It is commonly believed that habitat degradation and fragmentation decrease genetic variation within plant populations.  This hypothesis needs to be tested for those species that can withstand the harsh conditions left by desertification, a worldwide process in semiarid lands.  Malacomeles denticulata is a common shrub in the Rosaceae, which can grow in a large variety of habitats, including severely degraded and eroded areas, and displays a large phenotypic variation among populations.  We study the nature of such variations by means of a common garden experiment and single nucleotide polymorphisms (SNPs).  For the garden experiment we used a half-sib design. Three phenotypic characters were examined: leaf area, leaf indentation and relative growth rate in total leaf area per plant (RGR).  For the SNP study, an assay was designed from a partial M. denticulata DNA sequence 5.8S ITS 1-2 ribosomal RNA gene. The sequence was analysed with the Repeat Masker and Basic Local Alignment Search Tool (BLAST) software.  The assay included a pair of primers, and two fluorescent probes (VIC and FAM) using a real time PCR machine.  Eight and four populations of distinct habitats, including a barren area with deep gully incisions, were analyzed for SNPs and quantitative traits respectively

Results/Conclusions

Leaf area was the most variable character and had the highest genetic variance, followed by leaf indentation. RGR displayed little or non-detectable genetic variation in the studied populations.  Two populations had little or undetectable genetic variation for the quantitative traits and two were genetically variable including the population in the degraded area. We detected evidence of genetic differentiation among the four studied populations for leaf area and leaf indentation but not for relative growth rate.  Different habitat selection regimes for leaf area and indentation, and common selective pressures for RGR among localities may explain these results. The SNP design assay yielded recognizable homozygous and heterozygous genotypes.  Heterozygous individuals were only detected in the most degraded locality, where a negative value of inbreeding reveal an excess of heterozygosity (0.44) relative to Hardy Weinberg expectations (0.34).  Contrary to the above hypothesis, the population in the highly degraded area displayed high genetic variation (the highest for SNPs).  The high frequency of heterozygosity is probably due to heterosis in a gene in linkage disequilibrium with the SNP.  These results suggest the importance of genetic variation for survival in a highly degraded area