2.1 Sex chromosomes in A. palmeri

Data regarding the emergence of sex chromosomes as an evolutionary process in dioecious species is limited, especially for the Amaranthus genus, where both dioecious and monoecious species exist. It is likely that dioecy evolved independently from hermaphroditic ancestors in different plant families and genera (Ming et al., 2007). Using whole-genome approaches to assemble male-specific sequences, researchers identified a region of approximately 1.3–2 Mb associated with sex (Montgomery, Giacomini, Weigel, & Tranel, 2021; Neves et al., 2020), and have proposed a XY chromosomal model for sex-determination system in A. palmeri. In A. palmeri and A. tuberculatus, respectively, 345 and 2754 male-specific sequences, were identified designing sex-specific markers for both species with high degree of accuracy (Montgomery et al., 2019). The recent genome sequencing and explorations of male and female transcriptomes and genomes have provided male-specific Y molecular markers, insights on sex determination, and how dioecy evolved independently in A. tuberculatus and A. palmeri. Subsequently, a genome assembly of A. palmeri revealed that the male-specific Y region is approximately 1.3 Mbp long with 121 predicted genes, whereas in A. tuberculatus it is 4.6 Mbp and contains 147 genes (Montgomery, Giacomini, Weigel, & Tranel, 2021). Male A. palmeri and A. tuberculatus male-specific Y regions appeared to not be syntenic, suggesting that dioecy evolved independently in both species, leading to the hypothesis that dioecy has evolved multiple times in this genus (Montgomery, Giacomini, Weigel, & Tranel, 2021). The implications of these findings are not restricted to basic science and the evolution of sex-determination system. We will provide some suggestions later in this review about how the genetics of sex determination could be an ‘Achilles heel’ to target for RNAi gene silencing.

2.2 Hybridization with close Amaranthus species, molecular methods to explore differences among species

Identification of Amaranthus species based on morphology can sometimes be difficult due to similarities among them, especially in young growth stages and seeds (Matzrafi et al., 2025). However, it is highly important since different species have distinct mating systems and respond differently to management practices. Interspecific crosses have been demonstrated to occur under experimental conditions between several Amaranthus species, and hybridization deserves special attention as hybrids can result in the transfer of alleles associated with adaptation (e.g., herbicide resistance) from one species to another, making weed management more difficult. The rate of interspecific hybridization between A. Palmeri and other Amaranthus species has been investigated in many studies. Gaines et al. (2012) reported that, in field experiments, hybridization rates from 0.01% to 0.4% occurred between a glyphosate-resistant male A. palmeri and A. spinosus, less than 0.2% with A. tuberculatus, and less than 0.01% with A. hybridus. In greenhouse conditions, A. palmeri × A. spinosus hybridization rate was up to 1.4%. Increased 5-enolpyruvylshikimate-3-phosphate synthase (EPSPS) copy number, a mechanisms of glyphosate resistance, was found to be inherited in some of the hybrid offspring (detected through q-RT-PCR). Interestingly, hybrids between A. palmeri and A. hybridus show male and female plants, whereas hybrids of A. palmeri and A. spinosus show male and monoecious inflorescences (Gaines et al., 2012). Amaranthus palmeri × A. powellii and A. palmeri × A. retroflexus hybrids were not identified in the same study, and to the best of our knowledge, neither they have been reported elsewhere.

