Introduction
In insect-plant interactions, the selection of a host plant by a maternal insect can be a highly influential factor in parent-offspring relationships, where host plant suitability can affect both parental fecundity and offspring performance (Bernays & Chapman, Reference Bernays and Chapman1994). The interaction between maternal choice of host plant and subsequent offspring performance is often addressed in relation to the preference-performance hypothesis (PPH), which has received renewed interest recently (Gripenberg et al., Reference Gripenberg, Mayhew, Parnell and Roslin2010 and references therein). The PPH was first proposed by Jaenike (Reference Jaenike1978) and states that maternal insects will preferentially lay eggs on host plants that optimise the survival and performance of their offspring. The hypothesis particularly refers to insects whose larvae have limited or no ability to relocate and who are dependent on the maternal selection of host plant. In order to optimize offspring performance, the hypothesis predicts a strong association between the egg laying preferences of the mother and offspring performance (Mayhew, Reference Mayhew2001).
Many studies examining maternal oviposition preferences and offspring performance support the PPH (e.g. Craig et al., Reference Craig, Itami and Price1989; Heisswolf et al., Reference Heisswolf, Obermaier and Poethke2005; Staley et al., Reference Staley, Stewart-Jones, Poppy, Leather and Wright2009). Equally, though, linkages between egg laying preferences and offspring performance can be weak or go undetected (e.g. Rausher, Reference Rausher1979; Scheirs et al., Reference Scheirs, Zoebisch, Schuster and De Bruyn2004; Digweed, Reference Digweed2006; Gripenberg et al., Reference Gripenberg, Morrien, Cudmore, Salminen and Roslin2007). The presence of weak PPH linkages has led to a range of alternate hypotheses examining why maternal insects do not select the optimal host plant. These include optimal foraging, where maternal insects select host plants with superior nutritional quality without accounting for the suitability for offspring performance (Scheirs et al., Reference Scheirs, De Bruyn and Verhagen2000), and enemy free space (Thompson, Reference Thompson1988a,Reference Thompsonb), where preference-performance linkages are more strongly influenced by natural enemies (normally not incorporated in experiments).
To date, much of the research examining the PPH has incorporated insects with aboveground lifecycles. Equally, though, the hypothesis could be applied to maternal insects living aboveground that have soil-dwelling offspring with comparatively less capacity to relocate between plants (Johnson et al., Reference Johnson, Birch, Gregory and Murray2006). For example, the cabbage root fly (Delia radicum) preferentially lays eggs on plants with roots that had been pre-conditioned by existing larvae to be more suitable for their offspring (Baur et al., Reference Baur, Koštál, Patrian and Städler1996a,Reference Baur, Koštál and Städlerb). The same species, however, avoids laying eggs on plants that were in the vicinity of frass of its offsprings’ competitors (Jones & Finch, Reference Jones and Finch1987). The mechanisms and cues underpinning such linkages between aboveground maternal insects and belowground offspring are thought to be either plant- or soil-mediated (reviewed by Johnson et al., Reference Johnson, Birch, Gregory and Murray2006).
Here, we consider the PPH in an aboveground-belowground context in relation to two raspberry (Rubus idaeus) cultivars using the vine weevil (Otiorhynchus sulcatus (F.)) (Coleoptera: Curculionidae) as a model species. We selected Glen Ample and Glen Rosa as potentially good and poor hosts, respectively. Glen Rosa is generally more resistant to some insect pests (e.g. the large raspberry aphid, Amphorophora idaei: McMenemy et al., Reference McMenemy, Mitchell and Johnson2009) and shows less vigorous growth than Glen Ample. Moreover, Glen Rosa is particularly susceptible to leaf rust, which may deter insect feeding (S.N. Jennings, personal communication) in the field. Vine weevils are parthenogenetic and unfertilised eggs develop into new females without the need for males. The vine weevil is a highly suitable study species for investigating parent-offspring relationships in an aboveground-belowground context, as offspring are genetic clones of the adult. The adult weevil is highly polyphagous, feeding on over 150 different plant species (Smith, Reference Smith1932; Warner & Negley, Reference Warner and Negley1976). However, it is the root feeding larvae that cause most plant damage, decreasing plant vigour and growth and potentially causing death (Penman & Scott, Reference Penman and Scott1976; La Lone & Clarke, Reference La Lone and Clarke1981; Moorhouse et al., Reference Moorhouse, Charnley and Gillespie1992). Indeed, vine weevil larvae cause an estimated £8 million of damage to UK strawberry production every year (HDC, 2003).
