Quantitative webs as a means of assessing the impact of alien insects

Authors


Karsten Schönrogge, Centre for Ecology and Hydrology, CEH Dorset, Winfrith Technology Centre, Dorchester, Dorset, DT2 8ZD, UK. Tel: 01305 213587; Fax: 01305 213600; E-mail: ksc@ceh.ac.uk

Summary

1. We use quantitative linkage webs to investigate the impact of alien gall wasps on community structure. Britain has been invaded by four alien species of cynipid gallwasp, Andricus corruptrix, A. lignicola, A. kollari and A. quercuscalicis, over the last 150 years. To date, Britain can be divided into four zones from the north to the south with one, two, three and four invading species established in each zone.

2. The four species are naturalized in their new ranges and are locally the most abundant cynipid species, especially in their spring (sexual) generations. Like the native cynipid species they showed dramatic changes (up to three orders of magnitude) in density between generations, and the dominance structure of alien and native host species changed radically from generation to generation.

3. All four invading cynipid species were attacked by native parasitoid species. Using quantified linkage webs, we assess the contribution made by individual host gall species to each parasitoids population size. Although the parasitoid species have been described as broadly polyphagous, suggesting that the aliens should be richly linked with the native cynipid communities, we found that the galls of the invading species have become the main, and in a few cases the sole, contributors to local parasitoid populations, indicating major host shifts by the parasitoid species.

4. Within generations we found very little overlap among the parasitoid and inquiline communities associated with native and alien galls within generations. Similarly, the quantification of indirect interactions among cynipids between generations suggests that parasitoids and inquilines are not main factors in the dynamics of local cynipid communities. The recruitment of parasitoids and inquilines by the invading cynipid species is therefore unlikely to have a strong affect on native cynipid species.

Introduction

The rate of accidental and sometimes deliberate translocations of species around the world is increasing with the intensity of modern transport (Williamson 1996). In general the translocated or invading species will interact directly or indirectly with a community of native species, and it is of great importance to be able to assess the impact of an invasion on native communities. While direct interactions (competition, predation) are often readily observed (although long-term effects from such an interaction will be difficult to predict), indirect interactions (apparent competition/mutualism) are inherently difficult to identify in field situations (Holt & Lawton 1994). Where for instance apparent competition has been shown to be of importance (e.g. the variegated and grape leafhopper in California), the species involved represented a closed community with a specialist parasitoid species as the shared natural enemy (Settle & Wilson 1990). However, where natural enemies that recruited to an invading species are generalists, a community-wide approach is necessary to assess the impact of the invasion on the native community.

The first generation of food webs were intended as a graphic illustration of trophic relationships. For a given group of species, they showed their links to food resources and their natural enemies (Cohen 1978). The second generation of webs used the number of individuals (Askew 1961; Pimm 1982) or energy flows (Paine 1980) to show the relative importance of trophic links in the web. This technique can work well for a single species that is the focal point of a simple web, but can be misleading for larger assemblages particularly when species occur at radically different absolute densities. The equivalent in parasitoid webs would be hosts attacked by polyphagous parasitoid species, linking various host species in the web, and the most recent web studies have estimated the absolute population densities (i.e. per area) of all species involved, and the absolute attack rates (Memmott & Godfray 1994; Memmott, Godfray & Gould 1994; Müller et al. 1999). Here we present quantitative webs of a guild of eight native cynipid gall wasps, their inquilines and parasitoids, that has been invaded by four alien, bivoltine, host-alternating cynipids from south-east Europe. Our aim has been to use quantitative webs in an attempt to gauge the impact of these invasions on the structure and function of the native communities.

The theory underlying the assessment of interaction strength is well developed (May 1974; Yodzis 1988). Suppose that enough was known that a multispecies Lotka–Volterra model of dynamics could be constructed. Then influence is assessed from the Jacobian matrix showing the interaction parameters of each species with every other species in the community. Specifically, the eigenvalues of the Jacobian allow an assessment of the stability of the species assemblage at equilibrium in terms of first-order effects of all of the other species. This technique ignores time-lags and other higher order effects that might affect the dynamics of the web (for a review see Laska & Wootton 1998).

For most field systems we fall well short of this level of detail in our understanding of dynamics. We are seldom confident that the system is close to equilibrium and an invaded community, almost by definition, will not be at equilibrium. More seriously, we usually have little idea about the per capita impact of one species on another, often because we do not know their absolute densities, let alone their functional and numerical responses.

Our model system comprises host gall species that vary in abundance from > 100 per leaf (e.g. Neuroterus numismalis Geoffroy, agamic generation) to < 1 gall per 100 shoots (e.g. Andricus fecundator (Hartig), agamic generation), a density range of more than 5 orders of magnitude. Similarly, the galls vary in size from 1 to 2 mm in diameter (e.g. Andricus quercuscalicis Burgsdorf, sexual generation) to 60 mm diameter (e.g. Biorhiza pallida (Olivier), sexual generation), and even within species but between generations the differences are often striking (the agamic galls of A. quercuscalicis have a diameter of ∼25 mm). The natural enemies in this community are chalcidoid wasps, the vast majority of which are solitary ectoparasitoids in cynipid galls (Askew 1961; Schönrogge, Walker & Crawley 1999, 2000). Most of the parasitoid species that attack hosts in cynipid galls have been recorded from a number of different galls, all of which are on oak (Askew 1961). Therefore, the arrival of four new cynipid species in Britain can potentially affect native species by mechanisms such as apparent competition (Holt & Lawton 1993). The histories of the invasion of the British Isles by the four cynipid species Andricus corruptrix (Schlechtendahl), A. lignicola (Hartig), A. kollari (Hartig), and A quercuscalicis, and the history of the recruitment of natural enemies to the galls of the latter species, have been described in detail (Schönrogge, Stone & Crawley 1995, 1996; Stone et al. 1995), and some effects of the invasions on associated parasitoid populations have been documented (Schönrogge, Walker & Crawley 1999, 2000).

