Identification of the Population Structure of Myzus persicae (Hemiptera: Aphididae) on Peach Trees in China Using Microsatellites.

In this study, we characterized the genetic structure of Myzus persicae (Sulzer) (Hemiptera: Aphididae) populations in China using microsatellites. We expected that these data will reveal the genetic relationships among various populations of M. persicae and will be of value in the development of better methods for pest control. Four hundred sixty individuals from 23 areas over 13 provinces were collected in the early spring of 2010, all from their primary host, Prunus persicae. The markers analyzed were highly polymorphic, as demonstrated by the expected heterozygosity value (He = 0.861) and the Polymorphism Information Content (PIC = 0.847), which indicated that M. persicae maintains a high level of genetic diversity. Analysis of molecular variance revealed an intermediate level of population differentiation among M. persicae populations (F(ST) = 0.1215). Geographic isolation existed among these populations, and, consequently, the genetic structure of the populations was split into a southern group and a northern group divided by the Yangtse River.

Myzus persicae (Sulzer) (Hemiptera: Aphididae) is presumed to be of Asian origin, but it is globally distributed today and represents an economically important pest (Blackman and Eastop 2000). M. persicae is a polyphagous aphid species that feeds on over 400 plant species in 40 different families and is the most important vector of plant viruses, such as Potato leafroll virus (Van Emden et al. 1969, Blackman and Eastop 1984, Van den Heuvel et al. 1994).

Like many aphid species, M. persicae has a complex and variable life cycle. Zhang and Zhong (1983) reported that in China, M. persicae normally reproduces by cyclical parthenogenesis on its primary host, peach, Prunus persicae (L.) Batsch. The organism undergoes several generations of apomictic parthenogenesis followed by a single sexual generation. Mating takes place on the primary host, where the eggs are laid and undergo diapause over the winter. Parthenogenetic females hatch in the spring, and their descendants disperse to secondary host plants, where they undergo many parthenogenetic generations (Blackman and Eastop 2000).

Microsatellites are a class of codominant and hypervariable genetic markers (Jarne and Lagoda 1996) and have considerable potential for analyzing the population structures of aphid species (Llewellyn et al. 2003, Papura et al. 2003, Simon et al. 2003, Vorwerk and Forneck 2006, Cao et al. 2012). Fourteen microsatellite loci in M. persicae were separately described by Sloane et al. (2001). These loci have been used to study population structures by screening for genetic diversity (Wilson et al. 2002, Fenton et al. 1998, 2003, Guillemaud et al. 2003, Vorburger et al. 2003, Fuentes-Contreras et al. 2004, Vorburger 2006, John et al. 2009). Therefore, we expected that these loci could be used to analyze population structures in Chinese M. persicae.

Although M. persicae have long been recognized as one of the most important agricultural pests in China, little has been reported about its genetic diversity on peaches. Previous studies have focused on a few areas, different methods, and genetic variation in aphid populations on different host plants (Yang and Zhang 1999, Han et al. 2009, Liu et al. 2010). Furthermore, documentation of genetic structure and its distribution across the landscape is important to understanding the role of genetic variation in the success or failure of an invasion (Novak and Mack 2005, Lavergne and Molofsky 2007). Therefore, to obtain a thorough understanding of the M. persicae population structure on peaches and to better understand population phylogenetic relationship for forecasting and improving pest control, we analyzed the genetic diversity of various geographical populations of M. persicae on peaches using microsatellite markers.

Materials and Methods

Sampling Strategy

A total of 460 parthenogenetic aphids were collected from 23 areas distributed over 13 provinces in China mainly in early spring (late April to July) of 2010 (Fig. 1; Table 1). Because of the cyclical parthenogenetic life cycle of the aphid, the spring population has a large amount of variation that may decrease in later seasons due to natural selection, artificial selection, or drift (Guillemaud et al. 2003). Therefore, to fully reflect the genetic diversity among populations, sampling in early spring is necessary. Each aphid was sampled from different plants of peach, which were separated from each other by more than 50 m to minimize the risk of collecting organisms of the same clone. In most cases, the distance between any two populations was greater than 50 km. Samples from natural populations were preserved in 95% ethanol and stored at −20°C. Within each population, 20 individuals were sampled for analysis (Table 1).

