Impact of Gene Flow from Cultivated Beet on Genetic Diversity of Wild Sea Beet Populations

MEC769.fm Page 1733 Wednesday, September 29, 1999 9:43 AM

Molecular Ecology (1999) 8, 1733–1741

BlackwellImpact Science, Ltd of gene ﬂow from cultivated beet on genetic diversity of wild sea beet populations

DETLEF BARTSCH,* MARC LEHNEN,* JANET CLEGG,† MATTHIAS POHL-ORF,* INGOLF SCHUPHAN* and NORMAN C. ELLSTRAND† *Chair of Biology V, Ecology, Ecochemistry and Ecotoxicology, Aachen University of Technology — RWTH Aachen, D-52056 Aachen, Germany, †Department of Botany and Plant Sciences, University of California, Riverside, CA 92521–0124, USA

Abstract Gene flow and introgression from cultivated plants may have important consequences for the conservation of wild plant populations. Cultivated beets (sugar beet, red beet and Swiss chard: Beta vulgaris ssp. vulgaris) are of particular concern because they are cross- compatible with the wild taxon, sea beet (B.vs. ssp. maritima). Cultivated beet seed production areas are sometimes adjacent to sea beet populations; the numbers of flowering individuals in the former typically outnumber those in the populations of the latter. In such situations, gene flow from cultivated beets has the potential to alter the genetic composition of the nearby wild populations. In this study we measured isozyme allele frequencies of 11 polymorphic loci in 26 accessions of cultivated beet, in 20 sea beet accessions growing near a cultivated beet seed production region in northeastern Italy, and 19 wild beet accessions growing far from seed production areas. We found one allele that is specific to sugar beet, relative to other cultivated types, and a second that has a much higher frequency in Swiss chard and red beet than in sugar beet. Both alleles are typically rare in sea beet populations that are distant from seed production areas, but both are common in those that are near the Italian cultivated beet seed production region, supporting the contention that gene flow from the crop to the wild species can be substantial when both grow in proximity. Interestingly, the introgressed populations have higher genetic diversity than those that are isolated from the crop. The crop-to-wild gene flow rates are unknown, as are the fitness consequences of such alleles in the wild. Thus, we are unable to assess the long-term impact of such introgression. However, it is clear that gene flow from a crop to a wild taxon does not necessarily result in a decrease in the genetic diversity of the native plant. Keywords: Beta vulgaris, gene flow, genetic diversity, introgression, plant conservation genetics, genetically modified plants Received 7 March 1999; revision received 15 June 1998; accepted 15 June 1999

Introduction populations, gene flow may lead to significant evolutionary change in the recipient populations (e.g. Anderson 1949). Domesticated plants often have the potential to spon- Crop-to-weed gene flow will have important practical taneously hybridize with those wild relatives that are and economic consequences if it promotes the evolution growing in close proximity (Ellstrand et al. 1999). Such of more aggressive weeds (e.g. Anderson 1949; Barrett hybridization may lead to gene flow: ‘the incorporation 1983). Hybridization with domesticated species has also of genes into the gene pool of one population from one been implicated in the extinction of certain wild crop or more populations’ (Futuyma 1998). If new or locally relatives (e.g. Small 1984; Ellstrand & Elam 1993). Also, rare alleles from the domesticated plants persist in wild because gene flow tends to genetically homogenize populations (reviewed by Slatkin 1987) and because crops Correspondence: D. Bartsch. Fax: +49 (241) 8888 182; E-mail: are typically genetically depauperate compared to their [email protected] wild relatives (Ladizinsky 1985), overwhelming gene flow

MEC769.fm Page 1734 Wednesday, September 29, 1999 9:43 AM

1734 BARTSCH ET AL.

Fig. 1 Geographical distribution of sea beet accessions surveyed in this study (see Table 1).