While these studies were based on individual gall-former species and their associated parasitoids and inquilines, we try here to assess the impact of the four biological invasions on a community wide scale.

Methods

By 1994 the four invading gall wasps had different distributions in Britain due to different arrival times and dispersal ability. It was possible to define four geographical zones with all four invaders present in zone 1; A. lignicola, A. kollari and A. quercuscalicis in zone 2; A. lignicola and A. kollari in zone 3; and only A. kollari in zone 4 (Fig. 1). In each of the zones two sites were selected and local characteristics, such as the distribution and size structure of oak stands are given by Schönrogge, Walker & Crawley (1998).

Figure 1.

Locations of the eight sampling sites. The thick lines indicate the boundaries of the four zones according to the current distribution of the four invading gall wasps (see text).

The four alien cynipid species share an annual life-cycle with a sexual generation that develops in spring in galls induced on Quercus cerris L. and an agamic generation that develops in autumn on Q, robur L. (Schönrogge et al. 2000). In 1994 gall densities were measured on six trees of each of the host tree species (twelve trees in 1995) at each of the eight sample sites as described by Schönrogge et al. (1998). The relative densities per shoot were converted into absolute densities per area according to Schönrogge et al. (2000).

Sampling dates

At each site, samples of all cynipid species found were taken at five different times of the year. The density census for all galls was taken in April and early August before galls were matured (some galls drop off their host trees and a census at a later date could have underestimated densities). Collections for rearings were carried out in May, early June and late September so that all galls could be collected at maturity and rearing mortality could be minimized.

Gall collection and rearing

Buds of Q. cerris were dissected and any galls found were transferred to 2 mL glass tubes for incubation. Q. robur trees were searched for 4 person-hours at each site and all species found were collected. We attempted to collect and rear at least 150 galls of each species at every site. If more than 150 galls were collected, a subsample of 150 galls was taken and the excess was reared separately. In none of the rearing of excess galls did the increase in sample size result in an increased estimate of parasitoid and inquiline species richness, and a standard sample size of 150 galls allowed a direct comparison of species richness. For those species where experience suggested that rearing mortality would be high (e.g. Neuroterus spp.), a sample size of 400 galls was found to be more appropriate. For rarer species, every gall found was collected and reared. All rearings were stored in an outside insectary and emerging adults of gall wasps, inquilines and parasitoids were collected weekly and identified.

Pairwise interactions via parasitoids and inquilines

Many parasitoid and inquiline species associated with cynipid galls have been reported to make two or more generations per year. We assessed the potential for pairwise indirect interactions from generation to generation between cynipid gall wasps using the measure

image

to quantify the possible interactions between two hosts via all shared parasitoids, where αik is the strength of the link between host i and host k. The first term in the square brackets is the fraction of parasitoids of species k emerging from host i, while the second term denotes the fraction of parasitoids k that emerge from host j. dij summarizes the interactions between two hosts via all shared parasitoids, thus the outer summation is taken over all parasitoid species (Müller et al. 1999). If two galls, i and j, share no parasitoid or inquiline species, then dij = 0. If species i is attacked by one or more parasitoid species that show high attack rates in host j then dij→ 1, which would constitute a strong interaction particularly if host j is very abundant. This measure is discussed in detail by Müller et al. (1999).

Statistics

In tests of geographical gradients, cynipid and parasitoid species richness were used as covariates in log-linear models, along with ordnance survey grid co-ordinates. All explanatory variables, their interactions (product terms) and quadratic terms were fitted in a maximal model. Significance was assessed by stepwise deletion of terms until a minimal adequate model was reached where all terms included explained a significant amount of deviance (Crawley 1993). Consistency of ranking of species densities was assessed using Spearman's Rank Correlation.

Results

The local species pools of cynipids and parasitoids

The species pools of cynipids and parasitoids, for the eight sample sites over four generations, are shown in Tables 1 and 2. There was a significant decrease towards the north in the number of native cynipid species and in the species of parasitoids (log-linear models, cynipids: F1,7 = 21·89; P < 0·001; parasitoid species: F1,7 = 13·75; P < 0·01) and parasitoid species recorded. However, after controlling for the geographical gradient there was no significant correlation between parasitoid species richness and cynipid species richness across sites (GLM with Poison errors, χ2 = 0·17, P > 0·05).

