Present spatial diversity patterns of Theobroma cacao L. in the neotropics reflect genetic differentiation in pleistocene refugia followed by human-influenced dispersal

Abstract

Cacao (Theobroma cacao L.) is indigenous to the Amazon basin, but is generally believed to have been domesticated in Mesoamerica for the production of chocolate beverage. However, cacao's distribution of genetic diversity in South America is also likely to reflect pre-Columbian human influences that were superimposed on natural processes of genetic differentiation. Here we present the results of a spatial analysis of the intra-specific diversity of cacao in Latin America, drawing on a dataset of 939 cacao trees genotypically characterized by means of 96 SSR markers. To assess continental diversity patterns we performed grid-based calculations of allelic richness, Shannon diversity and Nei gene diversity, and distinguished different spatially coherent genetic groups by means of cluster analysis. The highest levels of genetic diversity were observed in the Upper Amazon areas from southern Peru to the Ecuadorian Amazon and the border areas between Colombia, Peru and Brazil. On the assumption that the last glaciation (22,000-13,000 BP) had the greatest pre-human impact on the current distribution and diversity of cacao, we modeled the species' Pleistocene niche suitability and overlaid this with present-day diversity maps. The results suggest that cacao was already widely distributed in the Western Amazon before the onset of glaciation. During glaciations, cacao populations were likely to have been restricted to several refugia where they probably underwent genetic differentiation, resulting in a number of genetic clusters which are representative for, or closest related to, the original wild cacao populations. The analyses also suggested that genetic differentiation and geographical distribution of a number of other clusters seem to have been significantly affected by processes of human management and accompanying genetic bottlenecks. We discuss the implications of these results for future germplasm collection and in situ, on farm and ex situ conservation of cacao.

Figure 1. Species richness of genus Theobroma.

Left: observed species richness in 10 minute grid cells and a circular neighborhood of 1 decimal degree; Right: modeled species richness in 2.5 minute grid cells.

Figure 2. Spatial variation of different genetic parameters, represented at a resolution of ten minute grid cells and a circular neighborhood of 1 degree.

Highest values are consistently observed in the extensive bean-shaped Amazonian area covering both the Peruvian-Brazilian border, and the southern part of the Colombian-Brazilian border, as well as Amazonian Ecuador.

Figure 3. Observed locally common alleles compared to past and current modeled distribution of cacao.

Upper: distribution of areas with modeled habitat suitability of cacao during the LGM; red dashed polygons show potential relatively isolated refugia associated with areas holding high levels of locally common alleles. Lower: changes in cacao habitat suitability from the LGM until present; red areas represent potential habitat suitability during LGM but no longer at present (high impact or restriction areas); green indicates areas with continued habitat suitability from LGM until present (low impact or stable areas); and blue indicates areas that were probably not suitable for cacao at the LGM, but are suitable at present (new or expansion areas).

Figure 4. Scores of sampled trees as projected on the first ordination axis of the biplot of a Spatial Analysis of Principal Components.

Location of the genetic cline coincides with the bean-shaped area (and particularly the southern part of it) where the highest values of the measured genetic parameters were observed (figure 2).

Figure 5. Overview of the different locations of the ten clusters identified by k-means clustering.

The three subclusters of cluster 6 are highlighted with different colours, clearly distinguishing the group that is largely composed of the Nacional cultivar of the Ecuadorean coastal plains (red colour).

Figure 6. Cluster richness, i.e. the number of different clusters shown in figure 5 that occur in a given area.

Figure 7. Complete linkage clustering based on Nei’s distance (cophenetic correlation = 0.92).

Figure 8. Ordination diagram of a Principal Coordinate Analysis applied on the cacao dataset, using Nei’s distance.

The first two axes show 69% of the variation in data. Environmental variables were added a posteriori through vector fitting. Arrows point in the direction of most rapid change in the variable and their length is proportional to the correlation between ordination and variable. According to the classification used by , Cluster1 = Purus; Cluster2 = Criollo; Cluster3 = Guiana; Cluster4 = Marañon-Amazon River; Cluster5 = Amelonado; Cluster6 = Contamana + Nacional (+Purus); Cluster7 = Marañon-Rondônia; Cluster8 = Iquitos (+Purus); Cluster9 = Nanay; Cluster10 = Curaray (alt = altitude; BIO1 = Annual mean temperature; BIO2 = Mean diurnal range (max temp – min temp) (monthly average); BIO3 = Isothermality (BIO1/BIO7) * 100; BIO4 = Temperature Seasonality (Coefficient of Variation); BIO5 = Max Temperature of Warmest Period; BIO6 = Min Temperature of Coldest Period; BIO7 = Temperature Annual Range (BIO5–BIO6); BIO8 = Mean Temperature of Wettest Quarter; BIO9 = Mean Temperature of Driest Quarter; BIO10 = Mean Temperature of Warmest Quarter; BIO11 = Mean Temperature of Coldest Quarter; BIO12 = Annual Precipitation; BIO13 = Precipitation of Wettest Period; BIO14 = Precipitation of Driest Period; BIO15 = Precipitation Seasonality (Coefficient of Variation); BIO16 = Precipitation of Wettest Quarter; BIO17 = Precipitation of Driest Quarter; BIO18 = Precipitation of Warmest Quarter; BIO19 = Precipitation of Coldest Quarter).

Figure 9. Observed locally common alleles compared to current and future modeled distribution of cacao.

Upper: predicted changes in cacao habitat suitability from present until 2050; red areas represent potential habitat suitability at present but no longer by 2050 (high impact or restriction areas); green indicates areas with continued habitat suitability from present until 2050 (low impact or stable areas); and blue indicates areas which are currently unsuitable for cacao, but may become suitable by 2050 (new or expansion areas) Lower: distribution of areas with modeled habitat suitability of cacao by 2050, overlaid with the location of currently existing protected areas.