Evidence for rapid spatiotemporal changes in genetic structure of an alien whitefly during initial invasion.

The sweetpotato whitefly Bemisia tabaci Q species is a recent invader and important pest of agricultural crops in China. This research tested the hypothesis that the Q populations that establish in agricultural fields in northern China each year are derived from multiple secondary introductions and/or local populations that overwinter in greenhouses (the pest cannot survive winters in the field in northern China). Here, we report the evidence that the Q populations in agricultural fields mainly derive from multiple secondary introductions. In addition, the common use of greenhouses during the winter in certain locations in northern China helps increase the genetic diversity and the genetic structure of the pest. The genetic structure information generated from this long-term and large-scale field analysis increases our understanding of B. tabaci Q as an invasive pest and has important implications for B. tabaci Q management.

Various factors, such as invasion history, pest dispersal ability, founder numbers, and passive dispersal by human activity1234, affect the genetic structure and diversity of invasive pests. Introduced populations often experience rapid genetic differentiation and microevolution as they adapt to the new environmental conditions45. Knowledge of the population genetics can be used for predicting invasiveness and the efficacy of alternative control efforts and improving management of invasive species5. Although a substantial body of research exists describing the structure and diversity of an alien species after establishment, data regarding diversification during initial invasion are limited. An important reason for this is that the introduction often remains unrecognised until the alien population becomes very large6. Population genetic structure and diversity can also be readily affected by spatial and temporal factors7891011, and several studies have highlighted the need for long-term and large-scale analyses121314.

The sweetpotato whitefly, Bemisia tabaci (Gennadius), is an important agricultural pest worldwide and is regarded as a species complex comprising at least 28 cryptic species151617. Some of these cryptic species have become invasive where they have been introduced. One species complex member, designated MED and commonly known as B. tabaci biotype Q (hereafter referred to as B. tabaci Q or Q), has been introduced to many countries from its original Mediterranean distribution, including China, over the past decade16. However, the various introduced B. tabaci Q populations have differed in invasive ability18192021 and other biological traits such as those related to virus transmission2223. Genetic analysis of the introduced populations is a promising strategy to discern the microevolutionary and ecological adaptations that underlie the ability of this insect pest to invade agro-ecosystems. Such knowledge is important for the design and optimisation of sustainable pest management strategies1234.

In China, B. tabaci Q was first detected in Yunnan, Beijing, and Henan provinces in 200318. During subsequent years, B. tabaci Q has gradually displaced the previously well-established populations of MEAM1 (commonly known as B. tabaci biotype B) and has been the dominant species in most regions of China since 20081920. Since 2005, the invasion process of B. tabaci Q has been well documented in many locations in Shandong Province in northern China242526, and since this initial detection, we have continuously sampled whitefly populations from numerous Shandong locations24. As is the case for other areas of China, the percentage of Q in B. tabaci populations has gradually increased, and Q has been the dominant species in most locations in Shandong Province since 20082526. B. tabaci Q cannot survive the winter in the field in Shandong27, but can survive in greenhouses, which are commonly used in the area of Shouguang. Despite elimination by low winter temperatures, Q has been present and abundant in the field during every growing season since its initial appearance in the province.

To explain how Q can appear in the field year after year in spite of its inability to overwinter in the field, we hypothesise that the initial populations detected in the field (i.e., from 2006 to 2012) derived from multiple, secondary introductions and/or from populations that overwintered in local greenhouses in Shouguang. The purpose of the present study is to test this hypothesis.

Results

Changes in genetic diversity and in microsatellite allele frequencies

From 2006 to 2012 in Shandong Province, the genetic diversity of B. tabaci Q (i.e. the value of He) increased over time only in one location (Shouguang) (Fig. 1) (R = 0.673, P = 0.024), which is the location where greenhouses are widely used during winter. However, the values of He in the six other locations were not significantly correlated with time (P > 0.05). In addition, the relationship between average He and time was not statistically significant (R = 0.486, P = 0.082) (Fig. S1).

