LEAF QUALITY AND THE HOST PREFERENCES OF GYPSY 273 Number of arboreal genera in MOTH IN THE NORTHERN DECIDUOUS FOREST the diets of North American Marlin J. Lechowicz macrolepidoptera - compiled
AssociaLe Professor of Biology, from Tietz ( 1972) McGill University, C: I :, 0 1205 Avenue Dr. Penfield, 0 MonLreal, Quebec >, () CANADA H3A lBl C: :,., .,O' u:
ABSTRACT
BoLh gypsy moLh hosL preferences and the foliage characLerisLics thaL have been imp'ii caLed as factors in hosL selecLion were moni tored from 1979 Lo 1982 in a Quercus-Acer-OsLrya foresL near Montreal, Quebec. The preliminary analyses of Lhese data suggesL Lhe hypothesis thaL gypsy moth larvae preferenLially attack 3 5 7 9 11 13 15 17 18 21 23 25 > 25 trees LhaL have high sugar:tannin ratios in 2 4 6 8 10 12 14 16 18 20 22· 24 Lheir young foliage.
Number of Arboreal Host Genera
The gypsy moth, LymanLria di spar L., is Figure 1. Numbers of genera of trees acceptable recognized as unusually polyphag0us boLh in iLs in the dieLs of NorLh American macro native Eurasian habitats and also in easLern lepidoptera·ns; compiled from Tietz North America where it is inLroduced. Of Lhe (1972). approximately 185 trees native to Europe (Polunin and Everard 1976) gypsy moLh larvae on evidence from la boraLory feeding tri als and have been reporLed to feed on the foliage of 75 neurophysiological moniLoring of sensilla ex (Kurir 1952; Gyorfi 1960). Similarly in Lhe posed Lo LesL compounds. DeLhier (1982) pro forests of northeasLern North America where Lhe vides an inLeresLing hisLorical review of Lhis gypsy moth has become esLablished, the larvae l i LeraLure. Ecological explanations fo cus aL can feed on al least 86 (Mosher 1915; Forbush tenL ion on the nutritional quality of poLential and Fernald 1896) of Lhe approximately 169 avai hosL foliage and assume that naLural selection lable native trees (LitLle 1979). Such a high has led to preferences for hosLs with relatively degree of polyphagy among Lree-feeding macro nuLrienL-rich foliage low in Loxic or di gesti lepidopLerans is matched by only a very few biliLy-reducing compounds. AlLhough such ecolo species (Fig. l). This poLent i al breadth of gical explanaLions have a long history (see diet combined with the facL LhaL not all Lrees FuLuyma 1983 for a recenL review) Lhey received are equally preferred as hosLs (Lechowicz and increased atLenLion after Feeny (1976) and Jobin 1983) makes Lhe gypsy moLh especially use Rhoades and Cates (1976) independently proposed ful in developing and LesLing hypotheses abouL a general Lheory of planL defense against herbi the interactions beLween woody planLs and the vores. These two types of explanation are noL herbivores LhaL feed on Lheir leaves. The gypsy mutually exclusive; evoluL ion should generally moth essentially provides a bioassay to call lead Lo proximaLe behavioral cues which result aLtenLion to Lraits that make trees better or in larvae feeding on hosLs of high nutritional worse hosts Lo defoliaLing insecLs. If we can quality. undersLand Lhe basis for gypsy moth hosL selec tion, we may hope Lo gain some general insights In the work describetree char Linguished as behavioral versus ecological ex acteri SL ics which have been suggested Lo in planations. Behavioral explanaLions focus aL fluence host selection by lepidopteran foli tention on proximaLe cues involving repel lenL vores. Here I describe Lhe sLudied forest, sum and aLLractant compounds that influence larval marize the hisLory of its in festat.ion by gypsy feeding behavior; this approach mosL ofLen draws mot!-\, explain Lhe meLhods which have been used
67 TABLE 1. PhyLosociological Summary of Lhe Tree SLraLum on Lhe Southern Faces of Lake Hill, MonL. SL. Hilaire, Quebec. Rel aL ive Importance Tree Species Common Name Acronym Frequencya/ DensiLyb/ DominanceC / Valued/
Acer pensylvanicum L. SLriped maple Apen 1 1 8.5 0.3 A. ---rubrum L. Red maple Arub l 3 67.7 0.5 A. saccharum Marsh. Sugar maple Asac 24 158 23L14 .4 17.0 A. spicatum Lam. MounLain maple Aspic l 4 39.2 0.5 Amelanchier sp. Serviceberry Amel 3 3 27.4 0.9 BeLul a papyrifera Marsh. While birch Bpao 9 47 696.3 5.5 B. luLea Michx. f. Yellow birch Bl uL 1 4 60.7 0.5 Carya cord iformis (Wang)K. Koch Yellowb ud hickory Carya 2 2 35.6 0.6 Fagus grandifolia Ehrh. Beech Fagus 7 78 1373.2 7.6 Fraxinus americana L. While ash Frax 17 56 778.5 8.1 Juglans cinerea L. BuLternuL Jug 2 2 71. 8 0. 7 OsLrya virginiana (Mill.)K. Koch Ironwood OsLrya 20 181 1835.3 15.7 Pious strobus L. White pine Pious 3 4 92.4 1.1 Populus grand idenLaLa Michx. Big-LooLh aspen Pgran 2 11 206.1 1.4 Prunus pensylvanic a L. f. Pi.n cherry Ppen l 3 27.2 0.5 P. seroLina Ehrh. Black cherry Pser 2 2 29.0 0.6 Quercus rub ra L. Red oak Qrub 22 341 6963. 8 33.3 ------Ulmus rubra Muhl. Slippery elm Urub 1 1 15.4 0.3 ---Ti1 ia americana L. Basswood Tilia 12 22 357.5 4.7
SummaLions 131 923 15030.0 99.8 -
2 a; Frequency: The numb er of 500 m quadrats in which each Lree species occurred ; Lota! sample w as 24 quadrats. b/ DensiLy: The number of sLems of each tree species found in all 24 quadraLs. C/ Dominance: The cumulaLive DBH in cm. for each Lree species summed in al.l 24 quadraLs. d/ The mean of frequency, d ensiLy, and dominance each relativized as a percenLage of Lhe respective toLal for all species; see Curlis (1959).
to quantify hosL preference, and deLail Lhe Lree characLerisLics which were moniLored over the period 1979 through 1981. I Lhen presenL a pre- 1iminary analysis of these data and suggest a Lentative explanation for the d ynamics of Lhe inLeracLion beLween Lhe gy psy moth and its hosL planLs.
The Lake Hill SLudy Sile
The work d escrib ed here concerns a gypsy moth infested foresL on MonL St. Hilaire, one of the eight Monteregian Hills which rise abrupLly from the plains of the SL. Lawrence River Valley in the viciniLy of Montreal, Quebec (Fig. 2). Mont St. Hilaire consisLs of seven low pe aks surround ing a small lake; Lhe peaks ri se to a m aximum of 416 meLers, abouL 355 meters above Figure 2. Aerial view of MonL St. Hilaire, the surround ing plai n which is a mosaic of vil- looking souLheasL toward Rougemont. lages, agricultural land, a nd remnant woodland . Lake Hill rises from Lhe south shore The mountain iLsel f is covered by forest which of Lac HerLel in the cenLer of Lhe in many areas h as been litLle or not al all figure. PhoLo courtesy of Dr. Luc disLurbed for over 300 years; topographic and Jobin, Laurentian ForesL Research edaphic heterogeneity contribute Lo a diversiLy CenLre, SLe. Foy, Quebec.
68 of habitats including old growth beech-maple forests on deep moist soils, hemlock on steep rocky slopes, yellow birch-red maple swamps in EXPANSION OF THE GYPSY MOTH depressions, oak-dominated forests on the dryer RANGE IN SOUTHWESTERN sites, and successional foresls wilh aspen and QUEBEC: ZONES OF MODERATE birch on recently disturbed siles (Phillips TO SEVERE OEFOLIATtON 1972). Comparable forests have existed in this area for the past 8000 years (Richard 1977). Maycock 0961) provides a thorough summary of tf•JIIIIOMI the geology, soils, climate, and flora of Mont St. Hilaire and Walther (1963) describes Lhe vegetation of all the Monteregian Hills.