Fig. 1.

Locations of the 23 populations sampled within China. Population codes are the same as in Table 1. Red line represents the Yagtse River. Blue line represents the Tian Shan Mountain.

Table 1.

Population information and genetic variability estimates based on data from 7 microsatellite loci in 23 populations of M. persicae

Code	Collection site	Latitude (N)/longitude (E)	n	Na	Ne	Ho	He	FIS
GYHX	Huaxi, Guiyang	26° 25' 29.20"/106° 40' 19.75"	20	9.571	4.176	0.493	0.690	0.291*
GYYL	Yongle, Guiyang	26° 35' 52.76"/106° 52' 27.22"	20	7.857	4.443	0.400	0.586	0.322*
GXGL	Guilin, Guangxi	24° 59' 21.65"/110° 51' 0.40"	20	10.571	4.519	0.607	0.774	0.220*
SCCD	Chengdu, Sichuan	30° 32' 29.67"/104° 18' 45.67"	20	7.571	4.031	0.729	0.733	0.006*
HBXG	Xiaogan, Hubei	31° 1' 35.77"/114° 5' 43.01"	20	9.571	5.616	0.664	0.827	0.201*
SXXY	Xianyang, Shanxi	34° 19' 44.27"/108° 44' 32.86"	20	8.286	4.720	0.557	0.774	0.285*
SXCA	Changan, Shanxi	34° 3' 22.59"/109° 3' 55.70"	20	8.571	5.035	0.671	0.776	0.137*
JSJR	Jurong, Jiangsu	31° 46' 9.37"/119° 11' 7.04"	20	8.143	4.404	0.514	0.571	0.102*
SDJZ	Jiaozhou, Shandong	36° 22' 3.69"/119° 57' 27.18"	20	8.286	4.555	0.593	0.755	0.219
SDPD	Pingdu, Shandong	36° 48' 21.65"/119° 36' 56.19"	20	8.000	5.539	0.757	0.812	0.070
HNNL	Ningling, Henan	34° 29' 54.75"/115° 18' 24.84"	20	8.143	4.052	0.664	0.728	0.089
HBSZ	Shenzhou, Hebei	37° 59' 34.74"/115° 31' 54.01"	20	8.143	5.395	0.771	0.801	0.037*
HBBD	Baoding, Hebei	38° 50' 31.03"/115° 7' 59.15"	20	9.857	5.926	0.736	0.801	0.083
HBCL	Changli, Hebei	39° 42' 46.14"/119° 9' 45.73"	20	8.714	5.255	0.771	0.793	0.028
HBLL	Lulong, Hebei	39° 53' 22.34"/118° 54' 58.73"	20	9.286	5.023	0.757	0.786	0.038*
BJCP	Changping, Beijing	40° 17' 49.50"/116° 13' 32.70"	20	9.857	5.974	0.700	0.819	0.149
BJPG	Pinggu, Beijing	40° 8' 3.43"/117° 1' 19.28"	20	9.571	5.879	0.736	0.813	0.097*
GSGL	Gaolan, Gansu	36° 20' 0.83"/103° 56' 49.50"	20	10.857	6.587	0.786	0.841	0.068
GSRS	Renshou, Gansu	36° 4' 34.55"/103° 45' 29.90"	20	10.857	6.866	0.800	0.848	0.058
LNLS	Lvshun, Liaoning	38° 48' 53.90"/121° 13' 2.67"	20	4.429	2.391	0.843	0.579	−0.473
LNWFD	Wafangdian, Liaoning	39° 37' 37.61"/121° 58' 46.57"	20	5.286	3.434	0.836	0.719	−0.168
XJSHZ	Shihezi, Xinjiang	44° 17' 39.43"/85° 51' 14.14"	20	9.429	5.244	0.800	0.795	−0.007
XJKEL	Kuerle, Xinjiang	41° 45' 39.51"/86° 9' 49.90"	20	10.857	6.074	0.629	0.821	0.239*
Means			20	8.770	5.006	0.688	0.758	0.011

n, number of individuals per population; Na, observed number of alleles per locus; Ne, effective number of alleles per locus; observed (Ho) and expected heterozygosity (He). FIS, inbreeding coefficient.

