ARTHROPODS IN RELATION TO PLANT DISEASES Sharpshooter X Wave: Correlation of an Electrical Penetration Graph Waveform With Xylem Penetration Supports a Hypothesized Mechanism for Xylella fastidiosa Inoculation

ELAINE A. BACKUS,1 WENDY J. HOLMES,2 FRED SCHREIBER,2 BRENDON J. REARDON,3,4 3 AND GREGORY P. WALKER

Ann. Entomol. Soc. Am. 102(5): 847Ð867 (2009) ABSTRACT Electrical penetration graph (EPG) monitoring is the most rigorous means of obser- vation and quantiÞcation of feeding by piercingÐsucking arthropods. Previous EPG studies with aphids and leafhoppers have demonstrated that the X wave identiÞes when the stylets of these phloem ßuid-ingesting insects make contact with their preferred plant vascular cell, phloem sieve elements. This article presents the Þrst direct evidence of an X wave identifying ingestion from a xylem tracheary element by a xylem ßuid-ingesting type of leafhopper Homalodisca liturata Ball (Hemiptera: Cicadel- lidae: Cicadellinae), whose waveforms are nearly identical to those of the glassy-winged sharpshooter, Homalodisca vitripennis (Germar). We document consistent association of the sharpshooter X wave with salivary sheath termini in xylem, especially ligniÞed secondary xylem cells, and absence of the X wave in the rare instance of ingestion from a nonxylem cell. The sharpshooter X wave is a complex, multicomponent waveform, composed of X wave-speciÞc variants of waveform subtypes B1w (rep- resenting salivation), B1s (representing precibarial valve movement for tasting), types C1 (a new waveform type that may represent egestion) and C2 (a new designation for the waveform type representing ingestion/cibarial pumping). It is proposed that the sharpshooter X wave represents a blended suite of behaviors that function to 1) physically seal stylet tips into the cell via sheath salivation, 2) repeatedly taste then eject (egest) chemical constituents of the cell to determine acceptability, and 3) mechanically test the strength of the stylet seal via trial cibarial pumping (ingestion). It is further hypothesized that the X wave represents vector behaviors that control inoculation of the PierceÕs disease bacterium, Xylella fastidiosa. The ingestion-(salivation and eges- tion) hypothesis is stated for the mechanism of transmission of X. fastidiosa.

KEY WORDS electrical penetration graph, Homalodisca spp., PierceÕs disease, probing, stylet pen- etration

Sharpshooters (Hemiptera: Cicadellidae: Cicadelli- tor of X. fastidiosa in California on host plants such as nae) are large, xylem ßuid-ingesting leafhoppers that oleander, almond, citrus, and nursery trees (Redak et are capable of transmitting (i.e., acquiring and inoc- al. 2004). These crops and others in California are at ulating) the bacterium Xylella fastidiosa, the causative risk of epidemic PD due to introduction of H. vitrip- agent of PierceÕs disease (PD) of grape (Vitis spp.) ennis. Efforts are underway to develop host plant re- and other leaf scorches into host plants. The glassy- sistance to the vectorÕs ability to transmit of X. fastid- winged sharpshooter, Homalodisca vitripennis (Ger- iosa; however, lack of fundamental understanding of mar), formerly Homalodisca coagulata (Say) (Takiya the transmission mechanism and its impact on PD et al. 2006), is an exotic, economically important vec- epidemiology currently impedes progress. For the past 6 yr, Backus and colleagues have been researching the basic question of how sharpshooter Mention of trade names or commercial products in this article is solely for the purpose of providing speciÞc information and does not feeding controls transmission (especially inoculation) imply recommendation or endorsement by the U.S. Department of of X. fastidiosa, with the goals of deÞning the speciÞc Agriculture. inoculation behaviors and identifying the instant of 1 Corresponding author: USDAÐAgricultural Research Service, San inoculation during the feeding process, in real-time. Joaquin Valley Agricultural Sciences Center, 9611 S. Riverbend Ave., Parlier, CA 93648 (e-mail: [email protected]). Our primary tool is electrical penetration graph 2 Department of Biology, California State University, Fresno, CA (EPG) monitoring, the most rigorous and detailed 93710. method of observing and quantifying the feeding of 3 Department of Entomology, University of California, Riverside, piercingÐsucking arthropods (Walker 2000). This CA 92521. 4 Current address: USDAÐAnimal and Plant Health Inspection Ser- study is the fourth in a series of articles that have 1) vice, 350 Corporate Blvd., Robbinsville, NJ 08691. characterized the main sharpshooter waveforms (rep- 848 ANNALS OF THE ENTOMOLOGICAL SOCIETY OF AMERICA Vol. 102, no. 5 resenting stylet penetration [probing]) of the blue- The uniqueness of the stylet activities during the green sharpshooter, Graphocephala atropunctata (Si- aphid X wave was supported by the discovery that the gnoret) (Almeida and Backus 2004), and the glassy- pds immediately preceding sieve element phase (later winged sharpshooter (Backus et al. 2005); 2) termed R-pds; Tjallingii and Gabrys 1999) were longer performed histological correlations to identify the in duration in some aphid species than pds elsewhere host plant tissues in which each waveform occurs in the probe. R-pds are thought to represent intracel- (Backus et al. 2005); and 3) performed correlations of lular punctures into phloem cells, as opposed to non- waveforms with speciÞc stylet activities by observing phloem cells along the pathway. Reese et al. (2000) stylet movements in transparent, artiÞcial diet (Joost also rediscovered the Þnding of McLean and Kinsey et al. 2006). (1967) that their “X waves” (actually pathway pds) are The present work builds on the previous studies by recognizable during pathway, in addition to immedi- characterizing more waveform types, completing our ately preceding sieve element phase. However, be- histological correlations, and using them to hypothe- cause these partial X waves were correlated with in- size the mechanism of X. fastidiosa inoculation. The tracellular punctures of mesophyll/parenchyma cells Backus et al. (2005) study left some unanswered ques- along the stylet pathway (McLean and Kinsey 1967), tions such as: 1) During which ingestion events in a rather than with sieve elements, Reese et al. (2000) probe are the stylets deÞnitely in and ingesting from felt their existence confused the deÞnition of X waves xylem? 2) Is there any point during stylet penetration, and did not advocate use of the term. before ingestion ensues, when the stylets penetrate In contrast, we believe that the concept of the X xylem cells? and 3) Can stylet penetration and inges- wave has great practical and heuristic value for EPG tion speciÞcally from xylem cells be reliably deter- researchers, if it is deÞned carefully and used as part mined via EPG waveforms alone, without histology? of a broader, behavioral/ecological concept for The speciÞc objectives of this study were to answer hemipteran probing behavior. McLean and Kinsey, the above-mentioned questions, by histologically cor- due primarily to the limitations of their instruments relating individual pathway and ingestion events in (Backus et al. 2000, Tjallingii 2000), used a “top-down” identiÞed probes with cell type penetrated. research approach of identifying broad concepts and The ultimate goal of this study was to search for a phases of stylet penetration, of necessity leaving the putative “sharpshooter X wave.” In EPG parlance, the details to later researchers. Tjallingii and colleagues X wave is a complex, stereotypically repeating pattern used the bottom-up approach of elegantly and pains- of waveforms that represents stylet contact and sub- takingly characterizing all waveform types and sub- sequent activities (such as ingestion or salivation) in types Þrst, then assembling their blended meanings a hemipteran speciesÕ preferred ingestion cell type; up into phases and concepts. Over time, these comple- to now, strictly phloem sieve elements. X waves are mentary styles have provided greater richness of in- important because they, or their components, have formation about aphid stylet activities than could ei- been shown to control inoculation of noncirculative ther method alone. In light of this ultimate success, we plant pathogens. Our study uses deductive, “bottom- have applied both top-down and bottom-up ap- up” reasoning, by Þrst identifying when xylem ele- proaches to our studies of sharpshooter waveforms. ments were penetrated, the stylet activities that oc- The current study employs the bottom-up approach. curred at that time, then searching for a stereotypical Based on the evidence here, we advocate restored use waveform that could be called the X wave, and Þnally, of the X wave concept but more precisely deÞned than hypothesizing its role in inoculation of X. fastidiosa. in previous literature. The concept of the X wave as a landmark waveform The X wave concept has proven invaluable for EPG was one of the earliest and most signiÞcant contribu- studies of leafhoppers and planthoppers. In recordings tions of the original inventors of EPG (McLean and of 10 species of phloem ßuid-ingesting Deltocephaline Kinsey 1965, 1967). McLean and Kinsey (1967) cor- leafhoppers (Triplehorn et al. 1984, Rapusas and Hei- related both the aphid X wave and the regular, long- nrichs 1990, Wayadande and Nault 1993) and one duration waveform following it with stylet tips in a species of delphacid planthopper, Nilaparvata lugens phloem sieve element, the preferred ingestion cell Stål (Velusamy and Heinrichs 1986, Kimmins 1989), type of aphids. Years later, the very important poten- researchers have identiÞed highly recognizable, spe- tial drop (pd) waveform was discovered and corre- cies-speciÞc, stereotypically repeating waveforms that lated with intracellular punctures by aphid stylets, of deÞnitively identiÞed sieve element penetration. both mesophyll/parenchyma cells along the stylet They also found that long-duration waveforms follow- pathway and of phloem sieve elements (Tjallingii ing the X wave represent stylet activities occurring 2000). Subtypes of the pd waveform have been Þrmly within the same sieve element. associated with acquisition and inoculation of non- None of the above-mentioned studies characterized persistent viruses via ingestion and salivation, respec- the component waveform types and subtypes within tively (Powell et al. 1995, Martin et al. 1997, Fereres and their X waves. However, in one study, the Þne struc- Collar 2001). These waveform concepts were brought ture of leafhopper X waves was correlated with virus together when Reese et al. (2000) identiÞed the pd plus inoculation (Wayadande and Nault 1993). X waves a return to pathway activities (including salivation) as were statistically similar in appearance among even the portions, or waveform components (representing distantly-related vectors of Maize chlorotic dwarf vi- associated stylet activities) of the aphid X wave. rus (family Sequiviridae, genus Waikavirus, MCDV), September 2009 BACKUS ET AL.: SHARPSHOOTER XWAVE 849 whereas nonvectors had statistically dissimilar X substrate voltage (AC or DC), type of signal process- waves. MCDV is a foregut-borne, semipersistently ing circuitry (AC or DC), and input impedance (Ri, transmitted virus whose inoculation mechanism prob- 106,107,108,109,or1013 ohms). The DC signal pro- ably involves egestion (also termed extravasation; cessing channel was used for all analyses. EPG record- McLean and Kinsey 1984). ings were made with an input impedance of 107 ohms Wayadande and Nault (1993) postulated that some (a few), 108 ohms (most) (experiment 1), or 109 ohms part of the X wave represents egestion (although it (experiment 2). These higher input impedances, com- was termed extravasation in their article). Thus, be- pared with the 106-ohms level published previously for cause of the potential role and value of the leafhopper glassy-winged sharpshooter (Backus et al. 2005) and X wave for understanding vector transmission, there is blue-green sharpshooter (Almeida and Backus 2004), a need to better understand the biological meanings of were chosen to enhance the amplitude of the inges- leafhopper X wave components and their represented tion waveforms of greatest interest for this study, stylet activities. Sharpshooters are excellent candi- while still balancing R- and emf-components (Backus dates for such studies, because their large body size and Bennett 2009). allows physiological studies not possible with smaller Plant tissue surrounding the salivary sheath (left insects (e.g., Dugravot et al. 2008). behind by each probe) was carefully marked with a In the present work, we studied a proxy for the permanent, Sharpie felt pen at the beginning of a quarantined glassy-winged sharpshooter, its closely probe during pathway phase (Backus et al. 2005), related native California congener Homalodisca litu- when the insect seemed less likely to be startled by the rata Ball (Hemiptera: Cicadellidae). We document for brief operator hand movements necessary to mark the the Þrst time the existence of an X wave for a xylem tissue. After recording was completed, all EPG wave- ßuid-ingesting leafhopper, and we also present the forms were viewed and measured with the Windaq ingestion-(salivation and egestion) hypothesis for waveform browser software (Dataq Instruments). transmission of X. fastidiosa by sharpshooters during The recorded waveforms of H. liturata were nearly the X wave. identical to published waveforms from the glassy- winged sharpshooter (Backus et al. 2005, Sandanayaka and Backus 2008) and blue-green sharpshooter Materials and Methods (Almeida and Backus 2004), with the minor excep- Insect and Plant Rearing. Adult H. liturata were tions noted in Discussion due to difference in input Þeld-collected on jojoba, Simmondsia chinensis impedance. (Schneid), in Riverside, CA, and maintained on one Sharpshooter Waveform Terminology. The current plant each of cowpea, Vigna unguiculata ssp. dekind- study uses the EPG terminology of Backus (2000) and tiana (Harms); sunßower, Helianthus annuus Nutt.; Backus et al. (2005). A probe is deÞned as all behaviors basil, Ocimum basilicum L.; and sorghum, Sorghum occurring from start of one stylet insertion until with- bicolor ssp. bicolor (L.) Moench, caged together. The drawal of the stylets from the plant; the overall action insects and plants were maintained under greenhouse is probing, synonymous with stylet penetration. A conditions at the USDAÐAgriculture Research Service waveform event is the duration within a probe during (ARS), San Joaquin Valley Agricultural Sciences Cen- which an uninterrupted occurrence of one waveform ter in Parlier, CA, as described in Dugravot et al. type or subtype is performed. Thus, each probe is (2008). Natural lighting was supplemented by artiÞ- comprised of an unbroken sequence of waveform cial light to achieve a photoperiod of 16:8 (L:D) h, events. In highly regular, repetitive waveforms such as with daytime temperatures from 24 to 29ЊC and over- C (cibarial pumping), each repeated unit is termed an night temperatures from 18 to 21ЊC. Test cowpea episode (Dugravot et al. 2008). plants were 3 and 2 wk old for experiments 1 and 2 (see Sharpshooter waveforms are hierarchically catego- below), respectively. rized. The highest level is the phase (general category;

