Phylogeny and Ecological Evolution of Gall-Inducing Sawflies (Hymenoptera: Tenthredinidae)

University of Joensuu, PhD Dissertations in Biology

No:6

Phylogeny and ecological evolution of gall-inducing sawflies (Hymenoptera: Tenthredinidae)

by Tommi Nyman

Joensuu 2000 Nyman, Tommi Phylogeny and ecological evolution of gall-inducing sawflies (Hymenoptera: Tenthredini- dae). – University of Joensuu, 2000, 92 pp. University of Joensuu, PhD Dissertations in Biology, n:o 6. ISSN 1457-2486

ISBN 951-708-962-7

Key words: coevolution, convergent evolution, extended phenotypes, gall morphology, gradualism, insect-plant interactions, Nematinae, phylogeny, Salix, sawfly, speciation

The nematine sawflies (Hymenoptera: Tenthredinidae) that induce galls on willows (Salix spp.) form one of the most abundant and speciose herbivore groups in the Holarctic region. The purpose of this thesis was to study the evolutionary history of the nematine gallers. The three main questions addressed were: (1) what is the sequence in which different gall types have evolved in the nematines; (2) how has the utilization of host plants evolved in the group; and (3) why has gall induction evolved in the nematines? The first question was studied by reconstructing the phylogeny of representative nematine species by using enzyme electrophoresis and DNA sequencing. According to the results, species that induce true closed galls evolved from species that induce leaf folds or rolls. Thereafter, several different gall types evolved gradually, and the evolution of each gall type was followed by a radiation to new host species. The phylogenies show that gall morphology is mainly determined by the insects, i.e., the gall represents an extended phenotype of the galler. The two phylogenies were also used to study the second question. The results show that parallel cladogenesis between willows and the gallers is not a sufficient explanation for the current host associations, since many willow species have been colonized multiple times by nematine species representing different gall types. Likewise, it seems that the overall chemical similarity of hosts does not determine the direction of host shifts. Consequently, other explanations, such as host phenology, geographical distribution, or habitat, are probably needed for explaining the patterns of host use in the nematine sawflies. Similar results were obtained on a smaller scale in an enzyme electrophoretic study of the species that induce bud galls. The third question was studied by comparing the chemical properties of sawfly galls to the chemical properties of the ungalled tissues of the willow hosts. The galls were found to contain comparatively low levels of most phenolic defense compounds that are found in other willow tissues. Thus, it appears that the galls are nutritionally beneficial for the sawfly larvae. The changes in chemistry occur in a highly coordinated pattern in all studied galler- host pairs, which suggests that the sawflies are able to manipulate the phenolic biosynthesis in their hosts. The end result is a striking convergence of the chemical properties of the galls both within and between host species.

Tommi Nyman, Department of Biology, University of Joensuu, P.O. Box 111, FIN-80101 Joensuu, Finland CONTENTS

LIST OF ORIGINAL PUBLICATIONS

1. INTRODUCTION 7 1.1. Gall-inducing insects 7 1.2. Purpose of this study 8

2. THE STUDY SYSTEM 8 2.1. Willows 8 2.2. The nematine gallers 9

3. THE EVOLUTION OF DIFFERENT GALL TYPES 11 3.1. Hypotheses 11 3.2. Allozyme results 12 3.3. DNA results 12 3.4. Combined evidence 14 3.5. Discussion 15

4. THE EVOLUTION OF HOST PLANT USE 16 4.1. Hypotheses 16 4.2. Host-plant relationships in the nematines 17 4.3. Euura mucronata: a polyphage or a sibling species complex? 18 4.4. Discussion 19

5. THE CHEMICAL ECOLOGY OF NEMATINE SAWFLIES 22 5.1. The Nutrition Hypothesis and previous studies 22 5.2. The chemical properties of Pontania galls on willows 23 5.3. Discussion 23

6. CONCLUDING REMARKS 26

ACKNOWLEDGEMENTS 28

REFERENCES 29 LIST OF ORIGINAL PUBLICATIONS

This thesis is based on the following articles, referred to in the text by their Roman numerals.

I Nyman, T., Roininen, H. & Vuorinen, J. A. 1998. Evolution of different gall types in willow-feeding sawflies (Hymenoptera: Tenthredinidae). Evolution 52: 465-474.

II Nyman, T., Widmer, A. & Roininen, H. 2000. Evolution of gall morphology and host- plant relationships in willow-feeding sawflies (Hymenoptera: Tenthredinidae). Evolu- tion 54: 526-533.

III Nyman, T. The willow bud galler Euura mucronata Hartig (Hymenoptera: Tenthredini- dae): one polyphage or many monophages? Manuscript.

IV Nyman, T. & Julkunen-Tiitto, R. 2000. Manipulation of the phenolic chemistry of willows by gall-inducing sawflies. Proc. Natl. Acad. Sci. USA 97: 13184-13187.

Some unpublished results are also presented. 7

1. INTRODUCTION nisms include mechanical damage, plant hormone analogs, and genetic manipulation 1.1. Gall-inducing insects (Hori 1992, Price 1992). Genetic manipulation is involved in the process of gall induc- Meyer (1987) defines galls as “all manifes- tion by the bacterium Agrobacterium tume- tations of growth, whether positive or nega- faciens. The bacterium inserts a portion of a tive, and of abnormal differentiation induced Ti plasmid into the genome of some of the on a plant by animal and plant parasites”. cells of its host, which eventually leads to Usually galls can be seen as abnormal abnormal growth (Davey et al. 1994). Al- growth caused by an increase in the number though it has been suggested that such ma- (hyperplasia) or size (hypertrophy) of plant nipulation may also occur in the case of in- cells at the site of the gall (Meyer 1987). sect gallers (Cornell 1983), no evidence for Galls can be induced on virtually all parts of such a mechanism has been found (Price plants, and their complexity varies from 1992). It is likely that the mechanism of gall simple pouches to highly structured organs induction varies between galler taxa, but with specifically differentiated tissues (Dre- currently it would seem that the production ger-Jauffret & Shorthouse 1992). of plant hormone analogs represents the The ability to induce galls has evolved most likely explanation (Hori 1992, Roskam convergently in many taxa, ranging from 1992). For example, McCalla et al. (1962) microbes and fungi to nematodes and arthro- found uric acid, glutamic acid, and adenine pods (Meyer 1987). In insects alone, gall derivatives in the female accessory glands of induction has evolved in at least the follow- the sawfly galler Pontania pacifica Marlatt. ing seven orders: Thysanoptera, Hemiptera, These compounds are closely related to cy- Homoptera, Lepidoptera, Coleoptera, Dip- tokinins, which are natural growth- tera, and Hymenoptera (Meyer 1987, Dre- promoting hormones in plants (Hartley ger-Jauffret & Shorthouse 1992). The num- 1992). High concentrations of cytokinins ber of independent origins of gall induction have been found in galls induced by Pon- in insects is, however, considerably higher, tania proxima Lepeletier, but apparently since within most of these orders there have they are produced by the plant (Leitch 1994). been several separate lineages leading to the Furthermore, studies on other galler/plant galling habit (Dreger-Jauffret & Shorthouse combinations have found that the relative 1992, Roskam 1992). concentrations of cytokinins are not neces- The fossil record of plant galls shows sarily higher in all galls (Mapes & Davies that unknown gall-inducing insects were 1998). Mapes and Davies (1998) found cy- already present during the Late Carbonifer- tokinins and IAA in larvae of the dipteran ous period, over 300 million years ago (La- stem galler Eurosta solidaginis Fitch, but it bandeira & Phillips 1996). Numerous galls is not clear whether the larvae produce these have been found in younger plant fossils, but hormones de novo, or sequester them from in most cases the identity of the gall inducer their diets. remains unknown (Larew 1992, Scott et al. Different galler taxa tend to specialize in 1994). However, it is evident that several different plant taxa, but some groups of gall-inducing insect groups were already in plants are attacked more than others (Ro- existence at the time of the initial radiation skam 1992). For example, relatively few of angiosperms during the mid-Cretaceous; insect species induce galls on ferns and possible taxa include aphids, midges, and gymnosperms (Roskam 1992). In contrast, cynipid wasps (Scott et al. 1994). over 90% of galls occur on dicots, and Although gallers are common, the within dicots the vast majority of galls occur mechanism by which insects induce galls on Fagaceae, Asteraceae, and Rosaceae remains largely unknown. Possible mecha- (Abrahamson & Weis 1987). In Britain, 8

Salix (Salicaceae) and Quercus (Fagaceae) two questions were studied by reconstructing species support the largest numbers of insect the phylogeny of representative nematines gallers (Kennedy & Southwood 1984). using allozyme electrophoresis and DNA A common observation is that gall- sequencing, and the last one was studied by inducing insect species are strictly host- analyzing the chemical properties of sawfly specific (Meyer 1987, Dreger-Jauffret & galls using high-performance liquid chro- Shorthouse 1992). This is probably a result matography. of the tight link between the gallers and their hosts, because gall induction may be possible only during a short period of time during 2. THE STUDY SYSTEM the growing season, and the available plant species (or even individuals or clones within 2.1. Willows species) may differ in their reactions to the gall-inducing stimuli (Smith 1970, Price Willows (Salix spp.) form one of the taxo- 1992, Whitham 1992, Craig et al. 1994, nomically and ecologically most diverse Peries 1994, Csóka 1994, Kokkonen 2000). plant genera in the Northern Hemisphere It is important to note that the degree of host (Argus 1973). The oldest willow fossils have specificity may have been underestimated, been found from North American deposits since galler species are often extremely dif- dating from the early Eocene, ca. 55-65 mil- ficult to identify by morphological methods, lion years ago (Collinson 1992). However, and many taxa have not been studied in Skvortsov (1968, 1999) argued that the depth (Smith 1968, Harris 1994). group originated in the warm temperate re- Obviously, the repeated evolution of the gion or the subtropics, with the main radia- galling habit in highly divergent insect taxa tion occurring later in the temperate region presents an excellent opportunity for study- and the arctic (see also Dorn 1976). There ing convergent evolution. Furthermore, the are currently over 350 willow species, most existence of various gall types makes it pos- of which occur in wet (riparian) habitats, or sible to study how gall morphology evolves, as colonizers in the early successional stages and whether the morphological and chemical of vegetation (Skvortsov 1968, 1999; Argus features of the galls are determined by the 1973). Most species are medium-sized host plants or by the gallers. Finally, the ex- bushes, but the size range is from creeping istence of groups of strictly monophagous species, less than one centimeter in height, to gallers on various host species facilitates large trees. comparative studies of the speciation pat- The taxonomy of willows is still debated. terns in the host and galler taxa. This is partly due to the extreme morphological variation in many species (Skvortsov 1.2. Purpose of this study 1968, 1999). A further complication is caused by hybridization between species, The purpose of this thesis was to study vari- although this is now considered to be less ous questions relating to the ecology and frequent than previously thought (Skvortsov evolution of galling insects by using the 1968, 1999; Argus 1973, Rechinger 1992). It nematine sawflies (Hymenoptera: Tenthre- is, however, clear that hybridization occurs dinidae) that induce galls on willows as a (Rechinger 1992, Brunsfeld et al. 1992, Har- model system. The main questions addressed dig et al. 2000), and in some areas hybrid in the thesis are: (1) what is the sequence in swarms can be found (Meikle 1992). Many which different gall types have evolved in species are polyploid (Dorn 1976, Argus the group; (2) how has the utilization of host 1973, Hämet-Ahti et al. 1998), but appar- plants evolved in the group; and (3) why has ently the possibility and extent of allopoly- gall induction evolved in the group? The first 9 ploid speciation in willows has not been insects (Ayres et al. 1997, Mutikainen et al. given serious study. 2000). Recently, the problems of willow taxon- It is probable that the presence of numer- omy have been tackled using molecular ous phenolic compounds in varying concen- methods, but they have not proved success- trations makes willows a challenging food ful in reconstructing the phylogeny of the source. In spite of this, willows are utilized group. Surprisingly, this is explained by a by a wide variety of mammalian and insect lack of informative variation in molecular herbivores (Kontuniemi 1960, Seppänen markers that are generally highly variable in 1970, Pasteels & Rowell-Rahier 1992). other plant genera (Leskinen & Alström- Some specialist insect herbivores that have Rapaport 1999, Azuma et al. 2000, C. K. overcome the chemical defense of willows Anttila, personal communication). use the phenolics as feeding or oviposition There is, however, agreement that mor- cues (Tahvanainen et al. 1985, Pasteels & phologically, the genus Salix can be divided Rowell-Rahier 1992, Kolehmainen et al. into three (Skvortsov 1968, 1999) or four 1994, 1995; Roininen et al. 1999a), or even (Argus 1997) subgenera, of which the sub- use the compounds for their own defense genus Salix is the most primitive (Skvortsov (Pasteels & Rowell-Rahier 1992, Köpf et al. 1968, 1999). The subgenus Salix is also the 1998). There are hundreds of insect species first one to appear in the fossil record adapted to use willows as food, and one of (Collinson 1992). The subgenus Chamaetia the most conspicuous and common groups is includes low arctic species, and the subgenus formed by the nematine gallers. Vetrix mostly bushes (Skvortsov 1968, 1999). The molecular studies indicate, how- 2.2. The nematine gallers ever, that the subgenus Salix is paraphyletic, and also that the subgenera Chamaetia and Within the large symphytan family Vetrix may not be monophyletic (Leskinen Tenthredinidae, gall induction has evolved & Alström-Rapaport 1999, Azuma et al. twice, in the subfamilies Blennocampinae 2000, see also Dorn 1976, Chong et al. and Nematinae (Price & Roininen 1993). Of 1995). these, the former includes only a couple of Chemically, willows are characterized by species that induce galls on Caprifoliaceae low-molecular weight phenolic glycosides species (Price & Roininen 1993). By con- that occur in various species-specific arrays trast, the subfamily Nematinae includes hun- (Palo 1984, Julkunen-Tiitto 1986, 1989; dreds of galling and non-galling species Shao 1991). Many of these phenolic com- (Smith 1979, Viitasaari & Vikberg 1985, pounds presumably have a defensive func- Gauld & Bolton 1988). The non-galling spe- tion in willows, since salicylates (Lindroth et cies have various larval habits. Most of them al. 1988, Lindroth & Bloomer 1991, Lin- have typical free-feeding (exophagous) lar- droth & Hwang 1996), cinnamic acid de- vae, whereas some have endophytous larvae rivatives (Matsuda & Senbo 1986, Summers feeding, for example, in fruits, leaf mines, & Felton 1994), and flavonoids (Shaver & berries, or willow catkins (Smith 1979, Lukefahr 1969, Elliger et al. 1980, Dreyer & Gauld & Bolton 1988, Zinovjev & Vikberg Jones 1981, Harborne & Grayer 1993) have 1998). been found to act as feeding or oviposition The gall-inducing nematines occur al- deterrents, growth inhibitors, and toxins most exclusively on Salix species; the few against various herbivores. Willows also exceptions feed on the closely related contain condensed tannins, a diverse class of Populus (Price & Roininen 1993). There are high-molecular weight phenolics, that may over 200 species of gall-inducing nematines, have detrimental effects on herbivorous which, like their host plants, are common in mammals (Hagerman & Butler 1992) and arctic and subarctic regions (Smith 1979, 10

Figure 1. The different gall types induced by nematine sawflies on willows. Arrows indicate the Price-Roininen hypothesis of how the gall types have evolved (Price 1992; see also Roininen 1991a, Price & Roininen 1993). The three genera of nematines are indicated to the left of the gall type figures.

