Susceptibility of Rhododendron to Azalea Lace Bug, Stephanitis pyrioides (Scott)¹

The invasive azalea lace bug, Stephanitis pyrioides (Scott), is one of the most damaging insect pests of the genus Rhododendron, especially evergreen azaleas, an especially popular subgenus of Rhododendron. Feeding by nymphs and adults remove chlorophyll from leaves, causing stippling on the top of the leaf, which reduces the aesthetic value of infested plants, and severe infestations can result in plant death. Rhododendron species exhibit a remarkable range of diversity in form, foliage and flower, and are prized worldwide in the areas that they can be grown. They are extensively hybridized in cultivation, with >28,000 cultivars listed in the International Rhododendron Registry held by the Royal Horticultural Society. Given the extensive use of Rhododendron species and varieties in ornamental plantings in both public and private landscapes, identifying those plants that are susceptible to azalea lace bug will allow control measures to be more effectively targeted against this invasive pest. Identifying resistant species or cultivars will assist breeding programs to develop Rhododendron plants that have desirable aesthetic qualities as well as resistance to azalea lace bug. Lastly, research focusing on resistant species and varieties may also shed light on the mechanisms of resistance in Rhododendron to azalea lace bug.

The non-native, invasive azalea lace bug, Stephanitis pyrioides (Scott) (Tingidae: Hempitera), is one of the most damaging insect pests of the genus Rhododendron, especially azaleas (Schuh and Slater 1995, Shrewsbury and Smith-Fiola 2000, Klingeman et al. 2001, Nair and Braman 2012). Feeding by nymphs and adult S. pyrioides removes chlorophyll from leaves, reducing rates of photosynthesis, and transpiration of infested plants (Buntin et al. 1996). Chlorotic stippling appears on the top of the leaf (Johnson and Lyon 1994), which causes severe economic damage to landscapes and cultivated Rhododendron in areas where they are grown (Shrewsbury and Smith-Fiola 2000, Klingeman et al. 2001). Stephanitis pyrioides is native to Japan, and was first reported in the United States in New Jersey where it was thought to have been introduced on infested nursery stock from Japan (Weiss 1916). It is now present throughout the eastern United States where azaleas are grown (Weiss and Headlee 1918, Drake and Ruhoff 1965). It was first detected in Washington in 2007 (Looney et al. 2016), and it is now established in Oregon, Washington, and British Columbia (Rosetta 2013, Lee et al. 2019, Garrison and Tobin 2022).

The biology and life history of S. pyrioides have been previously studied (Nair and Braman 2012). Briefly, life stages of S. pyrioides are generally found on the underside of leaves. Eggs are oviposited in the leaf generally along the midrib or along larger side veins (Neal and Douglass 1988). Nymphs are colorless at hatching and turn darker after each successive molt, and there are five instars. It undergoes 2-3 generations in the Pacific Northwest (Lee et al. 2019, Garrison and Tobin 2022). Eggs diapause over the winter and hatch in spring following the accumulation of the required number of degree days (Garrison and Tobin 2022).

Prior reports describe the occurrence of S. pyrioides on species of Rhododendron, as well as other species within the Ericaceae (Drake 1965, Nair and Braman 2012). Ericaceae is a large family with approximately 4,250 known species across 124 genera. It is also an economically important family, with some species having edible berries (e.g., species within Arctostaphylos, Gaylussacia, and Vaccinium; Venturini et al. 2017) and others cultivated as ornamental plants (e.g., species within Erica, Rhododendron, Arbutus, and Pieris; Yu et al. 2009). There are also limited reports of S. pyrioides attacking hosts of plant families other than Ericaceae; for example, damage by S. pyrioides have been observed on Eucryphia sp. Gay and Oemleria cerasiformis (Torr. & A. Gray ex Hook & Arn.) Landon, but it is not known if it can complete its life cycle on these genera.

