Delivering trait-enhanced varieties to African smallholders through a pangenomic breeding network

Delivering trait-enhanced varieties to African smallholders through a pangenomic breeding network | bioRxiv /* */ /* */ <!-- <!-- /*! * yepnope1.5.4 * (c) WTFPL, GPLv2 */ (function(a,b,c){function d(a){return"[object Function]"==o.call(a)}function e(a){return"string"==typeof a}function f(){}function g(a){return!a||"loaded"==a||"complete"==a||"uninitialized"==a}function h(){var a=p.shift();q=1,a?a.t?m(function(){("c"==a.t?B.injectCss:B.injectJs)(a.s,0,a.a,a.x,a.e,1)},0):(a(),h()):q=0}function i(a,c,d,e,f,i,j){function k(b){if(!o&&g(l.readyState)&&(u.r=o=1,!q&&h(),l.onload=l.onreadystatechange=null,b)){"img"!=a&&m(function(){t.removeChild(l)},50);for(var d in y[c])y[c].hasOwnProperty(d)&&y[c][d].onload()}}var j=j||B.errorTimeout,l=b.createElement(a),o=0,r=0,u={t:d,s:c,e:f,a:i,x:j};1===y[c]&&(r=1,y[c]=[]),"object"==a?l.data=c:(l.src=c,l.type=a),l.width=l.height="0",l.onerror=l.onload=l.onreadystatechange=function(){k.call(this,r)},p.splice(e,0,u),"img"!=a&&(r||2===y[c]?(t.insertBefore(l,s?null:n),m(k,j)):y[c].push(l))}function j(a,b,c,d,f){return q=0,b=b||"j",e(a)?i("c"==b?v:u,a,b,this.i++,c,d,f):(p.splice(this.i++,0,a),1==p.length&&h()),this}function k(){var a=B;return a.loader={load:j,i:0},a}var l=b.documentElement,m=a.setTimeout,n=b.getElementsByTagName("script")[0],o={}.toString,p=[],q=0,r="MozAppearance"in l.style,s=r&&!!b.createRange().compareNode,t=s?l:n.parentNode,l=a.opera&&"[object Opera]"==o.call(a.opera),l=!!b.attachEvent&&!l,u=r?"object":l?"script":"img",v=l?"script":u,w=Array.isArray||function(a){return"[object Array]"==o.call(a)},x=[],y={},z={timeout:function(a,b){return b.length&&(a.timeout=b[0]),a}},A,B;B=function(a){function b(a){var a=a.split("!"),b=x.length,c=a.pop(),d=a.length,c={url:c,origUrl:c,prefixes:a},e,f,g;for(f=0;f<d;f++)g=a[f].split("="),(e=z[g.shift()])&&(c=e(c,g));for(f=0;f<b;f++)c=x[f](c);return c}function g(a,e,f,g,h){var i=b(a),j=i.autoCallback;i.url.split(".").pop().split("?").shift(),i.bypass||(e&&(e=d(e)?e:e[a]||e[g]||e[a.split("/").pop().split("?")[0]]),i.instead?i.instead(a,e,f,g,h):(y[i.url]?i.noexec=!0:y[i.url]=1,f.load(i.url,i.forceCSS||!i.forceJS&&"css"==i.url.split(".").pop().split("?").shift()?"c":c,i.noexec,i.attrs,i.timeout),(d(e)||d(j))&&f.load(function(){k(),e&&e(i.origUrl,h,g),j&&j(i.origUrl,h,g),y[i.url]=2})))}function h(a,b){function c(a,c){if(a){if(e(a))c||(j=function(){var a=[].slice.call(arguments);k.apply(this,a),l()}),g(a,j,b,0,h);else if(Object(a)===a)for(n in m=function(){var b=0,c;for(c in a)a.hasOwnProperty(c)&&b++;return b}(),a)a.hasOwnProperty(n)&&(!c&&!--m&&(d(j)?j=function(){var a=[].slice.call(arguments);k.apply(this,a),l()}:j[n]=function(a){return function(){var b=[].slice.call(arguments);a&&a.apply(this,b),l()}}(k[n])),g(a[n],j,b,n,h))}else!c&&l()}var h=!!a.test,i=a.load||a.both,j=a.callback||f,k=j,l=a.complete||f,m,n;c(h?a.yep:a.nope,!!i),i&&c(i)}var i,j,l=this.yepnope.loader;if(e(a))g(a,0,l,0);else if(w(a))for(i=0;i (function(w,d,s,l,i){w[l]=w[l]||[];w[l].push({'gtm.start':new Date().getTime(),event:'gtm.js'});var f=d.getElementsByTagName(s)[0];var j=d.createElement(s);var dl=l!='dataLayer'?'&l='+l:'';j.src='//www.googletagmanager.com/gtm.js?id='+i+dl;j.type='text/javascript';j.async=true;f.parentNode.insertBefore(j,f);})(window,document,'script','dataLayer','GTM-M677548'); Skip to main content Home About Submit ALERTS / RSS Search for this keyword Advanced Search New Results Delivering trait-enhanced varieties to African smallholders through a pangenomic breeding network View ORCID Profile Fanna Maina , View ORCID Profile Jacques M. Faye , View ORCID Profile Alex W. Kena , View ORCID Profile Cyril Diatta , Moctar Issiakou Tankari , View ORCID Profile Ousseini Abdou Ardaly , Ousmane Seyni Diakite , Aissata Mamadou Ibrahim , View ORCID Profile Oumarou Abdoulye Moussa , View ORCID Profile Abdou Harou , View ORCID Profile Rabiou Abdou , View ORCID Profile Joseph Pascal Sene , Souleymane Bodian , Mbery Ndour , Diarietou Sambakhe , View ORCID Profile Bassirou Sine , View ORCID Profile Pana Kadanga , View ORCID Profile Eyanawa Akata , Nofou Ouedraogo , View ORCID Profile Benjamin Annor , View ORCID Profile Israel T. Tetteh , View ORCID Profile Muhammad Ahmad Yahaya , Abdoulaye Diallo , View ORCID Profile Ronald Kakeeto , View ORCID Profile Rachael K. Kisilu , View ORCID Profile Clarisse Pulchérie Kondombo , Tokuma Legesse , View ORCID Profile Lloyd Mbulwe , Joaquim Mutaliano , View ORCID Profile Emmanuel T. Mwenda , View ORCID Profile Gapili Naoura , View ORCID Profile Hortense Noëlle Apala Mafouasson , View ORCID Profile Kenneth Opare-Obuobi , View ORCID Profile Rekiya Otuchu Abdulmalik , View ORCID Profile Steven Runo , View ORCID Profile Tilal Sayed Abdelhalim , View ORCID Profile Louis Yalaukani , View ORCID Profile Joel Masanga , View ORCID Profile Chloee M. McLaughlin , View ORCID Profile Jesse R. Lasky , View ORCID Profile Avril M. Harder , View ORCID Profile Adam L. Healey , View ORCID Profile John T. Lovell , Portia Osei-Obeng , View ORCID Profile Sandeep R. Marla , View ORCID Profile Terry J. Felderhoff , Falalou Hamidou , View ORCID Profile Daniel Fonceka , View ORCID Profile Mohan Kumar Varma Chejerla , Baloua Nébié , View ORCID Profile Mark Nas Tamerlan , View ORCID Profile Amos Alakonya , View ORCID Profile Abhishek Rathore , Santosh Deshpande , View ORCID Profile Harish Gandhi , View ORCID Profile Linly Banda , View ORCID Profile Davina H. Rhodes , View ORCID Profile Clara Cruet-Burgos , View ORCID Profile Carl J. VanGessel , View ORCID Profile Geoffrey P. Morris doi: https://doi.org/10.1101/2025.08.07.667917 Fanna Maina 1 Institut National de la Recherche Agronomique du Niger , Niamey, Niger Find this author on Google Scholar Find this author on PubMed Search for this author on this site ORCID record for Fanna Maina For correspondence: fannamaina{at}inran.ne j.faye{at}cgiar.org Geoff.Morris{at}colostate.edu Jacques M. Faye 2 Institut Sénégalais de Recherches Agricoles , Thiès, Sénégal 3 International Maize and Wheat Improvement Center (CIMMYT) , Nairobi, Kenya Find this author on Google Scholar Find this author on PubMed Search for this author on this site ORCID record for Jacques M. Faye For correspondence: fannamaina{at}inran.ne j.faye{at}cgiar.org Geoff.Morris{at}colostate.edu Alex W. Kena 4 Department of Crop and Soil Sciences, Kwame Nkrumah University of Science and Technology , Kumasi, Ashanti, Ghana 5 Department of Soil & Crop Science, Colorado State University , Fort Collins, CO, USA Find this author on Google Scholar Find this author on PubMed Search for this author on this site ORCID record for Alex W. Kena Cyril Diatta 2 Institut Sénégalais de Recherches Agricoles , Thiès, Sénégal Find this author on Google Scholar Find this author on PubMed Search for this author on this site ORCID record for Cyril Diatta Moctar Issiakou Tankari 1 Institut National de la Recherche Agronomique du Niger , Niamey, Niger Find this author on Google Scholar Find this author on PubMed Search for this author on this site Ousseini Abdou Ardaly 1 Institut National de la Recherche Agronomique du Niger , Niamey, Niger Find this author on Google Scholar Find this author on PubMed Search for this author on this site ORCID record for Ousseini Abdou Ardaly Ousmane Seyni Diakite 1 Institut National de la Recherche Agronomique du Niger , Niamey, Niger Find this author on Google Scholar Find this author on PubMed Search for this author on this site Aissata Mamadou Ibrahim 1 Institut National de la Recherche Agronomique du Niger , Niamey, Niger Find this author on Google Scholar Find this author on PubMed Search for this author on this site Oumarou Abdoulye Moussa 6 Université André Salifou , Zinder, Niger Find this author on Google Scholar Find this author on PubMed Search for this author on this site ORCID record for Oumarou Abdoulye Moussa Abdou Harou 6 Université André Salifou , Zinder, Niger Find this author on Google Scholar Find this author on PubMed Search for this author on this site ORCID record for Abdou Harou Rabiou Abdou 6 Université André Salifou , Zinder, Niger Find this author on Google Scholar Find this author on PubMed Search for this author on this site ORCID record for Rabiou Abdou Joseph Pascal Sene 2 Institut Sénégalais de Recherches Agricoles , Thiès, Sénégal 7 West Africa Centre for Crop Improvement, University of Ghana , Accra, Ghana Find this author on Google Scholar Find this author on PubMed Search for this author on this site ORCID record for Joseph Pascal Sene Souleymane Bodian 2 Institut Sénégalais de Recherches Agricoles , Thiès, Sénégal Find this author on Google Scholar Find this author on PubMed Search for this author on this site Mbery Ndour 2 Institut Sénégalais de Recherches Agricoles , Thiès, Sénégal Find this author on Google Scholar Find this author on PubMed Search for this author on this site Diarietou Sambakhe 2 Institut Sénégalais de Recherches Agricoles , Thiès, Sénégal Find this author on Google Scholar Find this author on PubMed Search for this author on this site Bassirou Sine 2 Institut Sénégalais de Recherches Agricoles , Thiès, Sénégal Find this author on Google Scholar Find this author on PubMed Search for this author on this site ORCID record for Bassirou Sine Pana Kadanga 8 Institut Togolais de Recherche Agronomique , Lomé, Togo Find this author on Google Scholar Find this author on PubMed Search for this author on this site ORCID record for Pana Kadanga Eyanawa Akata 8 Institut Togolais de Recherche Agronomique , Lomé, Togo Find this author on Google Scholar Find this author on PubMed Search for this author on this site ORCID record for Eyanawa Akata Nofou Ouedraogo 9 Institut de l’Environnement et de Recherches Agricoles , Ouagadougou, Burkina Faso Find this author on Google Scholar Find this author on PubMed Search for this author on this site Benjamin Annor 4 Department of Crop and Soil Sciences, Kwame Nkrumah University of Science and Technology , Kumasi, Ashanti, Ghana Find this author on Google Scholar Find this author on PubMed Search for this author on this site ORCID record for Benjamin Annor Israel T. Tetteh 4 Department of Crop and Soil Sciences, Kwame Nkrumah University of Science and Technology , Kumasi, Ashanti, Ghana Find this author on Google Scholar Find this author on PubMed Search for this author on this site ORCID record for Israel T. Tetteh Muhammad Ahmad Yahaya 10 Institute for Agricultural Research, Samaru, Ahmadu Bello University , Zaria, Nigeria Find this author on Google Scholar Find this author on PubMed Search for this author on this site ORCID record for Muhammad Ahmad Yahaya Abdoulaye Diallo 11 Institut d’Economie Rurale, Centre Régional de Recherche Agronomique , Sotuba, Mali Find this author on Google Scholar Find this author on PubMed Search for this author on this site Ronald Kakeeto 12 National Agricultural Research Organization, National Semi-Arid Resources Research Institute , Soroti, Uganda Find this author on Google Scholar Find this author on PubMed Search for this author on this site ORCID record for Ronald Kakeeto Rachael K. Kisilu 13 Kenya Agricultural and Livestock Research Organisation , Kenya Find this author on Google Scholar Find this author on PubMed Search for this author on this site ORCID record for Rachael K. Kisilu Clarisse Pulchérie Kondombo 9 Institut de l’Environnement et de Recherches Agricoles , Ouagadougou, Burkina Faso Find this author on Google Scholar Find this author on PubMed Search for this author on this site ORCID record for Clarisse Pulchérie Kondombo Tokuma Legesse 14 Ethiopian Institute of Agricultural Research, Melkassa Agricultural Research Center , Melkasa, Ethiopia Find this author on Google Scholar Find this author on PubMed Search for this author on this site Lloyd Mbulwe 15 Zambia Agriculture Research Institute , Lusaka, Zambia Find this author on Google Scholar Find this author on PubMed Search for this author on this site ORCID record for Lloyd Mbulwe Joaquim Mutaliano 16 Instituto de Investigação Agrária de Moçambique , Ulónguè, Tete, Moçambique Find this author on Google Scholar Find this author on PubMed Search for this author on this site Emmanuel T. Mwenda 17 Tanzania Agricultural Research Institute , Dodoma, Tanzania Find this author on Google Scholar Find this author on PubMed Search for this author on this site ORCID record for Emmanuel T. Mwenda Gapili Naoura 18 Institut Tchadien de Recherche Agronomique pour le Développement , République du Tchad Find this author on Google Scholar Find this author on PubMed Search for this author on this site ORCID record for Gapili Naoura Hortense Noëlle Apala Mafouasson 19 Institute of Agricultural Research for Development , Yaoundé, Cameroon Find this author on Google Scholar Find this author on PubMed Search for this author on this site ORCID record for Hortense Noëlle Apala Mafouasson Kenneth Opare-Obuobi 20 Council for Scientific and Industrial Research-Savanna Agricultural Research Institute , Nyankpala/Tamale, Ghana Find this author on Google Scholar Find this author on PubMed Search for this author on this site ORCID record for Kenneth Opare-Obuobi Rekiya Otuchu Abdulmalik 10 Institute for Agricultural Research, Samaru, Ahmadu Bello University , Zaria, Nigeria Find this author on Google Scholar Find this author on PubMed Search for this author on this site ORCID record for Rekiya Otuchu Abdulmalik Steven Runo 21 Department of Biochemistry, Microbiology, and Biotechnology, Kenyatta University , Nairobi, Kenya Find this author on Google Scholar Find this author on PubMed Search for this author on this site ORCID record for Steven Runo Tilal Sayed Abdelhalim 22 Biotechnology and Biosafety Research Center, Agricultural Research Corporation , Khartoum North, Sudan Find this author on Google Scholar Find this author on PubMed Search for this author on this site ORCID record for Tilal Sayed Abdelhalim Louis Yalaukani 23 Department of Agricultural Research Services , Lilongwe, Malawi Find this author on Google Scholar Find this author on PubMed Search for this author on this site ORCID record for Louis Yalaukani Joel Masanga 24 Department of Biology, Pennsylvania State University , University Park, PA, USA Find this author on Google Scholar Find this author on PubMed Search for this author on this site ORCID record for Joel Masanga Chloee M. McLaughlin 24 Department of Biology, Pennsylvania State University , University Park, PA, USA 25 Genome Sequencing Center, HudsonAlpha