Tri-trophic interactions and the minimal effect of larval microsite and plant attributes on parasitism of Sphenella fascigera (Diptera: Tephritidae)

Abstract

Parasitoid insects are important natural biocontrols of insect herbivores. How parasitoids choose particular hosts among many larvae on different plants is not fully understood. Our study system was the coastal plant Senecio lautus and its tephritid herbivore Sphenella fascigera parasitised by an undescribed parasitoid wasp Pteromalus sp. We located S. fascigera larvae in the field and reared pupae to investigate factors that might promote variation in parasitism: host larval location on the plant (stem gall versus flower head), plant density and quality, host larval density, and presence of other insect herbivores. The overall rate of parasitism of S. fascigera was 25%. There was no significant effect of larval location, or any other recorded variable, upon parasitism. One explanation could be the evolution of an efficient search strategy by Pteromalus sp. Future research on whether S. fascigera larvae on congeneric plants experience different rates of parasitism could contribute to understanding the larger web of interactions involving these species.

Keywords:

• New Zealand
• parasitoid
• plant density
• Pteromalus
• Senecio lautus
• spatial distribution
• tephritid
• wasp

Introduction

About a quarter of all insects are parasites or parasitoids. Tri-trophic interactions between plants, their insect herbivores and parasitoids are among the most ubiquitous and important components of larger, more complex food webs (Schoonhoven et al. 2007). Parasitoids are capable of controlling populations of herbivorous insects. They have the potential to act as biocontrol agents for agricultural pests (i.e. crop-damaging insect herbivores) (Aebi et al. 2006) or they may inhibit the success of insect herbivores that act as biocontrol agents (Paynter et al. 2010). Despite studies of the interactions between parasitoids, their hosts and the host plant, a lack of knowledge remains for a great number of tri-trophic interactions (reviewed in Lewinsohn et al. 2005; Paynter et al. 2010). Major differences between ecologically similar or taxonomically related taxa have made it difficult to determine universal tri-trophic interaction patterns, so further species-specific work is needed (Sasakawa et al.2013).

Two steps of host location by parasitoids have been identified: first, finding the host plant of the host insect and second, locating suitable larval hosts on the plant (Vinson 1975, 1976). Both insect herbivores and parasitoids have been documented to use visual cues (Ferreira Santos de Aquino et al. 2012) and chemical cues (Segura et al. 2012) to locate their hosts. Insect herbivores typically face a choice between many potential host plants, and are likely to evaluate plant properties (plant size, health, plant architecture) to maximise offspring survival (Egan & Ott 2007). More structurally complex plants are thought to provide protection from enemies (i.e. parasitoids) by increasing the difficulty of the search for insect hosts (Obermaier et al. 2008). Potential insect hosts may escape from parasitism by seeking microsites that provide a refuge against discovery and attack. For example, the parasitoid wasp Venturia canescens Gravenhorst (Hymenoptera: Ichneumonidae) was unable to parasitise many potential host larvae of Plodia interpunctella Hübner (Lepidoptera: Pyralidae) due to them aggregating in a deep refuge (Begon et al. 1995). Galls are another means by which endophagous herbivores are thought to gain protection from parasitoids (Hawkins 1988; Price & Pschorn-Walcher 1988; Stone & Schönrogge 2003).

Theoretical studies have pointed to the importance of refugia from parasitism for maintaining quasi-stable, density-dependent control of herbivore populations by their parasitoids (Hassell et al. 1991; Hassell & Pacala 1991). The refugia need not provide absolute protection, as models demonstrate that spatial variation in parasitism rates (e.g. induced by structural variation of the host plant or density variations of the host plant and host insect populations) may be sufficient to maintain herbivore and parasitoid populations (Hassell et al. 1991; Hassell & Pacala 1991). Hence, identifying refugia has implications for evaluating the success of potential biological control agents, as discussed by Lynch et al. (1998).

