Prey Preference and Life Table of Amblyseius orientalis on Bemisia tabaci and Tetranychus cinnabarinus

Amblyseius orientalis (Ehara) (Acari: Phytoseiidae) is a native predatory mite species in China. It used to be considered as a specialist predator of spider mites. However, recent studies show it also preys on other small arthropod pests, such as Bemisia tabaci (Gennadius) (Hemiptera: Aleyrodidae). Experiments were conducted to investigate (1) prey preference of A. orientalis between Tetranychus cinnabarinus (Boisd.) (Acari: Tetranychidae) and B. tabaci, and (2) development, consumption and life table parameters of A. orientalis when reared on T. cinnabarinus, B. tabaci or a mix of both prey species. When preying on different stages of T. cinnabarinus, A. orientalis preferred protonymphs, whereas when preying on different stages of B. tabaci, A. orientalis preferred eggs. When these two most preferred stages were provided together (T. cinnabarinus protonymphs and B. tabaci eggs), A. orientalis randomly selected its prey. Amblyseius orientalis was able to complete its life cycle on B. tabaci eggs, T. cinnabarinus protonymphs, or a mix of both prey. However, its developmental duration was 53.9% and 30.0% longer when reared on B. tabaci eggs than on T. cinnabarinus and a mix of both prey, respectively. In addition, it produced only a few eggs and its intrinsic rate of increase was negative when reared on B. tabaci eggs, which indicates that B. tabaci is not sufficient to maintain A. orientalis population. The intrinsic rates of increase were 0.16 and 0.23 when A. orientalis was fed on the prey mix and T. cinnabarinus, respectively. These results suggest that although B. tabaci is a poor food resource for A. orientalis in comparison to T. cinnabarinus, A. orientalis is able to sustain its population on a mix of both prey. This predatory mite may thus be a potential biological control agent of B. tabaci when this pest co-occurs with the alternative minor pest T. cinnabarinus.

Introduction established with individuals collected from soybean fields of Changli Research Institution of Pomology, Hebei Academy of Agricultural and Forestry Sciences (119°09'E, 39°43'N), with the permission from Prof. Lichen Yu, and has been maintained on C. lactis, in a climate chamber at 25±1°C, 80%±5% RH and 16L: 8D for multiple years. The C. lactis colony was obtained from Prof. Dr. Qinghai Fan, Fujian University of Agricultural and Forestry Sciences. Bemisia tabaci was obtained from Assistant Prof. Yubo Wang, Dryland Farming Research Institute, Hebei Academy of Agricultural and Forestry Sciences, and was reared on 2-week-old bean (Phaseolus vulgaris) seedlings in climate boxes (22±1°C, 60%±5% RH and 16L: 8D). Tetranychus cinnabarinus was collected from strawberry fields of Changping Agricultural Technology Extension Station, Beijing (116°12'E, 40°13'N), with the permission from station director Baowen Liu, and was reared on 2-week-old bean seedlings at the same conditions as for rearing A. orientalis. Bean seedlings used for colony maintenance were reared at 25±1°C, 70%±5% RH and 16L: 8D.

Experimental unit
In both preference and life table experiments (experiments (1) and (2) respectively), A. orientalis was reared individually in 10 (dia.)×3 (h.) mm 3 arena, which was built with a transparent acrylic board (30×20×3mm 3 ) with a 10mm diameter hole in the center, sealed on one side with a piece of bean leaf disc (made of first leaves of 2-week-old bean seedlings as used for colony maintenance), a piece of filter paper, and a piece of rectangular glass (30×20×1mm 3 ), and on the other side with another piece of rectangular glass (30×20×1mm 3 ). The 5 layers were tightly clipped together on both ends to avoid mite escaping (Fig 1). Preliminary observations show that leaves keep fresh for ca. 2 days in our experimental unit. Experimental units were placed in climatic chambers at 25±1°C, 70%±5% RH and 16L: 8D.

