Mortality of the outbreak defoliator Ormiscodes amphimone ( Lepidoptera : Saturniidae ) caused by natural enemies in northwestern Patagonia , Argentina Mortalidad del defoliador epidémico Ormiscodes amphimone ( Lepidoptera : Saturniidae ) causada por enemigos naturales en el noroeste de la Patagonia

Outbreaks of the defoliator moth Ormiscodes amphimone are occurring more frequently in numerous Nothofagus forests of Patagonia. However, little is known about the life history of this species including its natural enemies as mortality agents. In this work we quantifi ed mortality by parasitoids and generalist predators at the egg, larval and pupal life stages of O. amphimone in a Nothofagus pumilio (Poepp. & Endl.) Krasser forest in northwestern Patagonia. Parasitism of eggs was relatively low (ca. 11 %), and we did not record signifi cant larval predation by insectivorous birds. However, we recorded elevated mortality caused by larval parasitoids (ca. 50 % in third instar larvae) and pupal predators (ca. 75 %), which suggests that these natural enemies could play a signifi cant role in regulating O. amphimone populations. Our research is an initial step towards understanding the infl uence of natural enemies on O. amphimone population dynamics.


INTRODUCTION
Ormiscodes caterpillars are Saturniid moths known to defoliate extensive areas of native southern beech (Nothofagus) forests as well as plantations of introduced pines (Pinus radiata D. Don) in southern Chile and Argentina (Baldini & Alvarado 2008).These outbreaks reduce radial tree growth (Paritsis et al. 2009), can kill saplings and apical shoots (Bauerle et al. 1997, Cogollor 2002), and have been suggested to cause partial crown dieback on Nothofagus stands if defoliation is severe (Veblen et al. 1996).In addition, Ormiscodes caterpillars are a nuisance for tourism and outdoor activities in general when they reach epidemic population levels due to their irritating hairs (Artigas 1972, Baldini & Alvarado 2008).
Over the past 30 years, outbreaks of O. amphimone, one of the most widespread Ormiscodes species in southern South America, have increased in frequency in many N. pumilio (Poepp. & Endl.)Krasser forests in Patagonia, probably due to climate warming (Paritsis & Veblen 2011).These outbreaks, which last a single growing season, can defoliate several thousand hectares of Nothofagus spp.forests, but N. pumilio is typically the most affected species (Baldini & Alvarado 2008, Paritsis et al. 2011).Recent O. amphimone outbreaks (2007, 2008 and 2009) on N. pumilio and N. antarctica (Forster) Oerst. in the Aysén region (Chile) generated concern among land owners and farmers, and the Agriculture Ministry of Chile took measures to monitor these outbreaks (Anonymous 2007, Baldini & Lanfranco 2008).
Despite the ecological impor tance of defoliation events caused by O. amphimone in Nothofagus forests in Patagonia, there are few specifi c studies on the ecology and natural history of this species.These include tree-ring reconstructions of past outbreaks (Paritsis & Veblen 2011), evaluation of larval performance to diet quality and temperature (Paritsis & Veblen 2010), and biophysical correlates with spatial patterns of defoliation events (Paritsis et al. 2011).However, there are no studies that have quantifi ed the incidence of natural enemies of O. amphimone in order to assess their potential role in the population dynamics of this outbreak defoliator.
Population ecologists interested in insect herbivores have often disagreed on the extent to which their population densities are controlled by food supply (bottom-up effects) or by the effects of their natural enemies (topdown effects) (Hunter & Price 1992, Schmitz 1994).Natural enemies are known to have great impacts on the population dynamics of forest Lepidoptera in various ecosystems (e.g., Klemola et al. 2002, Dwyer et al. 2004).Nevertheless, the relative infl uence of natural enemies on populations of forest Lepidoptera is highly variable depending on the species and location considered.For instance, while predators of pupae appear to play a signifi cant role in gypsy moth (Lymantria dispar [L.]) outbreak dynamics in North America (Liebhold et al. 2000), they do not seem to be a major factor in local outbreaks of the winter moth, Operophtera brumata L., in southern Finland (Heisswolf et al. 2010).In southern South America multiple species of natural enemies have been obser ved to attack Ormiscodes species at different stages of their life cycle (Silva 1917, Peigler 1994, Cogollor 2002), and it has been suggested that Ormiscodes populations could be controlled by the mortality associated with parasitoid attack (Silva 1917, Artigas 1972, Cogollor 2002).However, quantitative data on mortality of Ormiscodes species caused by parasitoids and generalist vertebrate predators, such as birds and rodents, are scarce or, as in the case of O. amphimone, non-existent.
Given the recent increase in the frequency of O. amphimone outbreak in multiple areas of Patagonia and the potential for more outbreak events to occur in the future (promoted by warming trends) (Paritsis & Veblen 2011), it is essential to quantify mortality caused by natural enemies on these populations in order to assess the potential role of parasitoids and predators on O. amphimone demography.Because N. pumilio appears to be the most extensively defoliated species (Baldini & Alvarado 2008, Paritsis et al. 2011), our objective is to quantify mortality of eggs, larvae and pupae of O. amphimone caused by natural enemies in N. pumilio forest stands in northwestern Patagonia.

