Impact of the pear psyllid Cacopsylla pyri host instar on the behavior and fitness 1 of the parasitoid Trechnites insidious 2 3

Pear is one of the most important fruit crops of temperate regions. The control of its mains pest, Cacopsylla pyri, is still largely based on the use of chemical pesticides, with all that this implies in terms of negative effects on the environment and health.Within the context of integrated pest management, innovative and ecologically sustainable strategies must be developed for. Although Trechnites insidiosus is the most abundant parasitoid of C. pyri, its biology and its potential as a control agent have been little studied.In this paper, we conducted experiments to evaluate the behavior of the specialist parasitoid T. insidiosus when exposed to different larval instars of the pear psyllid C. pyri, and to assess the quality of the next generation of parasitoids.We found that although T. insidiosus accepts all host instars for oviposition, the third and fourth instars were the most suitable host in terms of behavioral acceptance and progeny development.Our study is a first step for further studies on the interaction between psyllids and parasitoids and provides evidence that the specialist parasitoid T. insidiosus is a promising candidate for biological control strategies of the pear psyllid C. pyri.


Introduction 46
Pear (Pyrus communis L.) is one of the major cash crops in temperate climates, with 1,381,923 47 7 rate was calculated as the number of ovipositor insertions divided by the number of antennal 150 contacts. 151

The influence of host development on parasitism and offspring quality 152
After the behavioral bioassays, all psyllid larvae from a same replicate were placed on a same pear 153 tree for fourteen days to await the formation of mummies (i.e., dead psyllids containing a 154 developing parasitoid). We used in-vitro-cultivated pear trees (Williams cultivar) that were between 155 one and two years old, and between 0.75 and 1 meter in height. Plants were obtained from Battistini 156 Vivai (www.battisti-rebschule.it) and stored in individual cages in a climatic chamber at 24°C. 157 After fourteen days, pear trees were checked daily for the presence of mummies and adult psyllids. 158 Each mummy was then isolated in a falcon tube with a drop of honeydew until the parasitoid 159 emerged. Development time was measured as the number of days from oviposition to adult 160 emergence. Host suitability (number of mummies divided by the number of ovipositor insertions) 161 was calculated for each host instar to determine the developmental instar that provides the best egg-162 to-adult development. The emergence rate was also calculated as the number of emerging adult 163 parasitoid divided by the total number of mummies, for each treatment. Finally, the sex-ratio was 164 calculated as the number of males divided by the total number of emerging individuals. Three days 165 after emergence, parasitoids (males and females) were stored in a freezer at -20°C for future 166 measurements of tibia size and egg load. 167 The size of the tibia was used as a proxy for individual body size. The left hind tibia of each 168 emerging individual was measured using the software ImageJ 1.440 (Rasband, W.S., US National 169 Institutes of Health, Bethesda, MD, USA). To estimate their egg load, each emerging female was 170 squeezed from the abdomen beneath a cover slip on a microscope slide (Manfield and milles, 2002): 171 the female was placed on an object blade with a small amount of water and crushed with a 172 coverslip. To better extract the eggs, the pressure exerted on the coverslip started from the head 173 towards the abdomen. Only ellipsoidal mature eggs ( Figure 1) were counted. 174 8

Statistical analysis 175