The genus Rhododendron contains >1,000 species of woody ornamentals mostly known for their showy flowers, but also for its ecological importance (Paul et al. 2005, United States Department of Agriculture 2011). It is widely cultivated in temperate regions, and in the wild they form important components of montane ecosystems. The genus Rhododendron exhibits a wide range of form, foliage and flower (Nelson 2001). The majority of wild Rhododendron species occur in the Himalayas and southeastern Tibet, or in the mountain ranges that form the backbones of the archipelago stretching between mainland Asia and Australia, the islands of Java, Sumatra, Borneo, New Guinea, and the Philippines (Irving and Hebda 1993). The remaining species are distributed over the northern hemisphere, occurring in pockets that are isolated from one another in Japan, northwestern North America, and the Appalachian and Caucasus Mountains (Irving and Hebda 1993).

The current and most accepted hypothesis of its geographic distribution is that Rhododendron was more widely and continuously distributed across North America and Eurasia during the milder climate of the Tertiary geologic era, approximately 65 million years ago (MYA) (Nelson 2001). Before the onset of the glacial periods during the late Tertiary, Rhododendron could have extended more or less continuously from North America to Europe, and eastward into China and northeastern Asia (Nelson 2001). However, as the climate changed and conditions became less favorable to Rhododendron, populations became isolated in only those regions where survival was possible (Nelson 2001). The collision of India with Asia about 40 MYA created a region favorable to Rhododendron, a “region of extreme relief” (Nelson 2001) where deep valleys are clustered close together (Irving and Hebda 1993). At the onset of hostile climate conditions, Rhododendron populations at the margins of this area were able to persist in this region of extreme relief, taking advantage of newly developed and amiable conditions (Nelson 2001). From here, plants within the Rhododendron section Vireya are hypothesized to have spread into the high-island archipelago in the Indo-Malaysian region based upon a global genetic reconstruction of its phylogeny supporting this northeast Asian origin of Rhododendron (Shrestha et al. 2018). This work also suggests that Rhododendron first dispersed out of northeast Asia into North America in the mid-Eocene, followed by dispersal to South Asia and the Malay Archipelago in the late Eocene (Shrestha et al. 2018). Multiple dispersal events were inferred throughout the late Eocene and Oligocene between northeast Asia and the other regions, leading to paraphyletic assemblages in these regions (Shrestha et al. 2018). The ancestor of the Australian species likely occurred in the Malay Archipelago first, and dispersed to Australia approximately 10 MYA (Shrestha et al. 2018).

The genus Rhododendron can be separated into two categories: lepidote and elepidote. Lepidote species have specialized trichome structures on their leaves and elepidote species lack these structures. These structures in large part regulate the water supply of the plant and thus enable Rhododendron to withstand climatic extremes (Cowan 1950). Trichomes often take the form of scales or dense hairy layers called indumentum, but the architecture of the trichome varies within the genus nearly as widely as it does within the plant kingdom (Cowan 1950). Beyond the separation of lepidote and elepidote, further subdivision of Rhododendron has been, and still is, a complex and evolving process. In 1753, Linnaeus proposed two genera, Rhododendron and Azalea, into which 9 recognized species were placed (Cox and Cox 1997). Subsequent botanists such as Salisbury and Tate began to question the distinction between Azalea (which has 5 stamens) and Rhododendron (10 stamens), and in 1834, Azalea was incorporated into the genus Rhododendron (Don 1831). Over the years many botanists have further reorganized the taxonomy of Rhododendron, including the prominent 19^th and 20^th century botanists G. Don, C. J. Maximovicz, J. D. Hooker, and T. Nakai (Cox and Cox 1997). From 1916 onward, the classification of the genus has been largely carried out by botanists working in the herbarium of the Royal Botanic Garden, Edinburgh, Scotland (Cox and Cox 1997). From 1916 to 1922 Sir Isaac Bayley Balfour developed the system of Series (Cox and Cox 1997). In this system, groups of related Rhododendrons were placed in Series, and one species in each Series was selected as typical of its associates (Sleumer et al. 1978). The series were conglomerations, some with large numbers of extremely diverse species, others with a few species differing from those in another series by nuance (Sleumer et al. 1978). Although Balfour acknowledged that his system of Series was more of an expedient than a scientific statement, they did provide a conveniently ordered array for the scores of species that came pouring into Europe from the expeditions of Forrest, Kingdon-Ward, Farrer and others in the early years of the 20^th century (Sleumer et al. 1978, Cox and Cox 1997). Unfortunately, Balfour died before he could remedy the situation, and the authors' temporary stopgap measures became engraved in stone for many English-speaking horticulturalists. In 1937, Sleumer began to produce a practical, comprehensive and cohesive classification of Rhododendron, based on the work of 19^th century taxonomists and revised by his own research (Sleumer et al. 1978). Subsequently, the conclusions of a number of more narrowly focused morphological taxonomic studies (Sleumer 1966, Cullen 1980, Chamberlain 1982, Philipson and Philipson 1986, Judd and Kron 1995) were incorporated into an alternative Rhododendron classification (Chamberlain et al. 1996, Goetsch et al. 2005). This taxonomic system has been generally accepted by Rhododendron specialists (Cox and Cox 1997) because it embodies the findings of substantially all morphology-based Rhododendron systematic studies since 1980 (Goetsch et al. 2005).

