Metabolome Reprogramming During Fruit Ripening and Post-Harvest Storage in Ten Crop Species

Abstract

Background/Objectives: Plants alter metabolites of their fruits during the ripening process, leading to improved nutritional properties and taste. In addition, metabolite compositions continue to change on the shelf after harvest. However, the dynamics of these important processes are species-specific and so this study aimed to contrast the ripening dynamics of ten different fruit species simultaneously. Methods: Plant material was collected from the fruits of apple, banana, blueberry, kiwifruit, pear, plum, peach, strawberry, raspberry, and tomato at three different stages: unripe, fully ripe, and overripe fruits. Comparative metabolome analysis by GCMS was performed to identify differentially abundant metabolites across the species of this study and to examine their dynamics across ripening and post-harvest storage. These results were complemented by elemental compositions derived from a literature search. Results: In a first, this study demonstrated that both baseline metabolite abundances and their dynamics across ripening clustered species vary largely according to their phylogeny. Comparisons across ripe fruit identified differences in nutritional properties, highlighting species such as banana to be of especially high nutritional value and blueberry and peach to be prominent sources of antioxidants. Comparing the ripening dynamics of all species identified common patterns, such as the conversion of organic acids to sugars and cell wall dynamics, although species-specific responses were also acknowledged, in particular, kiwi and the Rosaceae berries, which may explain differences in post-harvest shelf-life. Conclusions: The observed inter- and intra-specific variation in nutritionally relevant metabolites and elements serves as a reference for both producers and consumers and emphasizes that consuming a variety of fruits, not only across species but also across cultivars within a species, can maximize the intake of beneficial phytonutrients, sugars, amino acids, and antioxidants.

1. Introduction

Fruits are widely recognized as essential components of a balanced diet due to their high content of minerals, especially electrolytes; vitamins, dietary fiber, and antioxidants, all of which contribute to maintaining overall human health [1,2]. According to the World Health Organization (WHO) and the Food and Agriculture Organization (FAO), a minimum daily intake of 400 g of fruits and vegetables per person is recommended to support overall well-being and reduce the risk of chronic diseases [3].

During fruit development and ripening, fruits undergo complex biochemical and physiological changes such as starch and chlorophyll degradation, accumulation of non-photosynthetic pigments (carotenoids and anthocyanins), remodeling of cell wall components leading to softening, and dynamic shifts in the composition of primary and secondary metabolites such as sugars, organic acids, amino acids, phenolics, flavonoids, and volatile compounds [4,5,6]. These metabolic trajectories can follow both linear and non-linear patterns over time and are largely driven by hormonal regulation. Climacteric fruits are governed by an ethylene burst after which they can continue to ripen off plants, whilst non-climacteric fruits are regulated by abscisic acid and other hormone pathways and cannot ripen once removed from the plant [7]. Understanding these metabolic transitions is crucial for improving fruit quality, flavor, and postharvest performance.

Recent metabolomic studies have provided valuable insights into the biochemical mechanisms underlying fruit ripening in several species individually, identifying metabolic patterns that contribute to fruit quality, flavor, and nutritional value in ripe fruit [1,8,9,10,11,12,13,14,15,16,17]. However, studies that compare multiple species simultaneously are underrepresented in the literature. Moreover, while metabolic changes during early and optimal ripening stages have been extensively studied, the dynamics of metabolism during late ripening and overripe stages, particularly under postharvest conditions, remain underrepresented. A more comprehensive analysis that includes multiple species at these overripe stages would therefore provide a deeper understanding of fruit physiology, quality traits, and nutritional composition.

Consequently, this study addressed this research gap by profiling metabolites across ten different fruit species during ripening and post-harvest storage. This study aimed to identify key metabolites and biochemical pathways associated with fruit ripening and to compare interspecific metabolic differences that influence postharvest physiology, nutritional quality, and flavor. By comparing multiple species across multiple ripening stages, this work provides a broader perspective than previous single-crop studies and highlights metabolites potentially important for fruit quality optimization of our diet.

