The Potential for Crop to Wild Hybridization in Eggplant

American Journal of Botany 102 ( 1 ): 129 – 139 , 2015 .

T HE POTENTIAL FOR CROP TO WILD HYBRIDIZATION IN EGGPLANT ( S OLANUM MELONGENA ; SOLANACEAE) IN SOUTHERN INDIA 1

P RIYA D AVIDAR 2 , A LLISON A . S NOW 3,6 , M UTHU R AJKUMAR 2 , R EMY P ASQUET 4 , M ARIE-CHRISTINE D AUNAY 5 , AND E VANS M UTEGI 3

2 Department of Ecology and Environmental Sciences, Pondicherry University, Kalapet, Pondicherry 605014, India; 3 Department of Evolution, Ecology, and Organismal Biology, Ohio State University, 318 W. 12th Avenue, Columbus, Ohio 43210 USA; 4 IRD, UR 072, LEGS 91198 Gif-sur-yvette, France; Université Paris-Sud 11 91400 Orsay, France; and 5 INRA, Unité de Génétique & Amélioration des Fruits et Légumes, UR1052, Domaine St Maurice, CS 60094 F-84143 Montfavet cedex, France

• Premise of the study: In India and elsewhere, transgenic Bt eggplant (Solanum melongena) has been developed to reduce insect herbivore damage, but published studies of the potential for pollen-mediated, crop- to- wild gene fl ow are scant. This information is useful for risk assessments as well as in situ conservation strategies for wild germplasm. • Methods: In 2010–2014, we surveyed 23 populations of wild/weedy eggplant ( Solanum insanum; known as wild brinjal), carried out hand-pollination experiments, and observed pollinators to assess the potential for crop- to- wild gene fl ow in southern India. • Key results: Wild brinjal is a spiny, low-growing perennial commonly found in disturbed sites such as roadsides, wastelands, and sparsely vegetated areas near villages and agricultural fi elds. Fourteen of the 23 wild populations in our study occurred within 0.5 km of cultivated brinjal and at least nine fl owered in synchrony with the crop. Hand crosses between wild and cul-

tivated brinjal resulted in seed set and viable F1 progeny. Wild brinjal ﬂ owers that were bagged to exclude pollinators did not set fruit, and fruit set from manual self-pollination was low. The exserted stigmas of wild brinjal are likely to promote outcrossing. The most effective pollinators appeared to be bees (Amegilla , Xylocopa , Nomia , and Heterotrigona spp.), which also were observed foraging for pollen on crop brinjal. • Conclusion: Our ﬁ ndings suggest that hybridization is possible between cultivated and wild brinjal in southern India. Thus, as part of the risk assessment process, we assume that transgenes from the crop could spread to wild brinjal populations that occur nearby.

Key words: brinjal; crop–wild hybridization; ﬂ owering times; genetic resources; India; pollinator observations; seed set; Solanum insanum ; transgenic; wild eggplant; wild germplasm.

Worldwide, governmental regulatory agencies typically re- and Wainwright, 2008 ). Many crops are known to exchange quire information about the extent to which a transgenic crop will genes with wild, weedy, or feral relatives, including rice, canola, hybridize with its wild or weedy relatives (e.g., Pilson and squash, sorghum, papaya, sunfl ower, alfalfa, beet, carrot, lettuce, Prendeville, 2004; Marvier and Van Acker, 2005). This informa- creeping bentgrass, and other species (e.g., Ellstrand, 2003 ; tion is used to determine whether transgenes may spread to un- Stewart et al., 2003 ; Ellstrand et al., 2013 ). Indeed, the commer- managed populations of the crop’s wild relatives and, if so, cialization of transgenic crops over the past two decades has whether the transgenes could persist and cause any unwanted triggered a large number of studies on the extent to which each outcomes (e.g., transferring herbicide resistance to agricultural crop species can contribute genes to neighboring populations of weeds; Lu and Snow, 2005 ). In addition, various stakeholders wild/weedy relatives (e.g., Ellstrand et al., 2013 ). When the po- have an interest in knowing whether transgenes are likely to tential for crop-to-wild gene fl ow has been established, further spread from crop to crop, crop to weeds, and crop to wild rela- research has been carried out to determine how novel transgenic tives for legal, ethical, and philosophical reasons (e.g., Mercer traits, such as resistance to insect herbivores or pathogens, affect the fi tness of recipient plant genotypes (e.g., Burke and Rieseberg, 1 Manuscript received 13 September 2014; revision accepted 1 December 2003 ; Snow et al., 2003 ; Sasu et al., 2009 ). 2014. To date, only a few commercialized transgenic crop species The authors thank the Lillian Goldman Charitable Trust for funding this have the potential to exchange genes with neighboring wild, research; H. Ponniah, S. Carr, M. Islam Najar, A. Rahim, and R. Lekshmi weedy, or feral populations. These include canola (Brassica na- for help with the fi eldwork; S. Puyravaud and H. Ponniah for photography; pu s), rice (Oryza sativa ), and squash (Cucurbita pepo ), and in D. Palik for statistical analyses; and J.-P. Puyravaud for creating the map in each case, research was carried out to understand the potential Fig. 1 and other logistical assistance. They also thank D. Palik, H. Chang, for gene fl ow (e.g., Ellstrand et al., 2013 ). The situation is quite and A. Maassen for helpful comments on the manuscript and R. Nayar for different for eggplant (Solanum melongena; Solanaceae). Ef- voucher specimen numbers. This project was undertaken as part of a memorandum of understanding between Pondicherry University and Ohio forts to develop transgenic eggplant in India have been under- State University. way for more than a decade (Padmanaban, 2009), and Bt 6 Author for correspondence (e-mail: [email protected]) eggplant with lepidopteran resistance recently was approved for commercial release in Bangladesh ( Chakraborty, 2014 ). doi:10.3732/ajb.1400404 However, few peer-reviewed papers address the potential for

