Broadening the narrative: reflecting on archaeological discourses on plant domestication

Introduction

In this discursive piece, we reflect on the ontological constitution of archaeological, including archaeogenomic, discourses on plant domestication, especially within debates concerning early agriculture. We argue that domestication discourses in archaeology have tended to be normative, linear and teleological in the construction of the past. Interpretations of domestication are not neutral reconstructions of the past, rather they tend to focus on specific types of crop and region, and normalise perspectives arising from these, while marginalising others. Often tracing specific phenotypic and genotypic traits associated with major crop lineages today, these reconstructions obscure the historical biogeography of crop wild relatives, cultiwilds and cultivar groups from which contemporary commercialised varieties arose. In doing so, domestication discourses risk relegating many plants, and the peoples who cultivated them, to the margins of archaeological reconstructions of the past.

We draw on our experiences working on vegetatively propagated plants in the tropics and subtropics to illustrate our reflections. We acknowledge that some discourses on domestication are already broadening and becoming more diverse, linear and historically complex. Our aims are to recontextualise domestication as a way point rather than an end point of long-term historical processes, and also to defocus our temporal perspective from a linear domestication pathway to highlight the temporally constituted genealogical array of plant diversity from which domesticated plants arose.

Resisting dominant narratives

In the past, archaeologists focussed upon a handful of crops in the study of domestication. These crops were considered the most significant, especially to western societies, and were archaeobotanically visible (Fuks et al., 2025). Additionally, they have tended to reflect the geographical regions with the greatest numbers of excavations (Stephens et al., 2019). Crops from other regions, especially the tropics, were understudied (Denham et al., 2007). These historical legacies underpin preconceptions about what are considered the most significant crops in world history – with so much focus on the “Big Three” and their regions of origin: wheat (Triticum spp.) from Southwest Asia; maize (Zea mays) from Mexico; and, rice (Oryza sativa ssp. japonica) from China. These cereals are among the three greatest contributors to contemporary global diets (Figure 1; FAOSTAT, 2016) and these crops and regions of origin retain a disproportionate influence upon debates concerning plant domestication and early agriculture (consider Harlan, 1971).

Yet, six of the top twelve contributors to global diets are vegetatively propagated field crops (FAOSTAT, 2016), comprising sugarcane (Saccharum cvs.) and bananas/plantains (Musa cvs.) from Island Southeast Asia-New Guinea (ISEA-NG); potato (Solanum tuberosum) from the Central Andes; manioc/cassava (Manihot esculenta) from the Amazon; sweet potato (Ipomoea batatas) from the Americas, and, pan-tropical yams (Dioscorea spp.) (Denham et al., 2020; Figure 1). These crops were all domesticated in the tropics and subtropics, have become global commodities, and continue to be major subsistence crops for communities living in those, as well as other regions today. Despite their significance to large percentages of the global population, the domestication histories for these crops are poorly documented archaeobotanically, although there are increasing insights from archaeogenomics (e.g., Kistler et al., 2025). Some regions of origin connected to these crops have been marginalised in mainstream debates on early agriculture, although these pervasive biases are being redressed (Denevan, 2001; Denham, 2018; Denham et al., 2007; Dillehay, 2011; Golson et al., 2017; Iriarte, 2024).

The domestication and globalisation histories of a wider range of crops from Africa and Eurasia are now research foci, including millets (small-seeded cereals) and tropical pulses (Fuller and Murphy, 2018; Mueller et al., 2022). There is a similar trend within some archaeogenomics, which has tended to focus on commercial crops; research now encompasses a wider range of regionally important crops, such as the pulse lablab (Lablab purpureus), the cereals fonio (Digitaria spp.), teff (Eragrostis tef), and finger millet (Eleusine coracana), and the vegecultural crop enset (Ensete ventricosum) in Africa (Burton et al., 2025; Castillo et al., 2025; Maass et al., 2017; Mekonnen et al., 2025). This shift in focus has occurred in collaboration with archaeologists who are becoming more interested in “underutilised crops”, often as part of multi-disciplinary research programmes for crop improvement and conservation. Genomic approaches can be particularly important for defining the relationship between wild relatives and crop groups, but only archaeology can ground timelines and help frame interpretations of crop historical biogeography.

There are now around 20 regions of early plant domestication identified (Fuller et al., 2014), multiple crop domestication pathways (Fuller et al., 2023), including conceptualisations of “non-centres as opposed to centres” and “landscape scale” domestication processes (Allaby et al., 2021; Harlan, 1971; Harris, 1990; also see Latinis, 2000; Levis et al., 2018; Rival, 2007; Yen, 1989). Yet, biases persist. Historical representations of crop domestication and early agriculture can literally write crops, places and peoples out of history; many are not considered significant, relegated to lesser importance, and denied historical efficacy. The way we construct subjects in historical discourse matters; what we marginalise and ignore in the past resonates with what we marginalise and ignore in the present. Those outside discourse have no voice, they cannot speak (Spivak, 1988).