2. Materials and Methods

2.1. Plant Material and Sampling

Fruit material from ten species was used in this study, including apple (Malus do-mestica cv. Golden Delicious), banana (Musa acuminata cv. Cavendish), blueberry (Vaccinium corymbosum cv. Elliott), kiwifruit (Actinidia deliciosa cv. Hayward Green), peach (Prunus persica cv. Evmolpiya), plum (Prunus domestica cv. Serdika), pear (Pyrus communis cv. Conference), raspberry (Rubus idaeus cv. Lyulin), strawberry (Fragaria × ananassa cv. Katrina), and two breeding lines of tomato (Solanum lycopersicum). From each species or variety, fruits were sampled across three ripening stages (unripe, fully ripe, and overripe). The BBCH-scale is a system for a uniform coding of phenologically similar growth stages of all mono and dicotyledonous plant species. BBCH codes for each sampling point and fruit type are provided in Supplementary Table S1. For each time point, three fruits/biological replicates were collected for each variety. Fruits were obtained either directly from plants (blueberry, raspberry, strawberry, and tomato) or by harvesting unripe fruit and allowing them to ripen to the desired stage at room temperature (apple, banana, peach, plum, kiwifruit, and pear). Fruit tissue was collected from each sample and immediately frozen in liquid nitrogen and stored at −80 °C until analysis.

Ripening stages were defined based on morphological and textural criteria rather than fixed time points. Fruits were categorized as unripe, fully ripe, or overripe according to visible color changes, surface smoothness, and firmness. For fruits collected directly from plants (blueberry, raspberry, strawberry, tomato), samples were harvested when they visibly matched the defined ripening stage. For these species, the transition from unripe to overripe stages was generally accompanied by clear changes in fruit color and texture. For the remaining climacteric fruits (apple, banana, peach, plum, pear, and kiwifruit), unripe fruits were harvested and allowed to ripen post-harvest at room temperature until they reached the ripe and overripe stages. In species where color changes were less pronounced, ripening stages were primarily assessed by firmness and smoothness of the fruit surface. Fruits from different ripening stages were stored separately and sampled independently to prevent cross-contamination.

The unripe stage (Stage 1) included firm fruits with incomplete color development, although the specific visual and textural cues varied among species. For fruits such as banana, apple, blueberry, raspberry, strawberry, plum, and tomato, unripe fruits were easily recognized by their color and firmness, whereas for fruits like pear, peach and kiwifruit, which exhibit no color change during ripening, firmness and surface smoothness were the main indicators. The fully ripe stage (Stage 2) included fruits with characteristic color, texture, and optimal eating quality, with the relative contribution of color or firmness depending on the species. The overripe stage (Stage 3) included fruits showing pronounced softening and early signs of senescence, again with species-specific differences in how these traits manifested. Ripening duration varied among species and was assessed based on these morphological and textural criteria rather than chronological time

2.2. Metabolome Analysis by GC-MS

Metabolite extractions were carried out with 50 mg of fruit tissue per sample, which was ground into fine powder in liquid nitrogen. The ground tissue was extracted with 750 µL of extraction buffer consisting of 100% methanol containing the internal standard ribitol (4 µg/mL). Samples were homogenized for 2 min at 20 Hz using a tissue homogenizer (Starbeater, VWR, Radnor, PA, USA) with precooled racks. Subsequently, 400 µL of chloroform was added, and samples were incubated at 1200 rpm for 10 min at 20 °C on an orbital shaker homogenizer (HM100-Pro, DLAB Scientific, Beijing, China). Phase separation was induced by the addition of 800 µL of HPLC-grade water, followed by centrifugation at 10,000 rpm for 10 min. The resulting polar phase was collected for further analysis.