Following the genetic analysis of Goetsch et al. (2005, 2011), these subgenera were further grouped into a higher level based on the discovery of three major clades (A, B, C) (Fig. 1). Subgenera Rhododendron and Hymenanthes are nested within clades A and B as monophyletic groups, respectively (Goetch et al. 2005). In contrast, subgenera Azaleastrum and Pentanthera were polyphyletic, and were divided between two clades. The four sections of Pentanthera were divided between clades B and C, with two each, while Azaleastrum had one section in each of A and C (Goetch et al. 2005). The subgenus Rhododendron was relatively untouched with regard to its three sections. Four other subgenera were eliminated and one new subgenus was created, leaving a total of five subgenera in all, compared to eight from Chamberlain et al. (1996). The discontinued subgenera are Pentanthera, Tsutsusi, Candidastrum and Mumeazalea, while a new subgenus was proposed by elevating section Azaleastrum section to subgenus rank (Goetch et al. 2005, 2011, Shrestha et al. 2018). In 2018, Shrestha et al. (2018) performed a phylogenetic reconstruction based on 423 species using 16 gene regions that fit well with the current understanding of the evolutionary relationships and time-scale of diversification of the genus (Goetsch et al. 2005, 2011).

Before 1860, the richness of the native flora of China was unsuspected and unknown; China was closed to Europeans and to foreign travel until the middle of the 19^th century. Foreigners were generally not allowed to go more than a few kilometers outside Canton and Macao, the only ports open to Europeans; thus, it was impossible for early collectors from Europe and elsewhere to explore the country (Davidian 1996). At the conclusion of the second Opium War, China was opened to foreign travel (Davidian 1996), and this opening and the exposure to the richness of the diversity of Rhododendron in the region of extreme relief led to the introduction and development of Rhododendron as “king of shrubs” (Nelson 2001). By the end of the 19^th century, a period of plant collecting in China by amateurs, missionaries, travelers, merchants and diplomats was over, and a new period of the professional or horticultural collector with E. H. Wilson and George Forrest began (Davidian 1996). These early 20^th century Rhododendron discoveries provided a vital link in the horticultural cultivation and dissemination of Rhododendron worldwide (Nelson 2001).

Differences in susceptibility of various Rhododendron species and cultivars to S. pyrioides have been previously studied. For example, deciduous Rhododendron spp. were found to be less suitable for adult feeding, oviposition, and nymphal development of S. pyrioides than the evergreen R. mucronatum variety ‘Delaware Valley White'; however, all species supported adult activity and oviposition in no-choice and free-choice tests (Braman and Pendley 1992). Another study reported that plant physical characteristics, such as bloom color and abaxial leaf texture, were not associated with host plant acceptance by S. pyrioides (Schultz 1993). Several possible mechanisms of resistance to S. pyrioides have been investigated, including the role of epicuticular waxes (Balsdon et al. 1995, Wang et al. 1998, 1999, Chappell and Robacker 2006). As S. pyrioides rests on the abaxial leaf surface, it comes into contact with these waxes, which are hypothesized to directly inhibit feeding or otherwise give a chemical signal that reduces feeding. Resistant and susceptible deciduous cultivars differ in components of the leaf-surface lipids, identified as n-alkanes and triterpenoids, and these lipids were negatively associated with S. pyrioides behavior as measured by oviposition, egg and nymphal development, nymphal survivorship and leaf area damaged (Nair and Braman 2012). Leaf wax extracts from resistant genotypes, when applied to susceptible ones, resulted in resistance to feeding and oviposition by S. pyrioides; in contrast, wax extracts from susceptible genotypes when applied to resistant ones increased susceptibility, indicating that leaf wax plays a role in resistance to S. pyrioides (Clark 2000, Chappell et al. 2004, Chappell and Robacker 2005, 2006, Chappell 2007, Nair and Braman 2012).