With respect to spatial variation in plant density, the resource concentration hypothesis (Root 1973) predicts a greater density of herbivores, and similarly more predators, in areas with a high amount of resources (i.e. high-density stands of plants). Nevertheless, many studies have shown the opposite effect, with fewer insect herbivores per plant in areas with high resource concentrations and more insects per plant on isolated plants (Karban & Courtney 1987; Hasenbank & Hartley 2015). Observations such as these led to the alternative resource diffusion hypothesis (Karban & Courtney 1987; Yamamura 1999; Otway et al. 2005; Hasenbank & Hartley 2015), which predicts that isolated plants experience a higher insect herbivore load than plants within dense patches. Furthermore, this preference for isolated plants by insect herbivores can also positively affect parasitism colonisation rates (Williams et al. 2001). In some cases, plant volatiles are species-specific for insect herbivores, and therefore attract parasitoids targeting these insects (Moraes et al. 1998).

Our study aimed to investigate the factors affecting parasitoid host choice at a range of scales. We studied a common coastal plant, Senecio lautus (Willd) (Asteraceae), which is frequently infested with Sphenella fascigera (Malloch) fly larvae which can either develop in galls in plant stems or within the flower heads. In turn, the S. fascigera larvae may or may not be parasitised by Pteromalus sp., an undescribed parasitoid wasp (Fig. 1).

Figure 1 Pteromalus sp. Photograph courtesy of F-R Schnitzler.

Specifically we asked whether:

1. Galls in plant stems provide a more protected refuge than flower heads,

2. Plant density (a proximate measure of resource concentration) is positively or negatively related to parasitism,

3. Plant properties (i.e. plant size, vigour and maturity) influence parasitism.

Materials and methods

Study species

Senecio lautus is a flowering annual or short-lived perennial herb with glabrous or sparsely hairy leaves that can be deeply divided and toothed, the leaves are heteroblastic (Burns 2005), and plants can vary tremendously in size (pers. obs.). Leaf shape is an adaptation to wind and abiotic conditions—leaves are succulent when exposed to salt and wind and are generally more deeply serrated in more exposed areas (Burns 2005). Ray florets vary between 0 (rare) and 7–13 (Webb et al. 1988). Plants flower year round (NZPC Network 2013) with a peak in summer. Typical habitats are coastal areas, especially shingle beaches and rocky areas (Webb et al. 1988). Like most plants in the senecioid tribe, Senecio spp. contain pyrrolizidine alkaloids (Jeffrey et al. 1977). Their primary function appears to be to deter vertebrate and invertebrate herbivores (Hol 2011). As a result of the alkaloids’ toxicity, however, some species of insect have evolved a general tolerance to alkaloids, and some specialist herbivores sequester them for their own defence (Nishida 1994; Nishida et al. 1994). Prominent examples of insect herbivores sequestering alkaloids from Senecio spp. include arctiid moths, e.g. Nyctemera spp. (Benn et al. 1979) and Tyria spp. (Naumann et al.2002).

Sphenella fascigera (Diptera: Tephritidae), formerly known as Tephritis fascigera (revised by Hancock & Drew 2003), is an endemic tephritid fly. Its larvae develop in the immature seeds of flower heads or through forming galls in the stems of Senecio plants (Martin 2013; pers. obs.). In captivity, S. fascigera can survive for several months as adult flies (pers. obs.). Sphenella fascigera adults have been observed on a range of Senecio species in New Zealand including: Senecio bipinnatisectus Belcher; Senecio diaschides D.G.Drury; Senecio esleri Webb; Senecio glomeratus Poiret; Senecio hispidulus A. Rich; Senecio lautus G. Forst. Ex Willd.; Senecio minimus Poiret; Senecio skirrhodon DC.; Senecio vulgaris (Martin 2013) and on Senecio elegans and Senecio sterquilinus (Orndorff) (pers. obs.). To our knowledge, no parasitic wasps have been previously recorded from Sphenella fascigera.

