Environmental Entomology, 47(3), 2018, 646–653 doi: 10.1093/ee/nvy033 Advance Access Publication Date: 31 March 2018 – Interactions Research

Behavioral Evidence for Host Transitions in Plant, Plant Parasite, and Insect Interactions

Dale A. Halbritter,1,2,7 Denis S. Willett,3,4 Johnalyn M. Gordon,5 Lukasz L. Stelinski,3 and Jaret C. Daniels1,6

1Entomology and Nematology Department, University of Florida, 1881 Natural Area Dr, Gainesville, FL 32611, 2Present address: Invasive Plant Research Laboratory, USDA-ARS, 3225 College Avenue, Fort Lauderdale, FL 33314, 3Citrus Research and Education Center, University of Florida, 700 Experiment Station Road, Lake Alfred, FL 33850, 4Center for Medical, Agricultural and Veterinary Entomology, USDA-ARS, 1600 SW 23rd Dr, Gainesville, FL 32608, 5Fort Lauderdale Research and Education Center, University of Florida, 3205 College Avenue, Davie, FL 33314, 6McGuire Center for and Biodiversity, Florida Museum of Natural History, 3215 Hull Road, Gainesville, FL 32611, and 7Corresponding author, e-mail: [email protected]

Subject Editor: Jared Ali

Received 3 November 2017; Editorial decision 26 February 2018

Abstract Specialized herbivorous have the ability to transition between host plant taxa, and considering the co-evolutionary history between and the organisms utilizing them is important to understanding plant insect interactions. We investigated the role of a parasite, dwarf mistletoe ( spp.) M. Bieb. : Viscaceae, in mediating interactions between (Lepidoptera: ) and pine , the butterflies’ larval hosts. Mistletoe is considered the butterflies’ ancestral host, and the evolutionary transition to pine may have occurred recently. In , United States, we studied six sites in pine forest habitats: three in (Felder and R. Felder, 1859) habitat and three in Neophasia terlooii Behr, 1869 habitat. Each site contained six stands of trees that varied in mistletoe infection severity. behavior was observed and ranked at each stand. Volatile compounds were collected from trees at each site and analyzed using gas chromatography- mass spectroscopy. Female butterflies landed on or patrolled around pine trees (i.e., interacted) more than males, and N. terlooii interacted more with pine trees than N. menapia. Both butterfly species interacted more with tree stands harboring greater mistletoe infection, and N. terlooii interacted more with heavily infected tree stands than did N. menapia. The influence of mistletoe onNeophasia behavior may be mediated by differences in tree volatiles resulting from mistletoe infection. Volatile profiles significantly differed between infected and uninfected pine trees. The role of mistletoe in mediating butterfly interactions with has implications for conservation biology and forest management, and highlights the importance of understanding an organism’s niche in an evolutionary context.

Key words: butterfly, evolution, mistletoe, pine, volatile compound

Specialized herbivorous insects can transition to different host biological communities and silviculture (Stevens and Hawksworth plant taxa through rapid genetic adaptation (Singer et al. 1993) 1970, Drummond 1982, Conklin 2000, Hoffman 2004). The pon- or via more gradual evolutionary processes (Dobler et al. 1996). derosa pine ecosystem occupies a significant portion of western Understanding the co-evolutionary history between plants and North America, ranging from extreme southwestern Canada to insects is key to understanding contemporary plant–insect interac- central (Little 1971) and the pines host dwarf mistletoes tions, especially when considering environmental stressors. Plant– (Hawksworth and Wiens 1996). Here, we investigate a tri-trophic insect interactions are critical to ecosystem function and developing interaction and a hypothesized host switch involving an insect her- successful habitat management strategies (Raffa et al. 2008, Soler bivore, its contemporary host, and the mistletoe parasite in a et al. 2012). Other organisms can influence these interactions and ponderosa-pine dominated ecosystem. may contribute to host transitioning. For example, plant pathogens Pine butterflies in the genusNeophasia Behr, 1869 (Lepidoptera: can influence the attractiveness of the plants to herbivorous insects Pieridae) belong to the subtribe Aporiina and, in the New World, (McLeod et al. 2005, Mauck et al. 2010, Mann et al. 2012). Dwarf members of Aporiina are primarily mistletoe (Santalales) feeders and mistletoes, Arceuthobium spp. (M. Bieb. Santalales: Viscaceae), are are restricted to South America (Braby and Nishida 2010). However, flowering plants and parasites of , and affect associated Neophasia and Eucheira Westwood, 1834 are two North American

© The Author(s) 2018. Published by Oxford University Press on behalf of Entomological Society of America. All rights reserved. For permissions, please e-mail: [email protected]. 646