Table 1.  Cynipid species recorded at the eight sample sites. Sex./agam. gives the location where the galls of the sexual/agamic generation of each species are induced: B = bud, L = leaf, C = catkin, S = Stem, R = root, A = acorn. The superscript c indicates that these galls are exclusively formed on Quercus cerris. All other galls are formed on the native oak species Q. robur and/or Q. petraea. * indicates galls that were present and their location on the tree are given according to Buhr (1965). The names of the invading cynipid species are printed in bold face
 Sex.Agam.SilwoodPuttenhamTattonRuffordErskineFalklandBeaulyDunrobin
Andricus albopunctatus
Andricus anthracinus
Andricus callidoma
Andricus corruptrix
Andricus curvator
Andricus fecundator
Andricus inflator
Andricus kollari
Andricus lignicola
Andricus nudus
Andricus quadrilineatus
Andricus quercuscalicis
Andricus quercusradicis
Andricus quercusramuli
Andricus seminationis
Andricus solitarius
Andricus testaceipes
Biorhiza pallida
Cynips divisa
Cynips longiventris
Cynips quercusfolii
Neuroterus albipes
Neuroterus numismalis
Neuroterus petioliventris
Neuroterus quercusbaccarum

B
C
Bc
L
C
S
Bc
Bc
C

C
L/S
C

C
L
B
B*/L
B*
B*
L
L
B
L
B
L
B
B
B/S
B
B
B
B
B
C/L
A
R
B
C
B
S
R
L
L
L
L
L
C
L
*
*
*
*
*
*
*
*
*
*
*
*
*
*
*
*
*
*
*
*
*
*
*
*
*
*
*
*
*
*
*
*
*
*
*
*
*
*
*
*
*
*
*
*
*
*
*
*
*
*
*
*
*
*
*
*
*
*
*
*
*
*
*
*
*
*
*
*
*
*
*
*
*
*
*
*
*
*
*
*
*
*
*
*
*
*
*
*
*
*
*
*
*
*
*
*
*
*
*
*
*
*
*
*
*
*
*
*
*
*
*
*
*
*
*
*
*
*
*
*
*
*
*
*
*
*
*
*
*
*
*
*
*
*
*
*
*
*
*
*
*
*
*
*
*
*
*
*
*
*
*
Species richness  2020212320171515
Table 2. . Parasitoid species recorded from the eight sample sites during autumn 1994 and spring 1995
 SilwoodPuttenhamTattonRuffordErskinFalklandBeaulyDunrobin
Torymidae
Megastigmus dorsalis
Torymus auratus
Torymus geranii
Torymus flavipes
M. dorsalis
T. auratus
T. geranii
T. flavipes
M. dorsalis
T auratus
T. geranii
T. flavipes
T. auratus
T. flavipes
Torymus cyanea
T. auratus
T. geranii
T. flavipes
T. flavipes
T. auratus
T. geranii
T. flavipes
M. dorsalis
T. auratus
T. geranii
T. flavipes
OrmyridaeOrmyrus punctiger       
Ormyrus nitidulus      
EurytomidaeEurytoma brunniventrisE. brunniventrisE. brunniventrisE. brunniventrisE. brunniventris  E. bunniventris
Sycophila biguttataS. biguttataS. biguttata     
PteromalidaeMesopolobus fasciiventrisM. fasciiventrisM. fasciiventrisM. fasciiventris  M. fasciiventrisM. fasciiventris
Mesopolobus xanthocerusM. xanthocerusM. xanthocerusM. xanthocerusM. xanthocerusM. xanthocerus M. xanthocerus
Mesopolobus sericeusM. sericeusM. sericeusM. sericeusM. sericeus M. sericeusM. sericeus
Mesopolobus dubiusM. dubiusM. dubiusM. dubius    
Mesopolobus fuscipesM. fuscipesM. fuscipesM. fuscipesM. fuscipesM. fuscipesM. fuscipesM. fuscipes
Mesopolobus tibialisM. tibialisM. tibialisM. tibialis M. tibialisM. tibialisM. tibialis
Mesopolobus amaenusM. amaenusM. amaenus    M. amaenus
Cecidostiba hilarisC. hilarisC. hilaris   C. hilaris
Cecidostiba
semifascia
C. hilaris
EupelmidaeEupelmus urozonusE. urozonusE. urozonusE. urozonus    
Ormocerus latusO. latus   Ormocerus
skianeurus
 O. skianeurus
Eulophidae     Aulogymnus
arsames
  
Pediobius lysisP. lysisP. lysisP. lysis P. lysis 
Species richness1718161386109

Host abundance in the 4 zones

The four alien and the four most abundant native cynipid species were ranked according to their absolute abundance averaged over the two sites in each zone (Table 3).

