Index | Search | Home | Table of Contents

Clement, C.R., R.M. Manshardt, C.G. Cavaletto, J. DeFrank, J. Mood, Jr., N.Y. Nagai, K. Fleming, and F. Zee. 1996. Pejibaye heart-of-palm in Hawaii: From introduction to market. p. 500-507. In: J. Janick (ed.), Progress in new crops. ASHS Press, Arlington, VA.

Pejibaye Heart-of-Palm in Hawaii: From Introduction to Market

Charles R. Clement, Richard M. Manshardt, Catherine G. Cavaletto, Joseph DeFrank, John Mood, Jr., Natalie Y. Nagai, Kent Fleming, and Francis Zee

  13. Table 1
  14. Table 2
  15. Table 3
  16. Table 4
  17. Fig. 1
  18. Fig. 2
  19. Fig. 3
  20. Fig. 4

Pejibaye or peach palm (Bactris gasipaes Kunth) was introduced into Hawaii in 1990 to provide fresh hearts-of-palm for the gourmet restaurant market (Clement et al. 1993). Although heart-of-palm is widely available as a canned product (Tabora et al. 1993), it is little known as a fresh vegetable. The species had been introduced into Hawaii earlier in this century, but no local information existed to guide potential commercial producers. The general objective of the introduction program reported here was to generate the information necessary to guide producers and consumers interested in this new crop.

There are three high-quality edible parts to a juvenile palm stem (Fig. 1). The heart-of-palm is composed of tender developing leaves enclosed within the spear leaf sheath immediately above the apical meristem. The heart accounts for 99% of world trade in vegetable pejibaye (1990 value: U.S.$40 million) and is produced mostly in Latin America, at present (Mora Urpí et al. 1991). The developing stem tissue, extending 10-20 cm below the apical meristem, is also edible. Little edible stem is currently utilized; only 1% of trade in fresh pejibaye consists of stem, mostly in the form of chunks in marinade. The edible leaf is similar to the heart but is not wrapped within the leaf sheath. Consequently, it is difficult to manage during processing, but is of very high quality. All edible leaf produced in Latin America is currently discarded. Chefs in Hawaii are enthusiastic about all three products.

Agronomic research done in Latin America provided a useful guide for the initial formulation of crop management practices in Hawaii, especially regarding plantation density, fertilization, plant management, harvesting, and post-harvest handling (Mora Urpí et al. 1984). Hawaiian agriculture operates under different constraints, however. The most important difference is the cost of labor, including social security, liability insurance, and other indirect costs. Consequently, the Latin American experience had to be modified to fit the limitations of the Hawaiian agroeconomic environment.

As part of the introduction program, growth and yield of half-sib progenies from the Benjamin Constant population of the Putumayo landrace (Brazil) were evaluated at three locations in Hawaii. The effect of planting density was also evaluated in Benjamin Constant progenies and progenies from the Yurimaguas population of the Pampa Hermosa landrace (Peru), using three densities at one location on the Island of Hawaii. Other project objectives were to describe juvenile palm morphology, design a planting scheme suitable for Hawaiian agriculture, compare conventional and organic weed control practices, characterize genetic relationships within the introduced germplasm, evaluate product quality, develop a cost-of-production model, and create a market for the fresh product.


Juvenile pejibaye plants are essentially all leaves, since no stem develops until just before harvest. There were fewer leaves (6-8) than expected (8-10) at all experimental sites, with the density trial site lowest of all (6-7) (Clement 1995). This may have been due to micronutrient deficiencies, since soils at two locations had been severely degraded by poor management. Adequate nitrogen fertilization masked the leaf deficiency until harvest, when no density effects were observed (see below).

Good sucker (off-shoot) production is essential for long-term survival of the plantation. Sucker number averaged 6-7 (range 0-18) at first harvest, but was substantially lower at second harvest (mean 2, range 0-10) for Benjamin Constant germplasm (Clement 1995). This may have been due to: (1) differences in sucker management, where intense competition among the full sucker complement up to first harvest may have caused abortion of second cycle sucker buds, or (2) different partitioning of resources between shoots and their offshoots during the second cycle, when shoot growth was much more vigorous.


There were no significant differences in individual plant dimensions or growth rates at 3333, 5000 (the commercial density in Costa Rica), or 6666 plants/ha in either Benjamin Constant or Yurimaguas germplasm at the Island of Hawaii site. There are four possible reasons for the lack of density effects on individual plant dimensions: (1) the 9-plant plots were too small, allowing sufficient light to penetrate the plot from the row spaces; (2) palm growth was sufficiently irregular that most plants received sufficient light from some side; (3) the reduced leaf number (noted above) reduced leaf area index (LAI) and therefore competition; or (4) the densities studied do not significantly affect vegetative growth of pejibaye (Clement 1995).

