Skip to main content
NCBI home page
Annals of Botany logoLink to Annals of Botany
. 2014 Mar 16;113(6):939–952. doi: 10.1093/aob/mcu016

Genetic and morphological contrasts between wild and anthropogenic populations of Agave parryi var. huachucensis in south-eastern Arizona

Kathleen C Parker 1,*, Dorset W Trapnell 2,3, J L Hamrick 3, Wendy C Hodgson 4
PMCID: PMC3997635  PMID: 24638822

Abstract

Background and Aims

At least seven species of Agave, including A. parryi, were cultivated prehistorically in Arizona, serving as important sources of food and fibre. Many relict populations from ancient cultivation remain in the modern landscape, offering a unique opportunity to study pre-Columbian plant manipulation practices. This study examined genetic and morphological variation in six A. p. var. huachucensis populations of unknown origin to compare them with previous work on A. parryi populations of known origin, to infer their cultivation history and to determine whether artificial selection is evident in populations potentially managed by early agriculturalists.

Methods

Six A. p. var. huachucensis and 17 A. parryi populations were sampled, and morphometric, allozyme and microsatellite data were used to compare morphology and genetic structure in purportedly anthropogenic and wild populations, as well as in the two taxa. Analysis of molecular variance and Bayesian clustering were performed to partition variation associated with taxonomic identity and hypothesized evolutionary history, to highlight patterns of similarity among populations and to identify potential wild sources for the planting stock.

Key Results A

p. var. huachucensis and A. parryi populations differed significantly both morphologically and genetically. Like A. parryi, wild A. p. var. huachucensis populations were more genetically diverse than the inferred anthropogenic populations, with greater expected heterozygosity, percentage of polymorphic loci and number of alleles. Inferred anthropogenic populations exhibited many traits indicative of past active cultivation: greater morphological uniformity, fixed heterozygosity for several loci (non-existent in wild populations), fewer multilocus genotypes and strong differentiation among populations.

Conclusions

Where archaeological information is lacking, the genetic signature of many Agave populations in Arizona can be used to infer their evolutionary history and to identify potentially fruitful sites for archaeological investigation of ancient settlements and cultivation practices. The same approach can clearly be adopted for other species in similar situations.

Keywords: Crop domestication, Agave parryi var. huachucensis, Agave parryi, genetic diversity, genetic structure, prehistoric agriculture, crop evolution, morphological traits

INTRODUCTION

Mesoamerica supports a rich native flora that comprises many arid and semi-arid species useful to humans for food, medicine, fibre and structural materials (Blancas et al., 2010). This area has been an important centre of plant domestication, ranging from prehistoric times when crops such as maize and agave (Smith, 1967) were first domesticated, to the present when many native plants are in various stages of domestication within traditional agricultural systems (Casas et al., 2007; Blancas et al., 2010). Although less diverse than other parts of Mesoamerica, the native Sonoran Desert flora also includes many ethnobotanically important plants cultivated by pre-Columbian residents (Ford, 1981; Nabhan, 1985), including several species of Agave L. (Fish and Nabhan, 1991). Intense focus on modern plant management in central and southern Mexico by Casas and his colleagues (Casas et al., 2007; Blancas et al., 2010) and others (Zizumbo-Villarreal et al., 2013) has greatly increased our understanding of cultivation effects on phenotypic and genetic diversity in areas with ongoing human habitation and artificial selection. We have less knowledge about domestication in the northern Sonoran Desert, particularly in areas where prehistoric cultures flourished prior to the 13th and 14th centuries.

Although plant domestication has classically been characterized as an ex situ process that reduces variability through selection of preferred traits in plants grown away from their natural habitat (Sauer, 1972; Doebley, 1989), human management of plants in Mesoamerica and the Sonoran Desert has been a more diverse process (Fish and Nabhan, 1991; Casas et al., 2007). Blancas et al. (2010) reported over 1500 plant species used by current residents of the Tehuacán–Cuicatlán Valley in Mexico. Of these, farmers managed over 400 species ex situ in agricultural fields and home gardens, and almost as many less intensively in situ by leaving certain naturally occurring individuals but modifying their environment to enhance production – with many species concurrently managed under both strategies. Casas et al. (2007) found that less intensively managed populations typically have reduced phenotypic and genetic diversity relative to wild populations, as expected with classical plant domestication models; however, more intense manipulation where farmers introduce new stock to their gardens and maintain a variety of landraces for different purposes often increases diversity (Casas et al., 2007; Parra et al., 2008).

In modern traditional agricultural systems, management intensity can be determined from interviews querying the energy investment of farmers, the complexity of their cultivation and unit crop production (Blancas et al., 2010). For prehistorically cultivated plant species without continuing management or direct cultural descendants to interview, we must rely on surrogate measures to infer intensity of management. The objective of this study was to use genetic information to better understand the prehistoric cultivation of Agave parryi var. huachucensis in south-eastern Arizona, where several Agave species were managed prehistorically. Although Apaches in the last 500–600 years harvested mature wild Agave plants, they were not agriculturalists (Gentry, 1982); therefore, relict populations in this area escaped ongoing anthropogenic selection after widespread abandonment in the late 13th century.

Agave has been important ethnobotanically to inhabitants of arid and semi-arid North America since ancient times (Smith, 1967; Colunga-GarcíaMarín and May-Pat, 1993; Good-Avila et al., 2006). In central Mexico Agave was an important part of the human diet as early as 8000 BP (Smith, 1967), before maize and beans were domesticated. In Arizona alone, where Agave was often planted in marginal fields to supplement less drought-tolerant annual crops, over 550 pre-Columbian Agave cultivation sites have been documented based on plant remains and artefacts indicative of cultivation and processing (Fish et al., 1985; Fish, 2000). At least seven Agave species were cultivated prehistorically in Arizona (Minnis and Plog, 1976; Hodgson and Slauson, 1995; Hodgson, 2001a; Hodgson and Salywon, 2013); collectively they were probably used for food, alcoholic beverages and fibre (Fish et al., 1985; Nabhan, 1992).

Parker et al. (2007, 2010) examined the effects of pre-Columbian cultivation in relict populations of three Agave species in central Arizona with known archaeological associations, including Agave parryi subsp. parryi. By comparing disjunct A. parryi populations associated with pre-Columbian ruins and core populations in the northern part of the species range, Parker et al. (2010) identified a genetic signature for cultivated populations. This study extends that work to closely related A. p. var. huachucensis populations in south-eastern Arizona, an area with a rich pre-Columbian history but less thoroughly analysed archaeologically. Like A. parryi further north, A. p. var. huachucensis has a core area of occurrence, as well as isolated populations outside the taxon's typical habitat. We examined both core and disjunct A. p. var. huachucensis populations whose influence by anthropogenic manipulation was uncertain to compare the two taxa and determine whether the genetic signature of cultivated agaves identified by Parker et al. (2010) could be used to infer the history of populations of unknown origin. Specific research questions were: (1) How do A. p. var. huachucensis populations compare with wild and cultivated A. parryi populations in terms of morphological and genetic diversity, genetic structure, and extent of asexual reproduction? Others have reported morphological contrasts between the two taxa (Gentry, 1982; Nobel and Smith, 1983), but little molecular work has focused on their phylogenetic relationship. We expected to confirm previous findings of morphological contrasts with our more extensive sampling; we hypothesized that these would be accompanied by genetic differences between the two taxa. (2) What wild populations potentially served as planting stock sources for the A. p. var. huachucensis populations with suspected human-management history? Based on our prior work with A. parryi, we expected that nearby wild populations would not be those most similar to anthropogenic populations; rather, planting stock was most likely obtained from more distant sources with some advantage other than geographical proximity. (3) Can the genetic signature of cultivation be used to infer the history of A. p. var. huachucensis populations, i.e. to indicate whether they were cultivated in ancient times? We anticipated that anthropogenic A. p. var. huachucensis populations would exhibit reduced genetic and multilocus genotypic diversity relative to wild populations, as well as greater morphological uniformity and population differentiation. Because these traits were linked to cultivation by Parker et al. (2010) and others examining species grown in traditional agricultural systems (Miller and Schaal, 2006), we felt that if detected in our populations, they would be a strong indication of active pre-Columbian cultivation.