Pteromalus sp. (Fig. 1) is an undescribed parasitoid wasp (pers. comm.; Early & Schnitzler 2012) of the Pteromalidae family. There are over 70 species of Pteromalidae in New Zealand (Landcare Research 2009). Most of the Pteromalidae are parasitoids and various types of parasitism such as egg and pupa parasitism are represented (Landcare Research 2009). Pteromalus spp. have been used as biocontrol agents, as in the case of the generalist pupal parasitoid Pteromalus puparum L. (Hymenoptera: Pteromalidae), which was introduced to New Zealand to control Pieris rapae L. (Lepidoptera: Pieridae), but also spread to other hosts (Barron 2007).

Known parasitoids of tephritids in New Zealand include the parasitoid species of the genus Megastigmus (Torymidae). This genus is known to parasitise Procecidochares alani (Tephritidae) on the mist flower Ageratina (= Eupatorium) riparia New Zealand (Schnitzler & Winks, pers. comm. 2012). Species of Megastigmus and Pteromalus have also been recorded in New Zealand and were found to parasitise Procecidochares utilis (Tephritidae) on Mexican devil weed Ageratina (= Eupatorium) adenophora. A Pteromalus sp. is known to parasitise Tephritis cassiniar (Tephritidae) on Cassinia sp. (Asteraceae) (Schnitzler & Winks, pers. comm. 2012).

Field sites

Populations of Senecio lautus were assessed at shingle beaches around Wellington, New Zealand (see Fig. 2) at Owhiro Bay, Owhiro Beach, Moa Point (near Lyall Bay), Makara Beach, Petone, Breaker Bay and Eastbourne Beach. At Moa Point, Petone and Eastbourne, populations of congeneric Senecio spp. (Senecio elegans and Senecio skirrhodon) were present within a few hundred metres of the sampling points. Other plant species commonly present at the field sites included Coprosma repens, Anagallis arvensis var. arvensis, Glaucium flavum, Gaznia rigens, Chrysanthemoides monilifera, Sonchus oleraceus, Lobularia maritima, Brassica spp., Crepis capillaris, Leontodon taraxacoides, Taraxacum officinale, Atriplex prostrata and Galium spp. The vegetation typically occurred in bands parallel to the tideline and the shore-washed debris. Often, coastal Senecio spp. root in debris from the tideline and the sections with Senecio lautus were usually the areas with less immediate public access.

Figure 2 Map showing the locations of the natural Senecio lautus populations sampled in this study (black points). White circles mark urban areas.

Data collection

Plants were surveyed between February and April 2012 when the majority of plants were mature. Quadrats (2 × 2 m), with an internal grid of 50 cm sub-quadrats, w laid out consecutively over the area where Senecio lautus was present. The number of Senecio lautus individuals and other congenerics in each subquadrat was counted, differentiating between juveniles, adults and dead individuals. Each Senecio lautus plant was searched (i.e. opening every flower head and every stem gall) and all S. fascigera pupae encountered were collected for rearing in the laboratory where they were kept in plastic tubes (4.5 cm height and 1 cm diameter) closed with cotton wool and stored at room temperature (17–22 °C). Larval stages of S. fascigera were also recorded, but not collected. The height and diameter of each plant was measured. Linear size was obtained using the formula:

(1)

This formula normalised the distribution of plant size by converting the value from a volume back to a linear value. The phenological state of the focal plant was assessed by counting the number of flowering buds (= 0), buds (= 1), mature flowers (= 2), post-flowering flower heads (= 3) and dried-out flower heads (= 4); the number of branches at the base of the plant, and plant vigour (categorised from 0 [dead] to 5 [very healthy]) and the presence of other insects on the plant (i.e. cecimidoyiid larva or Nyctemera annulata Boisduval [magpie moth] caterpillars) were also recorded. Each 2 × 2 m quadrat was nested within a larger 6 × 6 m quadrat. The number of Senecio plants outside the 2 × 2 m quadrats but within the 6 × 6 m quadrat was recorded. A total of 4740 flower heads, from 356 plants located in 68 small quadrats (2 × 2 m), were searched for pupae across all sites.

