New Protocol for Measuring Lepidoptera Wing Damage

Lepidoptera may acquire damage to their wings throughout their lifespan. The ability to quantify the accumulated damage is relevant when studying the impact of wing damage on territorial behavior, mating systems, predation, or assessing relative age of the insect, but unfortunately, methods for accurate insect wing damage quantification are scarce. The purpose of this paper is to introduce a new protocol that provides a simple method for accurately quantifying wing damage of live Lepidoptera, without the removal of them from the field. Using a combination of Adobe Photoshop® and Scion Image®, 3 photographs of wild Papilio (Pterourus) homerus butterflies with wing damage were analyzed and compared to an older method of visually estimating wing damage. Of the 12 individual wings analyzed, 7 were significantly different (p ≤ 0.05), and the new protocol yielded precise results. The newly described protocol is an inexpensive and accurate method for determining percent wing damage on insects without having to harm or remove them from the wild. Additional key words: wing area, Papilio homerus, wing assessment, predation, territoriality It is not unusual for Lepidoptera to acquire damage to their wings. Wing damage can accumulate from multiple factors such as conspecific territorial behavior (Pinheiro 1990; Freitas et al. 1993; Monge-Najera 1998), mating (Anderson & Keyel 2006), predation (Benson 1972; Shapiro 1974; Bowers et al. 1985; Mallet & Barton 1989; Lyytinen 2003; Langham 2004), and weather and daily wear (Carter 1992). Insect wing damage is commonly used as an indicator for age (Watt et al. 1977; Hayes et al. 1998; Kemp 2000; Pitts & Wall 2004) and to study predation associated with the evolution and development of eyespots and wing appendages in Lepidoptera, such as the False-head Hypothesis (Robbins 1981; Tonner et al. 1993). Severe wing damage can potentially hinder flight performance, impinging on mate-locating behavior and tactics (Koenig & Albano 1985), acquisition of food sources (Higginson & Barnard 2004), and predator evasion (Robbins 1981). The combination of digital cameras and visual software has led to improved quantification systems for measuring damage to biological entities, such as leaf area associated with multiple insect-plant interactions (James & Newcombe 2000; O’Neal et al. 2002) including host plant resistance and the effect of pesticide application (Hoy & Hall 1993). Although this technology has progressed in many different fields of study, there are few new applications for quantifying Lepidoptera wing damage. Previous methods of estimating wing damage were limited to categorical rankings (i.e., #1 = freshly emerged, no wing damage; #2 = slight wing damage; etc.) (Watt et al. 1977), nominal rankings (i.e., tears, missing areas, or notches) (Burkhard et al. 2002), or the use of a grid system or sectioning the wing to determine wing area (Tonner et al. 1993). The purpose of this paper is to introduce a simple protocol for the assessment of wing damage of Lepidoptera. This new protocol is an adjustment from a protocol described by O’Neal et al. (2002). To demonstrate the accuracy of the described protocol, it will be compared to the results of a survey where percent wing damage was visually estimated. Study Organism. The Homerus Swallowtail, Papilio (Pterourus) homerus Fabricius, 1793 (Papilionidae), is the largest swallowtail butterfly in the Western Hemisphere and is endemic to Jamaica (Emmel & Garraway 1990). Papilio homerus is protected as an Appendix I species by CITES and is listed by the IUCN as an endangered species, serving as a flagship species for the island’s natural wildlife heritage (Collins & Morris 1985). Photographs of wild P. homerus were taken within the Cockpit Country while estimating population size using mark-recapture protocol (Lehnert 2008). Wing wear (not associated with capturing technique) was noticed on captured and recaptured specimens, encouraging a study of wing wear analysis methods. MATERIALS AND METHODS Photographs of Papilio homerus were taken in the field using a Nikon Coolpix 8700 digital camera. The 8 mega-pixel camera was set to the highest resolution of 3264 × 2448 pixels. It was not necessary to have a camera with large mega-pixel capabilities, but a higher resolution gave more accurate results. Photographs were then transferred to a computer and opened in Adobe Photoshop®6.0 as JPEG images. The images used for analysis are shown in Fig. 1. Each wing was individually cut using the Lasso tool and pasted into a new file. In the new file, the Erase tool was selected and 3030 JOURNAL OF THE LEPIDOPTERISTS’ SOCIETY used to outline the wing as the presumed shape of an undamaged wing while erasing the remainder of the background. The image was cleaned using the Erase tool so that only the wing remained for analysis. Photographs of undamaged wings were used as a template when outlining the presumed shape of an undamaged wing on a wing with damage. The image was then saved as an undamaged wing JPEG file as a high quality image (10) with the format option set as baseline standard. The undamaged wing file was reopened and the Erase tool was used to outline the actual damaged wing, which was then saved as a damaged wing JPEG file. Both images were grayscaled by selecting the Image tab, then choosing the Mode option, which leads to the Grayscale option. The grayscaled images were saved as TIFF files. The undamaged wing TIFF file was opened in Scion Image® for analysis from the File menu. The Options menu was selected to ensure that the Grayscale tab was checked and the Threshold tab was also checked. The Threshold tab converted the image to black and white. The Map Box was opened from the windows menu and used to adjust the image so that the area used for analysis was completely black with a white background. If the area of the image needed for analysis was not becoming completely black on a white background or additional black spots appeared outside of the wing, the Paint tool and/or the Eraser tool was selected from the tools menu to adjust the image accordingly. The Wand tool was then selected and clicked on the black image (the wing) to highlight the area for analysis. If the image for analysis looked correct, the Measure option was chosen from the Analyze menu to reveal a pixel count of the image. The pixel count was given as the area in the Info Box and recorded. If the pixel count was not shown, it was then selected from the Set Scale option in the Analyze menu. The damaged wing TIFF file was then placed through the same procedure to retrieve the area (pixel count) of the image and recorded. By simply dividing the damaged wing area (pixels) by the undamaged wing area (pixels) and multiplying by 100, the percent wing area remaining was revealed. This number was subtracted from 100 to give the percent wing damage. The process is illustrated in Fig. 2. Each of the four wings in each photograph was analyzed 10 separate times to determine the efficiency of the described protocol. For comparison, a survey portraying each original photograph was given to ten people. Each person was asked to visually estimate the percent wing damage of every wing (left forewing (LFW), left hind wing (LHW), right forewing (RFW), and right hind wing (RHW)) in each photograph. A paired-sample t-test using SPSS 16.0 software was used for comparing differences in accuracy and precision of the results from Scion Image® analysis and the survey. FIG. 1. Photographs of Papilio homerus used for image analysis. From left to right, photograph #3889, 4060, and 3794. FIG. 2. Illustration of procedure used for wing analysis. In this example the left forewing (LFW) of individual 3889 is analyzed. The photograph on the far left (A) is the original picture opened as a JPEG file in Adobe Photoshop 6.0. The lasso tool was used to outline the LFW, which was saved into a new file. In the new file, the erase tool was selected and used to outline the presumed shape of an undamaged wing (B). The erase tool was used to then outline the damaged wing. Both photographs were grayscaled and saved as TIFF files. Using Scion Image Analysis, the photographs were opened and converted to completely black images on an all white background (C = presumed undamaged wing, D = damaged wing). The analysis was performed issuing a pixel count (C = 310818, D = 259192). By using the equation listed in the methods, it was estimated that there was 16.6% wing damage on the LFW. VOLUME 64, NUMBER 1 31