MATERIALS AND METHODS

Species biology and study area

Agave parryi Engelm. subsp. parryi (hereafter, A. parryi) and A. p. var. huachucensis (Baker) Little ex L.D. Benson (Agavoideae; Asparagaceae sensu APG III, 2009) are monocarpic succulent rosettes that occur primarily in Arizona and northern Mexico (Gentry, 1982). Gentry (1982) described different varieties within A. parryi, including A. p. var. huachucensis, but later Ullrich (1992) reduced A. p. var. huachucensis and weakly differentiated variants to synonymy under A. parryi subsp. parryi, as noted by Reveal and Hodgson (2002) – a distinction followed by US federal agencies [http://www.itis.gov/; http://plants.usda.gov/java/; both sites accessed March 2013]. However, IPNI (http://www.ipni.org/index.html) and Tropicos (http://www.tropicos.org/Home.aspx) list A. p. var. huachucensis separately (both sites also accessed March 2013). Classification of taxa within A. parryi has been based primarily on morphological traits rather than genetic relationships. While molecular approaches have led to the reclassification of higher order groups (i.e. the former Agavaceae; Bogler and Simpson, 1996; Good-Avila et al., 2006; APG III, 2009), varietal boundaries have not previously been similarly examined.

Although Agave includes many self-incompatible species (Slauson, 2000), the mating system of A. parryi has not been investigated. In addition to producing many small wind- and animal-dispersed seeds, suckering is common in both A. parryi and A. p. var. huachucensis (Reveal and Hodgson, 2002). Because of their longevity and frequent vegetative reproduction, many Agave populations that were cultivated prehistorically but unmanaged since European contact still exist in the modern landscape. Pinkava and Baker (1985) reported that A. parryi (as the synonomous A. parryi var. couesii) is a tetraploid (2n = 120). The presence of both balanced and unbalanced heterozygotes and the absence of fixed heterozygosity in wild populations are consistent with autopolyploidy with tetrasomic inheritance (Weeden and Wendel, 1989; Soltis and Soltis, 2000).

Agave p. var. huachucensis populations were sampled in south-eastern Arizona. Three core populations were from canyons at 1920–2175 m elevation in the Huachuca Mountains, dominated by oaks (Quercus gambelii, Q. arizonica, Q. hypoleucoides), pines (P. leiophylla, P. strobilis), Robinia neomexicana, Fraxinus velutina and Pseudotsuga menziesii. Three disjunct populations were from much lower elevations (1380–1584 m) in more open grass-dominated (Bouteloua gracilis, Bothriochloa languroides) terrain, with scattered woody plants (Q. emoryi, Baccharis pteronioides, Acacia greggii; Figs 1 and 2). Although all A. p. var. huachucensis populations were within 20 km of known pre-Columbian ruins, we had no specific archaeological information for any of the six populations (in contrast to Parker et al., 2010). We suspected an anthropogenic origin for the three open populations (hereafter referred to as anthropogenic) on the basis of their more isolated location relative to the mountainous core populations (hereafter referred to as wild) and their occurrence at low elevations, outside the typical elevation range Gentry (1982) described for this variety. A. parryi samples were collected from populations in the northern part of the species range in central Arizona. Wild populations (n = 9) were associated with woodlands of pinyon pine (Pinus edulis), juniper (Juniperus spp.) and ponderosa pine (Pinus ponderosa) at 1396–1954 m elevation. Cultivated populations (n = 8) were north-east of the Mogollon Rim at higher elevations (2005–2053 m) in pinyon–juniper–ponderosa pine woodlands adjacent to more open terrain (see Parker et al., 2010 for further details).

Fig. 1.

Fig. 1.

Open in a new tab

Location of Agave parryi and hypothesized wild and anthropogenic A. p. var. huachucensis study populations in Arizona.

Fig. 2.

Fig. 2.

Open in a new tab

Populations of Agave parryi var. huachucensis with different hypothesized evolutionary histories: wild populations in the Huachuca Mountains, LER (A) and SCO (B); anthropogenic populations at lower elevations in south-eastern Arizona, AZT (C) and BAB (D); photos by K.C.P.

Field sampling

Potential sample populations were identified from herbarium records at the Desert Botanical Garden (DES) in Phoenix, Arizona; Gentry (1982); personal communication with Mark Dimmitt (Arizona-Sonora Desert Museum, Tucson, AZ) and Chris Haas (Audubon Appleton-Whittell Research Ranch, Elgin, AZ); and field reconnaissance. A. p. var. huachucensis is uncommon, and several populations documented by DES voucher specimens no longer exist due to development.

We collected tips of younger, inner leaves from 48 individuals per population; these were kept at room temperature for ≤ 5 days until returned to the University of Georgia, where they were stored at 4 °C until protein extraction. We avoided multiple samples from tight spatial clusters. Universal Transverse Mercator (UTM) coordinates of each population were determined with a Garmin global positioning system (GPS) unit. A voucher specimen was prepared for each population and deposited in DES.

Twenty individuals from different clones were arbitrarily selected in each population for leaf-morphology measurements. Because these data were collected over several years subsequent to genetic sampling, the same individuals were not necessarily included in the morphological and genetic samples. For each individual, nine traits were measured: presence/absence of interstitial teeth; intertooth distance at the lower, middle and upper leaf margin; leaf length; leaf width; terminal spine length; number of basal offsets; and maximum leaf-width location.

Statistical analysis of morphological traits

Analysis of variance (ANOVA; PROC GLM 9·3, SAS Institute, 2011) was used to determine whether morphological traits differed among populations. We examined effects due to variety and suspected origin (i.e. hypothesized as wild vs. anthropogenic), as well as an interaction between these two main effects. Because eight traits were tested simultaneously (maximum leaf-width location was a categorical variable and therefore not tested), Bonferroni corrections were used in assessment of ANOVA significance. A neighbour-joining tree (Saitou and Nei, 1987) representing morphological similarities among populations was generated with NTSYSpc 2·2 (Rohlf, 2005), with relationships based on Euclidian distances.

Allozyme laboratory and statistical analysis

After protein extraction with a potassium phosphate buffer (Mitton et al., 1979), starch gel electrophoresis was used to assess allozyme variation following Parker et al. (2010). Four electrode buffer systems and 13 enzyme systems resolved 30 loci.

Standard genetic diversity indices were calculated for each population: percentage of polymorphic loci (PP), mean number of alleles per locus (A) and per polymorphic locus (PA), total number of alleles, and expected heterozygosity (He; Hamrick et al., 1979; Hamrick and Godt, 1989). He and Wright's (1965) fixation index (F) were estimated with Autotet, a program developed for tetraploids (Thrall and Young, 2000). Autotet accounts for the existence of multiple heterozygous states in autotetraploids and possible segregation of individual loci during meiosis by either chromosomes or chromatids. Nei's (1972) genetic identities were calculated for all population pairs of both taxa with GenAlEx 6·5 (Peakall and Smouse, 2006, 2012).

Multilocus allozyme genotype frequencies were determined for each population. Genotypic diversity (D) was calculated as D = 1 – Σ{[ni(ni – 1)]/[N(N – 1)]}, where ni is the number of individuals of the ith genotype and N is the sample size (Ellstrand and Roose, 1987). The formula used to calculate the number of unique multilocus genotypes possible (Ng) for tetraploids was modified from Cheliak and Pitel's (1984) diploid formula, as

graphic file with name M1.gif

where ai is the number of alleles detected at the ith locus, and L is the number of loci analysed.

Genetic structure was analysed hierarchically with both analysis of molecular variance (AMOVA) and Bayesian clustering to assess differentiation between varieties and between suspected cultivation histories. These analyses were also used to identify potential sources for the putative anthropogenic A. parryi var. huachucensis populations. Specifically, we used GenAlEx 6·5 to perform the AMOVA (Peakall and Smouse, 2006, 2012) in two different ways. The first involved all 23 populations, with allozyme variation partitioned between the two varieties, as well as within and among populations within each variety. We then used only the six A. parryi var. huachucensis populations to partition variation into among- and within-population and between-origin components. Each analysis was based on 999 total data permutations. We also used a Bayesian clustering approach (STRUCTURE 2·3·4; Pritchard et al., 2000; Falush et al., 2003) to estimate the number of genetically distinct populations (K), as well as levels of admixture within populations of different origins. To detect the optimum value of K, we made ten repeat runs at each K from K = 3 to 23, with a burnin length of 10 000, a run length of 100 000, and a model based on correlated allele frequencies with admixture permitted. We used Evanno et al.'s (2005) criterion of maximizing ΔK to select the most likely K value, but also considered the K value that maximized the estimated ln probability of the data, ln Pr(X|K), as well as biological interpretability (Pritchard et al., 2000).

In addition, pairwise FST values between A. p. var. huachucensis populations and all other populations were calculated. We used the allozyme data and SPAGeDi 1·2 (Hardy and Vekemans, 2002), which assumes tetrasomic inheritance for autotetraploids.