Institute for Biotechnology , Huntsville, AL, USA 26 Department of Horticulture and Landscape Architecture, Colorado State University , Fort Collins, CO, USA Find this author on Google Scholar Find this author on PubMed Search for this author on this site ORCID record for Chloee M. McLaughlin Jesse R. Lasky 24 Department of Biology, Pennsylvania State University , University Park, PA, USA Find this author on Google Scholar Find this author on PubMed Search for this author on this site ORCID record for Jesse R. Lasky Avril M. Harder 25 Genome Sequencing Center, HudsonAlpha Institute for Biotechnology , Huntsville, AL, USA Find this author on Google Scholar Find this author on PubMed Search for this author on this site ORCID record for Avril M. Harder Adam L. Healey 25 Genome Sequencing Center, HudsonAlpha Institute for Biotechnology , Huntsville, AL, USA Find this author on Google Scholar Find this author on PubMed Search for this author on this site ORCID record for Adam L. Healey John T. Lovell 25 Genome Sequencing Center, HudsonAlpha Institute for Biotechnology , Huntsville, AL, USA 27 US Department of Energy Joint Genome Institute , Berkeley, CA, USA Find this author on Google Scholar Find this author on PubMed Search for this author on this site ORCID record for John T. Lovell Portia Osei-Obeng 28 Department of Agronomy, Kansas State University , Manhattan, KS, USA Find this author on Google Scholar Find this author on PubMed Search for this author on this site Sandeep R. Marla 28 Department of Agronomy, Kansas State University , Manhattan, KS, USA Find this author on Google Scholar Find this author on PubMed Search for this author on this site ORCID record for Sandeep R. Marla Terry J. Felderhoff 28 Department of Agronomy, Kansas State University , Manhattan, KS, USA Find this author on Google Scholar Find this author on PubMed Search for this author on this site ORCID record for Terry J. Felderhoff Falalou Hamidou 29 International Crops Research Institute for the Semi-Arid Tropics – Sahelian Center , Niamey, Niger Find this author on Google Scholar Find this author on PubMed Search for this author on this site Daniel Fonceka 30 Centre de coopération internationale en recherche agronomique pour le développement , Montpellier, France Find this author on Google Scholar Find this author on PubMed Search for this author on this site ORCID record for Daniel Fonceka Mohan Kumar Varma Chejerla 3 International Maize and Wheat Improvement Center (CIMMYT) , Nairobi, Kenya Find this author on Google Scholar Find this author on PubMed Search for this author on this site ORCID record for Mohan Kumar Varma Chejerla Baloua Nébié 3 International Maize and Wheat Improvement Center (CIMMYT) , Nairobi, Kenya Find this author on Google Scholar Find this author on PubMed Search for this author on this site Mark Nas Tamerlan 3 International Maize and Wheat Improvement Center (CIMMYT) , Nairobi, Kenya Find this author on Google Scholar Find this author on PubMed Search for this author on this site ORCID record for Mark Nas Tamerlan Amos Alakonya 3 International Maize and Wheat Improvement Center (CIMMYT) , Nairobi, Kenya Find this author on Google Scholar Find this author on PubMed Search for this author on this site ORCID record for Amos Alakonya Abhishek Rathore 3 International Maize and Wheat Improvement Center (CIMMYT) , Nairobi, Kenya Find this author on Google Scholar Find this author on PubMed Search for this author on this site ORCID record for Abhishek Rathore Santosh Deshpande 31 International Crops Research Institute for the Semi-Arid Tropics , Patancheru, Telangana, India Find this author on Google Scholar Find this author on PubMed Search for this author on this site Harish Gandhi 3 International Maize and Wheat Improvement Center (CIMMYT) , Nairobi, Kenya 31 International Crops Research Institute for the Semi-Arid Tropics , Patancheru, Telangana, India Find this author on Google Scholar Find this author on PubMed Search for this author on this site ORCID record for Harish Gandhi Linly Banda 26 Department of Horticulture and Landscape Architecture, Colorado State University , Fort Collins, CO, USA Find this author on Google Scholar Find this author on PubMed Search for this author on this site ORCID record for Linly Banda Davina H. Rhodes 26 Department of Horticulture and Landscape Architecture, Colorado State University , Fort Collins, CO, USA Find this author on Google Scholar Find this author on PubMed Search for this author on this site ORCID record for Davina H. Rhodes Clara Cruet-Burgos 5 Department of Soil & Crop Science, Colorado State University , Fort Collins, CO, USA Find this author on Google Scholar Find this author on PubMed Search for this author on this site ORCID record for Clara Cruet-Burgos Carl J. VanGessel 5 Department of Soil & Crop Science, Colorado State University , Fort Collins, CO, USA Find this author on Google Scholar Find this author on PubMed Search for this author on this site ORCID record for Carl J. VanGessel Geoffrey P. Morris 5 Department of Soil & Crop Science, Colorado State University , Fort Collins, CO, USA Find this author on Google Scholar Find this author on PubMed Search for this author on this site ORCID record for Geoffrey P. Morris For correspondence: fannamaina{at}inran.ne j.faye{at}cgiar.org Geoff.Morris{at}colostate.edu Abstract Full Text Info/History Metrics Supplementary material Data/Code Preview PDF ABSTRACT Pangenomics has been promoted to accelerate breeding of orphan crops, but smallholder farmers in developing nations have seen little benefit so far. To address this gap, we built a global pangenomic breeding network, integrating African breeding programs, U.S. land grant universities, and international nonprofit research organizations. Here we demonstrate that pangenomics, when integrated with local crop improvement knowledge and global scientific partnerships, can facilitate breeding of drought and pest resilient varieties for smallholders. To breed trait-enhanced sorghum varieties with lgs1 - 1 resistance to witchweed ( Striga hermonthica) for smallholders in Niger, one of the world’s least developed nations, we used population genomics across local and global scales to develop lgs1 - 1 Striga resistance markers, and deployed them for rapid introgression of resistance into locally-preferred varieties. Genomic characterization, along with controlled experiments in laboratory, pot, field stations, and smallholder farms, confirmed lgs1 - 1 resistance was introgressed without loss of essential local-preference traits. New pangenomic resources, including global resequencing and graph pangenomes, further accelerated design of broadly-applicable markers. Unlocking the potential of pangenomics for stress-resilience breeding depended on stakeholder input, strong inference, South-led decision support software, and a dense collaborative network. The experience of the network provides a scalable roadmap for collaborative pangenomic breeding of trait-enhanced varieties for the world’s lowest-resourced farmers. Increasing the productivity and resilience of smallholder farming has been one of the most reliable and cost-effective ways of unlocking human flourishing over the past century 1 . Unfortunately, more than a billion people, mostly in sub-Saharan Africa, have yet to access benefits of improved agricultural production 1 , 2 . In response, major investments have been made in orphan crop genomics, but there has been little evidence of benefits to smallholders so far 3 – 5 . One approach to achieve the benefits of orphan crop genomics is empowering developing-country plant breeders to lead genomics-enabled crop improvement initiatives, as these scientists are best equipped to bridge local knowledge of smallholder communities with research capacity of the global scientific community 4 , 6 , 7 ( Fig. 1a ). The pangenomes of orphan crops harbor abundant diversity for useful traits and signatures of selection to guide crop improvement. Successive improvements in surveys of orphan crop pangenomes 3 , 5 —from genotyping-by-sequencing (GBS; ∼0.1–1% of SNPs), to whole-genome resequencing (∼100% of SNPs), and multiple de novo reference genomes (SNPs and structural variants)—have further enhanced the potential value of bridging pangenomics and breeding in developing countries. Download figure Open in new tab Fig. 1 Design and implementation of a global pangenomic breeding network for African smallholder farmers. a , Framework for a global pangenomic breeding network that meets smallholder farmer needs. African smallholder farmers and farmer representatives (left; smallholders in a Striga infested sorghum field) provide trait demand to African breeding programs (center; African breeders at INRAN’s Striga resistance breeding nursery) who establish collaborations with global scientific partners (right; Striga resistance testing in a US partner laboratory); the collaborations lead to development of trait technology packages for breeding and, finally, to trait-enhanced varieties for African smallholders (leftward arrows). b , The global pangenomic breeding network we have developed for African sorghum improvement. The network links African NARI with limited facilities to other NARI with extensive facilities and global research organizations (inset). The network bridges the West (green shades) and East African (yellow) sorghum belts, with founding NARI hubs (dark green) and early scaling partners (medium green) noted. CIMMYT research hubs, which support the NARS network, in West (“W”) and East (“E”) Africa, and US partners, are noted (inset). c , Pangenomic breeding plan to introgress lgs1-1 Striga resistance into locally-preferred varieties from a variety that harbors resistance but was not adopted by smallholder farmers. The recipient varieties span a range of genetic backgrounds targeted to a range of agricultural systems. d , Formalized applied science framework adopted in the network, linking goal-setting logic chains and scientific method logic chains (extending the strong inference approach 32 ; Extended Data Fig. 2 ). (Photo credits: G.P.M., G.P.M., F.M.) National agricultural research systems (NARS) in African countries are anchored by breeding programs at national agricultural research institutes (NARI; Extended Data Table 1 ), and bolstered by partnerships with local universities, farmer organizations, local seed companies, and grain processors. NARI breeding programs face an exceptionally difficult task: improving desired traits (under directional selection; e.g. yield, pest resistance, drought tolerance) and recovering acquired traits (under stabilizing selection; e.g. flowering time, height, pigmentation), without established elite breeding populations 6 , 8 – 10 and with few resources (e.g. 0.3–1.9 scientists per million tons of production in major sorghum producing countries) 11 . Failure to recover a single “must-have” trait, or to transfer the “winning trait” from a resistance donor, can cause rejection of new varieties by farmers 11 . We built our network around South-South (i.e. NARS-NARS) partnerships ( Fig. 1b ; Extended Data Fig. 1 ) to leverage complementary strengths among NARS. Contributors at U.S. land grant universities and nonprofit research organizations, including CGIAR centers (CIMMYT and ICRISAT), helped build network capacity (molecular breeding, pangenomics, strong inference) and supported development of trait technology packages (trait markers, adapted donor lines, and genotype-phenotype maps) ( Fig. 1a,b ; Extended Data Table 1 ). Outsourced marker genotyping platforms (such as Kompetitive Allele-Specific PCR; KASP) provide NARI breeders in the network the opportunity to deploy marker-assisted selection (MAS) 12 , 13 using markers that distill pangenomic discoveries into tractable tools. Download figure Open in new tab Extended Data Fig. 1 PCR markers are accurate in the INRAN breeding program, but are not tractable for MAS in Niger. a , INRAN attempted to introgress LGS resistance into locally-preferred varieties in the 2000s 22 , but was not completed due to various resource constraints. We relaunched the program, confirming the LGS1 status of INRAN breeding materials using two previously developed primers sets 29 were adapted for a duplex reaction. G2133 (non-deletion target) represents the common primer that amplifies all samples while PDstrigalgs5b (deletion target) represents the LGS1 deletion primer that amplifies samples without deletion at LGS1 gene (Sobic.005G213600). Ladder: 500 bp bands ladder; No DNA: negative control. BTx623, Mota Maradi, El Mota: Striga susceptible genotypes; SRN39: Striga resistant genotype. This experiment was done by F.M. during PhD training at Kansas State University (right panel; Photo: G.M.). b , INRAN-Niger is an example of an African NARI with limited capacity for molecular biological assays. At the launch of this initiative, the INRAN laboratory (left panel, Photo: G.M.) had some basic lab equipment (i.e. refrigerator, water bath, weigh scale, etc) but lacked standard molecular biology equipment (i.e. microcentrifuges, pipettors, spectrophotometer, PCR machine, gel electrophoresis system, ultra-low freezers) necessary for DNA extraction and PCR marker analysis and power outages were frequent at the facility and necessitated a battery backup system (right panel). c , ISRA-CERAAS in Senegal is an example of a NARI with high capacity with a dedicated molecular laboratory (Photo: G.M.) and other laboratories (for biochemistry, ecophysiology, nutritional analysis, bioinformatics, tissue culture, etc.). d , Recent investments (TWAS-UNESCO, FAO) have enhanced laboratory capacity at INRAN (e.g. microcentrifuge, pipettors, spectrophotometer, microscope, screenhouse; Photo: F.M.) so that INRAN scientists can contribute lab research activities within the pangenomic breeding network, in addition to critical field studies, such as those with local smallholders and/or e , local seed enterprises (AINOMA seed company trial, 2025; Photo: Aichatou Nasser). View this table: View inline View popup Download powerpoint Extended Data Table 1: Institutions participating in pangenomic breeding network We sought to address the gap between pangenomic resources and smallholder needs for sorghum ( Sorghum bicolor [L.] Moench), a multipurpose cereal crop (grain, forage, fiber) of smallholder farmers in Africa’s dry regions (semi-arid and subhumid savannah zones) 6 , 14 . A complex geographic mosaic of adaptation to