Adaptive Nature of Insect Galls

PETER W. PRICE,l.2 C. WILSON FERNANDES,l

AND CWENDOL YN L. W ARINCl

FORUM: Environ. Entomol. 16: 15-24 (1987)ABSTRACT Major hypotheses on the adaptive significance of insect gall formation arereviewed: nonadaptive, plant protection, mutual benefit, nutrition, microenvironment, andenemy hypotheses. We evaluate the validity of each, and find the first three to be withoutmerit because galls clearly have adaptive features for the insect, but few if any for the plant,and the galler has negative impact on the plant, making the relationship parasitic. Predic-tions are developed to enable testing of hypotheses, and tests are discussed. Nutrition andmicroenvironment hypotheses are supported, while the enemy hypothesis remains withseveral uncertain issues to be resolved. The evolution of the galling habit has followed twopathways, one via mining plant tissues and the other from sedentary external herbivoresthat then modify plant growth. In each route the sequence of selective factors was probablydifferent, but improved protection from hygrothermal stress and improved nutrition are ofpr~mary importance, and protection from enemies probably reinforced the galling habit.

KEY WORDS evolution. selection

THOSE INTERESTED IN insect galls have usually heldand expr~ssed their personal opinions on the adap-tive significance of the galling habit. Many opin-ions have thus been developed but little emphasishas been placed on testing them, and practicallyno studies have been designed explicitly to test thevalidity of several hypotheses simultaneously. Wewish to stiIIlulate more testing of hypotheses aboutgalling. In order to do this we first review most ofthe major ideas on galls as adaptations. We thenevaluate the evidence for and against each hy-potJ:1esis; we develop predictions derived from theviable hypotheses; and we present our approachto testing these predictions.

ly small numbers of galls which secrete honeydeware actually less preyed upon by parasites and oth-er enemies than the very many galls that lack suchalleged means of protection" (p. 109).

Adaptive Value for Plant

The most concerted arguments for galls as pro-tection for the plant have been made by Mani(1964). "In so far as the reaction of the plant fa-vours the survival of the organ attacked by thegall-inducing organism, the primary advantage inthe formation of the gall is to the plant. By pro-ducing the gall, the plant has in a sense localizedthe parasite in space and time and has forced it toextreme specialization" (p. 3). Galls are envisagedas something like the plant equivalent of encap-sulation: ". ..the ultimate object ...is the neu-tralization of the toxic effects of the cecidozoa onthe p.lant cells. In so far as the reaction of the plantpromotes this defensive effect, the primary signif-icance of the gall is for the plant and not for thececidozoa " (p. 227).

Hypotheses on Galls as Adaptations

No Adaptive Value

Bequaert (1924) argued that galls in generalcould not be shown to have any adaptive value,either for the plant or for the insect. He stated,". ..several peculiarities of galls are considered asnon.selective structures, originally evolved with-out definite purpose as mechanical or chemicalreaction of the plant tissues to the stimuli of. thegall maker ...there are many structures that areneither useful to the animal nor imperil its exis-tence, and in this category I should place the hon-eydew and resin secretingglands sometimes foundat the surface. The nectaries of galls, in particular,may well be regarded as extravagant or hypertelicstructures, such as are often encountered in themore highly evolvedanimals and plants" (p. 118)."It has never been demonstrated that the relative-

Adap_tive Value for Plant and Herbivore

As a link between arguments for galls beingprincipally an adaptation favoring the plant, ver-sus an adaptation favoring the insect, some authorshave argued that galls are of benefit to plant andherbivore- "Now the continued life and vitality ofthe plant is beneficial to the larva, and the largerand more perfect the gall, the greater the amountof available food- Hence natural selection will havepreserved and accumulated the gall-forming ten-dencies, as not only beneficial to the larvae, but asa means whereby the larvae can feed with leastharm to the plant"(Cockerell1890, 74). Zweigelt

'f';

1 Dep. of Biological Sciences, Box 5640, Northem Arizona Univ.,

Flagstaff, AZ 86011.2 Museum of Northem Arizona, Route 4, Box 720, Flagstaff,

AZ 86001.

1.~

Vol. 16, no.116 FqRUM: ENVIRONMENTAL ENTOMOI,.OGY

(1942) makes a similar argument, and Wetterhan(1889, 131) said «Natural selection evidently mayact in favor of each symbiont separately "

Bronner (1983) argues A:he case for "perfect coor-dination between the gall-inducing organism andthe gall-bearing plant.. (p. 66).

decreasing temperatures, the centers of the largebuds of horse chestnut, Aesculus hippocastanum,differed little from ambient temperature and thelag time was much less than 10 min. As Levitt(1980, 116) says, ..The plant's temperature usuallyclosely follows that of its environment, even in thecase of bulky, insulated tree trunks, although ofcourse, lagging behind it during periods of rise orfalI. " In addition, the gall-forming insects them-

selves are as well-adapted in their overwinteringstrategies as free-Iiving insects, so the gall is notnecessary as an insulator (e.g., see Baust et al. 1979,Ring 1981, 1982, Baust & Lee 1982). Uhler (1951)concluded that the importance of the gall is in ..theresistance of the gall to sudden changes of tem-perature and the protection it affords the larvaagainst actual body contact with rain, ice, snow,and sunlight" (p. 39).

Another factor that needs consideration is thewater relationships of the her~ivore. Several stud-ies indicate the limiting effects of low water in aherbivore's diet (reviewed by Scriber & Slansky1981). Many herbivores live in mines, leaf rolls,webbing, and curled leaves, all of which increasethe boundary layer and moisture content of themicroenvironment, among other things, and re-duce water loss from the herbivore.

The combination of moisture and high temper-ature, with the potential for hygrothermal stress,may also be important. Dry, hot environments re-sult inspecial adaptations in insects as well as otherorganisms, and dry, hot environments are partic-ularly limiting to all forms of life (e.g., Chapman1969). Temperature and humidity cannot be use-fully separated as factors in the environment. Thus,if the gall's primary selective advantage is "shelterfrom the elements," hygrothermal stress is likelyto be one of the important factors.