Kouki et al. 1994). These species have tradi- formation is initiated by a secretion that the tionally been classified in the subtribe females inject into the oviposition wound Euurina (Vikberg 1982). Euurina is divided together with the egg (McCalla et al. 1962, into three genera based on the type of gall Rey 1992, Leitch 1994, Higton & Mabberley that the species induce: Phyllocolpa species 1994). In some species, further gall devel- induce leaf folds or rolls, Pontania species opment requires additional stimulation by induce bean- or pea-shaped galls on leaves, the growing larva (McCalla et al. 1962, and Euura species induce galls on midribs, Smith 1970, Zinovjev & Vikberg 1998). The petioles, buds, or stems (Smith 1979, Price larvae grow during the summer and emerge & Roininen 1993). In all, about eight differ- in late fall to pupate, usually on the ground ent gall types can be recognized in the (Smith 1970, Price 1992, Kopelke 1998). nematine gallers (Fig. 1). Some species, especially leaf folders, can Typically, the galling sawflies emerge in produce multiple generations per summer the spring or early summer, mate, and the (Smith 1970, Price 1992). females then oviposit into young growing Most nematine gallers are fairly host- tissues of the willow hosts (Price 1992). Gall specific, but the degree of specificity may 11 vary between the three genera. Smith (1970) leaf gallers – midrib/petiole gallers – stem argued that the Phyllocolpa leaf fold- gallers – bud gallers (Fig. 1). The Price- ers/rollers are the least host-specific (oligo- Roininen hypothesis is based mainly on the or polyphages), whereas the Pontania gallers location of the gall on the host, so that the are more restricted in their host range, and oviposition site would have “wandered” to- Euura species are almost strictly restricted to wards the more central parts of the host plant one willow species (see also Zinovjev 1995). (Price & Roininen 1993). Such a generalization should, however, be Similar views had been proposed earlier treated with caution, since the species in by Smith (1970). Smith claimed that the Euurina are very difficult to identify by mor- most primitive galler species were the Phyl- phological methods, and in some cases ovi- locolpa leaf rollers, since the color of their position and/or crossing experiments may larvae resembles the larval coloration of the provide the only possibility for distinguish- nematine species that have free-feeding lar- ing between the species (Smith 1970, Ko- vae. Smith also proposed that the Pontania pelke 1990a, 1990b, 1991; Zinovjev 1993). species closest to the Phyllocolpa species Molecular methods might also provide tools belong to the P. proxima-group (leaf blade for species identification in the nematines, gallers). The first Euura species would be but so far they have been used only rarely the Euura (Gemmura) bud gallers, as their (Roininen et al. 1993, III). ovipositor shape is intermediate between Pontania and other Euura species. The next ones to evolve would be the Euura petiole 3. THE EVOLUTION OF DIFFERENT gallers, and the last ones the Euura stem GALL TYPES gallers. Although there are differences between 3.1. Hypotheses the two hypotheses, they share the view that each gall type has arisen only once, followed The existence of numerous different gall by a subsequent radiation to new willow types naturally raises the question of the or- species. The hypotheses are also essentially der in which these types have evolved. Theo- gradualistic in the sense that gall type shifts retically, multiple gall types could have would have occurred in small steps. The evolved independently, but most authors gradualistic view is, indeed, tempting, since have assumed that the galling species form a gall formation is probably dependent on the monophyletic group (Gauld & Bolton 1988), exact positioning of the gall-forming sub- although this view has also been disputed stance (McCalla et al. 1962, Kopelke (Zinovjev & Vikberg 1999). In addition, if 1990a). Thus, a change in gall type would the galling species are monophyletic as a require multiple, co-occuring changes in the whole, similar gall types could have evolved behavior of the ovipositing females and in independently on different willow species, or the shape of the ovipositor. In the initial each gall type could have arisen only once, stages of a gall type shift the population with subsequent radiations to multiple wil- must be polymorphic, which seems much low species (Smith 1970, Price & Roininen more probable if the gall types are relatively 1993). similar. This constraint is, however, some- Price (1992; see also Roininen 1991a, what relaxed by the fact that oviposition oc- Price & Roininen 1993) suggested that spe- curs during the early development of the cies inducing true closed galls evolved via willow shoots when different parts of the leaf folders/rollers, and that leaf blade gallers leaves are in close proximity, which can en- (P. proxima-type) would be the first true hance the possibility of oviposition jumps gallers to appear. Thereafter, the sequence that appear more dramatic than they really would have been apical leaf gallers – basal are (Price & Roininen 1993). 12

The evolution of different gall types was which could be rooted so that the ingroup studied by reconstructing the phylogeny of remains monophyletic (Fig. 2 in I). The two representative nematine species using allo- trees differ only in the structure of the Euura zyme electrophoresis (I) and DNA sequence clade. In addition, there is a polytomy at the data (II). The sequence in which the gall base of the Euura clade. If the polytomy is types evolved could then be inferred by resolved in all possible ways, there are six plotting the gall types on the resulting phy- equally parsimonious trees. The bootstrap logenies. supports of most branches were relatively low, and Decay indices ranged from one to 3.2. Allozyme results three. When the gall types are plotted on the The evolution of different gall types was first phylogeny (Fig. 4 in I), it can be seen that: studied by using allozyme electrophoresis to (1) leaf folders do indeed seem to be a basal reconstruct the phylogeny of 18 nematine group; (2) leaf blade gallers evolved inde- species (I). The selected species included 1-3 pendently of other true gallers (i.e., the spe- species per gall type. Four non-galling cies forming closed galls are polyphyletic); nematine species were used as an outgroup. (3) apical and basal leaf gallers are not the Eight informative enzyme loci were screened ancestors of petiole and bud gallers, but they using standard starch gel electrophoresis may share a common galling ancestor; (4) according to the protocols outlined in bud gallers evolved from midrib/petiole gal- Vuorinen (1984). lers; and (5) stem gallers are polyphyletic. The data was analyzed by maximum par- The results also show that gall type changes simony methods in PAUP version 3.1.1 occur less frequently than host shifts, and (Swofford 1993). The loci were treated as that there are restrictions affecting the evo- characters, and the allelic combinations as lution of different gall types: >95% of the character states (Character = Locus coding; randomized gall type characters are longer Buth 1984, Mabee & Humphries 1993, Mar- than the original character (Fig. 3 in I). dulyn & Pasteels 1994). The numbers of steps between character states were defined 3.3. DNA results using the step matrix option in PAUP. Gains and losses of alleles were weighted equally. In order to gain a more robust view of the Hypothetical ancestral states were created evolution of gall types, a second study was when needed, as described by Mardulyn and made, using DNA sequence data and a larger Pasteels (1994). Branch support was deter- number of sawfly species (II). In this study, mined by bootstrapping 100 times over loci 31 galler species representing seven different (Felsenstein 1985) and by determining the gall types were included (1-10 species per Decay index (Bremer 1988, 1994) for each gall type), and five non-galling species were branch. The evolution of gall types was in- used as an outgroup. An approximately 750- ferred by plotting the (unordered) gall types bp portion of the mitochondrial Cytochrome on the trees using MacClade version 3.01 b gene was amplified from each species us- (Maddison & Maddison 1992). Restrictions ing “universal” primers (Simon et al. 1994), in gall type evolution were tested by com- and the amplified fragments were sequenced paring the length of the original (ordered) using an ABI 310 automated sequencer gall type character to the character length (Perkin Elmer, Foster City, CA). The se- distribution from 200 random characters, in quences (678 bp) could be unambiguously which states had been shuffled between gall aligned, since no insertions/deletions were type groups. observed. The Character = Locus analysis produced Maximum-parsimony analyses of the two equally parsimonious trees, both of data set were performed using PAUP* ver- 13 sion 4.0b1 (Swofford 1998). The analyses high support for important branches (Fig. 1 were first performed using 3:1 transver- in II). The weakly supported branches are sion:transition weights (216 informative mainly within the outgroup and in the clade characters), and then transversions only (123 formed by the apical leaf gallers (P. vimina- informative characters). Unweighted maxi- lis-group). The trees could not be rooted so mum parsimony was not used, because a plot that the ingroup remains monophyletic, be- of transitions and transversions against se- cause Phyllocolpa anglica Cameron is quence divergence indicated that transitions grouped within the outgroup. However, this have become saturated (Fig. 2). Branch sup- grouping is not strongly supported, and if ports were again determined by boostrapping ingroup monophyly is enforced, the tree does (1000 replicates) and by calculating not become significantly less parsimonious Bremer’s (1988, 1994) support indices. In (Templeton’s test, P = 0.63). addition, the data was analyzed using quartet The phylogeny and the reconstruction of puzzling maximum likelihood analysis in the evolution of different gall types (Fig. 2 in PUZZLE version 4.0.1 (Strimmer & von II) indicate that: (1) the nematine gallers Haeseler 1999), with parameters estimated probably form a monophyletic group; (2) from the data and a neighbor joining tree. true closed galls evolved only once, via leaf Again, the evolution of different gall types folders; (3) the genus Euura and subgenus was inferred by plotting the (unordered) gall Eupontania Zinovjev (apical + basal leaf types on the resulting phylogeny, and the gallers) are monophyletic; (4) with the pos- possibility of restrictions in gall type evolu- sible exception of leaf rollers, all gall type tion was inferred by comparing the length of groups are mono- or paraphyletic; (5) similar the original (ordered) gall type character to gall types are closer on the phylogeny than the character length distribution of 1000 ran- would be expected by a random process; and domized characters, in which gall types had (6) there is an apparent evolutionary trend in been shuffled between groups (Fig. 3). galling site from the leaf edge towards the The different phylogenetic analyses re- more central parts of the host plant. sulted in essentially identical results with

80 100

80 60

60 40 40 Transitions

20 Transversions 20

0 0 0.0 0.1 0.2 0.0 0.1 0.2 Sequence divergence Sequence divergence

Figure 2. Numbers of transitions and transversions plotted against sequence divergence in pairwise comparisons of the DNA sequences used in (II). Note the saturation in transitions after ca. 10-15% sequence divergence. 14

zyme tree is probably erroneous, and if 80 monophyly of the closed galls is enforced in

60 the allozyme tree, the tree does not become significantly less parsimonious (Templeton’s 40 test, P = 0.32). Thus, it is likely that the spe- * cies inducing true closed galls are mono- 20 Number of replicates of Number 11 phyletic (as indicated by the DNA tree), and 0 that the first species to appear were leaf 10 15 20 25 30 blade gallers (P. proxima-group) or sausage Number of steps gallers (P. dolichura-group). In both of these gall types, an inner green layer can be found, Figure 3. Frequency distribution of the as in the rudimentary galls that form at the number of steps in 1000 random (or- site of egg-laying in some leaf rollers (Zi- dered) gall type characters when the tree novjev & Vikberg 1998). Furthermore, ovi- is as shown in Fig. 1 in (II). Gall types positing females in the P. dolichura-group were shuffled between gall type groups, make multiple insertions with their oviposi- but the outgroup state (= 0) was kept tor, and this habit is also present in the constant. Alpha = 0.05 level shown by Phyllocolpa leaf folders/rollers (Kopelke star, number of steps in the original gall 1985, 1998; Price & Roininen 1993). It is type character (= 11) shown by arrow also interesting to note that late-instar larvae (the P. dolichura samples and E. ameri- of some species in the P. proxima- and P. nae were excluded from the analysis). dolichura-groups occasionally come out of the gall to feed on the leaf edge (Roininen et al. 1999b). The allozyme and DNA trees both sup- 3.4. Combined evidence port the monophyly of the Euura + Eupon- tania clade, but it is unclear what the ances- Although there are slight differences be- tral gall type in the group was. Considering tween the allozyme tree and the DNA se- the similarities in the structure of the galls quence tree, they also have important com- induced by Pontania s. str. gallers (leaf blade mon features. For example, both trees agree gallers + sausage gallers) and Eupontania that the species that induce true closed galls basal leaf gallers, it seems very probable the evolved via leaf folders, as had been sug- ancestor induced galls on leaves, but a con- gested already by Smith (1970). As stated firmation of this will require further study. It above, the DNA tree could not be rooted so is possible that the ancestral gall type can be that the ingroup remains monophyletic, be- found by including more species in the cause Phyllocolpa anglica is grouped within analysis. The grouping of E. atra within the the outgroup. However, the exact location of Eupontania in the allozyme tree is clearly P. anglica does not affect the inferences on erroneous, and E. atra can be moved to the the sequence in which the gall types evolved. Euura clade without the allozyme tree be- It must be noted that Zinovjev and Vikberg coming less parsimonious (Templeton’s test, (1999) claimed that the genus Phyllocolpa is P = 0.18). The DNA tree indicates that the rather heterogeneous, and that the leaf- apical leaf gallers evolved from basal leaf rolling habit may be polyphyletic (see also gallers, and this view is not contradicted by Vikberg 1982). the allozyme evidence. Both the allozyme Both phylogenies leave open the question and DNA results contradict Smith’s (1970) of the next step in the evolutionary history. suggestion that Euura gallers evolved via the The inclusion of the P. proxima-group Euura (Gemmura) bud gallers. within the Phyllocolpa species in the allo- 15