The forest in which gypsy moth activiLf has been primarily monitored is on the southern and western faces of Lake Hi 11, one of the lower peaks of M::>nt SL. Hilaire which reaches an alti tude of only 297 meters. The composition of this forest was determined by randomly placing 2 ' twenty-four 500 m circular quadrals along 4 altitudinal isoclines at aboul 25 m intervals down from the ridge top. In each quadrat Lhe diameter ·al breast heighl (DBH) and Lhe species of all trees (DBH >.,.. 0. 8 dm) were recorded. Figure 3. Approximate t1m1ng of expansion of These dala are summarized in Table 1. The rela Lhe zone of serious gypsy moth in tive importance values (Curlis 1959) emphasize festation in Quebec; adapted from in Lhe dominance of Quercus rubra, Acer saccharum, formation in Brown (1967), Cardinal and Oslrya virginiana in this forest; Fraxinus (1967), Martineau el al. (1975, americana, Fagus grandifolia, Betula papyrifera, 1976), Lavallee el al. (1977, 1978), and Tilia americana are also substantial compo and Lachance el al. (1979, 1980, nentsof Lhe tree slralum but the remaining 1981). eleven tree species are of only minor impor tance. Tsuga canadensis, Populus tremuloides, .!'._. balsamifera, .!'._. deltoides and Betula populi There was no record of gypsy moLh, and folia occur sparsely on Lake Hill but were not certainly had be en no sub!ltantial defoliation, found in the random quadrats. on Mont St. Hilaire until 1977 when 10 hectares of Lhe forest on Burned Hill which is on the Compared Lo Lhe forests of the St. Lawrence southwest face of the mountain were severely Valley and southern Quebec in general (Grandtner defoliated l /. It is likely thal the infestation 1966; Bouchard and Maycock 1978) the Lake Hill had been newly established only in the preceding foresl is xeric. Detailed studies of Lhe annual few years; this slope of Lhe mountain fa ces a heat and water budgets of the south versus norlh public campground which may have been the source slopes of Lake Hil 1 are available (Rouse and of infection through vehicular Lrans port from Wilson 1969; Wilson 1970). The oak-dominated areas to the south or wesl. By 1978 Lhe infest foresl on the south slope is decidedly more al ion had spread Lo 259 heel ares of which 195 prone to summer drought stress than the beech were severely defoliated. This area included maple forest on the north slope. Such topogra 123 hectares of Lhe Lake Hill stu�y site most of phic juxtaposition of mesic and xeric forest which was heavily defoliated in 1978. It is communities is common on the Monteregian Hills noleworLhy thal serious infestalions were (Walther 1963). Xeric forests are generally limited to Lhe more xeric foresls on Mont St. more susceptible to atl ack by gypsy moth Hilaire in accord with ·Houston and Valentine's (Houston and Valentine 1977). (1977) comments on the ch�racteristics of sites mosl susceptible Lo gypsy moth infestations. As parL of hill doctoral re!learch Madrid (1979; History of Gypsy Molh al Mont St. Hilaire Madrid and SLewarL 1981) monitored gypsy molh populations on Lake Hill in 1977 and 1978; his The gypsy moth appears Lo have fi rsl become data provide a very useful supplement to my 1 established in Quebec during the mid-1960 s records which do not begin unlil 1979. In (Cardinal 196 7; Brown 196 7) and until the mid- 1970's serious infestations were limited prima rily Lo Huntingdon and Chaleauguay counties south and wesl of Montreal (Martineau et al. 1975). About 1975 �he zone of serious infesta l; Jobin, L. (1978). HisLorique el situation tion began a steady expansion (Fig. 3); in 1975 acLuelle de la spongieuse au Mont St. Hi traces of gypsy moth were noled on Monl. St. laire� Internal reporl, Laurentian Foresl Gregoire, one of the Monteregian Hills south of Research Centre, SLe-Foy, Quebec 13p. Mont SL. Hilaire (Martineau el al. 1976).
69 addition government agencies 2 / monitored the cribed in Lechowicz and Jobin (1983). This population on Lake Hill in 1978 and also in 1979 method measures preference as a function of the (Jobin 1982). Figure 4 summarizes the available availability and the utilization of foliage on information on gypsy moth population density at the different host trees in the forest. Empha Lake Hill from 1977 through 198 2. It is impor sis is on the host preferences of late instar tant to note that the population decline begin larvae which are primarily responsible for ning in 1979 was not due Lo Quebec's program of foliage losses (Valentine and Talerico 1980). spraying Bacillus thuringiensis (Jobin 1982 ); If dispersing gypsy moth larvae actually had no the Lake Hill study site is on land owned by preference for or against the foliage of a McGill University and protected as a UNESCO Man particular tree they would be expected to feed and Biosphere Ecological reserve in which spray on that tree in direct proportion to its abun ing has not been allowed. dance in the forest canopy. This null expecta tion is founded on the assumptions that the probability dispersing larvae will encounter a host is determined by the host's relative ab un dance in the canopy and that larvae without preferences will se ttle on the first host they 4.00 I] LAKE HILL STUDY encounter. At dispersal larval preferences in the field should be closely related to the pre D SITE: GYPSY MOTH ferences reported in the behavioral literature ,... ✓" Q) / ' / POPULATION which are determined in laboratory choice-trials ... / ' / ' tU / \ (Mosher 1915; Barbosa et al. 1979). The late ... \ NUMBERS 0 6 ' instar preferences reportedhere, however, arise Q) \ .c: \ not only from behavioral choices during dis ...... \ 1/) \ persal but also from possible differences in Q) early instar survival on different hosts. 1/) 1/) 3.00 Larval survival may be controlled not only by tU foliage qu•lity but also by bark roughness, E canopy architecture, and similar traits that 0) 0) influence vulnerability to predators and para ...... Q) sites. These late instar host preferences are perhaps best viewed as measures of host suscep 0) 0 tibility to defoliation by gypsy moth. ...J As discussed in detail by Lechowicz and Jobin (1983), the availability ..E.i. for host i is 2.00 measured by its contribution to the to tal DBH of the m different hosts in the sampled forest: 1977 1978 1979 1980 1981 1982 Figure 4. Gypsy moth population dynamics on eq.l) Lake Hill. The squares are 1977 and 1978 data from Madrid 0979). The 2 1978 estimate by the government /, where bij is the DBH of the jLh tree of species shown as a half-closed square, is i. Utilization is estimated fr om the mean of virtually identical to that by repeated counts of numbers of late instar larvae Madrid. The open triangles are data congregating. under tarpaper bands around each from Jobin (198 2) in 1978 and 1979. tree in the sampled quadrats; this is� standard All these plots were on the west end method to estimate numbers of late instar larvae of Lake Hill, while my data (closed feeding on a tree (Weseloh 1974). This estimate circles) also include the southern assumes that late instar migration between trees faces of Lake Hill. does not occur during the counting period; in low density populations this assumption appears to be met (Lechowicz and Jobin 198 3; Mauffette Estimates of Gypsy Moth Host Preferences and Lechowicz3 /; Wallner, this volume). The utilization .E.i of host is measured by the The host preferences of late inslar gypsy number of larvae on host i relative Lo the total moth larvae on Lake Hill have been monitored number of larvae on all the sampled hosts: from 1979 through 1982 using the method des-
3 ; Mauffette, Y. and M.J. Lechowicz. The in- 2 / Bordeleau, C., C. Gagnon, C. Theriault, L. fluence of host plant on the larval develop Jobin, C. Coulombe, and A. Caron (1980). ment rate of gypsy moth, Lymantria dispar Evaluation du programme de traitement au (L.), in the forest environment, manuscript B.T. effectue centre la spongieuse au Quebec in review. en 1979, Internal Report, Minislre de Agri culture, Quebec 37p.
