Galler-induced reduction of shoot growth and fruit production in the shrub Colliguaja integerrima ( Euphorbiaceae )

We evaluated experimentally the effect of a gall-maker insect on vegetative and reproductive traits of the shrub Colliguaja integerrima (Euphorbiaceae). We performed two experiments: (1) a chemical (insecticide) exclusion to prevent gall formation, and (2) a mechanical removal of new galls at the early stage to prevent gall growth. In the study area, galled shrubs were common (77 %). Because the pattern of insect attack may influence plant fitness, the distribution of egg clusters deposited by the galler and the number of galls among shrubs were also evaluated. Fruit production was inversely associated with the number of galls, but did not correlate with shrub height, shrub cover, and number of shoots. Ungalled shoots were longer than galled shoots after 12 months initiated the experiment. Chemical exclusion produced a delayed positive effect on plant reproduction. Fruit production was higher in experimental than control branches after 24 months. Mechanical removal of galls increased fruit production in comparison to control branches in the next reproductive season. These results indicate that the galler reduces shoot growth, and has a delayed detrimental impact on fruit production of C. integerrima. Because the chance of finding new galls was higher on the previously infected shrubs, it is possible that reinfection processes account for the cumulative negative effects of the bud-galling insect on plant fitness.


INTRODUCTION
Most studies on plant -herbivore relationships have focused on the fitness-impact of herbivores on annual plant species, where reduction in plant fecundity translates into a direct fitness loss across the individual lifetime (Crawley 1989, Marquis 1992a).In perennial plants, however, the impact of herbivory on plant growth and reproduction may be expressed in the same year in which damage actually occurs (Whitham & Mopper 1985, Weis & Sacchi 1996), in the following years (Primack & Hall 1990, Whigham 1990), or in both (Myers 1981), depending on the amount of resources acquired by the plant during the current and previous years (Obeso 1993).Because of the inherent difficulty to separate short-from long-term effects of herbivory on plant reproduction, most studies performed on perennial plants circumscribe to a short segment of the plant's life history (e.g., Marquis 1984, 1992b, Whitham & Mopper 1985, Paige & Whitham 1987, Crawley 1989, Weis & Sacchi 1996).This practice is unfortunate because plants may compensate or even overcompensate the immediate detrimental effect of herbivory when observed over a long period across reproductive seasons (e.g., Solomon 1983, Sacchi et al. 1988, Hendrix 1988, Messina et al. 2002).
In this report, we document the results of a three-year field study designed to evaluate the effect of a bud-galling insect on vegetative and reproductive traits of the native shrub Colliguaja integerrima Gill.et Hook.(Euphorbiaceae).Colliguaja integerrima is a native monoecious evergreen shrub that inhabits the Chilean Andean Range (Hoffmann 1998).Previous reports indicate that Colliguaja species suffer a low defoliation by chewing insects in comparison to other shrub species in central Chile (Montenegro et al. 1980), albeit they are commonly attacked by diverse native phloemfeeders and gall-makers insects (Martínez et al. 1992, Núñez & Sáiz 1994, Fuentes-Contreras et al. 1999, Gonzáles & Fernández 2000).Even though galling insects infect intensively C. integerrima (Martínez et al. 1992, Núñez & Sáiz 1994, Fuentes-Contreras et al. 1999), the extent to which insect damage translates into a lower reproductive success remains largely unknown.We investigated the impact of a bud-galling insect on vegetative growth (shoot growth) and reproduction (fruit set) of C. integerrima by using chemical and mechanical gall-exclusion experiments.

Study site and natural history
The fieldwork was carried out at Cerro El Roble, located in the Region Metropolitana, Santiago, Chile (32°58' S, 71°00' W, 1.600-1700 m elevation) from summer 2000 to summer 2002.Colliguaja integerrima reaches 1-2 m height and its flowering season occurs from September to November (Navas 1976).The vegetative growth occurs from September to February.Male flowers are terminal spikes and one or two female flowers are at the base of the male-spikes.The fruiting season occurs during March-April, and fruits are 3-seeded dry capsules (Navas 1976, Ravetta et al. 1991, Hoffmann 1998).In the study area, we did not record floral galls but spherical green galls within vegetative buds were common.Other herbivore insects (chewing and sucking) were relatively scarce in the study site.Even though the specific taxonomic description of this budgalling insect has to be done, it corresponds to a gall midge because the last instar larva has a sternal spatula (a dermal structure on the venter of the prothorax), which is exclusive of Cecidomyiidae (Gagné 1994).Its overall life cycle is similar to that described by Sáiz et al. (1999) in C. odorifera ("cecidia de copa").The bud-galling insect is a univoltine species whose hatching occurs approximately one week after oviposition.After that, the small neonate larvae move deeply into crevices of the vegetative bud to initiate gall formation.The gall grows from September to February.Each gall contains 1 to 30 separated chambers, with one larva per chamber.Pupation occurs within the galls at the end of the autumn.During the next spring, an adult emerges through one circular hole at the apical zone of each gall chamber.