The first possibility was recognized as a potential limitation before field planting. A shortage of germplasm, because of poor germination, did not permit larger plots. The second possibility is probably important too, since the growth rate of each plant is affected by its unique micro-environment. Also, as harvest progressed, the canopies became more irregular, since some plants were removed each month. The third possibility certainly existed during the experimental period, although the magnitude of its effect on LAI is unknown. The older leaves, which may have senesced prematurely due to micronutrient deficiency, would have a relatively small impact on LAI because they were small. The fourth reason is theoretically possible since Corley (1973) observed that vegetative growth in African oil palm (Elaeis guineensis Jacq.) was not reduced at planting densities higher than the optimum for fruiting, although fruit yield dropped dramatically.


Hawaiian farmers require mechanical access to all plants in the field to reduce labor costs. The lack of planting density effects suggested modifications to Latin American practices that would allow easy access without reducing the number of plants per hectare.

Palms are planted in offset double rows with 1-m spacing between plants. Service rows, planted with a legume or grass groundcover, alternate with double rows of palms and permit easy access. This design allows 5000 or more plants/ha. Rows should run along the contours, so that the service rows act to contain erosion. Because all plants are in border rows, the high within-row density should not reduce growth. This design is already being used commercially on the Island of Hawaii.


Weeds in pejibaye plantations in Hawaii are a major consideration. Consequently, the degree of control afforded at two levels each of oxyfluorofen, oryzalin, and paraquat were compared with the suppression provided by woven black polypropylene mat (as the control). Oxyfluorofen and polypropylene mat were equally good conventional weed controls (DeFrank and Clement 1995).

Groundcovers can also act to control weeds, and they fit well into pesticide-free organic production methods. Three vegetative groundcovers [two legumes (Arachis pintoi, Desmodium ovalifolium) and a grass (Chloris gayana)] and polypropylene mat were compared for efficacy in weed control. All of the vegetative groundcovers competed significantly with the palms, as evidenced by delayed harvest. Consequently, polypropylene mat or organic mulch in the tree rows is an essential complement to the groundcovers to avoid competition. The palms must also receive more fertilizer to compensate for that consumed by the groundcovers.

The plantation design in use on the Island of Hawaii integrates the polypropylene mat with Arachis pintoi. Two-meter-wide polypropylene mat is used to control weeds in the immediate vicinity of the palms in the double rows. A. pintoi is thereby restricted to the service row, where it does not compete with the palms. After full establishment of A. pintoi, it is an acceptable weed-suppressing groundcover, requiring only occasional mowing of weeds that penetrate above its canopy.


Relative growth rate (RGR) in the Benjamin Constant germplasm was highly correlated with earliness (r= -0.99) (Clement 1995), measured as days from nursery transplant to harvest (Table 1). An analysis of Unit Leaf Rate (EA) and Leaf Area Ratio (LAR), the product of which equals RGR, showed that most of the variation for earliness among Benjamin Constant progenies was due to variation in EA. Because individual plant heart-of-palm weights were not different among Benjamin Constant progenies (Clement 1995), earliness (as measured by either RGR or days to harvest) is the major component of annual yield in this germplasm.

Benjamin Constant progenies differed in partitioning to sucker number and sucker growth also (Clement 1995). Some partitioned early to sucker number, but not significantly to sucker growth. Others partitioned early to both number and growth. Others partitioned little to sucker number, but significantly to sucker growth. The latter partitioning pattern may be preferable in Hawaii because it will reduce labor costs for sucker pruning, while providing more rapid growth to the next harvest. However, the marked reduction in number of suckers produced by plants in the field between the first and second harvests may impose limitations on the optimum sucker number that is possible.


Mean annual yields of heart-of-palm were 500, 900, and 1000 kg/ha at 5000 plants/ha at the three Benjamin Constant experimental sites (Clement 1995). Yields at the two better sites are comparable to good tropical American figures. The lowest yield was obtained on a Histosol, essentially unweathered lava rock, with rainfall of 4800 mm per year, but where a week with no rain caused severe drought. The intermediate yield was obtained on an Andisol with 3500 mm of rain, degraded by nearly a century of sugarcane production. The highest yield was obtained on an Oxisol degraded by poor management, with 1000 mm of rain annually and supplemental irrigation. The high yield was probably due to high light intensity under cloudless conditions, and yield might have been higher with better irrigation technology. There was considerable variation in yield among Benjamin Constant progenies (Fig. 2).