70 plicated as possible factors determining the _1r· l ..lJ eq. 2) favorability of trees as hosts to lepidopteran folivores. Investigators have tended to focus on only one or a few fs1ctors in a given study where lij is the mean number of larvae counted despiLe the likelihood LhaL a number of factors on the jth tree of species i. interacL Lo conLrol the suscepLibility of a tree to herbivore altack. Certainly no single factor Since gypsy moth larval growth and develop has proven to be even a reasonably universal ment differs on differenL hosL species (SchmidL predictor of herbivore su sceptibility in woody 1956; Barbosa and Capinera 1977), the Liming of planLs. Also the predictable correlations larval counts to ac hieve a represenLative esti beLween cerLain traits that may arise from con mate of r i is important. If a count is taken straints imposed by tree physiology and archi too early, larvae on some hosts will not have tecture can only be understood in a multifac reached instar IV when the diurnal resting beha torial analysis. Therefore within logistic vior on which the tarpaper counts depend begins limitaLions we have atLempted to monitor a com (Leonard 1970). Conversely counts taken too prehensive set of foliage characLeristics (Table late will miss larvae that have ·already pupat 2 ) LhaL may play a role in hosL selecLion by the ed. MauffeLte and Lechowicz3 / have analyzed the gypsy moth. Sampling was concentrated on the time course of larval numbers on 29 hosL species spring foliage coincident with dispersal of at 13 sites in southwestern Quebec; in general gypsy moth and on the mid-summer foliage avail there is about a 2 to 3 week window during which able to the late insLar larvae. Samples were the mean of repeated counts will provide a rep taken from a toLal of 23 tree species on Lake resentative estimate of late instar larval num Hill but only 14 species both having adequate bers across host species. It is desirable lo replication (3 -6, mostly 5 or 6) and having been make a fairly long series of counts and Lhen and having been included among Lhe 19 species in discard ·the early and late entries which indi the quadrats analyzed for host preference pat cate that resting behavior is noL developed on terns are reported here. In Lhe following para all hos Ls or that substantial pupation has oc graphs I briefly rationalize my choice of curred on some hosts. The preferences reporLed foliage characteristics and describe the methods here are based on counLs made on June 26-27 and used to assay them. July 3-4, 1979; on June 22 and 30, 1980; on June 21, June 29, and July 5, 1981; and on June 22, June 30, July 6 and July 14, 1982. Leaf Toughness
Several differenL algorithms are available Feeny (1970) demonstrated that the peak to calculate preference from these measures of numbers of lepidopterans feeding on oak foliage availabiliLy and utilizaLion (Lechowicz 1982). coincided wiLh Lhe availability of tender, young Most give comparable results and differ prima leaves in the spring. For gypsy moth Hough and rily in convenience of interpretation. Here I Pimental (1 978) showed that larvae de velop less have used Vanderploeg and Scavia's (1979a, well on the older, tougher leaves of host 1979b) E* electivity index and their related trees. Consequently toughness was monitored by selectivity index W. The selectivity for tree F'eeny's (1970) method which measures the mass species i is: required to shear the leaf Lissue in an area between the main veins of the leaf. The shear m ing face of the penetrometer was 0.145 cm2. In (,E_i/ 2-i )/ E (,E_i /2., i) eq. 3) 1979 sand was used to add mass but subsequently i=l water delivered Lhrough a pipette· tip was found to give more reproducible results. Four repli and Lhe electivity for species i is: cate measurements of toughness were made on * freshly harvested leaves._ E· (W - 1/n)/(W + 1/n) eq. 4) l i i where n is the number of potential host spe Leaf Water ConLent cies. The selectivity W ranges from zero to one, a useful property in cerLain sLatisL ical Scriber and Feeny (1979) have shown a cor analyses but not clearly indicative of preferred relation between higher leaf water content and versus avoided hosts. In general, as a summary better larval growth in lepidopteran larvae preference index, the E* elecLivity is easiesL feeding on leaves of woody plants. Scriber and to interpret: E* is zero if Lhe larvae feed Slanksy 0980) review the importance of water randomly on a host, approaches minus one the content in effecting insect growth. The results more they avoid a hosL, and plus one the more of Hough and Pimental (1978) on gypsy moth they prefer a hosL. Lechowicz 0982) discusses larvae also indicate the favorability of high the interrelationships and properLies of all the water content. The fresh and oven-dry weights available preference indices. of newly harvested leaves were used to calculate waler conlenL as a pe rcentage of fresh weight at harvest; Lhis is the normal pr act ice in the Monitoring of Host Foliage Characteristics enLomological liLerature in conLrast Lo the expression of water as a% dry weight common in Many foliage characLerisLics have been 1m- Lhe hotanical liLeraLure.
71 Table 2. Foliage characteristics monitored from 1979 through 1981 at Lake Hill on Mont St. Hilaire, Quebec. Rows designate date and Julian day of actual samples and the tabled entry indicates how samples were combined to represent spring or summer leaf condition in each year. Unsampled characteristics are indicated by n.d. In addition daily pheno 1 og ical records were kept in 1980 and i981 during the period of canopy development.
Fol in-Denis Precipitable Leucoantho- Leaf Leaf Buffer Toushness % Water % Nitrogen Phenolics Tannins cianins J.!!__ Capacit)'.'. 1979
May 14 (Day 134) spring spring spring spring n.d. spring n. d. n.d, May 21 (Day 141) spring spring spring spring n.d. spring n.d. n,d, June 4 (Day 151) spring spring spring spring n.d, spring n.d. n.d,
July 18(Day 199) summer summer summer summer n.d, summer n.d. n.d.
1980
May 27 (Day 148) spring spring spring spring spring spring spring spring
June 18(Day 170) summer n.d. n.d. summer n.d, summer n. d. n.d. July 16(Day 198) summer n.d. n.d. n.d. n.d. n.d. n .d. n.d.
1981
June 1 (Day 152) spring spring spring spring spring spring spring spring
Leaf Nitrogen Concentration aggravated by the extreme chemical diversity of phenolic compounds (Ribereau-Gayon 1972; Swain McNeil and Southwood (1978) and Mattson 1979) and the inadequacies of available assays 0980) review the considerable support for the fo� different groups of phenolic compounds, The critical importance of nitrogen availability to latter problem has been recently reviewed but the growth of lepidopteran larvae feeding on the without a fully sa tisfactory conclusion (Horvath foliage of woody plants. The nitrogen concen 1981; Martin and Martin 1982; Tempel 1982). tration of oven-dry leaves was determined by a Despite these problems, the unresolved contro Kjeldahl analysis for total organic nitrogen versy on the role of al least some types of (Bradstreet 1965) involving a sulfuric acid phenolic compounds as defenses against herbi digestion in the presence of selenium catalyst vores requires their consideration here. and K2S04 without any predigestion to reduce inorganic nitrogen. The resulting digest was In the face of current methodological pro then Nesslerized (Middleton 1960) and assayed blems, I have settled on three assays for dif colorimetrically for ammonia. Pace et al. ferent phenolic fracl ions I ikely to have inter (1982) have reported problems with nitrate pretable relationships with observed gypsy moth reduction even in the normal Kjeldahl digestion activity: 1) a Folin-Denis assay for total phe but since tree leaves are exlremel y low in in nolics (Rosenblatt and Peluso 1941; Swain and organic nitrogen (Van Tuil 1965) the nitrogen Hillis 1959), 2) a leucoanthocyanin assay for assayed here should be only that in forms avail condensed tannins (Swain and Hillis 1959 as able to lepidopteran larvae. modified by Govindarajan and Mathew 1965), and 3) a newly developed assay for phenolic com pounds that bind to isinglass, a partially puri Leaf Concentrations of Various Phenolic Com fied collagen from fish swim bladders often used pounds to remove phenolic impurities from wine. The first two assays are well established. The Many authors have taken the view that phe isinglass assay is related to traditional tannin nolic compounds function as important defenses assays in the leather industry (Horowitz 197 0), against folivores, especially in woody plants Marigo's (1972) absorption on gelatin, or Martin (see particularly the reviews by Levin 197 1; and Martin's (1982) recent modification of Feeny 1976; Swain 1978; Rhoades 1979; and Bate-Smith's (1973) hemeanalysis. In all these Futuyma 1983). This view has not gone unchal methods Fol in-Denis assays are made on an ex lenged, particularly by those who emphasize the tract before and after exposure to a test pro many metabolic roles of phenolic compounds tein to which tannins will bind, hopefully leav (Seigler and Price 1976) and those who question ing only simple phenolics in solution. The dif the efficacy of such putative defenses (Fox and ference in initial and final absorptance is then McCauley 1977 ; Bernays 1978; Moran and Hamil ton taken as a me,1sure of biologically active tan 1980; Martin and Martin 1983). This debate is nins. Ideally the binding substrate should be
