Morphological and molecular characterization of Sri Lankan nutmeg (Myristica fragrans Houtt.)

Abstract

Background 

Nutmeg (Myristica fragrans Houtt.) is a commonly grown spice crop in Sri Lanka. The current study is the first systematic report on morphological and molecular characterization, and species distribution modeling of Sri Lankan nutmeg.

Results

The study examined 97 nutmeg trees from the main cultivating districts of Kandy, Matale and Kegalle in Sri Lanka, assessing 15 quantitative and 13 qualitative traits. A considerable variation was observed in both qualitative and quantitative traits, involving tree, leaf and fruit morphological characters. Correlation analysis revealed significant (p < 0.05) positive relationships considering quantitative (77.14%) and qualitative trait pairs (11.1%). However, none of the fruit traits (i.e. fruit dimensions, shape and pericarp) were found to be suitable predictors of economically important yield parameters i.e. dry weights of mace, nut and kernel. Though morphological and genetic variations were observed, cumulatively those led to a shallow divergence. This urges the need to expand the gene pool through introduction of new germplasms. Eight nutmeg trees were identified for their superiority based on economically important yield traits i.e. dry weights of mace, nut and kernel. Following further evaluations, these elite trees can be recommended for ex-situ conservation and as mother plants for nutmeg breeding programs. In addition to the main nutmeg cultivation districts, Kandy, Matale, and Kegalle, species distribution modeling identified some areas of Kurunegala and Nuwara Eliya administrative districts of Sri Lanka, as potential areas for cultivation of nutmeg.

Conclusion

The findings provide critical insights into the genetic diversity of nutmeg in Sri Lanka, aiding in germplasm conservation and crop improvement.

Supplementary Information

The online version contains supplementary material available at 10.1186/s12870-025-06840-z.

Background

Nutmeg (Myristica fragrans Houtt.), an important spice crop belonging to the family Myristicaceae, is native to Maluku province in Indonesia [1]. Nutmeg has spread to various tropical regions, including Sri Lanka, where it holds significant economic value. According to Sasikumar [2], nutmeg was introduced to Sri Lanka in the colonial era between the 18th and 19th century. The economically significant produce of nutmeg is its dried seeds and aril (mace), which are commonly used as a spice in various cuisines, traditional medicine, cosmetics, and in extraction of essential oil and butter for pharmaceutical and cosmetic industry [3, 4].

Sri Lanka, known for its rich spice heritage, has embraced nutmeg cultivation, and it has become an important crop contributing to the country’s agricultural economy [5]. In Sri Lanka, nutmeg trees are commonly found in the homegardens of Kandy, Matale and Kegalle administrative districts, where the annual average rainfall ranges between 1,500 and 2,500 mm and the average annual temperature between 20 and 30 ⁰C. In 2022, nutmeg was cultivated across an area of 2,936 hectares in Sri Lanka, resulting a production of approximately 2.878 million kilograms [5]. In the same year, the total export volume of nutmeg, including mace and kernel, reached approximately 2.347 million kilograms, generating an export revenue of approximately US$ 14.5 million [5]. Additionally, 63,000 kg of nutmeg oil were exported, bringing the total revenue from nutmeg exports to approximately US$ 24.2 million [5]. These figures highlight nutmeg as a valuable revenue-generating crop in Sri Lanka with significant potential for market expansion.

Nutmeg trees exhibit sexual dimorphism, occurring as dioecious (with separate male and female trees) and monoecious (having both male and female flowers in the same tree) [6]. Nutmeg is mainly propagated through generative seeds [7] from superior mother plants identified from homegardens. Despite being a high-price fetching crop commodity in local and export markets, the Sri Lankan nutmeg has undergone limited exploration, genetic characterization or improvement since its first introductions centuries ago. In order to improve nutmeg as an agricultural commodity, assessment of the existing nutmeg germplasm and identification of superior germplasms (i.e. high yield, better quality and abiotic/biotic stress tolerance/resistance) are key aspects that need timely attention.

Several previous studies have reported on morphological diversity of M. fragrans in India and Indonesia based on the tree, leaf, fruit, mace, and nut characteristics [8, 9, 11–14]. All these studies reported both qualitative (e.g. fruit shape, leaf size and tree architecture) and quantitative (e.g. fruit weight, pericarp thickness and mace dimensions) morphological characters and revealed considerable variations among the studied populations of nutmeg.