*Significance at the 5% nominal level.

DNA Isolation and PCR Assay

Total genomic DNA was extracted from a single aphid using the cetyltrimethylammonium bromide method (Doyle and Doyle 1987). An UV-visible spectrophotometer (Thermo Scientific NanoDrop 2000c) was used to detect the content and purity of the extracted DNA. All DNA samples were diluted to 40 ng/μl and stored at −20°C for future use.

Seven polymorphic microsatellite loci were used in this study (Table 2). Five of these, M35, M40, M49, M63, and M86, were previously identified in an Australian clonal lineage of M. persicae (Sloane et al. 2001). The additional two loci, myz2 and myz25, were identified in a British clone of M. persicae (G. Malarky, unpublished data). The full details of microsatellite testing and amplification were published in the report by Sloane et al. (2001).

Table 2.

Microsatellite primers and genetic variation among seven microsatellite loci of M. persicae in China

Locus	Repeat motif	Primer sequence (5'–3')	Ta (°C)	Na	Ne	Ho	He	FIS	PIC
M2	(GA)30	H-TGGCGAGAGAGAAGACCTGC	58	23	6.468	0.694	0.846	0.076	0.831
		TCGGAAGACAGAGACATCGAGA
M25	(AG)24	F-GAATCTGGAGAGCGGTTAATGC	55	23	4.278	0.657	0.767	0.010	0.742
		AACCCATCTCACTCGTCAGCC
M35	(AT)9-(AC)22	T-GGCAATAAAGATTAGCGATG	55	22	9.285	0.513	0.893	0.286	0.883
		TGTGTGTATAGATAGGATTTGTG
M40	(AC)17	H-ACACGCATACAAGAATAGGG	55	23	4.967	0.741	0.800	−0.088	0.779
		AGAGGAGGCAGAGGTCAAAC
M49	(AC)31	F-CCCATACATACCTCCAAGAC	49	45	16.155	0.772	0.939	0.033	0.935
		AGAGAGAAAATAGGTTCGTG
M63	(AC)29	T-GATTATGGTGCTCGGTGG	49	40	9.528	0.815	0.896	−0.002	0.887
		GCGGTTTTCTTTGTATTTTCG
M86X	(CA)23	H-TCCACTAAGACCTCAAACAC	55	27	8.567	0.622	0.884	0.173	0.872
		ATTTATTATGTCGTTCCGCC
Mean				29	8.464	0.688	0.861	0.070	0.847

The fluorophore is attached to the 5'-end of the forward primer. F, H, and T represent three different fluorophores. F (FAM) is blue fluorescence, H (HEX) is green fluorescence, and T (TAMRA) is yellow fluorescence. Ta, annealing temperature; Na, number of alleles; Ne, Effective number of alleles; Ho, observed heterozygosity; He, expected heterozygosity; FIS, inbreeding coefficient; PIC, polymorphic information content.

Data Analysis

Our results were interpreted using GeneMapper software (version 4.0) (Applied Biosystems, Foster City, CA), which calculates allele sizes at each microsatellite locus compared to the standard size. For each population and loci, the number of alleles (Na), the effective number of alleles (Ne), the observed heterozygosity (Ho), and the expected heterozygosity (He) were calculated by PopGene32 (version 1.31) (University of Alberta, Calgary, Canada) (Yeh et al. 1999). The Polymorphism Information Content (PIC) for each microsatellite locus was calculated according to Bostein et al. (1980). Arlequin version 3.11 (University of Bern, Bern, Switzerland) (Excoffier et al. 2005) was used to analyze Hardy–Weinberg equilibrium (HWE), linkage disequilibrium, and to calculate FIS values (Weir and Cockerham 1984).

Pairwise genetic distances and FST values were calculated in Arlequin version 3.11. We also used this software for analysis of molecular variance (AMOVA) (Ordonez and Kolmer 2007).

