KProceedingsoa looper fitness of the Hona wleavesaiian E andntomologic phyllodesal Sofociety Acaci (2013)a koa 45:141–147 141

Koa Looper Caterpillars ( paludicola, Geometridae) Have Lower Fitness on Koa ( koa, ) True than on Phyllodes

Kasey E. Barton1,3 and William P. Haines2 1Department of Botany, University of at Manoa, 3190 Maile Way, St. John 101, Honolulu, HI 96822; 2Department of and Environmental Protection Sciences, University of Hawaii at Manoa, 3050 Maile Way, Gilmore 310, Honolulu, HI 96822 3Corresponding author: [email protected]

Abstract. Native plant-herbivore interactions on islands remain understudied due to the widely discussed idea that island have weaker defenses than their con- tinental relatives. In Hawaii, the native , a specialist on the native , can undergo outbreaks that defoliate tens of thousands of acres of native koa , sometimes leading to massive stand mortality. Such extreme herbivory events are expected to exert strong selection pressure for defense in A. koa. Because mature A. koa often re-flush juvenile true leaves after defoliation, we predict that true leaves are better defended against S. paludicola than phyllodes, consistent with the phenomenon of induced resistance. A no-choice bioassay was conducted in the laboratory to compare S. paludicola development on true leaves vs. phyllodes. Consistent with our predictions, caterpillars reared on true leaves had a significantly higher mortality rate and took longer to pupate than caterpillars reared on phyllodes. Additional sources of variation in S. paludicola development included sex, phyllode age (young vs. mature), and host identity. Further research is needed to determine the mechanistic traits underlying A. koa resistance to S. paludicola, and to test whether true development does in fact contribute to a reduction in S. paludicola performance and population stability on previously defoliated trees.

Key words: , outbreaks, resistance, heteroblasty, induced defense

Introduction 2013). Because native island communi- Herbivory is a universal threat to plants, ties often lack herbivore guilds present in and one of the primary goals of ecology continental communities (such as grazing is to understand how plant traits mediate and browsing mammals), island plants are this interaction. A substantial body of predicted to have relatively weak defenses research has characterized the chemical, (Ziegler 2002, Carlquist 1970), and as a physical, and indirect defense traits of consequence, few studies have explored plants, documenting extensive variation island plant defenses. within and among species (Denno and Hawaii is an ideal setting for investigat- McClure 1983, Agrawal 2011). However, ing herbivory and the evolution of defense very little of this research comes from is- in native island plants because although land systems (Sardans et al. 2010, Peñuelas vertebrate herbivore guilds are largely et al. 2010, Funk and Throop 2010, Barton absent from the native fauna, others, most 142 Barton and Haines notably insect herbivores, are diverse and size of plants when they begin to develop abundant. One striking example of a na- phyllodes (Barton, personal observation), tive herbivore that is likely to exert strong suggesting genetic and/or environmental selection pressure for defense on its host variation in the timing of this heteroblas- is the koa moth, Scotorythra paludicola tic shift. Anatomical and physiological Butler (Geometridae). Endemic to , differences between true leaves and phyl- and Hawaii (Heddle 2003), S. palu- lodes are consistent with juvenile leaves dicola is a specialist folivore on the native maximizing growth and phyllodes having koa tree (Acacia koa A. Gray, Fabaceae). greater drought tolerance (Pasquet-Kok Populations of S. paludicola occasionally et al. 2010). While considerable research undergo massive increases, leading to has examined the ecophysiology of A. outbreaks that defoliate entire koa koa (Craven et al. 2010, Ares and Fownes (Haines et al. 2009). The largest outbreak 1999), nothing is known about its defenses on record is currently underway on the against S. paludicola, and whether defense island of Hawaii, where approximately or nutritional quality differ across the 70,000 acres of A. koa forest have been heteroblastic phase change. This is an defoliated (in some areas multiple times) important oversight, because extensive since the outbreak was first discovered research has shown that herbivory and in January 