Kusuma et al. [15] developed 25 simple sequence repeat (SSR) markers for the screening of diversity in nutmeg germplasm and found it to be an effective tool for assessing the genetic diversity of nutmeg populations [15]. Kusuma et al. [15], reported detection of up to eight alleles in total in the three study populations of nutmeg from Moluccas, Indonesia (from Ambon, Banda Besar and Banda Naira islands), expressing high allelic diversity. However, a comprehensive study targeting the assessment of morphological and molecular characterization in Sri Lankan nutmeg has not been conducted previously.

Species distribution models (SDM), such as maximum entropy models are popularly used for exploring suitable habitats and to determine environmental factors based on occurrence records [16–18]. Species distribution modeling is often used to identify new priority areas for conservation [19], monitoring and managing biological invasions and endangered species [20–22], discovering new species in unknown distributional areas [23, 24] and predicting impacts of climate change [25–28]. Species distribution modeling is also important for modeling crops to monitor changes in habitats due to natural and anthropogenic activities [29, 30], enabling sustainable planning and management of crops [31]. The approach has been previously used in crops such as Basella alba, Potentila indica, Camellia sinensis, Solanum tuberosum, and Cocos nucifera [32–36]. Currently in Sri Lanka, nutmeg cultivation is mainly limited to the three administrative districts Kandy, Matale and Kegalle, with potential to expand based on SDM.

The current study reports the first systematic exploration of Sri Lankan nutmeg germplasm, using both morphological and molecular characterization, urging for implementation of conservation and crop improvement efforts. Further, based on SDM the study also identifies potential areas to promote nutmeg cultivation as an economically important cash crop in Sri Lanka.

Materials and methods

Sample collection

The study panel consisted of randomly selected 97 mature cultivated nutmeg trees sampled from 51 homegarden locations in Kandy, Kegalle and Matale administrative districts of Sri Lanka (Supplementary Material 1: Table 1) with the consent of the homeowner. The study panel included only female/monoecious nutmeg trees, aged more than 10 years and with a previous fruit-bearing record.

Morphological characterization of Sri Lankan nutmeg

Based on the descriptors presented in Vikram [8], the 97 nutmeg trees in the study panel were characterized using 31 morphological traits covering tree, leaf, fruit and flower characters (Tree: foliage density (FD), branching pattern (BP); Leaf: leaf length (LL), leaf width (LW), shape of mature leaf (SL); Fruit: fruit length (FL), fruit breadth (FB), thickness of pericarp (TP), fresh fruit weight (FFW), fresh pericarp weight (FPW), dry pericarp weight (DPW), length of nut along the surface (NL), nut circumference (NC), fresh nut weight (FNW), dry nut weight (DNW), dry kernel weight (DKW), fresh mace weight (FMW), dry mace weight (DMW), shape of fruit (SF), shape of fruit base (SFB), shape of fruit apex (SFA), number of splits per fruit (NS), nature of mace (NtM), beakness of mace (BM), attachment of mace to nut (AtM), nature of groove (NtG) and Flower: corolla length (CL), corolla diameter (CD), pedicellus length (PL), number of petals (NoP), and shape of perianth (PS); Supplementary Material 2: Table 1). The data was curated by excluding the traits which had more than 25% missing data.

For the quantitative characters, descriptive statistics were calculated and for the qualitative characters, occurrence frequencies were computed using R V4.2.2 [37] in R Studio V2023.12.1 [38]. Anderson-Darling test was carried out to test the normality of the three economically important characters DNW, DKW and DMW, at a significance of p < 0.05 in Minitab V19 [39]. Since these parameters deviated from normal distribution, Johnson transformation was performed, and the residuals were tested for normality using the Anderson-Darling test. To test the significance of the traits DMW, DNW and DKW in the individual trees of the study panel, standard one-way ANOVA at p = 0.05 was performed. Mean separation was conducted using the Tukey method with 95% confidence and the nutmeg trees which were grouped into the highest DMW, DNW and DKW categories were identified as elite nutmeg trees for recommendation for further analysis.

Principal component analysis (PCA) was conducted for both qualitative and quantitative data using the PCAmixdata package in R V4.2.2. in RStudio V2023.12.1. A hierarchical cluster analysis was performed considering both quantitative and qualitative characters using neighbor-joining method based on the Gower distance matrix in PAST V4.03 [40].