The principal coordinate analysis (PCA) on a genetic distance matrix (Nei’s unbiased genetic distance; Nei 1972, 1978) was analyzed in GENEAlEX V6.1 (Australian National University, Canberra, Australia) (Peakall and Smouse 2006) for all populations. The genetic structure and an estimate of the most likely number of clusters (gene pools) were inferred by the Bayesian clustering method using STRUCTURE 2.0 (Oxford University, Oxford, UK) (Pritchard et al. 2000). The data set was analyzed using the admixture and uncorrelated allele frequencies models and K values 1–14 without incorporating population information. Four independent runs for each K were conducted with 100,000 iterations after a burn-in period of 10,000 iterations in each run.

A phylogenetic tree, based on Nei's unbiased genetic distance using the UPGMA clustering method (an unweighted pair-group method using arithmetical averages), was constructed using the PHYLIP 3.66 software (University of Washington, Seattle, WA) (Felsenstein 2004).

For detecting an association between genetic and geographic distances, Mantel’s test was implemented with 1,000 permutations using the program IBD version 1.5.2 (San Diego State University, San Diego, CA) (Bohonak 2002). The value of FST/(1 – FST) was calculated using Arlequin version 3.11, geographical distance between populations was calculated according to the latitude and longitude of the location of each population, and the natural logarithm of distance was used as the measure of geographical distance to reduce error (Sokal and Rohlf 1995). Alleles in Space (http://www.marksgeneticsoftware.net/) (Miller 2005) package including the Allelic Aggregation Index Analysis (AAIA) tested for nonrandom patterns of allele phenotype diversity across the landscape.

Results

Microsatellite Diversity

The seven microsatellite loci were highly polymorphic, with between 22 and 45 alleles per locus (Table 2) and an average of 29. The observed heterozygosity (Ho) was an average of 0.688 (range: 0.513–0.815). The expected heterozygosity (He) was an average of 0.861 (range: 0.767–0.939). In tests of HWE, we found that with the exception of M40 (FIS = −0.088) and M63 (FIS = −0.002), the loci exhibited a deficiency of heterozygosity (FIS > 0). PIC for each locus had a minimum value of 0.742 for the locus M25 and a maximum value of 0.935 for the locus M49 (Table 2).

Within Population Genetic Diversity

The observed number of alleles per locus for each population ranged from 4.429 (LNLS) to 10.857 (GSGL, GSRS, XJKEL), with an average of 8.770 (Table 1). The average effective number of alleles per locus was 5.006, with a maximum of 6.866 (GSRS) and a minimum of 2.391 (LNLS) (Table 1). The observed (Ho) and expected heterozygosity (He) values ranged from 0.400 (GYYL) to 0.843 (LNLS) (average = 0.688) and from 0.586 (GYYL) to 0.848 (GSRS) (average = 0.758), respectively (Table 1). FIS values ranged from −0.473 (LNLS) to 0.323 (GYYL) (average = 0.011), and only three populations had negative FIS values (LNLS, LNWFD, XJSHZ) (Table 1).

Genetic Differentiation Among Populations

The PCA was able to validate the differentiation that was apparent in the pairwise FST values (Fig. 2). The first two PC axes cumulatively accounted for 63.72% of the total variation. PC axis 1 accounted for 49.00% of the variation; along the first axis, populations from southern China (SCCD, GXGL, GYYL, GYHX and JSJR) were distinct from the majority of the collected samples. Populations were less divergent along PC axis 2, which accounted for 14.72% of the total variation. Along the second axis, a slight differentiation of LNLS, JSJR and XJKEL from other populations was observed.

Fig. 2.

PCA among 23 M. persicae populations.

With the exception of nine pairwise comparisons, all the tests for pairwise genetic differentiation among populations were significant (Table 3). Pairwise FST values ranged from 0.002 (between the GYHX and GYYL populations) to 0.410 (between the LNLS and JSJR populations), suggesting a low genetic differentiation among GYHX and GYYL and a high genetic differentiation among LNLS and JSJR. The genetic distance ranged from 0.061 (between GYHX and GYYL) to 2.873 (between LNLS and JSJR) reinforced the result above.

Table 3.