2013. Previous S. paludicola defense vary significantly across plant on- outbreaks have slowed tree growth and togeny in general (Barton and Koricheva caused as much as 35% mortality of 2010), and observations of the A. koa–S. mature defoliated A. koa trees (Stein and paludicola interaction suggest that leaf Scowcroft 1984), indicating the strong fit- form may play an important role during ness impact of this interaction on A. koa. outbreaks. Considering the specificity of this in- Following a 2004 S. paludicola out- teraction, the magnitude of S. paludicola break on Maui, mature A. koa trees were damage, and the documented fitness con- observed to flush juvenile true leaves, sequences, it is likely that strong selection primarily via epicormic growth from main pressure has resulted in well developed trunks or branches, rather than from ter- defenses in A. koa. Although damage minal shoots (Haines et al. 2009). Induced tolerance via survival and re-growth are juvenilization is not commonly observed expected to be important for plants that in response to herbivory, although it has experience herbivore outbreaks (Kessler been previously documented in poplar et al. 2012), we focus here on evidence and birch (Chapin et al. 1985). Whether for resistance traits that directly affect S. juvenile true leaves confer an advantage to paludicola caterpillars and, in particular, defoliated A. koa via increases in growth investigate whether resistance is associ- rate (i.e., tolerance mechanisms) or via ated with leaf heteroblasty. Like many induced resistance to further damage by species of Acacia, A. koa is heteroblastic, caterpillars (i.e., resistance mechanisms), with plants undergoing a striking shift or whether they develop simply as a con- from a juvenile stage characterized by sequence of anatomical and physiologi- bipinnately compound “true leaves” to a cal constraints following defoliation, is mature stage with phyllodes (Fig. 1). The unknown. transition from juvenile to mature stages To investigate the suitability of A. koa typically occurs within a few years of true leaves and phyllodes for S. paludicola (Pasquet-Kok et al. 2010), development, we performed a no-choice but there is considerable variation in the bioassay in which caterpillars were reared Koa looper fitness on leaves and phyllodes of Acacia koa 143

Figure 1. Acacia koa displaying both phyllodes and true leaves. Photo taken by K.E. Barton, Waimea Valley, Oahu, June 2013. on each of the two leaf types. Moreover, be high, but that performance will vary because resistance often changes during among leaf types consistent with observa- leaf expansion and maturation due to tions that caterpillars prefer phyllodes in decreases in defensive plant secondary the field (Haines, personal observation), metabolites and nutrients, and simultane- and in the laboratory (B. Peck personal ous increases in leaf toughness (Koricheva communication). and Barton 2012, Kursar and Coley 2003), we reared caterpillars on young as well as Materials and Methods mature phyllodes; true leaves were only A no-choice bioassay was conducted assessed at the mature stage due to the in the laboratory (St John Plant Sciences lack of availability of young true leaves. Building, University of Hawaii at Manoa We used S. paludicola performance as an campus) in July–August 2013. Scotorythra indirect measure of resistance. Caterpillar paludicola caterpillars were obtained survival, development time to pupation, from eggs laid by wild-caught female and pupal weight were compared among . Approximately 40 female moths groups reared on true leaves, young were collected from swarms in the Ocean phyllodes, and mature phyllodes. Lower View Estates subdivision (District of Kau, survival and pupal weight, and longer de- HI, N 19.142°, W 155.761) on 20 Jul 2013. velopment time, would reduce fitness ofS. Female S. paludicola moths insert eggs paludicola, reflecting a higher resistance into bark crevices and patches of moss in A. koa. As a specialist of A. koa, it is on the trunks of host trees (Zimmerman likely that S. paludicola performance will 1958), though it is unknown what specific 144 Barton and Haines cues they use to recognize host plants. To sample size was N=134 caterpillars, but obtain eggs, female moths were housed replication within leaf types varied as a together at 23°C in a screen cage contain- consequence of foliage availability (n=43 ing moss clumps, pieces of koa bark, and for true leaves; n=47 