Plant Volatiles Help Mediate Host Plant Selection and Attraction of the Spotted Lanternfly (Hemiptera: Fulgoridae): a Generalist With a Preferred Host

Abstract

Host plant volatiles play a key role in mediating plant–herbivore interactions. How an array of host plant volatiles guides host preference and attraction in the invasive polyphagous Lycorma delicatula (White), the spotted lanternfly (SLF), is largely unknown. A pernicious phloem feeder, SLF feeds on over 70 species of plants, some with high economic impact. To aid the development of detection and monitoring tools for SLF, we used a two-choice olfactometer to compare 14 host plant species for attraction, first to a blank control, and then to their preferred host Ailanthus altissima (Mill.) Swingle (Sapindales: Simaroubaceae), tree-of-heaven. SLF were significantly attracted to seven host plants compared to a blank control, but no host plant was more attractive than tree-of-heaven. We then used electroantennographic detection (EAD) to screen select host plants for EAD active compounds, hypothesizing that EAD-active plant volatiles act as kairomones and mediate SLF attraction to host plants. Out of 43 unique antennal responses, 18 compounds were identified and tested individually for attraction in a two-choice olfactometer against a blank control and then against methyl salicylate, the current best attractant. Eleven compounds were significantly attractive, and one, sulcatone, was more attractive than methyl salicylate. Blends of kairomones were then tested for attraction, revealing five blends that were significantly more attractive than methyl salicylate, and could be developed into lures for field testing. The presence of these kairomones in volatile profiles of 17 plant species is described. These findings support the hypothesis that the identified volatiles act as kairomones and function in attraction to host plants.

Lycorma delicatula, host plant volatile, kairomones, invasive species, electrophysiology

The spotted lanternfly (SLF) Lycorma delicatula (White) is a recent invasive insect in the Eastern United States. First discovered in Berks County, Pennsylvania in 2014, the population has grown and spread to more than a dozen counties in Pennsylvania (PDA 2018, VDACS 2018) and individual SLF have been confirmed in Delaware (Delaware Department of Agriculture 2019), Maryland (Maryland Department of Agriculture 2018), New Jersey (Wolfe 2018), New York (New York State Integrated Pest Management 2019), Connecticut (University of Connecticut College of Agriculture, Health, and Natural Resources 2018), Massachusetts (MDAR 2019), and Virginia (Lidholm 2019), at the time of this writing. As a generalist phloem feeder, L. delicatula has a wide host range, feeding on over 70 species of predominantly woody plants, including species with high economic impact such as grapevines, apple trees, stone fruit trees, ornamental trees, and valuable hardwoods (Dara et al. 2015). SLF inflict severe damage on these species via mass feeding that can simultaneously deprive the plant of nutrition and create wounds that facilitate pathogen transmission (reviewed in Bardner and Fletcher 1974). Additionally, SLF excrete copious amounts of honeydew when feeding, which coats all nearby surfaces and enables sooty mold to grow (Song 2010), which then blocks the sunlight required for photosynthesis (Tedders and Smith 1974, Filho and Paiva 2006). This can have severe impacts on agricultural systems like viticulture, where the proliferation of sooty mold on grapes prevents them from being used, causing large crop losses (Song 2010, Baker et al. 2019, Urban 2019). Unfortunately, highly effective tools such as traps and lures that aid in control and monitoring of this pest are relatively few and in the early stages of development (Cooperband et al. 2019).

Plant volatiles are known to mediate attraction to host plants in many phytophagous insects (Dethier 1982, Visser 1988, Bernays 1994). As no pheromones are known to mediate behavior in Fulgoridae, we initially focused on plant kairomones, starting with reports and observations of SLF feeding preferences for certain plant species. Without the aid of gas chromatography coupled with electroantennographic detection (GC-EAD), Cooperband et al. (2019) found that three volatile compounds found in Ailanthus altissima (Mill.) Swingle, Sapindales: Simaroubaceae (tree-of-heaven) and Vitis spp., Vitales: Vitaceae (wild grape) attracted SLF under laboratory conditions, and subsequent field tests revealed lures containing methyl salicylate increased trap captures on sticky bands. However, the composition of volatiles produced by plants is often notoriously complex (Peñuelas and Llusiá 2001), and discerning which compounds are important and mediate behavior for a given insect species can be incredibly difficult without the aid of electroantennography. To this end, we developed a GC-EAD method to accommodate the morphology of fulgorid antennae to screen their preferred host plant, tree-of-heaven for all EAD active compounds. These compounds could then be tested for attraction in a behavioral assay, and their presence or absence in other host plants evaluated. As insects likely respond to blends of compounds emitted by host plants, rather than singular components (Bruce and Pickett 2011), we hypothesized that artificial blends could be developed that would be more attractive than methyl salicylate. Whole host plants were evaluated in laboratory olfactory bioassays to determine overall host plant preference, and volatile profiles of several attractive host plant species were then compared to help design an optimal blend.

We hypothesized that EAD-active plant volatiles act as kairomones and mediate SLF attraction to host plants. With the main objective of developing a kairomone blend that is significantly more attractive than methyl salicylate alone, we 1) tested several host plants for attraction in behavioral bioassays; 2) characterized a profile of EAD-active compounds from tree-of-heaven and three other hosts; 3) tested candidate kairomones singly for attraction in behavioral bioassays; 4) selected candidate kairomones based on GC-EAD, behavioral responses, and availability, and tested new kairomone blends for attraction in an olfactometer behavioral bioassay; and 5) sampled and identified headspace volatiles of other host plants and characterized them as they relate to the kairomone profile of the preferred host, tree-of-heaven.