With two stems per plant after first harvest, cumulative 18-month yields of hearts-of-palm were 800, 1400, and 1500 kg/ha. When both edible stem and edible leaf were added, 18-month yields were 2.5, 4.5, and 4.8 mt/ha, respectively. Yurimaguas yields were consistently 10% higher than Benjamin Constant yields at the 4.5 mt/ha site, apparently because of lower fiber content.


There were strong indications of past inbreeding in the Benjamin Constant germplasm, in that narrow-sense heritabilities for both vegetative traits and growth rates, characters that usually have moderate to low heritabilities, were in excess of 1.0 (Clement 1995). When inbreeding was arbitrarily adjusted to 0.5, RGR, EA and earliness had moderate heritabilities.

Response to selection for RGR, EA or earliness depends upon the amount of variation in the population. This, in turn, depends upon the period over which the trait is measured. Three different intervals were examined: a uniform time period (e.g. 292 to 464 days in the field); a uniform developmental period (seedling to harvest); and a mixed category (seedling to 292 days in the field). RGR over the entire seedling-to-harvest period had the lowest variation, but is probably the correct evaluation period (Table 2), since it minimizes variation due to differing developmental morphologies (Clement 1995).

Conventional breeding methods are unlikely to yield rapid genetic gains with this germplasm, because of low variability. Consequently, vegetative propagation is being pursued through tissue culture to capture the best phenotypes for immediate use by farmers.


Averaging across 17 isozyme loci, mean allozyme heterozygosity in Benjamin Constant and Yurimaguas progenies was 0.074 (range 0.038-0.099) and 0.124 (range 0.098-0.153), respectively. In an allogamous crop, such low heterozygosity strongly suggests significant past inbreeding, substantiating results obtained in the quanitative genetic analysis. All Benjamin Constant progenies were very closely related genetically, as shown in the dendrogram based on Nei's (1978) genetic identities (Fig. 3). Allozyme heterozygosity was not correlated with growth rates or vegetative traits in the Benjamin Constant germplasm, probably because of the relative uniformity of this inbred germplasm (Clement 1995).


There is considerable within-progeny and within-plant variation for heart-of-palm quality characteristics, such as appearance, sweetness, tenderness, crispness, and moistness (Table 3). Total soluble solids varied from 3%-13%, with more variation between harvest dates than between progenies. Yurimaguas germplasm was always slightly sweeter than Benjamin Constant.

Negative sensory characteristics found in some samples were astringency and acridity (Table 3). Acridity was more common, occurring in 5% and 8% of Yurimaguas and Benjamin Constant germplasm, respectively. Human sensitivity to acridity ranges from no reaction to an extremely sensitive, perhaps even allergic reaction. Contrary to the finding of Tabora et al. (1993), most consumers in Hawaii do not like the "bite" (acridity) in fresh heart-of-palm. No biochemical assay is currently available to quantify acridity, which limits our ability to evaluate it or determine the relative importance of genetic and environmental factors. Nonetheless, these negative traits are receiving priority attention because of possible health risks which they may pose for a small minority of the population.

Hearts retain fresh flavor and appearance for two weeks when wrapped in plastic film and refrigerated immediately after harvesting. This shelf-life will allow shipping to and marketing in major U.S. markets. Further research is planned to extend shelf-life.


Establishment of a pejibaye heart-of-palm plantation in Hawaii is expensive ($51,000/ha, amortized over 10 yr), and yearly operating, harvesting, processing and marketing costs totalled $33,000/ha when in full production (3rd year) at a good rainfed site on old sugarcane land on the Island of Hawaii (Table 4). Each shoot (heart = 150-225g, stem = 340-500g, leaves = 45-135g) must be sold for $5.00 to break even. Each dollar received above that price produces a 15% return on investment. Current market price in Hawaii is $12/stem, but is expected to drop as more farmers start to produce. A hypothetical cash flow diagram, assuming $6 per shoot, is presented in Fig. 4. The full cost-of-production model is available upon request from Kent Fleming.


Harvests from the experimental plots were used to introduce the fresh heart-of-palm directly to chefs in Hawaii's finer restaurants and to consumers through participation in gourmet product expositions and food fairs. Interested farmers were urged to negotiate directly with chefs to develop strong relationships and favorable profit margins.