The comparatively limited ability of the larvae to move belowground means they are restricted to the host plant choice of the maternal weevil. Adult weevils live aboveground where they lay eggs at soil surface and on plants (although the latter usually fall to the soil) which subsequently develop into root feeding larvae. The life cycle has four distinct stages: eggs, larvae, pupae and adults. Each stage may occur concurrently (Schread, Reference Schread1972) and adults can, therefore, be feeding on plants aboveground, which are already exposed to root feeding larvae belowground.
The majority of studies investigating vine weevil preferences in relation to host plants have examined the relationship in terms of adult weevil oviposition and feeding behaviour (e.g. Shanks, Reference Shanks1980; Maier, Reference Maier1981; Nielsen & Dunlap, Reference Nielsen and Dunlap1981; Cram & Daubeny, Reference Cram and Daubeny1982; Van Tol et al., Reference Van Tol, Van Dijk and Sabelis2004). However, relatively few studies have considered the influence of host plants on larval performance. Strawberry (Fragaria×ananassa) has been demonstrated to enhance both the establishment of vine weevil populations and larval survival in comparison with Norway spruce (Picea abies), white spruce (Picea glauca), yew (Taxus baccata) and rhododendron (Rhododendron catawbiense) (Fisher, Reference Fisher2006). Additionally, larvae feeding on azalea (Rhododendron kiusianum) were smaller and had poorer survival on reaching adulthood compared to adults developing on strawberry (Fragaria× ananassa) or Taxus cuspidata (Hanula, Reference Hanula1988). Such examples provide evidence that the developmental stage of vine weevil larvae can be affected by the host plant species. However, these laboratory studies were conducted in controlled environments, and so the relevance of any such infestations to field populations is unknown.
Like foliar feeding insects, root feeding vine weevil larvae are likely to be influenced by the nutritional status of their host plant, where minerals have been shown to have beneficial, detrimental or neutral influences on insect herbivores (Awmack & Leather, Reference Awmack and Leather2002). Nitrogen (N) and phosphorus (P) are often deemed the most limiting factors in insect development (Mattson, Reference Mattson1980; White, Reference White1993; Elser et al., Reference Elser, Fagan, Denno, Dobberfuhl, Folarin, Huberty, Interlandi, Kilham, McCauley, Schulz, Siemann and Sterner2000; Huberty & Denno, Reference Huberty and Denno2006) due to their low concentrations in plants compared to insects. However, other minerals have been shown to significantly affect insect performance but are often overlooked. These include calcium (Ca) (Scutareanu & Loxdale, Reference Scutareanu and Loxdale2006), potassium (K) (Stamp, Reference Stamp1994), magnesium (Mg) (McKinnon et al., Reference McKinnon, Quiring and Bauce1999), Zinc (Zn) (Alyokhin et al., Reference Alyokhin, Porter, Groden and Drummond2005) and iron (Fe) (Thangavelu & Bania, Reference Thangavelu and Bania1990).
The objective of this study was to investigate vine weevil oviposition behaviour and performance, both aboveground and belowground, on two contrasting raspberry cultivars. The specific aims of this study were: (i) to determine how the two cultivars affected different larval abundance and body mass, and establish whether these traits were related with each other in terms of competition (e.g. high survival giving rise to competition and small body size); (ii) to determine whether either, or both, larval performance traits influenced oviposition behaviour by adults; and (iii) to assess whether these differences were reflected in the field over a two-year period.