Downloaded from https://academic.oup.com/ee/article-abstract/47/3/646/4957024 by University of Florida user on 07 June 2018 Environmental Entomology, 2018, Vol. 47, No. 3 647

aporiine genera that feed on conifers and madrone (Arbutus) (L. southern range limit of N. menapia and the northern range limit Ericales: Ericaceae) trees, respectively. This break in phylogenetic of N. terlooii (Fig. 1). We focused on habitats for N. menapia and conservatism via the exploitation of new larval host plants may N. terlooii in Arizona at the southern and northern range boundaries have facilitated the northward expansion of these two genera, along of each butterfly species, respectively. with adaptation to colder climates. When combined, the geographic From July through mid-August, adult N. menapia spend sunny ranges of Neophasia menapia (Felder and R. Felder, 1859) and days patrolling pine trees in search of mates and trees suitable for ovi- Neophasia terlooii Behr, 1869 span 30 degrees of latitude in western position. In northern Arizona, we observed N. menapia ovipositing North America, which overlap considerably with the ponderosa pine on Rocky Mountain ponderosa pine (ponderosa pine, henceforth), ecosystem (for butterfly distributions, see: Scott 1986, Bailowitz and ssp. scopulorum (Engelmann) (: ), Brock 1991, Lotts and Naberhaus 2015). Central Arizona marks the but a considerably greater diversity of conifers is utilized in the rest of

Fig. 1. Geographic distribution of the six sites in Arizona, each containing six tree stands. The blue shaded area indicates the distribution of Neophasia menapia at the county level. The red shaded area indicates the distribution of N. terlooii at the county level. The area in purple indicates the county where both species can be found, with the river marking the dividing line. The inset map shows the entire range of N. menapia in blue and that of N. terlooii in red. Arizona is purple because it contains both butterfly species. With the exception of a small area in SW , inset distribution data are at the state or province scale. Distribution data were obtained from Lotts and Naberhaus (2015). NJ = North Jacob Lake, SC = Schultz Pass, MO = Mormon Lake, BF = Barfoot Park, SW = Sawmill Canyon, and CR = Carr Canyon.

Downloaded from https://academic.oup.com/ee/article-abstract/47/3/646/4957024 by University of Florida user on 07 June 2018 648 Environmental Entomology, 2018, Vol. 47, No. 3

its range (see Discussion). N. terlooii has two flights per year, one in lengthwise through the center of each site. Each stand consisted of late spring/early summer and the primary flight, which is in October. two to six trees, where the two most distant trees in a stand were at Like its congener, N. terlooii patrols pine trees in search of mates and most 15 m apart from each other. Each stand was set apart from the trees suitable for oviposition. In the Arizona sky islands, we observed surrounding forest such that the tree crowns within the stands were N. terlooii ovipositing on Apache pine, Pinus englemannii Carrière, not touching the tree crowns of the surrounding forest matrix. The Arizona pine, P. ponderosa ssp. arizonica (Engelmann), ponderosa tree crowns within each stand were in contact. pine, and southwestern white pine, Pinus strobiformis Engelmann. Each site contained three stands that had visible dwarf mistletoe Southwestern dwarf mistletoe, Arceuthobium vaginatum subsp. infection and three stands with little to no visible infection. Stands eryptopodium (Engelmann) occurs throughout Arizona and para- were scored for infection on a scale from 0 to 6, with 0 being unin- sitizes ponderosa pine, Apache pine, and Arizona pine (Olsen 2003). fected and 6 being heavily infected. Scoring methods followed those Dwarf mistletoes induce a stress response that causes trees to alter of Hawksworth (1977), where the live crown of each tree in a stand their biochemistry (Nebeker et al. 1995). Stressed trees produce was divided into thirds, each third was visually scored as either 0 volatile chemicals such as monoterpenes, which have been shown (no visible infections), 1 (50% or less of the branches infected), or 2 to attract wood-boring insects (Costello et al. 2008). Four primary (greater than 50% of branches infected), and the sum of the thirds chemical isolates are emitted by infected Pinus contorta (Douglas was the score for the tree. We summed the scores of each tree in a ex Loudon) trees: α-pinene, β-pinene, α-terpineol, and tricyclene stand and took the average as the score for the stand. (Nebeker et al. 1995). Butterflies have odor receptors on their To investigate the potential for dwarf mistletoe infection in the antennae (Hansson 1995, Mercader et al. 2008), but the ability of surrounding forest to influence butterfly abundance at each of the Neophasia to detect airborne chemical cues has never been tested, six stands at each of the N. menapia sites, all ponderosa pine trees nor have the airborne volatiles been described for the specific pine within a 25-m radius of the centroid of each stand were scored for hosts of Neophasia in Arizona. dwarf mistletoe infection. The 25-m-radius circle was divided into We explored the role of a plant parasite in influencing plant– eight slices, starting with the slice containing trees between 0° (N) insect interactions, specifically how dwarf mistletoe mediates the and 45° (NE). Slice perimeters were marked with irrigation flags and interactions between butterflies and dwarf mistletoe-infected and trees were marked with chalk after they had been scored. This was uninfected pine trees. We hypothesized that the presence of para- repeated in a clockwise fashion for the remaining seven slices. The sitism and its severity would alter the volatile compound signatures scores of all trees in a 25-m radius circle were averaged. The aver- of pine trees. If this was the case, we expected differential behav- age infection scores were compared to butterfly abundance at each ior between butterflies flying among infected and uninfected trees. tree stand, i.e., the total number of butterflies seen per stand during Butterflies were predicted to interact more with infected trees and each visit. be more abundant in areas with greater dwarf mistletoe presence, which, in an evolutionary context, could reflect their origin from a Butterfly Abundance and Behavior mistletoe-feeding ancestor and subsequent switch to pine. N. menapia were observed from 22 July 2015 through 9 August 2015, and N. terlooii were observed from 18 October 2015 through Materials and Methods 28 October 2015. Each site was visited between one and five times during a field season between 9:00 a.m. and 3:00 p.m. Butterfly Field Sites and Tree Stands observations at each of the six tree stands at each site were com- Three field sites per Neophasia spp. were selected to observe the pleted in a random order, and each observation lasted for 30 min interactions between the butterflies and stands of their larval host with at least 15 min of full sun. If there was not at least 15 min of trees harboring various levels of dwarf mistletoe infection. For full sun, the stand was re-visited after the others were completed, N. menapia, one site on the Kaibab Plateau (South Jacob Lake) and time permitting (i.e., the last 30-min observation had to start by two sites on the Mogollon Rim (Schultz Pass and Mormon Lake) 2:30 pm). (Fig. 1) were visited. For N. terlooii, one site in the Chiricahua Butterfly behavior was ranked in a qualitative manner, from weak Mountains (Barfoot Park) and two sites in the Huachuca Mountains to strong interactions. If a butterfly passed by a stand and did not (Sawmill and Carr Canyons) (Fig. 1) were visited. Sites comprised significantly alter its flight direction, it was classified as ‘flyby’ (weak- roughly rectangular patches of pine forest, ranging from, 335- to est interaction). If a butterfly encountered a stand and patrolled the 715-m long and 130- to170-m wide. The latitude and longitude for canopy by circling the treetops, zigzagging up and down the canopy, each site are listed in Table 1. or both, it was classified as ‘patrol’. If a butterfly landed on a tree Within each site, six stands of trees were selected such that branch, it was classified as ‘land’ (strongest interaction for males. If each stand was at least 50 m from another stand running roughly a butterfly oviposited on a pine needle, it was classified as ‘oviposit’