FST (above diagonal) and Nei's unbiased genetic distance (below diagonal) among 23 populations of M. persicae on peaches in China

	1	2	3	4	5	6	7	8	9	10	11	12	13	14	15	16	17	18	19	20	21	22	23
1		0.002*	0.044	0.125	0.169	0.171	0.186	0.193	0.194	0.169	0.209	0.180	0.174	0.177	0.174	0.153	0.166	0.163	0.158	0.280	0.210	0.203	0.200
2	0.061		0.100	0.204	0.249	0.254	0.264	0.202	0.270	0.245	0.295	0.258	0.260	0.256	0.261	0.236	0.252	0.242	0.238	0.368	0.297	0.275	0.261
3	0.189	0.283		0.058	0.096	0.104	0.123	0.213	0.127	0.108	0.142	0.100	0.104	0.110	0.103	0.087	0.093	0.096	0.095	0.228	0.133	0.138	0.144
4	0.490	0.714	0.281		0.106	0.085	0.105	0.293	0.117	0.082	0.125	0.070	0.070	0.083	0.067	0.064	0.092	0.076	0.077	0.232	0.153	0.118	0.155
5	1.058	1.591	0.647	0.611		0.035	0.055	0.254	0.060	0.050	0.069	0.061	0.034	0.054	0.037	0.043	0.034	0.033	0.028	0.168	0.077	0.072	0.108
6	0.887	1.296	0.591	0.407	0.245		0.017*	0.295	0.027*	0.040	0.061	0.052	0.014*	0.029	0.019	0.021	0.033	0.029	0.025	0.181	0.073	0.099	0.134
7	1.043	1.455	0.743	0.517	0.356	0.143		0.308	0.029	0.036	0.113	0.057	0.049	0.037	0.049	0.033	0.052	0.045	0.045	0.200	0.089	0.116	0.143
8	0.555	0.462	0.819	1.514	1.547	1.968	2.484		0.317	0.283	0.329	0.293	0.297	0.301	0.296	0.274	0.288	0.276	0.270	0.410	0.327	0.293	0.267
9	1.037	1.418	0.717	0.555	0.352	0.170	0.179	2.449		0.033	0.083	0.073	0.039	0.036	0.044	0.042	0.049	0.055	0.052	0.207	0.098	0.103	0.149
10	1.000	1.396	0.713	0.431	0.381	0.261	0.243	2.160	0.210		0.074	0.058	0.034	0.036	0.039	0.028	0.054	0.042	0.041	0.179	0.102	0.094	0.114
11	1.093	1.646	0.757	0.553	0.363	0.288	0.556	2.435	0.368	0.375		0.109	0.022	0.084	0.039	0.074	0.073	0.062	0.055	0.211	0.138	0.124	0.137
12	1.074	1.534	0.618	0.356	0.439	0.316	0.340	2.359	0.401	0.392	0.572		0.052	0.040	0.049	0.040	0.036	0.034	0.043	0.186	0.070	0.092	0.128
13	1.012	1.587	0.649	0.354	0.270	0.142	0.301	2.548	0.234	0.258	0.144	0.345		0.024	0.003*	0.020	0.030	0.007*	0.005*	0.158	0.093	0.077	0.108
14	1.016	1.455	0.680	0.416	0.374	0.199	0.235	2.584	0.217	0.262	0.413	0.270	0.193		0.020	0.012*	0.041	0.028	0.039	0.172	0.093	0.097	0.121
15	0.959	1.481	0.612	0.329	0.269	0.156	0.289	2.180	0.247	0.271	0.206	0.311	0.100	0.171		0.019	0.033	0.016	0.022	0.173	0.104	0.096	0.124
16	0.853	1.301	0.555	0.337	0.344	0.176	0.233	1.962	0.257	0.239	0.385	0.295	0.189	0.151	0.178		0.029	0.031	0.028	0.153	0.087	0.084	0.111
17	0.965	1.530	0.590	0.489	0.282	0.227	0.324	2.388	0.288	0.388	0.372	0.266	0.240	0.285	0.238	0.248		0.030	0.024	0.122	0.056	0.085	0.129
18	1.043	1.565	0.683	0.427	0.308	0.220	0.311	2.420	0.338	0.347	0.334	0.280	0.137	0.236	0.170	0.283	0.271		0.003*	0.159	0.076	0.058	0.096
19	1.009	1.538	0.691	0.436	0.283	0.204	0.314	2.284	0.326	0.348	0.300	0.342	0.129	0.302	0.200	0.272	0.242	0.109		0.141	0.088	0.064	0.097
20	1.162	1.712	0.970	0.886	0.626	0.631	0.757	2.873	0.756	0.680	0.725	0.708	0.540	0.605	0.602	0.528	0.371	0.587	0.483		0.184	0.213	0.248
21	1.060	1.606	0.670	0.715	0.391	0.328	0.405	2.178	0.430	0.524	0.605	0.335	0.455	0.449	0.499	0.443	0.278	0.397	0.477	0.568		0.142	0.179
22	1.383	1.885	0.957	0.636	0.505	0.594	0.742	2.247	0.579	0.645	0.671	0.607	0.492	0.622	0.605	0.580	0.579	0.423	0.477	0.901	0.784		0.102
23	1.524	1.804	1.175	1.090	0.953	1.029	1.167	1.770	1.136	0.959	0.851	1.089	0.835	0.951	0.958	0.949	1.181	0.853	0.895	1.463	1.370	0.752