Microsatellite laboratory and statistical analysis

Four markers (D. W. Trapnell, University of Georgia, Athens) were developed and employed to assess microsatellite diversity in 18 of the 23 study populations (we were unable to sample populations CHR, DWT, MIL, LER and SCO). Twenty-four individuals used in the allozyme analysis for each population were analysed for microsatellite variation. PCR amplifications with primer P1-2F were performed in a 10-μL volume [1 µL PCR 10× buffer (supplied with Taq), 1 µL bovine serum albumin (BSA, 10×), 0·8 µL MgCl2 (25 mm), 0·6 µL dNTPs (2·5 mm each), 0·4 µL untagged primer (10 µm), 0·04 µL tagged primer (10 µm), 0·36 µL fluorescent dye-labelled primer (10 µm), 0·08 µL Sigma JumpStart Taq, 1 µL template DNA (3–10 ng) and 4·72 µL water] using an ABI 9700 thermal cycler. The tagged primer had a CAG tag (5′-CAGTCGGGCGTCATCA-3′) added to the 5′ end to allow use of a third fluorescently labelled primer that was identical to the CAG tag in the PCR reaction (Schable et al., 2002). PCR amplifications with primers P1-5G, P1-7A and P2-8D were performed in a 10-μL volume [1 µL PCR 10× buffer (supplied with Taq), 1 µL BSA (10×), 0·8 µL MgCl2 (25 mM), 0·6 µL dNTPs (2·5 mm each), 1 µL unlabelled primer (10 µm), 1 µL fluorescent dye-labelled primer (10 µm), 0·1 µL Sigma JumpStart Taq, 1 µL template DNA (3–10 ng) and 3·5 µL water] using an ABI 9700 thermal cycler. Touchdown thermal cycling programmes (Don et al., 1991) spanning 10 °C of annealing temperatures (60–50 °C for primers P1-2F, P1-5G and P1-7A and 65–55 °C for primer P2-8D) were used for amplification. Cycling parameters were one cycle of 95 °C for 2 min; six cycles of 95 °C for 30 s, highest annealing temperature for 30 s and 72 °C for 45 s; 21 cycles of 95 °C for 30 s, highest annealing temperature (decreased 0·5 °C per cycle) for 30 s and 72 °C for 45 s; 21 cycles of 95 °C for 30 s, lowest annealing temperature for 30 s and 72 °C for 45 s; one cycle of 72 °C for 10 min; and one cycle at 15 °C until samples were removed. PCR products were run on an ABI 3730 × l sequencer and sized with fluorescent ladder (CXR) of 60–400 bases (Promega, Madison, WI, USA). Results were analysed using GENEMAPPER version 3·7 (Applied Biosystems, Foster City, CA, USA).

Genetic diversity indices (He and the Shannon–Weiner Diversity Index, H′) were calculated from the microsatellite data with TETRASAT (Markwith et al., 2006), specifically designed for tetraploids, as described by Parker et al. (2010). Microsatellite genotypes were not analysed because the uncertainty of allele copy number in heterozygous individuals prevented identification of multilocus genotypes (Trapnell et al., 2011).

Genetic differentiation between varieties and among populations was examined in several ways with the microsatellite data. First, we scored the presence/absence (P/A) of microsatellite alleles (Becher et al., 2000), treating our co-dominant microsatellites as dominant markers, and performed an AMOVA in GenAlEx 6·5 (Peakall and Smouse, 2006, 2012) to test genetic differentiation. We were unable to partition genetic differentiation between A. parryi var. huachucensis populations of different putative origins because we lacked microsatellite data for the wild populations. Second, we used STRUCTURE 2·3·4 (Pritchard et al., 2000) to cluster populations and estimate levels of admixture based on microsatellite data, following the procedures detailed above for the allozyme-based analysis. Because those results were very similar, we elected to show only the more complete allozyme-based analysis.

RESULTS

Morphological characteristics of both taxa

Taxonomic variety and hypothesized population origin both significantly influenced leaf morphology and extent of basal offset formation (Table 1). In A. parryi intertooth distances were greater at all points along the leaf margin than in A. p. var. huachucensis, but leaves were shorter and narrower and had shorter terminal spines. Of the parameters examined, only leaf width did not differ significantly between populations grouped according to hypothesized origin. Anthropogenic populations tended to have more basal offsets and smaller leaves with more closely spaced teeth and a more pronounced obovate shape (i.e. maximum width closer to the tip) than wild populations. Traits were more uniform in anthropogenic than wild populations, as indicated by lower coefficients of variation (the ratio of the standard deviation to the mean) for all morphological variables examined. These contrasts between anthropogenic and wild populations were evident both within each variety and within the pooled sample, although interaction effects between variety and origin were only significant for leaf length and intertooth distance at the base and middle of the leaf (Table 1).

Table 1.

Mean values for morphological characteristics in Agave parryi var. huachucensis and A. parryi populations in Arizona

Inner-tooth distanceVOI Mid-tooth distanceVOI Outer-tooth distanceVO No. of basal offsetsVO Leaf lengthVOI Terminal spine lengthVO Leaf widthV Maximum width locationVO
A. parryi var. huachucensis populations
Hypothesized wild
MIL 0·7 (0·3) 1·7 (0·4) 2·3 (0·3) 5·7 (3·3) 31·6 (4·5) 3·0 (0·5) 10·2 (1·8) 1·8 (0·5)
LER 0·5 (0·2) 1·5 (0·3) 1·7 (0·4) 5·3 (3·4) 28·5 (4·2) 3·2 (0·4) 10·5 (1·6) 1·9 (0·3)
SCO 0·4 (0·2) 1·6 (0·3) 2·0 (0·4) 5·2 (3·7) 28·0 (4·3) 2·8 (0·4) 10·9 (1·8) 1·9 (0·4)
Mean 0·5 (0·2) 1·6 (0·4) 2·0 (0·4) 5·4 (3·4) 29·4 (4·5) 3·0 (0·5) 10·5 (1·8) 1·9 (0·4)
Hypothesized anthropogenic
ARR 0·7 (0·1) 1·4 (0·4) 1·8 (0·4) 5·8 (3·3) 29·2 (3·3) 2·4 (0·2) 13·3 (2·2) 2·5 (0·5)
AZT 0·8 (0·2) 1·4 (0·2) 1·5 (0·5) 10·2 (4·9) 26·7 (1·5) 3·1 (0·2) 11·4 (0·8) 2·4 (0·5)
BAB 0·8 (0·2) 1·6 (0·3) 1·7 (0·2) 9·2 (4·5) 30·2 (4·5) 3·0 (0·3) 11·2 (1·4) 2·5 (0·5)
Mean 0·8 (0·2) 1·5 (0·3) 1·7 (0·4) 8·4 (4·6) 28·7 (3·6) 2·8 (0·4) 12·0 (1·8) 2·5 (0·5)
Wild A. parryi populations
CCC 1·4 (0·6) 1·7 (0·4) 1·8 (0·4) 1·9 (2·8) 28·3 (5·3) 2·2 (0·4) 9·3 (1·6) 1·6 (0·8)
CHR 0·8 (0·2) 1·7 (0·5) 1·9 (0·4) 3·5 (2·4) 28·1 (4·6) 2·7 (0·4) 8·1 (1·4) 1·9 (0·4)
JER 1·1 (0·3) 1·4 (0·4) 1·9 (0·3) 8·2 (5·1) 23·2 (3·3) 2·2 (0·3) 6·8 (0·9) 1·1 (0·4)
OME 1·3 (0·5) 1·7 (0·6) 1·8 (0·6) 0·2 (0·7) 25·0 (3·4) 2·6 (0·4) 7·6 (0·9) 2·2 (0·7)
PIN 1·4 (0·5) 2·0 (0·8) 2·2 (0·6) 3·1 (3·3) 29·1 (5·9) 2·4 (0·6) 8·6 (1·5) 2·3 (0·9)
PRE 1·3 (0·4) 1·7 (0·4) 1·9 (0·4) 1·7 (1·9) 27·1 (4·3) 2·4 (0·5) 7·1 (1·2) 1·3 (0·7)
SAN 0·7 (0·3) 2·0 (0·4) 2·5 (0·3) 6·6 (3·8) 31·8 (4·0) 2·6 (0·5) 8·0 (1·1) 1·4 (0·6)
SEN 0·7 (0·3) 1·8 (0·5) 1·8 (0·5) 2·2 (2·8) 26·1 (4·5) 2·6 (0·3) 8·8 (1·3) 2·2 (0·7)
SHR 1·7 (0·8) 2·7 (0·7) 3·0 (0·8) 6·3 (4·5) 33·5 (5·2) 3·3 (0·5) 9·3 (1·6) 1·3 (0·5)
Mean 1·1 (0·6) 1·9 (0·6) 2·1 (0·6) 3·7 (4·1) 28·0 (5·4) 2·5 (0·5) 8·2 (1·5) 1·7 (0·8)
Cultivated A. parryi populations
JAK 0·8 (0·2) 1·3 (0·3) 1·9 (0·2) 19·4 (6·6) 19·4 (2·5) 2·4 (0·3) 8·1 (1·0) 2·4 (0·5)
JFI 0·7 (0·2) 1·2 (0·2) 1·6 (0·2) 1·1 (1·6) 16·9 (2·0) 1·9 (0·2) 5·4 (0·8) 2·0 (0·0)
POW 0·7 (0·2) 1·2 (0·3) 1·9 (0·3) 7·6 (3·9) 21·4 (4·1) 2·4 (0·2) 8·6 (1·5) 2·5 (0·5)
PUR 0·8 (0·2) 1·2 (0·3) 1·9 (0·3) 2·5 (3·5) 21·4 (3·1) 2·3 (0·2) 8·4 (1·0) 2·5 (0·5)
TIL 0·6 (0·3) 1·3 (0·3) 2·0 (0·3) 10·1 (6·9) 16·5 (2·9) 3·0 (0·3) 5·5 (0·8) 2·0 (0·5)
TRT 0·4 (0·1) 1·4 (0·3) 2·0 (0·4) 5·7 (3·2) 21·8 (4·1) 2·9 (0·3) 7·0 (1·3) 2·0 (0·0)
WCC 1·0 (0·2) 1·9 (0·3) 1·7 (0·2) 3·9 (2·4) 23·4 (1·2) 2·4 (0·2) 8·0 (0·4) 1·0 (0·0)
Mean 0·7 (0·3) 1·4 (0·4) 1·9 (0·3) 7·2 (7·2) 20·1 (3·8) 2·5 (0·4) 7·3 (1·6) 2·0 (0·6)
ANOVA F-statistic 45·01* 28·64* 12·60* 17·30* 106·25* 22·74* 149·05* 24·10*
Open in a new tab