Table 1. Locations of the field sites for the behavioral observations and volatile samples taken from pine trees

Habitat for Site Latitude Longitude

Neophasia menapia North Jacob Lake 36°42′14.88″N 112°15′49.08″W Schultz Pass 35°15′15.95″N 111°40′5.52″W Mormon Lake 34°59′22.03″N 111°30′28.32″W Neophasia terlooii Barfoot Park 31°54′59.04″N 109°16′48.66″W Sawmill Canyon 31°26′41.86″N 110°22′8.76″W Carr Canyon 31°25′45.17″N 110°18′14.84″W

GPS coordinates indicate the approximate center of each site.

Downloaded from https://academic.oup.com/ee/article-abstract/47/3/646/4957024 by University of Florida user on 07 June 2018 Environmental Entomology, 2018, Vol. 47, No. 3 649

(strongest interaction for females). Sex was determined by wing than Poisson models based on comparisons of log-likelihood val- color: males of both species are white with black markings, female ues, goodness of fit tests, and visual examination of the conditional N. terlooii are orange with black markings, and female N. menapia mean-variance relationship. All factors (i.e., mean dwarf mistletoe are pale yellow with heavier black markings than males, and have infection severity of trees surrounding a stand, number of trees sur- bright red dots along their outer hind wing margins. rounding a stand, date of observation, site, stand, and percentage of surrounding trees infected) as well as two- and three-level inter- Collection and Analysis of Tree Volatiles actions were evaluated for inclusion in the models. Model factors Airborne volatile compounds were collected from infected and unin- were selected for inclusion based on analysis of deviance, chi-square 2 fected trees within the habitats of both species of Neophasia. Each tests of log-likelihood values, pseudo-R values, goodness of fit, and infected tree had two samples taken from it: one branch that had examination of residual diagnostic plots. dwarf mistletoe on it and one that did not have dwarf mistletoe Internal factors (i.e., stand conditions) influencingN. menapia on it. Branches from ponderosa pines were sampled from sites in and N. terlooii behavior at each tree stand were examined using N. menapia habitat, and ponderosa pine, Apache pine, and Arizona cumulative link models (sometimes referred to as ordinal regression) pine branches were sampled from sites in N. terlooii habitat. Volatile with a logit link function. All factors (i.e., butterfly species, sex, and samples were collected by placing a living branch roughly 45-cm tree stand dwarf mistletoe infection severity) as well as two-factor long into a 48.2-cm × 59.6-cm oven bag (Reynold’s Oven Bags, interactions were evaluated for inclusion in the models. Model fac- Manufacturer # 1001090000510), which had been pre-baked for tors were selected for inclusion based on analysis of deviance, chi- 3.5 h at 121°C to remove unwanted volatiles. Bag openings were square tests of log-likelihood values, goodness of fit and examination gathered together and sealed around the basal stem of the branch of residual diagnostic plots. Tukey’s HSD tests were used to make with a plastic zip tie. Care was taken to ensure the bags were shel- post-hoc comparisons. tered from direct sunlight by suspending black plastic sheets in To investigate differences in tree volatile profiles potentially influ- branches above the bags to provide shade. One control bag of air enced by dwarf mistletoe infection, variable importance measures from each site was collected. There was a total of 25 volatile samples from random forests models were first used to identify compounds collected between the two habitats. of interest. Differences in abundances of those compounds corre- Branches were bagged at the start of a site visit for the behavioral sponding to dwarf mistletoe infection status were then modeled observations and branches were then cut off the tree in the after- using multivariate analysis of variance (MANOVA) after assur- noon, leaving about 15 cm of stem outside the bag. Approximately ing adherence to assumptions of normality and homoscedasticity. 