Flight Activity and Field Infestation Relationships for Coffee Berry Borer in Commercial Coffee Plantations in Kona and Kau Districts, Hawaii

Abstract

The coffee berry borer (CBB), Hypothenemus hampei Ferrari (Coleoptera: Curculionidae: Scolytinae) is a recent invader to Hawaii. To date, limited information regarding the seasonal phenology of this pest on the islands limits the implementation of integrated control strategies. As part of a coffee farmer training program, we monitored CBB flight activity in 15 coffee plantations (Kona and Kau Districts) over 10 mo with methanol-ethanol (3:1 ratio) baited traps. Concurrently, we quantified CBB infestation and penetration rates inside developing coffee berries through the end of harvest. Approximately 1 million CBB were captured, with the highest activity (e.g., >500 CBB/trap/wk) in December through February, coinciding with end of main regional harvesting periods. Relatively high activity (>250 CBB/trap/wk) was also observed during berry development, in May and June (Kona) and June and July (Kau). Field infestation rates were higher overall in Kau (9.6 ± 1.1%) compared with coffee plantations in Kona (4.7 ± 0.4%). Linear regression investigated relationships between CBB trap data and berry infestation rates. Trap catch data generally correlated better with the proportion of shallow entries (AB position) compared with deeper penetrations (CD position) or total infestation. Pearson correlation coefficients based on different parameters (i.e., region, altitude, and berry phenology) revealed positive and mostly significant correlations between these variables (R values 0.410 to 0.837). Timing peak flight activity of CBB with insecticide applications will help coffee growers improve pest control. The ability of trap data to calculate reliable economic (action) thresholds for the CBB is discussed.

Monitoring traps, ethanol, seasonal phenology, integrated pest management, Coffea arabica

In Hawaii, coffee, Coffea arabica L. (Gentianales: Rubiaceae), is grown on about 2,770 ha, which will produce an estimated 16.5 million kg of cherries valued at US$ 62.2 million in the 2016/2017 harvest (USDA-NASS 2017). Coffee farmers in the region are currently facing increasing challenges from the coffee berry borer (CBB), Hypothenemus hampei Ferrari (Coleoptera: Curculionidae: Scolytinae), which was detected in September 2010 (Burbano et al. 2011, Aristizábal et al. 2016). CBB is considered the most significant insect pest in coffee plantations worldwide based on its propensity to reproduce directly inside developing berries, reducing yields and the quality and price of harvested coffee (Duque and Baker 2003, Vega et al. 2015).

Understanding the seasonal phenology and dispersal behavior of CBB is needed to optimize both localized and area-wide integrated control strategies (Damon 2000, Jaramillo et al. 2006, Aristizábal et al. 2012). The flight activity of CBB is initiated when mated females abandon their native berry and colonize new host berries (Mathieu et al. 1997). This is the period when CBB are most vulnerable to insecticide applications. Alcohol baited traps, which mimic kairomones released by developing berries (Mendesil et al. 2009, Cruz and Malo 2013), have been used as lures in traps to monitor and help control CBB in various coffee-producing regions (Saravanan and Chozhan 2003, Silva et al. 2006, Dufour and Frérot 2008, Fernandes et al. 2011, Pereira et al. 2012, Aristizábal et al. 2015).

The CBB developmental dynamics correlate with regional conditions, including coffee phenology, management practices, and climate (Rodríguez et al. 2013). In Hawaii, most coffee is traditionally produced along the ‘Kona coffee belt,’ a low elevation and humid zone on the western side of the Island where growing conditions are ideal (Bittenbender and Easton Smith, 2008). The flight activity of CBB in Kona was reported by Messing (2012), who confirmed the effectiveness of a methanol:ethanol blend lure and reported peak flight activity in November and early January, during the height of coffee harvest. However, these studies were restricted to two plantations from a single District. Moreover, no attempt was made to correlate trap catch data with field infestation levels, which is required if action thresholds for management based on adult trapping data are to be developed.

Here, we monitored CBB flight activity across 15 commercial coffee plantations of Hawaii located in two regions (Districts). We also correlated trap data with berry infestation levels monitored in the field at various stages of crop development. One goal was to obtain additional information on the seasonal phenology of CBB in the region. We also correlated the relationships of trap catch and berry infestation levels in both Districts to explore the impact of several environmental parameters. Finally, we discuss how our data might be used to help develop action thresholds that can be used by coffee growers to improve CBB management practices.