Molecular characterization of Sri Lankan nutmeg

Genomic DNA was extracted from fresh flush of nutmeg trees using a modified CTAB method where polyvinylpyrrolidone (PVP) and 2% 2-mercaptoethanol was added for efficient extraction [41]. DNA was quantified using a Nano UV spectrophotometer (Nabi, Taiwan) and was normalized to a working stock of 50 ng/µL. Fourteen SSR markers (Myr10, Myr12, Myr13, Myr18, Myr23, Myr26, Myr27, Myr28, Myr29, Myr32, Myr36, Myr42, Myr43, and Myr47) designed by Kusuma et al. [15] for nutmeg, were assayed using DNA of a sub panel of 20 randomly selected trees to select the SSR markers that could best capture the allelic diversity of the current study panel. The PCR was conducted in a C1000 thermal cycler (Bio-Rad, USA), with each reaction comprising of a 15 µL final reaction volume containing 150 ng/µL template DNA, 1× Gotaq^® Green Master Mix (Promega Corporation, Madison, Wisconsin, USA), 0.66 µM forward and reverse primers, and volumed up using nuclease-free water (Promega Corporation, Madison, Wisconsin, USA). The PCR program consisted of an initial denaturation at 95 °C for 5 min, 30 cycles of denaturation at 95 °C for 23 s, annealing at 59 °C for 1 min 30 s, elongation at 72 °C for 1 min, and a final extension step at 72 °C for 10 min, following a modification to the annealing temperature reported by Kusuma et al. [15]. The PCR products were resolved on 3% (w/v) Agarose (Sigma Aldrich, Merck group, Germany), pre-stained with 5% (v/v) ethidium bromide (Sigma Aldrich, Merck group, Germany) and was visualized using a gel documentation system (UVCI-1100, Major Science, USA). The genotype for each SSR locus were scored manually, based on the amplicon sizes approximated based on a 50 bp/100 bp DNA ladder (Promega Corporation, Madison, Wisconsin, USA). The final set of SSR markers to be assayed on the entire study panel was selected based on the highest number of observed alleles for each marker in a subpanel of 20 trees and the allele numbers reported by Kusuma et al. [15]. The selected markers were assayed on the entire study panel based on the methods described above. The marker data was curated by removing individuals carrying more than 50% missing data.

For each marker locus the number of alleles reported, expected heterozygosity (H_e), observed heterozygosity (H_o), and polymorphic information content (PIC) were calculated in Cervus V3.0.7 [42]. A principal coordinate analysis (PCoA) was conducted with the codominant genotypic scores of the six SSR markers, adopting covariance standardized PCoA method in GenALEx V6.51b2 [43]. A hierarchical cluster analysis was conducted using an unweighted neighbor-joining method in DARwin V6.0.021 [44].

Trait correlation analysis

Pearson correlation and Spearman rank correlation analyses were conducted for quantitative and qualitative data to derive Pearson correlation coefficient (PCC) and Spearman rank correlation coefficient (SRCC), respectively in R V4.2.2 [37] in RStudio V2023.12.1 [38].

Species distribution modeling

A SDM was conducted based on positional coordinates (WGS84) of 432 nutmeg trees (Supplementary Material 1: Tables 1 and 2). The modeling was conducted using 12 bioclimatic variables including, mean diurnal range (BIO2), isothermality (BIO3), temperature seasonality (BIO4), temperature annual range (BIO7), mean temperature of driest quarter (BIO09), annual precipitation (BIO12), precipitation of driest month (BIO14), precipitation seasonality (BIO15), precipitation of wettest quarter (BIO16), precipitation of driest quarter (BIO17), precipitation of warmest quarter (BIO18), precipitation of coldest quarter (BIO19), along with slope, aspect and topographic wetness index (TWI) and a digital elevation model dataset (altitude data) downloaded from the WorldClim database [45], considering the time period of year 1970 to 2000 to predict potential nutmeg cultivation areas in Sri Lanka. The spatial resolution of the acquired datasets was approximately 1 km² (30 arc-seconds). All covariates were assessed using multidimensional scaling using ENM tools package [46] in R software to determine whether any climatic covariates tend to be highly correlated. A Pearson correlation coefficient value exceeding 0.75 [46], was considered as a strong correlation between two environmental covariates layers. The heat map created with 100 permutations using the enmtools.vip function in ENMtools package [46] in R V4.2.2 [37], was used to select the most suitable environmental covariate layers to be included in analysis. The model validation was conducted based on the area under the curve (AUC) according to Liu et al. [47]. Species distribution model was built based on the maximum entropy technique performed in Maxent V3.4.1 [48], together with presence-only data and the environmental covariates. The analysis was performed using the auto feature’s function in Maxent with 10,000 background points, 5,000 maximum iterations, 25% random test percentage and 15 replicates. Further, a sub sample method was used as the replicated run type, and other parameters were set as default. Based on the model predictions highly probable areas for growing nutmeg were interpreted. Further, the averaged model was reclassified based on the 10-percentile training presence logistic threshold to represent the suitable or unsuitable habitat using ArcGIS [49].

