Stink Bug Feeding Induces Fluorescence in Developing Cotton Bolls
- First Online:
Stink bugs (Hemiptera: Pentatomidae) comprise a critically important insect pest complex affecting 12 major crops worldwide including cotton. In the US, stink bug damage to developing cotton bolls causes boll abscission, lint staining, reduced fiber quality, and reduced yields with estimated losses ranging from 10 to 60 million dollars annually. Unfortunately, scouting for stink bug damage in the field is laborious and excessively time consuming. To improve scouting accuracy and efficiency, we investigated fluorescence changes in cotton boll tissues as a result of stink bug feeding.
Fluorescent imaging under long-wave ultraviolet light showed that stink bug-damaged lint, the inner carpal wall, and the outside of the boll emitted strong blue-green fluorescence in a circular region near the puncture wound, whereas undamaged tissue emissions occurred at different wavelengths; the much weaker emission of undamaged tissue was dominated by chlorophyll fluorescence. We further characterized the optimum emission and excitation spectra to distinguish between stink bug damaged bolls from undamaged bolls.
The observed characteristic fluorescence peaks associated with stink bug damage give rise to a fluorescence-based method to rapidly distinguish between undamaged and stink bug damaged cotton bolls. Based on the fluorescent fingerprint, we envision a fluorescence reflectance imaging or a fluorescence ratiometric device to assist pest management professionals with rapidly determining the extent of stink bug damage in a cotton field.
Phytophagous stink bugs (Hemiptera: Pentatomidae) comprise a critically important insect pest complex affecting worldwide food and fiber production. This group of closely related genera has a wide host range that includes fruit, vegetable, nut, fiber, and cereals in addition to numerous wild hosts . Preferential feeding sites are confined to the fruiting structures [2, 3], but some species feed on vegetative plant parts when fruiting structures are not available. Stink bugs have piercing/sucking mouthparts, and generalized feeding symptoms include abortion of young fruits, a predisposition to colonization by decay organisms, and cosmetic deformities. In southeastern US cotton production, feeding by stink bugs causes boll abscission, lint staining, reduced lint quality, and reduced yields [4, 5, 6, 7, 8]. More recent work has shown that the southern green stink bug, Nezara viridula (L.) (Hemiptera: Pentatomidae), is a competent vector of bacterial pathogens that causes seed and lint necrosis . Stink bug damage to the 2007 southeastern cotton crop was estimated at 11.6 million dollars .
The scientific basis for implementation of Integrated Pest Management or IPM  is that insect pest populations must be monitored during periods of plant susceptibility to make cost-effective decisions about pest management. The decision to intervene (i.e. make an insecticide application) should be based on a cost/benefit analysis: expected damage attributed to the insect population versus the cost of the insecticide application . Grower profits will be marginalized if the insect sampling procedure does not accurately represent the true insect density. For example, excessive spraying costs would result when damage estimates exceed the actual population density. Likewise, when the pest population is underestimated a portion of the producer's profits would be mitigated because the pests inflict excessive damage to the crop. Development of an effective sampling plan is the single most critical piece of information in the decision making process . However, development of an effective sampling plan cannot proceed without a rapid and accurate sampling method. Sampling for stink bugs and their associated damage in cotton fields is time-consuming, because the bugs are aggregated and the damage is often obscured on the outside of the boll. In cotton, the most reliable characteristic is to collect quarter-sized soft bolls and dissect them for internal feeding symptoms including punctures and warty growths on the inner boll wall, lint staining, and rotten locks. Toews et al.  compared traditional sampling methods for stink bugs including 50 sweeps with a 38.1 cm sweep net, 3.7-linear meters of row shaken over a white drop cloth, and internal examination of 20 quarter-sized bolls. Results show that internal examination was more than 10-fold more sensitive, but required more than seven minutes per sample set of 20 bolls compared to 97 seconds and 67 seconds for the sweep net and drop cloth, respectively. In detail, the time to collect 20 bolls (123.7 ± 1.6 seconds) and examine the same 20 bolls (445.0 ± 5.3 seconds) using the current internal detection method leads to an examination time of approximately 30 seconds per boll. Moreover, a large number of bolls is required for statistical accuracy. Reay-Jones et al. concluded that to obtain an estimate within 10% of the mean when there was 14.8% boll injury would require 112 samples of 20 bolls per sample (a total of 2240 bolls). Clearly, a new method that would reduce the examination time per boll is needed. In fact, recent efforts have been made towards characterizing changes in the production of volatile components by the cotton plant as a function of stink bug feeding [16, 17].