Materials and Methods

Insect Collection and Rearing

It was challenging to maintain large numbers of SLF in captivity due to their heavy phloem feeding requirements. Therefore, all stages of SLF were collected from within the infested area in Berks County, PA, and transported overnight to the Otis Laboratory Insect Containment Facility according to the conditions set by permits from Pennsylvania Department of Agriculture (PDA) (PP3-0123-2015) and U.S. Department of Agriculture (USDA) (P526P-15-00152 and P526P-17-04376). Areas with high SLF density that had not been treated with pesticide were located with the help of PDA. Collecting and shipping SLF usually occurred on Mondays, allowing the insects to be used in laboratory experiments on Tuesdays through Fridays. Cuttings of tree-of-heaven were collected locally in Barnstable County, Massachusetts, following conditions set in permits from the Massachusetts Department of Agricultural Resources (MDAR) (2016-PI18, 2017-PI21, 2018-PI21, and 2019-PI24). These cuttings were placed in tubes containing hydroponic solution (Maxigrow, GenyHydro Inc., Sebastopol, CA, prepared according to label) which were then placed inside a screened insect rearing cage (24.5 cm × 24.5 cm × 63 cm BugDorm, Taiwan). Cuttings of tree-of-heaven were replaced every 1–2 d. Containers were kept in a walk-in environmental chamber with 14.75:9.25 h L:D at 25^oC and 45–60% R.H. and placed ~100 cm below six full-spectrum fluorescent bulbs (40 W Gro-Lux, Osram Sylvania, Westfield, IN). Instars 1–4 were housed in groups of approximately 50–100 individuals per cage. Adults were separated according to sex by the diagnostic red valvifers of females (Dara et al. 2015) and kept in separate cages, up to ~20 individuals per cage.

Adults were categorized based on when they were collected: before mating was first observed in the field (early), between the first observation of mating and the first observation of oviposition (mid), and after the first oviposition was observed (late) (Cooperband et al. 2018, Baker et al. 2019). Early adults were collected between August 1 and September 13, mid adults were collected between September 14 and September 30, and late adults were collected in early October. Dates were selected based on field observations in 2017, 2018, and 2019, and retrospectively applied to individuals tested in 2016.

Host Plant Volatile Collections

Headspace volatiles of living plants were collected using a custom, portable, push–pull volatile collection system (Supp Fig. 1 [online only]). Briefly, air was pushed through a charcoal filter, into an oven bag (19 × 23.5 in, Reynolds Kitchens, Louisville, KY) enclosed around a plant, and then pulled out through a Hayesep Q adsorbent cartridge and a secondary overflow cartridge and through an air vacuum pump (12–15 LPM, 12W, 12V, D2028, Karlsson Robotics, Tequesta, FL) in a closed loop. Oven bags used in volatile collections were baked prior to use (150°C, 1–2 h) to remove caprolactam normally present in the bags (Stewart-Jones and Poppy 2006). The pump was powered by a 12 V rechargeable lead acid battery (ExpertPower, Compton, CA) and controlled by a potentiometer to adjust motor speed and resulting air flow to sample volatiles at 1–2 liter/min for 3–6 h. Charcoal filter cartridges were made from 2 ml glass pipettes filled with 20-60 mesh activated charcoal (Sigma–Aldrich, St. Louis MO) held in place with glass wool (Pyrex, Corning, NY). Sample cartridges were made the same way but utilized ~200 mg Hayesep Q (Sigma–Aldrich, St. Louis, MO) in place of activated charcoal. Volatiles retained by the sample cartridge were eluted with 1 ml pentane (Acros Organics, Fisher Scientific, Fair Lawn, NJ) for chemical analysis. These volatiles were tested for antennal responses using GC-EAD (described below). Leafy branches and trunks of tree-of-heaven, and additionally, leafy branches of black walnut, hops, and spicebush, were screened for EAD active compounds because high numbers of SLF were observed on them in the field (Murman et al. 2020; N.T.D. pers. obs.).

After identifying EAD-active compounds, volatiles were collected from 17 species of plants either in situ or growing in a greenhouse and compared to each other. We analyzed differences in relative volatile emissions between plant species without controlling for many factors (plant health, level of herbivory, type of herbivore, temperature, sun exposure, available water, soil differences, etc.) that were outside of the scope of this experiment. Plant species, the plant part sampled, location, and the date of sampling are provided in Supp Table 1 (online only). Species examined were Acer rubrum L. (Sapindales: Sapindaceae) (red maple), A. altissima (tree-of-heaven), Betula lenta L. (Fagales: Betulaceae) (black birch), Celastrus orbiculatus Thunb. (Celastrales: Celastraceae) (oriental bittersweet), Humulus lupulus L. (Rosales: Cannabaceae) (hops, variety: Cascade), Juglans nigra L. (Fagales: Juglandaceae) (black walnut), Lindera benzoin (L.) Blume (Laurales: Lauraceae) (spicebush), Liriodendron tulipifera L. (Magnoliales: Magnoliaceae) (tulip tree), Malus domestica (apple), Melia azedarach L. (Sapindales: Meliaceae) (chinaberry), Parthenocissus quinquefolia (L.) Planch. (Vitales: Vitaceae) (Virginia creeper), Prunus serotina Ehrh. (Rosales: Rosaceae) (black cherry), Quercus rubra L. (Fagales: Fagaceae) (red oak), Rhus typhina L. (Sapindales: Anacardiaceae) (staghorn sumac), Robinia pseudoacacia L. (Fabales: Fabaceae) (black locust), Vitis spp. (wild grape), and as an outgroup that is not considered a host, Picea abies L. (Pinales: Pinaceae) (Norway spruce). Volatiles were sampled from branches or vines with both bark and leaves. The percent abundance of volatiles of interest was calculated from GC-FID traces using the total peak area of all compounds in a run, starting after the solvent and ending with (E,E)- α-farnesene, which was the last major compound to elute. These percent abundances were then compared between species. In the case of apple trees, volatiles from branches and leaves were collected from trees not bearing apples, but the heavy SLF populations on apple in Pennsylvania have been observed on trees bearing ripe fruit, which are known as an abundant source of (E,E)-α -farnesene (Murray 1969, Sutherland et al. 1977, Suckling et al. 2010).

