Fruit flies (Diptera: Tephritidae) represent 1 of the most economically important insects in the tropical and sub-tropical regions of the world. Fruit fly larvae feed in the pulp of ripening host fruit, causing fruit spoilage (Aluja 1994). This damage has a serious impact on international marketing of fresh fruits and vegetables (Pierreval 2012). As a consequence, infestations of these insects require implementation of area-wide or national control programs in order to comply with sanitary and phytosanitary standard measures (Aluja 1994; IAEA 2003). Haiti is among the twenty greatest mango, Mangifera foetida Lour. (Anacardiaceae), producers in the world and the 6th largest mango exporter to the US market after Mexico, Peru, Ecuador, Brazil, and Guatemala (Bonicet 2013). Moreover, the Central Bank of Haiti estimates mango export at about US $10 million ( Pierreval 2012). In addition to its economic importance, mango is an important source of vitamin A for undernourished nations (Muoki et al. 2009). The Caribbean fruit fly, Anastrepha suspensa (Loew), and the Indian fruit fly, Anastrepha obliqua (Macquart) (Diptera: Tephitidae), are the primary pests of mango in Haiti (MARNDR 2009). In 2007, losses in mango production were estimated at more than US $4 million, equivalent to 40% of the value of mango exports (Pierreval 2012). In 2007, Haitian mango exports to the US were banned due to infestation with fruit fly larvae. Since 2008, a nation-wide program has been implemented to detect and control fruit flies, in order to protect the mango export economy in Haiti (MARNDR 2009).

In 2010 the National Mango Forum organized by 2 United States Agency for International Development-funded programs set a goal to help Haiti increase its export from 2.5 to 5.0 million cases of United States Department of Agriculture-certified mangoes by 2015 (USAID 2010). As a result, maintaining pesticide-free mango production is a crucial goal for the mango industry. However, a weakness of the mango industry in Haiti is the scarcity of commercial orchards. Instead, Haitian mango production relies primarily on individual mango trees dispersed throughout smallholder farms. Thus, mango production is not an important and permanent source of income to smallholders (Castañeda et al. 2011). Therefore, despite the fact that farmers with small or average orchards could improve production by applying sanitation and fruit fly control methods, only a few smallholders have shown interest in investing money for this purpose (Castañeda et al. 2011). This situation makes it difficult for the Division of Plant Protection to manage the National Program for Detection and Control of Fruit Flies. Given the distributed nature of mango production in Haiti, mass trapping and other non-insecticidal techniques may be better options for management of fruit flies in Haiti than broadcast insecticide applications (MARNDR 2014).

A mass trapping network established in Haiti has contributed to a significant reduction in Anastrepha species density (MARNDR 2014). These data demonstrated mass trapping to be an appropriate management method (Kogan & Jepson 2007). However, the density of McPhail traps (Great Lakes IPM, Inc., Vestaburg, Michigan) recommended for mass trapping methods to control fruit flies (25–50 traps per ha; Martinez-Ferrer et al. 2012) represents a financial cost that cannot be supported by smallholders (Burrack et al. 2008; Lasa et al. 2013; Malo & Zapien 1994). Thus, in order to be sustainable and to reduce food safety risks, new trapping methods and trap devices must be designed to control fruit flies in a more cost-effective manner.

In addition to attractive trap design and bait (Lasa et al. 2014a), the efficacy of mass trapping as a pest control method depends on trap density and distribution (El-Sayed et al. 2006). An adequate number of baited traps must be distributed in infested areas in order to minimize the adult fruit flies in the area (Martinez-Ferrer et al. 2012). Calkins et al. (1984) demonstrated that 20 to 25 traps provided a high probability of detecting A. suspensa in citrus groves in Florida. Standardized trapping guidelines for area-wide management of fruit flies recommend 20 to 25 traps per ha for detection surveys (IAEA 2003). However, Martinez- Ferrer et al. (2012) reported that a density of 25 traps per ha was sufficient as a stand-alone method to control the Mediterranean fruit fly (Ceratitis capitata Wiedemann; Diptera: Tephritidae) in Spain.

Our research was designed to evaluate the effectiveness of 2 inexpensive artisanal trap designs as an alternative to the standard McPhail trap for mass trapping fruit flies in Haiti. Our objectives were 1) to compare rates of fruit fly capture by 2 artisanal trap models and the McPhail trap, 2) to compare effectiveness of the better artisanal model and the McPhail trap for mass trapping fruit flies, and 3) to assess the optimal trap density for mass trapping fruit flies in mango orchards in Haiti (MARNDR 2009).