72 leaf protein from each of the tree species being tannin-protein complex occurs rapidly above pH 8 studied (Martin and Martin 1983) but this is for condensed tannins and above pH 5 for hydro virtually impossible in most studies. Recent lyzable tannins (Loomis and Battaile 1966; discussions (Martin and Martin 1982; Tempel Martin and Martin 1983). Thus high gut pH could 1982) agree Lhat at least some type of precipi conceivably increase the· leaf nitrogen available table tannin assay should be included in ecolo to caterpillars feeding on woody plant foliage. gical studies because such an assay directly measures the ability of tannins to bind protein, If this hypothesized mechanism is important the supposed basis of Lheir role in plant de in med iaL ing the cal erpi i I ar-host tree inter fense. In all three of the assays used here I act ion, trees Lhat have lower leaf pH and/or have expressed results as simple absorptance higher leaf buffer capacity should be less values for lack of appropriate calibration stan preferred as hosts. Caterpillars would incur dards (Martin and Marlin 1982). This does raise greater metabolic costs in mainta1n1ng their Lhe practically unavoidable possibility in this high gut pH to effectively digest more acidic cross-species comparison that absorptance wi 11 and/ or better buffered leaf Lissue and should not be an unbiased quantitative index of actual therefore Lend to avoid such host foliage. To phenolic concentrations because of qualitative Lest Lhis hypoLhesLs I measured both the pH of differences in the absorption spectra of consti freshly homogenized frozen leaf tissue and its tuent phenolics on Lhe different tr ee species. buffer ca pacity as Lhe �moles NaOH necessary to bring the aqueous homogenate to pH 8.75. One further aspect of these phenolic assays which has received too little consideration in the literature is the method of extraction of Leaf Phenology phenolic compounds from harvested leaf L issues. It is clear LhaL different extract ion solvents The temporal coordination of herbivore and will differ in both the quantities and quality hosl life cycles is obviously important, espe of phenolic compounds recovered (Ribereau-Gayon cially in insects like Lhe gypsy moth which 1972). In 1979 we extracted 0.1 gr oven-dry break diapause in concert with the emergence of ° (70-80 ) weight of powdered leaf tissue with an the tr ee leaves on which they feed. Barbosa et 80% methanol acid l % HCl sol ul ion for 24 hours al. (1979) have suggested that differences in in a soxhlet apparatus. In subsequent years the tree phenology may influence susceptibility to method of Swain (1979) was adopted in which 0.1 attack by gypsy moth larvae, To test this hypo ° gr oven-dry (4 0-50 Lo lessen degradation of thesis we made daily observations of canopy compounds during drying) weight of powdered leaf development on each tree from which leaves were ° tissue was extracted three Li mes in hol (80 ) harvesLed4 /. Here Lhe mean of the day a tr ee's 50% methanol, Test extractions of fresh frozen first buds burst and the day 90% of its leaves rather than dried leaf tissue did result in had expanded was taken as a measure of the higher abs or pt ion per unit dry weight in both timing of its spring canopy formation. the Folin-Denis and leucoanthocyanin assay buL the readings were signi fie ant ly positively cor related (FD: r = O.26, p = 0.029; LA: r = 0.35; Other Possible Factors p = 0.003). The problems and potential errors inherent in Laking truly parallel samples Lo A considerable number of additional fa ctors estimate Lhe equivalent dry weight of fro?.en L hat might have some inf 1 uence on gypsy moth samples makes the direct assay of dry tissue host selection were nol considered in these preferable in this comparative study. initial eKperiments. In some cases factors were rejected as of little or no importance on Lhe basis of available evidence. Foliage concentra Leaf pH and Buffer Capacity Li'ons of P, K, Mg, Na, Al, 'Ba, Fe, Sr, B, Cu, Zn, or Mn are unrelated Lo gypsy moth growth and Berenbaum (1980) has shown that caterpillar fecundity (Barbosa and Greenblatt 1979; Valen species feeding on leaves of woody plants have tine et al. 1983). Similarly variations in higher gut pH Lhan Lhose feeding on herbaceous foliage concentrations of 25 amino aci ds were plants. She reported a mean midgut pH of 8.67 not correlated with gypsy moth success for larvae feeding on woody plants, signifi (Valentine el al. 1983). These investigators do cantly higher than the pH 8.29 mean for species report significant positive regressions of pupal feeding on herbaceous foliage. She suggested weight on total free sugars and fr ee sugar:Ca Lhat the high gut pH of caLerpi llars feeding on ratios; but the ranges in sugar and Ca concen woody plant foliage would serve Lo break up trations sampled were based on bolh defoliated tannin-protein complexes Lhus rendering Lhe and undefoliated Lrees - other factors may well tannin-based leaf defenses postulated Lo be have contributed Lo these observed correla widespread in woody but not herbaceous plants Lions. Mosl earlier reports ('Beck and Reese less effective. On a dry weight basis Lhe hydrolyzable tannins are considerably more effective al complexing wilh proteins Lhan the condensed tannins (Swain 1978), but the hydro lyzable tannin-protein complexes are less stable 4; Lechowicz, M.J. The phenology of leaf emer than those formed by condensed tannins (Gold gence in a northern deciduous forest, manu stein and Swain 1965). DissociaL ion of Lhe script in preparation.
73 1976; Scriber and Slansky 1981) indicated that foliage carbohydrate concentrations are adequate to meet insecl requirements. Nonetheless, as discussed subsequently, some carbohydrate frac tions may provide token cues imporlant in host selection· and, considering the recent results of C Valentine el al., the possible importance of w a: carbohydrates in gypsy moth nutrition cannot be a: w dismissed. w a: 0. )( .,,., Gypsy Moth Host Preferences at Mont St. .!:
Hilaires: 1979-1982 >,
:E 0.00 The larval hosl preferences of gypsy molh "., in the Lake Hill forest were generally stable jjj during lhe decline of the outbreak in 1979 and 1980 (Fig. 5). Populus grandidentata, Quercus "'w w rubra, Ostrya virginiana, Amelanchier spp., and C Acer saccharum were consistently preferred as 0 > larval hosts while Tilia americana, Carya cordi < formis, Prunus pensylvanica, Fraxinus americana, Acer pensy 1van icum, · and Prunus serot ina were consistently avoided. With the except ions of Fagus grandifolia and Betula papyrifera, • trees lhat had erratic electivity values during 1979- 80 were poorly replicated in lhe available sam ples - Acer rubrum or Juglans cinerea, for example. ---"nie apparenl drop in preference for 1979 1980 1981 1982 1979 1980 1881 1882 beech in 1980 arises in part from a localized outbreak of nuclear polyhedrosis virus in a single quadrat; the increasing preference for Figure 5. The host preferences of late instar white birch is an interesl ing but unexplained larvae in the Lake Hill forest &om trend. In general the host preferences during 1979 through 1982. The calculations this period of relalively high larval numbers have al lowed for occasional missing are in accord with the earlier American and data; open circles represent electi European reporl s reviewed by Lechowic z and Jobin .vity values based on fewer than 6 (1983). trees. Most of the poorly replicated host species are plotted in the right In 1981-82 when larval numbers had dropped graph lo avoid confounding lhe inler to innocuous levels, there were some marked prelalion of trends in the generally shifts in larval preference. Only Populus better replicaled species in the left grandidentata was consistenlly slrongly pre gr·aph. ferred over the 1979-82 interval. Quercus rubra and Ostrya virginiana, allhough slill preferred, had lower electiviLy values in 1981 which re llecause of the inabi 1i ty of female gypsy lurned loward their 1979-80 levels in 1982. molh adults Lo fly and lhe limited mobi lily of Amelanchier dropped abruptly lo being a slightly dispersing larvae (Mason and McManus 1981), avoided hosl and Acer saccharum dee lined stead these shifls in larval preferences may be ex ily to also being """iwcided in 1982. In contrasl plained in part by patterns of egg deposition Fagus grandifolia which had been an avoided hosl ralher lhan simply la rval dispersal and host became slightly preferred and Belula papyrifera selection. IL is useful to compare Lhe observed had risen steadily until it was only slighlly eleclivilies for numbers of egg masses on a host avoided. Most hosts avoided in 1979-80 such as (calculaled analogously to lhat for larvae) to Tilia americana and Fraxinus americana were even lhe larval hosl preferences (Fig. 6). In less preferred by lhe endemic larval population general more preferred larval hosts also bear a in 1981-82. There is so�e indication here lhal grealer proporlion of the egg masses in the the few mosl preferred hosls are more heavily foresl; this is not parlicularly surprising used by an endemic popu lation and lhat lhe mosl considering lhat seleclion should favor larval avoided hosts are especially litlle used. preferences for hosts which will allow success Intermediale hosls see,n to undergo shifls to ful survival and reproduclion. ll does suggest greater or lesser preference that may poten that late-instar mi gralion lo avoided larval tially be explained by changes in fo L iage qua hosts for pupation (Rossiler 1980; Mauffelle and lity. It should be clearly recognized lhal even LechowiczS /) plays only a minor role in the al innocuous populaL io·n levels the gypsy :nolh population dynamics of gypsy molh on Lake Hil 1. population maintains a high degree of polyphagy On the other hand lhe cyclic lendencies apparent with virlually all hosl species allacked lo some in lhe rel alionships belween egg and larval host degree. eleclivities supporls lhe suggestion thal changes in foliage qu1.1lily may inf luence larval
74 host preferences. Consider, for example, Quer marize the primary interrelationships between cus rubra which had high larval and egg electi the measured foliage characterise ics. Biplot vityinl.979; in 1980 larval ac tivity was even analysis begins with a matrix of rows of objects slightly higher but egg electivity was sharply described by columns of their attributes, in reduced. This reduction is reflected in a this instance 14 tree species characterized by lowered larval electivity in 1981 but increased their mean values for the traits in Table 2. egg deposition. In turn by 1982 both larval and The matrix is first relativized by dividing all egg electivities have returned to 1979 levels. entries by the maximum value in their respective Such cyclic patterns raise the possibility that columns. The relativized matrix is then cen changes in foliage quality including· induced tered by subtracting the respective column mean responses such as those shown for birch, oak, from each entry; this results in comparison of and maple (Haukioja and Niemela 1979; Wallner all 'tree species to a hypothetical, mean tree and Walton 1979; Schlicting 1980; Schultz and species. Finally this relativized and centered Baldwin 1980) may influence the gypsy moth popu matrix is analyzed by canonical decomposition, a lation dynamics on host trees from year to year. generalization of singular value decomposition, to acquire the best possible representation in few dimensions of the information in the multi 1 ....------, .00 dimensional input matrix (see details in Gabriel
1979 1971). The reduced matrix can then be repre ..... sented graphically in two (hence biplot) or r••o three dimensions to aid interpretation of tr ends , ,'\ in the data. .. 1982 Acer In interpreting a biplot such as that of ► the foliage tr aits analyzed here (Fig. 7) a t- > number of general rules should be borne in j:: 0 mind. First consider the locations of the indi w ..1 0.00 w vidual host species in the two-dimensional graph. Recall that the analysis is centered on a hypothetical mean host plotted at the origin - c.? the dispersion of tr ee species around the origin c.? w is a graph of variation among hosts based on their foliage characteristics. Trees that are closer together are more similar than those further apart on the ba,is of the foliage traits used in the analysis.
