Old Parasitoids for New Mealybugs: Host Location Behavior and Parasitization Efficacy of Anagyrus vladimiri on Pseudococcus comstocki

Simple Summary Anagyrus vladimiri has been widely employed as a biological control agent (BCA) against the vine mealybugs Planococcus ficus but the knowledge about its employment against other mealybug species is limited. In this study, we investigated the potential efficacy of A. vladimiri for Pseudococcus comstocki management, considering the increasing threat represented by this mealybug pest in Mediterranean vineyards and fruit orchards. No-choice and two-choice tests were conducted to quantify parasitoid behavior against P. ficus and P. comstocki. Our results pointed out that A. vladimiri successfully parasitized both pests, showing no host preference between the two species. Our observations highlight that this parasitoid can be successfully deployed as BCA against P. comstocki populations. Abstract The Comstock mealybug, Pseudococcus comstocki (Hemiptera: Pseudococcidae) is a primary pest of orchards in the North and Northwest of China. This pest appeared recently in Europe, including Italy, where it is infesting mainly vineyards as well as apple and pear orchards. The present study investigated the efficacy of Anagyrus vladimiri, a known biological control agent (BCA) of Planococcus ficus, on P. comstocki to evaluate a potential use for the management of this new pest. No-choice tests were conducted to quantify the parasitoid behavior against P. ficus and P. comstocki. The parasitoid successfully parasitized both species (parasitization rate: 51% and 67% on P. comstocki and P. ficus, respectively). The A. vladimiri developmental time (19.67 ± 1.12 vs. 19.70 ± 1.07 days), sex ratio (1.16 ± 1.12 vs. 1.58 ± 1.07) and hind tibia length of the progeny showed no differences when P. comstocki and P. ficus, respectively, were exploited as hosts. Two-choice tests, conducted by providing the parasitoid with a mixed population of P. ficus and P. comstocki, showed no host preference for either of the two mealybug species (23 vs. 27 first choices on P. comstocki and P. ficus, respectively). The parasitization rate (61.5% and 64.5% in P. comstocki and P. ficus, respectively) did not differ between the two hosts. Overall, our study adds basic knowledge on parasitoid behavior and host preferences and confirms the use of this economically important encyrtid species as an effective BCA against the invasive Comstock mealybug.


Introduction
The Comstock mealybug, Pseudococcus comstocki (Kuwana) (Hemiptera: Pseudococcidae), is widely recognized as an important insect pest of fruit trees, especially pear trees, in many fruit-producing regions of the world, with special reference to North and Northwest Insect rearing and experimental assays were conducted in laboratory conditions at 23 ± 1 • C, 45 ± 5% RH and a 14:10 (L:D) photoperiod. A commercial strain of A. vladimiri, as well as its routine host P. ficus, were maintained as described by Romano et al. [14]. A field strain of P. comstocki, originally collected in the Emilia-Romagna region (Central Italy), was reared on potato sprouts, a common food substrate for mealybugs, as detailed by Islam and Copland [15]. The whole rearing apparatus was placed in dark rearing cages (180 × 90 × 90 cm) (Bugdorm ® , Megaview Science, Taiwan).
All experiments were carried out using 2-5-day-old A. vladimiri mated females fed ad libitum with a solution of honey and water (1:1, w:v) and never exposed to mealybug hosts before testing [15].

Oviposition Behavior, Host Preferences, Host Suitability and Quality of the Parasitoid Progeny
The host-seeking and oviposition behavior of A. vladimiri was quantified on P. comstocki young females following the method by Chong and Oetting [16] with a few modifications. The host-seeking and oviposition behavior of A. vladimiri on P. comstocki were observed using a new Petri dish for each test (hereafter, the arena, diameter 35 mm) under uniform daylight conditions. Two kinds of experiments were conducted: no-choice and two-choice tests, detailed below.