from crops is expected to deplete genetic diversity in wild natural crosses between sugar beet and sea beet in sugar populations (Ellstrand et al. 1999). beet seed production areas, with sea beet as the pollen Over the last decade, much attention has been focused parent. on crop-to-weed hybridization as a potential avenue for The genus Beta is endemic to the Old World. Cultiv- the escape of crop transgenes into natural populations ated beets have been known for more than 2000 years in the (e.g. Colwell et al. 1985; Dale 1994; Darmency 1994). eastern Mediterranean region (Ford-Lloyd & Williams Although that narrow issue has been the topic of numer- 1975). In Europe, wild B.vs. ssp. maritima is largely a coastal ous theoretical, empirical, and synthetic publications, the taxon, with a wide distribution from the Cape Verde general issue of the population genetic consequences of and Canary Islands in the west, northward along Europe’s the flow of alleles from domesticated plants (whether Atlantic coast to the North and Baltic Seas. It extends genetically engineered or not) to their wild relatives has eastward through the Mediterranean region into Asia received scant attention. where it occurs in Asia Minor, in the central and outer Empirical work has been largely focused in two areas: Asiatic steppes and desert areas as far as western India experiments addressing whether a crop and a wild rel- (Letschert 1993). There is no crossing barrier between ative are able to hybridize under field conditions (e.g. wild and cultivated forms of B. vulgaris (Bartsch & Pohl- Langevin et al. 1990; Arias & Rieseberg 1994; Arriola & Orf 1996). Ellstrand 1996; Darmency et al. 1998), and descriptive This study focuses on one of Europe’s most important studies addressing whether introgression from crops has sugar beet seed production districts in northeastern Italy occurred in populations of adjacent wild relatives (e.g. (Fig. 1). Domesticated beet seed production has been in Oka & Chang 1961; Wendel & Percy 1990; Whitton et al. progress in this region for more than 100 years, with 1997; Bartsch & Ellstrand 1999). However, we are not an intensification since the 1950s. All subspecies of B. aware of any study that addresses how population vulgaris are usually wind pollinated, but insect pollination genetic diversity of natural populations is altered under is also possible. In northeastern Italy, commercial sugar gene flow from a crop relative. beet seeds are produced on 4500 ha, each ha containing Therefore, the aim of this study is to examine how gene approximately 50 000 flowering plants. Furthermore, flow and introgression from cultivated beet (Beta vulgaris small farmers in the region grow red beet and Swiss ssp. vulgaris (DC.) HELM) to sea beet (B.vs. ssp. maritima chard for private seed production, which may be an addi- (L.) ARVANG) has impacted upon the genetic diversity tional source of gene flow. Wild sea beet populations of effected wild populations. To date, no population occur on the nearby coastal plain, sometimes within genetic study has been conducted in this gene flow 1000 m of the cultivated fields. All red beet and Swiss direction in Beta, but there are several reports of gene chard cultivars are diploid, as is the wild sea beet. The flow from wild to cultivar populations. Weed beets have most common varieties of sugar beet are diploid. How- appeared in sugar beet fields in France (Boudry et al. ever, some tetraploid sugar beet cultivar lines are used to 1993; Desplanque et al. 1999) and the UK (Hornsey & breed triploid varieties by crosses with diploids (Barocka Arnold 1979; Ford-Lloyd & Hawkes 1986) as a result of 1985). Previous studies demonstrated that genetically

MEC769.fm Page 1735 Wednesday, September 29, 1999 9:43 AM

GENE FLOW FROM CULTIVATED BEET TO SEA BEET 1735

Table 1a Cultivated Beta vulgaris accessions surveyed in this study

No. Taxon Variety/Type Accession Origin Ploidy N

01 B. vulgaris ssp. vulgaris sugar beet USDA FC172 USA 2 15 02 KWS-2N1009 Germany 2 30 03 KWS-Kavetina Germany 3 49 04 KWS-Rizor Germany 2 48 05 KWS-246 Italy 2 39 06 KWS-247 Italy 3 41 07 Betaseed-4035 USA 2 25 08 Betaseed-4581 USA 2 24 09 Betaseed-4776 USA 2 31 10 SES-HH103 USA 2 29 11 Spreckels-IV2R USA 2 30 12 Spreckels-NB2 USA 2 34 13 Spreckels-SS781 USA 2 85 14 Spreckels-VB7R USA 2 26 15 B. vulgaris ssp. vulgaris Swiss chard Dark Green USA 2 70 16 Fordhook USA 2 21 17 Lucullus Germany 2 21 18 Rhubarb USA 2 39 19 Glatter Silber Germany 2 20 20 B. vulgaris ssp. vulgaris Red beet USDA W300C USA 2 32 21 Burpee USA 2 78 22 Detroit Dark Red USA 2 16 23 Red Ball USA 2 25 24 Tall Top USA 2 29 25 Forono Germany 2 20 26 Rubia Germany 2 20

N, number of individuals examined.

based morphological markers that are crop specific occa- Materials and methods sionally occur in the wild populations, revealing that some gene flow occurs in this region from 2.25 × 108 flower- Plant materials ing sugar beets into populations of approximately 4 × 104 flowering wild beets over an area of 4000 km2 (Bartsch & We assayed allozyme diversity in 69 wild and cultivated Schmidt 1997; Bartsch & Brand 1998). Beta accessions. We obtained samples from seed com- We used allozymes to characterize the genetic vari- panies (mostly diploid, but also one triploid, sugar beet ation within accessions of cultivated beet (sugar beet, material), from international plant genetic resource collec- Swiss chard, and red beet), accessions of wild sea beet tions or from collecting directly from wild populations adjacent to the cultivated beet seed production region of (Table 1a,b). Accessions were selected to represent a wide northeastern Italy, and accessions of wild sea beet from geographical range of wild beets (Fig. 1). Precise loca- elsewhere in its distribution (but far from cultivated beet tions (in latitude and longitude) are available on request production areas). We first assessed the genetic relation- from D. Bartsch. We concentrated on selecting cultivated ship of these accessions and groups. We then identified accessions of sugar beet, Swiss chard, and red beet, alleles that are typically common in cultivated beet but because these were the types most likely to hybridize rare in sea beet accessions that are isolated from cultiv- with sea beet in northeastern Italy. ated beet, to permit us to identify which types of cultivated beet contributed to the introgression into the Italian Allozyme electrophoresis sea beet populations. Finally, we compared the genetic diversity of the sea beet populations with a history Fresh leaf material of wild and cultivated beet was of introgression from cultivated beet with the genetic extracted from greenhouse-grown plants. Starch gel elec- diversity of sea beet populations with no history of recent trophoresis was performed on crude protein extracts of contact with cultivated beet. young leaf tissue. Approximately 100 mg of tissue from