Leaf pubescence (Schultz 1993, Wang et al.1998), stomatal character (Kirker et al. 2008), and leaf moisture content (Wang et al.1998) have also been investigated. Because S. pyrioides feeds on the underside of the leaf, inserting their stylet through the stomata, leaf characteristics are hypothesized to affect the ability of S. pyrioides to insert its stylet to the palisade parenchyma where it feeds. In a study of 17 deciduous cultivars and one evergreen species, leaf water content and leaf hair density were not correlated with S. pyrioides damage, except in Rhododendron canescens, which had extremely high trichome density and was highly resistant to azalea lace bug feeding (Wang et al. 1998). Likewise, stomatal characters of 33 azalea cultivars were compared with their preference by S. pyrioides and, although stomata size differed significantly among the cultivars, they were not associated with S. pyrioides feeding preference (Kirker et al. 2008). However, across all these studies, a relatively few numbers of species and cultivars have been evaluated for their susceptibility to S. pyrioides.

The maritime climate of the Pacific Northwest is particularly well suited to growing Rhododendron, and most species, except the tropical section Vireya, can be grown outdoors in landscapes. During the winter months in the Pacific Northwest, mild temperatures, continual rainfall, and cloud cover are all conducive to the growing of Rhododendron, but during the summer months irrigation is often necessary (Nelson 2001). This beneficial climate and long history of Rhododendron cultivation in the Pacific Northwest provides an opportunity to study the damage caused by S. pyrioides in a wider variety of Rhododendron than has been done previously. In this paper, we sampled 71 Rhododendron species and cultivars over two years and quantified the damage done by S. pyrioides. Leaves were sampled during mid-summer when damage by S. pyrioides was generally the highest. Leaves were scanned after their collection, and the proportion of leaf area damaged was calculated for each sampled plant. Damage proportions were analyzed to determine if any patterns could be ascertained within the phylogenetic tree of Rhododendron, and we hypothesized that at some level of classification, differences in susceptibility would be detected.

This study was conducted at the Washington Park Arboretum in Seattle, Washington (47° 38′ 28.32″ N, 122° 17′ 36.996″ W), which contains >14,500 accessioned plant specimens including >4,000 different types of trees, shrubs and other plants native to 98 countries. Since the inception of the Washington Park Arboretum in 1936, Rhododendron spp. have been planted extensively, and there are >2,400 individual records for the genus Rhododendron. Plant records at the Washington Park Arboretum are georeferenced and maintained by the University of Washington Botanic Gardens.

Rhododendron plants to be sampled were first chosen by species or cultivars that had been measured in previous studies (e.g., Wang et al. 1998). Rhododendron spp. were also selected from available species and cultivars within the Washington Park Arboretum across the phylogenetic tree of Rhododendron, and these plants were selected based on the phylogenetic analysis by Goetch et al. (2005); this selection procedure resulted in 26 species and cultivars. Rhododendron spp. and cultivars were also selected across the taxonomic family tree from Cox and Cox (1997). Lastly, 41 species and cultivars were selected due to their phylogenic proximity to those already included for sampling. The 71 species and cultivars selected for this study are listed in Table 1, and full methodological details are available in Garrison (2020).

To quantify susceptibility of Rhododendron to S. pyrioides, a total of 80 leaves were collected from one plant of each of the 71 different Rhododendron species, varieties, and cultivars in August of 2018 and 2019. August was chosen so that at least one full generation of S. pyrioides would have been completed, with adults from the first generation and immatures from the second generation having the opportunity to feed on host plants (Garrison and Tobin 2022). Twenty leaves were collected from each plant once a week for 4 weeks. From each plant, four leaves were randomly collected from the north, south, east and west side of the plant, alternating from inner to outer canopy, and then four leaves were chosen at random from the plant. Leaves were placed in a re-sealable plastic bag, labeled, and stored at 4 C (39 F) for no more than one week to avoid leaf degradation. During preliminary data collection, it was observed that storage for one week at 4 C did not alter the samples, but that some leaves stored longer than two weeks became discolored, which could affect their analysis.

