Insights Into the Almond Domestication History

Abstract

Understanding crop domestication offers crucial insights into the evolutionary processes that drive population divergence and adaptation. It also informs the identification of genetically diverse wild germplasm, which is essential for breeding and conservation efforts. While domestication has been extensively studied in many Mediterranean fruit trees, the evolutionary history of the almond (Prunus dulcis) remains comparatively underexplored. To address this, we analyzed 209 wild and cultivated almond accessions sampled across Eurasia and genotyped with 23 microsatellite markers. Using population genetics and coalescent-based inference, we reconstructed the domestication history of P. dulcis and its relationships with wild relatives. Bayesian clustering revealed four genetically distinct clusters of cultivated almonds: Turkish, Caucasian-Central Asian, Southern Spanish, and European/North American. These groups were differentiated from wild almond species-including Prunus turcomanica, Prunus orientalis, Prunus fenzliana, and Prunus spinosissima-each forming its gene pool across the Middle East and Central Asia. Approximate Bayesian Computation (ABC) supported a single domestication event in the Middle East, originating from either P. orientalis or P. turcomanica, with subsequent gene flow from P. fenzliana and P. spinosissima into the Turkish and Central Asian cultivated gene pools, respectively. We also inferred reciprocal introgression from cultivated almonds back into wild populations. Notably, sharka resistance-caused by plum pox virus (PPV)-was identified in three P. dulcis clusters and P. fenzliana, suggesting that resistance may have arisen independently or been maintained through crop-wild introgression. Together, our results highlight a complex and protracted domestication history for almond, shaped by contributions from multiple wild relatives and recurrent gene flow. These findings enhance our understanding of perennial crop evolution and underscore the value of wild germplasm in breeding programs aimed at increasing resilience in fruit trees.

Keywords: Mediterranean Basin; Prunus; almonds; domestication; fruit trees; gene flow; genetic resources; sharka; virus.

FIGURE 1

Population genetic structure of Prunus dulcis (N = 89) and its wild relative species (Prunus fenzliana, N = 19, Prunus orientalis , N = 6, Prunus spinosissima , N = 8, and Prunus turcomanica , N = 5) inferred with STRUCTURE at K = 7. (A) Classification of almond and almond‐related species. Adapted from Browicz and Zohary (1996) and Grasselly (1976). Underlined, species included in this study. (B) Bayesian clustering results inferred with STRUCTURE at K = 7. Underneath the figure are depicted the names of the crop and wild Amygdalus populations as described in Table 2. Each individual is represented by a vertical bar, divided into K segments representing the amount of ancestry in its genotype corresponding to K clusters. (*) refers to Plum Pox virus‐resistant samples as listed in Table S1. Species names are provided primarily as contextual information to relate our genetic clusters to previously identified taxonomic classifications. (C) Spatial population genetic structure of Amygdalus clusters and its wild relatives along the Northern hemisphere. The world map can be downloaded under a free license at https://www.vecteezy.com/vector‐art/10961532‐world‐map‐vector‐illustration‐isolated‐on‐grey‐background‐flat‐earth‐globe‐or‐world‐map.

FIGURE 2

Factorial correspondence analyses (FCAs). (A) Including Prunus dulcis individuals (N = 98), colored as in Figure 1. (B) Including P. dulcis and related species individuals (N = 131) with membership coefficient ≥ 90% to a cluster in the STRUCTURE analysis at K = 7, colored as in Figure S5. The cluster numbers correspond to (1–4) P. dulcis , (5) P. fenzliana, (6) Prunus orientalis and Prunus turcomanica , and (7) Prunus spinosissima. Note that individuals with a probability of assignment to one of the seven clusters < 90% were removed.

FIGURE 3

Rooted weighted neighbor‐joining (NJ) tree among cultivated and wild almond species performed with DARwin. The 131 non‐admixed samples belong to six different Prunus species including P. communis (N = 4) in pink and P. dulcis (N = 89) in pink, light blue, and dark blue, P. fenzliana (N = 19) in yellow, P. orientalis (N = 6) in gray, P. spinosissima (N = 8) in green, and P. turcomanica (N = 5) in dark gray. In light gray (far right), P. kuramica which was used to root the tree. The NJ tree was built with DARwin, bootstrap support values were obtained from 30,000 repetitions. Bootstrap values when greater than 50% are shown above the branches. Species names are provided primarily as contextual information to relate our genetic clusters to previously identified taxonomic classifications.

FIGURE 4

The most likely scenario of domestication of the cultivated almond ( Prunus dulcis ) with parameter estimates for effective population size and divergence time. The mean posterior estimates for each parameter are written in bold and associated with 95% confidence intervals. The divergence time between populations X and Y, T _X–Y , was provided. N _X : effective population size of population X; m _X–Y : bidirectional arrows represent gene flow between populations X and Y. Each population detected with STRUCTURE is coded and colored accordingly to Table 2.