MEC769.fm Page 1736 Wednesday, September 29, 1999 9:43 AM

1736 BARTSCH ET AL.

Table 1b Wild Beta accessions surveyed in this study

No. Taxon Accession Origin Location N

27 B. vulgaris BGRC 54228 Ireland (WBN Standard) 26 28 ssp. maritima PI 518398 Ireland Kerry County, Dingle 37 29 (excl. NE Italy) PI 504172 Italy Reggio di Calabria County, Palmi 20 30 PI 504266 France Corsica, Ajaccio 26 31 PI 540575 France Gironde County, Andernons 21 32 PI 540588 France Charante Marit. County, Brouage 30 33 RWTH-31 France Seine-Marit. County, Fecamp 15 34 RWTH-32 France Cotentin County, Utah Beach 15 35 PI 518310 England East Sussex County 48 36 HHU Germany Oldenburg, Botanical Garden 85 37 RWTH–1 Greece Saloniki County, Chalkidiki 07 38 RWTH–2 Greece Heraklion County, Elounda 21 39 RWTH–3 Greece Kalamata County, Peleponnes 20 40 RWTH–4 Egypt Al Iskandariyah County, Alexandria 21 41 RWTH–5 Germany Hamburg County, Helgoland 30 42 RWTH–6 Germany Lübeck County, Fehmarn 30 43 RWTH–7 Germany Cologne County, Warden 30 44 RWTH–8 Denmark Storstrom County, Rodbyhavn 30 45 RWTH–9 Portugal Aveiro County, Aveiro 30 46 B. vulgaris RWTH–011 Italy (NE) Gorizia County, Grado 10 47 ssp. maritima RWTH–012 Italy (NE) Udine County, Auso Corno 5 48 NE Italy RWTH–013 Italy (NE) Venice County, Bibione 5 49 group RWTH–014 Italy (NE) Venice County, Torcello 20 50 RWTH–015 Italy (NE) Venice County, San Erasmo 25 51 RWTH–016 Italy (NE) Venice County, St. Michele 30 52 RWTH–017 Italy (NE) Venice County, Fusina 30 53 RWTH–018 Italy (NE) Venice County, Alberoni 20 54 RWTH–019 Italy (NE) Venice County, Porto Malamocco 20 55 RWTH–020 Italy (NE) Venice County, Pellestrina 60 56 RWTH–021 Italy (NE) Venice County, Chioggia 35 57 RWTH–022 Italy (NE) Rovigo County, Albarella 5 47 58 RWTH–023 Italy (NE) Rovigo County, Albarella 6 10 59 RWTH–024 Italy (NE) Rovigo County, Albarella 7 15 60 RWTH–025 Italy (NE) Rovigo County, Albarella 9 29 61 RWTH–026 Italy (NE) Rovigo County, Porto Levante 35 62 RWTH–027 Italy (NE) Rovigo County, Boccasette 40 63 RWTH–028a Italy (NE) Ravenna County, Cervia 1997 57 64 RWTH–028b Italy (NE) Ravenna County, Cervia 1998 106 65 RWTH–029 Italy (NE) Ancona County, Numana 43

N, number of individuals examined; WBN, World Beta Network, accessions are usually diploid.

each individual was ground in 0.5 mL of extraction buffer we used three different electrophoretic buffer systems: [0.1 m Tris-HCl pH 7, 4% polyvinylpyrrolidone (PVP), Tris–EDTA–borate pH 8.8 (Heywood 1980) for Gdh, Lap, 0.1% dithiothreitol (DTT), and 0.1% ascorbic acid]. Our and Udp, lithium-borate pH 8.0 (Rieseberg & Soltis 1989) nine enzyme systems revealed 12 loci: aspartate amino for Aat, Pgm, and Tpi, and morpholine-citrate pH 7.0 transferase (Aat; E.C. 2.6.1.1), aconitase (Aco; E.C. 4.2.1.3), (O’Malley et al. 1980) for Aco, Mdh, and Skd. Staining glutamate dehydrogenase (Gdh; E.C. 1.4.1.2), leucine techniques are as described by Devlin & Ellstrand (1989) aminopeptidase (Lap; E.C. 3.4.11.1), NAD+ malate dehydro- and Wendel & Weeden (1989). genase (Mdh1, Mdh2; E.C. 1.1.1.37), phosphoglucomutase Genetic interpretations of allozyme variation patterns (Pgm1, Pgm2; E.C. 5.4.2.2), shikimate dehydrogenase (Skd; were based on previously published reports for Beta (Abe E.C. 1.1.1.25), triose phosphate isomerase (Tpi1, Tpi2; & Tsuda 1987; Nagamine et al. 1989; Letschert 1993; E.C. 5.3.1.1), and uridine diphosphoglucose pyropho- Raybould et al. 1996). Loci encoding the less anodally sphorylase (Udp; E.C. 2.4.1.1). To resolve these isozymes, migrating allozyme for each enzyme system were