Nei's genetic distance (below diagonal) and FST (above diagonal); 1: GYHX; 2: GYYL; 3: GXGL; 4: SCCD; 5: HBXG; 6: SXXY; 7: SXCA; 8: JSJR; 9:SDJZ; 10: SDPD; 11: HNNL; 12:HBSZ; 13: HBBD; 14: HBCL; 15: HBLL; 16: BJCP; 17:BJPG; 18: GSGL; 19:GSRS; 20: LNLS; 21: LNWFD; 22: XJSHZ; 23: XJKEL.

*Nonsignificant values (P > 0.05).

From the AMOVA, we found that 12.15% (P < 0.05) of the genetic variation was between populations, and the remaining 79.46% (P < 0.05) of the genetic variation was between individuals, representing the main source of genetic variation. Genetic differentiation estimates reached an FST = 0.1215, suggesting moderate genetic differentiation (Table 4).

Table 4.

AMOVA of microsatellites in 23 M. persicae populations

Source of variation	df	Sum of squares	Variance components	Percentage of variation	Fixation indices	P
Among populations	22	387.972	0.3680	12.15	FST = 0.1215	0.0000
Among individuals within populations	437	1273.775	0.2542	8.39	FIS = 0.0955	0.0000
Within individuals	460	1107	2.4065	79.46	FIT = 0.2054	0.0000
Total	919	2768.747	3.02868

STRUCTURE analysis estimated the log likelihood and K values (Supp. Fig. 1 [online only]) (Evanno et al. 2005). The most probable division is K = 3, identifying three identifiable genetic clusters among the 23 populations (Fig. 3b). Each regional subspecies formed a unique set of clusters.

Fig. 3.

(a) Unrooted UPGMA consensus tree constructed from Nei's unbiased genetic distance depicting the relationships of 23 populations of M. persicae. (b) A population assignment test using STRUCTURE (version 2.3.2) software based on eight microsatellite loci produced from different groups (K = 3). The vertical lines are broken into colored segments showing the proportion of each individual assigned to each of the inferred K. Geographic regions from which the populations belong appear along the x-axis. 1: GYHX; 2: GYYL; 3: GXGL; 4: SCCD; 5: HBXG; 6: SXXY; 7: SXCA; 8: JSJR; 9: SDJZ; 10: SDPD; 11: HNNL; 12: HBSZ; 13: HBBD; 14: HBCL; 15: HBLL; 16: BJCP; 17: BJPG; 18: GSGL; 19: GSRS; 20: LNLS; 21: LNWFD; 22: XJSHZ; 23: XJKEL.

