SUMMARY
Equations were developed to predict the beginning of harvest and 50% harvest completion from cycling (pruning and fertilizing) for guava (Psidium guajava L. 'Beaumont') in Hawaii. Variables in the multiple regression equations were photoperiod and the sum of the growing degree days 150 days after cycling. harvest and weather data were obtained from two commercial farms on the islands of Hawaii and Kauai. Days to harvest and days to 50% harvest completion were predicted within mean deviations of 7 and 12 days, respectively. For the two sites photoperiod and temperature data were historical, not current season, means. Thus, predictions can be made for several years in advance.
INTRODUCTION
Guava (Psidium guajava L.) is an important crop in Hawaii with 15.3 million pounds produced in 2001, having a value of $2.2 million. Guava growing in the wild has two distinct harvest periods during the year. To insure consistent monthly commercial production all year-round the practice of cycling was introduced. Cycling involves pruning, partial defoliation, fertilization, and/or irrigation of the trees following the harvest period. This results in vegetative flushes upon which the flowers are produced. Harvesting begins approximately seven months after cycling though the time can vary greatly depending on the season and the weather.
Growers and processors need to know the monthly harvest of guava which depends in part on the time from cycling to harvest. To effectively manage guava orchards and processing plants, a method to estimate when harvest will begin and when the harvest peak will occur is needed. The objective of this study was to develop models to predict the beginning of harvest and 50% harvest completion of guava at two sites on the islands of Hawaii and Kauai.
PROCEDURE
Harvest and weather records were used from two commercial farms on the islands of Kauai and Hawaii, state of Hawaii. Data were analyzed on dates of cycling and daily harvest yields from (a) ten crop cycles in nine orchards on a farm on Kauai and (b) 31 crop cycles in ten orchards from a farm on Hawaii. Temperature records were obtained from the farms and a nearby official U.S. reporting station. Photoperiod (daylength in minutes) was calculated by interpolating from standard tables for 22° 12.5'N (Kauai) and 19° 30' (Hawaii). Linear, polynomial, and multiple regression analysis were performed.
RESULTS
Days to harvest and 50% harvest completion. Days to harvest from the date of cycling for the Kauai site showed a distinct sine wave curve. The curve resembled a photoperiodic response rather than a response to mean temperatures. The photoperiod at cycling correlated very strongly (r = 0.98) with the number of days to the beginning of harvesting and to 50% harvest completion. The term 50% harvest completion indicated when half of the total production was completed. A similar relationship was observed at the Hawaii site, but the equation developed from data at one site poorly predicted the number of days to harvest at the other site when compared with observed data.
Temperature and GDD. The effect of temperature was investigated by calculating growing degree days (GDD) by the standard formula with a base temperature of 15°C.
GDD - [(max + mini)/2] - 15
where max is the maximum daily temperature (°C) and min in the minimum daily temperature (°C).
GDD was calculated from historical weather records in 30-day intervals starting at the last day of cycling and summed at 60, 90, 120, 150, 180, and 210 days after cycling. Correlating these GDD sums with days to harvest, we found that the sum of GDD for the 150 days after cycling (150 GDD) had the highest correlation. However, when this term was entered in a multiple regression equation with photoperiod for a single site, it was not significant. Attempting to explain this, we correlated photoperiod at cycling and the 150 GDD at both sites and found that they were highly correlated (r = 0.99 and r = 0.98).
Time to harvest equation. When data from both sites were combined in a multiple regression equation, photoperiod and 150 GDD remained in the equation which explained 84% of the variation (R² = 0.84). This relationship was verified using data from ten cycled crops on Kauai that were not used in the development of the equations and from historical temperature data.
The equation for the number of days to harvest from cycling (H) was:
H = -606.5 + 0.35(150 GDD) - 0.0002(150 GDD)² + 2.1PP - 0.002PP²
where GDD is growing degree days and PP is daylength in minutes.
Time to 50 harvest completion equation. The equation for the number of days from cycling to 50% harvest completion (H50) was:
H50 = 169.5 + 0.31(150 GDD) - 0.0002(150 GDD)² + 0.0001PP²
where 150 GDD is the sum of GDD for 150 days after cycling and PP is the daylength in minutes.
DISCUSSION
We feel that the above equations will contribute to better management of cycled guava farms in Hawaii. Because the equations use historical data and not current season data, long-term planning is possible. Tables showing the cycling dates and the predicted dates of the beginning of harvest and 50% harvest completion can be developed for any location, provided historical temperature records are available.