Modelling response of Delottococcus aberiae (De lotto) male captures to pheromone emission rates

Abstract

BACKGROUND

An invasive mealybug from South Africa identified as Delottococcus aberiae (De Lotto) (Hemiptera: Pseudococcidae) is causing severe distortions and fruit size reduction in citrus areas of eastern Spain. The sex pheromone of this insect was recently identified and is being used as a powerful tool for pest monitoring and early detection purposes. These monitoring pheromone dispensers have been designed based on the pheromone loads used for other mealybug species due to the lack of specific studies. In the present work, to evaluate the response in male captures of different emission rates, several experimental dispensers have been prepared to release D. aberiae sex pheromone at different rates. Male captures of each emission rate were obtained through a randomized block design trial conducted under field conditions by installing cardboard sticky traps baited with the different experimental dispensers. Non‐linear regression was then employed to model the number of male captures according to the different emission rates.

RESULTS

The chromatographic studies revealed that the emission rates of the experimental dispensers under field conditions ranged from 0.09 ng/day to 1.71 mg/day. The number of captured males of D. aberiae increases with increasing doses until reaching a maximum threshold, which ranges from 0.4 to 40 μg/day. Exceeding emission levels result in a decline in captures.

CONCLUSIONS

For the first time, the emission levels to maximize the captures of D. aberiae males was described. The decrease on the male captures in higher emissions could be related to the triggering of non‐competitive mechanisms of mating disruption. © 2025 The Author(s). Pest Management Science published by John Wiley & Sons Ltd on behalf of Society of Chemical Industry.

Using increasing doses of the sex pheromone of D. aberiae, our model revealed that the emission levels to maximize the captures of males ranges from 0.4 to 40 μg/day.

1. INTRODUCTION

Pheromones offer significant advantages in pest management strategies, including high selectivity for target organisms and the lack of detrimental effects on the environment, especially when employed in confined devices or dispensers. ¹ , ² , ³ However, the use of these products for pest management is sometimes prohibitive due to the pheromone costs, which can sometimes account for nearly 90% of the dispenser's cost. ⁴ Consequently, following the identification of an insect pheromone, the study of its optimum emission rate becomes especially relevant in order to select and develop a pest management strategy. ⁵ In general, the costs associated with mating disruption are higher compared to those of a mass trapping or lure‐and‐kill strategies due to the higher pheromone amounts required. ⁶ , ⁷ , ⁸ The optimal emission rate of a pheromone as an attractant has been elucidated for many lepidopteran species in order to design mass trapping, attract‐and‐kill, or monitoring dispensers. ⁹ , ¹⁰ , ¹¹ , ¹² , ¹³ , ¹⁴ These studies additionally revealed that an optimum pheromone emission rate can be identified, so that rates above or below the optimum can reduce the total number of captures. ⁹ , ¹⁰ , ¹¹ , ¹² , ¹³ , ¹⁴

Delottococcus aberiae (De Lotto) (Hemiptera: Pseudococcidae) is an invasive mealybug pest in Mediterranean citrus in eastern Spain, native to South Africa. ¹⁵ Since its identification in 2009, this mealybug has caused serious damage to citrus production, as a result of the commercial depreciation of distorted and size‐reduced fruits. ¹⁵ In severe attacks, D. aberiae can produce decayed trees characterized by a reduction in leaf mass and the appearance of chlorotic leaves. According to the Farmers Association of Valencia (AVA‐ASAJA), losses in 2020 were estimated in more than 110 million €. ¹⁶ Since the initial detections of this pest in the north of the province of Valencia in 2009, its presence has extended rapidly, colonizing other citrus‐growing areas of eastern Spain. ¹⁷

Due to the typical biology and cryptic behavior of mealybugs, ¹⁸ there were serious difficulties in detecting and determining the extent of D. aberiae infestations by conventional methods such as visual inspection and monitoring for honeydew residue. ¹⁹ , ²⁰ Consequently, the development of monitoring trapping systems based on pheromones would offer increased sensitivity and reduce significantly the need for tedious manual sampling. ²¹ , ²² Pheromones are particularly valuable in areas where multiple mealybug species coexist. The sex pheromone of D. aberiae was recently identified as the irregular monoterpene (4,5,5‐trimethyl‐3‐methylenecyclopent‐1‐en‐1‐yl)methyl acetate and its activity has been tested in both laboratory and field experiments. ²³ Given the high efficacy of this novel pheromone in attracting males of D. aberiae in citrus orchards, this substance is a promising candidate for pest control. Since 2019, the pheromone has been employed for detection and monitoring networks throughout the citrus‐growing area of the Valencian Community, ¹⁷ whereas direct control methods are in the beginning stages.