Improved Protection Against Enemies. Manyauthors have noted the protective nature of gallsagainst natural enemies, particularly parasitoids,the importance of gall morphology, and the di-versity of gall types in any one community (Askew1961, 1980, Weis 1982a,b, Weis & Abrahamson1985, 1986, Weis et al. 1985). Cornell (1983) re-views much evidence suggesting that gall mor-phology is principally driven by parasitoids as se-lective agents.

Galls also protect against predators and diseases.Those with nectaries that attract ants reduce para-sitoid attack at least in one case (Washburn 1984)and probably predators also. As a concealed feed-ing place the gall provides plant protection fromenemies of the herbivore, as in the many othercases discussed bv Price et al. (1980).

Adaptive Value for Insect

Nutritional Improvements. Many authors havenoted the higher concentrations of potentially nu-tritive compounds in galls and have assumed thatthese enhance nutrition for the gall-maker (e.g.,Mani 1964, Palct & Hassler 1967, Malyshev 1968,Braun 1969, Shannon & Brewer 1980, Rohfritsch& Shorthouse 1982, Bronner 1983). The cytologyand histochemistry of the nutritive tissue lining thegall lumen appears to favor the herbi.vore (Roh-fritsch & Shorthouse 1982). Normal differentiationof plant tissues is completely disrupted by gall for-mation. There is little doubt that the radicalchanges in cell structure and content improve the-nutritional value of an attacked plant part for theherbivore, relative to what is present before attack.

Not only are nutrients increased but chemicaldefenses are decreased. Phenolic compounds de-cline in nutritive tissue during gall formation(Meyer 1957). In galls of Euura lasiolepis, tissuefed upon by larvae has total phenol levels about7 -fold lower than in ungalled stem tissue at anequivalen~ location on the shoot (G. L. Waring &P. W. Price, unpublished data). Galling Pemphi-gus aphids select and are mostsuccessful in loca-tions with low phenolic concentration (Zucker1982). ,The inquiline cynipid Synergus umbracu-lus actually initiates a gall in phenol-rich paren-chyma of the host gall, but nevertheless the nutri-tive tissue of the inquiline gall is phenol-free(Berland & Bernard 1951, Cornell1983). In a sur-vey of tannin concentrations in galled versus un-galled tissue in a .variety of species, Larew (1982)found that tannins were generally lower in galledtissue. Thus, the gall becomes a haven from plantdefenses and a rich food source.

Microenvironmental Improvements. Felt (1940)took it for granted that galls provided food andshelter to the gall-maker. He seems to emphasizeprotection from the harsh physical environmentwhen he mentions .the "most satisfactory shelterfrom the elements" (p. 3). The question remainsas to which specific physical factors are of impor-tance.

Cold temperatures are probably not of selectiveimportance in gall formation. Uhler (1951) foundonly slight protection against sudden changes intemperature in the Eurosta solidaginis gall. Forthe same galler Baust et al. (1979) concluded thatthe gall "provides a minimal thermal buffer to itsinhabitants" (p. 572). These findings are consistentwith the well-known similarity between temper-atures in plant tissues and ambient temperature(e.!!.- Levitt 1980). For examDle. even in raDidlv

Evaluatiou of Evidence on Hypotheses

The Nonadaptive Hypothesis

Several adaptive features of galls have been dis-cussed in this paper already: improved nutrition,reduced defensive chemicals. protection from harsh

Februarv 1987 PRICE ET AL.: GALLS AND ADAPTATION 1'1

Table I. Some phylogenetic relationships of gallformers and plant families in North America, based ondata in Felt (1940)

environments and enemies. Although Bequaert(1924) specifically stated that there was no adap-tive value to nectar secretion on galls, Washburn(1984) found that when ants were prevented from

reaching cynipid nectar-producing galls, parasit-ism almost doubled from 25% with ants present to48% in their absence. These findings lend no sup-port to the nonadaptive hypothesis, and it will notbe considered further in this paper.

Galling taxon

Plant familvGall Gall M.t

wasps midges Aphids El.es OtherCynip- Cecido- Aphidae

h r~ do- gallers

. d ..d p yl ae1 ae myll ae

Salicaceae

FagaceaeRosaceaeAceraceae

JuglandaceaeCompositaeTotal no. plant

families exploit-ed bv taxon

-

0"

92

5

O

O

2

-

7

6

8

2

2

2.~

-9ooo

41O

-

23

2

9

2

O

16

-

4

8

13

14

4

3

30II 71 17 41

a Numbers indicate the percentage of total species in a gal!ing

taxon that attack each plant family listed. AI! plant families at-tacked are not included so percentages do not add to 100. Notethat each taxon of gal!er attacks many plant families, and con-centration of gal!ing species is on a different plant family (init.li",,)

radiation of plant taxa with the potential for gallformation (see also Roskam 1986). The gallerclearly has the potentialadaptive advantage in theplant/herbivore relationship (see also Weis &Abrahamson 1986) and we know of no compel1ingarguments to the contrary.

The Mutual Benefit Hypothe8i8

There is no evidence in the literature that gallsimprove the fitness of the plant. Al1 the data shownegative impact of gal1ers on plant growth and/or fitness, as discussed, and positive reproductiverates in the gal1ers: they are parasites, except forthe gal1ing fig wasps (Agaonidae), which are mu-tualistic pol1inators of figs (e.g., Janzen 1979,Wiebes 1979). Thus a hypothesis arguing that gal1-ers and Dlants are mutualists is untenable.

The "Adaptive for Plant" or

Plant-protection Hypothesis

This hypothesis could be supported if advan-tages to the plant exceed those to the insect. Thisimbalance is not generally observed. Plants withgalls suffer severe decrements in growth and sex-ual reproduction, whereas insects reproduce effec-tively iri galls and multiply rapidly. Rhabdophagastrobíloides selects relatively large shoots for gallformation, and the gall, and the shoot it is on, actas a sink for photosynthate and nutrients (Weis &Kapelinski 1984). This is a commonly describedphenomenon' (Fourcroy & Braun 1967, Jankiewiczet al. 1970, Hartnett & Abrahamson 1979, Harris1980, Col1ins et al. 1983, Abrahamson & McCrea1985). Gal1s may even cause death of shoots andbranches and gal1ers may show preference for themost vigorous shoots (Craig et al. 1986). As Weis& Kapelinski (1984, 458) say, «The relationship isstrictly parasitic as the plant receives no benefit,and may even suffer a loss in reproductive output"(see also Stinner & Abrahamson 1979, Abraham-son & McCrea 1985). Gal1ers have been used forbicrtogical control of weeds, indjcating the strongnegative impact that some have on plant fitness(Hol1oway & Huffaker 1957, Harris 1977, ShQrt-house 1977a,b).

