How much do vegetation growth form, observer bias and data standardisation method affect repeatability of canopy cover estimates and power to detect change? (#17)