As demonstrated specifically for other coccoid pests, ¹⁸ , ²⁴ the optimization of the pheromone emission rate can significantly increase the number of captures in traps, thus increasing the efficacy of pest monitoring and pest control methods. The main objective of the present study was to determine the optimum emission rate that maximizes the number of trapped males. For this purpose, the captures of D. aberiae males according to different emission rates of its sex pheromone was investigated under field conditions. The emission rates of experimental dispensers designed for the tests were accurately determined by chromatographic techniques. The trapping efficacy of the different pheromone emission rates was assessed through a randomized block field trial. The number of captures of D. aberiae males according to the different emission rates was finally modelled by non‐linear regression in order to obtain the optimum emission rate leading to an estimated maximum of captures. Accordingly, the potential of mating disruption for controlling this pest is also discussed.

2. MATERIALS AND METHODS

2.1. Experimental sex pheromone dispensers

The experimental dispensers employed to release D. aberiae sex pheromone in the field trials consisted of 2‐mL gas chromatography (GC) amber glass vials with crimp top (Thermo Fisher Scientific, Waltham, USA). The vials were loaded with a solution of the sex pheromone (rac‐(4,5,5‐trimethyl‐3‐methylenecyclopent‐1‐en‐1‐yl)methyl acetate)) in decyl acetate, and sealed with a crimp cap equipped with a polymeric membrane, through which the pheromone solution is emitted by diffusion at 2 mg/day. Each vial was loaded with 0.5 mL of a solution with different concentrations of the pheromone dissolved in decyl acetate: D0 (only decyl acetate), D1 (0.07 μg/mL), D2 (6.67 μg/mL), D3 (0.67 mg/mL), D4 (0.07 g/mL), D5 (0.17 g/mL). This solvent was chosen given the similar chemical properties compared with the sex pheromone. The formulations with the different pheromone concentrations were previously prepared in the laboratory to obtain an expected theoretical emission rate for the formulation of 2 mg/day. According to the concentration of pheromone in the different formulations, the theoretical expected emission rates for the field assay ranged 0–500 μg/day.

2.2. Field trial

The field trial was conducted in a 1‐ha mandarin orchard (Citrus reticulata, cv. Clemenules), located in the municipality of Vila‐real (Castellón, Spain) (39° 55′ 19.34″ N, 0° 8′ 18.51″ W). The experimental design was based on six experimental blocks that consisted of rows of at least 21 trees, in which traps baited with the dispensers loaded with the different pheromone formulations were hung with a separation of 12 m, being the distance between blocks of at least 18 m, two rows (Fig. 1). The traps employed were white sticky cardboards of size 95 × 150 mm (EPA SL, Carlet, Spain), which were hung on the trees at 1.5 m height on the west side of the canopy. Each block comprised seven different emission rates: D0 (trap baited with a dispenser only with the solvent decyl acetate), D1 (trap with one D1 dispenser), D2 (trap with one D2 dispenser), D3 (trap with one D3 dispenser), D4 (trap with one D4 dispenser), D5 (trap with one D5 dispenser) and D6 (trap with two D5 dispensers).

Figure 1.

Schematic distribution of the experimental design and distances between each block (1 to 6) and each trap (red dots) in the experimental plot.

The dispensers were installed upside down tied to the center of the traps to release the pheromone through the polymeric membrane. The experiment started on June 6th 2023, and lasted 6 weeks. Once a week, the traps and dispensers were collected and rotated within blocks. Next, the cardboards were taken to the laboratory to count male catches under a stereomicroscope for the correct identification of individuals.

2.3. Dispenser release kinetic studies

2.3.1. Field aging and sample preparation

Pheromone emission rates from the experimental dispensers were quantified simultaneously to the field trial. Twenty‐four additional dispensers of each formulation were installed 100 m apart to the trial orchard (39° 55′ 22.10″ N, 0° 8′ 18.87″ W), with the aim of exposing them to the same environmental conditions and to collect the aged dispensers periodically. Four dispensers of each type were collected fortnightly and taken to the laboratory to quantify their residual pheromone load as a basis to determine their release profile. For the quantification, the content remaining on each dispenser was diluted in 10 mL toluene and stored at −20 °C until their analysis.