Prey preference between B. tabaci and T. cinnabarinus
In this experiment, the predator A. orientalis was exposed to three distinct treatments to estimate its prey preference. The first two treatments estimated preference of A. orientalis between two B. tabaci stages (eggs or 1 st instar larvae) and among three T. cinnabarinus stages (eggs, larvae, or protonymphs), respectively. A preliminary study with each prey species and stage provided in isolation suggest that later stages of these species are less preferred by A. orientalis and were therefore not tested in the present study [21]. In each treatment, 20 experimental units were prepared. In each experimental unit, 10 individuals of each stage were provided to a female A. orientalis adult. All females were starved for 24 hours before being used in preference experiments to standardize hunger levels. The number of prey consumed within 12 hours was recorded. Since prey were not able to escape from the experimental unit, all missing prey were considered consumed by the predator.
The 3 rd treatment was conducted to estimate A. orientalis preferences between the most preferred stages of each of the two prey species (B. tabaci eggs and T. cinnabarinus protonymphs, as obtained from the previous experiments). Three sets of prey density combinations were provided to A. orientalis adult females starved for 24 hours: (1) balanced densities with 10 individuals of each prey species, (2) 10 B. tabaci eggs and different numbers (15,20,25,30,35) of T. cinnabarinus protonymphs, and (3) 10 T. cinnabarinus protonymphs and different numbers (15,20,25,30,35) of B. tabaci eggs. A minimum of 15 replicates were prepared for each density. The number of each prey species consumed within 12 hours was recorded.
In each preference experiment, the prey preference index (α) was estimated using (Eq 1), modified from Chesson's preference index [27], where N i is the number of the i th prey type available, and N ai is the number of the i th prey type consumed. For each individual test, the preference index of the most preferred food type equals 1.
Mean prey preference indices to each prey type were compared with paired t-test and oneway ANOVAs when there were 2 or 3 choices, respectively. When ANOVA was conducted, multiple comparisons were also conducted with Tukey HSD test. All mean comparisons with p<0.05 were considered to have statistical significant differences. In the 3 rd experiment, linear regressions were conducted to estimate the impact of the proportion of T. cinnabarinus in the prey mix on the number of both prey species consumed by A. orientalis. All analyses were processed with SPSS 19.0.

Impact of prey type on population biology and life-table parameters of A. orientalis
A second experiment was performed to estimate A. orientalis life table parameters when preying on T. cinnabarinus only, B. tabaci only, and a mix of both prey, respectively. The experiment was started with 100, 65, and 80 synchronic eggs (laid within 6 hours, considered as 3 hours old on average when experiment started) for T. cinnabarinus, B. tabaci and the prey mix treatments, respectively. Fewer numbers of eggs were used for the B. tabaci and the prey mix treatments due to limited availability of B. tabaci eggs. After each A. orientalis larval emergence, 35 T. cinnabarinus protonymphs, 25 B. tabaci eggs, or a combination of 15 T. cinnabarinus protonymphs and 25 B. tabaci eggs were provided daily, for each of the three treatments, respectively. A smaller number of T. cinnabarinus protonymphs were provided in the prey mix treatment to match our objective to evaluate the performance of A. orientalis when B. tabaci is the major pest and T. cinnabarinus is a secondary pest. The amount of prey exceeded maximum daily consumption rate of the predator based on preliminary observations. Survival and stage of each A. orientalis individual was recorded every 12 hours during the immature stages. For individuals that entered the next stage between two observations, the molting timing was taken as the midpoint time between the two observations. Newly emerged adult females were paired with males for 24 hrs. Preliminary observations show that A. orientalis adults mate within two hours after being paired and mating usually lasts for two hours, similar to previous reports of other Phytoseiid mite species [28]. Daily consumption rate and fecundity of each female were recorded until it dies. Developmental and fecundity data were right-skewed, and were rank-transformed for further analyses [29][30][31].
The conversion rate from daily biomass intake (I) to the daily fecundity (F) is termed as the prey-offspring conversion rate (γ), and was estimated using (Eq 2), modified by Hayes (1988) from Beddington et al. (1976) [32][33], where M indicates prey consumed to maintain basal metabolism. Empirical estimates of γ and M were obtained through linear regressions of daily prey consumption against daily fecundity.
