The influence of planting density on the production of 'Goldfinger' (Musa spp., AAAB) in the subtropics

'Goldfinger', a tetraploid banana produced from the Fundación Hondureña de Investigación Agrícola (FHIA) breeding program, was released to the Australian industry in 1995. It was promoted as an apple-flavoured dessert banana with resistance to Fusarium wilt race 1 and subtropical race 4, as well as resistance to black and yellow Sigatoka (Mycosphaerella fijiensis and M. musicola, respectively). This study was initiated to provide agronomic information to the banana industry, which was under threat from Fusarium wilt, on a new cultivar which could replace 'Williams' (AAA, Cavendish subgroup) or 'Lady Finger' (AAB, Pome subgroup) in those areas affected by Fusarium wilt. Also few studies had reported on the production characteristics of the new tetraploid hybrids, especially from subtropical areas, and therefore two field sites, one a steep-land farm and the other a level, more productive site, were selected for planting density and spatial arrangement treatments. The optimum density in terms of commercial production, taking into account bunch weight, finger size, length of the production cycle, plant height and ease of management, was 1680 plants/ha on the steep-land site where plants were planted in single rows with 2.5 m × 2.5 m spacings. However on the level site a double-row triangular layout with inter-row distances of 4.5 m to allow vehicular access (1724 plants/ha) gave the best results. With this arrangement plants were in an alternate, triangular arrangement along a row and a spacing of 1.5 m between plants at the points of each triangle and between each block of triangles.

How accurate are the marker orders in crop linkage maps generated from large marker datasets?

Marker ordering during linkage map construction is a critical component of QTL mapping research. In recent years, high-throughput genotyping methods have become widely used, and these methods may generate hundreds of markers for a single mapping population. This poses problems for linkage analysis software because the number of possible marker orders increases exponentially as the number of markers increases. In this paper, we tested the accuracy of linkage analyses on simulated recombinant inbred line data using the commonly used Map Manager QTX (Manly et al. 2001: Mammalian Genome 12, 930-932) software and RECORD (Van Os et al. 2005: Theoretical and Applied Genetics 112, 30-40). Accuracy was measured by calculating two scores: % correct marker positions, and a novel, weighted rank-based score derived from the sum of absolute values of true minus observed marker ranks divided by the total number of markers. The accuracy of maps generated using Map Manager QTX was considerably lower than those generated using RECORD. Differences in linkage maps were often observed when marker ordering was performed several times using the identical dataset. In order to test the effect of reducing marker numbers on the stability of marker order, we pruned marker datasets focusing on regions consisting of tightly linked clusters of markers, which included redundant markers. Marker pruning improved the accuracy and stability of linkage maps because a single unambiguous marker order was produced that was consistent across replications of analysis. Marker pruning was also applied to a real barley mapping population and QTL analysis was performed using different map versions produced by the different programs. While some QTLs were identified with both map versions, there were large differences in QTL mapping results. Differences included maximum LOD and R-2 values at QTL peaks and map positions, thus highlighting the importance of marker order for QTL mapping