- 1.00 Perhaps the most useful quality of the biplol compared to a tec hnique li ke principal o.oo 1.00 - 1.00 component analysis (Seal 1964) to which it is LARVAL ELECTIVITY closely related is that unlike PCA the bi plot Figure 6. Electivity for egg masses versus that also directly illustrates the contributions of for late-instar larvae in the Lake individual traits in determining the relation Hill forest during 1979 through ships between trees. For convenience the traits 1982. Trends are shown only for the are represented by vectors originating from the eight well-replicated host species origin. The longer a vector, the more variable although all sa mpled hosts were used is that trait among the sample.d tr ees - note to calculate the electivities. however that small differences in a relatively invariant trait may still have important biolo gical effects. The biplot shows only the pat Foliage Characteristics at Mont St. Hilaire: tern of variability in ·traits, the investigator 1979-1982 must interpret that variation biologically.
A multivariate approach to preliminary ana Not only the length but a[so the direction lyses of the gypsy moth host interaction. It is of the Lr�iL vectors in the biplot are important not possible in this paper Lo present the de aids to biological interpretation. The cosine tailed results of all the assays of foliage of the angle between any two vectors ap proxi quality in relation to gypsy moth host selec mates their correlation. Thus acute angles tion. Instead I have used a multivariate sta suggest positive correlation and obtuse angles tistical technique that can· most concisely be negative correlation; trait vectors at right called bi plot analysis (Gabriel 1971) to sum- angles are uncorrelated.
Finally when any vector is extended across the biplot as an axis, the projections of tree species·on that axis rank the trees from low to 5/ Mauffette, Y. and M. Lechowicz. Utilization high in regard to thal trail in the direct ion of the la rval host tree as a pupation site by the vector points. In ot her words the .vectors the gypsy moth, Lymantria dispar L. (Lepi point in the direction that trees with high doptera: Lymantriidae), manuscript in review. values of their respective tr aits will be dis-
75 placed in the bi plot; the actual placement of provide a useful and needed overview which helps the tree species arises from the interaction of focus future research on specific hypotheses all the traits which "pull" the species in dif involving fewer traits. ferent directions with different strengths depending on its foliage characteristics. Biplot of Foliage Quality A cautionary note is in order: like many other multivariate techniques, biplot analysis The biplot of foliage traits alone (Fig. 7) cannot fully represent all Lhe informaLion in a emphasizes that many tr aits implicated in plant complex matrix in only a few dimensional graph. defense have fairly stable interrelationsh{ps The accuracy of the interpretal ions just des with one an other from year to year. For exam cribed hinges on the amount of variance in the ple, the suite of traits involving leaf acidity, original matrix represented in the biplot. Two buffer capacity, total phenolics, and precipit vectors apparently highly positively correlated able tannins is notably stable both in spring may, for example, actually diverge at a ri ght and summer and from year to year. Trees with angle in the third dimension. For this reason more acidic leaves tend to be consistently high biplot analysis can only be an exploratory tech er in total ph enolic concentrations. Foliage nique to provide initial help in puzzling out toughness and water content are generally nega complicated and little known biological situa tively correlated suggesting that these two tions. It must be supplemented by careful scru traits which have usually been- considered sepa tiny of ac tual· correlation coefficients between rately in the entomological literature may traits· and by traditional graphic comparisons of better be combined as a single measure of leaf interrelationships between selected traits. As sclerophylly. Leaf nitrogen concentrations an exploratory technique biplot analysis can remain relatively stable over time; although the only suggest, not test, hypotheses. Given _the correlations are weak, less sc lerophy l lous and large number of interacting traits which may less acidic leaves tend to 1:,e richer in nitro influence gypsy moth host selection it does gen. Time of leafing out varies relatively
CONDENSED TANNINS
AXIS I
• Cary a
PH
WATER
• Frax
Figure 7. Biplot of mean foliage characteristics for 14 tree species in the Lake Hi 11 forest. The two axes account for 70% of the variance in the relativized data matrix. See text for explana- tion of how_to interpret this graph.
76 little among the 14 tree species on Lake Hill; The host trees themselves fal 1 along two actual values have a range of only 8 days in general axes of variation in terms of the trails 1980 and 17 in 1981. measured: 1) coordinated variation in a sy ndrome of traits involving highly intercorrelated The annual pattern in condensed tannin con changes in acidity,· bu-ffer capacity, total centrations is in contrast to the generally phenolics, precipitable tannins, and nitrogen stable correlations among these other foliage and 2) in another syndrome involving to ughness, characteristics. Only total phenol ics and water content, and condensed tannins. The dis pricipitable tannins approach condensed tannins tribution of tree species in the biplot suggests in terms of variation between trees. Unlike two relatively distinct modes of leaf phenolic these other highly variable traits, condensed metabolism. Some trees like Populus grandi.den tannins also showed an interesting pattern of tata, •Ostrya virginiana, and Amelanchier are set annual variation. During the outbreak year 1979 apart by their relatively high concentrations of and 1980 the pattern of condensed tannin levels condensed tannins while others like Acer rubrum, between tree species was similar and condensed �- saccharum, Carya cordiformis, and �s tannins were higher in sclerophyllous leaves rubra have relatively low condensed tannin con (Fig. 7). In 1981 as the gypsy moth population centrations but high_ to tal phenolics and pr eci reached innocuous levels there was a shift in pitable tannins. The general validity of these this pattern of condensed tannins in the spring trends and their possible relevance to different leaves of the different host trees. modes of defense against folivores merit further consideration.
GYPSY MOTH PREFERENCES
HOST FOLIAGE TRAITS
� PRECIPITABLE .... -:,� TANNINS ,,. 0 <9 I 0 I I u'.o / / ,.,,, u'.o u' / 9 / BUFFER '0'},!'»'>I I CAPACITY: 9 '0;,., Spring 1980, 1981 -...... �qo Spring 1981 � TOTAL PHENOLIC$
Arub •
AXIS I • Carya BpaP. l!> g a� '\ 9'i-. --- OJ '<,9� \�Q, ,.._ � QI>, '"1,, ...._OJ Ppen QI<- 19 <:,9< o, '°�· iS'l'" q \�Q, -... -$ " • ��QI ,s I) � "> ,;, 0 l' q' Ap�n• co,._ ,,. ">19 ,9, PH WATER � co• -'slut •-':) 19 ,. 9;, 0q 0" Frax /' � ,,.