Requirements of individual chefs ranged from 4 to 18 kg/week of palm (including heart, stem, and leaf), so a farmer will need 0.10-0.50 ha to supply one chef. The local restaurant and resort market is estimated to require 80-160 ha of high density plantings. As more area comes into production and supply increases, prices will drop to a level that should allow the demand for hearts-of-palm to expand beyond its present gourmet niche.

California buyers are interested in expanding the area in cultivation in Hawaii to meet mainland and international export market demands. Preliminary research by the USDA/ARS Fruitfly Ecology Lab, Hilo, HI, indicated no fruitfly infestation problem in pejibaye palm hearts. This information is needed by USDA/APHIS to permit importation of heart-of-palm into the continental U.S. Once pejibaye heart-of-palm is cleared for importation, production area may expand beyond the above projections.


Table 1. Benjamin Constant progeny means for earliness (days to harvest), relative growth rate (RGR), unit leaf rate (EA), and leaf area ratio (LAR) for the period from nursery transplant to first harvest at the best rainfed site, and progeny mean correlation coefficients with earliness.

Progeny means
Progeny Earliness
(days to harvest)
B-0 898 11.71 2.66 10.71
B-1 932 11.33 2.59 10.69
B-2 982 10.82 2.45 10.71
B-3 972 10.87 2.51 10.67
B-5 1027 10.31 2.33 10.71
B-8 954 11.01 2.47 10.72
B-9 899 11.85 2.61 10.77
Correlation coefficientz -0.99*** -0.96*** -0.41
zCorrelation coefficiant of progeny means with progeny means for earliness. Critical values of r are: 0.75, 0.87, 0.95, for p = 0.05, 0.01, 0.001, respectively.

Table 2. Total variation and heritability for relative growth rates of Benjamin Constant progenies determined over different time intervals at the best rainfed site.

Time interval VP CVP h2
292 to 464 days in field 31.9 55.0 0.78z
Seedling to 292 days in field 10.7 24.5 1.44z
Seedling to first harvest 1.2 9.8 0.74z
zInbreeding suspected because of very high h2
VP = Total variance, CVP = coefficient of phenotypic variation, h2 = heritability (narrow sense)

Table 3. Variation in mean sensory evaluation scores for six quality characteristics of fresh heart-of-palm, comparing values for sib plants within a single Benjamin Constant progeny and for different sections of the same heart (the 1st section includes the apical meristem).

Evaluation score (1-5 scale)z
Plants or sections Sweetness Tenderness Crispness Moistness Astringency Acridity
Sib plants
1 2.1 2.7 1.8ay 2.8 1.5 1.8ab
2 2.4 2.4 2.2ab 3.0 1.4 1.7a
3 2.4 2.3 2.3b 2.9 1.6 2.3b
Heart sections
1st 2.6 2.8 1.9 2.7 1.8b 2.1b
2nd 2.1 2.2 2.1 2.9 1.3a 2.0ab
3rd 2.1 2.4 2.2 3.1 1.4a 1.6a
zCharacteristics were scored by a panel of 11 people, who rated each on a scale of 1 (low expression) to 5 (high expression).
yMean separation in columns by Duncan's new multiple range test, 5% level.

Table 4. Major components of a cost-of-production model for a 1-ha pejibaye heart-of-palm plantation on the Island of Hawaii.

Variable $/plant $/ha
Establishment and first year of management 7.907 50,370
Annualized ownership expense 1.091 6,974
Operating costs (5th year)
   Growing costs 1.035 6,615
   Harvesting, processing, shipping costs 2.357 15,067
   Total 3.392 21,682
Gross income (5th year)
   Heartz 2.880 18,410
   Edible stemy 3.168 20,251
   Edible leafx ? ?
   Total gross income 6.048 38,662
Gross margin 2.656 16,982
Ownership costs 1.626 10,454
Total cost of production 5.018 32,135
Economic profit (5th year)w 1.030 6,527
z0.22 kg/plant, $13.00/kg
y0.45 kg/plant, $7.00/kg
x0.05 kg/plant, $/kg unknown
wTotal gross income - Total cost of production

Fig. 1. The three potential palm shoot products.

Fig. 2. Mean annual heart-of-palm yields for seven Benjamin Constant progenies planted at three densities at the best rainfed site in Hawaii.

Fig. 3. Dendrogram based on Nei's (1978) unbiased genetic identities among progenies of the Benjamin Constant population in Hawaii. Cophenetic correlation = 0.69.

Fig. 4. Hypothetical cash flow for a pejibaye heart-of-palm plantation on the Island of Hawaii, assuming a planting density of 5000 plants/ha and a stem value of $6.00.

Last update August 22, 1997 aw