Results

Morphological and molecular characterization

Figure 1 illustrates the sampling locations of the 97 nutmeg trees in the current study panel (Supplementary Material 1: Tables 1 and 2) representing the administrative districts Kandy (51), Matale (30) and Kegalle (16) in Sri Lanka, where nutmeg is commonly grown. They belong to five agro-ecological regions (IM3a − 21, WL1a − 03, WL2b- 02, WM2b- 33, WM3a- 04 and WM3b −34).

Fig. 1.

Sampling locations of the 97 nutmeg trees from Kandy, Matale and Kegalle districts of Sri Lanka

Out of the 31 morphological traits evaluated in the study panel, due to data deficiency, the five flower traits; CL, CD, PL, NoP and PS was eliminated from further analysis (59% missing data; Supplementary Material 2: Table 1). The NtG character state ‘presence of grooves’ was monomorphic among the study panel, with all trees bearing grooved nuts (Supplementary Material 2: Table 1), therefore; the said character was also excluded from further analysis. Accordingly, for the final analysis 15 quantitative and 10 qualitative traits covering tree, leaf and fruit characters of nutmeg were used.

Table 1 provides descriptive statistics including mean, variance, coefficient of variation and range. The Anderson-Darling coefficient of the traits DMW, DNW and DKW, did not demonstrate a normal distribution (p < 0.05). Accordingly, to satisfy the assumptions of normality for subsequent parametric analyses, the data were transformed using the Johnson method (DNW: 0.143820 + 1.54727×Asinh((X − 8.11756)/2.90132), DKW: −0.0403652 + 1.64305×Asinh((X-5.20470)/2.65091) and DMW: −3.60935 + 3.52073×Asinh((X + 0.110945)/0.815728). Following the transformation, residuals were tested for normality using the Anderson-Darling test, and no significant deviations from normality were found for these traits (p > 0.05; Anderson-Darling coefficient of 1.80, 0.96 and 0.86, respectively for DMW, DNW and DKW).

Table 1.

Descriptive statistics of leaf, fruit, nut and mace characteristics of 97 nutmeg trees sampled from Kandy, Matale and Kegalle districts of Sri Lanka

Variable	Abbreviation	Mean	Variance	Coefficient of Variation	Minimum	Maximum
Leaf
Leaf length (cm)	LL	11.97 ± 2.85	8.10	23.78	6.98	31.26
Leaf width (cm)	LW	4.89 ± 0.850	0.73	17.46	3.24	7.40
Fruit
Fruit length (cm)	FL	8.40 ± 0.79	0.62	9.36	6.30	10.17
Fruit breadth (cm)	FB	7.91 ± 0.79	0.64	10.03	5.73	9.67
Thickness of pericarp (cm)	TP	1.17 ± 0.16	0.03	14.00	0.79	1.51
Fresh fruit weight (g)	FFW	66.88 ± 17.64	311.05	26.37	30.54	124.10
Fresh pericarp weight (g)	FPW	54.18 ± 15.54	241.40	28.68	23.20	107.93
Dry pericarp weight (g)	DPW	35.75 ± 14.14	199.91	39.55	11.30	96.95
Nut
Nut lengtha (cm)	NL	4.78 ± 0.47	0.22	9.83	3.52	5.87
Nut circumferencea (cm)	NC	7.89 ± 0.64	0.41	8.12	6.26	9.22
Fresh nut weight (g)	FNW	10.17 ± 0.24	5.43	22.91	4.60	16.50
Dry nut weight (g)	DNW	7.84 ± 2.06	4.24	26.25	2.22	13.30
Dry kernel weight (g)	DKW	5.29 ± 1.72	2.96	32.55	0.65	10.20
Mace
Fresh mace weight (g)	FMW	2.29 ± 0.86	0.74	37.61	0.69	5.32
Dry mace weight (g)	DMW	0.93 ± 0.36	0.13	38.79	0.28	2.09

^aMeasurements taken along the surface of the nut as described in Supplementary Material 2: Table 1

Figure 2 illustrates the frequency distribution of character states for the ten polymorphic qualitative traits. Among the trees in the study panel, the ‘spreading’ branching pattern was the prominent character state (40.2%). The majority of trees had ‘abundant’ foliage (69.2%) and the ‘elliptic’ leaf shape was the most common (39.2%). In fruits, the ‘oval’ fruit shape was most abundant (37.1%). Majority of fruits had ‘pointed’ fruit bases (63.9%) and ‘round’ fruit apexes (72.2%). In the study panel fruits splitting into two-halves at maturity were predominant (90.7%) and the ‘intermediate’ dissections on the mace was abundant (46.4%). Further, the mace was found mostly to be ‘loosely’ attached to the nut (57.7%).

