Phenotypic Diversity and Breeding Potential of Passiflora Germplasm Conserved Under Tropical Semi-Arid Conditions for Fruit Yield and Quality

Abstract

Passiflora germplasm represents an important genetic resource for improving fruit yield and quality in breeding programs targeting semi-arid environments. This study aimed to assess the phenotypic diversity, genetic parameters, and breeding potential of Passiflora accessions conserved in the Passion Fruit Active Germplasm Bank of Embrapa Semiárido. A total of 55 accessions, predominantly Passiflora cincinnata Mast., were evaluated using morphoagronomic descriptors related to plant, flower, and fruit traits. Quantitative data were analyzed using mixed linear models (REML/BLUP) to estimate genetic parameters, and multivariate analyses were applied to characterize phenotypic divergence. Substantial phenotypic variability was observed, particularly for fruit-related traits. Fruit weight ranged from 43.25 to 142.88 g, pulp weight ranged from 7.86 to 51.37 g, and pulp yield ranged from 17.06% to 40.27% among accessions. Broad-sense heritability estimates for key fruit traits were moderate to high, reaching 0.50 for fruit weight, 0.49 for pulp weight, and 0.36 for pulp yield, indicating favorable prospects for selection. Principal Component Analysis explained 66.0% of the total variation in the first two components, with fruit size, pulp-related traits, and seed number contributing most strongly to accession differentiation. Multivariate analyses consistently identified accessions 1 and 16 as superior for fruit weight and pulp yield, whereas accession 55 combined high fruit weight with elevated soluble solid content (up to 14.24 °Brix) but lower pulp yield. Overall, the observed variability highlights the relevance of Passiflora germplasm conserved under semi-arid conditions as a valuable resource for breeding programs focused on fruit yield, quality, and adaptation to water-limited environments.

1. Introduction

The genus Passiflora L. (Passifloraceae) comprises more than 500 species with pantropical distribution, with Brazil recognized as its main center of diversity and endemism [1,2]. Beyond its botanical relevance, Passiflora is strategically important in tropical and subtropical agriculture, especially due to the cultivation of sour passion fruit (Passiflora edulis Sims), one of the most economically important fruit crops in Brazil [3]. Passion fruit production is largely directed to the fresh fruit and juice industries, where fruit size, pulp yield, soluble solid content, and uniformity are key traits determining market value and profitability [4,5].

Despite its economic relevance, passion fruit production remains constrained by low average yields, susceptibility to biotic stresses, and limited adaptation to adverse environmental conditions, while the narrow genetic base of commercial cultivars represents a major challenge for crop improvement programs [6,7,8]. In Brazil, national yields remain well below the crop’s estimated productive potential, particularly in regions characterized by high temperatures, irregular rainfall, and prolonged drought periods [9,10]. These limitations highlight the urgent need to broaden the genetic base of cultivated passion fruit and to incorporate adaptive traits that enhance fruit production stability and quality under challenging environments.

Semi-arid regions represent one of the most critical scenarios for passion fruit cultivation, as water deficits, thermal stress, and high evaporative demand directly affect flowering, fruit set, and fruit development [11,12]. In this context, wild and native Passiflora species, especially those naturally occurring in the Brazilian Caatinga, constitute valuable genetic resources. Species such as Passiflora cincinnata Mast. exhibit remarkable adaptation to water-limited environments and harbor traits of agronomic interest, including tolerance to abiotic stresses, prolonged flowering periods, resistance to major diseases, and substantial variation in fruit-related traits [13,14]. These attributes make wild Passiflora species promising candidates for pre-breeding and for the development of cultivars better suited to semi-arid agriculture.

Reproductive biology plays a central role in fruit production efficiency in Passiflora. The predominance of allogamy, combined with floral traits such as herkogamy, androginophore length, and stigma positioning, directly influences pollination success and fruit set [15,16,17]. Recent studies have highlighted the relevance of functional herkogamy in Passiflora species, demonstrating that spatial variation between anthers and stigmas, as well as the occurrence of atypical floral morphologies (e.g., increased stigma number), can significantly affect pollination dynamics and fruit production [18]. These findings reinforce the importance of integrating floral phenotyping into breeding-oriented studies, particularly when fruit yield and quality are primary targets.

Plant genetic resources conserved in germplasm banks constitute the foundation of crop improvement programs; however, their effective utilization depends on comprehensive phenotypic characterization, which enables the identification of accessions with superior agronomic performance and breeding potential [19,20]. In Passiflora, although extensive germplasm collections have been established in Brazil, many accessions remain underexplored with respect to fruit-related traits and their genetic control, particularly under semi-arid conditions [13,21]. This knowledge gap limits the strategic use of germplasm in breeding programs aimed at improving fruit yield, quality, and environmental adaptation. In this context, the Passiflora Active Germplasm Bank maintained by Embrapa Semiárido represents a strategic collection composed predominantly of accessions originating from the Brazilian semi-arid region and conserved under field conditions for extended periods. The accessions used in this study include representatives of cultivated and wild Passiflora species, displaying wide phenotypic variation in plant architecture, floral morphology, and fruit size and quality traits. Although detailed agronomic characterization remains limited for most accessions, this germplasm provides a unique opportunity to evaluate phenotypic diversity related to plant, flower, and fruit traits and to identify accessions with superior fruit production potential and adaptation to semi-arid environments.

Plant phenotyping based on classical morphoagronomic descriptors remains a robust, accessible, and cost-effective approach for characterizing genetic diversity in germplasm collections [22,23]. When combined with genetic parameter estimation and multivariate analyses, phenotyping enables the identification of traits under strong genetic control, the structuring of phenotypic diversity, and the selection of divergent and superior accessions [24,25,26]. Such integrative approaches are particularly valuable for public breeding programs and for crops cultivated in marginal environments, where high-throughput phenotyping technologies are often unavailable.

Therefore, the objective of this study was to assess the phenotypic diversity, genetic parameters, and breeding potential of Passiflora accessions conserved in the Embrapa Semiárido germplasm bank using morphoagronomic descriptors, genetic parameter estimation, and multivariate analyses. By emphasizing fruit-related traits and reproductive characteristics, this work aims to support breeding strategies focused on improving fruit yield and quality and to contribute to the efficient and targeted use of Passiflora genetic resources in semi-arid agriculture.