It was hypothesised that: (i) vine weevil larvae feeding on Glen Ample would show improved performance (in terms of either abundance or body mass, or both) compared to larvae on Glen Rosa; (ii) adult vine weevils would preferentially lay more eggs on the cultivar that resulted in greatest larval performance, but would not lay excessively to avoid offspring competition; and (iii) field populations of adult vine weevils would be higher on the cultivar that increased larval performance and that was preferentially selected by ovipositing adults.
Methods and materials
Plants and insects
Raspberry plants (cvs. Glen Ample and Glen Rosa) were grown in plastic pots (12.5 cm diameter) containing a 2:1 mixture of insecticide-free compost (peat–sand–perlite mix containing 17N:10P:15 K; William Sinclair Horticulture Ltd, Lincoln, UK) and sand (Silver sand, J. Arthur Bowers, Lincoln, UK). Plants were grown in a greenhouse at optimum conditions (15–20°C, supplemented with artificial light). All experiments were conducted in controlled temperature environments at 21°C±2°C and 16:8 L:D photoperiod.
Ovipositing adult weevils were used from cultures maintained at 17°C±2°C and 16:8 L:D photoperiod fed on a mixture of strawberry cultivars. Melanised vine weevil eggs used in experiments were collected from the cultures ensuring egg viability (Smith, Reference Smith1932).
Larval performance
Ten plants (ca. 13 cm high) of each cultivar were treated with 30 vine weevil eggs (inserted into a small indentation in the soil 1 cm from the plant stem). After five weeks, plants were harvested. Roots were carefully teased apart to recover larvae. Individual larvae were counted and weighed on a microbalance (accuracy±0.01 mg). Root biometrics (root mass and maximum root length) were measured after washing and roots were then snap frozen in liquid nitrogen and stored at −18°C for subsequent chemical analyses.
Frozen root samples were milled to a fine powder for all chemical analyses. The %N and %C concentrations of 2-mg samples were determined by a combination of the Dumas and Pregl methods and were carried out using an Exeter Analytical CE440 Elemental Analyser. The percentage of C and N in the sample was calculated by comparison with known standards.
Measurement of other mineral elements was carried out as described in Johnson et al. (Reference Johnson, Hawes and Karley2009). In brief, root samples (0.05 g) were acid digested for 20 min at 180°C in 3 ml of 15.8 M HNO3 (Aristar grade, VWR International, Poole, UK) followed by oxidation with 1 ml of H2O2 for 20 min at 180°C in closed vessels within a MARS-Xpress microwave oven (CEM, Buckingham, UK). Digested samples were diluted to 50 ml using de-ionised water. Total mineral contents of calcium (Ca), phosphorus (P), magnesium (Mg), zinc (Zn), iron (Fe) and potassium (K) in the digested leaf samples were determined by inductively-coupled plasma mass spectrometry (Elan DRC-e, Perkin-Elmer, Beaconsfield, Bucks, UK).
Paired oviposition experiment with two raspberry cultivars
For the paired choice experiment, 20 Glen Ample (ca. 9 cm high with 14 leaflets) plants were selected and paired up with 20 Glen Rosa (ca. 8 cm high with 14 leaflets) plants according to size. Each plant pair was placed into a mesh cage (45×45×30 cm, height×length×width). The bases of the cage comprised a wooden base with two opposed holes (12.5 cm diameter) into which potted plants could be inserted. This ensured that plants were at least 15 cm apart from one another and were discrete units separated by non-soil substrate that would be unsuitable for oviposition. A fine mesh circular collar was placed around the stem of all plants and then covered with washed gravel (Coarse grit, J. Arthur Bowers, Lincoln, UK) (∼2–6 mm) to allow the retrieval of vine weevil eggs at the end of the experiment (see Johnson et al., Reference Johnson, Petitjean, Clark and Mitchell2010b).
One ovipositing weevil (ca. 1–2 months old) was introduced into each cage. Plants were harvested three weeks after the addition of the weevils. Weevils were recovered from the cages and plant biometrics were recorded (plant height, plant mass, number of leaves, leaf area and root mass). Eggs were recovered from the gravel by immersing it in a saturated KCl solution and gently stirring so that the eggs floated to the surface (see Johnson et al., Reference Johnson, Petitjean, Clark and Mitchell2010b).