Figure 1

Temporal changes in genetic diversity (He) in the populations of B. tabaci Q in Shouguang, Shandong Province during 2006–2012.

Analysis of genetic structure

Of the 45 collections, 44 exhibited significant departures from Hardy-Weinberg equilibrium and had high and positive mean values of Fis (0.287–0.552), indicating the existence of heterozygote deficiencies (Table 1). With respect to pairwise comparisons between collections, 841 of 990 values of Fst (84.8%) were associated with a significant exact test (Table S1).

Table 1

Characteristics of B. tabaci Q in seven locations in Shandong Province from 2006 to 2012. Note that only four and six of the seven locations were sampled in 2006 and 2007, respectively

Location	Code	Year	Host	N	Na	Ne	Ho	He	Nei	Fis
Liaocheng	06–LC	2006	Eggplant	15	3.60	2.134	0.413	0.439	0.424	0.413
Zibo	06–ZB	2006	Eggplant, cotton	30	4.60	2.733	0.373	0.529	0.520	0.373
Dezhou	06–DZ	2006	Cotton	15	3.20	2.061	0.480	0.407	0.393	0.480
Shouguang	06–SG	2006	Eggplant	14	2.60	1.931	0.386	0.443	0.426	0.386
Jinan	07–JN	2007	Eggplant	15	3.40	2.174	0.453	0.443	0.428	0.453
Liaocheng	07–LC	2007	Eggplant, cotton	30	3.40	2.497	0.490	0.495	0.487	0.490
Zibo	07–ZB	2007	Eggplant, cotton	30	3.00	2.052	0.473	0.433	0.426	0.473
Linyi	07–LY	2006	Eggplant, cucumber	30	3.40	1.866	0.287	0.290	0.285	0.287
Dezhou	07–DZ	2007	Eggplant, cotton	30	3.20	2.026	0.383	0.408	0.402	0.387
Shouguang	07–SG	2007	Cotton	15	4.20	2.319	0.388	0.476	0.459	0.388
Jinan	08–JN	2008	Cotton, Japanese hop	30	3.00	1.964	0.473	0.385	0.378	0.473
Zaozhuang	08–ZZ	2008	Eggplant, cotton	30	3.80	2.256	0.491	0.447	0.439	0.491
Liaocheng	08–LC	2008	Eggplant, cotton	30	3.80	2.345	0.472	0.491	0.483	0.472
Zibo	08–ZB	2008	Eggplant, cotton	30	4.80	2.453	0.527	0.501	0.493	0.527
Linyi	08–LY	2007	Eggplant, cotton	30	2.60	1.769	0.440	0.351	0.345	0.440
Dezhou	08–DZ	2008	Eggplant, cotton	30	3.20	2.007	0.493	0.396	0.389	0.493
Shouguang	08–SG	2008	Eggplant, cotton	30	3.6	2.001	0.440	0.435	0.427	0.440
Jinan	09–JN	2009	Eggplant, cotton	30	4.00	2.136	0.447	0.441	0.433	0.447