climate, stressors, and cultures 15 , 16 in this indigenous African crop provides abundant diversity for crop improvement, but hinders Green Revolution-style mega-varieties 11 , 14 . The top constraints on sorghum across Africa are drought 14 and witchweed ( Striga hermonthica ), an obligate hemiparasite of sorghum and other cereals in Africa that causes up to 50% yield losses or field abandonment 17 . Thus, developing drought and Striga resilient varieties is a top breeding program priority across Africa 17 , 18 (Supplementary Tables 1-4). Striga requires root-exuded host strigolactones to stimulate germination 19 , 20 and in the 1950s low germination stimulant (LGS) resistance was discovered in a West African landrace (Framida) 19 . Unfortunately, limited agroclimatic adaptation and/or non-preferred end-use traits of existing Striga -resistant varieties has hindered adoption (e.g. <0.1% adoption of Framida and Framida-derivative SRN39) 11 , 21 , 22 . Deploying our African-led GBS mapping of genomic diversity 23 – 27 , and leveraging other groups’ trait discoveries (mapping of stay-green Stg drought tolerance quantitative trait loci [QTL] 28 and cloning of the lgs1 Striga resistance gene 29 ), we launched a pangenomic breeding network for African smallholders ( Fig. 1b ). We targeted drought and Striga -resilient sorghum as initial cases to build, test, and scale the network ( Fig. 1c ). To address lessons on product delivery 30 and technology scaling 31 failures in global development, we used a framework linking formal breeder goal-setting and strong inference 32 scientific method ( Fig. 1d ; Extended Data Fig. 2 ). Here, we illustrate the approach for breeding of Stg drought tolerance ( Extended Data Fig. 3 ) and lgs1 Striga resistance (detailed below). Download figure Open in new tab Extended Data Fig. 2 Framework for formalization of goal setting and hypothesis testing. a , An example of logic chains that formally link breeder goal-setting and strong inference scientific methods. b , Rationale for each step and description of how each step is completed. The goal-setting logic chain (upper five steps) adapts one provided by hfp consulting (Heidelberg, Germany), adding the distinction between “development goals” (for technology) and “research goals” (for knowledge), and adding the “roadblock” identification to help breeders remain focused on research activities that are essential to breeding goals. When the breeding team believes no roadblock exists, the roadblock and research goal may be omitted, such that development goals are linked directly to hypothesis testing. The development goal or research goal is linked directly to hypothesis testing. The hypothesis-testing logic chain (lower four steps, with branching structure) is adapted from the strong inference approach to the scientific method 32 , 71 , which emphasizes the use of multiple working hypotheses, the awareness of the tendency to favor some hypotheses, and the need to design “crucial” (decisive) experiments. Our approach extends strong inference by (1) explicitly naming the “favored” and “unfavored” hypotheses, so that the team can address potential biases during experimental design, and (2) facilitating the design of crucial experiments by explicitly defining predictions under each hypothesis. Templates and training materials are available at www.gohy.org . Download figure Open in new tab Extended Data Fig. 3 Pangenome-enabled breeding for staygreen drought tolerance at ISRA-Senegal and INERA-Burkina Faso. Marker assisted selection for staygreen alleles ( Stg1 , Stg3A , Stg3B , and Stg5 ) in locally-preferred varieties and managed drought phenotyping in Senegal and Burkina Faso. a , Steps of marker-assisted selection (MAS) and marker-based quality control (MQC) with Stg KASP markers. b , Stg5 KASP marker (S1_01147523|v3.1, snpSB00540) scatterplot distinguishing segregating alleles in ISRA germplasm. c , Boxplot of drought response (SPAD) for BC 1 F 5 and BC 3 F 5 plants at physiological maturity conducted in a screenhouse at CERAAS-Thies in 2024. BC 1 F 5 and BC 3 F 5 lines were phenotype under well-watered (WW) and water-stressed pots (WS). The letters above indicate Tukey’s HSD. The number of Stg alleles, determined by KASP genotyping, indicates any combination of Stg1 , Stg3A , Stg3B , and Stg5 and demonstrated staygreen trait introgression to elite varieties. d , Genome-wide association study of drought stress index for grain weight identified loci contributing drought tolerance in the field at CNRA-Bambey during the 2023 off-season (February-June). Introgression lines from three backgrounds (Golobe, Nganda, Darou) were used, confirming the role of Stg and other loci in staygreen trait introgression. e , Post-flowering drought stress phenotyping of introgression lines in pots at CERAAS-Thies research station in 2024, data shown in panel c. f , Managed drought stress phenotyping representative of data in panel d at CNRA-Bambey during the 2022 late rainy season (Nov 2022-March 2023). BC 1 F 3 and BC 3 F 3 plants reached grain fill and maturity under post-flowering drought. g , Managed drought stress phenotyping at INERA-Kamboinse research station during the late growing season of 2022. Phenotyped RILs were enriched for Stg alleles by MAS. Developing markers using population genomics Striga resistance can be conferred by large structural variants ( lgs1-1 ; 34 kb, lgs1-2 , 29 kb; lgs1-3, 30 kb) that delete the sulfotransferase gene LGS1 (Sobic.005G213600) 29 ( Fig. 2a ). A previously developed lgs1 PCR marker (for lgs1-1 in SRN39) discriminates resistant and susceptible alleles in the INRAN-Niger breeding program, but was not tractable for MAS in Niger due to a lack of equipment and reliable power ( Extended Data Fig. 1 ; at network launch Niger was ranked 188th of 188 on UN Human Development Index) 33 . We reasoned we could develop an outsourced KASP marker for lgs1 , suitable for MAS at INRAN and other programs in the region, by identifying GBS SNPs in linkage disequilibrium (LD) with lgs1 deletions. Lines carrying an lgs1 deletion allele should have mostly missing data (80–100%) in the lgs1 deletion region (based on 21 GBS SNPs in the interval), while LGS1 genotypes should not. GBS-based inference suggested a high frequency of deletion alleles (e.g. 7% and 9% for initial focal countries of Niger and Senegal, respectively) and included known resistant genotypes, however missing SNPs may be due to low coverage (an inherent limitation of GBS 34 ) rather than lgs1 deletion (permutation test: 3% false positive rate; Fig. 2b ). Global germplasm ( N = 1724) 29 inferred to have large deletions indicated 224 African landraces are possible lgs1 carriers ( Fig. 2c ). Download figure Open in new tab Fig. 2 Population genomics analyses to develop an lgs1 marker to transfer Striga resistance. a , Genomic characterization of LGS1 region on Chr05 using genotyping-by-sequencing (GBS) SNPs. LGS1 region (69.40–70.02 Mb|v3.1) includes the lgs1-1 deletion region in SRN39, encompassing 5 genes including Sobic.005G213600 ( LGS1 gene), as well as the upstream (UFR, 69.40–69.97 Mb|v3.1) and downstream (DFR, 70.01–70.02 Mb|v3.1) flanking regions. b , Permutation test to estimate frequency of false positives using GBS-based inference of lgs1 deletion (21 missing GBS SNPs). c, Geographic origin of georeferenced accessions in Africa and West Africa color-coded based on the hypothesized LGS1 wild-type (blue) and lgs1 deletion (red, validated with known resistant genotypes SRN39: PI656027, 54.K.94: PI533752, and Framiola: PI533976). Deletion of LGS1 was inferred based on the number of missing GBS SNPs within the “ LGS1 region” (69.97–70.011 Mb). d , One-dimensional linkage disequilibrium analysis ( R 2 and D ’) for GNS-inferred lgs1 allele versus SNP alleles in the extended LGS1 region ( lgs1-1 deletion, UFR, DFR); orange stars indicate the position of single nucleotide polymorphisms (SNPs) tested for marker development. e , Identification of a SNP with perfect LD for the INRAN breeding program. f , KASP genotyping result for lgs1-1 marker (snpSB00246) on Striga susceptible (BTx623, El Mota, IRAT204, Mota Maradi, MR732) and resistant (SRN39) genotypes. Expected KASP calls for each sample are represented by upward triangles (predicted A:A), downward triangles (predicted G:G), diamonds (predicted A:G), and circles (Unknown or no a priori information) are a priori hypotheses based on known Striga resistance or susceptibility. To identify GBS SNPs in LD with lgs1 deletions, suitable for development of KASP markers, we calculated R 2 and D ’ for SNPs ( N = 342) in the genomic regions flanking the lgsl-1 deletions (UFR: 577 kb, DFR: 100 kb) across global sorghum germplasm ( N = 2,599) ( Fig. 2d , Extended Data Fig. 4 ). The D ’ ranged widely but was high for some SNPs (0–0.73), indicating some SNPs could be locally predictive of lgs1 deletions. However, R 2 of SNPs versus the GBS-inferred lgs1 deletion were all low (0–0.06), suggesting that no SNP can be converted into a globally-predictive lgs1 marker. We advanced 10 SNPs with contrasting alleles for parental genotypes to KASP marker conversion at our outsourced genotyping provider (Intertek Agritech, Sweden), and found eight markers that passed ‘SNP Quality Assessment’ (Supplementary Table 5). Finally, we advanced a single marker (snpSB00246; S5_69450954|v3.1), assaying a SNP at 535 kb upstream of lgs1 that was in perfect LD ( R 2 = D ’ = 1) for resistant ( N = 4) versus susceptible genotypes ( N = 9) within the INRAN program ( Fig. 2e , Extended Data Fig. 4 ). Due to the low R 2 of the target SNP with the GBS-based lgs1 deletion calls, this marker would not be expected to be trait-predictive in other global breeding programs. However, based on the high D ’ among INRAN’s parent lines, this marker should be locally trait-predictive and suitable to advance a marker-assisted backcross (MABC) program. Download figure Open in new tab Extended Data Fig. 4 Workflow for development of outsourced KASP marker. Description of steps towards development of KASP markers based on linkage disequilibrium of SNPs with GBS-inferred lgs1 deletion calls. Testing the KASP marker, we found genotyping calls for snpSB00246 were consistent with the expected allele for known susceptible and resistant genotypes ( Fig. 2f ). Further, for the artificial heterozygote samples, almost all samples (11/12) were called correctly as A:G heterozygotes, validating the marker for selection of heterozygotes during MABC. We validated the markers for MAS in two biparental breeding families, observing consistent SNP calling across technical replicates of parents and F 3 progenies (Supplementary Table 6). Note, given that weak clustering and some erroneous genotype calls were observed with snpSB00246, we later designed a reverse strand marker (snpSB00487, at the same SNP), which has better genotypic discrimination ( Extended Data Fig. 5 ). Further, we developed decision support software (panGenomeBreedr) 35 designed by developing-country molecular breeders (A.W.K. and others) to facilitate marker design and MAS. The functions incorporate strong inference principles for marker development and deployment, using color- and symbol-coding based on a priori hypotheses and predictions to highlight discrepancies in controls (i.e. known homozygotes of each class, artificial heterozygotes) ( Fig. 2f , Extended Data Fig. 6 ). Therefore, panGenomeBreedr quickly and clearly identifies failures in marker development, genotyping processes, or germplasm management, to drive improvement of breeding operations. Download figure Open in new tab Extended Data Fig. 5 KASP marker generated on the reverse strand improved genotyping calling. a , KASP marker snpSB00246, generated on the forward strand sequence of SNP S5_74564291-v5.1, showed incorrect calling of homozygous favored allele as heterozygotes due to close clustering of heterozygous and the favored allele clusters in the ITRA-Togo sorghum breeding program in 2021. b , KASP marker snpSB00487, generated on the reverse strand sequence of SNP S5_74564291-v5.1, showed three distinct clusters. Genotyping calls of the favored and alternate alleles matched the allele prediction, indicating improved marker prediction in the ITRA-Togo program in 2021. Download figure Open in new tab Extended Data Fig. 6 Incorporating strong inference principles into a decision support software for breeders (panGenomeBreedr). Comparison of allele/genotype discrimination cluster plots from a, standard software (SNPViewer) versus b-d , South-led decision support software (panGenomeBreedr) that incorporates strong inference. a , SNPViewer provides basic visualization with post hoc color-coding so erroneous genotype calls cannot be distinguished. b , By contrast panGenomeBreedr offers advanced functionality for accurate genotyping hypothesis testing by incorporating a priori predictions of known positive controls. c , Genotypes are represented using distinct symbols for inferred genotypes (homozygotes for each of two classes and heterozygotes). Notably, however, to facilitate hypothesis testing color overlays indicate the concordance between predicted and observed KASP genotype calls. Blue denotes concordant (accurate) genotypes, red indicates discordant calls (genotyping errors), and beige represents non-control samples for which verification is not possible. d , An additional barplot summarizes the genotype call accuracy across samples for the marker-plate combination to facilitate breeder decisions. Of the 94 samples, 32 were designated as positive controls; among these, 31 were accurately genotyped (blue) and 1 exhibited a genotyping error (red) (Panels c and d). Rapid introgression of lgs1-1 resistance To rapidly breed Striga -resistant sorghum varieties for smallholders, we launched MABC programs using three locally-preferred varieties as recurrent parents ( Fig. 3a and Extended Data Fig. 7 ). To transfer Striga resistance, we used the lgs1-1 KASP marker (snpSB00246 and snpSB00487) for several generations of MAS, as well as marker-based quality control during successive generations of line fixation. In an effort to rapidly test our favored hypothesis of successful marker-assisted trait introgression, we (F.M.) first used a laboratory germination assay at a Striga quarantine facility in the U.S. ( Fig. 3b , Upper inset: positive control, synthetic strigolactone GR24; Lower inset; DI water, negative control). As expected, root exudates of the susceptible parents (Mota Maradi, MR732, and SEPON82) induced Striga germination, while exudates of the resistant donor (SRN39) did not ( Fig. 3b ). Importantly, the lgs1-1 introgression lines (ILs) (Lines 7, 10, 11, 13, 15, 17, 19) induced almost no Striga germination ( Fig. 3b ) ( P > 0.01). To deploy strong inference within the breeding program, we had also retained sibling ILs carrying the LGS1 susceptible alleles for use as negative controls. As expected, almost all the LGS1 sibling ILs (Lines 14, 16, 18, 9, 12) induced Striga germination at the same rate (62–71%; P = 0.3) as the recurrent parents (though surprisingly two of the sibling ILs, Lines 6 and 8, did not induce Striga germination, either due to experimental noise, uncoupling of the marker and causative variant, or introgression of uncharacterized resistance alleles from SRN39). Download figure Open in new tab Extended Data Fig. 7 Marker-assisted introgression programs at INRAN-Niger and ITRA-Togo. Marker-assisted backcrossing of lgs1-1 deletion allele and testing for a , SEPON82 at INRAN-Niger b , MR732 at INRAN-Niger and c , Sorvato1 at ITRA-Togo. Steps of marker assisted selection (MAS) and marker assisted quality control (MQC) are noted. d , Field phenotyping of Striga infestation counts of lgs1 ILs and parent lines in naturally-infested Striga hotspot in Boufale, Togo (2023–2024). Note, this trial within the ITRA-Togo breeding program includes only lgs1 ILs and not LGS1 sibling ILs. Download figure Open in new tab Fig. 3 Effective introgression of Striga resistance into locally-preferred genetic backgrounds with lgs1 marker. a , Marker-assisted backcross scheme to breed Striga -resistant Mota Maradi via lgs1-1 introgression from SRN39. Steps of marker-assisted selection (MAS) and marker-based quality control (MQC) with lgs1 KASP markers (snpSB00246 or snpSB00487) are noted in red text. b , Germination rate for progenies (BC 3 F 3 and F 2 :BC 1 :F 3 ) including inbred lines. Genotypic classes are indicated in LGS1 (blue) representing G:G homozygote class and in lgs1 (red) representing A:A homozygote class. Inset images of preconditioned Striga seeds 72 hours after exposure to GR24 (top) and water (bottom). c , Striga resistance pot experiment in Niger using artificial infestation of Striga from Niger. Host plant comparison in infested versus uninfested control ( Striga -free) pot. d , Striga counts (45, 60, and 90 days after sowing) for Mota Maradi ILs, MR732 ILs, and parent lines in the pot assay. e , Use of qrlabelr RShiny app for barcode labeling in the pot experiment is an example of South-South capacity building. f , Dry biomass of sorghum (yellow material; inner columns) and Striga (gray material; outer columns) for representative LGS1 ILs (left) and lgs1-1 ILs (right). g , Sorghum biomass (above- and below-ground) and h , Striga biomass (above ground) for uninfested control pots (gray) versus Striga -infested pots (blue: lgs1 lines; red: LGS1 lines). (Photo credits: F.M.)\ Striga infestation may be affected by soil and climate factors, and Striga itself harbors ecotypic variation in response to host strigolactone profiles, and effectiveness of lgs1 varies by Striga ecotype 36 , 37 . Following strong inference, we considered the unfavored hypothesis that lgs1-1 resistance is ineffective in Niger. Over 3 years of trials, we tested the effectiveness of lgs1-1 introgression in pot assays in Niger, using Striga from Nigerien smallholder farms and local soil ( Fig. 3c ). A substantial reduction of the number of emerged Striga plants (62–71%, P < 0.001) was observed for lgs1-1 ILs compared to their LGS1 sibling ILs ( Fig. 3d ). Interestingly, however, lgs1-1 ILs had substantially higher Striga counts than donor parent SRN39 (90 days-after-sowing mean of 4.6 vs. 0.6), suggesting that SRN39 may harbor additional resistance QTL, not identified in the original mapping studies 38 , that are expressed in this environment. We also note, as a further example of South-South capacity-building, that as the scale of trials increased ( Extended Data Table 2 ), the INRAN breeding program (F.M. and staffers) adopted qrlabelr barcode-labeling RShiny app 39 ( Fig. 3e ). This app was designed by an African NARS scientist (A.W.K.) to help NARS implement foundational principles of data management (e.g. unique identifiers, standardization, traceability) and adopt open-source digitized breeding tools (e.g. BrAPI-compliant 40 software including digital field books 41 and breeding management systems 42 ). View this table: View inline View popup Download powerpoint Extended Data Table 2: Summary of phenotyping trials The reduction in Striga counts is promising, but not sufficient to conclude that lgs1-1 improves aspects of performance that are relevant to smallholder farmers. Encouragingly, we further observed substantially greater sorghum biomass (above and below ground) for lgs1-1 ILs compared to LGS1 sibling ILs ( Fig. 3f ; yellow plant material, Fig. 3g ), directly establishing a benefit of lgs1-1 for forage yield. Notably, the effect was observed only in Striga -infested pots, not control pots (gray bars), establishing that lgs1-1 confers Striga resistance per se (genotype-environment interaction). In striking contrast, we observed no difference in Striga biomass ( Fig. 3h , gray biomass in Fig. 3g ) between lgs1-1 ILs and LGS1 sibling ILs. This suggests that Striga ecotypes that are able to germinate on lgs1-1 genotypes perform well afterwards, raising the worrying possibility that large-scale lgs1-1 deployment could lead to rapid evolution of Striga that circumvent lgs1 , unless additional resistance genes are pyramided onto the lgs1 background 43 . Interestingly, there is further evidence that SRN39 harbors additional (non lgs1 ) Striga resistance alleles, as there was significantly less (58%; P < 0.01) Striga biomass in SRN39 pots compared to susceptible parent lines or the ILs. Together, these controlled-environment experiments established the effectiveness of lgs1-1 MABC for Niger, and provided a basis to advance the lgs1-1 ILs towards varietal release via multi-environment field trials. Varietal development from genome to farm Adoption of new crop varieties with desired traits often fails without recovery of background acquired traits, such as agronomic adaptation or end-use acceptability 44 . The use of foreground selection only in the early stages of MABC (due to a paucity of background KASP markers and decision support tools at the time) left open two major avenues for failure: recombination between the marker SNP and the lgs1-1 deletion (i.e. failed foreground selection for desired trait) and/or unacceptable alleles from the donor parent due to Mendelian segregation (i.e. failed recovery of acquired traits). We first characterized introgressions genome-wide in the ILs using a new mid-density genotyping assay (DArTag), designed to include highly-informative background markers and trait-predictive markers (see Methods), and panGenomeBreedr software ( Fig. 4a ). Introgressions of the SRN39 lgs1 - 1 haplotype were successful with the marker allele and deletion remaining in linkage phase ( Extended Data Fig. 8 ); future efforts using flanking markers and large families could recover more precise introgressions, while avoiding additional backcross generations that would delay varietal release. We assessed recurrent parent genome recovery at 16 loci that condition variation for traits classified in the regional market segments as must-have “basic traits” ( Tan, Ma, and Dw genes) or “value-added traits” ( Stg and bmr genes) (Supplementary Tables 1-4). Overall, recurrent parent haplotypes were observed at most loci (70% and 81% for Mota Maradi and MR732, respectively), suggesting that local preference traits had been recovered. Detailed examination of causative variant haplotypes for key traits provides insights on how pangenomic breeding could be optimized as more NARIs implement MAS and the number of known causative variants increases ( Extended Data Fig. 8 ). Download figure Open in new tab Extended Data Fig. 8 Genome-wide characterization of introgression lines guides pangenomic breeding strategy. Visualization of mid-density genotyping (DArTag) using panGenomeBreedr. a , Detailed view of selected acquired and desired trait loci in Mota Maradi. For instance, for the major height gene Dwarf3 72 , 73 , Mota Maradi carries the wild-type allele ( Dw3 ; tall) while the SRN39 carries the semi-dwarf allele ( dw3-ref ) that reduces forage yield. In this case, we were fortunate to recover the Mota Maradi haplotype in all the lgs1-1 ILs, but a recently developed pangenomic-derived KASP marker 73 could have easily ensured this outcome. A contrasting case is illustrated at Tannin1 , the major gene conditioning grain tannins, one the most important local adaptation traits 74 , 61 . At Tannin1 , one of the lgs1-1 ILs had an unintended introgression of the SRN39 haplotype, which might suggest this IL would be unacceptable as a variety. However, we know from the pangenomic characterization that Mota Maradi and SRN39 are isogenic at Tannin1 ( tan1b carriers), so MAS at this locus would not only have been unneeded, but would have unnecessarily limited genetic gain from segregation on this chromosome. b , An average 72% recurrent parent genome was recovered in Mota Maradi lgs1-1 ILs similar to the negative sibling ILs (77%, P = 0.56) and the expectation in the absence of selection (75%) ( t -test P = 0.62). c , Detailed view of select acquired and desired trait loci in MR732. d , An average 76% recurrent parent genome was recovered in MR732 lgs1-1 ILs similar to negative sibling ILs (80%, P = 0.79). Download figure Open in new tab Fig. 4 Precise introgression of Striga resistance and recovery of farmer-preferred traits. a , Genome-wide haplotypes for parent lines and marker-assisted backcross (MABC) progenies (F 2 :BC 1 F 7 ) from the Mota Maradi MABC program. The haplotypes were analysed with a mid-density genotyping assay (DArTag) and visualized in panGenomeBreedr, noting LGS1 and 16 genes/QTL that condition other traits that are specified in the target product profiles (Supplementary Table 2). b , Field phenotyping locations across Niger (color scale represents precipitation gradient) and panicles of parent and progeny lines collected on-station, showing recovery of recurrent parent phenotype in the ILs. c , Field phenotyping of Striga emergence (DAS: days after sowing) in controlled-infestation on-station trial in Niger (Konni) for Mota Maradi (MM) and MR732 ILs (BC 3 F 7 and F 2 :BC 1 :F 7 ) and parent lines, as a part of a set of multi-environment trials. d , Flowering time and, e , plant height phenotypes in the same experiment. f , Field phenotyping of Striga emergence, g , in a naturally-infested smallholder farm in Niger (Birni-N’Konni) for the same lines. h , Flowering time and, i , plant height phenotypes in the smallholder farm experiment. Height differences between recurrent parents represent traditional-style (tall: Mota Maradi) or Green Revolution-style (medium: MR732). (Photo credits: F.M.) Adaptive traits that are promising in controlled environments (laboratory or pot experiments) can fail under agronomically-relevant field conditions 45 , 46 ( Fig. 4b ). On-station field trials under controlled Striga infestation in Niger showed that lgs1-1 introgression confers resistance in both Mota Maradi and MR732 backgrounds, with 79% reduction of Striga counts in lgs1-1 ILs compared to LGS1 sibling ILs ( P < 0.001) ( Fig. 4c ). By contrast, sibling LGS1 ILs were heavily infested (similar to the recurrent parents, P = 0.3), confirming that improved ILs performance is due to lgs1 per se. Simultaneous to Striga phenotyping, on-station field trials determined the MABC scheme sufficiently recovered locally-preferred background traits. Flowering time is the most important local adaptation trait for cereals in regions with precipitation seasonality 14 . While the trait donor, SRN39, has similar flowering time to the recurrent parents, Mota Maradi and MR732, cryptic genetic heterogeneity between donor and recurrent parents could create transgressive segregation and result in failed adoption. On-station trials showed that lgs1 ILs recovered optimal flowering time (∼70 days), other than one (Line 10) that was unacceptably late (∼110 days) ( Fig. 4d ). Plant height is another trait that is essential for smallholder farmer acceptance, as a determinant of secondary (forage and building material) yield characteristics. The ILs all recovered acceptable plant height, matching the recurrent parent ( Fig. 4e ). Limited adoption of modern varieties by smallholder farmers 44 has been attributed in part to varietal testing in unrepresentative on-station trials (e.g. under favorable fertilization, irrigation, and pest management) rather than representative on-farm trials (e.g. under limited nutrient, water limitation, and pest pressure) 47 . From a genetics standpoint, it is critical to exclude the hypothesis of unfavorable genotype-environment interaction between the on-station environment and the on-farm target population of environments 14 . From a behavioral and economics perspective, ensuring similarity of improved varieties with locally-preferred varieties should also facilitate adoption. Thus, we tested Striga resistance and background trait recovery on smallholder farms near Birni-N’Konni, and near Zinder, over 400 km from Birni-N’Konni and 740 km from Niamey ( Fig. 4f ). Under natural Striga infestation in a smallholder’s field near Birni-N’Konni, lgs1-1 introgression reduced Striga counts by 88% compared to recurrent parents and LGS1 sibling ILs (ANOVA P < 0.001; Fig. 4g ), with all lgs1-1 ILs performing as well or better than they did on-station ( Fig. 4c ). Similarly, in smallholder fields near Birni-N’Konni ( Striga -infested) and Zinder (uninfested), phenotyping of flowering time ( Fig. 4h ), plant height ( Fig. 4i ), and grain yield components ( Extended Data Fig. 9 ) demonstrate the lgs1-1 ILs (other than Line 10, which was unacceptably late, as in on-station trials) recovered agronomic traits that are essential for varietal release and adoption. Download figure Open in new tab Extended Data Fig. 9 Field phenotyping trials on recovery of farmer-preferred traits. Box plots of flowering time, plant height, grain weight, panicle length, stem diameter and biomass evaluated on-station (Maradi, Niamey) or smallholder farmer fields (Matameye, Zinder) in Niger. Scaling pangenomic