12 h after a branch was bagged, the bag was affixed to a 1.5-gallon Canonical discriminant analysis was then used to identify relative shop vacuum (Stanley, Part # SL18125P1) calibrated using a rhe- influences of each compound. ostat (GE, Model # 18019) that was used to draw air out of the Data were collated in Microsoft Excel and analyzed using R ver- bag through an adsorbent 30-mg HayeSep Q filter (Volatile Assay sion 3.3.1 in the RStudio version 0.99.902 development environment Systems, VAS) at 160 ml/min for 1 h. Clean filtered air was allowed (R Core Team 2015). The following packages facilitated analysis: into the bag through a glass pipette filled with activated charcoal and dplyr (Wickham and Francois 2015) and tidy (Wickham 2014) for attached to an opposite corner of the bag. This allowed air to con- data formatting, xlsx (Dragulescu 2014) for the R-Excel interface, tinue flowing over the pine samples while minimizing external con- car (Fox and Weisberg 2011) and lsmeans (Lenth 2016) for regres- taminants. Filters containing volatiles were stored in airtight glass sion analysis, ordinal (Christensen 2015) for cumulative link models, vials and kept in a freezer (−15°C), later shipped on ice, and stored candisc (Friendly and Fox 2016) for canonical discriminant analysis, in a freezer until analysis. MASS (Venables and Ripley 2002) for negative binomial regression, Gas chromatography-mass spectroscopy (GC-MS) conducted randomForest (Liaw and Wiener 2002) for random forests models, at the Citrus Research and Education Center, University of Florida, and ggplot2 (Wickham 2009) and ggmap (Kahle and Wickham 2013) Lake Alfred, FL was used to identify the volatile compounds in our for graphics. Raw data and R scripts are available upon request. samples. Each sample was eluted from its filter into a 200-µl glass vial insert with two 75-µl rinses of methylene chloride. Following elution, 5 μl of 1.5 ng/μl nonyl acetate were added as an internal standard. Results One-μl aliquots of each sample were then run on a 30-m × 0.25- Butterfly Abundance mm-ID DB-5 capillary column in a Clarus 500 GC-MS (PerkinElmer, Greater numbers of trees surrounding each stand corresponded Waltham, MA). The column was held at 35°C for 3 min after injec- to greater numbers of N. menapia seen at each stand (χ2 = 14.1; tion and then increased 10°C/min until reaching 260°C, where it df = 1; P < 0.001) (Fig. 2). The number of trees surrounding each remained for an additional 5 min. Helium was used as a carrier gas stand ranged from 37 to 335, with the highest numbers indicating at a flow rate of 2 ml/min. Electron ionization spectra were compared many small saplings and lower numbers indicating fewer, but larger, with references found in the National Institute of Standards and mature trees. Holding the sampling date constant, for every unit Technology database and then confirmed with available standards. increase in the number of trees surrounding each stand, there was a Retention times, peak heights, peak areas, and the start and end times 0.6 % (95% CI: 0.3%, 0.9%) increase in numbers of butterflies seen for each peak were recorded and downloaded for analysis. at each stand. Butterfly abundance was affected by the observation date, with greatest numbers seen at the beginning and end of the Statistical Analyses summer sampling period (χ2 = 7.55; df = 1; P = 0.06). Dwarf mistle- External factors (i.e., surrounding forest condition) influencing toe infection of the trees surrounding each stand did not influence N. menapia abundance (i.e., total number of butterflies observed at butterfly abundance at each stand χ( 2 = 1.04; df = 1; P = 0.31). The a stand) were investigated with negative binomial regression using model incorporating these factors was better than the null model 2 generalized linear models with a log link function to account for (LR3 = 30.2; P < 0.0001), did not demonstrate a lack of fit χ( = 34.9; overdispersion in the abundance data. These models were better fits df = 30; P = 0.25), and explained 58.1% of the observed variance.

Downloaded from https://academic.oup.com/ee/article-abstract/47/3/646/4957024 by University of Florida user on 07 June 2018 650 Environmental Entomology, 2018, Vol. 47, No. 3

