Genotypic and Environmental Influence on Fresh Rhizome Yield of Turmeric (Curcuma longa L.)

Abstract

Turmeric (Curcuma longa) and related Curcuma species have been used traditionally in India, China, Hawaii, and other cultures for millennia. Today they are used around the world for spice, medicine, dye, and religious purposes. Recent biomedical studies have corroborated the long-known traditional medicinal values of turmeric and its constituent curcuminoid compounds, which have anti-inflammatory, antioxidant, and anticarcinogenic properties. As part of statewide research and extension efforts to support an expanding turmeric industry, we examined yield of 14 accessions across different climatic zones in Hawaii to observe and describe Genotype × Environmental influences. Fresh turmeric yield differed significantly among genotypes. The overall yields observed in this work ranged 11.3–57.22 t ha⁻¹ and generally agree with those in the literature. Data from the different sites suggest that fertility and water management are able to mitigate moderate stress imposed by climate change within a certain range, but suboptimal temperatures associated with high elevation in the tropics (>1000 m) are an important driver of lower yields. This suggests that high yielding turmeric varieties may possess wide adaptability and may perform well across diverse environments. However, site-specific evaluations will still be necessary, particularly in environments outside turmeric’s environmental optima and in the presence of high pest pressure.

1. Introduction

Turmeric (Curcuma longa L.) is the most commercially important species of its genus and one of the highly valuable crops belonging to the Zingiberaceae [1]. An herbaceous perennial, turmeric is primarily cultivated as an annual for its rhizomes. Turmeric and related Curcuma species have been used traditionally in India, China, other parts of Asia and the Pacific for millennia [2]. Today they continue to be used around the world for spice, medicine, dye, and religious purposes. Recent biomedical studies have investigated and, in some cases, corroborated what Ayurvedic and Chinese medical traditions have long purported; that turmeric and its constituent compounds demonstrate anti-inflammatory, antioxidant, anticarcinogenic and other properties important to human health [3,4]. Demand for turmeric has exploded in the last decade, presumably as a result of interest in the human health potential of this crop. For example, imports into the United States have increased over 5-fold between 2012 and 2021, from 11.53 million to 62.74 million [5] Although India has been and remains the largest producer and exporter of turmeric, there has been a push to expand turmeric production well beyond South and Southeast Asia. In the southern continental U.S., it has been shown that genotypic and environmental factors can interact to affect yield [6,7,8]. Agroecological approaches to turmeric production in these areas is especially important given that turmeric is a tropical crop and would need to be intensely managed to mitigate both supra- and sub-optimal temperatures and daylengths its normal range of production [9].

Despite the promising experimental success of shading, daylength modification and other environmental manipulation for the improved production of turmeric [6,7,8,9], the component of the agroecosystem that producers have perhaps the most control over is the selection of genotype. Of course, this assumes that a diverse pool of germplasm is available. Turmeric is a polyploid that is propagated by so-called “mother rhizomes” or “mothers” which are the swollen base of vegetative shoots from which primary rhizomes or “fingers” initiate and extend [2]. Although these “mothers” are invariably referred to as rhizomes, their swollen stems with dominant vertical vegetative growth are in fact more correctly called corms [10]. Tremendous genetic variability exists in turmeric largely as a result of clonal selection, although mutation breeding and conventional sexual recombination has been reported. As a result of these improvement efforts, largely occurring in India, there are at least dozens of improved genotypes, and hundreds of unique accessions [11]. In exploring the stability of turmeric yield components such as the fresh weight of fingers and mothers, multiple studies have demonstrated high heritability, suggesting a larger genotypic contribution and a relatively smaller contribution of the environment to phenotypic expression of these traits [12,13]. However, almost all of the work done to evaluate the performance of field-grown turmeric germplasm under multiple environments has been in India.

Our research objective for this work was to evaluate multiple turmeric genotypes for fresh finger and mother yield under contrasting environments across multiple locations and years in Hawai’i, USA.