77 Relationships of Gypsy Moth Activity to Foliage in this exploratory analysis ra ises a number of Quality hypotheses which require experimental tests. As a framework for continuing studies I have sum A second biplot (Fig. 8) adding the 1979 marized these hypotheses in a tentative model of through 1982 larval and egg selecLiviLy dala for gypsy moth host selection which is explained the 14· tree species to that on foliage qualiLy below. The model has two main tenets: 1) that was run Lo explore Lhe correlaLions beLween dispersing gypsy moth larvae preferentially Lhese plant trails and the palterns of gypsy settle on trees with higher sugar:tannin ratios moth host preference. This biplol relains the in Lheir young foliage and 2) that stress underlying relaLionships of Lhe plant Lraits induced variations in the sugar and tannin con among themselves. The two major axes of planl cenlrations of young foliage can lead to year to variation involving condensed Lannins versus year variation in a tree's susceptibility to precipitable Lannins and Lotal phenolics are atl�ck by gypsy molh. still apparent. PrecipiLable tannins, however, appear to be somewhal more associaLed with gypsy On a regional ba sis, the forests most moth preferences Lhan Lhe other plant traits susc ept ib le to infest at ion by gypsy moth occur involved in Lhe same sy ndrome of variaLion. on more arid, nutrient-poor sites (Houston and Similarly sclerophylly, parLicula·rly in the Valentine 1977) where sclerophyllous tree summer, appears more associaLed with gypsy molh species are mosl frequently found. Within these preferenc�s Lhan the concentrations of condensed susceptible forests, host trees with more · tannins. Leaf pH, buffer ca pacily, total pheno sclerophyllous malure foliage appear from the lics, and nilrogen appear to be only weakly or present analysis to be preferentially attacked not at all relaLed to larval hosl preferences. by gypsy moth larvae. This positive association Although leaf phenology varies relatively liLLle between sclerophylly and Lhe host preferences of among hosts there is some indication Lhat larvae gypsy moth provides a useful clue to the may prefer hosls that leaf out later. In possible basis of host selection by this· general Lhis biplol suggests that dispersing polyphagous folivore. gypsy moth larvae prefer trees Lhat later in the season wil.l have Lough, dry leaf Lissues and Sclerophylly has long been associated with that these preferences may be modified in part both arid, nutrient-poor environments and low by leaf concentral ions of condensed and, Lo a rates of ca rbon fixation (Smal 1 1973; Seddon lesser degree, precipitable tannins in the 1974). Sclerophylly is believed to improve spring and early summer. survival during the relatively frequent pe riods of stress typical of such habitats but at the The actual disLribution of hosl Lr ees in expense of reduced productivity when conditions the biplot also indicates that gypsy moth host are favorable. In general we might expect the preferences are not simply and entirely deler less productive sclerophyllous tree species to mined by the observed plant traits, but are also have more limited resources to allocate to influenced by trails not included in this dala defenses against folivores. In addition the set. For example, the two ·most consistently leaves of sclerophyllous trees tend to develop preferred hosts, Quercus rubra and Populus more slowly in the spring (Federer 1976). Thus grandidentata, represent the two different syn gypsy moth larvae feeding on the spring and dromes of leaf Lypes in the biplot of planl early summer foliage of a sclerophyllous host traits alone (Fig. 7). It is not immediately may enjoy a relatively longer period of access apparent what these two hosts have in common to higher quality, immature foliage; models of that leads to Lheir being attacked preferen gypsy moth larval development (Valentine and tially by gypsy moth. Quercus rubra in pa rLi Talerico 1980) suggesl that selection of hosts cular is singled out as highly preferred from offering even slightly longer periods of favor among trees like Acer saccharum, A. rubrum, and able foliage availability can improve larval Carya cordiformis �hich it is very similar in survival and fecundity. Host trees that later terms of the observed foliage traits. The in the summer are most sclerophyllous and then biplot (Fig. 8) thus suggests that factors other apparently lowest in foliage quality may actual than those tr aditionally supposed to determine ly have provided the longest periods of immature susceptibility to herbivore attack among Lrees foliage mosl favorable for larval development. are important in gypsy molh host selection. How then might dispersing larvae recognize the sclerophyl lous hosts offering such a favor A Model of the Gypsy Moth-Host Tree InteracLion. able opportunity for larval development? During larval dispersal al l Lhe available foliage is The diverse observational data analyzed relatively tender and moist and there are no here have al 1 owed elimination of certain plant consistent correlaLions beLween the sclerophylly traits as uncorrelated with gypsy moth prefer of young and old foliage; some cue other than ences and therefore unlikely to be important sclerophylly itself must be used by Lhe dispers factors in host selection. On the other hand, ing larvae. Soluble sugars, especially sucrose, although no clear and simple explanation of host which are known to be an important feeding selection has emerged, certain plant traits are stimulus in very many insects (Dethier 1982) implicated as factors involved in delermining seem the most li kely proximate control on ho st the probability that a tree will be attacked by selection by dispersing gypsy moth larvae. gypsy moth. The recognition of such key traits Transporl of so luble sugars to developing tree
78 leaves is high (Dickson and Larson 1981) and vores. When all tr ees have been weakened by a would continue longer in the more slowly deve stress such as a severe drought, limited re loping sclerophyllous species. Depending in serves of photosynthate may be diverted from part on the relative timing of eclosion and bud defense to other mor,e es1,ential functions. The break, dispersing larvae will thus be more polyphagous and widely dispersed gypsy moth likely to encounter sugar-rich foliage on the larvae in the subsequent spring then would find slower-developing, sclerophyllous hosts. Fur an unusually high number of favorable ho st trees ther study is necessary to test this hypothesis apparently high in sugar content for lack of that soluble sugar concentrations in host masking tannins. Larval success on these foliage during dispersal actually determine host drought stressed trees would be relatively high selection. and de foliation would thus reduce the recovery of �hotosynthate resources even if climatic Further investigations should also consider conditions were again favorable. Gradually the possible interaction between soluble sugar enhanced defenses, probably aided by predator and tannin concentrations. Dethier (1982) has and pathogen aLLacks on gypsy moth (Doane and reported that although gypsy moth larvae cannot McManus 1981), would reduce the larval numbers sense tannic acid � �• this hydrolyzable to low enough level·s for the normal pa tterns of tannin does suppress the larvae's electrophysio leaf development and defense to prevail. The logical response to sugars. Perhaps tannins endemic gypsy moth population then would again alter the apparent levels of soluble sugars in preferentially attack the more sclerophyllous the young foliage ..available Lo dispersing hosts which are generally less able ,to maintain larvae. If this is the case we may hypothesize effective tannin-based defenses and have the that gypsy moth preferentially attack trees that longer leaf development times favorable to gypsy have high sugar:tannin ratios in their young moth larval success. foliage. The model proposed here can potentially Tannins present in young foliage during explain two key aspects of the gypsy moth-host larval development may also influence year to interaction: l) the basis of host selection and year trends in the numbers of dispersing la rvae 2) the occurrence of periodic outbreaks. Addi that potentially settle on different host tional observational and experimental tests are trees. Because of the limited vagil ity of the necessary to test its Yalidity. later instars and female adul'ts of the gypsy moth (Doane and McManus 1981; Wallner, this volume), any reduction in, foliage quality on ACKNOWLEDGEMENTS attacked trees will act Lo limit larval su ccess on that tree and thereby reduce the local den Many people have worked with me on this sity of dispersing larvae the next year. Dat a project. I want to especially thank Luc Jobin, for lepidopteran larvae including the gypsy moth Yves Mauffette, and the following laboratory and (Feeny 1968; Karasov and Satarova 1973) have field assistants without whose help the pr oject shown that tannins can reduce foliage digesti would have been impossible: Claude Blais, bility. High tannin levels in the young foliage Patrick Blais, Pierre-Alain Blais, Anne Bruneau, critical for larval development can thus reduce Lyne Cossette, Susan Dudley, Ron Kara, Josee egg deposition by fe male larvae on these hosts. LaPierre, Dorothy Luk, Steve McCanny, David If in addition elevated levels of tannins can be Moscovitz, Danielle Pouliot, Chantal Reid, maintained in young foliage the next sp ring, Benoit St. Jacques, and Georgette Verebely. then the preference of dispersing larvae for the host may be further reduced by suppression of A long term project like this would be dif the neurophysiological response Lo sugar stimuli fi�ult without the continued support of the as hypothesized above. The maintenance of high staff al the McGill University Gault Estate and tann-in levels, however, depends on the availa permission from the Gault Estate Board to work bility of photosynthetic reserves which will in the Scientific Reserve at Mont St. Hilaire. tend Lo be lower in more sclerophyllous hosts. Funding from the Canadian National Sportsmen's Sclerophyllous hosts will therefore tend to be Fund, the Laurentian Forest Res.earch Centre of more susceptible to susl ained infest ations of Environmental Canada, Lhe Natural Sciences and gypsy moth larvae from year Lo year. Moreover, Engineering Research Council of Canada, the productive hosts better ahle Lo either constitu McGill University Computing Centre, and Cornell 'tively or facullatively maintain high levels of University have supported thi s research. I am tannins in their spring foliage could actually grateful to Brian Chabot, Paul Feeny, and shift large numbers of dispersing larvae to the Deborah Rabinowitz for pr oviding a productive less well-defended, sclerophyllous hosts. IL is and stimulating environment during my stay at through mechanisms such as these that the gypsy Cornell. Marcia Waterway, Bob Hagen, and Jean moth preference for more sclerophyllous hosts Chabot provided helpful critiques of the draft may be expressed and maintained. manuscript. Cathy Hughes prepared Lhe ca mera ready CO!>Y. This model is in accord with the recent revival and elaboration (Haukioja 1980) of the idea that climatic stress on host plants can release a serious outbreak of insect herbi-
79 LITERATURE CITED Forbush, Edward H; Fernald, Charles H. The Gypsy Moth Porthetria despar (Linn.). Wright Barbosa, Pedro; Capinera, John L. The influence and Potter Boston; 1896. 495p. of food on developmental characLerisLics of Fox, Laurel�; McCauley, B.J. Insecl grazing on Lhe gypsy moth, LymanLria di spar (L.). Can. Eucalyptus in response Lo variaLion in leaf J. Zool. 55: 1424-1429; 1977. tannins and nitrogen. Oecologia. 29: 145- Barbosa, Pedro; GreenblaLL, J.; WiLhers, W; 162; 1977. Cranshaw, W; HarringLon, E.A. Host-pl anl Fuluyma, Douglas J. Evolutionary interactions preferences and Lheir induction in larvae of among herbivorous insects and plants. Chap the gypsy moLh, LymanLria dispar. EnL. Exp. ter 10 in Futuyma, Douglas J; Slatkin, el Appl. 26: 180-188; 1979. Montgomerie (eds.) Coevolution. Cambridge, Barbosa, P; Greenblatt, J. SuitahiliLy, digesL Mass.: Sinauer; 1983. ability and assi�ilaLion of various hosl Gabriel, K.R. The biplot graphic display of plants ·of Lhe gypsy moth, Lymantria dispar matrices with application to principal compo L. Oecolgia. 43: 111-119; 1979 . nent analysis. BiomeLrika. 58: 453-467; Beck, S.D.; Reese, J.C. lnsect-plant inter- 1971. actions: n11tr1L1on and meLabolism. Rec. Goldstein, Judith L.; Swain, T. The inh ibition Adv. Phylochem. 10: 41-92; 1976. of enzymes by tannins. Phytochem. 4: 185- Bernays, E.A. Tannins: an alternaLive view- 192; 1965. point. Ent. Exp. et Appl. 24: 44-53; 1978. Govindarajan, V.S.; Mathew, A.G. Anthocyanidins Berenbaum, May.· Adaptive significance of midgut from leucoanLhocyanidins. Phytochemistry. 4: pH· in larval Lepidoptera. Amer. Natur. 115: 985-988; 1965. 138-146; 1980. GrandLner, Miroslav M. La vegetation foresLiere Bouchard, Andre; Maycock, Paul F. Les forets du Quebec meridional. Les presses de l 'Uni decidues et mixLes de la region Appalachienne versite Laval, Quebec: 1966. 216p. du Sud Quebecois. Natur. Can. 105: 383-4I5; Gyorfi, Janos. Contributions to the nutritional 1978. biology of Lhe gypsy moth (Lymantria dispar Bradstreet, R.B. The Kjeldahl method for or- L.). Erdeszeti Kutatasok. 56 (1-3): 279-291; ganic nitrogen. Academic Press, N.Y.; 1965. 1960. In Hungarian. 239p. Haukioja, Errki; Niemela, Pekka. Birch leaves Brown, G. Stuarl. The gypsy moth, PortheLria as a resource for herbivores: seasonal occur dispar L. A threal to Ontario horticulture rence of increased resistance in foliage and forestry. Proc. Entom. Soc. Ontario. 98: after mechanical damage of adj acenl leaves. 12-15; 1967. Oecologia. 39: 151-159; 1979. Cardinal, J. A. Lutle centre la spongieuse Haukioja, Errki. On Lhe role of plant defenses Porthetria dispar L. (Lepidoptera: Lyman- in the fluctuation of herbivore populations. tridae) du Quebec. Phytoprotection. 48: 91- Oikos. 35: 202-213; 1980. 100; 1967. Horowitz, W. (ed.). Official Methods of Curtis, J.T. The vegeLaLion of Wisconsin. Analyses of the Association of Official University of Wisconsin Press: Madison; Anal ylical Chemists, Washing ton, D.C.: AOAC. 1959. 657p. 11th edition; 1970. Dethier, Vincent. Mechanism of host-plant Horvath, P.J. The nutritional and ecological recognition. Ent. exp; and appl. 31: 49-56; significance of Acer-tannins and related 1982. polyphenols, M.S. thesis, Cornell University: Dickson, Richard E.; Larson, Philip R. 14c llhac a, New York; 1981. fixation, metabolic labeling palterns, and Hough, Judith A; Pimenlal, David. Influence of translocation profiles during leaf develop host foliage on developmenl, survival, and ment in Populus deltoides, Planta. 152: fecundity of the gypsy moth. Envir. Entom. 461-470. 7: 97-102; 1978. Doane, Charles C.; McManus, Michael L. The Houston, D.R.; Valentine, H.T. Comparing and gypsy molh: research toward integrated pesL predicting forest sLand susceptibility to managemenL. Washington, D.C.: Forest Ser gypsy moth. Can. J, For. Res. 7: 447-461; vice, U.S. Department of Agriculture, Tech 1977. nical Bulletin. 1584; 1981. 757p. Jobin, Luc. Results of aerial treatment with Federer, C.A. Differing diffusive resistance Dimilin and Jlacillus thuringiensis for gypsy and leaf development may cause differing moth (Lymantria dispar [L.]) conLrol in transpiration among har.dwoods in spring. Quebec. Can. For. Serv. Res, Notes 2(3):. Forest Sci, 22: 359-364; 1�76. 18-20; 1982. Feeny, Paul. Effecl of oak leaf Lann ins on Karasov, V.S.; Satarova, T.I. The effecLs of larval growth of the winter molh Operophlera tannins from Salix, Populus, and Quercus on brumaLa. J. Insect Physiol. 14: 805-817; Lhe developmenl of Porthetria dispar. 1968. Zakhyst Roslyn. 17: 44-46; 1973. In Feeny, Paul. Seasonal changes in oak leaf tan- Ukrainian. nins and nuLrients as a cause of spring feed Kurir, Anton. Die Frasspflanzen des Schwamm ing by winter moth caLerpil lars. Ecology. spinners (Lymantria di spar ·L.). Z. ang. 51: 565-581; 1970. enl. 34: 543-586; 1953. Feeny, Paul. Plant apparency and chemical defense. Rec. Adv. Phytochem. 10: 1-40; 1976.