Chemical Analysis

Aliquots (1 µl) of plant headspace samples were analyzed by gas chromatography–mass spectrometry (GC-MS). Samples were injected (splitless) on an Agilent 7890B GC coupled to a 5977A equipped with either a DB-5MS (30 m, 0.25 mm ID, 0.25 µm film thickness), operated at 40°C for 1 min, 10°/min to 280° and held for 10 min, or an RTX-1701 column of the same dimensions, operated at 40°C for 1 min, 10°/min to 250° and held for 15 min. The temperature of the inlet was 250°C, and the carrier gas (He) was at a constant flow of 1.2 ml/min. Peaks were identified through a combination of matches to the NIST11 mass spectral library (v11, Agilent Technologies, Santa Clara, CA), comparison to mass spectra, retention times of known standards where possible (Sigma–Aldrich, Fisher Scientific, Bedoukian Research, Inc.), and to Kovat’s retention indices calculated from the retention times of n-alkanes.

Electrophysiology

Antennal responses to plant volatiles were recorded using coupled gas-chromatography and electroantennographic detection (GC-EAD). An Agilent 6890 GC in splitless mode was equipped with either an HP-5 ms column (30 m × 0.320 mm I.D. × 0.25 μm film; Agilent Technologies, Inc., Santa Clara, CA) operated at 40°C for 1 min, 10°/min to 280° and held for 10 min, or an RTX-1701 column of the same dimensions, operated at 40°C for 1 min, 10°/min to 250° and held for 15 min. The inlet temperature was held at 250°C, the carrier gas (He) was a constant flow of 4 ml/min, and the flame ionization detector (FID) was held at 280°C. In addition to matching retention times with synthetic standards, the RTX-1701 column was also employed in an attempt to resolve early eluting peaks that were thought to be alcohols based on preliminary mass spectrometry data. The GC injector port was held at 250°C, and the effluent of the GC column was split 1:1 between the FID and the EAD using a glass Y-connector and equal lengths of deactivated fused silica columns. The EAD effluent was carried out of the GC via a temperature-controlled arm (Syntech Temperature Controller, Kirchzarten, Germany) held at 250°C, and into a glass stimulus delivery tube (7 mm i.d.) before exiting over the antennae. To carry the GC effluent to the antennae, humidified, charcoal-filtered air was pushed through the delivery tube at 0.3 liter/min.

To accommodate the structure of SLF antennae (Wang et al. 2018), the head was removed and mounted on the reference electrode, a flame-polished saline-filled glass capillary (TW150F-4, World Precision Instruments, Sarasota, FL). The apical arista and basal bulb of one antenna were then removed with fine forceps, and the saline-filled (7.5 g/liter NaCl, 0.21 g/liter CaCl₂, 0.35 g/liter KCl, and 0.2 g/liter NaHCO₃) recording electrode was brought into contact with the resulting cavity at the distal end of the antenna (Supp Fig. 2 [online only]). The entire preparation was positioned at the end of the odor delivery tube, such that the airstream carrying the sample effluent of the GC flowed over the antenna. Recording electrodes were fashioned from custom pulled glass capillaries (World Precision Instruments, Sarasota, FL; pipette puller PP-830, Narishige Scientific Instrument Lab, Japan). After pulling, capillary tips were cut to match the cavity left by the removal of the basal bulb. Electrodes were held in place and adjusted using micromanipulators (Signatone Corp., Gilroy, CA), secured magnetically to a stainless-steel platform (Syntech, Kirchzarten, Germany). The platform was placed on a lab jack for vertical placement of the antennal preparation at the end of the stimulus delivery tube. Recordings were made from third instar, fourth instar, and adult spotted lanternflies (minimum of three per host plant species tested). Antennal signals were amplified using a Grass model P55 amplifier (Astro-Med, Inc., West Warwick, RI) and integrated with the GC-FID signal using a PeakSimple 6-channel integrator (Model 302, SRI International, Menlo Park, CA). Data were captured and analyzed using PeakSimple software (v. 3.85; SRI International, Menlo Park, CA).

Preparation of Lures

Lures of synthetic compounds used in olfactometer bioassays were made by placing 1 mg of neat material, or 100 µl of a solution of compounds in hexane, into an open 0.25 ml microtube (Fisherbrand, Waltham, MA) containing a strip of filter paper (13 × 3 mm). Control microtubes contained only filter paper (for comparison with neat material) or filter paper with 100 µl of solvent. Neat material was weighed in a capillary pipette on a microbalance (Mettler Toledo, Columbus, OH). For testing blends, the desired amounts of components were weighed and prepared in hexane solutions, serially diluted, and added as blends in 100 µl aliquots to the filter paper. Newly prepared lures were immediately used in behavioral bioassay experiments. Blend names corresponded to the number of components in the blend followed by a letter that was unique for that specific combination of compounds. All individual compounds tested, and blend components and quantities per lure, are given in Tables 1 and 2, respectively.

Table 1.

Summary of responses to individual synthetic volatile compounds tested in the dual choice olfactometer and/or GC-EAD

Compound tested	Vs. control							Vs. Methyl salicylate							Behaviorally Active	Antennally Active	Retention index		Sourcee	Percent Purity
	first	second	third	fourth	Females	Males	Combined	first	second	third	fourth	Females	Males	Combined			HP-5 MS	RTX- 1701
1 (Z)-3-Hexenold	n.s.		n.s.	+	+		+					n.s.		n.s.	Y	N	846	974	S	>98
2 α-Pinene	n.s.						n.s.	-						-	N	Y	954	953	S	98
3 1-Octen-3-ol			+	n.s.			+				n.s.			n.s.	Y	Y	977	1086	F	98
4 Sulcatone	+		+		n.s.	+	+	n.s.		+	n.s.			+	Y	Y	981		S	99
5 β-Pinene			n.s.				n.s.								N	Y	987	1022	S	99
6 (Z)-3-Hexenyl acetate				n.s.			n.s.								N	Y	1006	1085	S	>98
7 β-Ocimene (mixed isomers)	n.s.	n.s.			n.s.		n.s.								N	Y	1050	1085	S	90
8 (Z)-3-Hexenyl propionatef																Y	1099	1176	S	>97
9 Methyl benzoate			n.s.	+			+	-				n.s.		-	Y	Y	1100	1208	F	99
10 Linalool (racemic)	+						+	n.s.a						n.s.	Y	Y	1101	1202	S	97
11 4,8-Dimethyl-(3E)-1,3,7-  nonatriene			n.s.		n.s.		+				-			-	Y	Y	1118	1154	B	98
12 (Z)-3-Hexenyl butyrate			n.s.				n.s.				n.s.			n.s.	N	Y	1187	1262	S	>97
13 4-Terpineol	+	+	n.s.				+	-a				n.s.		-	Y	Y	1188	1272	F	97
14 Methyl salicylated	+		+	+	+	n.s.	+								Y	Y	1204	1315	S	99
15 β-ylangene				+ b			+								Y	Y	1441	1465	E	90
16 Geranyl acetone				n.s.c			n.s.				-			-	N	N	1448		S	97
17 β-Caryophyllene	+		n.s.				+			n.s.				n.s.	Y	N	1450	1465	S	>98
18 (E,E)- α-Farnesened	n.s.		+	+	+		+					n.s.		n.s.	Y	Y	1513	1553	B	84