EPG. After a brief immobilization with CO2, adult the most coarse, compressed view of the waveform). sharpshooters were tethered with a 63.5-␮m (0.0025- The major phases of sharpshooter stylet penetration in. diameter gold wire (Sigmund Cohn Corp., Mt. are pathway (formation of the salivary sheath, search- Vernon, NY) by using silver print paint (LADD Re- ing for a xylem cell), ingestion (cibarial pumping and search Industries, Williston, VT; n-butyl acetate sol- swallowing), and interruption (two types: pathway- vent), as described in Dugravot et al. (2008). Wired like and nonpathway interruptions). Pathway-like in- insects were starved for 0.5Ð3 h and then placed on the terruptions represent the same behaviors as pathway test plant. phase. Until the current study, however, the biological Sharpshooter stylet penetration was recorded with meanings of nonpathway interruptions were un- WINDAQ Proϩ data acquisition software by using a known because waveform subtypes within could not DI 720 analog/digital board (Dataq Industries, Akron, be resolved using lower input impedances. Within the OH) at a sample rate of 100 Hz. Substrate voltage for phase are waveform families (medium level of com- all recordings was 20Ð25 mV AC. We used an AC-DC pression, assigned uppercase alphabetic designation correlation monitor (designed by W. H. Bennett [Uni- such as A, B, C, and so on). For example, B family versity of Missouri, retired] and E.A.B. and built by waveforms (B1 and B2) all occur when stylets are W. H. Bennett; Backus and Bennett 2009). The new deep within plant tissue and forming the salivary AC-DC monitor allows the user to choose the type of sheath. Within a family are waveform types (Þne- 850 ANNALS OF THE ENTOMOLOGICAL SOCIETY OF AMERICA Vol. 102, no. 5 structure view of the waveform, assigned an additive through seventh, providing recordings of 10 probes for numerical designation, e.g., B1). In some cases, even almost all C event orders). This allowed us to selec- subtypes have been assigned when correlations justify tively determine in which cell types the salivary sheath a different biological deÞnition. These are the most branch tips laid for each event order. In addition, highly expanded, ampliÞed views possible in Windaq; two probes were terminated during pathway and designated by additive lowercase letters (such as B1s three during waveform G. Thus, 75 probes from Ϸ40 and B1w). insects were artiÞcially terminated. Data were an- The most up-to-date summary of all waveform fam- alyzed in depth, providing most of the results de- ilies, types, and subtypes for sharpshooters, with their scribed here. biological meanings, is presented in Sandanayaka and Experiment 2. This was similar to experiment 1, ex- Backus (2008). More detailed waveform characteriza- cept 1) probing was terminated during (nonpathway tion is found in Backus et al. (2005; Table 1). interruption) N events, of Þrst through third orders The waveforms most important for this study are plus sixth (10 probes each); and 2) each recording summarized, below (from Backus et al. 2005, Joost et used a new cowpea plant. We attempted to use radio al. 2006, Backus et al. 2005, Holmes 2007, Dugravot et frequency cauterization (Fisher and Frame 1984) to al. 2008, Backus and Dugravot 2009). stylectomize sharpshooters during probing. Stylec- Pathway Phase. A1: Movement of both mandibular tomy was attempted on 40 insects; however, it was and maxillary stylets with simultaneous formation of only partially successful for most, due to the stoutness salivary sheath, especially the thick-walled trunk of of sharpshooter stylets. Ten stylet bundles were suc- the salivary sheath. cessfully severed, but only six were still intact after B1: Penetration of the maxillary stylets past the histological processing. Although the stylets fell out of trunk, further salivation (subtype B1w) and formation all six during processing, they all left distinctly hollow of sheath branches, with concomitant ßuttering of the salivary sheath branches to terminal cells. Thus, all precibarial valve to facilitate ßuid tasting (subtype B1s). such tissues were kept for analysis. Probing of an B2: Maxillary stylet sawing through tough tissues or additional 15 sharpshooters was artiÞcially terminated the walls of solid salivary sheath; indication of new in the same manner as for experiment 1; 12 more tissue salivary sheath branching or extension of an existing blocks containing salivary sheaths were intact after branch. B2 often occurs at the bottom of a steep drop processing, for a total of 18. Data were not analyzed in overall voltage level, termed a B2 trench. Such in-depth as for experiment 1 but used to answer re- trenches were correlated with partial stylet with- maining questions about N event correlations. drawal and branching of the salivary sheath, in H. Histology and Light Microscopy. Blocks of cowpea vitripennis recordings (Backus et al. 2005). stem tissue, 2Ð3 mm in length and each containing a Ingestion Phase. C: Cibarial pumping for active in- salivary sheath, were excised and Þxed for a minimum gestion. Correlated with cibarial dilator muscle move- of 24 h (the Þrst 2 h under light vacuum) in a modiÞed ments, C also portrays directionality of ßuid ßow via (Holmes 2007) CRAF 3 Þxative (Berlyn and Miksche streaming potentials (Dugravot et al. 2008). Once in- 1976) within 1Ð3 h of feeding. Samples were safely gestion phase is reached, ingestion (C) events alter- stored in the Þxative for up to 2 mo until enough nate with interruption (N) events. The order of a samples were accumulated for further processing. Tis- particular C event in this sequence is noted as 1stC, sue samples then were rinsed in HEPES wash buffer, 2ndC, and so on. dehydrated in a graded series of ethanol/tert-butanol, Interruption Phase. N: nonpathway interruption of inÞltrated with Paraplast X-tra (at 52ЊC), and then ingestion. embedded in Paraplast Ultra at 56ЊC (Berlyn and Mik- NA1, NB1, and/or NB2: Collectively designated sche 1976). ParafÞn blocks were serially thick-sec- pathway-type interruption; same as A1, B1, and B2, tioned at 10 ␮m by using a MICROM HM355 Inter- respectively, but performed after the Þrst C waveform national rotary microtome. Ribbons were mounted on event. This is a return to pathway activities to create slides coated with 5% (wt:vol) ammonium hydroxide a new salivary sheath branch. Similarly to C event and allowed to dry overnight on a slide warmer order, the order of interruption events is designated as (Fisher, Santa Clara, CA). All slides were dewaxed in 1stN, 2ndN, and so on, starting with the Þrst N after the 100% xylene, stained with aqueous safranin (0.5%), Þrst C event (Backus et al. 2005). For event order, N and then counterstained with ethanolic fast green is used to designate both pathway and nonpathway (0.01%) (Berlyn and Miksche 1976, Ruzin 1999). The interruptions. freshly stained slides were coverslipped using Per- R: A ßat-line waveform that probably represents mount mounting medium (Fisher). stylets motionless, resting shallowly in the plant. Forty-Þve tissue blocks were intact after histologi- Experimental Designs. Experiment 1. Each insect cal processing. Salivary sheaths were examined with a was allowed to perform one to four spatially separated, DM 5000B light microscope (Leica, DeerÞeld, IL) EPG-recorded probes (i.e., time periods comprising using both brightÞeld and polarized light, and imaged stylet penetration) on the main stem of the test plant. via a microscope-mounted Leica DC 500 digital cam- When an (ingestion phase) C event was observed, era and Adobe Photoshop 7.0 (Adobe Systems, Moun- probing was artiÞcially terminated by rapidly plucking tain View, CA). Polarized light caused ligniÞed cell the insect from the plant. Probing was artiÞcially ter- walls to shine brightly, allowing identiÞcation of non- minated 10 times for each C event order from Þrst ligniÞed protoxylem versus ligniÞed metaxylem, sec- September 2009 BACKUS ET AL.: SHARPSHOOTER XWAVE 851

1st X wave 2nd X3rd X4th X

b c

3rd 3rd 4th path 1stC 1stN 2ndC 2ndN C N C a -way

End of pathway 2ndN 2nd C C1 B1s * * * * * * * * * * *

b B1w c Fig. 1. (a) Four sharpshooter X waves. Each X wave phrase is composed of N and C events. This is a visually striking example of sharpshooter X waves because the waveforms were stable around 0 V., due to poor wire contact causing loss of voltage level information. This also caused the C2 plateaus to be unusually triangular. Windaq gain 1ϫ. (b and c) Expansions of boxes b and c in a, showing details of the end of pathway (“N”) and C portions of the X waves. Downward-pointing arrows, B1s spikelet bursts; *, C1 peaks; upward-pointing arrowheads, fB1w. The same symbols are repeated in other Þgures. Components are further described in Results. Windaq compression 12; 2.4 s/div. ondary xylem, and cortical Þber cells. It is likely that generally possible to determine the number of ligniÞed xylem was functional for translocation (T. branches and their cell termini by examining multiple Rost, personal communication), and therefore most sections of the sheath in relation to the EPG wave- useful for ingestion by a sharpshooter. Reference im- forms and to assign an estimated order for their for- ages of experiment 1 and experiment 2 plants, in both mation. brightÞeld and polarized light, are available in Holmes Statistical Analysis. We used a repeated measures (2007). analysis of variance (ANOVA) (restricted maximum As described in Backus et al. (2005), it was possible likelihood estimation) (PROC MIXED, SAS Institute to deÞnitively identify the terminal sheath branch 2001, Sandanayaka and Backus 2008) and least signif- (which marked the location of the stylet activities icant difference (LSD) test (LSMEANS option; SAS represented by the last waveform before artiÞcial ter- Institute 2001), to determine whether the durations of mination of the probe) via one of two methods. Either pathway and C events were signiÞcantly different the sheath was single-branched, or (in multibranched within each correlation. The dependent variable of sheaths) the terminal branch was hollow because the the model was mean waveform duration per event; the rapid removal of the insect precluded it from back- random effect was probe number. Log transformation Þlling that salivary sheath branch. In the case of multi- was performed before ANOVA to reduce variability branched sheaths, the last (i.e., hollow) sheath branch and improve homogeneity. For frequency measures that was produced before artiÞcial termination of (numbers of events or branches), we pooled data into feeding was termed the terminal branch; others were 2 by 2 contingency tables and used Fisher exact test termed abandoned branches. Only two of the 45 sal- (PROC FREQ, SAS Institute 2001). Means were usu- ivary sheaths had multiple, Þlled branches with no ally considered signiÞcantly different at ␣ ϭ 0.05, hollow branch. In those cases, less deÞnitive but still except in one case when P ϭ 0.0502 was accepted as strong correlation was made by comparing the EPG signiÞcant. waveform with the sheath branching pattern (see Correlation of Waveforms with Penetrated Cell Results Types). For salivary sheaths with fewer than six branches, As foreshadowed in Fig. 1, the sharpshooter X wave the order of formation of sheath branches was reliably is composed of a nonpathway interruption event (N) inferred from waveforms (see Correlation of Wave- followed by a short ingestion (C) event. Here, we forms with Penetrated Cell Types). For the one third ultimately provide type descriptions of the two phases of sheaths with more than six branches, it was common and four waveform types or subtypes (C1, fB1w, B1s, for the branches to lie very close together, contigu- and C2) that are components of the sharpshooter X ously. However, even in such extreme cases, it was wave. To best explain the evidence for our hypothesis, 852 ANNALS OF THE ENTOMOLOGICAL SOCIETY OF AMERICA Vol. 102, no. 5