American Journal of Botany 102 ( 1 ): 129 – 139 , 2015 ; http://www.amjbot.org/ © 2015 Botanical Society of America 129 130 AMERICAN JOURNAL OF BOTANY [Vol. 102 transgenes to spread to unmanaged populations of wild or poricidal anthers that shed pollen profusely when “buzzed” by bees, as is typi- weedy relatives of this crop, which is known locally as brinjal cal of many species in the Solanaceae (e.g., Buchmann and Hurley, 1978 ). In- (this common name is used below). To our knowledge, the dividual plants produce hermaphroditic (bisexual) fl owers with stigmas that extend beyond the anthers, as well as fl owers with a rudimentary ovary that are present study is the fi rst to show that extant populations of wild functionally male (referred to as “male” later). Likewise, in many crop acces- relatives can occur near cultivated brinjal and are likely to hy- sions of brinjal, fl owers at the base of the infl orescence are more likely to set bridize with the crop due to cross pollination by native bees. fruit, whereas more distal fl owers are functionally male (Pradeepa, 2002). The Initially, Bt brinjal was promoted by the United States fruits of wild brinjal are fl eshy, bitter, multiseeded, obovoid or round, and typi- Agency for International Development (USAID) and others for cally become golden orange-yellow at maturity. Fruits can be processed and deployment in India, where yields of cultivated brinjal are often used for food and Ayurvedic medicinal treatments (e.g., Kudlu and Stone, 2013). The seeds are probably dispersed by domestic livestock and wild verte- diminished by lepidopteran damage (Choudhary and Gaur, brates, as well as people. 2009 ; Padmanaban, 2009). The regulatory status of Bt brinjal in India is currently uncertain following a 2010 moratorium on its Study region— Our study focused on the Western Ghats region, which is a release and a great deal of controversy over its safety for human mountainous escarpment covering about 160 000 km2 , running parallel to the consumption and the environment ( Jayaraman, 2010 ; Choudhary west coast of India from about 8 °N to 21 °N latitude. This region has been rec- et al., 2014 ). Before the moratorium, the Indian Report of the ognized as a biodiversity hotspot due to high levels of endemic and threatened Expert Committee ( EC-II, 2009 ) examined the potential for species ( Myers, 1990 ). It represents an important in situ repository of germ- gene fl ow from Bt brinjal to various wild and weedy relatives plasm for cultivated plants, particularly spices, and is considered to be the cen- ter of origin for black pepper, Piper nigrum (Joy et al., 2007; Pandey et al., and concluded that such gene fl ow is unlikely. In addition, the 2008). The elevation of this area ranges from near sea level to ~2500 masl. report stated that “no instances of natural interspecifi c hybrid- Rainfall seasonality increases from south to north, infl uencing vegetation types ization with wild species have been reported for cultivated brin- ( Pascal, 1988 ) and tree diversity ( Davidar et al., 2005 ). The South West “sum- jal” (EC-II, 2009, p. 35). However, the scientifi c literature mer” monsoon brings rain to the southern coastline of Kerala around May and shows that several wild Solanum species have been used as June and then moves northward; it retreats in reverse. Later, the North East or germplasm in breeding programs to improve cultivated brinjal “winter” monsoon brings rain to the eastern fl ank of the Western Ghats during the months of October to December and is restricted to southern areas. (e.g., Venkataramani, 1946 ; references in Mutegi et al., 2015 ), and putative crop–wild hybrids have been observed in the fi eld (Viswanathan, 1975; Karihaloo et al., 1995). Studies of the ge- Distribution, fl owering phenology, and proximity to cultivated brinjal — netic diversity of cultivated and wild brinjal accessions also We studied wild brinjal populations at 23 sites located on private lands within 12 administrative districts and three states of southern India (Table 1, Fig. 1 ). The suggest that crop–wild hybridization may have occurred in the sites were chosen based on accessibility and our goal of including a large geo- past ( Karihaloo et al., 1995 ; Knapp et al., 2013 ; Mutegi et al., graphic area, spanning about 600 km from north to south. Most sites were lo- 2015 , and references therein). cated in dry regions along the eastern fl ank of the escarpment, where annual To determine whether crop–wild gene fl ow is likely to be rainfall is ~700–900 mm and the dry season persists for ~8–9 mo, including a taking place in extant populations, we focused on wild/weedy long dry season from January to June and a short dry season in September and Solanum insanum because it is very closely related to the crop October. Protected areas were not surveyed due to restrictions designed to pro- tect rare and endangered species. Our survey was not intended to be systematic, and is sometimes classifi ed as S. melongena var. insanum (its but our travels throughout the region suggest that the most typical habitats of nomenclature is complicated; e.g., see Lester and Hasan, 1990; wild brinjal are well represented by the 23 study sites. Daunay and Hazra, 2012 ; Knapp et al., 2013 ). In fact, previous We defi ned a population of wild brinjal as a discrete patch of plants that authors have speculated that populations of S. insanum in India were noncontiguous (>1 km) from other such populations. Groups of plants that may represent the crop’s wild progenitor as well as naturalized were <1 km apart are listed as subpopulations in Table 1 and Fig. 1 (e.g., 10a, populations derived from primitive cultivars (e.g., Knapp et al., 10b). Each population was geo-referenced using GPS, and site characteristics were noted ( Table 1 ). Photos were taken of each population, and voucher speci- 2013 ; references in Mutegi et al., 2015 ). For brevity, we refer to mens collected from sites 2, 3, 10, and 19 were deposited at the herbarium of S. insanum as “wild” brinjal, recognizing that these unman- the National Bureau of Plant Genetic Resources in New Delhi (specimen num- aged, free-living populations may represent a continuum of bers 21552, 21553, 21554, 21555). wild, weedy, and possibly feral plants that are not easily distin- To assess the potential for crop–wild hybridization, we visited the sites at guished (Deb, 1989; Karihaloo and Rai, 1995). Additional So- various times between July 2010 and January 2014 to observe fl owering phe- lanum species in southern India also may be able to hybridize nology and record their proximity to cultivated brinjal ( Table 2 ). For the latter, we surveyed an area of approximately 0.5 km around each wild population and with cultivated brinjal (EC-II, 2009; Samuels, 2011), but were recorded whether cultivated brinjal was present. This distance was chosen be- not included here. cause it was practical to record whether cultivated plants were present, keeping The main goals of our research in southern India were to ad- in mind that large bees are capable of foraging over much larger areas (see dress the following questions: (1) Do wild (S. insanum ) and Discussion). The presence of fl owers in both the wild and crop brinjal also was cultivated (S. melongena ) brinjal currently occur near each noted. The timing of our visits was not intended to be comprehensive or sys- other, and if so, do they share overlapping fl owering periods? tematic and refl ected dates when we traveled to collect DNA samples for a subset of 10 sites (Mutegi et al., 2015) at times when wild brinjal was fl owering (2) Is wild brinjal self-compatible, as occurs in cultivated brin- and/or setting fruit. Several populations in the Sigur plateau of Tamil Nadu jal (e.g., Sambandam, 1964)? (3) Does wild brinjal rely on pol- (sites 9, 10, 12, 16, and 19) were visited repeatedly to collect data on pollinator linator visits for seed set? (4) Do the same pollinator taxa visit visits and/or fl ower morphology. fl owers of both wild and cultivated brinjal? (5) Is pollen from cultivated brinjal capable of siring seeds on wild plants in hand- Flower morphology of wild and cultivated brinjal— Flowers of wild and pollination experiments? cultivated brinjal are very similar in appearance, and we hypothesized that small differences in stigma exsertion beyond the anthers could affect outcrossing rates, as occurs in other species (e.g., Georgiady et al., 2002; Takebayashi et al., 2006). We measured stigma exsertion and fl ower diameter for at least 10 MATERIALS AND METHODS plants of wild and cultivated brinjal at each of four locations (Table 3). One freshly opened hermaphroditic fl ower was chosen from each plant. We mea- Study species— Wild brinjal is a spiny, low-growing perennial that occurs sured the length of the stamen from the base of the staminal column to the tip in unshaded, disturbed habitats. Its large, purple fl owers are nectar-less, with of the anther lobes and the length of the style from its base to the stigma to the January 2015] DAVIDAR ET AL.—CROSS POLLINATION IN WILD EGGPLANT 131 ces elds Nearby habitat type/Vegetation season Local habitat Months of dry Tamil Nadu Electricity Board databases in regional offi Nadu Electricity Board databases in regional Tamil Annual rainfall (mm) rainfall Altitude (m a.s.l.) See Fig. 1 for map. Data on annual rainfall and duration of the dry season are approximate, based on data obtained from for map. Data on annual rainfall Fig. 1 See 1. Locations of 23 populations wild brinjal in southern India and site characteristics. Notes: ABLE T Site ID State District Site Latitude N Longitude E 1234 Nadu Tamil 5 Nadu Tamil 6 Tirunelveli Nadu Tamil 7 Tirunelveli Nadu Tamil 8 Kanyakumari Nadu Tamil 9 Kanyakumari Nadu Tamil 10a Sengamal Estate Theni Nadu Tamil 10b Thirukurungudi Sittar Palani Nadu Tamil 11 Kodayar Lower Coimbatore Nadu Tamil Nadu12a Tamil Coimbatore Nadu Tamil 12b 8.21487 Nilgiris Nilgiris13 Nadu Tamil 8.26344 Nilgiris Nadu Perur14 Tamil Foothills Meghamalai 8.30405 Nadu Ayyapulli Karamadai Tamil 15a 77.30322 Nilgiris Nilgiris15b 9.46496 Nadu 77.33384 Tamil Nilgiris16a Nadu 8.28444 Tamil Vazhaithottam 77.18395 Chadapatti Nadu16b Tamil Nilgiris Chadapatti 740 Nadu Tamil 17 77.26343 Nilgiris Nilgiris 127 Nadu Bokkapuram18 Tamil 77.17197 Mavanallah Nilgiris 11.12987 Nadu Tamil 19 273 10.02614 Mavanallah Nilgiris 10.59983 11.3149820 Kerala 560 Nilgiris Singara Road21 900 Karnataka 76.48265 11.32114 77.31342 Chemanatham22 97 800 Karnataka 76.54233 11.32130 76.42138 Masinagudi 140023 11.32436 Karnataka Masinagudi Camp Chamarajanagar Wayanad 76.41555 11.32616 Nadu Tamil Moyar Chamarajanagar 465 900 76.41549 11.32664 375 Karnataka Mangala Moyar ). ( http://www.tnebltd.gov.in Mysore 76.39057 424 11.33794 926 1200 Krishnagiri Karnataka Gundalpet 11.34139 8 11.34416 76.41095 891 76.41229 Tumkur 9 11.34195 Ambalavayal 76.38123 891 6 Shimoga 76.41250 980 800 76.38766 Bagalur 800 893 800 800 9 Nanjangud Wasteland 76.38484 891 6 11.35921 Roadside 800 950 11.39627 Rameshwaram 11.49800 Roadside, village 890 935 11.36034 800 Bhadravati 11.36966 800 935 76.41384 Roadside 800 76.38701 Dump site 9 76.41163 9 estate Private 76.42201 800 Village 76.12866 9 9 800 12.09326 12.49163 13.17183 Irrigated rice fi 800 800 9 1010 920 9 800 793 13.49606 Roadside 920 76.40876 77.48911 9 Roadside 974 77.27016 Dry deciduous Village Wasteland 9 9 Rubber plantation 75.41086 Wasteland 9 700 800 670 Wasteland 900 9 9 925 700 1200 Wasteland 800 scrub Thorny 9 Wasteland Coconut plantation scrub Thorny 600 Wasteland Agricultural land Roadside scrub Thorny Roadside Village 800 800 scrub Thorny 600 9 Agricultural land 9 Village 9 5 scrub Thorny 9 950 scrub Thorny Wasteland Wasteland Roadside, village Wasteland 9 edge Village 9 Roadside Roadside 9 Wasteland edge Village Dry deciduous 8 Roadside Wasteland Roadside Roadside Wasteland Dry deciduous Moist deciduous village Wasteland, scrub Thorny Moist deciduous scrub Thorny scrub Thorny 132 AMERICAN JOURNAL OF BOTANY [Vol. 102