Several PCR-based molecular markers have been developed to confirm the occurrence of interspecific hybridization in Amaranthus species. Internal transcribed spacer 1 (ITS1) and ITS2 sequences and restriction enzymes can successfully distinguish eight Amaranthus species (A. albus, A. arenicola, A. blitoides, A. hybridus, A. palmeri, A. powelli, A. spinosus, and A. retroflexus), with A. rudis and A. tuberculatus displaying unique digestion profiles from the others, but not distinct from each other (Wetzel et al., 1999). Similarly, a Kompetetive Allele Specific PCR (KASP) assay has been used to detect hybrids between A. palmeri and A. tuberculatus and gene flow of mesotrione resistance (Oliveira et al., 2018). The acetolactate synthase (ALS) gene is known to have species-specific polymorphisms and have been used for species determination and hybrids detection. For example, position 678 in this gene has a thymine in A. palmeri and a cytosine in A. tuberculatus (Tranel, & Wright, 2002). This has been used to differentiate both species as well as their hybrids through a restriction enzyme (Mfe1) at that locus (Franssen et al., 2001). Also, a KASP assay based on the same ALS single nucleotide polymorphism (SNP) has been used for species determination in a population of dioecious Amaranthus species introduced in Brazil (Küpper et al., 2017). Recently, an A. palmeri-specific SNP was identified that can be used to distinguish a seed of A. palmeri in up to 200 seeds of other mixed Amaranthus species (Brusa et al., 2021).

Cytogenetic tools have also been explored for Amaranthus identification. Flow cytometry identified a mean 2C genome size values of 1.04 and 1.34 pg, respectively, for A. hybridus and A. tuberculatus (Jeschke et al., 2003). Amaranthus palmeri genome size was estimated to be the smallest (0.96 pg) when compared to five other Amaranthus species (2C values ranging from 1.04 to 1.23 pg) (Rayburn et al., 2005).

With the reduction of costs of sequencing techniques, we expect that more sequence data becomes available for most Amaranthus species. Species-specific markers can then be developed to determine Amaranthus species and their hybrids, particularly in young plants or seeds, as demonstrated in the case of A. palmeri (Montgomery et al., 2019). In most of the cases described here, hybrids have been studied using bi-parental crosses with herbicide-resistant and -susceptible progenitors, since gene flow of resistance is relatively easy to assess phenotypically. However, the ability to detect hybrids occurring naturally, where more than one species is present, might be critical for designing effective control practices. Purely relying on herbicide resistance as a marker of putative hybrids has some limitations as it requires accurate application and phenotyping. Other limitations of these detection methods are that they depend on the mechanism of resistance and the dominance of that trait in heterozygous individuals (Gaines et al., 2012; Montgomery, Giacomini, & Tranel, 2021). It should also be noted that facultative apomixis can occur in A. palmeri, where tissue from the ovule is transformed into an embryo, avoiding the processes of meiosis and fertilization (Ribeiro et al., 2014), further complicating hybrid studies when studying gene flow from a different species into a female A. palmeri individual. Therefore, for interspecific hybridization diagnostics, we encourage utilizing multiple techniques for more robust conclusions.

4.1 Inheritance of herbicide resistance traits in A. palmeri

To date, A. palmeri has evolved resistance to nine herbicide SOAs: microtubule, ALS (Horak & Peterson, 1995; Nakka, Thompson, Peterson, & Jugulam, 2017), EPSPS (Culpepper et al., 2006), glutamine synthetase (GS) (Priess et al., 2022), protoporphyrinogen oxidase (PPO) (Salas et al., 2016), photosystem II (PSII) (Nakka, Godar, Thompson, et al., 2017), 4-hydroxyphenylpyaunate dioxygenase (HPPD) (Nakka, Godar, Wani, et al., 2017), and very long-chain fatty acid (VLCFA) (Brabham et al., 2019), microtubule inhibitors (Gossett et al., 1992) as well as auxin mimics (Shyam et al., 2022). A variety of resistance mechanisms have been identified in A. palmeri which may vary based on a population's specific evolutionary history, or the selection pressures that were applied (Table 1). In A. palmeri, target-site resistance (TSR) mechanisms include SNPs, codon deletions, and copy number variation of genes encoding herbicide target enzymes, and are commonly found for ALS, PPO, EPSPS and PSII inhibitors (Table 1). The major non-target site resistance (NTSR) mechanism identified in A. palmeri is increased herbicide detoxification, which was found to confer resistance to inhibitors of ALS, HPPD, PSII, VLCFA, PPO, and auxin mimic (Table 1). Understanding the resistance mechanisms for various modes of action is critical for unravelling aspects of weed evolution, and how these mechanisms spread. For instance, while mutations in nuclear genome-encoded genes can segregate following a single gene model, NTSR mechanisms can be quantitative (Délye, 2013). Furthermore, many NTSR mechanisms have not been fully characterized and, in most cases, their inheritance patterns remain unknown.