Material and Methods

Field Sites

Studies were conducted over 10 mo (May 2016 through February 2017) in 15 commercial coffee farms located in the Kona and Kau Districts (Fig. 1). Farms varied from 0.5–8 ha, with planting densities of 1,500–2,800 trees/ha and a spacing of 1.2–3.5 m between plants and rows. The following cultivars were grown, C. arabica var. Kona typical, Red Bourbon, Red Caturra, Yellow Caturra, and Maragogipe. Most pruning was done annually via the ‘Kona style,’ supporting multiple-age (4 to 10) verticals on each tree depending on prevailing site conditions (Bittenbender and Easton Smith 2008). In general, the management practices, including weed control and application of fertilizers were similar among locations. Since we hypothesized that altitude might influence CBB phenology, in Kona sampling sites encompassed three elevations, i.e., low (<313 meters above sea level [masl]), medium (426–434 masl), and high (> 573 masl), respectively. In Kau, altitude was not a criterion, because there was little variation among sites (400–600 masl). To control the CBB, farmers applied Beauveria bassiana (Botanigard ES and Mycotrol O, BioWorks, Inc., Victor, NY) either alone or combined with kaolin clay (Surround WP, NovaSource, Phoenix, AZ). The application of these products was based on monitoring data collected as described below. This resulted in an average of 5.25 (range 4–9) applications per site in the 2016/2017 coffee production cycle (data for Kona). Harvesting was conducted from August through December (Kona) and September through February (Kau), with picking intervals every 2 to 4 wk during these periods. On average, farms in Kona produced 5,670 kg (12,500 lbs) of cherry per ha, with peak harvesting yields in October, followed by September and November. Representative meteorological data were obtained from weather stations located in Kona medium elevation (19°32ʹN/155°56ʹW) and Kau Districts (19°19ʹN/155°49ʹW), respectively.

Fig. 1.

Sampling locations for Hypothenemus hampei (CBB) in coffee farms from Kona and Kau Districts, Big Island, Hawaii in 2016 and 2017.

Flight Activity of CBB

At each sample site, five red funnel interception traps (Brocap, CIRAD, Paris, France) were hung from trees within a 0.5–2 ha representative coffee plot (one trap per 200–500 trees). Traps were baited with a mixture of 3:1 methanol-ethanol enclosed in a semipermeable transparent plastic bag (10 × 21 cm) (ChemTica International, Costa Rica), which released about 186 mg/d (Borbón et al. 2002). Traps were placed at 1.2 m height based on earlier recommendations (Dufour and Frérot 2008) and located in each corner and the center of each plot (Fig. 2A). A plastic container containing 200 ml of soapy water (5% vol/vol) was placed in the base to capture CBB (Fig. 2C). Traps were placed after bloom and berry initiation, in May (Kona) and June (Kau). Traps were evaluated at 2- to 3-wk intervals through the end of harvesting and the numbers of CBB collected counted, or if more than 250, estimated with a calibrated vial (Fig. 2D). Only female CBB are usually collected because males do not fly and remain in the brood fruits (Damon 2000). The soapy water was changed after each evaluation and lures changed every 10–12 wk.

Fig. 2.

Brocap traps (CIRAD) baited with a 3:1 methanol-ethanol mixture (A); coffee berry borer (CBB) female colonizing a coffee green berry (B); plastic container with soapy water solution to capture CBB in the bottom of the trap (C); graded vial to quantify CBB (D). Photos by Luis F. Aristizábal.

Field Infestation Rate of CBB

Concurrently with the trapping evaluation, the proportion of berries infested with CBB was monitored inside coffee plots. We followed the ‘30-tree sampling plan’ developed in Colombia (Bustillo et al. 1998). Following a zig-zag path throughout lots, a representative tree was sampled every 10–15 m. For each tree, a branch in the middle was selected and all developing berries counted and examined for CBB entry holes. Sampled branches in May and June (12–16 wk after flowering) contained an average of 54.5 ± 3.4 and 56.8 ± 2.2 berries per branch, respectively, suggesting that fruit set was mostly complete at this time. Assessments were conducted every 2- to 3-wk intervals through the end of harvest periods.