Values provided are means with standard deviations in parentheses. Dimension measurements are indicated in cm. Superscripts following variable names indicate significant ANOVA effects as follows: V, varietal effects between A. parryi and A. p. var. huachucensis; o, origin effects between anthropogenic and wild populations; I, interaction effects between varieties and origins; asterisks following F statistics for ANOVA indicate significance at P < 0·001, with application of the Bonferroni correction.

Clustering of populations depicted by the neighbour-joining tree based on morphological characteristics (Fig. 3) separated populations according to both variety and hypothesized origin. The A. p. var. huachucensis populations were grouped somewhat diffusely (along with two wild A. parryi populations, SAN and SHR), with separation of wild and anthropogenic populations. Most wild A. parryi populations clustered together relatively closely. Cultivated A. parryi populations formed several groups, with some populations joining to others on relatively long branches.

Fig. 3.

Fig. 3.

Open in a new tab

Neighbour-joining phenogram based on morphological traits and Euclidian distances among hypothesized wild and anthropogenic Agave parryi and A. p. var. huachucensis populations in Arizona (see key).

Genetic diversity of A. p. var. huachucensis

The six A. p. var. huachucensis populations studied constituted a dichotomous group in terms of their allozyme-based genetic diversity and the extent of asexual reproduction. The three purportedly wild populations were significantly more genetically diverse than the anthropogenic populations; they had a greater percentage of polymorphic loci, number of alleles overall and per locus, and observed and expected heterozygosities (Table 2). Fixation indices (F) were negative in all six populations, indicating an excess of heterozygotes. Although fixed heterozygosity was common in hypothesized anthropogenic populations, it was non-existent in wild populations. Five allozyme alleles were restricted to A. p. var. huachucensis populations (vs. 11 alleles restricted to A. parryi); two of these were only found in SCO.

Table 2.

Measures of allozyme-based genetic diversity and multilocus genotypic diversity of Agave parryi var. huachucensis relative to wild and cultivated A. parryi populations in Arizona

PP PA No. of alleles Ho He FIS G/N Ng D
A. parryi var. huachucensis populations
Hypothesized wild
MIL 79·3 2·22 57 0·175 (0·210) 0·160 (0·178) –0·095 (0·110) 1·000 2·897 × 1018 1·000
LER 60·7 2·29 50 0·156 (0·223) 0·124 (0·179) –0·252 (0·052) 1·000 1·854 × 1014 1·000
SCO 70·0 2·38 59 0·174 (0·219) 0·144 (0·181) –0·215 (0·064) 0·958 2·433 × 1018 0·998
Population-level mean 70·0 (9·3) 2·30 (0·08) 55·3 (4·7) 0·168 (0·011) 0·143 (0·018) –0·187 (0·082) 0·986 (0·024) 1·777 × 1018 (1·556 × 1018) 0·999 (0·001)
Hypothesized anthropogenic
ARR 33·3 2·20 42 0·094 (0·185) 0·090 (0·159) –0·053 (0·123) 0·917 8·789 × 107 0·996
AZT 13·3 2·25 35 0·091 (0·235) 0·064 (0·166) –0·415 (0·043) 0·042 1·875 × 103 0·311
BAB 33·3 2·20 42 0·095 (0·192) 0·075 (0·144) –0·255 (0·081) 0·583 6·836 × 107 0·944
Population-level mean 26·7 (11·5) 2·22 (0·03) 39·7 (4·0) 0·093 (0·002) 0·076 (0·013) –0·241 (0·181) 0·514 (0·442) 5·208 × 107 (4·615 × 107) 0·750 (0·381)
Wild A. parryi populations
Population-level mean 47·8 (6·2) 2·37 (0·08) 49·7 (3·4) 0·112 (0·030) 0·114 (0·029) 0·028 (0·039) 0·956 (0·075) 4·593 × 1014 (1·289 × 1015) 0·998 (0·004)
Cultivated A. parryi populations
Population-level mean 25·0 (6·9) 2·05 (0·07) 37·9 (2·2) 0·106 (0·028) 0·079 (0·021) –0·341 (0·062) 0·260 (0·213) 2·422 × 106 (3·549 × 106) 0·721 (0·217)
Open in a new tab

PP, percentage of polymorphic loci; PA, mean number of alleles at polymorphic loci; No. of alleles, total number of alleles per population; Ho, observed heterozygosity; He, expected heterozygosity under random mating assuming some degree of chromatid segregation; FIS, fixation index indicating the departure of observed from expected heterozygosity; n = 48 individuals in all populations; G/N, number of multilocus genotypes per individual; Ng, number of possible multilocus genotypes; D, multilocus genotype diversity. Numbers in parentheses are standard deviations. See Parker et al. (2010) for values for individual wild and cultivated A. parryi populations.

Asexual reproduction was less common in wild than anthropogenic A. p. var. huachucensis populations (Table 2). In two of the three wild populations, all 48 individuals had unique multilocus genotypes; and in the third, only two individuals shared genotypes with other plants. In contrast, AZT, BAB and ARR included only two, 28 and 44 unique genotypes, respectively, and had far fewer unique multilocus genotypes possible than the wild populations.

These contrasts between hypothesized wild and anthropogenic A. p. var. huachucensis populations mirrored those for wild and cultivated A. parryi populations (Table 2; see Parker et al., 2010, for individual population values in the latter variety), with the exception of fixation indices. Unlike wild A. parryi populations, which had slightly positive F values, wild A. parryi var. huachucensis populations had negative F values, indicative of a heterozygote excess.

Microsatellite-based genetic diversity indices cast a similar picture, albeit less complete, given the absence of data for purportedly wild A. p. var. huachucensis populations and two of the A. parryi populations, as well as the P/A treatment of alleles. Only one microsatellite allele was unique to A. p. var. huachucensis populations (and none to a single population), which may reflect in part the absence of microsatellite representation for the more allozyme allele-rich wild populations (LER, MIL, SCO). Anthropogenic A. p. var. huachucensis populations were slightly more genetically diverse than anthropogenic A. parryi populations (Table 3).

Table 3.