A crucial factor that determines resistance inheritance is the location of the resistance gene, whether it is on the nuclear, chloroplastic, or mitochondrial genome. Nuclear encoded genes are transferred to the progeny after recombination, while the chloroplastic and mitochondrial genomes are typically inherited maternally. For instance, in a PSII inhibitor-resistant A. palmeri population, atrazine is conjugated with glutathione by a single nuclear gene (Nakka, Godar, Thompson, et al., 2017; Shyam et al., 2021). However, resistance to PSII inhibitors in other weed species is often caused by mutations in the chloroplastic gene psbA, including in some closely related Amaranthus, such as A. retroflexus, A. hybridus, and A. powellii (Davis et al., 2019; Dumont et al., 2016; Hirschberg & McIntosh, 1983; Park & Mallory-Smith, 2006). The maternal inheritance of chloroplastic alleles can lead to the rapid fixation of mutations.

Mutation in the ALS (A122, P197, T574, and S653) and PPO2 (A128, ΔG210, V361, and G399) have been found in resistant A. palmeri populations (Giacomini et al., 2017; Küpper et al., 2017; Larran et al., 2017; Montgomery, Giacomini, & Tranel, 2021; Nakka, Thompson, Peterson, & Jugulam, 2017; Nie et al., 2023; Rangani et al., 2019; Salas et al., 2016; Singh et al., 2019). The ALS and PPO2 are examples of enzymes that function in plastids and mitochondria, but are nuclear-encoded genes. The degree of dominance of ALS and PPO2 resistance alleles may differ among plant species and/or alleles (Foes et al., 1999; Hart et al., 1993; Sebastian et al., 1989; Wright & Penner, 1998), but these were demonstrated to be at least partially dominant (Patzoldt et al., 2006; Yu & Powles, 2010; Brusamarello et al., 2020), and can be selected for in heterozygous individuals (Giacomini et al., 2017; Küpper et al., 2017).

Enhanced herbicide detoxification, a type of NTSR, has been hypothesized to often be a quantitative trait (Délye, 2013). However, in an atrazine-resistant population the herbicide was found to be conjugated by a glutathione S-transferase enzyme, and further studies revealed that this trait followed a single nuclear gene segregation model (Nakka, Godar, Thompson, et al., 2017; Shyam et al., 2021). Inheritance studies have demonstrated that metabolism of chlorsulfuron and mesotrione are conferred by single and multiple nuclear genes, respectively (Shyam et al., 2021). A recently discovered dicamba-resistant A. tuberculatus trait is potentially polygenic, and although the resistance was predominantly inherited as a nuclear trait, minor maternal effects were not completely ruled out (Bobadilla et al., 2022). Given the complex and potentially polygenic inheritance of NTSR, predicting its inheritance is difficult and there are no rules-of-thumb that apply across all resistance cases.

Amaranthus palmeri stands out from many other weeds by the sheer amount of multiple resistance cases reported (Heap, 2023). It has been demonstrated that target-site resistance mechanisms can spread quickly within and across A. palmeri populations, due in large to copious amounts of wind-blown pollen and obligatory outcrossing in this dioecious species (Jhala et al., 2021; Sosnoskie et al., 2012; Ward et al., 2013). Although these traits are typically single genes, they may stack in one individual through time leading to multiple resistance.