Study area
Surveys and fi eld experiments were conducted at Paso Puyehue (40º43' S, 71º55' W; 1150 masl) in a valley bottom location at the foothills of the eastern slope of the Andes in Argentina.This is an area of relatively dense N. pumilio forest with an open understory of forbs (e.g., Adenocaulon chilensis Less.) and small shrubs (e.g., Ribes maguellanica Poer., Drymis winteri Forst., and Gaultheria mucronata [L.f.] Hook.& Arn.).To account for potential heterogeneity in predation levels within the study area, we selected three sub sites of 2 to 4 ha in size each, separated 0.5 to 2 km from each other, to conduct predation assessments.The climate in this region is characterized by cold and wet winters, and mild but dry summers, and the growing season occurs mainly from November to Februar y.Mean annual precipitation is approximately 3000 mm (Barros et al. 1983).

Study species
Ormiscodes amphimone is a native and polyphagous Saturniid moth that feeds preferentially on N. pumilio foliage (Baldini & Alvarado 2008, Paritsis et al. 2010).Outbreaks of this species generally do not cause widespread tree mor tality, probably because the defoliation event lasts only a single growing season.O. amphimone over winters in the egg stage, and lar vae emerge during early to late spring (Lemaire 2002).Larvae feed gregariously until mid to late summer, going through fi ve instars before pupating in the forest fl oor litter.Adults emerge after one to two months of pupation (Baldini & Alvarado 2008, personal obser vation) and, after mating, females lay their eggs in a single cluster around small diameter twigs of the host plant.
Multiple parasitoids have been described to attack Ormiscodes species.The most frequently cited include the egg parasitoids Horismenus ancillus Brèthes (synonym: Dirphiphagus ancilla) (Hymenoptera: Eulophidae) and Paridris chilensis Brèthes (Hymenoptera: Platygastridae), and the larval parasitoids Apanteles spp.(Hymenoptera: Braconidae) (Silva 1917, Cogollor 2002, Baldini & Alvarado 2008).Despite some early obser vations by Silva (1917) on the life history of parasitoids of Ormiscodes sp.(cited as Dirphia amphimone) in central Chile, little is known about the identity and the natural endemic-levels of parasitism of Ormiscodes species, especially in Nothofagus forests.Vertebrate predators of larvae and pupae, such as birds and small mammals, are known to exert signifi cant predation pressure on populations of eruptive Lepidoptera species in forests of the northern hemisphere (e.g., Liebhold et al. 2000, Barbaro & Battisti 2011).However, the potential ef fects of ver tebrate predators on Ormiscodes populations are largely unknown.

Egg parasitoids
During mid-February 2004 (i.e., ca.three months after egg hatching), we collected 15 egg clusters that were laid over fall 2003 (i.e., March-May) on N. pumilio saplings.Egg clusters laid in previous seasons (recognized by a pronounced weathering and discoloration of the egg shells) were not included in this analysis because it was not possible to accurately assess parasitism incidence.We counted the number of hatched eggs, parasitized eggs, and unhatched eggs in each egg cluster with a 70x magnifi cation dissection microscope.Parasitized eggs can be readily dif ferentiated from successful O. amphimone larval emergence by the size of the emergence opening, which is of significantly smaller diameter for the parasitoid (Fig. 1A) than for O. amphimone (Fig. 1B).We verifi ed that the smaller emergence holes were created by parasitoids by observing and photographing O. amphimone larvae and parasitoid emergence from eggs with a 70x magnifi cation dissection microscope on two fresh egg clusters collected earlier in the same season.To test for a potential infl uence of the total number of eggs per cluster on the number of parasitized eggs we conducted a linear correlation between these variables using SPSS (2007).