Fig. 2. Salivary sheaths left by stylet penetration (probing) of H. liturata on cowpea (waveforms not shown). (a) Large, multibranched sheath showing multiple red-then-blue branches off a central trunk. Blue saliva Þlled many cells through which the stylets penetrated. Multiple sections had to be examined because pieces of branches were located in several other sections, not shown. (b) A long sheath with one short, abandoned branch (*) and a second thinner and more tenuous terminal branch from which the stylets were pulled during a 1stC-terminated probe. The thinness of the terminal branch walls was typical of 1stC. Arrow indicates the protoxylem cell in which the terminal branch tip ended. (c) Triple-branched salivary sheath with large salivary ßange, from experiment 2 (waveforms not shown). Note the thicker branch walls. Waveform trace indicated partial stylet withdrawals and B2 trenches during pathway correlated with branches (numbered). Branch three waveforms showed X waves. Probe was terminated during 2ndN. 1. Pith Cell. 2. Pith cell. 3A. Immature, partially ligniÞed protoxylem cell, probably nonfunctional, correlated with abnormal X wave. 3B. LigniÞed protoxylem cell correlated with X wave and 2ndN. we present in the Þrst four sections of Results, a bot- H. liturata Salivary Sheaths tom-up approach to identifying the sharpshooter X Forty-Þve salivary sheaths from experiment 1 and 18 wave. First, the salivary sheaths that were examined sheaths from experiment 2 were histologically exam- from experiment 1 are described. Second, the new ined and correlated with waveforms. These H. liturata waveform types and subtypes discovered during our sheaths seemed nearly identical to those previously search for the X wave are characterized. Third, these observed from glassy-winged sharpshooter (Backus et waveforms are correlated with both xylem and non- al. 2005), only being smaller (because of smaller body xylem cell types penetrated by the stylets, for both size of smoke tree versus glassy-winged sharpshoot- terminal and abandoned salivary sheath branches ers) and having more branches. Because of this sim- from experiments 1 and 2. Fourth, we describe the ilarity, we provide here only a few images of smoke logical steps undertaken to formulate and test the X tree sharpshooter salivary sheaths. Further represen- wave hypothesis. For the Þrst three sections, the C and tative examples of two-, four-, and six-branched N events and their orders are referred to by their sheaths can be seen in Holmes (2007). Only sheaths/ standard sharpshooter terminology (e.g., 1stC, 2ndN, waveforms from experiment 1 are quantiÞed in the and so on). After introduction of the X wave concept following sections, unless otherwise stated. and terminology, those names are changed to reßect Stylets penetrated intracellularly, and the sheaths newly introduced terminology. Thus, description of left behind were distinctly separable into a main trunk results here starts at the Þnest level of analysis to and sheath branches. The trunk typically had a thick describe the logical building blocks of our hypothesis. wall composed of blobs of deeply red-staining sheath Nonetheless, it is worthwhile to keep in mind where saliva with a hollow lumen (Fig. 2a). A variable num- the analysis is ultimately headed, i.e., to a deeper ber of sheath branches protruded from the trunk; 77% understanding of sharpshooter feeding and pathogen of the sectioned sheaths had four or fewer branches, inoculation via the X wave. although as many as 15 branches could be identiÞed September 2009 BACKUS ET AL.: SHARPSHOOTER XWAVE 853

Table 1. Total number of branches (both abandoned and terminal) of salivary sheaths for 44 probes correlated with branches (see text)

No. branches No. No. branches whose tips ended in in sheath sheaths Pith Parenchyma Phloem Cambium Xylem 1 branch 10 0/10 4/10 0/10 0/10 6/10 2 branches 8 3/16 2/16 0/16 0/16 11/16 3 branches 10 7/30 4/30 1/30 1/30 17/30 4 branches 6 6/24 2/24 0/24 2/24 14/24 5 branches 2 1/10 2/10 0/10 3/10 4/10 6 branches 1 0/6 1/6 0/6 3/6 2/6 7 branches 4 9/28 3/28 0/28 0/28 16/28 9 branches 1 0/9 3/9 0/9 1/9 5/9 10 branches 1 4/10 0/10 1/10 1/10 4/10 15 branches 1 0/15 0/15 0/15 2/15 13/15 Total 44 30/158 21/158 2/158 13/158 92/158

Parenchyma includes cortical, fascicular and interfascicular parenchyma; xylem includes primary (protoxylem and metaxylem) and secondary xylem and does not imply degree of cell wall ligniÞcation or functionality. from a single sheath (Table 1). Sheath branches peaks, durations of C1 waveform events were divided stained a mixture of reddish pink and blue (Fig. 2a and into sporadic (1Ð2 peaks; Fig. 3a), short (3Ð14 peaks, c). A complete sheath was highly three-dimensional Fig. 3b), and extended (Ͼ14 peaks, Fig. 3c) durations. and varied in size from 60 to 250 ␮m in spread, re- C1 occurred during all phases of probing. quiring viewing of 6Ð25 tissue sections per probe (in C2. This waveform type is the new term for the total, 744 sections for experiment 1) to identify exact typical C waveform studied to date (see Discussion). branch termini. Figure 2a shows the divergent nature The waveform is regular and rhythmic (Fig. 4) and of some branches, especially at their bases near the occurs for highly variable durations per event, from trunk (red saliva) and the degree to which blue saliva seconds to many hours. Individual episodes (termed often Þlled the cell at the tips of abandoned, nonter- plateaus) usually have a rectangular or rounded-rect- minal branches. The usually blue terminal branch was angular appearance (Fig. 4a and b). (Rarely, as in Fig. hollow and often very thin-walled (especially for 1, they can be partially peaked due to problems with 1stC-terminated probes), as in Fig. 2b, which passed wire conductivity. In those cases, they are average- through broken small, secondary ligniÞed xylem cells sized and continue for many seconds, like C2 but Þnally terminating in protoxylem (arrow). unlike C1.) The average duration of a single plateau is Branch-tip cell types for all 158 branches were iden- Ϸ1 s, with a range from 0.5 s to 1.5 s. Size and shape tiÞed (Table 1) for 44 of 45 sheaths from experiment of C2 plateaus show moderate differences (Fig. 4), 1; the 45th sheath had too many contiguous, blended and the distance between successive plateaus is vari- branches to analyze their pattern. For experiment 2 able, in some cases interspersed with variably long sheaths, only the cell types for terminal branches (not periods of a ßat-line waveform (termed R). Variation abandoned) were identiÞed. Forty percent of the 158 in appearance, often consisting of peaklets at the lead- (experiment 1) branches ended in nonvascular cells ing or trailing edges of plateaus (Fig. 4d and e), is seen (i.e., pith, cortical or fascicular parenchyma, or cam- both within and among individual probes and among bium); 58.2% ended in some type of xylem, and only different insects. Also, a gradual transition in appear- 1.3% (two branches, both nonterminal) ended in ance of C2 occurs both within and among individual phloem cells (Table 1). Of the 92 branch tips in xylem, waveform events, during which the plateau amplitude 44% were terminal branches. The remainder were gradually increases and plateau repetition rate grad- abandoned, nonterminal branches to (usually) xylem ually decreases, until these properties stabilize. This cells that the insect had rejected en route to the “stable C2” is characteristic of sustained ingestion (see terminal cell (Table 1). X wave Characterization) and then continues for many minutes to hours. Only ingestion-phase C events containing C2 were counted for C event order, even Characterization of New Waveform Types though these sometimes consisted of a mixture of C1 The following new sharpshooter waveform types and C2. (C1, C2, and G) were characterized during this study C2 plateaus were occasionally seen (in only three and correlated (below) with terminal sheath cell probes, one each by three insects) during pathway types (see Discussion for similarity with waveforms phase. Such pathway C2 plateaus were often shorter from other studies). in both amplitude (25Ð75% of ingestion phase C2 C1. This waveform type has individual episodes height) and duration (0.2Ð0.5 s). Typically, a bout of (termed peaks) that are triangular or peaked-rectan- four to eight or more pathway C2 plateaus was pre- gle-like in appearance (Fig. 3). The amplitude of each ceded and followed by B1. peak is variable, from 50 to 300% the amplitude of C2 G. This waveform is frequently observed at the end (see below) from the same probe (Fig. 2). Peak du- of long-duration probes, often interspersed with R rations range from 0.5 to 1 s. Based on the number of (ßat-line waveform). Waveform G can follow imme- 854 ANNALS OF THE ENTOMOLOGICAL SOCIETY OF AMERICA Vol. 102, no. 5

* * * * * * *

a b c

First part of extended C1

d

Fig. 3. Example waveforms for C1. (a) Sporadic C1 (1Ð2 peaks), above 0 V. Windaq gain 4ϫ. (b and c) Short C1 (3Ð14 peaks), above 0 V. Windaq gain 32ϫ. (d) First few peaks of extended C1 (Ͼ15 peaks), above 0 V. Windaq gain 32x. Dashed lines represent 0 V. level in a and c. Windaq compression 3; 0.6 s/div in all cases. *, C1 peaks; arrows, B1s spikelet bursts. diately after pathway (Fig. 5), or after long durations nonpathway interruptions (N) between them (Fig. of C2 (other studies). G resembles a blend of a rect- 6bÐd). When voltage levels varied between pathway angular C2 plateau and tall B1s spikelet burst. Some- and ingestion phases during a probe, such a bout was times, it is nearly identical to B1s during late pathway preceded by a rise in voltage level to above 0 V that but occurs many seconds to hours later in the probe, occurred at the end of pathway, before 1stC (Fig. 6a during ingestion phase. and e). Seventy eight percent (29/37) of the sheaths correlated with these probes had multiple branches into many cell types (see further Þndings below). Correlation of Waveforms with Cell Types However, eight other salivary sheaths had single Penetrated by Sheath Branches branches, each to a terminal xylem cell. In these cases, Findings Supporting Correlations. As visual exam- the bouts of C events were terminated during 1stC ination and correlation of sheaths with recorded (three probes), 2ndC (one probe), 3rdC (three waveforms progressed, we made two observations that probes), and 4thC (three probes). One of these later proved important for discerning the X wave. probes, whose pathway lacked B2 trenches, is shown First, most probes correlated with terminal branches in Fig. 6a. Thus, one bout of repeated C-N events, all in xylem (see further Þndings below) had a single bout occurring at a steady voltage level, was correlated with of several, contiguous C events that occurred at the one sheath branch to a xylem cell, regardless of the same voltage level (Fig. 6a and e) with only short, number of C or N events in the bout.

a b

c d e Fig. 4. Example waveforms for C2. (a) Typical, rounded rectangle plateaus. Windaq gain 8ϫ. (b) Uncommon, thinner rectangles with unknown waveform D-like waves between them. Windaq gain 32ϫ. (cÐe) Less common rectangular plateaus with peaklets on edges. Windaq gains 32, 32, and 4ϫ, respectively. All waveforms are above 0 V (dashed line in e). Windaq compression 3; 0.6 s/div in all cases. September 2009 BACKUS ET AL.: SHARPSHOOTER XWAVE 855

Fig. 5. Example waveform for G. Windaq gain 8ϫ. Waveform was below 0 V. Windaq compression 3; 0.6 s/div.

