highly significantp < 0.0005
When we included latitude as a predictor in the model it explained additional variance to the temperature effect and both temperature, temperature 2 and latitude were highly significant independent predictors (e.g. for the January temperature effect the explained variance by temperature and temperature 2 was 16.8% (F 2,2648 = 268.0, p < 0.0005) and when latitude was included the explained variation increased to 20.1% (F 3,2647 = 221.98, p < 0.0005).