When a widespread frost severely damaged North Carolina’s peach crop in April 2007, James Rigby and Amilcare Porporato of the US state’s Duke University began thinking about how plants will deal with frost risk in a changing climate.

They realized that the risk of frost damaging flowers after buds have opened may not just be a function of the mean temperature. It might also depend on how widely the temperature varies, a feature statistically known as variance, and which is also expected to increase as a result of climate change.

"In the literature, many of the papers looking at the effects of climate change on plant phenology just take historical temperature records and shift up the mean temperature for the entire series," Rigby explains. "But natural ecosystems are adapted to the full temperature process, not just the mean temperature."

Rigby and Porporato took a simple model of how plants time the opening of their buds, which is largely in response to temperature, and used it to calculate when plants would open given a set of temperature data for Durham, North Carolina. They calculated a probability of frost damage faced by these model plants, based on the likelihood of a day of frost after budburst. Then they tried altering the temperature parameters, particularly the mean temperature and the variance.

The results showed that while increasing the mean temperature reduced the risk of frost damage, as expected, increasing the variance of temperature increased the risk. Specifically, an increase in temperature of 1°C increased a plant's chance of successfully fruiting by five per cent, but a 10% increase in temperature variance decreased the chance of good fruit by the same percentage. "If the basic degree-day models are correct," says Rigby "temperature variance has as much of an effect on plants as changes in the mean." If the temperature becomes more erratic, so that the temperature on any given day is less predictable from the temperature the previous day, this also increases the risk of frost damage to plants.

The finding opens up the possibility that climate change could affect species composition in natural ecosystems by increasing the risk of frost damage to plants even when temperatures are rising.

The result could also be significant for agriculture, in terms of assessing the financial risk from climate change. As well as spring frost damage, there are other risks to crops from crossing a threshold temperature, such as the danger to Californian grapes of excessive daytime temperatures in the summer. Could this risk also be increased by higher temperature variance?

Rigby is keen to verify his finding further by analysing some real data on plant flowering dates, coupled with temperature data, to see whether the variance is as important to plants as his model suggests.

The research is described in Geophysical Research Letters.