Providing methodological options

Archaeobotanical debates on plant domestication tend to focus on tracking the prevalence of phenotypic and, increasingly, genotypic traits in domesticated plants through time relative to the cognate wild plants (Fuller, 2007; Fuller et al., 2014, 2023; Meyer et al., 2012; Meyer and Purugganan, 2013). For instance, the non-shattering trait in cereals and pulses occurs at very low frequencies in wild plants and approaches 80%–100% “fixation” in domesticated plants. Archaeobotanical assemblages from multiple sites can demonstrate the time frame over which this domestication trait becomes dominant, namely, the domestication episode, with modern and potentially ancient DNA providing corresponding information on the prevalence of the genetic trait that regulates non-shattering. For rice (Oryza sativa ssp. japonica), archaeobotanical data in the Yangtze River drainage basin in China show increases in non-shattering from c. <20% to >80% over at least c. 1,500 years from c. 7,500–6,000 cal BP (Fuller et al., 2023), with genetic data correlating this phenotypic trait to qSH3 and Sh4 mutations (Alam et al., 2025; Ishikawa et al., 2017).

Until the last decade or so, plant domestication debates were dominated by a handful of crop groups – primarily cereals, pulses and some fruit trees – for which the emergence of domestication traits were macroscopically visible and could be readily traced in archaeobotanical assemblages. The remains of seeds, spikelet bases, fruit stones, nut shells and wood preserve in a variety of depositional environments in charred, desiccated and waterlogged form (Hastorf and Popper, 1988). Many other crops have not been considered on equal terms, partly because there are no clear signatures of domestication in the archaeobotanical record. For instance, many vegetatively-propagated crop plants in the tropics and subtropics require investigation using a suite of techniques including archaeological parenchyma, phytoliths and starch granule analysis (Hather, 1994). Phytolith analysis is well-established (Piperno, 2006), while the analysis of archaeological parenchyma is still under methodological development (Hather, 2000; Kubiak-Martens, 2016) and starch granule analysis is still of limited use for the investigation of most crop plants (Torrence and Barton, 2006). None of these techniques currently provides a reliable means to discriminate the wild and domesticated phenotypes of any vegetatively propagated crop in archaeobotanical assemblages, with most taxonomic identifications to species or genus, although there is future potential (Denham et al., 2020). Despite these methodological shortcomings, a complementary suite of archaeobotanical, archaeological (e.g., of cultivation practices), and genomic techniques provides more options for the investigation of a broader range of crop plants, as demonstrated for bananas (Perrier et al., 2011).

Although the domestication pathways and methods of analysis may vary by crop, the conceptual framing of tracking domestication in the past is mostly similar: namely, phenotypic transformations differentiating wild and domesticated plants are tracked through time to infer when domestication occurred. However, the tacking of change through time – although significant – tends to narrow our attention to certain types of change in certain types of plant. When a few, specific phenotypic traits are prioritised, what is being missed?

Recontextualising domestication discourse: the wild-domestic continuum

Plant domestication represents a way point on a much broader range of human-plant domesticatory relationships. As Harris (1990) alluded, there is a certain arbitrariness about domestication as a concept; it may apply more readily to some groups of plant than to others. The ways in which plants become domesticated are contingent upon a range of factors, varying from the biology of the plant, environmental transformations and the practices of cultivation, to the social significance of the plant within a given cultural context. We consider three elements arising from research on underground storage organs (USOs) that plausibly have wider salience: seeing domestication against much longer-term, human-plant domesticatory relationships; phenotypic plasticity and epimutations; and, ambiguity in the classification of wild and domesticated plants.

Harris (1990, 1996) advocated a focus on the long-term character of human-plant domesticatory relationships, as opposed to domestication per se. Drawing on multidisciplinary research on USOs, long histories of interaction between a plant and people can precede cultivation by millennia, and in some cases tens of millennia (Hather, 1996). During this time the phenotype of the plant will have undergone anthropic selection and resultant transformation. Against this long-term coevolutionary backdrop, formal “domestication”, however measured, need not mark an abrupt binary shift from “wild” to “domestic”, rather it represents a ratcheting up in the intensity of selective pressures on the plant along a spectrum that encompasses initial exploitation, management within habitat, resource intensification, translocation and replanting, extensive cultivation and more intensive forms of cultivation (Denham, 2018). From the perspective of long-term human-plant domesticatory relationships, overfocusing on “domestication” appears somewhat arbitrary: where on the temporal spectrum of human selective pressure is the beginning and end of the domestication episode, and why are these demarcated points considered the most significant as opposed to others?