No-Choice Tests
In the no-choice test, a single A. vladimiri mated female was provided with 8 P. comstocki (mean ± SD; length x width: 2.65 ± 0.38 × 1.57 ± 0.29 mm; weight: 0.0012 ± 0.0003 g) or 8 P. ficus young females (mean ± SD; length x width: 2.95 ± 0.14 × 1.78 ± 0.17 mm; weight: 0.0013 ± 0.0002 g). To easily track each mealybug during the behavioral assays, each insect position was noted on a separate paper sheet before the experiment, and further position changes were noted by an observer [16]. To reduce the potential influence of visual cues surrounding the testing arena on the insect behavior, the observer was dressed in a white coat, and the arena was surrounded by a white wall of filter paper (Whatman no. 1, height 30 cm) [17].
After introducing the parasitoid in the experimental arena, the occurrence and duration (s) of the following displays were noted and used to construct an ethogram: (i) latency (i.e., the time spent by A. vladimiri remaining stationary before starting host searching), (ii) host searching (i.e., the parasitoid walks around performing antennal tapping), (iii) host encounter (i.e., the parasitoid detects a potential host and stops close to it), (iv) antennal examination (i.e., the parasitoid remains still and performs antennal tapping on the host), (v) ovipositor probing (i.e., the parasitoid swiftly inserts its ovipositor in the host, if the event lasts more than 10 s, an oviposition succeeds; Benelli G., personal observation), (vi) and oviposition (i.e., the parasitoid keeps the ovipositor inside the host's body, laying an egg) [16]. (vii) Host dragging during oviposition (i.e., the host moves away dragging the ovipositing parasitoid around), as well as (vii) host defensive displays (i.e., the mealybug performs quick body movements against the parasitoid), were also recorded. The possible occurrence and duration of host feeding behavior was noted, following Bokonon-Ganta et al. [18]. The number of mealybugs encountered, examined, and probed by each parasitic wasp within the observation period was noted [19]. The observation period was 20 min; A. vladimiri females not starting any of the host-seeking displays detailed above within 10 min were discarded from the study [16]. Each no-choice observation was replicated 50 times.
Furthermore, we assessed the host suitability of P. comstocki young females for the development and survival of A. vladimiri, and the quality of progeny. Both in the P. comstocki and P. ficus no-choice experiments, the parasitoid was left in the arena for 24 h after the 20-min observation period, allowing the female to parasitize mealybugs. Then, the exposed hosts were monitored daily for 30 days using a stereomicroscope, noting the parasitism rate (i.e., % of mummified mealybugs) and development duration (i.e., the period between A. vladimiri parasitization and adult emergence). The number and sex ratio of A. vladimiri offspring on the two hosts were also determined [16].
Hind tibia length was used as a surrogate for body length [19]. Newly emerged parasitoids were stored in 70% ethanol 24 h after their emergence. Their left hind tibia was removed and mounted on a microscope slide. The left hind tibia length of all adult parasitoids was measured with an ocular micrometer at 60× to investigate the effect of host species on the fitness of parasitoids [16].

Two-Choice Tests
In the two-choice tests, a single A. vladimiri female was exposed to a mixed population of P. comstocki and P. ficus young females, transferring 4 individuals per species into Insects 2021, 12, 257 4 of 14 the experimental arena. The P. comstocki and P. ficus young females were alternately distributed close to the borders of the area, equally distanced each other. The A. vladimiri female was released in the center of the testing arena, equally distanced from all the mealybug individuals.
Following the method reported above for "No-choice tests", after the introduction of an A. vladimiri female into the arena, the parasitoid was visually tracked by an observer, and the following parameters were noted: (i) the A. vladimiri first choice (i.e., which mealybug was first approached with a successful parasitization during the observation time), (ii) the number of successfully parasitized hosts within the observation time, and (iii) the oviposition duration on the selected host (if multiple oviposition acts occurred, only the duration of the oviposition following the first choice on a given host was noted). For each replicate, the observation time was 20 min; 50 replicates were carried out. A. vladimiri females not starting any of the host-seeking displays within 10 min were removed from the study [16].
Furthermore, in each replicate A. vladimiri was left in the arena for 24 h after the 20 min observations period direct observation, allowing the parasitization of the mealybugs of both species. The successful parasitization of the exposed hosts was monitored daily for 30 days, noting the parasitism rate on P. comstocki and P. ficus (i.e., % of mummified mealybugs).