Data analysis

Linear mixed-effects models were used to investigate a variety of factors that might influence a parasitoid’s host selection at a range of nested scales. The response variable was binary—either ‘parasitised’ or ‘not parasitised’ for each fly pupa—hence the error distribution was specified as ‘binomial’. The independent variables were: the location of S. fascigera in the plant (stem or flower head); the total number of S. fascigera in the plant; plant size; number of flowers on the plant; density of conspecific Senecio lautus in the 50 cm, 2 m or 6 m quadrats; the presence of heterospecific Senecio plants; plant maturity; plant vigour; and the presence of other insects on the focal plant. Unique identifiers for each individual plant, nested within quadrat, nested within site, were added as random effects.

Stepwise backwards deletion of variables was used to obtain a parsimonious description of important variables that influenced parasitism (Table S1). The model with the lowest Akaike’s Information Criterion (AIC) was retained as the best model. A type II analysis of variance was performed for the model. In addition, all predictors were also tested individually in separate models to check the stability of the results in the face of potential cross-correlations between variables. All data analysis was conducted using the statistical software R (R Core Team 2012). Linear mixed-effect models were fitted using the package lme4 (Bates et al. 2012).

Results

A total of 1420 S. fascigera larvae and pupae were observed across 356 plants. Fifty-eight Pteromalus sp. were reared from 236 viable pupae giving an overall parasitism of 0.246 (Table 1). A total of 328 collected pupae produced neither viable S. fascigera nor Pteromalus sp. Thirty per cent of flower heads examined contained fly larvae or pupae. The greatest number of flies recorded on one plant was 39, from a plant at Owhiro Bay, 20 cm high and 28.5 cm in diameter with 72 flower heads and four major stems.

Table 1 Summary statistics of Sphenella fascigera and Pteromalus sp. incidence by site.

Site	No. of 2-m quadrats	No. of plants	No. of flower heads	Total no. of flies^1	No. of pupae collected^2	Reared flies^3	Reared wasps^3	Fly density per plant^4	Level of parasitism
Breaker Bay	4	7	50	14	6	4	0	2.00	0
Eastbourne	6	15	670	7	11	0	0	0.47	0
Makara Beach	2	2	15	1	0	0	0	0.50	0
Moa Point	3	9	183	6	3	0	0	0.67	0
Owhiro Bay 1	29	181	2336	660	486	91	32	3.65	0.260
Owhiro Bay 2	9	95	949	526	232	49	20	5.54	0.289
Owhiro Bay 3	4	15	57	39	7	1	1	2.60	0.5
Owhiro Beach	7	21	318	165	63	23	3	7.86	0.115
Petone	1	7	162	2	2	1	0	0.29	0
Total	68	356	4740	1420	823	178	58	3.99	0.246

¹ Total number of flies includes all larvae and pupae.

² Collected pupae are all the closed pupae found.

³ Reared from field collected S. fascigera pupae.

⁴ Fly density per plant is the total number of flies divided by the number of plants.

When tested individually, none of the measured variables had a significant influence on the rate of parasitism of S. fascigera larvae. With respect to our primary question, 29 of 145 pupae recovered from stems were parasitised, compared with 29 out of 90 pupae recovered from flower heads (Table 2). In a single-factor test, this difference in rates (20% versus 32%) was not significant (F = 3.392, df = 1, P = 0.066), and the variable was not retained in the minimum adequate model.

Table 2 Parasitism depending on the location of the pupa in the stem or flower head of the plant.