Leaves were removed from the plastic bag, and placed between two pieces of clear polycarbonate sheets. Both sides of the leaves were scanned. The scans of the top sides of leaves were used to calculate damage from S. pyrioides for use in subsequent analyses. The scans of the undersides of the leaves were only used to confirm that damage was indeed caused by S. pyrioides based on the presence of frass and life stages (i.e., nymphs or adults). Leaves were scanned with an Epson Perfection V800 Photo scanner at 600 dpi using SilverFast SE 8 scanning software (https://www.silverfast.com/) and saved as a JPEG file. Files were named with the accession number and species name of each Rhododendron species or cultivar. Leaf damage by S. pyrioides was estimated using Assess 2.0: Image Analysis Software for Plant Disease Quantification (Lamari 2008). This program uses a threshold level of leaf coloration that separates the area of chlorosis caused by S. pyrioides damage from the total leaf area. The protocol for measuring leaf area and the area damaged was as follows. The program has two modes, ‘Leaf' for measuring leaf area, and ‘Lesion' for measuring disease (in this case chlorotic areas). We used the 'Leaf' mode to select the entire leaf, the 'Lesion' mode to select the chlorotic area, from which we could estimate the proportion of the leaf area damaged. Of the 80 leaves collected for each Rhododendron species, varieties, and cultivars in each year, a mean of 77 provided usable data.

For each Rhododendron species, cultivar or variety, we used the mean proportion of leaf area damaged by S. pyrioides as the response variable. Mean proportions were calculated from leaves sampled in 2018 and 2019, and were transformed according to log_e (mean proportion / (1 - mean proportion)) to meet the assumption of normality in an Analysis of Variance (ANOVA). Prior to the transformations, means were adjusted by the addition of 0.001 to permit transformation of zero means. We tested the effect of trichome presence (lepidote) or absence (elepidote) on the transformed mean proportion of damage using ANOVA. We also examined the effect of Rhododendron subgenus and section on the transformed mean proportion of damage using ANOVA. Post-hoc tests, when appropriate, were done using Tukey's HSD (honestly significant difference). All statistical analyses were done in R (R Core Team 2018).

In 2018 and 2019, a total of 5,680 and 5,738 leaves were collected and scanned, respectively. The mean proportion of leaf area damaged for each Rhododendron species, cultivar or variety for both years is presented in Table 1. Overall, the proportion of the leaf damaged by S. pyrioides across all Rhododendron plants ranged from 0 to 0.3451. Rhododendron calendulaceum, and R. canescens were the only two species available in the Washington Park Arboretum that had been previously evaluated for their susceptibility to S. pyrioides (Wang et al. 1998). Rhododendron calendulaceum was described as less suitable for S. pyrioides feeding and oviposition by Braman and Pendley (1992). In this study, we observed similar findings as there was no measurable damage recorded on R. calendulaceum. Rhododendron canescens was described as resistant to S. pyrioides by Wang et al. (1998), and we observed a low proportion of leaf area damaged (<0.01) in this study.

Subgenus Azaleastrum (clade C, N=19 plants) had the highest overall level of damage with a mean proportion of 0.1408, and all samples were observed to have some level of stippling damage, which ranged from 0.0089 to 0.3451 (Table 1). The amount of leaf area damage was lowest in the subgenus Hymenanthes (clade B); samples from this subgenus (N=31 plants) had a mean proportion of leaf area damaged of 0.0080, and ranged from no damage to 0.0666 (Table 1). In addition, 20 of the species and cultivars from the subgenus Hymenanthes had no measurable damage. Samples from the subgenus Rhododendron (clade A, N=21) had intermediate levels of damage caused by S. pyrioides; in this group, the mean proportion of leaf area damaged was 0.0573, and ranged from 0 (which was recorded from seven species) to 0.2863 (Table 1). The compiled phylogenic tree of Rhododendron with the species matched to the corresponding proportion of leaf area damaged is presented in Figure 2, which highlights the variation in susceptibility among and within subgenera.

