Testing the fecundity advantage hypothesis with Sitobion avenae, Rhopalosiphum padi, and Schizaphis graminum (Hemiptera: Aphididae) feeding on ten wheat accessions

The fecundity advantage hypothesis suggests that females with a large body size produce more offspring than smaller females. We tested this hypothesis by exploring the correlations between life-history traits of three aphid species feeding on ten wheat accessions at three levels of analysis with respect to the host plant: overall, inter-accession, and intra-accession. We found that fecundity was significantly correlated with mean relative growth rate (MRGR), weight gain, and development time, and that the faster aphid develops the greater body and fecundity, depending on aphid species, wheat accession, and analyses level. Larger aphids of all three species produced more offspring overall; this held true for Sitobion avenae and Schizaphis graminum at the inter-accession level, and for S. avenae, Rhopalosiphum padi, and S. graminum for three, five, and eight accessions respectively at the intra-accession level. Only one correlation, between intrinsic rates of natural increase (rm) and MRGR, was significant for all aphid species at all three analysis levels. A more accurate statement of the fecundity advantage hypothesis is that cereal aphids with greater MRGR generally maintain higher rm on wheat. Our results also provide a method for exploring relationships between individual life-history traits and population dynamics for insects on host plants.

For aphids (Hemiptera: Aphididae), the correlations between reproductive potential (F) and body size or body weight may not be so straightforward 30 . Either F or r m were significantly negatively correlated with DT in each of three clonal lineages of the cotton aphid, Aphid gossypii living on six commercial cotton cultivars 38 and in the pea aphid Acyrthosiphon pisum living on 12 species of legumes 39 . A negative exponential relationship between the number of large embryos and adult weight was found for the green peach aphid Myzus persicae living on the sugar beet Beta vulgaris and potato Solanum tuberosum 40 , and, later, more than 90 aphid species living on 120 different host plant species 41 . However, the black bean aphid A. fabae did not exhibit significant linear correlations between growth rate or body size and reproductive output 42 .
The English grain aphid Sitobion avenae (Fab.), bird cherry-oat aphid Rhopalosiphum padi L., and greenbug aphid Schizaphis graminum (Rondani), are three important pests of wheat [Triticum aestivum (L.); Gramineae] and other cereals worldwide. Rhopalosiphum padi is a polyphagous insect that shows alternation of hosts; its winter hosts are Rosaceae, and its summer hosts are Gramineae 43 . Sitobion avenae and Schizaphis graminum are oligophagous insects and their hosts are mainly Gramineae 44 . All three aphid species have short life cycles and breed readily. Thus, the aphid-wheat system is an ideal biological model with which to study the influence of variations in host resistance to pests and the fecundity advantage hypothesis.
Our previous research estimated life history parameters for these three aphid species feeding on ten wheat accessions with different levels of resistance to aphids, and explored the correlations of five biological parameters among aphid species. We found that the wheat resistance to aphids has effects on the correlations between life-history traits of these three aphid species 17 . In this study, we used the same aphid species and wheat accessions to investigate the effects of wheat pest resistance on seven correlations: between F and DT, WG, and MRGR; between r m and DT, WG, and MRGR; and between WG and DT, all within an aphid species. We analyzed these effects at three levels: overall (all wheat accessions pooled), inter-accession (across accessions), and intra-accession (within an accession). Our goals were to test the fecundity advantage hypothesis; to partition overall aphid-wheat effects into the effects of host plant accession and aphid species on development, size, and population growth of aphids under standard laboratory conditions; and to establish a linkage between individual life-history traits and population dynamics for these insect species.
Data Collection. Our methods of sampling, dissection, and data collection and storage were in accordance to those described by Hu et al. (2013) 17 , using laboratory conditions of 20 ± 0.5 °C (day) and 18 ± 0.5 °C (night), a photoperiod of L16: D8 h, and 70 ± 10% relative humidity. Each combination of aphid species and wheat accession was one set of experiments; there were 30 sets of experiments in all, each with 30-31 replicates. One replicate consisted of a single first instar nymph transferred to a single seedling within 24 hours of birth. Five life-history traits were measured for each aphid individual: development time (DT), measured from birth to adult emergence + 0.5 d; weight gain (WG), where WG = Wa − Wn, and Wa is adult weight within 24 hours of emergence and Wn is the weight of the first instar nymph 24 hours after birth; fecundity (F), the number of offspring produced per female within a time period equal to development time; mean relative growth rate (MRGR), where MRGR = (ln Wa − ln Wn)/DT; intrinsic rate of natural increase (r m ), r m = 0.738 × ln (F)/DT 17,[45][46][47][48] . If any of the five parameters for an individual aphid were missing from the data set, the replicate was excluded. Less than 1% of S. avenae and

