Development!Of!Tools!

The$Pennsylvania$State$University$ $ The$Graduate$School$ $ College$of$Agriculture$ ! ! ! DEVELOPMENT!OF!TOOLS!TO!EVALUATE!THE!UTILITY!OF!GRAIN!LEGUME!! ! ROOT!ARCHITECTURE!! ! ! ! A$Dissertation$in$$ $ Horticulture$ $ by$$ $ James$D$Burridge$ $ $ $ ©$James$D$Burridge$ $ Submitted$in$Partial$Fulfillment$$ of$the$Requirements$ for$the$Degree$of$ $ Doctor$of$Philosophy$ $ December$2017$ The$dissertation$of$James$D$Burridge$was$reviewed$and$approved*$by$the$following:$ $ $ $ $ $ Jonathan$P.$Lynch$ Professor$of$Plant$Nutrition$ University$Distinguished$Professor$ Dissertation$Adviser$ Chair$of$Committee$ $ $ $ $ Kathleen$M.$Brown$ Professor$of$Plant$Stress$Biology$ ! ! ! $ Mark$J.$Guiltinan$ Professor$of$Plant$Molecular$Biology$ $ $ $ $ David$A.$Mortensen$ Professor$of$Weed$and$Applied$Plant$Ecology$ ! ! ! ! Erin$Connolly$ Professor$and$Head$of$the$Department$of$Plant$Science$ $ $ $ *Signatures$are$on$file$in$the$Graduate$School.$ ! ! $ ! ! !

$ ii$ ABSTRACT$ Grain$ legumes$ are$ important$ for$ smallTholder$ farmers$ and$ the$ socioTeconomic$ stability$of$the$world.$They$supply$proteins$to$humans$and$animals$and$contribute$ biologically$fixed$nitrogen$to$following$crops.$Low$phosphorus$(P)$availability$and$ drought$ are$ primary$ constraints$ to$ legume$ production$ in$ developing$ countries.$ Genetic$ variation$ in$ particular$ root$ architectural$ phenes$ of$ grain$ legumes$ are$ associated$ with$ improved$ acquisition$ of$ water$ and$ phosphorus.$ Quantitative$ evaluation$ of$ root$ architectural$ phenotypes$ of$ mature$ plants$ in$ the$ field$ is$ challenging.$Nonetheless,$in$situ$phenotyping$captures$responses$to$environmental$ variation$and$is$critical$to$improving$crop$performance$in$the$target$environment.$ This$ dissertation$ presents$ root$ architectural$ phenotyping$ techniques$ for$ common$ bean$ (Phaseolus) vulgaris)$ cowpea$ (Vigna) unguiculata)$ and$ several$ other$ critical$ grain$legumes.$Chapter$1$presents$the$phenotyping$techniques.$Chapter$2$connects$ cowpea$root$architecture$to$performance$through$genetic$coTlocalizations,$which$is$ a$ novel$ contribution$ for$ cowpea.$ Chapter$ 3$ presents$ results$ from$ phenotyping$ of$ other$grain$legumes,$identifies$root$architectural$patterns$in$and$among$species$and$ discusses$ opportunities$ and$ challenges$ for$ physiologists$ and$ plant$ breeders$ in$ developing$more$stress$tolerant$varieties.$$ ! ! ! !

$ iii$ ! Table!of!Contents! $ List$of$Figures$$…………………………………………….………………………………………v$ List$of$Photograph$Panels$…………………………….……………………………………..vi$$ Acknowledgments$$$…………………………….…………….……………………………….vii$ Introduction$………………………………………………….……………………………………$1$ $$$$$$$$References$$…..……………………………………….….…………………………………...5$ $ Chapter$1:$Legume$Shovelomics:$HighTThroughput$phenotyping$of$common$bean$ (Phaseolus)vulgaris$L.)$and$cowpea$(Vigna)unguiculata)subsp,$unguiculata))root$ architecture$in$the$field$$…………………………………………………………….…..……)9$ $$$$$$$$References$$$…..…………………………………………………………………………….19$ $ Chapter$2:$GenomeTwide$association$mapping$and$agronomic$impact$of$cowpea$root$ architecture$$………………………………………………………………………...….……….$21$ $$$$$$$$References$$$…..…………………………………………………………………………….31$ $ Chapter$3:$Comparative$Annual$Grain$Legume$Root$Architecture$……....$34$$$ $ $$$$$$$$References$$$….…………………………………………………………………………….50$ $$$$$$$$Figures$………………………………………………………………………………...... 61$ $ Epilogue$$$………………………………………………………………………………...... 113$ $$$$$$$$References$$$…..…………………………………………………………………………..121$ ! !

$ iv$ LIST$OF$FIGURES$ $ Introduction$ Page$2:$Figure$1.$Common$bean$root$system$classes$from$field$example$(left)$and$ cartoon$ example$ (right).$ Adventitious$ or$ hypocotyl$ roots$ emerge$ from$ the$ hypocotyl,$ basal$ roots$ emerge$ from$ the$ base$ of$ the$ hypocotyl,$ the$ primary$ or$ tap$ root$is$the$radical.$ $ CHAPTER$3:$Comparative$annual$grain$legume$root$architecture$$ Page$61:$Figure$1.$Drawings$depicting$differences$in$root$architecture$between$ epigeal$and$hypogeal$germinating$species.$ Page$62:$Figure$2.$Various$legumes$on$the$root$system$architectural$spectrum$(xT axis)$and$estimated$water$availability$in$domestication$environment$(yTaxis).$ Page$65:$Figure$3.$Violin$plot$of$branching$density$for$bean,$cowpea,$groundnut,$soy,$ tepary.$ Page$66:$Figure$4.$Violin$plot$of$BRGA$for$bean,$cowpea,$groundnut,$soy,$tepary.$ Page$67:$Figure$5.$Violin$plot$TD5$to$SD$for$bean,$chickpea,$cowpea,$groundnut,$soy,$ tepary.$ Page$68:$Figure$6.$Violin$plot$of$BRN$for$bean,$cowpea,$soy,$tepary.$ Page$69:$Figure$7.$Violin$plot$of$BRN$to$TD5$for$bean,$cowpea,$soy,$tepary.$ Page$70:$Figure$8.$Violin$plot$of$ARN$for$bean,$cowpea,$soy,$tepary.$ Page$71:$Figure$9.$Violin$plot$of$BRN$to$ARN$for$bean,$cowpea,$soy,$tepary.$ Page$72:$Figure$10.$Violin$plot$of$ARN$to$SD$for$bean,$cowpea,$soy,$tepary.$ Page$73:$Figure$11.$Violin$plot$of$BRGA$for$common$bean$gene$pools.$ Page$74:$Figure$12.$Violin$plot$of$ARN$for$common$bean$gene$pools.$ Page$75:$Figure$13.$Violin$plot$of$BRN$for$common$bean$gene$pools.$ Page$76:$Figure$14.$Violin$plot$of$TD5$for$common$bean$gene$pools.$ Page$77:$Figure$15.$Violin$plot$of$TD5$to$SD$for$common$bean$gene$pools.$ Page$78:$Figure$16.$Regression$between$TD5$and$BRN$in$Soy,$Tepary,$Cowpea$and$ Bean.$ Page$79:$Figure$17.$Regression$between$TD5$and$BRN$in$Bean.$ Page$80:$Figure$18.$Regression$between$TD5$and$BRN$plus$ARN$in$bean.$ Page$81:$Figure$19.$Regression$between$deep$and$shallow$scores$in$groundnut,$ soybean,$tepary$bean,$cowpea$and$common$bean.$ Page$82:$Figure$20.$Regression$between$deep$and$shallow$scores$in$common$bean.$ Page$83:$Figure$21.$Regression$between$deep$and$shallow$scores$in$cowpea.$ Page$84:$Figure$22.$Regression$between$deep$and$shallow$scores$in$groundnut.$ Page$85:$Figure$23.$Regression$between$deep$and$shallow$scores$in$soybean.$ Page$86:$Figure$24.$Regression$between$deep$and$shallow$scores$in$tepary$bean.$ Page$87:$Figure$25.$Regression$between$deep$and$shallow$scores$in$chickpea.$ Page$88:$Figure$26.$Violin$plot$of$sum$of$root$cross$sectional$area$relative$to$ hypocotyl$cross$sectional$area$in$bean,$cowpea,$groundnut,$soy$and$tepary.$ Page$ 89:$ Figure$ 27.$ Drawings$ depicting$ shallow,$ deep$ and$ dimorphic$ root$ architectural$phenotypes$of$epigeal$and$hypogeal$germinators.$ Page$90:$Table$1.$List$if$genotypes$phenotyped$by$location$and$species.$

$ v$ LIST$OF$PHOTOGRAPH$PANELS$ Chapter$3:$Comparative$annual$grain$legume$root$architecture$$ $ Page$63:$Photo$panel$1:$Representative$images$showing$variation$of$crown$root$ architecture$of$different$species.$$ Page$64:$Photo$panel$2:$Dimorphic$root$phenotypes.$ $ $

$ vi$ Acknowledgements! $ This$ research$ would$ not$ have$ been$ possible$ with$ the$ financial$ support$ and$ collaboration$of$various$foundations$and$institutions.$We$acknowledge$the$support$ of$ the$ Howard$ G$ Buffet$ Foundation,$ the$ United$ States$ Agency$ for$ International$ Development$ Global$ Hunger$ and$ Food$ Security$ Research$ Strategy:$ Climate$ Resilience,$ Nutrition,$ and$ Policy$ T$ Feed$ the$ Future$ Innovation$ Lab$ for$ Climate$ Resilience$in$Beans$Project$#$AIDTOAATAT00077,$The$McKnight$Foundation,$and$The$ Pennsylvania$State$University.$$ $ The$ findings$ and$ conclusions$ presented$ in$ this$ dissertation$ do$ not$ necessarily$ reflect$the$views$of$the$funding$agencies.$$ $

$ vii$ Introduction Root system architecture and the two resource problem Legumes are important for small-holder farmers and the socio-economic stability of the world. They supply proteins to humans and animals and contribute biologically fixed nitrogen to following crops. Common bean is the most important food legume for direct human consumption but average yield from 2011-2014 in South and Central American was 0.9 metric ton per hectare or 15% of yield potential and Sub-Saharan Africa was 0.8 metric ton per hectare or 13% of yield potential (Beaver et al., 2003; FAOSTAT, 2017). An estimated 73% of bean production is constrained by water availability and 50% by phosphorus (Beebe et al., 2009). Use of chemical fertilizers and irrigation systems are not practical for many small-holder farmers in Latin America and Sub-Sahara Africa because of lack of capital and infrastructure. Improvement of abiotic stress tolerance is critical for the food security of billions of people and legumes have the additional role of fixing nitrogen. Demand for food is estimated to increase 25%-70% by 2050 in response to a projected 9.7 billion people and increased affluence (Hunter et al., 2017; UN, 2015). Serious challenges face humanity as we face the need to achieve sustainable intensification and balance food production and environmental sustainability (Hunter et al., 2017). During this same period fertilizer will become more expensive as P becomes more scarce (Steen, 1998) and nitrogen fertilizer becomes more expensive as fuel prices rise. The situation is further compounded by challenges that are increasing in frequency and severity including; elevated temperature, altered precipitation patterns, drought, salinity, erosion, soil degradation, waterlogging, and increased cost of fertilizer (Groisman et al., 2005; St.Clair and Lynch, 2010). Varieties with superior stress tolerance must be paired with improved farming practices and cropping systems to regenerate soil and build organic matter. The precariousness of small-holder farmers and the local and regional networks they form the foundation of contributes to migration and regional instability (FAO, 2016; Maurel and Tuccio, 2016; Werz and Conley, 2012). The development and release of varieties with superior abiotic stress tolerance offers real possibilities to increase production in marginal environments (Comas et al., 2013; Lynch and Brown, 2012; Rogers and Benfey, 2015), which may help reduce instability. Grain legume experts highlight the need for physiologists to identify and introgress drought and low fertility tolerance phenes, or traits, and make use of emerging genetic tools (Mir et al., 2012). We follow York (York et al., 2013) in using the term ‘phene’ to refer to the units of the plant phenotype, rather than the more general ‘trait’. The term ‘metric’ is used for calculated scores describing the root system. Introgressing germplasm from wild types and relatives into modern breeding programs increases genetic diversity and may introduce favorable phenes (Muñoz et al., 2017; Porch et al., 2013; Souter et al., 2017). For these introgression efforts to be more efficient they must be targeted and that requires detailed knowledge relating plant phenotype to function.

! 1! In almost all soils phosphorus is more available in superficial strata due to natural processes of nutrient cycling, aeration, microbial activity and decomposition of organic matter over time. Water availability contrasts with P availability in a terminal drought environment, as soil water content develops a gradient from low to higher water availability with depth. This situation leads to tradeoffs between root deployment to deep or shallow soil zones because deep roots don’t take full advantage of the P enriched topsoil and shallow roots cannot access as much water as deep roots in a terminal drought environment. There are also temporal tradeoffs as soil dries, tortuosity increases and soil nutrients become less available. Root architecture is defined by the spatial configuration of the root system and involves a description of topology, distribution, and angle of root deployment by class (Lynch, 1995). As shown in figure 1, there are 3 major classes of roots in common bean; adventitious or shoot born roots emerge from the hypocotyl, basals emerge from the root and shoot interface and the radical or primary root (Zobel and Waisel, 2010). Basal roots emerge from the transition zone between shoot and root tissue at the base of the hypocotyl. One to four sets of basal roots emerge in distinct whorls of 4 due to a tetrarch xylem patterning, characteristic of bean. Basal root whorl number (BRWN) is easily phenotyped using the cigar roll method 4 days after imbibition begins and genetic variation as well as control for BRWN has been demonstrated (Miguel et al., 2013). Genotypes with greater BRWN also have greater basal root number (BRN) and an increased vertical range of soil exploration (Lynch, 2011).

adventitious

basal

tap

Fig 1. Common bean root system classes. Adventitious or hypocotyl roots emerge from the hypocotyl, basal roots emerge from the base of the hypocotyl, the primary or tap root is the radical. Image on left taken from video available here; http://plantscience.psu.edu/research/labs/roots/methods/field/shovelomics All root classes can develop laterals and root hairs in common bean. Adventitious roots are generally diagravitropic, or horizontal with respect to gravity. Basal roots are plagiogravitropic and have a growth angle range from zero degrees from horizontal to 80 degrees from horizontal and this deployment angle affects exploration of deep or shallow soil. The primary root usually grows directly down in respect to gravity and is assumed to contribute principally to water acquisition. Extensive variation for basal root growth angle (BRGA) and number of adventitious roots exists (Liao et al., 2004; Ochoa et al., 2006).

! 2! Root traits that enhance topsoil exploration contribute to phosphorus tolerance while traits that enhance deep soil exploration enhance terminal drought tolerance (Beebe et al., 2006; Zhu et al., 2010). Greater basal root whorl number, increased root hair length and density, extensive adventitious rooting and shallow BRGA are beneficial for low P tolerance (Ho et al., 2005; Liao et al., 2001a; Miguel et al., 2013; Walk et al., 2006). Deep basals and greater basal root number are helpful for drought tolerance (Ho et al., 2005). Traits that enhance early P acquisition are important because they potentially enable subsequent deep exploration, enhancing the plants ability to explore deep soil and avoid drought (Ho et al., 2005). Various aspects of root architecture can be controlled by P availability and are mediated by ethylene such as BRGA, ARN, and root hair length and density (Basu et al., 2007; Liao et al., 2001b). Rhizoeconomics, or the study of the relative costs and benefits of a particular root phene in a given time and place, is involved in the identification and evaluation of an efficient root system (Lynch, 2007; Lynch et al., 2005). Different root classes, as well as phenes with different anatomical and morphological characteristics, have different construction and maintenance costs. These costs can be revealed by specific root length and root respiration rate. Multiple observations of a root system over time can reveal how spatial root deployment changes. Past studies and modeling highlight the cost effectiveness of root hairs (Bates and Lynch, 2001; Yan et al., 2004) and indicate the potential advantage of early adventitious rooting (Walk et al., 2006). Research on root architecture is less developed than research on shoot traits and shoot architecture and the more complex questions considering the utility of combinations of traits for combined stress environments have received even less attention. Research has revealed that genotypes that excel under one type of stress perform poorly in the other type of stress (Ho et al., 2005). This trade-off is likely due to few root system phenotypes being able to efficiently explore both deep and shallow soil. However, there is reason to believe that root phenotypes that efficiently explore both deep and shallow soil exist and that this di-morphism can enable a plant to perform well under combined phosphorus and drought stress. Since critical edaphic resources are stratified in a contrasting fashion, a spatially and temporally dynamic RSA with increased range of soil exploration may enhance water and P acquisition. BRWN increases the vertical range of soil explored. Dimorphic root pairings target root deployment to both deep and shallow soil domains where phosphorus and water are available. Multiline pairings of contrasting root systems maximize resource acquisition of individuals, minimize interplant competition and act as an insurance policy if drought stress becomes severe since one line has superior water acquisition capacity. Any one of these 3 strategies (BRWN, dimorphism, multilines) for maximizing efficient resource acquisition can be incorporated into traditional plant breeding programs and can have impact on small holder and subsistence farmers. Other strategies

! 3! There may be other strategies for enhancing production in phosphorus and water stress environments. Research in Colombia indicates that elite lines selected under drought conditions also perform better than standard check lines under low phosphorus conditions (Beebe et al., 2008). This indicates that improvements are likely to exist for combined stress environments and that a vigorous root system is involved. However, a vigorous root system alone can only have limited contributions and may be inefficient compared to a root system with specific spatial and temporal characteristics breed for a particular environment. Understanding the types of responses to limited soil water availability will aid in identifying life strategies and this requires differentiating drought tolerance from drought avoidance. Drought resistance is a general term that may refer to both tolerance and avoidance. Drought tolerance is the ability of a plant to continue to function when available soil moisture drops below a particular threshold. It is often associated with sensitive stomatal control and adjustments in leaf area. Desiccation tolerance is a subcategory of drought tolerance and refers to the ability of a plant to withstand reduced water potential, often via osmotic adjustment. Desiccation recovery is the ability of a plant to return to normal function without substantial tissue damage following a period of time below a critical water potential threshold. Most of the root traits described above are meant to avoid limitations in water availability. We assume that maintaining plant water potential above a critical threshold is accomplished by increasing root length density and depth, thus increasing access to soil water. However, there is substantial discussion about the validity of the link between increased root length density (RLD) at depth and drought avoidance, which highlights lack of agreement between empirical studies on the connection between RLD and yield under drought (Vadez, 2014). Other traits related to water transport and use such as xylem characteristics, stomatal control and leaf area should be considered along with RLD (Vadez et al., 2013). Increased RLD at depth does not necessarily mean increased water acquisition and increased water acquisition does not necessarily lead to increased drought tolerance. Water use efficiency must be included in the calculus as the Passioura equation outlined many years ago (yield = water use x water use efficiency x harvest index (Passioura, 1983). In this logic increasing production in a drought environment includes balancing leaf area with the amount of soil water available, which is akin to resource use optimization and maintenance of an allometric trajectory (Anfodillo et al., 2016). Contrasting water use strategies have been identified in common bean that contribute to the discussion on the role of water acquisition and water use efficiency for drought tolerance. Genotypes showing both water saver and water spender strategies have been found to yield well in a single drought environment (Polania et al., 2016). This relationship could conceivably change in different environments, which could favor one strategy over the other but it is remarkable that different strategies are equally important in the same environment. ! !

! 4! References( Anfodillo,!T.,!Petit,!G.,!Sterck,!F.,!Lechthaler,!S.,!Olson,!M.E.,!2016.!Allometric! Trajectories!and!“Stress”:!A!Quantitative!Approach.!Front.!Plant!Sci.!7,!1681.! doi:10.3389/fpls.2016.01681! Basu,!P.,!Brown,!K.,!Lynch,!J.,!2007.!Ethylene!modulates!genetic,!positional,!and! nutritional!regulation!of!root!plagiogravitropism.!Funct.!Plant!Biol.!41–51.! doi:10.1071/FP06209! Bates,!T.R.,!Lynch,!J.P.,!2001.!Root!hairs!confer!a!competitive!advantage!under!low! phosphorus!availability.!Plant!Soil!236,!243–250.! doi:10.1023/A:1012791706800! Beaver,!J.S.,!Rosas,!J.C.,!Myers,!J.,!Acosta,!J.,!Kelly,!J.D.,!Nchimbi\Msolla,!S.,!Misangu,!R.,! Bokosi,!J.,!Temple,!S.,!Arnaud\Santana,!E.,!Coyne,!D.P.,!2003.!Contributions!of!the! Bean/Cowpea!CRSP!to!cultivar!and!germplasm!development!in!common!bean.! F.!Crop.!Res.!82,!87–102.!doi:10.1016/S0378\4290(03)00032\7! Beebe,!S.,!Rao,!I.M.,!Blair,!M.W.,!Butare,!L.,!2009.!Breeding!for!abiotic!stress!tolerance! in!common!bean:!present!and!future!challenges.!Aust.!Plant!Breeding;!SABRAO! Conf.!(14;!11;!2009,!Cairns,!Queensland,!Aust.!Proceedings.!Glob.!Partnersh.! Initiat.!Plant!Breed.!Capacit.!Build.!(GIPB),!Queensland,!AU.!11!p.! Beebe,!S.E.,!Rao,!I.M.,!Cajiao,!C.,!Grajales,!M.,!2008.!Selection!for!Drought!Resistance! in!Common!Bean!Also!Improves!Yield!in!Phosphorus!Limited!and!Favorable! Environments.!Crop!Sci.!48,!582.!doi:10.2135/cropsci2007.07.0404! Beebe,!S.E.,!Rojas\Pierce,!M.,!Yan,!X.,!Blair,!M.W.,!Pedraza,!F.,!Muñoz,!F.,!Tohme,!J.,! Lynch,!J.P.,!2006.!Quantitative!trait!loci!for!root!architecture!traits!correlated! with!phosphorus!acquisition!in!common!bean.!Crop!Sci.!46,!413–423.! doi:10.2135/cropsci2005.0226!

Burridge,!J.D.,!Jochua,!C.N.,!Bucksch,!A.,!Lynch,!J.P.,!2016a.!Legume!shovelomics:! High—Throughput!phenotyping!of!common!bean!(Phaseolus!vulgaris!L.)!and! cowpea!(Vigna!unguiculata!subsp,!unguiculata)!root!architecture!in!the!field.!F.! Crop.!Res.!192,!21–32.!doi:http://dx.doi.org/10.1016/j.fcr.2016.04.008.! Burridge,!J.D.,!Schneider,!H.M.,!Huynh,!B.\L.,!Roberts,!P.A.,!Bucksch,!A.,!Lynch,!J.P.,! 2016b.!Genome\wide!association!mapping!and!agronomic!impact!of!cowpea! root!architecture.!Theor.!Appl.!Genet.!1–13.!

Comas,!L.H.,!Becker,!S.R.,!Cruz,!V.M.!V,!Byrne,!P.F.,!Dierig,!D.!a,!2013.!Root!traits! contributing!to!plant!productivity!under!drought.!Front.!Plant!Sci.!4,!442.!

! 5! doi:10.3389/fpls.2013.00442! FAO,!2016.!Migration,!Agriculture!and!Rural!Development.!Rome.! FAOSTAT,!2017.!FAOSTAT![WWW!Document].!FAOSTAT.! Groisman,!P.,!Knigth,!R.,!Easterline,!D.,!Karl,!T.,!Hegerl,!G.,!Razuvaev,!V.,!2005.!Trends! in!Intense!Precipitation!in!the!Climate!Record.!J.!Clim.!18,!1326–1350.! Ho,!M.D.,!Rosas,!J.C.,!Brown,!K.M.,!Lynch,!J.P.,!2005.!Root!architectural!tradeoffs!for! water!and!phosphorus!acquisition.!Funct.!Plant!Biol.!32,!737–748.! doi:10.1071/FP05043! Hunter,!M.C.,!Smith,!R.G.,!Schipanski,!M.E.,!Atwood,!L.W.,!Mortensen,!D.A.,!2017.! Agriculture!in!2050:!Recalibrating!targets!for!sustainable!intensification.! Bioscience!67,!386–391.!doi:10.1093/biosci/bix010! Liao,!H.,!Rubio,!G.,!Yan,!X.,!Cao,!A.,!Brown,!K.M.,!Lynch,!J.P.,!2001a.!Effect!of! phosphorus!availability!on!basal!root!shallowness!in!common!bean.!Plant!Soil! 232,!69–79.!doi:10.1023/A:1010381919003! Liao,!H.,!Xiaolong,!Y.,!Rubio,!G.,!Beebe,!S.,!Blair,!M.,!Lynch,!J.,!2004.!Genetic!mapping! of!basal!root!gravitropism!and!phosphorus!acquisition!efficiency!in!common! bean.!Funct.!Plant!Biol.!91,!959–970.!doi:10.1071/FP03255! Liao,!H.,!Zhuang,!Y.,!Yan,!X.,!Brown,!K.M.,!Lynch,!J.P.,!2001b.!Ethylene!regulates! common!bean!root!growth!in!adaptation!to!P!deficiency.!Agriculture!550–551.! Lynch,!J.,!1995.!Root!Architecture!and!Plant!Productivity.!Plant!Physiol.!109,!7–13.! doi:10.1104/pp.109.1.7! Lynch,!J.P.,!2011.!Root!phenes!for!enhanced!soil!exploration!and!phosphorus! acquisition:!tools!for!future!crops.!Plant!Physiol.!156,!1041–9.! doi:10.1104/pp.111.175414! Lynch,!J.P.,!2007.!Rhizoeconomics :!The!Roots!of!Shoot!Growth!Limitations.! HortScience!42,!1107–1109.! Lynch,!J.P.,!Brown,!K.M.,!2012.!New!roots!for!agriculture:!exploiting!the!root! phenome.!Philos.!Trans.!R.!Soc.!Lond.!B.!Biol.!Sci.!367,!1598–1604.! doi:10.1098/rstb.2011.0243! Lynch,!J.P.,!Ho,!M.D.,!Phosphorus,!L.,!2005.!Rhizoeconomics:!Carbon!costs!of! phosphorus!acquisition.!Plant!Soil!269,!45–56.!doi:10.1007/s11104\004\1096\ 4!

! 6! Maurel,!M.,!Tuccio,!M.,!2016.!Climate!Instability,!Urbanisation!and!International! Migration.!J.!Dev.!Stud.!52,!735–752.!doi:10.1080/00220388.2015.1121240! Melka,!T.S.,!2010.!The!Moche!Lima!Beans!Recording!System,!Revisited.!Folklore!45,! 89–136.! Miguel,!M.A.,!Widrig,!A.,!Vieira,!R.F.,!Brown,!K.M.,!Lynch,!J.P.,!2013.!Basal!root!whorl! number:!a!modulator!of!phosphorus!acquisition!in!common!bean!(Phaseolus! vulgaris).!Ann.!Bot.!112,!973–82.!doi:10.1093/aob/mct164! Mir,!R.R.,!Zaman\Allah,!M.,!Sreenivasulu,!N.,!Trethowan,!R.,!Varshney,!R.K.,!2012.! Integrated!genomics,!physiology!and!breeding!approaches!for!improving! drought!tolerance!in!crops.!Theor.!Appl.!Genet.!125,!625–45.! doi:10.1007/s00122\012\1904\9! Muñoz,!N.,!Liu,!A.,!Kan,!L.,!Li,!M.,!Lam,!H.,!2017.!Potential!Uses!of!Wild!Germplasms!of! Grain!Legumes!for!Crop!Improvement.!Int.!J.!Mol.!Sci.!18,!1–28.! doi:10.3390/ijms18020328! Ochoa,!I.E.,!Blair,!M.W.,!Lynch,!J.P.,!2006.!QTL!Analysis!of!Adventitious!Root! Formation!in!Common!Bean!under!Contrasting!Phosphorus!Availability.!Crop! Sci.!46,!1609–1621.!doi:10.2135/cropsci2005.12\0446! Polania,!J.A.,!Poschenrieder,!C.,!Beebe,!S.E.,!Rao,!I.M.,!2016.!Effective!Use!of!Water! and!Increased!Dry!Matter!Partitioned!to!Grain!Contribute!to!Yield!of!Common! Bean!Improved!for!Drought!Resistance.!Front.!Plant!Sci.!7.! doi:10.3389/fpls.2016.00660! Porch,!T.G.,!Beaver,!J.S.,!Debouck,!D.G.,!Jackson,!S.A.,!Kelly,!J.D.,!Dempewolf,!H.,!2013.! Use!of!Wild!Relatives!and!Closely!Related!Species!to!Adapt!Common!Bean!to! Climate!Change.!Agronomy!3,!433–461.!doi:10.3390/agronomy3020433! Rogers,!E.D.,!Benfey,!P.N.,!2015.!Regulation!of!plant!root!system!architecture:! implications!for!crop!advancement.!Curr.!Opin.!Biotechnol.!32,!93–98.! doi:10.1016/j.copbio.2014.11.015! Souter,!J.R.,!Gurusamy,!V.,!Porch,!T.G.,!Bett,!K.E.,!2017.!Successful!Introgression!of! Abiotic!Stress!Tolerance!from!Wild!Tepary!Bean!to!Common!Bean.!Crop!Sci.!57,! 1160–1171.!doi:10.2135/cropsci2016.10.0851! St.Clair,!S.B.,!Lynch,!J.P.,!2010.!The!opening!of!Pandora’s!Box:!climate!change!impacts! on!soil!fertility!and!crop!nutrition!in!developing!countries.!Plant!Soil!335,!101– 115.!doi:10.1007/s11104\010\0328\z!

! 7! Steen,!I.,!1998.!Management!of!a!non\renewable!resource.!Phosphorus!and! potassium!25–31.! UN,!2015.!Department!of!Economic!and!Social!Affairs,!Population!Division.!World! Population!Prospects.!The!2015!Revision.!Key!Findings!and!Advance!Tables.! Vadez,!V.,!2014.!Root!hydraulics :!The!forgotten!side!of!roots!in!drought!adaptation.! F.!Crop.!Res.!165,!15–24.!doi:10.1016/j.fcr.2014.03.017! Vadez,!V.,!Kholova,!J.,!Zaman\Allah,!M.,!Belko,!N.,!2013.!Water:!The!most!important! “molecular”!component!of!water!stress!tolerance!research.!Funct.!Plant!Biol.!40,! 1310–1322.!doi:10.1071/FP13149! Walk,!T.C.,!Jaramillo,!R.,!Lynch,!J.P.,!2006.!Architectural!Tradeoffs!between! Adventitious!and!Basal!Roots!for!Phosphorus!Acquisition.!Plant!Soil!279,!347– 366.!doi:10.1007/s11104\005\0389\6! Werz,!M.,!Conley,!L.,!2012.!Climate!Change,!Migration!and!Conflict;!Addressing! complec!crisis!scenarios!in!the!21st!Centry.!Washington!D.C.! Yan,!X.,!Liao,!H.,!Beebe,!S.,!Blair,!M.,!Lynch,!J.,!2004.!QTL!mapping!of!root!hair!and! acid!exudation!traits!and!their!relationship!to!phosphorus!uptake!in!common! bean.!Plant!Soil!265,!17–29.!doi:10.1007/s11104\005\0693\1! York,!L.M.,!Nord,!E.!a,!Lynch,!J.P.,!2013.!Integration!of!root!phenes!for!soil!resource! acquisition.!Front.!Plant!Sci.!4.!doi:10.3389/fpls.2013.00355! Zhu,!J.,!Brown,!K.M.,!Lynch,!J.P.,!2010.!Root!cortical!aerenchyma!improves!the! drought!tolerance!of!maize!(Zea!mays!L.).!Plant.!Cell!Environ.!33,!740–9.! doi:10.1111/j.1365\3040.2009.02099.x! Zobel,!R.,!Waisel,!Y.,!2010.!A!plant!root!system!architectural!taxonomy:!a!framework! for!root!nomenclature.!Plant!Biosyst.! !

! 8!

Field Crops Research 192 (2016) 21–32

Contents lists available at ScienceDirect

Field Crops Research

jou rnal homepage: www.elsevier.com/locate/fcr

Legume shovelomics: High—Throughput phenotyping of common

bean (Phaseolus vulgaris L.) and cowpea (Vigna unguiculata subsp,

unguiculata) root architecture in the ﬁeld

a,1 a,b,1 c,d a,

James Burridge , Celestina N. Jochua , Alexander Bucksch , Jonathan P. Lynch ∗

Department of Plant Science, The Pennsylvania State University, 102 Tyson Building, University Park, PA 16802, USA

Instituto de Investigac¸ ão Agrária de Moc¸ ambique (IIAM), Av. FPLM no. 2698, P. O. Box 3658, Maputo, Mozambique

School of Interactive Computing, Georgia Institute of Technology, Tech Square Research Building, 85 Fifth Street NW, Atlanta, GA 30308, USA

School of Biology, Georgia Institute of Technology, 310 Ferst Drive, Atlanta, GA, 30332, USA

a r t i c l e i n f o a b s t r a c t

Article history: Low phosphorus (P) availability and drought are primary constraints to common bean and cowpea pro-

Received 26 January 2016

duction in developing countries. Genetic variation of particular root architectural phenes of common

Received in revised form 4 April 2016

bean is associated with improved acquisition of water and phosphorus. Quantitative evaluation of root

Accepted 4 April 2016

architectural phenotypes of mature plants in the ﬁeld is challenging Nonetheless, in situ phenotyping

Available online 30 April 2016

captures responses to environmental variation and is critical to improving crop performance in the tar-

get environment. The objective of this study was to develop ﬂexible high-throughput root architectural

Keywords:

phenotyping platforms for bean and cowpea, which have distinct but comparable root architectures. The

Root

Architecture bean phenotyping platform was speciﬁcally designed to scale from the lab to the ﬁeld. Initial labora-

Cowpea tory studies revealed cowpea does not have basal root whorls so the cowpea phenotypic platform was

Common bean taken directly to ﬁeld evaluation. Protocol development passed through several stages including compar-

Phenotyping isons of lab to ﬁeld quantiﬁcation systems and comparing manual and image-based phenotyping tools

of ﬁeld grown roots. Comparing lab-grown bean seedlings and ﬁeld measurements at pod elongation

stage resulted in a R of 0.66 for basal root whorl number (BRWN) and 0.92 for basal root number (BRN)

between lab and ﬁeld observations. Visual ratings were found to agree well with manual measurements

for 12 root parameters of common bean. Heritability for 51 traits ranged from zero to eighty-three, with

greatest heritability for BRWN and least for disease and secondary branching traits. Heritability for cow-

pea traits ranged from 0.01 to 0.80 to with number of large hypocotyl roots (1.5A) being most heritable,

nodule score (NS) and tap root diameter at 5 cm (TD5) being moderately heritable and tap root diameter

15 cm below the soil level (TD15) being least heritable. Two minutes per root crown were required to

evaluate 12 root phene descriptors manually and image analysis required 1 h to analyze 5000 images

for 39 phenes. Manual and image-based platforms can differentiate ﬁeld-grown genotypes on the basis

of these traits. We suggest an integrated protocol combining visual scoring, manual measurements, and

image analysis. The integrated phenotyping platform presented here has utility for identifying and select-

ing useful root architectural phenotypes for bean and cowpea and potentially extends to other annual

legume or dicotyledonous crops.

1. Introduction foods in many areas of the tropics and sub-tropics and make sig-

niﬁcant contributions to food security, income, and soil fertility

Global food production must double by the year 2050 in order (Beebe, 2012; Ehlers and Hall, 1997). Both legumes are impor-

to meet the projected demand (Tilman et al., 2011). Common bean tant crops for smallholder farmers with limited access to irrigation

(Phaseolus vulgaris L.) and cowpea (Vigna unguiculata) are staple and fertilizers and their production is critical for human nutrition

and agroecosystem productivity. Their production is challenged by

increasingly marginal soils and climate instability, amplifying the

need to develop high yielding cultivars under drought and low soil

∗ Corresponding author.

E-mail address: [email protected] (J.P. Lynch). fertility (Lynch, 2007; Wortmann et al., 1998).

These authors contributed equally.

http://dx.doi.org/10.1016/j.fcr.2016.04.008

22 J. Burridge et al. / Field Crops Research 192 (2016) 21–32

Root architecture is an important factor for the acquisition other media systems are able to extract complex 3D root architec-

of key soil resources including nitrogen and water (Lynch and ture traits (Clark et al., 2011; Iyer-Pascuzzi et al., 2010; Mairhofer

Brown, 2001; Lynch, 2013) and is particularly important for the et al., 2013; Metzner et al., 2015; Rellán-Álvarez et al., 2015; Schulz

highly immobile and commonly limiting nutrient phosphorus (P) et al., 2013; Topp et al., 2013). These imaging platforms implement a

(Lynch, 1995). Bean exhibits considerable genetic diversity for root non-invasive measurement procedure that allows capturing times

architectural traits related to growth in low P and water limited series of one individual root. However, pot size introduces root

environments (Lynch and Beebe, 1995; Bonser et al., 1996; Miller growth artifacts (Poorter et al., 2012) and to reduce these artifacts

et al., 2003; Miguel, 2004; Ochoa et al., 2006; Rubio et al., 2003; trials must be limited to timeframes of 2–3 weeks since roots that

Ho et al., 2005). Greater root hair length and density increase root hit the side of the pot should be discarded. At that stage some root

surface area and enhance phosphorus uptake but will not be dealt traits of bean and cowpea, e.g. hypocotyl root number and tap root

with here, as their quantiﬁcation requires a different sampling pro- diameter, have not fully developed.

cedure (Brown et al., 2013; Lynch, 2011; Miguel et al., 2015). Several In contrast, phenotyping of root architecture in real soils is

quantitative trait loci (QTL) of architectural traits have been iden- challenging because soils are diverse and opaque. For that rea-

tiﬁed in common bean using simple visual evaluations of phene son, ‘Shovelomics’ (Trachsel et al., 2011) complements laboratory

descriptors (‘phene’ is to ‘phenotype’ as ‘gene’ is to ‘genotype’, sensu platforms in that it permits phenotyping of mature roots in actual

(Lynch, 2011; Serebrovsky, 1925; York et al., 2013)) including basal soil in the ﬁeld. Shovelomics deﬁnes visual scores for 10 architec-

root growth angle (BRGA) (Ho et al., 2005; Liao et al., 2004), basal tural root phenes of maize crowns at the rate of 2 min per plot.

root whorl number (BRWN) (Miguel et al., 2013), and hypocotyl Shovelomics is a simple, robust and inexpensive tool for crop breed-

root number (Ochoa et al., 2006; Walk et al., 2006). Note that some ers to evaluate root systems and functional responses to varying

literature uses the term adventitious root rather than the more environments. However, the structurally and functionally dissim-

correct hypocotyl root (Zobel, 2011). The impact of genetic and ilar root architecture and development of maize and bean makes

phenotypic variation of cowpea root architectural traits on perfor- a unique legume protocol necessary. Key differences that distin-

mance has received less attention. guish legumes and grasses include the occurrence of secondary

Common bean and cowpea both have epigeal germination and growth and the long- term contribution of the embryonic root sys-

have embryonic root systems composed of the primary root and tem in legumes. A key example is the role of legume primary roots

basal roots. Following germination hypocotyl roots develop and latin extracting deeply available soil water, which involves signiﬁ-

eral roots develop from tap root, basal and hypocotyl roots (Lynch cant secondary thickening. In ﬁeld grown maize the primary root

and Beem, 1991). The number and growth angle of basal roots in is not always identiﬁably functional in mature plants and hypocotyl

bean vary and are organized in discrete whorls, each with capacity (nodal) roots become important for resource acquisition deeper in

for 4 basal roots (Basu et al., 2007). In bean, hypocotyl roots typically the soil proﬁle (Saengwilai et al., 2014).

grow horizontally from the hypocotyl (Bonser et al., 1996; Miller Recently, an automated image analysis platform, DIRT (Dig-

et al., 2003). In contrast, we have observed both hypocotyl and ital Imaging of Root Traits) has been validated for quantifying

basal roots in cowpea to have large variation in plagiogravitropism. root architecture of ﬁeld-excavated root crowns (Bucksch et al.,

Another major difference with common bean is that cowpea basal 2014). DIRT is an automated image analysis software developed

roots are not arranged in discrete whorls and are not always clearly for quantifying and differentiating crop root phenotypes. Over sev-

distinct from hypocotyl or primary root laterals. In common bean enty DIRT traits can be objectively and automatically extracted

the primary (tap) root is orthogravitropic (Basu et al., 2007) and from thousands of images in under an hour, shifting the bottle-

cowpea shows the same tendency. Variation between cowpea and neck to root excavation and washing. Several novel DIRT traits have

common bean represents phenotypic extremes along the spec- no manual analog due to their mathematical deﬁnition. However,

trum of root architectural phenotypes observed in grain legumes. the 2D projection to a digital image inhibits DIRT (Bucksch et al.,

Therefore, these two species can serve as models for other legumes 2014) to extract traits that are observable in the third dimension.

including soybean (Glycine max), tepary bean (Phaseolus acutifo- Here, manual phenotyping complements DIRT through mechanical

lious), fava bean (Vicia faba), chickpea (Cicer arietinum), pigeon pea testing of tissue ﬂexibility and visual color assessment. Together,

(Cajanus cajan), and groundnut (Arachis hypogaea). image-based and manual ﬁeld evaluations constitute an untapped

Phenotyping root traits related to edaphic stress tolerance is potential to conﬁrm phene utility, responses to environmental fac-

a bottleneck that limits genetic analysis and crop improvement tors and to appropriately match a phenotype to an environment.