If plants had the selective advantage we couldnot explain such conditions as the common devel-opment of a nutritive layer in the gal1; reductionsin plant fitness; widely divergent gal1 morphologyon the same plant species and even when derivedfrom the same plant part; and copious nectar pro-duction fromgal1s, when its attractiveness to antsprotects the gal1er from enemies.

Perhaps most importantly, this hypothesis sug-gests that a plant lineage that acquires the abilityfor gal1 formation would retain the trait and fre-quency of gal1ing would be better correlated withplant phylogenies than with insect phylogenies,However, while there are clearly defined affinitiesin plant family and gal1 species number (e.g., seeFelt 1940, Mani 1964), the phylogenetic relation-ships are much stronger on the arthropod side,with almost whole families with the habit (e.g.,Agaonidae, Cynipidae, and in themites Eriophyi-dae), or large parts of families (e.g., Cecidomyi-idae). The close links in the families cut acrossmany phylogenetic lineages of plants (Table 1).This pattern shows that principal1y there has beena radiation of gal1ers across plant taxa, rather than

Adaptive Value for Insects

The Nutrition Hypothesis. Morphological, de-velopmental, and chemical data support this hy-pothesis. The cytology and development of gallshas been reviewed by Rohfritsch & Shorthouse(1982) and Rohfritsch (1986). In Cecidomyiidae,one of the first reactions in the developing planttissue initiated by the larva is reduced cell-wallthickness, supression of cuticle synthesis, and cell-content modification. Within 3 or 4 d, nutritivetissue develops in response to larval feeding, andvascular tissue supplies these tissues, after whichthe cells become cytoplasmically dense. AlI of thecytological and morphological changes in the tis-sues close to the galler are beneficial to the feedingand development of the insect.

The development of cecidomyiid galls is regu-lated by the feeding larva and not by the plant. .Gall shape depends upon the typical feeding po-

Vol. 16. no. 1FORUM: ENVIRONMENTAL ENTOMOLOGY18

scarce in both the wet and dry tropics in CostaRica (D. H. Janzen, personal communication).Searches for gal1ing insects in the wet tropical for-ests of Costa Rica and Panama reinforce this view(P. Price, personal observation). The contrast be-tween wet tropical forest and deserts in NorthAmerica in terms of insect gal1ers is extreme, withhigh population density and species richness in thedesert (see test of predictions). Thus, in a qualita-tive way, distributions of gal1ers in Central andNorth America support the microenvironment hy-pothesis, and the role of hygrothermal stress.

The Enemy Hypothe8i8

The remarkable complexity of the communitybased on galls has been carefully documented inseveral systems. In fact, gall formers suffer fromsome of the richest faunas of natural insect ene-mies known to attack insects. There are manyspecies of both parasitoids and inquilines. Biorhizapallida has 20 species of parasitoids and inquilinesassociated with it, and on average cynipids on oakinBritain support a community of 11.3 species andthose of herbaceous plants and shrubs have a com-munity of 5.9 species (Askew 1980). Cecidomyiidgalls on the shrub, Atriplex, in California, typi-cally were attacked by 7.4 species of parasitoidand inquiline (10 gallspecies, 24 primary parasit-oid species, and 7 inquiline species) (Hawkins &Goeden 1984).

Even more impressive is the high mortality in-flicted by enemies on gall-former populations.Washburn & Cornell (1981) argued that parasit-oids were probably a major factor in the local ex-tinction of Xanthoteras politum. Of the Euuragallers in California, Smith (1970) states, "Sinceparasitism approaches 100%, it is difficult to getsingle individuais, much less a series" (p. 43). Fromleaf galls of Neuroterus quercus-baccarum, As-kew (1961) reared 8 gallers and 109 parasitoidsand inquilines from 63 galls in 1958, and 24 gallersand 178 parasitoids and inquilines from 117 gallsin 1959. Total parasitism of cecidomyiid gallers onAtriplex commonly reaches well over 50% of thegalling population (Hawkins & Goeden 1984). Insmall galls, Eurytoma gigantea may kill100% ofits hosts, Eurosta solidaginis (Weis & Abrahamson

1985).If gallers benefit from enemy protection they

shouldat least show reduced mortality in compar-ison with relatives feeding in other ways. This isnot the case in the nematine sawflies, for the leaffolders in the genus Phyllocolpa have much lowerparasitism than the gall-formers (Smith 1970).Galling tephritids were presumably derived fromfruit-infesting species, so there is little reason toassume that a gall provides more protection thana fleshy fruit. However, among the aphids, free-feeding species (e.g., Aphis, Acyrthosiphon) havemany parasitoid species, but gall formers in thegenera Pemiphigus and Adelges have no recorded

sition of the larva, with round galls produced byeven feeding around the galllumen, conical gallsby feeding at the bas~ of the gall, and lenticulargalls by activity at the lateral margins (Rbhfritsch& Shorthouse 1982).

Galls act as sinks for plant chemicals, benefitingthe insect by enabling rapid gall growth and con-centration of nutrients. Nutrients are concentratedin galls (Palct & Hassler 1967, Braun 1969, Jan-kiewicz et al. 1970, Shannon & Brewer 1980,Abrahamson & McCrea 1985) as well as carbonand energy (Fourcroy & Braun 1967, Hartnett &Abrahamsón 1979, Stinner & Abrahamson 1979,McCrea et al 1985).

Perhaps the most convincing evidence on theimproved nutritional status of galls comes fromcomparative studies of feeding efficiency of gall-ing and frf)e-feeding aphids (Llewellyn 1982).When energy consumed ( C) is compared withproduction (P, body growth and reproduction)galling aphids had a p /C ratio 71% higher thanaphids feeding on ungalled leaves. Also, produc-tion/ assimilation ratios for gallers were found tobe the highest of any fluid-feeding insects yet re-corded.