Second, the number of major sheath branches the sheath left by that probe had four branches: two above one was generally correlated with the number major branches (an abandoned branch into fascicular of B2 trenches during pathway. For example, the re- cambium as well as the hollow terminal branch into a cording in Fig. 6e had three B2 trenches (arrows), and large, secondary ligniÞed xylem cell), plus two more,

A1 B1 b c d *

a 1stXN * * 2ndXN* 3rdXN

1stXC nd 3rd XC b cd2 XC pathway phase ingestion & interruption phases

f g

e

C1 PC1fB fB B1C1 C2 1w 1w

*

f g Fig. 6. Traces from two probes demonstrating negative-to-positive voltage level shift and Þne structure of X waves. (a) Trace that begins positive (A1 peaks), then becomes mostly negative-going until the end of pathway and C events. Peaks above 0 V (dashed line) during pathway are sometimes C1 (*) but also B1s. Windaq compression 30; 6 s/div. Windaq gain 2ϫ. (b) Expansion of box b in a, showing end of pathway. (c) Expansion of box c in a, showing 1stN (2ndXN) (d) Expansion of box d in a, showing 2ndN (3rdXN) Windaq gains all 8ϫ for bÐd. (e) Trace in which pathway is entirely negative-going until the end of pathway. (f) Expansion of box f in part d, showing partial X wave (pathway XN portion only). Windaq gain 16ϫ. (g) Expansion of box g in d, showing the very tall C1 peaks in 1stXN (at the end of pathway), i.e., the Þrst part of the Þrst X wave. Windaq gain 8ϫ. Windaq compression 50; 10 s/div. All views Windaq compression 3; 0.6 s/div. Arrows, B1s spikelet bursts; *, C1 peaks; slanted arrows, B2 trenches; P, pathway. 856 ANNALS OF THE ENTOMOLOGICAL SOCIETY OF AMERICA Vol. 102, no. 5 very small abandoned branches occurring as off- Table 2. Numbers and mean durations (؎ SEM) of all naturally shoots of the terminal branch, one into a protoxylem terminated (i.e., nonterminal) waveform events cell and another into a small, secondary ligniÞed xylem Mean duration Type of event n cell. Eight probes (each correlated with multi- (s) Ϯ SEM branched sheaths with terminal branches in xylem) Pathway 43 296.8 Ϯ 41.4a* had two or three such bouts of C-N events, separated Pathway-like interruption 12 112.1 Ϯ 63.6b* by a pathway-like interruption containing at least one C2 in pathway 9 6.2 Ϯ 1.7c B2 trench (during which, the voltage level dropped C1 in pathway 93 5.7 Ϯ 1.3c below 0 V.). Five of these probes had two bouts each; C1 in pathway-like interruption 18 6.3 Ϯ 2.1c three probes had three bouts each. The sheaths of all 1stC 30 60.2 Ϯ 30.1d Þve probes had at least two branches (never fewer 2ndC 21 114.3 Ϯ 52.8de* [but sometimes more] than the number of B2 3rdC 15 221.2 Ϯ 116.6e* trenches). Many branches ended in xylem cells. This 4thC 7 45.0 Ϯ 21.0de 5thC 5 36.4 Ϯ 20.6de support of the original Þnding with H. vitripennis 6thC 1 45 Ϯ 0n/aa (Backus et al. 2005) (that numbers of branches is roughly correlated with numbers of B2 trenches) con- Pathway and pathway-like interruption durations exclude C1 and Þrmed with H. liturata that it was generally possible to C2 durations. SigniÞcant differences (ANOVA results: F ϭ 8.67, df ϭ predict the branching pattern of the salivary sheath by 10, P Ͻ 0.0001; see text for LSD results) among means within group- ings are denoted by different lowercase letters. Asterisk (*) denotes the appearance of the EPG waveform trace. Thus, that those event durations also were signiÞcantly different within the single- or multiple C-N bouts were correlated with row, therefore among probes. ␣ ϭ 0.05. single- or multiple (respectively) branches into xylem, a n/a, not applicable. some abandoned, some terminal. The above-mentioned two observations sup- Therefore, many, very short C1 and/or C2 events were ported cross-correlation of three sets of informa- embedded within the majority of less frequent, yet tion: 1) temporal waveform progression, especially much longer, pathway-like phases. C and N event order; 2) spatial sheath branching Forty-four sheath-correlated probes produced 114 patterns of abandoned versus terminal branches; interpretable, abandoned sheath branches, in addition and 3) cell type penetrated by each branch. For to their 44 hollow, terminal branches. These data were hollow, terminal branch tips, these correlations tabulated to compare numbers of pathway-like phases were deÞnitive. By a process of elimination, indirect with or without C1 and/or C2 versus xylem or non- correlations for abandoned, nonterminal branches xylem cells; frequencies were signiÞcantly different at were thus strengthened. P Ͻ 0.0001 (Table 4). Nearly half (48%; 55/114) of the Using these correlation tools, we tested two hy- abandoned branches were in some type of xylem, with potheses critical to X wave correlation. First, we hy- 82% (45/55) of these (including all three probes with pothesized that C1 and/or C2 during pathway might be correlated with abandoned branches into xylem. Late pathway waveforms were dominated by B1 and Table 3. Mean durations (؎ SEM) per probe for all, summed B2, with occasional C1 and very rare C2. We suspected events of pathway or pathway-like interruptions, for probes with or without C1 and/or C2 during those events; also, same for salivary pathway C1 and C2 because of their relative rarity and sheaths with one to five branches or six to 15 branches frequent association of C1 with C2 in early, xylem- correlated C events. Second, we hypothesized that C Waveform or Mean duration (s) Ϯ SEM event orders correlated with terminal branches into no. sheath n With C1 Without C1 n Total n xylem could be distinguished from those into nonxy- branches or C2 or C2 lem cells by some feature(s) of the C-N waveforms Pathwaya 34 346.0 Ϯ 57.0a* 11 176.5 Ϯ 47.6b 45 that would qualify them as X waves. Results supported Pathway-like 3 502.3 Ϯ 210.0a* 9 36.1 Ϯ 8.5b 12 both hypotheses. interruptiona C1 and C2 during Pathway-Like Phases Correlated 1Ð5 branchesb 22 256.1 Ϯ 29.0c 14 215.9 Ϯ 37.8c 36 With Abandoned Branches (Experiment 1). Pathway 6Ð15 branchesb 8 787.4 Ϯ 171.6d* Ñc Ñcd 8 events were signiÞcantly longer (P ϭ 0.0027) and more numerous, in general, than were events of path- SigniÞcant differences among the four means (within pathway or way-like interruption (Table 2). Also, mean event sheaths) are denoted by different lower-case letters. Asterisk (*) durations for C1 and C2 during pathway or pathway- denotes that those event durations were signiÞcantly different within the row, among probes. ␣ ϭ 0.05. like interruptions were signiÞcantly shorter than those a Pathway events were in the same probes as pathway-like inter- of both and pathway and pathway-like interruptions ruption events. Waveforms were measured in 45 probes. Durations (P Ͻ 0.0001 in all cases), and not signiÞcantly different were signiÞcantly different overall (ANOVA (F ϭ 10.03, df ϭ 53, P ϭ from one another (Table 2). For ease of description, 0.0026). See text for LSD and pooled frequency results. b A probe either left one to Þve sheath branches or six to 15 waveform events during either pathway phase or path- branches, never both. There were 44 probes whose branch numbers way-like interruption phase will be considered to- could be counted. Durations were signiÞcantly different overall gether when waveform correlations apply to both. (ANOVA (F ϭ 8.00, df ϭ 41, P ϭ 0.0072). See Results for LSD and They will be termed pathway-like phases. C1 and/or pooled frequency results. c For ANOVA analysis of durations only, a single probe of one C2 were performed during pathway-like phases in 71% second was artiÞcially added to this category, to give some value for of the 45 total probes (total n in row 1 in Table 3). this cell in the comparison. September 2009 BACKUS ET AL.: SHARPSHOOTER XWAVE 857

Table 4. Total number of waveform phases or events, of the types listed, correlated with cell types of nonterminal, abandoned sheath branches

Pathway-like C1 during C1 Ϫ N Ϫ Pathway C1 during C2 during C1 ϩ C2 Branch tip in this cell type interruption pathway-like C1 Ϫ C2 Total onlya pathway pathway (1stC) onlya interruption (1stC and 2ndC) Protoxylem 11 1 3 Metaxylem 11 Secondary xylem, small, nonligniÞed 1 1 Secondary xylem, small, ligniÞed 2 23 3 11 3 3 45 Secondary xylem, large, ligniÞed 23 5 Pith 26 2 1 29 Cortical parenchyma 8 7 15 Phloem 1 1 2 Fascicular cambium 4 6 1 2 13 Total 42 41 3 5 11 7 5 114

Bold numbers indicate some type of xylem cell. a Presence of only B1 or B2 during pathway or pathway N, not C1 or C2.

C2 during pathway) being in small, secondary, ligni- It may be that only extended C1 occurs in nonxylem, Þed xylem cells (Table 4). Eighty-nine percent (102/ and shorter C1 events are in xylem. However, we 114) of abandoned branches were correlated with could not determine this deÞnitively. pathway-like phases; the rest were correlated with Mean event durations of pathway-like phases were brief, isolated C1 or C2 events that were followed by signiÞcantly longer when they contained C1 or C2 pathway-like interruptions. Most importantly, 92% than when they lacked C1 or C2 (Table 3) (P ϭ 0.0091 (43/47) of probes (branches?) with pathway-like for pathway; P ϭ 0.0001 for pathway-like interrup- phases lacking C1 or C2 were correlated with aban- tions). Also, the signiÞcantly fewer (pooled frequency doned branches in nonxylem cells (pathway only, P ϭ 0.0018) pathway-like interruptions that contained Table 4). In contrast, when pathway-like phases con- C1 or C2 were (numerically, but not signiÞcantly) tained C1 and/or C2, their correlated, abandoned longer than the very common pathway events with C1 branches were 71% likely (39/55) to have occurred in or C2. The opposite was the case for both types of rejected xylem cells (Table 4). This supports that C1 pathway-like events lacking C1 or C2; the more com- and C2 during pathway-like phases often coincide mon pathway events were signiÞcantly (P Ͻ 0.0001) with xylem cell penetration and rejection. Yet, when longer than the fewer pathway-like interruption two probes from different insects were artiÞcially ter- events lacking C1 or C2 (Table 3). Thus, the longer the minated during extended C1 events in pathway (Fig. pathway-like interruption (i.e., after isolated C 7a and b), both terminal branch tips were in cortical events), the greater the likelihood it will have C1 or parenchyma cells. Thus, this correlation is not perfect. brief C2.

pathway phase

b

a

B1 C1 2

1

1 C2 b c 2 Fig. 7. (a) Trace from probe that was artiÞcially terminated during extended C1 during pathway (see narrative). Entire probe is below 0 V. Windaq gain 32ϫ. Dashed line represents 0 V. (b) Expansion of box b in a, showing end of pathway with extended C1. Windaq gain 64ϫ. (c) Trace from probe that lacked an X wave; signal jumped directly from pathway into 1stC. Inset box 1, expanded view of box 1, showing very smooth B1w. Inset box 2, expanded view of box 2, showing slightly more rippled B1w but still not as rhythmic as fB1w. Windaq gain 64ϫ; inset boxes 128ϫ. Windaq compression 3; 0.6 s/div in all parts. 858 ANNALS OF THE ENTOMOLOGICAL SOCIETY OF AMERICA Vol. 102, no. 5

Table 5. Total numbers of salivary sheaths whose terminal branch tips were in each cell type listed, after probes were artificially terminated during the cited C waveform event order (i.e., C events containing C2)

Waveform event order Terminal branch in this cell type 1stC 2ndC 3rdC 4thC 5thC 6thC 7thC Protoxylem 1 Metaxylem 2 (1) Secondary xylem, small, nonligniÞed 1 (1) Secondary xylem, small, ligniÞed 4 (4) 5 (1) 1 4 (1) 1 1 Secondary xylem, large, ligniÞed 2 (1) 4 (1) 2 4 1 4 Pith 1 (no X wave) Pith or Secondary xylem, small, ligniÞed (?) 1 (1) (X wave) Fascicular cambium (?) 1 (1) (X wave)

Numbers in parentheses indicate number of C events with mixed C1 and C2. Bold numbers indicate some type of xylem cell.