The main result of both phylogenetic spi & Worobey 1998). Interestingly, the analyses is that similar gall types tend to be galling habit seems to be preceded by leaf closer on the phylogenies than would be ex- folding also in the thrips that induce galls on pected by a random process. This indicates Acacia species (Crespi & Worobey 1998). that gall type changes have occurred in small This suggests that leaf folding can act as a steps, which is in accordance with the tradi- preadaptation that may lead to the evolution tional gradualistic view of evolution, and of true closed galls. seems reasonable, given the need for a Assuming that the sawfly gallers control polymorphic condition in the galler popula- the location and morphology of the galls, tion during a gall type shift. Both studies then why do these features change? The rea- also suggest that there has been a trend to- sons for changes in gall type could be adap- wards gall induction in the more basal parts tive. If parasitoids or predators search for of the host plants. galls according to stereotypical search patterns, changes in gall morphology could help 3.5. Discussion the gallers to avoid these mortality factors (Cornell 1983). This may well be the case in The fact that gall morphology reflects the the genus Pontania, where there are great phylogeny of the gallers, not the phylogeny differences between the parasitoid commu- of the willow hosts, demonstrates that the nities on different gall types (Kopelke 1985, location and shape of the gall is determined 1994). For example, in sympatric popula- mainly by the galler, not by the host. Thus, tions of P. samolad Malaise (an apical leaf gall morphology can be considered to repre- galler) and P. lapponica Malaise (a basal sent an extended phenotype (sensu Dawkins leaf galler) on Salix lapponum L., the para- 1982) of the galler. This interpretation is sitoid Adelognathus cubiceps Roman occurs essentially identical to the results obtained only in P. lapponica galls (Roininen et al. from phylogenetic studies of gall-inducing 1999c). psyllids (Yang & Mitter 1994), aphids (Stern It is also possible that shifts in gall type 1995), cynipid wasps (Stone & Cook 1998), are nonadaptive. In the initial polymorphic and thrips (Crespi & Worobey 1998), and stage, the galler could induce different galls the view has also been supported by some on different hosts. After speciation, the spe- nonphylogenetic studies (Dodson 1991). To cies with the novel gall type could radiate to date, the only exception seems to be in the new hosts but still retain the newly acquired fordinine aphids that induce galls on Pistacia gall type. Indeed, in some cases nematine species, since Wool (1997) suggested that in gallers have been found to induce abnormal this case the host plants may play a more galls on hosts that are not normally used by important role in the control of gall mor- the galler (Zinovjev 1994, 1995). This could, phology than the insects. Furthermore, some for example, be a result of differences in size studies indicate that the host plant can have or phenology between willows: the length of an effect on the size of the galls (Price & the sawfly ovipositor might not be suitable Clancy 1986, Weis 1996, Crespi & Worobey for oviposition at the appropriate site on a 1998). novel host. This could explain, among other Many common features are shared by the things, the transition from bud gallers to independent radiations of diverse insect gal- petiole gallers, since in both groups oviposi- lers on unrelated host plants. For example, tion occurs through the petiole base (Smith there are apparently constraints on how gall 1968, Roininen 1988, Kopelke 1998). morphology can change, since gall types However, the trend-like pattern observed seem to change gradually in sawflies (I, II), in the evolution of the galling site suggests psyllids (Yang & Mitter 1994), cynipid that adaptive explanations may be needed. wasps (Stone & Cook 1998) and thrips (Cre- There are several factors that might promote 16 oviposition in more basal parts of the hosts. 4. THE EVOLUTION OF HOST PLANT For example, the reason could be linked to USE the control of resource allocation in the plant. Galls have been shown to behave as 4.1. Hypotheses sinks for nutrients and photosynthetic as- similates in plants (Larson & Whitham 1991, Phytophagous insects have been used exten- 1997), and, thus, intraspecific competition sively to study various questions concerning could lead to unequal transition probabilities, the evolutionary relationships between ani- if galls can intercept nutrient flow to other mals and plants. In this respect, insects are a galls situated in more distal positions. This good model group, because they are abun- type of competition-driven evolution of dant, speciose, and they reproduce rapidly. galling site was also suggested by Yang and Furthermore, although some insect groups Mitter (1994) in the case of psyllid gallers on have been associated with the same host American Celtis species. The suggestion is plants for millions of years (Farrell et al. not entirely speculative: in pemphigine 1992, Wilf et al. 2000), it is evident that aphids, females fight for the basal parts of host-associated evolutionary changes can the leaves (Whitham 1979), where repro- occur very rapidly in some species (Bush ductive success is higher (Whitham 1978). In 1975, Carroll & Boyd 1992, Radkey & the oak leaf galler Cynips divisa Hartig, gall Singer 1995, Carroll et al. 1997, 1998; Gro- location in relation to other galls and the man & Pellmyr 2000). midrib affects gall size and growth (Sitch et Insect herbivores are commonly more or al. 1988, Hartley 1998) and galler mortality less restricted in their use of potential host (Gilbert et al. 1994). Fritz et al. (1986) dem- plants: most species feed on one or a few onstrated the possibility of asymmetric com- host species, and usually the hosts are petition between the stem galler E. lasiolepis closely related (Jermy 1984, Bernays & Gra- Smith and three other species representing ham 1988). This restricted diet is usually different gall types on Salix lasiolepis Ben- attributed to the “jack of all trades, master of tham. Obviously, this explanation is depend- none” hypothesis, which predicts that a ent on the abundance of the gallers, and it is given insect species cannot be adapted to the not known whether galler densities are nor- various chemical defenses of all potential mally high enough for serious competition to host plants (Dethier 1954, Fraenkel 1959, occur. Ehrlich & Raven 1964, Futuyma & Keese Competition, however, is not necessarily 1992). Restrictions in terms of host use can needed for selection favoring oviposition in also be caused, for example, by phenological more basal parts of the plant. Gall induction differences between hosts (Smith 1988, in basal positions could help in evading ab- Craig et al. 1993, Groman & Pellmyr 2000), scission reactions (Williams and Whitham differences in predation/parasitism on hosts 1986), and, as Larson and Whitham (1991) (Bernays & Graham 1988, Crespi & Sando- demonstrated, the sink strength of galls de- val 2000), and morphological or habitat dif- pends on their location. In their study of ferences between hosts (Futuyma & Keese aphid galls, a difference of 12 mm caused a 1992). fourfold difference in the amount of re- In fact, it appears that many phytopha- sources drawn to the gall from adjacent gous insects utilize only a small fraction of leaves. This kind of positional effect may be the host plants that are suitable as food for unimportant under normal conditions, but it them (Wiklund 1975, Smiley 1978, Roininen may, on the other hand, be important when & Tahvanainen 1989, Nylin & Janz 1999). resources are scarce. This has been explained by constraints on the information processing capacities of the insects: if some of the potential host plants 17 can be confused with unsuitable hosts, the The evolution of host use in the galling situation could lead to the evolution of an nematines was first studied by using the al- “unnecessarily” narrow host range (Levins & lozyme (I) and DNA phylogenies (II) to in- MacArthur 1969, Futuyma 1983, Courtney terpret the patterns of host associations in 1983, Janz & Nylin 1997). relation to the aforementioned hypotheses. The causes of host specificity are directly The evolution of host associations was then linked to the evolution of host shifts by further studied by analyzing host-associated monophagous insects. During a host shift, enzyme electrophoretic variation in the pre- the insect species must be at least oligopha- sumedly polyphagous bud galler Euura mu- gous and, thus, it is likely that the direction cronata Hartig (III). of host shifts is constrained by some of the aforementioned factors. For example, it has 4.2. Host-plant relationships in the nema- been suggested that a given insect species is tines more likely to colonize hosts that are chemically (Dethier 1954, Ehrlich & Raven 1964, The allozyme study (I) indicated that host Futuyma & McCafferty 1990, Becerra 1997, associations are rather labile in comparison Wahlberg 2000) or ecologically (Dobler et to the evolution of gall types, and this result al. 1996) similar to the original host. On the was confirmed by the results of the DNA other hand, specialist insects could speciate study (II). It is clear that many willow spe- simply by tracking the speciation events in cies have been colonized several times. For their hosts (Mitter & Brooks 1983, Mitter et example, S. lapponum L. has been colonized al. 1991). It has to be noted that speciation is at least three times independently by species not an automatic outcome of the colonization representing different gall types, and S. of a new host. Especially if the new host phylicifolia L. and S. pentandra L. at least closely resembles the original host, the end two times each (Fig. 1 in II). Evidently, such result will only be an expansion of the in- a pattern of host associations excludes any sect’s feeding range (Thompson 1994, Gro- coevolutionary hypotheses in the sense of man & Pellmyr 2000). gallers “tracking” speciation events in wil- Insect herbivores have also been studied lows (Fig. 4). Considering the host plant extensively in order to find out whether records of Eurasian galler species not in- sympatric speciation is possible in nature. cluded in this study (Table 2 in II, Zinovjev Many phytophagous insect species exhibit 1998), it seems likely that the numbers of features that are likely to enhance the possi- independent colonizations for many Salix bility of sympatric speciation: mating com- species are actually considerably higher. monly occurs on or near the host plant, the Furthermore, the results are not concor- adults determine the larval host, and host dant with the escape-and-radiate model of preference (and performance on different coevolution (Ehrlich & Raven 1964), since hosts) may be subject to simple genetic con- there cannot be any concordant clades in the trol (Bush 1975). Indeed, some of the best willow and galler phylogeny (Thompson putative cases of sympatric speciation are 1999). Likewise, there is no clear connection from insect herbivores (Bush 1969, 1994; between the phylogeny of the nematine gal- Tauber & Tauber 1989, Barraclough & Vo- lers and the overall chemical similarity gler 2000). Sympatric speciation has been (Julkunen-Tiitto 1986, 1989; Shao 1991) of considered unlikely (Mayr 1963, Futuyma willows. 1983), but mathematical models have shown In the DNA study (II), the subgenera of that it may be possible under some circum- the willow hosts were plotted on the galler stances (Diehl & Bush 1989, Dieckmann & phylogeny, but a randomization test did not Doebeli 1999). indicate any statistically significant conservatism in host use even at the level of willow 18