Table 3. . Ranked abundance of the four alien cynipid species and the four most abundant native species in each of the four zones in autumn 1994 and spring 1995. Abundance was averaged over the two sites in each zone. Some species were present at the sites, but below detectable densities during random sampling (–), whereas others where not present at all sites (n/a). The names of the invading cynipid species are printed in bold face
Zone 1
Species
Galls/100 m2Zone 2
Species
Galls/100 m2Zone 3
Species
Galls/100 m2Zone 4
Species
Galls/100 m2
Autumn 1994
Neuroterus quercusbaccarum14253·15N. quercusbaccarum70992·44N. quercusbaccarum284475·20N. quercusbaccarum175276·49
Neuroterus numismalis1043·56A. anthracinus36357·24A. anthracinus1992·22N. numismalis714·44
Andricus anthracinus1007·47C. divisa8477·45N. numismalis1863·60C. divisa338·38
Andricus lignicola488·87A. quercuscalicis1759·08C. divisa206·31A. anthracinus83·88
Andricus quercuscalicis161·67N. numismalis421·46A. lignicola0·03A. kollari2·08
Andricus corruptrix58·31A. lignicola181·03A. kollariA. lignicolan/a
Andricus kollari56·28A. kollariA. quercuscalicisn/aA. quercuscalicisn/a
C. divisaA. corruptrixn/aA. corruptrixn/aA. corruptrixn/a
Spring 1994
Andricus quercuscalicis17110·81A. quercuscalicis212731·78N. quercusbaccarum478778·76A. kollari27986·82
Andricus curvator9434·15A. lignicola43710·54A. quadrilineatus239494·85A. curvator17565·79
Andricus lignicola3856·58N. quercusbaccarum33468·46A. curvator30272·26A. quadrilineatus15516·01
Andricus corruptrix2124·45A. quadrilineatus18799·73A. lignicola20879·80N. quercusbaccarum1726·21
Neuroterus quercusbaccarum533·28A. kollari17467·30N. numismalis20383·02N. numismalis627·18
Andricus kollari483·08A. curvator11700·18A. kollari20378·49A. lignicolan/a
Andricus quadrilineatus125·15N. numismalis3184·68A. quercuscalicisn/aA. quercuscalicisn/a
Neuroterus numismalis9·31A. corruptrixn/aA. corruptrixn/aA. corruptrixn/a

In autumn 1994, Neuroterus quercusbaccarum L. was by far the most abundant cynipid species in all zones with its peak abundance in zone 3 (Table 3). The other three common native cynipid species showed no continuous trend in their abundance patterns, except that all of them were least abundant in the most northern zone (Table 3). In all four zones the alien species were less abundant than the most common native species. The abundance of A. quercuscalicis and A. lignicola decreased from south to north, and A. corruptrix was restricted entirely to the south-east of England. Ranked abundances were significantly correlated between the zones 2 and 3 (RS = 0·943; n = 6; P < 0·05), but variable among all other pairings.

In spring 1995, the alien cynipid species were present at abundances comparable to those of the most abundant native cynipids, and A. quercuscalicis was the most abundant species in zones 1 and 2 (Table 3). N. quercusbaccarum was most abundant in zone 3, while another invading species, A. kollari, was the most abundant cynipid in zone 4 (Table 3).

Dominance patterns between zones in spring were more variable than had been in the previous autumn. There were no significant correlations of ranked abundance between any of the zones.

Fully quantified linkage webs

The 16 linkage webs (8 sites; 2 generations) in Figs 2 and 3 show the quantified associations between inquilines, parasitoids and their host galls for the autumn generation in 1994 and the spring generation 1995. Only those cynipid species, from which parasitoids or inquilines emerged, and which were above detectable densities in our random sampling, are included in the webs.

Figure 2.

Figure 2.

Quantified linkage webs in autumn 1994 for eight sample sites: (a) Silwood, (b) Puttenham, (c) Tatton, (d) Rufford, (e) Erskin, (f) Falkland, (g) Beauly, (h) Dunrobin; see Fig. 1. See the Key for the interpretation of the quantified measures. Where a link is not represented by an area on the proportional bar, the contribution made is < 0·05. The names of the invading cynipid species are printed in bold face.

Figure 2.

Figure 2.

Quantified linkage webs in autumn 1994 for eight sample sites: (a) Silwood, (b) Puttenham, (c) Tatton, (d) Rufford, (e) Erskin, (f) Falkland, (g) Beauly, (h) Dunrobin; see Fig. 1. See the Key for the interpretation of the quantified measures. Where a link is not represented by an area on the proportional bar, the contribution made is < 0·05. The names of the invading cynipid species are printed in bold face.

Figure 2.

Figure 2.

Quantified linkage webs in autumn 1994 for eight sample sites: (a) Silwood, (b) Puttenham, (c) Tatton, (d) Rufford, (e) Erskin, (f) Falkland, (g) Beauly, (h) Dunrobin; see Fig. 1. See the Key for the interpretation of the quantified measures. Where a link is not represented by an area on the proportional bar, the contribution made is < 0·05. The names of the invading cynipid species are printed in bold face.

Figure 2.

Figure 2.

Quantified linkage webs in autumn 1994 for eight sample sites: (a) Silwood, (b) Puttenham, (c) Tatton, (d) Rufford, (e) Erskin, (f) Falkland, (g) Beauly, (h) Dunrobin; see Fig. 1. See the Key for the interpretation of the quantified measures. Where a link is not represented by an area on the proportional bar, the contribution made is < 0·05. The names of the invading cynipid species are printed in bold face.

Figure 2.

Figure 2.

Quantified linkage webs in autumn 1994 for eight sample sites: (a) Silwood, (b) Puttenham, (c) Tatton, (d) Rufford, (e) Erskin, (f) Falkland, (g) Beauly, (h) Dunrobin; see Fig. 1. See the Key for the interpretation of the quantified measures. Where a link is not represented by an area on the proportional bar, the contribution made is < 0·05. The names of the invading cynipid species are printed in bold face.