Fig. 2.

Frequency distribution of character states of the 10 polymorphic qualitative traits among the 97 nutmeg trees

According to Supplementary Material 3: Table 1, the mean separation of 97 nutmeg trees showed significant differences (p < 0.05) for the economically important yield parameters DMW, DNW and DKW. Based on the mean separation (Supplementary Material 3: Table 1), 16, 33 and 22 trees were identified, as the group carrying significantly (p < 0.05) highest means for DMW, DNW and DKW, respectively. Accordingly, eight trees were identified as best performers considering parameters DMW, DNW and DKW (Table 2). These eight-trees were presented as elite with respect to mace, nut and kernel yield and can be recommended for ex-situ conservation in orchards for the utilization in nutmeg breeding programs. Out of the eight trees NM023 was found to be carrying the highest mean DMW, DNW and DKW.

Table 2.

Mean of dry mace weight, dry nut weight and dry kernel weight of the selected eight elite nutmeg trees

Accession number	Mean dry mace weight (g)	Mean dry nut weight (g)	Mean dry kernel weight (g)
NM023	2.088	13.296	10.198
NM084	1.809	10.275	6.903
NM094	1.741	9.724	6.637
NM035	1.713	12.19	8.76
NM086	1.701	11.446	8.1377
NM046	1.495	9.612	6.758
NM013	1.45	12.38	9.91
NM085	1.361	11.61	6.47

The PCA mix model analysis conducted considering all morphological traits, detected 12 principal components (PC) explaining 71% of the total variability, where the first two components together explained 29% (Supplementary Material 4: Table 1). The PCA mix model analysis conducted considering only the fruit-related traits revealed nine PCs explaining 73% of the total observed variability, where the first two PCs explained a 37% cumulative variance (Supplementary Material 4: Table 2). Based on the Gower distance calculated considering all morphological traits (15 quantitative and 10 qualitative; Fig. 3a) and only considering fruit-related traits (13 quantitative and 7 qualitative; Fig. 3b), the 97 studied nutmeg trees were clustered, reflecting shallow morphological divergences among the sampled trees. Interestingly, out of the eight elite trees identified, all except NM013 were clustered much closer to each other (Fig. 3a-b). This indicates that out of the eight elite trees, NM013 carry more morphological differences compared to the rest.

Fig. 3.

Neighbour-joining dendrogram based on the Gower distance matrix for a all morphological traits b fruit-related traits among the 97 nutmeg trees. The nodes of the eight nutmeg trees identified as carrying the highest dry nut weight, dry kernel weight and dry mace weight are coloured in red

After curating for the missing data, 92 individual nutmeg trees were considered for SSR analysis. Based on the allele numbers reported in Kusuma et al. [15] and the observed allele numbers in the subpanel of nutmeg trees in the current study (Supplementary Material 5: Table 1), six markers namely, Myr13, Myr23, Myr26, Myr28, Myr29 and Myr43 were selected to be assayed on the entire study panel. According to Table 3, the number of alleles reported per locus in the current study panel ranged from five in marker Myr13 to seven in marker Myr28. The H_o, H_e and PIC in the study panel varied from 0.261 - 0.446, 0.729–0.800, and 0.681–0.763, respectively. According to Table 3, the allele frequency range was broadest in marker Myr28 (0.0163–0.4130) and narrowest in marker Myr43 (0.0326–0.2391).

Table 3.

Number of alleles, polymorphic information content (PIC), range of allele frequencies, expected heterozygosity (H_e), and observed heterozygosity (H_o) of selected six SSR markers on 92 nutmeg trees

Marker	Number of alleles	PIC	Range of allele frequencies	He	Ho
Myr43	6	0.763	0.0326–0.2391	0.8	0.359
Myr29	6	0.746	0.0435–0.3641	0.781	0.446
Myr23	6	0.713	0.0056–0.3167	0.758	0.433
Myr28	7	0.693	0.0163–0.4130	0.735	0.261
Myr26	6	0.688	0.0163–0.4076	0.733	0.435
Myr13	5	0.681	0.0326–0.4022	0.729	0.315

The PCoA analysis reported variation in the first two coordinates is presented in Fig. 4a. Based on the six markers assessed, the first two components explain up to 21.64% genetic variation of the study panel and cumulatively the first three components explain up to 30.49% genetic variation (Supplementary Material 5: Table 2). Further, when the trees are tagged according to the administrative district from where the samples were collected (Kandy, Matale or Kegalle), no clear distinction was observed owing to the sample collection locality (Fig. 4a). Though allelic variations existed among the studied 92 nutmeg trees at the six loci, cumulatively they led to a shallow genetic divergence, similar to that observed with morphological traits.