for young phyllodes; sprigs of koa foliage (to provide a potential n=44 for mature phyllodes). oviposition cue) for 5 days. Moths were Data were analyzed using SAS version fed honey water sprayed on damp paper 9.2 (Cary, North Carolina). Survival on towels draped over the cages. Moss and diets composed of the three leaf types bark containing eggs was removed daily was analyzed with a Fisher’s exact test. after overnight exposure to moths, and Days to pupation and pupal mass met as- placed in sealed 35-oz plastic containers sumptions of normality and were analyzed without foliage until the eggs hatched with mixed model ANOVA’s in which the (6–7 days). Caterpillars were presumably main factors of diet (3 levels: true leaves, from several different broods produced young phyllodes, mature phyllodes), S. by different mothers, and were randomly paludicola sex (2 levels), and tree iden- assigned to diet treatments. tity (10 levels) were tested. Interactions Diets consisted of a single leaf type between diet and sex were not significant throughout the development of the cat- and were removed from the final models. erpillars. Phyllodes and true leaves were Tree identity was considered a random collected weekly from A. koa in Manoa factor, the significance of which was tested Valley, Oahu. Trees were marked so that by running the models with and without caterpillars could be reared on foliage the random effect, and then calculating from the same individual tree throughout the log-likelihood ratio statistics, which the bioassay. Ten trees were used for the can be compared to a chi-square distribu- bioassay, and multiple caterpillars were tion with one degree of freedom (Littell reared on foliage from the same tree, pro- et al. 1996). The Kenward-Roger method viding the replication necessary for testing was used to calculate degrees of freedom whether S. paludicola performance varied (Kenward and Roger 1997). Tukey-adjust- significantly among A. koa trees due to ed least-square mean comparisons were genetic and/or environmental variation in used to interpret significant differences resistance. among diets. Caterpillars were reared singly in small plastic dishes fitted with moist filter paper. Results Fresh foliage and clean filter papers were Caterpillar survival varied signifi- provided daily, when caterpillar survival cantly among diets (Fisher’s exact test P could also be noted. Upon pupation, dishes = 0.03857), with less than half as many were set aside to allow pupal cases to caterpillars surviving to pupation on harden completely for at least one day, and true leaves (32.6%) as young (72.3%) or pupation date was recorded. Pupae were mature (71.1%) phyllodes. Larval devel- then weighed fresh to the nearest 0.01 mg opment time varied significantly among on a Mettler Toledo microbalance. The sex caterpillars reared on the three different of each pupa was determined by examin- diets (F2,73.6 = 30.38, P < 0.0001). Devel- ing differences in the genitalic bumps opment to pupation was fastest on young on the ventral surface of the terminal phyllodes (least-square mean 23.6 days abdominal segments, allowing us to assess ± 0.6 standard error), was intermediate whether performance differed between on mature phyllodes (least-square mean male and female S. paludicola. The total 27.7 ± 0.6 days), and took longest on true Koa looper fitness on leaves and phyllodes of Acacia koa 145 leaves (mean 31.5 ± 1.0 days). Least-square 2008, Nykänen and Koricheva 2004). mean comparisons of development time An ontogenetic decline in resistance is revealed that all three diets were sig- uncommon in tropical woody plants, nificantly different from each other (P < although it is common in boreal forest 0.0001). where ground-dwelling mammals exert Pupal biomass also depended sig- strong selection pressure for defense in nificantly on leaf type (F2,72.8 = 4.45, P = juveniles (Barton and Koricheva 2010, 0.0151). Pupae reared on young phyllodes Swihart and Bryant 2001). Furthermore, (least-square mean 0.07492 ± 0.00374 g) the tendency for some A. koa to produce were smaller than those reared on true more resistant true leaves through epi- leaves (least-square mean 0.07912 ± cormic growth following S. paludicola 0.00501 g), which were smaller than those outbreaks (Haines et al. 2009) may be reared on mature phyllodes (least-square considered a form of induced resistance mean 0.08532 ± 0.00380 g). However, in response to these extreme defoliations. pupal weight