2. Materials and Methods

A total of 14 accessions assembled from local collections were evaluated over 4 years and 4 locations for a total of 7 independent trials (called iterations from here on). Four cultivars were evaluated in all 7 iterations: ‘Hawaiian Red’, ‘BKK’, ‘Joy’ and ‘Olena’. ‘Hawaiian Red’ is a widely planted cultivar in Hawaii and was entered as the “check” or control cultivar and considered the industry standard. ‘BKK’ is reputed to be a high curcuminoid variety, and as such is in high demand. Both ‘Hawaiian Red’ and ‘BKK’ were originally obtained from Hawaii Clean Seed LLC (Pahoa, HI, USA). ‘Olena’ was obtained from a traditional Hawaiian herbal medicine practitioner (‘Olena is the Hawaiian word for turmeric). ‘Joy’ was obtained from an Asian market in Georgia. ‘Roma’ is an improved commercial variety originally from India and was included in all iterations except Maui Ag Park 2019 (not enough planting material was available at the time). Two other accessions, ‘Mystic’ and ‘Carib,’ were deemed similar to ‘Joy’ and were dropped after the first two trials. Additional accessions were included as they were made available. These were: 18-007; 18-008; 18-012; 18-013; 18-014; 18-023; ‘Wailua Gold.’ The numbered accessions were collected without names and ‘Wailua Gold’ is a local selection from Kauai and was included only at that location.

All trials were arranged in a RCB design with 4 replications. Seed pieces were sown in beds with 30 cm spacing within rows and 150 cm between bed centers. Border rows were planted on both sides of the field and border plants at the end of beds. Beds were drip irrigated to supplement rainfall and replace evapotranspiration. In each iteration, all genotypes were harvested on the same date, after senescence had progressed significantly (at least 50%) in all entries. 5 plants were harvested from each replication. Mother corms were separated from fingers, and all fresh marketable product was washed, dried, and weighed immediately. Site specific management practices reflect standard production practices at the location and are described below.

2.1. Waimanalo

Turmeric germplasm evaluations were conducted over three seasons (2018–2019, 2019–2020 and 2020–2021) at Waimanalo Research Station, (21°19′57″ N; 157°42′49″ W) located at 23 m elevation with a mean annual temperature of 22 °C and a 20-year mean annual rainfall of 1397 mm. The soil was a Waialua series (very fine, mixed, superactive, isohyperthermic Pachic Haplustolls). All trials in Waimanalo were managed in compliance with the USDA National Organic Program and plots were certified organic by a third-party certifier (Where Food Comes From, Inc., Castle Rock CO., USA). Fertilizer was incorporated to a depth of 15 cm preplant to supply 200 kg N ha⁻¹ as Sustane^® 8-2-4 fertilizer. Planting dates were 21 June 2018, 25 July 2019, and 12 July 2020. Beds were mulched with approximately 60 cm of wood chips to prevent weeds and incidental weeds were removed periodically as necessary. Harvest dates were 5 April 2019, 12 March 2020, 17 March 2021.

2.2. Kaua‘i

Turmeric germplasm evaluations were conducted over two seasons (2019–2020 and 2021–2022) at the Kaua‘i Agricultural Research and Extension Station (KARES), in Wailua, HI, USA (21°03′57″ N; 159°23′50″ W) located at 160 m elevation with a mean annual temperature of 22 °C. The total rainfall over the two growing seasons was 2102 mm (2019–2020) and 1424 mm (2021–2022). The soil was a Hāli‘i series (fine, ferruginous, isothermic, Anionic Acroperox). Both trials on Kaua‘i were managed in compliance with the USDA National Organic Program but were not certified. Fertilizer was incorporated at a depth of approximately 15 cm at planting to supply 200 kg N ha⁻¹ as Sustane^® 8-2-4 fertilizer in the 2019–2020 trial and 170 kg N ha⁻¹ in 2020–2021 as Sustane^® 4-6-4 fertilizer. Planting dates were 2020 June 2020 and 25 June 2021. Weeds were managed using weed mat between rows and hand weeding twice during the growing season for both trials. Harvest dates were 25 February 2020 and 24 February 2022.

