Impact of the arrival of Harmonía axyridis on adults of Coccinella septempunctata and Coleomegilla maculata ( Coleóptera : Coccinellidae )

The impact of the arrival of the exotic coccinellid Harmonia axyridis on the biological control and dispersion of apple mites and aphids, and on the vertical distribution of the established coccinellids Coccinella septempunctata and Coleomegilla macu­ lata have been studied. Assemblages of adults of the three coccinellid species were tested on apple saplings and on apple trees against Aphis citricola and Tetranychus urticae. The presence or substitution of adults of H. axyridis for adults of C. maculata or C. septempunctata did not reduce the predation impact on mite or aphid prey. H. axyridis was a more efficient predator of mites than was C. septempunctata and generated lower prey dispersal from host plants. Substitutions of adults of H. axyridis for those of C. maculata, and in some cases of C. septempunctata, resulted in greater reduction in number of aphids. On apple trees, the three cocci­ nellid species occurred at different heights. Neither C. maculata nor C. septempunctata modified their vertical distribution on apple tree over a 24 hour period in response to the presence of H. axyridis.

The arrival of a new coccinellid species may have important consequences on the guild of established spe cies.Both intruder and local coccinellids can interact directly or indirectly and may be implicated in intraguild predation (Agarwala & Dixon, 1992;Hironori & Katsuhiro, 1997;Lucas et al., 1998a;Dixon, 2000), or in exploitative competition (Evans, 1991).In alfalfa, maize and small grain crops, the establishment of Coccinella septempunctata L. was associated with significant changes in the local guilds of coccinellids (Elliot et al., 1996).The populations of Coccinella transversoguttata richardsoni Brown and Adalia bipunctata (L.) were dras tically reduced (more than 20 times less after invasion).The establishment of H. axyridis in natural and agricul tural ecosystems was also associated with significant population reductions of Brachiacantha ursine (F.), Cycloneda munda (Say) and Chilocorus stigma (Say) (Colunga-Garcia & Gage, 1998).
Such modifications in the dynamics and composition of the coccinellid guilds may also have significant impact on the biological control of prey.Previous experiments on larvae showed that interactions between C. septempunc tata and Coleomegilla maculata (DeGeer) did not signifi cantly affect the control of Myzus persicae (Sulzer) (Obrycki et al., 1998).However, no such studies have been conducted on adults.Adults have an important role in the outcome of competition as they are the only flying stage and will select the habitat of their offspring by oviposition.Adults are active predators and their contribu tion may be maximum early in the season when large larvae are absent.Moreover, no intraguild predation among adult coccinellids has been reported.
The aim of this study is to evaluate the potential impact of the arrival of H. axyridis adults on the predation effi cacy by adults of two local aphidophagous coccinellids, C. septempunctata and C. maculata.Three questions were formulated: 1) According to their different voraci ties and to the possibility of interference between H. axy ridis and C. maculata or C. septempunctata, may the overall impact on pest species be affected?, 2) May the arrival of H. axyridis modify pest dispersion generated by local coccinellid species?, 3) May H. axyridis displace spatially other coccinellid species by modifying their ver tical distribution?

MATERIAL AND METHODS
Coccinella septempunctata adults were collected in an apple orchard near Montreal (73°36'W, 45°30'N) (Quebec, Canada).Harmonia axyridis adults (population of Chinese origin) and Coleomegilla maculata came from mass rearing on Ephestia kuehniella Zeller eggs (Lepidoptera: Pyralidae) and an artificial diet of pork liver and ground pollen.All coccinellids were reared on the spirea aphid, Aphis citricola van der Goot, at 25°C, 70% R.H. and a 16L : 8D photoperiod, one month before tests.The two-spotted spider mites, Tetranychus urticae Koch, were mass reared on Lima bean plants, Phaseolus limensis.Aphis citricola were collected from apple orchards near Mont real.