For the GCMS analysis, derivatization was carried out on 100 µL of dried metabolite extract using 40 μL of 20 mg/mL methoxyaminehydrochloride (Merck, Rahway, NJ, USA) in pyridine and 70 μL trimethylsilyl-N-methyl trifluoroacetamide (Machinery-Nagel, Düren, Germany) [18]. Derivatized extracts were analyzed on a TSQ9000 GCMS (Thermo Fisher Scientific, Waltham, Massachusetts, USA) following a one-microliter injection. Helium was used as carrier gas at a constant flow rate of 2 mL.s⁻¹ and gas chromatography was performed on a 30 m DB-35 column with 0.32 mm inner diameter and 0.25 μm film thickness (Agilent Technologies, Santa Clara, CA, USA). The injection temperature was 230 °C and the transfer line and ion source temperatures were set to 250 °C. The initial temperature of the oven (85 °C) increased at a rate of 15 °C min⁻¹ up to a final temperature of 360 °C. After a solvent delay of 180 s, mass spectra were recorded at 20 scans s⁻¹ with 70–600 m/z scanning range. The chromatograms were processed in Xcalibur v.4.1.31 and peaks were annotated to an internal library of over 100 standards recently analyzed on the same instrument. In addition, prominent peaks not present in the internal library were manually annotated by matching spectra to the NIST Spectral Library. All metabolites were ranked by the certainty of their annotations, where a level of A refers to the most certain annotations, which were verified by an authentic standard. The level of B refers to peaks with spectral matches and parsimonious retention times but no reference standard. Finally, a level of C refers to putative annotations.

2.3. Statistical Analysis

All downstream data processing and analysis were performed in R v. 4.4.2 [19]. Firstly, the matrix of peak areas was normalized by a log2 transformation and corrected for signal drift with a linear model by subtracting the predicted value given the injection order. After correction, the data were normalized by the internal standard (ribitol).

For principal component analysis, the normalized abundances were further scaled and centered using the function prcomp under the default settings. To identify differentially abundant metabolites across the fruit at the ripe stage only, an analysis of variance (ANOVA) model was fit separately for each metabolite using the function aov(Abundance~Fruit). The resulting p-values were corrected for multiple comparisons by the Benjamini–Hochberg (BH) method and filtered to retain only metabolites with BH corrected p-values < 0.05. The scaled average abundances of these were plotted in a heatmap using the pheatmap package with samples and metabolites clustered via the Ward.D2 algorithm [20,21]. In all instances, data scaling was performed separately for each metabolite by subtracting values from their mean (across all samples) and dividing this difference by their standard deviation to produce a relative Z-score.

Following this, samples were analyzed at the level of their family group. To identify metabolites altered in a linear fashion across ripening, a linear model was fit on the scaled normalized abundances for each metabolite within each family group using the function lm(Abundance~Family_Group*Stage). Here, stage was treated as a numeric variable. After correcting p-values via the BH method, metabolites were identified with at least one BH-p-value < 0.05. The slopes of each family group for each metabolite were extracted from the linear model by summing the specific interaction estimate for a family group with the stage estimate (which represents the slope of the reference family group). The slopes for each family group were visualized using the pheatmap package with the default parameters. To identify the effect size of select metabolites, e.g., Unknown 511 mz, the model was re-fit as before, but this time on unscaled normalized abundances to extract the true gradient in log2 space.

Finally, metabolites differing across ripening stage within each family group were additionally identified by an ANOVA by treating ripening stage as a factor using the following model: aov (Abundance~Stage). Again, metabolites with at least one BH-corrected p-value < 0.05 were selected and their normalized averages within each family group were plotted across the three ripening stages.

2.4. Elemental Data Acquisition

Select element abundances were retrieved from previously published literature and the USDA FoodData Central database which are referenced in text. In all instances, the elemental composition of ripe fruit was selected.

3. Results

3.1. Fruit Ripening

Fruits from the ten different species, spanning a diverse phylogeny, were collected at the three ripening stages (unripe, fully ripe, and overripe fruits) and are presented in Figure 1. To collect samples that can be meaningfully compared across species, the ripening of each fruit species was followed. For some fruits, such as banana, blueberry, raspberry, strawberry, and tomato, the transition from unripe to fully ripe and then to overripe stages was concomitant with changes in fruit color and is clearly visible in Figure 1, while for other fruits this transition did not result in change in color but could be detected by the smoothness of the outside layer and the softness of the fruit. For example, pears were visibly undistinguishable across the three ripening stages but could be distinguished by their softness with the unripe fruits very hard and the overripe fruits very soft. By relying on these visual and textural cues, samples could be collected with comparable relative ripeness across the different species for metabolomic analysis.