MEC769.fm Page 1737 Wednesday, September 29, 1999 9:43 AM

GENE FLOW FROM CULTIVATED BEET TO SEA BEET 1737

designated ‘1’, with additional loci numbered sequentially Table 2 Mean allele frequencies for major groups of Beta vulgaris in order of increasing mobility. This nomenclature is and B. macrocarpa (with accession numbers) examined in this identical to Nagamine et al. (1989), but inverted relative to study. Locus/allele designation is described in the Materials and locus/allele designations for allozymes of Letschert (1993). methods Sugar Swiss Red Sea beet Sea beet Data analysis beet chard beet EU/Egypt NE Italy (1–14) (15–19) (20–26) (27–47) (48–65) Standard population genetic parameters were used to estimate genetic polymorphism and population genetic Aat1 –1 0.00 0.00 0.00 0.00 0.03 structure for individual accessions and groups of acces- –2 0.00 0.00 0.00 0.00 0.01 –3 0.54 0.50 0.49 0.42 0.53 sions, including the proportion of polymorphic loci (P), –4 0.00 0.00 0.01 0.18 0.05 the mean number of alleles among all loci (A) and among –5 0.46 0.50 0.50 0.40 0.38 polymorphic loci (AP), the mean number of alleles per Aco1 –1 0.93 0.28 0.27 0.92 0.40 accession Um and the absolute number of alleles per –2 0.07 0.72 0.73 0.08 0.60 accession group U, and Nei’s (1978) genetic diversity (H). Gdh1 –1 0.82 0.95 1.00 0.80 0.82 Genetic distances were computed according to Nei (1978). –2 0.18 0.05 0.00 0.13 0.18 Computations were facilitated by the PC-based program –3 0.00 0.00 0.00 0.07 0.00 Lap1 –1 0.40 0.74 0.75 0.75 0.81 popgene (HTTP://www.ualberta.ca/~fyeh/index.htm). –2 0.60 0.26 0.25 0.25 0.19 Mdh1 –1 0.75 0.62 0.92 0.92 0.86 Results –2 0.14 0.34 0.01 0.03 0.04 –3 0.11 0.04 0.07 0.05 0.10 For the nine enzyme systems, we resolved 12 loci and Mdh2 –1 0.52 0.00 0.01 0.02 0.10 36 alleles (≈ 3.0 alleles per locus). Allele frequencies for –2 0.48 0.99 0.95 0.93 0.89 individual accessions are available on request. A summary –3 0.00 0.01 0.04 0.05 0.01 Pgm1 –1 0.00 0.00 0.00 0.07 0.00 of the loci and alleles resolved in major accession groups –2 0.99 1.00 1.00 0.86 0.98 (taxonomically and geographically sorted) is provided in –3 0.01 0.00 0.00 0.07 0.01 Table 2. One locus (Tpi1) was monomorphic, three loci Pgm2 –1 0.00 0.00 0.00 0.01 0.01 (Aco1, Lap1 and Udp) had only two alleles per locus. –2 0.31 0.02 0.10 0.13 0.06 Three loci (Gdh1, Mdh1, Pgm1) were tri-allelic, and the –3 0.66 0.97 0.90 0.80 0.80 remaining loci were multiallelic, displaying up to ﬁve –4 0.03 0.01 0.00 0.06 0.14 alleles per locus. Skd1 –1 0.05 0.00 0.02 0.06 0.01 –2 0.01 0.00 0.03 0.09 0.00 As a species, our sample of B. vulgaris (wild and –3 0.83 0.96 0.88 0.78 0.71 domesticated) has a moderately high value for H (0.21). –4 0.11 0.04 0.06 0.07 0.28 Averaged across loci, the mean estimated heterozygosity Tpi1 –1 1.00 1.00 1.00 1.00 1.00 for individual groups of B. vulgaris accessions ranged Tpi2 –1 0.00 0.00 0.00 0.05 0.10 from a high of approximately 0.25 for sugar beet to a low –2 0.96 1.00 0.95 0.94 0.79 of 0.16 of red beet (Table 3). Of the cultivated forms, the –3 0.04 0.00 0.05 0.01 0.11 sugar beet is the most polymorphic. This trend was –4 0.00 0.00 0.00 0.00 0.01 Udp1 –1 0.58 0.44 0.75 0.30 0.14 evident in the summary statistics (Table 3) and allelic –2 0.42 0.56 0.25 0.70 0.86 frequencies for groups of B. vulgaris (Table 2). The cultivated forms posses fewer of the total 36 alleles found in B. vulgaris (sugar beet 29, Swiss chard 23, red beet 26) than both sea beet groups (33 each). beet accessions from areas isolated from sugar beet seed production; it is present in only four of the 19 accessions Alleles that characterize major groups of Beta vulgaris examined with a mean frequency of 0.02 over all acces- ssp. vulgaris sions (Table 4). Only one population had the allele in high We found no allele unique to cultivated beets. However, frequency (0.28); not surprisingly, this accession (no. 35) we found one allele (Mdh2–1) that occurs in all sugar beet is from the English Channel region, the region in which cultivars (at frequencies of 0.07–1.00, with a mean of 0.52), sea beet populations were originally sampled in the that was absent in our Swiss chard accessions, and that 18th century for the ancestor of sugar beet (Fischer 1989). was extremely rare in our red beet accessions (in only one In contrast, Mdh2–1 is much more common in our sea of seven accessions at a frequency of 0.07 with a mean of beet accessions from northeastern Italy growing close to 0.01 over all accessions). This allele is quite rare in sea sugar beet seed production; it is present in 12 of the 20