Fig. 4.

a Principal coordinate analysis and b dendrogram illustrating hierarchical clustering based on six SSR markers to analyze genetic similarities among the 92 nutmeg trees

Trait correlation analysis

Pearson correlation analysis ascertained significant (p < 0.05) positive correlations between 77.14% of quantitative trait-pairs (indicated in bold lettering in Fig. 5a) and no negative correlations were reported. The highest significant (p < 0.05) positive correlation was reported between the trait-pairs, FPW and FFW (PCC: 0.98) and was followed by FPW and FB (PCC: 0.93), FFW and FB (PCC: 0.93), DNW and DKW (PCC: 0.93), and FFW and FL (PCC: 0.92). Spearman rank correlation analysis revealed both positive and negative correlations for qualitative trait-pairs. Of which only 11.1% qualitative trait-pairs were found to be significant (p < 0.05). Based on the SRCC, the highest significant positive (p < 0.05) correlation was observed between the trait-pairs SF and SFB (SRCC: 0.47), followed by NS and SFA (SRCC: 0.40), BM and SFA (SRCC: 0.29), and SF and SL (SRCC: 0.28). Further, a significant (p < 0.05) negative correlation was only observed between the trait-pairs BP and AtM (SRCC: − 0.27).

Fig. 5.

a Pearson correlation coefficients among the 15 quantitative characters, and b Spearman rank correlation coefficients among the 10 qualitative characters. Coefficients denoted in bold letters are significant (p < 0.05)

Species distribution modeling

Different climatic variables were more or less important in modeling the distribution of nutmeg in Sri Lanka, as seen with the contribution to the model with 15 variables (Supplementary File 6: Table 1). Based on the values, the precipitation of driest month (BIO14: 48.1%) contributed most to the model. The annual range of temperature (BIO7: 25%), mean temperature of driest quarter (BIO09: 12.1%), and temperature seasonality (BIO4: 6.1%) contributed the highest to the distribution of nutmeg. These four variables together contributed 91.3% to the model’s prediction power. Accordingly, precipitation of the driest month and the temperature annual range are the most important variables for the prediction model.

Based on an area under the curve (AUC) value of 0.986, the species distribution model performed excellently in evaluating the potential distribution of M. fragrans in Sri Lanka. The map in Fig. 6a indicates the prediction probability, with red and green colours presenting high to low suitability conditions, respectively. Accordingly, highland areas of the country present potential for cultivation, where five administrative districts: Kandy, Kegalle, Matale, and parts of Kurunegala and Nuwara Eliya in Sri Lanka carry the best suited conditions to cultivate nutmeg. All together these areas represent 15 agro-ecological regions, namely, IU2, IL1a, IM1a, IM3a, IM3b, IM3c, IM1c, WL1a,WL2b, WM1a, WU2b, WM3a, WM2a, WM2b and WM3b. The rest of the administrative districts are not suitable for cultivating nutmeg. Based on the 10-percentile training presence logistic threshold (Fig. 6b), which represents the potential habitats for M. fragrans cultivation, habitats including part of Kandy, Matale, and Kegalle districts and a small proportion of Kurunegala and Nuwara Eliya districts are identified as most suitable for cultivation.

Fig. 6.

a Predicted probability of occurrence of M. fragrans in Sri Lanka, and b potentially suitable habitats for M. fragrans in Sri Lanka

Discussion

Morphological and molecular characterization

Assessing variation in germplasm is an initial step towards conservation and development of a breeding program aimed at improving nutmeg as a spice crop. The current study is the first systematic investigation of M. fragrans diversity in Sri Lanka using both morphological and molecular characterization, assessing the base population for their morphological and genetic variations. As shown in the descriptive statistics of quantitative traits (Table 1; Supplementary Material 3: Table 1) and the frequency distribution of qualitative trait character states (Fig. 2), the current study reveals variation among Sri Lankan nutmeg germplasms with respect to tree, leaf, and fruit morphology. Given nutmeg’s status as a cultivated crop, some of these morphological characters hold significant agronomic importance, particularly for improving the crop yield and quality through breeding programs.