(Varshney et al., 2014). A low-cost, ﬁeld-based phenotyping plat- Four development steps were taken to formulate a shovelomics

form would enable plant breeders with limited economic resources protocol for common bean and cowpea. First we determined root

to address local environmental constraints. Genotype by envi- trait scaling from laboratory to ﬁeld environments in bean. Sec-

ronment (GxE) and genotype by environment by management ond, two different rapid ﬁeld methods to evaluate root architectural

(GxExM) interactions are of particular interest to breeding pro- phenotypes were developed for bean and cowpea. Third, we com-

grams that require phenotyping in the agroecosystem of interest. pared a rating based system and a quantitative measurement

Greenhouse trials are typically reliant upon pots that generally limit system in bean. A ﬁnal analysis was carried out to compare visual

soil volume and restrict root growth and development (Poorter trait scores, manual trait measurements and image-based trait esti-

et al., 2012). which can obscure and constrain plant responses mation for bean and cowpea. On the basis of these results a protocol

to environmental stress. Many studies of root architecture uti- is recommended.

lize young plants in controlled environments in the interest of

lower cost and higher throughput. Frequently used experimental

techniques include paper germination rolls, pouches and sand or 2. Material and methods

media-ﬁlled pots. These platforms facilitate physiological studies

(York et al., 2013) and permit the phenotypic and genotypic charac- 2.1. Common bean

terization of large sets of genotypes (for a review see (Lynch, 2011))

but also do not represent natural soil conditions. Advances in imag- 2.1.1. Bean laboratory experiment

ing software have expanded the throughput of root phenotyping A customized panel of sixty-four genotypes was obtained from

platforms (Basu and Pal, 2012; Clark et al., 2013; De Smet et al., CIAT (Centro Internacional de Agricultura Tropical) were used in

2012; Galkovskyi et al., 2012; Zhu et al., 2011). MRI, X-Ray, CT, and this study (Table S1). Genotypes were selected based on diversity

J. Burridge et al. / Field Crops Research 192 (2016) 21–32 23

in root phenotypes, and tolerance to low P and drought conditions. In the Rock Springs, Pennsylvania 2014 trial 12 Recombinant

Genotypes G 19833 and DOR 364 were included as checks with Inbred Lines (RILs) from the DOR364 BAT477 population (Blair

contrasting adaptation to low P. G 19833 is an Andean genotype et al., 2012) were grown under non-limiting and water limiting

tolerant to low P (Beebe et al., 1997; CIAT, 1996) and has shallow conditions. The trial had 4 replications, between-row spacing of

basal roots (less than 25◦ from horizontal) (Liao et al., 2001; Lynch, 76 cm and in-row spacing of 10 cm. Irrigation was withheld from

1995), three basal root whorls (Basu et al., 2007) and many (more the water-limited plots 14 days after emergence using automated

than 25) hypocotyl roots (Miller et al., 2003). DOR 364, from the rainout shelters (i.e. the ﬁeld was protected from precipitation

Mesoamerican gene pool, has poor yield under P deﬁciency (Beebe by plastic greenhouse superstructures activated by precipitation

et al., 1997), steeper (greater than 30◦) basal roots (Liao et al., 2001) sensors but in dry weather were positioned off the ﬁeld) and irri-

and two basal root whorls (Basu et al., 2007). Five bean genotypes gation was applied to the well watered treatment when needed to

obtained from the Agricultural Research Institute of Mozambique maintain soil water potential near ﬁeld capacity. Weed and pest

(IIAM): Doutor, LIC-04-1-3, Diacol Calima, Ica Pijão and one com- management were applied as needed.

mercial variety, Bonus, were included as checks. Twenty accessions

evaluated in Rock Springs were a subset of the CIAT bean core col-

lection composed of accessions from different races and geographic

regions (Colombia, El Salvador, Guatemala, Mexico, Ecuador, Peru, 2.2. Cowpea ﬁeld experiments

Brazil and Haiti), and representing Andean and Middle American

gene pools (Beebe et al., 2000). Seeds of 188 lines of a cowpea diversity panel, a subset of

Four samples of each genotype in an individual germination a larger 422 entry diversity panel (Huynh et al., 2013), repre-

paper roll with 4 replications over time were evaluated in the senting worldwide diversity were obtained from University of

laboratory as a randomized complete block design (RCBD) at The California—Riverside (Table S3). Cowpea ﬁeld trials were conducted

Pennsylvania State University (PSU) in 2006. Seeds were surface- in 2012 and 2013 at Ukulima Root Biology Center (URBC), Limpopo

sterilized for 1–2 min with 10% (v/v) NaOCl, rinsed with deionized South Africa (24◦33′00.12 S, 28◦07′25.84 E, 1235 m.a.s.l). URBC has

water, mechanically scariﬁed with a razor and germinated in rolls of a loamy sand (Clovelly Plinthic Entisol) and was fertilized, irrigated

brown germination paper No 78 (Anchor Paper Company, St. Paul, and had pesticides applied to provide non-limiting growing con-

MN, USA). The rolls were placed upright in 5 l beakers containing 1 l ditions. The trials were arranged in RCBD with 4 blocks and each

of 0.5 mM CaSO4. Seeds were allowed to germinate in darkness at experimental unit consisted of a single 4 m row per genotype. Seeds

28 ◦C for 3–4 days. The seedlings were then placed in a plant culture were planted with a jab planter 4 cm below the soil surface, 30 cm

2 1

room at 26 ◦C for 4 days with 12 h of light (120 ␮m m− s− ). Basal apart in row and 76 cm between rows resulting in a plant density

root whorl number and total number of basal roots were counted of 100,000/ha. Two representative samples of the four excavated

on 8 day old seedlings. samples were visually evaluated for 11 roots traits and an image of

the root crown with identifying tag and scale marker was taken.

2.1.2. Bean ﬁeld experiments

Field trials were conducted at the IIAM Agriculture Research Sta-

tion of Chokwe, (24◦31′S; 33◦0′E, 40 m.a.s.l) in 2008 and 2009, the

Agriculture Research Station of Umbeluzi, Mozambique (26◦03′S; 2.3. Excavation and evaluation

32◦21′E, 64 m.a.s.l) in 2008, and the Russell Larson Agricultural

Research Station of The Pennsylvania State University in Rock Root crowns of bean and cowpea were excavated from all

Springs, Pennsylvania, USA (40◦44′N; 77◦53′W, 366 m.a.s.l.) in experiments between 35 and 45 days after planting (DAP) using

2010. The soil in Chokwe is a Mollic Ustiﬂuvent with silt-loam tex- a standard spade. Evaluation during the ﬂowering period mini-

ture (Mollic Fluvisols, FAO, 1988), while the soil at the Umbeluzi site mizes differences in phenology among lines that may affect root

is a Mollic Ustiﬂuvent with sandy-loam texture (Eutric Fluvisols, development. Also, at this stage the mature root phenotype is evi-

FAO, 1988). The P availability in Chokwe was 38 ppm (Olsen), with dent. Excavation volume was deﬁned by a cylinder with a radius

pH of 6.8 and 1.8% organic matter, and in Umbeluzi the P availabil- of 25 cm around the shoot and a depth of 25 cm. As much as possi-

ity was 20 ppm (Olsen). In Rock Springs, the genotypes were grown ble of the soil cylinder with the root inside was removed from the

in a Hagerstown silt loam soil (ﬁne, mixed, semi-active, medic surrounding soil and gently washed in water. The looser soils of

Typic Hapludult). The P availability at the ﬁeld in Rock Springs was Umbeluzi and URBC permitted only the removal of the root crown

10.5 ppm (P—Mehlich 3 extraction). with associated soil aggregates rather than the entire soil cylinder.

Thirty genotypes (Table S2) were planted in RCBD in Chokwe In the higher clay content soils of Rock Springs (30 g clay/kg soil)

in 2008 and 2009, and in Umbeluzi in 2008. The experiment con- 0.5% v/v detergent was added and the root cylinders were soaked

sisted of 4 replications and each experimental unit was composed prior to washing. Total time required from excavation to evalua-

of two rows of 5 m. Twenty-ﬁve seeds were sown in each row with tion varied from 4 min per plot in soils with sandy-loam texture

spacing of 0.7 m between rows and 0.2 m between plants in a row. to 11 min per plot in silt-loam soils with evaluation of 12 phene

Nitrogen in the form of urea was applied 25 days after planting at descriptors requiring 2 min for an experienced phenotyper in all

a rate of 30 kg N/ha in trials conducted in Chokwe and Umbeluzi. soils (Table 1). Root crowns evaluated in Rock Springs required

Phosphorus was not applied in any trial. Irrigation, weed and pest additional time (approximately 8.5 min) for soaking and washing

management were applied as needed. the roots while root crowns evaluated in the sandier soils of South

In 2010, twenty accessions of the bean core collection from dif- Africa and Mozambique required only a brief water rinse. Excavated

ferent gene pools (Beebe et al., 2000) (Supplementary materials roots were evaluated manually and using the image-based traits

Table S2) were evaluated under low phosphorus in Rock Springs estimation of DIRT. Manual measurements were taken as described

Pennsylvania in order to compare values of measured and visu- below. Images were taken and analyzed according the DIRT proto-

ally scored root traits. The experiment was planted in a RCBD with col (Bucksch et al., 2014). In brief, the additional requirements for

4 replications. Seeds of each genotype were sown in one row of image analysis includes taking a standard photograph of the root

1.6 m, and the space between rows was 0.7 m and in row plant crown on a ﬂat black background with a scale marker, recording

spacing was 0.2 m. Each experimental unit had 8 plants. Irrigation, the image number with plot identiﬁer information and uploading

weed and pest management were applied as needed. the image to the online DIRT platform.

24 J. Burridge et al. / Field Crops Research 192 (2016) 21–32

Table 1

Time required for field evaluation of 12 root phenes from one crown in different soil textures: Chokwe: Mollic Ustifluvent (silt-loam texture), Umbeluzi, Mollic Ustifluvents

(sandy-loam texture), Ukulima, Clovelly Plinthic Entisol (unstructured young sandy texture) and Rock Springs Typic Hapludalf (silt-loam texture). “-“ indicates root crowns

were not washed.

Activity Mollic Ustiﬂuvents (Chokwe) Mollic Ustiﬂuvents (Umbeluzi) Clovely Plinthic Entisol (Ukulima) Typic Hapludalf (Rock Springs)

Crown excavation 2.0 min 2.0 min 2.0 min 2.5 min

Soaking - - - 5.0 min

Washing - - 1.0 min 1.5 min

Evaluation 2.0 min 2.0 min 2.0 min 2.0 min

Total 4.0 min 4.0 min 5.0 min 11 min

Fig. 1. Photographs of common bean root crowns showing phene states. Trait deﬁnitions not in Table 2 are listed here: tap root diameter (TD) is measured 2 cm below the

basal roots; third order branching density (3BD) is a rating of overall root system branching density including 2nd and 3rd order laterals: nodulation score (NS) is a rating of

visible functional nodules that takes into account both number and size—high (30 + ), middle (10–30), and low (0–10), disease score D is a 1–9 rating of root disease where 1

is no visible symptoms of disease and 9 is severely infected.

J. Burridge et al. / Field Crops Research 192 (2016) 21–32 25

Table 2

Description of common bean root measurements used in Chokwe, Umbeluzi (2008 and 2009) and Rock Springs 2010.

Trait name Abbreviation Method Deﬁnition

Hypocotyl root number HRN Count number of visibly functional hypocotyl roots

Hypocotyl root length HRL Board 1 = 1 cm; 3 = 4–5 cm; 5 = 8–9 cm; 7 = 12–13 cm; 9 = 15–20 cm

≤

Hypocotyl root branching HRB Board 1 = no lateral branching; 5 = 2 orders of laterals; 9 = 4 orders of dense laterals

Basal root whorl number BRWN Count Count sets of 4 basal root whorls, typical range is 1–4

Basal root number BRN Count Number of basal roots

Basal root branching BRB Board 1 = no lateral branching; 5 = 2 orders of laterals; 9 = 4 orders of dense laterals

Basal root growth angle BRGA Board Approximate angle (degrees in groups of 10) where basal roots intersect 10 cm

arc on board when root origin placed in center. Zero is horizontal and 90◦ is

vertical.

Basal root length BRL Board 1 = 1 cm; 3 = 4–5 cm; 5 = 8–9 cm; 7 = 12–13 cm; 9 = 15–20 cm

≤

Primary root length PRL Board 1 3 cm; 3 = 7–9 cm; 5 = 13–15 cm; 7 = 19–21 cm; 9 = 25–30 cm (depth of ≤

excavation)

Primary root branching PRB Board 1 = no lateral branching; 5 = 2 orders of laterals; 9 = 4 orders of dense laterals

Nodulation NN or NS Rating Estimation of active nodule number 1 = > 80; 3 = 41–80; 5 = 21-40 nodules;

7 = 10-20; 9 = poor < 10

Root Rot RR Rating 1 = no visible symptoms; 3 = 10%, 5 = 25%; 7 = 50%; 9 = 75% or more of hypocotyl

and root with severe lesions

In Rock Springs 2014 we excavated four root crowns per plot

and evaluated two root crowns for 6 root traits (BRN, HRN, BRGA,

3BD, TBD, DS) using the phenotyping board technique described

above. We imaged all visually evaluated root crowns to obtain stem

diameter, tap root diameter, hypocotyl root number and branching

density using a proprietary plugin to Image J (available here). The

same images were used to estimate root traits with the DIRT plat-

form (Bucksch et al., 2014) and made available to the plant science

community at http://dirt.iplantcollaborative.org/ (Das et al., 2015).

2.4.2. Cowpea

The two most representative samples of the four roots exca-

vated were chosen for evaluation, based upon overall root system

size and branching density. Root crowns were evaluated for the

following 11 parameters both years using the phenotyping board

technique described above; tap root diameter 5 cm from soil surface

(TD5), tap diameter 10 cm below soil surface (TD10), tap diame-

Fig. 2. Annotated image of a common bean root crown highlighting important com-

ponents. ter 15 cm below soil surface (TD15), hypocotyl root number (HRN),

basal region root number (BRN), number of tap root laterals below

the basals to 10 cm from soil level (BD10), number of hypocotyl

roots with diameter greater than 1.5 mm 2 cm from origin (1.5A),

2.4. Evaluation of root traits number of basal roots with diameter greater than 1.5 mm 2 cm

from origin (1.5B), root system score for branching density (3BD),

2.4.1. Bean nodulation score (NS) and disease score (D) (Table 3, Figs. 3 and 4,).

In Mozambique in 2008 and 2009 9 bean traits were visu- In 2013, 6 additional parameters were evaluated: stem diam-

ally evaluated on a linear scale from 1 to 9, where 1 denotes the eter at soil surface (SD), tap diameter 20 cm below soil surface

minimum expression of a trait and 9 the maximum (Table 2 and (TD20), tap diameter 25 cm below soil surface (TD25), dominant

Fig. 1): hypocotyl root length (HRL); hypocotyl root branching (HB); hypocotyl root angle (ARGA), dominant basal region root angle

basal root growth angle (BRGA); basal root length (BRL); basal root (BRGA), and number of 1st order lateral roots 10–15 cm below

branching density (BB); primary root length (PRL); primary root soil surface (BD15). The root crown was placed on a non-reﬂective

branching density (PB); number of nodules (NN); and root rots (RR). black background and an image taken with a tripod mounted digital

Counts were taken for the total number of hypocotyl and basal camera for image analysis.

roots (HRN, BRN), and basal root whorls (BRWN). One represen-

tative sample, based upon overall size, branching density and root 2.5. Statistical analysis

deployment pattern was scored for each root trait per replication

after observation of 4 root crowns. Data from Mozambique and from greenhouse and ﬁeld exper-

In Rock Springs 2010 root crowns of 20 bean accessions were iments in Rock Springs 2010 were analyzed using Minitab 16

evaluated using the 1–9 scoring system described above. A phe- statistical software (Minitab Inc., State College, Pennsylvania, USA),

notyping board containing a length scale and protractor was used and Statistix, version 8 (Analytical Software, Tallahassee, FL, USA).

to obtain quantitative measures of root traits including length Analysis of variance was performed separately for laboratory and

of hypocotyl, basal and primary roots and basal root angle. Root ﬁeld experiments. Genotype and year were considered ﬁxed effects

branching density was determined by a count of lateral roots in a for experiments for Chokwe 2008 and 2009, while location was a

representative (based upon a visual scan of the entire sample) 2 cm ﬁxed effect for Umbeluzi and Chokwe 2008 and block was con-

segment of hypocotyl, basal and primary roots (Fig. 2). Stem and sidered random for both. For laboratory and ﬁeld experiments in

tap root diameter were measured using a standard digital caliper Rock Springs, genotypes were considered ﬁxed factors. Correlation

with 0.01 mm resolution, at the soil level and 2 cm below the base analysis was performed to determine relationship among phene

of the hypocotyl, respectively. descriptors, and to compare laboratory versus ﬁeld results as well

26 J. Burridge et al. / Field Crops Research 192 (2016) 21–32

Fig. 3. Cowpea root crowns showing phene states. Trait deﬁnitions are listed in Table 3.

visual scores and measured root phenes values. Broad sense her- where Vg is genotypic variance, Vgv is the genotypic by year variance

itability was calculated using Fehr’s method (see below) across and Ve is the residual variance (Fehr, 1993).

year for Chokwe 2008 and 2009 and by environment (location) in

Chokwe and Umbeluzi 2008.

3. Results

Data from cowpea trials were analyzed using R (R Core Team,

2014). Correlation analysis was performed using the Pearson

3.1. Common bean

method, and analysis of variance using the analysis of variance

function with genotype, year and genotype by year as ﬁxed fac-

3.1.1. Root phenotypic variability

tors and block as a random effect. The broad sense heritability H

Common bean genotypes differed signiﬁcantly for basal root

across seasons was estimated using Fehr’s method:

whorl number (BRWN) and basal root number (BRN) in 8 d old

seedlings (p 0.001; Table S1). We observed large phenotypic Vg ≤

variation for most root phene descriptors evaluated in the ﬁeld = Vgv Ve

+ #ofreps + #ofyears including BRWN, which ranged from 1 to 3.75 and BRN, which

J. Burridge et al. / Field Crops Research 192 (2016) 21–32 27

Table 3

Description of manual cowpea root architectural measurements used URBC 2012 and 2013.

Trait name Abbreviation Method Deﬁnition

Hypocotyl root growth angle HRGA Board approximate angle where hypocotyl roots intersect 10 cm arc on board when

root origin placed in center

Basal root growth angle BRGA Board approximate angle where basal roots intersect 10 cm arc on board when root

origin placed in center

Stem diameter SD Caliper stem diameter at soil level

Tap diameter 5 cm TD5 Caliper tap diameter 5 cm below soil surface or just below basal region

Tap diameter 10 cm TD10 Caliper tap diameter 10 below soil surface

Tap diameter 15 cm TD15 Caliper tap diameter 15 cm below soil surface

Basal root number BRN Count number of basal region roots

Hypocotyl root number HRN Count number of hypocotyl roots

Branching Density 10 BD10 Count number of 1st order lateral roots (not counting basal region roots) between 5

and 10 cm or from just below basal region to 10 cm below soil level

Branching density 15 BD15 Count number of 1st order lateral roots between 10 and 15 cm below soil surface

Hypocotyl roots1.5 mm or larger 1.5H Count number of hypocotyl roots with diameter greater than 1.5 mm 2 cm from

hypocotyl

Basal roots 1.5 mm or larger 1.5B Count number of basal roots with diameter greater than 1.5 mm 2 cm from hypocotyl

Number of large tap root laterals 1.5BD10 Count number of roots in region 5–10 cm below soil surface with diameter greater

than 1.5 mm 2 cm from hypocotyl

3rd order branching density 3BD Rating 3 order branching density score 1 = few, 9 = many

Nodulation NS Rating nodulation score 1 = none, 9 = many large nodules

Disease D Rating disease score 1 = healthy, 9 = severely affected

3.1.2. Field and laboratory evaluations were highly correlated

Bean BRWN evaluated in 8 d old seedlings in the laboratory was

highly correlated with BRN evaluated in 45 d old plants in the ﬁeld

in Chokwe in 2009 (R = 0.803, p < 0.001). BRWN and BRN evaluated

both in the laboratory and in the ﬁeld were also highly correlated

2 2

(R = 0.93 and R = 0.66, p < 0.001, respectively, Fig. S4). BRN eval-

uated in 8 d old seedlings in the laboratory was greater than the

BRN evaluated in 45 d old plants in the ﬁeld, suggesting root loss

in the ﬁeld. BRWN was strongly correlated with BRN when both

were evaluated in the laboratory (R = 0.949, p 0.001), or ﬁeld

2 ≤

(R = 0.867, p 0.001, Fig. S5).

≤

3.1.3. Root phenotypes were consistent across years and

environments

Fig. 4. Annotated image of a cowpea root crown highlighting important compo- Effects of genotype by environment and genotype by year inter-

nents. actions were evaluated using ANOVA for all 12 phene descriptors.

Besides HRN (p < 0.01) and root rot infection (p < 0.01) we did not

observe signiﬁcant interactions of genotype with year or environ-

ranged from 4 to 13.5 (Fig. S2 and S3). Signiﬁcant differences among ment on phenotypes (Table 4). Phene descriptors with moderately

genotypes within environment and within year were observed for high heritability in 2 years in the same environment and in 2 envi-

hypocotyl root number and branching, basal root growth angle, ronments in the same year include BRWN (86, 83), BRN (79, 74) and

BRWN, BRN, and primary root length (Table 4, summary statistics BRGA (67, 60) and one with moderately high heritability is HRN

in Table S4). (56, 53) (Tables 5 and 6). A graphical representation of the range,

Table 4

ANOVA for root traits in two environments (Chokwe and Umbeluzi) and two years (2008 and 2009 in Chokwe) in 30 genotypes. F values and signiﬁcance levels for the

effect of the environment (E), year, and genotype (G) within environment and year, and interactions G by E and G by year are shown for the following traits: hypocotyl root

number (HRN), hypocotyl root length (HRL), hypocotyl root branching (HRB), basal root whorl number (BRWN), basal root number (BRN), basal root length (BRL), basal root

branching (BRB), basal root growth angle (BRGA), primary root length (PRL), primary root branching (PRB), number of nodules (Nod) and root rot (RRot). Level of signiﬁcance:

*** = significant at p < 0.001, ** = significant at p < 0.01, * - significant at p < 0.05, ns = not significant. G = genotype and E = environment.

HRN HRL HRB BRWN BRN BRL BRB BRGA PRL PRB NN RR

E 301.8 8.58 14.05 0.92 1.48 23.07 0.50 65.4 170.47 3.70 4.55 15.39

*** ** *** ns ns *** ns *** *** ns * ***

Geno 6.80 1.23 1.44 19.71 12.04 1.96 1.14 6.18 1.72 1.74 1.03 1.47

*** ns *** *** * ns *** * * ns ns

G*E 1.78 0.92 1.49 0.16 0.86 1.28 0.61 0.56 1.06 0.73 0.71 1.72

* ns ns ns ns ns ns ns ns ns ns *

Year 10.45 20.52 0.15 0.71 19.62 1.84 12.97 0.10 8.75 24.94 1.85 1.79

** *** ns ns *** ns *** ns ** *** ns ns

Geno 7.26 1.26 2.07 23.87 16.81 1.57 1.43 8.58 1.83 1.36 1.16 1.09

*** ns ** *** *** * ns *** ** ns ns ns

G*Year 1.69 0.72 1.04 0.43 1.24 0.91 0.65 0.74 1.38 1.18 0.76 1.56

* ns ns ns ns ns ns ns ns ns ns *

28 J. Burridge et al. / Field Crops Research 192 (2016) 21–32

Table 5

Estimation of broad sense heritability (h2) from two years of data from one location,

Chokwe 2008 and 2009 100* Vg/[Vg + (Vgy/# of reps) + (Ve/# of years)].

Variance component (V)

2008 and 2009 Year Genotype G*Y Error H

HRN 2.282 21.748 5.401 31.36 0.56

HRL 0.3088 0.1267 0.133 1.8716 0.12

−

HRB 0.00323 0.05625 0.0047 0.4357 0.20

−

BRWN 0.0002 0.2758 0.0133 0.0942 0.86

−

BRN 0.1979 2.5148 0.0771 1.2918 0.79

BRL 0.0093 0.0996 0.0271 1.2004 0.14

−

BRB 0.0338 0.0321 0.0285 0.3291 0.17

−

BRGA 0.0078 1.4305 0.0947 1.4601 0.67

− −

PRL 0.0491 0.0448 0.0768 0.8006 0.10

PRB 0.1113 0.0129 0.0253 0.5621 0.04

NN 0.0029 0.0163 0.0196 0.3248 0.09

−

RR 0.00029 0.0088 0.0208 0.1488 0.12

− −

mean and median indicates which traits may be better suited to

differentiate genotypes (Fig. 5).

3.1.4. Scored vs. measured root traits in common bean

In order to validate our ﬁeld visual root scoring method values

of measured phenes were compared with values from visual scor-

ing. Signiﬁcant differences were found between genotypes for all

measured and visually scored root phenes, except for primary root

branching (Tables S5, S6). In addition, correlations between mea-

sured and visually scored phenes of twenty genotypes evaluated in

Rock Springs 2010 varied from low to high (0.31 for BRB to 0.76 for

BRGA) and all were statistically signiﬁcant (Table S7). High corre-

lations were found for BRGA (0.755), HRL (0.733), PRL (0.644), and Fig. 5. Graphical representation of coefﬁcient of variation of observations for com-

mon bean genotypes. CV was calculated by dividing the standard deviation by the

BRL (0.584).

mean, in this case for each genotype across blocks, environments and years. Larger

values indicate greater variation and less conﬁdence in the repeatability of the

3.2. Cowpea observation.

ANOVA of the two cowpea seasons revealed signiﬁcant variation

associated with genotype and year with generally normal distribu- Table 8

Estimation of broad sense heritability for cowpea root traits over two seasons at

tions (Fig. S6). For almost all traits both genotype and year had

URBC. 100*Vg/[Vg + (Vgy/# of reps) + (Ve/# of years)].

signiﬁcant effects while genotype by year interactions were only

signiﬁcant for AN, BN, 1.5A, 3BD and NS (Table 7). Moderate her- Trait Genotype Residual Genotype x Year Year H

itability was found for TD5 (27), HRN (27), 3BD (21), NS (44) and

TD5 0.339 1.806 0.1 0.0616 0.27

TD10 0.08 1.35 0.1 0.0001 0.10

TD15 0.0074 0.954 0.026 0.0675 0.02

Table 6

BRN 0.373 4.46 0.447 4.73 0.14

Estimation of broad sense heritability (h2) for two environments, Chokwe and

HRN 2.5 12.45 1.847 14.7 0.27

Umbeluzi 2008 100*Vg/[Vg + (Vge/# of reps) + (Ve/# of envirn)].

BD10 0.459 15.64 0.983 13.85 0.05

1.5H 6.682 2.176 2.364 2.781 0.80

Variance component (V)

1.5B 0.0893 0.8002 0.0454 0.0157 0.18

Chokwe and Umbeluzi Envirn Genotype G*E Error H 3BD 0.269 1.856 0.3368 1.063 0.21

NS 0.861 1.987 0.3114 0.2033 0.45

HRN 63.655 15.958 4.981 25.461 0.53

D 0.249 1.9634 0.0677 0.113 0.20

HRL 0.108 0.068 0.036 1.691 0.08

−

HRB 0.063 0.004 0.074 0.600 0.01

− −

BRWN 0.001 0.278 0.278 0.024 0.83

−

BRN 0.009 2.454 0.062 1.756 0.74

−

BRL 0.156 0.074 0.059 0.860 0.14

BRB 0.001 0.035 0.052 0.533 0.12

− − high heritability for 1.5A (79) (Table 8). A graphical representation

BRGA 0.684 0.891 0.140 1.267 0.60

− of coefﬁcient of variation in cowpea phene descriptors indicates

PRL 1.380 0.081 0.081 0.015 0.75

PRB 0.019 0.097 0.051 0.761 0.21 which may be better suited to differentiate genotypes and pheno-

−

NN 0.010 0.012 0.021 0.297 0.08 types (Fig. 6). Tap root diameter measurements deeper than 10 cm

−

RR 0.011 0.003 0.018 0.098 0.06

− − below the soil surface had very low heritability.

Table 7

ANOVA Table for cowpea root traits over two seasons showing F value and significance level. Significance values * = significant at 0.01, ** = significant at 0.001.

TD5 TD10 TD15 HRN BRN BD10 1.5H 1.5B 3BD NS D

Genotype 2.4 1.6 ** 1 4.4 ** 2.1 ** 2.2 ** 3.4** 2.0 ** 3.2 ** 4.8 ** 2.1

Year 51.3 3.1 49.3 756.8 662.6 551.0* 7.8** 21.6* 359.6 75.9* 32.9

Genotype x Year 1.2 1.2 1.1 1.6 ** 1.3 * 1.2 1.3* 1.2 1.5 ** 1.5 ** 1.2

J. Burridge et al. / Field Crops Research 192 (2016) 21–32 29

sides of the regression line. After the RANSAC conversion we found

coefﬁcients with p values <0.001 of 0.72 and 0.49 for all roots

and functional roots, respectively. These results indicate DIRT ade-

quately captures hypocotyl root number but disease and or drought

pressure complicates root counts as roots become dysfunctional

and die. Manual measurements take into account color variation

that indicates health and functionality while the binary image DIRT

uses cannot account for these subtle differences.

3.4. Bean and cowpea comparisons

Basal and hypocotyl roots show genetic variation for growth

angle but angles of basal roots vary less in cowpea than in com-

mon bean (Fig. 7). Cowpea adjusts the number and diameter of

lateral roots in a more gradual manner along the hypocotyl and

Fig. 6. Graphical representation of coefﬁcient of variation of observations for cow-

radical than common bean, which has a root system architecture

pea genotypes. CV was calculated by dividing the standard deviation by the mean,

in this case for each genotype across blocks, environments and years. Larger values dominated by clearly deﬁned basal roots. The ‘herringbone’ pat-

indicate greater variation and less conﬁdence in the repeatability of the observation. tern of cowpea may be more typical of other annual legumes. Tap

root strength, gauged by tap root diameter varies in both species

3.3. Image based (DIRT) vs. manual phenotyping

but a much higher median and larger range was found in cowpea.

Cowpea shoots are generally larger than bean and cross species

Four minutes were required to excavate and manually evaluate

comparisons should consider allometry. Here we use a t-test to

12 phenes from one root crown. Soil type and texture inﬂuenced

compare log of biomass and log of TD for bean and cowpea and

time required for root excavation and washing. Manual evaluation

find the slopes are significantly different. Correlation coefficients

requires between 1 and 2 min per crown and arranging a root for

between TD and plant biomass are moderate (0.36 for bean and 0.25

image acquisition requires approximately 30 s. Remote image anal-

for cowpea) indicating genetic variation exists for TD independent

ysis using DIRT allows more samples to be processed in a shorter

of shoot biomass.

period of time, which reduces the effects of secondary growth and

environmental factors. Image J analysis requires approximately

1 min per image whereas thousands of images can be analyzed 4. Discussion

by DIRT per hour. Signiﬁcant correlations exist between DIRT and

manual descriptors of bean root architecture indicating which We have developed a protocol combining manual and auto-

descriptors may be a reasonable substitute. Table 9 shows a com- mated image-based analysis of ﬁeld grown roots of common bean

parison of substitutes such as the tap root diameter (r = 0.71, and cowpea that takes advantage of the strengths of each tech-

P = 0.01) or basal root number (r = 0.51, P = 0.008). Additionally, nique. Automated evaluation is objective, has higher throughput,

DIRT provides manually inaccessible trait deﬁnitions that, so far, and is able to describe root architecture with higher-order phene

have proven to be distinctive for genotypes such as Dx and DSx descriptors in terms of mathematical functions and in relation to

values, and RTA and STA ranges. Dx and DSx values are functions other parameters of the individual root system. However, auto-

describing the rate of width accumulation over the depth, thus cap- mated evaluation lacks the ability to quantify color, ﬂexibility and

turing the shape of the root hull. RTA and STA ranges describe, in is limited by root occlusion and root placement on the imaging

relation to their respective means, variation within individual root board. Manual phenotyping is slower and more subjective but is

systems. better suited to gauge nodulation, disease, count ﬁne roots and

Given the relatively small sample size per genotype we accept a obtain information through manipulating and examining traits best

pair of genotypes to be distinguishable if their respective standard viewed in three dimensions, such as angle. A weakness of the pro-

error of the mean cover distinct numerical ranges. The image- tocol is that it samples only a portion of the entire root system and

based phenotyping could distinguish all possible combinations of may not capture ﬁne roots. However relationships between crown

bean genotypes. In contrast, the manually measured bean traits measurements and root length density have been shown in maize

failed to distinguish six genotype combinations in Chokwe and (Trachsel et al., 2013; Zhu et al., 2010). A related weakness is the dif-

all but 4 genotype combinations in Umbeluzi (for example dif- ﬁculty inherent with recovering small diameter roots, which make

ferentiation plot see Fig. S7). For the cowpea diversity panel, we up the majority of root length. The most time consuming steps

found that all 188 cowpea genotypes could be distinguished by at of either manual or image analysis are root excavation and wash-

least one image-based measurement and the manually measured ing, meaning combining both has a favorable cost to beneﬁt ratio.

phenes distinguished all but 5 genotype combinations (Bucksch This protocol can differentiate genotypes in a given environment

et al., 2014). and can quantify genotypic by environmental variation through

Investigation of an apparent lack of agreement between DIRT evaluation in multiple environments. A principle advantage of this

and manual measurements of BRGA and HRN in the Rock Springs protocol is that it directly evaluates expressed phenotypes in the

2014 experiment revealed some of the variation to be caused by target environment.

root ﬂexibility and color variation. Roots of plants with less sec- Observation of root architectural phenes over multiple grow-

ondary growth are more ﬂexible, increasing the potential inﬂuence ing seasons and locations permitted evaluation of environmental

of root placement on the phenotyping board on subsequent BRGA inﬂuence on root phenes and the utility of a given observation

measurements. For HRN, we compared an image-based manual to differentiate genotypes. Although we found differences in root

count of all hypocotyl roots and then all hypocotyl roots judged phenotypes across years and environments, they did not lead to

to be functional to DIRT results. Initial Pearson correlation coef- genotype by year, and genotype by environment interactions for

ﬁcients were 0.32 and 0.28 respectively and a robust regression most phenes. The exception is hypocotyl root number (Table 3),

estimator (RANSAC) was used to create inlier and outlier groups at which was likely due to differences in surface soil moisture. Differ-

a 99% conﬁdence interval that were uniformly dispersed on both ences in precipitation (64.6 mm in 2008 and 109.5 in 2009) likely

30 J. Burridge et al. / Field Crops Research 192 (2016) 21–32

reduced the number of observable hypocotyl roots as their more among genotypes for TD and BRN (Fig. 7) but less intra-genotypic

horizontal growth makes them susceptible to soil drying. The lim- variation than does bean (Fig. 6). Most cowpea phenes have lower

ited genotypic variation in root branching and length of hypocotyl, CV values than do comparable phenes in common bean. Cowpea’s

basal and primary roots is inﬂuenced by the excavation process greater range in TD and BRN indicates that in terms of root archi-

and we do not recommend using length measurements from ﬁeld- tecture cowpea offers greater phenotypic variation for some soil

grown root crowns. resource scavenging traits, particularly those hypothesized to be

We also validated a laboratory-based ‘roll-up’ phenotyping of beneﬁcial for deep water acquisition. A larger TD may enable a plant

BRWN and BRN, which simpliﬁes the evaluation for these impor- to extend deeper into the soil proﬁle and access water resources

tant phenes in common bean. We observed fewer basal roots in the that are inaccessible to shallower-rooted genotypes. This greater

ﬁeld compared to the laboratory, which we interpret as root loss range of TD and much greater median TD indicates that cowpea root

due to biotic and abiotic stresses (Fisher, 2002). In addition to BRWN architecture is taproot dominated. Measuring tap root diameter

and BRN having utility in low P, drought and combined low P and at multiple depths may help to differentiate allocation strategies.

drought environments, they may also have utility in environments Allometric comparisons indicate the greater median TD is not due

with signiﬁcant belowground biotic stress, in which more numer- only to allometry. The different slopes of log TD and log biomass

ous basal roots may compensate for root loss from herbivores and between cowpea and common suggest bean has divergent root

pathogens. strategies and that cowpea is more tap root dominated. Cowpea’s

strong taproot paired with variable BRN may help contribute to its

drought tolerance. An alternate approach to accessing more water

4.1. Cowpea and bean strategies

and nutrient resources is to extend many smaller basal roots over

a range of soil zones. While cowpea has a slightly greater range of

Based on the relatively small diversity panels included in this

BRN, common bean has greater median BRN and a greater BRGA

study we suggest that cowpea exhibits greater phenotypic variation

range as well as greater median HRN. This suggests common bean

has a basal root dominated system that has the ability to target

shallow or deep soil exploration. Its adaptive plasticity stems from

variation in BRGA, BRN and HRN. Both of these strategies seem to

make sense given cowpea’s purported evolution and domestica-

tion in drought-prone areas and common bean in nutrient limited,

highly competitive riparian zones. The higher CV values observed

for root traits in bean are suggestive of an adaptive strategy in which

plasticity itself is a beneﬁcial trait that allows a phenotype to match

and respond to a given environment as has been shown in maize

(Zhu et al., 2005).

4.2. Recommended protocol

We recommend the following ﬁeld-based shovelomics pro-

tocol for common bean and cowpea that combines manual- and

image-based phene descriptors. Tutorials are available online

(http://plantscience.psu.edu/research/labs/roots/projects/usaid-

Fig. 7. Comparison of variation between relevant manually accessed common bean

(Chokwe and Umbeluzi 2008, for TD ROS 2014 data used) and cowpea (URBC 2012 crb/resources/english/shovelomics-videos). Additionally, visual

and 2013) root parameters. Shaded area resprsents variation, line shows median, evaluation of seedlings grown in roll-ups offers a rapid method

square indicates mean. Note that common bean tap root diameter in ROS 2014 trial

to screen for BRWN, BRN and root hair length and density. At

was generally smaller than observed in other trials where a range up to 3 mm may

ﬂowering or pod elongation excavate 4–6 plants per plot and

be expected.

select the most representative crowns based on a quick visual

evaluation of health, vigor, symmetry, diameters and branching

Table 9

density. Select 2–4 plants for evaluation using manual and image

Correlations between comparable manual and DIRT observations for the PA ROS

2014 common bean data set on a per plot basis. Level of signiﬁcance: *** = signiﬁcant based measurements from at least 3 replications (Table 10). For

at p < 0.001, ** = significant at p < 0.01, * - significant at p < 0.05, ns = not significant. image acquisition and DIRT analysis position root crown naturally

on ﬂat ﬁnished black background approximately 40–50 cm from

DIRT Manual Pearson r

lens with identifying tag and circular scale marker. Do not allow

#Basal roots BRN 0.51**

roots to intersect with other objects on the board. Record image

#Basal roots TBD 0.78**

and record image number in spreadsheet with associated manual

angBas BRGA 0.64*

D30 BRGA 0.65*

D50 BRGA 0.66*

Table 10

D60 BRGA 0.67*

Recommended manual root observations for cowpea and common bean. See

D70 BRGA 0.66*

Tables 2 and 3 for details on how to measure.

D80 Brga 0.64*

DD90 max. TBD 0.65*

Trait name Abbreviation Method

DS10 3BD 0.67*

DS20 BRGA 0.73** Hypocotyl root number HRN Count

−

DS30 BRGA 0.6* Basal root number BRN Count

−

DS 40 BRGA 0.58* Basal root whorl number BRWN Count

−

Nr.RTPs 3BD 0.38 ns Basal root growth angle BRGA Measure using board

projected root area TBD 0.74** Hypocotyl root growth angle HRGA Measure using board

projected root area 3BD 0.59* Tap root branching density TBD Count or Image J count

STA 90% BRGA 0.61* Tap root diameter TD5 Caliper or Image J measure

STA 90% BRN 0.72** Nodulation score NS Score

−

Taproot dia TBD 0.71** Disease score DS Score

J. Burridge et al. / Field Crops Research 192 (2016) 21–32 31

data. Upload images to DIRT (http://dirt.iplantcollaborative.org/ References

), set parameters for automated analysis and submit for analysis.

Basu, P., Pal, A., 2012. A new tool for analysis of root growth in the spatio-temporal

Additional descriptions can be made at the researcher’s discretion,

continuum. New Phytol. 195, 264–274, http://dx.doi.org/10.1111/j.1469-8137.

such as measuring basal or hypocotyl root lateral root branching 2012.04149.x.

density, separating the disease score into radical and stem com- Basu, P., Brown, K., Lynch, J., 2007. Ethylene modulates genetic, positional, and

nutritional regulation of root plagiogravitropism. Funct. Plant Biol., 41–51,

ponents, counting the number of roots of a given class greater

http://dx.doi.org/10.1071/fp06209.

than a certain diameter threshold, or rating or counting nodules.

Beebe, S., Lynch, J., Galwey, N., Tohme, J., Ochoa, I., 1997. A geographical approach

Measuring tap root diameter of cowpea 10 cm below soil level to identify phosphorus-efﬁcient genotypes among landraces and wild

is recommended if time permits. Scores are given on a 1–9 scale ancestors of common bean. Euphytica 95, 325–336, http://dx.doi.org/10.1023/

A:1003008617829.

with 1 being very low and 9 being high. Angles are measured

Beebe, S., Skroch, P.W., Tohme, J., Duque, M.C., Pedraza, F., Nienhuis, J., 2000.

in increments of 10◦ from 0 to 90 with 0 being horizontal with Structure of genetic diversity among common bean landraces of

respect to gravity. Mesoamerican origin based on correspondence analysis of RAPD. Crop Sci. 40,

227–264, http://dx.doi.org/10.2135/cropsci2000.401264x.

Beebe, 2012, Common bean breeding in the tropics, In: Janick J., (Ed.), Plant

Breeding Reviews, pp. 357–426. 10.1002/9781118358566.

5. Conclusion

Blair, M.W., Galeano, C.H., Tovar, E., Mun, M.C., Beebe, S.E., Rao, I.M., 2012.

