Ecological impact of a secondary bacterial symbiont on the clones of Sitobion avenae (Fabricius) (Hemiptera: Aphididae)

Many insects harbor heritable endosymbionts, whether obligatory or facultative, and the role of facultative endosymbionts in shaping the phenotype of these species has become increasingly important. However, little is known about whether micro-injected endosymbionts can have any effects on aphid clones, which was measured using various ecological parameters. We examined the effects between symbiotic treatments and the vital life history traits generated by Regiella insecticola on the life table parameters of Sitobion avenae. The results showed that R. insecticola can decrease the intrinsic rate of increase (r), the finite rate of increase (λ) and birth rate and can increase the mean generation times (T) of S. avenae clones, suggesting that R. insecticola may decelerate the normal development of the hosts. No significant differences of these parameters were observed between the examined Sitobion avenae clones, and the symbiont treatment by genotype interaction affected only the net reproduction rate R0, pre-adult duration and total longevity but not the other parameters. Additionally, a population projection showed that R. insecticola decelerated the growth of the S. avenae clones. The evocable effects of R. insecticola on the S. avenae clones may have significant ramifications for the control of S. avenae populations under field/natural conditions.

the life history of R. insecticola in S. avenae using antibiotics to selectively cure infected lines (the elimination of secondary symbionts using 100 μ g ml −1 ampicillin, 50 μ g ml −1 cefotaxime and 50 μ g ml −1 gentomicin). However, a key component of this knowledge is the life table. The life table analysis is the most reliable method for evaluating the survival and reproduction of a population 28 , and some vital ecological parameters are derived from a life table analysis. For instance, the intrinsic rate of increase (r) represents the reproductive potential of a population in relation to its capacity to increase 29 ; the finite rate of increase (λ ) was defined as the rate of increase per individual per unit time; and the net reproductive rate (R 0 ) has been used to summarize the mean number of offspring that an individual can produce during its lifetime, with the mean generation time (T) defined as the period that a population needs to increase its size R 0 -fold 30 . At the same time, the inoculation (by injection or feeding) and elimination of secondary symbionts are complementary methods. It is often best to use both methods to demonstrate an effect of these bacteria, and adding and removing bacteria can present the predicted phenotypic effects. However, elimination of secondary symbionts with antibiotics may generate the concurrent harm to the primary symbiont, Buchnera, on which aphid performance and fecundity depends 15 . Therefore, for a more exact understanding of the impacts of R. insecticola on hosts, it is necessary to obtain S. avenae clones with R. insecticola inoculated via micro-injection, which would have the same genetic background as symbiont-free aphids. Additionally, different aphid species or different genetic backgrounds within one species in the field can harbor various endosymbionts, and it is not clear whether the genotypes of the hosts could impact the ecological effects of endosymbionts. Therefore, the study of the interactive effects between micro-injected endosymbionts and the genotypes of the hosts and the ecological effects of endosymbionts on the aphid population is warranted.
The objectives of the study are to investigate the ecological effects of R. insecticola on different S. avenae clones based on the life table data. The results that we present here might provide new insight for the future systematic study of endosymbionts and its ramifications in suppressing S. avenae populations.

Results
The effects of R. insecticola on population parameters in different aphid genotypes (clones). A two-way nested analysis of variance was used to analyze the effect of endosymbionts on the population parameters in the different genotypes of aphid host (Table 1). R. insecticola significantly impacted the population parameters (i.e., P < 0.001 for the intrinsic rate of increase r, mean generation times T, finite rate of increase λ and net reproductive rate R 0 ). Clone nested in 'R. insecticola' had no impact on r (P = 0.091), T (P = 0.236) and λ (P = 0.090), but significantly impacted R 0 (P = 0.001).