T ABLE 2. Observations of co-occurrence and overlapping ﬂ owering periods of wild and crop brinjal at 23 wild brinjal populations in southern India.

Crop within Simultaneous fl owering Years wild was Months wild was Approx. no. Site ID Site 0.5 km of wild and crop observed fl owering observed fl owering wild plants

1 Sengamal Estate No − 2012, 2013 July, January 6 2 Thirukurungudi YES YES 2012 February 120 3 Sittar No − 2012 September 100 4 Lower Kodayar No − 2012 August 20 5 Meghamalai No − − − 10 6 Ayyapulli YES No 2011 November 16 7 Perur No − 2011 November 34 8 Karamadai YES No 2011 November 21 9 Vazhaithottam YES YES 2010, 2011, 2013, 2014 July–December 10 10a Chadapatti No − 2012 July–December≈ 150 10b Chadapatti No − 2011, 2013, 2014 June-October >400 11 Bokkapuram YES YES 2012 September 23 12a Mavanallah YES YES 2012 July–December 166 12b Mavanallah YES YES 2011 July 200 13 Singara Road YES No 2012 September 6 14 Chemanatham No − 2012 September 26 15a Masinagudi YES YES 2012, 2013 September, June 15 15b Masinagudi Camp No − 2012 September 12 16a Moyar YES No 2012 August 3 16b Moyar YES YES 2011 July 25 17 Ambalavayal No − − − 10 18 Mangala No − − − 10 19 Gundalpet YES No 2011, 2012 July, January 10 20 Nanjangud YES YES 2012 January 25 21 Bagalur YES YES 2012 January 33 22 Rameshwaram YES YES 2012 January 21 23 Bhadravati YES No − − 28 Notes: See Fig. 1 for map. Month of ﬂ owering and approximate size of each wild population are shown. Flowering data were often based on a single visit to each site per year (see Methods).

nearest millimeter using Vernier calipers. The length to which the stigma was and emasculated; and (5) hand pollination with crop pollen from a neighboring exserted beyond the anther was calculated by subtracting the length of the sta- fi eld on fl owers that were bagged as described and emasculated. men from that of the style. We also measured the diameter of the fl ower at its One fl ower per plant was assigned to each treatment, for totals of 73–122 widest point. Using data from the three locations with both wild and cultivar plants per treatment across all sites and years. Within years, sample sizes were populations, we performed ANOVAs to test for signifi cant effects of fl ower variable and ranged from 9 to 25 ( Table 4 ), mainly because many plants were type (wild vs. cultivar), location, and the interaction between fl ower type and damaged by grazing animals. The fi rst hermaphroditic fl ower located at the location on these two traits. base of the infl orescence was chosen because these fl owers are likely to set fruit when pollinated (Pradeepa, 2002; P. Davidar, personal observation). Soft mesh Hand-pollination experiments— To examine the potential for selfi ng and netting was used to make bags that excluded pollinators, and unopend fl ower for crop–wild hybridization, we conducted experiments at three locations (sites buds were enclosed in bags. To prevent self-pollination, we emasculated fl ow- 10, 12, and 16) on privately owned land in the Sigur Plateau of the Nilgiris ers that were assigned to the outcrossing and crop–wild hybridization treat- District of Tamil Nadu. Locations, years, and sample sizes are shown in Table ments by gently removing the anthers before anthesis. 4. These experiments involved fi ve treatments: (1) natural pollination of fl ow- Fresh pollen was collected from the appropriate source and applied to the ers that were open to pollinators; (2) autogamous pollination of fl owers that stigmas on the fi rst morning of anthesis. Pollen was extracted by making an were bagged before anthesis and left untouched; (3) hand pollination with self incision along the length of the anther and collecting a mound of pollen on the pollen on fl owers that were bagged as described; (4) hand pollination with outcross lance tip, which was promptly transferred to the stigma of an emasculated pollen from another wild individual on fl owers that were bagged as described fl ower. To avoid any pollen contamination, the researcher’s hands and tools

T ABLE 3. Mean stigma exsertion and ﬂ ower diameter for wild vs. cultivated brinjal at four locations in southern India. See Table 1 for site details. N = number of ﬂ owers (each from a different plant).

Stigma Flower Flower type Site Site ID Date sampled exsertion (mm) N SE diameter (cm)N SE

Wild Vazhaithottam 9 7/13/2011 3.7 10 0.2 3.25 10 0.13 Mavanhalla 12 7/15/2011 3.4 10 0.2 3.32 10 0.16 Gundalpet 19 1/2/2012 2.0 10 0.3 3.22 10 0.16 Chadapatti 10 6/12/2013 4.1 20 0.2 2.22 25 0.04 All wild sites 3.3 4 0.5 3.00 4 0.26

Crop Vazhaithottam 9 7/13/2011 0.3 10 0.1 3.99 10 0.08 Mavanhalla 12 7/15/2011 0.5 10 0.1 3.96 10 0.02 Gundalpet 19 7/17/2011 0.9 10 0.1 3.73 10 0.11 All crop sites 0.6 3 0.6 3.89 3 0.08 January 2015] DAVIDAR ET AL.—CROSS POLLINATION IN WILD EGGPLANT 133

T ABLE 4. Percentage fruit set and mean number of seeds per fruit on wild brinjal plants in hand-pollination experiments with ﬁ ve experimental treatments at three populations in southern India. Sample sizes (N ) are indicated in parentheses. Data on numbers of seeds per fruit were pooled from subsamples from the three populations. See Table 1 for site details.