4.2 The unique inheritance of glyphosate resistance in A. palmeri

Glyphosate resistance was first identified in A. palmeri in 2005 (Culpepper et al., 2006) and quickly spread throughout the Southern US and across other regions (Heap, 2023). In 2010, a novel and unique TSR mechanism was discovered: EPSPS gene amplification, also known as copy number variation (CNV). In the case of EPSPS-CNV, resistant plants can have 30 to >150 copies of the gene. More copies of the EPSPS result in greater transcript and protein abundance such that the recommended dose of glyphosate no longer effectively inhibits all EPSPS in the cells (Gaines et al., 2010; Gaines et al., 2011). Despite glyphosate-resistant individuals displaying a massive copy number variation and EPSPS expression, no fitness penalty has been detected in greenhouse experiments (Giacomini et al., 2014; Vila-Aiub et al., 2014). However, in a cotton field, an experiment suggested that individuals possessing high EPSPS gene copy number exhibit reduced fitness (Cahoon et al., 2022). It is important to mention that this reduction was not translated into differences in interference with crop plants. Inheritance studies revealed that EPSPS-CNV is nuclear and dominant but does not follow the single gene segregation model (Gaines et al., 2010; Giacomini et al., 2019; Mohseni-Moghadam et al., 2013). Interestingly, fluorescence in situ hybridization (FISH) revealed that EPSPS copies are associated with all chromosomes in resistant plants (Gaines et al., 2010). Further studies revealed that about half of the EPSPS copies are present in the form of an extra chromosomal circular DNA (eccDNA), physically attached but not inserted into the genome (Koo et al., 2018). The EPSPS-eccDNA displays unique structural polymorphisms, with divergence occurring as consequence of duplication and deletion events. Furthermore, the EPSPS cassette carries an autonomous replication sequence, allowing it to duplicate independently from cell division (Gaines et al., 2019; Koo et al., 2018; Molin et al., 2020).

Various investigations into the origins and spread of the EPSPS-CNV in A. palmeri found that the EPSPS-eccDNA is conserved among resistant populations from geographically distant regions, suggesting a single evolutionary event that quickly spread to multiple parts of the US (Küpper, Manmathan, et al., 2018; Molin et al., 2017; Molin et al., 2018; Molin et al., 2020). It has also been observed that while the EPSPS-eccDNA is conserved between resistant plants, the genetic background is not, suggesting that eccDNA is inherited separately and able to backcross into locally adapted plants (Gaines et al., 2019; Molin et al., 2020). It has been hypothesized that glyphosate-resistant A. palmeri seeds are often dispersed via agricultural machinery, or wind, since they can travel long distances (Oliveira et al., 2018; Sarangi et al., 2017; Sosnoskie et al., 2012).

Amaranthus palmeri can also produce seed through apomixis (Ribeiro et al., 2014). Rates of apomixis may be governed by environmental conditions, thus, putative apomictic female plants are not always present in populations of dioecious species (Hojsgaard & Hörandl, 2019). Ribeiro et al. (2014) confirmed that reproductively isolated glyphosate-sensitive and -resistant female A. palmeri plants produced viable seeds, and that resistance was inherited in apomictic progenies, suggesting that A. palmeri can produce seeds both sexually and asexually (via apomixis), and that apomixis may enhance the stability of the glyphosate resistance trait in resistant populations in the absence of reproductive partners. These results suggested that inheritance of glyphosate resistance was more influenced by the female than the male parent, although the multiple copies of EPSPS are amplified in the nuclear genome. However, it is important to mention that the authors were not able to confirm apomictic progeny using DNA markers. In contrast, two independent studies conducted in glyphosate-resistant A. palmeri populations (multiple copies of EPSPS) by Mohseni-Moghadam et al. (2013) and Chandi et al. (2012) reported no differences in progeny from reciprocal crosses and concluded that EPSPS gene copy number is under nuclear control with no maternal effects. Later, the EPSPS-eccDNA was characterized, indicating polymorphisms and the presence of an autonomous replication sequence, and demonstrating the random segregation in somatic and gametic cells, regardless of the sex of progenitors (Koo et al., 2018).