Berry Penetration by CBB

Since the exposure of the CBB to insecticides declines as they progressively bore into the host fruit, we monitored their position inside infested coffee berries. On each sampling occasion, 50–100 infested berries were collected from 30 randomly selected trees (2–3 berries per tree). Berries were cut open with a pocket knife and categorized according to the position of the CBB (AB or CD position), after methods developed elsewhere (Bustillo et al. 1998). In an AB position, the CBB has bored a hole but not penetrated the endosperm and is still vulnerable to insecticides (Fig. 2B), whereas in a CD position, the CBB had penetrated the endosperm and was considered a poor target for insecticides.

Statistical Analysis

Linear regression was used to describe the relationship of trap catch and berry infestation data from each farm and sampling occasion. Trap data were averaged across the five traps and results expressed as #CBB/wk, to standardize comparisons among farms. Correlation coefficients were used to investigate the strength of these linear relationships according to the stage of berry infestation (AB or CD) and also when compared for several measured parameters (Faul et al. 2009). First, we compared regional effects by comparing farms from Kona and Kau separately to investigate the effect of District. Second, because farms in Kona spanned various elevations, we compared sites separately from low (268–313 m), medium (426–434 m), and high (483–624 m) elevations. We hypothesized that potential climatic effects associated with altitude (e.g., cooler temperatures and later planting at high altitudes) might influence the CBB dynamics. Finally, we compared data correlations across different phenological stages of the crop cycle; i.e., from samples obtained during early (8–18 wk post peak bloom periods), mid (19–28 wk), and late (30–40 wk) stages of berry development. During these periods, the majority of fruits were between stage 75 and 88 of the Biologische Bundesanstalt, Bundessortenamt und CHemische Industrie (BBCH) scale (Arcila-Pulgarín et al. 2002). These analyses were conducted using SPSS v.22 (IBM Corp. 2013). Count and proportion data transformed via ln (n+1) and arcsine, respectively, for analysis (Steel and Torrie 1980).

Results

Flight Activity of CBB

In total, 724,262 CBB were collected from nine coffee farms from Kona and 256,300 from six coffee farms in Kau Districts. Based on subsampling from some locations, we estimated that 95% of captured beetles were CBB, with other nontarget scolytids and other beetles (i.e., Hypothenemus obscurus F., Hypothenemus seriatus Eich., and Xyleborus compactus Eich.) generally, ≤5%. The number of captured CBB varied greatly, with up to 30,000 in a single trap. Broadly similar seasonal trends were observed in both districts. In Kona, moderate CBB flight activity (≈500 CBB/trap/wk) was observed during May and June but declined to a low level (<100 CBB/trap/wk) until late October, with a large peak (>2000 CBB/trap/wk) in December through early January (Fig. 3A). This peak corresponded with the late harvest season and followed elevations in humidity and rainfall (Fig. 4). In Kau, CBB activity remained evident throughout the summer (>150 CBB/trap/wk) declining in the late fall, with increased activity (reaching ≈1000 CBB/trap/wk) in December through February, corresponding with the later harvesting period in this region.

Fig. 3.

Number of CBB captured per trap (A), percent infestation of berries (B), and proportion of shallow entries (C) (mean ± SEM) on coffee farms sampled from Kona (n = 9) and Kau Districts (n = 6), Hawaii in 2016 and 2017.

Fig. 4.

Meteorological data from two coffee regions where CBB sampling was conducted in Hawaii.

Field Infestation Levels of CBB

CBB-infested berries were observed in all sampling months, although the proportion of infested berries was higher overall in Kau (9.6 ± 1.1%) compared with coffee plantations in Kona (4.7 ± 0.4%) (Fig. 3B). In Kona, infestation rates were high (>10%) in May but gradually declined throughout the summer months reaching a minimum in August (1.8 ± 0.3%). Infestation of CBB increased in the late fall, reaching a peak (16.3 ± 1.9%) late into the harvesting period. In Kau, relatively high berry infestation rates were observed throughout the summer, again with a late season increase observed starting in December and reaching a peak (18.4% ± 4.3) in January near the end of the harvest season.

Berry Penetration by CBB

The proportion of dissected berries in which CBB were considered susceptible to insecticides (AB position) declined through the growing season in both regions (Fig. 3C). In Kona, this proportion declined to ≈25% (nominal threshold for spraying) by early July but occurred about a month later in Kau, where the coffee was in a slightly later developmental phase. During the summer, most CBB were in the CD position. The proportion of CBB in the AB position increased in November, presumably when new generations of CBB started berry colonization. This increase in AB positions reached a peak in December and correlated with the increased proportions of infested berries recovered in our survey in both Kona and Kau.