Figure 2.

Figure 2.

Quantified linkage webs in autumn 1994 for eight sample sites: (a) Silwood, (b) Puttenham, (c) Tatton, (d) Rufford, (e) Erskin, (f) Falkland, (g) Beauly, (h) Dunrobin; see Fig. 1. See the Key for the interpretation of the quantified measures. Where a link is not represented by an area on the proportional bar, the contribution made is < 0·05. The names of the invading cynipid species are printed in bold face.

Figure 2.

Figure 2.

Quantified linkage webs in autumn 1994 for eight sample sites: (a) Silwood, (b) Puttenham, (c) Tatton, (d) Rufford, (e) Erskin, (f) Falkland, (g) Beauly, (h) Dunrobin; see Fig. 1. See the Key for the interpretation of the quantified measures. Where a link is not represented by an area on the proportional bar, the contribution made is < 0·05. The names of the invading cynipid species are printed in bold face.

Figure 2.

Figure 2.

Quantified linkage webs in autumn 1994 for eight sample sites: (a) Silwood, (b) Puttenham, (c) Tatton, (d) Rufford, (e) Erskin, (f) Falkland, (g) Beauly, (h) Dunrobin; see Fig. 1. See the Key for the interpretation of the quantified measures. Where a link is not represented by an area on the proportional bar, the contribution made is < 0·05. The names of the invading cynipid species are printed in bold face.

Figure 3.

Figure 3.

Quantified linkage webs in spring 1995 for eight sample sites: (a) Silwood, (b) Puttenham, (c) Tatton, (d) Rufford, (e) Erskin, (f) Falkland, (g) Beauly, (h) Dunrobin; see Fig. 1. See the Key for the interpretation of the quantified measures. Where a link is not represented by an area on the proportional bar, the contribution made is < 0·05. The names of the invading cynipid species are printed in bold face.

Figure 3.

Figure 3.

Quantified linkage webs in spring 1995 for eight sample sites: (a) Silwood, (b) Puttenham, (c) Tatton, (d) Rufford, (e) Erskin, (f) Falkland, (g) Beauly, (h) Dunrobin; see Fig. 1. See the Key for the interpretation of the quantified measures. Where a link is not represented by an area on the proportional bar, the contribution made is < 0·05. The names of the invading cynipid species are printed in bold face.

Figure 3.

Figure 3.

Quantified linkage webs in spring 1995 for eight sample sites: (a) Silwood, (b) Puttenham, (c) Tatton, (d) Rufford, (e) Erskin, (f) Falkland, (g) Beauly, (h) Dunrobin; see Fig. 1. See the Key for the interpretation of the quantified measures. Where a link is not represented by an area on the proportional bar, the contribution made is < 0·05. The names of the invading cynipid species are printed in bold face.

Figure 3.

Figure 3.

Quantified linkage webs in spring 1995 for eight sample sites: (a) Silwood, (b) Puttenham, (c) Tatton, (d) Rufford, (e) Erskin, (f) Falkland, (g) Beauly, (h) Dunrobin; see Fig. 1. See the Key for the interpretation of the quantified measures. Where a link is not represented by an area on the proportional bar, the contribution made is < 0·05. The names of the invading cynipid species are printed in bold face.

Figure 3.

Figure 3.

Quantified linkage webs in spring 1995 for eight sample sites: (a) Silwood, (b) Puttenham, (c) Tatton, (d) Rufford, (e) Erskin, (f) Falkland, (g) Beauly, (h) Dunrobin; see Fig. 1. See the Key for the interpretation of the quantified measures. Where a link is not represented by an area on the proportional bar, the contribution made is < 0·05. The names of the invading cynipid species are printed in bold face.

Figure 3.

Figure 3.

Quantified linkage webs in spring 1995 for eight sample sites: (a) Silwood, (b) Puttenham, (c) Tatton, (d) Rufford, (e) Erskin, (f) Falkland, (g) Beauly, (h) Dunrobin; see Fig. 1. See the Key for the interpretation of the quantified measures. Where a link is not represented by an area on the proportional bar, the contribution made is < 0·05. The names of the invading cynipid species are printed in bold face.

Figure 3.

Figure 3.

Quantified linkage webs in spring 1995 for eight sample sites: (a) Silwood, (b) Puttenham, (c) Tatton, (d) Rufford, (e) Erskin, (f) Falkland, (g) Beauly, (h) Dunrobin; see Fig. 1. See the Key for the interpretation of the quantified measures. Where a link is not represented by an area on the proportional bar, the contribution made is < 0·05. The names of the invading cynipid species are printed in bold face.

Figure 3.

Figure 3.

Quantified linkage webs in spring 1995 for eight sample sites: (a) Silwood, (b) Puttenham, (c) Tatton, (d) Rufford, (e) Erskin, (f) Falkland, (g) Beauly, (h) Dunrobin; see Fig. 1. See the Key for the interpretation of the quantified measures. Where a link is not represented by an area on the proportional bar, the contribution made is < 0·05. The names of the invading cynipid species are printed in bold face.