3.2. Metabolome Cross-Species Comparisons

To assess metabolomic changes associated with ripening across the different species, GCMS analysis was performed on polar metabolite extracts. In total, 96 metabolites could be quantified across the ten fruit types (Supplementary Tables S2 and S3). Principal component analysis (PCA) largely separated samples by species, with overlap observed primarily among closely related taxa (Figure 2A). Specifically, the two tomato varieties belonging to Solanum lycopersicum overlapped. Similarly, members of the Rosaceae clustered into three distinct but internally overlapping groups: pome fruits (apple and pear), stone fruits (Prunus spp.; peach and plum), and Rosaceae berries (strawberry and raspberry) (Figure 2B).

To identify metabolites that drive this separation, an ANOVA was performed on ripe fruit (Stage 2). This revealed 78 differentially abundant metabolites with BH-p-values < 0.05 (Supplementary Figure S1). The top 30 most significant metabolites are shown in a heatmap in Figure 3, as these resulted in the samples clustering according to phylogeny. Banana, the only Monocot, was clustered separately while the remaining fruits separated into rosids (Rosaceae) and asterids (Solanaceae, Ericaceae, and Actinidiaceae), with the Solanaceae correctly resolved as an outgroup within the asterid clade. Based on these patterns, and to simplify downstream analyses and visualization, species were subsequently grouped at the family level; however, the Rosaceae were further subdivided into three groups reflecting pome fruits, stone fruits, and berries (Figure 2B).

Despite these phylogenetic groupings, unique metabolic signals were observed within groups. For instance, the two tomato varieties differed in their contents of phenylalanine and nicotinic acid, both of which were the most abundant in either variety 23 or 32 across all species (Figure 3 and Figure S1). Within the stone fruit, peaches had the greatest levels of the essential amino acid methionine, whilst plums contained the highest levels of 4- and 5-caffeoylquinic acid across all species. However, peaches exhibited the second-highest levels of these caffeouylquinic acids and both species contained an above average level of 3-caffeouyl-quinic acid (Figure 3 and Figure S1).

Evidently, more pronounced differences were present between family groups. For example, banana showed elevated levels of disaccharides, including sucrose, relative to the other groups, as well as increased abundance of a putative dopamine derivative and the amino acids tyrosine, leucine and valine (Figure 3). In addition, blueberry contained a unique, unknown metabolite with a mass-to-charge ratio of 511 and the highest level of 3-caffeouyl-quinic acid. Other trends included elevated profiles of catechine, xylose and asparagine in all Rosaceae and higher glutamic acid content in the Solanaceae (Figure 3). It is worth pointing out that the two stone fruits (plums and peaches) had the lowest levels of glucose and particularly fructose, but above average levels of sucrose—second only to banana (Supplementary Figure S1). Adding to this, the highest levels of glucose and fructose were found in blueberry and the second highest in kiwifruit, which will be discussed below (Supplementary Figure S1).

3.3. Linear Trends of Metabolites Across Ripening

To identify metabolites with linear trends across the three stages of ripening, a linear model was fit for each scaled normalized metabolite abundance, including family group, ripening stage and their interaction as covariates; with ripening stage treated as a numeric value from one to three. The slopes of 23 metabolites, which had a significant BH-corrected p-value < 0.05 for their slope estimate of stage, or a significant stage by family group interaction for at least one family group, were plotted in the heatmap Figure 4. As observed for baseline metabolite differences, samples clustered largely according to phylogenetic relationships, indicating that ripening-associated metabolic trajectories are broadly conserved within evolutionary lineages. Minor deviations from the true phylogenetic relationships (Figure 1L) were observed, including closer clustering of tomato (Solanaceae) with Rosaceae relative to kiwi or blueberry, and the classification of the stone fruit (instead of the berries) as an outgroup within the Rosaceae.

Banana exhibited the strongest accumulation of the two most abundant fruit monosaccharides, glucose and fructose, across ripening. Consistent with fruit sweetening, these sugars increased in most other family groups, with the notable exception of kiwi and, to a lesser extent, the pomes. In contrast to all other species, banana uniquely showed a decline in pyroglutamic acid, glyceryl glycoside, and a putative dopamine derivative over ripening.