Relationship Between Trapping and CBB Infestation

Linear regressions revealed positive relationships between CBB trap and berry infestation data (AB position), with some differences among the tested parameters (Fig. 5). Overall, generally positive and significant correlations (Pearson) for CBB trap data and overall rates of berry infestation were observed (Table 1). When different stages of berry infestation were also compared in the model (i.e., AB, CD), the best correlations were obtained for AB position in eight out of nine comparisons; achieving R values between 0.410 and 0.837. AB position were therefore used to calculate the regression equations. Slightly better correlations (and coefficients of determination in Fig. 5A) were obtained from farms in Kona, when compared with Kau. Moreover, in Kona, the best correlations were obtained from farms at the high altitude, when compared with those at lower elevations (Fig. 5B). There was also a trend for improving correlations in the late season of berry phenology (30–40 wk post peak bloom), with the weakest correlations between trap catch and berry infestation rates obtained during the early season (8–18 wk post peak bloom) (Table 1, Fig. 5C).

Fig. 5.

Linear regression describing CBB infestation rate and trap catch data relationships for several parameters. (A) District shows farms from Kona (closed symbols, y = 0.0029x + 0.51, R² = 0.67) and Kau (open symbols, y = 0.0046x + 0.86, R² = 0.39), (B) shows farms from different altitudes (Kona only), i.e., high (closed symbols, y = 0.0029x + 0.56, R² = 0.85), medium (open symbols, y = 0.0027x + 0.90, R² = 0.27), and low elevations (triangles, y = 0.0025x + 0.14, R² = 0.54), and (C) shows samples from early (closed symbols, y = 0.0037x + 1.31, R² = 0.31), mid (y = 0.0067x + 0.11, R² = 0.73) and late (y = 0.0032x + 0.53, R² = 0.70) stages of berry phenology, respectively.

Table 1.

Pearson correlation coefficients between CBB trap catch and different stages of berry infestation sampled within 15 small coffee farms in Hawaii over 10 mo

Parametera		dfb	% infestation (R valuec)
			Total	AB position	CD position
All locations		159	0.576**	0.626**	0.472**
District	Kona	105	0.660**	0.676**	0.529**
	Kau	52	0.509**	0.590**	0.413**
Altitude (Kona)	High	49	0.785**	0.837**	0.666**
	Medium	24	0.680**	0.410*	0.628**
	Low	28	0.032	0.581**	−0.139
Berry phenology	Early	45	0.393**	0.430**	0.348*
	Mid	54	0.646**	0.720**	0.595**
	Late	56	0.673**	0.794**	0.534**

Correlations are compared separately for different regions and environmental parameters.

^aData fitted for individual farms and sample dates.^bdf calculated as number of correlations within model (n) minus 2.^ccorrelation based on #CBB/trap/wk (average of 5 traps) and % infestation (30 trees).*P < 0.05, **P < 0.01 (two tailed).

Discussion

We investigated the phenology of CBB in Hawaiian coffee farms. Although the number of seasonal CBB generations is unclear, we captured CBB in all months sampled, with mostly caught late November through February during and in the weeks following harvest. Other researchers noted a marked increase in CBB caught in traps during and shortly following main harvest periods or at times when overripe coffee fruits decline (Bustillo et al. 1999, Mathieu et al. 1999, Pereira et al. 2012). It is not clear why CBB was more problematic in Kau in our study, although factors including a longer growing season, inefficient harvesting, presence of berries on the ground, and trees grown with more than four verticals that reduce optimum spray coverage for B. bassiana may have contributed to the higher CBB pressure.

In Hawaii, Messing (2012) observed peak CBB flight activity in Kona on one farm in December and January, at the end of the harvest season. Trap catches exceeded 1,000 beetles per trap per day before stabilizing at about 100 beetles per trap per day. Trap catches on a second farm peaked at 400 beetles per trap per day in November but remained below 200 beetles per trap per day in the remaining months. Messing (2012) did not observe the elevated trap catches in May through early July when compared with August through mid-November, that we observed in 2016/2017, and which corresponded with green berry development. The reasons for this discrepancy is unknown but may reflect different harvesting practices, weather conditions, and the use of insecticide. Messing (2012) also noted that during the off-season (i.e., before the next crop is sufficiently developed for CBB infestation, between January and June depending on the region and altitude), a reduced but persistent number of CBB may be captured. The beetles captured during this period likely originated from the previous season’s berries that were not harvested. In 2017, we observed that many beetles emerged from old berries dropped on the ground and from dried berries (raisins) remaining in the trees. These beetles continued to emerge from old berries (in five out of six farms monitored in Kau and Kona districts in 2017) throughout the berry development period, as late as June in some cases (RGH, unpublished data).