Measures of microsatellite-based genetic diversity of anthropogenic Agave parryi var. huachucensis relative to wild and cultivated A. parryi populations in Arizona

n PP PA No. of alleles He H′
A. parryi var. huachucensis populations
ARR 24 100 4·25 17 0·659 (0·013) 1·749 (0·032)
AZT 24 100 2·75 11 0·600 (0·010) 1·389 (0·021)
BAB 24 100 4·00 16 0·607 (0·044) 1·510 (0·107)
Population-level mean 100 (0) 3·67 (0·80) 14·7 (3·2) 0·622 (0·033) 1·549 (0·150)
Wild A. parryi populations
Population-level mean 100 (0) 7·09 (1·26) 28·4 (5·0) 0·621 (0·034) 1·816 (0·151)
Cultivated A. parryi populations
Population-level mean 78·6 (30·4) 2·57 (0·51) 10·3 (2·1) 0·433 (0·197) 1·001 (0·455)
Open in a new tab

n, number of individuals sampled; PP, percentage of polymorphic loci; PA, mean number of alleles at polymorphic loci; No. of alleles, total number of alleles per population; He, expected heterozygosity; H′, Shannon-Weiner Diversity Index. Means for wild and cultivated A. parryi populations are based on eight and seven populations, respectively. Numbers in parentheses are standard deviations. See Parker et al. (2010) for further details about individual population sizes and values.

Genetic structure of both taxa

The AMOVA based on allozyme data for the 23 populations indicated significant differentiation between varieties and among populations (Table 4). Varietal contrasts accounted for 8 % and among-population differentiation accounted for 22 % of the total allozyme variance. Microsatellite-based AMOVA showed similar results, despite the absence of three A. p. var. huachucensis populations from the sample (Table 5). Nei's (1972) mean genetic identity for population pairs involving different taxa was only slightly lower (0·951) than for population pairs within each taxon (0·953 for A. p. var. huachucensis and 0·959 for A. parryi).

Table 4.

AMOVA of Agave parryi and A. p. var. huachucensis populations in Arizona based on allozyme data

Source of variation d.f. Estimated variance % of variance F statistic†
All populations (n = 23)
 Between varieties 1 0·220 8 FRT = 0·083*
 Among populations 21 0·587 22 FSR = 0·242*
FST = 0·305*
 Within populations 4393 1·834 69 –
 Total 4415 2·641
A. p. var. huachucensis populations (n = 6)
 Between origins 1 0·928 28 FRT = 0·278*
 Among populations 4 0·451 14 FSR = 0·187*
FST = 0·413*
 Within populations 1146 1·960 59 –
 Total 1151 3·339
Open in a new tab

† FRT, among-group variance expressed relative to the total variance; FSR, among-population variance relative to (within-population + among-population variance); FST, (among-population + between-group variance) relative to the total variance; *significant variation for that hierarchical level at P < 0·001, based on 999 permutations.

Table 5.

AMOVA of P/A microsatellite data for Agave parryi and A. parryi var. huachucensis populations in Arizona

Source of variation d.f. Estimated variance % of variance F statistic†
All populations
 Between varieties 1 1·103 9 FRT = 0·089*
 Among populations 16 3·644 29 FPR = 0·323*
FPT = 0·384*
 Within populations 403 7·625 62 –
 Total 420 12·372
Open in a new tab

† FRT, between-variety variance expressed relative to the total; FPR, among-population variance relative to (within-population + among-population variance); FPT, (among-population + between-variety variance) relative to the total; *significant variation for that hierarchical level at P < 0·001, based on 999 permutations.

When the allozyme-based AMOVA was restricted to the six A. p. var. huachucensis populations, 28 % of the genetic variance was accounted for by the hypothesized wild vs. anthropogenic distinction (Table 4). Differentiation among populations within the two groups accounted for 14 % of the total genetic diversity.

The grouping of populations with STRUCTURE (Pritchard et al., 2000) based on allozyme data was similar in some respects to the morphological neighbour-joining tree. ΔK was maximized at K = 5, indicating five as the optimum number of genetic clusters, following the procedure of Evanno et al. (2005). At this level, a large cluster included the three purportedly wild A. p. var. huachucensis populations (LER, MIL, SCO) and all but two of the wild A. parryi populations (SAN, SHR – the two populations that were not grouped closely with other wild populations morphologically; Fig. 4). Other clusters included three smaller groups of cultivated A. parryi populations (in one case, along with the other two wild populations) and a small group with the three possible anthropogenic A. p. var. huachucensis populations. Bar charts indicating the proportion of ancestry for each individual (Fig. 4) showed a marked contrast in the level of admixture for wild (group 2) vs. anthropogenic populations (groups 3, 4, 5 and most of group 1), regardless of variety. A secondary peak in ΔK at K = 14 [also the level of maximum ln Pr(X|K)] indicated further subdivision within these larger groups. At this level, the large group of wild populations was divided into primarily single populations; and most cultivated populations remained grouped with at least some others on relatively long branches: JAK, POW, PUR in one cluster; TIL and TRT in another; and DWT and JFI in a third (Fig. 5). Interestingly, the specific makeup of these smaller cultivated groups differed from that in the morphologically based tree. The A. p. var. huachucensis populations formed two separate clusters: one comprising wild populations on a short branch, and the other with possible anthropogenic populations (ARR, AZT, BAB) on a long branch. The greater admixture in wild than anthropogenic populations was still evident in the bar chart showing ancestry proportions at K = 14 (Fig. 5).

Fig. 4.

Fig. 4.

Open in a new tab

Results from Bayesian clustering analysis of Agave parryi and A. p. var. huachucensis performed with STRUCTURE 2·3·4 (Pritchard et al., 2000; Falush et al., 2003), K = 5. (A) Bar charts depicting the individual Q matrix, where Q = the estimated group membership coefficients for each individual in each of the five clusters, with each individual represented by a vertical line divided into five coloured segments depicting the respective estimated cluster membership fractions; black vertical lines separate populations. (B) Neighbour-joining tree of 23 populations based on allozyme variation, where each tip represents one or more population.

Fig. 5.

Fig. 5.

Open in a new tab

Results from Bayesian clustering analysis of Agave parryi and A. p. var. huachucensis performed with STRUCTURE 2·3·4 (Pritchard et al., 2000; Falush et al., 2003), K = 14. (A) Bar charts depicting the individual Q matrix, where Q = the estimated group membership coefficients for each individual in each of the 14 clusters, with each individual represented by a vertical line divided into 14 coloured segments depicting the respective estimated cluster membership fractions; black vertical lines separate populations. (B) Neighbour-joining tree of 23 populations based on allozyme variation, where each tip represents one or more population.

Wild A. p. var. huachucensis populations had three of the four lowest mean pairwise FST values (calculated for all possible pairs of the 23 populations analysed; Table 6), indicating a strong genetic similarity on average to the other populations. In contrast, hypothesized anthropogenic A. p. var. huachucensis populations had three of the four highest mean FST values, slightly higher than most of the cultivated A. parryi populations, which underlined their genetic distinctiveness. All three anthropogenic A. p. var. huachucensis populations were most similar to (i.e. lowest pairwise FST values) each other; beyond that they were generally most similar to the three wild A. p. var. huachucensis populations. The lowest pairwise FST values for the three anthropogenic populations ranged from 0·061 to 0·231 (vs. 0·024–0·029 for wild), and minimum values shared with wild populations ranged from 0·220 to 0·249. MIL (wild), located in a canyon on the east side of the Huachuca Mountains, was the most similar genetically to two of the three anthropogenic populations, based on having the lowest pairwise FST values, even though SCO was at least 7 km closer to all anthropogenic populations than were the other two wild populations. Of particular note, the wild populations shared their highest pairwise FST values with anthropogenic A. p. var. huachucensis populations; they were more similar genetically to most anthropogenic populations of the other variety than to hypothesized anthropogenic populations of their own variety.

Table 6.

Rank order of Agave parryi var. huachucensis and A. parryi populations in Arizona by allozyme-based mean pairwise FST values with all other populations

Population Description* Mean FST
LER huach. 0·142
SCO huach. 0·148
CCC parryi, wild 0·153
MIL huach. 0·155
CHR parryi, wild 0·165
PIN parryi, wild 0·174
OME parryi, wild 0·198
SHR parryi, wild 0·199
SEN parryi, wild 0·202
PRE parryi, wild 0·208
JER parryi, wild 0·223
POW parryi, cult. 0·228
WCC parryi, cult. 0·235
PUR parryi, cult. 0·246
SAN parryi, wild 0·261
DWT parryi, cult. 0·285
TIL parryi, cult. 0·285
TRT parryi, cult. 0·287
JFI parryi, cult. 0·295
ARR huach. 0·296
JAK parryi, cult. 0·303
BAB huach. 0·337
AZT huach. 0·343
Open in a new tab

* huach., A. parryi var. huachucensis; parryi, wild, wild A. parryi; parryi, cult., cultivated A. parryi.