Archaeobotanical investigations of plant domestication have tended to focus on plants in which specific domestication traits become “fixed” within assemblages. In cereals, a trait such as non-shattering is genetically controlled and is primarily considered a one-way transformation and forms part of putative domestication syndromes (Fuller, 2007; Hammer, 1984). In other plants, phenotypic traits characteristic of cultivated populations and potentially of domestication are not necessarily genetically-fixed in the same way nor readily archaeobotanically visible (Barron et al., 2022; Hather, 2000), rather they represent adaptive responses to growth environment, can become inherited through sexual and asexual modes of reproduction, and are domestication tendencies rather than syndrome traits (Denham et al., 2020; McKey et al., 2010). The transmission of desired traits can become genetically “fixed”, but often entails less well-defined processes of inheritance through environmentally-driven epimutations, potentially invoking mechanisms such as DNA methylation, chromatin (histone) modifications and RNA-mediated regulation (Rajpal et al., 2022). In contrast to the genetic fixation of phenotypic traits driving domestication documented in cereals, the plastic responses of USOs to different growth environments and cultivation practices probably drove phenotypic selection, epimutational change and the emergence of “domesticated” phenotypes.

Within USOs there may not always be a clear distinction between wild and domesticated plants, rather there may be clines and mosaics of genotype and phenotype that transcend this dichotomy, especially within the natural range of the species. For example, there is ongoing uncertainty of the taxonomy and relationship to wild relatives of cultivated indigenous yams in West Africa (Dansi et al., 2013). Some genomic studies have suggested Dioscorea praehensilis as the wild ancestor and others a hybridisation of D. abyssinica and D. praehensilis (Scarcelli et al., 2019; Sugihara et al., 2020). There has long been a debate whether there are two cultivated species (D. cayenensis and D. rotundata), or whether there are two sub-species of D. cayenensis; while in practice a species complex is often invoked (Condé et al., 2025; Gosline et al., 2023; POWO, 2023). The subspecies/species cannot be distinguished using above ground plant-parts; and although tuber colour, dormancy and cooking properties are often considered, they are not reliable (Coursey, 1967). Yet, cultivated yams can develop genetic changes and phenotypically distinctive cultivars even though they are clonally propagated (Lebot and Dulloo, 2021; McKey et al., 2010).

In West and East Africa, wild and feral yams are documented ethnographically as transplanted and incorporated into cultivated plots (Condé et al., 2025; Hildebrand, 2007). The emergence of desired traits in wild yams, or “ennoblement”, is widely reported by farmers as occurring within several years under cultivation (Zannou et al., 2006), yet yams can continue to grow feral once cultivation is abandoned and become difficult to distinguish from wild yams (Dumont and Vernier, 2000). Transplantation of wild and feral yams from forests potentially selected for varieties with greater plasticity (McKey et al., 2010). Furthermore, new yam plants can derive from sexual reproduction yielding novel genotypes that are then clonally propagated (Chaïr et al., 2010). The introduction of novel genotypes via the cultivation of wild, feral or seedling yams has been interpreted as creating a geographic and temporal mosaic of D. cayenensis-rotundata complex subgroups across West Africa (Dumont et al., 2005). This conceptualisation of temporally and spatially complex cultivar mosaics can be broadened to include Ethiopia, where a high diversity in historic cultivars and “ennoblement” suggests parallel domestication processes (Hildebrand, 2007; Kagnew et al., 2025; Worojie et al., 2021). Modern yam biogeography is currently being re-shaped by the displacement of historic cultivars and practices of “ennoblement” following the introduction of new “commercial” varieties; yet, this disappearing agrobiodiversity is critical for uncovering domestication processes at a continental scale (Condé et al., 2025). The Dioscorea cayenensis-rotundata species complex highlights how the boundary between wild and domesticated can blur, is agronomically contingent, and is genetically uncertain.

From linear domestication pathway to genealogical array: an example

Domestication discourses focus on tracking in time the emergence of phenotypic traits that are present in the major commercial crop today. There are exceptions to these types of domestication narrative, such as Timopheev's wheat (Triticum timopheevii; Filipović et al., 2024), a wheat species that was only identified in archaeobotanical assemblages in recent decades and has an independent genetic lineage. Another example is the reconstruction of the ancestry of a now extinct, local, six-row barley (Hordeum vulgare) in the Nile Valley that has a two-row phenotype due to two sterile lateral bracts, possibly an adaptation to aridity, which was identified through genomic analyses of desiccated, archaeobotanical remains (Palmer et al., 2009).

