**3. Results**

#### *3.1. Fluorescein Spray Results*

In total, 547 and 488 filter papers sprayed with the automated track sprayer were included for analysis with and without overlap points, respectively, and 475 and 375 filter papers were included for the manual spray.

Heat maps plotting the distribution of fluorescein deposition on the wall surface showed a general uniformity of spray from the automated track sprayer, with the majority of recorded values falling around the mid-point colour range; some visual variation was apparent between the Left, Centre, and Right swath positions (Figure 2A). Similar uniformity of deposition was not evident in the manual wall spray, which showed high variability of spray over the entire wall surface, particularly when fluorescein deposition at the top and bottom wall positions were compared (Figure 2B).

Descriptive statistical analyses confirmed that wall spraying using the track sprayer had higher median and mean fluorescein deposition, and lower standard deviation and percentage coefficient of variations in the track spray compared to the manual spray (Table 2, SD of 36.15 and 33.87 for track spray with and without overlaps, respectively, compared to 88.83 and 69.51 for manual spray with and without overlap, respectively, and coefficient of variation of 19.96% and 19.30% for track spray with and without overlap, respectively, vs. 58.03% and 53.54% for manual spray with and without overlap, respectively).


(**B**)

**Figure 2.** Heat maps of fluorescein deposition on wall surfaces using an automated track (**A**) and manual (**B**) spray. Filter papers to collect fluorescein deposited by each spray type were attached to walls in the configuration shown. Five swaths were present on each wall, measured at the Left (L), Centre (C), Right (R), and Overlap (O) positions. Letters A–E indicate the height position down the wall. Six trials were performed for the track spray and five were performed for the manual spray. Individual cells show the corrected fluorescence values for each spray trial replicate. Orange indicates higher fluorescein values, blue indicates midpoint values, and yellow indicates low values. A far higher degree of variation in fluorescein deposition is apparent in the manual spray compared to the track spray.

Including the overlap positions resulted in a significant different dataset for manual spraying (*p* ≤ 0.0001), but not for the track sprayer dataset (*p* = 0.6998). Both sets of manual spray data were significantly different to the track spray datasets (4 comparisons, *p* ≤ 0.0001 in all cases). Analysis of each combination of vertical and horizontal swath positions showed greater variation of fluorescein deposition in the manual spray at every wall position point compared to the track sprayer.


**Table 2.** Descriptive statistics for manual vs. track spraying using fluorescein. Results are given both with and without swath overlap positions.

#### *3.2. Insecticide Spray Results*

For both pirimiphos-methyl (PMM) and broflanilide (BRF), 60 filter papers were sprayed with the track sprayer and 60 with manual spraying. As a dual AI formulation, the concentration of both deltamethrin (DLT) and clothianidin (CTD) was determined in the same filter papers; 57 for manual spraying and 60 for the track sprayer. The concentration of active ingredient on each filter paper was visualised in boxplots (Figure 3). Similar to spraying with fluorescein in laboratory conditions, variation in insecticide application rate in the experimental huts was much greater for manual spraying compared to the track sprayer. For both PMM and BRF, Levene's test reported unequal variances between manual spraying and the track sprayer (F = 3.316, *p* < 0.001 and F = 4.3533, *p* < 0.001, respectively), indicating that the variation in insecticide application rate from top to bottom and from left to right of the wall was larger when spraying manually compared to using the automated track sprayer. Likewise, for CTD and DLT, the vertical variance between the two application methods was statistically different (F = 4.1735, *p* < 0.001 and F = 4.6389, *p* < 0.001 respectively).

Two of the active ingredients, PMM (Figure 3A) and BRF (Figure 3B), showed a significant difference (*p* < 0.01 and *p* < 0.0001, respectively) in the sprayed concentration when comparing track and manual spraying. For BRF, this resulted in a lower median concentration for the track sprayer (66.0 mg/m2, SD 23.8) compared to manual spraying (91.4 mg/m2, SD 49.7), whereas for PMM the median track sprayer concentration was higher (485.8 mg/m2, SD 193.4) compared to manual spraying (392.3 mg/m2, SD 268.6). Clothianidin (CTD) and deltamethrin (DLT) were sprayed together in the combination product Fludora Fusion and, unsurprisingly, the results for the two actives followed the same pattern and trend (Figure 3C,D). The amount of CTD sprayed by manual application (177.9 mg/m2, SD 51.3) was not significantly different from the amount of CTD sprayed using the track sprayer (165.4 mg/m2, SD 104.7), *p* = 0.5288. Likewise, the amount of DLT sprayed by manual application (18.4, SD 11.0) was not significantly different from the amount of DLT sprayed by using the track sprayer (19.1, SD 5.1), *p* = 0.6217.