The effects of R. insecticola on the development and longevity of different S. avenae genotypes. A two-way nested analysis of variance was used to analyze the effect of endosymbionts on fitness traits in the different genotypes of the aphid host (Table 3). R. insecticola clearly affected the birth rate (P < 0.001), pre-adult duration (P < 0.001) and total longevity (P = 0.001). Clone nested in 'R. insecticola' had no impact on birth rate (P = 0.091) but had a significant impact on pre-adult duration and total longevity (P = 0.018 and P = 0.004, respectively).
Fitness traits (i.e., birth rates, pre-adult duration, and total longevity) of the controls clones were compared to those corresponding to infected clones (treatments) ( Table 4). R. insecticola had a remarkable effect on birth rate, and birth rates in the treatments Linyi-1, Linyi-2, Neixiang-1 and Neixiang-2 were significantly lower (F = 30.404, P < 0.001) than that in the controls Linyi-NA and Neixiang-NA. No differences in birth rate were found between the treatments (Linyi-1, Linyi-2, Neixiang-1 and Neixiang-2) (P = 0.163). The pre-adult duration of the treatments Linyi-1, Linyi-2, Neixiang-1 and Neixiang-2 (8.32 ± 0.062, 7.51 ± 0.156, 7.80 ± 0.102 and 7.97 ± 0.259, respectively) were clearly longer than that of the controls Linyi-NA (7.03 ± 0.126) and Neixiang-NA Population projection. The simulated population growth and stage structure of the adult stages for 30 days, beginning with an initial population of 10 pairs of 1 st instar nymphs, are shown in Fig. 1. For the adults, after an early stable phase, the times in which the populations entered sustained population growth (outbreak times) for the treatments Linyi-1, Linyi-2, Neixiang-1 and Neixiang-2 occurred on the 17 th , 15 th , 16 th and 16 th day, respectively, and the corresponding controls Linyi-NA and Neixiang-NA both occurred on the 14 th day. This showed that the outbreak times of the treatments were delayed compared to those of the controls, and it suggested that R. insecticola might postpone the timing of the adult populations in the studied S. avenae clones.

Discussion
The diverse infections and functions of secondary endosymbionts of aphids (e.g., A. pisum, Aphis craccivora and Myzus persicae) have been extensively studied in recent decades [5][6][7][8][9]18,[31][32][33][34][35][36] . In addition, it has become clear that secondary endosymbionts have major effects on the ecology of the pea aphid and arthropods in general. For example, H. defensa can protect A. pisum from its natural enemies 1,16,18,24 . However, Łukasik et al. 31 reported that H. defensa has little effect on the resistance to the natural enemies in S. avenae, indicating that secondary endosymbionts might, to some extent, have different effects on related species. Therefore, it is necessary to study the effects of the secondary symbionts on different aphid clones using the life table.
A linear two-way nested analysis of variance (ANOVA) showed that the treatment by genotype interaction did not affect aphid fitness (e.g., r, T, λ and birth rate), though the interaction did affect the net reproductive rate (R 0 ), pre-adult duration and total longevity. For a comprehensive understanding of the treatment by genotype interaction, more studies are needed.
Understanding the entire life history plays a significant role in evaluating the population parameters that are influenced by secondary endosymbionts, and the key component of this analysis is the life table, which can provide a comprehensive description of the development, survival, and fecundity of a population 28 . Wang et al. 27 obtained the "curing of R. insecticola infected lines" through an oral administration of antibiotics, and they studied the change in the developmental time of 1 st to 4 th instar nymphs (DT1-DT4), the total developmental time of nymphs (DT5), and 10 d fecundity. However, their research did not focus on the population parameters. In life table research, the population parameters represent the intrinsic rate of increase r, the net reproductive rate R 0 , the mean generation times T and the finite rate of increase λ. In our study, the main life table parameters researched were the population parameters.