Development of a Mesoamerican intra-genepool genetic map for quantitative

Phaseolus

We suggest this legume-tailored, ﬂexible and low-cost pheno- trait loci detection in a drought tolerant 3 susceptible common bean (

vulgaris L.) cross. Mol. Breed. 29, 71–88, http://dx.doi.org/10.1007/s11032-

typing platform as an appropriate tool for in-ﬁeld identiﬁcation

010-9527-9.

and selection of bean and cowpea genotypes with desirable root Bonser, A.M., Lynch, J., Snapp, S., 1996. Effect of phosphorus deﬁciency on growth

phenotypes. Several root phenes are highly plastic and environ- angle of basal roots in Phaseolus vulgaris. New Phytol. 132, 281–288, http://dx.

doi.org/10.1111/j.1469-8137.1996.tb01847.x.

mental responses make ﬁeld evaluation in the target environment

Brown, L.K., George, T.S., Dupuy, L.X., White, P.J., 2013. A conceptual model of root

fundamental to effective characterization. Future research should hair ideotypes for future agricultural environments: what combination of

address the value of different phene descriptors to further focus and traits should be targeted to cope with limited P availability? Ann. Bot. 112,

317–330, http://dx.doi.org/10.1093/aob/mcs231.

streamline the crop improvement process. These phenotyping tools

Bucksch, A., Burridge, J., York, L.M., Das, A., Nord, E., Weitz, J.S., Lynch, J.P., 2014.

should be useful to traditional yield-based breeding programs and

Image-based high-throughput ﬁeld phenotyping of crop roots. Plant Physiol.

to those employing trait-based selection. In either case the phene 166, 470–486, http://dx.doi.org/10.1104/pp.114.243519.

CIAT, 1996. CIAT Annual Report for 1995. Cali, Colombia.

descriptors should lend themselves to GWAS. The demonstration

Clark, R.T., MacCurdy, R.B., Jung, J.K., Shaff, J.E., McCouch, S.R., Aneshansley, D.J.,

of manual and image based analysis on two distinct legume root

Kochian, L., 2011. Three-dimensional root phenotyping with a novel imaging

systems suggests general utility of customized protocols for phe- and software platform. Plant Physiol. 156, 455–465, http://dx.doi.org/10.1104/

pp.110.169102.

notyping tepary bean, soybean, pigeon pea, groundnut, chickpea

Clark, R.T., Famoso, A.N., Zhao, K., Shaff, J.E., Craft, E.J., Bustamante, C.D., McCouch,

and other pulse or even non-pulse roots such as tomato or potato.

S.R., Aneshansley, D.J., Kochian, L., 2013. High-throughput two-dimensional

The low cost, high-throughput and high reproducibility afforded by root system phenotyping platform facilitates genetic analysis of root growth

the combination of manual and image based analysis enables large and development. Plant. Cell Environ. 36, 454–466, http://dx.doi.org/10.1111/j.

1365-3040.2012.02587.x.

multi-location trials that are able to track genotypic responses and

R Core Team, 2014. R: A Language and Environment for Statistical Computing.

phenotypic utility across environments. We anticipate that broad Das, A., Schneider, H., Burridge, J., Karine, A., Ascanio, M., Topp, C.N., Lynch, J.P.,

application of the shovelomics technique will reveal biologically Weitz, J.S., 2015. Digital imaging of root traits (DIRT): a high-throughput

computing and collaboration platform for ﬁeld-based root phenomics. Plant

relevant phenes and aid in identifying their genetic control. We

Methods 11, 1–12, http://dx.doi.org/10.1186/s13007-015-0093-3.

expect this platform to be scaled up and automated, potentially De Smet, I., White, P.J., Bengough, a.G., Dupuy, L., Parizot, B., Casimiro, I., Heidstra,

including mechanized excavation and imaging. Integration with R., Laskowski, M., Lepetit, M., Hochholdinger, F., Draye, X., Zhang, H., Broadley,

M.R., Peret, B., Hammond, J.P., Fukaki, H., Mooney, S., Lynch, J.P., Nacry, P.,

SimRoot or other heuristic computer simulations would enable pro-

Schurr, U., Laplaze, L., Benfey, P., Beeckman, T., Bennett, M., 2012. Analyzing

jections of rooting patterns outside the crown root area and enable

lateral root development: how to move forward. Plant Cell 24, 15–20, http://

estimation of their physiologic contribution. Pairing in-vivo with dx.doi.org/10.1105/tpc.111.094292.

Ehlers, J.D., Hall, A.E., 1997. Cowpea (Vigna unguiculata L. walp). Field Crop. Res. 53,

in-silico approaches would be synergistic and would accelerate the

187–204, http://dx.doi.org/10.1071/PP9790265.

identiﬁcation of ideotypes based on actual ﬁeld observations.

Fehr, W., 1993. Principles of Cultivar Development. Macmillan Publishing

Company, New York.

Fisher, M.C.T., 2002. Lack of evidence for programmed root senescence in common

Acknowledgements bean (Phaseolus vulgaris) grown at different levels of phosphorus supply. New

Phytol. 153, 63–71, http://dx.doi.org/10.1046/j.0028-646X.2001.00285.x.

Galkovskyi, T., Mileyko, Y., Bucksch, A., Moore, B., Symonova, O., Price, C., Topp,

This research was supported by the McKnight Foundation

C.N., Iyer-Pascuzzi, A.S., Zurek, P.R., Fang, S., Harer, J., Benfey, P.N., Weitz, J.S.,

Collaborative Crop Research Program, the Howard G. Buffet Foun- 2012. GiA Roots: software for the high throughput analysis of plant root

system architecture. BMC Plant Biol. 12, 116, http://dx.doi.org/10.1186/1471-

dation, the USAID Dry Grain Pulses Collaborative Research Support

2229-12-116.

Program and the USAID Climate Resilient Beans Feed the Future

Ho, M.D., Rosas, J.C., Brown, K.M., Lynch, J.P., 2005. Root architectural tradeoffs for

Legume Innovation Laboratory, the NSF Plant Genome Research water and phosphorus acquisition. Funct. Plant Biol. 32, 737–748, http://dx.

doi.org/10.1071/FP05043.

Program, NSF 0820624 and the Center for Data Analytics, Georgia

Huynh, B., Close, T.J., Roberts, P.A., Hu, Z., Wanamaker, S., Lucas, M.R., Chiulele, R.,

Institute of Technology, Spatial Networks in Biology: Organizing

Cissé, N., David, A., Hearne, S., Fatokun, C., Diop, N.N., Ehlers, J.D., 2013. Gene

and Analyzing the Structure of Distributed Biological Systems. We pools and the geneteic architecture of domesticated cowpea. Plant Genome 6,

thank the International Center for Tropical Agriculture (CIAT) and 1–8, http://dx.doi.org/10.3835/plantgenome2013.03.0005.

Iyer-Pascuzzi, A.S., Symonova, O., Mileyko, Y., Hao, Y., Belcher, H., Harer, J., Weitz,

University of California Riverside for providing seed, Eric Nord for

J.S., Benfey, P.N., 2010. Imaging and analysis platform for automatic

help with Figs. 5–7, and Robert Snyder for the management of ﬁeld phenotyping and trait ranking of plant root systems. Plant Physiol. 152,

experiments in Penn State, Johan Prinsloo in South Africa and Alípio 1148–1157, http://dx.doi.org/10.1104/pp.109.150748.

Liao, H., Rubio, G., Yan, X., Cao, A., Brown, K.M., Lynch, J.P., 2001. Effect of

Simão and Marica Tomas in Mozambique.

phosphorus availability on basal root shallowness in common bean. Plant Soil

232, 69–79, http://dx.doi.org/10.1023/A:1010381919003.

Liao, H., Xiaolong, Y., Rubio, G., Beebe, S., Blair, M., Lynch, J., 2004. Genetic mapping

Appendix A. Supplementary data of basal root gravitropism and phosphorus acquisition efﬁciency in common

bean. Funct. Plant Biol. 91, 959–970, http://dx.doi.org/10.1071/FP03255.

Lynch, J., Beem, J.J.Van, 1991. Growth and architecture of seedling roots of common

Supplementary data associated with this article can be found, in

bean genotypes. Crop Sci. 33, 1253–1257.

the online version, at http://dx.doi.org/10.1016/j.fcr.2016.04.008.

32 J. Burridge et al. / Field Crops Research 192 (2016) 21–32

Lynch, J.P., Brown, K.M., 2001. Topsoil foraging—an architectural adaptation of Rubio, G., Liao, H., Yan, X.L., Lynch, J.P., 2003. Topsoil foraging and its role in plant

plants to low phosphorus availability. Plant Soil, 225–237, http://dx.doi.org/10. competitiveness for phosphorus in common bean. Crop Sci. 43, 598–607.

1023/A:1013324727040. Saengwilai, P., Tian, X., Lynch, J.P., 2014. Low crown root number enhances

Lynch, J., 1995. Root architecture and plant productivity. Plant Physiol. 109, 7–13, nitrogen acquisition from low-nitrogen soils in maize. Plant Physiol. 166,

http://dx.doi.org/10.1104/pp.109.1.7. 581–589, http://dx.doi.org/10.1104/pp.113.232603.

Lynch, J.P., 2007. Roots of the second green revolution. Aust. J. Bot. 55, 493–512, Schulz, H., Postma, J.a., van Dusschoten, D., Scharr, H., Behnke, S., 2013. Plant root

http://dx.doi.org/10.1071/BT06118. system analysis from MRI images. Commun. Comput. Inform. Sci., 411–425,

Lynch, J.P., 2011. Root phenes for enhanced soil exploration and phosphorus http://dx.doi.org/10.1007/978-3-642-38241-3 28.

acquisition: tools for future crops. Plant Physiol. 156, 1041–1049, http://dx.doi. Serebrovsky, A.S., 1925. Somatic segregation in domestic fowl. J. Genet. 16, 33–42.

org/10.1104/pp.111.175414. Tilman, D., Balzer, C., Hill, J., Befort, B.L., 2011. Global food demand and the

Lynch, J.P., 2013. Steep, cheap and deep: an ideotype to optimize water and N sustainable intensiﬁcation of agriculture. Proc. Natl. Acad. Sci. U. S. A. 108,

acquisition by maize root systems. Ann. Bot. 112, 347–357, http://dx.doi.org/ 20260–20264, http://dx.doi.org/10.1073/pnas.1116437108.

10.1093/aob/mcs293. Topp, C.N., Iyer-Pascuzzi, A.S., Anderson, J.T., Lee, C.-R., Zurek, P.R., Symonova, O.,

Mairhofer, S., Zappala, S., Tracy, S., Sturrock, C., Bennett, M.J., Mooney, S.J., Zheng, Y., Bucksch, A., Mileyko, Y., Galkovskyi, T., Moore, B.T., Harer, J.,

Pridmore, T.P., 2013. Recovering complete plant root system architectures Edelsbrunner, H., Mitchell-Olds, T., Weitz, J.S., Benfey, P.N., 2013. 3D

from soil via X-ray mu-computed Tomography. Plant Methods 9, http://dx.doi. phenotyping and quantitative trait locus mapping identify core regions of the

org/10.1186/1746-4811-9-8. rice genome controlling root architecture. Proc. Natl. Acad. Sci. U. S. A. 110,

Metzner, R., Eggert, A., van Dusschoten, D., Pﬂugfelder, D., Gerth, S., Schurr, U., E1695–E1704, http://dx.doi.org/10.1073/pnas.1304354110.

Uhlmann, N., Jahnke, S., 2015. Direct comparison of MRI and X-ray CT Trachsel, S., Kaeppler, S.M., Brown, K.M., Lynch, J.P., 2011. Shovelomics: high

technologies for 3D imaging of root systems in soil: potential and challenges throughput phenotyping of maize (Zea mays L.) root architecture in the ﬁeld.

for root trait quantiﬁcation. Plant Methods 11, 1–11, http://dx.doi.org/10.1186/ Plant Soil 341, 75–87, http://dx.doi.org/10.1007/s11104-010-0623-8.

s13007-015-0060-z. Trachsel, S., Kaeppler, S.M., Brown, K.M., Lynch, J.P., 2013. Maize root growth

Miguel, M., 2004. Genotypic variation in root hairs and phosphorus efﬁciency in angles become steeper under low N conditions. Field Crop. Res. 140, 18–31,

common bean (Phaseolus vulgaris L.). In: Dissertation. Penn State University. http://dx.doi.org/10.1016/j.fcr.2012.09.010.

Miguel, M.A., Widrig, A., Vieira, R.F., Brown, K.M., Lynch, J.P., 2013. Basal root whorl Varshney, R.K., Terauchi, R., Mccouch, S.R., 2014. Harvesting the promising fruits of

number: a modulator of phosphorus acquisition in common bean (Phaseolus genomics: applying genome sequencing technologies to crop breeding. PLoS

vulgaris). Ann. Bot. 112, 973–982, http://dx.doi.org/10.1093/aob/mct164. Biol. 12, http://dx.doi.org/10.1371/journal.pbio.1001883.

Miguel, M.A., Postma, J.A., Lynch, J., 2015. Phene synergism between root hair Walk, T.C., Jaramillo, R., Lynch, J.P., 2006. Architectural tradeoffs between

length and basal root growth angle for phosphorus acquisition. Plant Physiol. adventitious and basal roots for phosphorus acquisition. Plant Soil 279,

167, 1430–1439, http://dx.doi.org/10.1104/pp.15.00145. 347–366, http://dx.doi.org/10.1007/s11104-005-0389-6.

Miller, C., Ochoa, I., Nielsen, K., Beck, D., Lynch, J., 2003. Genetic variation for Wortmann, C., Kirkby, R., Eledu, C., Allen, D., 1998. Atlas of Common Bean

adventitious rooting in response to low phosphorus availability: potential (Phaseolus Vulgaris L.) Production in Africa. Centro Internationcal de

utility for phosphorus acquisition from stratiﬁed soils. Funct. Plant Biol. 30, Agricultura Tropical, CIAT publication, no 297, Cali, Colombia.

973–985, http://dx.doi.org/10.1071/FP03078. York, L.M., Nord, E., Lynch, a, 2013. Integration of root phenes for soil resource

Ochoa, I.E., Blair, M.W., Lynch, J.P., 2006. QTL analysis of adventitious root acquisition. Front. Plant Sci. 4, 355, http://dx.doi.org/10.3389/fpls.2013.00355.

formation in common bean under contrasting phosphorus availability. Crop Zhu, J., Kaeppler, S.M., Lynch, J.P., 2005. Mapping of QTL controlling root hair length

Sci. 46, 1609–1621, http://dx.doi.org/10.2135/cropsci2005.12-0446. in maize (Zea mays L.) under phosphorus deﬁciency. Plant Soil 270, 299–310,

Poorter, H., Bühler, J., Van Dusschoten, D., Climent, J., Postma, J.a., 2012. Pot size http://dx.doi.org/10.1007/s11104-004-1697-y.

matters: a meta-analysis of the effects of rooting volume on plant growth. Zhu, J., Zhang, C., Lynch, J.P., 2010. The utility of phenotypic plasticity of root hair

Funct. Plant Biol. 39, 839–850, http://dx.doi.org/10.1071/fp12049. length for phosphorus acquisition. Funct. Plant Biol. 37, 313–322, http://dx.doi.

Rellán-Álvarez, R., Lobet, G., Lindner, H., Pradier, P.-L., Sebastian, J., Yee, M.-C., org/10.1071/FP09197.

Geng, Y., Trontin, C., LaRue, T., Schrager-Lavelle, A., Haney, C.H., Nieu, R., Zhu, J., Ingram, P.A., Benfey, P.N., Elich, T., 2011. From lab to ﬁeld, new approaches

Maloof, J., Vogel, J.P., Dinneny, J.R., 2015. GLO-Roots: an imaging platform to phenotyping root system architecture. Curr. Opin. Plant Biol. 14, 310–317,

enabling multidimensional characterization of soil-grown root systems. Elife 4, http://dx.doi.org/10.1016/j.pbi.2011.03.020.

1–26, http://dx.doi.org/10.7554/eLife.07597. Zobel, R.W., 2011. A developmental gentic basis for deﬁning root classes. Crop Sci.

51, 1410–1413, http://dx.doi.org/10.2135/cropsci2010.11.0652. Theor Appl Genet (2017) 130:419–431 DOI 10.1007/s00122-016-2823-y

ORIGINAL ARTICLE

Genome-wide association mapping and agronomic impact of cowpea root architecture

James D. Burridge1 · Hannah M. Schneider1 · Bao-Lam Huynh2 · Philip A. Roberts2 · Alexander Bucksch3 · Jonathan P. Lynch1

Received: 25 June 2016 / Accepted: 3 November 2016 / Published online: 18 November 2016 © Springer-Verlag Berlin Heidelberg 2016

Abstract study identified 11 significant quantitative trait loci (QTL) Key message Genetic analysis of data produced by novel from manually scored root architecture traits and 21 QTL root phenotyping tools was used to establish relation- from root architecture traits phenotyped by DIRT image ships between cowpea root traits and performance indi- analysis. Subsequent comparisons of results from this root cators as well between root traits and Striga tolerance. study with other field studies revealed QTL co-localiza- Abstract Selection and breeding for better root phenotypes tions between root traits and performance indicators includ- can improve acquisition of soil resources and hence crop ing seed weight per plant, pod number, and Striga (Striga production in marginal environments. We hypothesized that gesnerioides) tolerance. The data suggest selection for root biologically relevant variation is measurable in cowpea root phenotypes could be employed by breeding programs to architecture. This study implemented manual phenotyping improve production in multiple constraint environments. (shovelomics) and automated image phenotyping (DIRT) on a 189-entry diversity panel of cowpea to reveal biologically important variation and genome regions affecting root Introduction architecture phenes. Significant variation in root phenes was found and relatively high heritabilities were detected Cowpea (Vigna unguiculata (L.) Walp) is a primary protein for root traits assessed manually (0.4 for nodulation and 0.8 source and food security crop for large portions of Africa, for number of larger laterals) as well as repeatability traits Asia and South America. Cowpea plays a significant role phenotyped via DIRT (0.5 for a measure of root width and in contributing nitrogen (N) to agroecosystems and fodder 0.3 for a measure of root tips). Genome-wide association to livestock especially in the low-input systems common in cowpea production zones (Ehlers and Hall 1997; Singh et al. 1997, 2003; Huynh et al. 2013). Cowpea is already Communicated by M. N. Nelson. cultivated in marginal environments commonly experienc- ing drought, low fertility and pest attack and these con- Electronic supplementary material The online version of this straints are likely to become more severe with the effect of article (doi:10.1007/s00122-016-2823-y) contains supplementary climate change (Yadav et al. 2015). Breeding efforts com- material, which is available to authorized users. monly target above-ground traits or incorporate studies on * Jonathan P. Lynch water balance of shoots. An underexploited breeding strat- [email protected] egy is trait-based selection focused on linking specific root traits to efficient resource acquisition (Cattivelli et al. 2008; 1 Department of Plant Science, The Pennsylvania State University, 221 Tyson Building, University Park, PA 16802, Lynch 2015). USA Phenes are the elementary unit of the plant phenotype 2 Department of Nematology, University of California, (Lynch 2011; York et al. 2013). The term was used as early Riverside, CA, USA as 1925 to describe phenotypic traits under genetic con- 3 Schools of Biology and Interactive Computing, George trol (Serebrovsky 1925). While “trait” is often used inter- Institute of Technology, Atlanta, GA, USA changeably with phene its use is ambiguous and can cross

1 3 420 Theor Appl Genet (2017) 130:419–431 several scales of biological organization. Phene is more such as occurs with greater concentration of Striga seeds precise and refers to an elementary unit at a given level of in the top 10 cm of the soil profile (Van Delft et al. 2000). biological organization (Lynch and Brown 2012). The term For these reasons root architectural phenes that affect spa- ‘metric’ is used in this article to describe some of the math- tial and temporal distribution of roots can have profound ematically derived observations generated using computer effects on efficient soil exploration and resource acquisition based image analysis. The term ‘trait’ is used in this article and avoidance of biotic or abiotic constraints (Lynch and to refer to groups comprised of phenes and metrics. Agro- Wojciechowski 2015). In maize and common bean, several nomically relevant traits of field grown root crowns can be root architectural phenes have been related to increased quantified using shovelomics (Burridge et al. 2016) a man- productivity under stressful conditions such as deep root- ual phenotyping method, or DIRT (Digital Imaging of Root ing for increased water acquisition in bean (Ho et al. 2005), Traits). DIRT is an image-based phenotyping software maize (Zhu et al. 2010), and wheat (Kirkegaard et al. (Bucksch et al. 2014) publically available online (Das et al. 2007). Shallow rooting has been shown to be advantageous 2015). Following identification of genes controlling root for immobile nutrients in maize (Bayuelo-Jiménez et al. phenes or metrics, marker-assisted selection (MAS) would 2011) and bean (Bonser et al. 1996; Miller et al. 2003; Ho enable trait-based selection orders of magnitude faster than et al. 2005; Beebe et al. 2006; Miguel et al. 2015). Given field-based phenotyping of mature plants (Varshney et al. the similarity between common bean and cowpea phenol- 2014). Modern breeding programs using these technologies ogy and root architecture we speculate that homologous can phenotype thousands of entries per three-month period phenes may have similar benefits. The limited body of cow- (an average growing season) while a field-based breeder pea root studies indicates deep rooting may be beneficial using shovelomics could phenotype only hundreds of for drought tolerance (Matsui and Singh 2003; Barros et al. entries in the same period. Indirect selection of root traits 2007; Agbicodo et al. 2009) but others highlight the lim- by MAS would also eliminate the need to phenotype roots ited payback when investing in deeper roots when very lit- at each generation during breeding advancement. tle additional water can be acquired (Hall 2012). Research Cowpea production is commonly limited by multiple conducted in soil cylinders suggests restricting water use is factors that may occur simultaneously, including biotic important for water use efficiency (Belko et al. 2012a, b, stress and limited access to water and nutrients. The para- 2014). Limited research on low P tolerance has been con- sitic weed Striga (Striga gesnerioides) is a constraint to ducted in cowpea, but roots may be important to increase P cowpea production and can be devastating in certain geo- efficiency (Kugblenu et al. 2014). graphic areas (Rubiales and Fernández-Aparicio 2012; Cowpea performance in low fertility environments Kamara et al. 2014). Several genome regions conferring could be related to efficient resource acquisition and use, Striga resistance have been identified in different environ- especially when low nutrient soils and water limitation co- ments and cowpea lines but resistance is not completely occur, as is the case for many production environments. In conferred by a single locus, indicating resistance may be this context, drought escape by accelerated phenology can highly quantitative and may depend on a variety of resist- be important, and in cowpea development of extra early ance mechanisms (Ouedraogo et al. 2001; Ouédraogo maturing varieties has been used as a strategy to avoid et al. 2002; Li et al. 2009; Noubissie Tchiagam et al. drought effects in West Africa (Hall 2012). However, trade- 2010), of which avoidance could be one (Van Delft et al. offs inherent in such strategies may include decreased time 2000). Furthermore, some previous studies to characterize to acquire immobile resources such as P and K (Bayuelo- genetic control of Striga tolerance used pot assays (Atok- Jiménez et al. 2011; Nord et al. 2011) and decreased yield ple et al. 1995; Noubissie Tchiagam et al. 2010; Omoigui potential and limited use of leaf fodder in cowpea (Hall et al. 2011), which may negate avoidance mechanisms and 2012) and yield penalty in bean (White and Singh 1991). amplify hypersensitive responses and chemical and physi- Plants with a strong, deep root system may be better at cal exclusion strategies. scavenging for available moisture and nutrients, contribut- Soil resources are frequently stratified, with immobile ing to stay-green. resources such as phosphorus (P) and potassium (K) being The acquisition of immobile resources is important more available in shallow soil strata (Lynch and Brown early in phenology before the topsoil becomes dry, which 2001) and mobile resources such as water being more increases the tortuosity of the diffusive pathway and available in deeper soil strata (Lynch and Wojciechowski decreases root growth rates, contributing to making the 2015). Constraints to root growth and resource acquisition acquisition of immobile soil resources more difficult (Bar- are often stratified, especially in oxisols, which are com- ber 1995). In other legumes such as common bean, P acqui- mon in many cowpea production areas, and in which acid- sition is related to efficient shallow soil exploration and ity and aluminum toxicity increase with depth (Lynch and may enable late season root growth into deep soil where Wojciechowski 2015). Biotic factors also may be stratified more water is available (Ho et al. 2005). As pods elongate

1 3 Theor Appl Genet (2017) 130:419–431 421 and seeds fill, grain sink strength and water acquisition Materials and methods become stronger regulators of successful maturation than new mineral resource acquisition, as significant amounts Genetic materials and experimental design of resources are redistributed within the plant (Rao et al. 2013). We hypothesize that in a terminal drought situation A diversity panel of 189 entries including traditional culti- the acquisition of immobile resources would be especially vars (landraces) and elite breeding lines representing world- important in the first four vegetative stages of phenology wide cowpea genetic diversity was assembled and seed- and water acquisition is critical throughout development. multiplied by University of California—Riverside (UCR) A dimorphic root system possessing phenes enabling both (Supplemental File). The panel was phenotyped in 2012 shallow and deep exploration may be particularly well and 2013 at Ukulima Root Biology Center (URBC), Lim- suited to environments characterized by dual limited water popo Province South Africa (24°33′00.12S, 28°07′25.84E, as well as P. Phenes that modulate access and rate of water 1235 masl). Both experiments were designed as rand- extraction from a given soil domain should be considered omized complete blocks with four replications each year. when developing ideotypes (Vadez et al. 2012; Belko et al. Each line was planted in a single-row plot consisting of ten 2014). Quantifying and understanding the utility of the plants per plot. Row width was 76 cm and distance between spatiotemporal deployment of roots in heterogeneous and plants within a row was 30 cm. URBC has a deep Clovelly dynamic soil as it relates to efficiently accessing resources loamy sand (Typic Ustipsamment). Experiments were fer- is important for production in environments constrained by tilized with complete N–P–K fertilizers before planting, multiple resources. Comparisons of cowpea lines with sim- treated with foliar feed during the growing season and ilar genetic background but varying in root architecture and irrigated regularly with a center pivot irrigation system to water use phenes will be important to assess the contribu- ensure non-limiting growing conditions. The agrochemical tion of root phenotypes to stress tolerance. Nemacur 400 EC was applied twice during the season to A consensus genetic map for cowpea (2n 2x 22) prevent interference by insects or nematodes. = = was constructed based on genotyping of 6 bi-parental recombinant inbred line (RIL) populations using an Illu- Root phenotyping mina GoldenGate assay for 1,536 EST-derived SNP markers (Muchero et al. 2009). The map resolution was Shovelomics was conducted 4–6 weeks after planting on improved by genotyping seven additional RIL populations, two plants per plot. Following excavation of six plants per which revealed further synteny with soybean and offers plot, the two most representative were selected for pheno- increased possibilities to identify gene function (Lucas typing. A complete description of the methodology is avail- et al. 2011). In addition, a core germplasm of traditional able in Bucksch et al. (2014) and Burridge et al. (2016). landraces across cowpea-growing regions in African and In brief, the method involves excavation of the root crown the world has been collected and characterized (Huynh with a shovel to 0.25 m average depth, rinsing of the root et al. 2013). Using these markers and genetic resources, crown in a bucket of water, followed by manual measure- quantitative trait loci (QTL) have been discovered for key ments using a phenotyping board and digital calipers. traits including tolerance to drought (Muchero et al. 2010; Eleven manual root architectural parameters were collected Muchero et al. 2013), seed quality (Lucas et al. 2013b) and in 2012 and 14 in 2013 (Table 1). Shoot biomass was col- resistance to root-knot nematodes (Huynh et al. 2016), root lected both years at the same time as the root crowns. pathogens including Macrophomina phaseolina (Muchero To acquire the image needed for automated image anal- et al. 2011) and Fusarium wilt (Pottorff et al. 2014; Pottorff ysis (DIRT), the washed root system was placed on a flat et al. 2012), insects (Huynh et al. 2015; Lucas et al. 2013a), black background with circular scale marker, plot identifier and the parasitic weed Striga (Ouedraogo et al. 2012). tag and a standard color image was obtained using a stand- These performance traits could be driven by root develop- ard digital camera mounted on a tripod. The color image ment but until now, no research on genetic mapping has was converted to a grayscale image, segmented into fore- been reported for root traits in cowpea due to challenges and background and the foreground images were separated in root phenotyping; the practical aspects of growing and into root, scale marker, and tag. Then the root image was excavating mature plant roots can be particularly challeng- used to calculate the root-width profile and root tip paths ing compared to above ground phenotyping. In this study, (RTPs) and the metrics derived from the profile and RTPs. we aimed to identify genomic regions controlling root The scale marker was used to transform pixels to millim- architecture in a cowpea diversity panel using state-of-the- eters and to correct for imperfect lens to board alignment. art root phenotyping systems and established relationships A complete set of images was taken in 2013 from which 47 between root trait QTL and yield, stress tolerance or resist- measures per root crown were extracted (see Bucksch et al. ance, and other agronomically important traits. 2014 for a full list of traits and descriptions).

1 3 422 Theor Appl Genet (2017) 130:419–431

Table 1 List of root trait names with abbreviation and deﬁnition Trait Deﬁnition

Adventitious root growth angle (ARGA) Average adventitious (hypocotyl) root growth angle relative to horizontal (0 is horizontal, 90 is vertical, scored in units of 10°) Adventitious root number (ARN) Number of 1st order lateral roots emerging from the hypocotyl Average root density Ratio between foreground and background pixels of the extracted root Basal root growth angle (BRGA) Average basal root growth angle relative to horizontal (0 is horizontal, 90 is vertical, scored in units of 10°) Basal root number (BRN) Number of roots in the basal region of the hypocotyl Branching density 10 (BD10) Number of 1st order laterals emerging from the primary root between 5 and 10 cm below soil level (tap root laterals) CPD 25, 50, 75, 90 Diameter at 25, 50, 75, 90% length of the central path diameter D10, D20, D50, D60 Percentage of width accumulation at x% depth Disease score (DS) Disease rating where 1 is severely infected with stem and root rots and 9 is extremely healthy DS10, DS20… DS80 Slope of the line made by two sequential Dx values Max width Maximum width of root calculated from root image Median width Median width of root calculated from root image Nodule score (NS) Rating of nodule size and number with 1 having no nodules and 9 having many large nodules Number of root tip paths (nr. of RTPs) Number of unique paths from top of stem to each root apex RTA range Range of all calculated root tip paths Stem diameter (SD) Diameter of stem (mm) at soil level Tap diameter (TD5, TD10, TD15, TD20) Tap root diameter (mm) 5, 10, 15 and 20 cm below the soil surface Third order branching density (3BD) Rating of number and order of laterals with 1 being very low branching and 9 having many orders and dense spacing of laterals 1.5BD5, 1.5BD10 Number of 1st order laterals with greater than 1.5 mm diameter 2 cm from origin emerging from 0–5 cm below soil level (1.5BD5) and from 5–10 cm below soil level (1.5BD10)

Broad-sense heritability on an entry mean basis was (BLUPs). BLUPs were calculated and used for subsequent calculated for the manually acquired root phenotypic data analysis. Statistical analyses were performed in R version from the two seasons. Repeatability on an entry mean basis 3.1.2 (R Core Team 2014). was calculated for the image-acquired root phenotypic data from four replications of each season (Fehr 1993). Association mapping Repeatability of DIRT traits was calculated as Genotypic data of 1536 SNP markers for the cowpea diver- σ 2(G) R2 = × 100, sity panel were derived from (Lucas et al. 2011). Version 6 2 E + 2 G σ ( ) σ ( ) of the consensus genetic map containing 1091 SNPs was where σ2(G) is the genotypic variance and σ2(E) is the error retrieved from HarvEST:Cowpea (http://harvest.ucr.edu/). variance. Genome-wide association study (GWAS) was performed Heritability on an entry mean basis was calculated for on root phenotypes collected via DIRT image analysis in the manual measurement traits with the following formula: 2013 and via manual measurements in 2012 and 2013. Genotypic data for GWAS included 1091 SNP mark- 2 2 σ (G) ers genotyped on 189 lines of the cowpea diversity panel. H = 2 2 × 100, σ (E) + σ (GY) σ 2(G) GWAS was performed using the mixed linear Q K model ry r + (MLM) (Zhang et al. 2010) implemented in the Genomic where σ2(G) is the genotypic variance, σ2(GY) is the geno- Association and Prediction Integrated Tool (GAPIT) R type by year variance, σ2(E) is the error variance, r is the package (Lipka et al. 2012) using the following model: number of replications and y is the number of years. y = Xβ + Wm + Qv + Zu + e, Normality of residuals was checked and data were power-transformed using the lambda identified by Box- where y is a vector of phenotypic observations; β is a cox transformations. Spearman and Pearson correlations vector of unknown fixed effects except the SNP marker between years, replications, and samples suggested data under testing, m is a vector of fixed marker effects (i.e. could be combined using best linear unbiased predictors SNP), v is a vector of subpopulation effects, u is a vector

1 3 Theor Appl Genet (2017) 130:419–431 423 of unknown random effects, and e is a vector of residual Root traits phenotyped manually for two years at URBC effects. Q is an incidence matrix of principal component with high heritability scores included 1.5BD5 (0.80), TD5 scores of marker-allele frequencies. X, W and Z are inci- (0.27), ARN (0.27), and NS (0.45) (Burridge et al. 2016) dence matrices of ones and zeros relating y to β, m and u, replotted in Table S2). Repeatability of root trait values respectively. The covariance of u is equal to KVA, where K acquired by one season of image analysis highlight D20, is the kinship matrix that was estimated with a random set Nr. of RTPs, and SD (DIRT) with repeatability over 0.25 of SNPs using the VanRaden method and VA is the addi- (Burridge et al. 2016) (replotted in Table S3). Allometry tive variance estimated with restricted maximum likelihood between root phenes and shoot biomass from the 2013 (REML). The kinship matrix estimation and the principal South African trial was modeled as a linear regression component analysis were performed in the GAPIT R pack- using the log10 transformed trait values to identify power- age. The optimum number of principle components/covari- law relationships between traits and shoot biomass. ates included in the model for each trait was determined by forward model selection using the Bayesian information Analysis of marker-trait association criterion. Additive effects were estimated relative to the minor Eleven significant QTL (LOD > 2.9) were identified for allele. SNPs with a LOD score greater than 2.9 were con- manually measured phenotypes including SD (manual), sidered to be significantly associated with the trait. BRGA, ARGA, TD10, and TD15 and 21 significant QTL were identified for DIRT measured phenotypes includ- QTL comparison of root architecture and other traits ing SD (DIRT), CPD25, CPD75, D10,20,50,60, RTA range, max width, median width and average root density QTL information for yield, plant biomass, seed weight, (Table 2). The most significant QTL were identified for stay-green, resistance or tolerance to biotic and abiotic median width (linkage group (LG) 8, 24.9 Cm), SD (DIRT stress was obtained from previous studies using GWAS and manual, LG 7, 47.4 cM and LG 3, 92.9 cM), ARGA and bi-parental mapping (Muchero et al. 2010, 2011, 2013; (LG 6, 48.4 cM, LOD 3.92, allelic effect 1.8) and BRGA − Lucas et al. 2013a, 2013b; Huynh et al. 2015, 2016; Pot- (LG 10, 2.3 cM, LOD 3.14, allelic effect 2.6). Median − torff et al. 2012, 2014; Ouedraogo et al. 2012). These QTL and max width co-localize on LG 8, 24.9 cM with LOD of locations were aligned on the cowpea consensus genetic 4.46 and 3.83 and allelic effects of 12.5 and 13.6, respec- map for identification of regions that coincided with QTLs tively. QTL for stem diameter identified from manual for root architecture. Root architecture SNPs with LOD measurements co-localized with QTL for stem diameter score equal or greater than 2.6 were considered significant identified from DIRT measurements of stem diameter on for correlations with agronomic traits. LGs 1, 3, 6, 7 (Table 2; Fig. 1). Six root architecture QTL identified using DIRT and manual measurements co-located with previously identi- Results fied regions for agronomically relevant traits including seed weight, seed number, pod number and Striga toler- Root phenotypic assessment ance measured in previous studies using the same genetic materials (Table 3). SNP marker 2227_693 on LG 6 affect- Significant genotypic and phenotypic variation of root ing root system median width is co-located with QTL for phenes is measureable using DIRT and manual shov- seed weight per plant (Muchero et al. 2013) in which the elomics and there is some correlation between DIRT and SNP haplotype conferring high root system median width manual measures of related traits (see Table 1 for trait defi- also significantly increased seed weight per plant. Mark- nitions and Table S1 for correlations). General phenotypic ers on LG1 affecting CPD25, a measure of hypocotyl size, results from DIRT and manual shovelomics have been co-localized with QTL for pod number (Muchero et al. discussed at length elsewhere (Bucksch et al. 2014; Bur- 2013), in which the SNP haplotype gave a negative effect ridge et al. 2016). However, several key points regarding on CPD25 but a slightly increased pod number effect. genotypic variation and heritability are highlighted in this Markers affecting root width accumulation metrics D20, paragraph to prepare the reader for novel genetic and allo- D50 and D60 were co-located on LG10, which coincided metric analysis that follow. All manually measured traits with QTL for seed number per plant (Muchero et al. 2013), had greater than three-fold range in values and the majority however, allelic effects for D20, D50, and D60 are small had greater than eight (Fig S1). Individual accessions are and vary in the direction of their effect. When analyzed by distinguishable from each other based on normalized mean individual trials, markers for median width and CPD 25 values of their root phenotypic descriptor data (Bucksch showed contrasting effects in different environments, sug- et al. 2014). Heritability for root traits had a wide range. gesting their environmentally specific trait utility (Tables 5,

1 3 424 Theor Appl Genet (2017) 130:419–431

Table 2 Root architecture QTL with LOD >2.9 Trait SNP marker Linkage group Position (cM) Allelica effect LOD Favorable allele Alternative allele

Stem diameter (manual) 13772_1075 1 22.4 0.5** 3.76 C G − Stem diameter (DIRT) 13772_1075 1 22.4 0.5** 3.81 C G − Stem diameter (DIRT) 5084_519 1 23.6 0.4** 3.26 C G − Stem diameter (manual) 5084_519 1 23.6 0.4** 3.25 C G − Stem diameter (manual) 4836_807 2 6.7 0.7** 3.17 G C Avg root density (DIRT) 10811_937 2 71 0.1** 3.13 A T − RTA range (DIRT) 2326_226 3 42.7 3.0** 3.94 G A TD10 (manual) 12501_343 3 76.7 0.2** 3.1 G A Stem diameter (DIRT) 139_439 3 92.9 0.6** 4.03 A G − Stem diameter (manual) 139_439 3 92.9 0.07** 2.92 A G − TD10 (manual) 5061_428 4 21 0.2** 3.34 A G − CPD75 (DIRT) 7102_965 4 21.8 0.03** 4.1 G A Avg root density (DIRT) 1004_587 5 21.2 0.2** 3.96 A G − Stem diameter (DIRT) 8969_1386 6 2.2 0.3** 3.36 A C − Stem diameter (manual) 8969_1386 6 2.2 0.3** 3.36 A C − Median width (DIRT) 2227_693 6 2.9 6.5** 3.05 G A CPD 25 (DIRT) 9645_589 6 5.4 0.03** 3.21 A G − CPD 25 (DIRT) 5428_339 6 6.1 0.03** 3.32 A G − D10 (DIRT) 3211_511 6 21.7 0.02** 3.49 G A D20 (DIRT) 3211_511 6 21.7 0.02** 3.71 G A Adventitious root angle (manual) 4749_1972 6 48.4 1.81** 3.92 A G − Stem diameter (DIRT) 11138_624 7 47.4 0.5** 4.25 G A Stem diameter (manual) 11138_624 7 47.4 0.5** 4.26 G A Max width (DIRT) 14604_737 8 24.9 13.6** 3.83 G A Median width (DIRT) 14604_737 8 24.9 12.5** 4.46 G A TD15 (manual) 13848_735 9 42.1 0.1** 3.29 G A Basal root angle (manual) 11851_914 10 2.3 2.6** 3.14 A G − D20 (DIRT) 4245_136 10 50.5 0.01** 2.91 G A D50 (DIRT) 4245_136 10 50.5 0.01** 2.99 G A D60 (DIRT) 4245_136 10 50.5 0.01** 3.01 A G − D50 (DIRT) 2391_614 11 25.8 0.01** 3.66 G A D60 (DIRT) 2391_614 11 25.8 0.01** 3.65 G A

Units for diameter measurements (SD, TD, CPD) are millimeters, Dx values are percent of total accumulated width at the specified depth meaning 0.01 corresponds to 1%, BRGA, ARGA, root tissue angle (RTA) are measures in degrees from 0 to 90, max and median width are measured in millimeters, Avg root density is the ratio of background to foreground pixels in the root image and ranges from 0.2 to 6.5. See Table 1 for abbreviation definitions a Marker additive effect, *, ** are significance level at P > 0.01 and 0.001, respectively; a positive allele effect indicates the favorable allele contributing to a positive phenotypic value, while a negative allelic effect indicates the favorable allele contributing to a negative phenotypic value

6). Median width positively correlated with seed weight in percentile for BRGA, bottom 5 percentile for ARN), a the Kano field site but negatively in the Kamboinse field relatively strong tap root, and low median and maximum site (Table 6). widths. Markers associated with BRGA and D60 on LG10 also co-located with QTL for Striga resistance (Ouédraogo et al. Allometry 2002). Interestingly, the positive alleles for BRGA and D60 are identical to SNP haplotypes of SuVita-2, which The slope of the regression line represents the scaling is a donor parent for Striga resistance (Ouédraogo et al. coefficient (α) (Niklas 1994) (Table 4). In our analysis, 2002) (Table S4). Phenotypic root evaluations of SuVita-2 isometry is established for α 0.33, because traits con- = found few and steep basal and adventitious roots (top 10 sidered have one-dimensional units whereas biomass has