2. Materials and Methods

2.1. Plant Material and Experimental Design

The study was conducted using accessions of Passiflora spp. conserved in the Active Germplasm Bank (AGB) of the Embrapa Semiárido (Petrolina, Brazil), located in Petrolina, Pernambuco State, Brazil (09°09′ S, 40°22′ W). The region is characterized by a semi-arid climate (BSh, Köppen classification), with a mean annual temperature around 26–27 °C, average minimum temperatures around 20–23 °C, and average maximum temperatures around 30–34 °C throughout the year. The mean annual rainfall is low, at approximately 430–445 mm, and relative humidity is generally moderate to low, reflecting the strong evaporative demand typical of semi-arid environments [9,11]. These climatic conditions provide context for the adaptation and performance of the evaluated accessions under field conditions. The Passion Fruit Active Germplasm Bank (AGB) comprises 74 accessions conserved under field conditions, from which 55 were selected for evaluation based on their economic relevance for fruit production and the availability of plant material during the assessment period. Fruit, plant, and flower descriptors were recorded for these accessions, as detailed in Table S1. The evaluated set comprised 51 accessions of Passiflora cincinnata Mast., two accessions of P. edulis Sims f. flavicarpa, one accession of P. setacea DC. and one accession of P. alata Curtis. The accessions originated from different areas of the Brazilian Caatinga biome and have been maintained under field conditions at Embrapa Semiárido since 2005.

The geographical origin of the accessions is shown in Figure 1, overlaid on the Caatinga biome delimitation proposed by Moro et al. [27]. The map illustrating the geographic context of the study area was generated using R software (v.4.3.1). The base shapefile was obtained from the publicly available dataset provided by [27] and deposited in Figshare. Map construction and graphical editing were performed using the sf, ggplot2, and ggspatial packages in R (v.4.3.1).

The experiment was established under a conventional cultivation system, following standard management practices for passion fruit in the semi-arid region. Plants were obtained from seeds derived from controlled pollination within each accession. Since 2005, the plants in the Active Germplasm Bank have been periodically renewed every five years using newly established seed-derived plants. Plants were grown on trellising systems and managed according to local technical recommendations, including irrigation, fertilization, pruning, and phytosanitary control [28].

The experimental design consisted of a randomized complete block design, with accessions considered as treatments. Each accession was represented by three plants per block, with three blocks, and evaluations were conducted during the productive cycle under uniform management conditions.

2.2. Phenotypic Characterization

Phenotypic characterization was performed using morphoagronomic descriptors, following standardized descriptor lists for Passiflora spp. proposed by Bioversity International and Embrapa [20,29]. Descriptors were grouped into plant, flower, and fruit categories.

2.2.1. Plant Descriptors

The evaluated descriptors and their coding are summarized in

Table 1. List of morphoagronomic descriptors used for the evaluation of Passiflora accessions.

Category	Code	Descriptor	Unit
Plant	CRA	Branch color	categorical
Plant	PAR	Presence of anthocyanin in branches	presence/absence
Plant	PHE	Presence of heterophylly	presence/absence
Plant	PES	Presence of stipules	presence/absence
Plant	CLF	Leaf blade length	cm
Plant	LMF	Maximum leaf width	cm
Plant	PRS	Sinus depth	cm
Plant	FOF	Leaf shape	categorical
Plant	BLF	Leaf blade margin	categorical
Plant	PPF	Presence of leaf pubescence	presence/absence
Plant	DLF	Leaf blade division	categorical
Plant	FBLF	Bullate shape of the leaf blade	categorical
Plant	COF	Leaf color	categorical
Plant	FAF	Leaf apex shape	categorical
Plant	PNE	Presence of nectaries on the leaf blade	presence/absence
Plant	NNE	Number of nectaries	count
Plant	PON	Position of nectaries on the leaf blade	categorical
Plant	COP	Petiole length	cm
Plant	NNP	Number of nectaries on the petiole	count
Plant	PNP	Position of nectaries on the petiole	categorical
Flower	NNS	Number of nectaries on the sepal	count
Flower	CAN	Androgynophore length	cm
Flower	DEC	Diameter of the corona extremity	cm
Flower	BFC	Banding (rings of different colors)	categorical
Flower	NAC	Number of colored rings	count
Flower	CAC	Predominant color of the filaments	categorical
Flower	CPE	Petal length	cm
Flower	CFO	Color of erect filaments	categorical
Flower	CPP	Predominant color of the perianth	categorical
Flower	PAB	Presence of anthocyanin in bracts	presence/absence
Flower	FHP	Hypanthium shape	categorical
Flower	CBR	Bract length	cm
Flower	PNB	Presence of nectaries on the bract	presence/absence
Flower	NNB	Number of nectaries on the bract	count
Flower	CSE	Sepal length	cm
Flower	LSE	Sepal width	cm
Flower	PNS	Presence of nectaries on the sepal	presence/absence
Fruit	WCA	Rind weight	g
Fruit	CACF	Rind color	categorical
Fruit	FFR	Fruit shape	categorical
Fruit	DFR	Fruit diameter	cm
Fruit	CFR	Fruit length	cm
Fruit	ECA	Rind thickness	mm
Fruit	RCL	Length-to-diameter ratio	ratio
Fruit	WFR	Fruit weight	g
Fruit	WSE	Seed weight	g
Fruit	NSF	Number of seeds per fruit	count
Fruit	WPO	Pulp weight	g
Fruit	CPO	Pulp color	categorical
Fruit	RES	Pulp yield	%
Fruit	SS	Soluble solids	°Brix
Fruit	TA	Titratable acidity	% (citric acid)
Fruit	RT	Ratio (SS/TA)	ratio

. Vegetative characterization included quantitative descriptors such as leaf blade length (CLF), maximum leaf width (LMF), petiole length (COP), as well as qualitative descriptors related to leaf shape, margin type, and degree of lobation (Figure 2).

2.2.2. Floral Descriptors

Floral characterization was carried out at anthesis and included quantitative descriptors such as sepal length (CSE), sepal width (LSE), petal length (CPE), bract length (CBR), androginophore length (CAN), and diameter of the corona extremity (DEC).

Qualitative descriptors related to corona morphology, filament coloration, and stigma number were also recorded. The complete list of floral descriptors is presented in Table 1, and representative floral morphologies are illustrated in Figure 3.