6.1 Selection pressure in agricultural systems and its effect on A. palmeri distribution

In agricultural fields, the frequency of hybridization events is determined by biological traits, such as species compatibility, seed viability and vigour, performance of progeny, and weed management strategy (i.e., prevention of new introduction and weed control prior to anthesis). How far a pollen grain can travel is dictated by environmental conditions such as relative humidity, wind speed and direction, temperature, physical characteristics of the A. palmeri donor and receptor (e.g., pollen size and density, plant size), and field obstacles along the way (e.g., a high versus low stature crop). Amaranthus palmeri is compatible with other species in the genus, such as A. spinosus, A. tuberculatus, and A. hybridus, although they display varied rates of hybridization (Federico et al., 2007; Gaines et al., 2012). Many studies have focused on the transfer of herbicide resistance alleles from A. palmeri to other closely related species. It has been demonstrated, for example, that copy number variation in EPSPS can confer glyphosate resistance in A. spinosus as a result of pollen transfer from glyphosate-resistant A. palmeri (Nandula et al., 2014).

Interspecific hybridization of A. palmeri with other closely related species is well documented. Progeny fertility largely depends on genomic compatibility. Most species within the Amaranthus genus exhibit 32 or 34 somatic chromosomes, and abnormal chromosome pairing can occur regardless of chromosome number. Amaranthus palmeri (2n = 34) and A. tuberculatus (2n = 32), for example, can intercross at low frequency, and produce a progeny that is broadly sterile, but not completely (Federico et al., 2007). The authors observed that surviving individuals were triploid and could be backcrossed with the A. tuberculatus parent. The mechanism leading to hybrid mortality was suggested to be hemizygosity or to hybrid dysgenesis. Another example is hybridization with A. hybridus, but also at low frequency (less than 0.01%) (Gaines et al., 2012). Conversely, crosses with A. spinosus (Gaines et al., 2012) and A. rudis (Wetzel et al., 1999) can result in fertile hybrids. Molin et al. (2016) found that an ALS-inhibitor resistant A. spinosus × A. palmeri population had a mutated ALS from the A. palmeri progenitor. Although hybrids have been extensively analysed in the context of herbicide resistance (Molin et al., 2016; Oliveira et al., 2018), much less is known about other fitness traits of the progeny.

While they are all members of the same subgenus Amaranthus, species have distinctive traits that associate with their centre of origin (Table 2). Amaranthus retroflexus and A. hybridus, for example, are weeds of riverbanks and adapted to moister regions (Sauer, 1967). Flowering time is a dominant trait in A. retroflexus and could be introgressed into the genetic background of a closely related species (Kulakow & Jain, 1985). As noted by Baker (1974), flowering time and duration are important weediness traits, because it can allow plants to escape human intervention and set seed prior to control events. Germination timing and requirements vary among Amaranthus species (Lawrence et al., 2004). Given interspecific hybridization with local Amaranthus species is possible, A. palmeri populations introduced to Europe are subjected to introgression of new traits that could hinder their management. For example, hybridization with species with a wider range of germination requirements could allow A. palmeri to exhibit a non-uniform germination pattern, making its management more difficult in agricultural systems.

Amaranthus palmeri, as previously discussed, is a dioecious species with open pollination by wind, with potential for long-distance pollen dispersal and interspecific hybridization. This mating system, in theory, favours populations to maintain a large genetic diversity because of (i) immense diversity of variants in the gene pool that natural selection can act upon, (ii) potential for evolutionary rescue by hybridization from new introductions (Carlson et al., 2014), and (iii) heterosis. High levels of genome-wide genetic diversity have been observed in a small amount of germplasm with populations from six US states (Küpper, Manmathan, et al., 2018). Although A. palmeri is an obligate outcrosser, populations can be maintained even when inbreeding occurs, as observed by Küpper, Manmathan, et al. (2018). Species with similar breeding system such as A. tuberculatus also exhibit this feature (Kreiner et al., 2019). Heterosis for enhanced biomass production has been reported in the intercross between A. cruentus and A. hypocondiacus (Lehmann et al., 1991). Much remains unknown about the early evolutionary mechanisms of adaptation following A. palmeri colonization of new environments. However, as discussed by Küpper, Manmathan, et al. (2018), A. palmeri populations from different geographic regions share high genetic similarity, which could be explained, in parts, by strong selection pressure caused by herbicides and, importantly, by intense gene flow between recent introductions and local populations.