Kiwifruit, besides decreasing monosaccharides such as glucose, fructose and galactose across ripening, was the only species to decrease tyrosine. Moreover, kiwifruit altered metabolites associated with the GABA shunt, namely: 4-Aminobutanoic acid (GABA), glycine, alanine and glyoxylic acid were increased whilst pyruvic acid and aspartic acid were decreased.

Blueberry contained a unique, putatively unidentified metabolite (511 mz) that exhibited the strongest slope across ripening amongst all metabolites. When inspected for effect size using the unscaled (log₂-space) abundances, the linear model estimated a slope of 3.8, corresponding to an approximately 14-fold increase in abundance per unit increase in ripening stage. Along with the two tomato varieties, blueberry was also the only species to degrade malic acid and maleic acid across ripening. The tomatoes were additionally characterized by a steep decline of trans-3-caffeoyl-quinic-acid and three N-containing metabolites, namely: GABA, 4-hydroxy-proline and β-Alanine.

Lastly, characteristic patterns present in the Rosaceae were the increase in gluconic acid and pyruvic acid in the stone fruit and pomes whilst the berries (strawberry and raspberry) increased glyceryl glycoside across ripening.

3.4. Non-Linear Trends of Metabolites Across Ripening

To identify metabolites following non-linear trends across ripening which may have been missed in the prior linear model, an ANOVA was performed for each species group treating ripening stage as a factor. When correcting the p-values for multiple comparisons, 35 unique metabolites were significantly (BH-p-value < 0.05) altered across the three stages of ripening in at least one species. Of these, 23 were not identified by the linear model and are plotted in Figure 5.

This analysis highlighted several cell wall derived sugars, including the disaccharides cellobiose and β-gentiobiose, and the monosaccharide xylose, as significantly altered during ripening. In most species, these compounds increased across ripening. However, especially for cellobiose, the rosaceous berries, pomes and blueberry showed a distinct pattern in which levels decreased or remained stable at the ripe stage before increasing during overripening. In contrast, banana exhibited increased accumulation of the cell-wall-derived disaccharides at the ripe stage, followed by a plateau into over-ripeness, while xylose showed a transient decrease specifically at the ripe stage for banana only.

Additional non-linear trends were observed for glucose-6-phosphate and fructose-6-phosphate, which declined sharply during overripening in kiwifruit and banana only. Interestingly, these species also presented the highest and lowest levels of myo-inositol respectively, with both species increasing their abundance across ripening.

Finally, this analysis also highlighted the dynamics of acids across ripening. Methyl-maleic acid tended to decrease linearly for all species except the rosaceous berries. In fact, when inspecting the other acids such as fumaric acid, glyceric acid and succinic acid, the rosaceous berries exhibited the strongest acid accumulation in over-ripe fruit out of all family groups (Figure 5). Only two acids were not increased in overripe fruit, namely 2-oxoglutaric acid (also known as α-ketoglutaric acid) and malonic acid. The former tended to follow a linear decrease across ripening and the latter showed a spike within ripe fruit for most species (Figure 5).

3.5. Elemental Composition of the Fruit

To complement our metabolomic study and provide a broader perspective on the nutritional value of the analyzed fruits, data on macro and microelement composition (including K, Ca, Mg, P, Fe, Zn, Cu, and Mn) were collected from previously published literature and the USDA FoodData Central database (

Table 1. The range of elemental composition of the fruit species of this study (mg/100 g of fresh weight). Bold indicates high levels compared with other species.