The mace, nut and kernel are the economically important plant parts of nutmeg and the market price is primarily determined by their dry weights. Accordingly, when identifying elite germplasms for conservation and breeding, the major focus is on DMW, DNW and DKW, as economically important yield parameters. In the current study, significant variations in DMW, DNW and DKW were observed among the nutmeg trees (Supplementary Material 3: Table 1). Based on these parameters, eight nutmeg trees (NM023, NM013, NM035, NM085, NM086, NM084, NM094 and NM046; Table 2) were selected as elite trees for further characterization and ex-situ conservation. Further, genetic analysis conducted using six SSR markers confirmed that these eight nutmeg trees are not genetically identical (Fig. 4b). Vikarm et al. [10], reported that weight of mace, kernel and nut, and 10 other traits (plant height, canopy spread (E-W), canopy spread (N-S), number of flowers per 10 cm², fruit set percentage, number of fruits per m², fruit weight, thickness of pericarp, ratio of nut to mace, and number of fruits per tree) can be used for characterizing of elite nutmeg trees. Accordingly, it is recommended to further characterize the eight nutmeg trees identified in the current study using the additional parameters described in Vikarm et al. [10]. Given that no released nutmeg varieties are currently existing in Sri Lanka, the findings of the current study will provide the knowledge and pave the way to establish a breeding program.

Apart from the weights of mace, nut and kernel, another important fruit character in nutmeg is AtM, where loose attachment of mace to the nut is preferred, as it facilitates easy detachment during processing compared to the tightly attached compact type. In the present study, majority of the nutmeg trees examined carried loosely attached mace, and six out of the eight identified elite trees also had loose attachment, adding further value to their selection.

Further, a key plant architectural trait that influences land use is the branching pattern. Among the branching patterns observed (Fig. 2), the ‘spreading type’ was the most prevalent in the study panel. From an agronomic perspective, spreading type branching present challenges for space utilization and intercropping in farmlands. This branching habit can lead to inefficient land use and excessive shading, which limits the use of nutmeg in intercropping systems; a cropping method which is often used to diversify production and enhance farm profitability. Therefore, improving plant architecture through breeding efforts to develop more compact and upright varieties could contribute to more sustainable agricultural practices.

Based on the qualitative and quantitative characters assessed in the current study, Sri Lankan nutmeg expresses morphological characteristics that are somewhat comparable to germplasm from India and Indonesia [8, 9, 11–14]. As an example, when comparing the Sri Lankan nutmeg with germplasms from other countries, the mean of economically important yield parameters DMW, DNW and DKW reported in the current study (Table 1), is comparable to those reported by Vikram et al. [9] and Kumar et al. [13] based on studies conducted in Kerala and Karnataka, India, respectively. However, the ranges reported in the current study (FMW: 0.86–2.29 g and DMW: 0.36–0.93 g) show considerable deviation from those reported by Priyanka et al. [11] based on nutmeg germplasm in Kerala, India.

The cumulative effect of genetic variations over extended periods drives the evolutionary process of divergence, leading to the formation of genetically distinct populations [50]. Though morphological diversity reflects the underlying genetic variations in populations, the morphological variation observed among the studied 97 nutmeg trees (Table 1; Fig. 2), showed a shallow morphological divergence (Fig. 3a). Shinde et al. [51] reported that fruit traits in nutmeg exhibit high heritability and genetic gain, indicating a stronger potential for capturing the genetic variability through these traits. However, when assessed using only the fruit-related traits of nutmeg (Fig. 3b), the resulting morphological divergence remained the same.

Concurrently, hierarchical clustering attempted considering the six selected SSR marker loci (Myr13, Myr23, Myr26, Myr28, and Myr29) revealed genetic variation. For these selected SSR markers, the current study reported PIC ranging from 0.684 - 0.767, capturing 5–7 alleles per marker. This indicates that the selected six markers were highly informative and have the potential to distinguish genetic differences among individuals and can be identified as suitable for investigating the genetic variability in nutmeg. Compared to Kusuma et al. [15], the detected number of alleles were in the same range for markers Myr23, Myr28 and Myr29, higher for markers Myr26 and Myr13, and slightly lower for the marker Myr43. Based on the PCoA (Fig. 4a), these six SSR markers were able to capture the genotypic differences between the trees in the study panel to a moderate to low level. Further, based on the H_o (proportion of individuals that are heterozygous at a given locus) and H_e (probability of an individual being heterozygous at a locus) reported from the SSR marker analysis (Table 3), it can be inferred that the study population is not in Hardy-Weinberg equilibrium. Deviation from the Hardy-Weinberg equilibrium is often seen in populations that undergo inbreeding and/or artificial selection, as these practices are known to reduce genetic diversity and increase homozygosity within populations, such as in agricultural systems [52–54].