The intrinsic rate of increase (r) of the treatments was significantly lower than that of the controls. The intrinsic rate of increase is a key demographic parameter that has been used to summarize the reproductive potential of an animal population in relation to its capacity to increase 29 . Southwood 37 stated that the intrinsic rate of increase is the most useful life table parameter for comparing the population growth potential under specific climatic and food conditions. Therefore, the negative effect of R. insecticola on the potential growth of S. avenae clones is intriguing. This notion was confirmed by the mean generation times (T) of the treatments, which were shorter than that of the controls. Oliver et al. 25 reported that H. defensa shortened the mean generation time (T) of A. pisum. Although R. insecticola might decelerate the potential growth of S. avenae populations, the finite rates of increase (λ) were all over 1 between the treatments and the controls, indicating that the studied S. avenae population increased continuously. The net reproductive rates (R 0 > 1) between the treatments and the controls also confirmed this observation. However, the finite rates of increase (λ) of the treatments with the secondary symbionts were remarkably less than that of the controls, suggesting that the daily increase of S. avenae clones in the treatment groups was significantly lower than that of the controls. This was confirmed by the birth rates of the treatment groups, which were significantly lower than that of the controls. In conclusion, R. insecticola may reduce the potential growth of S. avenae clones, and this is consistent with the findings of Wang et al. 27 , suggesting a possibility of a cost action to harbor this secondary endosymbiont. It was expected that the secondary endosymbiont would play a role against the infection of fungal pathogens 17 . Compared to the controls, we could conclude that R. insecticola increased the time of pre-adult duration. The results are consistent with the report by Wang et al. 27 . In addition, there was the positive effect of the secondary endosymbionts on the longevity of S. avenae clones compared to that of the Linyi control. However, there was no effect of the secondary endosymbionts on the clones of Neixiang-1, Neixiang-2 and Neixiang-NA, which may be explained by the difference of geographic populations.
The intrinsic rates of increase (r), finite rates of increase (λ), mean generation time (T) and birth rate demonstrated that R. insecticola infection had a negative effect on the natural development of S. avenae clones, and the impact levels differed among different genetic backgrounds. The difference between genotypes might be due to variations in host nutrients. For example, the effect of nitrogen content and protein quality might affect the fecundity of herbivores [38][39][40][41] . Therefore, to accurately and systematically assess the ecological effect of all endosymbionts on aphids, (1) the endosymbiont treatment and genotype interaction should be noted in future studies, (2) all endosymbionts (i.e., primary and secondary), such as S. symbiotica, H. defense and Rickettsia, need to be examined individually, and (3) more donor and recipient clones or species could be used, as only 2 aphid clones were used as donors and 2 aphid clones were recipients in the current study. Such efforts would provide us with a comprehensive perspective that would aid our understanding of the complex symbiotic relationships between aphids and bacterial endosymbionts.
In conclusion, the findings of this study indicate that the interaction effects between R. insecticola and the genotypes of S. avenae clones did not show any changes in fitness traits, although changes were observed with the net reproductive rate (R 0 ), pre-adult duration and total longevity. In addition, the significant reduction of fitness traits (e.g., r, λ and birth rate) and the clear increase of the mean generation times (T) indicated that R. insecticola might have a negative effect on the potential population growth of S. avenae.

Materials and Methods
Aphid sources. Aphid clones (considered genotypes in this paper) used in this study were established from individuals collected from winter wheat plants on private farm land, with the permission of the owners. Clones without R. insecticola were collected in Neixiang (33.03°N, 111.50°E), Henan Province and in Linyi (35.05°N, 118.35°E), Shandong Province in May 2014. Collected clones were genotyped at four microsatellite loci (S91, S24, S30, and Sm10) as described in the Supplementary Information (Tables S1 and S2). R. insecticola donor clones were collected in April 2014 in two quadrats, at a distance of 300 m from each other, in Yangpingguan (34.8°N, 105.6°E), Shaanxi. They were named Yangpingguan 1 and Yangpingguan 2. Each clone was derived from a single parthenogenetic S. avenae female, and the endosymbionts of the four clones were detected. The primers are shown in the Supplementary Information (Table S3). Clones were laboratory-reared at 20 °C under a 12 h light: 12 h dark cycle in separate cubical cages (side length = 25 cm) on fresh wheat (Triticum aestivum L.) seedlings (variety '1376') that were changed at ten-day intervals 32 .