Leaf consumption was calculated using a LI-3100C area meter (LI-COR Inc. Lincoln, Nebraska, USA) and digitally scanned leaf areas. Digital images were analysed to determine eaten leaf areas (see Johnson et al., Reference Johnson, Petitjean, Clark and Mitchell2010b). Previous work has established that leaf area was directly correlated with leaf mass in raspberry (Coyle et al., Reference Coyle, Clark, Raffa and Johnson2011).
Field experiment
The experiment was conducted in six separate and adjacent polytunnels at SCRI, Dundee, UK (56°27′N, 3°04′W). Protected cropping systems now provide >80% of UK soft fruit sold through supermarkets (McMenemy et al., Reference McMenemy, Mitchell and Johnson2009) and thus reflect the most realistic field environment for vine weevils feeding on raspberry. Each tunnel (22×8×3.3 m, length×width×height) was covered with Luminance THB polythene film (BPI, London, UK) and contained three raised beds of ca. 24 plants covered with polythene mulch. Three tunnels had been planted with Glen Ample and three with Glen Rosa in April 2005. In all three years of the experiment, the tunnels were left uncovered from October until June, according to commercial practice.
During April–May 2007, plants in each row (three rows per tunnel) were separated into plots of four and enclosed using a corrugated plastic (Correx®; DS Smith Plastics, Warwickshire, UK) barrier (3.25×1.25×0.60 m, length×width×height) that was dug ca. 10 cm into the soil (i.e. each row contained six plots). Weevil eggs collected from culture were applied to plants at regular intervals so that each row received 576 eggs in total (equivalent to ca. 24 eggs per plant). Inoculations were split over four separate occasions during August and September 2007 to facilitate inoculation of the 10,368 O. sulcatus eggs. Eggs were applied equally to the bases of the four plants in each plot.
Vine weevil adults were surveyed every 14 days (±2 days) at night (22:00–01:00) from mid-May until mid-October in 2008 and 2009, starting with the initial population in 2008. Weevils were dislodged on to white beating trays (110 cm×86 cm) (Watkins and Doncaster, Cranbrook, UK) held either side of the plants by shaking the two middle plants in each plot five times. Weevils were placed in labelled containers for counting on the following day after which they were returned to the base of the plants where they had been captured.
Data analysis
Larval mass was analysed by analysis of variance, with a plant as a block factor. Larval survival was analysed with a generalised linear model with a Binomial error structure and logit link function. The number of larvae recovered from each plant was analysed with a two-sample t-test. Relationships between root mineral element concentrations and larval performance were analysed using Pearson's product moment correlation. Mann-Whitney or t-tests (as indicated in text) were used to determine whether the two cultivars were significantly different in terms of specific minerals. Egg laying and feeding behaviour of adult weevils in relation to the paired cultivar experiment were analysed using paired t-tests with transformed data (log and log+1, see figure legends) to address non-normally distributed data. Egg laying in relation to feeding behaviour was examined using Spearman's rank correlations. All of the above analyses were conducted in Genstat version 11 (Payne et al., Reference Payne, Murray, Harding, Baird and Soutar2007).
Differences between the numbers of weevils caught on the two cultivars (Glen Ample and Glen Rosa) in the field were analysed using a generalised linear mixed model with Poisson error structure and log link function. Cultivar and year were included as fixed terms in the model. Tunnel and survey number were both included as random terms. The analysis was conducted using the ‘lme4’ package in ‘R’ programme (version 2.12.1, R Foundation for Statistical Computing).
Results
Larval performance
Vine weevil larvae were significantly more abundant on Glen Ample than on Glen Rosa (t 18=2.50, P=0.022) (fig. 1), with larval survival in terms of the original inoculation with eggs considerably higher (P=0.052) on Glen Ample (18%) than on Glen Rosa (8%). While survival rates were comparatively low compared to strawberry (Cowles, Reference Cowles2004), they were similar to levels reported in other woody perennial plants (Johnson et al., Reference Johnson, Barton, Clark, Gregory, McMenemy and Hancock2011). In contrast, larval mass was significantly higher on Glen Rosa than on Glen Ample (F1,65=1.14, P=0.001) (fig. 1).