2.2.3. Fruit Descriptors

Fruit characterization focused on traits directly related to production and quality, which are critical for fresh market and industrial use. The following quantitative descriptors were evaluated: fruit weight (WFR), fruit length (CFR), fruit diameter (DFR), rind thickness (ECA), rind weight (WCA), pulp weight (WPO), number of seeds per fruit (NSF), pulp yield (RES), soluble solid content (SS, °Brix), titratable acidity (TA), and SS/TA ratio (RT) (Figure 4). Soluble solid content was determined using a digital refractometer, while titratable acidity was measured according to standardized procedures described by the Instituto Adolfo Lutz [30].

2.3. Statistical Analyses

2.3.1. Analysis of Variance and Genetic Parameters

Quantitative data were analyzed using linear mixed models to test for differences among accessions. In these models, accessions were considered as random genotypic effects, blocks were considered as random environmental effects, and variance components were estimated using the REML/BLUP (Restricted Maximum Likelihood/Best Linear Unbiased Prediction) approach [23,24].

Although the evaluated materials do not represent structured genetic populations (e.g., families or breeding populations), each accession was treated as a distinct genotype, which is a standard procedure in germplasm bank evaluations. Under this framework, variance components were estimated to quantify genetic differences among accessions evaluated under uniform field conditions.

Genotypic variance (Vg) was estimated as the variance associated with accession effects (σ²g), residual variance (Ve) as the residual error variance (σ²e), and phenotypic variance (Vf) as the sum of genotypic and residual variances (Vf = Vg + Ve). Broad-sense heritability (h²g) was calculated as the ratio between genotypic and phenotypic variance (h²g = Vg/Vf).

In addition, selection accuracy, genotypic and residual coefficients of variation, and related genetic parameters were estimated based on the predicted genotypic values.

2.3.2. Multivariate Analyses

Multivariate analyses were performed to evaluate phenotypic divergence among Passiflora accessions using quantitative morphoagronomic descriptors. All analyses were conducted using the R programming language (v.4.3.1) within the RStudio (v.4.3.1) environment. The analyses were performed using the MultivariateAnalysis package [31], along with base R functions, including dist(), for the computation of dissimilarity matrices.

To investigate genetic divergence among accessions, three multivariate approaches were applied: Principal Component Analysis (PCA), Tocher clustering, and hierarchical cluster analysis using the Unweighted Pair Group Method with Arithmetic Mean (UPGMA). Prior to analysis, all variables were standardized to zero mean and unit variance (z-score transformation) to ensure comparability among traits, according to the following expression:

$$Z={\displaystyle \frac{X-\bar{X}}{s}}$$

where X is the original value, $\bar{X}$ is the mean of the variable, and s is the standard deviation.

Principal Component Analysis (PCA)

Principal Component Analysis (PCA) was applied to identify the variables that most strongly contributed to total phenotypic variability among accessions. This multivariate technique linearly combines the original variables and reduces them to a set of principal components that cumulatively explain the largest proportion of variance in the dataset. Eigenvalues, the percentage of variance explained by each component, and variable loadings were examined to interpret the contribution of plant, flower, and fruit descriptors to accession differentiation.

Dissimilarity Matrices and Cluster Analyses

Phenotypic dissimilarity among accessions was estimated using the squared standardized Euclidean distance, computed from the standardized data using the dist() function in R. Based on the resulting dissimilarity matrix, two complementary clustering methods were applied.

In the Tocher optimization method, accessions were grouped using the combined dataset of plant, flower, and fruit descriptors, according to the criterion that the mean intra-group dissimilarity must be smaller than the mean inter-group dissimilarity, allowing the identification of groups of more similar genotypes.

Hierarchical cluster analysis was performed using the UPGMA method, generating dendrograms that represent the hierarchical structure of similarity relationships among accessions and the dissimilarity levels at which groups are separated.

Cluster analyses were conducted separately for plant, flower, and fruit descriptors, as well as in an integrated analysis considering all variables, providing both trait-specific and overall perspectives of phenotypic diversity among accessions.

3. Results

3.1. Phenotypic Variability Among Passiflora Accessions

A pronounced phenotypic variability was detected among the evaluated Passiflora accessions for vegetative, floral, and fruit-related traits. Qualitative and quantitative descriptors were used to characterize vegetative and reproductive organs (plant, flower, and fruit).

Qualitative descriptors revealed marked variation in leaf morphology, floral structures, and corona characteristics, confirming extensive intra- and interspecific diversity within the germplasm collection. The qualitative leaf trait analysis of 40 passion fruit accessions allowed their classification into eight distinct groups based on leaf shape, number of lobes, margin type, and central lobe morphology, as shown in

Table 2. Classification of Passiflora accessions based on qualitative leaf descriptors and respective codes.

Group	No. of Accessions	Leaf Shape/Lobation	Leaf Margin	Central Lobe Shape *
A	11	Pentalobed	Serrated	Unidentate
B	14	Pentalobed	Serrated	Bidentate
C	4	Pentalobed	Serrated	Serrated
D	4	Pentalobed	Entire	—
E	2	Pentalobed	Crenate	—
F	2	Trilobed	Serrated	Serrated
G	2	Trilobed or irregular	Entire	—
H	1	Entire	Entire	—

* Central lobe shape recorded as a qualitative trait associated with leaf morphology.

and Table S2 and Figure 5.

Quantitative descriptors showed wide amplitude of variation, particularly for fruit-related traits. Descriptive statistics and analysis of variance indicated significant differences (p < 0.05) among accessions for fruit weight, pulp weight, pulp yield, fruit dimensions, rind thickness, number of seeds per fruit, soluble solid content, and titratable acidity. In contrast, vegetative and floral traits generally exhibited narrower ranges of variation, suggesting a stronger contribution of fruit traits to overall phenotypic divergence.

The results of the analysis of variance for 22 of the 53 evaluated traits are presented in

Table 3. Mean values of fruit yield- and quality-related descriptors evaluated in Passiflora accessions. Mean values followed by the same letter within each column do not differ significantly according to the Scott–Knott multiple comparison test (p ≤ 0.05). The overall mean, standard deviation (SD), coefficient of variation (CV), and significance levels from the analysis of variance are summarized at the bottom of the table for each descriptor. Measures of genetic parameters are provided in Table 4.