Thus the insect influences to a remarkable de-gree the improvement of plant tissues relative tothe ungalled plant part from which the gal1 de-velops; this is a form of resource modification.

The Microenvironment Hypothe8i8. Extraor-dinarily few published data are available that canbe used to support or refute this hypothesis. Pro-tection against extremes in temperature is minimaland we know of no compel1ing published data sug-gesting that protection against hygrothermal stressis a strongly adaptive feature. Perhaps such pro-tection has been regarded as so intuitively obviousas not to warrant study, but clearly more data onthe subject are needed.

Kiister (1911) found more galling species in hil1yand mountainous country than in the plains, a pat-tem especial1y evident in the Eriophyidae. On Java,the distributionpattern of gal1ing species numberseemed to be correlated mostly with plant speciesnumber, reaching a peak in wel1-developed trop-ical rain forest (Docters van Leeuwen-Reijnvaan &Docters van Leeuwen 1926). But within this gen-eral pattern mite galls became more abundantwhere evaporation was high andair móisture fluc-tuated strongly, and cecidomyiids dominated inthe moister, more equable, vegetation. Althoughthese verbal evaluations do not help much in re-solving the most important factors in distribution,they do indicate that physical factors may deter-mine to some degree the distribution of gallers,although the patterns discussed here do not sup-port the idea thathygrothermal stress is the strong-est selective factor.

The pattern in Java does not seem to be ref1ect-ed in Central America: Only three references togal1s are made in Janzen (1983) on natural historyof Costa Rica, and Janzen regards gal1s as ve.ry

February 1987 PRICE ETAL.: GALLS AND ADAPTATION 19

~~

~~CUPPED SPANGLE GALLNt,'"olo,

-~

~

~ ~ ~

SILK-BUTTON SPANGLE GALL SMOOTH SPANGLE GALLN.numismalis N.slblpas

Fig. I. Some agamic generation spangle galls onoak in England. which alI occur on the underside ofleaves. Note that gall shape reduces differences in thedistance a parasitoid must pierce to reach the larvalcharÍ1ber (see bar indicating minimum distance) (mod-ifjed from Darlington 1975).

with enemies later evolving the ability to exploitthem. This speculation is even more difficult toevaluate than arguments addressing contemporaryinteractions, and will be revisited when the pread-aptation and evolution are discussed.

Tests of Predictions from Hypotheses

From each hypothesis unique predictions can bedeveloped. Some of them are lÍsted in Table 2.Therefore it should be possible to test each hy-pothesis and accept or reject it.

If the plant-protection hypothesis is valid thereshould be clear phylogenetic patterns in plant taxafor gall-forming ability, and these should be clear-er than for the insect taxa. We should expect anarrower spectrum of plant taxa adapted for gall-ing and a broad range of insect taxa that are en-capsulated in galls. This is not the case. A smallnumber of insect taxa form galls across a widerange of plant taxa. Gall wasps (Cynipidae) useeight plantIamilies in North America (Felt 1940).The gall midges (Cecidomyiidae) use 71 families.Aphids (Aphidae) use 17 families, and mites (Er-iophyidae) attack 41 families. Other galling in-sects, including sawflies, tephritid flies, a few bee-tles and moths, and psyllids, use in aggregate 30plant families (Felt 1940). This evidence clearlyillustrates that a small number of insect lineageshave acquired the ability tó induce galls, and inseveral of these, radiation across plant taxa hasbeen extensive. Of the 291 plant families attackedby gallers, 38 (13%) contain only one genusthat isattacked, indicating that links in gall productionbetween genera within the same families are pooror absent. Clearly, the insect lineages carry thegalling trait, not the plants, and the plant-protec-tion hypothesis must be rejected.

The plant-protection hypothesis must also be re-jected on nutritional grounds. No evidence hasshown that nutrition in a gall is inferior to that inthe ungalled plant part, and several studies citedabove have documented increased concentrations

parasitoid species in North America (Krombein etal. 1979).

Several authors have emphasized that diver-gence of gall mor"hology among coexisting specieshas most likely been driven by parasitoid attack(Askew 1961, Price 1980, Cornell 1983). There isgood evidence for such divergence (Askew 1961,Cornell 1983) among closely related species, andno alternative hypothesis invoking adaptationseems to be viable. Cook (1923) recognized thatinsect galls "preserit the highest development andgreatest complexity of any of the plant galls" (p.13). CQrrelated with this is the presence of para-sitoids as enemies of insect gallers and their ab-sence in fungal, nematode, and .mite galls.

As counter-evidence for gall-divergence selec-tion by parasitoids and inquilines, eriophyid mitegalls on the same plant species show equally dis-tinct galJ morphologies, although they are not at-tacked by enem~es piercing the gall with oviposi-tors. For example, on beech leaves, Eriophyesnervisequus causes long, narrow "fitzgalls" on theveins, E. stenopis induces roll galls on leaf edges,and E. macrQrhynchus forms pouch galls betweenthe veins (Darlington 1975). Eriophyid leaf gallerson limes and sycamores show equally well devel-oped differences (Darlington 1975). In addition,differences in gall morphology may have little ef-fect on the parasitoids attacking gallers. For ex-ample, differences in morphology of spangle gallsof oak cynipids in the genus Neuroterus in En-gland actually reduce differences in access to para-sitoids, rather than increase theíh (Fig. 1), and areunlikely to cause differences in the parasitoidspe,cies that attack them. Even large differences ingall size have little impact on the number of para-sitoid species attacking the ganer in some cases.Cynips longiventris has a gan diameter of about4 mm and 10 parasitoids and inquilines attackingit (Askew 1961, 1980), but the much larger gall ofCynips quercusfollii, with a diameter of about 10mm, has 11 parasitoids and inquilines, 7 of whichare common to both ganers.