Species	K	Ca	Mg	Fe	Zn	P	Cu	Mn	References
Raspberry	71.8–156	1.14–28.3	15.9–21.1	0.45–1.33	0.22–0.37	5.7–27	0.052–0.68	0.43–4.06	[22,23,24,25]
Strawberry	51.2–161	2.2–24.6	8.78–12.5	0.26–1	0.11–0.13	6.6–23	0.02–0.035	0.368–0.41	[22,23,25]
Blueberry	70.1–110.5	8.2–20.3	4.9–11.9	0.34–0.544	0.09–0.607	8.6–13	0.024–0.733	0.171–0.582	[22,23,25,26]
Tomato	11.9–193	10–32.04	8.1–9.55	0.1–0.48	0.08–0.31	19–33.04	0.032–0.05	0.087–0.36	[22,27]
Kiwi	198–312	26–35	15.7–17	0.24–0.39	0.14–1.04	25.21–34	0.12–1.45	0.064–0.771	[22,28,29,30]
Banana	326–524	5–26	27–54	0.4–0.86	0.16–0.5	2–22	0.101–0.28	0.258–0.52	[22,31,32]
Peach	121–225	3.34–9.73	6.36–9.72	0.131–0.479	0.0752–0.23	1.4–22	0.0468–0.085	0.026–0.0465	[22,33]
Plum	157–186	4–6	6.6–7	0.25–0.17	0.05–0.1	16–19	0.054–0.057	0.04–0.052	[22,34]
Pear	79.9–122	4.61–10	5.7–6.5	0.106–0.25	0.07–0.11	12–13.2	0.058–0.081	0.054–0.03	[22,35]
Apple	95–118	2.48–6	3.9–5	0.1–0.15	0.02–0.04	2.2–11	0.024–0.07	0.029–0.06	[22,36]

). These reference ranges highlighted above average levels of potassium, zinc and copper in banana and kiwifruit. Banana and kiwifruit also contained above average contents of magnesium and calcium respectively. The berries (raspberry, strawberry and blueberry) exhibited the highest ranges of iron and manganese and raspberry exhibited particularly high zinc content (Table 1). Finally, tomato and kiwi had the largest range values for phosphorus.

4. Discussion

4.1. The Fruit Metabolome Reflects Phylogenetic Relationships

Our study provides a comparative overview of metabolome changes across ripening and post-harvest storage in ten crop species. We show that both baseline metabolite abundances and ripening-associated metabolic trajectories resulted in sample clustering that closely mirrored phylogenetic relationships. Such phylogenetically structured baseline metabolomes have been reported previously for specialized metabolites in crops [21,37]. Metabolic responses to ripening have in part also been linked to phylogeny in meta-analysis [4]. However, to our knowledge, this is the first time that such phylogenetic clustering has been demonstrated in a single study for the more evolutionarily conserved primary metabolites and their ripening dynamics, highlighting the importance of phylogenetic context when assessing fruit metabolism.

4.2. Evaluating Nutritional and Metabolic Composition Across Phylogeny

Dissecting these genetically associated metabolic differences revealed several lineage-specific metabolic fingerprints which characterize their dietary relevance. Notably, blueberry contained a unique, putatively unidentified metabolite (m/z 511) that exhibited the strongest positive ripening-associated trend across the dataset. Alongside the greatest abundance of the antioxidant trans-3-caffeouyl-quinic acid, it may play a specialized role in blueberry fruit development, flavor and its famed antioxidant power [38,39].

Similarly, the stone fruits (peach and plum) contained higher levels of three caffeouyl-quinic acids and catechin. This aligns with the work of previous studies which report that caffeoylquinic acids (aka. Chlorogenic acids) and catechin are the most abundant phenolics within stone fruit and that these are primarily responsible for its antioxidant capacity and differentiate stone fruit from other fruits such as the pomes [40,41,42]. This work extends these findings by highlighting these as a defining feature of the stone fruit amongst the ten species analyzed here, and that this signature likely underpins some of the flavor and health benefits of this group [39]. Adding to this, another signature of the stone fruit was seen in their above average sucrose concentrations whilst exhibiting lower glucose and fructose levels relative to other species. This suggests reduced sucrose hydrolysis during ripening and likely also contributes to the flavor and specifically the sweetness of this group.

More strikingly, banana exhibited the highest abundance of sucrose, in addition to other disaccharides, amongst the ten species of this study. Consistent with this, banana also has the highest caloric density of the fruits included in this study at approximately 88 kcal per 100 g [22], suggesting that its elevated disaccharide content contributes substantially to its energy value. Moreover, banana contained the highest levels of the essential amino acids leucine and valine and the second-highest level of phenylalanine (after tomato), positioning this fruit to be of high nutritional quality and of particular relevance to consumers relying on predominantly plant-based diets.