Results
Correlations between life-history parameters of aphid species. S. avenae. Table 2 presents the correlation coefficients between parameters of S. avenae at all three analysis levels.
At the overall level, F was significantly positively correlated with WG, MRGR, and DT; r m was significantly positively correlated with WG and MRGR, but negatively correlated with DT; and WG was significantly negatively correlated with DT. At the inter-accession level, F was significantly positively correlated with WG and MRGR; r m was significantly positively correlated with WG and MRGR; and F, r m , and WG were not correlated with DT. At the intra-accession level, there were significant correlations between F and DT, WG, and MRGR for five, three, and zero accessions respectively; there were significant correlations between r m and DT, WG, and MRGR for five, ten, and ten accessions respectively; there were significant correlations between WG and DT for seven accessions.
R. padi. The correlation coefficients between parameters of R. padi are shown in Table 3.
At the overall level, F was significantly positively correlated with DT, WG, and MRGR; r m was significantly positively correlated with WG and MRGR, and negatively correlated with DT; and WG was significantly negatively correlated with DT. At the inter-accession level, F was not significantly correlated with DT, WG, or MRGR; r m was positively correlated with WG and MRGR and significantly negatively correlated with DT; and WG was not significantly correlated with DT. At the intra-accession level, there were significant correlations between F and DT, WG, and MRGR for one, five, and two accessions respectively; there were significant correlations between r m and DT, WG, and MRGR for ten, six, and ten accessions respectively; there were significant correlations between WG and DT for only one accession.
S. graminum. Correlation coefficients between parameters of S. graminum are shown in Table 4.
At the overall level, F was positively correlated with WG and MRGR and significantly negatively correlated with DT; r m was positively correlated with WG and MRGR and significantly negatively correlated with DT; and WG was significantly negatively correlated with DT.  Table 2. Correlation coefficients recorded for S. avenae. Note: '*' indicates the correlation was significant at p < 0.05, '**' indicates the correlation was significant at p < 0.01. 'F' is fecundity, 'DT' is development time, 'WG' is weight gain, 'MRGR' is the mean relative growth rate, 'r m ' is the intrinsic rate of natural increase. The number in parentheses is the partial correlation coefficient controlling for DT. The notations in the following tables are the same. At the inter-accession level, F and r m were both significantly positively correlated with WG and MRGR; and F, r m, and WG were all significantly negatively correlated with DT. At the intra-accession level, there were significant correlations between F and DT, WG, and MRGR for four, seven, and eight accessions respectively; there were significant correlations between r m and DT, WG, and MRGR for all ten accessions; there were significant correlations between WG and DT for all ten accessions.

Comparison of aphid species based on their life-history correlations. Overall. At the overall level,
correlations for all seven life-history parameter pairs were significant for all three aphid species. Scatterplots of these data are shown in Figs 1, 2 and 3. Correlations between F and DT were strongly positive for S. avenae and R. padi, but strongly negative for S. graminum. F was significantly positively correlated with WG and MRGR for all three aphid species (Fig. 1). Correlations for r m were significantly negative with DT, and significantly positive with WG and MRGR (Fig. 2) for all three aphid species. There were significant negative correlations between WG and DT for all three aphid species (Fig. 3).
Inter-accession. Scatterplots for all seven life-history parameter correlations at the inter-accession level are shown in Fig. 4. F was significantly positively correlated with MRGR and WG for S. avenae and S. graminum, but not for R. padi. There was a positive correlation between r m and both WG and MRGR for all three aphid species. The correlations between WG and DT were also significantly negative for all three species. The correlation between F and DT was a strongly negative correlation for S. graminum, not for S. avenae and for R. padi. The correlation between DT and r m was significantly negative for R. padi and S. graminum, but not for S. avenae. Intra-accession. Scatterplots of the correlations between aphid species life-history parameters for each of the ten wheat accessions are shown in Figs 5, 6 and 7, and the appendix table.