In most domestication histories specific cultivar lineages of a crop are foregrounded and a wide array of cultivars and crop wild relatives that may have existed through time are backgrounded. Most domestication narratives are teleological and linear, they do not usually reflect the past mosaics of exploited plants of a species – including wild ancestors, cultiwild (asexually cultivated wild plants) and cultivated plants – nor the range of practices of exploitation that occurred through time. Here we reflect on the way domestication discourses are constructed with reference to modern sugarcane (Saccharum cvs.).

The geodomestication pathway for modern plantation sugarcane (Saccharum cvs.) can be traced back to the wild precursor species Saccharum robustum, which grows in the circum-New Guinea region (Brandes, 1956; Daniels and Roach, 1987; Denham, 2025; Garsmeur et al., 2025; Pompidor et al., 2021). The domestication process is multi-staged and has largely been constructed from phytogeography, linguistics and genetics (Figure 2; since Brandes, 1956). The archaeobotany of cane grasses within the region of domestication is limited (Pritchard et al., 2019), with the earliest dates reported for sugarcane being 1120–1210 AD at Quseir al-Qadim, an Islamic port on the Egyptian Red Sea (Van der Veen and Morales, 2011).

Saccharum robustum was initially exploited on New Guinea for food and non-food, including for boundary markers, construction, medicine and consumption as a vegetable (young stems and shoots) (Figure 3B; hereby following Denham, 2025). Saccharum robustum was likely transplanted as a cultwild, namely a phenotypically wild plant that is vegetatively transplanted by people (Figure 3C; following De Langhe et al., 2009). Three types of domesticate arose from the plant (Figure 3D): S. robustum for fencing, which was transplanted to the circum-New Guinea region; S. officinarum for sugar-rich pith, which became important for subsequent derivation of sugarcane; and, S. edule for a vegetable, which is a clonally propagated, sterile mutant that remained localised to the circum-New Guinea region. Subsequently, S. officinarum was taken into Southeast Asia and beyond, leading to three groups of derived cultivars, a more sugar-rich S. officinarum and two Saccharum officinarum x Saccharum spontaneum hybrids (Figure 3E). Saccharum spontaneum is a widely distributed grass within Southeast Asia and neighbouring areas; the hybrids are S. barberi, regionally important in South Asia and beyond, and S. sinense, regionally important in southern China. Modern, plantation sugarcane (Saccharum cvs.) was a product of deliberate crop breeding within the last 100 years or so and comprises interspecific hybridisation between cultivars of S. officinarum and S. barberi, as well as wild S. spontaneum, followed by backcrossing with S. officinarum (Figure 3F; Pompidor et al., 2021).

Although the chronological construction of the geodomestication history of Saccharum spp. traces from the past to the present, the actual line of methodological inference goes from the present to the past. Present inferences of cane grass phytogeography and genetics, as well as historical linguistics, track from present day distributions and characteristics into the past (Figure 4). These initially enable the approximate identification of genetic contributors to modern, plantation sugarcane (Saccharum cvs.) (Figure 4A), which can then be traced biogeographically (Figure 4B) back to the S. robustum populations on New Guinea that initially gave rise to sugar-rich variants (Figure 4C). These latter populations are now of considerable interest for crop improvement (Pompidor et al., 2021).

However, this line of historical inference tracks geographically across a vast array of wild plants, cultiwilds and cultivars that were co-present through time (Figure 4D). When we foreground the linear trajectory of domestication, we background the enormous diversity in cultivars, cutiwilds and crop wild relatives that formed and were distributed across space-time (Figure 4E). A linear interpretation of domestication (Figure 4B) cements teleological thinking about the past whereby we collapse history, namely a genealogical array of agrobiodiversity and crop wild relatives that existed through time, into what we value as significant in the present, namely, the linear ancestors of modern, plantation sugarcane. We tend to focus on what contributes to this linear pathway and ignore the rest (Figure 4F).