2.3.2. Pheromone quantification

Given the wide range of pheromone concentrations in the different dispenser formulations, two different procedures were employed for pheromone quantification. The higher dose dispensers (expected emission rates >10 μg/day) were analyzed by gas chromatography using a Clarus 590 GC (PerkinElmer Inc., Wellesley, MA, USA) equipped with a flame ionization detector (GC/FID). The solution that remained in each dispenser was diluted to 10 mL of with toluene; then, 25 μL of an n‐dodecane solution were added as internal standard to 1 mL of each sample. The resulting solutions were injected onto a ZB‐5MS column (30 m × 0.25 mm × 0.25 μm) (Phenomenex Inc., Torrance, CA). The oven of the GC was held at 130 °C for 3 min; then, raised by 1.5 °C/min up to 180 °C and finally raised at 30 °C/min up to 280 °C, held for 5 min. The carrier gas was helium at 1.5 mL/min. The amount of pheromone was estimated from their corresponding chromatographic areas by fitting a linear regression model, y = a + b·x, where x is the ratio between pheromone and n‐dodecane areas and y is the amount of pheromone.

The pheromone quantity remaining in the lower dose dispensers (expected emission rates <10 μg/day) was analyzed using a TSQ 8000 Evo triple quadrupole mass spectrometer (GC–MS/MS) coupled with a Thermo Scientific™ TRACE™ 1300 GC because pheromone concentrations were close or below the quantification limit of the GC/FID equipment. In this case, the internal standard employed was heptyl 4,4,5,5,6,6,7,7,8,8,9,9,9‐tridecafluorononanoate (TFN) to improve both sensitivity and selectivity for MS/MS method optimization. ²⁵ Twenty‐five μL of TFN were added to 100 μL of each dispenser sample, and 1 μL of the resulting solution was injected onto a ZB‐5MS (30 m × 0.25 mm × 0.25 mm) fused silica capillary column (Phenomenex Inc., Torrance, CA, USA). The oven was held at 60 °C for 1 min, then was raised by 10 °C/min up to 110 °C, maintained for 5 min, raised by 3 °C/min until 150 °C and finally raised by 35 °C/min up to 300 °C and held for 5 min. The carrier gas was helium at 1.5 mL/min. The MS/MS instrument operated in SRM (selected reaction monitoring) mode using electron ionization (EI +). For each target compound —internal standard TFN and the 4,5,5‐trimethyl‐3‐methylenecyclopent‐1‐en‐1‐yl)methyl acetate — the MS/MS method was optimized by selecting the precursor ion and the product ions that provided the highest selective and sensitive determinations (Table 1). Pheromone transition m/z 119 to 91 was the one employed to obtain the chromatographic area; the other transitions were monitored for confirmatory purposes. The amount of pheromone was estimated as a function of the corresponding chromatographic areas by fitting a linear regression model, y = a + b·x, where x is the ratio between the pheromone and TFN areas and y is the amount of pheromone.

Table 1.

Optimum values of the MS/MS parameters for each target compound.

Compound	Precursor Ion (m/z)	Product Ion (m/z)	Collision Energy (eV)
Pheromone	105	79	5
	119	91	10
	121	93	10
TFN*	393	373	5

^*Heptyl 4,4,5,5,6,6,7,7,8,8,9,9,9‐tridecafluorononanoate (TFN).

2.3.3. Pheromone release profiles

For each type of experimental dispenser, simple regression was applied to fit the residual pheromone loads (ng) quantified by GC, FID or MS/MS, as a function of the corresponding ageing period (days). The first derivative of the resulting linear equations (i.e., slope of the fitted line) accounts for the mean pheromone release rate of each type of dispenser per day.

2.4. Trap captures

After counting the number of captures in the traps at laboratory, statistical differences in the number of males captured per traps and week (MTW) by each dose were investigated by means of a multi‐factor ANOVA, considering the factors dose, week and block, followed by post‐hoc Fisher's least significant difference (LSD) test at P ≤ 0.05. Prior to the study, data were log‐transformed, log(x + 1), to normalize the distributions and to homogenize variance.