For each active ingredient, the amount delivered by each spray method was also compared to the target dose. For PMM the target dose is 1000 mg/m2, and both spray methods resulted in a significantly lower dose on filter papers (*p* < 0.0001). The concentration found in the liquid samples taken from the spray tanks prior to spraying was used to correct for possible mixing errors (see Table 3). This resulted in calculated concentrations of 568.5 mg/m<sup>2</sup> for manual spraying and 576.6 mg/m<sup>2</sup> for the track sprayer. Compared to the corrected target dose, manual spraying was significantly lower (*p* < 0.001), but there was no significant difference for the track sprayer (*p* = 0.07913).

**Figure 3.** Concentration of active ingredient on filter papers sprayed by manual spraying or the automated track sprayer. Filter papers were attached to plywood panels in a grid of 15 per swath. Four swaths were treated for each spray method per insecticide, resulting in 120 filter papers per insecticide. Letters A–D indicate the active ingredients; pirimiphos-methyl (**A**), broflanilide (**B**), clothianidin (**C**), and deltamethrin (**D**). Individual dots show the values for each filter paper. Boxplots indicate median, 25th and 75th percentile, extreme lines, and potential outliers. Dashed horizontal lines represent the target concentration for each insecticide.

For BRF, the amount of active ingredient sprayed on filter papers was not significantly different from the target dose for manual spraying (*p* = 0.4034, and *p* = 0.2403 for the corrected dose), but was significantly lower for the track sprayer (*p* < 0.0001). Similarly, for CTD, the dose applied to filter papers by manual spraying was not significantly different from the target dose (*p* = 0.0676, and *p* = 0.8863 for the corrected dose), but the dose applied by the track sprayer was significantly lower (*p* < 0.001, and *p* < 0.01 for the corrected dose). For DLT manual spraying was significantly different from the target dose of 25 mg/m<sup>2</sup> (*p* < 0.01) but not after correcting for the concentration in the spray tank (*p* = 0.9051). The track sprayer resulted in a significantly lower dose, both corrected and uncorrected (*p* < 0.0001).

#### *3.3. HPLC Analysis of Liquid Samples*

Samples of insecticide solutions were taken directly from the spray tank before and after spray application for both application methods to be able to detect non-homogeneous mixing. Deviation from the target concentration was calculated for each active ingredient (Table 3). The target concentrations for the liquid samples were calculated by taking the recommended dose per m2, divided by the application rate of 30 mL/m2, resulting in a target concentration of 33.33 mg/mL for PMM, 6.66 mg/mL for CLT, 0.83 mg/mL for DLT, and 3.33 mg/mL for BRF. Apart from the concentration of BRF before spraying, the concentration of active ingredient found in the spray solution was generally lower than the target concentration. The spray tank solution had a DLT concentration of 0.65 mg/mL before and 0.71 mg/mL after manual spraying (target 0.83), and a CTD concentration of 5.66 mg/mL before and 6.15 mg/mL after spraying (target 6.66 mg/mL). The average concentration of BRF was 3.44 mg/mL before and 3.06 mg/mL after spraying, both within 10% of the target concentration (3.33 mg/mL). The concentration of PMM in the tank solution was considerably lower than the target dose (33.3), ranging between 19.22 mg/mL (−36%) and 15.11 mg/mL ( −50%).

**Table 3.** HPLC results of liquid samples taken from the spray tank before and after spraying. The target dose in mg/m<sup>2</sup> is given for each active ingredient. HPLC results are represented as a percentage deviation from the target dose.


#### *3.4. Consistency of Wall Spraying across the Swath*

The nozzle delivered the most consistent fluorescein deposition from the track sprayer at the centre of the swath, displaying the lowest range of fluorescein deposited, the lowest standard deviation (13.43 vs. 24.25, 22.36 and 35.36 for Centre, Left, Right, and Overlap positions, respectively), and the lowest coefficient of variation (9.58% vs. 11.79%, 12.37% and 17.15% for Centre, Left, Right, and Overlap positions, respectively). However, significantly less fluorescein was deposited in this position compared to both the left and the right positions on the swath (*p* ≤ 0.0001 in both cases, Figure 4A). Further significant differences were seen between the Left and Right, Right and Overlap, and Centre and Overlap positions (*p* ≤ 0.0001 for all). No significant difference was seen between the Left and the Overlap positions (*p* ≥ 0.9999).

**Figure 4.** Box and whisker plots of horizontally stratified fluorescence. Track (**A**) and Manual (**B**) spray data was classified using left, centre, right, and overlap horizontal swath positions. The centre position on the track sprayer showed the most consistent fluorescein deposition. Greater differences were seen between horizontal positions on the track sprayer than the manual spray. Significant differences are indicated by different lowercase letters.

