Plant domestication dates back more than 10,000 years (Pickersgill 2016). It is an evolutionary process; a wild species is gradually transformed into a cultivated species via artificial trait selection instead of natural selection (e.g. Blanckaert et al. 2012; Chacón-Sánchez 2018; Gonçalves-Dias et al. 2023). Several categorization systems have been proposed to describe the different degrees of domestication, but the two most widely cited are Clement’s (1999) system, which includes wild, incidentally co-evolved, incipiently domesticated, semi-domesticated, and fully domesticated; and that of Meyer et al. (2012), who reduced the categories to domesticated, semi-domesticated, and undomesticated.

Other authors defined the concept of domestication syndrome as a set of morphological traits that differentiate domesticated plants from their wild ancestors and are usually modifications that resulted from human selection (Harlan 1971; Hammer 1984; Meyer et al. 2012; Denham et al. 2020; Pacheco-Huh et al. 2021). Typical features of a domestication syndrome, often resulting from selection for preferred characteristics include changes to secondary metabolism (e.g. reduced bitter compounds, increased sugars, and pigment changes), plant architecture (e.g. modified branching, stem or inflorescence structure, plant height, apical dominance), leaves (e.g. size, shape, arrangement), and fruits (e.g. increased seed retention, larger seeds/fruits, loss of seed dormancy, fruit/seed colour) (Brenner et al. 2000; Rana et al. 2005; Meyer et al. 2012; Joshi et al. 2018; Purugganan 2019; Denham et al. 2020; Mboujda et al. 2022). Plant species have undergone different degrees of domestication, usually in relation to their economic importance. Some are fully domesticated such as maize, which has large seeds that have lost their dormancy and dispersal capacity, and reduced branching and photoperiod sensitivity (Stetter et al. 2017). Other crops such as lentils, quinoa, sorghum, and grain amaranths apparently have poorly demarcated domestication syndromes (Li and Siddique 2018; Ruth et al. 2021; Chapman et al. 2022).

The domestication syndrome of the grain amaranths is still under debate. Some have noted that traits such as seed size and seed retention are shared with their wild relatives (Sauer 1967; Stetter et al. 2020). Stetter et al. (2017) reported that seed colour in A. caudatus is associated with domestication: most cultivated species have pale seeds, while wild amaranth species have dark brown seeds. These authors also revealed a significant loss of genetic variation in domesticated amaranths, suggesting strong agricultural and cultural selection. Based on domestication traits (inflorescence shape, seed shattering, and seed size), which are shared by both wild and cultivated species in Amaranthaceae, Stetter et al. (2017) suggested that the domestication of amaranths may not fit neatly into the traditional model of domestication for well-established crops, proposed by Hammer (1984). Moreover, the phylogenetic and taxonomic status of the putative ancestors of the grain amaranths, A. hybridus and A. quitensis Kunth, is in question (Kietlinski et al. 2014; Sánchez-del Pino et al. 2025). Stetter et al. (2017) concluded that weak artificial selection for seed colour and a high level of gene flow between A. caudatus and A. quitensis has counteracted the fixation of domestication traits.

Amaranthus cruentus is an annual plant native to Mexico and Guatemala but is also currently cultivated in the United States, Argentina, and China (Das 2016a; Wolosik and Markowska 2019). In Mexico, the cultivation of A. cruentus declined notably after the arrival and conquests of the Spaniards, relegating it to an underutilized crop with restricted cultivation and limited economic relevance (Joshi et al. 2018).

Archaeological findings, such as in the Coxcatlán caves of Tehuacán, Mexico, suggest that amaranth was crucial to the Aztec civilization and as such extensively cultivated in Mexico by the 15^th century. The domestication probably began more than 6000 years ago (Joshi et al. 2018; Turner et al. 2021). Sauer (1967: 123) mentioned: “Amaranthus cruentus evidently originated as domesticated grain crop somewhere in southern Mexico or Guatemala, the only region where it is found in aboriginal cultivation within the range of its probable progenitor, Amaranthus hybridus”. He also proposed that each grain amaranth has its own “progenitor” at a different centre of domestication. Although the precise origin of cultivated amaranths remains unclear, A. hybridus has been widely recognised as the most likely ancestor (Kietlinski et al. 2014; Stetter et al. 2020; Gonçalves-Dias et al. 2023).