2.5. Dose–response study

A series of different preliminary studies by means of ANOVA, followed by LSD tests were conducted in order to identify the most suitable type of transformation for the catches. This transformation enabled the normalization of residuals, the identification of any data anomalies, as well as the study of double interactions between the different factors (dose, block, and week). Following this preliminary statistical study, a non‐linear regression model was chosen to fit the number of captures obtained experimentally as a function of the pheromone dose. Finally, a study of variability for the optimal dose was conducted. These statistical analyses were conducted using Statgraphics Centurion 18 (Stat Point Technologies, Warrenton, VA, USA) and Excel 365 (Microsoft Corporation, Redmond, WA, USA).

3. RESULTS

3.1. Pheromone emission rates of the dispensers

The release profiles of all the experimental dispensers yielded a strong goodness‐of‐fit according to the fitted linear models, which indicates that they provided a rather constant mean emission rate throughout the period under study. Therefore, the slope of the resulting equations (Fig. 2) quantifies the mean pheromone release rate for each type of dispenser: 1.09 ng/day, 15.50 ng/day, 3.26 μg/day, 214.60 μg/day and 853.41 μg/day, for dispensers coded as D1, D2, D3, D4, and D5, respectively. For traps baited with two D5 dispensers (D6), which provided the highest emission rate, a double emission rate was assumed (i.e., 2 × 853.41 μg/day = 1.71 mg/day).

Figure 2.

Pheromone release profiles of the dispensers. Each emission rate was estimated as the decrease of the residual pheromone load contained in the dispensers (±SE) throughout the time (days) of field exposure (i.e., slope of the fitted line).

3.2. Trap captures

The factor dose significantly affected captures (F = 34.3; df = 6, 209; P ≤ 0.001). In the traps baited only with the solvent (D0), a mean of 40.64 ± 9.00 males per trap and week (MTW) were found. Among the lower doses, from D1 to D3, an increasing trend of MTW was found: 23.75 ± 8.65, 48.53 ± 10.78, and 201.42 ± 29.59 MTW, respectively. For the dispensers emitting more pheromone (D4, D5 and, D6) a decreasing trend on the captures was observed (47.11 ± 12.20, 16.58 ± 3.29, and 7.34 ± 0.83 MTW, respectively), with significantly lower MTW for D5 compared to D1. Additionally, the multifactor ANOVA revealed a significant effect by the factors week (F = 4.24; df = 5, 209; P ≤ 0.010) and block (F = 5.57; df = 5, 209; P ≤ 0.001).

3.3. Dose–response study

3.3.1. Preliminary analysis with ANOVA

The resulting data of trap captures were subjected to multifactor ANOVA. Three factors were considered: dose, block and week. The dose factor comprised seven levels (D0 to D6, from lowest to highest concentration of pheromone). The block factor comprised six variants, coded as B1 to B6. Finally, the weekly factor comprised six variants, W1 to W6. Traps were placed in the field on June 6th 2023. Consequently, code W1 corresponds to one week after this date, and so on until W6 for the captures collected 6 weeks later.

The total number of combinations between these three factors is: 7 × 6 × 6 = 252, which would be the total number of values (captures) for the statistical analysis. However, 29 out of these 252 values could not be obtained, primarily due to the significant losses of D5 dispensers.

The statistical analysis by means of ANOVA with the aforementioned factors, including the double interactions and excluding the data from D5, revealed that the main effect of the three factors was statistically significant (P ≤ 0.010), as well as the double interaction between dose and week (P ≤ 0.0001).

To check the normal distribution of the residuals, a histogram was plotted (Fig. 3(a)). This plot revealed a leptokurtic distribution (i.e., sharper than the Normal distribution), with the standardized kurtosis coefficient KC_std = 29.50. This result is attributed to the heterogeneity in the data variance (i.e., heteroscedasticity), because higher values of captures also imply a greater variability.

Figure 3.

Histograms of residuals from ANOVA (carried out with factors week, block and dose), considered as dependent variable: captures (a), square root of captures (b), or fourth root of captures (c). The density function of the normal distribution is also depicted for comparison and to discuss the degree of kurtosis.

