Winter chilling is a key environmental trigger of floral induction in temperate Eucalyptus. Over the past two decades, considerable site × eucalypt flowering interaction research has been undertaken over a range of high elevation (>1 100 m asl) forestry sites in the South African summer rainfall area. A practical method of accurately monitoring eucalypt bud temperature at remote sites in these areas was needed for this research. Utilisation of traditional methods of air and bud temperature measurement were not viable, due to significant risks of data and meteorological equipment loss posed by severe weather, vandalism and theft. Between 1996 and 2004, a robust structure for housing the miniature Hobo® temperature logger (hereafter termed ‘Hobo pole’) was designed and utilised for in-field monitoring of air temperature in the research trials. During 2009 and 2010, an experiment was conducted to investigate the relationship between E. nitens bud temperature (BudT), Hobo pole air temperature (HoboAT) and radiation screen air temperature (ScrnAT), and develop appropriate calibration models. Attempts to develop a single model for predicting hourly BudT from hourly HoboAT over the entire annual period yielded unsatisfactory results (maximum R 2 value 0.49). Separate winter and summer regression models were subsequently developed for predicting BudT from HoboAT, ScrnAT from HoboAT, and BudT from ScrnAT. In these regressions, R 2 values were generally slightly higher, and SE values lower, for mid-winter data than for mid-summer data. In mid-winter, BudT on HoboAT gave the highest R 2 value (0.99) and lowest SE value (0.49 °C) of all regressions. The Hobo logger/Hobo pole combination, together with developed regression models (presented in this paper), offers one practical, cost-effective solution for accurately monitoring eucalypt bud temperature at remote, high-elevation forestry sites in South Africa.
Southern Forests 78:105-113