Bagged, hand-pollinated (N ) Statistic or site Year Natural, open (N ) Autogamy, bagged (N ) Selfed Wild × Wild Wild × Crop

A) % Fruit set Pooled data (all sites) 2010–2014 74 (99) 0 (52) 13 (73) 52 (87) 26 (122) Populations Chadapatti (site 10) 2010 80 (10) no data 20 (10) 50 (10) 13 (15) Chadapatti (site 10) 2013 48 (29) 0 (20) 8 (27) 30 (23) 26 (27) Chadapatti (site 10) 2014 71 (17) 0 (10) 13 (16) 56 (9) 55 (9) Mavanallah (site 12) 2010–2011 70 (20) no data 10 (20) 45 (20) 20 (25) Moyar (site 16) 2011 100 (23) 0 (22) no data 80 (25) 17 (24) B) Mean number of seeds/fruit Pooled data (N ) 2010–2014 190 (28) no fruits 177 (4) 214 (32) 182 (9) 1 SE 13 52 17 36 Range 167–468 33–174 100–485 111–452 were rinsed in 90% alcohol between each successive hand-pollination. Flowers All fl ower visitors were recorded, and the effectiveness of pollination was de- from cultivated brinjal were collected from a nearby fi eld to obtain pollen for termined by noting whether (1) the insect was successful in collecting pollen the hybridization treatment. from the poricidal anthers, (2) the insect “buzzed” the anthers to release pollen, The mesh bags were removed as soon as fruit set was detected, and a subset of and (3) the insect contacted the stigma to affect pollen transfer. The total num- mature fruits was collected for seed counts. Many fruits were lost due to grazing ber of visits was divided by the total number of observation hours to calculate and other factors; sample sizes for characterizing the numbers of seeds per fruit the number of visits per hour for each type of fl ower visitor. are shown in Table 4B . Logistical problems prevented us from germinating seeds At Chadapatti, we captured 2–3 representative bees in each of three taxa to to compare fi tness-related traits of wild vs. crop–wild hybrid progeny, so a sepa- measure the intertegular wing span using Vernier calipers, as an indicator of rate study was carried out in France using germplasm from pre-existing acces- body size ( Table 5 ). These data were compared with those of Greenleaf et al. sions (M.-C. Daunay et al., unpublished manuscript). Nonetheless, a few hybrid (2007), who showed that larger bees forage over longer distances. For wild seeds germinated and grew to fl owering in the present study. brinjal plants, we also counted the numbers of male and hermaphroditic fl owers To assess the potential for outcrossing, we determined the index of self- open on a single day at peak fl owering for 45 plants in 2014 to characterize incompatibility (ISI) for wild brinjal. The ISI is calculated as the ratio of fl oral displays (number of open fl owers) and sex ratio. Floral longevity was fruit set through hand-augmented self-pollination over hand-augmented cross- assessed in 10 hermaphroditic and 10 male fl owers of wild brinjal on 20 differ- pollination ( Bawa, 1974 ). Higher ISI indices indicate higher levels of self- ent plants. compatibility.

Pollinator visits to wild and cultivated brinjal— We recorded pollinator visits to wild brinjal in Chadapatti and crop brinjal in Vazhaithottam during RESULTS June–October 2013 and May–June 2014, when both wild and crop brinjal were fl owering. These sites were about 2 km apart. Isolated patches of wild brinjal Distribution, fl owering phenology, and proximity to cultivated also were present close to crop fi elds, but because these were small (<10 plants), brinjal— Wild brinjal populations were found mainly in the drier pollinator observations were carried out at the larger population in Chadapatti. areas of Karnataka and Tamil Nadu, where the average annual rain- Data from three crop varieties (Green, Karamadai, and Neelam) in Vazhaithot- fall is ~800 mm and the dry season is ~8–9 mo long ( Fig. 1 ). The tam were pooled for analyses. Observations were carried out from 0900 hours to 1600 hours, when pollina- typical natural vegetation in this region is thorny scrub forest. We tors were most active, for a total of 20–27 hours at each location. A fl ower on a also found three wild populations on the western fl ank of the West- randomly selected plant was observed continuously for 15 min, after which ern Ghats, where the climate is wetter and the dry season is shorter another plant was selected for observation in a different part of the population. (Wayanad in Kerala; Sittar and Lower Kodayar in Tamil Nadu).

T ABLE 5. Pollinator visitation rates to wild and crop brinjal ﬂ owers in the Chadapatti-Vazhaithottam area.

Visits per hour (Mean ± 1 SD) Pollinating insects (Body size, mm) Family Year Wild brinjal Crop brinjal “Buzzing” ﬂ ower Pollen collection Stigma contact

Amegilla spp. (3.2–3.5 mm) Apidae 2013 0.60 ± 0.68 0.33 ± 0.80 Yes Good Yes 2014 3.40 + 2.68 2.33 + 2.59 Xylocopa spp. Apidae 2013 0.15 ± 0.49 0.14 ± 0.48 Yes Good Yes (7–8 mm or 10 mm) 2014 1.68 + 1.55 2.44 + 2.38 Nomia spp. (1.2 mm) Halictidae 2013 0.85 ± 1.87 0.33 ± 0.58 No Good Yes 2014 1.72 + 2.41 0.70 + 1.71 Heterotrigona iridipennis Apidae 2013 1.80 ± 2.78 4.76 ± 4.93 No Good Yes 2014 1.04 + 1.06 0.33 +0.78 Apis spp. Apidae 2013 0.05 ± 0.22 4.29 ± 2.49 No Poor Yes 2014 0.00 3.41 + 2.58 Syrphid ﬂ y Syrphidae 2013 1.40 ± 1.90 0.00 No Average Yes 2014 0.00 0.00 Notes: Numbers of hours of observation (N for means) = 20 or 21 h for wild vs. crop, respectively, in 2013, and 25 or 27 in 2014. The approximate body size for the ﬁ rst three species is listed; this is the intertegular span between wing bases; as in Greenleaf, et al. (2007) . 134 AMERICAN JOURNAL OF BOTANY [Vol. 102