Fig. 1. Differences in larval abundance (
) and average larval mass (□). Mean values±SE shown. Larval abundance and mass transformed prior to analysis (log+1 and log, respectively).
Overall, the number of larvae recovered per plant was positively correlated with root N concentrations (fig. 2 and table 1). Root C and root N concentrations were not significantly different between the two cultivars (C: t 16=0.90, P=0.384; N: t 16=0.22, P=0.831), nor did larval performance show any relationship with root C concentrations. The concentration of Mg in the roots was positively correlated with the number of larvae recovered per plant (table 1) but was not significantly different between Glen Ample and Glen Rosa (t 17=0.65, P=0.527). The number of larvae recovered was additionally negatively correlated with the concentration of Fe in the roots (table 1), but there was no difference in Fe root concentrations between Glen Ample and Glen Rosa (Mann-Whitney U test U=25.0, P=0.113). Larval abundance was not correlated with any other root mineral concentrations and larval masses showed no relationships with root mineral content (table 1). Root biometrics (root mass and maximum root length) were not significantly correlated with either the number of larvae recovered per plant or larval mass (results not shown).
Fig. 2. Relationship between root nitrogen concentrations and number of weevils per plant, Glen Rosa (○) and Glen Ample (●). Pearsons’ product moment correlation analyses shown in table 1 are for Glen Ample and Glen Rosa data collectively, linear regression line fitted to data is: y=0.524x–5.45.
Table 1. Summary of correlations for larval mass and abundance in relation to root nutritional quality. Significant relationships highlighted in bold where P<0.05. Correlations were calculated using Pearson's product moment correlations or Spearman's rank correlations as appropriate. Larval mass, n=17; larval abundance n=19.
Paired oviposition experiment on two raspberry cultivars
Adult weevils laid eggs on plants and the surrounding gravel only, with none being laid elsewhere in the cage. Weevils laid similar numbers of eggs on Glen Ample (64.6±4.8, mean±SE) and Glen Rosa (58.9±6.1, mean±SE) when allowed to choose between the plants (t 19=0.92, P=0.369, n=20). Oviposition was not related with any of the plant characteristics quantified (data not shown) and nor was it related to adult feeding in terms of leaf area eaten (rs=0.067, df=38, P=0.681) or proportion of plant eaten (rs=0.012, df=38, P=0.943). Adult weevil feeding preferences between Glen Ample and Glen Rosa were not detected in terms of either the leaf area consumed (t 19=0.71, P=0.488) or proportion of plant eaten (t 19=0.58, P=0.566).
Field experiment
In 2008, 817 weevils were captured across 12 sampling dates, whilst in 2009 2753 weevils were caught across 11 sampling dates. Weevils were significantly less abundant in 2008 (fig. 3a) than 2009 (fig. 3b) (Z=3.50, P<0.001). Moreover, weevils were significantly more abundant on Glen Ample than Glen Rosa (fig. 3) (Z=3.36, P<0.001). The difference between the two cultivars was particularly apparent in 2009 (fig. 3b), which was reflected in the statistically significant interaction between cultivar and year (Z=−7.15, P<0.001)
Fig. 3. Mean number of weevils captured per plot on Glen Ample and Glen Rosa in (a) 2008 and (b) 2009. Mean values±SE shown.
Discussion
Vine weevil larval performance (abundance and masses) differed significantly between Glen Rosa and Glen Ample, with Glen Rosa having fewer, but heavier, larvae than Glen Ample. Larger offspring are often deemed to show superior performance in comparison to smaller offspring (Stearns, Reference Stearns1992), which would suggest that larvae developing on Glen Rosa would be at an advantage. However, whilst the larvae on Glen Ample were smaller in terms of mass than those on Glen Rosa, they were more abundant. The specific reasons for these differences between Glen Rosa and Glen Ample were not established in this study, but Glen Rosa may be better at deterring first instar larvae (e.g. using chemical or physical mechanisms, see Johnson & Gregory, Reference Johnson and Gregory2006; Johnson et al., Reference Johnson, Hallett, Gillespie and Halpin2010a) leading to a higher early mortality. This in turn would lead to a decrease in competition, which could result in fewer but larger larvae. Differences in larval performance were not observed on different cultivars of strawberry (Cowles, Reference Cowles2004), although such differences are commonly observed between different plant species (Fisher, Reference Fisher2006).