breeding across Africa We initially scaled the pangenomic breeding network from hubs at ISRA-Senegal and INRAN-Niger to early-adopter programs, INERA-Burkina Faso ( Stg MABC; Extended Data Fig. 3 ) and ITRA-Togo ( lgs1 MABC; Extended Data Fig. 7 ), in the West African subregion ( Fig. 1b ). Based on promising findings in these countries (Extended Data. Fig. 3 , Figs. 2 – 4 ), we sought to expand the approach across almost all the NARI in the East and West African sorghum belts, as a part of the African Dryland Crop Improvement Network (ADCIN) ( Fig. 5a ; shaded regions). The countries represented in ADCIN account for 89% of African sorghum production area and quantity (and 62% and 40% global production area and quantity) 48 . Within ADCIN, the NARI and CIMMYT have formalized target product profiles that capture demand from local stakeholders (smallholder farmers, consumers, processors, seed enterprises), and developed prioritized regional market segments with a distributed NARS leadership structure (Supplementary Tables 1-4). Genotyping of the GBS-based lgs1 markers (snpSB00246/snpSB00487) across ADCIN showed high prevalence of lgs1- associated alleles (30% and 40% of West and East African germplasm, respectively). However, these are likely to mostly represent false positives for lgs1 , since the marker allele has a high MAF, and global LD between marker and lgs1 was low ( Fig. 2c , 2d). Therefore, we used new pangenome resources, including whole-genome sequencing (WGS; N = 2,144) and de novo reference genomes ( N = 33) capturing global diversity of georeferenced landraces and key improved varieties 49 , to characterize limitations of the first generation markers and design new markers that are scalable across Africa ( Fig. 5b ). Download figure Open in new tab Fig. 5 Using pangenomic resources to design markers that are effective from local to global scales. a , Continent-wide scaling of outsourced KASP genotyping across the African Dryland Cereals Improvement Network (ADCIN; 17 African NARI breeding programs, one African trait discovery program) demonstrates the feasibility of this MAS approach, but reveals limitations (high-false positive rate) of the GBS-based KASP markers (snpSB00246, snpSB00487) at a global scale. b , Whole-genome resequencing resources were used to characterize the limitations of the first generation markers (snpSB00246, snpSB00487) and develop improved markers (snpSB00806, snpSB00807). c , Improved pangenome-based KASP markers are more reliable in the global network, with high inferred lgs1 frequency found only in expected programs (region of origin: IAR-Nigeria and ITRAD-Chad; programs selecting lgs1 : ITRA-Togo and Kenyatta University). d , Tubemap graph of 33 pangenome references highlighting structural complexity of LGS1 locus. Structural variation leads to causative lgs1 resistance alleles and defines optimal regions for design of global markers (hatched boxes: black, considering all haplotypes; gray, ignoring the Tx430-like haplotype not observed in Africa). e , kmer-based inference of lgs1 status in georeferenced African landrace accessions (color-coded as in panel c) contrasting resistant lgs1-1 (red) genotypes versus susceptible LGS1 genotypes (cool shades). f , Linkage disequilibrium analysis of kmer calls of lgs1 versus resequencing SNPs for kmer-based design of improved lgs1 resistance markers. While no nearby SNP has high R 2 (≫50%) with lgs1-1 , many nearby SNPs have high D ’ (∼1), demonstrating that combinations of KASP markers could be used to accurately infer lgs1-1 presence using decision support tools. g , Integrated pangenomic technology links high-resolution pangenomic discoveries (black) on smallholder landraces to low-cost selection tools for breeders (green) via decision support software and various scales of genomic/genotyping tools to breed trait-enhanced varieties for smallholders (rightward arrow). LD analysis with a 1,676 WGS panel showed no association ( R 2 =0.001) at a global scale between WGS-inferred lgs1-1 and the snpSB00487 SNP (S5_74564291|v5.1; note we transition to the improved v5.1 reference 49 at this point). Further, resequencing data at S5_74564291|v5.1 (67% lgs1 -associated A allele) indicated LGS1- containing B35 (BTx642), Ajabsido (Sudan), Segaolane (Botswana), and Marupantse (Botswana) will be incorrectly called as lgs1 , since they harbor the lgs1 -associated A allele ( Fig. 5b ). Thus, we selected two flanking SNPs tightly linked with lgs1-1 (S5_75081793|v5.1 and S5_75132263|v5.1) for pangenome-based marker development ( Fig. 5b , Extended Data Fig. 10 ). Genotyping the INRAN program breeding lines for S5_75081793|v5.1 (snpSB00806) and S5_75132263|v5.1 (snpSB00807) showed three distinct clusters, homozygous lgs1-1 , homozygous LGS1 , and heterozygous ( Extended Data Fig. 10 ). Genotyping with snpSB00806 (MAF = 1%) suggested SRN39 contained lgs1-1 , while BTx623, El Mota, and Mota Maradi contained LGS1 . Genotyping of LGS1 -containing MR732 with snpSB00806 would erroneously suggest it harbors lgs1-1 , since it is clustered with empty controls. However, this can be explained by the pangenome data, which revealed that a flanking deletion covering S5_75081793|v5.1 in MR732 resulted in the loss of the primer annealing site and the clustering with empty controls ( Extended Data Fig. 10 ). Given that SRN39, El Mota, and Mota Maradi share the same allele at S5_75132263|v5.1 (MAF = 2%), snpSB00807 would incorrectly call the presence of lgs1-1 in El Mota and Mota Maradi, but it accurately calls LGS1 in BTx623, MR732, IRAT204, and Macia. Based on these genotyping results, we are using snpSB00806 to introduce lgs1-1 from SRN39 into El Mota and Mota Maradi, and snpSB00807 to introduce lgs1-1 from SRN39 into IRAT204 and Macia, major adopted varieties for East and West Africa, respectively 11 , 50 . Thus, even in a crop with a relatively stable genome such as sorghum (e.g. compared to maize 51 ), pangenomic maps of structural variants, along with pangenomic decision support tools, will be essential to guide marker design and deployment across global breeding networks. Download figure Open in new tab Extended Data Fig. 10 Pangenome-based development of markers suitable for introgression of lgs1-1 into a broad range of elite African varieties. a , Linkage disequilibrium of variants proximal to the lgs1-1 deletion inferred by manual inspection of deletion. SNP1 (snpSB00246/S5_69450954|v3.1), SNP2 (snpSB00806/S5_75081793|v5.1), SNP3 (snpSB00807/S5_75132263|v5.1). b , KASP genotyping of snpSB00806 in the Niger sorghum breeding program. Marker differentiated the favored ( lgs1-1 associated) vs. alternate allele accurately in BTx623, El Mota, and Mota Maradi, but failed in MR732 due to a large deletion surrounding S5_75081793|v5.1. c , KASP genotyping of snpSB00807 in the Niger sorghum breeding program. Marker differentiated the favored ( lgs1-1 associated) vs. alternate allele accurately in BTx623 and MR732 but failed in El Mota and Mota Maradi as the SNP at S5_75132263|v5.1 was conserved between SRN39, El Mota, and Mota Maradi. Finally, to leverage graph pangenome approaches for marker development, we characterized the LGS1 locus in the pangenome reference sequences, which harbors several derived variants that cause major Striga resistance ( Fig. 5c ), and projected them onto resequenced global georeferenced accessions ( Fig. 5d ). Several major structural variants on or around LGS1 are present among our reference genomes ( Fig. 5c ) including the rare 28 kb deletion typical of lgs1-1 resistant genotypes like SRN39 (red haplotype), a high-frequency (7/33 references; 21%) 14 kb deletion downstream of the LGS1 gene that interferes with flanking markers (blue; Mota Maradi), and a 29 kb deletion known in lgs1 resistant line RTx430 (orange) 52 , with breakpoints that correspond to the lgs1-2 allele 29 . The pangenome reference also revealed previously unknown alleles, including two large insertions downstream of LGS1 (13 kb, dark green; 8.9 kb, navy). Notably, these large structural variants must be avoided in KASP marker development to maximize utility of markers across populations ( Fig. 5d ; hatched boxes). To track these variants across global sorghum diversity, we counted exact matches to allele-specific kmer-dictionaries 49 and calculated ancestry probabilities for all six major structural alleles and 2,144 resequenced genotypes ( Fig. 5e ). The kmer analysis locates the origin of the lgs1-1 allele in landraces of the central Sahel (present-day northern Nigeria; red pie charts) from which it moved (via Framida, a Nigerian/Chadian landrace) to SRN39 in Sudan 29 , and on to the global breeding community. Notably, LGS1 proximal variants were in low LD with the SRN39 haplotype ( R 2 < 0.3, D ’ ≤ 1) ( Fig. 5f ), highlighting the need for multiple markers and decision support to track multiple causative alleles across global breeding networks. Conclusions Our next step for delivery from the pangenomic breeding network is the release of lgs1 - and Stg- enhanced varieties to the regional seed system 50 . Based on favorable evidence of Striga resistance ( Fig. 2 – 4 ) and recovery of must-have local-preference traits ( Fig. 4 ) the INRAN breeding program is advancing trait-enhanced varieties for release in Niger and nearby countries via harmonized regional seed regulation. Development of an lgs1 -enhanced hybrid would take additional steps. Since MR732 is a pollinator parent for the NAD-1 hybrid variety (not a variety itself), lgs1-1 is recessive, and the NAD-1 seed parent (ATx623) carries the dominant LGS1 susceptible allele ( Fig. 3f , 5c ), development of an lgs1-1 hybrid variety would require marker-assisted introgression into the maintainer BTx623, followed by backcross conversion of lgs1 -BTx623 to a male-sterile lgs1 -ATx623 seed parent. However, pangenomics findings on the prevalence of lgs1-1 resistance ( Fig. 5c ) suggest an alternative “prebreeding by sequencing” approach: using the pangenomic database to identify existing seed parents (A/B lines) that harbor lgs1 resistance and fast-tracking hybrids of these A-lines with MR732+ lgs1-1 ILs to performance trials. Deployment of lgs1-1 varieties is unlikely to eliminate Striga constraints, especially since Striga biomass, and presumably seed production, seems to be unaffected by lgs1-1 ( Fig. 3h ). A durable strategy may require pyramding traits 21 , 53 , 54 , deploying suicidal germination to reduce the parasite seed bank 55 , and scaling management practices that reduce infestation. Beyond the particular trait-enhanced varieties we developed, we demonstrated the feasibility and utility of a global pangenomic breeding network approach ( Fig. 5g ), which could be scaled to other staple smallholder crops (e.g. maize, millet, cowpea, common bean) that underlie food security. We demonstrated how pangenomics can be applied to design broadly-applicable outsourced markers, and integrate a range of best practices, including local stakeholder input, strong inference, and digital decision support software, to facilitate breeding trait-enhanced varieties for African smallholder farmers. In this case, we used backcross introgression as proof-of-concept for nascent programs, but the most effective long term approach would be a pangenomic forward breeding in locally developed elite germplasm 10 , 14 . Pangenomic forward breeding would combine: (1) foreground selection of trait-predictive markers for desired traits, (2) background selection of trait-predictive markers for must-have acquired traits, and (3) forward polygenic selection for yield using genomic prediction. This approach would require updated low- and mid-density genotyping tools, to distill pangenomic discoveries into low-cost services, and new decision support software, to optimize selection across complex marker-causative variant relationships, allelic heterogeneity, and genetic heterogeneity. Pangenomic forward breeding networks that incorporate local stakeholders knowledge 7 , 47 have the potential to deliver locally- and regionally-adapted varieties of staple crops that meet the demand of smallholder farmers and other stakeholders in developing regions. Methods Plant materials Two seed sources of the resistant genotypes SRN39 were used. The first source PI656027 was obtained from the United States National Plant Germplasm System (NPGS). The second source SRN39 was a product of one selfing of PI656027 (from NPGS) in our program. Biparental families were developed with SRN39 as the donor parent while susceptible parents used were: Mota Maradi (PI656050), MR732 (PI656051), and SEPON82 (PI656024). All four parents are released varieties in Niger that are included in the sorghum association panel 15 , 50 . We confirmed that the recurrent and donor parents carried the expected alleles ( LGS1 susceptible and lgs1-1 resistant, respectively) using the published PCR-based assay 29 . Briefly, genomic DNA of BTx623, Mota Maradi, El Mota, and SRN39 was extracted following the CTAB method and quantified using a NanoDrop spectrophotometer (Thermo Scientific). After normalization to 10 ng/μl, 2 μl of each sample was used for PCR. The first primer (G2133) has a predicted amplicon size of 351 bp on all genotypes assayed, while the second primer (PDstrigalgs5b) has a predicted amplicon size of 628 bp for only susceptible genotypes without LGS1 deletion. Population genomic analyses for lgs1 marker design GBS reads were aligned to the BTx623 sorghum reference genome version 3.1 56 using BWA-MEM 57 and the TASSEL 5 GBS pipeline 58 was used to call SNPs. We targeted a ∼0.5 Mb upstream flanking region (UFR; 69,400,000–69,977,147|v3.1; 577 kb) and a 100 kb downstream flanking region (DFR; 70,011,172–70,111,172|v3.1) on either side of the lgs1-1 deletion interval to identify candidate SNPs for conversion to KASP markers. The smaller DFR interval was chosen to account for the very high subtelomeric recombination rate on the right arm of Chr 05 in sorghum 15 which could increase the risk of uncoupling of causative and marker allele during breeding. A frequency of less than 80% missing data within the region was considered to account for polymorphism, not related to the deletion, present in the flanking regions. The extent of LD between the inferred deletion and flanking SNPs was quantified using R 2 and D ′ in TASSEL 5 software 59 . Allele frequencies were calculated using version 0.1.13 VCFtools 60 . SNPs in LD with the deletion were selected to develop KASP markers. Only