Zaozhuang	09–ZZ	2009	Eggplant, cotton	30	4.40	2.221	0.379	0.476	0.468	0.379
Liaocheng	09–LC	2009	Eggplant, cotton	30	3.80	2.434	0.447	0.486	0.478	0.447
Zibo	09–ZB	2009	Eggplant, tomato	30	3.40	2.179	0.552	0.438	0.431	0.552
Linyi	09–LY	2009	Eggplant, cotton	30	3.80	2.069	0.433	0.443	0.436	0.433
Dezhou	09–DZ	2009	Eggplant, Japanese hop	30	3.80	1.947	0.447	0.376	0.370	0.447
Shouguang	09–SG	2009	Eggplant, tomato	30	4.80	2.224	0.473	0.470	0.463	0.473
Jinan	10–JN	2010	Eggplant, cotton	30	4.80	2.283	0.401	0.525	0.516	0.401
Zaozhuang	10–ZZ	2010	Eggplant	15	3.80	2.507	0.421	0.567	0.548	0.421
Liaocheng	10–LC	2010	Eggplant, cotton	30	5.40	2.561	0.398	0.568	0.558	0.398
Zibo	10–ZB	2010	Eggplant, cotton	30	4.80	2.059	0.351	0.425	0.418	0.356
Linyi	10–LY	2010	Eggplant, cotton	29	3.80	2.536	0.402	0.587	0.576	0.400
Dezhou	10–DZ	2010	Eggplant, Japanese hop	30	3.60	2.267	0.420	0.552	0.543	0.420
Shouguang	10–SG	2010	Eggplant, tomato	30	4.40	2.362	0.482	0.566	0.557	0.479
Jinan	11–JN	2011	Eggplant, cotton	30	5.40	2.759	0.347	0.577	0.568	0.347
Zaozhuang	11–ZZ	2011	Eggplant, cotton	29	4.40	2.646	0.371	0.537	0.527	0.371
Liaocheng	11–LC	2011	Eggplant, cotton	29	5.40	2.605	0.489	0.579	0.569	0.489
Zibo	11–ZB	2011	Cotton, Japanese hop	30	5.00	2.572	0.407	0.590	0.580	0.407
Linyi	11–LY	2011	Eggplant, cotton	30	5.40	2.584	0.418	0.504	0.496	0.418
Dezhou	11–DZ	2011	Eggplant, cotton	29	4.40	2.461	0.463	0.510	0.501	0.463
Shouguang	11–SG	2011	Eggplant, tomato	30	5.20	2.214	0.414	0.507	0.498	0.412
Jinan	12–JN	2012	Eggplant, cotton	28	4.60	2.264	0.325	0.444	0.436	0.325
Zaozhuang	12–ZZ	2012	Eggplant, cotton	28	4.20	2.489	0.400	0.546	0.536	0.400
Liaocheng	12–LC	2012	Eggplant, cotton	30	4.40	2.402	0.340	0.503	0.495	0.340
Zibo	12–ZB	2012	Eggplant, cotton	30	4.00	2.072	0.433	0.429	0.422	0.437
Linyi	12–LY	2012	Eggplant, cotton	30	4.00	2.133	0.353	0.437	0.430	0.350
Dezhou	12–DZ	2012	Eggplant, cotton	30	4.60	2.059	0.333	0.464	0.457	0.361
Shouguang	12–SG	2012	Eggplant, cotton	30	5.80	2.735	0.444	0.599	0.589	0.444
Mean ± SD				27.8 ± 5.2	4.11 ± 0.79	2.275 ± 0.253	0.423 ± 0.057	0.478 ± 0.070	0.469 ± 0.069