In bioassays, SLF preferences between the compound tested and either a blank control or methyl salicylate were quantified. Kovat’s retention indices of compounds on nonpolar and semipolar columns are listed. Plus signs (+) indicate the compound tested was chosen significantly more than the alternative, whereas minus signs (−) indicate the alternative was chosen significantly more than the compound tested (χ ² test, α = 0.05, sample sizes provided in Supp Fig. 8 [online only]). Tests that were not significant are indicated as such (n.s.), and blank cells indicate no test was conducted. If significant differences in attraction were found it was concluded to be behaviorally active (Y), otherwise not (N). If repeatable antennal responses were elicited (Y) it was concluded to be antennally active, if not antennally active or results were inconclusive (N).

^aTests typically used 1 mg of neat material except for these which compared 1 mg of compound tested against 2 mg of methyl salicylate.

^bDue to a shortage of material, β-Ylangene was tested using 40 µg.

^cThis test used 0.5 mg of neat geranyl acetone versus control.

^dResults for these three compounds tested against controls are summarized from Cooperband et al. (2019).

^eSources of compounds are abbreviated as follows: S, Sigma–Aldrich (St. Louis, MO), F, Fisher Scientific Acros Organics (Geel, Belgium), B, Bedioukian Research (Danbury, CT), E, β-ylangene purified by Allard Cossé from Ylang Ylang Complete, Comoros essential oil purchased from Eden Botanicals (Petaluma, CA).

^f(Z)-3-Hexenyl propionate was only tested in bioassays as part of a blend, not individually.

Table 2.

Formulation of different blends of kairomones and the significant preferences compared to controls, each other, or to methyl salicylate

	Blend contents (mg/100 µl hexane)
	Blend Name
							Es					OH
Blend contents	2A	2B	2C	2D	3A	3B	3C	3D	3E	3F	3G	3H	5A	6A	6B	7A	7B	7C	8A	8B
(E,E)- α-Farnesene			0.25		0.33								0.5	0.50		0.50	0.50	0.50	0.50	0.50
Methyl salicylate				0.5		1.0	0.5	0.5	0.5	0.5	0.5		0.5	0.50	0.50	0.50	0.50	0.50	0.50	0.50
DMNT								0.5					0.5			0.20		0.20		0.20
(Z)-3-Hexenyl acetate	0.33	0.3			0.33	0.3								0.16	0.16	0.16	0.16	0.16	0.16	0.16
linalool (racemic)									0.5			0.5
Sulcatone				0.5
(Z)-3-Hexenol														0.16	0.16	0.16	0.16	0.16	0.16	0.16
Methyl benzoate	0.33	0.7	0.75		0.33	0.7	0.5						0.5	0.33	0.33	0.33	0.33		0.33	0.33
1-octen-3-ol										0.5		0.5	0.5	0.25	0.25	0.25	0.25	0.25	0.25	0.25
(Z)-3-Hexenyl butyrate							0.5	0.5	0.5	0.5	0.5						0.08	0.08	0.08	0.08
4-Terpineol											0.5	0.5
(Z)-3-Hexenyl propionate																			0.08
More attractive than … a		C					3G, 3H							1M	1M	1M	2M	4M
Less attractive than … a											3C	3C
N.S. compared to… b	3A		C	1M	2A	1M	3D, 3E, 3F 2M	3C	3C	3C			8B	7B	7A	6B, 8B	6A, 8A		7B	5A, 7A
Stages testedc	First	Fourth	Third	Fourth	Fourth	Fourth	First, second	First	First	First	First	First, second	mF	eF	eF	eF	eF	lF	eF	mF, eF

Each blend is named at the top of the column in which its components (mg) are listed. Results presented in the bottom 4 rows indicate behavioral responses of SLF in behavioral bioassays to blends listed at the top of each column relative to another blend, a control, or methyl salicylate.

^aSignificantly different, χ ² test (α = 0.05). Sample sizes for each comparison are listed in Fig. 4.

^bC, Control; M, methyl salicylate; the number preceding M indicates the factor of the dose of M compared: 1, same amount of methyl salicylate as contained in blend; 2, twice the methyl salicylate than contained in the blend; 4, four times the methyl salicylate than contained in the blend; DMNT , (E)-4,8-Dimethyl-1,3,7-nonatriene