SNPs that distinguish between the resistance line, SRN39, and the susceptible lines (e.g. BTx623) were selected to design KASP primers and SNP verification. A total of 10 SNPs within the UFR and DFR were used to design primers with 50 bp flanking sequences around each SNP obtained from https://phytozome.jgi.doe.gov for Sorghum bicolor v3.1, based on the physical position ( Fig. 2a , Supplementary Table 5) ranging between 41 to 534 kb upstream and downstream of the Sobic.005G213600 gene model. SNP verification of the 10 SNPs was conducted at Intertek-AgriTech (Alnarp, Sweden) to test that the putative SNPs selected are of good quality and differentiate between BTx623 and SRN39 ( Extended Data Fig. 4 ). After the SNP verification, selected markers (based on the SNP quality assessment) were tested in inbred lines and progenies derived from resistant and susceptible lines to determine how the markers classify individuals based on a priori predictions on the inbreds and parents used for biparental families. We mapped georeferenced African germplasm 23 – 25 , 61 , 62 for the deletion region on chromosome 5 ( N SNPs = 21,926). A frequency of less than 80% missing data within the region was considered to account for allelic series LGS1 referring as absence or presence of the deletion. Within the lgs1-1 interval, the number of SNPs is 21 in BTx623|v3.1, and SRN39 has all 21 missing SNPs. An expanded GBS dataset 62 was used to generate the allele map ( Fig. 2c ), containing 30 SNPs within the same deletion interval due to higher coverage. KASP genotyping assay for lgs1 Leaf tissues were collected from 14 day old plants grown in the greenhouse. A circular ¼ inch (∼ 6 mm) hole punch was used for tissue collection. Two leaf punches of each sample were placed into the respective well of a 96-well plate. Each accession has at least three individuals referred to as biological replicates. Samples from individual plants of the same seed lot were considered biological replicates, while samples from the same plant were considered technical replicates. “Artificial heterozygote” samples were generated by mixing leaf tissue from one leaf punch of the known resistant genotype with one leaf punch of the known susceptible genotype and used as positive controls for heterozygous genotype calling. Other inbred lines and progenies (F 2:3 ) were replicated either four or two times in the plate. Each plate contained three blank wells; one randomly assigned well and the last two wells (H11 and H12) were left empty as negative controls. To dry the leaf samples, the plate was placed in a plastic bag containing silica beads (“Dry & Dry” Premium orange indicating silica gel beads, Dry & Dry. Silicagel Factory, Brea CA) overnight at room temperature. Dried leaf tissues were sent for DNA extraction and KASP genotyping at Intertek AgriTech (Alnarp, Sweden). Samples were randomized on each genotyping plate. The biparental families of Mota Maradi × SRN39, SEPON82 × SRN39, MR723 × SRN39 (F 2 progeny) were grown in the greenhouse and tissue was collected 14 days after planting. F 3 progenies of Mota Maradi × SRN39 were grown in the field from June to October 2019 in Manhattan, Kansas. For each F 2:3 line, 50 seeds were sown within the plot. For the remaining F 2:3 lines, a single plant was genotyped. Using the markers, backcross lines were obtained using Mota Maradi, MR732, and SEPON82 as recurrent parents. Near isogenic lines (both positive and negative at snpSB00246) were selected. All experiments were done on ILs derived from BC 3 F 7 and F 2 :BC 1 F 7 . KASP genotyping was also conducted from individual plants in the field in Niger. Note, the lgs1 MABC material from INRAN was not included in the network-wide genotyping. Striga germination stimulant assay The germination assay was conducted in the quarantine lab in the Department of Biology at Pennsylvania State University following a published protocol for Striga hermonthica germination assay 36 . Briefly, Striga hermonthica seeds, collected in Siby, Mali (12°23′ N; 8°20′ W), were used for the germination assay. Striga seeds were sterilized with 0.5% sodium hypochlorite for 30 seconds and rinsed three times with sterile distilled water. Approximately 75 seeds (counted under a microscope) were transferred into a 12-well plate with 500 µl of deionized water (DI) in each well. Seeds were preconditioned in the dark for 10 days at 30℃ before root exudate application. Progenies (BC 3 F 3 and F 2 :BC 1 :F 3 ) derived from SRN39 and Mota Maradi, MR732, and SEPON82 ( n = 19) were used for the germination assay. Sorghum seeds were grown in sand and calcined clay mix (1:1 sand/calcined clay) for four weeks in the greenhouse. For each line, six biological replicates were grown. Plants were watered daily with fertigate (top-watering) with ¼ strength Miracle-Gro® Water Soluble All Purpose Plant Food (The Scotts Company, LLC. Marysville, OH). Four weeks after planting, each plant (root and shoot) was removed from the pot. Roots were washed with DI water to remove remaining potting debris. Plants were placed in a flask containing DI water using a 1:5 ratio of root:DI water, sealed with parafilm, and kept under darkness at room temperature for 48 hours. The solution was centrifuged at 4°C at 9,000 rpm for 10–15 minutes, until supernatant was clear, to remove residual root debris. Supernatant was placed on ice until the germination test. Root exudates (1.5 ml) were applied to preconditioned Striga seed. Three technical replicate assays were used for each biological replicate. The synthetic strigolactone GR24 (CAS Number 76974-79-3; ChemPep Inc, Wellington, FL) at 0.1 ppm and DI water were used as positive and negative controls, respectively. After 72 hours, germinated Striga seeds were counted under a stereomicroscope (Manufacturer: Amscope, specifications: 3.5–180X Manufacturing 144-LED Zoom Stereo Microscope with 10MP Digital Camera). Phenotyping for the Striga resistance program Pot experiments Pot experiments on Striga resistance of ILs were conducted in the research center at Niamey, Niger during the rainy seasons (June–October) in 2022, 2023, and 2024 ( Extended Data Table 2 ). Striga hermonthica seeds used for the experiments were collected from Striga -infestation hotspots in smallholder farmer sorghum fields of Niger (Bazaga; 13.8° N, 5.1° E, ∼320 km east of Niamey). Soil for the pot experiments was collected near the station in Niamey. Two Striga seeds inoculations were performed. The first inoculation of Striga seed was conducted at sowing sorghum seeds while the second inoculation was conducted after pre-condition of Striga seeds for 11 days; at this time sorghum plants reached 2 weeks old. The number of emerged Striga were counted weekly from the apparition of the first striga plant in each pot in the trial until physiological maturity or death of the sorghum plant. Sorghum biomass (above- and below-ground) and Striga biomass (above-ground) were collected for each pot. Striga plant biomass from individual pots were harvested, dried at room temperature, and weighed. Sorghum ground biomass was harvested and dried at room temperature along with carefully washed sorghum roots from individual pots. Field experiments The field experiments ( Extended Data Table 2 ) were conducted in Niger under artificial infestation, natural infestation, and uninfested control fields ( Striga -free fields) in six locations on (BC 3 F 7 and F 2 :BC 1 F 7 progenies of MR732 and Mota Maradi, respectively). Planting was done during the rainy season in 2023 and 2024 (June–October). In artificial infestation (on-station, sick plot), sorghum seeds were sown with Striga seeds (∼1g/hole) collected from 10 Konni and Bazaga villages with no fertilizer amendment during the trial. There were no additional inputs in the natural infestation (farmer’s field). For all the locations, phenotypic data recorded are days to flowering, plant height, and yield components. For Striga related traits, the number of striga plants at 45, 60, and 90 days after sowing were recorded. Field trials in Togo were conducted under natural Striga infestation in Boufale (Kara Region) in 2023 and 2024. Mid-density genotyping assay Development of mid-density amplicon-based genotyping assay (DArTag; Intertek Agritech, Sweden) for sorghum was initiated by collating sequencing data from 1,891 breeding lines collected from diverse programs across Asia, Africa, and international partners 63 . DNA was extracted from young seedling leaf tissue using a modified CTAB method and genotyped through Genotyping-by-Sequencing (GBS) with ApeKI enzyme 15 , 34 . SNPs were called using a hybrid pipeline combining GATK and BCFtools 64 , aligned to the Sorghum bicolor v3.1 reference genome 56 . After applying quality filters, mean depth ≥1, base quality ≥30, approximately 383,000 high-quality SNPs were retained. To construct the SNP panel, SNPs with >70% missing data were excluded, retaining only biallelic markers with MAF ≥5% and heterozygosity ≤5%. SNPs with high polymorphism information content (PIC > 0.25) were prioritized, and closely linked markers were removed. Marker specificity was validated via genomic alignment, and LD analysis was performed to ensure representative of markers across LD blocks. Functional annotation using SnpEff helped prioritize genic and near-genic SNPs. The final panel integrated criteria such as informativeness and trait relevance. Later we manually included curated SNPs available in literature linked to shoot fly, stay-green, Striga, aphid resistance, fertility restoration, and Al tolerance. Mid-density genotyping was used to test characterize the genome of the recurrent parent was recovered, Genotyping was conducted on introgression lines (BC 3 F 7 and F 2 :BC 1 F 7 progenies of MR732 and Mota Maradi, respectively) and as well as parent lines, as controls. Leaf tissues from each individual were collected in Niger with four replicates per genotype. Visualization and calculation of parent genome recovery percentages using the panGenomeBreedr package 35 . Marker development from resequencing To ensure data integrity, we excluded 468 libraries flagged as duplicates by kinship analysis and high missing-data rates, resulting in a curated dataset of 1,676 libraries for downstream analyses. SNPs flanking the LGS1 gene (600 kb upstream and 200 kb downstream) were extracted from the sorghum resequencing panel ( N = 1676) using the Samtools HTSlib program 64 , 65 . To tag the lgs1-1 deletion (34 kb) for LD analysis, an artificial SNP (S5_75078000|v5.1 A/T 99% reference vs. 1% favored allele) was introduced into the 1676 pangenome data set, with the T allele in lines with the lgs1-1 allele (Framida, SRN39, 54.K.94.WitchweedRes, and IS10234). LD between S5_75078000|v5.1 and other SNPs was conducted using the TASSEL package 59 . Two SNPs (S5_75081793|v5.1, C/T, 1% MAF; S5_75132263|v5.1, G/A, 2% MAF) in LD with the introduced SNP were selected for KASP marker development at Intertek. Favored versus alternate alleles were differentiated through the competitive binding of two allele-specific forward primers. KASP marker optimization was conducted on leaf samples submitted from the Nigerien program. Data visualization of LD analysis and KASP genotyping results was conducted using the R plot function. Pangenome reference analysis Pangenome graph To identify structural variants across reference haplotypes ( N = 33) around the LGS1 region and visualize them in a tubemap, we built a pangenome graph using Minigraph-Cactus 66 (v2.9.3) with default settings. Construction relied on the full v5.1 BTx623 Chr05 sequence 49 as the primary reference and sequences for the remaining 32 haplotypes were subset to include the LGS1 region plus 100 kb flanking sequence up- and downstream. For the LGS1 region, we further processed this graph with vg toolkit 67 and vcfwave (vcflib v1.0.10) 68 to reduce allelic complexity and retain only variants ≥5 kb in length. Note, BTx623 is within the Macia haplotype group but distinguished in tubemap for illustrative purposes. kmer genotyping The kmer genotyping methodology 49 extracts kmers (80mers in this instance) that are single copy within individual references and variable among members of the pangenome ( N -1 references; N = 32). The pipeline proceeds as follows: (1) Each reference genome is individually converted into kmers, where each kmer and its frequency in the assembly is stored in a hash table. Any non-single copy kmers are flagged to be ignored during downstream analyses. (2) Individual reference hashes are combined in a single hash (main_Hash), updating the flags such that only single copy kmers among all considered references will be used downstream. To isolate kmers that were diagnostic of the LGS1 haplotype structure, sequences were subsetted from the pangenome graph. Steps 1 and 2 above are repeated the subsetted sequences, generating a combined subsetted hash that contains local single-copy sequences from the LGS1 region across the pan-genome. (3) To ensure the local single copy markers do not occur elsewhere in the genome (which would prevent their use as a genotyping marker), the combined LGS1 subsetted hash is intersected with the main genome hash, updating flags such that any non-single copy kmers will be ignored when genotyping (genotyping_hash). (4) Illumina libraries ( N = 2,145) are genotyped by searching for all kmers in the genotyping_hash and counting their frequencies. A priori clusters were generated based on the haplotypes at LGS1 locus within the pangenome tubemap. Diagnostic kmers were sorted into non-overlapping bins from reference Illumina library kmer frequencies (Macia:21,545; Mota Maradi:10,966, IS12646:545, Ajabsido:10,414, SRN39:326, RTx430:80), then kmer matches within each resequencing Illumina library and reports a percentage match from each bin, regardless of which reference they were derived from (e.g. LibraryA|1400|2:93:1:3:1:0; where Library A matched 1400 total diagnostic kmers across six haplotype bins). Illumina library haplotypes were then assigned based on majority match (e.g. Haplotype 2 - 93% of matches) to haplotype bins. LD analysis For LD analysis of kmer calls, libraries were assigned a haplotype for lgs1-1 dependent on the proportion of kmer hits to SRN39 hashes. Libraries whose largest number of unique kmer hits were from the SRN39 haplotype were assigned a “1” and all other libraries were assigned a “0”. This kmer-inferred lgs1-1 call was used to calculate LD ( R 2 and D ’) with all