For each sample, the following are indicated: sampling location, population code, date of collection, host plant, sample size (N), average number of alleles per locus (Na), the effective number of alleles (Ne), the observed heterozygosity (Ho), the expected heterozygosity (He), Nei's expected heterozygosity (Nei), and the estimator of the fixation index (Fis). Significant values for heterozygote deficiency are in bold.

The results of a hierarchical analysis of molecular variance (AMOVA) revealed significant differences according to sample time (Table 2) and indicated no significant genetic difference between locations (Table 2). Further investigation showed a significant difference in the genetic structure of B. tabaci Q as time progressed from 2006 to 2012. Of the 21 values of FCT (difference among groups), only seven were not significant (Table S2).

Table 2

Hierarchical AMOVA table and corresponding values for FCT (difference among groups), FSC (differences among collections within groups), and FST (differences among all collections)

Hierarchical structure	FCT	FSC	FST
Location > Time	−0.00398	0.17006**	0.16676**
Time > Location	0.08570**	0.09948**	0.17666**

*P ≤ 0.05, **P ≤ 0.001.

Analyses using BAPS software identified ten genetic clusters within the 45 collections (Figs. 2 and S2). In each year, 2–4 genetic clusters were observed with samples from the different locations. In 2006, three distinct genetic clusters were present for all four collections that were sampled in that year, but in 2007, the 2006 clustering pattern was replaced by new cluster formations for all four locations. In 2008, the 2007 clusters were replaced by novel clustering patterns in three of six locations, and in 2009, the genetic clusters from 2008 persisted in all locations except Linyi. In 2010, the 2009 clusters were replaced by other clusters at all locations, and in 2011, the 2010 clusters had, once again, been replaced by other clusters at all locations. In 2012, the genetic clusters from 2011 had been replaced by other clusters in four of the seven locations.

Figure 2

B. tabaci Q sampling locations in Shandong Province during 2006–2012.

The color-coded assignment of populations to ten clusters identified by BAPS is shown in Figure S2 (B). In the graph, each color represents one cluster. The sampling locations are grouped by color to indicate which groups are likely to represent distinct populations. The map was generated using GeoMapApp (version 2) (http://www.geomapapp.org/).

Discussion

The specimens we collected during 2006–2012 provide an excellent opportunity to examine the changes in B. tabaci Q genetic diversity and structure. Because B. tabaci Q cannot overwinter in the field in northern China, we hypothesised that the initial populations detected in the field each year derived from multiple, secondary introductions and/or from populations that overwintered in local greenhouses. To test this hypothesis, we analysed the spatiotemporal genetic changes of field populations using 7 years of microsatellite data from samples collected annually at seven locations in Shandong Province, China.

Rapid spatiotemporal changes in genetic diversity

The microsatellite DNA data describing genetic diversity and allele frequency presented here reveal a significant genetic shift at the seven field locations during the initial invasion of the whitefly B. tabaci Q in Shandong Province, China. At most locations, the genetic diversity (He) remained constant from 2006 to 2009, increased substantially from 2010 to 2011, and decreased slightly in 2012 (Table 1). The values of microsatellite allele frequencies and average number of alleles per locus (Na) were consistent with the trends in He.

The changes in genetic variation in B. tabaci Q collections were unexpected. We expected that shortly after populations became established, those in different locations in Shandong Province would experience selection due to insecticide exposure and wintering in the greenhouse, and that these selective forces would reduce genetic diversity. In support of this expectation, Chu et al. (2008)28 found that selection by the insecticide thiamethoxam decreased the genetic diversity of B. tabaci under laboratory conditions. Franklin et al. (2010)10 reported that heterozygosity levels in greenhouse populations of Trichoplusia ni were slightly lower in the spring than in the summer and fall, perhaps as a result of winter cleanup operations.

The increases in the genetic diversity indices may be associated with the repeated yearly introduction of alien whiteflies from other regions as a consequence of natural dispersal or human activities. Our results also suggest that the number of alien whiteflies introduced was highly variable over time, i.e., the number introduced was small during 2006–2009 and in 2012, but was large during 2010–2011. An increase in genetic diversity caused by new introductions of alien whiteflies, however, could be countered by winter clean up in greenhouses.

An interesting finding in this study was that the location with abundant greenhouses was the only location in which genetic diversity was significantly correlated with time (expressed as years). We infer that populations overwintering in greenhouses may play an important role in increasing the genetic diversity of B. tabaci Q in Shouguang, northern China.

Rapid spatiotemporal changes in genetic structure

The Fst values reveal substantial genetic structure in the collections of B. tabaci Q at the seven locations. These results indicate that gene flow was low between locations and that inbreeding was prevalent at most locations, which is consistent with Tsagkarakou et al. (2007)29, who documented low gene flow between B. tabaci Q populations in Greece that were separated by only a few kilometres. These data suggest that gene flow between whitefly populations may be minimal and that dispersal by B. tabaci Q is usually limited. Low gene flow between populations may contribute to the genetic differentiation of the populations across the region. BAPS analysis revealed the existence of at least ten genetic clusters of B. tabaci Q in Shandong Province.