Our impression is that gan formers evolved andradiated in spite of heavy parasitoid attack. Thisheavy attack may have selected for divergence ofgan form, phenology, and position, but more care-ful comparative evaluation is needed on compar-ison of enemies and their impact on ganers andrelated nonganers; comparison of diversity of ganmorphology at the time of enemy attack, and con-sequent mortality; comparison of diversity of ganmorphology between coexisting species of insectsattacked by parasitoids, and mites not attacked byparasitoids; and comparison of enemy species ingalls with different morphology to assess the extentto which gan characters result in different com-munities of natural enemies.

Of course, the radiation of gan formers and theiradaptive gan traits, such as reduced kairomonalcues or protective or cryptic gan morphology, mayhave occurred durinl! a Deriod of enemv escaDe.

F.ORUM: ENVIRONMENTAL ENTOMOLOGY Vol. 16. no.120

Table 2. Predictions derived from hypotheses about galling

Plant ad-aptation Insect adaptation

Prediction

-

No

Yes

No

No

No

Yes

Yes

No

-

No

Yes

No

No

Yes

No

No

Yes

-

Yes4

No

No

Yes

Yes

No

Yes

No

-

No

Yes

Yes

No

Yes

No

Yes

No

1. Phylogenetic relationships among plant hosts stronger than among insect gallers2. Phylogenetic relationships among gallers stronger than among plant hosts3. Food quality in gall nutritive layer higher than for free-feeding insects on the

same plant4. Food quality in gall nutritive layer lower than for free-feeding insects on sameplant .

5. Gall species abundance even across environments, with equal plant species diversi-ty at the same latitude and altitude

6. Gall species abundance correlates with specific environmental variables, especially

evapotranspiration7. Gall morphology uniform on one plant species on one plant part8. Impact by natural enemies on gallers reduced, relative to .free-feeding insects on

the same host plant in the same environment

a Italics indicãte the uniqueness of each prediction.

the Sonoran Desert to the top of the San FranciscoPeaks at 3,843 m in northern Arizona, the totalnumber of galling species on trees, shrubs, andherbs declines rapidly with increasing altitude(regression; r2 = 0.66; P < 0.01; n = 26 sites). As

the sites become drier the diversity of gallingspecies increases. The result is not a simple declinein species, as predicted from general trends in al-titudinal/latitudinal species richness, because thetrend is not. significant for riparian sites sampledfrom 915 m to 2,440 m (regression; r2 = 0.006;P > 0.05; n = 13 sites) but over the same range a

significant correlation exists on dry sites away fromwater courses (regression; r2 = 0.36; P < 0.05; n =

13 sites). The patterns are even more distinct whenonly shrubs are considered, which support a largemajority of galling species on this gradient. Inde-pendent data collected by G. Waring on cecido-myiid gallers on creosote bush support this pattern,and sampling in Brazil by G. Wilson Fernandesshows the same trend in the relatively dry cerradovegetation. The pattern appears to be widespread.

Anybody interested in galls cannot fail to beimpressed by their large numbers in dry southwestNorth America. The correlation between the gall-ing habit and dry sites strongly implicates hygro-therma:l stress as a major factor selecting for gallformation. We anticipate that new studies on thebiogeography of gallers relating to climatic regionswill add significantly to an understanding of theevolution of the galling habit, and to the phylo-genetic and geographic relationships in galling lin-eages, as discussed by Gagné (1984).

The prediction that gall morphology should bemore or less uniform when made by related speciesfrom the same plant part is clearly not valid. Gallmorphology is often a better indicator of speciesdifferences than gall inhabitants, because siblingspecies are so often involved. Because each species'gall-inducing ability will evolve independently, weshould expect divergence of shape unless strong

of nútrients in galls. The nutrition hypothesis iswell supported by the evidence. It is well knownthat some chemicals that could be deleterious toherbivores accumulate in galls, but these concen-trate externally to the nutritive tissue (Rohfritsch& Shorthouse 1982, CorneII1983).

Plants do show reactions to galling that reduceinsect fitness, but resistance to galling and shed-ding of galled parts are prevalent defenses. Thefirst line of defense appears to be resistance to gallformation, resulting in an aborted attempt by theinsect (e.g., Whitham 1980). Euura lasiolepis formsthe sniallest galls on the most resistant plants andplant l:»lrts in which larvae rarely survive (P. Price,personal observation). In both cases, females usethe best possible resources for gall formation(Whitham 1978, 1980, Craig et al. 1986). Earlyabscission of galled leaves in poplars causes exten-sive mortality of Pemphigus aphid broods in galls(Williams & Whitham 1986). Shoot abscission inwillow causes death of stem gallers in the genusEuura (P. Price, personal observation). But notenough is known of these reactions to understandhow general they are or how uniform the reactionsare in plant phylogenies. Only isolated examplesare available, althoughsystematic study would bevaluable. Early abscission of leaves and shoots maybe very general traits induced in plant parts witha relatively weak positive, or negative, carbon bal-ance. Based on the scanty knowledge of this sub-ject it would not be parsimonious to argue thatentrapment of insects in galls by plants is a com-mon phenomenon. It appears to be an adaptiveresponse to gallers, rather than the primary adap-tive force selecting for gall formation.

Only the microenvironment hypothesis predictsthat galling species' frequency of occurrence shouldvary in response to environmental variables suchas moisture and temperature. G. Wilson Fer-nandes has collected new data supporting this pre-diction. On an altitudi~al gradient from 305 m in

21PRICE ET AL.:GALLS AND ADAPTATIONFebruarv 1987

FREE-FEEDING ACTIVEHERBIVORE

/ \

1\ II\ 1.PhYSiCal faclors 2. Melabolic cosi

3.Hidden from enemies

SEDENTARY SURFACE-FEEDING

~\

'"'(< í I.,

1.Nutrition" I=nAm;A~

Yf it\ \ ,.P.mphigu. P.mphigu. P.mphigu.

bur..rlu. .plroth.c.. Iy.im.chl..

Fig. 2. Three PemphigU8 aphid galls formed onLombardy poplar (Populus italica) in Britain, showingstructural differences, even though they are not attackedby parasitoids and most predators enter through the es-

cape aperture (after Darlington 1975).