^cAdult stage abbreviations: eF, early female; mF, mid female; lF, late female

Y-Plate Olfactometer Bioassays

Attraction of SLF to 1) live plant material or 2) synthetic lures was measured in custom-made Teflon Y-plate olfactometers (for details see below and also Cooperband et al. 2017, 2019). Filtered, humidified air was split and pushed past the stimulus on one side and a blank control lure on the other side, and into the two upwind arms of the olfactometer. After acclimating in an individual release cage for 30 min, an individual insect was released into the downwind end of the olfactometer and allowed 3 min to respond. If the insect passed the junction and traveled more than halfway up one of the arms of the Y, the choice was recorded. Insects that made no choice after 3 min were recorded as nonresponders. Insects were used only once. Sample sizes per treatment and instar are provided in Supp Fig. 7 (online only). In comparing responses to different live plant volatiles, a plant cutting of an apical growing tip was placed in a water-filled 2 ml glass autosampler vial exposing the apical 1–2 cm of the cutting, and the opening sealed with parafilm. This was placed into one of the two 50 ml flasks attached upwind of the two arms of the Y-plate, with the control (a water-filled 2 ml glass vial sealed with parafilm, but no plant) in another flask. As the preferred host, tree-of-heaven volatiles served as a positive control by which to directly compare the attractiveness of other plant volatiles. Comparing volatiles to blank controls allowed us to assess attraction as well as repellency. For testing synthetic compounds, the upwind flasks each received a lure as described above containing either 1 mg of neat material, 100 µl of a blend, or a control. Prior to each bioassay session, five insects were tested in the apparatus without any volatiles on either side to ensure no directional bias due to visual cues or contamination. Each session started with a clean bioassay apparatus and tested up to 20 individual SLF. Each lure type was tested in at least two sessions rotating the position of the lure and control. The entire apparatus was cleaned thoroughly between each session and disposable parts were replaced.

Statistical Analysis

Olfactometer data were analyzed using a χ ² test with a null hypothesis that both arms would be selected with equal frequency. Significant differences occurred when G was greater than or equal to 3.841 with α < 0.05.

Results

EAD Responses to Host Plant Volatiles

A representative GC-EAD trace of a volatile profile from tree-of-heaven branches containing leaves and the corresponding antennal responses are shown in Fig. 1. GC-EAD traces displaying responses to the tree-of-heaven trunk, hops, black walnut, and spicebush are provided in Supp Figs. 3–6 (online only). In total, host plant aerations produced 43 unique responses by SLF antennae, with some species sharing the same volatiles (Table 3). Of those, 7 antennal responses were elicited by volatiles from tree-of-heaven branches with leaves, 6 from tree-of-heaven trunks, 14 from black walnut, nine from hops, and 7 from spicebush. Of these antennally active compounds, 18 compounds were identified, acquired, tested to verify antennal activity, and tested for attraction in the dual choice olfactometer either alone or in blends or both (Tables 1 and 2). Those results are described below.

Table 3.

Percent (±SE) relative abundance (ratio) of the 13 compounds of interest found in the volatile profiles of 17 plant species (dashes indicate the compound was not detected in any of the samples)

13 Compounds	Tree-of- heaven (n = 3)	Black locust (n = 2)	Wild grape (n = 2)	Black walnut (n = 2)	Hops (n = 2)	Oriental Bittersweet (n = 2)	Red Oak (n = 2)	Virginia Creeper (n = 2)	Chinaberry (n = 2)	Staghorn Sumac (n = 2)	Black Cherry (n = 2)	Spicebush (n = 2)	Tulip Tree (n = 2)	Red Maple (n = 3)	Black Birch (n = 2)	Norway Spruce (n = 2)	Apple (n = 2)	No. of species
(E,E)- α- Farnesene	46.2 (±10.4)	48.8 (±36.9)	32.5 (±4.5)	3.6 (±1.5)	17.4 (±8.5)	49.7 (±36.3)	3.7 (±3.7)	31.6 (±5.0)	12.7 (±8.4)	43.5 (±2.9)	9.1 (±9.1)	1.7 (±1.7)	30.8 (±14.0)	62.6 (±31.5)	49.2 (±26.1)	32.4 (±19.8)	18.8 (±18.8)	17
β-Caryophyllene	4.1 (±2.2)	12.4 (±6.4)	2.5 (±0.3)	20.6 (±12.7)	12.4 (±0.3)	0.9 (±0.9)	3.3 (±3.3)	15.8 (±10.2)	0.6 (±0.1)	12.9 (±6.8)	38.4 (±33.9)	82.4 (±0.5)	3.4 (±3.4)	4.7 (±2.5)	18.0 (±2.2)	64.6 (±16.7)	47.9 (±9.3)	17
Methyl salicylate	1.4 (±1.2)	7.8 (±7.3)	9.8 (±8.5)	0.8 (±0.7)	19.6 (±18.6)	15.4 (±14.8)	3.1 (±3.1)	5.0 (±1.9)	7.1 (±2.1)	27.7 (±15.8)	1.7 (±1.7)	1.2 (±1.2)	–	3.4 (±3.4)	–	–	33.4 (±9.4)	14
(E)-4,8-Dimethyl-1,3,7- nonatriene	13.0 (±3.0)	10.5 (±10.5)	45.1 (±3.3)	30.2 (±21.9)	1.1 (±1.1)	18.4 (±15.9)	75.5 (±24.5)	35.7 (±8.1)	30.3 (±25.5)	0.6 (±0.6)	25.9 (±7.1)	9.3 (±2.6)	58.9 (±7.9)	–	20.1 (±15.6)	–	–	14
(Z)-3-Hexenyl acetate	24.0 (±8.8)	9.9 (±5.9)	2.1 (±2.1)	13.0 (±5.3)	37.3 (±37.3)	0.2 (±0.2)	13.3 (±13.3)	6.5 (±4.4)	41.3 (±19.5)	–	15.3 (±15.3)	–	–	29.2 (±26.2)	12.7 (±12.7)	–	–	12
Linalool	6.8 (±1.7)	5.9 (±3.4)	5.2 (±5.2)	–	0.9 (±0.9)	–	–	2.9 (±1.0)	6.2 (±6.2)	0.5 (±0.5)	7.2 (±1.7)	4.4 (±2.6)	3.5 (±0.8)	–	–	2.5 (±2.5)	–	11
β-Ylangene	0.8 (±0.4)	3.2 (±3.2)	0.02 (±0.02)	1.7 (±1.2)	1.0 (±0.9)	0.8 (±0.8)	0.1 (±0.1)	2.6 (±0.6)	–	2.0 (±2.0)	–	–	3.4 (±3.4)	–	–	–	–	10
sulcatone	0.4 (±0.4)	0.6 (±0.6)	0.02 (±0.02)	tr	–	–	0.4 (±0.4)	–	–	–	2.5 (±2.5)	0.8 (±0.8)	–	–	–	–	–	7
(Z)-3-Hexenol	0.9 (±0.9)	0.3 (±0.3)	–	28.9 (±28.9)	–	–	0.6 (±0.6)	–	–	–	–	0.2 (±0.2)	–	–	–	–	–	5
Methyl benzoate	2.4 (±2.4)	–	2.7 (±2.7)	–	–	7.3 (±0.3)	–	–	–	12.8 (±12.8)	–	–	–	–	–	–	–	4
1-Octen-3-ol	0.1 (±0.1)	0.6 (±0.6)	–	–	10.3 (±9.1)	7.1 (±7.1)	–	–	–	–	–	–	–	–	–	–	–	4
(Z)-3-Hexenyl butyrate	–	–	0.2 (±0.2)	–	3.3 (±3.3)	–	–	0.1 (±0.1)	1.8 (±1.8)	–	–	–	–	–	–	–	–	4
4-Terpineol	–	–	–	1.1 (±0.0)	–	–	–	–	–	–	–	–	–	–	–	0.6 (±0.6)	–	2
Percent 13 compounds	54.3 (±7.2)	45.8 (±6.3)	39.6 (±1.9)	28.8 (±13.7)	42.0 (±20.3)	33.7 (±10.7)	12.8 (±8.2)	42.7 (±13.7)	35.9 (±8.9)	22.5 (±16.8)	19.9 (±13.2)	40.8 (±6.4)	12.6 (±7.1)	16.3 (±11.7)	408. (±22.5)	3.3 (±2.2)	2.2 (±0.6)
Percent other compounds	45.7 (±7.2)	54.2 (±6.3)	60.4 (±1.9)	71.2 (±13.7)	58.0 (±20.3)	66.3 (±10.7)	87.2 (±8.2)	57.3 (±13.7)	64.1 (±8.9)	77.5 (±16.8)	80.1 (±13.2)	59.2 (±6.4)	87.4 (±7.1)	83.7 (±11.7)	59.2 (±22.5)	96.7 (±2.2)	97.8 (±0.6)
No. of 13 compounds	11	10	10	9	9	8	8	8	7	7	7	7	5	4	4	4	3

