The within-year distribution of temperature and precipitation could rearrange significantly, challenging both modelers and nature

The distribution of species and other ecological phenomena (e.g. vegetation types) may be affected by climate change. This impact is commonly investigated and predicted by predictive distribution models. The key components of these models are the so-called bioclimatic variables. The expected distributions predicted by the models depend on the values of bioclimatic variables (e.g. the temperature of the wettest quarter of the year). Meanwhile, the time period on which the variable is calculated (e.g. the wettest quarter) may shift within the year. This shift can easily be hidden, even though the ecological meaning of bioclimatic variables is highly dependent on the time period. For example, the wettest quarter may have been in May-June-July recently (this is true for a large part of Hungary), but this may shift to autumn or winter in the coming decades. It is not hard to see that this poses difficulties for distribution modeling. While in the recent past the bioclimatic variable describing the temperature of the wettest quarter characterized, in fact, the early summer period, in the future the same variable will describe the temperature of a completely different (in our example, much cooler) period of the year. However, to train the models that predict the future distribution (in our case, using autumn-winter temperatures), we have used the recent early summer temperatures.

Two researchers at the HUN-REN Centre for Ecological Research, Ákos Bede-Fazekas and Imelda Somodi, have already shown in a previous study that the so-called specific climate periods, such as the wettest quarter, used to calculate bioclimatic variables can not only theoretically shift by several months within the year, but that this can actually happen in the future in parts of Hungary according to climate models. The shift of the specific climatic periods reduces the reliability of the distribution models, so the modeler needs to recognize the problem and address it.

“Will Hungary be the only country in the future to be in a situation such unfortunate – from a modeling point of view? Or is the problem affecting the whole world, and perhaps some regions in particular? How far do the different global climate models agree on this issue? And the scenarios behind the climate models?,” lists Ákos Bede-Fazekas the research questions that have kept the two researchers busy.

“The questions we wanted to ask were given, as were the necessary climate data. We also knew that the results, whatever they would be, would not only be of interest to us, but would also provide important information to the large community of distribution modelers. The only thing left to do was to somehow synthesize the vast amount of data and the results that could be extracted from it, and present it to the scientific community in a form that would be accessible. I think that was the biggest challenge for us.”

Flowchart illustrating the main steps of the research, from input data to the calculation of specific climate periods and synthesizing analyses

In the end, the researchers succeeded, and their analysis of four climate models, four scenarios and four future time periods covering the whole Earth was published in the prestigious scientific journal Global Change Biology. The study highlights the areas most affected by the shift of specific climate periods.

In the map of intra-annual variability of precipitation (top) and temperature (bottom) blue polygons represent the areas that will be most exposed to shifts in the specific climate periods associated with precipitation and temperature

In addition to the map results, the two researchers from the HUN-REN Centre for Ecological Research also revealed which of the three important decisions that a distribution modeler must make when predicting the future distribution are the most and the least important ones. These decisions are the choice of the global climate model, the choice of the scenario, and the choice of the future period.

“We found that the modeler’s choice of the global climate model was the least important, while the choice of the future period was typically more important than the choice of the scenario,” reports Ákos Bede-Fazekas on the results. “The shift in specific climate periods becomes more pronounced over time and as more pessimistic scenarios are considered. However, global climate models could not be ranked in a clear order in this respect. From a modeling point of view, I find the result somewhat reassuring, as it is the choice of climate models that tends to cause the most difficulty for climate modelers, but this choice seems to be the least important.”

Unfortunately, from the point of view of ecology and the diversity of natural communities, the result is not nearly as reassuring. The researchers have found numerous examples of specific climate periods shifting by more than two months, and have also reported an expected shift of six months – the largest possible. Such a major shift in the within-year distribution of climatic features is something that is feared that many species will not be able to follow. To continue with our example, this means that plant and animal species that have adapted to the precipitation falling in the pleasant early summer heat over thousands of years, could face a major challenge if most of the precipitation falls in the autumn-winter months, when their life cycle makes it difficult for them to use this precipitation. This could lead to the migration or, in the worst case, the extinction of species.

“In the tropics, shifts in both temperature and precipitation related specific climate periods are expected in many areas. However, shifts related to the precipitation are also expected in many areas of the temperate and arctic zones,” summarizes Imelda Somodi the spatial analysis. “The combined shifts around the equator confirm the likelihood that a climate not known from elsewhere (non-analogue climate) will develop there in the future. Consequences of the development of such non-analogue climates are most difficult to grasp.”

In the conclusion of their study, the researchers from the HUN-REN Centre for Ecological Research point out that future predictive distribution models will need to take into account the shift of specific climate periods and incorporate this phenomenon into the modeling work if they are to provide reliable predictions.

Source: Precipitation and temperature timings underlying bioclimatic variables rearrange under climate change globally