In principle, the strategic use of CBB traps allows coffee farmers to identify periods when females are colonizing new berries and identify hotspots for treatment (Bustillo et al. 1998, Aristizábal et al. 2011). In our study, trap catch significantly positively correlated with field infestation rates in almost all cases. Similar relationships have been obtained in studies conducted in other regions (Mathieu et al. 1999, Fernandes et al. 2011, Pereira et al. 2012, Aristizábal et al. 2015). Pereira et al. 2012 suggested the use of baited traps to make a map and identify hotspots that require the need for early harvest or other management activities such as spraying. However, few attempts have been made to develop generalized action thresholds for CBB based on trapping studies. Factors including high dispersal ability, overlapping CBB generations especially later in the growing season, and the complexity of CBB–coffee–climate relationship make establishing reliable correlations problematic for management decisions (Baker 1999, Damon 2001).

Our data suggest possible refinements for using traps to develop action thresholds. We noted that trap data generally correlated well with early CBB infestation (AB position), which is the stage susceptible to insecticides prior to full berry penetration. We also observed differences in correlations among District, altitude, and stages of crop phenology. The reasons for these differences are not certain at present. However, we hypothesize that the poor fit of data associated with the early berry phenology stage might reflect the longer period (up to 2 mo) that many female beetles spend in the pericarp before berries become dry enough for penetration and gallery making. The better fit of upper elevation versus lower elevation (Kona) could reflect a more synchronized CBB population occurring at cooler temperatures while the reduced relative humidity at lower elevations might reduce insect movement and associated trap catches.

In Brazil, Pereira et al. (2012) attempted to develop an action threshold at 3% and 5% of field infestation, based on the numbers of CBB captured in alcohol traps. The authors correlated trap catch with infestation level recorded in berries at 2-wk intervals over a season. However, the authors were unable to derive the action threshold for traps that correlated with 3% and 5% berry infestation from their linear equation (i.e., y-intercept was ca. 15%), due to the high CBB infestation rates in their sampled experimental fields. Also in Brazil, Fernandes et al. (2011) used field data to derive economic injury level (EIL) for adult CBB. The EIL corresponds to the number of insects (amount of injury) that will cause yield losses equal to the insect management costs. When quantitative and qualitative approaches including insecticide control efficacy, labor costs, crop losses, and market prices were considered, an EIL of 4.3% of bored berries was estimated for both conventional and organic coffee. Associated EILs in the flowering, pinhead fruit, and ripening fruit stages were estimated at 426, 85, and 28 adult CBB per trap per 2 wk. Because different management practices (e.g., cheaper labor and different insecticide practices) may apply in Brazil, the validation of these estimates in the Hawaiian coffee system remains to be determined.

In conclusion, CBB has rapidly become a serious pest in Hawaii. Because the majority of coffee farms on Hawaii are small and have limited resources, ongoing research is needed to refine control recommendations for growers. Our findings provide useful information on the life history of this pest in Hawaii and the potential to improve the CBB management in the field by using sampling traps. Installing and maintaining monitoring traps might prove more acceptable to coffee farmers in Hawaii, when compared with the more labor-intensive process of inspecting fruits. Farmers can make their own trap to save money (Bittenbender et al. 2017) and/or place high densities (22 units per hectare) of traps (Dufour and Frérot 2008), in an attempt to control the CBB population. Additional efforts should further explore the development of action thresholds to facilitate decision-making and increase Integrated Pest Management (IPM) adoption for this pest.

Acknowledgments

This project was coordinated with the Hawaii Coffee Association, and all studies were done with the assistance of farmers and workers. The Synergistic Hawaii Agriculture Council (SHAC) funded this study (TASC Grant for CBB, T16GXCBB01, Control of CBB in Hawaii). We are grateful to technicians from the US Pacific Basin Agricultural Research Centre and the University of Hawaii Cooperative Extension for technical support. Special thanks go to the many coffee growers from Kona and Kau for participating in this study.