Statistical Analysis
In no-choice behavioral assays, differences in the duration of the following behavioral displays, i.e., latency, host searching, antennal tapping, probing and oviposition on the two hosts were evaluated by a General Linear Mixed Model (GLMM) with one factor [20]: y iw = µ + H i + ID w + e iw , in which y iw is the observation, µ is the overall mean, H i is the i-th fixed effect of the tested mealybug host (i = 1-2), ID w is the w-th random effect of the parasitoid over repeated host searching and parasitization events (w = 1-50) and e iw the residual error; p < 0.05 was used to assess significance of differences between means.
Differences in the number of parasitoids displaying host encounter, antennal tapping, probing, oviposition, host dragging, kicking and superparasitization on the two mealybug hosts and parasitoid emergence were evaluated using the Kruskal-Wallis test (p = 0.05), while parasitization rates were evaluated using the Wilcoxon test (p = 0.05).
Differences in the hind tibia length of parasitoids emerged from the two hosts were evaluated using a weighted generalized linear model (GLZ, Poisson distribution) with one fixed factor [21]: y = Xß + ε where y is the vector of the observations (the hind tibia length), X is the incidence matrix, ß is the vector of the fixed effect (the mealybug host) and ε is the vector of the random residual effects.
In two-choice tests, differences between the number of A. vladimiri first choices when parasitizing P. comstocki vs. P. ficus were evaluated using a likelihood ratio χ 2 tests, with Yates' correction [22]. Furthermore, differences in the number of successfully parasitized hosts within 20 min and 24 h on the two mealybug hosts were analyzed with a Wilcoxon test (p = 0.05). Differences in the A. vladimiri oviposition duration on the two mealybug hosts were evaluated using the Kruskal-Wallis test (p = 0.05). JMP 9 (SAS) was used for all the analyses.

No-Choice Tests
In no-choice tests, when A. vladimiri encountered a potential host, it started antennal tapping on the mealybug body to accept or reject the host (Figure 1). If the host was of interest, A. vladimiri turned itself, everted the ovipositor and attempted to probe.

No-Choice Tests
In no-choice tests, when A. vladimiri encountered a potential host, it started antennal tapping on the mealybug body to accept or reject the host (Figure 1). If the host was of interest, A. vladimiri turned itself, everted the ovipositor and attempted to probe. If the host was not suitable, the parasitoid moved away and started again the hostseeking activity. Oviposition occurred after a positive probing outcome, usually on the dorso-lateral side of the host ( Figure 2). Overall, the displays composing the host-seeking and parasitization behavior of A. vladimiri on P. comstocki were walking and drumming activity, arrestment close to the host, antennal tapping, probing, oviposition, and host dragging. In addition, a peculiar host defensive behavior was noted, i.e., a fast abdominal rocking movement against the parasitoid (kicking), coupled or not with the production of a viscous secretion against the parasitoid to impair its wings [23,24]. Host feeding behavior was not observed.
The ethograms of A. vladimiri parasitizing P. comstocki ( Figure 3a) and P. ficus ( Figure  3b) were built analyzing the first host-seeking and oviposition event observed in fifty A. vladimiri females. If the host was not suitable, the parasitoid moved away and started again the hostseeking activity. Oviposition occurred after a positive probing outcome, usually on the dorso-lateral side of the host ( Figure 2).