Transfer of R. insecticola and further detection. The micro-injection technique advocated by Chen
& Purcell 42 was used to inoculate R. insecticola into either 2 nd or 3 rd instar S. avenae individuals. To do this, the Yangpingguan clones 1 and 2 (from Shaanxi) that hosted R. insecticola were chosen as the donor clones. Body fluids from the Yangpingguan clones 1 and 2 were transferred into the Linyi (from Shandong) and the Neixiang   clones (from Henan), respectively, resulting in the creation of four new clones (Linyi-1 and Neixiang-1 from Yangpingguan clone 1 donor, Linyi-2 and Neixiang-2 from Yangpingguan clone 2 donor). Therefore, the Linyi-1 and Linyi-2 clones and the Neixiang-1 and Neixiang-2 clones have the same genetic background as the Linyi (Shandong) and the Neixiang clones (Shandong), respectively. The surviving aphids (approximately 20 individuals for clones Linyi-1, Linyi-2, Neixiang-1, and Neixiang-2) after the injection were transferred to fresh wheat plants (one aphid per plant). Twelve hours later, the dead aphids were discarded, and the remaining surviving aphids were reared till they became adults and produce nymphs. Next, one single offspring individual from each clone that tested "positive" was subjected to PCR using PCR primers described in Sandström et al. 9 , Taq DNA polymerase mix, and the following cycle parameters: 94 °C for 7 min and 35 cycles of 94 °C for 30 s, 57 °C for 45 s and 72 °C for 30 s. The resulting PCR products were separated in a 1% agarose gel and visualized under UV after gel staining with ethidium bromide for the presence of R. insecticola and were then placed on T. aestivum and allowed to reproduce for several days. At this time, adults were removed and 5 individuals were tested again for the presence of R. insecticola. One positive clone was then used to conduct the following study. To ensure that R. insecticola densities in S. avenae produced clones approached equilibrium, the original data of the life table were collected after a minimum of 10 generations 5 and statistically analyzed.
Life table study. Linyi-1 and Linyi-2 clones were compared with the Linyi clone control as they all shared a common genetic background. This clone was named Linyi-NA for convenience. Similarly, Neixiang-1 and Neixiang-2 clones were compared with the Neixiang clone control, and the control was named Neixiang-NA for convenience. A preliminary experiment showed that the control clone (no injection) and the negative control (injection of water) showed no difference in the growth and development of the studied S. avenae clones (unpublished data). Experiments were performed using three groups for each individual clone, with a total of 30 aphids in each group. Newly emerged 1 st instar nymphs were individually transferred within 24 h to a 9-cm Petri dish containing a fresh wheat leaf blade to start the experiment. The collection of life table data was initiated approximately 24 h later. Each aphid was observed daily at the same time until all individuals died. All offspring were removed daily, and the wheat leaves were replaced every 2 days. Molting and mortality events, as well as the number of produced nymphs, were recorded.
Life table analysis. The developmental time and daily fecundity of all individuals were analyzed according to an age-stage, two-sex life table 43,44 using the program TWOSEXMS Chart 45 . The intrinsic rate of increase (r) was estimated using an interactive bisection method and the following Euler-Lotka equation ( Experimental design and data analysis. A 2 × 3 factorial randomized in complete blocks with 30 replications was deployed to test R. insecticola on Sitobion avenae clones. All data were analyzed using IBM SPSS version 20. The differences in life history traits produced from the life table were separately subjected to a linear two-way nested analysis of variance (ANOVA) to analyze the effects of 'R. insecticola' and clone (i.e., 'Linyi' and 'Neixiang') nested in 'R. insecticola' . We used Tukey's procedure 49 to compare the differences among treatments, following the description of Sokal & Rohlf 50 . All mean separation tests were performed at P < 0.05. Mean values (± standard errors of the mean) were calculated and used in the above analyses. If needed, data were transformed to meet the required assumptions of normality and homoscedasticity. Graphs were generated using SigmaPlot 12.5.