1 3 Theor Appl Genet (2017) 130:419–431 425

Fig. 1 Chromosome map, showing signiﬁcant QTL for manually trait QTL measured by DIRT. Traits in white boxes refer to manu- measured and DIRT root architecture traits. Black dots represent root ally measured root QTL. Trait abbreviations and descriptions can be architecture QTL, grey dots represent root architecture QTL that co- found in Table 1 localize with agronomic traits. Traits in light grey boxes refer to root

Table 3 Marker locations, allelic effects and LOD scores for root traits with annotation indicating co-localization with agronomic traits

Traita SNP marker Linkage group Position (cM) Allelic effectb LOD Favorable allele Alternative allele

Median width (DIRT)1 2227_693 6 2.8579 6.5** 3.05 G A BRGA (manual)2,3 4510_497 10 50.1197 6.0* 2.63 T A D20 (DIRT)3 4245_136 10 50.5343 0.001** 2.91 G A D50 (DIRT)3 4245_136 10 50.5343 0.009** 2.99 G A D60 (DIRT)2,3 4245_136 10 50.5343 0.007** 3.01 A G − CPD 25 (DIRT)4 10905_418 1 45.1436 0.029* 2.59 A G − See Table 1 for abbreviation definitions a Colocation of QTL affecting root traits and agronomical traits measured in other studies: 1 seed weight per plant (Muchero et al. 2013); 2 Striga tolerance (Ouédraogo 2002); 3 seed number per plant (Muchero et al. 2013); 4 pod= number (Muchero et al. 2013) = = = b Marker additive effect, *, ** are significance level at P < 0.01 and 0.001, respectively; a positive allele effect indicates the favorable allele contributing to a positive phenotypic value for the root trait, while a negative allelic effect indicates the favorable allele contributing to a negative phenotypic value for the root trait a three-dimensional unit. Traits with a scaling coefficient to increase categorization precision. We discuss here only α > 0.33 plus 0.15 are positively allometric and indicate traits with significant P values (P < 0.001) for the regres- that the trait scales faster than shoot biomass. Traits with sion line. We calculated the adjusted R2 of the relation- α < 0.33 are negatively allometric and demonstrate trait ship as a measure of the predictive power of the trait in increases at a rate lower than expected from shoot bio- question. Five traits had a significant adjusted R2 and P mass. The value 0.15 0.33 was chosen to create a zone value of the slope. Two traits, TD10, and TD15, have ± where trait scaling was roughly proportional to biomass α > 0.33 0.15 indicating they are positively allometric. ± 1 3 426 Theor Appl Genet (2017) 130:419–431

Table 4 Allometric comparisons of correlations between plant bio- being related to drought and low P tolerance, respectively. mass and root traits This study integrates root phenotyping over two seasons in Trait Adjusted R2 Intercept Slope (α) P value (slope) the same environment with yield trials from multiple distinct environments. The effects of environmental variability TD10 0.16 0.47 0.61 <0.001*** − on root architecture, shoot growth, Striga infestation, and TD15 0.34*** 0.53 0.58 <0.001*** − ultimately yield introduced by these cross-location compar- TD 0.23*** 0.17 0.37 <0.001*** isons is undoubtedly large. Allelic effect and the number of SD (manual) 0.36*** 0.71 0.22 <0.001*** co-localizations between root traits and performance indi- Avg root density 0.05 0.076 0.16 0.004** − cators would likely increase if both root and performance CPD 75 0.12 0.24 0.14 <0.001*** − data were collected from the same environment. Further- Median width 0.02 1.97 0.12 0.011* more, the limited marker density available for cowpea Stem diameter 0.15*** 0.98 0.1 <0.001*** means that the SNPs used might not be perfect markers for (DIRT) traits and one should not place undue importance on weak CPD 25 0.07*** 0.026 0.079 <0.001*** allelic effects. Nevertheless, several interesting associations BRGA 0 1.51 0.071 0.07 were found between root traits and performance indicators Max width 0 2.35 0.037 0.16 that merit further investigation. ARGA 0 1.63 0.027 0.49 The co-localization between CPD25 and pod number RTA range 0 1.65 0.005 0.82 with a slightly negative allelic effect, indicated a smaller D60 0.02 0.16 0.01 0.4 − − hypocotyl was correlated with greater pod number. Analy- D50 0.02 0.19 0.02 0.17 − − sis by individual trials revealed that marker effect depends D20 0.02 0.33 0.046 0.02* − − on environment, suggesting the advantage of a given phe- D10 0.02 0.41 0.07 0.006** − − notype depends on environment. A large diameter hypoco- Allometric analysis was performed by plotting a linear regression of tyl could be useful as a reserve of carbohydrates for future log10 of each trait against log10 of total plant dry weight. The allomet- reallocation in a perennial, fodder or multiple harvest crop- ric scaling coefficient (α) of 0.33 with a margin of 0.15 was used ± ping system (Gwathmey et al. 1992). However, research as the threshold where trait response was proportional to a change in common bean indicates this “insurance” strategy con- in biomass. Traits with significant slopes greater than 0.45 exceeded proportional increase in biomass and traits with significant slopes less strains yield in terminal drought annual cropping systems, than 0.18 increased more slowly than if proportional to biomass. See which rewards a decisive shift to reproduction (Rao et al. Table 1 for abbreviation definitions 2009, 2013; Beebe 2012). Thus, a large diameter hypocotyl could constitute a misallocated reservoir of resources if those resources are not readily reallocated to reproduc- The non-significant R2 for TD10 indicates it is not strongly tive processes. QTL with both positive and negative allelic controlled by shoot biomass. TD and TD15 have highly effects were detected for a related measure, SD (accessed significant R2 and P value of slope. However, the slope (α) via DIRT and manual measurements), highlighting distinct of TD and SD are within 0.15 of 0.33 and are considered loci that are related to greater and smaller SD. This sug- ± to be nearly isometric with shoot biomass. Highly signifi- gests that selection may be possible to move towards either cant R2 for TD and SD, 0.23 and 0.36, respectively, were a smaller or larger hypocotyl. observed indicating greater predictive power. CPD 25 and The fact that QTL affecting root system shape (includ- stem diameter measured with DIRT had significant slopes ing BRGA, D20, D50 and D60) co-localized with QTL for close to zero and low R2, suggesting that the trait increases Striga resistance also indicates a possible role of root archi- independently from shoot biomass. tecture in avoiding Striga attachment and growth. An allele with a large positive effect for BRGA and an allele with a small negative effect for D60 both conferred Striga tol- Discussion erance and indicated that steeper basal root growth angles are related to Striga tolerance. Since the positive allele We suggest root phenotyping has utility for identifying for D20 and D50 is similar to the Striga susceptible allele mechanisms that drive performance. This statement is while the positive allele for D60 conferred a slightly nega- supported by moderate to high heritabilities of root traits tive allelic effect on Striga tolerance, we hypothesize that (Tables S2 and S3) and GWAS-based co-localizations D60 represents the inflection point where placing more root between root traits and performance indicators (Tables 2, 3; length above increases susceptibility and more root length Fig. 1). Greenhouse and field based studies in other crops below increases resistance. The implication is that less root have identified several root phenes, particularly those that length in the topmost layer of soil is related to Striga tol- lead to more efficient deep and shallow exploration, as erance. The greater abundance of Striga seeds in shallow

1 3 Theor Appl Genet (2017) 130:419–431 427

roots, the donor parent of this trait, SuVita-2, deploys very few roots in shallow soil zones compared to other lines in the diversity panel (see Fig. 2 for example of Striga attachment). The resultant root architecture resembles a narrow but deep cone with a predominate tap root, which places few roots in shallow soil zones and may also be related to greater water extraction from deep soil (Fig. 3). Similar biotic stress avoidance mechanisms may also play a role in other parasitic plant or nematode resistance. A potential tradeoff between Striga avoidance and arbuscular mycor- rhizal avoidance may exist as they occur in similar shallow soil and take advantage of similar chemical plant cues (Parniske 2008). The QTL co-localization and their positive effects on Fig. 2 Example of Striga attachment on cowpea (IT86D 1010) root both median root width and seed weight per plant (Muchero grown at URBC, note horizontal growth angle and relative enlarge- ment of host lateral root et al. 2013) suggest that a moderately large, broad cone- shaped root system is likely to be beneﬁcial in most environments. However, marker effects for median width soil suggests that avoidance of the shallow soil domain (marker 2227_693) and CPD 25 (marker 10905_418) on may prevent triggering Striga germination or avoid Striga performance indicators such as seed weight and seed num- attachment (Van Delft et al. 2000). Based upon phenotypic ber varied by environment (Table 5). Correlations between data on the number and angle of adventitious and basal median width and performance indicators also varied by

Fig. 3 From left to right SuVita-2 showing steep, narrow, tap root Local showing shallow root system with many adventitious roots and dominated architecture; TVu 9557 showing average root system with shallow adventitious and basal roots many adventitious and basal roots and steeper growth angle: Gorum

Table 5 Marker allelic effects Trait Marker 2227_693 (median Marker 10905_418 (CPD 25) on median width and CPD 25 width) and associated agronomic traits in three environments AA GG P value AA GG P value

2013_CPD25 1.3 1.3 0.267 1.4 1.3 0.002 2013_MedianWidth 119.6 139.1 0.001 132.5 129.6 0.541 2009_USA_Seed Weight 63.6 58.6 0.038 69.6 58.4 0.001 2008_Kamboinse_Seed Weight 21.8 17.9 0.020 25.4 18.1 0.001 2008_Kamboinse_SeedWtPlant 3.6 2.9 0.016 4.1 3.0 0.004 2008_Kamboinse_SeedNumberPlant 6.0 6.2 0.502 5.5 6.3 0.032 2007_Kano_Seed Weight 11.4 16.0 0.006 10.6 15.0 0.028 2009_Pobe_Seed Weight 21.1 19.3 0.089 23.4 19.4 0.002

USA, Kamboinse, Kano and Pobe refer to different locations of trials. SeedWTPlant means seed weight per plant. SeedNumberPlant means seed number per plant

1 3 428 Theor Appl Genet (2017) 130:419–431

Table 6 Trait correlations between median width and CPD 25 with The discrepancy in the allometric relationships between performance indicators in three environments seemingly comparable traits such as TD and CPD 25, and Correlation coefficients 2013_MedianWidth 2013_CPD25 stem diameter measured manually and using DIRT lies in the trait definition. TD is a diameter measurement at a 2013_CPD25 0.139 − defined location below the soil line. TD traits capture dif- 2007_Kano_Seed Weight 0.252** 0.106 ferential secondary growth at a given point in time inde- 2008_Kamboinse_SeedNum- 0.050 0.047 pendent of total plant biomass or total root system size. berPlant In contrast, CPD traits measure diameter increments rela- 2008_Kamboinse_Seed Weight 0.250** 0.212* − tive to excavated taproot length meaning CPD 25 does not 2008_Kamboinse_SeedWtPlant 0.240** 0.182* − always measure the same point as TD. Because CPD traits 2009_Pobe_Seed Weight 0.024 0.074 may measure diameter at different points than TD traits 2009_USA_Seed Weight 0.067 0.229* − they may be under different genetic control, as suggested *, ** are significance at P < 0.05 and 0.01, respectively. USA, Kam- by GWAS. Stem diameter measured manually and using boinse, Kano and Pobe refer to different locations of trials. SeedWT- DIRT differs slightly from each other. Stem diameters Plant means seed weight per plant. SeedNumberPlant means seed measured with DIRT resulted in α values and adjusted R2 number per plant about half than achieved for SD measured manually. Man- ual stem diameter is taken at a point determined by subtle environment (r 0.25 in Kano, r 0.25 in Kamboinse) color variations indicating soil level. DIRT measures stem = = − (Table 6). This suggests that median width and CPD 25 are diameter at the topmost point of the excised stem, which related to enhanced performance in some environments and does not always coincide with soil level. While the allo- related to inferior performance in others. Tradeoffs between metric relationships are not consistent, significant SNPs for costs and benefits are in general extremely dependent upon manually and image accessed measures of stem diameter environment, particularly when soil resources and environ- co-locate on the genome suggesting both have utility for mental constraints occur in contrasting spatial or temporal trait-based selection. Identification of genetic control using arrangements. different methods indicates that both manual and image A positive allelic effect on both median width and analysis based tools have utility for breeding programs. yield components suggests that a broad root system con- Our introductory study posits a root ideotype with large fers an advantage that leads to greater yield. We hypoth- median width and steeper BRGA. As this root system may esize that this root system architecture offers a balanced efficiently explore both shallow and deep zones it is termed foraging strategy that maximizes benefits relative to costs dimorphic. This type of root system may increase tolerance and efficiently explores both shallow and deep soil zones. to multiple edaphic stresses while avoiding Striga parasit- In an environment in which water is most limiting and ism in shallow soil. The ideal stay-green phenotype for a the ideal strategy is to escape drought by rapid matura- combined stress environment may be built upon optimized tion or deep soil exploration a minimalistic or steep root root and hypocotyl phenes, including a balanced but deep system may be best. Future research needs to address root architecture and more parsimonious water uptake. the complexities of matching phenotype to environment Combining these phenes may permit an increased period of highlighted by these environmentally dependent allelic water extraction (Vadez et al. 2013) with a positive balance effects. between benefit and costs of resource acquisition. It is a central but often overlooked task of physiolo- Ongoing work is investigating the mechanisms regulat- gists, geneticists and breeders to distinguish independent ing soil water use and in particular the effects of root ana- fundamental phenes from auto-correlations and allometric tomical phenes such as xylem diameter on conductance relationships. This study found SD and TD to be isometri- and soil moisture budgets. One example of this is tap root cally related to plant biomass and TD15 to be positively diameter, which seems intuitively linked to greater total allometrically related to plant biomass. TD10 is positively xylem area and to increased ability to extend into deep soil allometric but the insignificant R2 indicates it can scale zones in a terminal drought situation. We hypothesize that independently from shoot biomass. This suggests that it a strong tap root with large number of small diameter ves- may be possible to select for a strong tap root on a rela- sels offers a balance that extracts large volumes of water in tively small plant. We speculate that TD10 affects deep a metered fashion without depleting soil water before the soil exploration more than TD because a larger diameter growing cycle is over. root may be able to access deeper soil domains, sup- We present data indicating that both manual and ply greater quantities of water and contribute to drought image-based phenotyping of mature, field-grown cowpea avoidance. root systems detected QTL describing root architectural

1 3 Theor Appl Genet (2017) 130:419–431 429 phenes and some of these co-located with QTL related to References efficient resource acquisition and performance in suboptimal environments. Phenotyping of mature roots under Agbicodo EM, Fatokun CA, Muranaka S, Visser RGF, Linden van field conditions is important because it more closely der CG (2009) Breeding drought tolerant cowpea: constraints, accomplishments, and future prospects. Euphytica 167:353–370 approximates agricultural conditions than do greenhouse Atokple IDK, Singh BB, Emechebe AM (1995) Genetics of resistance or laboratory based phenotyping platforms. Our data sug- to Striga and Alectra in cowpea. J Hered 86:45–49 gest that root architectural traits form part of the suite of Barber S (1995) Soil nutrient bioavailability: a mechanistic approach. traits conferring abiotic stress tolerance and Striga resist- Wiley, New York Bayuelo-Jiménez JS, Gallardo-Valdéz M, Pérez-Decelis VA, Magda- ance. We propose that traits contributing to a cone-shaped leno-Armas L, Ochoa I, Lynch JP (2011) Genotypic variation architecture should be investigated further and potentially for root traits of maize (Zea mays L.) from the Purhepecha Pla- integrated into Striga management strategies such as those teau under contrasting phosphorus availability. Field Crops Res outlined elsewhere (Franke et al. 2006). Phenotyping root 121:350–362 Beebe SE (2012) Common bean breeding in the tropics. In: Jan- architecture may help to identify additional mechanisms ick J (ed) Plant breeding reviews, vol 36. Wiley, New York, pp related to increased productivity in sub-optimal environ- 357–426 ments and accelerate the release of climate adapted culti- Beebe SE, Rojas-Pierce M, Yan X, Blair MW, Pedraza F, Muñoz F, vars. Modeling and ideotype approaches could be useful Tohme J, Lynch JP (2006) Quantitative trait loci for root architecture traits correlated with phosphorus acquisition in common to inform physiology studies to investigate trait integra- bean. Crop Sci 46:413–423 tion and tradeoffs. Knowledge of allometric relationships Belko N, Zaman-allah M, Cisse N, Ndack Diop N, Zombre G, Ehlers should inform breeding targets and root ideotype devel- JD, Vadez V (2012a) Lower soil moisture threshold for transpi- opment. Similarities in root architecture between cow- ration decline under water deficit correlates with lower canopy conductance and higher transpiration efficiency in drought-toler- pea, soybean, common bean, tepary bean and other leg- ant cowpea. Funct Plant Biol 39:306–322 umes suggest genetic control may be homologous and Belko N, Zaman-Allah M, Diop NN, Cisse N, Zombre G, Ehlers JD, agronomic impact of certain root architectures may be Vadez V (2012b) Restriction of transpiration rate under high vapour comparable. pressure deficit and non-limiting water conditions is important for terminal drought tolerance in cowpea. Plant Biol 15:304–316 Belko N, Cisse N, Diop NN, Zombre G, Thiaw S, Muranaka S, Ehlers Author contribution statement JB phenotyped roots JD (2014) Selection for postflowering drought resistance in and led writing, analysis, figure preparations and revisions. short- and medium- duration cowpeas using stress tolerance indi- HS performed GWAS on root traits, correlated performance ces. Crop Sci 54:25–33 Bonser AM, Lynch JP, Snapp S (1996) Effect of phosphorus defi- indicators to root traits, as well as contributed to writing, ciency on growth angle of basal roots in Phaseolus vulgaris. analysis, figure preparation, interpretation and revisions. New Phytol 132:281–288 BLH contributed expertise to genetic analysis, writing and Bucksch A, Burridge J, York LM, Das A, Nord E, Weitz JS, Lynch JP interpretation. PR contributed interpretation, analysis and (2014) Image-based high-throughput field phenotyping of crop roots. Plant Physiol 166:470–486 manuscript revisions. AB developed the automated image Burridge J, Jochua CN, Bucksch A, Lynch JP (2016) Legume shov- phenotyping platform and contributed to interpretation, elomics: high—throughput phenotyping of common bean (Pha- writing and revising the manuscript. JL contributed to con- seolus vulgaris L.) and cowpea (Vigna unguiculata subsp, unguic- ceptual design, interpretation, writing and revisions of the ulata) root architecture in the field. Field Crops Res 192:21–32 Cattivelli L, Rizza F, Badeck FW, Mazzucotelli E, Mastrangelo AM, manuscript. Francia E, Marè C, Tondelli A, Stanca AM (2008) Drought tolerance improvement in crop plants: an integrated view from breeding to genomics. Field Crops Res 105:1–14 Acknowledgements This work was supported by the Howard G. Das A, Schneider H, Burridge J, Karine A, Ascanio M, Topp CN, Buffet Foundation, the USAID Feed the Future Innovation Laboratory Lynch JP, Weitz JS, Bucksch A (2015) Digital Imaging of Root for Climate Resilient Beans, and the Feed the Future Innovation Lab Traits (DIRT): a high-throughput computing and collaboration for Collaborative Research on Grain Legumes. Genotyping was sup- platform for field-based root phenomics. Plant Methods 11:1–12 ported by the CGIAR Generation Challenge Program. This work was de Barros I, Gaiser T, Lange F-M, Romheld V (2007) Mineral nutri- also supported by the USDA National Institute of Food and Agricul- tion and water use patterns of a maize/cowpea intercrop on ture, Hatch Project 4372, the NSF Plant Genome Research Program, a highly acidic soil of the tropic semiarid. Field Crops Res NSF 0820624 and the Center for Data Analytics, Georgia, Institute of 101:26–36 Technology, Spatial Networks in Biology: Organizing and Analyzing Ehlers JD, Hall AE (1997) Cowpea (Vigna unguiculata L. Walp). the Structure of Distributed Biological Systems. Any opinions, find- Field Crops Res 53:187–204 ings, conclusions, or recommendations expressed in this publication Fehr W (1993) Principles of cultivar development. Macmillan Pub- are those of the author(s) and do not necessarily reflect the view of lishing Company, New York the National Institute of Food and Agriculture (NIFA) or the United Franke AC, Ellis-Jones J, Tarawali G, Schulz S, Hussaini MA, Kureh States Department of Agriculture (USDA). I, White R, Chikoye D, Douthwaite B, Oyewole BD, Olanrewaju AS (2006) Evaluating and scaling-up integrated Striga hermon- Compliance with ethical standards thica control technologies among farmers in northern Nigeria. Crop Prot 25:868–878 The authors declare that they have no conflict of interest.

1 3 430 Theor Appl Genet (2017) 130:419–431

Gwathmey OC, Hall AE, Madore MA (1992) Adaptive attributes of Miguel MA, Postma JA, Lynch JP (2015) Phene synergism between cowpea genotypes with delayed monocarpic leaf senescence. root hair length and basal root growth angle for phosphorus Crop Sci 32:765–772 acquisition. Plant Physiol 167:1430–1439 Hall AE (2012) Phenotyping cowpeas for adaptation to drought. Front Miller C, Ochoa I, Nielsen K, Beck D, Lynch JP (2003) Genetic varia- Physiol 3:1–8 tion for adventitious rooting in response to low phosphorus avail- Ho MD, Rosas JC, Brown KM, Lynch JP (2005) Root architectural ability: potential utility for phosphorus acquisition from stratified tradeoffs for water and phosphorus acquisition. Funct Plant Biol soils. Funct Plant Biol 30:973–985 32:737–748 Muchero W, Diop NN, Bhat PR, Fenton RD, Wanamaker S, Pottorft Huynh B, Close TJ, Roberts PA, Hu Z, Wanamaker S, Lucas MR, M, Hearne S, Cisse N, Fatokun C, Ehlers JD, Roberts PA, Close Chiulele R, Cisse N, David A, Hearne S, Fatokun C, Diop NN, TJ (2009) A consensus genetic map of cowpea [Vigna unguicu- Ehlers JD (2013) Gene pools and the genetic architecture of lata (L) Walp.] and synteny based on EST-derived SNPs. Proc domesticated cowpea. Plant Genome 6:1–8 Natl Acad Sci USA 106:18159–18164 Huynh B-L, Ehlers JD, Ndeve A, Wanamaker S, Lucas MR, Close Muchero W, Ehlers JD, Roberts PA (2010) Restriction site polymor- TJ, Roberts PA (2015) Genetic mapping and legume synteny phism-based candidate gene mapping for seedling drought tol- of aphid resistance in African cowpea (Vigna unguiculata L. erance in cowpea [Vigna unguiculata (L.) Walp.]. Theor Appl Walp.) grown in California. Mol Breeding 35:36. doi:10.1007/ Genet 120(3):509–518. doi:10.1007/s00122-009-1171-6 s11032-015-0254-0 Muchero W, Ehlers JD, Close TJ, Roberts PA (2011) Genic SNP Huynh B-L, Matthews WC, Ehlers JD, Lucas MR, Santos JRP, Ndeve markers and legume synteny reveal candidate genes underlying A, Close TJ, Roberts PA (2016) A major QTL corresponding to QTL for Macrophomina phaseolina resistance and maturity in the Rk locus for resistance to root-knot nematodes in cowpea cowpea [Vigna unguiculata (L) Walp.]. BMC Genomics 12:8. (Vigna unguiculata L. Walp.). Theor Appl Genet 129(1):87–95. doi:10.1186/1471-2164-12-8 doi:10.1007/s00122-015-2611-0 Muchero W, Roberts PA, Diop NN, Drabo I, Cisse N, Close TJ, Kamara AY, Ekeleme F, Jibrin JM, Tarawali G, Tofa I (2014) Agri- Muranaka S, Boukar O, Ehlers JD (2013) Genetic architecture culture, Ecosystems and environment assessment of level, extent of delayed senescence, biomass, and grain yield under drought and factors influencing Striga infestation of cereals and cowpea stress in cowpea. PLoS One 8:1–10 in a Sudan savanna ecology of northern Nigeria. Agric Ecosyst Niklas KJ (1994) Plant allometry: the scaling of form and process. Environ 188:111–121 University of Chicago Press, Chicago Kirkegaard JA, Lilley JM, Howe GN, Graham JM (2007) Impact of Nord EA, Shea K, Lynch JP (2011) Optimizing reproductive phe- subsoil water use on wheat yield. Aust J Agric Res 58:303–315 nology in a two-resource world: a dynamic allocation model of Kugblenu YO, Kumaga FK, Ofori K, Adu-Gyamfi JJ (2014) Evalua- plant growth predicts later reproduction in phosphorus-limited tion of cowpea genotypes for phosphorus use efficiency. J Agric plants. Ann Bot 108:391–404. doi:10.1093/aob/mcr143 Crop Res 2:202–210 Noubissie Tchiagam JB, Bell JM, Guissai Birwe S, Gonne S, Youmbi Li J, Lis KE, Timko MP (2009) Molecular genetics of race-specific E (2010) Varietal response of cowpea (Vigna unguiculata (L.) resistance of cowpea to Striga gesnerioides (Willd.). Pest Manag Walp.) to Striga gesnerioides (Willd.) vatke race SG5 infesta- Sci 65:520–527 tion. Not Bot Hort Agrobot Cluj 38:33–41 Lipka AE, Tian F, Wang Q, Peiffer J, Li M, Bradbury PJ, Gore MA, Omoigui LO, Ishiyaku MF, Ousmane B, Gowda BS, Timko MP Buckler ES, Zhang Z (2012) GAPIT: genome association and (2011) Application of fast technology for analysis (FTA) for prediction integrated tool. Bioinformatics 28:2397–2399 sampling and recovery of deoxyribonucleic acid (DNA) for Lucas MR, Diop N-N, Wanamaker S, Ehlers JD, Roberts PA, Close TJ molecular characterization of cowpea breeding lines for Striga (2011) Cowpea–soybean synteny clarified through an improved resistance. Afr J Biotechnol 10:19681–19686 genetic map. Plant Genome 4:218–225 Ouedraogo J, Maheshwari V, Berner D, St Pierre C-A, Belzile F, Lucas MR, Ehlers JD, Huynh B-L, Diop N-N, Roberts PA, Close TJ Timko M (2001) Identification of AFLP markers linked to resist- (2013a) Markers for breeding heat-tolerant cowpea. Molecular ance of cowpea (Vigna unguiculata L.) to parasitism by Striga Breeding 31(3):529–536. doi:10.1007/s11032-012-9810-z gesnerioides. Theor Appl Genet 102:1029–1036 Lucas MR, Huynh B, Vinholes S, Cisse N, Drabo I, Ehlers JD, Rob- Ouedraogo JT, Ouedraogo M, Gowda BS, Timko MP (2012) Devel- erts PA, Close TJ (2013b) Association studies and legume syn- opment of sequence characterized amplified region (SCAR) teny reveal haplotypes determining seed size in Vigna unguicu- markers linked to race-specific resistance to Striga gesne- lata. Front Plant Sci 4:95. doi:10.3389/fpls.2013.00095 rioides in cowpea (Vigna unguiculata L.). Afr J Biotechnol Lynch JP (2011) Root phenes for enhanced soil exploration and 11:12555–12562 phosphorus acquisition: tools for future crops. Plant Physiol Ouédraogo JT, Tignedre J-B, Timko MP, Belzile FJ (2002) AFLP 156:1041–1049 markers linked to resistance against Striga gesnerioides race Lynch JP (2015) Root phenes that reduce the metabolic costs of soil 1 in cowpea (Vigna unguiculata). Genome 45(5):787–793. exploration: opportunities for 21st century agriculture. Plant Cell doi:10.1139/G02-043 Environ 38:1775–1784 Parniske M (2008) Arbuscular mycorrhiza: the mother of plant root Lynch JP, Brown KM (2001) Topsoil foraging—an architectural endosymbioses. Nat Rev Microbiol 6:763–775 adaptation of plants to low phosphorus availability. Plant Soil Pottorff M, Wanamaker S, Ma YQ, Ehlers JD, Roberts PA, Close TJ 237:225–237 (2012) Genetic and Physical Mapping of Candidate Genes for Lynch JP, Brown KM (2012) New roots for agriculture: exploit- Resistance to Fusarium oxysporum f.sp. tracheiphilum Race 3 in ing the root phenome. Philos Trans R Soc Lond Ser B Biol Sci Cowpea [Vigna unguiculata (L.) Walp]. PLoS One 7 (7):e41600. 367:1598–1604 doi:10.1371/journal.pone.0041600 Lynch JP, Wojciechowski T (2015) Opportunities and challenges Pottorff MO, Li G, Ehlers JD, Close TJ, Roberts PA (2014) Genetic in the subsoil: pathways to deeper rooted crops. J Exp Bot mapping, synteny, and physical location of two loci for Fusar- 66:2199–2210 ium oxysporum f. sp. tracheiphilum race 4 resistance in cowpea Matsui T, Singh BB (2003) Root characteristics in cowpea related to [Vigna unguiculata (L.) Walp]. Mol Breeding 33(4):779–791. drought tolerance at the seedling stage. Exp Agric 39:29–38 doi:10.1007/s11032-013-9991-0

1 3 Theor Appl Genet (2017) 130:419–431 431

R Core Team (2014) R: a language and environment for statistical Vadez V, Kholová J, Yadav RS, Hash CT (2013) Small temporal dif- computing ferences in water uptake among varieties of pearl millet (Pen- Rao IM, Beebe SE, Polania J, Grajales M, Cajiao C, Garcia R, nisetum glaucum (L.) R. Br.) are critical for grain yield under Ricaurte J, Rivera M (2009) Physiological basis of improved terminal drought. Plant Soil 371:447–462 drought resistance in common bean: the contribution of photo- Van Delft GJ, Graves JD, Fitter AH, Van Ast A (2000) Striga seed synthate mobilization to grain. Interdrought III: the 3rd interna- avoidance by deep planting and no-tillage in sorghum and maize. tional conference on integrated approaches to improve crop pro- Int J Pest Manag 46:251–256 duction under drought-prone environments, Shanghai, pp 11–16 Varshney RK, Terauchi R, McCouch SR (2014) Harvesting the prom- Rao I, Beebe S, Polania J, Ricaurte J, Cajiao C, Garcia R, Rivera M ising fruits of genomics: applying genome sequencing technolo- (2013) Can Tepary Bean be a model for improvement of drought gies to crop breeding. PLoS Biol 12(6):e1001883 resistance in Common Bean? Afr Crop Sci J 21:265–281 White JW, Singh SP (1991) Sources and inheritance of earliness in Rubiales D, Fernández-Aparicio M (2012) Innovations in parasitic tropically adapted indeterminate common bean. Euphytica weeds management in legume crops. A review. Agron Sustain 55:15–19 Dev 32:433–449 Yadav SS, Hunter D, Redden B, Nang M, Yadava DK, Habibi AB Serebrovsky AS (1925) ‘Somatic segreation’ in domestic fowl. J (2015) Impact of climate change on agriculture production, food, Genet 16:33–42 and nutritional security. Crop wild relatives and climate change, Singh BB, Chambliss OL, Sharma B (1997) Recent advances in cow- pp 1–23 pea breeding. In: Singh BB, Mohan Raj DR, Dashiell KE (eds) York LM, Nord EA, Lynch JP (2013) Integration of root phenes for Advances in cowpea research. Copublication of International soil resource acquisition. Front Plant Sci 4:1–15 Institute of Tropical Agriculture (IITA) and Japan International Zhang Z, Ersoz E, Lai C-Q, Todhunter RJ, Tiwari HK, Gore MA, Research Center for Agricultural Sciences (JIRCAS). IITA, Bradbury PJ, Yu J, Arnett DK, Ordovas JM, Buckler ES (2010) Ibadan, pp 30–49 Mixed linear model approach adapted for genome-wide associa- Singh BB, Ajeigbe HA, Tarawali SA, Fernandez-Rivera S, Abubakar tion studies. Nat Genet 42:355–360 M (2003) Improving the production and utilization of cowpea as Zhu J, Brown KM, Lynch JP (2010) Root cortical aerenchyma food and fodder. Field Crops Res 84:169–177 improves the drought tolerance of maize (Zea mays L.). Plant Vadez V, Soltani A, Sinclair TR (2012) Modelling possible beneﬁts Cell Environ 33:740–749 of root related traits to enhance terminal drought adaptation of chickpea. Field Crops Res 137:108–115

1 3 CHAPTER 3 Comparative annual grain legume root architecture Abstract Grain legume production is fundamental to many smallholder and subsistence farmers and to local and regional economies. Suboptimal water and phosphorus availability are primary limitations to production, which typically have contrasting availabilities in the soil profile when both are limiting. Root architecture is important for water and phosphorus acquisition but tradeoffs may mitigate the benefits of breeding for single stress environments. Panels of common bean (Phaseolus vulgaris), tepary bean (Phaseolus acutifolius), cowpea (Vigna unguiculata), soybean (Glycine max), chickpea (Cicer arientinum), and groundnut (Arachis hypogaea) were evaluated for a variety of root architectural characteristics. A smaller collection including lima bean (Phaseolus lunatus) and faba bean (Vicia faba) and minor Phaseolus species were also evaluated. We found that legume root systems can be placed on a root system architecture (RSA) spectrum according to the dominance of superficial versus primary roots. This classification corresponds to water availability in their putative domestication environment. Root system architecture can be grouped into root architectural categories corresponding to epigeal or hypogeal germination. Hypogeal germinators have a root system composed of epicotyl roots, the primary root and lateral roots. Epigeal germinators have distinguishable hypocotyl and basal roots as well as a primary root and primary root laterals. Epigeal and hypogeal categories of root systems present different adaptive mechanisms, which may be complemented by a particular life strategy. For instance, the chickpea root system may be best suited for an elongated phenology but parsimonious water use strategy. The tepary root system may be best suited for accelerated phenology with dynamic but potentially elevated water use. All strategies likely involve balancing tradeoffs and opportunity costs. We identified inverse relationships between investment in different root classes in most species and between indicators of deep and shallow exploration in all species. Bean and tepary showed particularly strong tradeoffs in investment patterns, while chickpea and groundnut show less pronounced tradeoffs. These tradeoffs may be formed by interactions between resource availability, resource acquisition strategy and life strategy in a given domestication environment. Using an economic analysis to understand the advantages and disadvantages of various root architectures may help to maintain and expand the range of these important food security grain legumes. We highlight instances of dimorphic root architectures that may co- optimize resource acquisition in environments with contrasting resource availability profiles. Introduction Food production must increase dramatically within the next 50 years to support rapidly growing demands in spite of increases in the incidence and severity of drought (Hunter et al., 2017; IPCC, 2014; Lobell et al., 2011, 2008). The 1st Green Revolution and GM crops have failed to benefit subsistence farmers in

! 34! many regions and resolve issues related to excessive fertilization (Lynch, 2007a). Part of the reason for this failure was because food security crops such as grain legumes and low input cropping systems were not targeted. Low input cropping systems are characterized by suboptimal water and nutrient availability, which have always been fundamental factors affecting plant productivity. Root phenotypes are key adaptations to these conditions and present opportunities for crop improvement (Lynch, 2007a; Rogers and Benfey, 2015). Information on root architecture, domestication environment and adaptive strategies may help guide breeding programs to pair a root system with a life strategy and production environment and thus increase system resiliency (Bullock et al., 2017). We focus on common bean (Phaseolus vulgaris), cowpea (Vigna unguiculata), chickpea (Cicer arientinum), and groundnut (Arachis hypogaea) because they are among the most important crops for food security of smallholder farmers (Graham and Vance, 2003). We include tepary bean (Phaseolus acutifolius) because it has similar seed characteristics to common bean but with much increased heat and drought tolerance, and because of this is receiving attention as a climate resilient crop (Porch et al., 2013). We include soybean (Glycine max) because of its global importance for food, feed and oil. Other key grain legumes such as pea (Piscum sativum L.) and lentil (Lens culinaris L.) are important crops, would have introduced interesting diversity to the study and would have complemented Vicia and Cicer in the Galegoid family, but were not included due to space and time constraints. Legumes are placed phylogenetically in the Fabaceae family and the majority of grain legumes are in the Papilonaceae sub family (Smýkal et al., 2015).The Phaseolus, Soya and Vigna species are in the Phaseolid clade and are relatively closely related to each other, share extensive genomic regions (Gujaria-verma et al., 2016; Kim et al., 2015; Lucas et al., 2011; McClean et al., 2010). The other legumes studies are from more disparate ancestries. Vicia is in the Vicea tribe and Galegoid clade. Cicer is also in the Galegoid clade but it falls in the Cicereae sub-clad. Arachis is from the Dalbergia clade, which has the smallest number of cultivated legumes (Iwata et al., 2013; Smýkal et al., 2015). Domestication environments The grain legumes we treat as crops today are in the most recent stage of 4- 10,000 year process of adaptation (Scott, 2017). The principal agronomic traits that humans have modified include reducing shattering/dehiscence, shifting towards determinant non-climbing growth, reducing photoperiodicity, and reducing seed dormancy and hardness (Purugganan and Fuller, 2009). Modern (i.e. since 1950) plant breeding has had additional but not completely understood effects on grain legume phenotypes as we have developed new varieties for high input, monoculture, mechanized farming systems that are an anomaly in their history. Domestication bottlenecks restrict the genetic variation available to breeders, and agronomic requirements restrict the strategies employable in a given cropping system. Understanding what favorable root phenotypes are present in extant germplasm may help breeders to accelerate their adaptation to changing environments.