DISCUSSION

Varietal contrasts in morphological traits and genetic diversity

A number of contrasts were evident between A. p. var. huachucensis populations that occurred in south-eastern Arizona and A. parryi populations studied by Parker et al. (2010). The morphological differences documented herein were in keeping with previous descriptions of A. p. var. huachucensis as being more robust plants with larger leaves (Gentry, 1982; Reveal and Hodgson, 2002). In contrast, Nobel and Smith (1983) reported shorter leaves in A. p. var. huachucensis than in A. parryi, but their comparison was based on a much smaller sample of plants from only one location representing each taxon.

Previous discussions of the taxonomic status of A. p. var. huachucensis and A. parryi focused on morphological rather than genetic traits; our results indicated that intervarietal genetic contrasts were also evident, with taxonomic identity accounting for close to 10 % of the genetic variation apparent in the overall sample, whether for the allozyme-based (23 populations) or microsatellite-based (18 populations) AMOVA. Although this represents statistically significant differentiation between varieties, there is no absolute genetic threshold that defines distinct taxa (Morgan-Richards and Wolff, 1999). In our analysis, genetic identities were slightly greater for population pairs of the same taxon than for pairs involving the two taxa. Crawford (1985) and Gottlieb et al. (1985) reported that populations of different taxa often have only minimally different mean genetic identities for same-taxon and different-taxon population pairs, particularly if divergence has been relatively rapid and recent, as has been the case with Agave (Good-Avila et al., 2006). While our data reported modest but significant genetic and morphological contrasts between A. parryi and A. p. var. huachucensis, expanded analysis with more extensive sampling and additional molecular markers is necessary to identify conclusively phylogenetic relationships between these taxa.

Evolutionary history of A. p. var. huachucensis populations

The A. p. var. huachucensis populations differed morphologically and genetically according to their purported origin. Because characteristics of the three disjunct lower elevation populations were consistent with those of cultivated A. parryi populations (Parker et al., 2010), we infer that they were probably of anthropogenic origin. These populations were more uniform morphologically, with traits probably favoured by human selection, such as shorter terminal spines and greater production of basal offsets relative to wild populations. Reduced spine length would have facilitated harvesting and processing, which entailed removing the developing inflorescence and leaves and baking the remaining heart, or plant core, in roasting pits to convert starch to sugar (Hodgson, 2001b). Inflorescence production in many species (including A. parryi and A. p. var. huachucensis) stimulates the formation of basal offsets that replace the parent once it flowers and dies; plants with greater basal offset production may have experienced artificial selection because offsets facilitated ongoing restocking of Agave fields and maintenance of preferred traits through asexual reproduction (Parker et al., 2007; Zizumbo-Villarreal et al., 2013; but see González et al., 2003; Gil-Vega et al., 2006).

Anthropogenic A. p. var. huachucensis populations had low genetic diversity, maintaining an even smaller subset of the genetic variation found in wild populations (53 %) than has been reported for other cultivated species, including A. parryi (70 %; Parker et al., 2010) and a number of long-lived tree and columnar cactus species (88–100 %; Miller and Schaal, 2006). Several factors may have contributed to this marked loss of diversity, including selection for preferred traits by farmers, a predominance of asexual reproduction, and the isolated nature of populations and consequent reduced gene flow and sexual recruitment.

Whether crops in traditional agricultural systems experience diversifying or unifying selection depends in part on their specific use (Cruse-Sanders et al., 2013). Where farmers manage for many uses, they enhance phenotypic and genetic diversity (Pujol et al., 2005; Casas et al., 2006; Miller and Schaal, 2006). Modern mescal farmers in central Mexico grow several Agave landraces for a traditional industry that favours subtly different flavours and aromas to produce a variety of mescals (Zizumbo-Villarreal et al., 2013). Farmers manage other crops, however, for more uniform traits, often resulting in less genetic diversity relative to wild stock [e.g. tropical trees in Peru (Hollingsworth et al., 2005) or Agave in more commercialized operations (Vargas-Ponce et al., 2009)].

While we can determine management goals of modern farmers from interviews, we have less information about ancient farming practices. Agave was used in Mesoamerica for food, medicine, alcoholic beverages and fibre at the time of Spanish contact (Zizumbo-Villarreal et al., 2013); and at least seven south-eastern Arizona Agave species were used prehistorically by indigenous cultures (Hodgson, 2001b). Historically, the native O'odham in Arizona who harvested Agave from the wild selected different species for different uses (Radding, 2012). If prehistoric use was tailored to different species, rather than to landraces within the same species as has been reported for potatoes, mescal and some other crops (Brush et al., 1995; Zizumbo-Villarreal et al., 2013), the effect would be to reduce phenotypic and genetic variation within species, as reported herein and by Parker et al. (2007, 2010) for other cultivated Agave species in Arizona.

Parker et al. (2007, 2010) found that spatial patterns of genetic similarity among prehistorically cultivated Agave populations seen in the landscape today reflect anthropogenic selection and gene flow occurring against a backdrop of natural processes. Farmers often asexually propagate individuals with certain traits to increase their preferred stock and eliminate plants lacking desirable traits. In addition to reducing genetic variation within populations (Miller and Schaal, 2006), this often increases genetic differentiation among populations (Nabhan, 1985), which we reported for A. p. var. huachucensis. Movement of planting stock among locations as farmers migrate or trade with others creates a complex pattern of gene flow that differs from that resulting from non-anthropogenic pollen and seed movement. For example, in virtually all cases for A. p. var. huachucensis and A. parryi (except one population of the former taxon), anthropogenic populations were most similar to a more geographically distant wild population. For both taxa, the wild population most similar to a number of anthropogenic populations had the greatest genetic diversity. In addition, anthropogenic A. p. var. huachucensis populations were more genetically divergent from the most similar wild populations than was reported for A. parryi (Parker et al., 2010), despite their closer proximity to potential wild sources. Two of three anthropogenic A. p. var. huachucensis populations were within 30 km of wild populations, whereas the closest wild sources to cultivated A. parryi populations were 33–60 km away (Parker et al., 2010). More in-depth archaeological information would help to explain the idiosyncrasies of the specific patterns observed.

Post-abandonment conditions may also have affected patterns of genetic diversity and structure in the relict populations we examined. Genetic drift or non-anthropogenic selection may have led to differential loss of some genotypes from extant populations following active cultivation. Additional genetic diversity could have been lost due to destruction of populations by recent development. Although fixed heterozygosity has been reported for modern traditional crops (Pujol et al., 2005), we cannot rule out passive proliferation of heterozygotes through repeated asexual reproduction over centuries as an explanation for the frequent fixed heterozygosity observed in anthropogenic A. p. var. huachucensis populations.

Wild and cultivated populations may also have differed in the frequency of gene flow and sexual recruitment, particularly with greater isolation of populations in more open habitats at lower elevations (although, as noted above, these populations were less geographically isolated than the A. parryi populations studied by Parker et al., 2010). The distance between most anthropogenic and wild A. p. var. huachucensis populations is within the foraging range of the lesser long-nosed bat (Leptonycteris yerbabuenae) and Mexican long-tongued bat (Choeronycteris mexicana), which feed on A. parryi and A. palmeri in south-eastern Arizona (Lindsay et al., 2012). Slauson (2000) found, however, that bats' sporadic overlap with flowering makes other pollinators more important for A. palmeri. If A. p. var. huachucensis is most successfully serviced by more narrowly ranging pollinators than bats, the isolation of anthropogenic populations could have contributed to their reduced genetic diversity.

CONCLUSIONS

Patterns of morphological and genetic variation in the A. p. var. huachucensis populations reflected their evolutionary history; they also showed distinct contrasts with A. parryi. Populations occurring at lower elevations in more open habitats were probably the result of pre-Columbian agriculturalists moving plant stock from wild populations in the nearby mountains to more accessible locations, where they remain as relicts in the landscape today. For most modern crops, ongoing selection over millennia with continuing use has obscured details of prehistoric domestication (Hughes et al., 2007). In contrast, sites like the study area, with pre-Columbian cultivation followed by centuries of little or no anthropogenic manipulation, are some of the most fruitful for examining early domestication practices. Although many questions remain, genetic analyses can help to identify sites that merit further archaeological investigation of early farming practices. They also improve our understanding of how indigenous cultures shaped prehistoric landscapes – a legacy that remains today in many areas.