Fruit
Accession	TA (%)	CFR (cm)	DFR (cm)	ECA (cm)	WCA (g)	WFR (g)	WPO (g)	WSE (g)	NSF -	RCL -	RES (%)	RT	SS (°Brix)
1	6.08 b	6.44 b	6.32 b	0.42 c	41.18 b	129.96 a	51.37 a	24.27 a	380.66 a	1.01 c	39.49 a	1.71 c	10.43 b
4	6.98 a	5.8 c	5.53 d	0.45 c	33.35 c	82.48 c	21.43 d	17.77 b	306.91 b	1.05 b	26.29 c	1.64 c	9.71 b
6	3.34 c	6.84 a	5.79 c	0.47 c	42.78 b	101.58 b	26.56 c	16.52 b	262.59 b	1.17 a	25.92 c	2.25 c	7.55 b
7	4.36 c	5.79 c	5.78 c	0.37 d	31.37 c	97.77 b	30.71 c	19.47 b	313.83 b	1.01 c	31.51 b	2.29 c	9.83 b
8	4.88 c	5.66 c	5.16 d	0.37 d	24.67 c	75.08 c	22.35 d	15.51 c	281.66 b	1.09 b	29.63 b	2.36 c	11.33 a
9	6.62 a	5.08 d	4.59 d	0.43 c	18.95 c	48.47 d	13.03 e	10.5 c	137.83 c	1.1 b	27.23 c	1.84 c	12.03 a
10	4.01 c	4.54 d	4.54 d	0.47 c	19.18 c	45.2 d	7.86 e	9.33 c	155.66 c	1 c	17.06 c	2.57 b	10.33 b
11	3.65 c	4.39 d	4.97 d	0.38 d	18.43 c	43.25 d	10.8 e	6.71 c	93.33 c	0.88 d	24.66 c	3.66 a	13.23 a
12	5.17 b	5.58 c	5.93 c	0.53 c	37.5 b	91.77 b	24.73 d	20.18 b	335.5 a	0.94 d	27.18 c	2.09 c	10.71 b
13	3.98 c	4.98 d	5.46 d	0.48 c	28.45 c	70.33 c	16.09 e	18.62 b	260.58 b	0.91 d	21.24 c	3.23 a	11.96 a
15	4.48 c	5.44 c	5.72 c	0.43 c	31.71 c	91.01 b	30.54 c	17.51 b	261.22 b	0.94 d	33.25 b	2.66 b	11.93 a
16	5.13 b	6.03 b	6.94 a	0.47 c	47.4 b	142.88 a	43.91 a	24.15 a	435 a	0.87 d	30.55 b	1.83 c	9.51 b
17	5.2 b	5.18 d	5.71 c	0.38 d	27.82 c	82.2 c	25.96 c	16.9 b	288.77 b	0.91 d	31.08 b	2.24 c	11.61 a
19	6.52 a	5.01 d	5.6 c	0.34 d	25.16 c	78.72 c	26.59 c	19.21 b	267.88 b	0.89 d	33.24 b	1.69 c	10.87 b
20	7.37 a	5.75 c	6.31 b	0.54 c	39.11 b	104.81 b	29.78 c	27.4 a	358 a	0.91 d	28.29 c	1.74 c	12.86 a
21	5.95 b	5.36 c	5.61 c	0.43 c	28.56 c	80.63 c	22.99 d	21.43 b	350 a	0.95 c	29.66 b	2.02 c	11.78 a
22	4.22 c	5.25 d	5.4 d	0.48 c	24.3 c	79.76 c	20.17 d	17.24 b	277.66 b	0.97 c	24.62 c	1.78 c	7.53 b
23	6.64 a	5.33 c	5.24 d	0.44 c	24.96 c	68.53 c	21.55 d	12.62 c	264.5 b	1.01 c	32.11 b	1.78 c	11.75 a
24	5.83 b	5.55 c	5.96 c	0.45 c	35.64 b	94.88 b	30.03 c	19.65 b	268.27 b	0.93 d	31.57 b	1.97 c	11.6 a
25	5.23 b	4.79 d	5.06 d	0.47 c	24.08 c	60.69 d	15.04 e	12.14 c	221.5 c	0.95 c	23.44 c	1.99 c	10.3 b
26	6.34 a	5.88 c	5.87 c	0.47 c	30.2 c	88.08 b	31.89 c	17.75 b	302.33 b	1 c	36.38 a	1.95 c	12.36 a
27	5.72 b	5.56 c	5.69 c	0.5 c	30.56 c	83.71 c	23.85 d	18.02 b	250.72 c	0.98 c	28.12 c	1.87 c	10.71 b
28	4.46 c	5.37 c	5.4 d	0.38 d	24.71 c	75.23 c	27.56 c	14.78 c	237 c	0.99 c	36.69 a	2.52 b	11.26 a
30	5.49 b	5.53 c	4.94 d	0.36 d	24.32 c	62.58 d	18.22 d	13.08 c	283.66 b	1.11 a	28.51 c	1.8 c	9.9 b
31	4.48 c	5.23 d	5.8 c	0.47 c	34.03 c	88.45 b	26.13 c	16.84 b	290 b	0.9 d	29.35 b	2.6 b	11.63 a
32	6.03 b	4.73 d	4.95 d	0.34 d	18.64 c	57.33 d	11.91 e	13.42 c	231 c	0.95 c	20.66 c	1.81 c	10.93 b
33	4.3 c	4.75 d	5.15 d	0.32 d	18.82 c	58.55 d	14.43 e	11.57 c	234.16 c	0.92 d	24.88 c	2.63 b	10.11 b
34	4.57 c	5.15 d	5.6 c	0.37 d	27.05 c	78.61 c	21.18 d	18.98 b	281.77 b	0.92 d	26.45 c	2.29 c	10.27 b
36	4.35 c	5.14 d	5.74 c	0.59 c	29.61 c	76.06 c	19.46 d	12.79 c	192.33 c	0.89 d	25.21 c	2.31 c	10.26 b
38	-	5.96 b	5.91 c	0.5 c	42.27 b	107.86 b	27.82 c	27.07 a	362.33 a	1.01 c	25.93 c	2.46 b	10.4 b
40	5.4 b	5.59 c	6.2 c	0.46 c	32.89 c	109.1 b	43.62 a	18.11 b	294.66 b	0.9 d	39.88 a	2.19 c	11.86 a
44	5.29 b	5.79 c	5.32 d	0.4 d	26.98 c	81.04 c	23.76 d	20.67 b	327.26 a	1.09 b	29.87 b	1.99 c	10.02 b
48	5.69 b	4.94 d	4.92 d	0.37 d	23.36 c	63.17 d	13.4 e	14.23 c	266.66 b	1 c	21.18 c	1.91 c	10.86 b
49	5.74 b	6.07 b	6.07 c	0.45 c	37.03 b	108.54 b	35.92 b	22.74 a	340.83 a	0.99 c	32.68 b	1.93 c	11.08 b
50	5.38 b	5.32 c	5.37 d	0.46 c	26.48 c	67.42 c	17.95 d	11.95 c	190.22 c	0.99 c	26.57 c	1.88 c	10.05 b
52	4.66 c	5.71 c	5.64 c	0.41 d	32.49 c	91.88 b	23.16 d	20.4 b	275 b	1.01 c	25.2 c	1.96 c	9.16 b
53	5.4 b	5.01 d	5.96 c	0.24 d	23.5 c	91.41 b	33.96 c	23.13 a	367.66 a	0.84 d	37.05 a	2.28 c	12.23 a
54	5.18 b	5.25 d	6.02 c	0.37 d	31.34 c	91.56 b	32.08 c	18.85 b	305.16 b	0.88 d	35.57 a	2.16 c	11 b
55	4.33 c	7.37 a	7.13 a	0.95 a	73.57 a	134.59 a	30.69 c	11.52 c	220.22 c	1.03 b	21.01 c	3.29 a	14.24 a
56	6.83 a	4.91 d	5.24 d	0.34 d	25.19 c	75.44 c	27.41 c	14.91 c	289.66 b	0.94 d	36.05 a	1.55 c	10.56 b
59	4.99 c	5.58 c	5.3 d	0.34 d	27.74 c	74.94 c	19.46 d	15.94 c	293.66 b	1.05 b	25.51 c	1.88 c	9.43 b
64	5.21 b	6.05 b	6.42 b	0.41 d	40.79 b	109.11 b	34.96 c	18.67 b	258.16 b	0.94 d	32.01 b	1.86 c	9.76 b
65	3.87 c	5.16 d	5.23 d	0.38 d	22.47 c	70.62 c	20.64 d	12.75 c	199.5 c	0.99 c	28.9 c	2.31 c	8.95 b
66	3.84 c	7.19 a	6.14 c	0.71 b	45.81 b	97.12 b	20.79 d	7.59 c	191.59 c	1.15 a	19.89 c	3.51 a	13.39 a
68	5.69 b	5.76 c	5.76 c	0.48 c	27.22 c	93.02 b	37.82 b	18.94 b	385 a	1 c	40.27 a	1.76 c	9.96 b
72	5.57 b	6.66 a	5.81 c	0.39 d	35.64 b	104.99 b	36.71 b	19.18 b	312.25 b	1.14 a	35.14 a	1.93 c	10.73 b
73	5.08 b	5.58 c	5.76 c	0.44 c	29.73 c	82.24 c	26.54 c	13 c	216.11 c	0.95 c	32.57 b	2.18 c	10.71 b
74	4.5 c	5.1 d	5.17 d	0.43 c	26.47 c	64.08 d	14.68 e	15.52 c	288.6 b	0.97 c	23.07 c	2.45 b	10.72 b
Mean	5.19	5.52	5.63	0.44	30.70	84.51	25.15	16.86	275.18	0.98	29.01	2.18	10.82
CV (%)	11.55	6.98	6.43	15.03	20.27	18.43	26.45	19.47	20.95	5.27	15.27	15.79	11.22
Significance	**	**	**	**	**	**	**	**	**	**	**	**	**