SNP and InDel variants flanking LGS1 (Chr05, 74850000–75350000) using PLINK1.9. Pangenome-enabled breeding for staygreen drought tolerance Marker development and testing Developing improved versions of farmers’ preferred varieties with increased drought tolerance, while maintaining acquired traits for acceptability, was a common goal of the ISRA-Senegal and INERA-Burkina Faso sorghum breeding programs. To achieve this goal within a five year timeframe, we took advantage of discoveries made by the international scientific community 28 , 69 , 70 to identify Stg1 , Stg3A , Stg3B , and Stg5 markers for use in oligogenic MAS in ISRA and INERA breeding programs ( Extended Data Fig. 3a ). Markers at Stg2 and Stg4 were not included in the MABC where BTx642 (B35) was used as donor because their favorable alleles are harbored by local cultivars instead. We used the 1,676 whole genome resequencing panel 49 and 1,362 existing KASP assays (LGC Biosearch Technologies, Middlesex, UK) to identify polymorphic markers at Stg3A and Stg3B genomic regions suitable to use and scalable across other programs, specifically INERA, ITRA, and INRAN. After validation, two markers at Stg3A (S2_56954578|v3.1, S2_70591587|v3.1) and two markers at Stg3B (S2_69739036|v3.1, S2_71419274|v3.1) were retained for use in marker-assisted backcrossing activities in the selected ISRA and INERA farmers’ preferred varieties. For Stg1 , the KASP marker S3_66366589|v3.1, which is carried by BTx642, was used in ISRA and INERA population development. In addition, with the discovery of another stay-green Stg5 locus which co-localizes to the dhurrin biosynthetic gene cluster and confirmed to contribute post-flowering drought tolerance 69 , additional KASP markers were developed. Marker assisted selection Following validation two flanking markers around Stg5 (S1_01147523|v3.1, S1_01287238|v3.1) were used for MABC selection in the ISRA pilot program, then in INERA breeding programs ( Extended Data Fig. 3b ). Backcross breeding at ISRA consisted of one backcross for Nganda × BTx642 (advanced to the BC 1 F 8 ) and three backcrosses for both Darou × BTx642 and Golobe × BTx642 (advanced to the BC 3 F 8 ). After each backcross generation, progeny were genotyped and those carrying at least one stay-green QTL ( Stg1 , Stg3A , Stg3B , or Stg5 ) were retained. At INERA, the elite variety IRAT204 was crossed with BTx642. Three hundred F 2 progenies were planted in the field at Kamboinse Research Station, and leaf samples were collected and sent for KASP marker genotyping at Intertek Lab, Sweden. Superior lines carrying staygreen alleles were marked and self-pollinated to produce F 3 recombinant inbred lines (RILs) before being advanced for several generations to obtain a total of 168 F 5 progenies ( Extended Data Fig. 3g ). Phenotyping for the drought tolerance program Field trials Two off-seasons managed drought stress experiments were performed at CNRA research station in Bambey (14°42′N; 16°28′W), Senegal in 2022 (Nov 2022-March 2023) and 2023 (Feb 2023-June 2023). A total of 30 NILs were used in the 2022 field experiment and 234 NILs in the 2023 field experiment. The 2023 managed drought stress field experiment was conducted during the hot off-season (February-June) at the CNRA research station in Bambey, Senegal (14°42′N, 16°28′W) ( Extended Data Fig. 3d ). The experiment included 234 BC 1 F 5 and BC 3 F 5 lines from ISRA, parental checks (Golobe, Nganda, Darou, BTx642) as well as reference checks (non-staygreen Tx7000 and drought tolerant CE145-66) for a total of 240 genotypes. A 20 × 12 triple alpha lattice design was used. Each plot consisted of two rows of 2 m in length separated by 0.7 m with 0.2 m between plants within a row. Four to six seeds were sown per hill, and seedlings were thinned to one plant per hill at approximately 20 days after sowing (DAS), achieving a planting density of ∼65,000 plants/ha. Two irrigation regimes were applied, well-watered (WW) from sowing to maturity and post-flowering water-stressed (WS), with irrigation withheld in the WS treatment when 75% of the entries (genotypes) had reached flowering until physiological maturity. A similar managed drought stress field experiment at CNRA-Bambey in 2022 is pictured in Extended Data Fig. 3f . At INERA-Kamboinse research station, Burkina Faso (12°27′N, 1°32′W), rain-fed field screening was conducted for the IRAT204 × BTx642 F 6 progenies at the end of the growing in 2022 ( Extended Data Fig. 3g ). A total of 19 lines were selected based on agronomic performance, earliness, and grain hardness and vitreosity. The 19 lines were KASP genotyped, among which nine carry one or more Stg alleles. A final set of four lines was retained for field evaluation in three different stations and in the farmers’ fields surrounding each research station. Pot trials A managed drought stress pot experiment was carried out in the screenhouse at CERAAS in Thies, Senegal (14°45′N, 16°53′W) during 2024 hot off-season (March – June) ( Extended Data Fig. 3c,e ). A total of 16 genotypes (12 ISRA introgression lines and the four parental checks) were used, with three biological replications (pots) per line and three replications for each treatment (WW pots and WS pots) for a total of 288 plants. Water deficit was applied at 50% flowering of plants for a given genotype for a period of 10 days. SPAD chlorophyll content was measured at different time points including physiological maturity. Data availability Reference genome assembly and annotation files of all pangenome members are available at https://phytozome-next.jgi.doe.gov/ . All raw sequence reads have been deposited in the NCBI SRA database under BioProject accessions listed in the pangenome reference paper 49 . Genotype and phenotype data is available at https://github.com/CropAdaptationLab/PangenomicBreedingNetwork . Description of Supplementary Information Supplementary Information | Document with Supplementary Tables 1-6 and Supplementary Data 1-3. Supplementary Table 1 | Market Segment Descriptions for West and Central Africa Supplementary Table 2 | Market Segment Descriptions for East and South Africa Supplementary Table 3 | Target Product Profile Descriptions for West and Central Africa Supplementary Table 4 | Target Product Profile Descriptions for East and South Africa Supplementary Table 5 | Summary of genotyping for 10 candidate SNPs for lgs1 KASP markers Supplementary Table 6 | KASP genotype calls of snpSB00246 in INRAN germplasm Supplementary Data 1 | GBS data of Sorghum Association Panel and West African Sorghum Association Panel on Chr05 Supplementary Data 2 | kmer genotyping results for resequencing panel Supplementary Data 3 | Linkage disequilibrium of resequencing variants with kmer-inferred lgs1-1 deletion Code availability Scripts and analysis methods are available at https://github.com/CropAdaptationLab/PangenomicBreedingNetwork and https://github.com/awkena/panGenomeBreedr . Author contributions G.P.M., D.F., F.M., J.M.F, and H.G conceived and organized the research and development initiative. F.M., J.M.F., C.D., E.A., N.O., S.R.M., and T.J.F. conducted the molecular breeding activities. F.M., J.Ma., C.M.M, M.I.T., O.A.A. O.S.D., A.M.I., O.A.M., A.H., R.A., P.K., and E.A. conducted the Striga resistance experiments. J.M.F., J.P.S., S.B., M.N., D.S., B.S., C.D., and N.O. conducted the drought tolerance experiments. F.M., J.M.F., S.D., A.R., H.G., P.O.-O., S.R.M., G.P.M., and T.J.F developed the genotyping assays. F.M., J.M.F., A.M.H., A.L.H., C.M.M., J.T.L., P.O.-O., S.R.M., C.J.V., and C.C.-B. conducted population genomic and pangenomic analyses. A.W.K, I.T.T, B.A., F.M., J.M.F., G.P.M., L.B., and C.C.-B. developed panGenomeBreedR. F.M., J.M.F., C.D., O.A.A., O.S.D., A.M.I., E.A., N.O., M.A.Y., A.D., R.K., R.K.K., C.P.K., T.L., L.M., J.Mu., E.T.M., G.N., H.N.A.M., K.O.-O., R.O.A., S.R., T.S.A., and L.Y. contributed germplasm for network-wide genotyping. M.K.V.C., B.N., M.N.T, A.A., J.M.F., F.M., T.J.F., S.R.M., C.J.V., and H.G. coordinated and analyzed network-wide genotyping. G.P.M., F.H. D.F., A.M.I., F.M., J.M.F., C.D., B.S., E.A., N.O., A.W.K., J.T.L., T.J.F., J.R.L., C.C.-B, D.H.R, and H.G. acquired funding and supervised the research. F.M. and G.P.M. wrote the manuscript. J.M.F, S.R.M., J.T.L., A.R., and C.J.V. contributed text, figures, and/or revisions. All authors edited and approved the manuscript. Competing interests The authors declare no competing interests. Acknowledgements This paper is made possible by the support of the American People provided to the Feed the Future Innovation Lab for Collaborative Research on Sorghum and Millet (SMIL) through the United States Agency for International Development (USAID) under Cooperative Agreement No. AID-OAA-A-13-00047. The contents are the sole responsibility of the authors and do not necessarily reflect the views of USAID or the United States Government. We thank the USAID-SMIL management entity and external advisory board for guidance on developing the network. The work was funded by the Gates Foundation through “Mining useful alleles for climate change adaptation from CGIAR gene banks (INV-030574)” (G.P.M., J.R.L.), “Accelerated Varietal Improvement and Seed Delivery of Legumes and Dryland Cereals in Africa” (H.G), and “Green Evolution - Accelerating Dryland Cereals Improvement for Africa (INV-053669)” (G.P.M., D.H.R.). The Striga resistance phenotyping in Niger was supported by TWAS-UNESCO. Portions of this manuscript were adapted from F.M.’s Ph.D. dissertation at Kansas State University. We thank colleagues who contributed to aspects of the network which were not addressed in this paper due to space limitations. The work (proposals: 503014; Award DOIs: 10.46936/10.25585/60001093) conducted by the U.S. Department of Energy Joint Genome Institute ( https://ror.org/04xm1d337 ), a DOE Office of Science User Facility, is supported by the Office of Science of the U.S. Department of Energy operated under Contract No. DE-AC02-05CH11231. Funder Information Declared United States Agency for International Development , AID-OAA-A-13-00047 Bill & Melinda Gates Foundation , INV-053669 , INV-030574 Footnotes https://phytozome-next.jgi.doe.gov/ https://github.com/CropAdaptationLab/PangenomicBreedingNetwork https://github.com/awkena/panGenomeBreedr References 1. ↵ Ejeta , G . African Green Revolution needn’t be a mirage . Science 327 , 831 – 832 ( 2010 ). OpenUrl Abstract / FREE Full Text 2. ↵ Food and Agriculture Organization of the United Nations, African Union Commission, United Nations Economic Commission for Africa & World Food Programme . Africa - Regional Overview of Food Security and Nutrition 2023. ( FAO; AUC; ECA; WFP , Accra, Ghana , 2023 ). 3. ↵ Chapman , M. A. , He , Y. & Zhou , M . Beyond a reference genome: pangenomes and population genomics of underutilized and orphan crops for future food and nutrition security . New Phytologist 234 , 1583 – 1597 ( 2022 ). OpenUrl PubMed 4. ↵ Hendre , P. S. et al. African Orphan Crops Consortium (AOCC): status of developing genomic resources for African orphan crops . Planta 250 , 989 – 1003 ( 2019 ). OpenUrl PubMed 5. ↵ Ye , C.-Y. & Fan , L . Orphan crops and their wild relatives in the genomic era . Molecular Plant 14 , 27 – 39 ( 2021 ). OpenUrl CrossRef PubMed 6. ↵ Kane , N. A. , Foncéka , D. Dalton , T. J. Faye , J. M. , Diatta , C. , Maina , F. & Morris , G. P. Past, present, and future of West African sorghum improvement: Building a roadmap for climate-adaptive, farmer-adopted varieties . in Crop Adaptation and Improvement for Drought-Prone Environments (eds Kane , N. A. , Foncéka , D. & Dalton , T. J. ) ( New Prairie Press , 2022 ). 7. ↵ Tonapi , V. A. Haussmann , B. I. G. et al. Tackling Key Issues for Smallholder Farmers: The Farmer Research Network (FRN) Approach . in Sorghum in the 21st Century: Food – Fodder – Feed – Fuel for a Rapidly Changing World (eds Tonapi , V. A. et al. ) 315 – 329 ( Springer , Singapore , 2020 ). doi: 10.1007/978-981-15-8249-3_13 . OpenUrl CrossRef 8. ↵ Atlin , G. N. , Cairns , J. E. & Das , B . Rapid breeding and varietal replacement are critical to adaptation of cropping systems in the developing world to climate change . Global Food Security 12 , 31 – 37 ( 2017 ). OpenUrl PubMed 9. Cobb , J. N. , Biswas , P. S. & Platten , J. D . Back to the future: revisiting MAS as a tool for modern plant breeding . Theor Appl Genet ( 2018 ) doi: 10.1007/s00122-018-3266-4 . OpenUrl CrossRef 10. ↵ Fall , S. T. et al. Genomic approaches to build de novo elite breeding gene pools from locally-adapted landraces . 2025.06.30.662425 Preprint at doi: 10.1101/2025.06.30.662425 ( 2025 ). OpenUrl Abstract / FREE Full Text 11. ↵ Walker , T. S. & Alwang , J. Ndjeunga , J. , Mausch , K. & Simtowe , F . Assessing the effectiveness of agricultural R&D for groundnut, pearl millet, pigeonpea and sorghum in West and Central Africa and East and Southern Africa . in Crop improvement, adoption, and impact of improved varieties in food crops in sub-Saharan Africa (eds Walker , T. S. & Alwang , J. ) 123 – 147 ( CABI , Wallingford , 2015 ). doi: 10.1079/9781780644011.0123 . OpenUrl CrossRef 12. ↵ Thomson , M. J . High-throughput SNP genotyping to accelerate crop improvement . Plant Breeding and Biotechnology 2 , 195 – 212 ( 2014 ). OpenUrl CrossRef 13. ↵ Excellence in Breeding. KASP low density genotyping Platform . https://excellenceinbreeding.org/toolbox/tools/kasp-low-density-genotyping-platform . 14. ↵ Haussmann , B. I. G. et al. Breeding strategies for adaptation of pearl millet and sorghum to climate variability and change in West Africa . Journal of Agronomy and Crop Science 198 , 327 – 339 ( 2012 ). OpenUrl 15. ↵ Morris , G. P. et al. Population genomic and genome-wide association studies of agroclimatic traits in sorghum . PNAS 110 , 453 – 458 ( 2013 ). OpenUrl Abstract / FREE Full Text 16. ↵ Westengen , O. T. et al. Ethnolinguistic structuring of sorghum genetic diversity in Africa and the role of local seed systems . PNAS 201401646 ( 2014 ) doi: 10.1073/pnas.1401646111 . OpenUrl Abstract / FREE Full Text 17. ↵ Integrating New Technologies for Striga Control: Towards Ending the Witch-Hunt. ( World Scientific , 2007 ). 18. ↵ Runo , S. & Kuria , E. K . Habits of a highly successful cereal killer, Striga . PLoS Pathog 14 , ( 2018 ). 19. ↵ Williams , C. N . Resistance of sorghum to witchweed . Nature 184 , 1511 – 1512 ( 1959 ). OpenUrl 20. ↵ Al-Babili , S. & Bouwmeester , H. J . Strigolactones, a novel carotenoid-derived plant hormone . Annu. Rev. Plant Biol . 