Durations differed signiÞcantly (P ϭ 0.0330) for the Almeida and Backus 2004) indicate acceptability of pooled number of correlated sheath branches (Table the penetrated cell. Thus, in most probes, insects lo- 3), because the number of sheath branches was cated and accepted an ingestion cell by 2ndC or 3rdC. strongly related to duration of pathway-like phases. Those that could not, returned to pathway activities Pathway-like phases without C1 or C2 always left during a subsequent interruption. sheaths with few (1Ð5) branches. Sheaths with fewer Forty C events of Þrst to seventh order, all com- branches were equally likely to have or lack C1 or C2. posed of either 1) C1 and C2 in the same event or 2) In contrast, larger, more branching sheaths were al- C2 alone were correlated with terminal sheath branch ways correlated with C1 or C2 (Table 3). tips (Table 5). Of those 40 events, 37 branches pen- In summary, our Þndings on C1 and C2 during etrated into some type of primary xylem (i.e., pro- pathway-like phases support that when C1 was toxylem or metaxylem; 8%) or secondary xylem (92%) present, it usually (but not always) indicated pene- cell. Eighty-nine percent (33/37) of those deÞnitively tration of xylem. The shorter the duration of a C1 found in xylem were in ligniÞed secondary xylem cells; event during pathway-like phases, the more likely it half were in large and half in small cells (Table 5). represented xylem penetration; longer durations may Only once did penetration and ingestion deÞnitively occur in nonxylem cells. Isolated C2 during pathway- occur in a nonxylem (pith) cell. Two other nonxylem like phases may more strongly indicate xylem pene- termini are questionable (see further results under tration; however, a sample size of three is not deÞn- Sharpshooter X Waves). itive. Presence of C1 and/or C2 during pathway-like Sharpshooters performed C2 ingestion much more phases was strongly correlated with longer phase du- commonly (82% of combined abandoned and terminal rations (especially of pathway-like interruptions) and branches correlated with C2) from ligniÞed (thus, multiple, abandoned branches, many in xylem. Thus, probably mature) secondary xylem than any other C1 and/or C2 waveforms become more common after plant cell type. Small secondary xylem cells, even an insect has rejected cells penetrated during the Þrst ligniÞed ones, were rejected 74% of the time they were one or two sheath branches, because the insect renews sampled (45/61, i.e., abandoned branches as percent- its search for an acceptable xylem cell. However, it was not possible to precisely correlate an individual path- age of total in small cells); in contrast, large, ligniÞed way C1 event with an individual branch into xylem. secondary xylem cells were accepted 74% of the time C Event Orders Correlated With Both Abandoned (17/23) once encountered (Tables 4 and 5). and Terminal Branches (Experiment 1). Seventy- As mentioned above, mixed C1 and C2 were per- nine naturally terminated C events were performed in formed in all cell types sampled by the insect, corre- the 45 correlated probes, before the terminal C event lated with both abandoned sheath branches (Table 4) in each (n in bottom half of Table 2). Mean durations and terminal branches (Table 5). Twelve isolated 1stC of naturally terminated C events showed that most C (or 1stC and 2ndC) events (all containing a mixture of event orders were short (Յ1 min in duration), except C1 and C2) were correlated with abandoned sheath for 2ndC and 3rdC (Table 2). Mean duration of 3rdC branches; ten in the xylem (Table 4, columns 7 and 8). was numerically longer than all other C events, but Nine of 10 of these branches were in ligniÞed second- only signiÞcantly longer than 1stC (P ϭ 0.0502), due ary xylem cells and six of the nine in small cells. to low sample size of C event orders greater than third. However, the 10 abandoned branches represented Duration of 2ndC was intermediate, not signiÞcantly only 21% (10/47) of the total number of sheath different from any others. Numbers of events in each branches correlated with xylem (terminal plus aban- C event order were highest in the early C events, and doned). The remaining 79% of xylem branch tips (all lower with each successive C event; 84% (66/79) of terminal) had only C2 in their correlated C events the C events were 1stC to 3rdC (Table 2). Interrup- (Table 5). This is demonstrated in Table 6, showing tions, which mark the end of C events, became less that C1 nearly disappeared after 3rdC, with one rare common after 2ndC, i.e., the C events became longer. exception of a single probe in 4thC. Thus, C2-con- Long ingestion events (termed sustained ingestion, taining ingestion events were most often performed in September 2009 BACKUS ET AL.: SHARPSHOOTER XWAVE 859

Table 6. Numbers of various (ingestion phase) waveforms for general, the signal was positive during the earliest each C event order pathway activities (especially waveform A1, forming the sheath trunk, as in Fig. 6a), although sometimes No. C and G events for each event order, even A1 would be negative (Fig. 6e). Then the signal C event order including terminal and nonterminal C1 ϩ C2 C2 only G would either remain positive throughout the probe (43% of the 40 artiÞcially terminated probes, as in Figs. 1stC 25 16 2 3b and d and 4aÐd), or trend negatively as pathway 2ndC 7 24 3rdC 4 17 progressed into B1 (45%, as in Fig. 6a and e). Eighty- 4thC 1 14 two percent of negative-trending signals would rise 5thC 6 1 into positive during C1 or C2 in pathway (Fig. 3a and 6thC 5 c) or at the end of pathway just before 1stC (Fig. 6a, 7thC 1 Total 37 83 3 b, and e). C events were then positive, whereas N events were negative (Fig. 6aÐd). Ten percent of the 40 probes remained negative throughout the probe, even during C (data not shown). One probe lost all acceptable xylem cells, in which eventually only C2, voltage level information and waveforms were stable not C1, was performed. around 0 V (Fig. 2). This occurs occasionally when an G Waveform. Two probes were terminated during insect is poorly wired. 1stG (i.e., the Þrst postpathway event was waveform G) (Table 6); both branch tips were in cortical pa- renchyma cells. One other G event was terminated in Sharpshooter X Waves 5thG (i.e., the Þrst four events contained C2 and only Developing and Testing the Hypothesis. During our the Þfth was G). This branch terminated in a large, waveform-salivary sheath correlations, we made two ligniÞed secondary xylem cell. Thus, it is likely that G more observations important for the X wave hypoth- can represent activities in either nonxylem or xylem esis. First, we observed that B1 waveforms during cells, depending on whether it occurs during pathway nonpathway interruption (N) events associated with or after C2 ingestion. However, deÞnitive determina- xylem ingestion had different appearances from B1 tion awaits a larger sample size. during pathway. Second, a pattern of waveform sub- N Event Orders Correlated With Terminal types similar in appearance to a nonpathway inter- Branches (Experiment 2). All 18 probes that were ruption was nearly always visible at the end of path- artiÞcially terminated during 1stN (i.e., the Þrst N way phase, just before the 1stC event of ingestion after 1stC), 2ndN, 3rdN, or 6thN events resulted in phase. This pattern was often distinctly different from hollow, terminal sheath branches. In all 18, branch tips the preceding pathway waveforms. When such an were in some type of xylem (Table 7), although due end-of-pathway “interruption” (“N”) (although not to the test plantsÕ younger age (compared with ex- technically called that until now) was visually com- periment 1), diverse xylem cell types were more bined with the subsequent C and N events, an alter- evenly represented among the branch termini. This nating sequence of “N”ÐCÐNÐC events became clear. strengthens the earlier conclusion that N events be- When each NÐC sequence was considered a unit, they tween C events occur in xylem, and likely in the same were seen to repeat several times, sequentially (Fig. xylem cell as the preceding C event. 1). We therefore hypothesized that we had discerned a new type of X wave, for xylem ingestion by sharp- Voltage Levels of EPG Waveforms at Input shooters, which would be composed of bouts of NÐC Impedance 108 Ohms units (or “phrases,” as in Wayadande and Nault [1993]; hereafter, each part of an X wave phrase is termed XN The overall voltage levels of waveforms in this study and XC, respectively). We then tested to determine were a mixture of positive and negative signals. In whether this putative X wave could be used to predict xylem versus nonxylem ingestion. Table 7. Numbers of terminal branch tips (i.e., hollow or Accordingly, we studied the waveforms in 120 end- single) in each cell type listed, after probes were artificially termi- of-pathway phases and nonpathway interruptions nated during the cited order of nonpathway interruption (N) events from all 40 C-terminated probes (the other Þve were (experiment 2) not terminated during C events) and categorized their

a waveform types and appearances. In particular, the 40 Branch tip in this cell type 1stN 2ndN 3rdN 6thN sets of waveforms at the end of each probeÕs pathway Protoxylem, ligniÞed 311phase (i.e., putative 1stXN) were studied for presence Protoxylem, nonligniÞed 22 of common, consistent features. Distinctive X waves, Metaxylem 2 Secondary xylem, ligniÞed 221 matching the description below, occurred in all but Secondary xylem, nonligniÞed 2 one unusual case (Table 5; Fig. 7c), in which the probeÕs hollow sheath branch deÞnitely terminated in Bold numbers indicate some type of xylem cell. a a pith cell. Its end-of-pathway waveforms completely Called 1stN, 2ndN, 3rdN and 6thN in this table, according to the lacked any of the necessary components of an X wave naming convention of Backus et al. (2005), used in the Þrst part of this paper. Hereafter, these events are termed 2ndXN, 3rdXN, 4thXN and (described below), merely transitioning straight from 7thXN, respectively (see terminology for X waves). pathway into C1 then C2, after a abrupt rise in voltage 860 ANNALS OF THE ENTOMOLOGICAL SOCIETY OF AMERICA Vol. 102, no. 5