2.3. Maui

2.3.1. Kula Agricultural Park

A turmeric germplasm evaluation was conducted over one season (2018–2019) at the Kula Agricultural Park (KAP) in Kula, HI, USA (20°47′46″ N; 156°21′35″ W) located at 428 m elevation with a mean temperature of 22 °C and 478 mm total rainfall over the growing season. The soil was a Keahua series (fine, kaolinitic, isohyperthermic Ustic Haplocambids). 220 kg N ha⁻¹ was supplied by monthly fertigation until 1 November 2018 as CaNO₃ (15.5-0-0), KNO₃ (13.75-0-46) and Urea (46-0-0). Planting date was 1 May 2018. The plot was covered by 4.6 m wide weed mat and cut for planting. Weeds were removed by hand as necessary. Harvest date was 25 February 2019.

2.3.2. Maui Agricultural Research Center

A second turmeric germplasm evaluation was conducted on the island of Maui over one season (2019–2020) at the Maui Agricultural Research Center (MARC) in Kula, HI, USA (20°45′22″ N; 156°19′07″ W) located at 990 m elevation with a mean temperature of 18 °C and 660 mm total rainfall over the growing season. The soil was a Kula series (medial, amorphic, isothermic Humic Haplustands). Fertilizer was incorporated to a depth of 15 cm preplant and by monthly fertigation until 1 November 2019 to supply 220 kg N ha⁻¹ as CaNO₃ (15.5-0-0) and 20-10-20. Planting date was 18 July 2019. The plot was covered by 4.6 m wide weed mat and cut for planting. Weeds were removed by hand as necessary. Harvest date was 16 June 2020.

2.4. Statistical Analysis

Data analysis for each iteration was performed as a RCBD. For the 5 cultivars commonly planted in six iterations, data were analyzed as a split plot with iteration as the main plot and genotype as the split plot. ANOVA was conducted using Statistix statistical package. When treatment effects were significant, means were separated by Least Significant Difference (LSD). Statistical significance (α) was set a priori at 5% (0.05).

Yield data across all genotypes and environments was analyzed using an R package, PPBstats 0.26 [14], which is based on a Bayesian model: Pr(θ|y) ∝ Pr(θ)Pr(y|θ), with Pr(θ|y) the posterior, Pr(y|θ) the likelihood and Pr(θ) the prior. The distribution of the parameter of interest is proportional to the distribution of the prior × the information brought by the data. For mean comparison across all genotypes and environments, the residual variance of a trial is estimated using all information available from total data (including all trials) rather than using data from that trial only. Therefore, the mean of each genotype is compared to the mean of each other genotype. H0: µij ≥ µi’j H1: µij < µi’j if either H0 or H1 had a high posterior probability. If Pr{H0|y} > 1 − α or Pr{H1|y} > 1 − α, where α was some specified threshold, the difference (µij − µi’j) between the means of germplasm i and population i’ in environment j was considered as significant. The difference was considered as not significant otherwise.

3. Results

Significant differences in fingers and mother corms were found among genotypes in each of the 7 iterations (

Table 1. Rhizome (finger) and corm (mother) weight of turmeric cultivars grown at four locations over 4 years in Hawai’i. Letters within columns of each year/location combination indicate Fisher protected LSD means separation, where means having the same letter are not significantly different from each other. Estimated yield is based on a plant population of 22,222 plant per hectare based on spacing of 30 cm within rows and 150 cm between rows, including access pathways.