Butterfly Behavior hosts and conifer parasites like dwarf mistletoe as larval hosts may Butterfly species, average dwarf mistletoe infection severity of each be fairly common in Lepidoptera because the mistletoes are grow- tree stand, and the interaction between butterfly sex and dwarf ing on the trees very close to the pine needles on which caterpil- mistletoe infection severity influenced butterfly behaviorTable 2 ( ). lars feed (Mooney 2003). Braby and Trueman (2006) hypothesized Females interacted significantly more with trees than did males that this close tree-to-parasite proximity could be responsible for (P < 0.001; Fig. 3). Interestingly, N. terlooii interacted more with the numerous shifts between caterpillar host preference of trees and trees than did N. menapia (P = 0.0002) (Fig. 3). An increase in dwarf their associated mistletoes observed in Aporiina. Our behavioral evi- mistletoe infection severity increased butterfly interactions with trees dence suggests that Neophasia has an affinity for pine trees infected for both males (P = 0.012) and females (P = 0.048) (Fig. 4). with dwarf mistletoe. It is possible that some element of mistletoe host-seeking behavior has been retained in extant Neophasia as an Volatiles evolutionary relic. N. terlooii showed a stronger preference for infected trees than Variable importance measures from random forests models identified N. menapia, suggesting that N. terlooii may have retained more of seven primary volatile compounds that differed based on dwarf mis- its hypothesized ancestral traits. From a biogeographic context, the tletoe infection status (F = 3.73; df = 21,42; P < 0.0001). Volatile pro- distribution of N. terlooii is closer geographically to the distribu- files of branches from infected trees with dwarf mistletoe present on tions of remaining New World Aporiina, which are predominantly the branches were different from the profiles of branches from unin- Neotropical (Braby et al. 2007), while the distribution of N. menapia fected trees or uninfected branches from infected trees (t = −6.765; 1 has shifted farther north. A host plant shift away from mistle- df = 18; P < 0.0001) (Fig. 5a). Volatile profiles of branches of infected toe opens a new niche for Neophasia, as conifers are more abun- trees without dwarf mistletoe present on the branches were margin- dant than mistletoes, in terms of biomass, as larval food sources. ally different from those of branches from uninfected trees (t = −2.38; N. menapia is fairly polyphagous within the Pinaceae, feeding on df = 18; P = 0.05) (Fig. 5a). Germacrene-D and δ-cadinene were the Pinus, Pseudotsuga Carrière, Abies Mill., Tsuga Carrière, and Picea primary contributors to the volatile index that provided resolution to Mill. (Evenden 1926, Cole 1971, Scott 1986, Robinson et al. 2002). detect differences in infection status (Fig. 5b). Less is known about N. terlooii, but they may be more limited in host plant options and have thus far been documented on only two Discussion genera: Pinus and Picea (Arizona Game and Fish Department 2001). Dwarf mistletoes can be found on all aforementioned tree genera We are potentially witnessing an evolutionary transition of host (Hawksworth and Wiens 1996). Assuming the latter host-parasite plant utilization in Neophasia. Shifts between conifers as larval relationships have persisted, and that ancestral Neophasia special- ized on a more limited number of dwarf mistletoe species, this could explain why N. terlooii is more restricted in its pine hosts. N. terlooii has a greater affinity for pines that are parasitized by specific dwarf mistletoe species, while N. menapia is more likely to utilize other pines over a greater geographic range that are parasitized by more species of dwarf mistletoe. Dwarf mistletoe infection has a significant effect on Neophasia behavior. Neophasia interacted more with trees that were heavily infected. Male behavior predominantly consisted of patrolling, which is typical mate-seeking behavior (Scott 1986, Lotts and Naberhaus

Fig. 2. Tree and butterfly abundance were positively correlated. The number of butterflies indicates the total number of butterflies observed at each tree stand. The line and shaded area indicate the model fit and 95% confidence intervals, respectively. Results were averaged across sampling date at each stand.

Table 2. Analysis of deviance for models of Neophasia behavior

Source χ2 df P

Speciesa 21.0 1 <0.0001 Sex 0.3 1 0.567 Hawksworthb 13.5 1 0.0002 Fig. 3. behavior with respect to sex and species. Flyby, Patrol, Sex:Hawksworth 14.0 1 0.0002 Neophasia Landing, and Oviposition indicate potential interactions with host trees Total 56.5 7 <0.0001 irrespective of infection status. A Flyby is a weak interaction while Oviposition Residual 1.1 3 0.777 is a strong interaction. Points and error bars denote mean behavior and 95% confidence intervals from cumulative link models ofNeophasia behavior, a Species are N. menapia and N. terlooii. respectively. Letters A, B, C, and D indicate significant differences (P < 0.05) bHawksworth is the index of mistletoe infection severity. based on Tukey’s HSD test.

Downloaded from https://academic.oup.com/ee/article-abstract/47/3/646/4957024 by University of Florida user on 07 June 2018 Environmental Entomology, 2018, Vol. 47, No. 3 651

Fig. 4. Effect of dwarf mistletoe infection severity on Neophasia behavior. The Hawksworth Score ranges from 0 (uninfected) to 6 (heavily infected). A Flyby is a weak interaction while Oviposition is a strong interaction. Lines and shaded areas denote mean behavior and 95% confidence intervals from cumulative link models of Neophasia behavior, respectively.