Life tables were constructed from the observed age-specific survival rate (l x ) and age specific fecundity rate (m x ) of A. orientalis females. The following life table parameters were estimated: the intrinsic rate of population increase (r m ), the finite rate of population increase (t), the net reproductive rate (R 0 ), the mean generation time (T), and the doubling time (t) using Eqs (3a-3e) [34][35].
X 1 x¼0 e Àr m ðxþ1Þ l x m x ¼ 1 ð3aÞ l ¼ e r m ð3bÞ Variances of M, γ, and life table parameters were estimated using the bootstrap technique with 1000 repeated samples [36][37]. Bootstrap resamplings were conducted with R version 3.2.0. [38]. One-way ANOVAs were conducted to compare mean developmental durations (rank-transformed data), fecundity (rank-transformed data), and life table parameters of A. orientalis in the three treatments. Two-way ANOVA was conducted to estimate the impact of both prey type and predator growth stage on A. orientalis consumption rate. Multiple comparisons were also conducted with Tukey HSD test. All mean comparisons with p<0.05 were considered to have statistical significant differences. Mean comparison analyses were processed with SPSS 19.0.
When both prey species were provided together, the overall preference index of A. orientalis to T. cinnabarinus protonymphs did not differ significantly from that of A. orientalis to B. tabaci eggs (t = -1.940, df = 201, p = 0.054) ( Table 3). Among all density combinations, a significant preference to B. tabaci eggs was observed when 10 B. tabaci eggs and 15 T. cinnabarinus protonymphs were provided (t = -2.934; df = 14; p = 0.011) ( Table 3, Fig 2). The number of each prey type consumed matched their available proportion (Fig 2). The results suggest that A. orientalis randomly preys B. tabaci eggs and T. cinnabarinus protonymphs.

Impact of prey type on population biology and life-table parameters of A. orientalis
Among all eggs used in this experiment, 93, 76, and 57 developed to adults when reared on T. cinnabarinus protonymphs, the prey mix and B. tabaci eggs, respectively, among which, 43, 44 and 32 were females. Biological characteristics and life table parameters were estimated based on data obtained from female individuals. Amblyseius orientalis was able to complete its life cycle on all the three treatments. Developmental duration of eggs did not differ significantly among the three treatments, while for other immature stages and the pre-oviposition stage, the shortest developmental duration was observed when T. cinnabarinus protonymphs were provided, followed by the prey mix and B. tabaci eggs treatments (Table 4).
Amblyseius orientalis that used B. tabaci eggs as prey had the shortest oviposition duration and longevity, and the lowest daily fecundity and cumulative fecundity. Among the two other treatments, the oviposition duration of A. orientalis on the prey mix was 28% longer than that on T. cinnabarinus protonymphs. The post-oviposition duration and female longevity were 66% and 40% shorter, respectively, when reared on the prey mix than on T. cinnabarinus protonymphs (Table 4). Daily fecundity of A. orientalis on T. cinnabarinus protonymphs was 47% higher than that on the prey mix, while the cumulative fecundity of these two treatments does not differ significantly (Table 4). No significant difference was observed between the proportions of A. orientalis female offspring when reared on T. cinnabarinus protonymphs and on the prey mix, but both were higher than that on B. tabaci eggs (Table 4). Fig 3(a) summarizes A. orientalis female age-specific survival rate (l x ), while Fig 3(b) summarizes its age-specific fecundity curves (m x ) for the 3 treatments. The peak daily fecundity occurred 7, 10 and 3 days after adult emergence of A. orientalis reared on T. cinnabarinus protonymphs, the prey mix, and B. tabaci eggs, respectively.
The total prey consumption of A. orientalis was significantly affected by both prey type (F = 169.208; df = 2, 659; p<0.001), predator growth stage (F = 437.264; df = 5, 659; p<0.001), and the 2 way interaction (F = 62.374; df = 10, 659; p<0.001). For each of the three treatments, daily consumption rate of A. orientalis was higher at oviposition duration and pre-oviposition durations than at other stages (Table 5). When A. orientalis fed on T. cinnabarinus, its daily consumption rates at oviposition and pre-oviposition durations were ca. twice as high as that in the other two treatments, while much smaller differences were observed for other stages across the three treatments (Table 5). Table 6 summarizes estimated daily metabolism and prey-offspring conversion rate of A. orientalis in the three treatments. The estimated amount of prey consumed for basal metabolism was higher when T. cinnabarinus protonymphs were provided than when B. tabaci eggs or the prey mix were provided. When the prey mix were provided, A. orientalis showed higher conversion rate of each prey species (0.15 for B. tabaci, 0.16 for T. cinnabarinus) than that when each prey species was provided separately. The lowest conversion rate (0.02) was observed when B. tabaci eggs were provided.