Because we did not survey the wetter habitats of Kerala, we are not Flower morphology of wild and cultivated brinjal— Wild and able to report whether wild populations occur there. The average cultivated brinjal have gamopetalous fl owers (fused petals), pro- elevation where wild brinjal populations were recorded was around duced singly or in small clusters of ~2–4 fl owers at the apex of 900 m a.s.l., with a range of 97–1010 m a.s.l. We did not fi nd wild stems (Figs. 2, 3) . The fl owers are fl at and usually fi ve-lobed, al- brinjal at higher elevations, despite searches. though some male fl owers are four-lobed, and varied from a deep All 23 populations and nearby subpopulations were associated violet to pale purple in both taxa. The fl owers produce no nectar with villages, human settlements, roadsides, and/or barren unuti- and pollen is the only reward for pollinators. As noted already, wild lized wasteland ( Table 1 ). Usually the plants did not occur di- brinjal is andromonecious, with hermaphroditic and male fl owers rectly within the villages, but were found in their immediate on the same plant. In hermaphroditic fl owers, the lobed, poricidal vicinity. Roadside populations were linear, while those in waste- anthers form a tube through which the style and stigma protrude, land and village lands were irregular in their spatial distribution. whereas male fl owers have very small ovaries that typically do not Large populations (>100 plants) were rare, and most had fewer develop into fruits. Many fl owers of cultivated brinjals in Vazhai- than 50 plants ( Table 2 ). Wild brinjal is a perennial, with the thottam did not have visible stigmas suggesting that they were growth habit of a small shrub that is sometimes prostrate. The functionally male (P. Davidar, personal observation), as is often plants are often smaller after being grazed by vertebrates such as seen in other cultivars. Wild brinjal often produced about 5–15 cattle, goats, wild pigs, and spotted dear (Axis axis ), despite hav- open fl owers per plant, while cultivated brinjal produced ~10–15 ing sharp spines on the leaves, stems, and fl oral calyx. Above- fl owers, although fl ower number varied widely depending on the ground stems and branches usually died back during the late dry size of the plant. At Chadappati, wild brinjal plants had an average season due to drought and resprouted after the premonsoon of 2.8 ± 0.3 (1 SE) hermaphroditic fl owers and 5.8 ± 0.6 (1 SE) showers in April–May. In contrast, the crop brinjal is cultivated as male fl owers per day in 2014 ( N = 45 plants). Anthesis occurred in an annual. In wild populations, mature fruits typically turned the early morning hours, and the fl owers closed by late afternoon from green to golden yellow-orange, but some plants produced to reopen the next morning. The hermaphroditic fl owers lasted for immature fruits that were white (at sites 21 and 22). about 2 d and male fl owers for 3 d. Fourteen of the 23 wild populations (61%) occurred within Wild plants had greater stigma exsertion and fl owers that 0.5 km of cultivated brinjal populations (Fig. 1) , and fl owering were somewhat smaller than those of cultivated fl owers in the times of wild and cultivated plants overlapped for at least nine sampled populations ( Table 3 ). ANOVA for data from 2011 of these populations (Table 2; note that subpopulations are and 2012 at three sites showed signifi cant differences in grouped together). We made only single visits to the six popula- stigma exsertion due to fl ower type (wild vs. cultivated; F 1, 54 = tions where cultivated and wild plants occurred within 0.5 km 236.84, P < 0.001), location (F 2, 54 = 6.40, P < 0.01), and the of each other and did not fl ower synchronously, so it is possible interaction between them (F 2, 54 = 18.98, P < 0.001). For the that they overlapped at other times of the year. three sampled wild populations, mean stigma exsertion was

Fig. 1. Map of study sites for wild brinjal (also known as wild eggplant) in southern India, showing which populations occurred within 0.5 km of cultivated brinjal. See Table 1 for site details. January 2015] DAVIDAR ET AL.—CROSS POLLINATION IN WILD EGGPLANT 135

suggesting partial incompatibility between these closely related taxa, but also showing that hybridization is possible. The number of seeds per fruit varied from 33–485 seeds across treatments ( Table 4B ). Sample sizes were generally too small for statistical comparisons among treatments ( Table 4B ). Natural pollination resulted in a mean of 190 seeds per fruit (N = 28), while hand-outcrossing resulted in a mean of 214 seeds, N = 32). It is noteworthy that crop pollen resulted in an average of 182 seeds per fruit (N = 9), indicating that crop–wild hybridization is possible. A few F 1 crop–hybrid progeny were germinated and grown until ﬂ owering, demonstrating that hybrid genotypes were vigorous. We assume that these progeny were true hybrids because the bagged ﬂ owers were emasculated before hand-pollination with crop pollen, and the F1 plants had phenotypes that were in- termediate between wild and cultivated parents (similar to M.-C. Daunay et al., unpublished manuscript).

Pollinator visits— Pollinator species were studied at one wild population and one cultivated population, <2 km apart, over several months. Insects that actively collected pollen and made contact with the stigma were designated as pollinators. Five of the six pollinator species visited both wild and cultivated fl owers ( Table 5 , Fig. 3 ), indicating the potential for crop–wild hybridization where both are present (as observed at Vazhaithottam, for example). Ame- Fig. 2. Wild brinjal in wasteland area near Chadapatti in southern gilla spp. and Xylocopa spp. (including X. latipes ) buzzed the fl ow- India. Note syrphid fl y visiting the fl ower. ers for pollen and were likely to be the most effective pollinators on a per visit basis. Xylocopa spp. were the largest bees (7–10 mm 2.0–3.7 mm, compared with means of 0.3−0.9 mm from three intertegular span) to visit the fl owers, followed by Amegilla spp. cultivated populations growing near them (Table 3). In a (~3–4 mm) and Nomia spp. (~1.2 mm). Nomia spp. extracted pol- fourth wild population, Chadapatti, the mean stigma exsertion len with their proboscis. Heterotrigona iridipennis was able to ex- was 4.0 mm. tract pollen by manipulating the anthers. These pollinator taxa Flower diameter also differed between types, with wild pop- were the most common fl ower visitors on wild brinjal and also ulations averaging 3.2–3.3 cm wide and cultivar populations visited crop brinjal, whereas Apis spp., including A. cerana, were averaging 3.7–4.0 cm wide. For this trait, ANOVA showed a much more common on cultivated than wild plants. Apis cerana did not seem to be effective at collecting pollen and spent time signifi cant main effect of fl ower type (F 1, 54 = 30.02, P < 0.01), while location and the interaction between type and location chewing the anthers to try to access the pollen. Very few pollen were not signifi cant (P > 0.40). grains were observed on Apis visitors. Syrphid fl ies were also observed on wild brinjal, but were not seen on the crop fl owers. For all pollinators combined, crop brinjal fl owers were vis- Hand-pollination experiments— Similar hand-pollination ited nearly 10 times per hour on average, and wild brinjal results were found across locations and years (Table 4A). Be- fl owers were visited about 5 times per hour. In contrast, visits cause sample sizes were small at some locations, the pooled by the buzz-pollinating taxa were observed less than once an data set is reported here. Fruit set was highest for natural polli- hour on the crop fl owers, and about 1.6 times per hour on wild nations (74%, N = 99 fl owers), and no fruit set occurred from plants. bagged fl owers in the autogamy treatment (used at two locations; N = 52), indicating that fruit set requires pollinator visits. Outcrossed fl owers that were bagged and emasculated set fewer DISCUSSION fruits (52%, N = 57) than those from natural pollination, per- haps due to damage associated with anther removal, effects of Several fi ndings from our study demonstrate the potential for bagging, or differences in the quality of the pollen source, al- cross-pollination between cultivated and wild brinjal in southern though seed set from hand-outcrossed fl owers was high (mean India. First, many of the 23 wild brinjal populations that we ob- of 214 seeds per fruit, N = 32). served occurred near cultivated brinjal and had overlapping fl ow- For the hand-pollinated fl owers, we found that outcross ering periods. We also found that wild brinjal depends on insect hand-pollination resulted in high levels of fruit set, followed by pollinators for seed set and shares pollinator taxa with the crop. lower fruit set using pollen from cultivated brinjal, while self Buzz-pollinating bees such as Xylocopa spp., in particular, are pollen resulted in the lowest levels of fruit set overall (Table 4). known to forage over distances of several kilometers (e.g., Pasquet Flowers that were emasculated, bagged, and cross-pollinated et al., 2008 ). Furthermore, pollen from cultivated brinjal was ca- by hand had 52% fruit set, compared with only 13% fruit set pable of siring seeds on wild plants in hand-pollination experi- from hand-selfi ng (N = 87 and 73, respectively). Using pooled ments. Next, we interpret these results in greater detail. data, the index of self-incompatibility was 0.35 for Chadapatti and 0.22 for Mavanallah, indicating relatively low levels of Proximity and overlapping fl owering times— Small popula- self-compatibility and high potential for outcrossing. Fruit set tions of wild brinjal were common in disturbed habitats such as from crop–wild hand-pollinations was only 26% (N = 122), roadsides and fi eld margins around the outskirts of villages. For 136 AMERICAN JOURNAL OF BOTANY [Vol. 102