tr—trace amount detected (< 0.01%)

Fig. 1.

GC-FID chromatogram of tree-of-heaven aeration and corresponding GC-EAD traces from fourth instar SLF antennae. Labeled peaks are: (Z)-3-hexenyl acetate (6), β-ocimene (7), linalool (10), DMNT (11), methyl salicylate (14), β-caryophyllene (17), and (E,E)- α-farnesene (18).

Attraction to Live Plant Volatiles

Nymphs of available instars of SLF were offered choices between volatiles released from host plants and blank controls. In general, when presented with a choice between any plant and no plant, nymphs and adults were attracted by the plant volatiles, and 7 of 14 plant species tested were significantly more attractive than blank controls by χ ² test (P < 0.05, Fig. 2). Not all stages could be tested to all plants due to limitations in time and availability. Stages that were significantly attracted to plants tested were: first instars to tulip tree and black walnut, second instars to wild grape, chinaberry, hops, black walnut, and tree-of-heaven, third instars to wild grape, chinaberry, and tree-of-heaven, and fourth instars to wild grape and staghorn sumac. Stages that did not significantly choose the plant over the controls were first instars to black cherry and Norway spruce, second instars to Medicago sativa L. (Fabales: Fabaceae) (alfalfa), Armoracia rusticana Gaertn (Brassicales: Brassicaceae) (horseradish), and oriental bittersweet, and third instars to hops, Ocimum basilicum L. (Lamiales: Lamiaceae) (sweet basil), and Asclepias syriaca L. (Gentianales: Apocynaceae) (common milkweed) (Supp Fig. 7 [online only]).

Fig. 2.

Frequency and direction of choice made by first through fourth instar SLF nymphs in response to choice between volatiles produced by various plants and either blank controls or volatiles from tree-of-heaven. Asterisks indicate that there was a significant difference (α < 0.05) between the two choices (χ ² test). Numbers inside bars represent number of insects responding. Numbers in parentheses indicates total number of insects tested, including nonresponders.

When nymphs were offered a choice between volatiles of tree-of-heaven and another host plant, the majority chose tree-of-heaven (Fig. 2; Supp Fig. 7 [online only]). Tree-of-heaven was chosen significantly more than chinaberry and Norway spruce by first instars, more than chinaberry, wild grape, and hops by second instars, more than wild grape, hops, and milkweed by third instars, and more than staghorn sumac by fourth instars (Supp Fig. 7 [online only]).

Attraction to Synthetic Compounds

Responses to synthetic compounds are summarized in Fig. 3 and detailed for each stage tested in Table 1 and Supp Fig. 8 (online only). Individual compounds that were significantly attractive to at least one stage were (Z)-3-hexenol, 1-octen-3-ol, sulcatone, methyl benzoate, racemic linalool, (E)-4,8-dimethyl-1,3,7-nonatriene (DMNT), 4-terpineol, methyl salicylate, β-ylangene, β-caryophyllene, and (E,E)- α-farnesene. None of the individual compounds tested were more attractive in comparison to methyl salicylate with the exception of sulcatone (Table 1, Fig. 3, also see Supp Fig. 8 [online only]). Each of these compounds was tested at 1 mg except for β-ylangene which was tested at 40 µg due to limited availability, yet is was still among the most highly attractive compounds. Although β-ocimene elicited an antennal response, and was present in volatile profiles of all 17 plant species examined, the β-ocimene tested in behavioral bioassays was a mix of isomers (Sigma–Aldrich) and was not found to be attractive.

Fig. 3.

Frequency and direction of choice made by all SLF stages combined in response to individual compounds and either blank controls or volatiles from tree-of-heaven. Asterisks indicate that there was a significant difference (α < 0.05) between the two choices (χ ² test). Numbers inside bars represent the number of insects responding. Numbers in parentheses indicate the total number of insects tested, including nonresponders. DMNT, 4,8-dimethyl-1,3,7-nonatriene.