In contrast, there was no significant difference seen between the left and right swath positions when spraying manually (*p* ≥ 0.9999, Figure 4B). However, significant differences were seen between Left vs. Centre (*p* = 0.0041), Right vs. Centre (*p* = 0.0027), Left vs. Overlap (*p* = 0.0001), and Right vs. Overlap (*p* ≤ 0.0001) positions on the swath.

To discern whether the greater variation between fluorescein deposition at different wall heights in the manual spray was obscuring horizontal differences, data were further split into vertical and horizontal spray position categories. Except for comparisons including Overlap data, only one comparison showed a significant difference (CB vs. RB, *p* = 0.048). Plotting the stratified data showed a general trend for mean values from the Centre position to be lower than either Left or Right position data, indicating that this variation exists independently of the method of spray application.

Similar to results with fluorescein, the amount of insecticide sprayed with the track sprayer was most consistent in the middle position of a swath (Table 4). For both CLT and DLT, the middle position also had the lowest concentration deposited (*p* < 0.0001 for all combinations). However, this was not observed for PPM or BRF. For three out of the four insecticides, the left position on a swath resulted in a higher concentration of active ingredient applied compared to either the centre or right positions.

**Table 4.** Horizontally stratified application rates of insecticides. Manual and track spray data was classified using left, centre, and right horizontal swath positions. Median application rate ± standard deviation is indicated for each horizontal position. Comparisons are done between horizontal positions by spray method. Significant differences are indicated by different lower-case letters.


Greater variation in spraying was observed when spraying manually for all horizontal positions for all active ingredients, except for PMM in the left position. For PMM and BRF, there was a trend from left to right with a lower amount deposited at the left compared to the right position on a swath (*p* < 0.05 for PMM and *p* < 0.0001 for BRF). No significant differences were found for DLT. For CLT, the left position showed a significantly higher concentration of active ingredient applied compared to the middle position on a swath (*p* < 0.05).

#### *3.5. Consistency of Wall Spraying along the Swath*

Although the difference was slight, there was a general trend for fluorescein deposition from the track sprayer to be greater at the top of walls, decreasing with each vertical wall position (Figure 5). There were no significant differences between deposition onto filter papers immediately above or below each other. However, a significant difference between deposition onto filter papers at the uppermost and lowermost wall positions was evident (high vs. low, *p* = 0.0166).

Fluorescein deposition using a manual spraying (Figure 3B) showed significant differences between the two upper positions and the two lowest positions on the wall (four comparisons, *p* ≤ 0.0001). Significant differences in fluorescence were seen between the centre and the lower middle position (*p* = 0.0294), centre and bottom position (*p* ≤ 0.0001), and lower middle and bottom position on the wall (*p* = 0.0017). No significant differences were seen between the three upper positions. When the direction of spray (upwards or downwards swath) and total swath spray time were added to the analysis, the pattern of variation was preserved only in downwards swaths. Upwards swaths showed no significant difference between fluorescein deposition at the tops and bottoms of walls (*p*= 0.9555), but significant differences were seen between the top and centre of the wall (*p* = 0.0164) and the upper middle, middle, and lower middle compared to the bottom of the wall (*p* = 0.0014; *p* ≤ 0.0001; *p* = 0.0356, respectively). No significant difference was found between the time taken to spray a downwards swath vs. an upwards swath (*p* = 0.8779), indicating that it is spray rhythm, rather than spray time, that differs with spray direction.

The minor trend towards decreasing fluorescence with vertical positions was not shown when spraying with insecticides. No significant differences were found between vertical positions sprayed with the track sprayer for either PMM or BRF. For DLT and CLT, only the upper middle position was significantly different, compared to the lower middle position (CLT; *p* < 0.05) or the centre position (DLT; *p* < 0.05).

Variation in insecticide application rate when sprayed manually was generally larger by vertical position compared to the track sprayer. Although some significant differences were found, there was no common trend between insecticides (Table 5). For PMM, the centre position had a significantly lower concentration on filter papers compared to the upper middle and lowest position (two comparisons; *p* < 0.05). No significant differences were found for BRF. For DLT, only the highest position was significantly different, compared to upper middle, centre, and lower middle positions (three comparisons, *p* < 0.01) and lowest position (*p* < 0.001). Similarly, for CLT the highest position was significantly different, compared to upper middle and centre positions (two comparisons; *p* < 0.01) lower middle position (*p* < 0.05), and lowest position (*p* < 0.001).

**Table 5.** Wall height analysis for track and manual spray for insecticides. Median application rate ± standard deviation is indicated for each vertical position. Comparisons are done between vertical positions by spray method. Significant differences are indicated by different lowercase letters.