MEC769.fm Page 1738 Wednesday, September 29, 1999 9:43 AM

1738 BARTSCH ET AL.

Table 3 Genetic diversity statistics for major groups of genus Beta

Beta (accession numbers) NNa AAP PHUa U

B. vulgaris (1–65) 65 33.5 1.60 2.17 0.56 0.21 20.7 36 B. vulgaris ssp. vulgaris (1–26) 26 36.9 1.52 2.10 0.54 0.21 19.8 29 Sugar beet (1–14) 14 36.1 1.60 2.11 0.60 0.25 20.8 27 Swiss chard (15–19) 5 34.8 1.51 2.11 0.52 0.18 19.6 23 Red beet (20–26) 7 31.9 1.43 2.08 0.46 0.16 18.6 26 B. vulgaris ssp. maritima (27–63) 39 31.5 1.56 2.18 0.53 0.19 20.3 36 Sea beet Europe/Egypt (27–45) 19 28.5 1.54 2.20 0.50 0.17 20.0 33 Sea beet NE Italy (46–65) 20 32.1 1.58 2.17 0.57 0.21 20.6 33

Abbreviations for gene diversity statistics include N (number of accessions examined per group), Na (mean number of plants sampled per accession), A (mean number of alleles per locus), Ap (mean number of alleles per polymorphic locus), P (proportion of polymorphic loci ), H (Nei’s (1978) estimated heterozygosity), Ua (mean number of alleles per accession) and U (total number of alleles per accession group).

Table 4 Gene flow from cultivated beet to wild beets. Frequency variance of Mdh2–1 (common sugar beet allele) and Aco1–2 Genetic diversity of sea beet: comparison of ‘seed (common Swiss chard/red beet allele) and frequency (F) per production’ accessions with ‘control’ accessions group In terms of Nei’s estimated diversity (H) and the fraction Variety/Wild location Mdh2–1 F (%) Aco1–2 F (%) of polymorphic loci, sea beet accessions from the beet seed production area had a higher diversity than the con- Sugar beet 0.07–1.00 100 0.00–0.28 57 trol sea beet accession (Table 3). The number of alleles per Swiss chard 0.00–0.00 0 0.34–0.97 100 accessions of the seed production wild populations was Red beet 0.00–0.06 14 0.50–1.00 100 also slightly higher with 20.6 alleles compared to 20.0 Sea beet (Europe/Egypt) 0.00–0.28 21 0.00–0.68 53 alleles in the control accessions, leading to an average of Sea beet (NE Italy) 0.00–0.58 65 0.11–1.00 100 1.58 alleles per locus in the sea beets affected by gene flow and 1.54 in the control group, respectively. Only the average number of polymorphic alleles in both groups was in the same range with approximately 2.2 alleles. accessions sampled with a mean frequency of 0.10 over As a group, B. vulgaris included 36 alleles at 12 loci, all accessions. These data support the contention that with an average of 1.6 alleles per locus (2.2 alleles per introgression has occurred from flowering sugar beet polymorphic locus). Seven of these alleles (Aat1–1, Aat1–2, into at least some of the nearby natural populations of Gdh-3, Pgm1–1, Pgm2–1, Tpi2–1, Tpi2–4) were unique to sea beet. sea beet. The diversity of B. vulgaris alleles depends We also found one allele (Aco1–2) that occurs in sub- overall on the wild sea beet subspecies B. vulgaris ssp. stantial frequencies in all examined accessions of Swiss maritima, in which all of the 36 alleles could be found. The chard and red beet (at a frequency of 0.34–1.00 with accessions that comprise the seed production area popu- means of 0.72 in Swiss chard and 0.73 in red beet). This lations showed the highest overall diversity within the B. allele is typically absent or rare in most sugar beet cul- vulgaris group with 33 total alleles/12 loci and a per

tivars (in only eight of the 14 accessions, with a mean of accession average A = 1.6, AP = 2.2 and H = 0.21. The wild 0.07 over all accessions). This allele is quite rare in sea control accession group had slightly less polymorphism

beet accessions from areas isolated from sugar beet seed with 33 alleles/12 loci (A = 1.5, AP = 2.2, H = 0.17). The production; it is present in only 11 of the 19 accessions mean number of alleles per accession was higher for the examined with a mean frequency of 0.08 over all acces- seed production area (20.6 alleles) than the number in sions (Table 4). In contrast, Aco1–2 is much more com- the control sea beet accessions (20.0 alleles). mon in our sea beet accessions from northeastern Italy growing close to the beet seed production region; it is Discussion present in all of the accessions sampled with a mean frequency of 0.60 over all accessions. These data support the We found substantial genetic evidence for gene flow from contention that introgression has occurred from flower- domesticated beet seed production fields into nearby wild ing Swiss chard and/or red beet into the nearby natural sea beet populations in northeastern Italy. Two alleles populations of sea beet. that are common in cultivars, but otherwise typically