Location	Parasitised	Not parasitised	n	% of parasitism
Flower head	29	61	90	0.32
Stem	29	116	145	0.20
Total	58	177	235	0.25

With respect to plant density, both the fine-scale (50 cm) and coarse-scale (6 m) measures of conspecific density had a weak (non-significant, 0.05 < P < 0.1) effect of increasing parasitism (Table 3). Sphenella fascigera density had no significant effect on Pteromalus sp. parasitism.

Table 3 Results of a multi-variable (stepwise deletion) minimum adequate model and single-variable linear models for parasitism of Sphenella fascigera by Pteromalus sp.

	Minimum adequate model		Variables tested one by one
Variable	β coefficient	P-value	β coefficient	P-value	Significance
1. Location on plant
Flower head-vs-stem	dropped		−0.634	0.066	–
2. Other plant-level factors
Total number of flies	−0.182	0.473	−0.220	0.403	NS
Other insects	NA	0.447	NA	0.306	NS
3. Plant attributes
Maturity	0.517	0.079	0.483	0.085	–
Vigour	peaked at vig. = 3	0.324	NA	0.244	NS
Total flowers	dropped		−0.078	0.696	NS
Size	dropped		−0.393	0.116	NS
4. Neighbourhood plant densities
No. of Senecio lautus within 50 cm	dropped		0.565	0.051	–
No. of Senecio lautus within 2 m	dropped		0.299	0.319	NS
No. of Senecio lautus within 6 m	0.456	0.230	0.728	0.056	–
No. of other Senecio plants within 2 m	dropped		0.009	0.995	NS

NA, single value β coefficients are not applicable for multi-level factors; NS, not significant; –, marginal significance when 0.05 < P < 0.1.

Plant maturity was the only variable to show marginal significance in both the single-variable test and the parsimonious model: older plants hosted pupae with slightly higher parasitism. Plant size, plant vigour and the number of branches showed no significant influence on parasitism (Table 3).

None of the variables was significant in the full model. With backwards stepwise elimination of variables the AIC of the model was reduced by 10.5 (Table S1). The most parsimonious model included the following predictor variables: the total number of insects in the plant (all development stages of S. fascigera), density of Senecio lautus in the 6 m quadrat, plant maturity, plant vigour, and the presence of other insects (Table 3). As in the single-variable models, a positive trend was found for the influence of plant maturity (P = 0.079) (Table 3), i.e. parasitism was slightly higher in S. fascigera recovered from more mature plants.

Discussion and conclusions

We tested a range of factors for their influence on parasitism of S. fascigera larvae. These factors ranged from fine-scale differences in locations on the plant (e.g. stem versus flower head), properties of the plant (such as plant height and maturity) and finally neighbourhood effects of the density of other conspecifics and congeneric plants. None of the variables showed a significant effect on parasitism of S. fascigera by Pteromalus sp. We found some weakly supported trends (0.05 < P < 0.1) for the influence of stem versus flower head, plant maturity and plant density when tested in single-factor models.

In general, insect larvae that inhabit galls are thought to benefit from increased protection from parasitoid attack—especially if the gall is thick relative to the length of the parasitoid’s ovipositor (Ito & Hijii 2004). Although the level of parasitism of S. fascigera larvae recovered from stem galls was two-thirds of the parasitism level for larvae in flower heads, this difference was not statistically significant with our sample size. The variable for ‘stem gall-versus-flower head’ was eliminated from the minimum adequate model, suggesting that it has only a small influence on parasitoid attraction. Hence, there is little evidence that stem galls provide significant protection in comparison to larval development in flower heads, for S. fascigera.

Throughout this study only one fly larva was found per flower head of Senecio lautus, with the exception of a single instance of two larvae. Whether this is the result of S. fascigera only laying one egg per flower head or only one larva surviving is unknown, although if it were the case of competitive exclusion we might expect to see more instances of two immature larvae per flower head. In the closely related Senecio sterquilinus, however, it is not uncommon to observe two S. fascigera larvae within one single flower head (S. Krejcek pers. obs.). Given that Senecio sterquilinus has larger flower heads, and is associated with bird guano, which could lead to a higher nutrient status, the distribution of only single larvae of S. fascigera in Senecio lautus flower heads might be a necessity of resource limitation. The use of stems might be a strategy to avoid intra-specific competition when S. fascigera densities are high.