1.Physical fa7tors 2.Enemies

MINING PLANTTISSUE

" i

GALL FORMATION

gy in some cases, a point that needs some study ifthe enemy hypothesis is to be tested. But the casesof generalist parasitoids cited above suggest thatthis will be an tinimportant difference. We alsopredict that galls should be more different whentogether than when they occur apart, as a resultof character displacement in relation to enemy-free space. Alternatively, gallers with similar mor-phology should not occur together. These predic-tions have never been tested. But our impressionis that galls on different plant species, and in iso-lated geographic distributions, are as different asgalls on the same plant species, for example, Pem-phigus galls on Populus species. Thus, we do notfind the enemy hypothesis a compelling one. AI-though it remains as a viable possibility, there areinconsistencies in the empirical evidence. Moresystematic comparative studies are needed andmany avenues of research that have been suggest-ed in this discussion would be rewarding.

In general, we find support for the nutrition hy-pothesis and the microenvironment hypothesis, al-though this support is largely correlational in na-ture. Ambiguous evidence is available for theenemy hypothesis, and the plant-protection hy-pothesis has clear evidence to the contrary. Withthese arguments in mind we can undertake a hy-pothetical journey through the steps in the evolu-tion of the galling habit.

Preadaptations and the Evolutionof Galling

The routes followed by insects to the gallinghabit from the free-feeding herbivore are via plantmining and sedentary surface feeding (Fig. 3).

selection pressures work to preserve a particulargall shape. We know of no selective factor thatshould woTk in this way. So perhaps the predictionis unrealistic because we should really expect thateither genetic drift will cause fixation of some gallcharacters and loss of others, leading to diver-gence, or natural enemies will cause disruptive se-lection on gall characters during a speciation event,such that gall morphology will diverge more rap-idly than insect morphology.

These alternatives are exceedingly difficult totest. We know little of the genetics of particularinsect gall characters and how the genes are ex-pressed, or which plant genes are involved, al-though real progress has been made by Weis &Abrahamson (1986). For every case of gall-shapedivergence under the influence of parasitoid at-tack, especially the Cynips example discussed byAskew (1961), Price (1980), and Cornell (1983),and the directional selection by parasitoids on Eu-rosta gall size (Weis & Abrahamson 1985, Weis etal. 1985), there are other cases where gall shapeseems to make little difference to protection againstenemies. The primary parasitoid Rileya tegularisattacks six species of gallers on Atriplex in spite ofthe distinct differences in the outward appearanceof the galls. In the same system Tenuipetiolusmedicaginus attacks seven species. So parasitoidscan remain quite generalized in spite of differentgall morphologies. This is true for parasitoids onAsphondylia galls on creosote bush also (G. War-ing, personal observation). Many guilds of gallersproduce some galls too similar to prevent attackby a species of parasitoid common to both. Hick-ory leaf galls produced by Caryomyia and the var-ious spangle galls on oak leaves initiated by Neu-roterus, are examples. Conversely, why arePemphigus galls so different when no parasitoidsare known to attack the species (Fig. 2)?

It is possible that galls' chemical cues used bysearching parasitoids differ as much as morpholo-

FORUM: ENVIRONMENTAL ENTOMOLOGY Vol. 16. no.122

Acknowledgment

We thank Warren Abrahamson, Howard Comell,NanGy Moran, and Arthur Weis for their insights on thissubject and their reviews of the manuscript. Supportedby N.S.F. grant BSR-8314594 and a Brazilian NationalScholarship to G. Wilson Femandes (Conselho Nacionalde Desenvolvimento Científico e Tecnológico 200.747 /

84Z0).

Plant miners include tephritid flies and lepidop-teran stem borers, both with some gall-formingspecies. Sedentary surface feeders include thehemimetabolous gallers (aphids, psyllids, andthrips), mites, the holometabolous taxa of sawflies,and cecidomyiids.

In either case, the initiation of swelling in planttissues should have an immediate positive effecton insect fitness for the trait to continue and fordirectional selection to w.ork in favor of larger plantgrowths. In the plant miners, protection fromphysical factors is unlikely to be of much selectivevalue because these herbivores are already con-cealed. However, swelling of tissue would proba-bly induce the production of parenchyma cells ofhigher nutritional value than differentiated cells,and dead cells in the pith of stems. AIso, protectionagainst enemies will increase. In this case the pri-mary selective advantages will be nutrition andprotection against enemies. In the sedentary sur-face feeders, persistent feeding in one location islikely to result in differential growth of tissues,

'Uepressions and rolling in leaves, and an immedi-ate effect on the boundary layer of the leaf. Theherbivore becomes better protected from physicalfactors, especially moisture stress. The evolution ofmore potent stimulating mechanisms would causedeepening of depressions and ultimate gall for-mation. Such a scenario may be easily envisionedfor aphids and cecidomyiid latvae. Here the majorchange of adaptive value seems to be primarily inrelation to physical factors and secondarily to nu-trition and enemies. These herbivores were al-ready well adapted to the nutrition of the plant,and initialleaf distortion would not provide goodprotection against enemies. Nutrition may wellhave improved as more cells retained the paren-chymatous type, and as meristematic activity andthe associated nutrient sink were prolonged. Butsuch improvement would have been gradual, andmay well have been secondary to the change inmicroclimate. As the gall closed, protection wouldhave become more effective. Thus the two routeshave different selective regimes associated withthem, and the nutrition hypothesis, microenviron-ment hypothesis, and enemy hypothesis are alllikely to be valid at some stage in the evolution ofthe galling habit.

We hope that this discussion will revive the olddebate on the adaptive significance of galls (e.g.,Hollis 1889, 1890, McLachlan 1889, Mivart .l889,Romanes 1889, 1890, Wetterhan 1889). After al-most 100 yr we have much more information ongallers and their communities of enemies, and wethink we can confidently reject some of the hy-potheses discussed earlier in this century and inthe last. However, adequate evaluations of somehypotheses, especially the microenvironment andenemy hypotheses, are yet to be completed. If westimulate any further discussion and research onthis topic we will have achieved our purpose.

References Cited

Abrahamson, W. G. & K. D. McCrea. 1986. Nutrientand biomass allocation in Solidago altissima stemgall insect-parasitoid guild food chain. Oecologia(Berlin) 68: 174-180.