Larval parasitoids
To assess mortality caused by parasitoids and to determine parasitism per instar, we collected groups of larvae from the fi eld at their fi rst, second, third and fourth instar and reared them until pupation or parasitoid emergence.We did not examine parasitism at the fi fth instar because larvae are mostly solitary during this instar, which makes it diffi cult to obtain a suffi ciently large sample size.We fi rst located groups of larvae in the fi eld and monitored them at weekly intervals until they reached the desired instar.Instars were identifi ed by the approximate width of the head capsule (Llanderal 1993) and from previous knowledge of their relative body size.Larval collection in the fi eld was conducted in two different summers.In the fi rst summer (2003-2004; hereinafter 2003) we collected a total of 28 groups of larvae from the three sub sites, totaling 560 larvae.When collected, two groups were in fi rst instar (29 larvae), fi ve groups were in second instar (117 larvae), eight groups were in third instar (143 larvae), and 13 groups were in fourth instar (271 larvae).In the second summer (2006-2007; hereinafter 2006) we collected a total of 14 groups of larvae (474 larvae total).At the time of collection, two groups were in fi rst instar (31 larvae), seven groups were in second instar (292 larvae) and fi ve groups were in third instar (151 larvae).No fourth instar larvae were collected in the second summer.Groups of larvae were reared in plastic containers in the laboratory at room temperature (18 to 20ºC) with a 14:10 L:D photoperiod and constant supply of fresh N. pumilio leaves until pupation or parasitoid emergence.We quantifi ed the number of larvae killed by parasitoids per group and instar.Parasitoid specimens were preserved in 70 % ethanol and were taxonomically classifi ed using the keys and/or taxa descriptions in Townes & Townes (1966) (Ichneumonidae), Johansen (2010) and Silva (1917) (Braconidae), Hansson (2009) (Eulophidae), and Poinar (1983) (Nematoda).To assess temporal variation and variation in the incidence of different parasitoid species in larval mortality we used the data for third instar larvae (i.e., the most representative instar for estimating parasitism in this system).These data were analyzed with a two-way analysis of variance, with year, parasitoid species and their interaction as the main effects using SPSS (2007).

Bird predation on larvae
Bird predation on larvae was estimated by quantifying lar val mor tality on N. pumilio saplings with bird exclosures and on saplings with free access to birds.In late December 2003 we identifi ed 24 N. pumilio saplings 1.5-3.5 m in height in the study area, which had one O. amphimone egg cluster attached.On January 23, 2004, when larval groups were in second to third instar, we randomly selected 12 saplings and covered each of them with a bird-proof mesh (mesh size 2 × 2 cm) and left the 12 remaining saplings uncovered.Bird-proof meshes allow parasitoids and other insect predators free access to larvae, but protect larvae from bird predation.After bird-proof mesh placement we counted the number of larvae per group on each sapling, which ranged from 12 to 150 (67 ± 12, mean ± SE) larvae each.Bird-proof meshes were installed on the saplings for seven days.During the sampling period, the commonest birds feeding on insects on canopy foliage in these forests are the thorn-tailed rayadito, Aphrastura spinicauda Gmelin (Furnaridae); the white-crested elaenia, Elaenia albiceps Orbigny & Lafresnaye (Tyranidae), and the house wren Troglodytes aedon, Vieillot (Deferrari et al. 2001, personal observation).Seven days after mesh installation and before larvae reached the fourth instar and actively disperse, we re-counted larvae in each group to estimate mortality (or disappearance) in both covered and uncovered saplings.Bird predation was estimated as the difference in mortality between treatments (i.e., mesh-covered vs. uncovered).Because during the second time we counted the larvae we were unable to fi nd three groups of larvae on uncovered saplings and one group on a mesh-covered sapling, the fi nal sample size was reduced to 11 bird exclusion samples and nine control samples.The effect of bird exclosures on larval mortality was assessed with an independent samples t-test using SPSS (2007).