WCA = rind weight (g); WSE = seed weight (g); NSF = number of seeds per fruit; WPO = pulp weight (g); RES = pulp yield (%); DFR = fruit diameter (cm); CFR = fruit length (cm); ECA = rind thickness (cm); WFR = fruit weight (g); RCL = length-to-diameter ratio; SS = soluble solids (°Brix); RT = SS/TA ratio; TA = titratable acidity (%). ** indicate significance at p ≤ 0.01.

, Tables S3 and S4. Significant differences (p < 0.01) were detected for all analyzed traits, indicating high phenotypic variability among the accessions and highlighting their importance as a valuable source of genetic diversity for breeding programs.

The coefficient of variation (CV%) differed among traits. Pulp weight (WPO) showed the highest CV (26.45%), whereas the length-to-diameter ratio (RCL) exhibited the lowest value (5.27%), reflecting differences in the degree of phenotypic dispersion among the evaluated characters.

As shown in Table 3, Tables S3 and S4 differences were observed among the mean values obtained for the evaluated descriptors. For clarity and conciseness, mean comparisons for plant and flower descriptors are provided in the Supplementary Material (Tables S3 and S4), while fruit-related traits are presented in Table 3. The traits exhibiting the greatest variability were those related to fruit characteristics (WCA, WFR, WPO, NSF, and WSE), reinforcing their potential for selection, as they are directly associated with yield and the commercial quality of passion fruit. This variability indicates the presence of accessions with superior agronomic performance, which can be exploited in breeding programs aimed at improving fruit productivity and quality.

Marked differences were observed among accessions for fruit-related traits. Accessions 1 and 16 consistently exhibited superior performance, combining high fruit weight, pulp weight, and pulp yield with intermediate rind thickness, a trait combination considered favorable for both fresh consumption and industrial processing. In contrast, a distinct group of accessions showed significantly lower fruit weight and pulp-related traits, indicating reduced productive potential.

Accession 55 also displayed high fruit weight and stood out for fruit quality attributes, particularly elevated soluble solid content. However, this accession differed from accessions 1 and 16 by exhibiting greater rind thickness and comparatively lower pulp weight and pulp yield, reflecting a trade-off between fruit quality and processing efficiency. Despite this limitation, accession 55 represents a promising source of quality-related traits, such as sweetness and rind characteristics.

Regarding seed-related traits, accessions with superior fruit size generally exhibited higher seed numbers, reinforcing the association between seed content and fruit development. Overall, the results highlight accessions 1 and 16 as the most balanced genotypes in terms of yield and processing efficiency, whereas accession 55 represents a complementary genotype with superior quality attributes.