ACKNOWLEDGEMENTS

We thank Albert J. Parker for field assistance; Cecile Deen and the Hamrick lab personnel for laboratory assistance; Stephen Plog for recommendations of A. parryi populations and Mark Dimmitt and Chris Haas for recommendations of A. p. var. huachucensis populations to sample; Doug Ruppel for logistical assistance with sampling on the Babacomari Ranch; Julie Solometo, Kate Spielman, Jeffrey Clark, Suzy Fish, Paul Fish and Patrick Lyons for valuable discussions about the archaeological context; Scott Markwith for assistance with TETRASAT; Robin Rodgers for help with calculation of interpopulation distances; and Tommy Jordan for assistance with ArcView. This work was supported by the National Science Foundation (#BCS-0216832 to K.C.P. and J.L.H.).

LITERATURE CITED

  1. APG III (The Angiosperm Phylogeny Group) An update of the Angiosperm Phylogeny Group classification for the orders and families of flowering plants: APG III. Botanical Journal of the Linnean Society. 2009;161:105–121. [Google Scholar]
  2. Becher SA, Steinmetz K, Weising K, et al. Microsatellites for cultivar identification in Pelargonium. Theoretical and Applied Genetics. 2000;101:643–651. [Google Scholar]
  3. Blancas J, Casas A, Rangel-Landa S, et al. Plant management in the Tehuacán–Cuicatlán Valley, Mexico. Economic Botany. 2010;64:287–302. [Google Scholar]
  4. Bogler DJ, Simpson BB. Phylogeny of Agavaceae based on ITS rDNA sequence variation. American Journal of Botany. 1996;83:1225–1235. [Google Scholar]
  5. Brush S, Kesseli R, Ortega R, et al. Potato diversity in the Andean center of crop domestication. Conservation Biology. 1995;9:1189–1198. doi: 10.1046/j.1523-1739.1995.9051176.x-i1. [DOI] [PubMed] [Google Scholar]
  6. Casas A, Cruse-Sanders J, Morales E, Otero-Arnaiz A, Valiente-Banuet A. Maintenance of phenotypic and genotypic diversity in managed populations of Stenocereus stellatus (Cactaceae) by indigenous peoples in Central Mexico. Biodiversity and Conservation. 2006;15:879–898. [Google Scholar]
  7. Casas A, Otero-Arnaiz A, Peréz-Negrón E, Valiente-Banuet A. In situ management and domestication of plants in Mesoamerica. Annals of Botany. 2007;100:1101–1115. doi: 10.1093/aob/mcm126. [DOI] [PMC free article] [PubMed] [Google Scholar]
  8. Cheliak WM, Pitel JA. Electrophoretic identification of clones in trembling aspen. Canadian Journal of Forest Research. 1984;14:740–743. [Google Scholar]
  9. Colunga-GarcíaMarín P, May-Pat F. Agave studies in Yucatan, Mexico. I. Past and present germplasm diversity and uses. Economic Botany. 1993;47:312–327. [Google Scholar]
  10. Crawford DJ. Electrophoretic data and plant speciation. Systematic Botany. 1985;10:405–416. [Google Scholar]
  11. Cruse-Sanders JM, Parker KC, Friar EA, et al. Managing diversity: domestication and gene flow in Stenocereus stellatus Riccob. (Cactaceae) in Mexico. Ecology and Evolution. 2013;3:1340–1355. doi: 10.1002/ece3.524. [DOI] [PMC free article] [PubMed] [Google Scholar]
  12. Doebley J. Isozymic evidence and the evolution of crop plants. In: Soltis DE, Soltis PS, editors. Isozymes in plant biology. Portland, OR: Dioscorides Press; 1989. pp. 165–191. [Google Scholar]
  13. Don RH, Cox PT, Wainwright BJ, Baker K, Mattick JS. ‘Touchdown’ PCR to circumvent spurious priming during gene amplification. Nucleic Acids Research. 1991;19 doi: 10.1093/nar/19.14.4008. 4008. [DOI] [PMC free article] [PubMed] [Google Scholar]
  14. Ellstrand NC, Roose ML. Patterns of genotypic diversity in clonal plant species. American Journal of Botany. 1987;74:123–131. [Google Scholar]
  15. Evanno G, Regnaut S, Goudet J. Detecting the number of clusters of individuals using the software STRUCTURE: a simulation study. Molecular Ecology. 2005;14:2611–2620. doi: 10.1111/j.1365-294X.2005.02553.x. [DOI] [PubMed] [Google Scholar]
  16. Falush D, Stephens M, Pritchard JK. Inference of population structure using multilocus genotype data: linked loci and correlated allele frequencies. Genetics. 2003;164:1567–1587. doi: 10.1093/genetics/164.4.1567. [DOI] [PMC free article] [PubMed] [Google Scholar]
  17. Fish SK. Hohokam impacts on Sonoran Desert environment. In: Lentz DL, editor. Imperfect balance: Landscape transformations in the precolumbian Americas. New York: Columbia University Press; 2000. pp. 251–280. [Google Scholar]
  18. Fish SK, Nabhan GP. Desert as context: the Hohokam environment. In: Gumerman GJ, editor. Exploring the Hohokam: Prehistoric desert peoples of the American Southwest. Albuquerque, NM: University of New Mexico Press; 1991. pp. 29–60. [Google Scholar]
  19. Fish SK, Fish PR, Miksicek C, Madsen J. Prehistoric agave cultivation in southern Arizona. Desert Plants. 1985;7:107–112. 100. [Google Scholar]
  20. Ford RI. Gardening and farming before A.D. 1000: patterns of prehistoric cultivation north of Mexico. Journal of Ethnobiology. 1981;1:6–27. [Google Scholar]
  21. Gentry HS. Agaves of continental North America. Tucson, AZ: University of Arizona Press; 1982. [Google Scholar]
  22. Gil-Vega K, Díaz C, Nava-Cedillo A, Simpson J. AFLP analysis of Agave tequilana varieties. Plant Science. 2006;170:904–909. [Google Scholar]
  23. González G, Alemán S, Infante D. Asexual genetic variability in Agave fourcroydes II: selection among individuals in a clonally propagated population. Plant Science. 2003;165:595–601. [Google Scholar]
  24. Good-Avila SV, Souza V, Gaut BS, Eguiarte LE. Timing and rate of speciation in Agave (Agavaceae) Proceedings of the National Academy of Sciences, USA. 2006;103:9124–9129. doi: 10.1073/pnas.0603312103. [DOI] [PMC free article] [PubMed] [Google Scholar]
  25. Gottlieb LD, Warwick SI, Ford VS. Morphological and electrophoretic divergence between Layia discoidea and L. glandulosa. Systematic Botany. 1985;10:484–495. [Google Scholar]
  26. Hamrick JL, Godt MJW. Allozyme diversity in plant species. In: Brown AHD, Clegg MT, Kahler AL, Weir BS, editors. Plant population genetics, breeding and genetic resources. Sunderland, MA: Sinauer Associates; 1989. pp. 43–63. [Google Scholar]
  27. Hamrick JL, Linhart YB, Mitton JB. Relationships between life history characteristics and electrophoretically detectable genetic variation in plants. Annual Review of Ecology and Systematics. 1979;10:173–200. [Google Scholar]
  28. Hardy OJ, Vekemans X. SPAGeDi: a versatile computer program to analyse spatial genetic structure at the individual or population levels. Molecular Ecology Notes. 2002;2:618–620. [Google Scholar]
  29. Hodgson WC. Taxonomic novelties in American Agave (Agavaceae) Novon. 2001a;11:410–416. [Google Scholar]
  30. Hodgson WC. Food plants of the Sonoran Desert. Tucson, AZ: University of Arizona Press; 2001b. [Google Scholar]
  31. Hodgson WC, Slauson L. Agave delamateri (Agavaceae) and its role in the subsistence patterns of pre-Columbian cultures in Arizona. Haseltonia. 1995;3:130–140. [Google Scholar]
  32. Hodgson WC, Salywon A. Two new Agave species (Agavaceae) from central Arizona and their putative pre-Columbian domesticated origins. Brittonia. 2013;65:5–15. [Google Scholar]
  33. Hollingsworth PM, Dawson IK, Goodall-Copestake WP, et al. Do farmers reduce genetic diversity when they domesticate tropical trees? A case study from Amazonia. Molecular Ecology. 2005;14:497–501. doi: 10.1111/j.1365-294X.2005.02431.x. [DOI] [PubMed] [Google Scholar]