We report here the observation of an unusual and strong fluorescent emission in cotton boll tissue that has been damaged by stink bug feeding. The objectives of this study were (1) to investigate differences in fluorescent emission between stink bug damaged and undamaged cotton bolls, (2) to find the optimum excitation and emission wavelength ranges of both stink bug-related auto fluorescence and normal tissue auto fluorescence of cotton bolls, and (3) to characterize the potential of fluorescence measurements to differentiate undamaged and damaged cotton boll tissue.
Results and Discussion
In light of the observation of fluorescence related to needle punctures, the question of specificity arises. It is conceivable that compression causes fluorescence emission with a similar spectrum to the fluorescent spots seen around piercing wounds. The main difference is the intensity, which is higher around piercing wounds, and the shape. Piercing wounds under fluorescent imaging are round and have a distinct black center (see Figures 2 and 3). Shallow scratch marks are elongated. Suitable image analysis methods are available that can eliminate those false-positives. Furthermore, cotton bolls develop on a peduncle from the extra-axillary bud at the base of the cotton leaf petiole . Although it is possible that one boll could physically contact a close neighboring boll during extreme weather conditions, the stiff peduncle and bracts would preclude penetration of the boll wall tissues and symptoms observed with stink bug feeding. Similarly, boll trauma from agricultural machinery would include boll abrasion and crushing as opposed to puncturing.
Furthermore, a new peak (unlabeled) at λex = 320 nm and λem = 430 nm became evident, which was insignificant in its intensity compared to the A-dome in Figure 6A. This wavelength range is typical for protein autofluorescence .
The solvent, i.e., pure 70% spectroscopy-grade ethanol was also scanned in the same way to examine potential solvent background emission. Although the solvent also emitted some very weak blue-green fluorescence (~340 nm to 440 nm), it was excited by shorter wavelengths ranging from 300 nm through 320 nm and was therefore easily distinguishable from the tissue peaks. Furthermore, there was no ethanol emission when the excitation wavelength was longer than 320 nm. We conclude that the ethanol solvent does not affect the fluorescence detection of the relevant tissue components and was a suitable solvent for this study.
Comparison of Detection Accuracy with Fluorescence and with Conventional Inspection
Accuracy and error rates for detecting damaged cotton bolls as a result of stink bug feeding.
Time since damage (d)
(number bolls/total bolls)
(number bolls/total bolls)
Fluorescence spectroscopy based methods have been widely used in food and agricultural produce quality assessment and constituent identification [21, 22, 23]. Imaging methods that use either fluorescent staining or auto fluorescence can be used to visualize key constituents of the target object and have the potential to provide superior image contrast. For example, Kuensting et al. used an autofluorescence imaging method to visualize and highlight the internal structural details in soybeans. By using a fluorescence staining imaging method, Ogawa et al. developed a fluorescence-based technique to visualize the three-dimensional distribution of constituents in rice grains. Herein, we report a measurable and visible fluorescence emission associated with stink bug feeding of cotton bolls.
Both fluorescence reflectance imaging and epifluorescence microscopic examination conducted in this study indicated that the piercing action that is associated with stink bug feeding on cotton bolls produced a characteristic blue-green fluorescence when excited by long-wave UV exposure. Red emission from chlorophyll recedes at the same time (most prominently seen in Figure 5B). This fluorescence is unusually strong in the inner carpal wall and in affected lint, but it can also be detected from the exterior of the cotton boll as shown in Figure 5.
Our spectral analysis shows that the characteristic fluorescence is not only unique with respect to its bright intensity, but also with respect to its wavelength. Most prominently, the fluorescence emission peak at 420 nm with an excitation of 350 nm (marked with the letter A in Figure 6) differs strongly between intact boll tissue and pierced boll tissue. Moreover, we observed receding chlorophyll emission. In fact, it appears as if the intensity ratio I(λ = 420nm) = I(λ = 680nm) at an excitation of near 350 nm could serve as an indicator for the presence of piercing damage.