Generalized linear models (GLMs) were fitted to the data to test the potential influence of host 176 instar (explanatory variable, with five levels) on female parasitoids behaviors and emerging 177 parasitoid quality. Response variables were the number of host feeding events (Poisson 178 distribution), the time spent host feeding (Gaussian distribution), the time spent grooming 179 (Gaussian distribution), the time spent walking (Gaussian distribution), the time spent resting 180 (Gaussian distribution), the number of antennal contacts (Poisson distribution), the number of 181 ovipositor insertions (Poisson distribution), the host acceptance rate (Gaussian distribution), the 182 number of mummies (Poisson distribution), the host suitability (Gaussian distribution), the 183 emergence rate (Binomial distribution), and the egg load (Poisson distribution). 184 We also used a GLM (Gaussian distribution) to test the potential influence of the sex, of the host 185 instar, and of their interaction, on the tibia size and the development of emerging parasitoids. All 186 significant GLMs were followed by Tukey post hoc tests to compare each level of the same factor 187 (host instar and sex). In addition, Spearman correlation tests were performed between the tibia size 188 and the egg load at the emergence of each female, for each host instar. Finally, using χ ² tests, we 189 compared the experimental results of sex ratio obtained for each larval instar to a 50/50 theoretical 190 sex ratio. 191 Statistical analyses were performed using R version 3.3.3 Copyright (C) 2016 The R Foundation for 192 Statistical Computing for Mac. All tests were applied under two-tailed hypotheses, and the 193 significance level was set at 0. 05. 194 195 Results 196

The influence of host development on parasitoid behavior 197
The number of antennal contacts varied significantly with host developmental instar (χ2=800. 30,198 DF=4, P<0.001). The minimum value was observed for T. insidious females exposed to second 9 instar psyllids, while the maximum was observed for individuals exposed to third and fourth larval 200 instars (Table 1)  P<0.05). In the presence of fifth instar larvae, the parasitoid spent significantly more time walking, 208 than in the presence of second instar larvae (Table 1) (S1). Host developmental instar had a 209 significant impact on the time spent resting (F=5.50, DF=4, P<0.01). Parasitoids exposed to third, 210 fourth and fifth instar larvae spent less time resting than those exposed to second instar larvae 211 (Table 1) (S1). 212 The average number of host-feeding behaviors was very low; for each developmental instar 213 about 1 out of 200 larvae were killed and then eaten by a parasitoid for an average duration of 214 0.10% of the total experimental time. No significant differences between developmental instars 215 were observed for the occurrence (X²=0.19, DF=4, P=0.10) or the duration of host feeding (F=0.72, 216 DF=4, P=0.58) (Table 1) (S1). Grooming accounted for an important part of the behavior expressed 217 by the parasitoid over thirty minutes, and was expressed in a similar proportion in all developmental 218 instars tested (around 42%) (F=0.90, DF=4, P=0.47) (Table 1) (S1). 219

The influence of host development on parasitism and offspring quality 220
The average number of mummies was significantly different between developmental instars 221 (X²=111.22, DF=4, P<0.001), with a higher mean number of mummies produced when parasitizing 222 third-and fourth instar psyllids than first-and second instar psyllids (Table 2) (S1). Host suitability 223 was significantly influenced by psyllid developmental instar (F=8.50, DF=4, P<0.001), with a lower 224 suitability of first instar hosts than third instars hosts. Fifth instars showed a null suitability as they 225 10 produced no mummies (Table 2) (S1). The emergence rate did not vary significantly with host 226 developmental instar (χ²=44.463, DF=3, P>0.05). Of the 162 mummies obtained, 155 resulted in 227 emergence. All mummies resulted in emergence when development occurred in the third instar 228 (72/72), whereas when development of parasitoids occurred in first and fourth instar larvea, two 229 mummies did not emerge (2/14 and 2/60, respectively) and for development in second instar larvae, 230 three mummies did not emerge (3/16). 231 Parasitoids emerging from the first, second, and third instar larvae had a balanced sex ratio 232 .48, P>0.2, respectively), whereas individuals emerging from the 233 fourth instar larvae had a sex ratio that was largely skewed in favor of females (37 females for 8 234 males) (χ²=8.52, P<0.01) (Table 2) (S1). Tibia lengths of parasitoids differed significantly between 235 the two sexes, males being smaller than females (0.33 mm vs. 0.35 mm, respectively), regardless of 236 the host developmental instar (F=43.35, DF=1, P<0.001) (Table 2) (S1). Tibia length also varied 237 with host developmental instar (F=3.33, DF=3, P<0.05). Male and female parasitoids from second 238 instar larvae were in average smaller than those developed from other developmental instars (Table  239 2) (S1). No interaction was detected between sex and developmental instar factors (F=0.96, DF=3, 240 P=0.41). 