No-Choice Tests
In no-choice tests, when A. vladimiri encountered a potential host, it started antennal tapping on the mealybug body to accept or reject the host (Figure 1). If the host was of interest, A. vladimiri turned itself, everted the ovipositor and attempted to probe. If the host was not suitable, the parasitoid moved away and started again the hostseeking activity. Oviposition occurred after a positive probing outcome, usually on the dorso-lateral side of the host ( Figure 2). Overall, the displays composing the host-seeking and parasitization behavior of A. vladimiri on P. comstocki were walking and drumming activity, arrestment close to the host, antennal tapping, probing, oviposition, and host dragging. In addition, a peculiar host defensive behavior was noted, i.e., a fast abdominal rocking movement against the parasitoid (kicking), coupled or not with the production of a viscous secretion against the parasitoid to impair its wings [23,24]. Host feeding behavior was not observed.
The ethograms of A. vladimiri parasitizing P. comstocki ( Figure 3a) and P. ficus ( Figure  3b) were built analyzing the first host-seeking and oviposition event observed in fifty A. vladimiri females. Overall, the displays composing the host-seeking and parasitization behavior of A. vladimiri on P. comstocki were walking and drumming activity, arrestment close to the host, antennal tapping, probing, oviposition, and host dragging. In addition, a peculiar host defensive behavior was noted, i.e., a fast abdominal rocking movement against the parasitoid (kicking), coupled or not with the production of a viscous secretion against the parasitoid to impair its wings [23,24]. Host feeding behavior was not observed.
The ethograms of A. vladimiri parasitizing P. comstocki ( Figure 3a) and P. ficus (Figure 3b) were built analyzing the first host-seeking and oviposition event observed in fifty A. vladimiri females.  As detailed in Figure 3, A. vladimiri showed comparable host seeking and oviposition sequences towards both mealybug hosts. However, in our no-choice tests, A. vladimiri detected a slightly higher number of P. comstocki individuals over P. ficus (68% vs. 54%, respectively) although this does not mean that one was detected more than the other. Most of the selected hosts were subjected to antennal examination, lasting less than 10 s on both hosts. Probing was observed in 32% of parasitoids attacking P. comstocki, while probing on P. ficus was 40%. The oviposition rate was 22% on P. comstocki and 20% on P. ficus. The above-described host defensive behavior showed by P. comstocki and P. ficus against A. vladimiri was observed in 3% and 0.0015% of the parasitoid-host interactions, respectively. A. vladimiri superparasitization acts occurred on both hosts with a comparable rate (P. comstocki 0.1% and P. ficus 0.13%).
On the other hand, no significant difference in the developmental time of the offspring of A. vladimiri from the hosts was found (χ 2 = 0.088, d.f. = 1, p = 0.7675) (Figure 7). The first adult emerged after 17 days from P. comstocki (with a mean emergence time of 19.67 ± 1.12 (mean ± SD)) and 18 days from P. ficus (with a mean emergence time of 19.70 ± 1.07 (mean ± SD)). Moreover, no difference in the sex ratio was noted between parasitoid progeny emerging from P. comstocki and P. ficus (adult sex ratio (ASR): 1.16 and 1.58 for female P. comstocki and P. ficus, respectively).
The length of the hind tibia of emerged A. vladimiri was used as a surrogate of body length to evaluate the fitness of the offspring [19]. No significant difference was found between the two hosts (χ 2 = 0.12, d.f. = 1, p = 0.720) (Figure 8), confirming that P. comstocki is a highly suitable host for A. vladimiri.  host defensive behavior (χ 2 = 4.77, d.f. = 1, p = 0.029). However, comparing latency (χ 2 = 0.86, d.f. = 1, p = 0.35), host dragging (χ 2 = 0.0085, d.f. = 1, p = 0.92) and superparasitization (χ 2 = 3.15, d.f. = 1, p = 0.076), the results showed no significant differences between P. comstocki and P. ficus ( Figure 5). The parasitization rate, percentage of emerged parasitoids, developmental time and sex ratio of A. vladimiri developed on P. comstocki were compared with those from wasps developed on P. ficus ( Figure 6). Significant differences between the percentage of parasitized hosts were found (χ 2 = 11.808, d.f. = 1, p = 0.0006) ( Figure 6). The number of parasitoids emerged from the two hosts was not statistically different (χ 2 = 0.141, d.f. = 1, p = 0.707). On the other hand, no significant difference in the developmental time of the offspring of A. vladimiri from the hosts was found (χ 2 = 0.088, d.f. = 1, p = 0.7675) (Figure 7). The first adult emerged after 17 days from P. comstocki (with a mean emergence time of 19.67 ± 1.12 (mean ± SD)) and 18 days from P. ficus (with a mean emergence time of 19.70 On the other hand, no significant difference in the developmental time of the offspring of A. vladimiri from the hosts was found (χ 2 = 0.088, d.f. = 1, p = 0.7675) (Figure 7). The first adult emerged after 17 days from P. comstocki (with a mean emergence time of 19.67 ± 1.12 (mean ± SD)) and 18 days from P. ficus (with a mean emergence time of 19.70 ± 1.07 (mean ± SD)). Moreover, no difference in the sex ratio was noted between parasitoid progeny emerging from P. comstocki and P. ficus (adult sex ratio (ASR): 1.16 and 1.58 for female P. comstocki and P. ficus, respectively). The length of the hind tibia of emerged A. vladimiri was used as a surrogate of body length to evaluate the fitness of the offspring [19]. No significant difference was found between the two hosts (χ 2 = 0.12, d.f. = 1, p = 0.720) (Figure 8), confirming that P. comstocki is a highly suitable host for A. vladimiri.

Two-Choice Tests
Anagyrus vladimiri did not show a significant preference for P. comstocki over P. ficus in terms of first parasitization choice (23 vs

Discussion
This study aimed to understand whether A. vladimiri could be used to effectively manage P. comstocki based on the knowledge that this wasp is a reliable parasitoid of P. ficus [12,25]. However, there was no evidence about the suitability of P. comstocki as a host for A. vladimiri, despite the growing importance of Comstock mealybugs [2]. The present study proved the successful development of A. vladimiri on P. comstocki. An accurate comparison between the two hosts was provided by the parasitization rate and supported by the quantification of frequencies and durations of the characteristic behaviors involved in the host location and parasitization sequence (i.e., host searching, encounter, antennal tapping, probing, oviposition, host dragging, kicking, and superparasitization). Although some displays were more frequent or prolonged on one host species over the other, the parasitization rates as well as the percentage of parasitoids emerged from the respective hosts, were fully comparable. Earlier research on closely related Anagyrus species showed comparable parasitization rates on different hosts of agricultural importance. For example, testing Anagyrus kamali Moursi (Hymenoptera: Encyrtidae) on Maconellicoccus hirsutus