Overall, larval abundance was found to be positively correlated with root N concentrations, which is consistent with the reliance of young insect larvae on an N rich source (White, Reference White1993). Potentially, root N concentration could play an important role in the performance of vine weevil larvae on host plants. Additionally, larval abundance was positively correlated with root Mg concentrations, but negatively correlated with root Fe concentrations. Foliar concentrations of Mg have been associated with both increases (e.g. Thangavelu & Bania, Reference Thangavelu and Bania1990) and decreases (Clancy & King, Reference Clancy and King1993) in insect performance, yet the exact role that Mg plays in insect nutrition remains unclear. Fe content in rice plants was found to detrimentally impact the growth and development of the white backed planthopper (Sogatella furcifera) (Horváth), causing lower nymphal survival and prolonged nymphal development (Rath, Reference Rath2004). Consequently, root mineral concentrations may be influential in determining the development of vine weevil larvae.
In our study, the presence of a trade off in larval performance parameters between the two raspberry cultivars may have complicated the decision of the adult weevil. The results showed no evidence of any link between adult weevil oviposition and the performance of vine weevil larvae belowground. The inability of maternal adult insects to select a host plant for oviposition which maximises the survival and development of subsequent offspring has been considered several times in relation to the preference-performance hypothesis (Jaenike, Reference Jaenike1978; Denno et al., Reference Denno, Larsson and Olmstead1990; Price, Reference Price1991; Scheirs & De Bruyn, Reference Scheirs and De Bruyn2002). A subtle decision between increased abundance or larval masses may simply prove too complex for the highly polyphagous vine weevil. The neural restraints hypothesis (Levins & Macarthur, Reference Levins and Macarthur1969; Bernays, Reference Bernays2001) states that insects have limited capabilities to process information. Consequently, generalist insects are believed to make poorer decisions regarding their choice of host plants, in comparison to specialist insects, due to difficulties in assessing multiple host plant options. Offspring competition may also explain why adults laid similar quantities of eggs on both cultivars. For instance, if Glen Ample were a more suitable host in terms of larval survival, the consequent increased competition (possibly explaining smaller body size, see above) may have caused adults to lay initially on Glen Ample but then switch to Glen Rosa to avoid excessive offspring competition. For pragmatic reasons, our study did not measure the sequence of egg laying, but this is at least a tenable argument.
Typically, the preference-performance hypothesis is studied in a controlled environment devoid of factors that may influence the relationship between mother and offspring, for instance the presence of enemies (Thompson, Reference Thompson1988a,Reference Thompsonb). In this study, the preference-performance hypothesis was not investigated directly in the field (due to difficulties in the collection of both eggs and larvae); however, the build up of a vine weevil population on the two cultivars was monitored for two consecutive years. Larval abundances were significantly higher on Glen Ample than Glen Rosa in the laboratory, which corresponded with the significant difference in adult population sizes recorded on the two cultivars in the field. This suggests that, while there wasn't a strong preference-performance linkage, there is indirect evidence that larval survival is more strongly related to adult abundance than larval body size.
Acknowledgements
The authors would like to thank Sheena Lamond, Carolyn Mitchell, Lindsay McMenemy, Ali Vaughan, Steve Gellatly, Pauline Martin, Antoine Alliaume and Richard McGonigle for assistance with surveying weevils. The authors would also like to thank Jackie Thompson for assistance with running the ICP-MS. The study was supported financially by a NERC CASE studentship NER/S/C/2006/14263 and the Scottish Government's Rural and Environment Directorate.