number of trees surrounding the stands rather than infection severity. There is evidence to suggest N. menapia on the Kaibab Plateau formed a population distinct from conspecifics found on the Mogollon Rim (D. A. Halbritter, unpublished data), suggesting the latter two regions would comprise the geographic scale necessary to make inferences of how forest structure affects population dynamics. While dwarf mis- tletoe infection may influence butterfly behavior within a tree stand, overall abundances of butterflies seem to be affected primarily by differences in other attributes of available host resources (i.e., the number and size of trees in the area). However, replicated popula- tion-scale surveys of dwarf mistletoe infection, tree sizes and abun- dance, and butterfly population size estimates are needed before any conclusions can be drawn about how spatial characteristics of forests and dwarf mistletoe infections affect Neophasia populations. The success of Neophasia larvae is at least partially dependent on the nutritional content of the pine needles onto which they were placed as ova. In a related species of tree, Pinus contorta, infected trees were found to have significantly lower starch, total nitrogen, and free amino-nitrogen composition in their phloem (Nebeker et al. 1995). Phloem samples in the latter study were taken from tree trunks at breast height, but dwarf mistletoe infections can have more localized effects, causing infected branches to become nutrient sinks Fig. 5. The dwarf mistletoe infection status of a tree influences its volatile chemical profile. (a) Differences in volatile profiles based on infection at the expense of the tree as a whole (Hawksworth and Wiens 1996). status. Black points and error bars denote mean and 95% bootstrapped Eventually tree health declines and the tree can die from an infection. confidence intervals, respectively. Grey points denote observed values. (b) Butterflies in our study seemed to randomly explore the forest matrix, Relative weighting of each compound used to construct the volatile index. but then spent more time searching the canopies of specific trees. The magnitude of arrows denotes relative weighting while direction denotes Within close proximity to those trees, butterflies may select branches the effect of contributions (i.e., Germecrene-D had higher levels in volatile profiles from uninfected branches, while d-longifolene had higher levels in on infected trees that have higher nutritional content, which may be profiles from infected branches). reflected in the different volatile profiles detected at close range. In addition to signaling nutritional quality, tree volatiles released 2015). Despite the low abundance of females and greater variabil- as a result of mistletoe infection and other stressors may be cues to ity compared to those of males, females still interacted significantly Neophasia indicating compromised tree defenses. We did not docu- more with infected trees, tending to land and oviposit on them more ment feeding damage from other insects or infections from other often as compared with uninfected counterparts. pathogens, the latter of which are also known to alter terpene blends Although butterfly behavior was affected by dwarf mistletoe in pines (Nebeker et al. 1995). Pathogen-infected host plants are infection in the tree stands from where the butterflies were observed, often more attractive to the insect vectors of these pathogens than butterfly abundance at these stands was influenced more by the uninfected counterparts (Mauck et al. 2010, Mann et al. 2012). For

Downloaded from https://academic.oup.com/ee/article-abstract/47/3/646/4957024 by University of Florida user on 07 June 2018 652 Environmental Entomology, 2018, Vol. 47, No. 3