The most interesting result in our study was the rapid change in population genetic structure, which was significantly associated with time rather than location according to hierarchical AMOVA (Table 2). According to BAPS analysis for the data from 2006 to 2012, specific genetic clusters persisted for only 1–3 years in any one location (Fig. 2), which suggests that genetic composition changed rapidly at those locations. This result is consistent with the genetic changes in B. tabaci B over a short time frame in the Lockyer Valley, Queensland, Australia11. In that study, Dinsdale and colleagues used eight microsatellite loci and detected a significant temporal change in the local genetic composition of B. tabaci B during the growing seasons of 2006–2007.

The mechanism underlying the rapid spatiotemporal changes in genetic structure of B. tabaci Q in Shandong Province is consistent with the hypothesis that the populations detected in the field from 2006 to 2012 were derived from multiple, secondary introductions and from populations that overwintered in local greenhouses. In the present study, host and climate factors affecting the genetic composition were similar across the study area because the specimens were collected from a limited number of host types and within the same region; thus it seems unlikely that host or climate can explain the changes and differences in genetic structure of B. tabaci Q. At the Liaocheng location, for example, the host plants were eggplant and cotton throughout the sampling period (2007–2012) but the genetic structure changed. In support of the inference that there were multiple introductions of the pest, some new genetic clusters suddenly appeared in different locations in the same year, e.g., three new genetic clusters appeared in 2010 or 2011. Local whitefly populations that overwinter in greenhouses may also affect the genetic structure, e.g., the population in Shouguang in 2012 (12–SG), whose genetic cluster differed from those of the other six locations in 2012, may be a mixture of the local population and external whitefly individuals. As a typical location with greenhouses where whiteflies can survive the winter, Shouguang may have numerous local whitefly individuals that survive the winter.

The application of insecticides cannot be ruled out as a factor influencing genetic change because insecticides are widely used in the study region. Prior studies suggest that the application of some insecticides can cause population bottlenecks that enhance genetic differentiation102830.

Implications for B. tabaci Q management and for future research

The finding of rapid genetic changes in field populations of B. tabaci Q during the initial invasion provides useful information concerning the insect's invasion mechanism and its control. First, the results indicate that both secondary introductions from other regions and overwintering individuals from local greenhouses can contribute to the substantial genetic variability, which would provide a genetic basis for the adaptation of B. tabaci Q to new environments. Second, a change in genetic cluster may be associated with a change in biological and ecological characteristics. Understanding the relationship between the genotypes and these characteristics, and the monitoring of genotypes, can help guide management of B. tabaci181920212223. For example, if a new genetic cluster involves a high level of resistance to specific insecticides, the types of insecticides can be adjusted accordingly.

Several questions remain regarding the genetic changes of invasive B. tabaci Q in China. First, we do not know the geographic distribution of the rapid genetic changes of B. tabaci Q populations; additional research is required to determine whether the changes in the genetic clusters in Shandong Province reflect the changes at the national scale. Second, the key factors affecting the spatiotemporal genetic changes remain to be determined, although many ecological factors have been explored or discussed for other species101130. Finally, the potential effects of population genetic changes on the biological characteristics of the populations should be explored. Two important characteristics concern the harbouring of endosymbionts and the transmission of plant viruses. The presence of the endosymbiont Rickettsia in whiteflies enhances whitefly fitness and spread31, and variation in the percentage of invasive B. tabaci Q with secondary endosymbionts was associated with temporal genetic changes in B. tabaci Q in China4. Similarly, the acquisition of certain plant viruses by whiteflies can modify whitefly feeding behaviour32.