Harvest Year	Location	Cultivar	Fingers	Mothers	Estimated Finger Yield
			kg plant−1		t ha−1
2019	Maui Ag. Park	BKK	1.03 b	0.30 b	22.89 b
		Hawaiian Red	1.65 a	0.47 a	36.59 a
		Joy	1.58 a	0.53 a	35.18 a
		Mystic	1.50 a	0.50 a	33.33 a
		Olena	1.37 ab	0.30 b	30.37 ab
2019	Waimānalo	BKK	1.10 b	0.11 c	24.36 b
		Caribbean	1.15 b	0.27 b	25.64 b
		Hawaiian Red	1.41 b	0.28 b	31.35 b
		Joy	1.23 b	0.33 b	27.41 b
		Mystic	1.22 b	0.35 ab	27.17 b
		Olena	2.19 a	0.33 b	48.77 a
		Roma	1.99 a	0.46 a	44.23 a
2020	Kaua‘i	BKK	0.52 d	0.18 d	11.61 d
		Hawaiian Red	1.12 ab	0.31 bc	24.78 ab
		Joy	0.93 bc	0.39 ab	20.56 bc
		Olena	1.25 a	0.22 cd	27.67 a
		Roma	1.26 a	0.42 a	27.89 a
		Wailua Gold	0.70 cd	0.17 d	15.50 cd
2020	MARC	BKK	0.16 b	0.07 c	3.64 b
		Hawaiian Red	0.26 ab	0.13 b	5.69 ab
		Joy	0.33 a	0.17 a	7.29 a
		Olena	0.33 a	0.08 c	7.35 a
		Roma	0.23 b	0.05 c	5.20 b
2020	Waimānalo	18-007	1.13 bcd	0.27 ab	25.11 bcd
		18-008	1.19 bc	0.23 abc	26.53 bc
		18-012	0.83 cd	0.21 bc	18.44 cd
		18-013	1.32 b	0.21 bc	29.33 b
		18-014	1.12 bcd	0.20 c	24.80 bcd
		BKK	0.80 d	0.19 c	17.87 d
		Hawaiian Red	0.97 bcd	0.28 a	21.64 bcd
		Joy	1.29 b	0.22 abc	28.58 b
		Olena	1.10 bcd	0.27 a	24.40 bcd
		Roma	1.87 a	0.24 abc	41.51 a
2021	Waimānalo	18-007	1.48 bcd	0.24 ab	32.89 bcd
		18-008	1.20 d	0.22 bc	26.61 d
		18-013	1.50 bcd	0.27 a	33.33 bcd
		18-014	1.98 b	0.20 bcd	43.89 b
		18-023	1.36 cd	0.18 cd	30.11 cd
		BKK	0.92 d	0.16 de	20.39 d
		Hawaiian Red	1.14 d	0.18 cd	25.33 d
		Joy	1.10 d	0.20 bcd	24.39 d
		Olena	1.80 bc	0.13 e	40.06 bc
		Roma	2.58 a	0.22 abc	57.22 a
2022	Kaua‘i	18-013	1.48 a	0.36 a	32.89 a
		18-023	1.26 ab	0.28 ab	27.89 ab
		BKK	0.51 d	0.09 e	11.39 d
		Hawaiian Red	1.07 bc	0.17 cde	23.72 bc
		Joy	1.02 bc	0.20 bcd	22.56 bc
		Olena	0.97 c	0.13 de	21.50 c
		Roma	1.34 a	0.26b c	29.72 a
		Wailua Gold	0.97 d	0.11 e	21.50 d
Iteration (I)			p < 0.001	p < 0.001	p < 0.001
Genotype (G)			p < 0.001	p < 0.001	p < 0.001
I × G			p < 0.001	p < 0.001	p < 0.001

). Yield of marketable fresh fingers ranged 1.03–1.65 kg plant⁻¹ in Maui 2019; 1.10–2.19 kg plant⁻¹ in Waimanalo 2019; 0.52–1.26 kg plant⁻¹ for Kauai 2020; 0.16–0.33 kg plant⁻¹ at the Maui Agricultural Research Center (MARC) 2020; 0.80–1.87 kg plant⁻¹ in Waimanalo 2020; 0.92–2.58 kg plant⁻¹ in Waimanalo 2021; 0.51–1.48 kg plant⁻¹ for the Kauai 2022 trial. Plants were generally pest free with the notable exception of the MARC 2020 iteration, where rhizomes of all genotypes were damaged by severe rot. Binucleate Rhizoctonia, Pithium and Meloidogyne nematodes were isolated from the diseased samples.

There were five cultivars that were common to 6 of the 7 iterations. When analyzed as a split plot with iteration (I) as the main plot and genotype (G) as the split plot, iteration, genotype and the I × G interaction were significant. When viewed by iteration (Figure 1), limited interaction between iteration and genotype was observed, with the exception of Olena, which occasionally changed ranks in each iteration. This suggests the interaction between genotype and environment is largely one of magnitude rather than direction. Across iterations, the standard cultivar ‘Hawaiian Red’ is generally a mid-level yielder (0.92 kg plant⁻¹) with ‘Roma’ consistently yielding the highest (1.39 kg plant⁻¹) and ‘BKK’ yielding the lowest (0.64 kg plant⁻¹) (Figure 2).