C1 B1 C1 fB1w C1 fB1w C 2

* *

a

B C1 C2 1

* * * * *

b c Fig. 8. Examples of the variability in X wave duration and composition; aÐc are above 0 V. (a) Long X wave. fB1w is prevalent by itself. B1 denotes short pathway event before start of 1stXN (at C1 second from left). Windaq gain 8ϫ. (b) Medium, typical X wave. fB1w is seen between all C1 peaks and C2 plateaus. Inset box shows expanded fB1w. Windaq gain 8ϫ; inset 16ϫ. (c) Short X wave. Windaq gain 8ϫ. Windaq compression 3; 0.6 s/div in all parts. Downward-pointing arrows, B1s spikelet bursts; *, C1 peaks; upward-pointing arrowheads, fB1w. level (Fig. 7c). The 37 branch tips that were deÞnitely correlated with abandoned branches in xylem 70% of in xylem cells (Table 5) also were correlated with the time. The later in the probe this occurred (espe- distinct X waves, including 1stXN events at the end of cially during pathway-like interruptions), the more pathway. Two other probes with distinct X waves, likely that C1 during pathway XN represented xylem terminated during 1stC and 2ndC, had questionable penetration. If an XN event was paired with an XC branch termini (Table 5). The 1stC sheath branch containing C2, together they were considered true X penetrated at least to the fascicular cambium. How- waves, because they were 100% correlated with xylem. ever, broken xylem cells nearby resembled the result Therefore, in the case of sharpshooters, pathway XN of stylet penetration rather than histological process- is not considered a true X wave, but indicates that ing damage. The 2ndC sheath branch terminated in stylets sampled and rejected cells, usually but not either of two closely adjoining cells: a pith cell and a always xylem. In true X waves, the insects sampled and small, secondary ligniÞed xylem cell. We concluded accepted cells, always xylem. that both of these questionable salivary sheaths actu- Characterization (Type Description) of X Wave ally terminated in xylem, but this could not be deÞn- Components. The X wave comprises unique variants itively determined by histology due to processing ar- of the following four waveform components (B1w, tifacts. Therefore, at least 95% (but likely 100%) of all B1s, C1, and C2) (Fig. 6). At least two of the Þrst three terminal sheath branches in xylem were correlated components must be present during 1stXN for X wave with X waves; in the rare case when a terminal branch status to be assigned. The composition and durations was not in xylem, there was no X wave. All C events of XN events were highly diverse, exhibiting variants of orders Ͼ1stC, and their associated N events, were whose durations were long (rarely occurring; Fig. 8a), also parts of X waves. These Þndings support our medium (most common; Fig. 8b), or short (very rare; hypothesis that a sharpshooter X wave exists and that Fig. 8c), due to different waveform components it provides a deÞnitive landmark for C2 ingestion from and their durations. XC events are much less variable a xylem cell. in composition; they contain at least C2, but often Furthermore, partial X waves, consisting of an iso- also C1. lated XN (hereafter, pathway XN), were sometimes C1. One to several triangular peaks often (but not seen during pathway-like phases (Fig. 6f). These were always) occur just before the end of pathway phase, highly recognizable due to the presence of XN com- usually marking the beginning of 1stXN, therefore of ponent waveforms, including C1, described below. the Þrst X wave phrase (Fig. 8b). One to several C1 Close examination of all pathway XN events could not peaks also are found at the beginning of some XC distinguish by waveform structure alone those corre- events (Figs. 8aÐc and 6f and g). Amplitudes of these lated with xylem versus nonxylem cells. However, C1 peaks are often much higher than adjoining C2 when pathway XN events included C1, they were plateaus (Fig. 8f and g). The C1 and C2 peaks in XC September 2009 BACKUS ET AL.: SHARPSHOOTER XWAVE 861 events are separated by fuzzy B1w (see below) (Fig. the signals (E.A.B., unpublished data). Backus et al. 8b and c), not smooth B1w as occurs during pathway (2005) used 106 ohms input impedance, whereas re- (Fig. 7c). C1 is the most variable (in both duration and cordings for the present work primarily used 108 or 109 amplitude) of the X wave components; an X wave may ohms input impedance. It is outside the scope of this contain either extended C1 or lack C1 entirely. article to present a complete sharpshooter waveform Fuzzy (f) B1w. This component is the key charac- library at all input impedances. For now, we empha- teristic of the X wave, and varies so much from typical size that the same waveform types and subtypes as in B1w that we chose to name it separately. fB1w is found Backus et al. (2005) were recognizable and interpret- in all XN events, and nowhere else during stylet pen- able for this article. etration. Its presence during pathway is an indicator of C Waveform Types in Relation to Past Terms. pathway XN. The usually ßat or slightly humped wave- Almeida and Backus (2004) introduced the C wave- form in pathway between B1s bursts (Fig. 7c) be- form family and characterized C1 and C2 waveform comes highly rhythmic and regular (Fig. 8b, inset) types. C1 was a rhythmically regular waveform whose during the X wave. fB1w is a high-frequency (11.5Ð13 Þne structure was variably triangular, rectangular, or Hz.) burst that is distinguished from B1s by its very steeply peaked. C2 was similar to C1, but superim- short amplitude and slightly lower frequency (Fig. posed on a “variable-amplitude, low-frequency coarse 8b). Duration of fB1w is highly variable. structure.” Although it was understood at the time that B1s. The X wave variant of this waveform is similar C1 was still quite broad, there was not enough infor- to spikelet bursts during pathway, but often higher in mation to divide it further. Here, we are redeÞning amplitude and much shorter in duration, with longer both C1 and C2, because subsequent work has shown fB1w events interspersed (Fig. 8a and b). B1s often that the low-frequency coarse structure of C2 in disappears from later XN events, leaving only fB1w Almeida and Backus (2004) was actually baseline (Fig. 6d). noise from which it was difÞcult to resolve C wave- C2. Short (as opposed to very long) events of these form Þne structure at the low input impedance used rectangular plateaus are unique to the X wave. Also, in that study. At the higher input impedances of this like C1, C2 plateaus in X waves often have fB1w study, resolution of C was excellent. Therefore, we between them, unlike during sustained ingestion (Fig. have separated the triangular and steeply peaked C 6g). When C2 is interrupted by any of the other three waves as C1 and the rectangular or rounded-rectan- waveforms (usually C1 or B1w), the end of the C2 gular waves as C2. event marks the end of XC, i.e., of that particular X Biological Meaning of C2. A recent EPG study with wave phrase (Fig. 2). Because bouts of X wave phrases smoke tree sharpshooters correlated C2 with electro- continue during a probe, the durations of each XC myography of cibarial dilator muscle contractions and portion become longer, until XN interruptions be- video observations of cibarial diaphragm movements come quite rare and ingestion continues on a long- (Dugravot et al. 2008). It showed that C2 represents term, sustained basis. This gradual transition is the true ingestion, i.e., cibarial pumping and swallowing of shift from trial to sustained ingestion. ßuid past the true mouth (at the anterior end of the esophagus) (Snodgrass 1935). They also demon- strated empirically for the Þrst time in EPG studies Discussion that streaming potentials, as hypothesized by Walker (2000), are the biopotentials (emf component) that Characterization and Biological Meanings of underlie the Þne structure of C2. In a C2 plateau, the Waveform Types initial voltage rise occurs during the onset of inward Past versus Present Published Waveform Appear- ßuid ßow, caused by the start of the diaphragmÕs mus- ances. There are several differences in the coarse and cular uplift. The ßat top of the plateau corresponds to Þne structure appearances of sharpshooter waveforms the steady electrical state of ßuid ßowing inward (an- in this article compared with the waveforms in earlier teriorward) as the cibarium is Þlling, until the dia- studies (Almeida and Backus 2004, Backus et al. 2005, phragm reaches its maximum uplift. The rapid voltage Joost et al. 2006). For example, in waveform images in drop at the end of the C2 plateau occurs as the dia- Backus et al. (2005): 1) overall voltage levels of probes phragm is released by its dilator muscles. Cuticular are strictly positive-going, compared with the positive- elasticity plus muscular release provide a strong force or negative-going or both, as here; 2) A1 and B2 are in the opposite direction. Due to a groove in the large in amplitude and detailed, compared with the bottom of the sharpshooter cibarium and the abrupt reverse, as here; 3) B1 and C are very small and curvature of the esophagus immediately after the difÞcult to resolve, compared with the reverse, as here; cibarium (Snodgrass 1935; Almeida et al. 2005; E.A.B., and 4) B2 trenches, representing partial stylet with- unpublished observations), ßuid ßows posteriorward, drawals, as indicated by decreasing voltage levels are and could be pushed either into the esophagus (i.e., very deep and obvious, compared with the partial swallowed), or back into the precibarium (the narrow ßattening of voltage levels, as here. channel that conveys ßuid from the stylet food canal Research in the Backus laboratory since Backus et to the cibarium; Backus and McLean, 1983). Ordi- al. (2005) has shown that these differences in wave- narily, the closure of the precibarial valve would pre- form appearance among studies are strictly due to vent ßuid backßow into the stylets (Dugravot et al. changes in input impedance of the monitors recording 2008, Backus and Dugravot 2009). Thus, cibarial Þlling 862 ANNALS OF THE ENTOMOLOGICAL SOCIETY OF AMERICA Vol. 102, no. 5 and swallowing, the critical acts of ingestion, only fect system voltage, such as electrode potential occur during C2. (Tjallingii 2000, Walker 2000). Alternatively, variation Biological Meaning of C1. The peaks of C1 are in voltage levels can also represent biological phe- essentially the same as a C2 plateau but without the ßat nomena, as has been well demonstrated in EPG studies top. Based on the directionality of streaming potentials of aphids and other insects that probe intercellularly. described above, the rapid rise and fall of voltage If so, then voltage level variation during a probe may during C1 is probably caused by a similarly rapid rise occur because of a combination of input impedance but immediate drop of the cibarial diaphragm, without and recording/wiring conditions that make EPG more Þlling the cibarium. Only a relatively small amount of or less sensitive to emf components, thought to be the ßuid would be taken up, probably insufÞcient for swal- electrical origin of charge separation in aphids. When lowing. Rapid release of the diaphragm would there- sharpshooters are recorded at input impedance 106 fore propel ßuid backward into the precibarium and ohms, all signals are positive (Backus et al. 2005). At stylets, and thence into the plant. The amplitude of a 109 ohms, a majority of signals are negative-then-pos- C1 peak would be proportional to the degree of uplift itive (Miranda et al. 2009). At the present 108 ohms, of the diaphragm. Thus, higher peaks probably rep- the best-wired insects often showed both negative resent more forceful expulsion of ßuid from the stylets. and positive voltage levels. We therefore propose the hypothesis that C1 is the In aphids, negative voltage level is interpreted as waveform that represents egestion behavior, long- intracellular stylet position, positive levels as extra- sought by vector entomologists. Studies are underway cellular (Walker 2000). The interior of a mature, apo- to deÞnitively test this hypothesis. plastic xylem cell is also extracellular. Although sharp- Biological Meaning of B1s. Another recent corre- shooter pathway activities are intracellular, the lation study using electromyography of the precibarial presumed tightness of the salivary sheath seal into a valve muscle shows that B1s represents muscular xylem cell (to prevent cavitation, see below) may movements (termed ßuttering) of the precibarial allow similar charge separation to be sustained during valve (Backus and Dugravot 2009). It was concluded ingestion, indicating extracellular stylet tip position in that precibarial valve ßuttering sweeps ßuids previ- xylem (Miranda et al. 2009). If so, then researchers ously brought into the precibarium back and forth could consistently use C1 occurring at a positive volt- across the two different sets of precibarial chemosen- age level to deÞnitively ascertain xylem penetration silla that are separated by the valve (Backus and during pathway, as is proposed in the study of Brazil- McLean 1982, 1983, 1985; Backus 1988). Thus, B1s ian sharpshooters by Miranda et al. (2009). It is there- corresponds to tasting of the internal ßuids of the fore provocative that C1 even during pathway XN is plant. This vital function explains why B1s (along with often positive in our study, supporting that xylem is B1w; see below) is ubiquitous throughout the pathway Þrst penetrated during partial XN, not as late as XC. phase of sharpshooter stylet penetration. However, quality of charge separation may vary from Biological Meaning of fB1w. Joost et al. (2006) cor- probe to probe. Our Þndings show that voltage level related B1w with secretion of saliva into artiÞcial diet. variation does not occur in every sharpshooter probe, Using input impedance 108 ohms, the current study at least using an input impedance of 108 ohms. Un- was clearly able to distinguish between pathway-type fortunately, it is not in the scope of this article to B1w (smoothly humped and waved) and the higher deÞnitively test whether C1 at positive versus negative frequency fB1w. Like B1w, fB1w is interspersed voltage levels is correlated with xylem penetration. among B1s spikelet bursts but in the unique location This hypothesis may be tested in future research. of X waves. Therefore, fB1w represents salivation plus Thus, although voltage level variation is generally sup- some presently unknown, invisible but rhythmically portive of our waveform interpretations, it may not be repeating behavior, occurring simultaneously. Re- as reliable as X waves in indicating xylem penetration search is underway to determine this additional phe- during an individual probe. X waves are visible at all nomenon, unique to X waves. input impedances (107,108, and 109 ohms), whereas Biological Meaning of G. Waveform G looks like a voltage level shifts are most visible at 109 ohms (E.A.B., blend of B1s and C1 waveforms. We hypothesize that unpublished data). its appearance represents its meaning, i.e., simulta- neous ßuttering of the precibarial valve and rapid Sharpshooter X Waves uplift and downward push of the cibarial diaphragm. If so, then it could bring ßuids into the precibarium for Biological Meaning. Assembling the known and hy- brief tasting then expulsion. G apparently occurs out- pothesized information about the meanings of the side of xylem when it directly follows pathway, but waveform components of the X wave (above), we perhaps occurs in xylem when it directly follows C2. propose the following scenario for what occurs during Regardless, it seems that the insect is resting with its the sharpshooter X wave. XN represents the earliest stylets in the plant (during R) but occasionally tastes stages of sampling and acceptance of a (usually) xylem and expels ßuid. cell. C1 often, but not always, begins XN. We hypoth- Biological Meaning of Voltage Level Differences esize that the insect takes up small amounts of ßuid during Probing. The variations in overall voltage lev- into the precibarium and then quickly egests them els for sharpshooter probes could be due to the quality out. This action would physically scrub out the of wiring and/or other nonbiological factors that af- precibarium of clogging microbes, bioÞlm, or plant September 2009 BACKUS ET AL.: SHARPSHOOTER XWAVE 863 precipitates that can block the proper functionality of We hypothesize that mechanical testing of the sal- the precibarial valve (Backus and McLean 1983, ivary seal (via cibarial pulling and swallowing of xylem Almeida and Purcell 2006) and chemosensilla, which ßuid) is the likely function of the XC portion of a true have been shown to be necessary for judging accept- X wave. During the Þrst one to four XC events, a xylem ability of xylem cells (Backus and McLean 1985). ßuid-saliva mix is again pulled into the precibarium It is known that during fB1w, sheath saliva is se- and even more forcefully egested (C1 at the start of creted and sculpted by the stylets (and perhaps also XC is usually the highest amplitude C1), probably the rush of xylem ßuid, as stylets pierce the xylem cell) further scrubbing the precibarium clean. This is prob- into a smooth seal that is conßuent between the out- ably necessary to fully free the precibarial valve and side and inside of the cell, forming a salivary sheath the basin around it, so that the valve can function as lining in the cell (Backus et al. 2005, and references a pressure-sensitive check valve to allow proper swal- therein). We propose that the strength and durability lowing (Backus and McLean 1983, Dugravot et al. of this stylet seal is critical for sharpshooters, as it is 2008). Several to a few dozen pulls of the cibarial probably this seal that prevents occurrence of cavita- diaphragm (C2 plateaus) are then performed, prob- tion (formation of an air embolism in xylem that acts ably to test the strength of the stylet seal and the as a hydraulic signal for plant water stress, causing success of pulling against the high xylem tension. If the stomatal closure; Salleo et al. 2000). A preliminary seal is strong, sustained ingestion can ensue. If the seal study of cavitation in relation to sharpshooter feeding, is weak, the insect returns to the XN behaviors to via EPG combined with magnetic resonance imaging strengthen the salivary seal. Sometimes, one or two of the plant, suggested that sharpshooters do not cav- isolated true X waves are produced, followed by a itate xylem cells (Perez-Donoso 2006). Cavitation pathway-like interruption when new sheath branches would preclude ingestion from a xylem cell, and there- are made to new cells. This supports that a cell can be fore is the most challenging impediment in xylem for rejected even at the XC stage (probably based on the sharpshooter to overcome. mechanical criteria) and the sheath branch aban- Saliva secreted during fB1w must become mixed doned. with xylem ßuid, because the food and salivary canals Do the Present Findings Meet the X Wave Criteria? in sharpshooter stylets do not conjoin at the tips The current study for the Þrst time identiÞes a hemipteran X wave as a landmark for penetration of (Leopold et al. 2003), as do those of aphids (Fereres a xylem tracheary element instead of a phloem sieve and Collar 2001). Thus, all saliva is Þrst ejected directly element. Although a Þrst, this is probably because no into the plant, and when ßuid uptake (during C1) other xylem-ingesting hemipteran has been EPG-re- occurs immediately after salivation (during fB1w), per corded and studied so extensively, before our work. To force a mixture of saliva and plant ßuid must be pulled conclusively prove that our sharpshooter X wave is a into the precibarium by cibarial diaphragm uplift (Du- marker for xylem ingestion, two criteria must be ful- gravot et al. 2008). That ßuid then would be swished Þlled. First, all ingestion waveforms from histologi- across the precibarial chemosensilla by ßuttering of cally identiÞed xylem cells must be preceded by an X the valve (interspersed B1s bursts) (Backus and Du- wave. Our Þndings strongly support this criterion. gravot 2009). Valve ßuttering may also sweep some of Second, all ingestion waveforms from identiÞed non- this ßuid back out of the stylets, perhaps contributing xylem cells must lack an X wave preceding them. Our to some egestion, as suggested by Backus et al. (2005). Þndings are sparse on this point, because we have a Thus, present evidence supports that C1-B1s-fB1w sample size of one probe in which this occurred. We component waveforms of XN represent behaviors that counter this argument, however, with our (and oth- work together in rapidly alternating sequence. These ersÕ) Þndings that Homalodisca spp. sharpshooters behaviors blend to accomplish multiple functions: 1) have very strong ingestion Þdelity to xylem cells; those clearing out the precibarium for optimum tasting and that have been EPG-recorded are quite unerring in pulling functionality, 2) sampling and tasting a mixture their ability to locate and accept those cells, although of xylem cell contents and saliva to judge acceptability (as shown by the present results) on some hosts it may of the cell, and 3) tightly sealing the stylet tips into the require considerable time and effort. Ingestion from xylem cell to prevent cavitation and ensure stronger nonxylem cells is a very rare event in those proconiine pulling against high xylem tension. It is likely that XN species, perhaps also in others. This idea can be tested behaviors function in gustation of cell constituents by comparison of recordings of cicadelline sharp- and mechanosensation of xylem tension (via stylet shooters, e.g., Graphocephala atropunctata (Signoret). proprioceptors; Backus and McLean 1982, Leopold et Previous recordings of that species (Almeida and al. 2003), tasks that are crucial for decision-making Backus 2004) found several putative ingestion wave- and, ultimately, xylem cell acceptance. Pathway XN forms that are quite different in appearance from C represents tasting and rejecting of an unacceptable waveforms. At the time, the authors speculated that cell (nonxylem, or too-small or nonligniÞed xylem these might represent ingestion from phloem or me- cell) followed by abandonment of that sheath branch. sophyll/parenchyma cells. Future studies can examine The XN portion of a true X wave represents tasting and this possibility, to determine whether this last crite- accepting of an appropriate xylem cell, as well as steps rion for X waves is fully realized. Nonetheless, we taken to overcome the challenge of cavitation via maintain that the behaviors underlying the X wave build-up of the sheath saliva seal. components must be performed in all xylem cells; 864 ANNALS OF THE ENTOMOLOGICAL SOCIETY OF AMERICA Vol. 102, no. 5 therefore, they are an appropriate justiÞcation for the ies of host acceptance and pathogen transmission be- X wave label. haviors in Hemiptera, ultimately leading to improved Do the Present Findings Match the Present Con- ecological and epidemiological modeling of many im- cepts/Definitions of X Waves? Some may question portant pest species. whether the term X wave is appropriate for the sharp- shooter host acceptance behaviors we describe here, New Definition of the X Wave because the details of sharpshooter behaviors differ from those of aphids, for whom X waves were Þrst We propose the following update for the X wave named. Yet, the original X wave deÞnition was devoid deÞnition. The X wave is a complex, multi-component, of reference to underlying behaviors (McLean and stereotypically repeating pattern of waveform types Kinsey 1967). It simply is a waveform that deÞnitively and subtypes whose species-speciÞc coarse and Þne identiÞed ingestion (ultimately) from a phloem sieve structures are visible even in highly time-compressed element (Reese et al. 2000). This deÞnition has been EPG recordings. The X wave encompasses a transition appropriately applied to several species of intracellu- among several waveform phases. A true (as opposed to lar-probing, nonsharpshooter leafhoppers for many partial) X wave represents stylet penetration of the years (e.g., Wayadande and Nault 1993), even though preferred ingestion cell type for a vascular-specializ- the behavioral details underlying those X waves may ing species, either phloem sieve elements or xylem be different from both aphids and sharpshooters. tracheary elements. The stylet activities (either in- Aphid X waves represent a series of intracellular gestion or salivation) that begin within the last X wave punctures (i.e., into the interior of sieve elements), and proceed from it occur in the vascular cell pene- during the aphidÕs intercellular mode of penetration, trated during the X wave. followed by salivation and stylet withdrawal back into Not every EPG researcher studying a new species the extracellular cell wall matrix outside the sieve has time to similarly tease apart the Þne-structure element. Once the Þnal penetration into the sieve correlations for every waveform type within the X element has been made (in the middle of the last pd, wave, to build such a bottom-up approach as we have or Þnal part of the X wave), E1 salivation into the cell demonstrated here. Our work is intended to provide commences, then eventually E2 ingestion. Sharp- a model that can be relied on for a more top-down shooters, as intracellular probers, cannot go from an approach that is warranted in such cases. Nonetheless, extracellular environment between cell walls to an based on the evidence here, we advocate restored use intracellular environment within the cell; in fact, they of the X wave concept. do the reverse (Miranda et al. 2009). Also, the second half of the sharpshooter X wave represents ingestion Implications for the Inoculation Mechanism of in the cell, not salivation outside the cell. Despite these X. fastidiosa differences, there are broad similarities between aphids and sharpshooters in X wave-represented be- It has been known for many years that the xylem- haviors. limited bacterium X. fastidiosa, the causative agent of In general, X wave behaviors function to 1) senso- PD, is the only known noncirculative yet propagative rially test, judge, and accept the internal environment plant pathogen transmitted by an insect vector (for of a potential ingestion cell; 2) secure the styletsÕ Þrm review, see Almeida et al. 2005; Backus et al. 2005, and attachment to the cell; and 3) begin overcoming the references therein). This is because X. fastidiosa forms challenges to sustained ingestion presented by the dense mats of bacterial bioÞlm in the precibarium and cellular environment, i.e., cavitation of xylem cells for cibarium of sharpshooters and other xylem ßuid-in- sharpshooters or callose plugging/chemical defense of gesting hemipterans (Brlansky et al. 1983, Almeida and sieve elements for aphids (Will et al. 2008). Thus, the Purcell 2006). Bacteria are inoculated directly into X wave represents a complicated, species-speciÞc rep- xylem cells during some until-now unknown part of ertoire of behavioral-physiological building blocks the stylet penetration process, but which all research- that represent that speciesÕ cell acceptance process, ers agree probably involves the precibarial valve taking into consideration its mode of stylet penetra- (Backus and McLean 1983, Brlansky et al. 1983, Purcell tion. 1989, Backus et al. 2005, Almeida and Purcell 2006). In an ecological consideration of the host accep- We hypothesize that the sharpshooter X wave rep- tance process, the X wave represents the critical tran- resents the “insect-associated probing behavior re- sitions from Þnding to accepting to consuming food sponsible for inoculation events” (Chatterjee et al. (Miller and Strickler 1984) and is therefore crucial to 2008) or the inoculation behavior for X. fastidiosa, each speciesÕ survival. This cell acceptance process is because: 1) the X wave is strictly associated with also crucial to the mechanisms of transmission of plant xylem; and 2) it has been correlated with (or hypoth- pathogens. We maintain that the X waveÕs chief utility esized to represent) all behavioral building blocks is in allowing our recognition of these specialized likely to be involved in inoculation, including move- cell-acceptance behaviors, in this broad behavioral ments of the precibarial valve and cibarial diaphragm and ecological sense. Understanding that, it should be which must be the sole means of propelling egestion possible to test whether every sheath-producing spe- of plant ßuids. Furthermore, based on these biological ciesÕ EPG recordings reveal an X wave. If so, then the meanings, we propose a new hypothesis for one mech- X wave can be a tool for comparative behavioral stud- anism of noncirculative pathogen transmission by in- September 2009 BACKUS ET AL.: SHARPSHOOTER XWAVE 865 sect vectors: the ingestion-(salivation and egestion) plant rearing and to Dr. Thomas Rost (University of Cali- hypothesis (Backus 2007). fornia, Davis) for several useful conversations on plant anat- As in all other hypotheses for vector transmission, omy. This research was done in partial fulÞllment of the ours accepts that the acquisition portion of transmis- requirements of a Master of Science degree by W.J.H. at sion is via ingestion from infected plant cells. Our new California State University, Fresno. This work was supported by a pass-through grant from USDAÐCSREES via the Uni- idea is that one mechanism of inoculation can be via versity of California PierceÕs Disease Research Program, and a combination of salivation and egestion. Thus, we in-house ARS funding. unify the existing two hypotheses, the ingestion-eges- tion hypothesis (Harris, 1977, Harris and Harris, 2001) and ingestion-salivation hypothesis (Martin et al. 1997; References Cited both hypotheses reviewed in Fereres and Collar 2001). SpeciÞcally, we propose that the saliva brought Alhaddad, H. 2008. Salivary secretions of Homalodisca vit- ripennis and their relation to Xylella fastidiosa inocula- into the precibarium during the X wave not only acts tion. M.S. thesis, California State University-Fresno, as a solvent for plant chemical cues for gustation by the Fresno. precibarial chemosensilla but also acts enzymatically Almeida, R.P.P., and E. A. Backus. 2004. Stylet penetration to break down the adhesive properties that bind X. behaviors of Graphocephala atropunctata (Signoret) fastidiosa cells to the cuticular intima of the (Hemiptera: Cicadellidae): EPG waveform characteriza- precibarium (Chatterjee et al. 2008). Once the bac- tion and quantiÞcation. Ann. Entomol. Soc. Am. 97: 838Ð teria have been loosened from their mooring on the 851. cuticle, they are then swept out the stylets by the Almeida, R., and A. H. Purcell. 2006. Patterns of Xylella combined movements of the precibarial valve and fastidiosa colonization on the precibarium of sharp- shooter vectors relative to transmission to plants. Ann. cibarial diaphragm. Entomol. Soc. Am. 99: 884Ð890. Our hypothesis is supported by the recent, intrigu- Almeida, R., M. Blua, J. Lopes, and A. H. Purcell. 2005. ing Þnding of Ramirez et al. (2008) that 40% of salivary Vector transmission of Xylella fastidiosa: applying funda- deposits made by X. fastidiosa-inoculative glassy- mental knowledge to generate disease management strat- winged sharpshooter during labial exploration of a egies. Ann. Entomol. Soc. Am. 98: 775Ð786. hard surface contained X. fastidiosa bacteria. How- Backus, E. A. 1985. Anatomical and sensory mechanisms of ever, frequency of X. fastidiosa detection was not leafhopper and planthopper feeding behavior, pp. 163Ð correlated with titer of X. fastidiosa in the heads. 194. In L. R. Nault and J. G. Rodriguez [eds.], The leaf- Salivation during surface exploration is thought to hoppers and planthoppers. Wiley, New York. Backus, E. A. 1988. Sensory systems and behaviours which dissolve plant epidermal compounds for subsequent mediate hemipteran plant-feeding: a taxonomic over- uptake and tasting by the precibarial chemosensilla view. J. Insect Physiol. 34: 151Ð165. (Backus 1985). Backus, E. A. 2000. Our own jabberwocky: clarifying the Recent observations show that glassy-winged sharp- terminology of certain piercing-sucking behaviors of ho- shooter salivates as a reßex response to physical mopterans. In G. P. Walker and E. A. Backus [eds.], stimulation of the labium and other facial surfaces Principles and applications of electronic monitoring and (Alhaddad 2008). Therefore, the only plausible expla- other techniques in the study of homopteran feeding nation of the X. fastidiosa Þndings by Ramirez et al. behavior. Thomas Say Publications in Entomology, En- (2008) is that saliva is secreted during surface explo- tomological Society of America, Lanham, MD. Backus, E. A. 2007. Feeding behaviors of the glassy-winged ration and then is taken up to the precibarium. There, sharpshooter that control inoculation of Xylella fastid- it may dislodge a few X. fastidiosa bacteria from what iosa, pp. 116Ð119. In T. Esser [ed.], Proceedings of the could be a large or small colony in the precibarium and 2007 PierceÕs Disease Research Symposium. California cibarium. The salivaÐbacteria mixture is then egested Department of Food and Agriculture, 12Ð14 December out the stylets to mix with freshly secreted saliva. 2007, San Diego, CA. Several research projects in the Backus laboratory Backus, E. A., and W. H. Bennett. 2009. The AC-DC corre- are currently testing key aspects of our ingestion- lation monitor: new EPG design with ßexible input resistors (salivation and egestion) hypothesis with H. vitripen- to detect both R and emf components for any piercing- nis and X. fastidiosa. Ultimately, these Þndings will be sucking hemipteran. J. Insect Physiol. 55: 869Ð884. Backus, E. A., and S. Dugravot. 2009. Precibarial valve used to develop 1) a comprehensive model of the role muscle potentials of Homalodisca spp. sharpshooters of feeding behavior and physiology in X. fastidiosa (Hemiptera: Cicadellidae) correlated with an EPG path- transmission efÞciency; and 2) a stylet penetration way waveform. J. Insect Physiol. (in press). index (Serrano et al. 2000), an experimental technique Backus, E. A., and D. L. McLean. 1982. The sensory systems for rapid, nondestructive screening of host plants to and feeding behavior of leafhoppers. I. The aster leaf- analyze natural or transgenic resistance to vector in- hopper, Macrosteles fascifrons Stål (Homoptera: Cicadel- oculation. lidae). J. Morphol. 172: 361Ð379. Backus, E. A., and D. L. McLean. 1983. The sensory systems and feeding behavior of leafhoppers. II. A comparison of Acknowledgments the sensillar morphologies of several species (Ho- moptera: Cicadellidae). J. Morphol. 176: 3Ð14. We thank Holly Shugart (ARS, Parlier, CA) for instructing Backus, E. A., and D. L. McLean. 1985. Behavioral evidence W.J.H. in histology and light microscopy and for aiding im- that the precibarial sensilla of leafhoppers are chemo- measurably in the correlation analysis. Appreciation also is sensory and function in host discrimination. Entomol. extended to Jose Gutierrez (ARS, Parlier) for insect and Exp. Appl. 37: 219Ð228. 866 ANNALS OF THE ENTOMOLOGICAL SOCIETY OF AMERICA Vol. 102, no. 5

