternative technologies may offer specific advantages depending on the compounds studied. These results open up avenues for optimizing citrus processing methods and valorizing co-products rich in bioactive compounds. Future studies using standardized protocols and more in-depth mechanistic approaches will be necessary to better understand the transformations induced by drying processes.

5. Conclusion

Drying processes are important strategies for the preservation, stabilization, and valorization of citrus matrices. However, their effects on nutritional and functional attributes remain highly variable depending on the drying technology used, the operating conditions applied, and the characteristics of the matrices studied. This multilevel meta-analysis provides quantitative data on the impact of drying processes on the biochemical composition, antioxidant properties, and colorimetric quality of citrus matrices. Overall, drying significantly reduces the levels of the main bioactive compounds, including vitamin C, total phenolic compounds, and total flavonoids, confirming their sensitivity to thermal and oxidative stresses. Among the technologies evaluated, freeze-drying showed the best capacity for retaining bioactive compounds, while solar drying and hot air drying were associated with the greatest losses. However, responses for specific flavonoids and antioxidant activity varied with the compounds and the analytical methods used, suggesting that drying-induced changes result from both degradation and changes in compound extractability. Retention rates exceeding 100%, observed for certain parameters such as narirutin content and some antioxidant activities, should therefore be interpreted as apparent increases related to concentration effects or improved extractability rather than actual compound synthesis. Colorimetric parameters generally appear more stable, although some variations may occur depending on processing conditions. The significant heterogeneity observed between studies highlights the important influence of citrus matrices, drying technology parameters, and analytical methodologies, emphasizing the need for more standardized experimental protocols. Overall, these results provide scientific evidence to guide the selection of appropriate drying strategies to preserve functional quality and enhance the valorization potential of citrus products and by-products.

Conflict of Interest Statement

The authors received no specific funding for this work and declared no conflicts of interest.

Author Contributions

Ana-kpan Dome Vincent Bekuoné Somé: Conceptualization, Methodology, Investigation, Data curation, Formal analysis,Writing original draft; Mamounata Diao: Conceptualization, Methodology, Validation, Supervision, Writing review & editing; Rocksane Octavia Kiswendsida Nikiema: Investigation, Data curation, Literature screening, Writing review & editing;

David Bazié: Methodology, Validation, Writing review & editing;

Désiré Ouaro: Investigation, Data curation, Risk of bias assessment, Writing review & editing;

Zakaria Dindané: Investigation, Data curation, Risk of bias assessment, Writing review & editing;

Roger Dakuyo: Data curation, Validation, Writing review & editing;

Fréderic Anderson Konkobo: Investigation, Data curation, Writing review & editing

Hemayoro Sama: Validation, Writing review & editing; Samson Guenné: Validation, Writing review & editing; Mamoudou Hama Dicko: Supervision and Validation

Supplementary

References