Attraction to Synthetic Blends

Results of behavioral bioassays examining 20 different blend formulations for attraction are summarized in Table 2 and Fig. 4. Blend 2B, a 70:30 mixture of methyl benzoate and (Z)-3-hexenyl acetate, was significantly more attractive than controls for fourth instars. Blend 3C (containing three esters) was significantly more attractive than Blend 3G and Blend 3H (containing three alcohols). Blends 6A, 6B, and 7A, which each contained 0.5 mg methyl salicylate and either 5 or 6 additional compounds, were all significantly more attractive than 0.5 mg of methyl salicylate alone for early females. Blend 7B was significantly more attractive than 1 mg of methyl salicylate, which is double the amount of methyl salicylate contained in the blend. Blend 7C, which contained 0.5 mg methyl salicylate plus 6 other compounds, was significantly more attractive than 2 mg of methyl salicylate alone, which is four times the amount of methyl salicylate in the blend. Overall, blends with more compounds attracted more SLF than blends with fewer compounds.

Fig. 4.

Choices made by SLF in a dual choice olfactometer walking bioassay when presented with choices between blends of different compounds or methyl salicylate at different multiples compared to the contents of the competing blend. Asterisks indicate that there was a significant difference (α < 0.05) between the two choices (χ ² test). Numbers inside bars represent the number of insects responding. Numbers in parentheses indicates the total number of insects tested, including nonresponders.

Comparison of Host Plant Volatiles

The 13 compounds that were found to produce attraction were qualitatively and quantitatively examined in 17 host plants, including Norway spruce as a nonhost outgroup, and host plants were found to share many EAD active volatiles (Table 3, Fig. 6). Representation of each kairomone by percentage of relative abundance in each species is provided in Table 2 and Fig. 6. Present in all species were (E,E)- α -farnesene and β-caryophyllene, whereas methyl salicylate, DMNT, (Z)-3-hexenyl acetate, linalool, and β-ylangene were present in 10 or more of the 17 species analyzed. Relative to the total volatile profiles of each species, tree-of-heaven contained the highest proportion (54.3%) and number (11) of the 13 volatiles found to be attractive alone or in a blend, followed by black locust (45.8%, 10), wild grape (39.6%, 9), and black walnut (28.8%, 9). On the other end of the spectrum were Norway spruce (3.3%, 4) and apple (2.2%, 3) volatile profiles which were represented by the fewest and lowest percentage of the 13 identified kairomones (Table 3, Fig. 6).

Discussion

Our experiments provide evidence that host plant volatiles are perceived by SLF and help mediate attraction to, and selection of, potential host plants (Figs. 1 and 2). Host plant volatiles were clearly perceived and attractive to SLF when isolated as an olfactory stimulus, apart from visual cues, as shown in olfactometer bioassays (Fig. 3). These findings directed us toward GC-EAD experiments to identify the specific antennally active volatiles which mediate their directional choice. When compared head to head, no plant was more attractive than tree-of-heaven (Fig. 2), which was more attractive than hops, wild grape, chinaberry, staghorn sumac, milkweed, and Norway spruce. This aligns with strong evidence that although SLF are polyphagous, they feed preferentially on tree-of-heaven, and that preference starts off less pronounced and increases through the season as they develop from first instar nymph to sexually mature adult (Lee et al. 2009, Park et al. 2009, Kim et al. 2011, Song et al. 2018, Liu 2019, Murman et al. 2020). Their feeding plasticity suggests that if their preferred host, tree-of-heaven, might be unavailable or declines below a critical point of suitability due to overfeeding, a number of alternative hosts would be attractive, especially if they produce the same kairomones in similar ratios or abundance. Common host plant volatiles as a mechanism for host-switching has been described previously in the apple fruit moth, Argyresthia conjugella (Bengtsson et al. 2006), and may also play a role in the formation of host races in Rhagoletis flies (Linn et al. 2003, 2005).

Qualitative analyses and comparison of relative proportions of headspace volatiles from host plants revealed many shared kairomones, and host plants had different numbers and combinations of these compounds (Tables 3 and 4, Fig. 6). Although we were not able to control for all environmental factors which can produce variation in plant volatile emissions, these analyses may help to explain strong attraction in two hosts that share only a few kairomones, such as tree-of-heaven and tulip tree. Tulip tree is native to eastern North America is a new association for the invasive SLF, and was attacked heavily by SLF in the field despite being a poor host (Murman et al. 2020). Curiously, tulip tree volatiles were the most attractive when tested against tree-of-heaven volatiles in behavioral bioassays (Fig. 2). The initial attraction of SLF to nonviable host plants (on which they cannot complete development, including tulip tree, wild grape, and staghorn sumac, which are native to North America), may be due to key volatile cues being shared with viable host plants from their native range (such as tree-of-heaven, chinaberry, and oriental bittersweet) (Murman et al. 2020). Of the 13 attractive compounds, only five were detected in tulip tree volatile profiles, which lacked methyl salicylate (Table 3, Fig. 2). One possible explanation is that the blend of those five compounds is extremely attractive to SLF. The presence of comparable amounts of β-ylangene and DMNT in tulip tree and tree-of-heaven supports this hypothesis. Another possibility is that some of the roughly 20 remaining unidentified compounds that elicited antennal responses might be present in tulip tree and contribute to its attraction. More research is needed to identify and test the remaining antennally active unknown host plant volatiles that were found (Supp Table 2 and Figs. 3–6 [online only]).