Figure 4. The evolution of host associations in nematine sawflies, plotted using TreeMap version 1.0b (Page 1995). The phylogeny of gallers is from (II) (T. arquata has been excluded), the willow phylogeny is mainly according to the phenetic tree of Argus (1997). Locations of host species not included in Argus (1997) are according to Skvortsov (1968, 1999). Potential cospeciation events are shown by black dots on the nodes of the sawfly tree. Hosts are abbreviated by using only the three first letters of the species name (e.g., S. lan = Salix lanata). subgenera. This could, however, partly be a records, E. mucronata induces galls on over result of the limited sampling of species 30 willow species in the Palearctic and feeding on the subgenera Salix and Chamae- Nearctic Regions (Smith 1970, Hartley 1992, tia. When all available host plant records are Price & Roininen 1993, Kopelke 1998, H. considered, there are signs of conservatism Roininen & A. G. Zinovjev, unpublished in host use (Table 2 in II). data). Thus, the evolution of bud galling would have been accompanied by a switch 4.3. Euura mucronata: a polyphage or a from monophagy to extreme polyphagy. sibling species complex? Although switches from monophagy to polyphagy can and do occur (Thompson The results from the phylogenetic analyses 1994, Kelley & Farrell 1998, Nylin & Janz (I, II) raise a problem: the monophyletic ge- 1999), the case of E. mucronata should be nus Euura contains several sibling species interpreted with caution. In many cases, in- complexes comprised of strictly monopha- sect species presumed to be polyphagous gous species (Smith 1968, 1970; Kopelke have been found to consist of numerous 1996), but the rather derived bud-galling morphologically indistinguishable mono- Euura mucronata is considered to be polyphagous sibling species or host races (Mayr phagous. According to the available host 1963, Diehl & Bush 1984) that may be iden- 19 tifiable only by using molecular markers Cockerham (1984) using the Arlequin ver- (see, e.g., Menken 1981, Sturgeon & Mitton sion 2.000 program (Schneider et al. 2000). 1986, Feder et al. 1988, McPheron et al. The results show that: (1) in northern 1988, Herbst & Heitland 1994, Emelianov et Fennoscandia, “E. mucronata” comprises at al. 1995, Condon & Steck 1997, Groman & least three mono- or oligophagous species or Pellmyr 2000). Furthermore, even the stem- host races, but occasional hybridization be- galling Euura atra complex was previously tween the species may occur; (2) the pattern considered to be one polyphagous species of host use excludes the possibility of paral- until it was discovered that it is probably a lel cladogenesis between the bud gallers and sibling species complex (Roininen et al. their hosts; (3) the overall chemical similar- 1993, Kopelke 1996). E. mucronata has been ity of the hosts does not explain the pattern split into three species based on morphologi- of host use; and (4) simple allopatric specia- cal features, but the feeding ranges of the tion does not seem to provide a sufficient species, and even their existence, is unclear explanation. (Viitasaari & Vikberg 1985, Kopelke 1998, Zinovjev & Vikberg 1998). This is partly a 4.4. Discussion result of extreme size variation in the bud gallers (Benson 1958). Both the overall nematine phylogeny (II) and To reveal the specific status of E. mucro- the pattern of host use in E. mucronata (III) nata, enzyme electrophoretic variation was exclude an explanation based on simple studied in bud gallers collected from differ- tracking of host speciation. This does not, ent willow species occurring in the sur- however, imply that host phylogeny has no roundings of the Abisko research station in effect. For example, species in the mono- Sweden and the Kilpisjärvi research station phyletic subgenus Eupontania Zinovjev in Finland. The distance between the two (apical + basal leaf gallers) are almost exclu- locations is approximately 120 km, which sively associated with willow species in the should be well above the dispersal capabili- subgenera Chamaetia and Vetrix (Zinovjev ties of individual sawflies. Collections were 1993, 1995, 1998). Furthermore, there are made from randomly selected willow indi- differences in how separate groups of nema- viduals in six host species (S. lanata L., S. tines use the three willow subgenera (Table 2 glauca L., S. lapponum L., S. phylicifolia L., in II), which can be taken as an indication of S. myrsinifolia Salisb., and S. hastata L.). In constraints and phylogenetic inertia in host both locations bud galls can also be found on use. A similar combination of conservatism S. myrsinites L., but not enough larvae were and promiscuity in host use has been ob- found on this host for analyses. The larvae served in galling thrips (Crespi et al. 1997). were reared to adults, which were sexed and It must be noted that strictly concordant then frozen until electrophoresis. phylogenies between phytophagous insects Seven variable enzyme systems were and their hosts are probably rare (Mitter et surveyed using standard starch gel electro- al. 1991). In fact, the best demonstrated phoresis, mainly according to the protocol cases of parallel cladogenesis come from outlined in Vuorinen (1984). The data was mammals and their ectoparasites (Hafner & first analyzed by performing preliminary Nadler 1988), and insects and their endo- clustering analyses using the TFPGA version symbionts (Moran et al. 1993, Clark et al. 1.3 program (Miller 1997). Population 2000). The close association between the structure was then further studied by calcu- galling insects and their hosts would seem to lating hierarchical F-statistics (Wright 1978) predispose the system for parallel cladogene- within the subgroups formed by the cluster- sis, but, on the other hand, it would be un- ing analyses. Fixation indices were calcu- likely if the mechanism of gall induction is lated according to the method of Weir & 20

(A) Females (B) Males

40 S. lan (K) S. lan (K) 10

0 0 80 S. lan (A) S. lan (A) 20

0 0 S. gla (K) S. gla (K) 40 30

0 0 S. gla (A) S. gla (A) 60 30

0 0 40 S. lap (K) S. lap (K) 20

0 0 100 S. lap (A) 60 S. lap (A)

0 0 40 S. phy (K) S. phy (K) 10

0 0 10 S. phy (A) S. phy (A) 5 Number of individuals Number of individuals 0 0 20 S. myr (K) 5 S. myr (K)

0 0 S. myr (A) S. myr (A) 20 5

0 0 S. has (K) S. has (K) 20 10

0 0 40 S. has (A) S. has (A) 10

0 0 0246810 0246810 Day Day

Figure 5. Emergence times of the E. mucronata adults used in (III). On the x-axis, 1 = the day when the first individual emerged. Hosts are abbreviated by using only the three first letters of the species name (e.g., S. lan = Salix lanata). Locations are shown in parenthe- ses after the host name: K = Kilpisjärvi, A = Abisko. 21 general enough to produce galls on phyloge- ferent host species, and the resulting possi- netically distant host species. bility of allochronic speciation (Craig et al. The studies also show that the evolution 1993, Feder et al. 1997a, 1997b; Filchak et of host use in the nematines is not connected al. 2000, Groman & Pellmyr 2000, Tikkanen with the overall chemical similarity 2000). It is possible that phenological factors (Julkunen-Tiitto 1986, 1989; Shao 1991) of influence the pattern of host use in the the willow hosts. Similar results were earlier nematines, but this is more difficult to study. found by Roininen et al. (1993) in the Euura The possibility should, however, be taken atra complex. This is not surprising consid- seriously, and at least in the case of E. mu- ering that the chemical properties of galls cronata there seem to be differences in the may differ radically from that of the host emergence times of adults collected from plants (Hartley 1992, 1998; IV). Further- different willow species (Fig. 5). The ovi- more, if diet chemistry is not important, then position period of E. mucronata lasts for the oviposition behavior of females may be weeks (Roininen 1991b), but this does not determined by only a few chemical com- mean that timing is irrelevant. For example, pounds. Results indicating that this is the the weight of the larvae may generally be case have been found in the stem gallers lower in the more distal positions on the Euura amerinae L. (Kolehmainen et al. shoots (Fig. 6), which may have conse- 1994) and E. lasiolepis Smith (Roininen et quences if fitness correlates positively with al. 1999a). adult weight, as has been demonstrated in Recent studies have emphasized the im- many insect species (see, e.g., Charnov et al. portance of phenological adaptation to dif- 1981, Haukioja & Neuvonen 1985, Ueno 1999). If there are phenological differences between willows, it is possible that the most “profitable” host species changes during the summer.

1 Another possibility is that the sizes of the oviposition target tissues vary between willow species. Gall induction in the nematines require exact positioning of the egg in the 0 host tissue (McCalla et al. 1962, Kopelke 1990a), which means that size differences between willow species could restrict the

Log weight (mg) range of available hosts for an individual -1 sawfly. The possibility is especially relevant in certain Euura species that are extremely 0.2 0.4 0.6 0.8 1.0 variable in size (Benson 1958, Smith 1970). Relative location It should be noted that if the willow host affects the size of adult sawflies (via the size Figure 6. Log-transformed weight (mg) of of galls), and if sawflies exhibit size- E. mucronata larvae in different relative dependent assortative mating, then repro- positions on S. phylicifolia shoots col- ductive isolation between host races could lected on August 28, 1998, from the sur- evolve almost automatically. The result roundings of the Kilpisjärvi Research would essentially be analogous to the “con- Station. Relative position = number of ditioning to the larval host” effect, which can the galled bud on the shoot (counting up enhance the likelihood of sympatric specia- from the base of the shoot) / the total tion (Bush 1975). Indeed, the galling habit number of buds on shoot. (r = -0.35, n = may even predispose insects to “accidental” 122, P < 0.001). sympatric speciation. 22

5. THE CHEMICAL ECOLOGY OF should be lowered in galls. The support for NEMATINE SAWFLIES this prediction is currently vague and contra- dictory. Although Price et al. (1986) found 5.1. The Nutrition Hypothesis and previ- low levels of phenolics in galls induced by ous studies the nematine stem galler Euura lasiolepis Smith on Salix lasiolepis Bentham, many The convergent evolution of the galling habit galls have in fact been found to contain rela- in highly diverse organisms suggests that tively high concentrations of phenolics galls are beneficial for the organisms that (Abrahamson et al. 1991, Gopichandran et induce them. Price et al. (1986) summarized al. 1992, Hartley 1992, 1998, 1999), espe- the possible adaptive reasons that have pro- cially condensed tannins (Cornell 1983, An- moted gall induction, and concluded that the anthakrishnan & Gopichandran 1993, Yang three most probable hypotheses are the Mi- et al. 1998). croenvironment Hypothesis (“galls create an However, the interpretation of most optimal microenvironment”), the Enemy studies is rather problematic. Many studies Hypothesis (“galls protect gallers from have analyzed whole galls (e.g., Shannon & predators/parasitoids”), and the Nutrition Brewer 1980, Andersen & Mizell 1987, Hypothesis (“galls are nutritionally of better Brewer et al. 1987, Gopichandran et al. quality that the host”). 1992, Hartley 1992, Hartley & Lawton 1992, Of these three hypotheses, the Nutrition Yang et al. 1998), although it is known that Hypothesis has been studied most fre- the chemical properties of the outer layers quently, and it is evident that the chemical can differ considerably from the chemistry properties of galls may differ from the of the inner layers (Cornell 1983, Abraham- chemistry of the normal tissues of the host son & McCrea 1986, Bronner 1992). Since plant. For example, Brewer et al. (1987) the larvae feed only on the inner layers of the found that the galls induced by five species gall, chemical analyses of whole galls can of cecidomyiid midges on the needles of lead to severely misleading results. Many Pinus edulis Engelm. generally had lower studies have also analyzed compounds that concentrations of nitrogen than ungalled are of unknown importance to phytophagous needles, but in other inorganic nutrients (P, insects (e.g., most inorganic nutrients). S, Ca, Mg, Na, Zn, Mn, Cu, B) the pattern When defensive compounds have been ana- was not as clear. Low levels of nitrogen have lyzed, all studies have been performed with also been found in galls induced by homop- low chemical resolution, i.e., they have ana- terans (Andersen & Mizell 1987), cynipid lyzed total concentrations of various chemi- wasps (Hartley & Lawton 1992, Hartley cal classes (e.g., phenolics) rather than the 1998) and dipterans (Hartley 1998). On the concentrations of individual compounds. other hand, in some galler-host pairs, nitro- Obviously, such analyses can obscure ecol- gen levels are higher in galls than in compa- ogically meaningful variation in the levels of rable plant tissues (Abrahamson & McCrea individual compounds (Berenbaum 1995, 1986, Hartley 1998). Several studies have Keinänen et al. 1999). Thus, for a chemical found increased levels of starch (Shannon & analysis to be meaningful, it should be car- Brewer 1980, Bronner 1992, De Bruyn et al. ried out separately for different parts of the 1998) and proteins (Price et al. 1986, Bron- galls, it should measure compounds with ner 1992, Gopichandran et al. 1992, De- known importance, and it should analyze Bruyn 1994, De Bruyn et al. 1998) in galls, individual compounds rather than totals of which seems to support the Nutrition Hy- chemical classes. pothesis. The purpose of the chemical study (IV) The Nutrition Hypothesis also predicts was to gain insights into the chemical ecol- that the levels of plant defense compounds ogy of Pontania gallers by following the 23 three aforementioned guidelines: (1) the (Seigler 1998), and the total concentrations analyses were performed separately for the of the different chemical categories were outer and inner layers of the galls; (2) the calculated for each sample class. Differences analyzed phenolic compounds have demon- in the totals were tested by a repeated- stratedly significant effects on insects; and measures analysis of variance (RM- (3) individual compounds were analyzed by ANOVA) design in SPSS version 8.0 (SPSS, High-performance liquid chromatography Inc., Chicago, IL), using the samples as (HPLC), and totals were calculated by sum- within-subjects factors and species as be- ming the concentrations of the individual tween-subjects factors. The overall chemical phenolics. similarities of samples and sample classes were analyzed using clustering and ordina- 5.2. The chemical properties of Pontania tion methods in PC-ORD version 4.01 (MjM galls on willows Software Design, Gleneden Beach, OR). The results show that: (1) the concentra- Apical leaf galls were collected from six tions of most low-molecular weight phe- willow species from the surroundings of the nolics are clearly lower in gall interiors than Kilpisjärvi research station. Samples were in the leaves of the respective host willows, collected from ten individuals per host spe- and the levels of virtually all analyzed com- cies. From each willow individual, a leaf pounds are lower in gall interiors than in gall with a gall, and the leaf immediately below cortices; (2) the levels of condensed tannins the galled one, were collected. Each gall was are elevated in gall interiors, but the highest cut open, and the larva was removed. There- concentrations are found in the gall cortices; after the tissue samples were dried at room (3) the pattern of the chemical changes indi- temperature, and then frozen until analysis. cates that the phenolic biosynthesis of wil- Before the chemical analyses, the outer and lows is redirected in gall tissues; (4) within inner layers of the galls were separated using each host species, the overall variation in the a scalpel. chemical properties of gall interiors is mark- The concentrations of various phenolic edly lower than in the leaf samples collected compounds in the samples were analyzed from the same host individuals; and (5) using High-performance liquid chromatog- similarly, the between-species variation in raphy (HPLC), essentially according to the willows is drastically reduced in gall interior protocol outlined in Julkunen-Tiitto et al. samples. Indeed, the maximum level of (1996). This allowed identification and variation in gall interiors is comparable to quantification of 36 low-molecular weight (or lower than) within-species variability in phenolic compounds in the samples. Only leaf samples. compounds that could be reliably scored in all species were analyzed, but the phenolics 5.3. Discussion included represent on average over 90% of the absorbance in the chromatograms at 270 The results clearly suggest that the galls in- nm. In addition, the concentration of con- duced by Pontania gallers on willows are densed tannins was determined using a col- nutritionally beneficial for the sawfly larvae, orimetric test (Hagerman & Butler 1994). because the concentrations of most phenolic The data was first analyzed by calculat- compounds are clearly lower in gall interiors ing the mean concentrations of different than in the leaves of the host plants. As compounds in each of the four sample stated above, there is a substantial amount of classes (gall cortex, gall interior, galled leaf, evidence indicating that the individual com- and ungalled leaf) in each galler-host pair. In pounds that were measured have detrimental addition, compounds were classified into effects on insects. On the other hand, the different categories based on their structure benefits conferred by the reduction in the 24 levels of low-molecular weight phenolics aphids (W. J. Mattson, personal communi- could be counterbalanced by the increase in cation). Although very little is known about the concentration of condensed tannins in the the chemical properties of these other insect galls. It must be noted, however, that con- galls, the abundance of condensed tannins in densed tannins have been found to have a galls induced by various gallers on highly rather low level of toxicity in some feeding diverse host plants is a pattern that should be tests (Ayres et al. 1997), but evidently fur- further studied. It could reflect some com- ther studies on the relative toxicity and other mon principle in the gall-induction process, harmful effects of the various phenolics are such as, for example, a general response of needed. plants to gall-inducing stimuli. On the other Additional benefits for the larvae could hand, the pattern could be caused by conver- be caused by the reduction in the number of gent selection pressures on the insects, which different phenolics present in the gall tissues. may be able to control the chemical proper- It is probable that different defensive com- ties of the galls. pounds can have synergistic effects (Lin- In the sawfly galls, different phenolic droth et al. 1988, Harborne & Grayer 1993, categories react in a highly coordinated pat- Berenbaum 1995, Berenbaum & Zangerl tern in all the studied galler-host pairs. Since 1999). Consequently, it may be easier for an the analyzed compounds are all produced via insect to adapt to one or a few toxins. This is the phenylpropanoid pathway in willows also supported by the fact that many com- (Seigler 1998, Cooper-Driver & Bhattacha- mercial pesticides have lost their effective- rya 1998), this suggests that the biosynthesis ness in one or two decades (McKenzie & of phenolic compounds is redirected in gall Batterham 1994). Conversely, the applica- tissues. The observed pattern could result tion of a mixture of compounds can delay from a specific blocking of only a few the evolution of resistance, or prevent it al- branching points in the phenylpropanoid together (Pimentel & Bellotti 1976, Beren- pathway (Fig. 7), which raises the possibility baum & Zangerl 1999). that the mechanism of manipulation is rather It is interesting to note that the concen- simple. For example, the specific enzymes trations of virtually all analyzed compounds acting at the branching points could be af- are higher in the outer layers of the galls than fected somehow in gall tissues. On the other in the gall interiors, and especially the high hand, the mechanism could also operate by concentrations of condensed tannins in the an overstimulation of the production of con- cortices are striking. The pattern is, however, densed tannins, which would lead to a de- not so surprising considering that the gall pletion of intermediate substrates for the must be protected against attack by other other phenolic classes. Theoretically, the phytophages (Schulz 1992), especially if the observed pattern could arise as a conse- gall interiors are of a high nutritional quality. quence of the rapid growth of the gall tis- High concentrations of tannins have also sues, since there may be a trade-off between been found in the outer layers of galls in- growth and the production of defensive duced by cynipid wasps (Cornell 1983), chemicals in plants (Herms & Mattson where they may protect the gall from fungal 1992). However, such a trade-off cannot by invasion (Taper et al. 1986, Taper & Case itself explain the observed change in alloca- 1987). tion to different compounds. In addition to the sawfly (IV) and cynipid The changes in chemistry are so great wasp galls (Cornell 1983), high concentra- that they probably have far-reaching conse- tions of condensed tannins have also been quences for the evolutionary relationships found in galls induced by thrips (Ananthak- between the nematine gallers and their wil- rishnan & Gopichandran 1993), cecidomyiid low hosts. The most interesting possibility midges (Yang et al. 1998), and adelgid concerns the causes of host specificity and 25 the evolution of host shifts by the gallers: phenylalanine since the galls induced on chemically divergent host species are chemically almost identical, then why are the gallers monophagous? It is unlikely that the constraints in host use are caused by an inability to adapt salicylates cinnamic acid to variable diet chemistry. Instead, specialization may be caused by the other possible causes mentioned above (phenological dif- -cinnamic acid ferences between hosts, differential preda- derivatives 4-coumaryl-CoA -chlorogenic tion/parasitism, sensory constraints, plant acid responses to gall-inducing stimuli, etc.). Evidently, if the causes of host specificity flavones change, then there should also be conse- -apigenin derivatives flavanone quences for the evolution of host use on a -luteolin derivatives larger scale. Thus, studies on gall-inducing flavonols insects can provide valuable insights into the -quercetin derivatives -myricetin derivatives (co)evolutionary relationships between in- dihydroflavonol sects and plants. ampelopsin Although it seems that the results support the Nutrition Hypothesis, one should be cau- tious about stating that the evolution of gall flavan-3-ol flavan-3,4-diol induction in the nematines was originally caused by nutritional benefits. The Pontania species used in the study represent a rather derived group within the gall-inducing condensed tannins nematines (I, II). Thus, more information is needed on the chemical composition of the galls induced by the P. proxima-type leaf blade gallers, and on the chemistry of the Figure 7. A simplified scheme of the phen- leaf folds induced by Phyllocolpa species. ylpropanoid pathway, along which dif- Leaf folding probably does not markedly ferent phenolic compounds are pro- alter the quality of the leaf as food (but see duced in willows (modified from Sei- Sagers 1992), but it may help in avoidance gler (1998) and Keinänen et al. (1999)). of predation or parasitism (Damman 1987, Crossed arrows indicate branching Sipura 1999, but see Murakami 1999). This points that seem to be partially blocked is also supported by the fact that many lepi- in Pontania galls (IV). dopteran larvae make leaf rolls mechanically (Seppänen 1970). Thus, it is possible that the original cause for the evolution of gall induction was a reduction in parasitism and/or predation, and more sophisticated host manipulation evolved only subsequently in the Pontania gallers. The results also reinforce the importance of analyzing individual compounds rather than total concentrations of chemical classes. If only total phenolics had been measured, the result would have been that the total con- 26