Attempting to fulfill the normality of residuals, an additional ANOVA was conducted employing the square root of the captures. Residuals were once again saved, and represented in a histogram (Fig. 3(b)), which persisted in revealing a deviation from a normal distribution (KC_std = 8.17). Consequently, a subsequent ANOVA was carried out with the fourth root (i.e., captures^0.25). In this case, the histogram of residual data suggested a reasonable normal pattern (Fig. 3(c)) with KC_std = 1.93, comprised between −2 and 2, which implies that data can be regarded as mesokurtic (i.e., the case in samples extracted from a normal population). Therefore, this transformation seems adequate to fulfill the hypotheses of normality and homoscedasticity.

Fig. 4 shows a similar pattern between the different blocks: A maximum was reached with dose D3, and the next higher dose D4 significantly reduced the number of captures. However, a slightly different pattern is observed for block B6, because a similar quantity of catches was obtained with D3 and D4. Consequently, it was decided to exclude the data from block B6, as its pattern of captures was clearly different from the rest.

Figure 4.

Interaction plot with 95% LSD intervals corresponding to the interaction between dose and block, for the multifactor ANOVA fitted with the fourth root of captures.

After discarding data from B6 and repeating the ANOVA using the fourth‐root transformation, the double interaction was not statistically significant (P > 0.2), simplifying the interpretation of the results. In Fig. 4, which represents a means plot with LSD intervals for block and week, it can be observed that LSD intervals corresponding to blocks B3, B4 and B5 overlap with each other, indicating that their means did not differ significantly. By contrast, B2 exhibited a significantly higher mean compared to B4 and B5, at a significance level of 5% (Fig. 5(a)). Likewise, the means plot for the week factor shows an overlap for the LSD intervals corresponding to weeks 2, 4, and 5 (Fig. 5(b)). According to these results, the subset of data from blocks 3, 4, and 5, corresponding to weeks 2, 4 and 5, can be considered as a homogeneous set, comprising 57 values (i.e., 25.6% of the total). This subset was subsequently used to develop a non‐linear regression model able to fit the number of captures as a function of the pheromone dose.

Figure 5.

Means plot with 95% LSD intervals for the factor block (a) and week (b), from the ANOVA carried out with three factors (block, week, and dose), using the fourth‐root transformation of captures. Data of dose D5 and block B6 were excluded.

3.3.2. Statistical analysis with non‐linear regression

In order to develop a non‐linear regression model suitable for this case, it seemed convenient to visualize in a scatterplot the fourth root of captures for the subset of 57 values as a function of the dose. Unfortunately, such plot did not provide clear information because the range of pheromone doses tested experimentally was unevenly distributed, not equidistantly, in a somewhat logarithmic manner: 1.1 ng/day (D1), 15.5 ng/day (D2), 3.3 μg/day (D3), 214.6 μg/day (D4), 853.4 μg/day (D5), and 1706.8 μg/day (D6). This fact implied a drawback for regression models, as the capture values at the highest dose would presumably act as influential points for the estimation of parameters. Moreover, it would be necessary to use an asymmetric mathematical function with enough parameters to be fitted conveniently to the observed pattern. Attempting to avoid these drawbacks, a transformation to the emission dose was applied (Fig. 6).

Figure 6.

Scatterplot of captures as a function of pheromone dose^0.1, for a subset of 57 homogeneous data, not showing differences between week nor experimental block. The dependent variable (vertical axis) is the fourth root of captures (a) or the square root (b). The sequence of blue dots corresponds to the fitted non‐linear regression model (Eqn [1]).

After performing several tests, the transformation that appeared to be most convenient was raising the independent variable to the power of 0.1 (i.e., dose^0.1). This transformation ensured that the dose D3 remained approximately equidistant from both D2 and D4, which is convenient because D3 was the dose leading to the maximum number of captures among the experimentally tested doses (Fig. 6). As appears in Fig. 5(a), the fourth‐root number of captures for the subset of 57 homogeneous values yielded a symmetrical pattern of captures as a function of dose^0.1, suggesting the suitability of a non‐linear model with a symmetrical shape, such as a Gaussian bell curve.

