Three preliminary whitefly releases allowed for a rapid and large build-up of pest populations (Table 3-1). Whiteflies, obtained from Applied Bionomics Corporation, B. C., Canada, were introduced either by placing fourth instar pupae onto the lowest leaf of tomato plants as for the first introduction (August 16) or by direct introduction of adults in the case of the later two introductions (August 29, September 1). Together, these introductions ensured the presence of a large developing whitefly population. By the third whitefly introduction, whitefly had colonized the entire height of the tomato plants. On September 24^th, whitefly densities were judged sufficiently large to support the persistence of natural enemies that were introduced either that day or the following day. An initial release of E. formosa and D. hesperus made in late September was followed by one larger parasitoid and two smaller predator releases in October (Table 3-1). Encarsia formosa was first released as adults received from Biobest® Biological systems (Westerlo, Belgium) and subsequent introductions were made by hanging parasitized whitefly cards (Applied Bionomics).

Dicyphus hesperus insects were obtained from a laboratory colony at the PARC research centre, which was initially established from field collected individuals in 1999 from the foothills of the Sierra Nevada Mountains (elevation ca. 300 m, near Woody, CA, USA). These insects were originally found on stem hedge nettle, Stachys albens A. Gray (Lamiaceae). Colonies were maintained in screened wooden cages (38 x 51 x 53 cm) containing tobacco plants at 24.8 ± 1.5˚C (± SD), and under a 16 h photoperiod, and were fed eggs of Ephestia kuehniella Zeller (Lepidoptera: Pyralidae) from Biobest® Biological systems (Westerlo, Belgium). In experiments, four day old adult D. hesperus were added in a 50/50 ratio of males and females.

The first application of BotaniGard® followed the first release of parasitoids and predators (Table 3-1). A second and third application of this pathogen was followed 5 and 26 days after the initial application. This third and final application was applied during a period of high whitefly densities.

Before treatment, in order to determine fungal deposition rate (conidia per mm²), 5% water agar blocks were pinned on the underside of three randomly selected leaves from two plants in each of the eight BotaniGard® compartments. Pinned agar blocks held in place by a piece of plasticine were collected after microbial application and the number of conidia in a set ocular area counted. Entomopathogen applications were made using a hand held pressurized sprayer (11.4 L Model # 65010, Hudson & Industrial, Chicago IL, USA). A small quantity of the fungal suspension was taken from the sprayer early, mid and late during the course of each treatment in order to determine the actual conidial concentration of a suspension through their enumeration using a haemocytometer (Table 3-1). Conidia viability was also determined by spraying the spore suspension onto three 0.005% Benlate® (benomyl, wettable powder 50%, E.I. duPont de Nemours and Co., Wilmington, DE) amended potato dextrose agar (PDA) plates during application. Benlate®-amended plates facilitate counting germinated conidia as benlate halts cell division during mitosis that otherwise causes overgrowth to occur (Goettel and Inglis, 1997). Plates were sealed with Parafilm and incubated at 25ºC for 2 days, after which the proportion of 500 conidia that had germinated from each sample was determined. Viability of conidia was established to be of 97.5, 96.3 and 93.2 % for BotaniGard® treatments 1, 2 and 3, respectively.

Each class has specific morphological characteristics that we used for identification. Whitefly eggs were identified as small black dots from a distance, that appear spindle shaped when examined with a hand lens. N1-2 whiteflies were also identified using a hand lens. N1 and N2 whitefly instars had characteristic small (<0.5 mm), white, flattened bodies. N3-4 whiteflies were larger in size, and were characterized by thickened bodies and a fringe of hairlike protrusions around the periphery of their waxy cover. When infected with B. bassiana , the colour of immature whitefly generally changed from opaque white to red or brown. In certain cases infected whitefly pupae were identified by the presence of mycelium or hyphae, which protruded around their periphery. Late in the infection process, whiteflies appeared ddehydrated and flattened. Adult whiteflies were mostly found concentrated at the top of tomato plants, but were also occasionally found on lower leaves following emergence.

Adult E. formosa parasitoids measure approximately 0.6 mm in length and are easily detected with the naked eye. Parasitized whiteflies appeared black as they normally do two weeks following the deposition of the parasitoid egg within the host pupae. Early parasitoid development could not be visually detected.

Adult D. hesperus are relatively large insects with a length of approximately 7 mm and are therefore easily detectable at a distance. Dicyphus hesperus consumed prey are left behind on leaves, and thus could be counted by observing the empty whitefly exoskeletons. Though predator consumed whitefly could be confounded with the emptied pupal exoskeletons of emerged adult whitefly, the top surface of empty pupae resulting from predation remained intact whereas emerging adult whitefly from pupae cause a distinct tear or dislocate the top surface of the pupal shell.