When comparing between lepidote (e.g., with trichomes) and elepidote (e.g., without trichomes) species, there were no statistically significant differences in the proportion of leaf damage due to S. pyrioides (F = 0.07, df = 1,69, P = 0.80, Fig. 3A). However, there were significant differences in the proportion of leaf damage by Rhododendron subgenera (F = 35.99, df = 2, 68, P < 0.01). Post hoc tests revealed that among the subgenera, all were significantly different from each other, with Azaleastrum being the most susceptible and Hymenanthes being the least susceptible to S. pyrioides (Fig. 3B). We also observed significant differences in the proportion of leaf damage by Rhododendron section (F = 19.39, df = 4,66, P < 0.01), with Tsutsusi (evergreen azaleas) being the most susceptible to S. pyrioides (Fig. 3C).

Rhododendron species can be subdivided through the presence or absence of trichomes (i.e., lepidote or elepidote, respectively), after which it can be subdivided by a hierarchy of clade/subgenus, section, subsection, and species (Chamberlain 1996, Goetch et al. 2005). When comparing between lepidote and elepidote species, there were no differences in the proportion of leaf area damaged (Fig. 3A). Thus, among the species and cultivars tested in this study, trichome structures on Rhododendron leaves do not predict susceptibility to S. pyrioides, and their presence did not deter S. pyrioides life stages from feeding on host plants. However, differences were observed among Rhododendron subgenera (Fig. 3B). The two subgenera that are lepidote, subgenus Azaleastum (clade C) and subgenus Hymenanthes (clade B), were observed to have the highest and lowest levels of damage by S. pyrioides, respectively, which reinforces that the presence of trichomes is not a predictor of susceptibility to S. pyrioides. However, there can be considerable variation in the architecture of Rhododendron trichomes (Cowan 1950), and although trichome presence might not be a predictor of susceptibility, future work that considers, for example, details of morphological structure in trichomes, and their possible production of chemical defenses against herbivores, might provide greater insight into the role that trichomes could play in conferring resistance to S. pyrioides.

Although we did observe some Rhododendron species and cultivars that are highly susceptible to S. pyrioides, it is not trivial to suggest that growers simply avoid planting susceptible ones (e.g., evergreen azaleas from the section Tsutsusi) and plant instead resistant species or varieties (e.g., plants from the section Ponicum) (Fig. 3C). Evergreen azaleas, which are highly susceptible to S. pyrioides, are also highly valued for their generally small stature and year-round foliar interest, and exhibit a range of color and form that may not be available in other subgenera (Kobayashi 2013). They are also culturally important in Japan, Korea, and China (Lee 1978), and they are a featured commodity of historic landscapes in western Washington, such as Azalea Way in the Washington Park Arboretum. Also, species and cultivars within the subgenus Azaleastrum were the most susceptible to S. pyrioides and all tested plants within this subgenus had some level of stippling damage; however, they are noted for their diversity of form and flower and a common choice in landscaping. The results from this research suggest that if susceptible species and cultivars are desired, a management program for S. pyrioides will be necessary. However, the identification of resistance species and cultivars, such as those within the subgenus Hymenanthes and the section Ponicum, also provide useful information for growers that might desire plants that require less management.

The identification of susceptible Rhododendron species and cultivars in botanical gardens can also be valuable for the preservation of Rhododendron that are threatened or endangered in their native range. Approximately 70% of Rhododendron species are classified as vulnerable, threatened, endangered or critically endangered, and about 25% of all Rhododendron taxa are under threat of extinction in the wild (Gibbs et al. 2011). Global trade is thought to have resulted in the introduction of S. pyrioides to many places outside of its native range (Nair and Braman 2012). Identification of susceptible species ex situ would allow susceptible populations to be monitored more intensively for S. pyrioides and to take measures to mitigate the introduction of S. pyrioides to areas in which Rhododendron species are endemic. Given the susceptibility of azaleas (subgenus Azaleastrum), and that S. pyrioides is already established in most areas of the United States where azaleas can be grown, careful consideration should be made when they are used. One possible avenue of future research is whether or not resistant species can be strategically planted among more susceptible species as a form of cultural control against S. pyrioides to reduce the overall damage in landscapes.