Amaranthus cruentus produces seeds (pseudo-grains, Sánchez-del Pino et al. 2025) that are consumed like a cereal grain and are of high value (Aguilar et al. 2015; Das 2016b; Adeniji 2018; Manyelo et al. 2020; Araujo-León et al. 2023). The International Union for the Protection of New Varieties of Plants (UPOV 2024) reports 29 varieties of A. cruentus from Mexico, Russia, France, the European Union, among others. The National Plant Germplasm System of the United States Department of Agriculture (USDA) holds 428 accessions of A. cruentus (364 active, 349 available), including “landraces” [sic] from Mexico, Guatemala, the United States, India, and China. Additionally, Kauffman (1992) described “morphological groups” of A. cruentus from Guatemala and Mexico, which are clusters of accessions with distinctive traits resulting from traditional selection and might each represent distinct “landraces” [sic] (National Plant Germplasm System 2025).

Three local races of A. cruentus (Mexican, African, and Guatemalan) have also been described, each with specific “phenological” and phenotypic characteristics generated through artificial selection in diverse cultural contexts (Kauffman 1992; Espitia-Rangel et al. 2010). Kauffman’s (1992) “morphological groups” refer to variation at the accession or “landrace” level, whereas the three local races correspond to broader, geographically structured categories recognised later. These local races illustrate how hybridization and selection have shaped the diversity and adaption of A. cruentus across different regions (Espitia-Rangel 2018).

Although we introduced features that are associated with the domestication syndrome in plants, only seed colour has consistently been identified as a domestication trait for grain amaranths (Sauer 1967; Stetter et al. 2017, 2020; Sánchez-del Pino et al. 2025). Attempts to define a domestication syndrome for the grain amaranths have mostly described generalities, neglecting that each crop has its own evolutionary history and thus its own domestication syndrome. Martínez-Núñez et al. (2019) proposed a method to use amaranths as model plants based on their life cycle. We propose that observations and measurements of traits at different phenological stages can be used to identify a set of characters that pertain to the domestication syndrome.

Studies to assess morphological traits of cultivated amaranths (e.g. Hauptli and Jain 1978; Sogbohossou and Achigan-Dako 2014; Thapa and Blair 2018) did not systematically evaluate traits that resulted from domestication in any of the proposed domestication centres. Our work is the first comparative morphometric analysis of A. cruentus in its postulated centres of domestication (Mexico and Guatemala) including both wild (A. hybridus) and cultivated accessions. Using populations of A. cruentus from Mexico and Guatemala, we sought to identify a set of characteristics based on measurements of morphological characters at different developmental stages of the plant. We then tested their usefulness, in particular considering Sauer’s (1950, 1967) main hypothesis, by evaluating whether this set of traits can reveal different degrees of domestication in the putative centres of domestication proposed by Sauer. We also formulated the following questions: Can we identify a set of traits that unambiguously separates wild, ancestral A. hybridus from cultivated A. cruentus? Can this set be used to determine the level of domestication in populations from different centres of domestication of A. cruentus?

Among the 221 individuals belonging to 60 accessions from Mexico and Guatemala that were used, 54 accessions were Amaranthus cruentus (169 individuals) and six accessions were its wild relative Amaranthus hybridus (52 individuals). Fifty-one accessions of A. cruentus were obtained from the National Plant Germplasm System of the United States Department of Agriculture-Agricultural Research Service (National Plant Germplasm System 2025), and three were collected in situ. Thirty-six accessions are from Mexico and 18 from Guatemala. Material of the wild species A. hybridus is challenging to obtain due to limited representation in seed banks and herbarium collections, and because it is often considered a weed, it has received less attention. The six accessions of A. hybridus were collected in Mexico and served as reference species (voucher information: Suppl. material 1; distribution: Fig. 1). Plant identifications were based on the original sources (seed banks and personal collections). Once the plants were cultivated and reached maturity, the original source was checked using the diagnostic keys developed by Sánchez-del Pino et al. (2019). Herbarium specimens of the collected plants were deposited in CICY and UVAL (Thiers 2025).