example, the bark beetle, Hylurgopinus rufipes (Eichhoff 1868), is References Cited preferentially attracted to elm trees infected with the Dutch elm dis- Allen, C. D., and D. D. Breshears. 1998. Drought-induced shift of a for- ease pathogen (Ophiostoma novo-ulmi) compared with uninfected est-woodland ecotone: rapid landscape response to climate variation. elms (McLeod et al. 2005). Although not vectors of mistletoe, ances- Proc. Natl. Acad. Sci. USA 95: 14839–14842. tral Neophasia spp. likely fed on mistletoe and therefore would have Arizona Game and Fish Department. 2001. Neophasia terlootii. Unpublished had a direct dependence on it and would later retain an association abstract compiled and edited by the Heritage Data Management System, with infected trees, potentially responding to kairomones in the pine Arizona Game and Fish Department, Phoenix, AZ. forest matrix that signal compromised trees. Bailowitz, R. A., and J. P. Brock. 1991. Butterflies of Southeastern Arizona. Pine branches infected with mistletoe were characterized by an Sonoran Studies, Inc, Tucson, AZ. overall greater quantitative release of volatiles, and also qualita- Braby, M. F., and K. Nishida. 2010. The immature stages, larval food plants and biology of Neotropical mistletoe butterflies (Lepidoptera: Pieridae). tively released a blend of volatiles characterized by specific terpenes, II. The group (: Aporiina). J. Nat. Hist. 44: 1831–1928. such as germacrene-D and δ-cadinene, as compared with uninfected Braby, M. F., N. E. Pierce, and R. Vila. 2007. Phylogeny and historical biogeog- branches. The mistletoe itself may have contributed to the volatile raphy of the subtribe Aporiina (Lepidoptera: Pieridae): implications for profile of branches with mistletoe on them as well. Additionally, indi- the origin of Australian butterflies. Biol. J. Linn. Soc. 90: 413–440. vidual volatile compounds can be emitted by non-host tree species, Braby, M. F., and J. W. H. Trueman. 2006. Evolution of larval host plant asso- thereby adding to the complex milieu of forest volatiles. For these ciations and adaptive radiation in pierid butterflies. J. Evolution. Bio. 19: reasons, host recognition is likely dependent on a particular blend 1677–1690. of volatiles (Witzgall et al. 2005). Additional behavioral assays with Christensen, R. H. B. 2015. Ordinal—regression models for ordinal butterflies in the laboratory and field will be necessary to determine data. R package version 2015. 6–28. http://www.cran.r-project.org/ the specific volatile signals that influence host-seeking inNeophasia . package=ordinal/ As a result of mistletoe infection, it may be an overall quantitative Ciesla, W. 1974. Forest insect damage from high-altitude color-IR photos. Am. Soc. Photogramm. 40: 683–689. increase in volatile production, a specific blend of volatiles, or both, Cole, W. E. 1971. Pine butterfly. USDA For. Serv. For. Pest Leaflet 66: 3 pp. that influence host selection.Neophasia may be drawn to volatiles Conklin, D. A. 2000. Dwarf mistletoe management and forest health in the from the mistletoe itself, or the butterflies may be attracted to trees Southwest. US Forest Service, Albuquerque, NM. afflicted with other diseases or insect damage. Costello, S. L., J. F. Negrón, and W. R. Jacobi. 2008. Traps and attractants for wood-boring insects in ponderosa pine stands in the Black Hills, South Conclusions Dakota. J. Econ. Entomol. 101: 409–420. Dwarf mistletoe and Neophasia butterflies are two conspicuous Dobler, S., P. Mardulyn, J. M. Pasteels, and M. RowellRahier. 1996. Host-plant members of the ponderosa pine community that utilize the pine switches and the evolution of chemical defense and life history in the trees as hosts and have been shown here to interact with each beetle Oreina. Evolution. 50: 2373–2386. other, comprising a multitrophic ecological interaction. Pine Dragulescu, A. A. 2014. xlsx: read, write, format Excel 2007 and Excel forests in the western United States are facing threats from fires 97/2000/XP/2003 files. R package version 0.5.7. http://CRAN.R-project. org/package=xlsx exacerbated by climate change (Allen and Breshears 1998), insect Drummond, D. B. 1982. Timber loss estimates for the coniferous forests of damage (Kenaley et al. 2006), synergistic effects of fire and insects the United States due to dwarf mistletoes. Fort Collins, CO: USDA Forest (McHugh et al. 2003), and from interactions between insects and Service, Forest Pest Management, Methods Application Group Rpt. tree parasites (Wagner and Mathiasen 1985). Neophasia are gen- No.83-2. erally not detrimental to pine forests, but there have been occa- Evenden, J. C. 1926. The pine butterfly, Neophasia menapia Felder. J. Agr. Res. sional, localized population irruptions that result in defoliation 33: 339–344. (Ciesla 1974, Young 1986). Neophasia are abundant and easily Fox, J., and S. Weisberg. 2011. An {R} companion to applied regression, detectable as adults, making them effective representatives of the Second Edition. Thousand Oaks, CA: Sage. http://socserv.socsci.mcmaster. ponderosa pine community. Changes in Neophasia population ca/jfox/Books/Companion dynamics could reflect changes in dwarf mistletoe populations Friendly, M., and J. Fox. 2016. candisc: visualizing generalized canonical dis- criminant and canonical correlation analysis. R package version 0.7-0. and consequently changes in forest health. Presence of butter- https://CRAN.R-project.org/package=candisc flies may be indicative of new or ongoing mistletoe infestations. Hansson, B. S. 1995. Olfaction in Lepidoptera. Experientia (Basel). 51: Understanding the ecology of Neophasia in a community context 1003–1027. will therefore be important to inform forest management and con- Hawksworth, F. G. 1977. General Technical Report RM-48. U.S. Dept. of servation practices. Studies such as ours, which integrate under- Agriculture, Forest Service, Rocky Mountain Forest and Range Experiment standing of evolutionary histories and field studies on multitrophic Station, Fort Collins, CO. ecological interactions, will be important for the improvement of Hawksworth, F. G., and D. Wiens. 1996. Dwarf mistletoes: biology, pathology, conservation science. and systematics. U.S. Dept. of Agriculture, Forest Service, Washington, DC. Hoffman, J. T. 2004. Management guide for dwarf mistletoe Arceuthobium spp., USDA For. Serv. 14 pp. https://www.fs.usda.gov/Internet/FSE_ Acknowledgments DOCUMENTS/stelprdb5187427.pdf. We thank Alan Yanahan, the Merriam-Powell Research Station, and the Kahle, D., and H. Wickham. 2013. ggmap: spatial visualization with ggplot2. University of Arizona Entomology Department for providing lodging and lab- R J. 5: 144–161. oratory space in Arizona, and Matthew Standridge for handling volatile sam- Kenaley, S. N. C., R. L. Mathiasen, and C. M. Daugherty. 2006. Selection ple shipments. We extend our gratitude to Fort Huachuca and the National of dwarf mistletoe-infected ponderosa pines by Ips species (Coleoptera: Forest Service and for their collaboration. Funds from the William C. and Scolytidae) in northern Arizona. West. N. Am. Naturalist. 66: 279–284. Bertha M. Cornett Fellowship, the Florida Museum of Natural History Travel Lenth, R. V. 2016. Least-squares means: the R package lsmeans. J. Stat. Softw. Award, and an Entomology and Nematology Department Scholarship, the lat- 69: 1–33. ter awarded to author J.M.G., through the University of Florida were used to Liaw, A., and M. Wiener. 2002. Classification and regression by randomForest. cover travel and lodging expenses. R News 2: 18–22.