GENE FLOW FROM CULTIVATED BEET TO SEA BEET 1739

Table 5 Nei’s genetic identity (I, above the Sea beet (I) Sea beet (II) diagonal) and genetic distance (D, below Sugar beet Swiss chard Red beet (EU/Egypt) (NE Italy) the diagonal) for major groups of cultivated beet (genus Beta) and related wild taxa Sugar beet **** 0.88 0.89 0.89 0.84 Swiss chard 0.13 **** 0.98 0.87 0.91 Red beet 0.12 0.02 **** 0.88 0.90 Sea beet (I) 0.11 0.14 0.13 **** 0.96 Sea beet (II) 0.17 0.10 0.11 0.04 ****

quite rare in wild beets, were found in unusually high contained less variation than the previous one, we might frequencies in the natural populations. Our data are expect an erosion in diversity in the wild populations supported by a previous study (Bartsch & Schmidt 1997) receiving gene flow. But if the new cultivars were well which reported that these wild populations had a sub- differentiated from the prior ones, we might see at least a stantial number of individuals displaying morphological temporary accumulation of different alleles in the wild traits that are common in cultivars but typically rare in populations. the wild subspecies. Third, in this particular system, sea beets have received Contrary to predictions that crop-to-wild gene flow gene flow from both sugar beet and red beet/Swiss chard, should result in decreased genetic variation (Ellstrand as our data have demonstrated. These different groups et al. 1999), we found, for most parameters, a slight of cultivars are genetically distinct (Table 5). A population increase in per-accession genetic variation in our Italian receiving gene flow from two well-differentiated sources wild beet populations compared with their counterparts would be expected to evolve more diversity than one from elsewhere in the range of the wild subspecies. The receiving gene flow from a single source. difference is most profound in Nei’s H, which averaged Finally, the system might not be in equilibrium. Under approximately 30% higher than that of the accessions high levels of unidirectional gene flow from a source growing far from the seed production region. fixed for a novel allele, recipient populations would experi- It is possible that northeastern Italy represents a ‘centre ence a transient increase in variation as that allele accumu- of diversity’ for Beta vulgaris ssp. maritima and that the lated over generations. Fixation of that allele would diversity we observed has nothing to do with gene flow eventually occur if not opposed by natural selection, from the crop. However, it is unusual for a small portion mutation, or gene flow from an alternative source. of a wild taxon’s range (on the order of a few hundred Whether these populations are in equilibrium or not, square kilometres, Table 1) to hold as much as or more we do note that a century of crop-to-wild gene flow has diversity than the vast majority of its range (encompass- had a limited evolutionary effect on the wild popula- ing tens of thousands of square kilometres). Given that tions. They have not been so overwhelmed by gene flow sea beets in northeastern Italy grow so close to over- as to evolve into cultivated beets. Although natural selec- whelming numbers of cross-compatible, wind-pollinated tion undoubtedly plays a role in limiting the establish- crop plants, gene flow is probably the most parsimonious ment of certain domesticated alleles, we acknowledge explanation. that gene flow may be somewhat limited as well. A previ- Indeed, we can suggest several reasons why gene flow ous study has shown that prevailing winds would more from the crop has not lead to the erosion of genetic divers- frequently carry pollen from the wild plants to the crop ity in this particular system. First, although most crops than the other way around (Bartsch & Schmidt 1997). examined have low genetic diversity compared with their Many descriptive genetic studies on other crop species wild relatives (Ladizinsky 1985; Doebley 1989), beet cul- have provided evidence for crop-to-wild introgression tivars typically hold a level of genetic diversity approxim- (reviewed by Doebley 1989; Ellstrand et al. 1999). How- ately equivalent to that of their wild progenitor. In fact, ever, we are not aware of any previous studies that are allozyme diversity of sugar beet cultivars is substantially comparable to this study, that is, any that directly com- higher than that of the wild beet accessions (Table 3). pare levels of genetic diversity in natural populations Thus, if evolution from sugar beet gene flow continued to with a history of hybridization vs. those that have no such equilibrium, we would expect diversity to increase to the history. Such studies on other crops would be valuable to level of the crop. determine the generality of our findings. Second, sea beets in Italy have received gene flow We have evidence that a few of our ‘control’ accessions from many cultivars over the last century as new varieties of sea beet might also be contaminated with gene flow have emerged (Van Geyt et al. 1990). If each new cultivar events from cultivated beet. In particular, accession no. 29