Prey consumption and dispersion effect on apple saplings
The first experiment was carried out on four-to six-week old apple saplings (McIntosh cultivar) grown in a greenhouse and carefully washed to clear all arthropods before the experiments.The saplings with five equal-sized and undamaged leaves were fixed with Plasticine in a hole pierced in the bottom of a plastic container (8,1 cm high with a 15,0 cm diameter).This container was placed in a second plastic container with water in which soaked the sapling roots.The adult coccinellids were fed ad libitum and then starved for 24 h before the experiment.Adult mites (n = 60) or 3rd instar nymphs of the aphid (n = 100) were placed on the apple sapling leaves.After 1 h, predators were added on the plant.The set up was hermetically sealed using a transparent tightened plastic film placed on the top of first con tainer and held by the plastic cover pierced at its center.The set up was placed in an incubator (25°C, 70% RH, 16L : 8D).Two sets of treatments were carried out, one with aphids and one with spider mites.The treatments were as follows: 1) two adults of C. septempunctata, 2) two adults of H. axyridis, and 3) one adult of C. septempunctata and one of H. axyridis.All treat ments were replicated 28 times.After 24 h, uneaten prey on and out of the sapling tree were counted with a binocular micro scope.The number of prey eaten as well as the proportion of dispersed prey (surviving individuals that had left the plant compared to the total of the survivors) were then compared using a one-way ANOVA followed by an LSD Fisher's pro tected post-hoc test (Sokal & Rohlf, 1981).Controls without predators were included to evaluate natural prey mortality.Data were adjusted accordingly.All analyses were performed with SuperAnova ® (Abacus Concepts, 1989).
In order to evaluate heterospecific treatment results, an expected consumption value (E) was calculated for each repli cate using the number of prey consumed in the monospecific treatments: E = (C + H)/2 where: C = number of prey consumed by 2 adults of C. septempunc tata, H = number of prey consumed by 2 adults of H. axyridis.
The expected mean number of prey consumed in the hetero specific treatment was then compared to the mean of observed prey consumed in the heterospecific treatment by a Wilcoxon-Mann Whitney test.Non parametrical statistics were used because of the unequal variances in the treatments.The same formula and test were used to compare the observed and the expected proportion of prey that left the plant.Analyses was performed with JMP IN ® (Sall & Lehman, 1996).

Prey consumption and vertical distribution on apple trees
The second experiment was done on potted apple trees in con trolled conditions (23°C, 70% RH, 24L : 0D).Dwarf apple trees (180 cm) were caged with a closed muslin.Nine aphid colonies were established at three heights on the tree, low (> 100 cm), medium (110-130 cm) and high (>170 cm) height on the tree.At each height, three aphid colonies of 10 4th instar nymphs of A. citricola were established.In this experiment, adults of C. maculata were included with those of H. axyridis and/or C. sep tempunctata in the experiment design.After 24 h, starved cocci nellids were introduced in the system, at the base of the tree.Seven treatments were considered: 1) 6 adults of C. maculata, 2) 6 adults of C. septempunctata, 3) 6 adults of H. axyridis, 4) 3 adults of C. maculata + 3 adults of C. septempunctata, 5) 3 adults of C. maculata + 3 adults of H. axyridis, 6) 3 adults of C. septempunctata + 3 adults of H. axyridis, and 7) 2 adults of C. maculata + 2 adults of C. septempunctata + 2 adults of H. axy ridis.Fifteen replicates were carried out.Control trees without predators were included and data were adjusted accordingly.After 24 h, the number of living aphids at the different heights and the height position of coccinellids on the tree were recorded.The number of predated aphids was compared on the entire tree (all levels pooled), and then for each vertical level with a one-way ANOVA followed by a LSD Fisher's protected post-hoc test.The mean height position of each coccinellid spe cies was compared among the three monospecific treatments also by using a one-way ANOVA.The height position of each species was compared also by an ANOVA in monospecific and multispecific assemblages to evaluate whether the presence of another coccinellid species may result in modification of a par ticular species height position.