(A) (B)

400 200

0 0 S. phy S. myr S. gla S. lap S. bor S. ret S. phy S. myr S. gla S. lap S. bor S. ret Species Species Gall cortex Gall interior Gall Galled leaf Galled leaf Ungalled leaf Ungalled leaf Concentration (mg/g dry weight) dry (mg/g Concentration

Figure 8. (A) Total phenolics (condensed tannins + other phenolics) in different sample classes in (IV); (B) Total phenolics concentrations when galls are treated as one sample. The mean concentration in galls (B) was calculated by using the weighted average of the concentrations in the cortex and interior samples for each individual gall. centration is only slightly reduced in galls (1) The phylogenetic relationships within (Fig. 8A). More importantly, if whole galls the Nematinae. One direction would be to had been used in the analyses, the conclusion go “up” on the phylogenies (I, II) by making would have been that phenolic levels do not separate phylogenetic studies of the main really change at all in galled tissues (Fig. groups. The phylogeny in (II) demonstrates 8B). Since all possible compounds were not the existence of three to five independent measured in the study, these results must be radiations on the genus Salix, at least in the treated with caution, but they illustrate the Pontania s. str., the Eupontania, and in the point that measuring compounds in bulk can Euura. Separate phylogenetic analyses of severely affect the conclusions that are these groups would facilitate a comparative drawn. analysis of the radiation patterns in the groups. Combining the separate studies would also provide additional insights into 6. CONCLUDING REMARKS the evolution of different gall types. The current phylogenies provide a rough view of It is evident that the nematine sawflies and the evolution of gall morphology and loca- their willow hosts offer an excellent model tion in the nematine sawflies, but the study system for testing various evolutionary hy- should be continued. There are still some potheses, since both the hosts and gallers are Nearctic gall types that are relatively un- abundant and speciose in northern ecosys- known. Molecular methods should also be tems. However, it is clear that many inter- used more often to study the specific compo- esting questions remained outside the scope sition of problematic nematine groups. An- of this thesis, and that further studies are other promising direction would be to go needed. There are four main topics that “down” on the phylogeny by adding species should be studied: to the base. There are numerous different 27 larval habits (and very divergent host asso- analyses of the nematine sawflies, and ciations) in other groups within the Nemati- knowing the chemical properties of Salix nae, and a broader sampling of species could species could also help in reconstructing the provide interesting insights into the ecologi- phylogeny of willows. cal evolution in these sawflies. (4) The chemical composition of the galls (2) The phylogenetic relationships within induced by the nematine gallers. Detailed the genus Salix. Currently, the interpretation knowledge about the chemistry of the willow of possible macroevolutionary patterns in the hosts would also be valuable when the evolution of host use in the nematines is chemistry of galls is studied. It is likely that limited by the lack of knowledge about the the galls induced by, for example, leaf blade relationships between the willow hosts. gallers are chemically quite different from Since morphological studies have not suc- those induced by the apical leaf gallers ceeded in elucidating the phylogenetic rela- studied in (IV). This can be suspected since tionships within the genus Salix, one can the coloration of the leaf blade galls is usu- only hope that molecular methods will be ally greenish, in contrast to the pale galls more successful. Thus far, the results have induced by the apical leaf gallers. Likewise, not been promising, but hopefully something the color of Euura galls is also usually will emerge from the intensive research that greenish. The interesting question is whether is currently being carried out on molecular different galler groups produce chemically Salix taxonomy. It should be noted that similar galls, or whether there are differences knowing the phylogeny of the Salix would between the galler groups. help in many other research areas as well. Evidently, the end goal must be a situa- (3) The chemical properties of Salix spe- tion in which detailed information on the cies. With improved methods of chemical phylogeny of the nematines can be combined analyses, the identification and quantifica- with information on the phylogeny of the tion of compounds has become easier. Cur- willow hosts, and with information on the rently it is possible to reliably measure many chemical properties of both the hosts and the more compounds than in the pioneering galls. This would make it possible to study studies of Julkunen-Tiitto (1986, 1989). A how the speciation patterns evolve in rela- thorough chemical analysis of northern wil- tion to host chemistry and phylogeny, and low species would be extremely valuable for how the chemistry of the galls changes dur- interpreting the results from the phylogenetic ing the evolutionary history of the sawflies. 28

ACKNOWLEDGEMENTS

I wish to extend special thanks to my advisors, professor Jorma Tahvanainen and Dr. Heikki Roininen, for their advice and comments during the work for my thesis. Alexey Zinovjev has been an invaluable provider of samples, information, and criticism throughout the process. I also thank my “extra advisors” from the Graduate School of Biology and Biotechnology of Forest Trees, professors Outi Savolainen and Tuomas Sopanen, for their interest and support. Liselotte Sundström and Jaakko Lumme kindly pre-examined the thesis. The research has been done mainly at the Department of Biology, University of Joensuu, and at the Geobotanical Institute, ETH Zürich. I thank professor Heikki Hyvärinen, the head of the Department of Biology in Joensuu, for providing the excellent facilities. The Kilpis- järvi and Abisko research stations provided accommodation during the collection trips. It has been a pleasure to work with my coauthors Jukka Vuorinen, Alex Widmer, and Riitta Julkunen-Tiitto, and I thank Outi Nousiainen and Elke Karaus for their skillful work in the laboratory. My friends and colleagues in the “Herbivory Research Group”, and in the “Discussing and Critical Scientific Community” (= room 173b) have created a pleasant working atmosphere. The days have been made a lot easier by the relaxing coffee table con- versations in cafeteria Kuutti, in which especially Olli-Pekka Tikkanen, Mervi Kunnasranta, Johanna Witzell, and Susanna Nuutinen have amazed me with their ability to avoid any serious subjects. Life in general has been made easy by the support from my parents. Funding for this thesis was provided by the University of Joensuu Faculty of Mathemat- ics and Natural Sciences, the Emil Aaltonen Foundation, and the Finnish Academy (Finnish Centre of Excellence Program 2000-2005). 29

REFERENCES Becerra, J. X. 1997. Insects on plants: macroevolutionary chemical trends in Abrahamson, W. G. & McCrea, K. D. 1986. host use. Science 276: 253-256. Nutrient and biomass allocation in Soli- Benson, B. R. 1958. Hymenoptera 2. Sym- dago altissima: effects of two stem gall- phyta. Handbooks for the Identification makers, fertilization, and ramet isolation. of British Insects VI(2c): 139-252. Oecologia 68: 174-180. Berenbaum, M. R. 1995. The chemistry of Abrahamson, W. G. & Weis, A. E. 1987. defense: theory and practice. Proc. Natl. Nutritional ecology of arthropod gall Acad. Sci. USA 92: 2-8. makers. In: Nutritional Ecology of In- Berenbaum, M. & Zangerl, A. 1999. Genetic sects, Mites, Spiders, and Related Inver- variation in cytochrome P450-based re- tebrates (F. Slansky, Jr. & J. G. Rod- sistance to plant allelochemicals and in- rigues, eds.), pp. 235-258. John Wiley & secticides. In: Herbivores: Between Sons, New York. Plants and Predators (H. Olff, V. K. Abrahamson, W. G., McCrea, K. D., Whit- Brown & R. H. Drent, eds.), pp. 55-84. well, A. J. & Vernieri, L. A. 1991. The Blackwell, Cambridge. role of phenolics in goldenrod ball gall Bernays, E. & Graham, M. 1988. On the resistance and formation. Biochem. Syst. evolution of host specificity in phytopha- Ecol. 19: 615-622. gous arthropods. Ecology 69: 886-892. Ananthakrishnan, T. N. & Gopichandran, R. Bremer, K. 1988. The limits of amino acid 1993. Chemical Ecology in Thrips-Host sequence data in angiosperm phyloge- Plant Interactions. Science Publishers, netic reconstruction. Evolution 42: 795- Enfield, NH. 803. Andersen, P. C. & Mizell, R. F, III. 1987. Bremer, K. 1994. Branch support and tree Physiological effects of galls induced by stability. Cladistics 10: 295-304. Phylloxera notabilis (Homoptera: Phyl- Brewer, J. W., Bishop, J. N. & Skuhravý, V. loxeridae) on pecan foliage. Environ. 1987. Levels of foliar chemicals in in- Entomol. 16: 264-268. sect-induced galls (Dipt., Cecidomyii- Argus, G. W. 1973. The Genus Salix in dae). J. Appl. Ent. 104: 504-510. Alaska and the Yukon. Nat. Mus. Canada, Bronner, R. 1992. The role of nutritive cells Ottawa. in the nutrition of cynipids and cecido- Argus, G. W. 1997. Infrageneric classifica- myiids. In: Biology of Insect-Induced tion of the genus Salix (Salicaceae) in the Galls (J. D. Shorthouse & O. Rohfritsch, New World. Syst. Bot. Monogr. 52: 1- eds.), pp. 118-140. Oxford Univ. Press, 121. New York. Ayres, M. P., Clausen, T. P., MacLean, S. F., Brunsfeld, S. J., Soltis, D. E. & Soltis, P. S. Jr., Redman, A. M. & Reichardt, P. B. 1992. Evolutionary patterns and proc- 1997. Diversity of structure and antiher- esses in Salix sect. Longifoliae: evidence bivore activity in condensed tannins. form chloroplast DNA. Syst. Bot. 17: Ecology 78: 1696-1712. 239-256. Azuma, T., Kajita, T., Yokoyama, J. & Bush, G. L. 1969. Sympatric host race for- Ohashi, H. 2000. Phylogenetic relation- mation and speciation in frugivorous flies ships of Salix (Salicaceae) based on rbcL of the genus Rhagoletis (Diptera, Tephri- sequence data. Am. J. Bot. 87: 67-75. tidae). Evolution 23: 237-251. Barraclough, T. G. & Vogler, A. P. 2000. Bush, G. L. 1975. Sympatric speciation in Detecting the geographical pattern of phytophagous parasitic insects. In: Evo- speciation from species-level phyloge- lutionary Strategies of Parasitic Insects nies. Am. Nat. 155: 419-434. and Mites (P. W. Price, ed.), pp. 187-206. Plenum Press, New York. 30