6.2 How can we use characteristics in its breeding system to control A. palmeri

The breeding system of A. palmeri provide unique management opportunities. Recently, WeedOUT (https://www.weedout-ibs.com) has developed a sterile pollen-based weed control strategy for A. palmeri. WeedOUT's irradiated pollens compete successfully with naturally occurring A. palmeri pollen and fertilizes the weed ovule, which leads to the formation of deformed light weight seeds. Consequently, WeedOUT's novel technology reduces seed germination of treated plants to zero. This technology is based on the previously developed sterile insect technique (SIT), where irradiation is used to sterilize mass-reared insects so that, while remaining sexually competitive, they cannot produce any offspring (Knipling, 1959). It remains to be seen the efficacy of this technique in a practical settings and the evolutionary of resistance/escape by A. palmeri populations.

Genetic manipulation of weedy traits is a controversial topic with much interest as it promises to alter adaptive traits at the population level (Barrett et al., 2019). Gene drive, for example, is a genetic manipulation where normal Mendelian inheritance is altered, biasing the segregation to an allele of interest by spreading the fitness-reducing genetic modifications through artificial populations in the wild (Neve, 2018). Although the use of gene drive for weed management is at its infancy, other disciplines have been using this technology commercially. For example, a proprietary self-limiting gene is being used in the Florida Keys, Florida, USA, individuals die before reaching adulthood when containing the gene drive (Najjar et al., 2017). It should be noted that gene drives are currently being developed for testing in malaria-carrying mosquitoes and for invasive rats in New Zealand (Champer et al., 2021; Gantz et al., 2015). Dioecious Amaranthus species are particularly good candidates for gene drives as they rely on sexual reproduction and cross pollination. The National Academy of Sciences of the US suggested A. palmeri could be a lead candidate for genetic manipulation using gene drive to fix glyphosate susceptibility alleles in the population (National Academies of Sciences, 2016). Alternatively, a gene drive could be developed to reduce the frequency of female plants in the population, eventually leading to local extinction. Gene drives in natural populations may have unexpected ecological consequences associated with altering populations, because native population may be ecologically important, although mechanisms to limit spillovers have been developed (Barrett et al., 2019; Greenbaum et al., 2021). Since A. palmeri is an endemic species in the United States, if gene drives that shift an entire population to extinction are more effective than expected, they may eradicate A. palmeri from its native range, disrupting the natural ecosystem and possibly causing irreparable damage. Amaranthus palmeri may be the prime example for this technology in plants; however, like all technologies, gene drive may also be thwarted by A. palmeri's rapid adaptation, such as the evolution of apomictic populations (Section 4.2).

Another technology being evaluated for weed control is spray-induced gene silencing also known as ‘foliar applied RNA interference (RNAi)’. This non-GMO technology relies on the uptake of small silencing RNAs that interfere with the production of target messenger RNAs (mRNA). RNAi is a natural process that all plants perform for gene expression regulation; however, it can be manipulated in certain ways to lead plants to death. For instance, RNAi molecules could target key mRNAs in the plant metabolism to disrupt metabolic pathways, leading to weed control or turn off herbicide resistance genes, reverting once resistant plants into susceptibility (Mezzetti et al., 2020; Sammons et al., 2015). Another possibility for dioecious species like A. palmeri is to target genes that determine sex, and thereby manipulating plants to all be male, leading to reduced seed production. The major advantage of this technology is the lack of synthetic chemicals and off-target effects as the RNAi technology is specific to the exact allele of the target mRNA. Furthermore, if this selects for polymorphisms in the target mRNA, RNAi molecules can be quickly modified to match the mutation. The major challenge currently being addressed for effective RNAi is how to practically deliver such molecules into plant cells, given plants have multiple protective layers outside of the plasma membrane and small RNAs are typically very charged (Bennett et al., 2020; Dalakouras et al., 2016; Jiang et al., 2014). Some of these limitations, such as rapid degradation by nucleases, could be overcome by use of FANA ASO (2′-Fluoro-arabinonucleic Acid antisense oligonucleotides), as reviewed elsewhere (El-Khoury & Damha, 2021).