The autumn generations

The most conspicuous and long-established of the alien galls, Andricus kollari, was below detectable densities at most sites. Rearing the agamic galls showed that all the alien species were attacked by parasitoid and inquiline species, but only three species of inquilines emerged from galls of A. lignicola and then only in small numbers. These galls contributed very little to the total parasitoid and inquiline populations at any site. In contrast, the agamic galls of A. quercuscalicis were a focal point for parasitoid and inquiline species at three of the four sites where this species was present (Fig. 2a,b,d). Inquiline species, such as Synergus gallaepomiformis de Fonscolombe, emerged almost exclusively from these galls, along with parasitoid species like Mesopolobus sericeus (Förster) and Eurytoma brunniventris Ratzeburg. These species were also relatively abundant, although the agamic galls of A. quercuscalicis were not the most abundant galls at any of the four sites. Only at Tatton Park did these parasitoids and inquilines emerge also in high numbers from other galls, particularly Cynips divisa Hartig (Fig. 2c). Of all the native species, the agamic galls of C. divisa supported the most species-rich parasitoid and inquiline communities.

The linkage webs for the northern sites (Fig. 2e–h) are much simpler than those from the south (Fig. 2a–d), which is mainly due to a decrease in abundance as indicated for the species listed in Table 3. No parasitoids or inquilines emerged from the galls of some species collected at the sites in zone 3 and zone 4 (Fig. 2e–h). Since sample sizes for all cynipid species shown in the webs were standardized, this also suggests fewer incidences of parasitoid and inquiline attack.

The spring generations

Rearings of the small spring galls of the four invading cynipid species yielded only parasitoid species and no inquilines. Of the four parasitoid species, Mesopolobus dubius (Walker), M. fuscipes (Walker), M. xanthocerus (Thomson) and M. tibialis (Westwood), which attacked the alien cynipid species in the southern sites, only the latter was reared from native cynipid species, such as Neuroterus quercusbaccarum and Andricus curvator Hartig (Fig. 3a–d). Otherwise the parasitoid communities attacking galls of the invading cynipid species on Quercus cerris and on Q. robur were completely separate. At the four northern sites only M. xanthocerus and M. fuscipes were recorded from the alien galls, and there was no overlap between the parasitoid species attacking galls on Q. cerris and Q. robur (Fig. 3e–h).

At all eight sites two or more of a set of four native cynipid species were recorded during the random sampling: Andricus curvator, Biorhiza pallida, Neuroterus numismalis and N. quercusbaccarum. All other species shown in Table 1 were very rare. The largest number of parasitoid and inquiline species was associated with the galls of Biorhiza pallida, where this species was present. Where B. pallida was too rare to be detected during the random sampling, either Neuroterus quercusbaccarum or Andricus curvator supported the most species-rich, local parasitoid and inquiline communities. The galls of the spring generations of A. curvator and N. quercusbaccarum are both globular structures on the leaves with a diameter of about 1 cm. Interestingly, depending on which of the two species was more abundant at the sampling sites, it also supported the larger parasitoid and inquiline community (Fig. 3a–c,e–h). At Tatton and Rufford (Fig. 3c,d), where both species were equally abundant (Table 3), both species also have the same (± 1 species) that attack their galls.

Summaries from the webs and indirect interactions

We selected a number of parameters taken from the quantified linkage webs in Figs 1 and 2 that illustrate the relationship between native and invading cynipid species in the eight local communities (Table 4). In autumn 1994, the number of alien galls was less than 10% of all galls that were recorded. In contrast, the spring generations of the invaders were more abundant than the native species and reached at Puttenham and Dunrobin more than 80% of all the galls at those sites. The number of links per species in the autumn generation were higher for the alien galls at Puttenham and Silwood, which was mainly due to the large number of parasitoid and inquiline species emerging from the galls of A. quercuscalicis (Fig. 2a,b). Further north and at all sites during spring the linkage of the alien galls was lower than that of the native species. The links per species squared showed very little variation among sites in autumn (range: 0·04–0·08) and similarly in spring (range: 0·06–0·1; Table 4). However, since we did not dissect galls to identify the trophic links within them, these values do not represent the connectance of the food webs, which could be considerably higher.

Table 4.  Summary of parameters taken from the linkage webs (Fig. 1a–h; 2a–h)
Generation
Figure
Autumn 94
1a
1b1c1d1e1f1gSpring 95
1 h
2a2b2c2d2e2f2g2 h
  1. Li = links; P. + I. sp. = parasitoid and inquiline species; aliens = number of invading species present at the site above detectable densities; natives = number of native species present at that site; Σ= number of individuals/100 m2.

