Materials and methods

The study was conducted in Puerto Rico at the U.S. Department of Agriculture, Agricultural Research Service Research Farm in Isabela (Coto clay: clayey, kaolinitic isohyperthermic Typic Hapludox), at the Corozal Agricultural Expt. Sta. of the University of Puerto Rico (UPR) (Corozal clay: clayey, mixed, isohyperthermic Aquic Haplohumults), and at the UPR-Juana Diaz Agricultural Expt. Substation (San Anton: fine-loamy, mixed isohyperthermic Cumulic Haplustoll). Soil and climatic characteristics are described in Tables 1 and 2.

Table 1.Average preplant soil characteristics at three carambola test sites in Puerto Rico measured to a depth of 12 inches.

View Fullscreen

Table 2.Historical weather data at three carambola test sites in Puerto Rico (n = 58 years).

View Fullscreen

Six-month-old trees of cultivars Arkin, B-10, B-16, B17, Kajang, Kari, Lara, Sri Kembangan, and Thai Knight grafted onto ‘Golden Star’ rootstocks were transplanted to the field 9 Feb. 1999 (Isabela), 17 Mar. 1999 (Corozal), and 3 June 1999 (Juana Diaz), and were arranged in a randomized complete-block design with six replications at each location. Planting holes about 1.5 ft deep were dug with an auger connected by a drive shaft to the power take-off unit of a tractor. At transplanting, each plant received 11 g granular P provided as triple superphosphate. Within a replication, plots for each cultivar contained two trees spaced 12 ft apart and 18 ft between adjacent rows in a triangular array. The experiments were surrounded by two guard rows of ‘Golden Star’ seedlings. Irrigation based on tensiometer readings was provided with spinner jets (model DXMAG368X; Maxijet, Dundee, FL) spaced 12 ft apart and providing 13.5 gal/h at 20 psi. Fertilization was provided every 3 months using a 15N–2.2P–16.3K–1.8Mg commercial mixture at a rate of 498 kg·ha⁻¹. Herbicide for weed control was applied only in strips within the planting row.

Harvests were initiated in Aug. 2001 at all locations. At each harvest, number and weight of marketable and nonmarketable fruit were recorded and weighed. Deformed fruit and those measuring less than 10 cm in length and 5.5 cm in diameter were considered nonmarketable. Representative fruit totaling 10% of those harvested were used to determine fruit length and diameter. Soluble solids readings were also recorded with a temperature-compensated Packet PAL-1 digital refractometer (ATAGO Co., Ltd., Tokyo, Japan) when the fruit ripened, about 2 to 3 d after harvest. The experiment was ended in Aug. 2005.

Analysis of variance was carried out using the GLM procedure of SAS (release 9.0 for Windows; SAS Institute, Cary, NC). After significant F test at P ≤ 0.05, means separation was performed with lsd.

Results and discussion

Location, cultivars, and year showed highly significant effects (P ≤ 0.01) on most fruit parameters measured in the study. Exceptions were number of marketable fruit and marketable weight per hectare, which did not show a significant location effect. The cultivar × location interaction was significant for total number of fruit, number of nonmarketable fruit, nonmarketable fruit yield, and fruit soluble solids but not for number of marketable fruit, marketable fruit yield, total yield, fruit length, and fruit diameter (Table 3). Except for fruit length and diameter, the year × cultivar interaction was also highly significant for most fruit variables. Overall, cultivars exhibited an increase in the number of marketable fruit and weight during the first 3 years of production and then leveled off on the fourth year (Table 4). This response was expected as trees increased in age. However, the magnitude of this response varied among cultivars as expected by the significant year × cultivar interaction (Table 3). ‘Arkin’, ‘B-10’, ‘B-16’, ‘Kajang’, ‘Lara’, and ‘Thai Knight’ had an increase in the number of marketable fruit and marketable yield during the first 3 years and then declined or leveled off. In contrast, these variables increased in ‘B-17,’ ‘Kari,’ and ‘Sri Kembangan’ throughout the experimental period (Table 4).

Table 3.Yield and fruit quality traits of nine carambola cultivars planted at three locations in Puerto Rico.^z

View Fullscreen

Table 4.Number and yield of marketable fruit of nine cultivars of carambola grown in Puerto Rico.