Backus, E. A., M. J. Devaney, and W. H. Bennett. 2000. lation of electronically recorded waveforms with aphid Comparison of signal processing circuits among seven AC probing activities. Ann. Entomol. Soc. Am. 60: 400Ð405. electronic monitoring systems for their effects on the emf McLean, D. L., and M. G. Kinsey. 1984. The precibarial and R components of aphid (Homoptera: Aphididae) valve and its role in the feeding behavior of the pea aphid, waveforms, pp. 102Ð143. In G. P. Walker and E. A. Backus Acyrthosiphon pisum. Bull. Entomol. Soc. Am. 30: 26Ð31. [eds.], Principles and applications of electronic monitor- Miller, J. R., and K. L. Strickler. 1984. Finding and accepting ing and other techniques in the study of homopteran host plants, pp. 127Ð157. In W. J. Bell and R. T. Carde´ feeding behavior. Entomological Society of America, [eds.], Chemical ecology of insects. Chapman & Hall, Lanham, MD. London, United Kingdom. Backus, E., J. Habibi, F. Yan, and M. Ellersieck. 2005. Stylet Miranda, M. P., A. Fereres, B. Appezzato-Da-Gloria, and penetration by adult glassy-winged sharpshooter on J.R.S. Lopes. 2009. Characterization of electrical pene- grape: EPG waveform characterization, tissue correla- tration graphs of Bucephalogonia xanthophis, a vector of tions, and possible implication for transmission of Xylella Xylella fastidiosa in citrus. Entomol. Exp. Appl. 130: 35Ð fastidiosa. Ann. Entomol. Soc. Am. 98: 787Ð813. 46. Berlyn, G., and J. Miksche. 1976. Botanical microtechnique Perez-Donoso, A. 2006. The development of PierceÕs dis- and cytochemistry. Iowa State University Press, Ames, IA. ease in grapevines: the relationship of water conduit func- Brlansky, R. H., L. W. Timmer, W. J. French, and R. E. tion to pathogen spread and disease symptoms. In Plant McCoy. 1983. Colonization of the sharpshooter vectors, sciences. University of California, Davis, Davis, CA. Oncometopia and Homalodisca, by xylem limited bacteria. Powell, G., T. Pirone, and J. Hardie. 1995. Aphid stylet ac- Phytopathology 73: 530Ð535. tivities during potyvirus acquisition from plants and an in Chatterjee, S., R.P.P. Almeida, and S. Lindow. 2008. Living vitro system that correlates with subsequent transmission. in two worlds: the plant and insect lifestyles of Xylella Eur. J. Plant Pathol. 101: 411Ð420. fastidiosa. Annu. Rev. Phytopathol. 46: 243Ð271. Purcell, A. H. 1989. Homopteran transmission of xylem-in- Dugravot, S., E. A. Backus, B. Reardon, and T. A. Miller. habiting bacteria, pp. 243Ð266. In K. F. Harris [ed.], Ad- 2008. Correlations of cibarial muscle activities with EPG vances in Disease Vector Research. Springer-Verlag, Inc., waveforms and excretion in sharpshooter (Hemiptera: New York. Cicadellidae: Cicadellinae) vectors of Xylella fastidiosa. Ramirez, J. L., P. T. Lacava, and T. A. Miller. 2008. Detec- J. Insect Physiol. 54: 1467Ð1478. tion of the bacterium, Xylella fastidiosa, in saliva of glassy- Fereres, A., and J. L. Collar. 2001. Analysis of noncirculative winged sharpshooter, Homalodisca vitripennis. J. Insect transmission by electrical penetration graphs, pp. 87Ð109. Sci. 8: 1Ð7. In K. F. Harris, O. P. Smith, and J. E. Duffus [eds.], Rapusas, H. R., and E. A. Heinrichs. 1990. Feeding behavior Virus-insect-plant interactions. Academic, New York. of Nephotettix virescens (Homoptera: Cicadellidae) on Fisher, D. B., and J. M. Frame. 1984. A guide to the use of rice varieties with different levels of resistance. Environ. the exuding-stylet technique in phloem physiology. Entomol. 19: 594Ð602. Planta 161: 385Ð393. Redak, R. A., A. H. Purcell., J.R.S. Lopes, M. A. Blua, R. F. Harris, K. F. 1977. An ingestion-egestion hypothesis of non- Mizell, and P. C. Anderson. 2004. The biology of xylem- circulative virus transmission, pp. 165Ð220. In K. F. Harris ßuid feeding insect vectors of Xylella fastidiosa and their and K. Maramorosch [eds.], Aphids as virus vectors. Ac- ademic, New York. relation to disease epidemiology. Annu. Rev. Entomol. 49: Harris, K. F., and L. J. Harris. 2001. Ingestion-egestion the- 243Ð270. ory of cuticula-borne virus transmission, pp. 376. In K. F. Reese, J., W. F. Tjalingii, M. Van Helden, and E. Prado. 2000. Harris, O. P. Smith, and J. E. Duffus [eds.], Virus-insect- Waveform comparisons among AC and DC electronic plant interactions. Academic, New York. monitoring systems for aphid (Homoptera: Aphididae) Holmes, W. J. 2007. EPG waveforms of sharpshooters cor- feeding behavior, pp. 70Ð101. In G. P. Walker and E. A. related with Þrst penetration of xylem. M.S. thesis. Cal- Backus [eds.], Principles and applications of electronic ifornia State University, Fresno. monitoring and other techniques in the study of ho- Joost, P. H., E. A. Backus, and F. Yan. 2006. Correlation of mopteran feeding behavior. Entomological Society of stylet activities by the glassy-winged sharpshooter, Ho- America, Lanham, MD. malodisca oagulate (Say), with electrical penetration Ruzin, S. 1999. Plant microtechnique and microscopy. Ox- graph (EPG waveforms). J. Insect Physiol. 52: 327Ð337. ford University Press, New York. Kimmins, F. M. 1989. Electrical penetration graphs from Sandanayaka, W.R.M., and E. A. Backus. 2008. Quantitative Nilaparvata lugens on resistant and susceptible rice vari- comparison of stylet penetration behaviors of glassy- eties. Entomol. Exp. Appl. 50: 69Ð79. winged sharpshooter, Homalodisca vitripennis, on four Leopold, R. A., T. P. Freeman, J. S. Buckner, and D. R. crop plants important in New Zealand and the USA. J. Nelson. 2003. Mouthpart morphology and stylet pene- Econ. Entomol. 98: 787Ð813. tration of host plants by the glassy-winged sharpshooter, SAS Institute. 2001. The SAS userÕs guide. Statistical analysis Homalodisca coagulata (Homoptera: Cicadellidae). Arth. systems. SAS Institute, Cary, NC. Struct. Func. 32: 189Ð199. Salleo, S., A. Nardini, F. Pitt, and M. A. Lo Gullo. 2000. Martin, B., J. L. Collar, W. F. Tjallingii, and A. Fereres. 1997. Xylem cavitation and hydraulic control of stomatal con- Intracellular ingestion and salivation by aphids may cause ductance in laurel (Laurus nobilis L.). Plant Cell Environ. the acquisition and inoculation of non-persistently trans- 23: 71Ð79. mitted plant viruses. J. Gen. Virol. 78: 2701Ð2705. Serrano, M. S., E. A. Backus, and C. Cardona. 2000. Com- McLean, D. L., and M. G. Kinsey. 1965. IdentiÞcation of parison of AC electronic monitoring and Þeld data for electrically recorded curve patterns associated with estimating tolerance to Empoasca kraemeri (Homoptera: aphid salivation and ingestion. Nature (Lond.) 205: 1130Ð Cicadellidae) in common bean genotypes. J. Econ. En- 1131. tomol. 93: 1796Ð1809. McLean, D. L., and M. G. Kinsey. 1967. Probing behavior of Snodgrass, R. E. 1935. Principles of insect morphology. the pea aphid, Acyrthosiphon pisum. I. DeÞnitive corre- McGraw-Hill, New York. September 2009 BACKUS ET AL.: SHARPSHOOTER XWAVE 867