If plant properties were a crucial factor for parasitoid attraction, an influence of plant size and plant vigour on parasitism would have been expected, supporting the plant vigour hypothesis (Price 1991). However, that was not the case because only plant maturity showed a positive trend for increasing parasitism. This is consistent with the findings of Souza-Filho et al. (2009), who showed with their study on fruit flies (Tephritidae and Lonchaeidae) that seasonality is important for some host–parasitoid interactions. A succession of insect herbivores throughout the season, with insect herbivores such as S. fascigera occurring later in spring, peaking in summer after aphids and leaf miners have mostly disappeared, has been observed (Krejcek, unpubl. data, 2013). A positive relationship between numbers of seed head predators and plant maturity is expected as larvae develop within flower heads.

When the effect of local plant density was tested at different spatial scales around each focal plant (50 cm, 2 m and 6 m) a marginally significant trend was observed indicating a positive relationship between conspecific plant density (at 50 cm and 6 m scales) and rates of wasp parasitism. This runs counter to the resource dilution hypothesis (Yamamura 1999; Otway et al. 2005), which predicts that isolated patches are more heavily infested with insect herbivores and so more likely to be searched by generalist parasitoids (Thomas 1989). Another explanation for the possible attraction of insect herbivores to isolated patches could be to spread the risk of feeding on host plants from large vertebrate herbivores (Williams et al. 2001). Pteromalus sp. is an undescribed species (Early & Schnitzler, pers. comm. 2012) and to our knowledge there are no previous records of other hosts. Nevertheless, there is the potential that it uses other insect host species on other plants, which might influence distributions of wasps amongst S. fascigera larvae. Anecdotal evidence from a congeneric species (Duan et al. 1996) suggests a close relationship between Pteromalus spp. and their tephritid hosts.

Despite taking into consideration plant spatial distribution, plant properties and host larvae location, none of the factors provided a significant explanation of the variation in parasitism rates. It is possible that we did not investigate the most influential factor for Pteromalus sp. attraction, but some authors (Vinson 1976; Xu et al. 2010) have suggested that parasitoid host location can also follow a random distribution. In other words, it appears that the combination of environmental cues and foraging behaviours used by Pteromalus sp. has adapted it to exploit all available S. fascigera hosts in Senecio lautus with almost equal probability, regardless of where the host is located on the plant, the size and quality of the plant, or the density of host plants.

Taking this research one step further, it would be interesting to compare rates of parasitism of S. fascigera located on other Senecio plants, both native and introduced species. Sphenella fascigera has been recorded on Senecio sterquilinus, Senecio elegans, Senecio skirrhodon, Senecio glastifolius and Senecio vulgaris (Krejcek et al., unpubl. data, 2013). A recently adopted host plant shift by S. fascigera to an introduced plant species might yet reveal a refuge of S. fascigera larvae from parasitism by Pteromalus sp. Understanding tri-trophic relationships within the context of novel food webs of native and introduced species is an area of growing ecological importance as the indirect effects of introduced species may be mediated by altered parasitoid–herbivore interactions.

Supplementary data

Supplementary file: Table S1. Model selection through Akaike’s Information Criterion values.

Associate Editor: Dr Rob Cruickshank.

Acknowledgements

We wish to acknowledge the support received from Victoria University of Wellington, the Wellington Botanical Society and F.-Rudi Schnitzler for his help with Pteromalus sp. identification and comments on the manuscript. Chris Winks and John Early also helped with wasp identification. Jon Sullivan provided helpful comments on S. fascigera biology. We also thank the Associate Editor of the New Zealand Journal of Zoology, as well as two anonymous reviewers for their constructive criticism and insights, which allowed us to improve this manuscript.