Significant effects of prey types on all the life table parameters were observed (Table 7). When B. tabaci was used as the prey, both intrinsic rate of increase and doubling time of A. orientalis were negative (Table 7). Among the other two treatments, faster population increase was expected when T. cinnabarinus protonymphs was used as the prey, but A. orientalis population also increased on the prey mix.

Discussion
In this study, we aimed at measuring preference of the predatory mite A. orientalis between mites (T. cinnabarinus) and whiteflies (B. tabaci), and evaluating whether A. orientalis is able to establish its population when fed on whiteflies or on a mix of spider mites and whiteflies. Based on our results, A. orientalis randomly chose its prey between T. cinnabarinus and B. tabaci, although B. tabaci appeared to have a smaller nutritional value than T. cinnabarinus. Indeed A. orientalis population failed to increase when it fed on B. tabaci, but a prey mix with only 37.5% individuals of T. cinnabarinus resulted in A. orientalis producing approximately the same amount of offspring as when fed on T. cinnabarinus only. Preferences of generalist predators are often tradeoffs between multiple factors, especially prey nutritional quality and the ease of attacking, this latter being influenced by a few factors, including the ease to detect and access the prey, prey abundance, and the defensive behavior of the prey, etc. [26,[39][40]. Some Amblyseisus mites prefer prey with higher nutritional levels. For example, Xu and Enkegaard (2010) reported higher prey preference of A. swirskii to 1 st instar larvae of western flower thrips, Frankliniella occidentalis, the prey with higher nutritional value [41], than to nymphs of two-spotted spider mite Tetranychus urticae [42]. Under natural conditions, phytoseiid mites often locate and choose prey patches based on host plant and prey specific herbivore-induced plant volatiles. These phytochemicals provide information to arthropod predators about prey quality [43]. However, within a small arena as the experimental unit used in the present study, volatiles induced by different prey species are mixed and might not impact prey selection. In addition, the small enclosed arena caused longer patch visits. Zhang and Sanderson (1993) investigated the searching behavior of three Phytoseiidae mite species, Phytoseiulus persimilis, Metaseiulus (= Typhlodromus) occidentalis, Amblyseius andersoni, and showed that the number of prey encountered and attacked per predator are positively correlated with the duration of patch visits [44]. They also suggested that starvation lead to longer patch visits of the predatory mites. In this case, it is possible that A. orientalis spent long time at each prey patch it randomly hit, and maximized its food intake regardless of the most abundant prey species in the patch. Some other generalist predators show similar foraging behaviors, such as Macrolophus pygmaeus. Previous studies show this predator has strong preference to the prey with higher relative density, regardless of its nutritional value [45][46].