80 Lachance, D.; Benoit, P; Bonneau, G; Desaul Martineau, R.; Lavallee, A.; Beique, R.; niers, R; Laflannne, G. lnsectes et maladies Davidson, J.G. lnsectes et maladies des des arbres Quebec 1978. Foret Conservation. arbres Quebec - 1975. Foret Conservation. 45(4): 1-23; 1979. 42(4): 1-23; 1976 •. Lachance, D.; Benoit, P.; Bonneau, G.; Laflamme, Mason, Conrad, J.; McManus, Michael L. Larval G. Insectes et maladies des arbres Quebec - dispersal of the gypsy moth, pp. 161-202 in 1979. Foret ConservaLion. 46(10): 1-23; Doane, Charles C.; McManus, M.L. (eds.). The 1980. Gypsy Moth: Research toward integrated pest Lachance, D.; BenoiL, P.; Bonneau, G.; Laflamme, management. Washington, D.C.: U.S. Depart G. Insectes et maladies des arbres Quebec - ment of Agriculture, Forest Service. Tech 1980. Foret Conservation. 47(9): 1-19; 1981. nical Bulletin 1584; 1981; 757p. Lavallee, A.; Beno'it, P.; Beique, R.; Davidson, Mattson, W.J. Jr. Herbivory in relation to J.-G. Insectes et maladies des arbres: plant nitrogen content. Ann. Rev, Ecol. Quebec - 1976. Foret Conservation. 43(4): Syst. 11: 119-161; 1980. 1-23; 1977. Maycock, PauL F. Botanical studies on Mont Lavallee, A.; Benoit, P.; Lachance, D.; Beique, St. Hilaire, Rauville County, Quebec. Can. R.; Davidson, J.-G. Insectes el maladies des J. Bot. 1293-1325;· 1961. arbres Quebec 1977. Foret Conservation. 45 McNeil, S.; Southwood, T.R.E. The role of (1): 1-23; 1978. nitrogen in the development of insect/plant Lechowicz, Martin J. The sampling character- relationships, pp. 77-98 in Harborne, J.B. istics of electivity indices. Oecologia. 52: (ed.). Biochemical aspects of plant and 22-30; 1982. animal coevolution. Proc. Phytoc·hem. Soc. Lechowicz, Martin J.; Jobin, Luc. Estimating Europe 15; 1978. the susceptibility of Lree species to attack Middleton, K.R. New Nessler reagent and its use by the gypsy moth, Lymantria di spar. Ecol. in the direct Nesslerization of Kjeldahl Entomol., in press; 1983. digests. J. Appl. Chem. 10: 281-286; 1960. Leonard, David E. Feeding rhythm in larvae of Moran, Nancy; Hamilton, W.D. Low nutritive the gypsy moth. J. Econ. Entom. 63: 1454- quality as defense against herbivores. J. 1457; 1970. Theor. Biol, 86: 247-254; 1980. Levin, Donald A. The chemical defenses of Mosher, F.H. Food plants of the gypsy moth in plants to pathogens and herbivores. Ann. America. Bulletin. No. 250. Washington, Rev. Ecol. Syst. 7: 121-159; 1976. D.C.: U.S. Department of Agriculture, Bureau Little, Elbert Jr. Checklist of United States of Entomology. 1915: 39p. Trees Agriculture Handbook No. 541. Washing Pace, G.M.; Mackown, C.T.; Volk, R.J, Mini- ton, D.C.: U.S. Department of Agriculture, mizing nitrate reduction during Kjeldahl Forest Service; 1979. 375p. digestion of plant tissue extracts and stem Loomis, W.D.; Battaile, J. Plant phenolic com exudates. Plant Physiol. 69: 32-36; 1982,. pounds and the isolation of plant enzymes. Phillips, W.F. An evaluation of the relative Phytochem. 5: 423-438; 1966. importance index to the stu dy of forest vege Madrid, Francisco J, Laboratory and field tation on Mont St. Hilaire, Quebec, Canada. studies on Lymantria dispar L. (Lepidoptera: MSc. thesis, McGill University, Montreal, Lymantriidae) in Quebec. Ph.D. thesis, Quebec; 1972. Macdonald College of McGill University, Polunin, Oleg; Everard, Barbara. Trees and Montreal, Quebec; 1979. 190p. bushes of Europe. Oxford University Press, Madrid, F.J.; Stewart, R.K. The influence of London; 1976. 208p. some environmental factors on the development Rhoades, David F.; Cates, Rex G. Towards a and behavior of the gypsy moth (Lymantria general theory of pl ant · ant iherbi vore dispar L.) in Quebec. Ann. Soc, ent. �hemistry. Rec. Adv. Phytochem. 10: 168-213; Quebec. 26: 191-211; 1981. 1976. Marigo, G. Sur une methode de fractionnement et Ribereau-Gayon, P. Plant Phenolics, Hafner, d'estimation des composes phenoliques chez New York; 1972. 254p. les vegetaux. Analusis. 2: 106-110. Richard, Pierre. Histoire postglaciare de la Martin, Joan S.; Martin, Michael M. Tannin foret decidue du Quebec Naturaliste assays in ecological studies: lack of corre Canadien. 104: 135-141; 1977. · lation between phenolics, proanthocyanidins Rosenblatt, M.; Peluso, .J.V. Determination of and protein-precipitating constituents in tannins by photocolorimeter. Assoc. Off. mature foliage of six oak species. Agric. Chum. 24: 170-181; 1941. Oecologia. 54: 205-211; 1982. Rossiter, Mary C. Enhanced pathogen resistance Mart in, Joan S.; Mart in, Michae 1 M. Tannin ' as a result of a host shift by the gypsy assays in ecological studies. Precipitation moth. Second International Congress of of ribulose-1, S-bisphosphate carboxylase/ Systematic and Evolutionary Biology Ab- oxygenase by tannic acid, quebracho, and oak stracts: p328; University of British foliage extracts. J. Chem. Ecol. 9: 285-294; Columbia, Vancouver; 1980. 1983. Rouse, W.R.; Wilson, R,G. Time and space varia Martineau, R.; Lavallee, A.; Beique, R.; tion in the radiant energy fluxes over slop Davidson, J.G. lnsectes et maladies des ing forested terrain and their influence on arbres, region de Quebec 1974. Foret seasonal heat and water balance at a middle Conservation. 41(4): 1-23; 1975. latitude site. Geografiska Annaler 51A: 160-175.
81 Schlicting, Carl D. Wound-induced responses of Van Tuil, H.D.W. Organic salts in plants· in forest trees Lo gypsy moth herbivory. M.S. relation to nutrition and growth. Agricul thesis, Stale University of New York at Stony tural Res. Rep. No: 657. Wageningen, Nether Brook; 1980. lands: PUDOC Centre for Agricultural Publica Schmidt, Lea. The influP.nce of food on the tions and Doc�mentation; 1965. 83p. development of gypsy moth (Lymantria dispar Wallner, W.E.; Walton, G.S. Host defoliation: a L.). Annales Exper. Foreslicis Jugoslavia possible determinant of gypsy moth population Glasnik. 12: 105-166; 1956. In Serbo- quality. Ann. Entom. Soc. Amer. 72: 62-67; Croatian. 1979. Schultz, Jack C.; Baldwin, I.T. Oak leaf qua Walther, A. Forest ecological studies of the lity declines in response to defoliation by l1onteregian Hills of southern Quebec. MSc. gypsy moth larvae. Science. 217: 149-151; thesis, McGill University, Montreal, Quebec; 1980. 1963. 234p. Scriber, J.M.; Feeny, P·. Growth of herbivorous Weseloh, R.M. Relationships between different caterpillars in relation to fe eding speciali sampling procedures for gypsy moth, Por zation and Lo the growth form of their food thetria dispar (Lepidoptera: Lymantriidae) plants. Ecology. 60: 829-850; 1979. and its natural enemies. Can. Ent. 106: Scriber, J.M.; Slansky, F. Jr. The nutritional 225-231; 1974. ecology of immature insects. Ann. Rev. Wilson, R.G. Topographic influences on a fo rest Entomol. 26: 183-212; 1981. microclimate. Climatological Research Seal, Hilary. Multivariate statistical analysis Series No. 5; Montreal, Quebec: McGill for biologists, Methuen, London; 1964; 207p. University, Department of Geography; 1970. Seddon, G. Xerophytes, xeromorphs and sclero- 109p. phylls: the history of some concepts in ecology. Biol. J. Linn. Soc. 6: 65-87; 1974. Seigler, D.; Price, P.W. Secondary compounds ·in plants: primary functions. Amer. Natur. 110: 101-105; 1976. Small, Ernest. Xeromorphy in plants as a pos- sible basis for migration between arid and nutritionally-deficient environments. Bot. Notiser. 126: 534-539; 1973. Swain, T. Phenolics in the environment. Rec. Adv. Phylochem. 12: 617-640; 1978. Swain, T. Tannins and lignins. pp 657-682 in Rosenthal, G.A.; Janzen, D.H. (eds.). Herbi vores: their interaction with secondary plant metabolites. Academic Press, N.Y.; 1979. Swain, T.; Hillis, W.E. • The phenolic consti- tuents of Prunus domestica. I. The quantita tive analysis of phenolic consliluerits. J. Sci. Food Agric. 10: 63-68; 1959. Tempel, Alice S. Tannin measuring techniques: a review. J. Chem. Ecol. 8: 1289-1298·; 1982. Tietz, Harrison M. An index to the described life histories, early· stages and hosts of the macrolepidoptera of the co ntinental United States and Canada. Sarasota, Florida: Allyn Museum of Entomology; 1972. 104lp. in 2 volumes. Valentine, Harry T.; Talerico, Robert L. Gypsy moth larval growth and consumption on red oak. For. Sci. 26: 599-605; 1980. Valentine, H.T.; Wallner, W.E.; Wargo, P.M. Nutritional changes in host foliage during and after defoliation, and their relation to the weight of gypsy moth pupae. Oecologia: · in press; 1983. Vanderploeg, H.A.; Scavia, D. Two electivity indices for feeding with special reference to zooplankton grazing. J. Fish. Res. Board Can. 36: 362-365; 1979a.· Vanderploeg, H.A.; Scavia, D. Calculation and use of selectivity coefficients of feeding: zooplankton grazing. Ecol. Modelling. 7:· 135-149; 1979b.
82