only differed significantly To determine whether induced resistance between groups reared on mature vs. via true leaf development contributes to young phyllodes (P = 0.0040), and pupal local declines in S. paludicola outbreaks weight did not differ significantly between requires further research. Field studies true leaves and either young (P = 0.4045) that quantify S. paludicola population or mature (P = 0.2218) phyllodes. dynamics on trees that vary in their de- Significant variation among trees in velopment of true leaves vs. phyllodes their suitability as host plants was detected following outbreak defoliations would be for pupal mass (χ2 = 11.1, P = 0.0004), particularly enlightening. but not for development time (χ2 = 0.7, Caterpillar development was also af- P = 0.2013). Sex significantly influenced fected by phyllode age when reared on

S. paludicola development time (F1,71.7 = phyllodes. Scotorythra paludicola cater-

19.53, P < 0.0001) and pupal mass (F1,67.3 pillars took longer to develop on mature = 86.08, P < 0.0001), with females taking than on young phyllodes, but they were approximately five days longer to pupate also ultimately larger, even after taking and weighing nearly twice as much as into account the strong sexual dimorphism male pupae. in size. Although a longer larval stage poses additional risk for predation and Discussion parasitism, this cost may be outweighed In this study, we provide clear evidence by the benefit gained in pupal size, given that A. koa leaves differ in their suit- the strong relationship between pupal size ability for S. paludicola development. and reproductive fitness in Lepidoptera Caterpillars reared on true leaves suffered (Honek 1993). from higher mortality and took longer to Faster development time on young phyl- develop than those reared on phyllodes. lodes has important implications regard- This suggests that juvenile A. koa trees ing the causes of S. paludicola outbreaks, that retain their true leaves are more re- which are as yet undetermined (Haines et sistant than mature trees to herbivory by al. 2009). Caterpillars of S. paludiciola this specialist herbivore, though young are known to be impacted by native and trees are likely less tolerant to defoliation, non-native predators, parasitoids, and due to lower capacity to store resources pathogens (Haines et al. 2009), but it is in the trunk and roots and slower growth unknown to what extent these top-down rates following damage (Stevens et al., pressures affect population fluctuations. 146 Barton and Haines Our data demonstrate that bottom-up fac- University of Hawaii at Manoa and to tors are potentially important regulatory WPH by the State of Hawaii, Department factors as well. Although much more eco- of Land and Natural Resources. logical data will be required to elucidate the causes of S. paludicola outbreaks, Literature Cited they could be caused by rare interactions Agrawal, A.A. 2011. Current trends in the that optimize conditions for caterpillar evolutionary ecology of plant defence. survival. For example, an increased avail- Functional Ecology 25: 420–432. Ares, A., J.H. Fownes. ability of young phyllodes could cause a and 1999. Water sup- ply regulates structure, productivity, and sudden population increase, especially water use efficiency ofAcacia koa forest in if it happens to be synchronized with an Hawaii. Oecologia 121: 458–466. unrelated release from top-down regula- Barton, K.E. 2013. Prickles, latex and toler- tory factors such as parasitism or disease. ance in the endemic Hawaiian prickly poppy Although it is clear that S. paludicola (Argemone glauca): Variation between caterpillars are strongly affected by A. koa populations, across ontogeny and due to leaf type and age, the traits underlying phenotypic plasticity. Oecologia, in press. this resistance remain unknown. Further Barton, K.E., and J. Koricheva. 2010. The ontogeny of plant defense and herbivory: research is needed to investigate whether characterizing general patterns using meta- A. koa resistance is chemical (i.e., pheno- analysis. American Naturalist 175: 481–493. lics, alkaloids), physical (i.e., toughness, Carlquist, S.J. 1970. Hawaii: a natural history. trichomes), or nutritional (i.e. nutrient Pacific Tropical Botanical Garden, Lawai, content or availability), and to demonstrate , Hawaii. how these traits differ among leaf types. Chapin, F.S. III, J.P. Bryant, and J.F. Fox. The significant variation detected among 1985. Lack of induced chemical defense in trees for pupal mass (but not development juvenile Alaskan woody plants in response time) indicates that besides the variation to simulated browsing. Oecologia, 67: 457–459. due to leaf type, A. koa trees vary in their Craven, D., S. Gulamhussein, and G.P. resistance to S. paludicola, suggesting the Berlyn. 