Cluster 1 (M1) contained almost all the lineages from GYHX, GYYL, and JSJR. Cluster 2 (M2) is characterized by the lineages collected from SCCD, XJSHZ, XJKEL, and a large portion of GXGL, HNNL, GSGL, and GSRS were assigned to Cluster 2. All the other populations were assigned to Cluster 3 (M3) (Fig. 3b).

A similar pattern of differentiation among populations was constructed with Nei’s unbiased genetic distance using the UPGMA clustering method (Fig. 3a). The phylogenetic tree showed that four southern Chinese populations (GYYL, GYHX, GXGL, and JSJR) clustered together and then clustered with other populations from north and northwest China.

Isolation by Distance Among Populations

The geographical distance between any two populations ranged from 27.55 to 3,209.23 km. Most of the distances between two populations were greater than 50 km except the distance between GYHX and GYYL, SXXY and SXCA, HBCL and HBLL, GSGL and GSRS. For the Mantel test, the value of FST/(1 − FST) was calculated using Arlequin 3.11; the natural logarithm of distance was used as the measure of geographical distance to reduce error. A pattern of isolation by distance (IBD) was evident by the positive regression between linearized FST and geographic distance (r = 0.2864, P = 0.0110) (Fig. 4). In testing the microsatellite allelic data for spatial patterns due to gene flow, the AAIA test for nonrandom genotypic patterns among populations displayed non-significant evidence of a spatial pattern of allele distribution (spatial aggregation of sampled points: r = 0.7010, P = 0.2210). For the genetic landscape interpolation, the genetic distance between populations from the northwest to the southern existed big variations, whereas from northeast to southeast, the genetic distance tends to be stable although with small fluctuations (Fig. 5).

Fig. 4.

Isolation-by-distance plot of FST/(1 − FST) versus the natural log of geographical distance (km). The solid line represents the best-fit linear regression based on all points.

Fig. 5.

Genetic landscape interpolation of M. persicae. X, Y coordinates are not altitude/longtitude coordinates. Z coordinate is genetic distance.

Discussion

All seven microsatellite loci showed a high polymorphism: Na ranged from 22 to 45, average PIC was 0.847(0.5), and locus M49 showed the greatest variability. This phenomenon was similar to many previous studies (Wilson et al. 2002, Fuentes-Contreras et al. 2004, John et al. 2009) and illustrated that sample collection (Maudet et al. 2002) and loci selection were at a sensible level.

In our study, high levels of polymorphism were identified in all loci within each population. According to Zhang and Zhong (1983), M. persicae in China is cyclically parthenogenetic (holocyclic); therefore, a high genetic diversity, a heterozygote deficit, and a positive FIS value were expected in its primary host the peach. As Wilson et al. (2003) proved, a high genetic diversity may be the result of genetic recombination which has taken place in mating once a year. Delmotte et al. (2002) examined how reproductive mode shapes genetic structure of sexual (cyclically parthenogenetic) and asexual (obligately parthenogenetic) populations of the aphid Rhopalosiphum padi by comparing microsatellite and allozyme data sets. Microsatellites indicated that sexual populations have high allelic polymorphism and heterozygote deficits (possibly because of population subdivision, inbreeding or selection). Wilson et al. (2002) considered that heterozygote excess has been found in asexual lineages is attributed either to ancient loss of sexuality and the consequence of accumulated mutations or to a hybrid origin. We think heterozygosity in aphid may be related to reproduction mode and molecular inheritance marker method. In our study, the Ho values in most populations were smaller than the He values, with the exception of the LNLS, LNWFD, and XJSHZ populations; and the FIS values of these three populations were negative, especially of LNLS, which was strongly negative (−0.473). Many explanations for this phenomenon have been discussed in previous studies, including selection, clonal expansion, the Wahlund effect, inbreeding, and other population effects (Fenton et al. 2003, John et al. 2009). Within the category of selection, natural selection (e.g., selective pressures from temperature, rainfall, and natural predators) and human selection (e.g., insecticides, plant trade, and transport) have been commonly used to explain our observed phenomenon in different aphid species (Delmotte et al. 2002, Fenton et al. 2003, Vorburger 2006). Wahlund effect of sampling from distinct gene pools in the same population may also contribute to the homozygote excess (Fenton et al. 2003, John et al. 2009).