Abrahamson, W. G. & A. E. Weis. 1986. The nutri-tional ecology of arthropod gall-makers. In F. Slan-sky & J. G. Rodriquez [eds.], The nutritional ecologyof insects, mites and spiders. Wiley, New York.

Askew,R. R. 1961. On the biology of the inhabitantsof oak galls of Cynipidae (Hymenoptera) in Britain.Trans. Brit. Soc. Entomol. 14: 237-268.

1980. The diversity of insect communities in leaf-mines and plant galls. J. Anim. Ecol. 49: 817-829.

Baust, J. G. & R. E. Lee. 1982. Environmental trig-gers to cryoprotectant modulation in separate pop-ulations of the gall fly, Eurosta solidaginis (Fitch).J. Insect Physiol. 28: 431-436.

Baust, J.G., R. Grandee, G. Condon & R. E. Morrissey.1979. The diversity of overwintering strategies uti-lized by separate populations of gall insects. Physiol.Zool. 52: 572-580.

Bequaert, J. i 924. Galls that secrete honeydew. Acontribution to the problem as to whether galls arealtruistic adaptations. Buli. Brooklyn Entomol. Soc.19: 101-124.

Berland, L. & F. Bernard. 1951. Ordre des Hyme-nopteres, pp. 771-1276. In P. Grassé [ed.], Traité dezoologie, No.10. Masson, Paris.

Braun, A. C. 1969. Abnormal growth in plants, pp.379-420. In F. C. Steward [ed.], Plant physiology: atreatise, vol. VB. Analysis of growth: the responsesof cells and tissues in culture. Academic, New York.

Bronner, R. 1983. Adaptation insect-plant in cynipidgalls, pp. 61-68. In N. S. Margaris, M. Arianoutsou-Faraggitaki & R. J. Reiter [eds.], Plant, animal andmicrobial adaptations to terrestrial environments.Plenum, New York.

Chapman, R. F. 1969. The insects: structure andfunction. American Elsevier, New York.

Cockerell, T. D. A. 1890. The evolution of insect-galls. Entomologist (London) 23: 73-76.

Collins, M., M. J. Crawley & G. C. McGavin. 1983.Survivorship of the sexual and agamic generations ofAndricus ~ercuscalicis on Quercus cerris and Q.robur. Ecol. Entomol. 8: 133-138.

Cook, M. T. 1923. The origin and structure of plantgalls. Science 57: 6-14.

Cornell, H. V. 1983. The secondary chemistry andcomplex morphology of galls formed by the Cynipi-nae (Hymenoptera): why and how? Am. Midl. Nat.110: 225-234.

Craig, T. P.., P. W. Price & J. K. Itami. 1986. Re-source regulation by a stem-galling sawfly on the ar-royo willow. Ecology 67: 419-425.

Darlington, A. 1975. The pocket encyclopaedia of plantgalls in color. Blandford, Poole, England.

Docters van Leeuwen-Reiinvaan, J. & W. M. Docters

February 1987 PRICE ET ALo: GALLS AND ADAPTATION

van Leeuwen. 1926. The zoocecidia of the Neth-erlands East Indies. Drukkerij de Unie, Java, Batavia,Java.

Felt, E. P. 194{). Plant galls and gall makers. Com-stock, Ithaca, N. Y.

Foureroy, M. & C. Braun. 1967. Observations sur Iagalle de rAulax glechomae L. sur Glechoma hedera-cea L. II. Histologie et rôle physiologique de Ia coquesclerifiée. Marcellia 34: 3-30.

Gagné, R. J. 1984. The geography of gall insects, pp.305-322. ln T. N. Ananthakrishnan [ed.], Biology ofgall insects. Oxford & IBH, New Delhi.

Harris, P. 1977. Biological control of knapweed withUrophora affinis and U. quadrifasciata: a summary,pp. 265-27Ó. ln Proceedings, Knapweed Sympo-sium, Vancouver, B.C.

1980. Effects of Urophora affinis Frt1d. and U. quad-rifasciata (Meig.) (Diptera: Tephritidae) on Centau-rea diffusa Lam. and C. maculosa Lam. (Composi-tae).. Z. Angew. Entomol. 90: 190-201.

Hartneu, D. C. & W. G. Abrahamson. 1979. Theeffects of stem gall insects on life history patterns inSolidago canadensis. Ecology 60: 910-917.

Hawkins, B. A. & R.. D. Goeden. 1984. Organizationof a parasitoid community associated with a complexof galls on Atriplex spp, in southern California. Ecol.Entomol. 9: 271-292.

Hollis, W. A. 1889. Galls. Nature (London) 41: 131.1890. Galls. Nature (London) 41: 272.

Holloway, J. K. & C. B. Hu:ffaker. 1953. Establish-ment ofa root borer and a gall t1y for control ofklamath weed. J. Econ. Entomol. 46: 65-67.

Jankiewiez, L. S., H. Flieh & M. Antoszewski. 1970.Preliminary studies on the translocation of "C-la-belledassimilates and 32PO, towards the gall evokedby Cynips (Diplolepis) quercus-folii L. on oak leaves.Marcellia 36: 163-172.

J~zen, D. H. 1979. How to be a fig. Annu. Rev. Ecol.Syst. 10: 13-51.

1983. Costa Rican natural history. Univ. of Chicago,Chicago.

Krombein, K. V., P. D. Hurd & D. R. Smith [eds.].1979. Catalog of Hymenoptera in America northof Mexico, vol. 3. Indexes. Smithsonian InstitutionPress, Washington, D. C.

Kiister, E. 1911. Die Gallen der Pt1anzen. Ein Lehr-buch der Botaniker und Entomologen. Hirzel, Leip-zig.

Larew, H. 1982. A comparative anatomical study ofgalls caused by the major cecidogenetic groups, withspecial emphasis on the nutritive tissue. Ph.D. dis-sertation, Oregon State Univ., Corvallis.

Leviu, J. 1980. Responses of plants to environmentalstresses, vol. 1. Chilling, freezing, and high-temper-ature stresses. Academic, New York.