Discussion
This study aimed to understand whether A. vladimiri could be used to effectively manage P. comstocki based on the knowledge that this wasp is a reliable parasitoid of P. ficus [12,25]. However, there was no evidence about the suitability of P. comstocki as a host for A. vladimiri, despite the growing importance of Comstock mealybugs [2]. The present study proved the successful development of A. vladimiri on P. comstocki. An accurate comparison between the two hosts was provided by the parasitization rate and supported by the quantification of frequencies and durations of the characteristic behaviors involved in the host location and parasitization sequence (i.e., host searching, encounter, antennal tapping, probing, oviposition, host dragging, kicking, and superparasitization). Although some displays were more frequent or prolonged on one host species over the other, the parasitization rates as well as the percentage of parasitoids emerged from the respective hosts, were fully comparable. Earlier research on closely related Anagyrus species showed comparable parasitization rates on different hosts of agricultural importance. For example, testing Anagyrus kamali Moursi (Hymenoptera: Encyrtidae) on Maconellicoccus hirsutus (Green) (Hemiptera: Pseudococcidae), [26] a parasitization rate of 65% was found, while Aenasius bambawalei Hayat (Hymenoptera: Encyrtidae) parasitized 52% of Phenacoccus solenopsis Tinsley (Hemiptera: Pseudococcidae) [27]. However, concerning the emergence rate, Sagarra and Vincent [26] found much lower values than ours (between 19% and 48.5% depending on the age of the host), while Zhang et al. [27] obtained an emergence rate (81.78%) similar to our results. Of note, in our study, the number of males emerged from P. ficus was higher than females.
A key feature when assessing host suitability is the evaluation of the fitness of the parasitoid progeny. As stressed by Sagarra et al. [19], body length and hind tibia length are linearly related, thus the hind tibia length can be a precise and rapid tool to evaluate the overall size of the parasitoid and, therefore, its fitness. In our results, the tibia length of A. vladimiri progeny emerged from the two host species was not statistically different, showing that the fitness of the newly emerged parasitoids was not compromised when P. comstocki was exploited as a host. Of note, the two-choice tests carried out by introducing mated A. vladimiri in the arena with a mixed population of four P. comstocki and four P. ficus young females revealed no preferences for a particular host, thus confirming that the parasitoid accepted both hosts. The parasitization rate of the two species after 24 h of host exposure to the female wasp was similar as well.
The capability of A. vladimiri to successfully parasitize different species of invasive mealybugs makes it a highly adaptable BCA. On the other hand, this may represent a relative risk of high likelihood of non-target effects if the insect needs to be introduced in regions where it is not native [28,29]. However, since A. vladimiri is naturally present in many fruit-producing regions worldwide [11,12], mass releases may boost the local population of the parasitoid, positively enhancing IPM and biocontrol programs. This is confirmed by our results, as well as by other papers where this encyrtid has been evaluated against various mealybugs, e.g., P. ficus, Planococcus citri (Risso), Pseudococcus calceolariae (Maskell), Pseudococcus viburni (Signoret), and Phenacoccus peruvianus Granara de Willink (Hemiptera: Pseudococcidae) [30]. For instance, in this study, it is shown that A. vladimiri can complete its development on all the hosts although with different success rates. Further tests have shown significant differences in the behavioral patterns of host recognition, host handling, and the level of host acceptance [31]. In our research, the lack of differences in the sequence of events leading to oviposition, the main behavioral parameters as well as to the parasitization success in all the performed tests, support the use of A. vladimiri as effective BCA for P. comstocki management. Moreover, the utilization of both species suggests that A. vladimiri can be released to manage the infestation of a single mealybug pest, as well as in scenarios where both mealybug species are present simultaneously. This is a common situation in many fruit orchards of North and Central Italy (G. Benelli, pers. observ.).

Conclusions
Our no-choice experiments showed several differences in the frequency and duration of selected displays characterizing the host-seeking and oviposition of A. vladimiri on P. comstocki and P. ficus. However, both mealybug species were equally suitable as hosts for A. vladimiri and supported the production of progeny with similar body size. Furthermore, the results from two-choice tests highlighted that P. comstocki was preferred by A. vladimiri females in a comparable manner to its classic host P. ficus. Overall, our findings showed that A. vladimiri successfully parasitized and developed on P. comstocki, therefore, highlighting that this encyrtid species may have general utility in biological control programs with one or more mealybug species.