No abstract

Compositional changes of skin and seed phenolic compounds and berry glycosylated aroma precursors were measured in Vitis vinifera L. cv. Cabernet Sauvignon onto 1103P and SO4 rootstocks, in three irrigation regimes (FI, 100% of evapotranspiration; DI, 50% of evapotranspiration; and NI, non-irrigated). The study was conducted in a commercial vineyard of central Greece, in a factorial experiment during two growing seasons (2005-2006). Grape samples were obtained at commercial harvest. The deficit water supply decreased berry size but did not affect the skin/pulp weight ratio. Water limitation, especially pre-veraison, caused a substantial increase of skin anthocyanin concentration, and this effect was independent of water deficit-induced reductions in berry size and vine vigor. Among individual anthocyanins, malvidin-3-O-glucoside was mostly affected by water supply. The rootstock genotype did not affect berry growth parameters and skin polyphenol concentrations. The irrigation regime (mainly post-veraison) and rootstock genotype affected total flavan-3-ol monomers in seed tissue, mainly as a result of variations in the catechin amount. The lower seed phenolic concentration was found in non-irrigated and SO4-grafted vines, probably as a result of the restriction of scion vigor caused by these treatments, thereby altering cluster exposure. Skin and seed tannins were not affected by either rootstock or irrigation. The limited water supply was associated with increased aroma potential at harvest.

No abstract

Effects of nitrogen (M) and water supply on grapevine leaf transmittance and canopy reflectance were studied over 2 years (2009)(2010) in two vineyards planted with cvs. Cabernet Sauvignon and Xinomavro (Vitis vinifera L.) located in northern Greece. Three N (0, 60 and 120 kg ha -1 ) and two irrigation (irrigated at 70 % of crop evapotranspiration and non-irrigated) treatments were managed in triplicate in randomized blocks. Measurements with two transmittance-based chlorophyll meters (CM: SPAD-502 and CCM-200) and two active canopy sensors (Crop Circle ACS-210; amber band and Crop Circle ACS-430; red and red edge bands) were conducted on four and two growth stages, respectively. Fertilization increased leaf N and chlorophyll concentrations, as well as CM readings and vegetation indices. The CCM and SPAD values did not vary across growth stages and better estimated leaf N and chlorophyll concentrations in Cabernet Sauvignon than in Xinomavro. The SPAD readings provided better predictions of leaf N concentration. The vegetation indices of the ACS-430 sensor correlated with leaf N and chlorophyll concentrations and dormant wood in Cabernet Sauvignon; the red edge-based indices were generally more efficient than the red-based indices. Consistent relationships were observed for all ACS-430 indices with yield and for red edge-based indices with total phenols early in the season for both vineyards. The correlation between ACS-210 sensor indices and vine properties was low. These results showed that the two technologies examined could provide information on grapevine performance early in the season but have limitations that may restrict their use in vineyards because the acquired relationships were not consistent across vineyards and instruments.

We provide a unifying perspective for two decades of work on cost-sensitive Boosting algorithms. When analyzing the literature 1997-2016, we find 15 distinct costsensitive variants of the original algorithm; each of these has its own motivation and claims to superiority-so who should we believe? In this work we critique the Boosting literature using four theoretical frameworks: Bayesian decision theory, the functional gradient descent view, margin theory, and probabilistic modelling. Our finding is that only three algorithms are fully supported-and the probabilistic model view suggests that all require their outputs to be calibrated for best performance. Experiments on 18 datasets across 21 degrees of imbalance support the hypothesis-showing that once calibrated, they perform equivalently, and outperform all others. Our final recommendation-based on simplicity, flexibility and Editors: Thomas Gärtner, Mirco Nanni, Andrea Passerini, and Celine Robardet. Learn (2016) 104:359-384 performance-is to use the original Adaboost algorithm with a shifted decision threshold and calibrated probability estimates. Electronic supplementary material