Discussion
Correlations between fecundity and other biological parameters. Although F of most insect taxa increases with WG or body size 25-28 , we found that correlations between F and other biological parameters varied depending on aphid species, host wheat accession, the interaction between aphid species and host accession, and the level of the analyses (overall, inter-accession, or intra-accession). Previous work reported significant correlations between F and MRGR for R. padi at the overall level for five host species 49 ; and between F and DT for three A. gossypii clonal lineages across six commercial cotton cultivars 38 ; for A. gossypii, Brevicoryne brassicae (L.), and R. padi feeding on plants treated with sublethal doses of insecticides [50][51][52] ; and for S. avenae feeding on wheat infected with barley yellow dwarf virus 53 . However, the significant correlation we found between F and DT for S. avenae did not agree with what Özder (2002) 54 or Wojciechowicz-Zytko & van Emden (1995) 42 reported. These data indicate that larger aphids produced more offspring at the overall level for all three aphid species. At the inter-accession level, large S. avenae and S. graminum produced more offspring than small individuals did, but large R. padi did not produce more offspring than small R. padi. At the intra-accession level, whether larger aphids produced more offspring depended on the wheat accession on which they fed.

Correlations between r m and other biological parameters. That r m significantly positively correlated
with MRGR in nearly all cases in this study agrees with previous findings for R. padi at the overall level for five host plant species 49 and for A. fabae with V. faba cultivars ' Aquadulce' and 'Relon' though not with seven other cultivars 42 . Our finding that r m is significantly negatively correlated with DT agrees with previous reports for cotton aphid A. gossypii clonal lineages across six commercial cotton cultivars 38 and for the pea aphid A. pisum on 12 species of legumes 39 .
The equations for MRGR and r m both have a denominator of DT. To remove the effect of DT, we calculated the partial correlation coefficients that control for DT. We found that the partial correlation between r m and MRGR was significant for S. avenae, R. padi, and S. graminum for five, five, and six wheat accessions respectively. This means the correlations between r m and MRGR depended on both DT of aphids and wheat accession.
The accessional resistance effect on the correlations. The host plant's resistance to aphids can affect the aphid individual and population traits 10,[14][15][16][17][18][19][20]34,38,46,47 . We found that accessional resistance has influence on the life-history traits of S. graminum. For example, of the ten wheat accessions, the ' Amigo' accession, which has a gene for resistance to S. graminum biotypes B and C 55 , had the lowest nymphal survival, WG, MRGR, F, and r m and longest DT for this aphid 17 . These results are similar to those reported for hypersensitive apple trees that can rapidly necrose tissue at aphid feeding sites (a resistance reaction), which induced lower F and MRGR for the rosy apple aphid Dysaphis plantaginea compared to susceptible apple trees 56 . Accessional resistance did not have broad    influence on the life-history traits of S. avenae and R. padi. For example, '98-10-30' , which is resistant to S. avenae due to a high level of hydroxamic acid 17,57 had the lowest WG and MRGR for this aphid, but its F and r m were not the lowest and DT not the highest. 'Xiaoyan22' has a gene for resistance to R. padi 17 ; the WG was lowest but F was high. The correlations between aphid life-history traits could help define and differentiate the mechanisms of wheat accession resistance to different aphid species.

Conclusion
In summary, the fecundity advantage hypothesis is not supported in the aphid-wheat systems studied. For these aphid species, larger aphids produce more offspring only at the overall level; for S. avenae and S. graminum this is also true at the inter-accession level, but not for R. padi. At the intra-accession level of analysis, we found that the resistance characteristics of wheat accessions significantly affect the correlations between aphid life-history traits that link the individual to the population. A more accurate statement is that aphids that are larger and develop more quickly generally maintain higher population growth rates.
The time period used to determine WG, MRGR, and DT was from nymphae birth to adult emergence, but that used for F and r m was the entire lifespan. Host plants may become weak or die during the experiment in the laboratory, due to lack of fertilizer or constraints on root growth, leading to experimental failure. Based on our results, we conclude that one may use the parameters that can be determined in a short amount of time to calculate parameters that would need more time to be measured directly. For example, we can use WG to calculate r m for S. avenae, use DT to calculate r m for R. padi, and use DT or WG to calculate r m for S. graminum. Our results also provide a method for exploring relationships between individual life-history traits and population dynamics for insects on host plants. Figure 7. (a-j) Correlations between fecundity (F) and mean relative growth rate (MRGR), and (k-t) between intrinsic rates of natural increase (r m ) and MRGR at the intra-accession level within ten wheat accessions.