! 35! Recent evidence suggests that Glycine max, soybean, was first domesticated from G. soya in the Huang-Huai Valley in Central China between the Yellow and Huai Rivers (Sedivy et al., 2017). The domestication process took place over a long period of time, an estimated 9,000 to 5,000 years ago in a variety of locations and probably multiple types of agro-ecosystems (Sedivy et al., 2017) around the same time that millet and rice were beginning to be cultivated (Zong et al., 2017). Speculation on the domestication environment should be considered cautiously but we may guess that first adopters of soybean cultivation were those living in relatively more accessible, temperate and fertile valleys and that intensity of drought and fertility constraints were balanced with each other. Chickpea (Cicer arientinum) is one of the ancient founder crops of the Fertile Crescent (Zohary et al., 2012) and is globally the second most important pulse crop with great importance to smallholder farmers in the Mediterranean, Middle East and India (Varshney et al., 2013). Wild relatives have been found in South Eastern Turkey growing in mixed environments with several other species (Abbo et al., 2010). Chickpea has undergone several severe bottlenecks before and during its domestication, which increase the need to fully explore existing genetic diversity as well as to categorize and exploit land races, wild varieties and related species (Abbo et. al., 2003). Terminal drought stress has historically been a primary constraint in chickpea production environments (Abbo et al., 2003b; Purushothaman et al., 2016) and modern production sometimes relies on stored soil moisture (Abbo et al., 2003a). Cowpea (Vigna unguculata) is especially important to food security, income generation and soil fertility in Western and Southern Africa (Boukar et al., 2016; Munos-Amatriain et al., 2017). Its native environment is thought to be savannah edges on acidic, low fertility soils with ephemeral water availability. The most recent research that two distinct centers of diversity in Southern and Western Africa suggesting separate domestication processes (Huynh et al., 2013). Some of the earliest uses are thought to be as animal fodder (Ng, 1995) and the oldest indicator of human consumption of seed is approximately 3,400 years BP (D’Andrea et al., 2017). Current practices likely resemble traditional farming systems and often use landraces grown in polyculture with millet or sorghum (Singh et al., 1997). This reduces direct competition with other cowpea plants. Breeding efforts have focused on one of two basic strategies, the first focusing on rapidly maturing genotypes with high harvest index. The other strategy targets production systems where young leaves and or dry forage as well as seed are used as fodder but there is not necessarily a tradeoff between seed and haulm production (Samiryeddypalle, 2017). The Phaseolus genus offers unique opportunities to study evolution as several species have been domesticated (Rendón-Anaya et al., 2017) and extensive genetic synteny exists among related Phaseolus species as well as other legumes (Bitocchi et al., 2017; Kaplan, 1965). Recent evidence suggests that common bean and cowpea are remarkably closely related (Munos-Amatriain et al., 2017). Common bean had two centers of domestication, one in Meso- America and one in the Andes, (Bitocchi et al., 2012; Rendón-Anaya et al., 2017;

! 36! Schmutz et al., 2014). These two gene pools are still distinct although gene flow between pools still happens and wild types occur from present day Mexico to Argentina (Mamidi et al., 2013). Evidence of its use dates to around 8,000 years BP (Bellucci et al., 2014; Gepts and Debouck, 1991; Kaplan et al., 1973; Kaplan and Lynch, 1999). Paleolinguistic evidence dates common bean to more than 5000 years BP in various locations in the Americas (Brown et al., 2014). Its native environment is thought to be mesic riparian zones, where there may be more intense competition for light than phosphorus (Beebe et al., 1997). Bean would presumably have germinated on the soil surface after explosive dehiscence and climbed on other vegetation to access light (Gepts and Debouck, 1991). The increase in whorl number and root hair length in cultivated beans suggests that domestication has extended its range to areas with less favorable nutrient availability that implies domestication has made bean more phosphorus efficient (Beebe et al., 1997). Lima bean has perhaps the widest range of adaptation of the Phaseolus genus (Gepts, 2014) and has two centers of domestication, the first in South America west of the Andes and another in Mexico containing two gene pools (Andueza- Noh et al., 2013). The large seeded Andean types were a staple food of the Moche and other South American civilizations who cultivated it in arid but irrigated areas west of the Andes, possibly as long ago as 5600 years BP (Kaplan and Lynch, 1999), if not longer (Kaplan et al., 1973). Details on lima bean domestication environments are scarce and information on early cropping systems even more so but given at least one of the Mesoamerican gene pools has been traced to tropical dry forests and we can assume efficient acquisition and utilization of soil water was a priority (Serrano-Serrano et al., 2012). Tepary bean (Phaseolus acutifolius) is native to arid areas of the Southwest US and Northwest Mexico where it has been grown by indigenous farmers for thousands of years (Nabhan and Felger, 1978). Wild types occur in both upland and floodplain areas characterized by extreme temperatures and paucity of available soil water. It has likely arisen from a single domestication event (Garvin and Weeden, 1994) and then adapted to distinct geographic regions (Gujaria- verma et al., 2016). Traditional cultivation systems seem to have been based on floodplain agriculture that take advantage of relatively nutrient rich alluvial soil but ephemeral water availability (Pratt and Nabhan, 1988). Planting timed to the monsoon rains works in concert with rapid germination, emergence and phenology to complete its life cycle before the soil water profile dries. Groundnut, Arachis hypogaea, native to Southern Bolivia and Northern Argentina has become an important smallholder and subsistence crop in Africa, India and Southeast Asia (Janila et al., 2013). Early domestication environments in South America varied widely but may have included sandbanks along streams and rivers (Hammons, 1994) and also grasslands or grassy patches in forests (Bertioli et al., 2011). However the tetraploid nature of cultivated groundnut limits the genetic variation available to breeders (Mallikarjuna et al., 2010). Geocarpy, the somewhat rare practice of subterranean seed development, contributes to heat and drought tolerance (Bertioli et al., 2011). These factors combine to make

! 37! groundnut improvement challenging but potentially useful genetic variation is likely available. Drought tolerance research has involved both water capture via increased root length density at depth (Koolachart et al., 2013) and more efficient water use (Krishnamurthy, 2007). Yard long bean, Vigna unguiculata spp sesquipedalis, derived from wild or early domesticated cowpea, is cultivated in South East Asia where it developed its characteristic long pods and viney growth habit (Kongjaimun et al., 2012). Mung bean, Vigna radiata was domesticated in Persia or Northwest India with early adoption in India (Fuller, 2007). Vigna umbellata or rice bean has been historically planted following rice in some South East Asian cropping systems. Tephrosia vogelii or fish bean is native to Africa but is used as fish poison and acaricide both in Africa and Southeast Asia. Psophocarpus tetragonolobus or winged bean is thought to be native to Africa but underwent significant domestication in tropical South East Asia (Harder and Smartt, 1995). Faba bean, Vicia faba was most likely domesticated in the Mediterranean region (Hebblethwaite, 1983). Faba bean is either grown at a winter crop or planted in early spring in accord with the local climate and is well adapted to late season water limitation (Hebblethwaite, 1983). Pigeon pea, Cajanus cajan is a widely adapted legume with exceptional heat, low soil fertility and drought stress tolerance native to the Indian subcontinent but with wild relatives throughout Southeast Asia and into Australia (Khoury et al., 2015). Lablab purpurenus (L.) Sweet has become popular as a drought tolerant cover crop. Recent population genetic analysis on lablab suggests an East African origin and range of drought tolerance (Robotham and Chapman, 2017). Soil resource availability and root architecture In epigeal germination, hypocotyl roots originate between the radical and the cotyledons, which are lifted out of the soil. Hypogeal germinators retain the cotyledons below the soil surface and have epicotyl-borne roots that emerge from shoot tissue above the subterranean cotyledons. Below the subterranean cotyledons is the primary root. Epigeal germinators have three classes of roots including adventitious, primary or taproot and basal roots (Figure 1). Basal roots originate at the base of the hypocotyl and above the taproot (Zobel and Waisel, 2010; Zobel, 2011). Common bean is unique in that the basal roots can be divided into discrete whorls (BRWN) (Miguel et al., 2013). Understanding RSA and its connections to resource capture is complicated by the opaque and heterogeneous nature of soil. Shovelomics quantifies root system architecture and can identify several important phenes affecting root distribution (Burridge et al., 2016a; Trachsel et al., 2011). Understanding the advantages and disadvantages of different RSA as well as understanding interactions between root phenes is aided by the use of economic concepts. Rhizoeconomics involves the application of micro-economics concepts such as risk, tradeoffs, diminishing returns and cost: benefit analysis to the spatio-temporal evaluation of root system architecture (Lynch et al., 2005). Rhizoeconomics is the study of optimizing resource caption in order to maximize growth and is useful to guide hypothesis testing. A basic tenant of rhizoeconomics involves opportunity costs, which

! 38! describes how allocation of resources in a particular way precludes implementation of other strategies and root deployment patterns. The two most fundamental constraints for nitrogen fixing legumes are water and phosphorus (P) (Beebe et al., 2009; Lynch and Brown, 2012; Wortmann et al., 1998) although low soil N and high soil acidity can also severely limit growth (Beebe et al., 2014). This influences the design of a life strategy involving acquiring adequate amounts of P and water over a period of time. Life strategy is logically related to both resource acquisition strategy and root architecture, as they must work in concert to be most efficient. An accelerated and decisive phenology may be appropriate for a high fertility but short growing season constrained by water availability or temperature (Polania et al., 2016; Rao et al., 2013). A nutrient-limited environment with adequate soil moisture would favor long growth pattern allowing increased P acquisition, P utilization, and yield (Nord and Lynch, 2009). Both of these strategies contrast to a perennial, conservative and biphasic strategy exhibited by some wild crop progenitors and crop relatives that may be appropriate for combined low fertility and seasonally dry environments with long or bimodal growing seasons (Beebe et al., 2008). Different drought adaptation strategies such as avoidance, escape or tolerance (Varshney et al. 2011) would be best paired with different sets of root phenes. We use the term ‘phene’ to refer to the units of the plant phenotype (Serebrovsky, 1925; York et al., 2013). The term ‘metric’ is used for calculated scores describing the root system. An avoidance strategy may involve deep roots to capture water. An escape strategy may involve a minimalist root system that requires limited investment paired with rapid phenology. A tolerance strategy may combine dimorphic root architecture that can capture some deep water and take advantage of shallow water deposited in light precipitation events with a parsimonious water use strategy. The life strategy of neighbors and competition with them may also affect the utility of an individual’s drought adaptation strategy. Research has shown that herringbone root architecture minimizes competition while maximizing soil resource acquisition (Fitter et al., 1991). In several dicot species under nutrient limitation the root system acquires a more herringbone structure (Fitter and Stickland, 1991; Taub and Goldberg, 1996). We suggest plastic allocation is also very important in resource limited environments as the plant attempts to maintain an allometric trajectory (Anfodillo et al., 2016; Weiner, 2004). A root system’s ability to respond to environmental cues such as increasing water content with depth has been noted in several crops (Vadez et al., 2008) and has been linked to drought tolerance. Greater rooting depth has been linked to increased yield when subjected to terminal drought (Comas et al., 2013) in chickpea (Gaur et al., 2008; Kashiwagi et al., 2008), soybean (Fenta et al., 2014), wheat (Lopes and Reynolds, 2010; Rich et al., 2016), groundnut (Jongrungklang et al., 2012; Koolachart et al., 2013) and in maize (Chimungu et al., 2014; Hund et al., 2009; Zhu et al., 2010). Vadez questions the conclusion that water acquisition and root architecture are most fundamental and posits that transpiration efficiency (TE) and water use efficiency (WUE) is critical to ensure

! 39! adequate water availability at key plant growth stages (Vadez, 2014). Merely acquiring and transpiring a large amount of water is no guarantee of desiccation avoidance, let alone maintaining yield under limited water availability. The efficient control of stomata to reserve water stores for later in phenology and transforming photosynthates to reproductive structure have been shown to be very important in lysimeter systems (Belko et al., 2014, 2012; Vadez et al., 2013). While an experimental system that limits available soil volume may bias results towards WUE and water saving strategies the importance of considering WUE is well taken. The consideration of water acquisition and WUE points to the importance of looking at the integrated life strategy of the genotype or species and understanding cost and benefit ratios related to the spatial and temporal aspects of soil exploration (Bloom et al., 1985; Lynch, 2007b). In most agricultural soils, resources are stratified, with P being more available in shallow soil strata and water generally more available in deep soil in the terminal drought environments that most farmers face (Lynch and Brown, 2001). Root phenes such as rooting depth and root distribution are modulated by root system architecture and are important mechanisms that determine the spatial and temporal placement of roots, which can affect resource acquisition (Lynch and Brown, 2001; Lynch and Wojciechowski, 2015). Common bean has genetic variation for BRGA that leads to root phenotypes contrasting for root length density by depth (Liao et al., 2004). Phenotypes that place more root length in shallow soil tend to have greater tolerance to low phosphorus and root phenotypes that place more root length in deep soil tend to be more drought tolerant (Ho et al., 2005). Other root architectural phene states of common bean that have been linked to greater productivity in soils with low soil fertility include greater basal root whorl number (BRWN) (Miguel et al., 2013), and greater adventitious root number (Miller et al., 2003; Ochoa et al., 2006). Large root hair length and density are also positively linked to phosphorus acquisition (Bates and Lynch, 2001) and tolerance to low P soils (Yan et al., 2004). Combining phene states of BRGA and root hair length has positive synergistic effects (Miguel et al., 2015). The common reality of contrasting edaphic resource availability leads to potential tradeoffs and opportunity costs between root architectural phenotypes best suited for water or for phosphorus acquisition (Ho et al., 2005). Tradeoffs between deep and shallow exploration have not been well documented in other grain legume crops. One way to avoid compromising phosphorus acquisition for water acquisition, or vica versa, is to deploy a dimorphic root architecture. A dimorphic architecture is characterized by root phenes that promote both shallow and deep exploration. Dimorphic architecture may reduce tradeoffs between acquisition of deep and shallow resources such as water and phosphorus and enable a plant to optimize production in multiple environments and environments with multiple edaphic limitations. Selecting for phenes related to resource acquisition requires a clear understanding of the root system of a given species and the possible advantages and interactions of that root system in dynamic and variable environments (Lynch, 2015). This goal requires the conceptualization of a four-dimensional

! 40! fitness landscape as environmental constraints stimulate plant responses and plant responses modulate resource allocation. The above discussion on tradeoffs between foraging strategies and the primacy of phosphorus acquisition or water acquisition or water use efficiency highlights the need to consider root architecture as part of an integrated resource acquisition and use strategy. This strategy must also be contextualized within the plant’s life strategy, which seeks to consider tolerance and avoidance strategies independently. The objective of this article is to shed light on differences in root architecture among grain legume species, speculate on how their domestication environment and life strategy complements their root architecture and identify opportunities to improve their root architecture. Analyzing domestication environment, life strategy and shoot and root architectural components together is termed an integrated phenotype. By understanding how an integrated phenotype makes agro-ecological sense within a given environment we may be able to generate new hypotheses matching an integrated phenotype to a particular agro-ecosystem. Observations on individual and linked selection criteria can be passed on to breeders so the information can be used to inform their crossing and selection process. Materials and Methods Plant material and field sites Two common bean diversity panels representing the Mesoamerican (n=120) and Andean (n=110) gene pools were phenotyped at R5 at the Apache Root Biology Center (ARBC), Wilcox, Arizona in 2016. The research station is located at 32˚02’00N, 109˚41’28W, 1321 masl, and has a Grabe loam soil (coarse-loamy, mixed, thermic Typic Torrifluvent) and 26˚C mean growing season temperature. A sub-set of the Tepary Diversity Panel was phenotyped at flowering (R5) in 2014 (n=110) and 2015 (n=98) in ARBC using a randomized complete block (RCB) design with 3 replications. Two hundred and eighty-six groundnut genotypes representing worldwide diversity from the ICRISAT core collection were evaluated at R5 at the Ukulima Root Biology Center (URBC) Limpopo Province, South Africa in 2014. This station is located at 24˚33’00S, 28˚07’25E, 1235 masl, and has a deep loamy sand (Clovelly Plinthic) and 24˚C mean growing season temperature. A chickpea diversity panel (n=172) was evaluated in 2013 at URBC. One hundred and eighty-eight genotypes assembled by the cowpea research team at UC Riverside representing global diversity were evaluated at URBC 2012 and 2013 using a 4 block RCB design. Forty-one parents from the SoyNAM panel were planted and evaluated at the ARBC site at R5 for 10 root architectural parameters in 2013. Thirty-six desert adapted legumes obtained from Native Seed Search, including Phaseolus maculatus, Phaseolus acutifolius (including spp tenufolius and latifolius) were evaluated the same year. Plot size at URBC and ARBC was 2 rows by 3 meters and the most representative 3 plants out of 6 excavated were phenotyped except in the case of chickpea where only one plant per plot was phenotyped. Phaseolus species and soybean were planted at 10cm spacing in row and 76cm between row while cowpea was planted at 30cm in row and 76cm between rows.

! 41! Single genotypes of minor legumes including pigeon pea (Cajanus cajan), macuna (Macuna pruriens), mung bean (Vigna radiata), Jack bean (Canavalia ensiformis), lablab (Lablab pupureus), winged bean (Psophocarpus tetragonolobus), rice bean (Vigna umbellata), fish bean (Tephrosia vogelii) and faba bean (Vicia faba) were evaluated in Pennsylvania 2012 in single row plots at a spacing of 20cm in row and 60cm between rows. The Pennsylvania research site is located at 40˚49’14N, 77˚52’08W, 316masl and has Hagerstown silt loam (fine, mixed, mesic, Typic Hapludalf) and 21.5˚C mean growing season temperature. All trials were irrigated, fertilized and had agro-chemicals applied to provide non-limiting growing conditions. The shovelomics protocol described by Burridge et. al. 2016 was used to quantify the root system architecture of all species with some modifications for particular species Number of basal roots or basal root growth angle was not observable for hypogeal germinators which do not have basal roots. When appropriate, counts of epicotyl roots were taken separately from primary root laterals and the dominant angle of epicotyl roots was taken to compare to basal root growth angle of epigeal germinators. To facilitate comparisons of RSA across legume species we developed a simple RSA comparison framework based on ratios of one phene to another. These ratios are also used to control for allometric relationships between plant size and number and size of roots. Ratios between stem and taproot diameter were used for estimating relative investment in root classes and in exploration of deep and shallow zones. Having a smaller ratio between stem and taproot diameter indicates more resources being allocated to the tap root. Ratios between traits on linked organs such as hypocotyl root number (HRN) to stem diameter (SD) or basal root number (BRN) to tap root diameter (TD5) control for the situation that a larger stem may support more HRN and allow comparisons between species of different sizes. Deep and shallow scores were calculated from shovelomics data and allow comparisons across species and evaluation of tradeoffs. The untransformed mean plot values were used for analyses. All statistical analysis and plotting were performed using base R and ggplot2 (R Core Team, 2014). Pearson correlation coefficients and simple linear models (x~y) were used to evaluate tradeoffs between metrics. The Tukey test was used to compare means and variance. Throughout this analysis significance of P values are indicated by * for 0.01, ** for 0.001 and *** when the P value is less than 0.000. Results We focused analysis on the five species for which the greatest amount of data is available, common bean, chickpea, cowpea, soy, groundnut and tepary. Glycine max, the Vigna species and Phaseolus species have epigeal germination and identifiable basal roots although the transition from hypocotyl borne to basal root can be difficult to identity in soybean. The hypogeal species, faba bean, chickpea and groundnut did not have observable basal roots when excavated from the field but groundnut lateral root growth angle was recorded as basal root growth angle (BRGA). Groundnut and chickpea data are excluded from some of the comparisons because counts of hypocotyl and basal roots could not be made.

! 42! The shape and distribution of the violin plots visually demonstrate that root phenotypes varied among different species. Tukey tests indicated the mean value of individual species were significantly different from others, which may relate to their root deployment strategies. Considering the median, range and the distribution of values for a particular species can be illustrative of trends and strategies and is depicted in violin plots. For BRGA, bean had the broadest and most even distribution of angles from zero to eighty degrees from horizontal (Figure 4). Tepary showed a wide range of BRGA values but was weighted toward steeper angles. Groundnut showed a uniform BRGA distribution but slightly shifted towards shallower angles. Cowpea and soy showed a more limited range of BRGA, with cowpea being steeper than soy. Analysis of other phenes illustrates variable degrees of tradeoffs between investment patterns within and among species. Since stem diameter (SD) has been shown to be allometric with plant size (Burridge et al., 2016) we looked at the tap root diameter to stem diameter ratio (TD5 to SD) and found bean, groundnut and cowpea had the broadest ranges (Figure 5). Based on the mean of the distributions, groundnut, cowpea and tepary favored greater investment in primary root relative to stem diameter. Bean had a distribution shifted towards reduced TD. The tendency for cowpea and tepary to favor primary root investment over investment in other root classes is further supported by observations of basal root number (BRN) observations (Figure 6). For BRN, cowpea had a wide range but the smallest mean, and tepary had the second smallest mean. Bean had the greatest range and soy had the largest mean BRN. The ratio of basal root number (BRN) to tap root diameter (TD5) was plotted in the interest of accounting for allometry and showed the same trend, with cowpea and tepary investing more in the primary root compared to basal roots. The relationship between primary versus hypocotyl or epicotyl investment was investigated by comparing tap root diameter (TD5) to the number of hypocotyl borne roots (BRN plus HRN). Plotting this relationship showed that bean exhibits the strongest tradeoff between hypocotyl and primary root investment with a R value of -0.35, significant at P=0.001, while soybean a slight tradeoff with a R value of -0.18, significant at P=0.001. Groundnut actually had a very slight positive association between primary and epicotyl investment (R=0.12**) suggestive of allometry between root system components. For adventitious root number (ARN), soy had the greatest mean and cowpea has the widest range, followed by soy (Figure 8). Bean and tepary were comparable in range but the tepary distribution was shifted towards fewer ARN. The basal root number to adventitious root number (BRN to ARN) ratio revealed cowpea has strategies contrasting to other grain legumes studies. Cowpea has a much greater range in which some genotypes displayed more adventitious than basal roots while other genotypes displayed more basal roots than adventitious roots (Figure 9). The ratio of ARN to SD was plotted to account for allometry and revealed soy having a very different strategy with a more adventitious roots relative to stem diameter and tepary with the fewest relative number of adventitious roots (Figure 10).

! 43! Observations of Andean and Mesoamerican common bean suggest contrasting strategies between gene pools. The Andean gene pool had a greater range for basal root growth angle (BRGA) (Figure 11), ARN (Figure 12), and BRN, for which it also had a greater mean (Figure 13). The Mesoamerican gene pool had a greater range for tap root diameter (TD5) (Figure 14) and tap to stem diameter ratio (TD5 to SD) (Figure 15) with the mean shifted to larger values than the Andean gene pool. This suggests that the Mesoamerican gene pool favored taproot investment while the Andean gene pool favored basal root investment. Across bean, cowpea, tepary and soy there was a negative relationship between basal root number (BRN) and tap root diameter (TD5) (R = -0.31***), suggesting a tradeoff between investment in tap roots and basal roots (Figure 16). Bean had a more pronounced tradeoff than the other species (Figure 17) and this tradeoff is evident when TD5 is plotted against BRN plus ARN (Figure 18). Groundnut did not show a tradeoff between BRN to TD5 and actually had a slightly positive association between BRN plus ARN relative to TD5 suggesting those traits scale together. Comparing deep and shallow scores indicated all species had a statistically significant tradeoff between deep and shallow exploration (Figures 19-25). However, groundnut and chickpea had very small R values for the relationship between deep and shallow scores that are only slightly significant (R values of -0.13** and -0.13* respectively). Comparisons of cross-sectional area of different root classes showed that all species had greater root cross sectional area than hypocotyl cross sectional area. Tepary however, has the lowest ratio indicating the sum of root diameter is closest to stem diameter. (Fig 26). Comparisons among less intensively phenotyped species are discussed below and are based primarily upon observations of single genotypes. Among the Vigna species Mung bean and yardlong bean appear similar to cowpea in terms of RSA but based upon visual evaluation, V. umbellata shows finer hypocotyl- borne roots. Based on number and fineness of roots, faba bean has an RSA intermediate between chickpea and groundnut. Lablab seemed to have basal roots and could be placed between lima bean and cowpea based on dominance of primary and basal roots. Pigeon pea showed a very strong preference for primary root investment. Phaseolus vulgaris and acutifolius species have similar root architectures except that P. acutifolius does not have basal roots organized into whorls and P. acutifolius tends to have greater allocation to the primary root. Discussion We applied the shovelomics method to a variety of legumes, developed a framework for comparisons across legume species, and found that they can be categorized on an RSA spectrum with functional implications for acquisition of soil resources. Species that have greater dominance of epicotyl or superficial roots are on one end of the RSA spectrum and species with stronger primary root dominance on the other (Figure 2). In the middle are species with basal root dominance but only common bean has basal roots organized into whorls. Along this spectrum we observe two categories that correspond to epigeal or hypogeal germination. One is the basal and primary root-dominated architecture of epigeal species visible in Phaseolus, Vigna and Glycine. The other is the epicotyl-root

! 44! dominated system of hypogeal species visible in Cicer, Vicia and Arachis, which is similar to the herringbone system suggested by Fitter (Fitter et al., 1991). Results from across species suggest common tendencies involving opportunity costs among investment in individual root classes and tradeoffs between deep and shallow exploration. The observation that tradeoffs and opportunity costs exist has relevance for breeding efforts involving trait-based selection. Results suggest that investment in one class of the root system limits resources available to another. This tradeoff is evident in negative relationships between metrics estimating investment in primary and hypocotyl or epicotyl-borne root systems, such as between BRN and TD5 and the TD5 to SD ratio. Tradeoffs between deep and shallow scores indicate that most species and most genotypes cannot extensively explore both deep and shallow soil zones. The existence of the opportunity cost between deep and shallow exploration may lead to specialization in one type of single stress environment environment. A genotype native to a terminal drought environment may develop a deeper root system while one native to a low fertility environment may develop a shallower root system. These strategies may be paired with phenology such that a drought avoiding deep root system may be paired with a shorter phenology and a shallower immobile resource scavenging system may be paired with a longer phenology. However, a drought avoidance and short phenology strategy would incur significant tradeoffs in a low fertility environment and the low fertility adapted root system and phenology would underperform in a higher fertility but terminal drought environment. These tradeoffs or opportunity costs between drought and low fertility tolerance may be accentuated by domestication simultaneously introducing novel constraint combinations by extending the zones where plants are cultivated and limiting their available adaptive mechanisms. While we have greatly extended their range on both meso scale (i.e. from isolated riparian zones for bean or a savanna edge for cowpea) and macro scales (i.e. cultivation on multiple continents), domestication has also stripped important reproductive strategies from wild plants, including biphasic or perennial growth, and extended the duration of flowering. Additionally, it is possible that early domestication environments did not have as many contrasting limitations as those experienced in modern agroecosystems. In order to optimize production and increase resiliency, plant breeders should consider how root system architecture can be used to co-optimize acquisition of edaphic resources. One avenue to co-optimize edaphic resource acquisition is through a dimorphic root system architecture that reduces tradeoffs and opportunity costs between deep and shallow exploration. We recognize various types of dimorphic root systems. Common bean becomes dimorphic through strategic investment in different root classes including adventitious, basal, and primary roots, by increasing whorl number and by modulating the growth angle of basal roots (Bonser et al., 1996; Lynch and Brown, 2012; Miguel et al., 2013; Miller et al., 2003b; Walk et al., 2006). Basal root number (BRN) and basal root growth angle (BRGA) in large part determine the range of soil exploration for common bean. Of the legumes studied, bean has the greatest range of BRGA, from horizontal to 80 degrees from horizontal. Plasticity of BRGA in a single plant and in a stand may be

! 45! another mechanism for achieving a dimorphic root system. This may occur as intrinsic variation or dynamic plasticity among the four to twelve individual basal roots on a plant. Deploying roots at multiple angles on a single plant increases the range of soil explored and increases the plant’s ability to acquire resources from both deep and shallow soil. Increasing the responsiveness of BRGA to the environment may be advantageous for two reasons. For one, interplant competition may be reduced and secondly the increased range and plasticity of deployment reduces the risk of failing to encounter edaphic resources. Bean has a large range in the basal root number (BRN) to tap root diameter (TD5) ratio (Figure 7)suggesting high genetic diversity for basal and tap root system investment. Its greater mean for the BRN to TD5 ratio suggests that bean favors investment in BRN over taproots. A negative correlation between BRN and TD5 (Figure 17) and the most negative correlation between deep and shallow scores further support the existence of this tradeoff. However, significant variation also exists for number of adventitious roots and strength of the primary root, which could combine to form various dimorphic root systems. Adventitious roots are relatively cheaper than basal roots due to their lower construction costs but investment in adventitious reduces resources available for basal roots (Miller et al., 2003b; Walk et al., 2006). The utility of adventitious roots for phosphorus acquisition from shallow soil is diminished in a terminal drought environment when shallow soil no longer has adequate soil moisture. As soil dries and tortuosity of the soil diffusion pathway increases, nutrients become less available and horizontally deployed adventitious roots have diminishing returns due to limited nutrient availability in drying topsoil. Pairing a strong adventitious root system with a strong tap root system is one option for a dimorphic root system (Fig 27). This pairing would optimize P acquisition from shallow soil early and then minimize tradeoffs as the tap root compensates for the reduced utility of adventitious roots as soil dries from the top down. Another dimorphic phenotype would involve increasing RSA plasticity of basal root growth angle and lateral branching density both spatially and temporally. Enhancing the plant’s ability to respond to its environment may not always be beneficial as transient environmental cues may stimulate changes that are disadvantageous in the long term. Attempts to optimize the root system of any plant should be tempered by a historical awareness of the differences between wild and domesticated gene pools. Domesticated common bean is more phosphorus efficient than wild common bean, suggesting humans have already selected for traits related to improved phosphorus acquisition (Beebe et al., 1997) such as greater BRWN and longer root hairs, which tend to have higher values in cultivated varieties (Miguel et al., 2015, 2013). Cowpea has a large range for BRN to TD5 ratio suggesting large genetic diversity and ability of the species to adapt to many environments (Figure 7). However, the TD5 to SD ratio (Figure 5) and the BRN count (Figure 6) indicate cowpea and tepary favor taproot investment while bean and soy invest more in basal roots. This apparent preference for deep soil exploration may be related to adaptation to environments where water is the most limiting resource, and this water is typically more available deeper in the soil profile. If this phene were not

! 46! complemented by others, it would incur significant tradeoffs in a well-watered but nutrient-limited environment. An example of a phene that reduces tradeoffs when paired with the taproot dominated root architecture of cowpea is its potential for many hypocotyl-borne laterals. The less negative correlation between deep and shallow scores also suggests that cowpea mitigates tradeoffs between deep and shallow exploration. Cowpea adventitious roots emerge in the relatively P- enriched topsoil and tend to grow an angle greater than 45 degrees from horizontal. Many ARN but steep adventitious root growth angle (ARGA) may facilitate both P and water acquisition, meaning the hypocotyl borne roots do not incur diminishing returns as soil moisture recedes as do the more horizontal hypocotyl roots of common bean. A different type of dimorphic cowpea root system may pair many shallow 1st order laterals in the upper 5cm of soil with a strong and deep taproot. While tepary has a relatively strong taproot and generally steeper BRGA it has the potential to respond to favorable water conditions by increasing investment in adventitious roots. Both tepary and bean modulate BRGA much more than cowpea and soy, suggesting they use this mechanism to adapt to a given environment. They also have the most negative relationships between TD5 and BRN and between deep and shallow scores, indicating strong tradeoffs between root classes and between deep and shallow exploration. Tepary has less genetic variation for greater ARN and shallow BRGA and this may be due to its domestication environment. The alluvial floodplains where tepary may have been domesticated would be more likely to have nutrients homogenously mixed throughout the soil profile. This mixing diminishes the reward of shallow exploration and means that deep root systems could access both nutrients and water. Additionally, in drier environments nutrients do not weather or leach as much and soils are not depleted, so a deeper root system incurs fewer tradeoffs than a shallow root system. For tepary bean production to expand to areas with less extreme water limitation and more heterogeneously distributed nutrients, its root system should be selected to have a crown root architecture more like common bean with more adventitious roots and increased potential for shallow BRGA. Alternatively, selection for long and dense root hairs on the existing tepary root system may be just as effective and simpler for improving adaptation to heterogeneous and nutrient-limited soils (Bates and Lynch, 2001; Ma et al., 2001). Observations of the soy panel indicated that the strengths of basal, adventitious and primary roots are relatively balanced in this species. Soy may rely more heavily on variation of branching density and number of adventitious roots for environmental adaptation. Balancing relative investment in root classes can be considered dimorphic, which enables soy to be responsive to moderate drought or low fertility conditions. While no domestication environment removes all constraints to plant growth, the soy domestication environment may have presented less extreme constraints than other grain legumes studied. The soy root system seems well suited for a range of soil nutrient and soil water availability regimes. To further increase its drought avoidance capacity breeders could consider using a trait based selection strategy to increase allocation to the

! 47! tap root system, like that of tepary. Increasing allocation to the primary root may come at the expense of basal and adventitious roots, but the benefit of increasing primary root depth, water acquisition and drought avoidance should outweigh the cost of shallow nutrient acquisition in well fertilized environments. Delving into differences between the two common bean gene pools offers insights into divergent strategies that may have parallels in the Western and Southern gene-pools in cowpea, the gene pools of lima bean and in the Kabuli and Desi gene pools of chickpea. For both TD5 and the TD5 to SD ratio the Mesoamerican gene pool has a greater range with a higher mean. The Andean gene pool has a greater range in BRN with a higher mean than the Mesoamerican gene pool. The two gene pools are similar in range and distribution for other phenes. This suggests that the Mesoamerican gene pool favors primary root investment while the Andean favors basal root investment (Jochua and Lynch, 2013). The observation that the two gene pools evolved distinct strategies for co-optimizing resource acquisition should inform RSA breeding goals, especially since the Mesoamerican gene pool is generally considered more drought sensitive than the Andean gene pool (Beebe et al., 2013; Polania et al., 2016). Breeders could incorporate a primary root focus to the Andean gene pool and a basal root focus to the Mesoamerican gene pool to increase resiliency within each of the gene pools. The RSA spectrum may be useful for considering how to integrate resource acquisition, use and transport strategies with phenology and overall reproductive strategy. Optimization of resource acquisition in the two different categories (i.e. hypogeal vs. epigeal) of root systems requires different approaches. It also requires a basic understanding of their domestication environments as that affects their life strategy, resource acquisition strategy and rhizoeconomic balance. Species native to very dry environments with ephemeral water availability, such as tepary, may best be served by a root system with few axes and high plasticity. That strategy would maximize rooting depth and minimize investment needed to acquire resources from relatively enriched alluvial soils. The riparian and disturbed edges where wild common bean and lima bean are often found may be best exploited by a basal dominated but highly plastic root system that can respond to a variable and dynamic environment. The epigeal species surveyed in this study have in common the ancestral tendency for viney growth that enables them to climb over neighbors to reach the top of the canopy. This suggest that shoot competition for light and space may have been greater than competition for edaphic resources in their wild and or domestication environments. The native environment of the hypogeal species, groundnut, fava and chickpea, seems to include niche segregation by space or time and may not require aggressive canopy coverage. Thus these species have evolved a less sprawling shoot habit and corresponding root system. Their growth strategy, focused on localized exploitation, could have been favored by slowly mineralizing soil nutrients and deeply available water. An approach to co- optimizing resource acquisition in epigeal species would involve modulation of growth angles and investment among adventitious, basal and taproot classes.

! 48! Improvement of the hypogeal species would involve balancing epicotyl root redundancy with primary root elongation. Groundnut, faba bean and chickpea have a herringbone like root systems formed by a primary root and an epicotyl with numerous axial roots. Groundnut and chickpea have extensive variation in lateral root growth angle and TD5. More importantly, they do not show tradeoffs between deep and shallow exploration. The herringbone root systems of these hypogeal species may be better suited for longer growth periods in low fertility environments with slowly receding soil water, where parsimonious water use conserves soil water until maturation. Species such as chickpea were likely domesticated in a low fertility and terminal drought environment. A successful strategy in this environment could include a longer growth period that would optimize P acquisition and parsimonious water use that would optimize water use efficiency. This strategy may be best complemented by a root system with many axes. Having many axes ensures deployment of root tips to both shallow and deeper soil zones and be a way to co-optimize water and nutrient acquisition. Species with longer duration growth cycles may also be exposed to more biotic stress, which may increase the utility of many root tips. In the case that some roots are lost, others may be able to compensate and negative impacts on overall plant fitness would be minimized. An ideal dimorphic root system for the herringbone type of root architecture could expand their range and increase performance within their existing range. One type of dimorphic root architecture could involve few but shallow epicotyl borne roots with high secondary branching density paired with a strong and deep primary root (Figure 27). This RSA would optimize shallow and deep exploration and limit competition between shallow roots. Another strategy could involve upper epicotyl borne roots with a shallow growth angle paired with steeper primary root laterals. We suggest phenotypes able to deploy roots to both deep and shallow soil would increase adaptation to combined drought and low fertility environments. Species all along the spectrum could be improved by increasing root hair length and density. Increasing root hair length, density and P mobilizing exudates may be the most easily accessible and implementable breeding improvement that would increase resilience to low fertility soils with minimal costs and tradeoffs. Matching a root phenotype to an environment is complicated by two additional factors. Firstly, multiple effective strategies for responding to a given set of environmental conditions may be employed by the same species. Secondly, genetic variation and plastic responses of individual plants dynamically modulating root phenes or even deploying multiple root growth angles on a single plant may be positively related to resiliency. This resiliency borne of genetic variation and plasticity may be useful both in competition avoidance and individual plant fitness optimization. We have identified examples of dimorphic root architecture and efficient exploration of deep and shallow in five legume species (photo panel 2). These are the combinations physiologists and breeding programs should consider in conjunction with use efficiency traits to make grain legumes more productive and resilient in multiple constraint environments.

! 49! References Abbo,!S.,!Berger,!J.,!Turner,!N.C.,!2003a.!Evolution!of!cultivated!chickpea:!four! bottlenecks!limit!diversity!and!constrain!adaptation.!Funct.!Plant!Biol.!30,!1081.! doi:10.1071/FP03084! Abbo,!S.,!LevPYadun,!S.,!Gopher,!A.,!2010.!Agricultural!Origins:!Centers!and! Noncenters;!A!Near!Eastern!Reappraisal.!CRC.!Crit.!Rev.!Plant!Sci.!29,!317–328.! doi:10.1080/07352689.2010.502823! Abbo,!S.,!Shtienberg,!D.,!Lichtenzveig,!J.,!LevPYadun,!S.,!Gopher,!A.,!2003b.!The! Chickpea,!Summer!Cropping,!and!a!New!Model!for!Pulse!Domestication!in!the! Ancient!Near!East.!Q.!Rev.!Biol.!78,!435–448.!doi:10.1086/378927! AnduezaPNoh,!R.H.,!SerranoPSerrano,!M.L.,!Chacón!Sánchez,!M.I.,!del!Pino,!I.,! CamachoPPérez,!L.,!CoelloPCoello,!J.,!Mijangos!Cortes,!J.,!Debouck,!D.G.,!MartínezP Castillo,!J.,!2013.!Multiple!domestications!of!the!Mesoamerican!gene!pool!of! lima!bean!(Phaseolus!lunatus!L.):!evidence!from!chloroplast!DNA!sequences.! Genet.!Resour.!Crop!Evol.!60,!1069–1086.!doi:10.1007/s10722P012P9904P9! Anfodillo,!T.,!Petit,!G.,!Sterck,!F.,!Lechthaler,!S.,!Olson,!M.E.,!2016.!Allometric! Trajectories!and!“Stress”:!A!Quantitative!Approach.!Front.!Plant!Sci.!7,!1681.! doi:10.3389/fpls.2016.01681! Bates,!T.R.,!Lynch,!J.P.,!2001.!Root!hairs!confer!a!competitive!advantage!under!low! phosphorus!availability.!Plant!Soil!236,!243–250.! doi:10.1023/A:1012791706800! Beebe,!S.,!Lynch,!J.,!Galwey,!N.,!Tohme,!J.,!Ochoa,!I.,!1997.!A!geographical!approach!to! identify!phosphorusPefficient!genotypes!among!landraces!and!wild!ancestors!of! common!bean.!Euphytica!95,!325–336.!doi:10.1023/A:1003008617829! Beebe,!S.,!Rao,!I.M.,!Blair,!M.W.,!Butare,!L.,!2009.!Breeding!for!abiotic!stress!tolerance! in!common!bean:!present!and!future!challenges.!Aust.!Plant!Breeding;!SABRAO! Conf.!(14;!11;!2009,!Cairns,!Queensland,!Aust.!Proceedings.!Glob.!Partnersh.! Initiat.!Plant!Breed.!Capacit.!Build.!(GIPB),!Queensland,!AU.!11!p.! Beebe,!S.E.,!Rao,!I.M.,!Blair,!M.W.,!AcostaPGallegos,!J.A.,!2013.!Phenotyping!common! beans!for!adaptation!to!drought.!Front.!Physiol.!4,!35.! doi:10.3389/fphys.2013.00035! Beebe,!S.E.,!Rao,!I.M.,!Cajiao,!C.,!Grajales,!M.,!2008.!Selection!for!Drought!Resistance! in!Common!Bean!Also!Improves!Yield!in!Phosphorus!Limited!and!Favorable! Environments.!Crop!Sci.!48,!582.!doi:10.2135/cropsci2007.07.0404!

! 50! Beebe,!S.E.,!Rao,!I.M.,!Devi,!M.J.,!Polania,!J.,!2014.!Common!beans,!biodiversity,!and! multiple!stresses:!challenges!of!drought!resistance!in!tropical!soils.!Crop! Pasture!Sci.!65,!667–675.! Belko,!N.,!Cisse,!N.,!Diop,!N.N.,!Zombre,!G.,!Thiaw,!S.,!Muranaka,!S.,!Ehlers,!J.D.,!2014.! Selection!for!Postflowering!Drought!Resistance!in!ShortP!and!MediumP!duration! Cowpeas!Using!Stress!Tolerance!Indices.!Crop!Sci.!54,!25–33.! doi:10.2135/cropsci2012.12.0685! Belko,!N.,!ZamanPallah,!M.,!Cisse,!N.,!Ndack!Diop,!N.,!Zombre,!G.,!Ehlers,!J.D.,!Vadez,!V.,! 2012.!Lower!soil!moisture!threshold!for!transpiration!decline!under!water! deficit!correlates!with!lower!canopy!conductance!and!higher!transpiration! efficiency!in!droughtPtolerant!cowpea.!Funct.!Plant!Biol.!39,!306–322.! doi:10.1071/FP11282! Bellucci,!E.,!Bitocchi,!E.,!Rau,!D.,!Rodriguez,!M.,!Biagetti,!E.,!Giardini,!A.,!Attene,!G.,! Nanni,!L.,!Papa,!R.,!2014.!Genomics!of!Origin,!Domestication!and!Evolution!of! Phaseolus!vulgaris,!in:!Tuberosa,!R.,!Graner,!A.,!Frison,!E.!(Eds.),!Genomics!of! Plant!Genetic!Resources.!Springer!Dordrect,!Dordrecht,!pp.!483–507.! doi:10.1007/978P94P007P7572P5_20! Bertioli,!D.J.,!Seijo,!G.,!Freitas,!F.O.,!Valls,!J.F.M.,!LealPBertioli,!S.C.M.,!Moretzsohn,!M.C.,! 2011.!An!overview!of!peanut!and!its!wild!relatives.!Plant!Genet.!Resour.!9,!134– 149.!doi:10.1017/S1479262110000444! Bitocchi,!E.,!Bellucci,!E.,!Giardini,!A.,!Rau,!D.,!Rodriguez,!M.,!Biagetti,!E.,!Santilocchi,!R.,! Zeuli,!P.S.,!Gioia,!T.,!Logozzo,!G.,!Attene,!G.,!Nanni,!L.,!Papa,!R.,!2012.!Molecular! analysis!of!the!parallel!domestication!of!the!common!bean!(!Phaseolus!vulgaris! )!in!Mesoamerica!and!the!Andes.!New!Phytol.!197,!300–313.! doi:10.1111/j.1469P8137.2012.04377.x! Bitocchi,!E.,!Rau,!D.,!Bellucci,!E.,!Rodriguez,!M.,!Murgia,!M.L.,!Gioia,!T.,!Santo,!D.,! Nanni,!L.,!Attene,!G.,!Papa,!R.,!2017.!Beans!(!Phaseolus!ssp!.)!as!a!Model!for! Understanding!Crop!Evolution.!Front.!Plant!Sci.!8,!1–21.! doi:10.3389/fpls.2017.00722! Bloom,!A.,!Chapin,!F.,!Mooney,!H.,!1985.!Resource!limitation!in!plantsPPan!economic! analogy.!Annu.!Rev.!Ecol.!Syst.!16,!363–392.! doi:10.1146/annurev.ecolsys.16.1.363! Bonser,!A.M.,!Lynch,!J.,!Snapp,!S.,!1996.!Effect!of!Phosphorus!Deficiency!on!Growth! Angle!of!Basal!Roots!in!Phaseolus!vulgaris.!New!Phytol.!132,!281–288.! doi:10.1111/j.1469P8137.1996.tb01847.x!