Predation of pupae
To assess predation of pupae in the field we placed pupae in the forest litter from February 24 to March 12, 2004 (i.e., 17 days) -which coincides with a fraction of the time pupae are naturally found in this area-and quantified predation.To test for potential ef fects of density of pupae on pupal predation we established two sets of plots with dif ferent pupal densities.We defined eight 2 × 1 m high pupal density plots and placed 18 uniformly distributed pupae in each plot (i.e., six pupae per square meter).We also defined eight 1 × 1 m low pupal density plots where we placed two pupae on each (i.e., two pupae per square meter).These eight low density plots were paired with the eight high pupal density plots (i.e., at 10 to 15 m from these).Plot pairs (high-low densities) were at least 40 m away from any other pair of plots.By February 24, 2004, when O. amphimone pupae are typically found in these forests, pupae were buried in the forest litter at a depth of 1 cm at the base of a wooden stick which indicated their location within the plot.Predation of pupae was estimated by quantifying missing or partially eaten pupae after 17 days in the fi eld.When partially eaten pupae were found, we discarded potential adult emergence by examining the pupal case for lines of weakness typical of successful adult emergence.Common species that could predate on pupae in our study area are ground birds, such as the chucao tapaculo, Scelorchilus rubecula Kittlitz (Rhinocriptydae) (Correa et al. 1990), and various small mammals, such as Geoxus valdivianus Thomas (Cricetidae) (Meser ve et al. 1988, R. Sage, personal communication, 2011).

Egg parasitoids
Egg clusters ranged from 87 to 229 eggs and had 167.8 ± 10.9 eggs on average.The only species of egg parasitoid we obser ved emerging from fresh eggs was Horismenus ancillus Brèthes (Hymenoptera: Eulophidae); consequently, it is likely that this species is responsible for most of the parasitism we observed on O. amphimone egg clusters (Fig.   ).However, it is possible that other species of egg parasitoids (e.g., Paridris chilensis Brèthes) might have also contributed to the egg parasitism documented here.Mean egg parasitism per egg cluster was 11.5 ± 7.5 % (Fig. 2A) and ranged from 0 % to 92.5 %.Mean percentage of eggs per egg cluster that failed to hatch was 24.4 ± 10.7 % (Fig. 2A) and ranged from 0 % to 98.2 %.Twelve out of the 15 egg clusters collected had less than 7 % parasitism, while three egg clusters had more than 27 % (and up to 92.5 %) parasitism (Fig. 2B).There was no signifi cant linear correlation between the number of parasitized eggs and the total number of eggs per cluster (r = -0.08;n = 15; P = 0.8).

Larval parasitoids
Ormiscodes amphimone larvae were parasitized by a total of three parasitoid species in our study area and parasitism varied substantially; not only among species of parasitoids (F 2,39 = 8.9; P = 0.001), but also between survey years (as demonstrated by a signifi cant parasitoid species by year interaction; F 2,39 = 12.0; P < 0.001) (Fig. 3).An undescribed species of Hyposoter Förster (Hymenoptera: Ichneumonidae) (D. Wahl, personal communication, 2011) was the most common parasitoid af fecting O. amphimone larvae during 2003, while Apanteles sp.Forster (Hymenoptera: Braconidae) and an unidentified species of mermithid nematode (Mermithida: Mermithidae) caused markedly less parasitism (Fig. 3).In 2006, the nematode species was the most common parasitoid attacking O. amphimone larvae, while Hyposoter sp.considerably decreased its parasitism and Apanteles sp. was absent from our samples (Fig. 3).Parasitism by Hyposoter sp.varied markedly among groups of lar vae ranging from 0 % to 96 %.Hyposoter sp.parasitized O. amphimone larvae as early as during their fi rst instar (Fig. 3A) and larvae were typically killed during their fourth instar when the parasitoid pupated inside the mummified lar va (Fig. 4A).However, we also obser ved individuals of Hyposoter sp.emerging from pupae of O. amphimone.Because Hyposoter sp.emergence from pupae was substantially lower compared to the emergence from fourth instar mummifi ed lar vae, we did not include these parasitized pupae in the analyses.Hyposoter sp.proved to be a solitary endoparasitoid, as only one adult (Fig. 4B) emerged per O. amphimone larva (or pupa).Apanteles sp., the rar est of the thr ee documented parasitoids in our study area, parasitized O. amphimone larvae as early as in their third instar (Fig. 3B).Multiple larvae (ca.15 to 25) of Apanteles sp.emerged from single O. amphimone larvae when these were in their fi fth instar, and each formed a silk cocoon to pupate attached to the body of the larva.The nematode species parasitized O. amphimone larvae as early as in their second instar (Fig. 3C) and emerged and killed the larvae during their third or fourth instar.Typically, one single nematode emerged per larva of O. amphimone, but in rare occasions we also recorded two individuals emerging from the same larva.

Bird predation on larvae
There was no signifi cant difference in lar val disappearance between groups of lar vae on saplings with (9.4 ± 2.3 %) and without (12.8± 4.7 %) bird-proof mesh (t 18 = 0.7; P = 0.5), which suggests that bird predation on O. amphimone larvae was not signifi cant over the seven-day study period (Fig. 5).