SLF antennae respond to numerous volatiles from a wide range of host plants (Fig. 5, Supp Table 2 and Figs. 3–6). In total, SLF antennae responded to ~40 unique compounds across tree-of-heaven, black walnut, hops, chinaberry, grape, and spicebush. This broad antennal sensitivity to host plant volatiles has precedent in other polyphagous herbivores, like the moth Spodoptera littoralis, where differences in the proportion of EAD active host plant volatiles allow discrimination between host species (Conchou et al. 2017). In this study, the EAD active compounds included green leaf volatiles (GLVs), esters, monoterpenes, sesquiterpenes, and phenylpropanoids, among others (Fig. 5), including many compounds previously studied in the context of plant–insect interactions (Bruce et al. 2005). The GLVs and terpenoids are well studied in the context of tri-trophic interactions, where they often function as signals of herbivore damage to host or prey-seeking parasitoids and predators (Turlings and Erb 2018, and references therein), or as chemicals which enhance the response of insects to sex pheromones (Dickens et al. 1990, Landolt and Phillips 1997). From volatiles produced by tree-of-heaven, the strongest antennal responses were seen to linalool, DMNT, and methyl salicylate (Fig. 1). These were not the most abundant antennally active compounds in headspace samples, indicating that the antenna has been ‘tuned’ to these compounds, relative to others. As antennal responses do not impart information about behavior (Evans and Allen-Williams 1992), behavioral bioassays were needed to determine whether these compounds are attractive or repellent, or only elicit a behavior when part of a blend.

Fig. 5.

Chemical structures of 13 host plant volatiles that were found to be attractive either individually or in a blend.

Of the 45 compounds that elicited antennal responses, and some chosen based on prevalence in aeration samples, 18 compounds were identified and obtained for testing in olfactometer bioassays (Table 1). No individual compound tested had a repellent effect. This contrasts with research in other polyphagous Hemipterans, like the black bean aphid Aphis fabae, where components of an attractive blend of host plant volatiles are repellent when presented individually (Webster et al. 2010). Here, when tested individually against a blank control, 11 of 18 compounds stimulated attraction. Only sulcatone was more attractive than methyl salicylate when compared directly (Table 1, Fig. 3). Initial field experiments using sulcatone as a lure did not find it to be more attractive than methyl salicylate (MFC, unpublished). The results of these comparisons support the idea that methyl salicylate is a key kairomone that mediates SLF attraction to host plants (Cooperband et al. 2019), although the attraction to tulip tree volatiles (Fig. 3) which do not contain methyl salicylate (Fig. 6) suggests there are redundant volatiles which can mediate attraction.

Fig. 6.

Average percentage of 13 identified kairomones (and all other compounds combined) with relation to the total volatile profile composition of 17 host plant species. Above each column is the number of the 13 kairomones that were detected in each plant species.

Experiments testing the attractiveness of blends demonstrated that SLF prefer multi-component blends over methyl salicylate alone, even when methyl salicylate is presented in up to 4x higher amounts than in the blend. While it was not feasible to compare all blend combinations or each blend to increasing doses of methyl salicylate, we found 5 blends that were preferred over methyl salicylate alone. These blends all contained methyl salicylate, 1-octen-3-ol, (Z)-3-hexenol, and (Z)-3-hexenyl acetate, and varied in the presence of (E,E)- α-farnesene, methyl benzoate, DMNT, and (Z)-3-hexenyl butyrate. Blend 7B was arguably the best performing blend in olfactometer testing. This blend may be a good candidate for further development into lures and field testing, an endeavor that will pose unique formulation challenges based on the different volatilities of the constituents.

SLF exhibit both generalist and specialist tendencies in feeding upon a diverse array of host plants, but clearly having a preference for tree-of-heaven over the other species. Herbivore generalists and specialists may process and respond to arrays of host plant volatiles in vastly different ways, with generalists perhaps having more behavioral plasticity and influence of learning than specialists (Blust and Hopkins 1987, Gols et al. 2012). In our studies, SLF were able to discriminate and significantly choose tree-of-heaven odors over 6 of 12 other potential host plant species tested (Fig. 2), a strong indication of host preference for such a generalist herbivore. Studies of other diurnal insect herbivores like butterflies show that olfactory information is not always enough to discern potential hosts, even in specialist species, pointing to the importance of visual cues in addition to olfaction (Schäpers et al. 2015).

We have observed that SLF frequently probe plant and other surfaces with their proboscis, on which they have numerous gustatory, olfactory, and mechanosensory receptors (Hao et al. 2016). Upon locating a host plant, the decision to either stay and feed or move on is likely guided by these sensory inputs in addition to phloem pressure, nutritional requirements, and physiological state. One such example is mating status, which has been shown to influence both flight propensity and performance in adult females, through a negative correlation with weight (Wolfin et al. 2019). Research examining the anemotactic flight capabilities of adult SLF showed that SLF primarily orient to wind direction, using an ‘aim then shoot’ strategy to launch themselves from tall objects, gradually descending and being incapable of generating enough lift to gain altitude (Baker et al. 2019, Myrick and Baker 2019, Wolfin et al. 2019). Myrick and Baker (2019) found no evidence for the role of olfaction in guiding adult flight bouts. Our study provides contrasting evidence by demonstrating that all stages of SLF are capable of perceiving specific host plant volatiles (Supp Figs. 1, 3–6 [online only]) and orienting to olfactory cues from host plants in olfactometer bioassays (Figs. 2–4). While our experiments evaluated walking responses (which apart from jumping, is the only mode of locomotion for all immature life stages), future research could evaluate flight responses in a wind tunnel bioassay to better understand how plant volatiles and visual cues mediate long-range host detection in adult SLF, as has been studied in variety of pest insects (Rojas and Wyatt 1999, Brévault and Quilici 2010). How host plants may alter volatile emissions as a result of SLF feeding damage, and if those changes are exploited by other SLF, or by predators or parasitoids, is another topic of ongoing investigation.

Acknowledgments

We thank the Pennsylvania Department of Agriculture and numerous community members in the Lehigh Valley for helping to locate SLF infested properties, granting access to property, and collecting SLF that were used in this study. We thank Allard Cossé for providing purified β-ylangene and for helpful comments on the later stages of this manuscript. This project was funded by USDA Farm Bill Section 10007 in 2015 (6.0380.04, 6.0703.00), 2016 (3.0187.01, 3.0187.03), 2017 (6R.3.0320.01, 6R.3.0320.03), and 2018 (6R.0679.02). Mention of a commercial product does not constitute its endorsement by the USDA.