! 51! Boukar,!O.,!Fatokun,!C.A.,!Huynh,!B.,!Roberts,!P.A.,!Close,!T.J.,!2016.!Genomic!Tools!in! Cowpea!Breeding!Programs :!Status!and!Perspectives.!Front.!Plant!Sci.!7,!1–13.! doi:10.3389/fpls.2016.00757! Brown,!C.H.,!Clement,!C.R.,!Epps,!P.,!Luedeling,!E.,!Wichmann,!S.,!2014.!The! Paleobiolinguistics!of!the!Common!Bean!(Phaseolus!vulgaris!L.).!Ethnobiol.!Lett.! 5,!104–115.!doi:10.14237/ebl.5.2014.203.105! Bullock,!J.M.,!DhanjalPAdams,!K.L.,!Milne,!A.,!Oliver,!T.H.,!Todman,!L.C.,!Whitmore,! A.P.,!Pywell,!R.F.,!2017.!Resilience!and!food!security:!rethinking!an!ecological! concept.!J.!Ecol.!105,!880–884.!doi:10.1111/1365P2745.12791! Burridge,!J.D.,!Jochua,!C.N.,!Bucksch,!A.,!Lynch,!J.P.,!2016a.!Legume!shovelomics:! High—Throughput!phenotyping!of!common!bean!(Phaseolus!vulgaris!L.)!and! cowpea!(Vigna!unguiculata!subsp,!unguiculata)!root!architecture!in!the!field.!F.! Crop.!Res.!192,!21–32.!doi:http://dx.doi.org/10.1016/j.fcr.2016.04.008.! Burridge,!J.D.,!Schneider,!H.M.,!Huynh,!B.PL.,!Roberts,!P.A.,!Bucksch,!A.,!Lynch,!J.P.,! 2016b.!GenomePwide!association!mapping!and!agronomic!impact!of!cowpea! root!architecture.!Theor.!Appl.!Genet.!1–13.! Chimungu,!J.G.,!Brown,!K.M.,!Lynch,!J.P.,!2014.!Reduced!Root!Cortical!Cell!File! Number!Improves!Drought!Tolerance!in!Maize.!Plant!Physiol.!166,!1943!LPP 1955.! Comas,!L.H.,!Becker,!S.R.,!Cruz,!V.M.!V,!Byrne,!P.F.,!Dierig,!D.!a,!2013.!Root!traits! contributing!to!plant!productivity!under!drought.!Front.!Plant!Sci.!4,!442.! doi:10.3389/fpls.2013.00442! D’Andrea,!A.C.,!Kahlheber,!S.,!Logan,!A.L.,!Watson,!D.J.,!2017.!Early!domesticated! cowpea!(Vigna!unguiculata)!from!Central!Ghana.!Antiquity!81,!686–698.! Fenta,!A.B.,!Beebe,!E.S.,!Kunert,!J.K.,!Burridge,!D.J.,!Barlow,!M.K.,!Lynch,!P.J.,!Foyer,! H.C.,!2014.!Field!Phenotyping!of!Soybean!Roots!for!Drought!!Stress!Tolerance.! Agron.!.!doi:10.3390/agronomy4030418! Fitter,!A.,!Stickland,!T.,!1991.!Architectural!analysis!of!plant!root!systems.!2.! Influence!of!nutrient!supply!on!architecture!in!contrasting!plant!species.!New! Phytol.!118,!383–389.!doi:10.1111/j.1469P8137.1991.tb00019.x! Fitter,!A.H.,!Stickland,!T.R.,!Harvey,!M.L.,!Wilson,!G.W.,!1991.!Architectural!analysis!of! plant!root!systems!1.!Architectural!correlates!of!exploitation!efficiency.!New! Phytol.!118,!375–382.!doi:10.1111/j.1469P8137.1991.tb00018.x!

! 52! Fuller,!D.Q.,!2007.!Contrasting!Patterns!in!Crop!Domestication!and!Domestication! Rates:!Recent!Archaeobotanical!Insights!from!the!Old!World.!Ann.!Bot.!100,! 903–924.!doi:10.1093/aob/mcm048! Garvin,!D.F.,!Weeden,!N.F.,!1994.!Isozyme!Evidence!Supporting!a!Single!Geographic! Origin!for!Domesticated!Tepary!Bean.!Crop!Sci.!34,!1390–1395.! doi:10.2135/cropsci1994.0011183X003400050045x! Gaur,!P.M.,!Krishnamurthy,!L.,!Kashiwagi,!J.,!2008.!Improving!DroughtPAvoidance! Root!Traits!in!Chickpea!(!Cicer!arietinum!L!.)!PCurrent!Status!of!Research!at! ICRISAT!11,!3–11.! Gepts,!P.,!2014.!Beans:!Origins!and!Development,!in:!Smith,!C.!(Ed.),!Encyclopedia!of! Global!Archaeology.!Springer!New!York,!New!York,!NY,!pp.!822–827.! doi:10.1007/978P1P4419P0465P2_2169! Gepts,!P.,!Debouck,!D.,!1991.!Origin,!Domestication!and!Evolution!of!the!Common! Bean!(Phaseolus!vulgaris!L.),!in:!Schoonhoven,!A.!van,!Voysest,!O.!(Eds.),! Common!Beans:!Research!for!Crop!Improvement,.!CAB!Inetrnational!and!CIAT,! pp.!7–43.! Graham,!P.H.,!Vance,!C.P.,!2003.!Legumes:!Importance!and!Constraints!to!Greater! Use.!Plant!Physiol.!131,!872–877.! GujariaPverma,!N.,!Ramsay,!L.,!Sharpe,!A.G.,!Sanderson,!L.,!Debouck,!D.G.,!Tar’an,!B.,! Bett,!K.E.,!2016.!GenePbased!SNP!discovery!in!tepary!bean!(Phaseolus! acutifolius)!and!common!bean!(P!.!vulgaris)!for!diversity!analysis!and! comparative!mapping.!BMC!Genomics!17,!1–16.!doi:10.1186/s12864P016P 2499P3! Hammons,!R.O.,!1994.!The!origin!and!history!of!the!groundnut,!in:!Smartt,!J.!(Ed.),! The!Groundnut!Crop.!Chapman!and!Hall,!London,!pp.!24–39.! Harder,!D.K.,!Smartt,!J.,!1995.!Winged!Bean,!in:!Smartt,!J.,!Simmonds,!N.W.!(Eds.),! Evolution!of!Crop!Plants.!Longman!Scientific!and!Technical,!Essex!England.! Ho,!M.D.,!Rosas,!J.C.,!Brown,!K.M.,!Lynch,!J.P.,!2005.!Root!architectural!tradeoffs!for! water!and!phosphorus!acquisition.!Funct.!Plant!Biol.!32,!737–748.! doi:10.1071/FP05043! Hund,!A.,!Ruta,!N.,!Liedgens,!M.,!2009.!Rooting!depth!and!water!use!efficiency!of! tropical!maize!inbred!lines!,!differing!in!drought!tolerance.!Plant!Soil!311–325.! doi:10.1007/s11104P008P9843P6!

! 53! Hunter,!M.C.,!Smith,!R.G.,!Schipanski,!M.E.,!Atwood,!L.W.,!Mortensen,!D.A.,!2017.! Agriculture!in!2050:!Recalibrating!targets!for!sustainable!intensification.! Bioscience!67,!386–391.!doi:10.1093/biosci/bix010! Huynh,!B.,!Close,!T.J.,!Roberts,!P.A.,!Hu,!Z.,!Wanamaker,!S.,!Lucas,!M.R.,!Chiulele,!R.,! Cissé,!N.,!David,!A.,!Hearne,!S.,!Fatokun,!C.,!Diop,!N.N.,!Ehlers,!J.D.,!2013.!Gene! Pools!and!the!Geneteic!Architecture!of!Domesticated!Cowpea.!Plant!Genome!6,! 1–8.!doi:10.3835/plantgenome2013.03.0005! IPCC,!2014.!Climate!change!2014:!Impacts,!adaptation,!and!vulnerabilty.!Clim.! Chang.! Iwata,!A.,!Greenland,!C.M.,!Jackson,!S.A.,!2013.!Cytogenetics!of!Legumes!in!the! Phaseoloid!Clade.!Plant!Genome!6.!doi:10.3835/plantgenome2013.03.0004! Janila,!P.,!Nigam,!S.N.,!Pandey,!M.K.,!Nagesh,!P.,!Varshney,!R.K.,!2013.!Groundnut! improvement:!use!of!genetic!and!genomic!tools.!Front.!Plant!Sci.! doi:10.3389/fpls.2013.00023! Jochua,!C.N.,!Lynch,!J.P.,!2013.!Deploying!root!traits!for!African!bean!breeding.!The! Pennsylvania!State!University.! Jongrungklang,!N.,!Toomsan,!B.,!Vorasoot,!N.,!Jogloy,!S.,!Boote,!K.J.,!Hoogenboom,!G.,! Patanothai,!!a.,!2012.!Classification!of!root!distribution!patterns!and!their! contributions!to!yield!in!peanut!genotypes!under!midPseason!drought!stress.!F.! Crop.!Res.!127,!181–190.!doi:10.1016/j.fcr.2011.11.023! Kaplan,!L.,!1965.!Archeology!and!Domestication!in!American!Phaseolus!(Beans).! Econ.!Bot.!19,!358–368.!doi:10.1007/BF02904806! Kaplan,!L.,!Lynch,!T.F.,!1999.!Phaseolus!(Fabaceae)!in!Archaeology:!AMS.!Econ.!Bot.! 53,!261–272.!doi:10.1007/BF02866636!

Kaplan,!L.,!Lynch,!T.F.,!Smith,!C.E.,!1973.!Early!Cultivated!Beans!(Phaseolus!vulgaris)! from!an!Intermontane!Peruvian!Valley.!Science!(80P.!).!179,!76!LPP77.! Kashiwagi,!J.,!Krishnamurthy,!L.,!Gaur,!P.M.,!Chandra,!S.,!Upadhyaya,!H.D.,!2008.! Estimation!of!gene!effects!of!the!drought!avoidance!root!characteristics!in! chickpea!(C.!arietinum!L.).!F.!Crop.!Res.!105,!64–69.! doi:10.1016/j.fcr.2007.07.007! Khoury,!C.K.,!CastañedaPAlvarez,!N.P.,!Achicanoy,!H.A.,!Sosa,!C.C.,!Bernau,!V.,!Kassa,! M.T.,!Norton,!S.L.,!van!der!Maesen,!L.J.G.,!Upadhyaya,!H.D.,!RamírezPVillegas,!J.,! Jarvis,!A.,!Struik,!P.C.,!2015.!Crop!wild!relatives!of!pigeonpea![Cajanus!cajan!(L.)!

! 54! Millsp.]:!Distributions,!ex!situ!conservation!status,!and!potential!genetic! resources!for!abiotic!stress!tolerance.!Biol.!Conserv.!184,!259–270.! doi:https://doi.org/10.1016/j.biocon.2015.01.032! Kim,!K.!Do,!El!Baidouri,!M.,!Abernathy,!B.,!IwataPOtsubo,!A.,!Chavarro,!C.,!Gonzales,! M.,!Libault,!M.,!Grimwood,!J.,!Jackson,!S.A.,!2015.!A!Comparative!Epigenomic! Analysis!of!PolyploidyPDerived!Genes!in!Soybean!and!Common!Bean.!Plant! Physiol.!168,!1433!LPP1447.! Kongjaimun,!A.,!Kaga,!A.,!Tomooka,!N.,!Somta,!P.,!Vaughan,!D.A.,!Srinives,!P.,!2012.! The!genetics!of!domestication!of!yardlong!bean,!Vigna!unguiculata!(L.)!Walp.! ssp.!unguiculata!cv.Pgr.!sesquipedalis.!Ann.!Bot.!109,!1185–1200.! doi:10.1093/aob/mcs048! Koolachart,!R.,!Jogloy,!S.,!Vorasoot,!N.,!Wongkaew,!S.,!Holbrook,!C.C.,!Jongrungklang,! N.,!Kesmala,!T.,!Patanothai,!!a.,!2013.!Rooting!traits!of!peanut!genotypes!with! different!yield!responses!to!terminal!drought.!F.!Crop.!Res.!149,!366–378.! doi:10.1016/j.fcr.2013.05.024! Liao,!H.,!Xiaolong,!Y.,!Rubio,!G.,!Beebe,!S.,!Blair,!M.,!Lynch,!J.,!2004.!Genetic!mapping! of!basal!root!gravitropism!and!phosphorus!acquisition!efficiency!in!common! bean.!Funct.!Plant!Biol.!91,!959–970.!doi:10.1071/FP03255! Lobell,!D.B.,!Burke,!M.B.,!Tebaldi,!C.,!Mastrandrea,!M.D.,!Falcon,!W.P.,!Naylor,!R.L.,! 2008.!Prioritizing!Climate!Change!Adaptation!Needs!for!Food!Security!in!2030.! Science!(80P.!).!319,!607–610.! Lobell,!D.B.,!Schlenker,!W.,!CostaPRoberts,!J.,!2011.!Climate!Trends!and!Global!Crop! Production!Since!1980.!Science!(80P.!).!333,!616!LPP620.! Lopes,!M.,!Reynolds,!M.,!2010.!Partitioning!of!assimilates!to!deeper!roots!is! associated!with!cooler!canopies!and!increased!yield!under!drought!in!wheat.! Funct.!Plant!Biol.!147–156.! Lucas,!M.R.,!Diop,!N.PN.,!Wanamaker,!S.,!Ehlers,!J.D.,!Roberts,!P.A.,!Close,!T.J.,!2011.! Cowpea–Soybean!Synteny!Clarified!through!an!Improved!Genetic!Map.!Plant! Genome!4,!218–225.!doi:10.3835/plantgenome2011.06.0019! Lynch,!J.P.,!2015.!Root!phenes!that!reduce!the!metabolic!costs!of!soil!exploration:! ppportunities!for!21st!century!agriculture.!Plant,!Cell!Environ.!38,!1775–1784.! doi:10.1111/pce.12451! Lynch,!J.P.,!2007a.!Roots!of!the!second!green!revolution.!Aust!J!Bot!55.! doi:10.1071/BT06118!

! 55! Lynch,!J.P.,!2007b.!Rhizoeconomics :!The!Roots!of!Shoot!Growth!Limitations.! HortScience!42,!1107–1109.! Lynch,!J.P.,!Brown,!K.M.,!2012.!New!roots!for!agriculture:!exploiting!the!root! phenome.!Philos.!Trans.!R.!Soc.!Lond.!B.!Biol.!Sci.!367,!1598–1604.! doi:10.1098/rstb.2011.0243! Lynch,!J.P.,!Brown,!K.M.,!2001.!Topsoil!foraging!–!an!architectural!adaptation!of! plants!to!low!phosphorus!availability.!Plant!Soil!237,!225–237.! doi:10.1023/A:1013324727040! Lynch,!J.P.,!Ho,!M.D.,!Phosphorus,!L.,!2005.!Rhizoeconomics:!Carbon!costs!of! phosphorus!acquisition.!Plant!Soil!269,!45–56.!doi:10.1007/s11104P004P1096P 4! Lynch,!J.P.,!Wojciechowski,!T.,!2015.!Opportunities!and!challenges!in!the!subsoil:! pathways!to!deeper!rooted!crops.!J.!Exp.!Bot.!66,!2199–2210.! doi:10.1093/jxb/eru508! Ma,!Z.,!Walk,!T.C.,!Marcus,!A.,!Lynch,!J.P.,!2001.!Morphological!synergism!in!root!hair! length!,!density!,!initiation!and!geometry!for!phosphorus!acquisition!in! Arabidopsis!thaliana :!A!modeling!approach.!Simulation!221–235.! Mallikarjuna,!N.,!Senthilvel,!S.,!Hoisington,!D.,!2010.!Development!of!new!sources!of! tetraploid!Arachis!to!broaden!the!genetic!base!of!cultivated!groundnut!(Arachis! hypogaea!L.).!Genet.!Resour.!Crop!Evol.!58,!889–907.!doi:10.1007/s10722P010P 9627P8! Mamidi,!S.,!Rossi,!M.,!Moghaddam,!S.M.,!Annam,!D.,!Lee,!R.,!Papa,!R.,!McClean,!P.E.,! 2013.!Demographic!factors!shaped!diversity!in!the!two!gene!pools!of!wild! common!bean!Phaseolus!vulgaris!L.!Heredity!(Edinb).!110,!267–76.! doi:10.1038/hdy.2012.82! McClean,!P.E.,!Mamidi,!S.,!McConnell,!M.,!Chikara,!S.,!Lee,!R.,!2010.!Synteny!mapping! between!common!bean!and!soybean!reveals!extensive!blocks!of!shared!loci.! BMC!Genomics!11,!184.!doi:10.1186/1471P2164P11P184! Miguel,!M.A.,!Postma,!J.A.,!Lynch,!J.,!2015.!Phene!synergism!between!root!hair!length! and!basal!root!growth!angle!for!phosphorus!acquisition.!Plant!Physiol.!167,! 1430–1439.!doi:10.1104/pp.15.00145! Miguel,!M.A.,!Widrig,!A.,!Vieira,!R.F.,!Brown,!K.M.,!Lynch,!J.P.,!2013.!Basal!root!whorl! number:!a!modulator!of!phosphorus!acquisition!in!common!bean!(Phaseolus! vulgaris).!Ann.!Bot.!112,!973–82.!doi:10.1093/aob/mct164!

! 56! Miller,!C.R.,!Ochoa,!I.,!Nielsen,!K.L.,!Beck,!D.,!Lynch,!J.P.,!2003a.!Genetic!variation!for! adventitious!rooting!in!response!to!low!phosphorus!availability:!Potential! utility!for!phosphorus!acquisition!from!stratified!soils.!Funct.!Plant!Biol.!30,! 973–985.!doi:10.1071/FP03078! Miller,!C.R.,!Ochoa,!I.,!Nielsen,!K.L.,!Beck,!D.,!Lynch,!J.P.,!2003b.!Genetic!variation!for! adventitious!rooting!in!response!to!low!phosphorus!availability:!potential! utility!for!phosphorus!acquisition!from!stratified!soils.!Funct.!Plant!Biol.!30,! 973–985.!doi:10.1071/FP03078! MunosPAmatriain,!M.,!Mirebrahim,!H.,!Xu,!P.,!Wanamaker,!S.,!Luo,!M.,!Alhakami,!H.,! Alpert,!M.,!Atokple,!I.,!Batieno,!B.,!Boukar,!O.,!Bozdag,!S.,!Ndiaga,!C.,!Drabo,!I.,! Ehlers,!J.D.,!Farmer,!A.,!Fatokun,!C.,!Gu,!Y.Q.,!Guo,!Y.,!Huynh,!B.,!Jackson,!S.,!Kusi,! F.,!Layley,!C.,!Lucas,!M.,!Ma,!Y.,!Timko,!M.,!Wu,!J.,!You,!F.,!Barkley,!N.A.,!Roberts,! P.A.,!Lonardi,!S.,!Close,!T.J.,!2017.!Genome!resources!for!climatePresilient! cowpea,!an!essential!crop!for!food!security.!Plant!J.!89,!1042–1054.! doi:10.1111/tpj.13404! Nabhan,!G.P.,!Felger,!R.S.,!1978.!Teparies!in!Southwestern!North!America.!New!York! Bot.!Gard.!Socienty!Econ.!Bot.!3–19.! Nord,!E.!a,!Lynch,!J.P.,!2009.!Plant!phenology:!a!critical!controller!of!soil!resource! acquisition.!J.!Exp.!Bot.!60,!1927–37.!doi:10.1093/jxb/erp018! Ochoa,!I.E.,!Blair,!M.W.,!Lynch,!J.P.,!2006.!QTL!Analysis!of!Adventitious!Root! Formation!in!Common!Bean!under!Contrasting!Phosphorus!Availability.!Crop! Sci.!46,!1609–1621.!doi:10.2135/cropsci2005.12P0446! Polania,!J.,!Rao,!I.M.,!Cajiao,!C.,!Rivera,!M.,!Raatz,!B.,!Beebe,!S.,!2016.!Physiological! traits!associated!with!drought!resistance!in!Andean!and!Mesoamerican! genotypes!of!common!bean!(!Phaseolus!vulgaris!L!.).!Euphytica!210,!17–29.! doi:10.1007/s10681P016P1691P5! Porch,!T.G.,!Beaver,!J.S.,!Debouck,!D.G.,!Jackson,!S.A.,!Kelly,!J.D.,!Dempewolf,!H.,!2013.! Use!of!Wild!Relatives!and!Closely!Related!Species!to!Adapt!Common!Bean!to! Climate!Change.!Agronomy!3,!433–461.!doi:10.3390/agronomy3020433! Pratt,!R.C.,!Nabhan,!G.,!1988.!Evolution!and!Diversity!of!Phaseolus!acutifolius!Genetic! Resources,!in:!Gepts,!P.!(Ed.),!Genetic!Resources!of!Phaseolus!Beans.!Kluwer! Academic!Publishers,!pp.!409–440.! Purugganan,!M.D.,!Fuller,!D.Q.,!2009.!The!nature!of!selection!during!plant! domestication.!Nature!457,!843–8.!doi:10.1038/nature07895!

! 57! Purushothaman,!R.,!Krishnamurthy,!L.,!Upadhyaya,!H.D.,!Vadez,!V.,!Varshney,!R.K.,! 2016.!Field!Crops!Research!Shoot!traits!and!their!relevance!in!terminal!drought! tolerance!of!chickpea!(Cicer!arietinum!L!.).!F.!Crop.!Res.!197,!10–27.! doi:10.1016/j.fcr.2016.07.016! R!Core!Team,!2014.!R:!A!Language!and!Environment!for!Statistical!Computing.! Rao,!I.,!Beebe,!S.,!Polania,!J.,!Ricaurte,!J.,!Cajiao,!C.,!Garcia,!R.,!Rivera,!M.,!2013.!Can! Tepary!Bean!be!a!model!for!improvement!of!drought!resistance!in!Common! Bean?!African!Crop!Sci.!J.!21,!265–281.! RendónPAnaya,!M.,!MonteroPVargas,!J.M.,!SaburidoPÁlvarez,!S.,!Vlasova,!A.,!CapellaP Gutierrez,!S.,!OrdazPOrtiz,!J.J.,!Aguilar,!O.M.,!VianelloPBrondani,!R.P.,!Santalla,!M.,! Delaye,!L.,!Gabaldón,!T.,!Gepts,!P.,!Winkler,!R.,!Guigó,!R.,!DelgadoPSalinas,!A.,! HerreraPEstrella,!A.,!2017.!Genomic!history!of!the!origin!and!domestication!of! common!bean!unveils!its!closest!sister!species.!Genome!Biol.!18,!60.! doi:10.1186/s13059P017P1190P6! Rich,!S.M.,!Wasson,!A.P.,!Richards,!R.A.,!Katore,!T.,!Prashar,!R.,!Chowdhary,!R.,!Saxena,! D.C.,!Mamrutha,!H.M.,!Zwart,!A.,!Misra,!S.C.,!Sai!Prasad,!S.V.,!Chatrath,!R.,! Christopher,!J.,!Watt,!M.,!2016.!Wheats!developed!for!high!yield!on!stored!soil! moisture!have!deep!vigorous!root!systems.!Funct.!Plant!Biol.!43,!173–188.! doi:10.1071/FP15182! Robotham,!O.,!Chapman,!M.,!2017.!Population!genetic!analysis!of!hyacinth!bean! (Lablab!purpureus!(L.)!Sweet,!Leguminosae)!indicates!an!East!African!origin! and!variation!in!drought!tolerance.!Genet.!Resour.!Crop!Evol.!64,!139–148.! Rogers,!E.D.,!Benfey,!P.N.,!2015.!Regulation!of!plant!root!system!architecture:! implications!for!crop!advancement.!Curr.!Opin.!Biotechnol.!32,!93–98.! doi:10.1016/j.copbio.2014.11.015! Schmutz,!J.,!Mcclean,!P.E.,!Mamidi,!S.,!Wu,!G.A.,!Cannon,!S.B.,!Grimwood,!J.,!Jenkins,!J.,! Shu,!S.,!Song,!Q.,!Chavarro,!C.,!TorresPtorres,!M.,!Geffroy,!V.,!Moghaddam,!S.M.,! Gao,!D.,!Abernathy,!B.,!Barry,!K.,!Blair,!M.,!Brick,!M.A.,!Chovatia,!M.,!Gepts,!P.,! Goodstein,!D.M.,!Gonzales,!M.,!Hellsten,!U.,!Hyten,!D.L.,!Jia,!G.,!Kelly,!J.D.,!Kudrna,! D.,!Lee,!R.,!Richard,!M.M.S.,!Miklas,!P.N.,!Osorno,!J.M.,!Rodrigues,!J.,!Thareau,!V.,! Urrea,!C.A.,!Wang,!M.,!Yu,!Y.,!Zhang,!M.,!Wing,!R.A.,!Cregan,!P.B.,!Rokhsar,!D.S.,! Jackson,!S.A.,!2014.!A!reference!genome!for!common!bean!and!genomePwide! analysis!of!dual!domestications.!Nat.!Genet.!46,!707–713.!doi:10.1038/ng.3008! Scott,!J.C.,!2017.!Against!the!Grain,!A!Deep!History!of!the!Earliest!States.!Yale! University!Press.!

! 58! Sedivy,!E.J.,!Wu,!F.,!Hanzawa,!Y.,!2017.!Soybean!domestication :!the!origin!,!genetic! architecture!and!molecular!bases.!New!Phytol.!214,!539–553.! doi:10.1111/nph.14418! Serebrovsky,!A.S.,!1925.!“Somatic!segreation”!in!domestic!fowl.!J.!Genet.!16,!33–42.! SerranoPSerrano,!M.L.,!AnduezaPNoh,!R.H.,!MartínezPCastillo,!J.,!Debouck,!D.G.,! Chacón!S,!M.I.,!2012.!Evolution!and!Domestication!of!Lima!Bean!in!Mexico:! Evidence!from!Ribosomal!DNA.!Crop!Sci.!52,!1698–1712.! doi:10.2135/cropsci2011.12.0642! Singh,!B.B.,!Chambliss,!O.L.,!Sharma,!B.,!1997.!Recent!advances!in!cowpea!breeding,! in:!Singh!BB,!Mohan!Raj!DR,!Dashiell!KE,!J.L.!(Ed.),!Advances!in!Cowpea! Research.!Copublication!of!International!Institute!of!Tropical!Agriculture!(IITA)! and!Japan!International!Research!Center!for!Agricultural!Sciences!(JIRCAS),! Ibadan,!Nigeria,!pp.!30–49.! Smýkal,!P.,!Coyne,!C.J.,!Ambrose,!M.J.,!Maxted,!N.,!Schaefer,!H.,!Blair,!M.W.,!Berger,!J.,! Greene,!S.L.,!Nelson,!M.N.,!Besharat,!N.,!Vymyslický,!T.,!Toker,!C.,!Saxena,!R.K.,! Roorkiwal,!M.,!Pandey,!M.K.,!Hu,!J.,!Li,!Y.H.,!Wang,!L.X.,!Guo,!Y.,!Qiu,!L.J.,!Redden,! R.J.,!Varshney,!R.K.,!2015.!Legume!Crops!Phylogeny!and!Genetic!Diversity!for! Science!and!Breeding.!CRC.!Crit.!Rev.!Plant!Sci.!34,!43–104.! doi:10.1080/07352689.2014.897904! Taub,!D.R.,!Goldberg,!D.,!1996.!Root!system!topology!of!plants!from!habitats! differing!in!soil!resource!availability.!Funct.!Ecol.!10,!258–264.! Trachsel,!S.,!Kaeppler,!S.M.,!Brown,!K.M.,!Lynch,!J.P.,!2011.!Shovelomics :!high! throughput!phenotyping!of!maize!(!Zea!mays!L!.)!root!architecture!in!the!field.! Plant!Soil!341,!75–87.!doi:10.1007/s11104P010P0623P8! Vadez,!V.,!2014.!Root!hydraulics :!The!forgotten!side!of!roots!in!drought!adaptation.! F.!Crop.!Res.!165,!15–24.!doi:10.1016/j.fcr.2014.03.017! Vadez,!V.,!Kholová,!J.,!Yadav,!R.S.,!Hash,!C.T.,!2013.!Small!temporal!differences!in! water!uptake!among!varieties!of!pearl!millet!(Pennisetum!glaucum!(L.)!R.!Br.)! are!critical!for!grain!yield!under!terminal!drought.!Plant!Soil!371,!447–462.! doi:10.1007/s11104P013P1706P0! Vadez,!V.,!Rao,!S.,!Kholova,!J.,!Krishnamurthy,!L.,!Kashiwagi,!J.,!Ratnakumar,!P.,!2008.! Root!research!for!drought!tolerance!in!legumes :!Quo!vadis ?!21,!77–85.! Varshney,!R.K.,!Song,!C.,!Saxena,!R.K.,!Azam,!S.,!Yu,!S.,!Sharpe,!A.G.,!Cannon,!S.,!Baek,!J.,! Rosen,!B.D.,!Tar’an,!B.,!Millan,!T.,!Zhang,!X.,!Ramsay,!L.D.,!Iwata,!A.,!Wang,!Y.,!

! 59! Nelson,!W.,!Farmer,!A.D.,!Gaur,!P.M.,!Soderlund,!C.,!Penmetsa,!R.V.,!Xu,!C.,!Bharti,! A.K.,!He,!W.,!Winter,!P.,!Zhao,!S.,!Hane,!J.K.,!CarrasquillaPGarcia,!N.,!Condie,!J.!a,! Upadhyaya,!H.D.,!Luo,!M.PC.,!Thudi,!M.,!Gowda,!C.L.L.,!Singh,!N.P.,!Lichtenzveig,!J.,! Gali,!K.K.,!Rubio,!J.,!Nadarajan,!N.,!Dolezel,!J.,!Bansal,!K.C.,!Xu,!X.,!Edwards,!D.,! Zhang,!G.,!Kahl,!G.,!Gil,!J.,!Singh,!K.B.,!Datta,!S.K.,!Jackson,!S.!a,!Wang,!J.,!Cook,!D.R.,! 2013.!Draft!genome!sequence!of!chickpea!(Cicer!arietinum)!provides!a!resource! for!trait!improvement.!Nat.!Biotechnol.!31,!240–6.!doi:10.1038/nbt.2491! Walk,!T.C.,!Jaramillo,!R.,!Lynch,!J.P.,!2006.!Architectural!Tradeoffs!between! Adventitious!and!Basal!Roots!for!Phosphorus!Acquisition.!Plant!Soil!279,!347– 366.!doi:10.1007/s11104P005P0389P6! Weiner,!J.,!2004.!Allocation,!plasticity!and!allometry!in!plants.!Perspect.!Plant!Ecol.! Evol.!Syst.!6,!207–215.!doi:10.1078/1433P8319P00083! Wortmann,!C.,!Kirkby,!R.,!Eledu,!C.,!Allen,!D.,!1998.!Atlas!of!common!bean!(Phaseolus! vulgaris!L.)!production!in!Africa,!Centro!Internationcal!de!Agricultura!Tropical.! CIAT!publication,!no!297,!Cali,!Colombia.! Yan,!X.,!Liao,!H.,!Beebe,!S.,!Blair,!M.,!Lynch,!J.,!2004.!QTL!mapping!of!root!hair!and! acid!exudation!traits!and!their!relationship!to!phosphorus!uptake!in!common! bean.!Plant!Soil!265,!17–29.!doi:10.1007/s11104P005P0693P1! York,!L.M.,!Nord,!E.!a,!Lynch,!J.P.,!2013.!Integration!of!root!phenes!for!soil!resource! acquisition.!Front.!Plant!Sci.!4.!doi:10.3389/fpls.2013.00355! Zhu,!J.,!Brown,!K.M.,!Lynch,!J.P.,!2010.!Root!cortical!aerenchyma!improves!the! drought!tolerance!of!maize!(Zea!mays!L.).!Plant.!Cell!Environ.!33,!740–9.! doi:10.1111/j.1365P3040.2009.02099.x! Zobel,!R.,!Waisel,!Y.,!2010.!A!plant!root!system!architectural!taxonomy:!a!framework! for!root!nomenclature.!Plant!Biosyst.! Zobel,!R.W.,!2011.!A!Developmental!Genetic!Basis!for!Defining!Root!Classes.!Crop! Sci.!51,!1410–1413.!doi:10.2135/cropsci2010.11.0652! Zohary,!D.,!Hopf,!M.,!Weiss,!E.,!2012.!Domestication!of!plants!in!the!Old!World:!the! origin!and!spread!of!domesticated!plants!in!Southwest!Asia,!Europe,!and!the! Mediterranean!Basin,!4th!ed.!Oxford!University!Press,!Oxford,!UK.! Zong,!Y.,!Yao,!S.,!Crawford,!G.W.,!Fang,!H.,!Lang,!J.,!Fan,!J.,!Sun,!Z.,!Liu,!Y.,!Zhang,!J.,! Duan,!X.,!Zhou,!G.,!Xiao,!T.,!Luan,!F.,!Wang,!Q.,!Chen,!X.,!Jiang,!H.,!2017.!Selection! for!Oil!Content!During!Soybean!Domestication!Revealed!by!XPRay!Tomography! of!Ancient!Beans.!Sci.!Rep.!doi:10.1038/srep43595

! 60! Hypogeal# Epigeal# Cotyledons#

epicotyl# hypocotyl#

Seed#placement#

Primary#root#

Basal#root#

A.#hypogaea# P.#acu.folius#

Figure 1. Simplified drawing depicting differences in root architecture between epigeal and hypogeal germinators. Solid vertical line indicates primary root. Dashed lines indicate lateral roots. Basal roots emerge at transition zone between hypocotyl and primary root. Dotted lines on epigeal model indicate hypocotyl (adventitious) roots.

! 61!

Root#System#Architectural#Spectrum# +" Water## availability#

Epicotyl#root# Basal#or#primary#root#dominant# dominant# Epigeal#germina.on# Hypogeal#germina.on#

G.#max# P.#vulgaris# A.#hypogaea# P.#lunatus#

V.#faba# Vigna# C.#arien.num# P.#acu.folius#

#" Uppermost#roots#dominant# Primary#root#dominant#

Figure 2. Simplified drawings of crown root architecture of various legumes arranged on the root system architectural spectrum (x-axis) according to dominant root class and estimated water availability in domestication environment (y-axis). Historical weather data acquired from World Bank 1901- 1930 averages for SE Jalisco, Mexico (P. lunatus), Northern Argentina (A. hypogea), Guerrero, Mexico (P. vulgaris), North-Central Nigeria (Vigna unguiculata), SE Turkey (C. arientinum), East China (G. max), Tucson, Arizona (P. acutifolius), Mt. Carmel, Israel (V. faba). http://sdwebx.worldbank.org/climateportal/index.cfm?page=country_historical_cli mate&ThisCCode=MEX

! 62!

Photo panel 1: Representative images showing species level variation of crown root architecture. Roots were grown in the field, excavated manually and washed with water before photographing. Circular scale has diameter of 25mm in all images.

! 63!

Photo panel 2: Examples of dimorphic root phenotypes. Description with each photo identifies the dimorphic strategic employed. Deployment of roots to both deep and shallow soil may allow dimorphic root architectures to co-optimize edaphic resource acquisition.

! 64!

Figure 3. Violin plot showing distribution of branching density for bean, cowpea, groundnut, and soybean. Branching density was measured by counting the number of 2nd order lateral roots in a 2cm segment on a representative 1st order root (basal root for epigeal germinators). Data was taken from field grown samples using the shovelomics method. Horizontal lines inside violins indicate median, 25% quartile and 75% quartile.

! 65!

Figure 4. Violin plot showing distribution of basal root growth angle (BRGA) for bean, chickpea, cowpea, groundnut, soy and tepary. BRGA is measured in degrees from horizontal. For the hypogeal germinators, that do not have basal roots, the angle of the larger epicotyl roots was measured. Horizontal lines inside violins indicate median, 25% quartile and 75% quartile.

! 66!

Figure 5. Violin plot showing distribution of tap root (TD5) to stem diameter (SD) ratio for bean, cowpea, groundnut, soy, tepary. This ratio is presented as an indicator in relative investment to deep or shallow investment where a higher mean indicates greater preference to the primary root. Horizontal lines inside violins indicate median, 25% quartile and 75% quartile.

! 67!

Figure 6. Violin plot showing distribution of basal root number (BRN) for bean, cowpea, soy and tepary. Horizontal lines inside violins indicate median, 25% quartile and 75% quartile.

! 68!

Figure 7. Violin plot showing distribution of basal root number (BRN) to tap root diameter (TD5) for bean, cowpea, soy and tepary. This ratio is presented as an indicator of investment in hypocotyl (including basal roots) or primary root investment where higher values indicate greater investment in the hypocotyl based classes of roots. Horizontal lines inside violins indicate median, 25% quartile and 75% quartile.

! 69!

Figure 8. Violin plot showing distribution of adventitious or hypocotyl root number (ARN) for bean, cowpea, soy and tepary. ARN is a count of adventitious roots. Horizontal lines inside violins indicate median, 25% quartile and 75% quartile.

! 70!

Figure 9. Violin plot showing distribution of basal root number (BRN) to adventitious root number (ARN) for bean, cowpea, soy and tepary. This metric distinguishes between investments in the two classes of hypocotyl borne roots in epigeal germinators with higher values indicating greater investment in basal roots relative to adventitious roots. Horizontal lines inside violins indicate median, 25% quartile and 75% quartile.

! 71!

Figure 10. Violin plot showing distribution of ratio of adventitious root number (ARN) to stem diameter (SD) for bean, cowpea, soy and tepary. This metric was used to compare investment in the adventitious root system to stem diameter. Higher values indicate greater investment in the adventitious system relative to stem diameter. Horizontal lines inside violins indicate median, 25% quartile and 75% quartile.

! 72!

Figure 11. Violin plot showing distribution of basal root growth angle (BRGA) for the Andean, grown at ARBC in 2016 (B.And.ARBC.2016) and Mesoamerican, grown at ARBC in 2016 (B.Meso.ARBC.2016) common bean gene pools. The mini-diversity panels were composed of 120 and 110 entries, respectively. Horizontal lines inside violins indicate median, 25% quartile and 75% quartile.

! 73!

Figure 12. Violin plot showing distribution of adventitious root number (ARN) for the Andean, grown at ARBC in 2016 (B.And.ARBC.2016) and Mesoamerican, grown at ARBC in 2016 (B.Meso.ARBC.2016) common bean gene pools. The mini-diversity panels were composed of 120 and 110 entries, respectively. Horizontal lines inside violins indicate median, 25% quartile and 75% quartile.

! 74!

Figure 13. Violin plot showing distribution of basal root number (BRN) for the Andean, grown at ARBC in 2016 (B.And.ARBC.2016) and Mesoamerican, grown at ARBC in 2016 (B.Meso.ARBC.2016) common bean gene pools. The mini- diversity panels were composed of 120 and 110 entries, respectively. Horizontal lines inside violins indicate median, 25% quartile and 75% quartile.

! 75!

Figure 14. Violin plot showing distribution of TD5 for the Andean, grown at ARBC in 2016 (B.And.ARBC.2016) and Mesoamerican, grown at ARBC in 2016 (B.Meso.ARBC.2016) common bean gene pools. The mini-diversity panels were composed of 120 and 110 entries, respectively. Horizontal lines inside violins indicate median, 25% quartile and 75% quartile.

! 76!

Figure 15. Violin plot showing distribution of tap diameter (TD5) to stem diameter (SD) for the Andean, grown at ARBC in 2016 (B.And.ARBC.2016) and Mesoamerican, grown at ARBC in 2016 (B.Meso.ARBC.2016) common bean gene pools. The mini-diversity panels were composed of 120 and 110 entries, respectively. Horizontal lines inside violins indicate median, 25% quartile and 75% quartile.

! 77! Tradeoﬀ(between(deep(and(shallow(investment(in( (groundnut,(soy,(tepary(cowpea(and(bean( n(obs( Tradeoﬀ(between(deep(and(shallow(investment(in(tepary(

n(obs(

Tradeoﬀ(between(tap(and(basal(investment(in(soy,(( n(obs( Tradeoﬀ(between(tap(and(hypocotyl(investment(in(bean( n(obs( tepary,(cowpea(and(bean(

Figure 16. Regression between tap root diameter (TD5) and basal root number (BRN) shows a negative relationship (R = -0.31***) for soy, tepary, cowpea and bean data sets. Circle size indicates number of observations at a particular value.

! 78! ! Tradeoﬀ(between(tap(and(basal(investment(in(bean( n(obs( Tradeoﬀ(between(deep(and(shallow(scores(in(bean( n(obs(

Figure 17. Regression between tap root diameter (TD5) and basal root number (BRN) in bean shows a negative relationship between TD5 and BRN (R=-0.40***) and suggests the existence of a tradeoff. Circle size indicates number of observations at a particular value.

! 79! Tradeoﬀ(between(deep(and(shallow(investment(in( (groundnut,(soy,(tepary(cowpea(and(bean( n(obs( Tradeoﬀ(between(deep(and(shallow(investment(in(tepary(

n(obs(

Tradeoﬀ(between(tap(and(basal(investment(in(soy,(( n(obs( Tradeoﬀ(between(tap(and(hypocotyl(investment(in(bean( n(obs( tepary,(cowpea(and(bean(

Figure 18. Regression between tap root diameter (TD5) and basal root number (BRN) plus adventitious root number (ARN) (R= -0.35***) indicates a tradeoff between hypocotyl and primary root investment. Circle size indicates number of observations at a particular value.

! 80! ! Tradeoﬀ(between(deep(and(shallow(investment(in( (groundnut,(soy,(tepary,(cowpea(and(bean( n(obs( Tradeoﬀ(between(deep(and(shallow(investment(in(tepary(

n(obs(

Figure 19.. Regression between deep and shallow scores indicates a tradeoff between deep and shallow exploration in groundnut, soy, tepary, cowpea and bean (R= -0.36***). Deep score calculated using the equation (10*TD5) + BRGA. Tradeoﬀ(between(tap(and(basal(investment(in(soy,(( n(obs( ShallowTradeoﬀ(between(tap(and(hypocotyl(investment(in(bean( score calculated using the equation (10*ARN) +(90-BRGA). Circlen(obs size( tepary,(cowpea(and(bean( indicates number of observations at a particular value.

! 81! Tradeoﬀ(between(tap(and(basal(investment(in(bean( n(obs( Tradeoﬀ(between(deep(and(shallow(scores(in(bean( n(obs(

Figure 20. Regression between deep and shallow scores indicates a tradeoff between deep and shallow exploration in common bean (R= -0.47***). Deep score calculated using the equation (10*TD5) + BRGA. Shallow score calculated using the equation (10*ARN) +(90-BRGA). Circle size indicates number of observations at a particular value.