Impact of galls on plant growth and reproduction
We estimated gall prevalence from the percentage of infected shrubs using eight 50-meter linear and parallel transects in January 2000 (n = 102 shrubs).In the same area, we chose randomly 55 shrubs (at least 5 m apart), and recorded shrub height (m) and cover (length x width, m 2 ), and the number of galls and vegetative buds in four branches per shrub.In addition, we recorded fruit production in the same branches.We regressed all the descriptive variables on fruit production in a multiple regression analysis.To determine how gall formation affected shoot growth, in November 2000 we tagged galled and ungalled apical vegetative buds on randomly chosen shrubs.A t-test for independent samples was performed to compare the shoot growth after 12 months (n = 50 shoots) (Sokal & Rohlf 1995).
We carried out two experiments to evaluate the effect of bud galling insects on plant reproduction.First, a non-systemic pesticide was applied to prevent new gall formation.We sprayed a solution of Malathion (171 µg/L -1 of active ingredient) on one branch of C. integerrima shrubs on a weekly basis, leaving a second branch of the same plant as control (n = 30 plants).Experimental and control branches were isolated and covered with plastic bags during pesticide application.The chemical was applied throughout the galling insect emergence season (which occurred from October 15 to November 30, 2000).We recorded the number of fruits and new gall formation in experimental and control branches during two consecutive reproductive seasons of C. integerrima (summer 2001 and summer 2002).A t-test for dependent samples was performed to compare the paired branches during the two years (Sokal & Rohlf 1995).Second, in order to assess the consequences of mechanical gall abortion on the future reproductive success of the host plant, two branches per shrub were randomly chosen in December 2000 (n = 37 galled shrubs).On one branch, all the recently developed galls were mechanically damaged with entomological pins and the second branch was left intact.A t-test for dependent samples was performed to compare the number of fruits/branch on each shrub during the summer 2002.
We assessed whether previous gall attack influenced subsequent oviposition and gall pattern distribution among host shrubs.First, we looked for egg clusters deposited by the galler and recorded the number of vegetative buds and galls from the previous year on randomly chosen shrubs (n = 48 plants) during the maximal activity date of adult emergence (last week of October 2000).We run a logistic regression to estimate the chance of finding at least one egg cluster, using the number of vegetative buds and galls from the previous year as predictor variables.Second, to determine whether the amount of previous galls influenced the occurrence of new galls, we randomly chose 49 shrubs in January 2001, and evaluated the relationship between the number of galls formed in the previous year and the presence of new galls.