No. of species182517201077131821171812131717
No. of links20321725723122026202113162023
ΣNatives171161474812294110200256230000033114014162700396594519145000180001040000340005640
ΣAliens41499711582722240574865054716103738600012000700002210030000
%Aliens2·46·38·62·40·080·0010·00·474·144·078·189·040·06·339·484·1
Li/native sp.1·51·33·254·52·00·50·43·54·54·04·675·55·04·346·04·4
Li/alien sp.5·77·02·03·51·00·00·05·02·82·52·02·51·51·51·01·0
Links/species20·060·050·060·060·070·040·060·080·060·060·060·060·080·10·070·10
P. sp.10148932151011887599
I. sp.343532423342332
Li/P. + I. sp.1·541·781·551·791·21·01·01·31·671·861·671·751·442·01·672·09
P. + I. sp. att. alien sp.11153610443442211
P. + I. sp. shared between
aliens + natives00520000110001
ΣP. + I. from aliens171314912387871011284377257130712700149001290011014191
ΣP. + I. from natives553729299849572461074451920205224316083121063500362000192041000

The number of links per parasitoid or inquiline species suggest some degree of polyphagy, which had a trend to decrease from the south to the north in the autumn, while there was no such trend for the communities in spring (Table 4). Again, the sites Puttenham and Silwood showed an exceptionally large number of parasitoid and inquiline species that attacked the galls of the invading species and the vast majority of those were recorded from the agamic galls of A. quercuscalicis (Fig. 2a,b). Possibly most striking is the very small number of parasitoid and inquiline species that were shared, within each of the two generations, between alien and native galls (Table 4). The largest number of shared parasitoid and inquiline species was five recorded at Tatton. The number of parasitoids and inquilines emerging from alien or native galls in autumn largely reflect the prevalence of the agamic galls of A. quercuscalicis as a focus for parasitoid attack, and the dominance structure in the abundance patterns between invading and native species in the spring generations (Table 4).

Twenty species pairs of cynipid galls in autumn and in spring that shared one or more parasitoid or inquiline species could be identified (Table 5). Only at Falkland, none of the autumn galls shared any parasitoid or inquiline species with the spring galls (Fig. 2f). Although most of the invading cynipid species and some of the native species, such as N. quercusbaccarum and N. numismalis, were present in both generations, on only one occasion were the autumn and the spring galls of the same cynipid species, A. corruptrix, attacked by the same parasitoid species, Mesopolobus dubius. Thirty-eight cases of shared parasitism or inquilinism were identified at all eight sample sites. The dij-values for of these pairwise, indirect interactions were never > 0·5, only three values were > 0·4, and 22 of the 38 values were < 0·1 (Table 5). Furthermore, the links and Figs 1 and 2 represent associations of parasitoid and inquiline species with the cynipid galls, rather than trophic links. It is therefore possible that some of the interactions recorded here do not affect the dynamics of the cynipid host, for instance where a parasitoid attacks an inquiline that is not lethal to the cynipid gall former. Where this situation occurs, however, the dij-values in Table 5 are overestimates of the interaction between the two cynipid species. This strongly suggests that there are no strong interactions via parasitoids or inquilines between the cynipid species recorded here at any of the eight locations.

Table 5. . dij-values for the indirect interactions among local cynipid populations between generations. The table includes only the galls of those cynipid species, which community share at least one parasitoid or inquiline species with a gall in the next generation. No parasitoid or inquiline species was shared between the autumn and spring galls collected at Falkland. The names of the invading cynipid species are printed in bold face
SiteAutumn gallsSpring galls   
Silwood
A. corruptrix
A. lignicola
A. quercuscalicis
A. curvator
0·2222

0·0630
B. pallida
0·0344
0·2840
0·1091
  
Puttenham A. corruptrixA. curvatorB. pallidaN. quercusbaccarum
A. corruptrix0·01140·00110·01490·0327
A. quercuscalicis0·00150·00020·3586
N. numismalis0·07450·19570·1957
N. quercusbaccarum0·4348
Tatton A.curvatorB. pallidaN. quercusbaccarum 
A. lignicola0·1415 
A. quercuscalicis0·0228 
C. divisa0·03920·01330·0615 
Rufford A. curvatorN. quercusbaccarum  
A. anthracinus0·32520·0012  
A. quercuscalicis0·00070·0180  
C. divisa0·05040·0329  
Erskin A. curvatorN. quercusbaccarum  
A. lignicola0·0856  
C. divisa0·05080·1795  
Beauly A. curvator   
C. divisa0·4902   
Dunrobin A. curvatorA. quadrilineatusN. quercusbaccarum 
C. divisa0·01740·27080·0700 
N. numismalis0·47000·28000·2900 

Discussion

Food webs are depictions of the trophic relationships within a community and as such a tool to identify the potential for indirect relationships between members of the community. A wealth of theory about food web structures, such as food-chain length or connectance, has developed, although more recently authors have called for more ‘purpose-built’ webs, collected with a common set of techniques to allow further progress to be made in this field (Pimm 1982; Cohen, Briand & Newman 1990; Pimm, Lawton & Cohen 1991; Cohen et al. 1993; Laska & Wootton 1998). Since then published webs tended to become more inclusive (Hawkins, Martinez & Gilbert 1997). However, another feature called for, to quantify trophic relationships, has not been answered readily. Although the theory and statistics to analyse interaction strength in quantified food webs is developing, we know of only a few semi-quantified and two fully quantified webs that involve communities of herbivorous insects and their associated parasitoids (Askew 1961; Goldwasser & Roughgarden 1993; Memmot et al. 1994; Wootton 1997; Laska & Wootton 1998; Müller et al. 1999).