Bush, G. L. 1994. Sympatric speciation in Cornell, H. V. 1983. The secondary chemis- animals: new wine in old bottles. Trends try and complex morphology of galls Ecol. Evol. 9: 285-288. formed by the Cynipinae (Hymenoptera): Buth, D. G. 1984. The application of elec- why and how? Am. Midl. Nat. 110: 225- trophoretic data in systematic studies. 234. Annu. Rev. Ecol. Syst. 15: 501-522. Courtney, S. P. 1983. Models of hostplant Carroll, S. P. & Boyd, C. 1992. Host race location by butterflies: the effect of radiation in the soapberry bug: natural search images and search efficiency. Oe- history with the history. Evolution 46: cologia 59: 317-321. 1052-1069. Craig, T. P., Itami, J. K., Abrahamson, W. G. Carroll, S. P., Dingle, H. & Klassen, S. P. & Horner, J. D. 1993. Behavioral evi- 1997. Genetic differentiation of fitness- dence for host-race formation in Eurosta associated traits among rapidly evolving solidaginis. Evolution 47: 1696-1710. populations of the soapberry bug. Evolu- Craig, T. P., Itami, J. K., Horner, J. D. & tion 51: 1182-1188. Abrahamson, W. G. 1994. Host shifts Carroll, S. P., Klassen, S. P. & Dingle, H. and speciation in gall-forming insects. In: 1998. Rapidly evolving adaptations to The Ecology and Evolution of Gall- host ecology and nutrition in the soap- Forming Insects (P. W. Price, W. J. berry bug. Evol. Ecol. 12: 955-968. Mattson & Y. N. Baranchikov, eds.), pp. Charnov, E. L., Los-den Hartogh, R. L., 194-207. USDA Forest Service, St. Paul, Jones, W. T. & van den Assem, J. 1981. MN. Sex ratio evolution in a variable envi- Crespi, B. & Worobey, M. 1998. Compara- ronment. Nature 289: 27-33. tive analysis of gall morphology in Aus- Chong, D. K. X., Zsuffa, L. & Aravanopou- tralian gall thrips: the evolution of ex- los, F. A. 1995. Genetic relationships tended phenotypes. Evolution 52: 1686- between Salix exigua and other North 1696. American willows (Salix L.): evidence Crespi, B. J. & Sandoval, C. P. 2000. Phylo- from allozyme variation. Biochem. Syst. genetic evidence for the evolution of Ecol. 23: 767-771. ecological specialization in Timema Clark, M. A., Moran, N. A., Baumann, P. & walking-sticks. J. Evol. Biol. 13: 249- Wernegreen, J. J. 2000. Cospeciation 262. between bacterial endosymbionts Crespi, B. J., Carmean, D. A. & Chapman, (Buchnera) and a recent radiation of T. W. 1997. Ecology and evolution of aphids (Uroleucon) and pitfalls of testing galling thrips and their allies. Annu. Rev. for phylogenetic congruence. Evolution Entomol. 42: 51-71. 54: 517-525. Csóka, G. 1994. Variation in Quercus robur Collinson, M. E. 1992. The early fossil his- susceptibility to galling wasps (Hy- tory of Salicaceae: a brief review. Proc. menoptera: Cynipidae) linked to tree R. Soc. Edinb. 98B: 155-167. phenology. In: The Ecology and Evolu- Condon, M. A. & Steck, G. J. 1997. Evolu- tion of Gall-Forming Insects (P. W. tion of host use in fruit flies of the genus Price, W. J. Mattson & Y. N. Baran- Blepharoneura (Diptera: Tephritidae): chikov, eds.), pp. 148-152. USDA Forest cryptic species on sexually dimorphic Service, St. Paul, MN. host plants. Biol. J. Linn. Soc. 60: 443- Damman, H. 1987. Leaf quality and enemy 466. avoidance by the larvae of a pyralid Cooper-Driver, G. A. & Bhattacharya, M. moth. Ecology 68: 88-97. 1998. Role of phenolics in plant evolu- Davey, M. R., Curtis, I. S., Gartland, G. M. tion. Phytochemistry 49: 1165-1174. A. & Power, J. B. 1994. Agrobacterium- induced crown gall and hairy root dis- 31

eases: their biology and application to Dreger-Jauffret, F. & Shorthouse, J. D. 1992. plant genetic engineering. In: Plant Diversity of gall-inducing insects and Galls: Organisms, Interactions, Popula- their galls. In: Biology of Insect-Induced tions (M. A. J. Williams, ed.), pp. 9-56. Galls (J. D. Shorthouse & O. Rohfritsch, Oxford University Press, New York. eds.), pp. 8-33. Oxford Univ. Press, New Dawkins, R. 1982. The Extended Phenotype. York. Oxford Univ. Press, Oxford, U.K. Dreyer, D. L. & Jones, K. C. 1981. Feeding DeBruyn, L. 1994. Life cycle strategies in a deterrency of flavonoids and related phe- guild of dipteran gall formers on the nolics towards Schizaphis graminum and common reed. In: Plant Galls: Organ- Myzus persicae: aphid feeding deterrents isms, Interactions, Populations (M. A. J. in wheat. Phytochemistry 20: 2489-2493. Williams, ed.), pp. 259-281. Oxford Ehrlich, P. R. & Raven, P. H. 1964. Butter- Univ. Press, New York. flies and plants: a study in coevolution. DeBruyn, L., Vandevyvere, I., Jaminé, D. & Evolution 18: 586-608. Prinsen, E. 1998. The effects of gall for- Elliger, C. A., Chan, B. C. & Waiss, A. C., mation by Lipara lucens (Diptera, Chlo- Jr. 1980. Flavonoids as larval growth in- ropidae) on its host Phragmites australis hibitors. Structural factors governing (Poaceae). In: The Biology of Gall- toxicity. Naturvissenschaften 67: 358- Inducing Arthropods (G. Csóka, W. J. 360. Mattson, G. N. Stone & P. W. Price, Emelianov, I., Mallet, J. & Baltensweiler, W. eds.), pp. 173-187. USDA Forest Service, 1995. Genetic differentiation in Zeira- St. Paul, MN. phera diniana (Lepidoptera: Tortricidae, Dethier, V. G. 1954. Evolution of feeding the larch budmoth): polymorphism, host preferences in phytophagous insects. races or sibling species? Heredity 75: Evolution 8: 33-54. 416-424. Dieckmann, U. & Doebeli, M. 1999. On the Farrell, B. D., Mitter, C. & Futuyma, D. J. origin of species by sympatric speciation. 1992. Diversification at the insect-plant Nature 400: 354-357. interface: insights form phylogenetics. Diehl, S. R. & Bush, G. L. 1984. An evolu- BioScience 42: 34-42. tionary and applied perspective of insect Feder, J. L., Chilcote, C. A. & Bush, G. L. biotypes. Annu. Rev. Entomol. 29: 471- 1988. Genetic differentiation between 504. sympatric host races of the apple maggot Diehl, S. R. & Bush, G. L. 1989. The role of fly Rhagoletis pomonella. Nature 336: habitat preference in adaptation and spe- 61-64. ciation. In: Speciation and its Conse- Feder, J. L., Roethele, J. B., Wlazlo, B. & quences (D. Otte & J. A. Endler, eds.), Berlocher, S. H. 1997a. Selective main- pp. 345-365. Sinauer, Sunderland, MA. tenance of allozyme differences among Dobler, S., Mardulyn, P., Pasteels, J. M. & sympatric host races of the apple maggot Rowell-Rahier, M. 1996. Host-plant fly. Proc. Natl. Acad. Sci. USA 94: switches and the evolution of chemical 11417-11421. defense and life history in the leaf beetle Feder, J. L., Stolz, U., Lewis, K. M., Perry, genus Oreina. Evolution 50: 2373-2386. W., Roethele, J. B. & Rogers, A. 1997b. Dodson, G. N. 1991. Control of gall mor- The effects of winter length on the ge- phology: tephritid gallformers (Aciurina netics of apple and hawthorn races of spp.) on rabbitbrush (Chrysothamnus). Rhagoletis pomonella (Diptera: Tephriti- Ecol. Entomol. 16: 177-181. dae). Evolution 51: 1862-1876. Dorn, R. D. 1976. A synopsis of American Felsenstein, J. 1985. Confidence limits on Salix. Can. J. Bot. 54: 2769-2789. phylogenies: an approach using the bootstrap. Evolution 39: 783-791. 32

Filchak, K. E., Roethele, J. B. & Feder, J. L. Hagerman, A. E. & Butler, L. G. 1992. Tan- 2000. Natural selection and sympatric di- nins and lignins. In: Herbivores: Their vergence in the apple maggot Rhagoletis Interactions with Secondary Plant Me- pomonella. Nature 407: 739-742. tabolites, 2nd ed., Vol. I. The Chemical Fraenkel, G. S. 1959. The raison d’être of Participants (G. A. Rosenthal & M. R. secondary plant substances. Science 129: Berenbaum, eds.), pp. 355-388. Aca- 1466-1470. demic Press, New York. Fritz, R. S., Sacchi, C. F. & Price, P. W. Hagerman, A. E. & Butler, L. G. 1994. As- 1986. Competition versus host plant phe- say for condensed tannins or flavonoid notype in species composition: willow oligomers and related flavonoids in sawflies. Ecology 67: 1608-1618. plants. Meth. Enzymol. 234: 429-437. Futuyma, D. J. 1983. Evolutionary interac- Harborne, J. B. & Grayer, R. J. 1993. Flavo- tions among herbivorous insects and noids and insects. In: The Flavonoids: plants. In: Coevolution (D. J. Futuyma & Advances in Research since 1986 (J. B. M. Slatkin, eds.), pp. 207-231. Sinauer, Harborne, ed.), pp. 589-618. Chapman & Sunderland, MA. Hall, London. Futuyma, D. J. & McCafferty, S. S. 1990. Hardig, T. M., Brunsfeld, S. J., Fritz, R. S., Phylogeny and the evolution of host plant Morgan, M. & Orians, C. M. 2000. Mor- associations in the leaf beetle genus phological and molecular evidence for Ophraella (Coleoptera, Chrysomelidae). hybridization and introgression in a wil- Evolution 44: 1885-1913. low (Salix) hybrid zone. Mol. Ecol. 9: 9- Futuyma, D. J. & Keese, M. C. 1992. Evolu- 24. tion and coevolution of plants and phy- Harris, K. M. 1994. Gall midges (Cecido- tophagous arthropods. In: Herbivores: myiidae): classification and biology. In: Their Interactions with Secondary Plant Plant Galls: Organisms, Interactions, Metabolites, 2nd ed., Vol. II. Ecological Populations (M. A. J. Williams, ed.), pp. and Evolutionary Processes (G. A. 201-211. Oxford University Press, New Rosenthal & M. R. Berenbaum, eds.), pp. York. 439-475. Academic Press, New York. Hartley, S. E. 1992. The insect galls on wil- Gauld, I. & Bolton, B. 1988. The Hymenop- low. Proc. R. Soc. Edinb. 98B: 91-104. tera. Oxford Univ. Press, Oxford. Hartley, S. E. 1998. The chemical composi- Gilbert. F., Astbury, C., Bedingfield, J., En- tion of plant galls: are levels of nutrients nis, B., Lawson, S. & Sitch, T. 1994. The and secondary compounds controlled by ecology of the pea galls of Cynips divisa. the gall-former? Oecologia 113: 492- In: Plant Galls: Organisms, Interactions, 501. Populations (M. A. J. Williams, ed.), pp. Hartley, S. E. 1999. Are gall insects large 331-349. Clarendon, Oxford. rhizobia? OIKOS 84: 333-342. Gopichandran, R., Peter, A. J. & Subrama- Hartley, S. E. & Lawton, J. H. 1992. Host- niam, V. R. 1992. Age-correlated bio- plant manipulation by gall-insects: a test chemical profiles of thrips galls in rela- of the nutrition hypothesis. J. Anim. Ecol. tion to population density of thrips. J. 61: 113-119. Nat. Hist. 26: 609-619. Haukioja, E. & Neuvonen, S. 1985. The re- Groman, J. D. & Pellmyr, O. 2000. Rapid lationship between size and reproductive evolution and specialization following potential in male and female Epirrita host colonization in a yucca moth. J. autumnata (Lep., Geometridae). Ecol. Evol. Biol. 13: 223-236. Entomol. 10: 267-270. Hafner, M. S. & Nadler, S. A. 1988. Phylo- Herbst, J. & Heitland, W. 1994. Genetic dif- genetic trees support the coevolution of ferentiation among populations of the parasites and hosts. Nature 332: 258-259. sawfly-species Platycampus luridiven- 33