The most striking metabolic feature of banana was evident in the higher abundance of an unknown dopamine derivative and tyrosine. Banana is known to synthesize dopamine and melatonin, which lead to the characteristic brown spotting on banana peel during ripening [43]. These are synthesized from derivatives such as L-DOPA (dihydroxyphenylalanine) or tyramine, which in turn are synthesized from tyrosine. L-DOPA has been observed to decrease across ripening and given that we observe the same here for this unknown compound alongside an increase in tyrosine, this supports that it is indeed a rapidly converted intermediate within this pathway [43]. Given the strong antioxidant capacity of melatonin, this process likely buffers oxidative stress in overripe fruit and further highlights the nutritional value of banana in overripe fruit [44].

Additional patterns of nutraceutical relevance were observed between the two tomato varieties. Variety 23 contained the highest levels of nicotinic acid (vitamin B3), whereas variety 32 showed the highest abundance of the essential amino acid phenylalanine. These differences indicate that nutritionally important metabolites can vary substantially even among cultivars within the same species, which is in accordance with other comparisons across tomato cultivars [45]. Taken together with the broader interspecific variation observed here, these findings suggest that consumers seeking to maximize the intake of beneficial phytonutrients may benefit not only from consuming a diversity of fruit species but also from selecting across varieties within a given species.

4.3. Metabolic Changes Across Ripening and Senescence

Consistent with classical models of fruit ripening, the monosaccharides glucose and fructose increased across ripening in most species, reflecting starch breakdown and/or sucrose cleavage and contributing to fruit sweetening. Kiwi represented a notable exception, showing relatively stable or declining monosaccharide levels despite undergoing ripening. Declining glucose across ripening has previously been reported for kiwi [46]. Within this dataset, kiwi contained the second-highest levels of glucose and fructose and so potentially has reduced the selective pressure for further sugar accumulation during late ripening stages.

Alternatively, carbon may be rerouted away from soluble sugars in kiwi toward the GABA shunt. In plants, GABA catabolism proceeds predominantly via pyruvate- and glyoxylate-dependent GABA transaminases, thereby linking GABA turnover to the production of alanine and glycine, respectively [41]. The coordinated accumulation of GABA, alanine, glycine, and glyoxylate in kiwi is therefore consistent with enhanced engagement of GABA-associated metabolism during ripening. The decline in pyruvate suggests that it may be consumed by GABA transamination more rapidly than it is replenished through glycolysis, potentially relating to the concurrent decrease in glucose and fructose that serve as glycolytic substrates. Moreover, declining aspartic acid supports a broader reconfiguration of nitrogen metabolism, which may influence glutamate availability required for GABA biosynthesis from α-ketoglutarate [47]. Together, these patterns indicate a rerouting of carbon and nitrogen metabolism during kiwi ripening, potentially contributing to the limited accumulation of soluble sugars observed in this species.

Moreover, kiwi also exhibited the highest abundance of myo-inositol, a sugar alcohol derived from glucose-6-phosphate, suggesting additional metabolic routing of carbon away from free hexoses. In contrast, banana, which displayed the strongest accumulation of glucose and fructose across ripening, contained the lowest levels of myo-inositol. Myoinositol has recently been shown to alter plant sugar accumulation in a light-dependent manner [48]. Thus, this inverse relationship could suggest a trade-off between monosaccharide accumulation and sugar-alcohol metabolism that may reflect species-specific strategies for carbon allocation during ripening.

In several species, organic acids such as methyl-maleic acid and 2-oxoglutaric acid (also known as α-ketoglutaric acid) decreased across ripening. Within the Solanaceae and blueberry, malic acid and maleic acid also declined across ripening while monosaccharides increased, displaying opposing slopes. This pattern is consistent with the well-documented conversion of organic acids, particularly those from the tricarboxylic acid (TCA) cycle, into sugars via gluconeogenesis or their use as respiratory substrates during ripening [49].