Assessing the impact of alien gall wasps on the native cynipid communities in britain

The communities of cynipid gall wasps in Britain, native and alien, have been the subject of extensive ecological studies and we have a reasonable understanding of the biology of all species involved (Askew 1961; Schönrogge, Stone & Crawley 1996; Schönrogge, Walker & Crawley 1998, 1999, 2000). Most of the parasitoid species in the system are idiobionts and have been described as polyphagous and facultative hyper- or autoparasitoids (Askew 1961, 1975). Similarly, inquiline species are known to develop in the galls of a range of different cynipid species, where their presence is often lethal to the gall former (Askew 1961). Thus, the biology of both parasitoid and inquiline species would suggest that once the invading species have recruited parasitoids and inquilines these would be the main links for interactions with native cynipid species. Direct interactions, such as intra- or interspecific competition, are believed to be weak among cynipids, because of their ability to control and shape the development of the gall tissue from omnipotent tissue anywhere in the tree, and we know of only one example that has been reported (Gilbert et al. 1994; Schönrogge, Harper & Lichtenstein 2000). This is in contrast to aphid or sawfly galls, where modification of the auxin transport in the plant is associated with changes in rates of cell-proliferation and tissue physiology, and thus sites of auxin synthesis in the plant become valuable resources to compete for (Whitham 1980; Leitch 1994).

Many of the parasitoid and inquiline species emerged from a range of host galls within and among sites during our study, which would be the information contained in a non-quantified web. However, we found that local populations were much more specialized than the non-quantified data would suggest. If a parasitoid or inquiline species was shared between host galls, more than 90% emerged usually from one gall and often less than 5% from the alternative host gall although both host galls were abundant. Thus, the potential for apparent competition is very much diminished. This was also reflected in the dij-values calculated for links between generations. All the values were found to be < 0·5 and 22 out of 38 were < 0·1 suggesting a high degree of specialization by parasitoid and inquiline species (no shared parasitoids: dij = 0; the majority of parasitoids emerging from species j attack species i: dij→ 1; see also Müller et al. 1999). The linkage webs presented suggest that the invading species have little effect on native cynipid species through shared parasitoids or inquilines, nor that there are strong indirect interactions between native species. However, it is important to remember that the number of cynipid species included in the webs was generally one-third of the species pools recorded for the sites. Two-thirds of the cynipid species were too rare to be detected and it is possible that the rarer species are most affected by small increases in parasitoid pressure.

Many parasitoid species appear to have shifted their attack to the invading species. This is particularly true for those attacking the sexual generations of the alien species (Schönrogge et al. 2000). Further affects observed in parasitoid populations of species associated with the invading galls included a strong male bias in species emerging from the sexual galls of A. quercuscalicis, and apparent bottom up controlled dynamics by species attacking A. kollari in the north of Scotland where parasitoid attack rates were inversely correlated with gall density (Schönrogge et al. 1999).

Müller et al. (1999) listed seven potential sources of bias in the collection of quantified food web data, most of which apply to the webs presented here as well. Here we want to emphasize two that we believe to be particularly important in the system of cynipids and their parasitoids.

Sampling efficiency

To estimate absolute gall densities 2160 shoots taken from 24 trees were examined for galls per site and generation. During these surveys we recorded between four and seven cynipid species (Figs 1 and 2). However, during 4 man-hour searches of the sites we recorded between 15 and 20 cynipid species (Table 1), which underlines the trade-off between quantification and the inclusiveness of food webs that was pointed out by Müller et al. (1999). However, in the context of the assessment of indirect interactions between the invading species and native cynipids, it also means that we missed particularly rarer species where the asymmetric interactions could be strongest.

Resolution

As we mentioned before, the webs presented here are linkage webs rather than food webs, and the links represent associations of parasitoid and inquiline species with the galls rather than trophic links. We therefore refrained from referring to parameters such as connectance in the webs, since the trophic interactions within the galls can be of considerable complexity (Askew 1961; Schönrogge et al. 1995). Also, the number of links per parasitoid species must be an underestimate if they attack different hosts in the same host gall. Most importantly, the dij-values are overestimates, because we treated all parasitoids and inquilines emerging from a host galls as if they all attacked the cynipid larvae. We found indirect interactions between host galls to be weak and thus those between invading cynipid species and natives are likely to be even weaker. However, where strong interactions are indicated, functional species or pooled higher taxonomic units as used in some food webs will need to be resolved to avoid overestimating interactions.

The quantification of webs can reveal characteristics of interactions among species that are otherwise not apparent. If it were possible to collect sufficiently long time series it would be possible to quantify interaction strength between all member species or alternatively, as here, to concentrate on indirect interactions between hosts (Laska & Wootton 1998; Müller et al. 1999). For assessing the impact of biological invasions on native communities, quantified webs could potentially be a valuable tool, but the information gained will be directly proportional to the effort spent on the collection of data.

Acknowledgements

This study would have not been possible without the help of numerous people and Institutions. We would like to thank particularly R.R. Askew and J.L. Nieves Aldrey for their help with the identification of parasitoids and inquilines, and G.N. Stone, C.B. Müller, H.C.J. Godfray and A. Hildrew for their much appreciated comments and discussions. This study was funded by the Natural Environmental Research Council.

Received 3 December 1999;revisionreceived 17 March 2000

Ancillary

Advertisement