The study was conducted at the Centro de Investigación Científica de Yucatán (CICY) from January to June 2023. Fifteen seeds for each accession (Suppl. material 1) were germinated under controlled conditions in a greenhouse (22–27°C). After 15 days, up to five of the most-developed seedlings of A. cruentus and a maximum of 10 of A. hybridus were transplanted into a terraced area in a randomized block design to minimise the effect of differences in natural light, that could influence plant size. Plants were observed and measured at three stages: vegetative, anthesis, and seed maturation (Fig. 2; Martínez-Núñez et al. 2019). Although amaranths are predominantly autogamous, to prevent cross-pollination between the cultivated and wild species, we covered each inflorescence of both species with a 30 × 38 cm glassine bag during anthesis. In the greenhouse, plants were irrigated every third day throughout their life cycle. No fertilizers were applied. Pest control was implemented only when infestations were observed (for further details, see Xingú-López 2024).

All 221 plants available for the 60 accessions were measured at the vegetative stage (20 days post-seeding), anthesis stage (60 and 150 days post-seeding), and seed maturation (90 and 180 days post-seeding; Fig. 2).

Eight quantitative and two qualitative traits that are associated with domestication syndromes were considered. Quantitative traits were 1) plant height, 2) basal stem diameter, 3) leaf length, 4) leaf width, 5) terminal inflorescence length, 6) duration of the vegetative stage, 7) dry mass of inflorescence, and 8) seed yield (total mass). Qualitative traits were inflorescence and seed colour. Plant height (1) was measured with a standard tape measure using the method of Vazquez et al. (2011) and Ortiz-Torres et al. (2018). Basal stem diameter (2) was measured 5 cm above the ground using a digital micrometre (0–0.25 mm, Fowler, Massachusetts USA) as specified by Reinaudi et al. (2011). Leaf length (3) and width (4) were measured at the vegetative stage and anthesis. Starting at anthesis on, the leaves begin to senesce, then fall. Three to four randomly selected, well-developed leaves were marked to track length and width during both developmental stages. If the marked leaf had abscised, the closest leaf was used. Terminal inflorescence length (5) (from the last foliage leaf to the tip of the inflorescence) was measured at anthesis and seed maturation (Vazquez et al. 2011). Duration of the vegetative stage (6) was based on daily observations until the plant reached anthesis. Dry mass of inflorescences (7) was measured at seed maturation. The inflorescences were harvested individually and dried in a climate-controlled room at 38 ± 1°C and 25–30% relative humidity. After drying under controlled room conditions (approximately 12% relative humidity), seed and dried inflorescences were weighed using a precision balance (220 g/ 0.1 mg, Ohaus Explorer, EX224, USA). Seed yield (total mass) (8) was weighed when seeds were mature. All seeds were removed manually from the inflorescences on one plant and weighed using a precision balance (220 g/ 0.1 mg, Ohaus Explorer, EX224, USA). Values were recorded in grams of total seed per plant.

The qualitative traits were assessed using the guidelines in the “Graphic Handbook for Variety Description in Amaranth (Amaranthus spp.)” by the National Seed Inspection and Certification Service (CP-SNICS 2006).