A Gaussian equation with a horizontal asymptote would seem adequate, because zero captures were expected for a null dose of pheromone. However, Fig. 4 shows that the number of captures was significantly higher than zero for the control dose D0 (without the pheromone). The reason is that the solvent (decyl acetate) used for the D0 control yielded a certain attractant effect, something which was not expected. Consequently, the regression model should predict a non‐zero value of captures for Dose = 0 and should asymptotically approach zero for a dose tending to infinity. Based on these requirements, the function proposed is the following, being the independent variable x the dose of pheromone raised to 0.1.

$${\mathit{\text{captures}}}^{0.25}=\mathit{a}·e^{-\mathit{b}·x}+\mathit{c}·e^{-\mathit{d}·{\left(x-\mathit{m}\right)}^2}$$ (1)

This mathematical function is based on five parameters with a rather straightforward interpretation. The dependent variable (Y) is estimated as the sum of two terms: The first one takes the value ‘a’ when x = 0, and decreases exponentially for higher values. Therefore, ‘a’ would be approximately the average value of Y for a null dose of pheromone. The parameter ‘b’ is related to the speed of descent of this asymptote. The second term is the mathematical equation of a Gaussian bell curve, being ‘m’ the axis of symmetry. This is the most important parameter of Eqn (1), because ‘m’ would be nearly coincident with the value of x leading to a maximum number of captures. Finally, this Gaussian bell curve is characterized by two parameters: ‘c’ and ‘d’. The former is related to the size of the bell curve, so that a higher value leads to a greater value of Y. The latter is inversely related to the width of the bell curve.

This model was implemented in a spreadsheet of Microsoft Excel, and the optimization tool Solver was used to find the values of these five parameters leading to the minimum value of the sum of squared residuals, for the subset of 57 homogeneous captures. The resulting values of these parameters that yielded the best fit are indicated in Table 2.

Table 2.

Values of the regression coefficients in Eqn (1) that lead to the best goodness‐of‐fit of the model to estimate the number of captures, considering the subset of 57 homogeneous values

Parameter in Eqn (1)	Fourth root of captures	Square root of captures
a	1.64	3.065
b	0.064	0.106
c	1.405	5.505
d	5.62	6.337
m	1.25	1.20

The number of captures was transformed in two different ways.

As mentioned above, ‘m’ is the axis of symmetry of the Gaussian bell curve, which closely corresponds to the value of x that yields the relative maximum. Therefore, m = 1.25 = dose^0.1. To undo the transformation, this value is elevated to the tenth power, resulting in an optimal dose of 1.25¹⁰ = 9.4 μg/day. For the 57 values (Fig. 6(a)), the coefficient of determination is R² = 37.4%, indicating a moderate goodness‐of‐fit. The same procedure was applied to the square root of captures. The regression coefficients are also indicated in Table 2. In this case, the parameter m becomes nearly the same, leading to an optimal dose of: 1.20¹⁰ = 6.2 μg/day.

Fig. 6a shows that the fourth‐root transformation yields a more symmetrical distribution of the observed points around the fitted curve (i.e., the distribution of residuals tends to be normal, as also obtained with ANOVA). However, a slightly positive asymmetrical distribution of the residuals is observed with the square root (Fig. 6(b)) though, curiously, this transformation yields a slightly higher goodness‐of‐fit (R ² = 42.3%). Given that both methods basically lead to the same optimal dose, it was decided to proceed with the fourth root in the next section, as the normality of residuals is required in least‐squares regression models.

3.3.3. Study of variability for the optimal dose

The results obtained by fitting the model to the subset of 57 values provide a preliminary estimate of the optimal pheromone concentration at which the captures are maximized. However, the model still needs to be applied to the remaining experimental values, and it is also of interest to obtain a confidence interval for that optimal concentration. For this purpose, the model was fitted to each subset of data available in each block and each week, with the exception of block B6 (i.e., the fourth‐root of captures corresponding to the seven pheromone levels: D0 to D6). The Solver optimization tool implemented in Microsoft Excel was used to compute the regression coefficients yielding the best fit. The parameters ‘a’ and ‘b’ were kept constant, setting the optimal values obtained with the subset of 57 data (a = 1.64 and b = 0.064), so that the three parameters adjusted with Solver were ‘c’, ‘d’, and ‘m’.

Using this procedure, 30 values of m were obtained (i.e., 5 blocks × 6 weeks). By plotting these values on a normal probability plot, a good fit was found to a normal distribution, with a mean = 1.1741 and standard deviation s = 0.1314. This result is consistent with the mean value of 1.25 estimated with the 57 captures (Table 2). This mean value = 1.1741 = (optimal dose)^0.1. To undo the transformation, the value must be raised to the tenth power: 1.1741¹⁰ ≈ 5 μg/day.