Fig. 3. Examples of pollinators visiting wild and cultivated brinjal in southern India. Amegilla on a male fl ower (A) and hermaphroditic fl ower (B) of wild brinjal; (C) Amegilla buzzing a hermaphroditic fl ower of cultivated brinjal; (D) Xylocopa buzzing a fl ower (left); Nomia on a hermaphroditic fl ower (right) of wild brinjal; (E) Nomia extracting pollen on a male fl ower of wild brinjal; (F) Heterotrigona on a male fl ower of wild brinjal; (G) Apis on a hermaphroditic fl ower of cultivated brinjal. January 2015] DAVIDAR ET AL.—CROSS POLLINATION IN WILD EGGPLANT 137 our 23 wild study populations, 61% were located within 0.5 km Sicard and Lenhard, 2011 ). Because wild brinjal has exserted stig- of cultivated brinjal and some occurred side by side, as in mas and exhibits relatively low levels of self-compatibility, with an Vazhaithottam (site 9). This was not surprising given that wild ISI of 0.22–0.35 in two study populations, we assume that out- brinjal thrives in disturbed sites near human habitation, and crossing is common. Average stigma exsertion varied from a low many households are located near small farm fi elds or kitchen of 2.0 mm in Gundalpet to a high of 4.1 mm in Chadapatti. In a gardens. We also observed large fi elds of cultivated brinjal in companion paper using measures of population structure (F ST ) Karamadai, near site 8, along the eastern foothills of the Nilgiri from 14 microsatellite DNA markers, we also found evidence for Mountains, close to the dry forest edges where wild brinjal is variation in outcrossing rates (t ) among 10 wild brinjal popula- found. In contrast, other wild populations were not located near tions, ranging from ~5–33% ( Mutegi et al., 2015 ). One of these cultivated fi elds, and many farmed areas were far from wild was the large population of >400 plants at Chadapatti (site 10), brinjal. On the eastern plains, we observed large-scale brinjal with an estimated outcrossing rate of 33%. Therefore, we conclude farms that were intensively managed and replanted frequently, that wild brinjal has a mixed mating system, with variable out- with fewer gaps in the vegetation for perennial weedy/wild crossing rates. Self-pollination may occur both within and between brinjal to become established. fl owers on the same plant, when bees visit multiple fl owers per At least 9 of the 14 wild populations that occurred near the foraging bout, but we observed relatively low fruit set from hand- crop had fl owering times that overlapped with cultivated brin- selfi ng. Wild brinjal populations with greater levels of outcrossing jal, providing opportunities for hybridization (Table 2). As noted may be more likely to hybridize with nearby fi elds of cultivated already, this value is likely to be an underestimate because it brinjal, while those with a high degree of selfi ng may be more in- was not feasible to visit the populations more than a few times sulated from the transmission of crop alleles. during the course of this project. Flowering in wild brinjal typically began in April–May, after the onset of the premonsoon Shared pollinators and their foraging distances— Solanum and South West monsoon rainfall, and continued sporadically fl owers have distinctive poricidal anthers that are buzz-pollinated until January–February over about 7–8 mo ( Table 2 ; P. Davidar, (e.g., Buchmann and Hurley, 1978 ). Both the wild and crop fl ow- personal observation). ers were visited by buzzing bees, such as Xylocopa and Amegilla , Most of the cultivated brinjal fi elds we visited were irrigated, as well as Nomia spp. and Heterotrigona iridipennis , which did not which facilitates year-round cultivation and fl owering. The buzz the anthers but made contact with the stigmas ( Table 5 ). We cropping cycle of cultivated brinjal typically is ~4–6 mo de- consider these taxa to be the most effective pollinators of wild brin- pending on the region and cultivar, and the plants start fl ower- jal at our study sites. Apis bees were common on the crop, but very ing about 1 mo after planting (Choudhary and Gaur, 2009). In rare on wild brinjal and did not appear to be very effective as pol- the Sigur plateau, the fi rst cropping cycle was initiated in May, linators. Although pollinator visits were quantifi ed at only two sites the plants started fl owering in June–July, and the next cropping studied in 2013 and 2014, we often saw the same pollinator taxa at cycle started soon after in September–October. There was often other populations of wild and cultivated brinjal when the plants a third cropping cycle in December. Therefore, given the ex- were in bloom. tended fl owering of wild brinjal, the proximity of crop and wild The transmission of crop genes to wild relatives via pollen also populations, and the overlap in fl owering times, we conclude depends on the foraging distances of pollinators and the proximity that there is ample potential for gene fl ow between the two. of the fl owering pollen donors and recipients. Here, we used These results are similar to other studies that documented prox- 0.5 km as a convenient distance for recording whether wild popu- imity and overlapping fl owering times between cultivated and lations occurred near enough to be pollinated by cultivated brinjal. wild populations of sunfl ower (Burke et al., 2002), rice (Cohen The foraging range of small solitary bees such as Heterotrigona et al., 2008 ), and sorghum ( Tesso et al., 2008 ), as a prerequisite iridipennis may be only about 100 m, whereas the ranges of larger for the occurrence of crop-to-wild gene fl ow. Nomia spp. can exceed 2–11 km (Greenleaf et al., 2007). Greenleaf et al. (2007) showed that pollinators’ foraging distances Potential for outcrossing in wild brinjal— Hand-pollination increase disproportionately with body size based on studies of 62 treatments showed that pollinator visits are required to effect bee species. In our study, Amegilla spp. had an intertegular (IT) pollen transfer in wild brinjal, and outcross pollen resulted in span of about 3.2–3.5 mm, which suggests foraging distances of >2 nearly 5-fold greater percentage fruit set than self pollen (Table km and possibly much farther (Greenleaf et al., 2007). Xylocopa 4 ). Most importantly, crop pollen resulted in viable seeds on species (carpenter bees) were the largest pollinators that visited wild wild plants at each of the three study populations, consistent and cultivated brinjal. We observed several species of Xylocopa with assessments of researchers who consider the crop (S. mel- with IT spans of 7–10 mm, which suggests that they may be ongena ) and S. insanum to be conspecifi c (e.g., Karihaloo, et al., capable of foraging over as distances as great as 10 km ( Greenleaf 1995 ). Although fruit set from crop pollen was relatively low et al., 2007). Foraging distances of 7.5 km were typical for and similar to that from self-pollen, it is possible that hetero- Xylocopa virginica in the United States (Greenleaf et al., 2007; sis of F 1 crop–wild hybrids (e.g., M.–C. Daunay et al., unpub- IT span of 5.6 mm). In Kenya, radio tracking of Xylocopa fl a- lished manuscript) could facilitate crop-to-wild introgression vorufa indicated that median fl ight distances were ~720 m, and under fi eld conditions. Our sample sizes for the hand-pollination the maximum was >6 km (Pasquet et al., 2008). Thus, visits to treatments were somewhat limited in any given year or wild brinjal from Xylocopa and Amegilla species may be most location (~10–29 plants per treatment, year, and location), likely to result in long-distance crossing between the crop and but treatment effects were consistent across sites and years its wild relative, due to their large size and “buzzing” behavior ( Table 4 ). to extract pollen from the anthers. In general, outcrossing rates of self-compatible species or popu- In summary, we conclude that Xylocopa , Amegilla , and Nomia lations are expected to be positively correlated with the degree of spp. were capable of effecting crop-to-wild pollen transfer over stigma exsertion, as documented in other bee-pollinated species distances of >0.5 km, and possibly over distances of several ( Hermanutz, 1991 ; Barrett et al., 2000 ; Takebayashi et al., 2006 ; kilometers. General predictions of foraging distances based on 138 AMERICAN JOURNAL OF BOTANY [Vol. 102 body size provide a helpful starting point, but many additional LITERATURE CITED factors can infl uence the abundance and behavior of pollinating bees, including densities and locations of nectar and pollen B ARRETT , S. C. H. , L. K. JESSON , AND A. M. BAKER . 