PERMANOVA was used to test for significant differences in values for the morphological traits at each developmental stage between A. cruentus and A. hybridus, and among A. cruentus populations from different regions, because it is a non-parametric approach that does not rely on distributional assumptions, such as normality or homogeneity of variances. This method is particularly useful for analysing complex, high-dimensional datasets or when the assumptions of an ANOVA are not met (Dwivedi et al. 2019; Hamidi et al. 2019), as was the case for our data. Our samples comprised 169 individuals of A. cruentus and 52 of A. hybridus, which might cause a bias when using ANOVA; therefore, the small number of A. hybridus accessions did not influence the outcome of this study results when using PERMANOVA. Moreover, PERMANOVA was used to detect differences between Mexican and Guatemalan accessions of A. cruentus.

Minimum, maximum, and mean values and standard deviations were determined for each trait and compared between A. cruentus and A. hybridus at each of the three developmental stages using PERMANOVA in the R package vegan v.2.6-10 (Oksanen et al. 2025).

Nonmetric multidimensional scaling (NMDS) in the R package vegan v.2.6-10 (Oksanen et al. 2025) was used to visualise the similarity structure among samples of 1) A. cruentus and A. hybridus, 2) Mexican and Guatemalan populations of A. cruentus, in a lower-dimensional space based on the morphological traits at each stage. Pearson’s correlation analysis in R package stats v.4.4.0 (R Core Team 2024) was used to analyse how the traits are correlated within and between both species (Morales Saavedra et al. 2019; Ciccarelli and Bona 2022).

Plant height, basal stem diameter, leaf length and width, terminal inflorescence length, inflorescence dry mass, and seed yield were included in multivariate statistical analyses. Duration of the vegetative stage, inflorescence and seed colour were evaluated as qualitative descriptors and not included in the multivariate tests. The frequency of the occurrence of the qualitative characters was determined in each accession.

Our results revealed substantial morphological differences between the cultivated and wild species in plant height, leaf length and width at the vegetative stage; inflorescence length and duration of the vegetative stage at anthesis; and plant height, dry mass of the inflorescence, and seed yield (total mass) at seed maturation (Table 1). At the vegetative stage, plants of A. cruentus were taller and had larger and wider leaves compared to A. hybridus. The duration of the vegetative stage in A. hybridus was 49 days after transplant, compared to 100 days for A. cruentus. Amaranthus cruentus developed significantly larger inflorescences than A. hybridus. At seed maturation, mean plant height for A. cruentus was significantly greater than for A. hybridus and A. cruentus had substantially greater inflorescence dry mass and seed yield (total mass) compared to A. hybridus.

The PERMANOVA results (Table 2) showed the greatest differentiation between the two species at anthesis (F = 103.61; p < 0.001). At seed maturation, the differences remained significant (F = 42.049; p < 0.001), but the variance decreased to 20.1%. The vegetative stage had the lowest level of differentiation (F = 30.226; p < 0.001), with only 12.18 % of the variance explained. The NMDS scatter diagram (Fig. 3) illustrates the variation between the two species across the growth stages: the traits of the cultivated species appear more clustered than those of the wild relative.

Several positive correlations were found between morphological traits of individual plants for the two Amaranthus species at different stages of growth (Table 3), reflecting trait variation across the two species. At the vegetative stage, plant height, stem diameter, and leaf length and width were positively correlated. At anthesis, positive correlations persisted between plant height and leaf length and width. At seed maturation, inflorescence dry mass and seed yield were strongly correlated, and stem diameter was also positively correlated with inflorescence dry mass and seed yield.

Furthermore, correlations among the morphological traits were examined within each species separately to explore trait interrelationships. Plant height, stem diameter, leaf length, and leaf width were strongly positive correlated at the vegetative stage in A. cruentus and A. hybridus. At anthesis, positive correlations among plant height, leaf length, and leaf width persisted in both species except for stem diameter, which was only positively correlated at this stage for A. cruentus. During the seed maturation stage, positive correlations among plant height, stem diameter, terminal inflorescence length, inflorescence dry mass, and seed mass were present in both species. However, the correlation between terminal inflorescence length and seed mass was comparatively weaker in A. hybridus (Suppl. material 2).