For floral descriptors, significant variation was mainly associated with flower size, with differences in the diameter of the corona extremity contributing most strongly to floral phenotypic diversity among accessions. Vegetative traits also varied among accessions, with leaf size representing the main source of differentiation, although these traits showed a lower contribution to agronomic performance when compared with fruit-related descriptors.

3.2. Analysis of Variance and Genetic Parameters

The analysis of variance confirmed the existence of significant genetic variability for most quantitative traits evaluated. Estimates of genetic parameters obtained using REML/BLUP (

Table 4. Estimates of genetic and phenotypic parameters for morphoagronomic traits evaluated in passion fruit accessions, obtained using the REML/BLUP method. Where Vg = genotypic variance; Vbloc = environmental variance among blocks; Ve = residual variance; Vf = phenotypic variance; h²g = broad-sense heritability; C²bloc = coefficient of determination of block effects; h²mc = adjusted heritability; Acclon = selection accuracy; CVgi (%) = genotypic coefficient of variation; CVe (%) = residual coefficient of variation; CVr = relative coefficient of variation (CVgi/CVe); Mean = overall experimental mean.

Category	Descriptor	Vg	Vbloc	Ve	Vf	h2g	C2bloc	h2mc	Acclon	CVgi (%)	CVe (%)	CVr	Mean
Plant	CLF	0.8143	0.0328	0.7397	1.5868	0.5132	0.0207	0.6877	0.8293	12.772	12.173	1.0492	7.0653
	COP	0.3309	0.0004	0.3382	0.6695	0.4943	0.0005	0.6618	0.8135	18.576	18.78	0.9892	3.0968
	LMF	2.8228	0.0039	1.6182	4.4449	0.6351	0.0009	0.7772	0.8816	16.366	12.392	1.3207	10.266
Flower	CAN	0.0983	0.0052	0.3757	0.4792	0.2051	0.0109	0.3435	0.5861	26.214	51.251	0.5115	1.196
	CBR	0.0851	0.0003	0.0686	0.154	0.5526	0.0021	0.7128	0.8443	10.387	9.3234	1.1141	2.8083
	CPE	0.1697	0.0003	0.1555	0.3255	0.5214	0.001	0.6859	0.8282	10.106	9.6717	1.0449	4.0766
	CSE	0.1241	0.002	0.1455	0.2717	0.4569	0.0073	0.6304	0.794	8.667	9.3848	0.9235	4.065
	DEC	3.6984	0.1113	54.408	58.218	0.0635	0.0019	0.1197	0.346	19.707	75.586	0.2607	9.7586
	LSE	0.016	0.0004	0.0109	0.0273	0.5861	0.0152	0.7462	0.8638	8.0256	6.6189	1.2125	1.5756
Fruit	WCA	84.085	12.195	65.035	161.31	0.5212	0.0756	0.7211	0.8492	30.878	27.156	1.1371	29.697
	WSE	15.331	0.0191	19.938	35.288	0.4345	0.0005	0.606	0.7784	23.281	26.549	0.8769	16.819
	NSF	2746	19.069	5711.9	8477	0.3239	0.0022	0.4902	0.7001	18.983	27.378	0.6934	276.05
	WPO	64.605	1.504	65.629	131.74	0.4904	0.0114	0.6632	0.8143	30.967	31.211	0.9922	25.956
	RES	21.954	3.7519	35.328	61.034	0.3597	0.0615	0.5541	0.7444	15.466	19.619	0.7883	30.295
	DFR	0.2407	0.0022	0.2032	0.4461	0.5397	0.0049	0.7033	0.8386	8.7051	7.9966	1.0886	5.6365
	CFR	0.3416	0.0014	0.2367	0.5797	0.5893	0.0023	0.7427	0.8618	10.57	8.7987	1.2013	5.5299
	ECA	0.0109	0.0017	0.0071	0.0196	0.5541	0.086	0.7549	0.8688	24.843	20.022	1.2408	0.4199
	WFR	379.06	5.2165	369.11	753.39	0.5031	0.0069	0.6726	0.8201	22.96	22.657	1.0134	84.797
	RCL	0.005	0.0073	0.0051	0.0174	0.2897	0.4182	0.6648	0.8154	6.9732	7.0023	0.9958	1.0168
	SS	0.9798	0.0022	2.6313	3.6133	0.2712	0.0006	0.4268	0.6533	9.0658	14.857	0.6102	10.918
	RT	0.1776	0.0208	0.2139	0.4123	0.4308	0.0504	0.6242	0.79	18.386	20.177	0.9112	2.292
	TA	0.6736	0.0996	0.7783	1.5515	0.4341	0.0642	0.6338	0.7961	16.612	17.857	0.9303	4.9405

CLF = leaf blade length (cm); COP = petiole length (cm); LMF = maximum leaf width (cm); CAN = androgynophore length (cm); CBR = bract length (cm); CPE = petal length (cm); CSE = sepal length (cm); DEC = diameter of the corona extremity (cm); LSE = sepal width (cm); WCA = rind weight (g); WSE = seed weight (g); NSF = number of seeds per fruit; WPO = pulp weight (g); RES = pulp yield (%); DFR = fruit diameter (cm); CFR = fruit length (cm); ECA = rind thickness (cm); WFR = fruit weight (g); RCL = length-to-diameter ratio; SS = soluble solids (°Brix); RT = SS/TA ratio; TA = titratable acidity (%).

) revealed moderate to high heritability values for key fruit-related traits, including fruit weight, pulp weight, pulp yield, fruit length, and fruit diameter.

These traits also presented high genotypic coefficients of variation and favorable accuracy values, indicating strong genetic control and high reliability of selection based on phenotypic performance. Conversely, some vegetative and floral traits showed lower heritability estimates, reflecting greater environmental influence on their expression.

The predicted genetic gains (Tables S5–S7) further demonstrated that selection based on fruit traits would result in substantial improvements, particularly for fruit weight, pulp weight, pulp yield, and number of seeds per fruit.

Fruit-related traits exhibited the highest heritability estimates (h²g and h²mc) and selection accuracy (Acclon). In particular, rind weight (WCA), fruit diameter (DFR), fruit length (CFR), rind thickness (ECA), and fruit weight (WFR) showed heritability values (h²g and h²mc) above 0.50, selection accuracy greater than 0.80, and relative coefficients of variation (CVr) above 1.0, indicating a strong genetic control of these traits.