Prey consumption and dispersion effect on apple saplings
Two adults of C. septempunctata together consumed significantly fewer mites in 24h tests than did two adults of H. axyridis or two adults of the heterospecific treat ment (F2,s3 = 4.54, P = 0.013) (Fig. 1).Adults in the het erospecific treatment (22.8 mites) consumed significantly more mites than expected (20.9 mites) (Wilcoxon, Z = 2.19, P = 0.029).H. axyridis adults also consumed sig nificantly more aphids than did C. septempunctata adults (F2.86 = 5.89, P = 0.004).Adults in the heterospecific treatment (30.3 aphids) consumed similar number of aphids as expected (30.9 aphids) (Wilcoxon, Z = 0.38, P = 0.702).After 24 h, a mean of 26% of the mites had left the sap ling in the control treatment without predators.A greater percentage of mites left the sapling in the presence of C. septempunctata adults (75.7%) than in the presence of H. axyridis (54.0%) (F2.s3 = 3.32, P = 0.041) (Fig. 2).No dif ference was recorded between observed (66.8%) and expected (64.3%) mean proportion of mites that left the plant in the heterospecific treatment (Wilcoxon, Z = 0.38, P = 0.702).After 24 h, only 5.4% of the aphids had left the plant in the control treatment.In the presence of adult coccinellids, the percentage of aphids that left the apple sapling was similar among the different treatments (F2J4 = 0.01, P = 0.993).As with mites, expected (15.4%) and observed (18.2%) percentages of aphids leaving the plant  for the heterospecific treatment were similar (Wilcoxon, Z = 0.33, P = 0.735).
Considering the entire apple tree, in monospecific treat ments, C. septempunctata and H. axyridis consumed more aphids than C. maculata (F6,i03 = 22.68, P < 0.001) (Fig. 3).The presence of H. axyridis or C. septempunc tata in the same tree as C. maculata significantly increased the number of aphids consumed by the guild compared to the trees where C. maculata occurred alone.Considering each vertical level, this pattern was observed in high (F6j 03 = 13.21,P < 0.001) and medium height (F6,i03 = 7.85, P < 0.001) in the tree (Fig. 4).No differ ence was observed at the lower height (F6j 03 = 7.38, P < 0.001; LSD P > 0.05) (Fig. 4).At all levels, the presence of both C. septempunctata and H. axyridis in the tree did not significantly affect the number of aphids consumed, compared to trees where only C. septempunctata or only H. axyridis occurred.Finally the trispecific combination generated a stronger impact on aphid populations than the C. maculata treatment at each height.

