Range shift of the species in response to global warming have been extensively studied by using CTI (Community Temperature Index). However, There are limitations. First, there are geological gaps. Many studies were conducted in South America or Europe, however, none have been done in Asia. Second, the validity of CTI (Community Temperature Index), which is widely used in this line of studies, have not been sufficiently confirmed. CTI represents temperature preference for a given community. Although many studies showed the increase in CTI and suggested the communities thermophilization, few have examined how well the CTI change can evaluate the rate of community thermophilization.
In this study, we analyzed the CTI changes in the 55 forest sites in Japan during 2004–2017, using tree survey data by the Monitoring Sites 1000 Project (the Ministry of the Environment). We calculated CTI based on species' temperature preferences estimated by the temperature of the all occurrence records in Japan, obtained from GBIF, for each species. Each pair of consecutive surveys (interval, hereafter) in a site were categorized into either CTI increased or decreased. We also examined the relationship between the average CTI through the period and the actual temperature in each site.
Results/Conclusions
CTI increased at 60 % of the intervals and the average rate of the CTI change was 0.0037 deg / year. These values were higher than 50 % (p < 0.01, binominal test) and zero (p < 0.001, one sample t-test), respectively, suggesting the community thermophilization of Japanese forests. However, the rate is lower than 0.019 deg/year (p< 0.001, one sample t-test), the rate of temperature increase in Japan since 1996.
We confirmed that CTIs at the cold and the warm edge of study sites were closer tothe median CTI of all sites than expected by actual temperature. Such bias could be generated by the used species distribution data that was limited to Japan. Because we did not use any distribution information out of Japan, extreme climates were cut off in our species' temperature preferences’ calculation. This could bias CTI of each site to the median CTI, leading to underestimated variance in CTI among sites and among intervals. We suggest that CTI does not perform well to estimate the rate of community thermophilization unless one use the complete species distribution data including the both norther and southern edges.