241 No impact of host instar was observed on female egg load, which showed an average of 242 10.94 ± 9.00 mature eggs over all experimental conditions (X²=549.71, DF=3, P>0.2) ( Table 2) 243 (S1). However, a significant correlation between the tibia length and female egg load was observed 244 for females that developed in fourth instar hosts (Spearman's R = 0.50, P <0.001, n=46), but not for 245 females that emerged from the other developmental instars (R = 0.38, P> 0.05, n=9, R = 0.66, P> 246 0.05, n=9, and R = 0.12, P> 0.05, n=35, for instar 1, 2, and 3, respectively) ( Figure 2). The 247 developmental time of parasitoids was significantly different among host instars (F = 36.11, DF = 3, 248 P<0.001). Parasitoid eggs laid in first instar hosts took longer to emerge from the mummies than 249 parasitoid eggs laid in other developmental instars (Table 2) (S1). There was no significant 250 11 difference in development time between sexes (F=0.14, DF=1, P=0.71), and no interaction occurred 251 between the two factors (F=1.52, DF=3, P=0.21) ( According to Armand et al. (1991Armand et al. ( , 1990, and Booth (1992), T. insidiosus tends to oviposit in the 255 fourth and fifth larval instars of the pear psyllid C. pyri, whereas McMullen (1966) suggest that this 256 parasitoid oviposits mainly in the first three larval instars. In our study, we observed that this 257 parasitoid accepted all host instars for oviposition. However, they had a lower acceptance rate for 258 fifth instar hosts, in which no parasitoid could develop, and higher mummy production was found 259 when eggs were laid in third and fourth instar larvae. Parasitoids spent more time resting and less 260 time exploring the patch when exposed to the first two larval instars, and first instar hosts presented 261 the lowest suitability for T. insidiosus of the instars in which mummy development was possible. 262 Second instar hosts received few antennal contacts and as few ovipositor insertions as fifth instar 263 hosts. First and second instar larvae together accounted for only 20 % of the total number of 264 mummies produced in this experiment. In general, low parasitism rates of young host instars are 265 associated with higher mortality of the host larvae, which are more susceptible to oviposition 266 injuries from stinging and/or venom (Colinet et al., 2005). In addition, the mortality rate of young 267 parasitized instars could be high because hosts are more likely to die between successive molts. All 268 parameters taken together, we suggest that third and fourth instar psyllid larvae are the most suitable 269 hosts for the development of T. insidiosus, both qualitatively and quantitatively. 270 We found that Trechnites insidiosus was more motivated to forage for hosts in presence of 271 third, fourth and fifth instar larvae, with more time spent moving and less time spent resting than 272 when exposed to the first and second instar larvae. Clues left by older psyllid larvae (e.g. honeydew, 273 exuviae and chemical volatiles) could stimulate the locomotor activity of the parasitoid and thus 274 increase its probability of finding hosts. This phenomenon was previously observed in the encytrid 275 parasitoid Psyllaephagus pistaciae whose searching time, locomotion, antennal drumming, and 12 ovipositor probing behaviors were increased by the presence of pistachio psylla honeydew 277 (Mehrnejad and Copland, 2006). The antochorid predator Orius sauteri tends to forage more and to 278 lay more eggs in the presence of the pear psylla Cacopsylla chinensis honeydew (Ge et al., 2019). 279 Our results suggest that the amount and/or the quality of the clues present in the environment may 280 be important stimulating cues for the parasitoid. Determining the nature of these clues influencing 281 the exploratory behavior of T. insidiosus could be an additional step to unravel the factors driving 282 the interactions between psyllids and parasitoids. 