Predation of pupae
Pupal predation was the highest compared with predation and parasitism in the egg and larval life stages.Nevertheless, pupal density did not signifi cantly affect predation levels (t 7 = 0.9; P = 0.4) (Fig. 6).Most predated pupae were likely consumed by vertebrate predators because they disappeared leaving no remains (Frank 1967, Heissenwolf et al. 2010).We obser ved only four (i.e., 3 %) partially eaten pupae, which may have been predated by invertebrates.However, the signifi cance of invertebrate predators is  Porcentaje de mortalidad (media ± EE, estimada de la desaparición de larvas) en grupos de larvas sobre renovales sin redes y con redes antipájaros.The relatively high parasitism we recorded confirms previous qualitative obser vations indicating a high incidence of parasitoids on Ormiscodes populations (Silva 1917, Cogollor 2002).In addition, the high predation levels of pupae, a stage with possibly a much higher per capita reproductive value than earlier larval stages (Caswell 2001), suggests that pupal predation may also be important in regulating O. amphimone population dynamics.Egg parasitoids are ef fective controls of herbivor y because they kill lar vae before these star t damaging the host plant (e.g., Trichogramma spp; Smith 1996).Therefore, despite the relatively low levels of egg parasitism recorded here (ca.11 %) egg parasitoids could regulate herbivor y levels caused by O. amphimone in N. pumilio forests.In addition, a portion of the unhatched eggs could have also been parasitized by Horismenus ancillus, but the adult parasitoids failed to emerge, and thus egg parasitism may have been higher than what was obser ved here.Our preliminar y data also suggest that egg parasitism follows an aggregated distribution pattern in which a few egg clusters are heavily parasitized while the majority suffers minimal or no parasitoid attack (Fig. 2B); however, a larger number of egg masses is needed to confi rm this trend.
L a r v a l p a r a s i t i s m w a s s u b s t a n t i a l l y higher than egg parasitism, but it was highly dependent on the timing (i.e., instar) that larvae were collected in the fi eld.Collection of early instar larvae typically resulted in lower levels of parasitism than collection of more advanced instars (Fig. 3), most likely due to the shorter time period that lar vae had been exposed to natural enemies in the fi eld (Benrey & Denno 1997).In addition, parasitism by Hyposoter sp. and the nematode species documented for four th instar lar vae (Fig. 3) is likely underestimated given that these parasitoids generally killed O. amphimone lar vae during their four th instar (and also third instar in the case of the nematode).Consequently, a considerable number of fourth instar lar vae may have died due to parasitoid emergence before being collected in the fi eld and therefore the propor tion of parasitized four th lar vae in the fi eld was probably higher than in our samples.These observations should be taken into account when quantifying larvae parasitism in O. amphimone populations, because collection of early (i.e., fi rst and second) or late (i.e., fourth and fi fth) instar larvae only, could result in underestimation of parasitism levels.Thus, based on our results, collecting third instar larvae should provide the most realistic estimates of larval parasitism in O. amphimone for the parasitoid species evaluated here.
T wo of the three species documented here that parasitize O. amphimone larvae can be effective at decreasing herbivore damage in these forests.Both Hyposoter sp. and the mermithid nematode species typically killed lar vae during the fourth instar at the latest.Ormiscodes amphimone larvae start increasing foliage consumption exponentially at their fourth instar.During the fourth and fi fth instar, lar vae consume ca.90 % of the total foliage they consume over their lifespan (J.Paritsis, unpublished data).Therefore, the relatively high mortality caused by these natural enemies (i.e., ca.50 % in third instar lar vae, Fig. 3) in combination with the timing of parasitoid emergence implies that mortality caused by both species of natural enemies may have impor tant implications in controlling levels of herbivory by O. amphimone in N. pumilio forests.
The substantial change in the relative abundances of parasitoids between years implies that the parasitoid community of O. amphimone is highly dynamic over time.Multiple causes may be responsible for the marked reduction in the parasitism incidence by Hyposoter sp. in 2006.First, climate variability and stochastic population fl uctuations are well-known causes of changes in rates of parasitoid attack and may explain the observed variation in relative parasitoid abundance.Second, a low severity O. amphimone outbreak occurred during the 2001-2002 austral summer in the study area (i.e., two years before we collected the 2003 samples) (Paritsis et al. 2009).Consequently, the relatively high Hyposoter sp.parasitism values recorded in 2003 could be associated with a postepidemic (declining) population phase of O. amphimone and the values in 2006 may be more representative of an endemic population phase.Finally, the higher incidence of the nematode species in 2006 may have caused a failure of emergence of Hyposoter sp.The nematode species generally killed O. amphimone larvae earlier than Hyposoter sp., which may have caused a decline in successful emergence of Hyposoter sp.