This observation gives rise to possible detection instruments. The ideal excitation wavelength is near the emission maxima of solid-state UV lasers and high-power UV light-emitting diodes (both 365 nm). A dual photodiode - ideally an avalanche photodiode for its higher sensitivity - would serve as the detection element. One photodiode would be sensitized with a bandpass filter for 420 nm, and the other would measure chlorophyll emission at 680 nm. The entire assembly could be housed in a wand to be used in the field. Alternatively, a CCD or CMOS imaging element would acquire fluorescence from a larger area of an individual boll. One challenge is the suppression of environmental light. Here, the detection or imaging element could be inserted into an enclosure that reduces environmental light, and further increase of the sensitivity can be achieved by employing the lock-in principle. In the field, a cotton boll would be inserted into the box for measurement. Likely, there would be no need to pluck the boll from the plant. Since fluorescence measurements take fractions of a second, an enormous time advantage could be gained over the manual examination of bolls that involves breaking bolls open.
Development of damaged cotton bolls
Stink bug feeding damage to cotton bolls was created using 5th instars of the southern green stink bug, Nezara viridula (L.) (Hemiptera: Pentatomidae). The insect colony was founded with ~ 50 adults collected from Tift County, Georgia in April 2007. The resulting colony was maintained in the lab following the methods of Harris and Todd  on fresh green beans, shelled green peanuts, and field corn. Adults were maintained in 37.9-liter glass aquaria while immatures were held in ventilated Petri dishes and small plastic dishes (part no. JSS16-89PP, Olcott Plastics, St. Chas, IL) at 25.0°C and 65% relative humidity. Previous research  suggested that these colonies may decline in vigor and viability so additional feral individuals were introduced annually.
Damaged cotton bolls were generated by caging immature stink bugs on cotton bolls of a known age in the greenhouse. Since manual examination of the fiber quality is destructive, no additional samples were grown in parallel to maturity to test for fiber quality. Briefly, picker cotton (FM 9063 B2F) was grown in a greenhouse maintained at 21 to 35°C with a 14:10 (L:D) photoperiod. Individual seeds were sown in 11.35-liter plastic pots filled with Metro Mix 300 growing medium (Sun Gro Horticulture, Bellevue, WA) and fertilized bimonthly with Osmocote 14-14-14 and Micromax 90505 (The Scotts Co. LLC, Marsville, OH). Following the methodology of Bundy et al. , individual white flowers were tagged daily and the subsequent bolls were allowed to develop normally for a period of 10-14 days. Then, a 30 cm long by 20 cm wide sleeve cage, containing three fifth instar stink bugs (treatment), was tightly sealed around the boll and subtending leaf for 72 hours. The age of the bolls at harvest was based on external boll diameter, and generally these bolls were 13 to 17 days after white flower. Bolls were excised from the plant immediately after exposure to the stink bugs, removed from the bag, and brought into the laboratory for examination. Undamaged bolls were prepared exactly as described above except that no insects were introduced into the sleeve cages.
To further examine the origin of the fluorescence emission, we used bolls that were not exposed to stink bugs and punctured them with a sterile syringe needle (31 Ga, 8 mm long, Beckton-Dickinson, product no. 328418). Puncturing was done manually, and care was taken that the needle penetrated into the lint tissue. The punctured bolls were either harvested immediately, or kept on the plant for 1, 2, 3, 4, 6, or 7 days after puncture before being harvested. Bolls with needle punctures were processed in the same manner as the other bolls.
To examine reflectance at the microscopic level, epifluorescent imaging was conducted by using a compound microscope (Olympus IX-71) with a 10× objective (total magnification 100×) and the Deep Blue filter set, which has an excitation wavelength range in the violet with a peak at 405 nm.
Epifluorescence images were taken in an identical fashion for normal and lesion tissue of cotton bolls.
Optimal excitation and emission wavelengths of undamaged and damaged cotton boll tissue were determined through spectral analysis. An analytical spectrofluorometer (FluoroMax-3, Horiba Jobin Yvon, Edison, NJ, USA) was used to analyze tissue spectral properties. The capabilities of the instrument include emission scanning where the selected excitation wavelength is held constant while the emission intensity is obtained as a function of the wavelength, excitation scanning where the emission wavelength is kept fixed while emission intensity is obtained as a function of excitation wavelength, and a matrix scan where fluorescence intensity is determined as a two-dimensional function of excitation and emission wavelength. The matrix scan allows to exhaustively characterize fluorescent properties of an unknown material. The result of a matrix scan is generally represented in 3-dimensional space with one fluorescence intensity axis, one excitation wavelength axis, and one emission wavelength axis.