283 We found that third and fourth instar psyllids represent the most suitable hosts for 284 oviposition, as 80 % of the mummies obtained in this experiment resulted from these two 285 developmental instars. Although they are larger and therefore more difficult to handle by the 286 parasitoid than first and second instar hosts, they appear to be the most suitable candidates for the 287 female parasitoid, given the trade-off between the nutritional quality of the host and its behavioral 288 and immune defense capabilities. From a biological control perspective, by parasitizing the fourth 289 instars of C. pyri, T. insidiosus will mainly parasitized psyllid individuals that have escape to all 290 other mortality factors at the end of their developmental cycle and just before the reproduction of 291 the pest occurs. Such a feature gives the parasitoid a potentially important control efficiency on the 292 population dynamics of its host with an immediate impact on the resulting imaginal population, and 293 thus on the next generation of psyllids (Armand et al., 1991). 294 Our results showed a lower attraction and acceptance of T. insidiosus to fifth instar psyllids, 295 as few antennal contacts and ovipositors insertion were performed, and no mummies were 296 produced. Fifth instar larvae are probably too large and too advanced in ontogeny to allow proper 297 development of T. insidiosus. Indeed, advanced larval instars of psyllids are able to escape the 298 parasitoid more easily than the earlier developmental instars (Villagra et al., 2002). Such differences 299 in escape behavior between host instars are commonly reported in aphid-parasitoid interactions, in 300 which more mature hosts also generally have a higher capacity to encapsulate parasitoid eggs 301 (Colinet et al., 2005). For example, it has also been shown that the last instar of the aphid Toxoptera 302 13 citricida exhibits a greater immune response to parasitism than younger instars (Walker and Hoy, 303 2003). The absence of mummies for the fifth instar of psyllids could therefore be explained by a 304 combination of behavioral and immune responses of the psyllid to parasitoid attack (Colinet et al., 305 2005). Fifth instar larvae of C. pyri are therefore completely unsuitable for parasitoid development. 306 Grooming accounts for nearly half of the activity of T. insidiosus, regardless of the host 307 instar encountered. Psyllids, especially larvae, produce large amounts of honeydew (Civolani, 308 2012), which is highly concentrated in sugar (Le Goff et al., 2019). After an ovipositor insertion, 309 residues of honeydew left on the cuticule probably promote bacterial and/or fungal infections on the 310 parasitoid's body. Selection likely favored individuals that spent a lot of time cleaning themselves 311 (legs, ovipositor, antennae), because this behavior may not only help parasitoids to live longer, but 312 also contributes maintain high levels of locomotor activity and host detection ability (Zhukovskaya 313 et al., 2013). For psyllids, high honeydew production could also be a protection against parasitoids. 314 Indeed, it has been observed that the honeydew of the pear psylla Cacopsylla chinensis limits the 315 foraging behavior of its predators and could provide a physical defense for the psyllid (Ge et al., 316 2019). Moreover, T. insidiosus has been observed attempting to oviposit in in honeydew droplets, 317 allowing time for psyllid larvae to escape. When attacking aphids, parasitoids also waste time 318 manipulating and inserting their ovipositors into aphid exuviae (Muratori et al., 2008). Finally, the 319 time T. insidiosus spends grooming itself is time that is not spent searching for a host. An 320 experiment to analyze the behavior of the parasitoid when faced with exuviae of different instars 321 and/or honeydew could be conducted to clarify the role that psyllid waste might play in its defense 322 against parasitoids, in terms of the parasitoid's time budget. 323 Regarding the parasitoid fitness indicators obtained in our experiments, parasitoids 324 distribute their offspring in a balanced sex ratio in the first three host instars, while fourth instar 325 psyllids were chosen by the parasitoid to lay a majority of females. It has already been shown that 326 host size/instar can influence the sex ratio of parasitoid offspring: eggs leading to the emergence of 327 females tend to be laid in large hosts (Bernal et al., 1997;Jervis and Kidd, 1986;Van Den Assem et 328 14 al., 1982). This strategy is consistent with the host size distribution model, which assumes that the 329 amount of resources available for parasitoid development determines its fitness (Charnov, 1976;330 Charnov and Skinner, 1985). Thus, it is more profitable for a female parasitoid to lay female eggs in 331 large hosts that provide more resources (Jervis and Kidd, 1986), so they ultimately have a higher 332 egg load. Our experiments were conducted with solitary females, but it would be interesting to test 333 whether this species produces more males under conditions of competition for large hosts, as 334 predicted by the theory of local mate competition (Hamilton, 1967). 