In this study, strawberry and raspberry were the only species to show an accumulation of methyl-maleic acid in overripe fruits and showed the most pronounced accumulation of fumaric acid, succinic acid and glyceric acid specifically at the overripe stage. These metabolites are closely associated with the TCA cycle, suggesting a reconfiguration of respiration during senescence in overripe fruit. In tomato, failure to consume TCA cycle intermediates, particularly malic acid, during ripening has been associated with accelerated fruit softening and reduced fruit quality [50,51]. Similarly, enhanced expression of TCA cycle genes and succinic acid is associated with water core disorder in overripe pears [52]. Taken together, this suggests that the accumulation of TCA intermediates in overripe Rosaceous berries may reflect enhanced senescence and fruit softening in these highly perishable berries.

Fruit softening during ripening is driven by extensive cell wall remodeling, and this was reflected in the accumulation of cell-wall-derived sugars [53,54]. The disaccharides cellobiose and to a lesser extent β-gentiobiose increased markedly during ripening across many species. These present the end products of cellulose and β-glucan depolymerization respectively and so their increase is indicative of fruit softening. The increase in xylose levels across ripening in most species likely reflects further hemicellulose (xylan) solubilization and continued cell wall disassembly during fruit ripening/softening. Interestingly, banana showed the strongest increase in these disaccharides in ripe fruit but was the only fruit species to decrease xylose at the ripe stage.

Adding to this, with the exception of the tomatoes, all species groups increased 4-hydroxyproline across ripening. Given its role as a key component of hydroxyproline-rich glycoproteins and arabinogalactan proteins, this increase may reflect enhanced turnover or remodeling of structural cell wall proteins [55]. Taken together, these observations indicate that distinct yet coordinated polysaccharide and glycoprotein remodeling processes underpin fruit softening. Such differences in cell wall dynamics across fruit types may account for differences in textures and softening dynamics across the fruit species.

4.4. Integrating Elemental Composition into Fruit Nutritional Profiles

In addition to species-specific metabolite profiles, fruits also represent important dietary sources of essential mineral elements that contribute to human nutrition and health. As summarized in Table 1, the analyzed fruit species differ substantially in their contents of macroelements such as potassium (K), calcium (Ca), magnesium (Mg), and phosphorus (P), as well as trace elements including iron (Fe), zinc (Zn), copper (Cu), and manganese (Mn). Notably, either banana or kiwifruit exhibited the highest or above average concentrations of most elements included in this study, namely: K, Zn, P, Cu, Ca, Mg and P. This implies they could be of particular value for promoting electrolyte balance, cardiovascular health, immune health, bone health and enzymatic function in humans [56,57]. Taken together with their increased content of sugars and essential amino acids in these fruits, these results reinforce their high nutritional value overall.

The second notable pattern that was derived from comparing the elemental composition across fruits showed that the berries, namely raspberry, strawberry and blueberry, showed the highest levels of iron and manganese. These elements are involved in redox regulation and antioxidant enzyme activity, which could contribute to the increased antioxidant potential of berries such as blueberry [38,58]. These mineral profiles complement the observed metabolomic signatures described in this study and further emphasize the nutritional diversity among fruit species. Overall, the combination of diverse metabolites and essential macro and microelements underscores the importance of consuming a broad range of fruits to support overall human health.

5. Conclusions

In conclusion, this study provides a broad comparative analysis of primary and several specialized metabolites across ten fruit species and three ripening stages. These findings are complemented by a resource comparing elemental composition. Our results highlight the strong influence of phylogeny, the diversity of species-specific metabolites, and the conserved trends in sugar and acid and cell wall dynamics. Notable species-specific metabolic signatures were observed in blueberry and the pome fruits exhibiting antioxidant compounds, which for the former was supported by elemental data. In addition, banana showed high disaccharides and essential amino acids and kiwi demonstrated a rerouting of carbon away from sugars, possibly through the GABA shunt. The elemental analysis further highlighted the nutritional value of these two fruits. Lastly, the Rosaceae berries accumulated organic acids in overripe fruit, which may relate to their perishability. Understanding these metabolic shifts enhances our knowledge of fruit quality, nutritional value, and postharvest physiology, and emphasizes the importance of fruit diversity in the human diet.