2000 . The evolution sources, availability of nest sites, and other habitat characteris- and function of stylar polymorphisms in fl owering plants. Annals of tics (e.g., references of Greenleaf et al., 2007; Pasquet et al., Botany 85 : 253 – 265 . B AWA , K. S. 1974 . Breeding systems of tree species of a lowland tropical 2008 ). community. Evolution 28 : 85 – 92 . B UCHMANN , S. L. , AND J . P . H URLEY. 1978 . Biophysical model for buzz pollination in angiosperms. Journal of Theoretical Biology 7 2 : Implications and future research— Future research could 639 – 657 . focus on whether fi ndings from our study populations are repre- B URKE , J. M. , K. A. GARDNER , AND L. H. RIESEBERG . 2002 . The poten- sentative of others within the broad distribution of wild brinjal tial for gene fl ow between cultivated and wild sunfl ower (Helianthus across much of India and elsewhere (e.g., Deb, 1989; Lester annuus ) in the United States. American Journal of Botany 8 9 : and Hasan, 1991 ; Karihaloo and Rai, 1995 ). Also, further re- 1550 – 1552 . search is needed to confi rm the expectation that repeated epi- B URKE , J. M. , AND L. H. RIESEBERG . 2003 . Fitness effects of transgenic sodes of hybridization lead to introgression, i.e., the long term disease resistance in sunfl owers. Science 300 : 1250 . persistence, of crop alleles in wild brinjal populations (e.g., C HAKRABORTY , T. 2014 . Release of Bt-Brinjal in Bangladesh. Economic and Political Weekly 49 : 25 . Ellstrand et al., 2013 ). We predict that if transgenic brinjal is C HOUDHARY , B . , AND K . G AUR. 2009 . The development and regulation widely grown in southern India, transgenes are likely to fi nd of Bt Brinjal in India (eggplant/aubergine). International Service for their way into some wild, weedy, or feral populations, depend- the Acquisition of Agri-Biotech Applications (ISAAA), Brief no. 38. ing on local conditions such as the relative sizes and proximity ISAAA, Ithaca, New York, USA. Available at http://www.isaaa.org/ of the crop and wild brinjal populations. The persistence of crop resources/publications/briefs/38/default.asp . alleles, whether transgenic or not, would depend on the fi tness C HOUDHARY , B. , K. M. NASIRUDDIN , AND K . G AUR . 2014 . The status of effects of these genes, as well as linked genes, within wild pop- commercialized Bt brinjal in Bangladesh. International Service for ulations (e.g., Snow et al., 2003 ). the Acquisition of Agri-Biotech Applications (ISAAA), Brief no. 47. Although we did not measure gene fl ow directly, this study ISAAA, Ithaca, New York, USA. http://www.isaaa.org/resources/ fi lls a major gap in the scientifi c literature by demonstrating the publications/briefs/47/default.asp . C OHEN, M. B., A. A. S NOW , S . A RPAIEA , L . P . L AN , AND L. M. CHAU . 2008 . potential for crop genes from transgenic brinjal to disperse into Shared fl owering phenology, insect pests, and pathogens among neighboring populations of S. insanum if Bt brinjal is adopted in wild, weedy, and cultivated rice in the Mekong Delta, Vietnam: the future. Contrary to the conclusions of the EC-II report (EC-II, Implications for transgenic rice. Environmental Biosafety Research 2009 ), our results suggest that transgenes would be introduced 7 : 73 – 85 . into populations of wild relatives of brinjal in southern India. Ef- D AUNAY , M. C. , AND P . H AZRA . 2012 . Eggplant . In P. Hazra and K. V. Peter forts are under way to introduce Bt eggplant into many countries [eds.], Handbook of vegetables, 257–322. Studium Press, New Delhi, in Southeast Asia, and commercial release was recently approved India. by the government of Bangladesh ( Choudhary et al., 2014 ). The D AV I DA R , P . , J . P . P UYRAVAUD , AND E. G. L EIGH. 2005 . Changes in rain ecological and evolutionary consequences of such transgene “es- forest tree diversity, dominance, and rarity across a seasonality cape”, if any, will be infl uenced by the scale over which gene gradient in the Western Ghats, India. Journal of Biogeography 3 2 : fl ow occurs and whether the transgenes are associated with fi t- 493 – 501 . D EB , D. 1989 . Solanum melongena, Solanum incanum versus Solanum ness benefi ts, greater weediness, or any harmful or unwanted ef- insanum (Solanaceae). Taxon 38 : 138 – 139 . fects on valued nontarget organisms (e.g., Snow et al., 2005). In EC-II . 2009 . Report of the Expert Committee (EC-II) on Bt Brinjal Event EE-1. the case of Bt brinjal, additional research could be carried out to Developed by M/s Maharashtra Hybrid Seeds, University of Agricultural examine such effects of the Bt transgene on wild brinjal popula- Sciences (UAS), and Tamil Nadu Agricultural University; submitted to the tions, similar to previous studies of Bt sunfl ower and Bt rice (e.g., Genetic Engineering Approval Committee, Ministry of Environment and Snow et al., 2003 ; Yang et al., 2011 ). Forests, Government of India, New Delhi. Available at http://www.moef. Meanwhile, social and political issues surrounding Bt brinjal nic.in/sites/default/fi les/Report%20on%20Bt%20brinjal_2.pdf . in India and elsewhere are complex, given the continuing debate E LLSTRAND , N. C. 2003 . Current knowledge of gene fl ow in plants: over whether it is safe to eat (e.g., Shantharam, 2010 ; Samuels, Implications for transgene fl ow. Proceedings of the Royal Society, B, 2013 ; Kudlu and Stone, 2013). Similar biosafety questions are Biological Sciences 358 : 1163 – 1170 . likely to arise in other parts of Asia, where the regulatory status E LLSTRAND , N. C. , P. MEIRMANS , J . R ONG , D . B ARTSCH , A . G HOSH , T. J. D ELONG , P . H ACCAU , ET AL . 2013 . Introgression of crop alleles into of transgenic crops is still uncertain, and our results about the wild or weedy populations. Annual Review of Ecology and Systematics potential for pollen-mediated gene fl ow to wild relatives may 44 : 325 – 345 . be relevant in these regions. During a recent trip to Thailand, G EORGIADY , M . , R . W HITKUS , AND E . L ORD . 2002 . Genetic analysis of traits for example, we found that a small-fruited S. melongena , lo- distinguishing outcrossing and self-pollinating forms of currant tomato, cally known as makheua chae, is a common volunteer food Lycopersicon pimpinellifolium (Jusl.) Mill. Genetics 161 : 333 – 344 . plant that is allowed to persist in disturbed habitats such as G REENLEAF , S. S. , N. M. WILLIAMS , R . W INFREE , AND C . K REMEN . 2007 . roadsides and in small garden plots, adjacent to cultivated brin- Bee foraging ranges and their relationship to body size. Oecologia jal (M.-C. Daunay and R. Pasquet, personal observations). Be- 153 : 589 – 596 . fore deregulating transgenic crops, regulatory agencies in the H ERMANUTZ , L. 1991 . Outcrossing in the weed, Solanum ptycanthum USA and elsewhere typically consider (1) the potential for gene (Solanaceae) — A comparison of agrestal and ruderal populations. American Journal of Botany 78 : 638 – 646 . fl ow via both pollen and seeds, (2) the trait being introduced, J AYARAMAN , K. 2010 . Bt brinjal splits Indian cabinet. Nature Biotechnology (3) the receiving environment, and (4) the likelihood of any un- 28 : 296 . wanted, unintended effects. Therefore, baseline information J OY , N . , Z . A BRAHAM , AND E. V. SONIYA . 2007 . A preliminary assessment about gene fl ow and its consequences can contribute to risk as- of genetic relationships among agronomically important cultivars of sessments of this popular and widely grown crop. black pepper. BMC Genetics 8 : 42 . January 2015] DAVIDAR ET AL.—CROSS POLLINATION IN WILD EGGPLANT 139