The AMOVA revealed an intermediate level (0.05 < FST ≤ 0.15; Wright 1978) of population differentiation (FST = 0.1215) among populations of M. persicae in China. The results of the STRUCTURE analysis also revealed a moderate level of differentiation among the populations of M. persicae. Similarly, moderate population genetic differentiation was suggested by the UPMGA tree and the PCA test too. Meanwhile, the high within individual and intermediate inter-population genetic diversity of M. persicae reflect some gene flow among populations. The inferred gene flow among some populations based on the FST values revealed were high (Nm values ranging from 0.360 to 124.75) (data not shown).

The pairwise FST was not significant over a small geographic distance (<50 km) and was similar to the results observed by Guillemaud et al. (2003) (<60 km). As an explanation for this phenomenon, Guillemaud et al. discussed previous findings that interpopulation differentiation and aphid species had a clear relationship and that large differences in migration capacities existed between aphid species. In addition, the effects of natural forces on migration should be taken into account. Our primary research demonstrated that southeast monsoons originating from the Pacific Ocean in May strongly influence the migration of the Spiraea aphid (Cao et al. 2012). Similarly, the influence of topographic factors (e.g., mountains and rivers) of population genetic differentiation should be considered.

The Yangtse River divides China’s geography and distinct climates into northern and southern regions. As we know, geographic isolation, such as mountains and rivers, is a major factor contributing to genetic differentiation. They can lead to different geographic populations cannot freely exchanged. The UPMGA phylogenetic tree and the PCA demonstrated a clear divide between a southern and northern group separated by the Yangtse River. In the northern group, the LNLS population, which is seaside, was widely separated from other populations. Compared with XJSHZ population, XJKEL population is further from other northern population. Because, Tian Shan Mountain is the boundary of northern and southern Xinjiang, XJSHZ population in northern Xinjiang is convenient to exchange to other populations. This finding illustrated that the geographic isolation between northern and southern populations as well as differences in the climate between the north and south accelerated genetic differentiation among populations of M. persicae to some extent.

The Bayesian clustering and admixture analysis indicated that each regional subspecies formed a unique set of clusters (or gene pools). The genetic variation partitioned into three clusters. Those indicated that M. persicae populations were moving to three different evolution directions. For those phenomena, migration of aphids should firstly be considered. In addition, to some extent, this effect can be caused by anthropogenic activity (John et al. 2009), such as the trading of plants. M. persicae is ideally suit for this, as its primary host, the peach tree, has been spread throughout the nation and has had an immediate impact on the evolution of pest populations.

In this study, there was an obscure relationship between genetic and geographic distance (r = 0.2864, P = 0.0110). However, our previous research identified a significant correlation between genetic and geographic distance (r = 0.6392, P = 0.0070) (Cao et al. 2012). Andrew et al. (2009) reported that genetic isolation did not significantly correlate with geographic distance (R = 0.02; P = 0.24) in North American populations of Aphis glycines Matsumura. We speculated the possibility that geographic isolation exists among M. persicae populations in China, but isolation by distance could not fully explain our results. Rivers and mountains in China may, to a certain extent, create an effect on the separation of different geographic populations that linear distance did not reflect. For the genetic landscape interpolation, the emergence of genetic distance peak means the appearance of geographic barriers. The big genetic distance variations between populations from the northwest to the southern may be related directly to the complex geological structure locally (such as plateaus, basins and mountainous). The stable genetic distance among east populations indicates extensive gene flow. Factors that influence gene flow, such as transportation, trade, and migration, were also not expressed by the linear distance. And geographic isolation may also correlate with specific aphid species.

Our study was the first to comprehensively analyze the population structure of M. persicae (Sulzer) in China using microsatellite markers. We found that the northern and southern populations had clear genetic differences, but the genetic relationships between pairwise populations were difficult to directly survey and could only be indirectly speculated by the genetic information. Degree to which factors affect population structures is difficult to measure accurately. In future studies, additional molecular markers will be adopted to obtain more accurate and reliable genetic information. Such as mitochondrial DNA markers, so that can make sure the exchange between populations and migratory situation.