Materials and Methods

DATA COLLECTION AND ANALYSIS

For the first experiment, traps were sampled twice weekly, and the traps in each group were rotated clockwise to minimize any effect of trap location. The bait was replaced weekly, i.e., every second sample time. Fruit flies were collected from the traps, placed in labeled vials with 75% alcohol, and identified to species. Traps caught many insects from various orders, of which non-tephritid dipterans and wasps were predominant. Only 2 fruit fly species were identified: A. obliqua and A. suspensa. These fruit flies were identified to species and sorted by sex. Statistical analysis was performed using ANOVA (PROC General Linear Model), followed by Fisher LSD mean separation procedure (α = 0.05) for significant results, in order to compare trap models on their average number of fruit flies captured (SAS 2009). A binomial test for equal proportions (α = 0.05; Bonferroni correction for 3 comparisons, α = 0.017) was performed to compare the trap models on the proportion of fruit flies captured.

Fig. 1.

Trap models developed and used for evaluation of mass trapping efficacy for fruit flies in a mango orchard in Haiti. AT1, yellow-bottomed artisanal trap; AT2, clear artisanal trap; and MP, commercial McPhail trap.

For the second and third experiments, traps were sampled weekly, and the bait was replaced at the same time. Trapped insects were placed in labeled vials with 75% alcohol, and identified. Statistical analysis was performed using a binomial test for equal proportions (α = 0.05) to compare the number of fruit flies trapped by the two methods (SAS 2009).

For the third experiment, statistical analysis was performed using SAS (SAS 2009). Because of differences in the variances of the treatments, data were transformed using the negative exponential function: y = asymptote (1^{-(curve(x + shift))}). The Gauss-Newton iterative method was used to estimate the asymptote, curve, and shift. Nonlinear regression analysis (Proc NLIN; nonlinear model - negative exponential) was performed to assess the relationship between the dependent variable (y = number of fruit flies caught) and the independent variable (x = trap density). This algorithm regresses residuals onto the partial derivatives of the model with respect to the parameters until the estimates converge (SAS 2009). Convergence criteria were met when the sum of squares was minimized. The initial parameter estimates for the nonlinear regression were asymptote = 50, curve = -0.40, and shift = -14.

Results

No significant differences were observed in the average number of fruit flies captured between the yellow-bottomed artisanal trap (AT1; 8.9 ± 2.6), the clear artisanal trap (AT2; 13 ± 2.9), and the commercial McPhail trap (MP; 16 ± 4.1) (F = 1.2, df = 2, 57, P = 0.31). However, significant differences were observed between the 3 trap models in total number of fruit flies caught (P <0.001). The total number of fruit flies trapped was highest in McPhail traps (319 flies), followed by the clear artisanal traps (253 flies) and lowest in the yellow-bottomed artisanal traps (178 flies; Fig. 2).

The proportion of A. obliqua was significantly higher than that of A. suspensa in all 3 trap models. Anastrepha obliqua was represented by 99%, 100%, and 100% of specimens in the yellow-bottomed artisanal, clear artisanal, and McPhail traps, respectively; A. suspensa was represented by only 1%, <1%, and <1% in the 3 trap models, respectively.

The percentage of females trapped in the yellow-bottomed artisanal trap (74%), the clear artisanal trap (63%), and the McPhail trap (63%) was significantly greater than that of males captured (Fig. 3). Further analysis demonstrated that the proportion of females caught in the yellow-bottomed artisanal trap was significantly greater than that of the clear artisanal trap (Z = 2.5, P = 0.01), and that of the McPhail trap (Z = 2.7, P = 0.008). However, the proportions of females trapped in the clear artisanal trap and the McPhail trap were not different (Z = 0.099, P = 0.92) (Fig. 3).

Fig. 2.

Total number of fruit flies caught for each trap model in a mango orchard in Haiti. Bars labeled with the same letter were not significantly different (ANOVA, Fisher LSD α = 0.05).

Given the higher number of fruit flies captured by the McPhail traps compared with the clear artisanal traps in the first experiment, we used 36 AT2 traps per ha and 24 MP traps per ha for the second experiment. During the 7-week mass-trapping period, 236 and 239 fruit flies were captured using 36 AT2 per ha and 24 MP per ha, respectively. This difference in the total number of flies caught was not significant (Z = 0.13, P = 0.73) (Fig. 4).

The analysis of the number of fruit flies caught every week using both trap densities (24 MP per ha and 36 AT2 per ha) showed that the number fruit flies increased as mango fruits matured (Fig. 4). Fruit fly densities were low initially, because food and water sources in the orchard were limited during this driest season of the year and the beginning of the mango production season. The number of fruit flies captured related to trap density was borderline significant (F = 3.64, df = 2, 15, P = 0.051; Table 1; Fig. 5). Moreover, the parameter estimates (i.e., asymptote = 80 ± 27, curve = -0.11 ± 0.11, shift = -3.75 ± 2.7) (Table 2) showed a significant correlation that permitted development of a predictive equation to assess the relationship between the number of fruit flies caught and the trap density: y = 80^{(10.112(x - 3.745)}). The inflection point for the rate of fruit flies caught per additional traps occurred at 24 to 25 traps per ha (Fig. 6). Few additional fruit flies were captured at higher densities.