Takiya, D., S. Mckemey, and D. Cavichiol. 2006. Validity of moptera: Delphacidae) on resistant and susceptible rice Homalodisca and of H. vitripennis as the name for glassy cultivars. Environ. Entomol. 15: 678Ð682. winged sharpshooter (Hemiptera: Cicadellidae: Cicadel- Walker, G. P. 2000. A beginner guide to electronic moni- linae). Ann. Entomol .Soc. Am. 99: 648Ð655. toring of homopteran feeding behavior, pp. 14Ð20. In G. P. Tjallingii, W. F. 2000. Comparison of AC and DC systems Walker and E. A. Backus [eds.], Principles and applica- for electronic monitoring of stylet penetration activities tions of electronic monitoring and other techniques in the by homopterans, pp. 41Ð69. In G. P. Walker and E. A. study of homopteran feeding behavior. Thomas Say Pub- Backus [eds.], Principles and applications of electronic lications in Entomology, Entomological Society of Amer- monitoring and other techniques in the study of ho- ica, Lanham, MD. mopteran feeding behavior. Entomological Society of Wayadande, A. C., and L. R. Nault. 1993. Leafhopper prob- America, Lanham, MD. ing behavior associated with maize chlorotic dwarf virus Tjallingii, W. F., and B. Gabrys. 1999. Anomalous stylet transmission to maize. Phytopathology 83: 522Ð526. punctures of phloem sieve elements by aphids. Entomol. Will, T., W. F. Tjallingii, A. Thonnessen, and A.J.E. Van Exp. Appl. 91: 97Ð103. Bel. 2008. Molecular sabotage of plant defense by Triplehorn, B. W., L. R. Nault, and D. J. Horn. 1984. Feed- aphid saliva. Proc. Natl. Acad. Sci. U.S.A. 104: 10536Ð ing behavior of Graminella nigrifrons (Forbes). Ann. En- 10541. tomol. Soc. Am. 77: 102Ð107. Velusamy, R., and E. A. Heinrichs. 1986. Electronic moni- toring of feeding behavior of Nilaparvata lugens (Ho- Received 1 October 2008; accepted 17 June 2009.