For plant and flower descriptors, maximum leaf width (LMF), petal length (CPE), sepal length (CSE), and sepal width (LSE) also exhibited relatively high heritability and accuracy values (>0.60 and >0.80, respectively). Although to a lesser extent, these traits also contribute to the differentiation among accessions.

The analysis of genetic parameters reinforced the existence of accessions with superior performance. For fruit-related traits, accessions 1, 16, and 55 were particularly noteworthy. However, accession 55 showed the highest genetic gain for rind thickness (0.48) and rind weight (39.08), ranking first in the selection order (Table S5). In contrast, this accession did not rank among the top ten for pulp weight (WPO), highlighting an inverse relationship widely reported in the literature, in which fruits with greater rind thickness tend to exhibit lower pulp weight.

3.3. Performance of Accessions for Fruit Production and Quality Traits

Marked differences were observed among accessions regarding fruit production and quality attributes (Table 3). Accessions 1 and 16 exhibited the highest mean values for fruit mass and pulp weight, as well as superior pulp yield, indicating a favorable balance between fruit size and edible portion. These accessions also showed high numbers of seeds per fruit, which is generally associated with improved fruit filling and development.

Accession 55 also presented high fruit weight; however, it differed from accessions 1 and 16 by exhibiting greater rind thickness and comparatively lower pulp yield. Despite this limitation, accession 55 stood out for its elevated soluble solid content (°Brix), indicating superior sweetness and potential sensory quality (Table 3).

The remaining accessions displayed intermediate or lower performance for most fruit-related traits, reinforcing the distinction between high-yielding and quality-oriented genotypes within the collection.

3.4. Principal Component Analysis

Principal Component Analysis (PCA) was performed separately for fruit, plant, and flower descriptors to investigate patterns of phenotypic variation among accessions (Figure 6). The analyses revealed that fruit-related traits played a prominent role in accession differentiation, as reflected by the proportion of variance explained by the first principal components. In the PCA biplot, the directionality of the axes reflects gradients of trait contribution, with accessions distributed according to the sign and magnitude of their scores along PC1 and PC2.

For fruit descriptors (Figure 6), the first two principal components explained 66.0% of the total variation, while the first four components accounted for approximately 87.0%. The first principal component (PC1; 40.8%) was primarily associated with fruit size, weight, and seed-related traits (fruit diameter, fruit length, fruit weight, rind weight, seed mass, pulp weight, and number of seeds per fruit). Accessions positioned toward the negative direction of PC1 were characterized by higher mean values for these yield-related traits, whereas those located in the positive direction exhibited lower values, indicating reduced productive potential. Accordingly, accessions 1 and 16 clustered in the negative region of PC1, reflecting superior fruit size and pulp-related performance, while accessions 10 and 11 were positioned in the opposite region, associated with lower mean values for these traits.

The second principal component (PC2; 25.17%) was mainly influenced by pulp yield, the SS/TA ratio, and rind thickness, representing a quality-related gradient. Along this axis, accessions 55 and 66 were separated from the remaining accessions, reflecting higher rind thickness and ratio values combined with comparatively lower pulp yield. This positioning indicates a trade-off between fruit quality attributes and processing efficiency.

For floral descriptors, the first two principal components explained 78.9% of the total variation. PC1 was primarily associated with flower size-related traits, whereas PC2 was mainly influenced by sepal width. Accessions positioned along the positive direction of PC2 showed higher mean values for sepal width, while those with high positive PC1 scores were associated with reduced flower size. In this context, accessions 34 and 44 were clearly separated due to their larger floral structures, whereas accession 9 was associated with smaller flower size descriptors.

For plant descriptors, the first two principal components explained 93.52% of the total variation, with PC1 accounting for 73.15% and PC2 for 20.37%. All vegetative traits were strongly associated with PC1, which represented a gradient of leaf and petiole size. Accessions with negative PC1 scores exhibited larger leaf blade length, petiole length, and maximum leaf width, whereas accessions positioned in the positive direction showed reduced vegetative development. Accordingly, accessions 9, 73, and 55 were associated with larger leaf dimensions, while accession 33 was positioned at the opposite extreme, reflecting the lowest mean values for plant traits.

Overall, PCA clearly discriminated accessions based on fruit yield, fruit quality, floral morphology, and vegetative traits. The directionality of PC1 and PC2 facilitated the identification of accessions with contrasting phenotypic profiles, supporting the classification of genotypes with distinct agronomic and breeding potential, as illustrated in Figure 6.

3.5. Cluster Analysis and Grouping of Accessions

Cluster analysis using the Tocher method grouped the passion fruit accessions into 11 distinct clusters, confirming high genetic diversity among the evaluated germplasms. Group 4 was the largest, comprising 28 accessions, followed by Group 3 with 11 accessions (

Table 5. Clustering of Passiflora accessions obtained using Tocher’s method. Each number corresponds to an individual accession.

Group	Accession Number
Group 1	18, 51, 39, 63, 42
Group 2	5, 41
Group 3	17, 21, 31, 22, 48, 65, 36, 56, 33, 49, 40
Group 4	4, 27, 52, 59, 8, 30, 74, 50, 25, 28, 7, 54, 24, 12, 26, 72, 64, 68, 6, 73, 53, 23, 44, 15, 20, 38, 34, 13
Group 5	10, 11
Group 6	55, 66
Group 7	1
Group 8	9
Group 9	16
Group 10	19
Group 11	32

). Highly divergent accessions (1, 9, 16, 19, and 32) formed single-member clusters, while accessions 55 and 66 grouped together, forming a distinct cluster (Group 6). Low intracluster distances and higher intercluster distances (0.50–1.84) indicate substantial genetic divergence among clusters.

Cluster analyses based on fruit descriptors resulted in the formation of well-defined groups of accessions. For the fruit dendrogram (Figure 7), the formation of small subclusters was observed. Accessions 55 and 66 stood out due to their high dissimilarity, being positioned separately from the remaining accessions, which confirms interspecific differentiation as these accessions belong to P. edulis. Another highly divergent group was formed by accessions 10 and 11, as well as by accessions 1, 72, 16, and 38. In contrast, most accessions formed compact clusters, indicating close genotypic similarity.

In the dendrogram based on plant descriptors (Figure 8), a similar pattern was observed, with most accessions forming clusters with low levels of dissimilarity. Accessions 55 and 66 stood out due to their greater distance from the remaining accessions. Accession 55 clustered with accessions 9 and 73; however, accessions 9 and 73 were more closely related to each other and more similar to the majority of accessions, whereas accession 55 remained more distant.