Fig. 3.

Average proportion of female and male fruit flies caught per trap model in a mango orchard in Haiti. Bars labeled with identical letters were not significantly different after comparisons among trap models (ANOVA, Fisher LSD α = 0.05).

Fig. 4.

Total number of fruit flies caught at 24 McPhail per ha (MP/ha) and 36 clear artisanal trap per ha (AT2/ha) in a mango orchard in Haiti. Bars labeled with the same letters were not significantly different (Binomial test for equal proportion, α = 0.05).

Discussion

Our data demonstrated that 36 clear artisanal traps per ha performed as well as 24 standard McPhail traps per ha but at a much lower cost. Furthermore, a density of 25 traps per ha was optimal for mass trapping, as found for previous studies in different orchard crops in different locations (e.g., IAEA 2003, Martinez-Ferrer et al. 2012). This density corresponds to the current recommendation for mass trapping fruit flies in Haiti.

Certain intrinsic characteristics of each trap model could have contributed to the observed difference in number of fruit flies caught between the 3 trap models. First, the entrance hole in the McPhail trap (6 cm diameter) is 6 times larger than the entrance hole in the artisanal trap models (AT1 and AT2 are 1 cm diameter), which might facilitate a better diffusion of the attractant and entrance of the flies (F. M, personal observation). Second, in addition to the chemical cues released from the food bait, the yellow bottom in the McPhail and the AT1 is considered a visual cue that attracts flies when they approach the traps (Aluja & Rull 2009, Cytrynowicz et al. 1982, Sivinski 1990). This coloration plays a key role in the McPhail traps effectiveness by guiding flies through the entrance hole in the underside of the trap whereas it may have had a negative effect on the performance of the yellow artisanal trap by distracting flies from entering the holes, which were not located in the yellow portion of this trap. Furthermore, the volume of the McPhail trap (3.4 L) is approximately 6 times larger than the artisanal trap models (0.59 L), which might prevent flies from escaping the former after entering (F. M., personal observation). These factors likely contributed to the larger number of flies in the McPhail trap compared with the clear artisanal trap and to the higher number of flies in the clear artisanal trap compared with the yellow-bottomed artisanal trap.

Table 1.

Regression analysis (based on data transformation with the Gauss- Newton iteration method: negative exponential) for the effect of trap density on fruit fly capture in a mango orchard in Haiti.

Fig. 5.

Numbers of fruit flies caught by trap density in a mango orchard in Haiti.

Our results indicated that the population of A. obliqua might be much higher in this mango growing area of Haiti than that of A. suspensa, based on the ≤1% representation of A. suspensa in samples. Previous studies have shown that the McPhail trap is effective against both fruit flies species (Burditt 1982). Furthermore, we confirmed that all 3 trap models are biased toward females, as shown in earlier studies (Lasa et al. 2014a). Female fruit flies need more protein than males, because of their requirements for oviposition (Aluja & Rull 2009, Bateman 1972). As a result, more females are typically attracted to sources of protein (e.g., Torula yeast) than males (Aluja & Rull 2009, Bateman 1972).

A comparison of the construction costs of these 3 trap models demonstrated that the artisanal models are much less expensive than the McPhail model. The cost of 20 McPhail traps used for the experiment (US $220) is 3 times the cost of 20 yellow-bottomed artisanal traps (US $75) and more than 3 times the total cost of 20 clear artisanal traps (US $70) (Table 1). These results demonstrated that the clear artisanal trap (AT2) was the most cost-effective model, followed by the yellow artisanal trap model (AT1). The importance of inexpensive traps was emphasized by Lasa et al. (2014b). Given the number of traps necessary for mass trapping, the cost per trap becomes a primary criterion for determining the economic injury level (Stern et al. 1959). If the cost of a management plan is too high, crop yield and market value may be insufficient to warrant its use. In Haiti, mass trapping with commercial McPhail traps requires a prohibitively-expensive number of traps per ha, making a fruit fly control program financially unfeasible for small growers. Fortunately, our data demonstrate that the clear artisanal trap model (AT2) is an efficient, but less expensive, alternative to the McPhail trap.

Table 2.

Regression analysis of parameter estimates for a model of the effect of trap density on fruit fly capture in a mango orchard in Haiti.

Fig. 6.

Predicted number of fruit flies caught relative to trap density in a mango orchard in Haiti during Jan 21 to Mar 3, 2015.