K ARIHALOO, J. L., S. B RAUNER , AND L. D. G OTTLIEB. 1995 . Random P ILSON , D . , AND H. R. PRENDEVILLE . 2004 . Ecological effects of transgenic amplified polymorphic DNA variation in the eggplant, Solanum crops and the escape of transgenes into wild populations. Annual Review melongena L. (Solanaceae). Theoretical and Applied Genetics 9 0 : of Ecology and Systematics 35 : 149 – 174 . 767 – 770 . P RADEEPA , G. L. 2002 . Fruit set in relation to fl oral morphology of egg- K ARIHALOO , J. L. , AND M . R AI . 1995 . Signifi cance of morphological vari- plant (Solanum melongena L.). Tropical Agricultural Research and ability in Solanum insanum L. ( sensu lato ). Plant Genetic Resources Extension 5 : 12 – 16 . Newsletter (Rome, Italy) 103 : 24 – 26 . S AMBANDAM , C. N. 1964 . Natural cross pollination in eggplant. Economic K NAPP , S. , M. S. VORONTSOVA , AND J . P ROHENS . 2013 . Wild relatives of the Botany 18 : 128 – 131 . eggplant (Solanum melongena L.: Solanaceae): New understanding of S AMUELS , J. 2011 . Bt brinjal, wild relatives and biodiversity. Current species names in a complex group. PLoS ONE 8 : e57039 . Science 100 : 603 – 604 . K UDLU , C . , AND G. D. STONE . 2013 . The trials of genetically modifi ed S AMUELS , J. 2013 . Bt brinjal: A risk assessment worth taking? Current food: Bt eggplant and Ayurvedic medicine in India. Food Culture & Science 104 : 571 – 572 . Society 16 : 21 – 42 . S ASU , M. A. , M. J. FERRARI , D . D U , J. A. WINSOR , AND A. G. STEPHENSON . L ESTER , R. N. , AND S. M. Z. HASAN . 1990 . The distinction between 2009 . Indirect costs of a nontarget pathogen mitigate the direct ben- Solanum incanum L. and Solanum insanum L. (Solanaceae). Taxon efi ts of a virus-resistant transgene in wild Cucurbita. Proceedings of 39 : 521 – 523 . the National Academy of Sciences, USA 106 : 19067 – 19071 . L ESTER , R. N. , AND S. M. Z. HASAN . 1991 . Origin and domestication of the S HANTHARAM , S. 2010 . Setback to Bt brinjal will have long-term effect on eggplant, Solanum melongena , from Solanum incanum , in Africa and Indian science and technology. Current Science 98 : 996 – 997 . Asia . In J. G. Hawkes, R. N. Lester, M. Nee, and N. Estrada [eds.], S ICARD , A . , AND M . L ENHARD . 2011 . The selfi ng syndrome: A model for Solanaceae III: Taxonomy, chemistry, evolution, 369–387. Royal studying the genetic and evolutionary basis of morphological adapta- Botanical Gardens, Kew, UK. tion in plants. Annals of Botany 107 : 1433 – 1443 . L U , B . , AND A. A. SNOW . 2005 . Gene fl ow from genetically modifi ed rice S NOW , A. A. , D. A. ANDOW , P . G EPTS , E. M. HALLERMAN , A . P OWER , J . and its environmental consequences. Bioscience 55 : 669 – 678 . M. TIEDJE , AND L. L. WOLFENBARGER . 2005 . Genetically engineered M ARVIER , M . , AND R. C. V AN ACKER. 2005 . Can crop transgenes be organisms and the environment: Current status and recommendations. kept on a leash? Frontiers in Ecology and the Environment 3 : Ecological Applications 15 : 377 – 404 . 93 – 100 . S NOW , A. A. , D. PILSON , L. H. RIESEBERG , M . P AULSEN , N . P LESKAC , M. R. M ERCER , K. L. , AND J. D. WAINWRIGHT . 2008 . Gene fl ow from transgenic R EAGON , D. E. WOLF , AND S. M. SELBO . 2003 . A Bt transgene reduces maize to landraces in Mexico: An analysis. Agriculture, Ecosystems herbivory and enhances fecundity in wild sunfl owers. Ecological & Environment 123 : 109 – 115 . Applications 13 : 279 – 286 . M UTEGI , E. , A. A. SNOW , M . R AJKUMAR , R . P ASQUET , H . P ONNIAH , M.-C. S TEWART , C. N. JR ., M. D. HALFHILL , AND S. I. WARWICK . 2003 . Transgene D AUNAY , AND P . D AVIDAR . 2015 . Genetic diversity and population introgression from genetically modifi ed crops to their wild relatives. structure of wild/weedy eggplant (Solanum insanum, Solanaceae) in Nature Reviews. Genetics 4 : 806 – 817 . southern India: Implications for conservation. American Journal of T AKEBAYASHI , N. , D. E. WOLF , AND L. F. DELPH . 2006 . Effect of variation Botany 102 : 139 – 147 . in herkogamy on outcrossing within a population of Gilia achilleifo- M YERS , N. 1990 . The biodiversity challenge: Expanded hotspot analysis. lia. Heredity 96 : 159 – 165 . Environmentalist 10 : 243 – 256 . T ESSO , T. T. , I. KAPRAN , C . G RENIER , A. A. SNOW , P. M. SWEENEY , J. F. P ADMANABAN , G. 2009 . Bt brinjal—Ban or boon? Current Science 97 : P EDERSEN , D . M ARX , ET AL . 2008 . The potential for crop-to-wild gene 1715 – 1716 . fl ow in sorghum in Ethiopia and Niger: a geographic survey. Crop P ASCAL , J. P. 1988 . Wet evergreen forests of the Western Ghats of India: Science 48 : 1425 – 1431 . Ecology, structure, fl oristic composition and succession. Institut V ENKATARAMANI , K. S. 1946 . Breeding brinjals ( Solanum melongena ) in Francais de Pondichery, Pondicherry, India. Madras. Hybrid vigour in brinjals. Proceedings of the Indian Academy P ANDEY , A. , A. K. TOMER , D. C. BHANDARI , AND S. K. PAREEK . 2008 . of Sciences, Section B 23 : 266 – 273 . Towards collection of wild relatives of crop plants in India. Genetic V ISWANATHAN , T. V. 1975 . On the occurrence of natural hybridization Resources and Crop Evolution 55 : 187 – 202 . between Solanum incanum L. and Solanum melongena L. Current P ASQUET , R. S. , A. PELTIER , M. B. HUFFORD , E . O UDIN , H . S AULNIER , L . P AUL , Science 44 : 134 . J. T. KNUDSEN , ET AL . 2008 . Long-distance pollen fl ow assessment Y ANG , X . , X . H IU , F . W ANG , J . S U , A. A. SNOW , AND B.-R. LU . 2011 . through evaluation of pollinator foraging range suggests transgene Transgenes for insect resistance reduce herbivory and enhance fe- escape distances. Proceedings of the National Academy of Sciences, cundity in advanced generations of crop-weed hybrids of rice. Evolu- USA 105 : 13456 – 13461 . tionary Applications 4 : 672 – 684 .