RESULTS
The prevalence of galled shrubs was 77 %.A multiple regression analysis showed that while fruit production decreased with the number of galls, the height and cover of shrubs as well as the number of shoots had no influence on shrub fecundity (Table 1).Regarding vegetative growth, shoots were longer in ungalled than Multiple regression analysis to estimate the mean fruit production per branch of Colliguaja integerrima using as predictor variables plant traits and the number of galls from vegetative buds (summer 2000).R 2 model = 18.9 %, F 5,50 = 2.93, P < 0.029 Análisis de regresión múltiple para estimar la producción promedio de frutos por rama de Colliguaja integerrima usando como variables independientes rasgos de la planta y el número de agallas de yemas vegetativas (verano del 2000).R 2 modelo = 18.9 %, F Experimental branches had a lower number of galls than control branches in the summer 2001 (experimental = 0.98 ± 0.23, control = 9.65 ± 1.03, t 29 = 8.93, P < 0.001).In spite of that, fruit production did not vary significantly between treatments in 2001 (t 29 = 0.544, P = 0.589, Fig. 1).In the summer 2002, the number of galls was also lower in experimental than control branches (experimental = 0.33 ± 0.21, control = 19.07 ± 3.07, t 29 = 6.19,P < 0.001).Unlike the situation observed in 2001, however, in the summer 2002 experimental branches had a higher fruit production than control branches (t 29 = 3.158, P < 0.01, Fig. 1).The mechanical damage applied to new galls resulted in an increased fruit production in experimental branches in the next reproductive season as compared to control branches (experimental = 5.95 ± 0.98, control = 3.49 ± 0.90, t 36 = 2.20, P < 0.034).
Gall infection correlated to past gall infection on shrubs (r = 0.74, t 47 = 7.59, P < 0.001), and the probability of finding an egg cluster increased significantly with the number of galls present in the previous year (logistic regression analysis, Wald = 15.5, P < 0.001).The number of vegetative buds had no effect on the chance of finding egg clusters (logistic regression analysis, Wald = 0.23, P = 0.63).& Connor 1999).Our results provide additional evidence that insect attack has dramatic consequences for fruit production, and stress that detrimental effects on plant fecundity may occur over long periods across reproductive seasons in perennial plants.Even though the chemical treatment reduced gall infection in the summer 2001, fruit production did not differ significantly between experimental and control branches.However, this result does not rule out that short-term effects may occur on other fitness related traits, such as seed size or seed viability, which were not evaluated in this study.Interestingly, experimental branches produced a higher number of fruits than control branches in the summer 2002.Mechanical abortion of new galls at an early stage increased fruit production in the next reproductive season.Both plant phenology and the pattern of plant resource allocation may explain the delayed negative effect of the gall infection on plant reproduction.Flowering and fruiting of C. integerrima occurred at the beginning of the growing season, previous to vegetative growth.Arroyo et al. (1981) showed that high mountain plants present different flowering strategies.In the high Andean Cordillera of central Chile (2300-3500 masl), the flowering and fruiting season of most shrubs are often preceded by a long period of shoot and leaf growth.However, it is well known that the flowering of some shrub species occurs before leaf emergence in early spring (Arroyo et al. 1981, Rathcke & Lacey 1985) such as in C. integerrima (Navas 1976).Colliguaja integerrima leaves pre-formed floral buds for the next reproductive season at the end of summer.In alpine areas, most early and mid season flowering plants present flowerpreformation, indicating that flower initiation occurs in the previous season (Körner 1999).Thus, the delayed impact of gall attack on plant reproduction could be explained if the current plant reproductive outcome is a consequence of nutrient production, storage and allocation to reproductive structure during the previous growing season (Obeso 1993, Ehrlen & Groenendael 2001).
In this study, gall infection reduced the length of vegetative shoots.The leaves of galled shoots were small and crowded due to the highly reduced internodes that resulted from gall infection as compared to unaffected vegetative shoots.In addition, galled shoots inhibited the production of new shoots in the following years.This effect on plant architecture may affect plant reproduction.Since reproductive buds are located in the apical portion of shoots (Navas 1976, Hoffmann 1998), it is quite possible that flower production decrease in the following years.A similar result has been reported for C. odorífera, where gall infection on vegetative buds inhibits shoot development and future production of new shoots and fruits (Martínez et al. 1992).However, unlike C. integerrima (which is galled mainly on vegetative buds), gall infection in C. odorífera frequently occurs in both male flowers and vegetative buds (Martínez et al. 1992, Fuentes-Contreras et al. 1999, Sáiz et al 1999) and dissect the impact of each gall type is more complex in the absence of an experimental approach.Other studies have documented a decreased ramet production, and biomass allocation to leaves, rhizomes, and inflorescences in attacked ramets (e.g., Hartnett & Abrahamson 1979, Abrahamson & Weis 1997, but see Craig et al. 1990).It is possible that gall makers modify plant tissues that would otherwise serve to other function, therefore competing for assimilates with plant tissues (Larson & Whitham 1991, Stinner & Abrahamson 1979, 1997, Fay et al. 1996).
The negative association between the number of galls and fruit production in the summer 2000 was not in agreement with the absence of short-term effect of gall attack on fruit production observed in the summer 2001.This situation may be explained, at least partially, by the reinfection pattern of the galling insect.Even though the mechanistic explanation for reinfection is beyond the scope of this paper, it is possible that gallers require maximizing the availability of plant reactive tissue to get a successful gall establishment (see also Weis et al. 1988, Gagné 1994), and consecutive gall attacks on the previously galled hosts may contribute to cumulative negative effects on plant reproduction.In conclusion, the bud galler insect caused a direct reduction on shoot length and a delayed negative effect on the fruit production of C. integerrima.Cumulative negative effects probably result from the reinfection pattern of C. integerrima.These results represent the first step to a fully understanding of the relevance of an herbivorous insect on the ecology of C. integerrima.Future work must take into account the role of defensive plant traits against insect attack and the factors that influence the observed reinfection pattern.