The amount of variance explained by each term in an additive main-effects and multiplicative interaction (AMMI) model is presented in Figure 3, which includes all 14 genotypes across all 7 trials. Here, we see the greatest magnitude effect is location (31.8%) followed by genotype (21.6%). While the interaction terms are significant, they explain much less (4.8–7.8%) of the variation than other effects. This is similar to the pattern seen in the fully replicated data (Figure 1), where there are very few genotypes that appear to interact (change rank) in different site/year combinations. AMMI models can extract maximum information from the multi-environment trials (MET) by seeking the best estimator of each genotype’s mean yield in a given environment [15]. The model identifies 5 cultivars as potentially producing fresh finger yield statistically similar to the high yielding ‘Roma’: 18-013; 18-014; 18-023; ‘Carib’; ‘Mystic’ (Figure 4). Conversely 18-012 and ‘Wailua Gold’ are identified as statistically similar to the low yielding ‘BKK’.

4. Discussion

Both genotype and environment contributed significantly to yield variance, with environment being somewhat more influential than genotype. This influence of environment can be attributed primarily to the very low yields observed in MARC 2020. The optimum temperature range for turmeric is 18.2–27.4 °C [1]. Average minimum temperatures during crop growth at MARC in 2020 were 12 °C with 8.3 °C measured as the lowest temperature during the crop cycle. These sub-optimal temperatures likely predisposed the crop to naturally occurring pathogens that were promoted by the cool wet conditions. Pythium, Rhizoctonia and Meloidogyne are all important pests of turmeric [15,16,17]. These pests were not observed on turmeric grown in the other iterations. Management differences between iterations also likely contributed to the environmental variance and included fertilizer type, mulch type (organic vs. plastic) and the presence or absence of raised beds.

Fresh turmeric yield differed significantly among genotypes. Excluding the outlying MARC 2020 iteration, the overall yields observed in this work ranged 11.3–57.22 t ha⁻¹ and generally agree with those in the literature. In a Nigerian trial of 15 accessions, fresh turmeric yields of 21.5–45.0 t ha⁻¹ were reported [18], while another trial using organic production methods in India found the mean yield of improved cultivars (32.94 t ha⁻¹) superior to local varieties (12.45 t ha⁻¹) [19]. In a recent study, use of 25% shade netting increased turmeric yields from an average of 21 t ha⁻¹ to 51–61 t ha⁻¹ [20]. ‘Roma’ is the only genotype used in our trial that has yields reported elsewhere in the literature. In previously published studies, fresh ‘Roma’ yields range 10.2–21.6 t ha⁻¹ [21,22,23], while we report yields of 27.9–57.2 t ha⁻¹, excluding the outlying MARC 2020 ‘Roma’ yield of 5.20 t ha⁻¹.

One challenge in comparing our results with other published reports is the difference in plant spacing. Our wider spacing to account for mechanical planting and harvest results in a plant population of 22,222 plants ha⁻¹ compared to 50,000–164,300 [20,21,22,23,24] reported primarily from India where planting and harvest is by hand. Furthermore, it is important to note that the planting densities used in many of these studies are artificially high because they are calculated from small experimental plots and do not take into account space for walkways or aisles [22]. Nevertheless, commercial production spacing in India where most of these studies are from is likely 2–3× what we used in our trials. One result of closer spacing is lower fresh rhizome yield per plant [25]. This is probably the reason that our fresh weight per plant (0.52–2.58 kg plant⁻¹) generally exceeds previously published plant yields (0.09–0.80 kg plant⁻¹) [22,24,26,27,28,29]. Lower yields per plant at closer spacings are compensated for by higher plant numbers per hectare, resulting comparable yields when compared at the t ha⁻¹ level.

5. Conclusions

Both genotype (G) and environment (E) independently contribute to the variance in turmeric fresh yield across diverse environments. In our trial, environment accounted for the largest variance, very likely due in part to suboptimal temperatures and pest pressure at the MARC site. The GxE interaction was significant but relatively small compared to the independent effects. This may explain the similarity in yields reported here with those in the literature. This suggests that high yielding turmeric varieties may possess wide adaptability and may be expected to preform well across diverse environments. However, site-specific evaluations will still be necessary, particularly in environments outside turmeric’s environmental optima and in the presence of high pest pressure.