View Fullscreen

‘Sri Kembangan’, ‘Thai Knight’, and ‘B-17’ had the three highest 4-year mean for number and yield of marketable fruit (Table 4). ‘Sri Kembangan’ had the highest number of marketable fruit and marketable yield in 2 of the 4 years the experiment lasted, and the highest 4-year mean for these variables. ‘B-16’ had the lowest 4-year mean for number of marketable fruit and marketable yield.

Total number of fruit was significantly different among locations, with more fruit produced in Corozal (793,735 fruit/ha) and less at Isabela (479,759 fruit/ha; Table 3). A possible explanation on why more fruit were produced at Corozal than at other locations may be that this site is less windy than Isabela and Juana Diaz. Windy conditions have been reported to be detrimental to carambola fruit production (Crane, 2005; Galan-Sauco et al., 1993), but the effects of wind on physiological processes of carambola are still unknown (Marler et al., 1994). The number of marketable fruit was not significantly different among locations averaging 258,761 fruit/ha. At Corozal, the number of marketable fruit constituted only 31% of the total fruit harvested whereas at Juana Diaz and Isabela it was 41% and 55% respectively (Table 3). Chemical fruit or flower thinning to reduce crop load and increase fruit size is not a standard cultural practice in carambola production. The potential of fruit thinning in carambola is being studied by the author to find out if this practice is a potentially profitable one for this crop. Studies in other crops have shown that fruit thinning often reduces returns to the growers (Davis et al., 2004). The average marketable fruit weight was also similar among locations averaging 30,978 kg·ha⁻¹. Average fruit length and diameter were significantly greater at Isabela than at other locations whereas average soluble solids concentration was significantly higher at Juana Diaz (Table 3).

At all locations, total number of fruit was significantly higher in ‘Arkin’ and ‘Thai Knight’ than in other cultivars and significantly lower in ‘Kari’ (Table 3). The number of marketable fruit produced by each cultivar varied among locations. At Corozal, there were no significant differences in marketable fruit production among top-producing cultivars B-17, Thai Knight, and Kari, which averaged 308,582 fruit/ha. ‘Kajang’ and ‘B-17’ were the highest producers of marketable fruit at Juana Diaz, averaging 320,392 fruit/ha. The number of marketable fruit produced by each cultivar at Isabela was not significantly different (Table 3). ‘B-16’ produced the lowest number of marketable fruit at all locations, but differences between ‘B-16’ and ‘Lara’ at Corozal and ‘Arkin’ and ‘B-10’ at Juana Diaz were not statistically significant (Table 3). As a percentage of total fruit production, ‘Arkin’ produced a significantly higher percentage (75%) of nonmarketable fruit at all locations (data not shown) than other cultivars. This cultivar normally produced a large number of fruit, which may result in high sink demand for assimilates and smaller fruit.

Significantly higher yield of marketable fruit was obtained by ‘Kajang’, ‘B-17’, and ‘Sri Kembangan’ at Juana Diaz and ‘B-17’ at Corozal. ‘Lara’, ‘B-16’, and ‘Kajang’ produced significantly lower weight of marketable fruit per hectare than others at Corozal, whereas ‘Arkin’ and ‘B-16’ did so at Juana Diaz. Weight of marketable fruit at Isabela was not significantly different among cultivars.

Individual weight of marketable fruit averaged over locations and cultivars was 120.2 g (data not shown). However, there was considerable deviation from this weight. At all locations individual weight of marketable fruit was significantly higher in ‘B-10’ and ‘B-16’, averaging 132.3 g, whereas ‘Kari’ and ‘Thai Knight’ produced marketable fruit that weighed only an average of 111 g (data not shown).

‘Kari’ produced significantly longer fruit than other cultivars at all locations, whereas ‘B-16’ produced the shortest fruit overall (Table 3). ‘B-10’ and ‘Lara’ produced fruit with significantly greater fruit diameter than other cultivars at all locations, whereas those of ‘Kari’ and ‘Sri Kembangan’ produced fruit with the smallest diameter (Table 3). The highest soluble solids concentration values were obtained from fruit of ‘B-17’ at all locations, whereas the lowest values were obtained from ‘Arkin’ fruit.

In this study nine carambola cultivars were evaluated for the first time at three locations during 4 years of production. Overall, cultivars were highly adaptable to the diverse agroenvironments in which they were planted, although significantly higher yields of marketable fruit were obtained at Isabela (Oxisol) and Juana Diaz (Mollisol). The fact that ‘B-17’ had a high production of marketable fruit, high marketable yield, and significantly higher soluble solids concentration values than the rest of the cultivars at all locations makes this cultivar suitable for planting on various agroenvironments.