The proportion of infected whitefly was determined under laboratory conditions by sampling one leaflet with at least 30 immature whiteflies from each of the 8 compartments per glasshouse. These leaflets were placed within a 15 cm diameter Petri dish on a moistened filter paper, and incubated at 25ºC. Leaflets were kept moist by regularly adding water as needed. Following a 5-day incubation period, infected whitefly that were either sporulating or were red in colour due to B. bassiana ’s production of oosporein, were enumerated and the proportion of infected whitefly calculated over the total number of whitefly per leaflet was determined.

Within all BotaniGard® treated compartments, the density of infected immature whitefly was significantly different over time compared to control compartments (Table 3-3). Infected whitefly were consistently more abundant in treated compartments, and increased in proportion as soon as the first sample was taken following an initial BotaniGard® treatment (Figure 3-2). Infected pupae were either red in colour, had developed fungal hyphae or displayed both of these characteristics.

Figure 3-1 Effect of BotaniGard® treatment on the density per leaf of a) eggs; b) N1-2; c) N3-4 and d) adults of the whitefly Trialeurodes vaporariorum . Values represent the mean density (± SEM) of samples (N= 16) taken at each of four plant canopy levels. Vertical lines indicate BotaniGard® treatments. The x-axis represents the sample date whereby the first number stands for weeks and the second for the bi-weekly repetition of sampling.

Though the impact of infection on E. formosa during development in the whitefly host was not determined in our study, other parasitoid-pathogen interactions have been studied and should be considered (Rombach and Gillespie, 1988; Fransen and van Lenteren, 1993; 1994; Mesquita et al ., 1997; Lacey et al ., 1997; Furlong, 2004). In these studies factors such as the temporal separation of species and intraguild interactions seem to contribute to the outcome of parasitoid-pathogen interactions. A study of the interaction between the aphelinid parasitoid, Aphelinus asychis Walker (Hymenoptera: Aphelinidae) and the entomopathogenic fungus Paecilomyces fumosoroseus Wize (Deuteromycotina: Hyphomycetes) exploiting the Russian wheat aphid in barley fields showed that these natural enemies are compatible as they work additively to enhance aphid control (Mesquita et al . 1997). In this study, no difference in the number of mummies recovered per plant was found between the A. asychis parasitism treatment alone and the combined effect of parasitism and infection by P. fumosoroseus together. Also, the comparable A. asychis emergence in both of these treatments observed suggests the absence of interference between pathogen and parasitoid. In contrast, laboratory tests showed that adult parasitoids of A. asychis were susceptible to fungal infection by P. fumosoroseus (Lacey et al ., 1997). These contradictory results highlight the importance of evaluating interactions between fungus and parasitoid under both laboratory and field conditions.

Timing the application of a pathogen during the use of a parasitoid is often an important factor contributing to the success of biological control. Furlong (2004) showed that B. bassiana was detrimental to the endolarval parasitoid Plutella xylostella L. (Lepidoptera: Plutellidae) developing within the immature stages of Diadegma semiclausum Hellen (Hymenoptera: Ichneumonidae). Emergence of parasitoids was reduced and infection of parasitoid larvae by B. bassiana increased when the concentration of the pathogen increased (Furlong, 2004). However, parasitoid mortality was avoided through the careful timing of pathogen applications. Beauveria bassiana may be effectively applied at least one day before the predicted parasitoid cocoon formation, thus avoiding infection to parasitoids. Temporal separation was also an important factor in the interaction of E. formosa and the entomopathogen Aschersonia aleyrodis (Fransen and van Lenteren, 1993; 1994). Infection of parasitoids diminished as developing parasitoids became older. Fungal spores of A. aleyrodis did not infect parasitized whiteflies with immature parasitoids older than three days (Fransen and van Lenteren, 1993). When appropriate measures of temporal separation are applied, the potential use of E. formosa fungi such as Verticillium lecanii Zimm. (Deuteromycotina: Hyphomycetes) (Rombach and Gillespie, 1988) or P. fumosoroseus (van de Veire and Degheele, 1996), may also be effective.

