Juglans regia L : genetic variation and provenance performance
A range-wide collection of Juglans regia seeds was undertaken in autumn 1997 from 12 countries, including 25 provenances and 375 half-sib progenies. 2200 seedlings were produced using innovative nursery techniques. The seedlings were planted in three provenance trials in southern England in 1999, the largest of which acted as a combined provenance/progeny trial. After one growing season, survival was 98.9 %, mean height growth 35 cm, and mean stem diameter increment 5 mm. Provenance differences for both height and stem diameter increment were highly significant (p<0.001). There were no significant genotype × environment interactions. Flushing assessments revealed few significant differences between provenances and flushing was complete by early April. Family heritability for tree height was 0.19 at one site and, with combined selection, genetic gain was estimated at 8 %. The effects of three types of treeshelter and a stumping treatment on walnut establishment were tested over three growing seasons. Treeshelters were found beneficial to height increment. However, 120 cm tall shelters promoted early flushing, and consequent risk of increased frost damage, and caused more stem die-back than 75 cm shelters. Stumping promoted rapid early height increment but gave no longer-term benefit. The crown (cd) and stem (dbh) diameter at breast height relationship of open growing trees in Britain was assessed and was highly significant (r2 = 0.96, p<0.001). The regression equation (cd = 2.71 + 17.6dbh) permitted the estimation of suitable planting densities for the provenance trials and the calculation of a thinning regime. Isozyme analysis of the 375 genotypes identified 20 loci in 15 enzyme systems with seed embryo extracts. Using young leaf extracts, the polymorphic locus Pgm-1 indicated low expected heterozygosity of 0.06 both within populations and at the species level. FST and GST estimates, both 0.05, indicated high uniformity among populations. Genetic distance estimates did not identify significant clustering consistent with geographic origin.