Yet, if we build our interpretations from the past forward, we see that the historical processes involved are not linear. Rather, we see considerable diversity in exploited plants and practices across the Island Southeast Asian-New Guinea regions at different points in the past. Mosaics of different wild plant, cultiwild and cultivar assemblages and associated practices of management, transplantation and cultivation, as well as the resultant gradients of domestication, characterised agriculture across the Island Southeast Asia and New Guinea regions during the mid-to-late Holocene (Barton and Denham, 2011, 2018; Denham, 2005, 2011). These shared domesticatory dispositions created the mosaics of agrobiodiversity that were the agronomic context from which major sugarcane lineages arose (see Garsmeur et al., 2025). Many cultiwilds and cultivars have remained locally important, or faded from history as they were replaced with others. This genealogical array constitutes the greatest majority of the domestication history for cane grasses, but is often sidelined in discourses that focus on one or a few lineages. Comparable mosaics across these regions were important for understanding the emergence of major lineages of banana cultivars (Musa cvs.) (Martin et al., 2023; Perrier et al., 2011; Sardos et al., 2022), with renewed interest in local cultivar diversity and crop wild relatives of Musa spp. across New Guinea and Southeast Asia (Eyland et al., 2021).

Rewilding and resocialising the past

Each of the sections above has sought to contrast more mainstream archaeological discourses on domestication with alternative and emerging perspectives within the discipline drawn from our own research on vegetatively propagated crops in the tropics and subtropics. Although there continues to be a heavy focus on a handful of crops from a few global regions, especially cereals, pulses and some fruits that are highly visible in the archaeobotanical record, there is now a growing body of research on more diverse crops and regions that bring subaltern histories to a global audience, such as the domestication of root and tuber crops, bananas and cane grasses. For some, these intellectual activities are committed and deliberately engaged activities to resist dominant narratives and give voice to those marginalised and disenfranchised in our archaeological reconstructions of the past. Others have sought to develop methods that bring archaeobotanical visibility to groups in the past, such as archaeological parenchyma, thereby broadening the scope of the plants, associated technologies and types of plant exploitation practice that can be investigated and brought into archaeological discourse.

At the same time, a conceptual reframing seeks to recast domestication as a way point, rather than the end point, along a backdrop of long-term, human-plant domesticatory relationships. This conceptual recasting is illustrated by the often ambiguous, phenotypic and genotypic discrimination of wild, cultivated and feral yams in West Africa. Yam plants can readily move among these categories rather than being readily defined by them.

Another perspectival shift aims to highlight the genealogical array of agrobiodiversity and crop wild relatives in the past, rather than collapse this temporal and geographical diversity into a linear and genealogical, domestication pathway through time-space. Defocusing domestication brings into view past and present crop wild relatives and agrobiodiversity beyond the major cultivar groups and modern crop lineages. A better understanding of past agrobiodiversity directly informs a better understanding of present-day agrobiodiversity. Whilst the conservation and diversification of cultivated crops is increasingly being advocated (Ulian et al., 2020), a wide range of agrobiodiversity continues to be lost and remains at risk of loss especially among local subsistence communities in the tropics and subtropics (Khoury et al., 2022).

Looking forwards, and beyond our scope here, these reflections bring into focus a multi-dimensional, ethno-ecological context for conceptualising human-plant domesticatory relationships (Figure 5; see Martin, 2004). Most archaeobotanical literature on domestication focusses on the phenotypic, as well as increasingly genotypic, traits of useful plants in the past, often divorcing them from: associative layers of cultural meaning, including the people, their knowledge (from practical to symbolic), places and practices; the agroecology of the plant's growth environment, including the landscape and soil; and, the Indigenous technical knowledge, or practical knowledge, associated with plant exploitation, including managing/cultivating, harvesting/gathering, storage/processing, and cooking/use. The multi-dimensionality of plants, their ethno-ecological contexts, are often poorly articulated, investigated and discussed in archaeological discourses of domestication. Partly these absences reflect the archaeological invisibility of cultural meanings for many societies in the past, yet these social contexts should not be ignored where possible, because they provide clues as to the “how” and “why” plants were cultivated by local communities in the past. Thus, not only are plants of increasing interest for agricultural heritage, crop improvement and Indigenous connections, they also bring to the fore the Indigenous Technical Knowledge (ITK), ways of life and cultures of people who have hitherto been living beyond the margins of archaeological discourse.

Generative AI statement

The author(s) declared that generative AI was not used in the creation of this manuscript.

Any alternative text (alt text) provided alongside figures in this article has been generated by Frontiers with the support of artificial intelligence and reasonable efforts have been made to ensure accuracy, including review by the authors wherever possible. If you identify any issues, please contact us.

Publisher's note

All claims expressed in this article are solely those of the authors and do not necessarily represent those of their affiliated organizations, or those of the publisher, the editors and the reviewers. Any product that may be evaluated in this article, or claim that may be made by its manufacturer, is not guaranteed or endorsed by the publisher.

Footnotes

Edited by: Huw Barton, University of Leicester, United Kingdom

Reviewed by: Héctor Andrés Peña, University of the Basque Country, Spain

Laura Rival, University of Oxford, United Kingdom