Host discrimination by E. formosa adults may also have played a role in population dynamics of the developing parasitoid. The higher abundance of parasitized whitefly observed in treated compartments (Figure 3-2) may be influenced by the selective oviposition of females (Fransen and van Lenteren, 1993; 1994). Encarsia formosa rarely chooses to oviposit whitefly pupae infected by Aschersonia aleyrodis (Fransen and van Lenteren, 1993; 1994). In this light, it is believed that oviposition of pathogen infected hosts is generally avoided by E. formosa adults. Jazzar and Hammad (2004) demonstrated that ovipositing females of E. formosa discriminated between healthy and Verticillium lecanii infected Bemisia argentifolii whitefly and therefore reduced interference between these natural enemies. The combined use of E. formosa and V. lecanii resulted in a 70.7% whitefly mortality. Of this amount, 10% of whitefly mortality was due exclusively to parasitism and 58.3% due to infection. Lacey et al ., (1996) reported that E. formosa under field conditions resulted in a synergistic interaction with the naturally occurring V. lecanii . In our study, the absence of interference between the pathogen and the parasitoid suggested by the greater abundance of parasitized whitefly in treated compartments signals the compatibility of the parasitoid and the pathogen (Figure 3-4). Parasitoids were occasionally found dead on leaves following the application of both water or pathogen treatments (Labbe, personal observation). It was proposed that these parasitoids would have died of drowning, a factor that may be considered equally important in all compartments.

Competition for healthy prey may be an important factor contributing to changes in the distribution of predator D. hesperus . In many ecosystems, a change in distribution may occur when an organism inhabits an area or a time where its optimal food resource is limited (Polis et al ., 1989). In this study the distribution of D. hesperus was positively correlated to that of whitefly so that predators were likely aggregated in regions where a high density of immature whitefly were found, regardless of whether these were infected (Labbé, unpublished data). This suggests that when the whitefly resource is limited, predators may have little option but to prey upon infected whitefly. In a study by Dixon (1998), a similar distribution was observed when examining the interactions between different species of aphid parasitoids. Notably, the patchiness of aphid hosts favoured the aggregation of aphid natural enemies around them, resulting in intra- and interspecific competition for this same limited resource. While it appears normal that a predator or parasitoid would aggregate around its primary resource, results presented in the previous chapter of this thesis suggest that it should not be the case for predator D. hesperus feeding upon infected whitefly. In particular, I reported that during laboratory experiments, D. hesperus avoided feeding on B. bassiana infected prey showing the presence of hyphae or of oosporein, a red pigment produced by the fungus.

A number of field studies in invertebrate communities have shown that though most herbivores are attacked by large guilds of natural enemies, complex and unexpected interactions between predators, parasitoids and pathogens may significantly reduce their impact on herbivore populations (Brodeur and Rosenheim, 2000). Predators, parasitoids and pathogens have been previously considered to have parallel and largely independent effects on herbivore populations, and potentially significant interactions between these natural enemies were neglected (Brodeur and Rosenheim, 2000). Recent attempts to extend current population models by adding interactions between natural enemies have contributed to the emergence of a more dynamic and reliable understanding of the factors that regulate the density of animal populations (see Polis et al ., 1989; Polis, 1991; Strong, 1992; Kareiva, 1994; Polis and Strong, 1996; Losey and Denno, 1998; Rosenheim, 1998).

Multispecies assemblages of biological control organisms may provide a balance between organisms in greenhouse agroecosystems that is necessary for consistent pest suppression. When species with known biological control properties are chosen to interact together, the shortcomings and limitations of individual species may be compensated for within the same guild. For example, whitefly suppression by polyphagous predators are rarely effective to lower pest populations below an acceptable level (Gerling, 1990), but they may be useful in containing their exponential growth by conservation, augmentation or even inoculation in the environment (Gerling, 1990). In the case of the specific parasitoid, E. formosa , during some periods of the year when peak high or low temperatures are observed, whitefly populations may become difficult to control solely by this species. In such situations, biological control practitioners resort to the use of alternate control measures such as provided by the use of generalist pathogens (Perdikis and Lykouressis, 2002). Our study showed that while the application of the entomopathogen B. bassiana caused an important reduction in whitefly populations, other natural enemy populations were maintained. These findings promote the use of multispecies complexes in greenhouse biological control that may serve to recreate a complete trophic web. Such a web is composed of a number of interactions each having an important role in community dynamics.

In contrast, a large number of recent field studies in arthropod communities have shown that while most herbivores are attacked by large guilds of natural enemies, some unexpected and complex interactions between predators, parasitoids and pathogens may significantly reduce their impact on herbivore populations (Brodeur and Rosenheim, 2000). Community dynamics are affected by factors such as foraging behaviour, indirect mutualism or intraguild predation and may be the cause of trophic cascades that appear over time. However, intraguild predation may be effectively used to offset competition during prey shortage (Hochberg and Lawton, 1990). In our study, B. bassiana controlled an imminent pest outbreak with a minor disruption on biological control. It is clear that further research on the role of intraguild interactions in the greenhouse environment is warranted, and particularly so for the most currently applied biological control organisms. This research has identified the compatibility of whitefly natural enemies, and would be further enriched by studies of simple and paired interactions between natural enemies with the whitefly, as well as between other natural enemy species.