In the dendrogram based on flower descriptors (Figure 9), high dissimilarity was observed between accession 09 and the remaining accessions. Accessions 34 and 44 were more similar to each other but showed high dissimilarity relative to most accessions. Considering a cut-off value between 0.4 and 0.6, distinct clusters were formed, indicating genetic variability among the accessions.

The cluster analysis based on the integrated set of plant, flower, and fruit descriptors resulted in the formation of two main groups (Figure 10). The first group, located on the left side of the dendrogram, comprised accessions 5, 41, 42, 63, 39, 18, and 51, which were closely related to each other but clearly separated from the remaining accessions, indicating a high level of phenotypic divergence.

The second main group was further subdivided into two distinct subgroups. One subgroup included accessions 55, 66, 32, 56, 33, 48, 31, 17, 21, 36, 22, 65, 16, 40, and 49, which showed intermediate dissimilarity among themselves and were clearly differentiated from the first main group. The other subgroup comprised the majority of the remaining accessions, which exhibited high phenotypic similarity and low dissimilarity values, indicating a more homogeneous group.

Overall, the integrated clustering highlighted contrasting phenotypic patterns among accessions and reinforced the presence of distinct groups within the germplasm collection when plant, flower, and fruit descriptors were jointly considered.

3.6. Identification of Promising Accessions for Breeding

Based on the combined evaluation of phenotypic performance, genetic parameters, predicted genetic gains, and multivariate analyses, accessions 1 and 16 were identified as the most promising materials for improving fruit yield and industrial efficiency.

These accessions combine high fruit weight, high pulp yield, and favorable fruit morphology, making them suitable candidates for direct selection or use as parents in breeding programs. Accession 55, although less efficient in terms of pulp yield due to thicker rind, represents a valuable source of quality-related traits, particularly soluble solid content, and may be strategically used in crosses with high-yielding accessions to combine productivity and fruit quality.

4. Discussion

The phenotypic variability observed among the evaluated Passiflora accessions is consistent with the broad diversity reported for the genus in Brazil and confirms the relevance of morphoagronomic descriptors for discriminating germplasm with breeding potential [32]. Previous evaluations involving P. edulis, P. cincinnata, and interspecific populations have consistently demonstrated wide variation in fruit size, pulp-related traits, soluble solids, and yield components, highlighting the importance of these traits for selection and genetic improvement [22,33]. The magnitude and structure of variability detected in the present study fall within the ranges reported in these studies, reinforcing the reliability of the observed patterns.

Fruit-related traits were the main contributors to phenotypic divergence, a result that strongly agrees with earlier multivariate analyses of Passiflora germplasm. Studies applying PCA, clustering, and distance-based methods have shown that fruit weight, fruit length and diameter, pulp yield, and rind characteristics explain most of the total variation, whereas acidity-related traits usually contribute less to overall divergence [34]. This predominance reflects not only the high discriminatory power of fruit descriptors but also their direct economic relevance, as fruit size and pulp yield are decisive attributes for fresh market acceptance and juice processing efficiency [35].

The genetic parameter estimates obtained in this study further support the feasibility of selection based on fruit traits. Moderate to high heritability values for fruit weight, pulp weight, pulp yield, and fruit dimensions agree with reports from breeding populations evaluated using mixed linear models. Studies applying REML/BLUP in sour and sweet passion fruit populations have demonstrated that these traits exhibit sufficient genetic control to allow effective selection under field conditions [36,37]. In particular, REML/BLUP has been highlighted as an efficient approach in Passiflora breeding because it improves the prediction of genotypic values and supports the more reliable ranking of genotypes in unbalanced datasets typical of perennial crop evaluations [34,35,36,37].

From a breeding standpoint, the identification of contrasting accessions represents one of the most relevant outcomes of this work. Accessions 1 and 16 showed superior performance for fruit weight and pulp yield, corresponding to an ideotype oriented toward yield efficiency and industrial processing. Similar ideotypes have been proposed in previous studies, where large fruits with high pulp proportion were identified as priority breeding targets due to their direct impact on juice yield and processing profitability [5]. In contrast, accession 55 exhibited higher soluble solid content, indicating superior sweetness and potential sensory quality. This pattern mirrors results reported in earlier evaluations, where genotypes with high soluble solids often displayed lower pulp yield due to thicker rind, revealing a trade-off between quality and processing efficiency [34].

Multivariate analyses provided a robust framework for interpreting diversity and guiding parent selection. The concordance among PCA, Tocher optimization, and UPGMA clustering indicates that the grouping patterns are biologically meaningful and suitable for breeding purposes. Similar convergence among multivariate methods has been reported in Passiflora studies, where integrated analyses based on multiple descriptors were effective in identifying divergent groups and supporting the selection of parents for recombining complementary traits [35]. Thus, the identification of accessions belonging to distinct multivariate groups provides a rational basis for crosses aimed at broadening the genetic base and maximizing genetic gains.

Although fruit traits were the primary drivers of divergence, reproductive traits may also influence yield indirectly. Studies on P. cincinnata and related species have demonstrated that floral morphology and pollination biology affect fruit set and seed number, particularly under environments where pollinator availability or environmental constraints limit reproductive success [14,18]. These findings indicate that variation in floral traits can contribute to differences in fruit performance and should be considered as complementary information in breeding programs focused on yield stability.

Overall, the results corroborate previous findings and reinforce the value of classical morphoagronomic evaluation combined with mixed-model and multivariate analyses for identifying promising genetic resources in Passiflora. The phenotypic patterns observed here are consistent with those reported in breeding and diversity studies based on the same methodological framework, supporting the conclusion that the evaluated germplasm represents a valuable genetic resource for passion fruit breeding programs [32,34].

5. Conclusions

This study demonstrates that the Passiflora germplasm conserved at Embrapa Semiárido harbors substantial phenotypic and genetic diversity with direct relevance to fruit production and quality. Accessions 1 and 16 emerge as priority materials for improving yield and industrial efficiency, whereas accession 55 represents a key source of quality-related traits, particularly soluble solid content. Together, these accessions provide valuable opportunities for strategic crosses aimed at developing high-yielding, high-quality, and passion fruit cultivars resilient to semi-arid climatic conditions.