tris, associated with different Alder spe- nolic metabolism of silver birch: effects cies (Hymenoptera: Tenthredinidae). of fertilization, defoliation, and genotype. Entomol. Gener. 19: 39-48. Ecology 80: 1970-1986. Herms, D. A. & Mattson, W. J. 1992. The Kelley, S. T. & Farrell, B. D. 1998. Is spe- dilemma of plants: to grow or to defend. cialization a dead end? The phylogeny of Q. Rev. Biol. 67: 283-335. host use in Dendroctonus bark beetles Higton, R. N. & Mabberley, D. J. 1994. A (Scolytidae). Evolution 52: 1731-1743. willow gall form the galler’s point of Kennedy, C. E. J. & Southwood, T. R. E. view. In: Plant Galls: Organisms, Inter- 1984. The number of species of insects actions, Populations (M. A. J. Williams, associated with British trees: a reed.), pp. 301-312. Oxford Univ. Press, analysis. J. Anim. Ecol. 53: 455-478. New York. Kokkonen, K. 2000. Mixed significance of Hori, K. 1992. Insect secretions and their plant vigor: two species of galling Pon- effect on plant growth, with special ref- tania in a hybridizing willow complex. erence to hemipterans. In: Biology of In- OIKOS 90: 97-106. sect-Induced Galls (J. D. Shorthouse & Kolehmainen, J., Roininen, H., Julkunen- O. Rohfritsch, eds.), pp. 157-170. Oxford Tiitto, R. & Tahvanainen, J. 1994. Im- Univ. Press, New York. portance of phenolic glucosides in host Hämet-Ahti, L., Suominen, J., Ulvinen, T. & selection of shoot galling sawfly, Euura Uotila, P. 1998. Retkeilykasvio (Field amerinae, on Salix pentandra. J. Chem. Flora of Finland), 4th ed. Finnish Mu- Ecol. 20: 2455-2466. seum of Natural History, Botanical Mu- Kolehmainen, J., Julkunen-Tiitto, R., Ro- seum, Helsinki, Finland. ininen, H. & Tahvanainen, J. 1995. Phe- Janz, N. & Nylin, S. 1997. The role of fe- nolic glucosides as feeding cues for wil- male search behaviour in determining low-feeding leaf beetles. Entomol. Exp. host plant range in plant feeding insects: Appl. 74: 235-243. a test of the information processing hy- Kontuniemi, T. 1960. Suomen sahapistiäis- pothesis. Proc. R. Soc. Lond. B 264: 701- toukkien ravintokasvit (Die Futterpflan- 707. zen der Sägewespenlarven (Hymenop- Jermy, T. 1984. Evolution of insect/host tera: Symphyta) Finnlands). Anim. Fenn. plant relationships. Am. Nat. 124: 609- 9: 1-104. 630. Kopelke, J.-P. 1985. Über die Biologie und Julkunen-Tiitto, R. 1986. A chemotax- Parasiten der gallenbildenden Blatt- onomic survey of phenolics in leaves of wespenarten Pontania dolichura (Thoms. Northern Salicaceae species. Phyto- 1871), P. vesicator (Bremi 1849) und P. chemistry 25: 663-667. viminalis (L. 1758) (Hymenoptera: Julkunen-Tiitto, R. 1989. Phenolic constitu- Tenthredinidae). Faun. Oekol. Mitt. 5: ents of Salix: a chemotaxonomic survey 331-344. of further Finnish species. Phytochemis- Kopelke, J.-P. 1990a. Wirtsspezifität als Dif- try 28: 2115-2125. ferenzierungskriterium bei Gallenbilden- Julkunen-Tiitto, R., Rousi, M., Bryant, J., den Blattwespenarten der Gattung Pon- Sorsa, S., Keinänen, M. & Sikanen, H. tania O. Costa (Hymenoptera: Tenthre- 1996. Chemical diversity of several Be- dinidae: Nematinae). Mitt. Dtsch. Ges. tulaceae species: comparison of phe- Allg. Angew. Ent. 7: 527-534. nolics and terpenoids in northern birch Kopelke, J.-P. 1990b. Der taxonomische stems. Trees 11: 16-22. Status von Pontania dolichura (Thomson Keinänen, M., Julkunen-Tiitto, R., Muti- 1871)(Insecta: Hymenoptera: Tenthre- kainen, P., Walls, M., Ovaska, J. & dinidae). Senckenb. Biol. 70: 271-279. Vapaavuori, E. 1999. Trade-offs in phe- 34

Kopelke, J.-P. 1991. Die Arten der vimina- Leitch, I. J. 1994. Induction and develop- lis-Gruppe, Gattung Pontania O. Costa ment of the bean gall caused by Pontania 1859, Mittel- und Nordeuropas (Insecta: proxima. In: Plant Galls: Organisms, Hymenoptera: Tenthredinidae). Senck- Interactions, Populations (M. A. J. Wil- enb. Biol. 71: 65-128. liams, ed.), pp. 283-300. Oxford Univ. Kopelke, J.-P. 1994. Der Schmarotzerkom- Press, New York. plex (Brutparasiten und Parasitoide) der Leskinen, E. & Alström-Rapaport, C. 1999. gallenbildenden Pontania-Arten (Insecta: Molecular phylogeny of Salicaceae and Hymenoptera: Tenthredinidae). Senck- closely related Flacourtiaceae: evidence enb. Biol. 73: 83-133. from 5.8 S, ITS 1, and ITS 2 of the Kopelke, J.-P. 1996. Die Euura atra- und rDNA. Pl. Syst. Evol. 215: 209-227. amerinae-Gruppe in Nord- und Mitteleu- Levins, R. & MacArthur, R. 1969. An hy- ropa (Insecta: Hymenoptera: Tenthre- pothesis to explain the incidence of dinidae: Nematinae). Senckenb. Biol. 76: monophagy. Ecology 50: 910-911. 93-113. Lindroth, R. L. & Bloomer, M. S. 1991. Bio- Kopelke, J.-P. 1998. Eiablage-Strategien bei chemical ecology of the forest tent cater- gallenbildenden Arten der Blattwespen- pillar: responses to dietary protein and Gattungen Pontania, Euura und Phyllo- phenolic glycosides. Oecologia 86: 408- colpa (Hymenoptera: Tenthredinidae: 413. Nematinae). Entomol. Gener. 22: 251- Lindroth, R. L. & Hwang, S.-Y. 1996. Di- 275. versity, redundancy, and multiplicity in Kouki, J., Niemelä, P. & Viitasaari, M. 1994. chemical defense systems of aspen. Re- Reversed latitudinal gradient in species cent Adv. Phytochem. 30: 25-56. richness of sawflies (Hymenoptera, Lindroth, R. L., Scriber, J. M. & Hsia, M. T. Symphyta). Ann. Zool. Fenn. 31: 83-88. S. 1988. Chemical ecology of the tiger Köpf, A., Rank, N. E., Roininen, H., swallowtail: mediation of host use by Julkunen-Tiitto, R., Pasteels, J. M. & Ta- phenolic glycosides. Ecology 69: 814- hvanainen, J. 1998. The evolution of 822. host-plant use and sequestration in the Mabee, P. M. & Humphries, J. 1993. Coding leaf beetle genus Phratora (Coleoptera: polymorphic data: examples from allo- Chrysomelidae). Evolution 52: 517-528. zymes and ontogeny. Syst. Biol. 42: 166- Labandeira, C. C. & Phillips, T. L. 1996. A 181. Carboniferous insect gall: insight into Maddison, W. P. & Maddison, D. R. 1992. early ecologic history of the Holome- MacClade: analysis of phylogeny and tabola. Proc. Natl. Acad. Sci USA 93: character evolution. Ver. 3.01. Sinauer, 8470-8474. Sunderland, MA. Larew, H. G. 1992. Fossil galls. In: Biology Mardulyn, P. & Pasteels, J. M. 1994. Coding of Insect-Induced Galls (J. D. Shorthouse allozyme data using step matrices: de- & O. Rohfritsch, eds.), pp. 50-59. Oxford fining new original states for the ances- Univ. Press, New York. tral taxa. Syst. Biol. 43: 567-572. Larson, K. C. & Whitham, T. G. 1991. Ma- Mapes, C. C. & Davies, P. J. 1998. Plant nipulation of food resources by a gall- hormones and gall formation by Eurosta forming aphid: the physiology of sink- solidaginis on Solidago altissima. In: The source interactions. Oecologia 88: 15-21. Biology of Gall-Inducing Arthropods (G. Larson, K. C. & Whitham, T. G. 1997. Csóka, W. J. Mattson, G. N. Stone & P. Competition between gall aphids and W. Price, eds.), pp. 161-172. USDA For- natural plant sinks: plant architecture af- est Service, St. Paul, MN. fects resistance to galling. Oecologia Matsuda, K. & Senbo, S. 1986. Chlorogenic 109: 575-582. acid as a feeding deterrent for the Salica- 35

ceae-feeding leaf beetle, Lochmaeae Murakami, M. 1999. Effect of avian preda- capreae cribrata (Coleoptera: Chryso- tion on survival of leaf-rolling lepidop- melidae) and other species of leaf bee- terous larvae. Res. Popul. Ecol. 41: 135- tles. Appl. Ent. Zool. 21: 411-416. 138. Mayr, E. 1963. Animal Species and Evolu- Mutikainen, P., Walls, M., Ovaska, J., tion. Harvard Univ. Press, Cambridge, Keinänen, M., Julkunen-Tiitto, R. & MA. Vapaavuori, E. 2000. Herbivore resis- McCalla, D. R, Genthe, M. K. & Howanitz, tance in Betula pendula: effect of fertili- W. 1962. Chemical nature of an insect zation, defoliation, and plant genotype. gall growth factor. Plant Physiol. 37: 48- Ecology 81: 49-65. 103. Nylin, S. & Janz, N. 1999. The ecology and McKenzie, J. A. & Batterham, P. 1994. The evolution of host plant range: butterflies genetic, molecular, and phenotypic con- as a model group. In: Herbivores: Be- sequences of selection for insecticide re- tween Plants and Predators (H. Olff, V. sistance. Trends Ecol. Evol. 9: 166-169. K. Brown & R. H. Drent, eds.), pp. 31- McPheron, B. A., Smith, D. C. & Berlocher, 54. Blackwell, Cambridge. S. H. 1988. Genetic differences between Page, R. D. M. 1995. TreeMap. Ver. 1.0b. host races of Rhagoletis pomonella. Na- Institute of Biomedical and Life Sci- ture 336: 64-66. ences, University of Glasgow, U.K. Meikle, R. D. 1992. British willows; some Palo, R. T. 1984. Distribution of birch (Be- hybrids and some problems. Proc. R. tula spp.), willow (Salix spp.), and poplar Soc. Edinb. 98B: 13-20. (Populus spp.) secondary metabolites and Menken, S. B. J. 1981. Host races and sym- their potential role as chemical defense patric speciation in small ermine moths, against herbivores. J. Chem. Ecol. 10: Yponomeutidae. Ent. Exp. Appl. 30: 280- 499-520. 292. Pasteels, J. M. & Rowell-Rahier, M. 1992. Meyer, J. 1987. Plant Galls and Gall Induc- The chemical ecology of herbivory on ers. Gebrüder Borntraeger, Berlin. willows. Proc. R. Soc. Edinb. 98B: 63- Miller, M. P. 1997. Tools for population 73. genetics analyses (TFPGA) 1.3: A Win- Peries, I. D. R. 1994. Studies on host-plant dows program for the analysis of allo- resistance in rice to gall midge Orseolia zyme and molecular population genetic oryzae. In: Plant Galls: Organisms, In- data. Computer software distributed by teractions, Populations (M. A. J. Wil- author. liams, ed.), pp. 231-243. Oxford Univ. Mitter, C. & Brooks, D. R. 1983. Phyloge- Press, New York. netic aspects of coevolution. In: Coevo- Pimentel, D. & Bellotti, A. C. 1976. Para- lution (D. J. Futuyma & M. Slatkin, site-host systems and genetic stability. eds.), pp. 65-98. Sinauer, Sunderland, Am. Nat. 110: 877-888. MA. Price, P. W. 1992. Evolution and ecology of Mitter, C., Farrell, B. & Futuyma, D. J. gall-inducing sawflies. In: Biology of In- 1991. Phylogenetic studies of insect- sect-Induced Galls (J. D. Shorthouse & plant interactions: insights into the gene- O. Rohfritsch, eds.), pp. 208-224. Oxford sis of diversity. Trends Ecol. Evol. 6: Univ. Press, Oxford. 290-293. Price, P. W. & Clancy, K. M. 1986. Interac- Moran, N. A., Munson, M. A., Baumann, P. tions between three trophic levels: gall & Ishikawa, H. 1993. A molecular clock size and parasitoid attack. Ecology 67: in endosymbiotic bacteria is calibrated 1593-1600. using the insect hosts. Proc. R. Soc. Price, P. W. & Roininen, H. 1993. Adaptive Lond. B 253: 167-171. radiation in gall induction. In: Sawfly Life 36