DISCUSSION
We draw three primary conclusions from our experi ments.First, the replacement of C. maculata adults or C. septempunctata adults by H. axyridis adults did not reduce the impact of the coccinellids on spider mites or on aphids.Second, mite dispersal from plants was lower in the presence of H. axyridis than in the presence of C. septempunctata.Third, in our short-term experiment, nei ther C. maculata nor C. septempunctata modified their vertical distribution on apple trees in response to the pres ence of H. axyridis.
In our study, adults of H. axyridis or adults of hetero specific combinations consumed more mites than did C. septempunctata adults.Furthermore, more mites were consumed by adults in heterospecific combination than expected, suggesting that some kind of facilitation or stimulation may have occurred.In apple orchards, several mite species including the two spotted spider mites, T. urticae, and the European red mites, Panonychus ulmi (Koch), can become important pests (Van de Vrie, 1985).Among large coccinellids, several species attack these mites but, except for C. maculata, mites appear to consti tute a low-quality prey (Putman, 1957;Lucas et al., 1997bLucas et al., , 1998b)).Previous studies on C. septempunctata and H. axyridis have shown that both species can con sume mites but have a significant preference for aphids (Lucas et al., 1997b).Furthermore, despite its large size, C. septempunctata had a very low voracity confirming its inefficiency as a biological control agent against T. urticae (Lucas et al., 1997b) and P. ulmi (Lucas et al., 1998b).Thus, the arrival of H. axyridis may possibly improve the control of mites but only during periods of aphid scarcity.
Our results also revealed that replacing C. maculata adults with H. axyridis adults improved the impact of coccinellids on aphid populations in apple trees, and that replacing C. septempunctata by H. axyridis improved or maintained their impact.These results were consistent among the different heights in the tree.The overall voracity of the coccinellid assemblage was probably enhanced when H. axyridis was included because of the greater voracity of this species.Moreover, the number of aphids consumed by adults in heterospecific combination was as expected, suggesting that no unusually strong exploitative interspecific competition (i.e., superior to intraspecific competition between two individuals) occurred between C. septempunctata and H. axyridis.The three coccinellid species are primarily aphidophagous and prey on numerous aphid species in natural and agricul tural systems (Blackman, 1965 ;Hodek, 1973;Angalet et al., 1979 ;Lucas, 1994).Thus, because of their preference for aphids which constitute their essential food source (allowing egg maturation and complete development), the arrival of H. axyridis may have a greater impact on aphid than on mite control.
Adults of C. septempunctata caused greater mite dis persal from the plants than did H. axyridis adults.Aphids responded similarly to both predators.Prey dispersal as a defensive response to predation has been poorly studied but may be important from a biological control perspec tive (Roitberg & Myers, 1978, 1979;Roitberg et al., 1979;Evans, 1991).One should believe that a predator with a low capture success and a high disturbing effect may generate a strong dispersal response effect of the prey without significantly reducing prey number.Since C. septempunctata has a low efficacy on mites, its overall effect may be to propagate the infestation in other parts of the plant or of the crop.Furthermore, by dispersing prey, the predator may reduce the local densities of mites such that the consequent numerical and functional response of other predators could be delayed or weaken.In our situa tion, H. axyridis consumed more mites than did C. sep tempunctata and generated less dispersal by the mites, indicating greater potential of this beetle against this prey.These results should be confirmed in field or semi-field experiments.
In our short-term study, the mean vertical position of H. axyridis in the tree was different from those of the other species.Such different vertical positions may reduce potential interaction; in addition, in the fields the three species exploit a considerable array of different habitats (Angalet et al., 1979;Mareida et al., 1992;LaMana & Miller, 1996;Colunga-Garcia & Gage, 1998;Obrycki et al., 1998).According to our results and also to the fact that C. maculata is mainly encountered in the herbaceous stratum (Coderre & Tourneur, 1986), competition between H. axyridis and C. maculata should be less important than between H. axyridis and C. septempunc tata.In maize, while some other indigenous species such as Hippodamia tredecimpunctata tibialis (Say) were severely affected, C. maculata has remained a predomi nant species despite the arrival of other coccinellids such as, Propylea quatuordecimpunctata (L.) and C. septempunctata (Coderre unpubl.).In our study, neither C. septempunctata, nor C. maculata modified their vertical position in the presence of H. axyridis; however a long term study in the field may generate a different response.For example, according to Iablokoff-Khnzorian (1982), Adalia bipunctata (L.) in Siberia avoid plants where H. axyridis is present, mainly Salix sp.
This study is an initial evaluation of the impact of H. axyridis on the aphidophagous and acarophagous guilds and on biological control of aphids and mites in apple orchards.It would be risky to speculate on the potential long-term impact of H. axyridis in the field.However, H. axyridis is eurytopic and polyphagous and may be impli cated in exploitative or interference competition as well as in intraguild predation with local predatory species (Evans, 1991;Agarwala & Dixon, 1992;Hironori & Katsuhiro, 1997;Lucas et al., 1997aLucas et al., , 1998a;;Dixon, 2000).In the field, its arrival in natural and agricultural ecosystems is associated with significant reductions of several coccinellid species populations (Colunga-Garcia & Gage, 1998).To obtain an overall pattern of the effect of the arrival of H. axyridis on the local guild of coccinellids and consequently on their prey, the next step would be to evaluate the impact on immature stages of local ladybeetles (selection of oviposition sites, selection of pupating and molting sites, interactions among larvae, etc.) and to carry out in situ studies in different agricultural and natural systems.

Fig. 1 .
Fig. 1.Mean number of prey consumed by adult coccinellids in 24 h on apple saplings.C. 7 = Coccinella septempunctata, H. ax.= Harmonia axyridis.Different letters indicate a significant difference among the different combinations of coccinellids for the same prey (P < 0.05).

Fig. 2 .
Fig. 2. Mean percentage of prey that had left the apple sapling after 24 h.C. 7 = Coccinella septempunctata, H. ax.= Har monia axyridis.Different letters indicate a significant difference among the different combinations of coccinellids for the same prey (P < 0.05).

Fig. 3 .
Fig. 3. Mean number of prey consumed by adult coccinellids in 24 h on the entire apple tree, (all vertical levels pooled).C. 7 = Coccinella septempunctata, H. ax.= Harmonia axyridis, C. mac.= Coleomegilla maculata.For details about the treatments, see Material and Methods section.Different letters indicate a significant difference among the different combinations of coccinellids (P < 0.05).

Fig. 4 .
Fig. 4. Mean number of prey consumed by adult coccmellids in 24 h on apple tree, at upper, medium and lower level.C. 7 = Coccinella septempunctata, H. ax.= Harmonia axyridis, C. mac.= Coleomegilla maculata.For details about the treatments, see Mate rial and Methods section.Different letters indicate a significant difference among the different combinations of coccinellids (P < 0.05).

Table 1 .
Mean vertical position of adult coccinellids on apple trees.