Downloaded from https://academic.oup.com/ee/article-abstract/47/3/646/4957024 by University of Florida user on 07 June 2018 Environmental Entomology, 2018, Vol. 47, No. 3 653

Little, E. L. Jr. 1971. Atlas of United States trees, volume 1: conifers and Raffa, K. F., B. H. Aukema, B. J. Bentz, A. L. Carroll, J. A. Hicke, M. G. Turner, important hardwoods. Misc. Publ. 1146, U.S. Department of Agriculture, and W. H. Romme. 2008. Cross-scale drivers of natural disturbances Washington, DC. prone to anthropogenic amplification: the dynamics of bark beetle erup- Lotts, K., and T. Naberhaus. 2015. Butterflies and Moths of North America. tions. Bioscience. 58: 501–517. http://www.butterfliesandmoths.org/ Robinson, G. S., P. R. Ackery, I. J. Kitching, G. W. Beccaloni, and L. M. Hernández. Mann, R. S., J. G. Ali, S. L. Hermann, S. Tiwari, K. S. Pelz-Stelinski, 2002. Hostplants of the moth and butterfly caterpillars of America north of H. T. Alborn, and L. L. Stelinski. 2012. Induced release of a plant-defense Mexico. The American Entomological Institute, Gainsville, FL. volatile ‘deceptively’ attracts insect vectors to plants infected with a bac- Scott, J. A. 1986. The butterflies of North America: a natural history and field terial pathogen. PLoS Pathog. 8: e1002610. guide. Stanford University Press, Palo Alto, CA. Mauck, K. E., C. M. De Moraes, and M. C. Mescher. 2010. Deceptive chem- Singer, M. C., C. D. Thomas, and C. Parmesan. 1993. Rapid human-induced ical signals induced by a plant virus attract insect vectors to inferior hosts. evolution of insect host associations. Nature. 366:681–683. PNAS USA. 107: 3600–3605. Soler, R., W. H. Van der Putten, J. A. Harvey, L. E. M. Vet, M. Dicke, and McHugh, C. W., T. E. Kolb, and J. L. Wilson. 2003. Bark beetle attacks on T. M. Bezemer. 2012. Root herbivore effects on aboveground multi- ponderosa pine following fire in northern Arizona. Environ. Entomol. 32: trophic interactions: patterns, processes and mechanisms. J. Chem. Ecol. 510–522. 38: 755–767. McLeod, G., R. Gries, S. H. von Reuss, J. E. Rahe, R. McIntosh, W. A. König, Stevens, R. E., and F. G. Hawksworth. 1970. Insects and mites associated with and G. Gries. 2005. The pathogen causing Dutch elm disease makes host dwarf mistletoes, pp. 1–12. In Research Paper RM-59, USDA For. Serv., trees attract insect vectors. Proc. Biol. Sci. 272: 2499–2503. Rocky Mountain Forest & Range Experiment Station, Fort Collins, CO. Mercader, R. J., L. L. Stelinski, and J. M. Scriber. 2008. Differential antennal Venables, W. N., and B. D. Ripley. 2002. Modern applied statistics with S, sensitivities of the generalist butterflies Papilio glaucus and P. canadensis Fourth Edition. Springer, New York. to host plant and non-host plant extracts. J. Lepid. Soc. 62: 84–88. Wagner, M. R., and R. L. Mathiasen. 1985. Dwarf mistletoe Arceuthobium Mooney, K. A. 2003. Promylea lunigerella glendella Dyar (Pyralidae) feeds vaginatum ssp erytopodium and Pandora moth Coloradia pandora pan- on both conifers and parasitic dwarf mistletoe (Arceuthobium spp.): one dora interaction and its contribution to ponderosa pine Pinus ponderosa example of food plant shifting between parasitic plants and their hosts. J. mortality in Arizona USA. Great Basin Nat. 45: 423–426. Lepid. Soc. 57: 47–53. Wickham, H. 2009. ggplot2: elegant graphics for data analysis. Springer- Nebeker, T. E., R. F. Schmitz, R. A. Tisdale, and K. R. Hobson. 1995. Chemical Verlag, New York. and nutritional-status of dwarf mistletoe, Armillaria root-rot, and Wickham, H. 2014. Tidy data. J. Stat. Softw. 59: 23 pp. Comandra blister rust infected trees which may influence tree susceptibil- Wickham, H., and R. Francois. 2015. dplyr: a grammar of data manipulation. ity to bark beetle attack. Can. J. Bot. 73: 360–369. R package version 0.4.3. http://CRAN.R-project.org/package=dplyr Olsen, M. W. 2003. AZ1309, Arizona Cooperative Extension. http://exten- Witzgall, P., L. Ansebo, Z. H. Yang, G. Angeli, B. Sauphanor, and M. Bengtsson. sion.arizona.edu/sites/extension.arizona.edu/files/pubs/az1309.pdf 2005. Plant volatiles affect oviposition by codling moths. Chemoecology. R Core Team. 2015. R: a language and environment for statistical comput- 15: 77–83. ing. R Foundation for Statistical Computing, Vienna, Austria. https:// Young, R. M. 1986. Mass emergences of the pine white, Neophasia menapia www.R-project.org/ menapia (Felder & Felder), in (Pieridae). J. Lepid. Soc. 40: 314.

Downloaded from https://academic.oup.com/ee/article-abstract/47/3/646/4957024 by University of Florida user on 07 June 2018