! 82! Tradeoﬀ(between(tap(and(basal(investment( Tradeoﬀ(between(deep(and(shallow(investment(in(cowpea( in(soy,(tepary,(cowpea(and(bean( n(obs(

n(obs(

! Figure 21. Regression between deep and shallow scores indicates a slight Tradeoﬀ(between(deep(and(shallow(investment(in(groundnut( tradeoffTradeoﬀ(between(deep(and(shallow(investment(in(soy( between deep and shallow exploration in cowpea (R= -0.23***). Deep score calculated using the equation (10*TD5) + BRGA. Shallow score calculated n(obs( using the equation (10*ARN) +(90-BRGA). Circle size indicates number of observations at a particular value.

n(obs(

! 83! Tradeoﬀ(between(tap(and(basal(investment( Tradeoﬀ(between(deep(and(shallow(investment(in(cowpea( in(soy,(tepary,(cowpea(and(bean( n(obs(

n(obs(

Tradeoﬀ(between(deep(and(shallow(investment(in(groundnut( Tradeoﬀ(between(deep(and(shallow(investment(in(soy( n(obs(

n(obs(

Figure 22. Regression between deep and shallow scores indicates a weak tradeoff between deep and shallow exploration in groundnut (R= -0.13***). Deep score calculated using the equation (10*TD5) + BRGA. Shallow score calculated using the equation (10*ARN) +(90-BRGA). Circle size indicates number of observations at a particular value.

! 84! Tradeoﬀ(between(tap(and(basal(investment( Tradeoﬀ(between(deep(and(shallow(investment(in(cowpea( in(soy,(tepary,(cowpea(and(bean( n(obs(

n(obs(

Tradeoﬀ(between(deep(and(shallow(investment(in(groundnut( Tradeoﬀ(between(deep(and(shallow(investment(in(soy( n(obs(

n(obs(

Figure 23. Regression between deep and shallow scores indicates a slight tradeoff between deep and shallow exploration in soy (R= -0.28***). Deep score calculated using the equation (10*TD5) + BRGA. Shallow score calculated using the equation (10*ARN) +(90-BRGA). Circle size indicates number of observations at a particular value.

! 85! Tradeoﬀ(between(deep(and(shallow(investment(in( (groundnut,(soy,(tepary(cowpea(and(bean( n(obs( Tradeoﬀ(between(deep(and(shallow(investment(in(tepary(

n(obs(

Figure 24. Regression between deep and shallow scores indicates a tradeoff between Tradeoﬀ(between(tap(and(basal(investment(in(soy,(( deep and shallow exploration in tepary (R= -0.46***). Deep score calculated using the equation (10*TD5) + BRGA. Shallow score calculated n(usingobs( Tradeoﬀ(between(tap(and(hypocotyl(investment(in(bean( n(obs( the equation (10*ARN) +(90tepary,(cowpea(and(bean(-BRGA). Circle size indicates number of observations at a particular value.

! 86!

Figure 25. Regression between deep and shallow scores in chickpea (R= -0.13*). Deep score calculated using the equation (10*TD5) + BRGA. Shallow score calculated using the equation (10*ARN) +(90-BRGA).

! 87!

Figure 26. Violin plot showing distribution of sum of root cross sectional area relative to hypocotyl cross sectional area. Data acquired using the line tool in ImageJ from a subset of 120 images of root crowns. Higher values indicate greater root cross sectional area relative to hypocotyl diameter at the soil level.

! 88!

Shallow' EPIGEAL'

Dimorphic'

Deep'

Shallow' HYPOGEAL'

Dimorphic'

Deep'

Figure 27. Simplified drawings depicting examples of deep, shallow and dimorphic root architectural phenotypes of hypogeal and epigeal germinators. Solid vertical line indicates primary root. Vertical dashed line indicates hypocotyl or epicotyl. Other dashed lines indicate lateral roots. Dotted lines on epigeal model indicate adventitious roots. Dotted lines on hypogeal model indicate lateral roots emerging from epicotyl borne roots.

! 89! Table l. Lists of grain legume genotype 44 IT95K 1093 5 names and where and when phenotyped. 45 IT95K 1095 4 46 IT95K 1105 5 47 IT95K 1479 48 IT 95K 1491 List of cowpea genotypes phenotyped 49 IT96D 602 51 IT97K 499 35 2012 and 2013 at URBC.! 52 IT97K 556 6 53 IT97K 819 132 ID# Genotype 54 IT98K 128 2 1 Apagbaala 55 IT98K 317 2 2 Bombay 21 56 IT98K 428 3 3 Cam 7-29 57 IT98K 498 1 4 Cam 12-58 58 IT98K 555 1 5 CB 27 59 IT98K 598 2 6 CB 46 60 IT98K 698 2 7 CC 27 61 IT98K 205 8 8 CC 36 62 IT98K 1111 1 9 CC 85 2 63 IT98D 1399 10 CC 110 2 65 IT99K 124 5 11 CRSP Niebe 66 IT99K 407 8 12 Danila 67 Sh49 10 4 13 Early Scarlet 68 Su Vita 2 14 French Bean 69 Vya 15 Gorum Local 70 Yacine 17 Ife Brown 72 IT93K 693 2 18 Iron Clay 73 IT93K 2046 19 KVx 61 1 74 IT94K 437 1 20 KBx 396 75 IT95M 190 21 KVx 403 76 IT95M 303 22 KVx 421 25 79 UCR 11 23 KVx 525 80 UCR P 24 24 Lori Niebe 81 UCR 41 25 Lyg 321 1 82 UCR 162 26 Marfo Tuya 83 UCR 231 27 Melakh 84 UCR 288 28 Montiero 85 UCR 707 29 Petite n Green 86 UCR 739 30 Mouride 87 UCR 779 31 IT82E 18 88 UCR 830 32 IT83D 442 89 UCR 1340 33 IT84S 2049 90 UCR 1342 36 IT85F 3139 91 UCR 2567 37 IT86D 364 92 UCR 3249 38 IT86D 1010 93 UCR 3301 39 IT89D 288 94 UCR 3310 40 IT90K 284 2 95 UCR 3326 41 IT93K 93 10 96 UCR 3399 42 IT93K 503 1 97 UCR 4539 43 IT95K 1090 2

! 90! 98 UCR 5353 150 TVu 8883 99 UCR 5272 151 TVu 9469 100 UCR 5279 153 TVu 9506 101 UCR 5275 154 TVu 9557 102 UCR 5373 155 TVu 9801 103 UCR 5383 156 TVu 9848 105 UCR 5280 157 TVu 10366 106 UCR 5329 158 TVu 10394 107 UCR 5385 159 TVu 10466 108 TVu 53 160 TVu 12565 109 TVu 264 161 TVu 12710 110 TVu 374 162 TVu 12746 111 TVu 408 163 TVu 12802 112 TVu 415 164 TVu 12923 113 TVu 467 165 TVu 12937 114 TVu 1004 166 TVu 13463 115 TVu 1037 167 TVu 14190 116 TVu 1059 168 TVu 14195 117 TVu 1124 169 TVu 14224 119 TVu 1474 170 TVu 14272 120 TVu 1656 171 TVu 14253 121 TVu 1727 173 TVu 14633 122 TVu 1780 174 TVu 14691 123 TVu 1886 175 TVu 14759 124 TVu 2398 176 TVu 14875 125 TVu 2548 177 TVu 14890 126 TVu 2769 178 TVu 15143 127 TVu 2280 179 TVu 15391 128 TVu 2845 180 TVu 15445 129 TVu 2933 181 TVu 15610 130 TVu 2971 182 TVu 15639 131 TVu 3310 183 TVu 15653 132 TVu 3346 184 TVu 15995 133 TVu 3360 185 TVu 16220 134 TVu 3830 186 TVu 16403 135 TVu 3947 187 TVu 16408 136 TVu 4316 188 TVu 16454 137 TVu 4535 189 TVu 16594 138 TVu 4622 190 24 125B 1 139 TVu 4632 191 1393 1 2 3 (-) 140 TVu 5444 192 524 B 141 TVu 6365 193 58 53 142 TVu 6439 194 58 57 143 TVu 6837 196 IT82E 16 144 TVu 7155 197 IT85F 867 5 146 TVu 7778 1 198 CB 5 147 TVu 8262 200 Sasque 148 TVu 8656 201 TVu 10179 149 TVu 8877 202 IT97K 207 15

! 91! 205 N'Diambour PI507681B NAM 46 206 IT00K 901 6 PI518751 NAM 48 ! PI561370 NAM 50 PI404188A NAM 54 ! PI574486 NAM 64 List!of!Soybean!genotypes!phenotyped!at! ! ARBC!2013! ! ! Parent' test' TN05-3027 NAM 2 4J105-3-4 NAM 3 5M20-2-5-2 NAM 4 CL0J095-4-6 NAM 5 CL0J173-6-8 NAM 6 HS6-3976 NAM 8 Prohio NAM 9 LD00-3309 NAM 10 LD01-5907 NAM 11 LD02-4485 NAM 12 LD02-9050 NAM 13 Magellan NAM 14 Maverick NAM 15 S06-13640 NAM 17 NE3001 NAM 18 Skylla NAM 22 U03-100612 NAM 23 LG03-2979 NAM 24 LG03-3191 NAM 25 LG04-4717 NAM 26 LG05-4292 NAM 27 LG05-4317 NAM 28 LG05-4464 NAM 29 LG05-4832 NAM 30 LG90-2550 NAM 31 LG92-1255 NAM 32 LG94-1128 NAM 33 LG94-1906 NAM 34 LG97-7012 NAM 36 LG98-1605 NAM 37 LG00-3372 NAM 38 LG04-6000 NAM 39 PI398881 NAM 40 PI427136 NAM 41 PI437169B NAM 42

! 92! List!of!groundnut!genotypes!phenotyped! 40' ICG'2031' ARBC!2014! 41' ICG'2106' 42' ICG'2286' genotype' geno.name' 43' ICG'2772' 1' ICG'36' 44' ICG'2773' 2' ICG'76' 45' ICG'2777' 3' ICG'81' 46' ICG'2857' 4' ICG'111' 47' ICG'2925' 5' ICG'115' 48' ICG'3027' 6' ICG'118' 49' ICG'3053' 7' ICG'163' 50' ICG'3102' 8' ICG'188' 51' ICG'3140' 9' ICG'297' 52' ICG'3240' 10' ICG'311' 53' ICG'3312' 11' ICG'332' 54' ICG'3343' 12' ICG'334' 55' ICG'3421' 13' ICG'397' 56' ICG'3584' 14' ICG'405' 57' ICG'3673' 15' ICG'434' 58' ICG'3681' 16' ICG'442' 59' ICG'3746' 17' ICG'513' 60' ICG'3775' 18' ICG'532' 61' ICG'3992' 19' ICG'721' 62' ICG'4111' 20' ICG'862' 63' ICG'4156' 21' ICG'875' 64' ICG'4343' 22' ICG'928' 65' ICG'4389' 23' ICG'1137' 66' ICG'4412' 24' ICG'1142' 67' ICG'4527' 25' ICG'1274' 68' ICG'4538' 26' ICG'1399' 69' ICG'4543' 27' ICG'1415' 70' ICG'4598' 28' ICG'1487' 71' ICG'4670' 29' ICG'1519' 72' ICG'4684' 30' ICG'1534' 73' ICG'4729' 31' ICG'1569' 74' ICG'4746' 32' ICG'1668' 75' ICG'4750' 33' ICG'1699' 76' ICG'4764' 34' ICG'1703' 77' ICG'4798' 35' ICG'1711' 78' ICG'4906' 36' ICG'1823' 79' ICG'4911' 37' ICG'1834' 80' ICG'4955' 38' ICG'1973' 81' ICG'4998' 39' ICG'2019' 82' ICG'5016'

! 93! 83' ICG'5195' 126' ICG'8285' 84' ICG'5221' 127' ICG'8490' 85' ICG'5236' 128' ICG'8517' 86' ICG'5286' 129' ICG'8567' 87' ICG'5327' 130' ICG'8751' 88' ICG'5475' 131' ICG'8760' 89' ICG'5494' 132' ICG'9037' 90' ICG'5609' 133' ICG'9249' 91' ICG'5662' 134' ICG'9315' 92' ICG'5663' 135' ICG'9362' 93' ICG'5745' 136' ICG'9418' 94' ICG'5779' 137' ICG'9449' 95' ICG'5827' 138' ICG'9507' 96' ICG'5891' 139' ICG'9666' 97' ICG'6022' 140' ICG'9777' 98' ICG'6057' 141' ICG'9809' 99' ICG'6201' 142' ICG'9842' 100' ICG'6263' 143' ICG'9905' 101' ICG'6375' 144' ICG'9961' 102' ICG'6394' 145' ICG'9987' 103' ICG'6402' 146' ICG'10010' 104' ICG'6407' 147' ICG'10036' 105' ICG'6643' 148' ICG'10053' 106' ICG'6646' 149' ICG'10092' 107' ICG'6654' 150' ICG'10185' 108' ICG'6667' 151' ICG'10384' 109' ICG'6703' 152' ICG'10474' 110' ICG'6766' 153' ICG'10479' 111' ICG'6813' 154' ICG'10566' 112' ICG'6888' 155' ICG'10701' 113' ICG'7153' 156' ICG'10890' 114' ICG'7181' 157' ICG'10950' 115' ICG'7190' 158' ICG'11088' 116' ICG'7243' 159' ICG'11109' 117' ICG'7867' 160' ICG'11144' 118' ICG'7883' 161' ICG'11219' 119' ICG'7897' 162' ICG'11249' 120' ICG'7906' 163' ICG'11322' 121' ICG'7963' 164' ICG'11386' 122' ICG'7969' 165' ICG'11426' 123' ICG'8083' 166' ICG'11457' 124' ICG'8106' 167' ICG'11515' 125' ICG'8253' 168' ICG'11542'

! 94! 169' ICG'11651' 212' ICG'15233' 170' ICG'11687' 213' ICG'15234' 171' ICG'11855' 214' ICG'15236' 172' ICG'11862' 215' ICG'15287' 173' ICG'12000' 216' ICG'15309' 174' ICG'12189' 217' ICG'15405' 175' ICG'12235' 218' ICG'15415' 176' ICG'12276' 219' ICG'15419' 177' ICG'12370' 220' Gangapuri' 178' ICG'12509' 221' ICGS'44' 179' ICG'12625' 222' ICGV'01232' 180' ICG'12665' 223' ICGV'01276' 181' ICG'12672' 224' ICGV'01328' 182' ICG'12682' 225' ICGV'02022' 183' ICG'12697' 226' ICGV'02038' 184' ICG'12879' 227' ICGV'02148' 185' ICG'12921' 228' ICGV'02189' 186' ICG'12988' 229' ICGV'02194' 187' ICG'12991' 230' ICGV'02266' 188' ICG'13099' 231' ICGV'02271' 189' ICG'13491' 232' ICGV'02286' 190' ICG'13603' 233' ICGV'02290' 191' ICG'13723' 234' ICGV'02446' 192' ICG'13787' 235' ICGV'86011' 193' ICG'13856' 236' ICGV'86325' 194' ICG'13858' 237' ICGV'86326' 195' ICG'13895' 238' ICGV'86590' 196' ICG'13982' 239' ICGV'87160' 197' ICG'14106' 240' ICGV'87354' 198' ICG'14118' 241' ICGV'87378' 199' ICG'14127' 242' ICGV'87921' 200' ICG'14466' 243' ICGV'88145' 201' ICG'14475' 244' ICGV'91116' 202' ICG'14482' 245' ICGV'91278' 203' ICG'14523' 246' ICGV'92206' 204' ICG'14630' 247' ICGV'92234' 205' ICG'14705' 248' ICGV'92267' 206' ICG'14710' 249' ICGV'93470' 207' ICG'14834' 250' ICGV'94169' 208' ICG'14985' 251' ICGV'94361' 209' ICG'15042' 252' ICGV'95377' 210' ICG'15190' 253' ICGV'96466' 211' ICG'15232' 254' ICGV'96468'

! 95! 255' ICGV'97182' ICC'7315' Iran' 256' ICGV'97183' ICC'8195' Pakistan' 257' ICGV'98294' KABALAK' Turkey' 258' ICGV'99001' RPIP'12A069A00159' India' 259' MA13' RPIP'12A069A00468' India' 260' ICG'11605' ICC'6571' India' RPIP'12A069A00559' India' 261' ICG'14788' RPIP'12A069A01483' India' 262' ICGV'90087' RPIP'12A069A01506' India' 263' ICGVASMA86021'' RPIP'12A069A01574' India' 264' ICGVASM'00537' ICC'9137' Iran' 265' ICGVASM'01513' ICC'13187' Iran' 266' ICGVASM'01514' RPIP'12A071A02994' Iran' 267' ICGVASM'01711' ICC'6877' Iran' 268' ICGVASM'01731' ICC'13462' Iran' 269' ICGVASM'02536' RPIP'12A071A06322' Iran' 270' ICGVASM'03516' RPIP'12A069A00785' India' 271' ICGVASM'03519' RPIP'12A069A01390' India' 272' ICGVASM'03543' RPIP'12A071A03046' Iran' 273' ICGVASM'03701' RPIP'12A071A03679' Iran' 274' ICGVASM'05521' RPIP'12A071A03866' Iran' RPIP'12A071A04026' Iran' 275' ICGVASM'05680' RPIP'12A071A05117' Iran' 276' ICGVASM'05701' RPIP'12A071A05153' Iran' 277' ICGVASM'05723' RPIP'12A071A06655' Iran' 278' ICGVASM'05756' RPIP'12A071A07052' Iran' 279' ICGVASM'06516' ICC'10341' India' 280' ICGVASM'06596' ICC'10393' India' 281' ICGVASM'06661' ICC'10399' India' 282' ICGVASM'06684' ICC'1052' India' 283' ICGVASM'07531' ICC'10755' India' 284' ICGVASM'07543' ICC'1083' India' 285' ICGVASM'90704' ICC'10885' India' 286' ICGVASM'95533' ICC'10945' India' 287' ICGVASM'95741' ICC'1098' India' 288' ICGVASM'99541' ICC'11121' India' ICC'11198' India' 289' JL'24' ICC'11284' India' ! ICC'11378' India' List!of!chickpea!genotypes!phenotyped! ICC'11584' India' URBC!2013! ICC'1161' India' ICC'11627' India' PLANTID' COUNTRY' ICC'1164' India' ICC'6279' India' ICC'11664' India' ILC'213' Peru' ICC'11764' India' ICC'6293' Iran' ICC'1180' India'

! 96! ICC'11879' India' ICC'15333' India' ICC'1194' India' ICC'15406' India' ICC'11944' India' ICC'15435' India' ICC'12028' India' ICC'15510' India' ICC'12037' India' ICC'15518' India' ICC'1205' India' ICC'15567' India' ICC'12155' India' ICC'15606' India' ICC'12299' India' ICC'15610' India' ICC'1230' India' ICC'15612' India' ICC'12307' India' ICC'15618' India' ICC'12328' India' ICC'15697' India' ICC'12492' India' ICC'15802' India' ICC'12537' India' ICC'15868' India' ICC'12654' India' ICC'15888' India' ICC'12726' India' ICC'15996' India' ICC'12824' India' ICC'16207' India' ICC'12851' India' ICC'16261' India' ICC'12866' India' ICC'16269' India' ICC'12916' India' ICC'16374' India' ICC'12928' India' ICC'16487' India' ICC'12947' India' ICC'16524' India' ICC'12968' India' ICC'16796' India' ICC'1356' India' ICC'16903' India' ICC'13816' India' ICC'16915' India' ICC'13863' India' ICC'1710' India' ICC'13892' India' ICC'1715' India' ICC'1392' India' ICC'1882' India' ICC'1397' India' ICC'1915' India' ICC'1398' India' ICC'1923' India' ICC'14051' India' ICC'2065' India' ICC'14077' India' ICC'2072' India' ICC'14098' India' ICC'2210' India' ICC'14199' India' ICC'2242' India' ICC'1422' India' ICC'2263' India' ICC'1431' India' ICC'2277' India' ICC'14402' India' ICC'2507' India' ICC'14595' India' ICC'2580' India' ICC'14669' India' ICC'2629' India' ICC'14778' India' ICC'2720' India' ICC'14799' India' ICC'283' India' ICC'14815' India' ICC'2884' India' ICC'14831' India' ICC'2919' India' ICC'11498' India' ICC'2969' India' ICC'1510' India' ICC'2990' India' ICC'15264' India' ICC'3218' India' ICC'15294' India' ICC'3230' India'

! 97! ICC'3325' India' ICC'762' India' ICC'3362' India' ICC'7668' India' ICC'3421' India' ICC'7819' India' ICC'3512' India' ICC'7867' India' ICC'3631' India' ICC'791' India' ICC'3761' India' ICC'8058' India' ICC'3776' India' ICC'8151' India' ICC'3946' India' ICC'8318' India' ICC'4182' India' ICC'8350' India' ICC'440' India' ICC'8384' India' ICC'4418' India' ICC'8522' India' ICC'4463' India' ICC'8607' India' ICC'4495' India' ICC'8621' India' ICC'4533' India' ICC'867' India' ICC'456' India' ICC'8740' India' ICC'4567' India' ICC'8855' India' ICC'4593' India' ICC'8950' India' ICC'4639' India' ICC'9402' India' ICC'4657' India' ICC'95' India' ICC'4814' India' ICC'9586' India' ICC'4841' India' ICC'9643' India' ICC'4872' India' ICC'9755' India' ICC'4918' India' ICC'9848' India' ICC'4948' India' ICC'9862' India' ICC'4973' India' ICC'9895' India' ICC'506' India' ICC'9942' India' ICC'5135' India' ! ICC'5337' India' ICC'5383' India' ! ICC'5434' India' ICC'5504' India' ICC'5613' India' ICC'5639' India' ICC'5845' India' ICC'5878' India' ICC'6306' India' ICC'6263' India' ICC'637' India' ICC'67' India' ICC'708' India' ICC'7184' India' ICC'7255' India' ICC'7272' India' ICC'7441' India' ICC'7554' India' ICC'7571' India'

! 98! List!of!Common!bean!genotypes!phenotyped!ARBC!2016!

ADP$ID$ Sub)ID$ Genotype$ gene$pool$ ADP)0004$ TZ)4$ KILOMBERO$ Andean$ ADP)0007$ TZ)7$ BUKOBA$ Andean$ ADP)0010$ TZ)10$ CANADA$ Andean$ ADP)0010$ Canada$ Andean$ ADP)0013$ TZ$ )13$ KIBUMBULA$ Andean$ ADP)0014$ TZ)14$ KIANGWE$ Andean$ ADP)0016$ TZ)16$ GOLOLI$ Andean$ ADP)0018$ TZ)18$ SODAN$ Andean$ ADP)0031$ TZ)31$ RH$No.$11$ Andean$ ADP)0033$ Kijivu$ Andean$ ADP)0035$ TZ$ )35$ Kokola$ Andean$ ADP)0037$ TZ)37$ W6$16488$ Andean$ ADP)0038$ TZ)38$ Moono$ Andean$ ADP)0041$ TZV)41$ MRONDO$ Andean$ ADP)0041$ Mrondo$ Andean$ ADP)0054$ TZV$ )55$ W6$16447$ Andean$ ADP)0055$ TZV)56$ KABUKU$ Andean$ ADP)0055$ KABUKU$ Andean$ ADP)0056$ TZV$ )57$ SOYA$ Andean$ ADP)0057$ TZV)58$ KIJIVU$ Andean$ ADP)0064$ TZV)65$ W6$16500$ Andean$ ADP)0069$ TZV)70$ SOYA$ Andean$ ADP)0080$ TZV)84$ KABLANKETI$ Andean$ ADP)0098$ AF)3$ Selian$97$ Andean$ ADP)0099$ AF)4$ Bwana$Shamba$ Andean$ ADP)0102$ AF)7$ Jesca$ Andean$ ADP)0102$ Jesca$ Andean$ ADP)0109$ AFV$ )14$ Kablanketi$ Andean$ ADP)0113$ AFV)18$ OPS)RS4$ Andean$ ADP)0114$ AFV)19$ OPS)RS1$ Andean$ ADP)0118$ AFV)23$ Werna$ Andean$ ADP)0119$ AFV)24$ A193$ Andean$ ADP)0121$ AFV)26$ Kranskop$HR)1$ Andean$ ADP)0123$ AFV)28$ Jenny$ Andean$ ADP)0166$ AF)30$ NABE$4$$ Andean$ ADP)0168$ AF)32$ KANYEBWA$$ Andean$ ADP)0186$ CC)3$ G$1368$ Andean$ ADP)0192$ CC)7$ G$2377$ Andean$ ADP)0247$ CC)29$ G$9975$ Andean$ ADP)0368$ CC)56$ G$23093$ Andean$

! 99! ADP)0390$ USC)4$ PI$307808$ Andean$ ADP)0428$ PR)2$ Colorado$del$Pais$ Andean$ ADP)0433$ PR)7$ PR9745)232$ Andean$ ADP)0434$ PR)8$ PR0737)1$ Andean$ ADP)0439$ PR)13$ 754)3$ Andean$ ADP)0458$ EC)12$ INIAP$483$ Andean$ ADP)0472$ EAL)14$ PI527537)B$ Andean$ ADP)0475$ EAL)17$ PI319706$ Andean$ ADP)0481$ EAL)23$ PI449428$ Andean$ ADP)0508$ AG)1$ Calembe$ Andean$ ADP)0516$ AG)9$ Mantega,$Kibala$ Andean$ ADP)0519$ AG)12$ Katarina,$Cela$ Andean$ ADP)0523$ AG)16$ Canario,$Cela$ Andean$ ADP)0526$ CL)3$ CAL$$143$ Andean$ ADP)0532$ CL)9$ A$$$$$$197$ Andean$ ADP)0536$ CL)13$ CAL$$$96$ Andean$ ADP)0537$ CL)14$ AFR$$619$ Andean$ ADP)0538$ CL)15$ RWR$$221$ Andean$ ADP)0540$ CL)17$ AFR$$708$ Andean$ ADP)0546$ CL)23$ RED$CANADIAN$WONDER$ Andean$ ADP)0546$ REDCANAD$ Andean$ ADP)0549$ $CL)26$ RWR$$$10$ Andean$ ADP)0551$ CL)28$ AFR$$612$ Andean$ ADP)0553$ CL)30$ AND277$ Andean$ ADP)0556$ CL)33$ BRB194$ Andean$ ADP)0557$ CL)34$ COS16$ Andean$ ADP)0566$ CL)43$ G5686$ Andean$ ADP)0567$ CL)44$ G$$4523$ Andean$ ADP)0571$ CL)48$ NUA45$ Andean$ ADP)0574$ CL)51$ RADICAL$CERINZA$ Andean$ ADP)0576$ CL)53$ SAB618$ Andean$ ADP)0584$ CL)61$ SAB659$ Andean$ ADP)0585$ CL)62$ SAB686$ Andean$ ADP)0586$ CL)63$ SAB691$ Andean$ ADP)0588$ CL)65$ SAP$$1$ Andean$ ADP)0590$ CL)67$ SEQ11$ Andean$ ADP)0591$ CL)68$ VELAZCO$LARGO$ Andean$ ADP)0592$ CL)69$ AND$1005$ Andean$ ADP)0597$ CL)74$ G23829$ Andean$ ADP)0607$ BC)154$ NY$105$ Andean$ ADP)0611$ BC)283$ Pompadour$B$ Andean$ ADP)0612$ BC)285$ ICA$Quimbaya$ Andean$ ADP)0626$ BCV)34$ Badillo$ Andean$

! 100! ADP)0628$ BCV)501$ H9659)27)7$ Andean$ ADP)0631$ OACInfer$ Andean$ ADP)0632$ BC$ )505$ TARS$HT$1$ Andean$ ADP)0636$ BC)58$ Montcalm$ Andean$ ADP)0643$ BC)108$ Cardinal$ Andean$ ADP)0656$ BC)253$ Royal$Red$ Andean$ ADP)0658$ BC)255$ Blush$ Andean$ ADP)0660$ BC)261$ Krimson$ Andean$ ADP)0660$ Krimson$ Andean$ ADP)0663$ $ USCR)CBB)20$ Andean$ ADP)0665$ $ USWK)CBB)17$ Andean$ ADP)0667$ $BC)277$ VA)19$ Andean$ ADP)0674$ UCD0704$ Andean$ ADP)0676$ $BC)377$ CELRK$ Andean$ ADP)0683$ BCV)272$ IJR$ Andean$ ADP)0687$ PinkPanther$ Andean$ ADP)0747$ $ PI638816$ Andean$ $ ADP)0784$ PS11)006C)8)B$ Andean$ ADP)0786$ $ USCR)14$ Andean$ ADP)0790$ $PR0633)10$ Red$mottled$ Andean$ ADP)0791$ PR1146)123$ Yellow$$ Andean$ ADP)0792$ PR1146)124$ Yellow$$ Andean$ black$line$ Krimson$ Andean$ green$line$ $ Rosetta$ Andean$ red$line$ $ Alpena$ Andean$ XX$36$ $ G19833$x$RBR$211RIL$36$ Andean$ XX$41$ $ G19833$x$RBR$211RIL$41$ Andean$ 1$ $ Aifi$Wuriti$SS$ Meso$ 2$ $ ALB$213$ Meso$ 3$ $ ALB$74$ Meso$ 4$ $ ALS$0532)6$ Meso$ 5$ $ Amadeus$77$ Meso$ 6$ $ B12724$ Meso$ 7$ $ B14302$ Meso$ 8$ $ B14303$ Meso$ 9$ $ B14311$ Meso$ 10$ $ BAT$477$ Meso$ 11$ $ BAT$881$ Meso$ 12$ $ Beniquez$ Meso$ 13$ $ BFS$139$ Meso$ 14$ $ BFS$29$ Meso$ 15$ $ BFS$81$ Meso$ $

! 101! 16$ BFS$87$ Meso$ 17$ $ BFS$94$ Meso$ 18$ $ BFS$95$ Meso$ 19$ $ BIOF$2)106$ Meso$ 20$ $ BIOF$4)70$ Meso$ 21$ $ BK9)2$ Meso$ 22$ $ Bribri$ Meso$ 23$ $ BRT$103)182$ Meso$ 24$ $ BRT$943)20$ Meso$ 25$ $ CALIMA:$Check)3$ Meso$ 26$ $ Cardenal$ Meso$ 27$ $ Carioca$ Meso$ 28$ $ Carrizalito$ Meso$ 29$ $ Cedrón$ Meso$ 30$ $ Centa$Pipil$ Meso$ 31$ $ Croissant$ Meso$ 32$ $ DEHORO$ Meso$ 33$ $ DOR$364$ Meso$ 34$ $ DOR$390$(ck)$ Meso$ 35$ $ DPC40$ Meso$ 36$ $ FBN$1203)43$ Meso$ 37$ $ FBN$1203)47$ Meso$ 38$ $ FBN$1205)31$ Meso$ 39$ $ FBN$1210)48$ Meso$ 40$ $ G21212$ Meso$ 41$ $ GN9)4$ Meso$ 42$ $ I9365)31$ Meso$ 43$ $ ICA$Pijao$ Meso$ 44$ $ ICB$301)204$ Meso$ 45$ $ ICTAZAM$ML$ Meso$ 46$ $ IJR$ Meso$ 47$ $ INB$835$ Meso$ 48$ $ INB$841$ Meso$ 49$ $ INTA$Precoz$ Meso$ 50$ $ Jamapa$ Meso$ 51$ $ Matambú$ Meso$ 52$ $ Matterhorn$ Meso$ 53$ $ MEN$2201)64$ML$ Meso$ 54$ $ MER$2212)28$ Meso$ 55$ $ MHN$322)49$ Meso$ 56$ $ Morales$ Meso$ 57$ $ MSU$BNF$Entry$86$ Meso$ 58$ $ NCB$280$ Meso$ $

! 102! 59$ Paraisito$ Meso$ 60$ $ Pérola$ Meso$ 61$ $ PR$0401)259$ Meso$ 62$ $ PR$0443)151$ Meso$ 63$ $ PR0806)81$ Meso$ 64$ $ PR1147)1$ Meso$ 65$ $ PR1147)3$ Meso$ 66$ $ PR1147)6$ Meso$ 67$ $ PR1147)8$ Meso$ 68$ $ PR1165)3$ Meso$ 69$ $ PR1217)1$ Meso$ 70$ $ PR1217)16$ Meso$ 71$ $ PR1418)15$ Meso$ 72$ $ PR1483)105$ Meso$ 73$ $ PR9920)171$ Meso$ 74$ $ Quimbaya$ Meso$ 75$ $ RCB$593$ Meso$ 76$ $ Rosetta$ Meso$ 77$ $ RRH$336)28$ Meso$ 78$ $ Sayaxché$ML$ Meso$ 79$ $ SB$747$ Meso$ 80$ $ SB$754$ Meso$ 81$ $ SB$761$ Meso$ 82$ $ SB$770$ Meso$ 83$ $ SB$774$ Meso$ 84$ $ SB$781$ Meso$ 85$ $ SB$787$ Meso$ 86$ $ SB$793$ Meso$ 87$ $ SB$815$ Meso$ 88$ $ SB$DT1$ Meso$ 89$ $ SEF$10$ Meso$ 90$ $ SEF$13$ Meso$ 91$ $ SEF$15$ Meso$ 92$ $ SEF$16$ Meso$ 93$ $ SEF$17$ Meso$ 94$ $ SEF$71$ Meso$ 95$ $ SEQ$342)39$ Meso$ 96$ $ SEQ$342)89$ Meso$ 97$ $ SER$113$ Meso$ 98$ $ SER$118$ Meso$ 99$ $ SER$125$ Meso$ 100$ $ SER$16$ Meso$ 101$ $ SER$78$ Meso$ $

! 103! 102$ SJC$730)79$ Meso$ 103$ $ SMC$137$ Meso$ 104$ $ SMR$138$ Meso$ 105$ $ SMR$139$ Meso$ 106$ $ SXB$405$ Meso$ 107$ $ SXB$412$ Meso$ 108$ $ TARS$HT)1$ Meso$ 110$ $ TARS)LFR1$ Meso$ 111$ $ TARS)MST1$ Meso$ 109$ $ TARS09)RR029$ Meso$ 112$ $ Tepary$G40001$ Meso$ 113$ $ Tepary$Tep)22$ Meso$ 114$ $ Tio$Canela$75$ Meso$ 115$ $ USRM$20$ Meso$ 116$ $ VAX$6$ Meso$ 117$ $ Verano$ Meso$ 118$ $ XRAV)40)4$ Meso$ 119$ $ Zenith$ Meso$ 120$ $ Zorro$ Meso$ ! $

List!of!desert!adapted!legumes!phenotyped!ARBC!2013! geno.id$ name$ genus$ species$ spp$ PT005$ Sacaton$White$ Phaseolus$ acutifolius$ FV001$ Tarahumara$Habas$ Vicia$ faba$ $ PT099$ Paiute$Mixed$ Phaseolus$ acutifolius$ $ PC043$ Tarahumara$Choliwame$ Phaseolus$ vulgaris$ $ PW105$ Kitt$Peak$Wild$Tepary$ Phaseolus$ acutifolius$ tenuifolius$ $ PC014$ Tarahumara$Purple$Ojos$ Phaseolus$ vulgaris$ PC079$ Hopi$White$ Phaseolus$ vulgaris$ $ PC098$ Frijol$Gringo$ Phaseolus$ vulgaris$ $ PW095$ Chihuahua$Wild$Tepary$ Phaseolus$ acutifolius$ tenuifolius$ $ PT120$ Santa$Rosa$Brown$ Phaseolus$ acutifolius$ latifolius$ PC083$ Chihuahua$Ojo$de$Cabra$ Phaseolus$ vulgaris$ PW096$ Tiburon$Island$Wild$ Phaseolus$ acutifolius$ $ PL009$ Hopi$Red$Lima$ Phaseolus$ lunatus$ $ PL073$ Hopi$White$Hatilko$ Phaseolus$ lunatus$ $ PL080$ Hopi$Gray$ Phaseolus$ lunatus$ $ PT107$ Cocopah$Brown$ Phaseolus$ acutifolius$ $ PT006$ Sonoran$White$ Phaseolus$ acutifolius$ $ PL011$ Pima$Orange$ Phaseolus$ lunatus$ $ PT118$ Colonia$Morelos$Speckled$ Phaseolus$ acutifolius$ $ $

! 104! NA$ NA$ NA$ NA$ PT003$ Yoreme$White$ Phaseolus$ acutifolius$ $ $ssp.$ PW091$ Wild$Cocolmeca$ Phaseolus$ maculatus$ maculatus$ V009$ Pima$Bajo$ Vigna$ unguiculata$ PT108$ Cocopah$White$ Phaseolus$ acutifolius$ $ PC095$ Tarahumara$Ejotero$Negro$ Phaseolus$ vulgaris$ $ PC059$ Tarahumara$Star$ Phaseolus$ vulgaris$ $ PT078$ Yoeme$Brown$ Phaseolus$ acutifolius$ $ PC063$ Tohon$O'odham$Pink$ Phaseolus$ vulgaris$ $ FV022$ Cuarteles$ Vicia$ faba$ $ PT075$ Tohono$O'odham$Brown$ Phaseolus$ acutifolius$ $ V006$ Tohono$O'odham$ Vigna$ unguiculata$ $ Mountain$Pima$Ojo$de$ $ PC021$ Cabra$ Phaseolus$ vulgaris$ PC054$ Tarahumara$Ojo$de$Cabra$ Phaseolus$ vulgaris$ $ PT004$ Sacaton$Brown$ Phaseolus$ acutifolius$ $ FV014$ Truchas$Habas$ Vicia$ faba$ $ PT074$ Pinacate$ Phaseolus$ acutifolius$ latifolius$ $ List!of!Tepary!genotype!names!phenotyped!2014!

entry$num$ ID$#$ Culture$ source$ year$ 1$ 30387$ 13IS)7500$ PR$ 2014$ 2$ 30393$ 13IS)7501$ PR$ 2014$ 3$ 30399$ 13IS)7502$ PR$ 2014$ 4$ 30403$ 13IS)7503$ PR$ 2014$ 5$ 30409$ 13IS)7504$ PR$ 2014$ 6$ 30413$ 13IS)7505$ PR$ 2014$ 7$ 30417$ 13IS)7506$ PR$ 2014$ 8$ 30421$ 13IS)7507$ PR$ 2014$ 9$ 30431$ 13IS)7508$ PR$ 2014$ 10$ 30441$ 13IS)7509$ PR$ 2014$ 11$ 30455$ 13IS)7510$ PR$ 2014$ 12$ 30459$ 13IS)7511$ PR$ 2014$ 13$ 30461$ 13IS)7580$ PR$ 2014$ 14$ 30463$ 13IS)7513$ PR$ 2014$ 15$ 30467$ 13IS)7514$ PR$ 2014$ 16$ 30475$ 13IS)7515$ PR$ 2014$ 17$ 30483$ 13IS)7516$ PR$ 2014$ 18$ 30493$ 13IS)7517$ PR$ 2014$ 19$ 30495$ 13IS)7518$ PR$ 2014$ 20$ 30497$ 13IS)7599$ PR$ 2014$ 21$ 30509$ 13IS)7520$ PR$ 2014$ 22$ 30523$ 13IS)7521$ PR$ 2014$ 23$ 30525$ 13IS)7522$ PR$ 2014$

! 105! 24$ 30531$ 13IS)7523$ PR$ 2014$ 25$ 30535$ 13IS)7524$ PR$ 2014$ 26$ 30539$ 13IS)7525$ PR$ 2014$ 27$ 30541$ 13IS)7526$ PR$ 2014$ 28$ 30543$ 13IS)7527$ PR$ 2014$ 29$ 30547$ 13IS)7528$ PR$ 2014$ 30$ 30551$ 13IS)7529$ PR$ 2014$ 31$ 30555$ 13IS)7530$ PR$ 2014$ 32$ 30557$ 13IS)7531$ PR$ 2014$ 33$ 30561$ 13IS)7532$ PR$ 2014$ 34$ 30563$ 13IS)7533$ PR$ 2014$ 35$ 30567$ 13IS)7534$ PR$ 2014$ 36$ 30286)2$ 13IS)7682$ PR$ 2014$ 37$ 30290)2$ 13IS)7692$ PR$ 2014$ 38$ 30290)3$ 13IS)7693$ PR$ 2014$ 39$ 30301)4$ 13IS)7704$ PR$ 2014$ 40$ 30303)1$ 13IS)7706$ PR$ 2014$ 41$ 30304)1$ 13IS)7711$ PR$ 2014$ 42$ 30304)5$ 13IS)7715$ PR$ 2014$ 43$ 30307)5$ 13IS)7725$ PR$ 2014$ 44$ 30308)1$ 13IS)7726$ PR$ 2014$ 45$ 30308)2$ 13IS)7727$ PR$ 2014$ 46$ 30308)5$ 13IS)7730$ PR$ 2014$ 47$ 30313)3$ 13IS)7733$ PR$ 2014$ 48$ 30313)5$ 13IS)7735$ PR$ 2014$ 49$ 30314)1$ 13IS)7736$ PR$ 2014$ 50$ 30314)2$ 13IS)7737$ PR$ 2014$ 51$ 30314)3$ 13IS)7738$ PR$ 2014$ 52$ 30314)4$ 13IS)7739$ PR$ 2014$ 53$ 30314)5$ 13IS)7740$ PR$ 2014$ 54$ 30315)1$ 13IS)7741$ PR$ 2014$ 55$ 30315)2$ 13IS)7742$ PR$ 2014$ 56$ 30315)3$ 13IS)7743$ PR$ 2014$ 57$ 30315)4$ 13IS)7744$ PR$ 2014$ 58$ 30317)1$ 13IS)7746$ PR$ 2014$ 59$ 30320)1$ 13IS)7751$ PR$ 2014$ 60$ 30320)3$ 13IS)7753$ PR$ 2014$ 61$ 30320)4$ 13IS)7754$ PR$ 2014$ 62$ 30320)5$ 13IS)7755$ PR$ 2014$ 63$ 30325)3$ 13IS)7773$ PR$ 2014$ 64$ 30325)4$ 13IS)7774$ PR$ 2014$ 65$ 30329)2$ 13IS)7777$ PR$ 2014$ 66$ 30329)3$ 13IS)7778$ PR$ 2014$ 67$ 30329)5$ 13IS)7780$ PR$ 2014$ 68$ 30330)4$ 13IS)7784$ PR$ 2014$ 69$ 30331)1$ 13IS)7786$ PR$ 2014$