The preference of A. orientalis between spider mites and whiteflies in our experiment may have been influenced by differences in the predator capacity to attack and handle these prey. Gan et al. (2001) reported that it took A. orientalis multiple attacks to kill a P. citri female adult, which lead to higher energy cost in preying on spider mites and possibly lead to a reduced preference [47]. However, spider mite protonymphs are much smaller in size than female adults and show much lower levels of defensive behaviors. Amblyseius orientalis also randomly chose its prey when eggs, larvae, and protonymphs of T. cinnabarinus were provided together, similar to the prey preferences of some other Phytoseiidae mites, such as Neoseiulus californicus, and A. swirskii, to different stages of T. urticae [48]. Bemisia tabaci eggs are of comparable sizes as T. cinnabarinus eggs, and both are smaller than T. cinnabarinus protonymphs. It will be interesting to further investigate prey preference of A. orientalis to the eggs of the two species when provided together. If a significant preference to T. cinnabarinus eggs is observed, the random selection between T. cinnabarinus nymphs and B. tabaci eggs might be attributed to a tradeoff between prey nutritional value and prey defensive behavior. Otherwise, random selection between the two types of eggs will suggest that prey nutritional value may not affect prey preference of A. orientalis.   Learning is another factor that might affect prey selection. Some predatory mite adults that experienced alternative food during their immature stages accept alternative food faster than naïve adults, eg. P. persimilis and N. californicus [49][50]. In our study, A. orientalis individuals used in the preference experiment were reared on C. lactis and naïve to B. tabaci, while A. orientalis individuals used in the B. tabaci and the prey mix treatments of the life table experiment experienced B. tabaci during their immature stages. If learning positively affects the predator capacity to recognize and handle prey, we would expect higher daily predation rates of B. tabaci in the life table experiment than in the preference experiments. However, A. orientalis female adults consumed 14.10 B. tabaci eggs per day in preference experiments, but only 8.05 B. tabaci eggs per day in the life table experiments. A possible reason is that adults in the preference experiments had been starved for 24 hours before being tested but not in the life table experiment. Predatory mites used in preference studies are often starved for a minimum of 24 hours to standardize their hunger level, with the impact of starvation on prey preference often considered as negligible [42,51]. It is valuable to further investigate whether learning affects prey preference of A. orientalis or was confounded with the impact of starvation in the present study. To further estimate the potential of A. orientalis in controlling whiteflies, we compared its development, longevity, fecundity, and intrinsic rate of increase with estimates of these parameters of A. swirskii, the commercially available biological control agent of B. tabaci, in Table 8. The intrinsic rate of increase of A. orientalis when fed on the prey mix is lower than that of A. swirskii on B. tabaci, but higher than that of A. swirskii on spider mites. An interesting result is that longevities, mainly owing to the post oviposition duration (Table 4), of A. orientalis that fed on T. cinnabarinus or on prey mix are much longer than either A. orientalis that fed on B. tabaci, or A. swirskii. Previous studies showed that longevity of predatory mite females are sometimes affected by their mating status. For example, Ji et al. (2007) indicated that Neoseiulus cucumeris that mated only once have longer post-oviposition durations and longevities than those that mated multiple times [52]. In addition, A. orientalis is also highly tolerant to starving. A previous study showed that it survived up to 16 days without food supply [53]. These facts suggested that A. orientalis might sustain its population for a long time in field trials with low prey density.
To compare the performance of the two predators when spider mites and whiteflies coexist, we also performed a preliminary experiment to estimate the predation of the two prey by A. swirskii, at similar combinations of prey densities as used in the preference study of A. orientalis. The methods and results of this experiment were summarized in supporting information (S1 Side Experiment). We found that when A. swirskii was reared on B. tabaci eggs, it consumed ca. 11.13 whiteflies during 12 hours, ca. 15% higher than that of A. orientalis (9.8). The mean preference index for A. swirskii to B. tabaci eggs did not differ significantly from that to T. cinnabarinus protonymphs (t = -0.669, df = 133, p = 0.504). A similar pattern was observed for A. swirskii consumption rate of each prey as affected by the prey proportion (Fig 2 and  S1 Fig).
Overall, it is reasonable to consider A. orientalis as a potential biological control agent of B. tabaci, when whiteflies co-occur with T. cinnabarinus. For example, in vegetable greenhouses in Beijing, China and neighboring provinces (32- 42°N and 110-120°E), T. cinnabarinus outbreaks often occur in spring (April to May), while B. tabaci outbreaks occur in early summer (June). Based on results of the present study, we believe it is valuable to evaluate strategies with releases of a specialist predator of T. cinnabarinus, such as P. persimilis, in combination or sequentially to releases of A. orientalis. The specialist predator is expected to control T. cinnabarinus before its outbreak, while A. orientalis is expected to establish its population before T. cinnabarinus is completely controlled, and to control B. tabaci in early summer. The ease of A. orientalis mass rearing allows this species to be mass released in greenhouses [Patent: CN 201110456703]. To achieve successful field applications, it will be necessary to further estimate possible impacts of abiotic and biotic factors on biological control efficiency of B. tabaci, including temperature, relative humidity, and interactions with other coexisting pests or predators.