(Begon et al. 1997).This suggestion is reinforced by the observation that in 2006 fi ve out of the six groups of larvae that were parasitized by Hyposoter sp.(with at least 5 % parasitism) were simultaneously parasitized by the nematode.Susceptibility of parasitic Hymenoptera to simultaneous use of the host with entomopathogenic nematodes is a wellknown cause of parasitoid emergence failure in other host-parasitoid systems (Zaki et al. 1997, Lacey et al. 2003).
In contrast to parasitoids, birds did not exert signifi cant predation pressure on second and third instar O. amphimone lar vae.Previous studies conducted in our study area that evaluated the ef fect of birds on folivor y by insect chewers on N. pumilio saplings showed contrasting trends.Studies by Mazía et al. (2004Mazía et al. ( , 2009) ) found that bird exclusion over the entire growing season increased folivor y, suggesting that birds may reduce populations of insect chewers such as O. amphimone; however, Garibaldi et al. (2010) did not fi nd a signifi cant effect of bird exclusion on folivor y levels by insect chewers.It is likely that the potential contradiction in the role of birds as predators of O. amphimone larvae among studies (including ours) stems from the temporal variability of bird predation pressure.Furthermore, in our study some larvae may have dispersed during these early instars adding random variability to our dataset.Hence, further research assessing bird predation across all larval instars and over longer periods of time should provide more defi nitive answers regarding the role of birds as predators of O. amphimone larvae.
Predation on advanced life stages, such as pupae, has a crucial infl uence on population dynamics, given its immediacy to the reproductive stage (Caswell 2001).Predators of pupae can exer t significant control of forest Lepidoptera during endemic population phases maintaining populations at stable levels (Tanhuanpää et al. 1999, Liebhold et al. 2000).When generalist predators decrease predation pressure on pupae due to availability of alternative food items, lack of pupal predation could favor the onset of an outbreak, as has been suggested to occur with small mammal predators and the gypsy moth, L ymantria dispar, in North America (Liebhold et al. 2000).Consequently, predation of pupae may be a critical top-down control regulating outbreak dynamics of O. amphimone.The high levels of pupal predation documented in our study (i.e., > 70 %) warrants further research on the identity and food habits of the potential pupal predator species.
Despite the limited spatiotemporal extent of our study, this research constitutes a fi rst step towards understanding the potential role of natural enemies in regulating populations of O. amphimone.Because the mechanisms responsible for O. amphimone outbreaks remain largely unknown, there is a need to explore how parasitoids and generalist pr edators (mainly of pupae) influence changes in O. amphimone populations.The high predation levels documented here suggest that natural enemies could be a key factor infl uencing outbreak dynamics of O. amphimone.The validity of our results, however, needs to be verifi ed at spatiotemporal scales r elevant to outbr eak dynamics.Therefore, to better understand the dynamics in population fl uctuations of this key herbivore in Nothofagus forests, future research should focus on assessing temporal variability in mortality due to multiple natural enemies attacking egg, lar vae, pupae and adult stages across larger areas of the geographic distribution of O. amphimone.

Fig. 2 :
Fig. 2: (A) Percentage (± SE) of successfully hatched, parasitized, and unhatched O. amphimone eggs per egg cluster collected in the fi eld three months after caterpillar emergence.(B) Frequency of egg clusters according to their percentage of parasitism.

Fig. 3 :
Fig. 3: (A) Mean (± SE) larval parasitism for groups of caterpillars collected in the fi eld in their fi rst to fourth instar in 2003 and 2006 caused by Hyposoter sp., (B) Apanteles sp., and (C) a nematode species."ND" indicates that no data are available for that instar/year.Note the change in the scale of the axis for plot B. (A) Parasitismo medio (± EE) larval por instar en 2003 y 2006 causado por Hyposoter sp., (B) Apanteles sp.y (C) la especie de nemátodo."ND" indica que no existen datos disponibles sobre ese instar/año.Nótese la escala diferente del eje en el gráfi co B.

Fig. 5 :
Fig. 5: Percent of mortality (mean ± SE, estimated from larval disappearance) of larval groups on uncovered (no mesh) saplings and saplings covered with bird-proof mesh.