1740 BARTSCH ET AL.

has the highest frequency of the Aco1–2 alleles (0.7) in the J. R. Stander (Betaseed), L. Panella (Fort Collins), R. Whitkus control group and was originally identified as a landrace (UC Riverside), L. Frese (FAL Braunschweig), E. Biancardi (ISC of leaf beet (see HTTP://ars-grin.gov/cgi-bin/npgs/html/ Rovigo), Cancelliere Rosario (Isola Albarella), B. Heinrich (Helgoland), C. Morak, H. Gluth, B. Witte, S. Driessen, T. Mücher, acchtml pL?1399108 (USDA-ARS GRIN database on P. R. Hesse, C. von Soosten (RWTH-Aachen), U. Lansing, PI504172), but our morphological examination showed a M. Lansing (Dorsten), M. Zayed (El-Menoufia University, Egypt). stronger affinity to sea beet. Accession no. 44 has a weed This study was supported by The German Ministry of Science beet origin and is therefore known to have a hybridiza- and Technology (Grant no. 310532 and no. 0310785) and Univer- tion history with sugar beet (D. Bartsch, unpublished data). sity of California DANR Competitive Grant (no. 1997–980069). A history of gene flow might account for the occurence of our cultivar ‘markers’ (Mdh2–1 and Aco1–2) in a few of References our other control accessions. However, the rarity of those common cultivar alleles in the control accessions clearly Abe J, Tsuda CH (1987) Genetic analysis for allozyme variation demonstrates the extraordinary level of gene flow in the in the section Vulgares, genus Beta. Japanese Journal of Breeding, northeastern Italy accessions. 37, 253–261. Sugar beet is an important crop of Europe, North Anderson E (1949) Introgressive Hybridization. John Wiley & Sons, New York. America, the Near East, Chile, and Japan (see HTTP:// Arias DM, Rieseberg LH (1994) Gene flow between cultivated and apps.fao.org/(FAOSTAT statistics database)). Conse- wild sunflowers. Theoretical and Applied Genetics, 89, 655–660. quently, it has become an object of transformation by Arriola PE, Ellstrand NC (1996) Crop-to-weed gene flow in the recombinant DNA technology. For example, two dif- genus Sorghum (Poaceae): spontaneous interspecific hybridization ferent sugar beet cultivars engineered for herbicide between johnsongrass, Sorghum halepense, and crop sorghum, tolerance have been recently deregulated in the United S. bicolor. American Journal of Botany, 83, 1153–1160. States by USDA-APHIS [see HTTP://www.aphis.usda.gov/ Barocka KH (1985) Zucker- und Futterrüben. In: Lehrbuch der Pflanzenzüchtung Landwirtschaftlicher Kulturformen, Bd, 2, bbep/bp/petday.html (USDA-APHIS Current Status Spezieller Teil (eds Fischbeck G, Plarre W, Schuster W), of Petitions)]. There has been concern that transgenic pp. 245–287. Paul Parey, Berlin/Hamburg. traits may cause unwanted effects after they escape via Barrett SCH (1983) Crop mimicry in weeds. Economic Botany, 37, hybridization into sea beet populations (Bartsch & Pohl 255–282. 1996; Bartsch et al. 1996; Dietz-Pfeilstetter & Kirchner Bartsch D, Pohl-Orf M (1996) Ecological aspects of transgenic 1998). Transgenes may be more likely to alter the fitness sugar beet: Transfer and expression of herbicide resistance in 91 of hybrid or introgressed individuals than supposedly hybrids with wild beets. Euphytica, , 55–58. Bartsch D, Schmidt M (1997) Influence of sugar beet breeding on neutral alleles such as allozymes. Therefore, the introgres- populations of Beta vulgaris ssp. maritima in Italy. Journal of sion of transgenes into wild populations may change their Vegetation Science, 8, 81–84. niche relationships (Ellstrand 1992). Given that crop alleles Bartsch D, Brand U (1998) Saline soil condition decreases apparently move with ease into populations of the sea rhizomania infection of Beta vulgaris. Journal of Plant Pathology, beets of northeastern Italy, we suggest that these popula- 80, 219–223. tions should be monitored after seed production of trans- Bartsch D, Ellstrand NC (1999) Genetic evidence for the origin genic beets starts in this region. Our study here provides of Californian wild beets (genus Beta). Theoretical and Applied Genetics, in press. base line data on the current genetic diversity prior to the Bartsch D, Schmidt M, Pohl-Orf M, Haag C, Schuphan I (1996) introduction of genetically modified organisms (GMOs). Competitiveness of transgenic sugar beet resistant to beet It is difficult to judge the ecological genetic impact of a necrotic yellow vein virus and potential impact on wild beet century of gene flow from traditionally bred cultivated beets populations. Molecular Ecology, 5, 199–205. into the sea beet populations of northeastern Italy. We Boudry P, Mörchen M, Saumitou-Laprade P, Vernet P, have not quantified the crop-to-wild gene flow rates in this Van Dijk H (1993) The origin and evolution of weed beets: system, nor have we quantified the fitness consequences consequences for the breeding and release of herbicide- resistant transgenic sugar beets. Theoretical and Applied Genetics, of crop alleles in the wild. Therefore, we do not have the 87, 471–478. parameters to assess the long-term evolutionary impact of Colwell RK, Norse EA, Pimentel D, Sharples FE, Simberloff D unilateral crop gene flow into the wild beet populations (1985) Genetic engineering in agriculture. Science, 229, 111–112. (Ellstrand et al. 1999). However, it is clear that gene flow Dale PJ (1994) The impact of hybrids between genetically from a crop to a wild relative does not necessarily result modified crop plants and their related species: general in a decrease in the genetic diversity of the wild plant. considerations. Molecular Ecology, 3, 31–36. Darmency H (1994) The impact of hybrids between genetically modified crop plants and their related species: introgression Acknowledgements and weediness. Molecular Ecology, 3, 37–40. Darmency H, Lefol E, Fleury A (1998) Spontaneous hybrid- We thank the following for assistance during early phases of this izations between oilseed rapes and wild radish. Molecular project: K. Meyerholz (KWS Einbeck), A. Schröter (KWS Italia), Ecology, 7, 1467–1473.