In a normal distribution, the interval between the mean ± 2·s comprises approximately 95% of the data. In this case: 1.1741 ± 2 · 0.1314 = [0.911; 1.437]. Raising these values to the tenth power, it turned out that the optimal dose of pheromone emission at which the maximum captures are achieved, with a confidence level of 95%, would be 0.39 to 37.5 μg/day, which can be rounded up to the range of 0.4 to 40 μg/day.

Finally, to determine whether the regression coefficient ‘m’ significantly varied across the different blocks or weeks, a multifactorial ANOVA was carried out. This analysis revealed that the main effect of block was not statistically significant (P = 0.120), and the same turned out for the effect of week (P = 0.480). This result increases the robustness of the confidence interval obtained for ‘m’.

4. DISCUSSION

In this study, the relation of the sex pheromone emission rate and male captures for D. aberiae mealybug has been described for the first time. After a new pheromone is identified and synthesized, to gain knowledge on the emission value to evaluate the response of the male insect to maximize the captures is a key element to develop and optimize tools for an effective control strategy. ⁵

In view of the results and the statistical analysis carried out with the male captures obtained by all the doses tested, we have determined a correlation between D. aberiae sex pheromone emission rates and the number of male captures using a set of dispensers that provided different emission rates of the sex pheromone.

Results show a similar trend in all blocks during the whole trial period, which suggest that no overlapping effect of traps with high emission rates between neighboring traps was produced. This is a key point for the study because we need to reach a compromise between the distances of the traps: close enough to avoid spatial variations due to the clumped pest distribution but far enough away to avoid trap overlapping.

Not expected but not surprisingly, the solvent used to prepare the pheromone formulations elicited a response to D. aberiae males comparable to the lowest dose of pheromone tested, which can be rationalized by the fact that decyl acetate is a common volatile compound found in many vegetal species, particularly in citrus species. ²⁶ , ²⁷ Thus, D. aberiae males could be slightly attracted by this compound, emitted in much higher doses compared with the sex pheromone. Despite this, the dosage values that elicited the highest male captures would range from 0.4 to 40 μg/day, with a mode around 5 μg/day.

A limited number of works have studied the effect of different doses and release rates on male captures for hemipterans. Specifically, there is only one study available for two stink bugs, ²⁸ several studies about other mealybugs ¹⁸ , ²⁹ , ³⁰ , ³¹ and one available for the scale Aonidiella aurantii Maskell (Hemiptera: Diaspididae). ²⁴ In the A. aurantii study, the relative maximum of male captures at which captures begin to decrease corresponded to a release rate of approximately 300 μg/day, ²⁴ a higher value than the one obtained in this study for D. aberiae.

In the case of mealybugs, a saturation effect was observed for doses between 200 and 800 μg in Planococcus citri (Risso) studies, but with no reduction registered in male captures in the higher tested doses. ³² Millar et al. did not find differences between the captures of Planococcus ficus (Signoret) males obtained by different rubbers impregnated with increasing doses of their pheromone, from 10 μg to 1 mg, and suggested mealybug insensitiveness to pheromone concentration from a threshold. ¹⁸ However, these authors worked with pheromone loads but they did not estimate the actual emission rates in field trials. According to the abovementioned study of Vacas et al., it was found that a rubber septa impregnated with 3 mg of pheromone reached a constant emission of 29.75 μg/day over a 6 week period. ²⁴ With this information, it could be roughly estimated that 1 μg of pheromone load would represent the release of 0.01 μg/day. This rough estimation could suggest that Millar et al. emission rates would range from 0.1 to 10 μg/day, values within the lower emission rates tested in our study. Zada et al. studied the effect of increasing sex pheromone doses in P. citri male captures and found that male captures increased statistically with increasing release rates from 0.25 to 2 μg/day and did not change when pheromone release rates increased from 2 until 10 μg/day. ³¹ A similar study carried out by Flores et al. found that males of Pseudococcus calceolariae (Maskell) were similarly attracted to rubbers with 10, 30, and 50 μg of pheromone; only a dose of 100 μg increased significantly the male captures in the field. ²⁹ Waterworth et al. studied the effect of increasing doses of racemic pheromone on trap captures in two different species, Pseudococcus longispinus (Targioni Tozzetti) and Pseudococcus viburni (Signoret), impregnating gray rubber septa. ³⁰ The study revealed an increase in captures for P. viburni in doses ranging from 1 to 10 μg, with an increasing trend until the highest tested dose, 320 μg. For P. longispinus, an increasing trend was observed from 1 to 3.2 μg, and the increase was kept until the highest dose, 320 μg. In the latter, despite the absence of a maximum dosage threshold capable of decreasing the captures, a comparison of old versus fresh grey rubber septa revealed that fresh dispensers reached fewer captures than 2, 4, 8, and 12 aged lures. The authors attributed this finding to a deterrent effect carried out by higher doses/release rates.