Methods

Sample collection

During July to September of 2006–2012, adult B. tabaci were collected from host plants including four field crops (cotton, tomato, eggplant, and cucumber) and one weed (Japanese hop) at seven representative locations (Dezhou, DZ; Liaocheng, LC; Jinan, JN; Shouguang, SG; Zibo, ZB; Zaozhuang, ZZ; and Linyi, LY) in Shandong Province, China (Table 1). The geographical locations are shown using GeoMapApp v. 233 (Fig. 2). Among the seven locations, Shouguang is considered China's main vegetable producer, and the abundant greenhouses in Shouguang are used to produce vegetables throughout the winter (http://en.wikipedia.org/wiki/Shouguang). Each B. tabaci collection involved sampling whiteflies from every second available plant until at least 100 whitefly adults had been collected from each host. The living B. tabaci individuals were stored in 95% ethanol at −20°C until DNA extraction. Four collections were missing (Jinan and Linyi were not sampled in 2006, and Zibo was not sampled in 2006 or 2007), and there were, therefore, 45 rather than 49 collections in total.

DNA extraction, PCR amplification, and species determination

Genomic DNA was extracted from individual females (14–30 individuals) from each collection (one collection was defined as the samples from a single location in a single year), as described by Frohlich et al. (1999)34. The extracted genomic DNA was stored at −20°C and used as template for PCR amplification. Our long-term field survey indicated that only two species from the B. tabaci species complex (B. tabaci B and Q) were represented in Shandong Province3536. B. tabaci species was determined using cleaved amplified polymorphic sequences (CAPS) of the mitochondrial cytochrome oxidase I (mtCOI) gene (about 620 bp), which was amplified with primers (C1-J-2195/R-BQ-2819), as reported by Chu et al. (2011)26. The amplified mtCOI fragment was cleaved by the restriction endonuclease VspI37; mtCOI fragments that can be cut by VspI belong to B. tabaci Q and those that cannot be cut belong to B. tabaci B26.

Microsatellite genotyping

Microsatellite markers have been widely used to analyse the population structure of B. tabaci11263839. In this study, primers for a suite of five microsatellites (BEM6, BEM11, BEM25, BEM31, and BEM37) were used to amplify loci from 1,251 B. tabaci Q individuals collected in Shandong Province during 2006–2012, as described by De Barro et al. (2003)40. Amplification products of the microsatellite loci were run on an ABI 3730xl DNA analyser. Allele size was determined by comparing the mobility of the PCR products to that of the GeneScanTM 400HD size standard (Applied Biosystems).

Data analysis

For each of the 45 microsatellite data sets (seven locations) of B. tabaci Q, the following genetic diversity indexes were calculated using POPGENE v.1.31 software41: observed number of alleles (Na), effective number of alleles (Ne), expected heterozygosity (He), observed heterozygosity (Ho), and Nei's expected heterozygosity (Nei). The fixation index within collection (Fis) was calculated to quantify the inbreeding coefficient within each collection with GENEPOP v.3.442. Conformity to Hardy–Weinberg equilibrium was assessed with exact tests in GENEPOP v.3.4 with Markov chain parameters of 10,000 dememorisation steps, followed by 1,000 batches of 10,000 iterations per batch. Deviations from heterozygosity deficits were assessed using GENEPOP v.3.442.

The fixation index between collections (Fst) was estimated with GENEPOP v.3.442. A hierarchical AMOVA was performed with ARLEQUIN v.3.5 software to compare the values for FCT (difference among groups), FSC (difference among collections within groups), and FST (differences among all collections)43. In this analysis, ‘groups' refers to data grouped according to location or time. AMOVA computations were performed with 10,000 permutations to test for significance.

Genetic clusters within the 45 B. tabaci Q collections and estimated individual coefficients of ancestry were determined using BAPS v.4.14 software44. For estimating individual ancestry coefficients via admixture analyses, the recommended values were used as follows: the number of iterations used to estimate the admixture coefficients for the individuals was 100; the number of reference individuals from each collection was 200; and the number of iterations used to estimate the admixture coefficients for the reference individuals was 20.

Author Contributions

D.C. and Y.J.Z. designed the experiment. D.C. and D.G. performed the experiment. Y.L.T., J.L. and D.F.J. contributed reagents/materials. D.C. and Y.J.Z. wrote the paper.