History Adaptations to Woody Plants (M. Roininen, H., Ohgushi, T., Zinovjev, A., Wagner & K. F. Raffa, eds.), pp. 229- Nyman, T., Vikberg, V., Matsushita, K. 257. Academic Press, New York. & Nakamura, M. 1999b. External feeding Price, P. W., Waring, G. L. & Fernandes, G. of leaf galler: intermediate behaviour W. 1986. Hypotheses on the adaptive between leaf folders (Phyllocolpa spp.) nature of galls. Proc. Entomol. Soc. and leaf gallers (Pontania spp.). XIV Wash. 88: 361-363. Population Ecology Symposium on Radkey, R. R. & Singer, M. C. 1995. Re- plant-herbivore interactions and inter- peated reversal of host-preference evolu- specific interactions in tropical regions. tion in a specialist insect herbivore. October 22-24, 1999, Takashima-gun, Evolution 49: 351-359. Shiga, Japan. Rechinger, K. H. 1992. Salix taxonomy in Roininen, H., Danell, K., Zinovjev, A. G. & Europe – problems, interpretations, ob- Vikberg, V. 1999c. The community servations. Proc. R. Soc. Edinb. 98B: 1- structure, survival and mortality factors 12. in Arctic populations of Pontania leaf Rey, L. A. 1992. Developmental morphol- gallers. Submitted. ogy of two types of hymenopterous galls. Roskam, J. C. 1992. Evolution of the gall- In: Biology of Insect-Induced Galls (J. D. inducing guild. In: Biology of Insect- Shorthouse & O. Rohfritsch, eds.), pp. Induced Galls (J. D. Shorthouse & O. 87-101. Oxford Univ. Press, New York. Rohfritsch, eds.), pp. 34-49. Oxford Roininen, H. 1988. Field test of resource Univ. Press, New York. regulation by the bud-galling sawfly, Sagers, C. L. 1992. Manipulation of host Euura mucronata, on Salix cinerea. Hol. plant quality: herbivores keep leaves in Ecol. 11: 136-139. the dark. Funct. Ecol. 6: 741-743. Roininen, H. 1991a. The ecology and evolu- Schneider, S., Roessli, D. & Excoffier, L. tion of host plant relationships among 2000. Arlequin. Ver. 2.000. A software willow-feeding sawflies. Univ. Joensuu for population genetics data analysis. Publ. Sci. 23: 1-21. Genetics and Biometry Laboratory, Uni- Roininen, H. 1991b. Temporal change in the versity of Geneva, Switzerland. location of egg-laying by a bud-galling Schultz, B. B. 1992. Insect herbivores as sawfly, Euura mucronata, on growing potential causes of mortality and adapta- shoots of Salix cinerea. Oecologia 87: tion in gallforming insects. Oecologia 90: 265-269. 297-299. Roininen, H. & Tahvanainen, J. 1989. Host Scott, A. C., Stephenson, J. & Collinson, M. selection and larval performance of two E. 1994. The fossil record of leaves with willow-feeding sawflies. Ecology 70: galls. In: Plant Galls: Organisms, Inter- 129-136. actions, Populations (M. A. J. Williams, Roininen, H., Vuorinen, J., Tahvanainen, J. ed.), pp. 447-470. Oxford Univ. Press, & Julkunen-Tiitto, R. 1993. Host prefer- New York. ence and allozyme differentiation in Seigler, D. S. 1998. Plant Secondary Me- shoot galling sawfly, Euura atra. Evolu- tabolism. Kluwer Academic Publishers, tion 47: 300-308. London. Roininen, H., Price, P. W., Julkunen-Tiitto, Seppänen, E. J. 1970. Suomen suurperhos- R., Tahvanainen, J. & Ikonen, A. 1999a. toukkien ravintokasvit (Die Nahrungs- Oviposition stimulant for a gall-inducing pflanzen der Großschmetterlingsraupen sawfly, Euura lasiolepis, on willow is a Finnlands). Anim. Fenn. 14: 1-179. phenolic glucoside. J. Chem. Ecol. 25: Shannon, R. E. & Brewer, J. W. 1980. Starch 943-953. and sugar levels in three coniferous insect galls. Z. Ang. Ent. 89: 526-533. 37

Shao, Y. 1991. Phytochemischer Atlas der Smith, E. L. 1979. Symphyta. In: Catalog of Schweizer Weiden. Ph.D. Thesis, ETH Hymenoptera in America North of Mex- Zürich, Switzerland. ico (K. V. Krombein, P. D. Hurd, D. R. Shaver, T. N. & Lukefahr, M. J. 1969. Ef- Smith & B. D. Burks, eds.), pp. 3-137. fects of flavonoid pigments and gossypol Smithsonian Institution Press, Washing- on growth and development of the boll- ton, DC. worm, tobacco budworm, and pink boll- Stern, D. L. 1995. Phylogenetic evidence worm. J. Econ. Ent. 62: 643-646. that aphids, rather than plants, determine Simon, C., Frati, F., Beckenbach, A., Crespi, gall morphology. Proc. R. Soc. Lond. B B., Liu, H. & Flook, P. 1994. Evolution, 260: 85-89. weighting, and phylogenetic utility of Stone, G. N. & Cook, J. M. 1998. The mitochondrial gene sequences and a structure of cynipid oak galls: patterns in compilation of conserved polymerase the evolution of an extended phenotype. chain reaction primers. Ann. Entomol. Proc. R. Soc. Lond. B 265: 979-988. Soc. Amer. 87: 651-701. Strimmer, K., & von Haeseler, A. 1999. The Sipura, M. 1999. Tritrophic interactions: PUZZLE Software. Maximum likelihood willows, herbivorous insects and insec- analysis for nucleotide, amino acid, and tivorous birds. Oecologia 121: 537-545. two-state data. Ver. 4.0.1. Max-Planck Sitch, T. A., Grewcock, D. A. & Gilbert, F. Institute for Biochemistry, München. S. 1988. Factors affecting components of Sturgeon, K. B. & Mitton, J. B. 1986. Allo- fitness in a gall-making wasp (Cynips di- zyme and morphological differentiation visa Hartig). Oecologia 76: 371-375. of mountain pine beetles Dendroctonus Skvortsov, A. K. 1968. Willows of the USSR. ponderosae Hopkins (Coleoptera: Sco- A Taxonomic and Geographic Revision. lytidae) associated with host tree. Evolu- Publishing Office “Nauka”, Moscow (in tion 40: 290-302. Russian). Summers, C. B. & Felton, G. W. 1994. Pro- Skvortsov, A. K. 1999. Willows of Russia oxidant effects of phenolic acids on the and adjacent countries. Taxonomical and generalist herbivore Helicoverpa zea geographical revision. University of Jo- (Lepidoptera: Noctuidae): potential mode ensuu Faculty of Mathematics and Natu- of action for phenolic compounds in ral Sciences Report Series 39: 1-307. plant anti-herbivore chemistry. Insect Smiley, J. 1978. Plant chemistry and the Biochem. Molec. Biol. 24: 943-953. evolution of host specificity: new evi- Swofford, D. L. 1993. PAUP: phylogenetic dence from Heliconius and Passiflora. analysis using parsimony. Ver. 3.1.1. Il- Science 201: 745-747. linois Natural History Survey, Cham- Smith, D. C. 1988. Heritable divergence of paign, IL. Rhagoletis pomonella host races by sea- Swofford, D. L. 1998. PAUP*: phylogenetic sonal asynchrony. Nature 336: 66-67. analysis using parsimony (and other Smith, E. L. 1968. Biosystematics and mor- methods). Ver. 4.0b1. Sinauer, Sunder- phology of Symphyta. I. Stem-galling land, MA. Euura of the California region, and a new Tahvanainen, J., Julkunen-Tiitto, R. & Ket- female genitalic nomenclature. Ann. En- tunen, J. 1985. Phenolic glycosides gov- tomol. Soc. Amer. 61: 1389-1407. ern the food selection pattern of willow Smith, E. L. 1970. Biosystematics and mor- feeding leaf beetles. Oecologia 67: 52- phology of Symphyta. II. Biology of gall- 56. making nematine sawflies of the Califor- Taper, M. L. & Case, T. J. 1987. Interactions nia region. Ann. Entomol. Soc. Amer. 63: between oak tannins and parasite com- 35-51. munity structure: unexpected benefits of 38

tannins to cynipid gall-wasps. Oecologia population structure. Evolution 38: 1358- 71: 254-261. 1370. Taper, M. L., Zimmerman, E. M. & Case, T. Weis, A. E. 1996. Variable selection on Eu- J. 1986. Sources of mortality for a rosta’s gall size, III: can an evolutionary cynipid gall-wasp (Dryocosmus dubiosus response to selection be detected? J. (Hymenoptera: Cynipidae)): the impor- Evol. Biol. 9: 623-640. tance of the tannin/fungus interaction. Whitham, T. G. 1978. Habitat selection by Oecologia 68: 437-445. Pemphigus gall aphids in response to re- Tauber, C. A. & Tauber, M. J. 1989. Sym- source limitation and competition. Ecol- patric speciation in insects: perception ogy 59: 1164-1176. and perspective. In: Speciation and its Whitham, T. G. 1979. Territorial behaviour Consequences (D. Otte & J. A. Endler, of Pemphigus gall aphids. Nature 279: eds.), pp. 307-344. Sinauer, Sunderland, 324-325. MA. Whitham, T. G. 1992. Ecology of Pemphi- Thompson, J. N. 1994. The Coevolutionary gus gall aphids. In: Biology of Insect- Process. Univ. of Chicago Press, Chi- Induced Galls (J. D. Shorthouse & O. cago, IL. Rohfritsch, eds.), pp. 225-237. Oxford Thompson, J. N. 1999. What we know and Univ. Press, New York. do not know about coevolution: insect Wiklund, C. 1975. The evolutionary rela- herbivores and plants as a test case. In: tionship between adult oviposition pref- Herbivores: Between Plants and Preda- erences and larval host plant range in tors (H. Olff, V. K. Brown & R. H. Papilio machaon L. Oecologia 18: 185- Drent, eds.), pp. 7-30. Blackwell, Cam- 197. bridge. Wilf, P., Labandeira, C. C., Kress, W. J., Tikkanen, O.-P. 2000. Adaptation of a Gen- Staines, C. L., Windsor, D. M., Allen, A. eralist Moth, Operophtera brumata, to L. & Johnson, K. R. 2000. Timing the Host Plants. Ph.D. Thesis, University of radiations of leaf beetles: hispines on Joensuu, Finland. gingers from latest Cretaceous to recent. Ueno, T. 1999. Host-size-dependent sex ra- Science 289: 291-294. tio in a parasitoid wasp. Res. Popul. Ecol. Williams, A. G. & Whitham, T. G. 1986. 41: 47-57. Premature leaf abscission: an induced Viitasaari, M. & Vikberg, V. 1985. A plant defense against gall aphids. Ecol- checklist of the sawflies (Hymenoptera, ogy 67: 1619-1627. Symphyta) of Finland. Not. Entomol. 65: Wool, D. 1997. The shapes of insect galls: 1-17. insect control, plant constraints and phy- Vikberg, V. 1982. Notes on the taxonomy logeny. In: Ecology and Evolution of and nomenclature of some mainly Fen- Plant-Feeding Insects in Natural and noscandian sawflies (Hymenoptera: Man-Made Environments (A. Raman, Symphyta). Not. Entomol. 62: 61-65. ed.), pp. 203-212. International Scientific Vuorinen, J. 1984. Electrophoretic expres- Publications, New Delhi. sion of genetic variation and duplicate Wright, S. 1978. Evolution and the Genetics gene activity in vendace, Coregonus al- of Populations. Vol. 4. Variability Within bula (Salmonidae). Hereditas 101: 85-96. and Among Natural Populations. Univ. Wahlberg, N. 2000. The Ecology and Evolu- of Chicago Press, Chicago, IL. tion of Melitaeine Butterflies. Ph.D. The- Yang, C.-M., Yin, M.-H., Chang, K.-W., sis, University of Helsinki, Finland. Tsai, C.-J., Huang, S.-M. & Yang, M.-M. Weir, B. S. & Cockerham, C. C. 1984. Esti- 1998. Analysis of pigment-protein com- mating F-statistics for the analysis of plexes in two cecidomyiid galls. In: The Biology of Gall-Inducing Arthropods (G. 39

Csóka, W. J. Mattson, G. N. Stone & P. Zinovjev, A. G. 1995. The gall-making spe- W. Price, eds.), pp. 188-192. USDA For- cies of Pontania subgenus Eupontania est Service, St. Paul, MN. (Hymenoptera, Tenthredinidae) of East- Yang, M.-M. & Mitter, C. 1994. Biosys- ern Fennoscandia and their host plant tematics of hackberry psyllids (Pachyp- specificity. Acta Zool. Fennica 199: 49- sylla) and the evolution of gall and lerp 53. formation in psyllids (Homoptera: Psyl- Zinovjev, A. G. 1998. Palaearctic sawflies of loidea): a preliminary report. In: The the genus Pontania Costa (Hymenoptera: Ecology and Evolution of Gall-Forming Tenthredinidae) and their host-plant Insects (P. W. Price, W. J. Mattson & Y. specificity. In: The Biology of Gall- N. Baranchikov, eds.), pp. 172-185. Inducing Arthropods (G. Csóka, W. J. USDA Forest Service, St. Paul, MN. Mattson, G. N. Stone & P. W. Price, Zinovjev, A. G. 1993. Subgenera and Pa- eds.), pp. 204-225. USDA Forest Service, laearctic species groups of the genus St. Paul, MN. Pontania, with notes of the taxonomy of Zinovjev, A. G. & Vikberg, V. 1998. On the some European species of the viminalis- biology of Nematinae with hiding larvae. group (Hymenoptera: Tenthredinidae). Beitr. Ent. 48: 145-155. Zoosyst. Rossica 2: 145-154. Zinovjev, A. G. & Vikberg, V. 1999. The Zinovjev, A. G. 1994. Taxonomy and biol- sawflies of the Pontania crassispina- ogy of two related species of gall-making group with a key for the genera of the sawflies from the Pontania viminalis- subtribe Euurina (Hymenoptera: Tenthre- group (Hymenoptera: Tenthredinidae). dinidae, Nematinae). Ent. Scand. 30: Ent. Scand. 25: 231-240. 281-298.