Llewellyn, M. 1982. The energy economy of t1uid-feeding herbivorous insects, pp. 243-251. ln J. H.Visser & A. K. Minks [eds.], Insect-plant relation-ships. Center for Agricultural Publishing and Docu-mentation, Wageningen.

Malyshev, S. .I. 1968. Genesis of the Hymenopteraand the phases of their evolution. Methuen, London.

Mani, M. S. 1964. Ecology of plant galls. W. Junk,The Hague.

MeCrea, K. D., W. G. Abrahamson & A. E. Weis.1985. Goldenrod ball gall effects on Solidago altis-sima: l'C translocation and growth. Ecology 66: 1902-1907.

McLachlan, R. 1889. Galls. Nature (London) 41: 131.Meyer, J. 1957. Cécidogenése comparée de quelques

galles d. Arthropodes et évolution cytologique des tis-sus nourriciers. Théses, Univ. of Strasbourg, Stras-bourg.

Mivart, S. G. 1889. Galls. Nature (London) 41: 174-175.

Palct, J. & J. Hassler. 1967. Concentration of nitro-gen in some plant galls. Phyton 12: 173-176.

Price, P. W. 1980. Evolutionary biology of parasites.Princeton Univ., Princeton.

Price, P. W ., C. E. Bouton, P. Gross, B. A. McPheron,J. N. Thompson & A. E. Weis. 1980. Interactionsamong three trophic levels: influence of plants oninteractions between insect herbivores and naturalenemies. Annu. Rev. Ecol. Syst. 11: 41-65.

Ring, R. A. 1981. The physiology and biochemistryof cold tolerance in arctic insects. J. Therm. Biol. 6:219-229.

1982. Freezing-tolerant insects with low supercoolingpoints. Comp. Biochem. Physiol. A 73: 605-612.

Rohfritsch, O. 1986. Pattems in gall development.In J. D. Shorthouse & o. Rohfritsch [eds.]. Biologyof insect- and acarina-induced galls. Praeger. NewYork.

Rohfritsch, 0. & J. D. Shorthouse. 1982. Insect galls,pp. 131-152. In G. Kahl & J. S. Schell [eds.]. Molec-ular biology of plant tumors. Academic, New York.

Romanes, G. J. 1889. Galls. Nature (London) 41: 80.174.

1890. Galls. Nature (London) 41: 369.Roskam, J. C. 1986. Evolution of the gall-inducing

guild. In J. D. Shorthouse & 0. Rohfritsch [eds.],Biology of insect- and acarina-induced galls. Praeger,New York.

Scriber, J. M. & F. Slansky. 1981. The nutritionalecology of immature insects. Annu. Rev. Entomol.26: 183-211.

Shannon, R. E. & J. W. Brewer. 1980. Starch andsugar levels in three coniferous insect galls. z. An-gew. Entomol. 89: 526-533.

Shorthouse, J. D. 1977a. Developmental morphologyof Urophora affinis galls, pp. 188-195. In Proceed-ings, Knapweed Symposium, Vancouver, B.C.

1977b. Developmental morphology of Urophoraquadrifasciata galls. pp. 196-201. In Proceedings,Knapweed Symposium, Vancouver, B.C.

Smith, E. L. 1970. Biosystematics and morphology ofSymphyta. II. Biology of gall-making nematine saw-flies in the Califomia region. Ann. Entomol. Soc. Am.63: 36-51.

Stinner, B. R. & W. G. Abrahamson. 1979. Ener-getics of the Solidago canadensis-stem gall insect-parasitoid guild interaction. Ecology 60: 918-926.

Uhler, L. D. 1951. Biology and ecology of the gold-enrod gall fly, Eurosta solidaginis (Fitch). CornellUniv. Agric. Exp. Stn. Mem. 300.

Washburn, J. 0. 1984. Mutualism between a cynipidgall wasp and ants. Ecology 65: 654-656.

Washburn, J. 0. & H. V. Corne". 1981. Parasitoids,patches, and phenology: their possible role in thelocal extinction of a cynipid gall wasp population.Ecology 62: 1597-1607.

Weis, A. E. 1982a. Use of a symbiotic fungus by thegall maker Asteromyia carbonifera to inhibit attackby the parasitoid Torymus capite. Ecology 63: 1602-1605.

1982b. Resource utilization pattems in a community

FoRuM: ENVIRONMENTAL ENTOMOLOGY Vol. 16. no. I24

1980. The theory of habitat selection: examined andextended using Pemphigus aphids. Am. Nat. 115:449-466.

Wiebes, J. T. 1979. Coevolution of figs and theirinsect pollinators. Annu. Rev. Ecol. Syst. 10: 1-12.

Wi"iams, A. G. & T. G. Whitham. 1986. Prematureleaf abscission: an induced plant defense against gallaphids. Ecology 67: 1619-1627.

Zucker, W. V. 1982. How aphids choose leaves: theroles of phenolics in host selection by a galling aphid.Ecology 63: 972-981.

Zweigelt, F. 1942. Beitriige zur Kenntnis der Blatt-lausgallen. Biologia Generalis 16: 554-572.

Received for publication 7 February 1986; accepted6 October 1986.

of gall-attacking parasitoids. Environ. Entornol. 11:809-815.

Weis, A. E. & W. G. Abrahamson. 1985. Potentialselective pressures by parasitoids on a plant-herbi-vore interaction. Ecology 66: 1261-1269.

1986. Evolution of host plant rnanipulation by gall-Iilakers: ecological and genetic factors in the Soli-dago-Eurosta systern. Am. Nat. 127: 681-695.

Weis, A. E. & A. Kapelinski. 1984. Manipulation ofhost plant developrnent by the gall-rnidge Rhab-dophaga strobiloides. Ecol. Entornol. 9: 457 -465.

Weis, A. E., W. G. Abrahamson & K. D. MeCrea.1985. Host gall size and oviposition success by theparasitoid Eurytoma gigantea. Ecol. Entornol. 10:341-348.

Wetterhan,D. 1889. Galls. Nature (Lóndon) 41: 131.Whitham, T. G. 1978. Habitat selection by Pemphi-

gus aphids in response to resource lirnitation andcompetition. Ecology 59: 1164-1176.