Miller and Gut described different mechanisms able to trigger two types of Mating Disruption (MD), competitive MD and non‐competitive MD. While competitive MD is triggered when a pheromone source competes with a female, non‐competitive MD triggers when there are physiological impairments in males, the females or the signal and the males become insensitive to the pheromone sources. ³⁵

In the case of Lepidopterans, in this type of study the reduction of male captures with increasing pheromone doses has been suggested to be a consequence of the triggering of non‐competitive MD due to the disorientation of males due to their desensitization. ⁹ , ¹⁰ , ¹¹ , ¹² , ³³ , ³⁴ In the case of hemipterans, Vacas et al. suggested a relationship of such concentrations to mating disruption doses used for MD of A. aurantii using mesoporous dispensers. ²⁴ Franco et al. stated that competitive MD are most likely to occur in scales, although there are no specific studies. ³⁶ Trials conducted with P. calceolariae found no experimental evidence of habituation in males and a decrease in female detection in a competitive trial without a decrease of male searching activity, suggesting competitive disruption. ³⁶ , ³⁷ Nevertheless, the reduction of male captures to near‐zero close to of a single high release rate pheromone source do not allow us to discard the role of non‐competitive MD.

Although this statement must be demonstrated with deeper and more specific studies, this study would provide the first evidence of the disorientation effect in mealybugs. In fact, the absence of this effect in the mealybug studies abovementioned could be explained by the lack of sufficiently high doses used in those studies. Most of the aforementioned studies conducted to relate different pheromone doses to the number of captures in mealybugs were carried out using rubber septa dispensers. Despite their utility in monitoring insect populations, the lifespan and lack of regulation parameters of this type of dispenser made this technology not accurate enough for the development of successful control methods such as mating disruption, mass trapping or attract and kill.

The development of devices that better allow the regulation of the sex pheromone emission in our study by the use of liquid formulations, facilitate the use of more precise emission levels, enhancing the accuracy and the information provided. It is important to highlight that the attractive effect exerted by the solvent used in the present work, together with a polymeric membrane used to regulate the emission rates, implies that the results of this work should be validated with another solvent before being considered as absolute values. On the other hand, it is of interest to keep in mind that the pheromone used in this study was a racemic mixture, being nowadays the only cost‐effective product for implementing these techniques. Possibly, the use of only the natural enantiomer emitted by females will determine if our emission rates are equally valid, due to the different effects exerted by both enantiomers separately. ³⁸

In conclusion, this study reports, for the first time, the effect of different emission levels of the sex pheromone on male captures for D. aberiae. The results indicate that sex pheromone emission rates between 0.4 to 40 μg/day of the designed devices maximize captures and clearly indicate a threshold from which males' captures decline, pointing to a possible disorientation effect that could be used for triggering mating disruption in this species.

AUTHOR CONTRIBUTIONS

V.N.‐L., S.V., and I.N.F conceived and designed the research. J.M.B. synthetized and supplied the pheromone. J.M.B. and I.N.F. formulated the pheromone dispensers. V.N.‐L. S.V. and A.G. conducted the experiments. M.Z. analyzed the data. A.G. wrote the original draft of the manuscript. All authors revised and edited the manuscript giving their explicit consent to submit. Funding acquisition: V.N.‐L. All authors have read and agreed to the published version of the manuscript.

CONFLICT OF INTEREST

The authors declare that they have no known competing financial interests or personal relationships that could have appeared to influence the work reported in this paper.

DATA AVAILABILITY STATEMENT

All data supporting the findings of this study that are not available within the manuscript can be requested from the corresponding authors, subject to the restrictions imposed by the project‐funding agency.