  34. Hughes CE, Govindarajulu R, Robertson A, et al. Serendipitous backyard hybridization and the origin of crops. Proceedings of the National Academy of Sciences, USA. 2007;104:14389–14394. doi: 10.1073/pnas.0702193104. [DOI] [PMC free article] [PubMed] [Google Scholar]
  35. Lindsay DL, Edwards CE, Jung MG, Bailey P, Lance RF. Novel microsatellite loci for Agave parryi and cross-amplification in Agave palmeri (Agavaceae) American Journal of Botany. 2012;99 doi: 10.3732/ajb.1200033. e295–e297. doi:10·3732/ajb.1200033. [DOI] [PubMed] [Google Scholar]
  36. Markwith SH, Stewart DJ, Dyer JL. Tetrasat: a program for the population analysis of allotetraploid microsatellite data. Molecular Ecology Notes. 2006;6:586–589. [Google Scholar]
  37. Miller AJ, Schaal BA. Domestication and the distribution of genetic variation in wild and cultivated populations of the Mesoamerican fruit tree Spondias purpurea L. (Anacardiaceae) Molecular Ecology. 2006;15:1467–1480. doi: 10.1111/j.1365-294X.2006.02834.x. [DOI] [PubMed] [Google Scholar]
  38. Minnis PE, Plog SE. A study of the site specific distribution of Agave parryi in east central Arizona. Kiva. 1976;41:299–308. [Google Scholar]
  39. Mitton JB, Linhart YB, Sturgeon KB, Hamrick JL. Allozyme polymorphisms detected in mature needle tissue of ponderosa pine. Journal of Heredity. 1979;70:86–89. [Google Scholar]
  40. Morgan-Richards M, Wolff K. Genetic structure and differentiation of Plantago major reveals a pair of sympatric sister species. Molecular Ecology. 1999;8:1027–1036. [Google Scholar]
  41. Nabhan G. Native crop diversity in Aridoamerica: conservation of regional gene pools. Economic Botany. 1985;39:387–399. [Google Scholar]
  42. Nabhan GP. Threatened Native American plants. Endangered Species Update. 1992;9(1–4) 9. [Google Scholar]
  43. Nei M. Genetic distance between populations. American Naturalist. 1972;106:283–292. [Google Scholar]
  44. Nobel PS, Smith SD. High and low temperature tolerances and their relationships to distribution of agaves. Plant, Cell and Environment. 1983;6:711–719. [Google Scholar]
  45. Parker KC, Hamrick JL, Hodgson WC, et al. Genetic consequences of pre-Columbian cultivation for Agave murpheyi and A. delamateri (Agavaceae) American Journal of Botany. 2007;94:1479–1490. doi: 10.3732/ajb.94.9.1479. [DOI] [PubMed] [Google Scholar]
  46. Parker KC, Trapnell DW, Hamrick JL, Hodgson WC, Parker AJ. Inferring ancient Agave cultivation practices from contemporary genetic patterns. Molecular Ecology. 2010;19:1622–1637. doi: 10.1111/j.1365-294X.2010.04593.x. [DOI] [PubMed] [Google Scholar]
  47. Parra F, Pérez-Nasser N, Lira R, et al. Population genetics and process of domestication of Stenocereus pruinosus (Cactaceae) in the Tehuacán Valley, México. Journal of Arid Environments. 2008;72:1997–2010. [Google Scholar]
  48. Peakall R, Smouse PE. GENALEX 6: genetic analysis in Excel. Population genetic software for teaching and research. Molecular Ecology Notes. 2006;6:288–295. doi: 10.1093/bioinformatics/bts460. [DOI] [PMC free article] [PubMed] [Google Scholar]
  49. Peakall R, Smouse PE. GenAlEx 6·5: genetic analysis in Excel. Population genetic software for teaching and research – an update. Bioinformatics. 2012;28:2537–2539. doi: 10.1093/bioinformatics/bts460. [DOI] [PMC free article] [PubMed] [Google Scholar]
  50. Pinkava DJ, Baker MA. Chromosome and hybridization studies of agaves. Desert Plants. 1985;7:93–100. [Google Scholar]
  51. Pritchard JK, Stephens M, Donnelly P. Inference of population structure using multilocus genotype data. Genetics. 2000;155:945–959. doi: 10.1093/genetics/155.2.945. [DOI] [PMC free article] [PubMed] [Google Scholar]
  52. Pujol B, David P, McKey D. Microevolution in agricultural environments: how a traditional Amerindian farming practice favours heterozygosity in cassava (Manihot esculenta Crantz, Euphorbiaceae) Ecology Letters. 2005;8:138–147. [Google Scholar]
  53. Radding C. The children of Mayahuel: Agaves, human cultures, and desert landscapes in northern Mexico. Environmental History. 2012;17:84–115. [Google Scholar]
  54. Reveal JL, Hodgson WC. Agave Linnaeus. In: Flora of North America Editorial Committee., editor. Flora of North America Vol. 26. New York: Oxford University Press; 2002. pp. 442–461. [Google Scholar]
  55. Rohlf FJ. Port Jefferson, NY: Applied Biostatistics, Inc; 2005. NTSYSpc: Numerical taxonomy and multivariate analysis system, version 2·20b. [Google Scholar]
  56. Saitou N, Nei M. The neighbor-joining method: a new method for reconstructing phylogenetic trees. Molecular Biology and Evolution. 1987;4:406–425. doi: 10.1093/oxfordjournals.molbev.a040454. [DOI] [PubMed] [Google Scholar]
  57. Sauer C. Seeds, spades, hearths, and herds. The domestication of animals and foodstuffs. Cambridge, MA: MIT Press; 1972. [Google Scholar]
  58. Schable NA, Fischer RU, Glenn TC. Tetranucleotide microsatellite DNA loci from the dollar sunfish (Lepomis marginatus) Molecular Ecology Notes. 2002;2:509–511. [Google Scholar]
  59. Slauson LA. Pollination biology of two chiropterophilous agaves in Arizona. American Journal of Botany. 2000;87:825–836. [PubMed] [Google Scholar]
  60. Smith CE., Jr. Plant remains. In: Byers DS, editor. The prehistory of the Tehuacan Valley. Vol. 1. Austin, TX: University of Texas Press; 1967. pp. 220–256. [Google Scholar]
  61. Soltis PS, Soltis DE. The role of genetic and genomic attributes in the success of polyploids. Proceedings of the National Academy of Sciences. 2000;97:7051–7057. doi: 10.1073/pnas.97.13.7051. [DOI] [PMC free article] [PubMed] [Google Scholar]
  62. Thrall PH, Young A. AUTOTET: a program for analysis of autotetraploid genotypic data. Journal of Heredity. 2000;91:348–349. doi: 10.1093/jhered/91.4.348. [DOI] [PubMed] [Google Scholar]
  63. Trapnell DW, Hamrick JL, Parker KC, Braungart KW, Glenn TC. Evaluating the utility of microsatellites for investigations of autopolyploid taxa. Journal of Heredity. 2011;102:473–478. doi: 10.1093/jhered/esr045. [DOI] [PubMed] [Google Scholar]
  64. Ullrich B. On the history of Agave asperrima and A. scabra (Agavaceae) as well as some taxa of the Parryanae. Sida. 1992;15:257–259. [Google Scholar]
  65. Vargas-Ponce O, Zizumbo-Villarreal D, Martínez-Castillo J, Coello-Coello J, Colunga-GarcíaMarín P. Diversity and structure of landraces of Agave grown for spirits under traditional agriculture: a comparison with wild populations of A. angustifolia (Agavaceae) and commercial plantations of A. tequilana. American Journal of Botany. 2009;96:448–457. doi: 10.3732/ajb.0800176. [DOI] [PubMed] [Google Scholar]
  66. Weeden NF, Wendel JF. Genetics of plant isozymes. In: Soltis DE, Soltis PS, editors. Isozymes in plant biology. Portland, OR: Dioscorides Press; 1989. pp. 46–72. [Google Scholar]
  67. Wright S. The interpretation of population structure by F-statistics with special regard to systems of mating. Evolution. 1965;19:395–420. [Google Scholar]
  68. Zizumbo-Villarreal D, Vargas-Ponce O, Rosales-Adame JJ Colunga-García-Marín. Sustainability of the traditional management of Agave genetic resources in the elaboration of mezcal and tequila spirits in western Mexico. Genetic Resources and Crop Evolution. 2013;60:33–47. [Google Scholar]

Articles from Annals of Botany are provided here courtesy of Oxford University Press

RESOURCES