2010. Physiological and anatomical possibility for further evolution by natural responses of Acacia koa (Gray) seedlings selection of resistance traits. Thus, the A. to varying light and drought conditions. koa–S. paludicola interaction presents an Environmental and Experimental Botany ideal system for studying the evolution in 69: 205–213. action of island plant defenses, and herbi- Denno, R.F., and M.S. McClure. 1983. Vari- vore offenses. able plants and herbivores in natural and managed systems. New York, Academic Press. Acknowledgments Funk, J.L., and H.L. Throop. 2010. Enemy The authors thank Robert Peck, Paul release and plant invasion: patterns of de- Banko, and Aaron Shiels for advice on fensive traits and leaf damage in Hawaii. the study design, insights from field Oecologia 162: 815–823. observations and unpublished data from Haines, W.P., M.L. Heddle, P. Welton, and laboratory trials. Kelsey Coleman, Robert D. Rubinoff. 2009. A recent outbreak of the Guerrero, Tayler Pave, Casey Jones, and Hawaiian koa moth, Scotorythra paludicola Lydia Lam assisted with the bioassay and (Lepidoptera: Geometridae), and a review of outbreaks between 1892 and 2003. Pacific caterpillar rearing. Comments from an Science 63: 349–369. anonymous reviewer improved the manu- Heddle, A. 2003. Systematics and phyloge- script. Financial support was provided to netics of the endemic Scotorythra KEB by the College of Natural Sciences, (Lepidoptera: Geometridae) in the Hawai- Koa looper fitness on leaves and phyllodes of Acacia koa 147 ian Islands, PhD, University of California, Pasquet-Kok, J., C. Creese, and L. Sack. Berkeley. 2010. Turning over a new ‘leaf’: multiple Honek, A. 1993. Intraspecific variation in body functional significances of leaves versus size and fecundity in —a general phyllodes in Hawaiian Acacia koa. Plant relationship. Oikos 66: 483–492. Cell and Environment 33: 2084–2100. Kenward, M.G., and J.H. Roger. 1997. Small Peñuelas, J., J. Sardans, J. Llusia, S.M. sample inference for fixed effects from Owen, J. Silva, and U. Niinemets. 2010. restricted maximum likelihood. Biometrics Higher allocation to low cost chemical de- 53: 983–997. fenses in invasive species of Hawaii. Journal Kessler, A., K. Poveda, and E.H. Poelman. of Chemical Ecology 36: 1255–1270. 2012. Plant-induced responses and herbi- Sardans, J., J. Llusia, U. Niinemets, S. Owen, vore population dynamics. In: P. Barbosa, and J. Peñuelas. 2010. Foliar mono- and D.K. Letourneau, and A.A. Agrawal (eds.), sesquiterpene contents in relation to leaf Insect outbreaks revisited. West Sussex, economic spectrum in native and alien spe- Wiley-Blackwell. cies in Oahu (Hawaii). Journal of Chemical Koricheva, J., and K.E. Barton. 2012. Ecology 36: 210–226. Temporal changes in plant secondary me- Stein, J.D., and P.G. Scowcroft. 1984. Growth tabolite production: patterns, causes and and refoliation of koa trees infested by the consequences. In: G.R. Iason, M. Dicke, koa moth Scotorythra paludicola (Lepi- and S.E. Hartley (eds.), The ecology of doptera, Geometridae). Pacific Science 38: plant secondary metabolites. Cambridge, 333–339. Cambridge University Press. Stevens, M.T., E.L. Kruger, and R.L. Kursar, T.A., and P.D. Coley. 2003. Con- Lindroth. 2008. Variation in tolerance to vergence in defense syndromes of young herbivory is mediated by differences in leaves in tropical rainforests. Biochemical biomass allocation in aspen. Functional Systematics and Ecology 31: 929–949. Ecology 22: 40–47. Littell, R.C., G.A. Milliken, W.W. Stroup, Swihart, R.K., and J.P. Bryant. 2001. Im- and R.D Wolfinger.1996. SAS System for portance of biogeography and ontogeny mixed models. Cary, NC, SAS Institute, Inc. of woody plants in winter herbivory by Nykänen, H., and J. Koricheva. 2004. mammals. Journal of Mammalogy 82: 1–21. Damage-induced changes in woody plants Ziegler, A.C. 2002. Hawaiian natural history, and their effects on insect herbivore per- ecology, and evolution. Honolulu, University formance: a meta-analysis. Oikos 104: of Hawaii Press. 247–268. Zimmerman, E.C. 1958. Macrolepidoptera. Honolulu, University of Hawaii Press. 148 Barton and Haines