335 The third and fourth larval instars of psyllids produced larger individuals than second 336 instars, probably because these developmental instars exhibit more abundant reserves that allow for 337 better growth of the parasitoid. More surprisingly, females that were laid in first instar hosts 338 appeared to be as large as those that developed in third and fourth instar hosts. One mechanism that 339 would explain these observations is that when an egg is laid in a first instar host, the egg does not 340 begin to develop until the psyllid larva reaches a more advanced development instar (Colinet et al., 341 2005). This hypothesis is supported by the fact that individuals from a first instar host take longer to 342 develop than individuals from other instars. It is also possible that parasitoid larvae develop less 343 rapidly in such a host in order to keep it alive longer, thus ensuring completion of their 344 development. These hypotheses could be confirmed by dissecting second, third and fourth instar 345 larvae that were parasitized during the first development instar, and identifying the developmental 346 instar of the parasitoid. 347 We observed a fairly low egg load in T. insidiosus females regardless of the host instar in 348 which they developed, suggesting that this species is synovogenic and produces eggs throughout its 349 life (approximately 20 days when fed under laboratory conditions (Berthe, 2018)). Furthermore, it is 350 generally observed in parasitoids that the larger the female, the greater the egg load (Jia and Liu, 351 2018). In our study, this relationship was only observed for individuals from four instar hosts, 352 confirming that this larval instar is the most suitable for parasitoid development. Yet, some large 353 females did have no or few mature eggs. Yet, some large females had no or few mature eggs. It is 354 15 possible that a stimulus such as psyllids honeydew, host feeding, or simply the presence of the host, 355 is necessary to stimulate egg production (Aung et al., 2012;Pan et al., 2017). 356 Finally, although host feeding behavior was not affected by the host developmental instar, 357 they may play an important role in the ecology of T. insidiosus and its interaction with the host. 358 Host feeding is the consumption of host fluids exuding from oviposition wounds by the adult 359 female parasitoid (Heimpel and Collier, 1996). This behavior has already been described in other 360 encyrtidae species (Aung et al., 2012) but never in T. insidiosus. The number of host-feeding events 361 that were observed in our experiments was very low, likely because the females we used were fed, 362 hydrated and full of eggs. Therefore, their only optimal foraging strategy under these conditions 363 was probably to lay as many eggs as possible. To better understand under what conditions host-364 feeding behavior is expressed, further experiments should be conducted on fertilized females with 365 poor access to food, and/or with low egg loads. Such an experiment would highlight the trade-off 366 between feeding to replenish their energetic reserves/egg load, and to lay eggs, when a femane meet 367 a psyllid larvae. It is also possible that T. insidiosus is able to discriminate between a parasitized 368 and a non-parasitized host, as is the case with most parasitoid species (van Alphen and Visser, 369 1990). Thus, a female arriving in a patch already visited by a conspecific would be more prone to 370 express host-feeding behavior on a host parasitized by a competitor, and thus decrease the 371 competition pressure encountered by its offspring. However, this hypothesis has yet to be tested. 372 The purpose of this study was to determine key elements in the interaction between pear 373 psyllid C. pyri and the specialist parasitoid T. insidiosus, including the most profitable host instars 374 for its development. We showed that third and fourth instar larvae are the most suitable hosts, both 375 behaviorally and physiologically, for the parasitoid to produce high quantity and quality offspring.