At anthesis, inflorescences of A. cruentus had a wider range of colours (shades of green, purple, pink, and red) than those of A. hybridus, which are green (Fig. 4). However, the most common inflorescence colour in A. cruentus was green (Fig. 5A).

Mature seeds of A. cruentus exhibited a wide range of colours (white to yellow, brown, and black), but white was most common. Seeds of A. hybridus were always black (Figs 5B, 6).

The comparative morphometric analysis of the traits at three growth stages showed significant differences between the accessions from Mexico and Guatemala (Table 4) in leaf length and width at the vegetative stage; plant height, basal stem diameter, leaf length, leaf width, and terminal inflorescence length during anthesis; and plant height, basal stem diameter, terminal inflorescence length, inflorescence dry mass, and seed yield at seed maturation stage. During the vegetative stage, Mexican accessions had larger and wider leaves than the Guatemalan (Table 4). At anthesis, Mexican accessions had taller plants, thicker basal stems, larger leaves, and considerably longer terminal inflorescences (43 vs 37 cm in Guatemalan accessions; Table 4). At seed maturation, Mexican accessions consistently produced taller plants (262.6 vs 213.0 cm in the Guatemalan accessions; Table 4), thicker basal stems, and longer and heavier inflorescences, resulting in greater dry biomass (51 vs 37 g, respectively; Table 4). The Mexican accessions also had higher seed yields (mean 13.5 g, max 43.5 g) compared to the Guatemalan accessions.

The PERMANOVA analysis (Table 5) comparing the Mexican and Guatemalan accessions of A. cruentus revealed significant differences during the vegetative stage (F = 7.98; p < 0.004) and seed maturation (F = 5.57; p < 0.012), but not during anthesis (F = 2.39; p > 0.1). In the NMDS ordination analysis (Fig. 3), individuals of A. cruentus from Mexico and Guatemala during the vegetative stage were not clearly separated by country (Fig. 3D). In contrast, at anthesis (Fig. 3E) and seed maturation (Fig. 3F), individuals were more dispersed in the ordination space, reflecting increased morphological variation within the species.

Several positive correlations among the accessions of A. cruentus were observed (Table 6): at the vegetative stage, plant height, stem diameter, and leaf length and width; at anthesis, plant height, stem diameter and leaf length and width; at seed maturation inflorescence dry mass with seed yield, terminal inflorescence length and stem diameter with inflorescence dry mass and seed yield.

In the NMDS plot of the 10 characters, the points for A. hybridus are in a very dispersed cloud, whereas the points for A. cruentus are concentrated inside the larger cloud for A. hybridus (Fig. 3). The higher number of A. hybridus points reflects our sampling strategy, in which up to 10 individuals were selected per accession, so that traits for 52 individuals were measured throughout their life cycle. This finding suggests that there is an overall distinction between the two species. Amaranthus hybridus and A. cruentus share some characters, which should not be surprising since A. hybridus is considered to be the ancestor of A. cruentus. However, several domestication-linked characters distinguish A. cruentus from A. hybridus (Sánchez-del Pino et al. 2025: fig. 8I–K). According to our results (Table 2), the vegetative stage exhibits the lowest level of differentiation between the two species, suggesting that in the earlier developmental stages there is little or no morphological distinctions between the two species studied. In A. cruentus, the vegetative phase appears to retain ancestral traits as observed for other crops (Jisha et al. 2011).

At anthesis and seed maturation, the NMDS analysis (Fig. 7) showed a more dispersed cloud for the overall set of characters in A. cruentus from Mexico than in A. cruentus from Guatemala, suggesting that the two populations are distinct. At the vegetative stage, both populations are dispersed without distinction, which is not surprising considering that the relevant characters for domestication are mostly characters linked to the inflorescence or characters that are realised late in the life cycle. It should be noted that the Mexican dataset includes more accessions than the Guatemalan one, which may be related to the lower collection rates in Guatemala compared with Mexico (Ivonne Sánchez-del Pino pers. comm.).