Similar to the findings of the morphological traits, the genetic variations captured by the six SSR markers also supported a shallow genetic divergence among the 92 trees analysed (Fig. 4b). This indicates that though variations exist at morphological and molecular level, they are not notably significant to result in a divergence to form genetically distinct populations. Nutmeg was introduced to Sri Lanka in the 18th − 19th century [2] and thereafter no considerable germplasm introductions were made in Sri Lanka to enrich the existing genetic stock. The observed shallow divergence could be due to the limited time span since introduction of nutmeg to Sri Lanka and the limited genetic diversity in the material used at the introductions. Accordingly, the need for genetic improvement of Sri Lankan nutmeg, through a national breeding program, with exotic germplasm introductions to enrich the diversity of the base population is highlighted in the current study.

Trait correlations and species distribution modeling

The economically important yield components (mace, nut and kernel) are located inside the fruit and are covered by a fleshy pericarp. Therefore, through correlation analysis, it was attempted to determine if any of the fruit parameters can be directly used to predict the important yield parameters DMW, DNW and DKW. The FFW shows a significant positive correlation only with DNW, with a moderate PCC of 0.60 (Fig. 3). Given the reported correlation is moderate, it is not strong enough to support the effective use of FFW to predict DNW. For the qualitative traits, only few significant correlations were observed (Fig. 3), and even those were associated with low SPCC. Susi et al. [14] previously reported a moderate correlation between the fruit weight and the nut weight (r = 0.531), which is somewhat comparable to the PCC reported in the current study (FFW and FNW r = 0.49 and FFW and DNW r = 0.6). Accordingly, based on the present analysis, no external fruit trait can be recommended as a reliable predictor of the economically important yield parameters DMW, DNW and DKW.

Urbanization, land subdivision and climate change are likely to pose significant challenges to the cultivation of nutmeg in the future. The current study is the first attempt of SDM to identify potential cultivation areas for nutmeg in Sri Lanka. The integration of bioclimatic variables and topographic data identified specific regions, primarily in Kandy, Kegalle, and Matale districts, as highly suitable for nutmeg cultivation. These findings align with the existing cultivation zones and also point towards potential expansion areas particularly in parts of the Kurunegala and Nuwara Eliya administrative districts of Sri Lanka (Fig. 6). Among the key environmental factors influencing nutmeg cultivation, precipitation during the driest month emerged as a critical driver (Supplementary Material 6: Table 1). Other notable factors include the annual temperature range and the mean temperature of the driest quarter. The delineation of suitable areas for nutmeg cultivation using SDM offers guidance for agricultural practices, including potential expansion zones and areas optimal for new plantations. This information could assist policymakers and farmers in making informed decisions regarding nutmeg cultivation, potentially leading to increased production and economic returns.

The current study provides valuable insights into the genetic diversity and ecological niche of Sri Lankan nutmeg; while highlighting limitations that need urgent attention. It is recommended to define standardized measurement protocols to enhance comparability in characterizations conducted across study populations and regions. Further, it is recommended to improve nutmeg genetic resources to facilitate development of molecular tools to be effectively used in improvement of nutmeg in crop breeding programs.

Conclusions

The current study reports shallow morphological and genetic divergence in Sri Lankan M. fragrans, despite the presence of notable variations observed at both morphological and molecular levels. This urges the necessity of genetic introductions, conservation and establishment of a national breeding program for nutmeg in Sri Lanka. Species distribution modeling identified potential new areas in Kurunagala and Nuwara Eliya administrative districts of Sri Lanka in addition to the Kandy, Matale and Kegalle districts where nutmeg is extensively cultivated. Eight nutmeg trees superior in terms of economically important yield parameters were identified for further investigation, to be recommended as elite trees for ex-situ conservation and for the use in future nutmeg breeding efforts in Sri Lanka as a spice crop.

Authors’ contributions

D.J., A.P., and T.K. were involved in conceptualization and methodology. D.J., A.P., T.K., and H.D. acquired funding, D.J., A.P., L.S., and S.K. carried out supervision. D.J. and L.A. did project administration and finances, L.A., T.K., S.P., I.M., D.J., A.P., H.D., and T.W. conducted data collection, A.P. conducted formal taxonomic identification of trees, L.A., T.K., S.P., D.J., A.P., I.M., and S.K. conducted investigation and analysis. T.K., and L.S., conducted statistical analysis. L.A and D.J. wrote the manuscript, L.A., T.K., and I.M. prepared the figures. L.A., D.J., T.K., I.M., L.S, S.K, and A.P edited the manuscript.

Funding

Research was conducted with funding received by University of Peradeniya, Sri Lanka (Multidisciplinary research grant 263).

Data availability

Data is provided within the manuscript or supplementary information files and any additional data can be provided upon request.

Declarations

Ethics approval and consent to participate

Not applicable.

Consent for publication

Not applicable.

Competing interests

The authors declare no competing interests.