66 , 161 – 186 ( 2015 ). OpenUrl CrossRef PubMed 21. ↵ Yohannes , T. et al. Marker-assisted introgression improves Striga resistance in an Eritrean farmer-preferred sorghum variety . Field Crops Research 173 , 22 – 29 ( 2015 ). OpenUrl 22. ↵ Kapran , I. , Grenier , C. & Ejeta , G. Introgression of genes for Striga resistance into African landraces of sorghum . in Integrating New Technologies for Striga Control: Towards Ending the Witch-Hunt 129 – 141 ( 2007 ). 23. ↵ Olatoye , M. O. , Hu , Z. , Maina , F. & Morris , G. P . Genomic signatures of adaptation to a precipitation gradient in Nigerian sorghum . G3: Genes, Genomes, Genetics 8 , 3269 – 3281 ( 2018 ). OpenUrl 24. Faye , J. M. et al. Genomic signatures of adaptation to Sahelian and Soudanian climates in sorghum landraces of Senegal . Ecology and Evolution 9 , 1 – 14 ( 2019 ). OpenUrl 25. ↵ Maina , F. et al. Population genomics of sorghum (Sorghum bicolor) across diverse agroclimatic zones of Niger . Genome 61 , 223 – 232 ( 2018 ). OpenUrl CrossRef 26. Faye , J. M. et al. A genomics resource for genetics, physiology, and breeding of West African sorghum . The Plant Genome 14 , e20075 ( 2021 ). OpenUrl PubMed 27. ↵ Olatoye , M. O. et al. Dissecting adaptive traits with nested association mapping: genetic architecture of inflorescence morphology in sorghum . G3: Genes, Genomes, Genetics 10 , 1785 – 1796 ( 2020 ). OpenUrl 28. ↵ Borrell , A. K. et al. Stay-green alleles individually enhance grain yield in sorghum under drought by modifying canopy development and water uptake patterns . New Phytol 203 , 817 – 830 ( 2014 ). OpenUrl CrossRef PubMed 29. ↵ Gobena , D. et al. Mutation in sorghum LOW GERMINATION STIMULANT 1 alters strigolactones and causes Striga resistance . PNAS 114 , 4471 – 4476 ( 2017 ). OpenUrl Abstract / FREE Full Text 30. ↵ Cobb , J. N. et al. Enhancing the rate of genetic gain in public-sector plant breeding programs: lessons from the breeder’s equation . Theor Appl Genet 132 , 627 – 645 ( 2019 ). OpenUrl CrossRef PubMed 31. ↵ Woltering , L. , Fehlenberg , K. , Gerard , B. , Ubels , J. & Cooley , L . Scaling – from “reaching many” to sustainable systems change at scale: A critical shift in mindset . Agricultural Systems 176 , 102652 ( 2019 ). OpenUrl CrossRef 32. ↵ Platt , J. R . Strong Inference: Certain systematic methods of scientific thinking may produce much more rapid progress than others . Science 146 , 347 – 353 ( 1964 ). OpenUrl FREE Full Text 33. ↵ United Nations Development Programme . Human Development Report 2015 . https://hdr.undp.org/system/files/documents/global-report-document/2015humandevelopmentreport1.pdf ( 2015 ). 34. ↵ Elshire , R. J. et al. A robust, simple genotyping-by-sequencing (GBS) approach for high diversity species . PLoS ONE 6 , e19379 ( 2011 ). OpenUrl CrossRef PubMed 35. ↵ Kena , A. awkena/panGenomeBreedr: Helpers for pangenome-enabled crop breeding . ( 2025 ). 36. ↵ Bellis , E. S. et al. Genomics of sorghum local adaptation to a parasitic plant . PNAS ( 2020 ) doi: 10.1073/pnas.1908707117 . OpenUrl Abstract / FREE Full Text 37. ↵ Bellis , E. S. , McLaughlin , C. M. , dePamphilis , C. W. & Lasky , J. R. The geography of parasite local adaptation to host communities . Ecography 44 , 1205 – 1217 ( 2021 ). OpenUrl 38. ↵ Satish , K. , Gutema , Z. , Grenier , C. , Rich , P. J. & Ejeta , G . Molecular tagging and validation of microsatellite markers linked to the low germination stimulant gene (lgs) for Striga resistance in sorghum [Sorghum bicolor (L.) Moench] . Theor Appl Genet 124 , 989 – 1003 ( 2012 ). OpenUrl PubMed 39. ↵ Kena , A. et al. Introducing qrlabelr : Fast user-friendly software for machine- and human-readable labels in agricultural research and development . Gates Open Research ( 2024 ) doi: 10.12688/gatesopenres.15268.1 . OpenUrl CrossRef 40. ↵ Selby , P. et al. BrAPI—an application programming interface for plant breeding applications . Bioinformatics 35 , 4147 – 4155 ( 2019 ). OpenUrl PubMed 41. ↵ Rife , T. W. & Poland , J. A . Field book: an open-source application for field data collection on android . Crop Science 54 , 1624 ( 2014 ). OpenUrl CrossRef PubMed Web of Science 42. ↵ Ribaut , J.-M. , Delannay , X. , McLaren , G. & Okono , F . Molecular breeding platforms in world agriculture . in Encyclopedia of Sustainability Science and Technology 6692 – 6720 ( Springer , New York, NY , 2012 ). doi: 10.1007/978-1-4419-0851-3_237 . OpenUrl CrossRef 43. ↵ Mundt , C. C . Pyramiding for resistance durability: theory and practice . Phytopathology 108 , 792 – 802 ( 2018 ). OpenUrl CrossRef PubMed 44. ↵ Walker , T. S. & Alwang , J. Crop Improvement, Adoption and Impact of Improved Varieties in Food Crops in Sub-Saharan Africa. ( CABI , 2015 ). 45. ↵ Blum , A. Plant Breeding for Water-Limited Environments. ( Springer Publishing , New York, 2010 ). 46. ↵ Nelson , R. , Wiesner-Hanks , T. , Wisser , R. & Balint-Kurti , P . Navigating complexity to breed disease-resistant crops . Nature Reviews Genetics 19 , 21 – 33 ( 2018 ). OpenUrl CrossRef PubMed 47. ↵ Etten , J. van et al. Crop variety management for climate adaptation supported by citizen science . PNAS 116 , 4194 – 4199 ( 2019 ). OpenUrl Abstract / FREE Full Text 48. ↵ Food and Agriculture Organization of the United Nations . FAOSTAT . https://www.fao.org/faostat/en/#home . 49. ↵ Morris , G. , et al. Developing future resilience from signatures of adaptation across the sorghum pangenome . ( 2025 ). 50. ↵ FAO Plant Production and Protection Division . West African Catalogue of Plant Species and Varieties. ( Food and Agriculture Organization of the United Nations , Rome, Italy , 2008 ). 51. ↵ Schnable , J. C. , Springer , N. M. & Freeling , M . Differentiation of the maize subgenomes by genome dominance and both ancient and ongoing gene loss . PNAS 108 , 4069 – 4074 ( 2011 ). OpenUrl Abstract / FREE Full Text 52. ↵ Yoda , A. et al. Strigolactone biosynthesis catalyzed by cytochrome P450 and sulfotransferase in sorghum . New Phytologist 232 , 1999 – 2010 ( 2021 ). OpenUrl CrossRef PubMed 53. ↵ Shi , J. et al. Resistance to Striga parasitism through reduction of strigolactone exudation . Cell 0 , ( 2025 ). 54. ↵ Li , C. et al. Maize resistance to witchweed through changes in strigolactone biosynthesis . Science 379 , 94 – 99 ( 2023 ). OpenUrl CrossRef PubMed 55. ↵ Uraguchi , D. et al. A femtomolar-range suicide germination stimulant for the parasitic plant Striga hermonthica . Science 362 , 1301 – 1305 ( 2018 ). OpenUrl Abstract / FREE Full Text 56. ↵ McCormick , R. F. et al. The Sorghum bicolor reference genome: improved assembly, gene annotations, a transcriptome atlas, and signatures of genome organization . The Plant Journal 93 , 338 – 354 ( 2018 ). OpenUrl CrossRef PubMed 57. ↵ Li , H. & Durbin , R . Fast and accurate long-read alignment with Burrows-Wheeler transform . Bioinformatics 26 , 589 – 595 ( 2010 ). OpenUrl CrossRef PubMed Web of Science 58. ↵ Glaubitz , J. C. et al. TASSEL-GBS: A high capacity genotyping by sequencing analysis pipeline . PLoS ONE 9 , e90346 ( 2014 ). OpenUrl CrossRef PubMed 59. ↵ Bradbury , P. J. et al. TASSEL: software for association mapping of complex traits in diverse samples . Bioinformatics 23 , 2633 – 2635 ( 2007 ). OpenUrl CrossRef PubMed Web of Science 60. ↵ Danecek , P. et al. The variant call format and VCFtools . Bioinformatics 27 , 2156 – 2158 ( 2011 ). OpenUrl CrossRef PubMed Web of Science 61. ↵ Lasky , J. R. et al. Genome-environment associations in sorghum landraces predict adaptive traits . Science Advances 1 , e1400218 ( 2015 ). OpenUrl FREE Full Text 62. ↵ Hu , Z. , Olatoye , M. O. , Marla , S. & Morris , G. P . An integrated genotyping-by-sequencing polymorphism map for over 10,000 sorghum genotypes . The Plant Genome 12 , 1 – 15 ( 2019 ). OpenUrl 63. ↵ Deshpande , S. , Rathore , A. , Gandhi , H. , Selvanayagam , S. & Gupta , R . Development of Quality Control (QC) marker set, and mid-density marker set in Sorghum using Genotyping-by-Sequencing (GBS) based SNPs data of a large set of about 1900 breeding lines across several breeding programs . ICRISAT Dataverse doi: 10.21421/D2/YU3ALS ( 2021 ). OpenUrl CrossRef 64. ↵ Danecek , P. et al. Twelve years of SAMtools and BCFtools . GigaScience 10 , giab008 ( 2021 ). OpenUrl CrossRef PubMed 65. ↵ Bonfield , J. K. et al. HTSlib: C library for reading/writing high-throughput sequencing data . GigaScience 10 , giab007 ( 2021 ). OpenUrl CrossRef PubMed 66. ↵ Hickey , G. et al. Pangenome graph construction from genome alignments with Minigraph-Cactus . Nat Biotechnol 42 , 663 – 673 ( 2024 ). OpenUrl CrossRef PubMed 67. ↵ Garrison , E. et al. Variation graph toolkit improves read mapping by representing genetic variation in the reference . Nat Biotechnol 36 , 875 – 879 ( 2018 ). OpenUrl CrossRef PubMed 68. ↵ Garrison , E. , Kronenberg , Z. N. , Dawson , E. T. , Pedersen , B. S. & Prins , P . A spectrum of free software tools for processing the VCF variant call format: vcflib, bio-vcf, cyvcf2, hts-nim and slivar . PLOS Computational Biology 18 , e1009123 ( 2022 ). OpenUrl 69. ↵ Hayes , C. M. et al. Discovery of a dhurrin QTL in sorghum: co-localization of dhurrin biosynthesis and a novel stay-green QTL . Crop Science 56 , 104 – 112 ( 2016 ). OpenUrl 70. ↵ Mwamahonje , A. et al. Introgression of QTLs for drought tolerance into farmers’ preferred sorghum varieties . Agriculture 11 , 883 ( 2021 ). OpenUrl 71. ↵ Kinraide , T. B. & Denison , R. F . Strong Inference: The Way of Science . The American Biology Teacher 65 , 419 – 424 ( 2003 ). OpenUrl FREE Full Text 72. ↵ Multani , D. S. et al. Loss of an MDR transporter in compact stalks of maize br2 and sorghum dw3 mutants . Science 302 , 81 – 84 ( 2003 ). OpenUrl Abstract / FREE Full Text 73. ↵ Marla , S. R. , et al. Mining sorghum pangenome enabled identification of new dw3 alleles for breeding stable-dwarfing hybrids . G3 Genes|Genomes|Genetics 15 , jkaf054 ( 2025 ). OpenUrl 74. ↵ Wu , Y. et al. Presence of tannins in sorghum grains is conditioned by different natural alleles of Tannin1 . PNAS 109 , 10281 – 10286 ( 2012 ). OpenUrl Abstract / FREE Full Text View the discussion thread. Back to top Previous Next Posted August 07, 2025. Download PDF Supplementary Material Data/Code Email Thank you for your interest in spreading the word about bioRxiv. NOTE: Your email address is requested solely to identify you as the sender of this article. Your Email * Your Name * Send To * Enter multiple addresses on separate lines or separate them with commas. You are going to email the following Delivering trait-enhanced varieties to African smallholders through a pangenomic breeding network Message Subject (Your Name) has forwarded a page to you from bioRxiv Message Body (Your Name) thought you would like to see this page from the bioRxiv website. Your Personal Message CAPTCHA This question is for testing whether or not you are a human visitor and to prevent automated spam submissions. Share Delivering trait-enhanced varieties to African smallholders through a pangenomic breeding network Fanna Maina , Jacques M. Faye , Alex W. Kena , Cyril Diatta , Moctar Issiakou Tankari , Ousseini Abdou Ardaly , Ousmane Seyni Diakite , Aissata Mamadou Ibrahim , Oumarou Abdoulye Moussa , Abdou Harou , Rabiou Abdou , Joseph Pascal Sene , Souleymane Bodian , Mbery Ndour , Diarietou Sambakhe , Bassirou Sine , Pana Kadanga , Eyanawa Akata , Nofou Ouedraogo , Benjamin Annor , Israel T. Tetteh , Muhammad Ahmad Yahaya , Abdoulaye Diallo , Ronald Kakeeto , Rachael K. Kisilu , Clarisse Pulchérie Kondombo , Tokuma Legesse , Lloyd Mbulwe , Joaquim Mutaliano , Emmanuel T. Mwenda , Gapili Naoura , Hortense Noëlle Apala Mafouasson , Kenneth Opare-Obuobi , Rekiya Otuchu Abdulmalik , Steven Runo , Tilal Sayed Abdelhalim , Louis Yalaukani , Joel Masanga , Chloee M. McLaughlin , Jesse R. Lasky , Avril M. Harder , Adam L. Healey , John T. Lovell , Portia Osei-Obeng , Sandeep R. Marla , Terry J. Felderhoff , Falalou Hamidou , Daniel Fonceka , Mohan Kumar Varma Chejerla , Baloua Nébié , Mark Nas Tamerlan , Amos Alakonya , Abhishek Rathore , Santosh Deshpande , Harish Gandhi , Linly Banda , Davina H. Rhodes , Clara Cruet-Burgos , Carl J. VanGessel , Geoffrey P. Morris bioRxiv 2025.08.07.667917; doi: https://doi.org/10.1101/2025.08.07.667917 Share This Article: Copy Citation Tools Delivering trait-enhanced varieties to African smallholders through a pangenomic breeding network Fanna Maina , Jacques M. Faye , Alex W. Kena , Cyril Diatta , Moctar Issiakou Tankari , Ousseini Abdou Ardaly , Ousmane Seyni Diakite , Aissata Mamadou Ibrahim , Oumarou Abdoulye Moussa , Abdou Harou , Rabiou Abdou , Joseph Pascal Sene , Souleymane Bodian , Mbery Ndour , Diarietou Sambakhe , Bassirou Sine , Pana Kadanga , Eyanawa Akata , Nofou Ouedraogo , Benjamin Annor , Israel T. Tetteh , Muhammad Ahmad Yahaya , Abdoulaye Diallo , Ronald Kakeeto , Rachael K. Kisilu , Clarisse Pulchérie Kondombo , Tokuma Legesse , Lloyd Mbulwe , Joaquim Mutaliano , Emmanuel T. Mwenda , Gapili Naoura , Hortense Noëlle Apala Mafouasson , Kenneth Opare-Obuobi , Rekiya Otuchu Abdulmalik , Steven Runo , Tilal Sayed Abdelhalim , Louis Yalaukani , Joel Masanga , Chloee M. McLaughlin , Jesse R. Lasky , Avril M. Harder , Adam L. Healey , John T. Lovell , Portia Osei-Obeng , Sandeep R. Marla , Terry J. Felderhoff , Falalou Hamidou , Daniel Fonceka , Mohan Kumar Varma Chejerla , Baloua Nébié , Mark Nas Tamerlan , Amos Alakonya , Abhishek Rathore , Santosh Deshpande , Harish Gandhi , Linly Banda , Davina H. Rhodes , Clara Cruet-Burgos , Carl J. VanGessel , Geoffrey P. Morris bioRxiv 2025.08.07.667917; doi: https://doi.org/10.1101/2025.08.07.667917 Citation Manager Formats BibTeX Bookends EasyBib EndNote (tagged) EndNote 8 (xml) Medlars Mendeley Papers RefWorks Tagged Ref Manager RIS Zotero Tweet Widget Facebook Like Google Plus One Subject Area Genomics Subject Areas All Articles Animal Behavior and Cognition (7629) Biochemistry (17660) Bioengineering (13881) Bioinformatics (41912) Biophysics (21436) Cancer Biology (18578) Cell Biology (25482) Clinical Trials (138) Developmental Biology (13372) Ecology (19889) Epidemiology (2067) Evolutionary Biology (24302) Genetics (15599) Genomics (22483) Immunology (17728) Microbiology (40365) Molecular Biology (17163) Neuroscience (88540) Paleontology (666) Pathology (2830) Pharmacology and Toxicology (4821) Physiology (7637) Plant Biology (15130) Scientific Communication and Education (2045) Synthetic Biology (4290) Systems Biology (9818) Zoology (2269)