GENE FLOW FROM CULTIVATED BEET TO SEA BEET 1741

Desplanque B, Boudry P, Broomberg K, Saumitou-Laprade P, Nei M (1978) Estimation of average heterozygosity and genetic Cuguen J, Van Dijk H (1999) Genetic diversity and gene flow distance from a small number of individuals. Genetics, 89, 583– between wild, cultivated and weedy forms of Beta vulgaris 590. L. (Chenopodiaceae), assessed by RFLP and microsatellite O’Malley DM, Wheeler NE, Guries RP (1980) A manual for starch markers. Theoretical and Applied Genetics, 96, 1194–1201. gel electrophoresis. University of Wisconsin Madison, Depart- Devlin B, Ellstrand NC (1989) Transmission genetics of allozyme ment of Forest Staff Paper Series. loci in Raphanus sativus (Brassicaceae): stress dependent non- Oka HI, Chang WT (1961) Hybrid swarms between wild and Mendelian segregation. American Journal of Botany, 76, 40–46. cultivated rice species, Oryza perennis and O. sativa. Evolution, Dietz-Pfeilstetter A, Kirchner M (1998) Analysis of gene inherit- 15, 418–430. ance and expression in hybrids between transgenic sugar beet Raybould AF, Mogg RJ, Clarke RT (1996) The genetic structure of and wild beets. Molecular Ecology, 7, 1693–1700. Beta vulgaris ssp. maritima (sea beet) populations: RFLPs and Doebley J (1989) Isozymic evidence and the evolution of crop allozymes show different patterns of gene flow. Heredity, 77, plants. In: Isozymes in Plant Biology (eds Soltis DE, Soltis PS), 245–250. pp. 87–105. Dioscorides Press, Portland, Oregon, USA. Rieseberg LH, Soltis DE (1989) Assessing the utility of allozyme Ellstrand NC (1992) Gene flow by pollen: implications for plant number for determining ploidal level: evidence from Helianthus conservation genetics. Oikos, 63, 77–86. and Heliomeris (Asteraceae). Aliso, 12, 277–286. Ellstrand NC, Elam DR (1993) Population genetic consequences Slatkin M (1987) Gene flow and the geographic structure of of small population size: implications for plant conservation. natural populations. Science, 236, 787–792. Annual Review of Ecology and Systematics, 24, 217–242. Small E (1984) Hybridization in the domesticated-weed-wild Ellstrand NC, Prentice HC, Hancock JF (1999) Gene flow and complex. In: Plant Biosystematics (ed. Grant WF), pp. 195–210. introgression from domesticated plants into their wild relat- Academic Press, Toronto, Canada. ives. Annual Review of Ecology and Systematics, in press. Van Geyt JPC, Lange W, Oleo M, De Bock ThSM (1990) Natural Fischer HE (1989) Origin of the ‘Weisse Schlesische Rübe’ (white variation within the genus Beta and its possible use for breed- Silesian beet) and resynthesis of sugar beet. Euphytica, 41, 75– ing sugar beet: a review. Euphytica, 49, 57–76. 80. Wendel JF, Percy RG (1990) Allozyme diversity and intro- Ford-Lloyd BV, Hawkes JG (1986) Weed beets, their origin and gression in the Galapagos Islands endemic Gossypium darwinii classification. Acta Horticultura, 82, 399–401. and its relationship to continental G. barbadense. Biochemical Ford-Lloyd BV, Williams JT (1975) A revision of Beta section Systematics and Ecology, 18, 517–528. Vulgares (Chenopodiaceae), with new light on the origin of Wendel JF, Weeden NF (1989) Visualization and interpretation of cultivated beets. Botanical Journal of the Linnean Society, 71, 89– plant allozymes. In: Isozymes in Plant Biology (eds Soltis DE, 102. Soltis PS), pp. 5–45, Dioscorides Press, Portland, Oregon, Futuyma DJ (1998) Evolutionary Biology. 3rd edn. Sinauer Press, USA. Sunderland, MA, USA. Whitton J, Wolf DE, Arias DM, Snow AA, Rieseberg LH (1997) Heywood JS (1980) Genetic correlates of edaphic differentiation and The persistence of cultivar alleles in wild populations of endemism in Gaillardia. PhD Thesis. University of Texas, Austin. sunflowers five generations after hybridization. Theoretical Hornsey KG, Arnold MH (1979) The origin of weed beet. Annals and Applied Genetics, 95, 33–40. of Applied Biology, 92, 279–285. Ladizinsky G (1985) Founder effect in crop-plant evolution. Economic Botany, 39, 191–199. D. Bartsch is a plant ecologist with general interest in plant invas- Langevin S, Clay K, Grace JB (1990) The incidence and effects of iveness and the ecological behaviour of genetically modified hybridization between cultivated rice and its related weed red species. His working group for practical biosafety research on rice (Oryza sativa L.). Evolution, 44, 1000–1008. transgenic organisms includes a plant ecologist (M. Lehnen) and Letschert JPW (1993) Beta section Beta: biogeographical patterns a plant physiologist (M. Pohl-Orf). I. Schuphan is an ecotoxico- of variation and taxonomy. Wageningen Agricultural University logist working on the environmental impact of chemicals and Papers, 93, 1–153. transgenic organisms. J. Clegg is an isozyme analytic specialist, Nagamine T, Catty JP, Ford-Lloyd BV (1989) Phenotypic poly- N. C. Ellstrand is an applied population geneticist interested in morphism and allele differentiation of allozymes in fodder gene flow between wild and cultivated species as well as the role beet, multigerm sugar beet and monogerm sugar beet. Theoret- of gene flow in plant population genetics. ical and Applied Genetics, 77, 711–720.