! 106! 70$ 30344)2$ 13IS)7802$ PR$ 2014$ 71$ 30344)3$ 13IS)7803$ PR$ 2014$ 72$ 30345)2$ 13IS)7807$ PR$ 2014$ 73$ 30347)1$ 13IS)7811$ PR$ 2014$ 74$ 30350)1$ 13IS)7816$ PR$ 2014$ 75$ 30350)2$ 13IS)7817$ PR$ 2014$ 76$ 30369)2$ 13IS)7857$ PR$ 2014$ 77$ 30369)3$ 13IS)7858$ PR$ 2014$ 78$ 30369)4$ 13IS)7859$ PR$ 2014$ 79$ 30370)1$ 13IS)7861$ PR$ 2014$ 80$ 30370)4$ 13IS)7864$ PR$ 2014$ 81$ 30371)1$ 13IS)7866$ PR$ 2014$ 82$ 30375)3$ 13IS)7873$ PR$ 2014$ 83$ 30375)4$ 13IS)7874$ PR$ 2014$ 84$ 30375)5$ 13IS)7875$ PR$ 2014$ 85$ Tep.$1$ PI)440801$ PR$ 2014$ 86$ Tep.$2$ PI)440802$ PR$ 2014$ 87$ Tep.$3$ PI)440799$ PR$ 2014$ 88$ Tep.$4$ Neb$T)1)s$ PR$ 2014$ 89$ Tep.$5$ TB1$ PR$ 2014$ 90$ Tep.$6$ PI)502217)s$ PR$ 2014$ 91$ Tep.$7$ PR$ 2014$ 92$ Tep.$8$ $ PR$ 2014$ 93$ Tep.$9$ $ PR$ 2014$ 94$ Tep.$10$ $ PR$ 2014$ 95$ Tep.$11$ $ PR$ 2014$ 96$ Tep.$12$ $ PR$ 2014$ 97$ Tep.$13$ $ PR$ 2014$ 98$ Tep.$14$ $ PR$ 2014$ 99$ Tep.$15$ $ PR$ 2014$ 100$ Tep.$16$ $ PR$ 2014$ 101$ Tep.$17$ PR$ )101$ PR$ 2014$ 102$ Tep.$18$ PR)102$ PR$ 2014$ 103$ 30387$ PI)477033$ PR$ 2014$ 104$ Tep.$21$ PR)104$ PR$ 2014$ 105$ Tep.$22$ PR)105$ PR$ 2014$ 106$ Tep.$23$ PR)106$ PR$ 2014$ 107$ Tep.$24$ PR)107$ PR$ 2014$ 108$ Tep.$25$ PR)108$ PR$ 2014$ 109$ Tep.$26$ PR)109$ PR$ 2014$ 110$ Tep.$27$ PR)110$ PR$ 2014$ 111$ Tep.$28$ PR)111$ PR$ 2014$ 112$ Tep.$29$ PR)112$ PR$ 2014$ 113$ Tep.$30$ PR)113$ PR$ 2014$ 114$ Tep.$31$ PR)114$ PR$ 2014$ 115$ Tep.$32$ PR)115$ PR$ 2014$

! 107! 116$ G40177E1$ G40177E1$ PR$ 2014$ 117$ G40042$ G40042$ PR$ 2014$ 118$ G40103$ G40103$ PR$ 2014$ 119$ Mitla$speckled$ Colorado$ 2014$ 120$ $ Sonora$ Colorado$ 2014$ 121$ $ Santa$Rosa$White$ Colorado$ 2014$ 122$ $ Tucson$brown$ Colorado$ 2014$ ! $

List!of!Tepary!genotype!names!phenotyped!2015! geno$ID$ geno$name$ geno$code$ year$ 1$ G40001$ TDP)1$ 2015$ 2$ G40006B$ TDP)2$ 2015$ 3$ G40008$ TDP)3$ 2015$ 4$ G40009$ TDP)4$ 2015$ 5$ G40010$ TDP)5$ 2015$ 6$ G40012$ TDP)6$ 2015$ 6$ G40012$ TDP)6$ 2015$ 7$ G40013A$ TDP)7$ 2015$ 8$ G40013D$ TDP)8$ 2015$ 9$ G40018$ TDP)9$ 2015$ 10$ G40019$ TDP)10$ 2015$ 10$ G40019$ TDP)10$ 2015$ 11$ G40020$ TDP)11$ 2015$ 12$ G40021$ TDP)12$ 2015$ 13$ G40022A$ TDP)14$ 2015$ 14$ G40023$ TDP)15$ 2015$ 15$ G40025$ TDP)16$ 2015$ 16$ G40028$ TDP)17$ 2015$ 17$ G40029$ TDP)18$ 2015$ 18$ G40030$ TDP)19$ 2015$ 19$ G40032$ TDP)20$ 2015$ 20$ G40034$ TDP)21$ 2015$ 21$ G40035$ TDP)22$ 2015$ 22$ G40036$ TDP)23$ 2015$ 23$ G40037$ TDP)24$ 2015$ 24$ G40037A$ TDP)25$ 2015$ 25$ G40038$ TDP)26$ 2015$ 26$ G40039$ TDP)27$ 2015$ 27$ G40040$ TDP)28$ 2015$ 28$ G40041$ TDP)29$ 2015$ 29$ G40042$ TDP)30$ 2015$ 30$ G40043$ TDP)31$ 2015$

! 108! 31$ G40057$ TDP)45$ 2015$ 32$ G40058$ TDP)46$ 2015$ 33$ G40060$ TDP)48$ 2015$ 34$ G40061$ TDP)49$ 2015$ 35$ G40062$ TDP)50$ 2015$ 36$ G40066A$ TDP)51$ 2015$ 37$ G40067$ TDP)52$ 2015$ 38$ G40110$ TDP)88$ 2015$ 39$ G40111$ TDP)89$ 2015$ 39$ G40111$ TDP)89$ 2015$ 40$ G40119$ TDP)97$ 2015$ 41$ G40120$ TDP)98$ 2015$ 42$ G40122$ TDP)99$ 2015$ 43$ G40125$ TDP)100$ 2015$ 44$ G40127$ TDP)101$ 2015$ 45$ G40128$ TDP)102$ 2015$ 46$ G40129$ TDP)103$ 2015$ 46$ G40129$ TDP)103$ 2015$ 47$ G40130$ TDP)104$ 2015$ 48$ G40138$ TDP)105$ 2015$ 49$ G40140$ TDP)106$ 2015$ 49$ G40140$ TDP)106$ 2015$ 50$ G40141$ TDP)107$ 2015$ 51$ G40142$ TDP)108$ 2015$ 52$ G40142A$ TDP)109$ 2015$ 53$ G40143$ TDP)110$ 2015$ 54$ G40144$ TDP)111$ 2015$ 55$ G40144A$ TDP)112$ 2015$ 56$ G40144B$ TDP)113$ 2015$ 57$ G40144C$ TDP)114$ 2015$ 58$ G40145$ TDP)115$ 2015$ 59$ G40146$ TDP)116$ 2015$ 60$ G40147$ TDP)117$ 2015$ 60$ G40147$ TDP)117$ 2015$ 61$ G40148$ TDP)118$ 2015$ 62$ G40149$ TDP)119$ 2015$ 63$ G40150$ TDP)120$ 2015$ 64$ G40151$ TDP)121$ 2015$ 65$ G40152$ TDP)122$ 2015$ 66$ G40153$ TDP)123$ 2015$ 67$ G40154$ TDP)124$ 2015$ 68$ G40156$ TDP)125$ 2015$ 69$ G40157$ TDP)126$ 2015$

! 109! 70$ G40158A$ TDP)127$ 2015$ 71$ G40159$ TDP)128$ 2015$ 72$ G40160$ TDP)129$ 2015$ 73$ G40161$ TDP)130$ 2015$ 74$ G40162$ TDP)131$ 2015$ 75$ G40163$ TDP)132$ 2015$ 76$ G40164$ TDP)133$ 2015$ 77$ G40165$ TDP)134$ 2015$ 78$ G40172$ TDP)139$ 2015$ 79$ G40173$ TDP)140$ 2015$ 80$ G40173A$ TDP)141$ 2015$ 81$ G40173B$ TDP)142$ 2015$ 82$ G40173C$ TDP)143$ 2015$ 83$ G40174$ TDP)144$ 2015$ 84$ G40175$ TDP)145$ 2015$ 84$ G40175$ TDP)145$ 2015$ 85$ G40176$ TDP)146$ 2015$ 86$ G40177A$ TDP)147$ 2015$ 87$ G40177A1$ TDP)148$ 2015$ 88$ G40177A2$ TDP)149$ 2015$ 89$ G40177B1$ TDP)150$ 2015$ 90$ G40177C$ TDP)151$ 2015$ 91$ G40177D$ TDP)152$ 2015$ 92$ G40177E2$ TDP)155$ 2015$ 156$ TB$1$ TDP)314$ 2015$ 1$ G40001$ TDP)1$ 2015$ 2$ G40006B$ TDP)2$ 2015$ 3$ G40008$ TDP)3$ 2015$ 4$ G40009$ TDP)4$ 2015$ 5$ G40010$ TDP)5$ 2015$ 6$ G40012$ TDP)6$ 2015$ 6$ G40012$ TDP)6$ 2015$ 7$ G40013A$ TDP)7$ 2015$ 8$ G40013D$ TDP)8$ 2015$ 9$ G40018$ TDP)9$ 2015$ 10$ G40019$ TDP)10$ 2015$ 10$ G40019$ TDP)10$ 2015$ 11$ G40020$ TDP)11$ 2015$ 12$ G40021$ TDP)12$ 2015$ 13$ G40022A$ TDP)14$ 2015$ 14$ G40023$ TDP)15$ 2015$ 15$ G40025$ TDP)16$ 2015$ 16$ G40028$ TDP)17$ 2015$

! 110! 17$ G40029$ TDP)18$ 2015$ 18$ G40030$ TDP)19$ 2015$ 19$ G40032$ TDP)20$ 2015$ 20$ G40034$ TDP)21$ 2015$ 21$ G40035$ TDP)22$ 2015$ 22$ G40036$ TDP)23$ 2015$ 23$ G40037$ TDP)24$ 2015$ 24$ G40037A$ TDP)25$ 2015$ 25$ G40038$ TDP)26$ 2015$ 26$ G40039$ TDP)27$ 2015$ 27$ G40040$ TDP)28$ 2015$ 28$ G40041$ TDP)29$ 2015$ 29$ G40042$ TDP)30$ 2015$ 30$ G40043$ TDP)31$ 2015$ 31$ G40057$ TDP)45$ 2015$ 32$ G40058$ TDP)46$ 2015$ 33$ G40060$ TDP)48$ 2015$ 34$ G40061$ TDP)49$ 2015$ 35$ G40062$ TDP)50$ 2015$ 36$ G40066A$ TDP)51$ 2015$ 37$ G40067$ TDP)52$ 2015$ 38$ G40110$ TDP)88$ 2015$ 39$ G40111$ TDP)89$ 2015$ 39$ G40111$ TDP)89$ 2015$ 40$ G40119$ TDP)97$ 2015$ 41$ G40120$ TDP)98$ 2015$ 42$ G40122$ TDP)99$ 2015$ 43$ G40125$ TDP)100$ 2015$ 44$ G40127$ TDP)101$ 2015$ 45$ G40128$ TDP)102$ 2015$ 46$ G40129$ TDP)103$ 2015$ 46$ G40129$ TDP)103$ 2015$ 47$ G40130$ TDP)104$ 2015$ 48$ G40138$ TDP)105$ 2015$ 49$ G40140$ TDP)106$ 2015$ 49$ G40140$ TDP)106$ 2015$ 50$ G40141$ TDP)107$ 2015$ 51$ G40142$ TDP)108$ 2015$ 52$ G40142A$ TDP)109$ 2015$ 53$ G40143$ TDP)110$ 2015$ 54$ G40144$ TDP)111$ 2015$ 55$ G40144A$ TDP)112$ 2015$ 56$ G40144B$ TDP)113$ 2015$

! 111! 57$ G40144C$ TDP)114$ 2015$ 58$ G40145$ TDP)115$ 2015$ 59$ G40146$ TDP)116$ 2015$ 60$ G40147$ TDP)117$ 2015$ 60$ G40147$ TDP)117$ 2015$ 61$ G40148$ TDP)118$ 2015$ 62$ G40149$ TDP)119$ 2015$ 63$ G40150$ TDP)120$ 2015$ 64$ G40151$ TDP)121$ 2015$ 65$ G40152$ TDP)122$ 2015$ 66$ G40153$ TDP)123$ 2015$ 67$ G40154$ TDP)124$ 2015$ 68$ G40156$ TDP)125$ 2015$ 69$ G40157$ TDP)126$ 2015$ 70$ G40158A$ TDP)127$ 2015$ 71$ G40159$ TDP)128$ 2015$ 72$ G40160$ TDP)129$ 2015$ 73$ G40161$ TDP)130$ 2015$ 74$ G40162$ TDP)131$ 2015$ 75$ G40163$ TDP)132$ 2015$ 76$ G40164$ TDP)133$ 2015$ 77$ G40165$ TDP)134$ 2015$ 78$ G40172$ TDP)139$ 2015$ 79$ G40173$ TDP)140$ 2015$ 80$ G40173A$ TDP)141$ 2015$ 81$ G40173B$ TDP)142$ 2015$ 82$ G40173C$ TDP)143$ 2015$ 83$ G40174$ TDP)144$ 2015$ 84$ G40175$ TDP)145$ 2015$ 84$ G40175$ TDP)145$ 2015$ 85$ G40176$ TDP)146$ 2015$ 86$ G40177A$ TDP)147$ 2015$ 87$ G40177A1$ TDP)148$ 2015$ 88$ G40177A2$ TDP)149$ 2015$ 89$ G40177B1$ TDP)150$ 2015$ 90$ G40177C$ TDP)151$ 2015$ 91$ G40177D$ TDP)152$ 2015$ 92$ G40177E2$ TDP)155$ 2015$ 156$ TB$1$ TDP)314$ 2015$ !

! 112! Epilogue The shovelomics method that quantifies root architecture has developed over time. In 2009 we began by using a protractor like visual aid with various lines and dots drawn on a cutting board to reduce the objective nature of root architectural observations. That first year a trial in Honduras motivated changes to the protocol as we started noticing how difficult measuring basal root growth angle (BRGA) was given that all basal roots don’t grow in the same plane or at the same angle. By 2011 I had more fully grasped how difficult measuring angles was and how variable the growth angle of various roots can be on a single plant. I began measuring minimum and maximum angles rather than trying to visually assign an angle. Later I measured minimum, maximum and assigned a visually determined average by more heaving weighting the angle of the larger diameter roots. I also explored greater subjectivity as in a trial in 2013 where I resorted to assigning a 1-3 score for rooting depth. I currently am using a system that takes into account the shallow and the deep range of BRGA, a visual score of the dominant BRGA and a calculated BRGA based on the deep and shallow range. I then combine BRGA with adventitious root number to create a shallow rooting score and combine BRGA with primary root diameter to create a deep rooting score such that a higher score indicates that a greater allocation of resources to either shallow or deep exploration. This does a reasonable job classifying the primary areas of investment of root systems based on the crown architecture, but it remains difficult to find associations with biomass or other performance indicators. This problem rests on the relatively high variation of root traits in a plot. Where does the root architectural research go next? In a 1984 article highlighting the common occurrence of pseudo-replication in ecological research Stuart Hurlbert suggests that true replication is necessary to account for chance events and preexisting gradients that can be dealt with by control treatments, increased replication or check lines (Hurlbert, 1984). He contrasts this, in a tongue in cheek fashion, to demonic intrusion that can only be dealt with through external vigilance, exorcism and human sacrifice. I like to say that I’ve experienced both demonic and non-demonic intrusion. Several trials were lost or confounded by factors know or unknown. My first field season was confounded by the legacy effects of the previous trial. I’ve had drought experiments lost to rain, low fertility experiments fertilized, seedlings grazed by protected species of deer, trials lost to incorrectly applied herbicide, a season lost to damage from atrazine mobilized by liming, trials confounded by disease amongst other more mundane problems like excessively high coefficient of variances. I’ve struggled to move forward when the architectural patterns I'm trying to detect don’t seem to be observable. This last category mentioned above, “excessively” high variation is by far the most interesting because it suggests that there are unacknowledged patterns or unknown factors affecting the experiments. The influence of these factors far outweighs the influence of the simplified root phenotypes to which I was

! 113! attempting to attribute my results. I believe I have been too hasty to attribute this lack of consistency to demonic influence rather than to approach the problem of understanding plant responses to environments. The question of understanding the drivers of variation and the variety of ways that plants respond to environmental stimuli is precisely where I should have directed my research, and where root research should go next. My understanding of plasticity, compensation and dynamism is much more developed now than it was when I began. I started out with what I see now as a cartoon-like vision of root architecture. I expected root architecture to be easily classifiable as deep or shallow, with genotypes exhibiting consistent and clearly observable phenes across locations and years. After some years of challenges with weather and agronomy that delayed a truly intensive analysis of root architecture I came to a phase of acknowledging the extensive variation apparent within a given genotype in a given field and when comparing multiple fields and years. I regret that I so easily moved on from trials that were compromised or produced data I didn’t know how to interpret. Gathering what lessons I could have from those trials may have led me to classify the variation I saw not as noise but rather as biologically significant variation. I’m beginning to appreciate the complexity of plant and environment interactions. I'm grasping that root architecture in one environment does not necessarily have a strong correlation to phenotype or performance in a different environment. I wish I had paid more attention to and encouraged the research of a former lab- mate Claire Lorts as she sought to understand how maternal growth environment and even position in the pod could affect offspring. I think this was a good idea. It seems reasonable that understanding maternal effects would shed some light on the variation observed when comparing performance of a given genotype across years or when comparing plants grown from seeds sourced from different lots. The “problem” of excessively high coefficients of variation is really a failure of perception. I believe the variation is real and something that we should be paying attention to. Part of the problem may be due to categorizing phenotypic responses of root architecture to environments as stress responses rather than as adaptive adjustments (Anfodillo et al., 2016). Exploring variation as an adaptive response along with greater awareness of allometry between parts of the root system may enable a deeper understanding of the dynamic suites of traits related to performance and yield. Pursuing an understanding of the dynamic nature of plant growth, compensation and response to environment is where root architectural research should move. While this is being pursued at molecular and ecosystem levels bridging the links between organ (i.e. root), organism, and community to generate an integrated understanding of the form and function remains a challenge. Addressing the question of variation is a future opportunity for the author’s future work on compensation and dynamism as well as on spatial variation. I believe with shovelomics we’re just starting to understand root systems, albeit in an oversimplified way. A limitation of the field phenotyping approach that gathers root architectural data at only one point in time is somewhat analogous to the

! 114! type of error generated by aliasing in which periodic sampling doesn’t match or adequately represent the actual dynamic nature of the object or phenomenon being measured. Measuring characteristics of the root crown is essentially a snapshot of the past combined with a hint of what has transpired. One may see the effects of individual roots reaching farther or having greater branching farther down their axis in the form of increased secondary growth. One may see a decaying root or the scar of a root that has since decayed due to disease or soil drying and infer that other roots must have compensated. One may find signs of nematode damage but can only wonder how that has affected branching density and axial elongation. One can observe a root rot complex or fusarium invading the hypocotyl but can only speculate on how the plant is managing to transport water and explore the soil. Moving down the row one may notice a small plant next to a large plant, a rock under the third, a weed growing next to the fourth, a scrawny diseased plant struggling to sit up, the next growing through a patch of residue from the previous crop, and then finally one with the type of root architecture we expect to see and are equipped to measure so we stop there and take our observations. From afar the field may look homogenous but as the level of detail we examine increases so too does diversity and our field based root architectural sampling protocols ignore that diversity. There are limitations of time and energy that can be expended on any single study but these constraints can shift as new tools become available. Performing non-destructive serial measurements would open up a new window to understanding the dynamic nature of root, soil and plant interactions. The future of root research is in understanding the dynamic nature of growth and its relationship to resource use. To do this we need to increase sampling frequency in order to identify and focus on the stages of growth and measurements that are most frequently indicative of performance. In order to identify when, where and how to sample we need to first implement a saturated sampling protocol. We need to phenotype many more aspects of the root system and understand how interactions with the rest of the plant are mediated. The next stage is in pairing root and shoot from, function and strategy. The challenge here is first to identify what root and shoot parameters should be gathered and when. The second part of the challenge is to make the observations in a way that facilitates linking them to root form and function. Concurrently we should seek to understand plot and stand level interactions. As has become evident, not all plants in a plot have the same or even comparable root crown architectures. It is yet to be determined to what degree plants in a single stand may interact, be that in competition, cooperation, compensation, avoidance or some combination of all four. Mutual aid, cooperation and community survivorship have been systematically underemphasized in agricultural research and may offer revealing insights into plant behavior (Kropotkin, 1902). Examples of mutual aid, or facilitation as it is more commonly called in and across plant and animal communities abound (Brooker et al., 2008; Jordano et al., 2003). Research on the root system architecture of the three sisters offers a unique scientific insight into an understudies aspect of niche complementarity (Zhang et al., 2014). It is possible

! 115! that this type of root based complementarity may have been much more common in the thousands of years between crop domestication and the last 70 years of modern industrial agriculture. Differences in planting times and phenology amongst the three sisters may likely an important part of the above-ground complementarity but this strategy is not longer used by humans on a large scale. The transition from climbing or prostrate growth habit to determinate bush beans was one of the most significant achievements of modern plant breeding (Kelly, 2001). CIAT work on selecting beans for decisive shifts to reproductive stage growth is an attempt to distance common beans from their perennial or multi- season origins (Rao et al., 2013). This effort acknowledges the more primitive tendencies of common bean to reserve some resources. This phenomenon was easily observable in ARBC 2016 when a second flush of vegetative growth and flowers initiated following late season rains when the plants were already physiologically mature. Domesticated plants may have a set of goals, means and a genetic vocabulary that does not always coincide with our species’ interest as single season monoculture agriculturalists. Their goal may be long-term community survivorship, which is different than maximizing yield at the plot or field level (Weiner, 2017). We should reconsider the assumption we make about the crops we cultivate having completely bought into the industrial agricultural model and being intent only upon maximizing single season reproductive capacity. While genetically pure scientifically generated seed is widely available for purchase some farmers operate in conditions where significant gene flow between wild– weed–crop mixes still occurs (Beebe et al., 1997). We err if we think that by decreasing seed dormancy, reducing explosive dehiscence and increasing seed size we’ve made beans into some sort of automated food production system. They have a lineage that far precedes modern industrial agricultural interventions. If observations on the dynamic nature of crown root systems offer any bearing on the lack of uniformity, repeatability and dynamic nature of resource acquisition strategies then we should also reconsider our assumptions on the completeness of domestication. As I’ve thought about where agro-ecological research moves next I think about how it may be comparable to human societies and to other networks. Attempting to understand the interactions among plant organs, soil and atmospheric processes, competition and cooperating with neighbors, and the relative utility of divergent and dynamic life strategies is an extremely complex problem which reductionist scientific processes may not adequately address. Trying to conceptualize this level of detail and dynamic interaction is reminiscent of the hyperobject concept popularized by Timothy Morton (Morton, 2013). A hyperobject is something so complex, multifaceted and that operates at multiple scales that humans cannot fully understand it. An example of this is climate change and the difficulty we have in disentangling our effects on it and its effects on us, which is comparable to the challenges of phenotyping and predicting the utility of those phenotypes. In a sense the conundrum of making inferences about the utility of a soil foraging strategy based on measurements of a particular organ and a particular time is analogous to the parable of the blind men trying to

! 116! describe an elephant. While we do have scientific basis for measuring the particular things that we measure we fall short of completely understanding the complexity of the interactions, dependencies and dynamic nature of the organism in its environment. In the parable the blind investigators, none of them make inaccurate observations but they all fail to understand or describe the organism. If they were able to effectively communicate with each other and accurately describe all the parts of the elephant and how they fit together, understanding what an elephant is in relation to the rest of the world and what makes an elephant successful is a much more complex matter that requires much more observation. Returning to the world of legume roots, we fail when we reduce and simplify interactions and processes to the point that we ignore not only the drivers of the variation but the fact that it's the variation that makes the system work. As stated above, we need sampling protocols with much higher density and frequency to more effectively guide our selection of when and which phenes to measure. Advanced non-destructive phenotyping techniques such as ground penetrating radar paired with analysis and simulations performed by artificial intelligence may help to resolve the extremely complex dynamic interactions of agro-ecosystems. I propose the following research plan to answer the question of where investigations into root phenotypes goes next. Research should be directed to understand the reason why an effect of multilines is hard to detect, why differentiating genotypes by root phenotypes requires such large sample size, why grouping a single genotype based on its root phenotype in a single season in one field requires elimination of multiple putative outliers, and why stable architectural phenotypes across fields and season is even more challenging. At a later stage, or perhaps even concurrently with the proposed agenda the author would be interested to conduct factorial experiments investigating genotypes with contrasting root architectures and water use efficiency (WUE). This could help to disentangle the WUE vs water acquisition question and possibly identify additive or synergistic effects. I hypothesize that the reason the effect of multilines is not consistently observable is the same reason I have such trouble identifying stable root system architecture (RSA) phenotypes of deep, shallow and dimorphic genotypes, even in a single season and environment. Application of Occom’s razor leads to the conclusion that a single genotype can and does exhibit multiple phenotypes even in a single season and environment. Simply put, the variation is real. It is not an artifact of the experimental system, the field soil or the measurement technique. Root architecture may have intrinsic and dynamic variation in response to multiple environmental factors, between neighbors whether mediated by resource availability or other factors and interactions between those factors and plant phenology. I propose the term multimorphic for this type of RSA phenotypic plasticity and suggest it should be studied as an adaptive strategy. Multimorphic refers to the observation that a single genotype can exhibit various root architectural phenotypes in a single environment. This environment can be a single field, plot and row. The various root architectural phenotypes may be composed by varying the amount of investment in a given root class and in a

! 117! particular root. For example, the number of hypocotyl and basal roots may vary, but basal root number tends to vary much less than hypocotyl root number. Samples from a genotype in a single row may have from between five and fifteen hypocotyl borne roots. The relative investment between hypocotyl, basal and primary root classes may also shift in a single plot. Quantifying that investment would best be accomplished by collecting the entire root length from each class, which is not feasible in a field trial and so we must estimate. Angle of basal roots is also highly variable on both a plot and on an individual plant level. While a given genotype does tend to deploy roots within a range of angle not all roots have the same angel. It is possible that that this characteristic arises because it benefits the plant by avoiding competition with itself or that the differences arise naturally from micro-variations in soil bulk density. Variation in basal root diameter and presumably length can also vary on a single plant. The chance encounter with a bio-pore or easy to follow fracture between soil peds may facilitate the expansion and resource acquisition of an individual root leading to a feedback loop in which it receives more resources because it contributes more due to its greater metabolic activity. Disease, effects of pests and compaction influence root architecture directly and indirectly as the plant responds to the stimuli. It is of course highly plausible to assume that when the primary root, for example, becomes infected with a disease or is damaged by herbivory another root’s relative sink strength increases and in effect acts to compensate for loss of function in the damaged root. This is commonly observable when a basal root seems to behave like a primary root, although the progression of that compensation is unclear. It is also known that hormonal signals resulting from a damaged root tip can lead to increased lateral branching behind that root. Damage to the hypocotyl from bean stem maggot or soil pathogens potential has regularly been observed to be associated with increased hypocotyl root number. Pairing multimorphic individuals in a multiline may be less effective than hypothesized since a given genotype already has divergent RSA and divergent strategies. A multimorphic RSA already limits competitions between the individual and neighbors and potentially co-optimizes deep and shallow exploration. We have failed to appreciate the hypothesis that this variable phenotype deployment is part of a strategy individuals and populations employ to reduce competition, optimize population level performance and ensure reproductive ability of at least some individuals. Exhibiting a consistently plastic and multimorphic root phenotype may not always be beneficial, particularly in low biotic stress environments with predictable soil and resource availability characteristics. In these environments variable deployment of root numbers, root angle and variable investment in different root classes may misallocate resources. If the environment could be anticipated a plant with the ideal root architecture and responses to its environment would deploy exactly the number, type and angle of roots to maximize resource acquisition. However, the number of variables involved in controlling soil texture and its effects on abiotic resource availability and movement as well as biotic

! 118! factors both in and outside of the soil are enormous. While we may be able to acquire the computational power to analyze the data, gathering the data in an effective and meaningful way remains the biggest challenge. In conclusion, we really do need to phenotype the soil, but not just the soil. We need to phenotype and model the entire growth environment. A brief outline to investigate the extent, origin and potential utility of root architectural plasticity is given below. Example work plan Part 1: From Genotype by sequencing (GBS) to whole genotype sequencing and epigenetics A coarse work plan is outlined to frame the proposal to use new technologies to understand the variability inherent in our system. This variation is difficult to approach using current data acquisition, image processing and data analysis tools. New phenotyping and genotyping tools should be implemented to understand the effects of parental provisioning for which GBS would not be appropriate. Emerging forms of small RNAs, histone modification, chromatin reorganizations and DNA methylation mapping may be useful to approach the most primal form of environmental variation, those imparted via maternal effects and “stress memory.” The plan would start with selection of three common bean genotypes (L88 57, DOR364 and SEQ7) from different backgrounds. GBS and or DNA methylation mapping on 100 entries of each genotype would be performed using leaf tissue. Measurements of seed dimensions and weight of each seed along with details on days to emergence and to first trifoliate may help explain subsequent phenotypic variation. If after GBS genetic variation is observed in the large batch of any of the sets of seed that does not align with having a couple foreign seeds mixed in we could argue for non-DNA modification i.e. epigenetics or environmental/ maternal effects or that cross pollination or mutation was involved. Then we could ask the question, do different genotypes have different rates of genetic mutation or different degrees of maternal effects or different types of maternal effects i.e offspring responding differently to the same maternal environment? To investigate this question we could genotype different batches of the same genotype to see what season had the biggest change or contamination or maternal effects. If it looks like maternal effects may be involved then more rigorous genotyping and analysis may be in order to understand the nature of the genetic or epigenetic changes and what these changes could affect. Part 2: From standard multi-location phenotypic comparisons to Artificial Intelligence The project would involve using the common bean genotypes L88 57, DOR 364 and SEQ 7 in all the different trials that I have data and photos for. We would assemble all these photos and pair with distribution plots for all potentially relevant traits. Then we would assemble photos of all of one location, then of one season and one location. Then of one plot in one field in one season. Each

! 119! display would be followed by distribution plots for that scale (collection of seasons-locations-plots). These collections would help the human eye detect patterns, perhaps through the use of standard statistics and plotting procedures. The second stage of this project involves developing a new version of DIRT (DIRTv2) that does a better job of following individual root paths by predicting their trajectory. This version will be accompanied by collection of environmental data in much greater detail and care. Data that would be collected includes centimeter scale measurements of soil bulk density, soil texture, soil temperatures, water and resource availability. We may not yet have the technology needed to collect this type of data. Agronomic parameters must also be collected as well as measuring distance between all plants in a row, making a mark on the stem indicating in-row orientation and taking two 2-D images 90˚ apart to use for analysis of a 3D reconstruction, or at least make 2 analyses of the in-row and between row sides of the plant, which should then be both analyzed individually and combined in some way. It must also be more effective at segmenting the root crown into root classes. The large set of data described above would used for training Artificial Intelligence (AI) on large sets of DIRTv2 analyzed images from ARBC 2015 and 2016. The initial goal would be to determine if genotypes can be differentiated. The questions that we will seek to answer include the following: Can the eye differentiate genotypes? Can shovelomics be used to differentiate the genotypes? Can analysis of single or combined DIRT metrics tell the genotypes apart in any meaningful way that is not related to allometry? Could AI tell the genotypes apart in a given block, in a given experiment, in a given location over multiple seasons? Can the eye or shovelomics or DIRT be used to detect a pattern across seasons in a given location or across locations? Can AI detect a genotypic or environmental pattern? Collecting additional environmental data should be considered to control for variation and help detect patterns including: soil moisture by vertical and horizontal units, soil type, soil texture, tractor paths, cropping history including tillage implements used and direction of tillage, air and soil temperature, relative humidity in and above the canopy and evapotranspiration. With these additions I hope that important new insights can be made on the role of root architecture in abiotic stress tolerance.

! 120! References( Anfodillo,!T.,!Petit,!G.,!Sterck,!F.,!Lechthaler,!S.,!Olson,!M.E.,!2016.!Allometric! Trajectories!and!“Stress”:!A!Quantitative!Approach.!Front.!Plant!Sci.!7,!1681.! doi:10.3389/fpls.2016.01681! Beebe,!S.,!Toro,!O.,!González,!A.!V,!Chacón,!M.I.,!Debouck,!D.G.,!1997.!WildZweedZcrop! complexes!of!common!bean!(Phaseolus!vulgaris!L.,!Fabaceae)!in!the!Andes!of! Peru!and!Colombia,!and!their!implications!for!conservation!and!breeding.!Genet! Resour!Crop!Ev!44.!doi:10.1023/A:1008621632680! Brooker,!R.W.,!Maestre,!F.T.,!Callaway,!R.M.,!Lortie,!C.L.,!Cavieres,!L.A.,!Kunstler,!G.,! Liancourt,!P.,!Tielbörger,!K.,!Travis,!J.M.J.,!Anthelme,!F.,!Armas,!C.,!Coll,!L.,! Corcket,!E.,!Delzon,!S.,!Forey,!E.,!Kikvidze,!Z.,!Olofsson,!J.,!Pugnaire,!F.,!Quiroz,! C.L.,!Saccone,!P.,!Schiffers,!K.,!Seifan,!M.,!Touzard,!B.,!Michalet,!R.,!2008.! Facilitation!in!plant!communities:!the!past,!the!present,!and!the!future.!J.!Ecol.! 96,!18–34.!doi:10.1111/j.1365Z2745.2007.01295.x! Hurlbert,!S.H.,!1984.!Pseudoreplication!and!the!Design!of!Ecological!Field! Experiments.!Ecol.!Monogr.!54,!187–211.! Jordano,!P.,!Bascompte,!J.,!Olesen,!J.M.,!2003.!Invariant!properties!in!coevolutionary! networks!of!plant–animal!interactions.!Ecol.!Lett.!6,!69–81.!doi:10.1046/j.1461Z 0248.2003.00403.x! Kelly,!J.,!2001.!Remaking!bean!plant!architecture!for!efficient!production.!Adv.! Agron.!71,!109–143.!doi:https://doi.org/10.1016/S0065Z2113(01)71013Z9! Kropotkin,!P.,!1902.!Mutual!Aid,!A!Factor!of!Evolution.!William!Heinemann,!London.! Rao,!I.,!Beebe,!S.,!Polania,!J.,!Ricaurte,!J.,!Cajiao,!C.,!Garcia,!R.,!Rivera,!M.,!2013.!Can! Tepary!Bean!be!a!model!for!improvement!of!drought!resistance!in!Common! Bean?!African!Crop!Sci.!J.!21,!265–281.! Weiner,!J.,!2017.!Applying!plant!ecological!knowledge!to!increase!agricultural! sustainability.!J.!Ecol.!105,!865–870.!doi:10.1111/1365Z2745.12792! Zhang,!C.,!Postma,!J.A.,!York,!L.M.,!Lynch,!J.P.,!2014.!Root!foraging!elicits!niche! complementarityZdependent!yield!advantage!in!the!ancient!“three!sisters”! (maize/bean/squash)!polyculture.!Ann.!Bot.!114,!1719–1733.! ! Anfodillo,!T.,!Petit,!G.,!Sterck,!F.,!Lechthaler,!S.,!Olson,!M.E.,!2016.!Allometric! Trajectories!and!“Stress”:!A!Quantitative!Approach.!Front.!Plant!Sci.!7,!1681.! doi:10.3389/fpls.2016.01681!

! 121! Beebe,!S.,!Toro,!O.,!González,!A.!V,!Chacón,!M.I.,!Debouck,!D.G.,!1997.!WildZweedZcrop! complexes!of!common!bean!(Phaseolus!vulgaris!L.,!Fabaceae)!in!the!Andes!of! Peru!and!Colombia,!and!their!implications!for!conservation!and!breeding.!Genet! Resour!Crop!Ev!44.!doi:10.1023/A:1008621632680! Brooker,!R.W.,!Maestre,!F.T.,!Callaway,!R.M.,!Lortie,!C.L.,!Cavieres,!L.A.,!Kunstler,!G.,! Liancourt,!P.,!Tielbörger,!K.,!Travis,!J.M.J.,!Anthelme,!F.,!Armas,!C.,!Coll,!L.,! Corcket,!E.,!Delzon,!S.,!Forey,!E.,!Kikvidze,!Z.,!Olofsson,!J.,!Pugnaire,!F.,!Quiroz,! C.L.,!Saccone,!P.,!Schiffers,!K.,!Seifan,!M.,!Touzard,!B.,!Michalet,!R.,!2008.! Facilitation!in!plant!communities:!the!past,!the!present,!and!the!future.!J.!Ecol.! 96,!18–34.!doi:10.1111/j.1365Z2745.2007.01295.x! Hurlbert,!S.H.,!1984.!Pseudoreplication!and!the!Design!of!Ecological!Field! Experiments.!Ecol.!Monogr.!54,!187–211.! Jordano,!P.,!Bascompte,!J.,!Olesen,!J.M.,!2003.!Invariant!properties!in!coevolutionary! networks!of!plant–animal!interactions.!Ecol.!Lett.!6,!69–81.!doi:10.1046/j.1461Z 0248.2003.00403.x! Kelly,!J.,!2001.!Remaking!bean!plant!architecture!for!efficient!production.!Adv.! Agron.!71,!109–143.!doi:https://doi.org/10.1016/S0065Z2113(01)71013Z9! Kropotkin,!P.,!1902.!Mutual!Aid,!A!Factor!of!Evolution.!William!Heinemann,!London.! Rao,!I.,!Beebe,!S.,!Polania,!J.,!Ricaurte,!J.,!Cajiao,!C.,!Garcia,!R.,!Rivera,!M.,!2013.!Can! Tepary!Bean!be!a!model!for!improvement!of!drought!resistance!in!Common! Bean?!African!Crop!Sci.!J.!21,!265–281.! Weiner,!J.,!2017.!Applying!plant!ecological!knowledge!to!increase!agricultural! sustainability.!J.!Ecol.!105,!865–870.!doi:10.1111/1365Z2745.12792! Zhang,!C.,!Postma,!J.A.,!York,!L.M.,!Lynch,!J.P.,!2014.!Root!foraging!elicits!niche! complementarityZdependent!yield!advantage!in!the!ancient!“three!sisters”! (maize/bean/squash)!polyculture.!Ann.!Bot.!114,!1719–1733.! !

! 122! Vita

James D. Burridge

Professional Preparation University of Dayton B.A. International Studies 2004 Pennsylvania State University M.Ag Horticulture 2008 Pennsylvania State University Ph.D. Plant Science 2017

Professional Experience Program Manager USAID Climate Resilient Bean 2013-present Ph.D Student Pennsylvania State University 2009-present

Statement My focus is to understand the unique adaptations and tradeoffs of legume root architecture to combined low fertility and water limitation. I have explored common bean most extensively while other projects have involved tepary bean, cowpea and groundnut. Collaborative research trials in Zambia, Malawi, Honduras and Colombia with extended research stays in South Africa and Mozambique have broadened my experience and deepened my ability to work in different environments. Managing a USAID Legume Innovation Lab has brought me into closer collaboration with partners worldwide and helped me to develop administrative and time management skills.

Publications Burridge, J. D., Schneider, H. M., Huynh, B. L., Roberts, P. A., Bucksch, A., & Lynch, J. P. (2016). Genome-wide association mapping and agronomic impact of cowpea root architecture. Theoretical and Applied Genetics, 1-13.

James Burridge, Celestina N. Jochua, Alexander Bucksch, Jonathan P. Lynch, Legume shovelomics: High—Throughput phenotyping of common bean (Phaseolus vulgaris L.) and cowpea (Vigna unguiculata subsp, unguiculata) root architecture in the field, Field Crops Research, Volume 192, June 2016, Pages 21-32, ISSN 0378-4290, http://dx.doi.org/10.1016/j.fcr.2016.04.008.

Bucksch, A., Burridge, J., York, L.M., Das, A., Nord, E., Weitz, J.S., Lynch, J.P., 2014. Image-based high-throughput field phenotyping of crop roots. Plant Physiol. 166, 470– 486. doi:10.1104/pp.114.243519

Das, A., Schneider, H., Burridge, J., Ascanio, A. K. M., Wojciechowski, T., Topp, C. N., & Bucksch, A. (2015). Digital imaging of root traits (DIRT): a high-throughput computing and collaboration platform for field-based root phenomics. Plant methods, 11(1), 1. 10.1186/s13007-015-0093-3

Fenta, B. A., Beebe, S. E., Kunert, K. J., Burridge, J. D., Barlow, K. M., Lynch, J. P., & Foyer, C. H. (2014). Field phenotyping of soybean roots for drought stress tolerance. Agronomy, 4(3), 418-435.

McClean Phillip E., Burridge Jimmy, Beebe Stephen, Rao Idupulapati M., Porch Timothy G. (2011) Crop improvement in the era of climate change: an integrated, multi- disciplinary approach for common bean (Phaseolus vulgaris). Functional Plant Biology 38, 927–933.