Small solid tissue samples from the interior boll walls of n = 3 bolls were excised under a dissecting microscope while the subject was illuminated under long-wave UV. Only strongly fluorescing tissue was cut from the treated bolls, while similar masses of undamaged tissues were also excised from undamaged bolls for comparison. A total of 5 mg of fluorescent and non-damaged tissues were pooled separately for analyses. A glass cuvette was used to contain the sample for spectral measurement. Samples were soaked in ~5 ml of 70% spectroscopic-grade ethanol (Sigma-Aldrich, St. Louis, MO) for 48 h to extract the fluorescent materials and then 3 ml of the resulting solution were examined with the spectrofluorometer. Matrix scan 3-D graphs were created from the scan data with standard surface-rendering techniques .
Comparison of Detection Accuracy with Fluorescence and with Conventional Inspection
To determine the potential for using fluorescence as a method for detecting stink bug damage, we acquired fluorescent images of 56 bolls as described above. Images were examined by a trained technician on a computer monitor. The technician had no knowledge about the treatment (control versus infested). Subsequently, the bolls were opened and examined for visible damage (lint staining, warts, puncture marks) . We define detection accuracy as the sum of true-positives and true-negatives relative to the total number of bolls. To observe the development of the fluorescent regions over time, 26 bolls were examined within less than one day after exposure to stink bugs, and 30 bolls were kept on the plant for up to 7 days before harvesting and subsequent imaging.
We appreciate excellent technical support from Jessica Corbett, Blake Crabtree, and Darcy Lichlyter. Cotton seed for this project was graciously donated by Bayer CropScience. Mention of trade names or commercial products in this publication is solely for providing specific information and does not imply recommendation or endorsement by the University of Georgia.
- 2.Barbour KS, Bradley JR Jr, Bacheler JS: Phytophagous stink bugs in North Carolina cotton: an evaluation of damage potential. Proceedings-Beltwide Cotton Production Research Conferences (USA) 1988.Google Scholar
- 8.Turnispeed S, Sullivan M, Hagerty A, Gibson A: Boll damage by the stinkbug/plant bug complex in South Carolina cotton. roceedings of the Beltwide Cotton Conferences, January 2003, Nashville, TN, National Cotton Council of America, Memphis, TN 2003.Google Scholar
- 10.Williams MR: Cotton insect losses-2007. Proceedings of the 2008 Beltwide Cotton Conference 2008, 929-979.Google Scholar
- 13.Southwood TRE: Ecological methods, with particular reference to the study of insect populations. 1978, 524.Google Scholar
- 14.Toews M, Greene J, Reay-Jones F, Reeves R: A comparison of sampling techniques for stink bugs in cotton. Proceedings of the Beltwide Cotton Conferences 2008, 1193-1203.Google Scholar
- 17.Degenhardt DC, Greene JK, Khalilian A, Reeves RB: Volatile emissions from developing cotton bolls in response to hemipteran feeding damage. Journal of Entomological Science 2011, in press.Google Scholar
- 18.Hancock NI: Relative growth rate of the main stem of the cotton plant and its relationship to yield. Journal of the American Society of Agronomy 33: 590 1941., 602:Google Scholar
- 21.Swatland HJ: Measurement of the gristle content in beef by macroscopic ultraviolet fluorimetry. Journal of Animal Science 1987, 65: 158.Google Scholar
- 25.Ogawa Y, Sugiyama J, Kuensting H, Ohtani T, Hagiwara S, Liu XQ, Kokubo M, Yamamoto A, Kudoh K, Higuchi T: Advanced technique for three-dimensional visualization of compound distributions in a rice kernel. Journal of Agricultural and Food Chemistry 2001,49(2):736-740. 10.1021/jf000997pCrossRefGoogle Scholar
- 26.Harris VE, Todd JW: Comparative fecundity, egg fertility and hatch among wild-type and three laboratory-reared generations of the southern green stink bug, Nezara viridula (L.)(Hemiptera: Pentatomidae). Journal of the Georgia Entomological Society 1980,15(3):241-245.Google Scholar
- 27.Bundy CS, McPherson RM, Herzog GA: An examination of the external and internal signs of cotton boll damage by stink bugs (Heteroptera: Pentatomidae). Journal of Entomological Science 2000,35(4):402-410.Google Scholar
- 28.Haidekker MA: Image Visualization. In Advanced Biomedical Image Analysis. Hoboken, NJ: John Wiley & Sons; 2011:413-440.Google Scholar
This article is published under license to BioMed Central Ltd. This is an Open Access article distributed under the terms of the Creative Commons Attribution License (http://creativecommons.org/licenses/by/2.0), which permits unrestricted use, distribution, and reproduction in any medium, provided the original work is properly cited.