Quantitative reconstruction of past climates using extension of modern climate-pollen relations based on the MAT method

Instrumental records span only a tiny fraction (<107) of the Earth’s climatic history. Therefore, indirect source of climate data must be used. Subfossil pollen and plant macrofossil data derived from sediment profiles can provide quantitative information on past climate. Recognizing the relationship between pollen assemblages and climatic variables and then interpreting fossil pollens base on relationships between them is required to quantitative reconstructions in different regions. Despite pollen-based quantitative reconstruction has begun several decades ago; studies on Iranchr('39')s past climate have been mainly qualitative and comparative. In this study we used a modern pollen dataset consisting of 119 modern surface samples along significant ecological and climatic gradients over the Zagros Mountains and west of Iran to develop transfer functions and quantitative reconstruction of climatic variables.  

In this study collected 59 samples, also 60 samples were collected by wright (1967) were used. Five subsamples of 15*15 cm2 within an area of 100 m2 were mixed into on sample at each site. Their geographic locations and altitude were determined using GPS. The altitude gradient of these samples is 150 to 4130 m ASL where mean annual temperature ranges from 25.94 to 2.1 C and mean annual precipitation ranges from 203 to 748 mm. Samples were prepared using combination standard techniques of Fægri and Iversen (1975) and Erdtman (1951). A Hund WETZLAR microscope at a magnification of 400 was used for pollens identification and counting. 300 pollen grains were counted for each sample. A total of 58 pollen taxa were recognized from the surface samples of the 119 study sites. Modern climatic data of 195 meteorological stations of western Iran were used for spatial interpolation. Eight environmental variables were chosen: altitude, mean annual precipitation, mean annual temperature, mean summer temperature, mean winter temperature, mean winter precipitation, mean summer precipitation, mean spring precipitation. Modern Analog Technique (MAT) method was used to reconstruct climatic variables in this study. The MAT is a calibration method for reconstructing a past environmental variable (e.g. temperature) from faunal assosications. It works by finding modern sites with faunal associations close to those in fossil samples. Environmental data from the modern sites are then used to estimate the environment fossil. In order to remove the effects of high collinearity among environmental variables in the process of analyses, we examined the Variance Inflation Factors (VIFs) for each environmental variable. If the VIF value of a variable was larger than 10, the variable was assumed to be almost perfectly correlated with the other variables and it captures little variance. A sediment core that used in this study, (63-J) of Zaribar lake with 25 meters long from the western shore of the lake, were taken by Wright in 1963 and pollinated and published. All radiocarbon dates obtained from Lake Zaribar sediments and published by authors show that this core covers 22500 years before present.  

Transfer functions, based on Modern analog technic (MAT), were developed for mean annual temperature (R2=0.6, RMSEP=3.74), mean annual precipitation (R2=0.79, RMSEP=82.74 mm) and mean precipitation of spring season (R2=0.62, RMSEP=21.1 mm). Our results confirm that pollens assemblages can provide reliable estimates of the climatic parameters. The results of quantitative reconstructions of the temperature of the last glaciation of the Earth showed that the air temperature in western Iran was on average 4 to 5 degrees lower than the current period. Also, the mean annual precipitation has been on average 20% less than the current period. At the beginning of the Holocene, the temperature rises significantly, so that the temperature was 1 to 2 degrees higher than the current period. In the Middle Holocene, the temperature dropped by about two degrees, and this situation with low fluctuations has remained almost constant until today. Prediction errors of pollen-climate transfer functions that were calculated in this study is MAPRMSEP=82.7 mm, ca. 15.1% of the MAP range; MATRMSEP=3.74 C, ca. 15.5% of MAT range. A number of factors can influence the prediction error of the calibration dataset such as topographical variation that affected temperature and precipitation variations. Moreover, human impact on vegetation may be another important factor that complicates many pollen-based climate reconstructions. It is important to note that the small number of samples also increased the amount of prediction error.  

For pollen base quantitative reconstruction in different regions, a reliable number of samples should be collected in different regions, especially drier or wetter regions, as well as colder or warmer regions. Among the climatic variables, the mean annual precipitation has a more significant relationship with the pollen assemblages of the samples and has a better ability for quantitative reconstruction than other variables. Researchs in the Middle and South Zagros shows that during the last glacial period, the vegetation type characterized by high values of chenopodium and Artemisia. This type of vegetation indicates the prevalence of cold and dry weather conditions in this area. The results of quantitative reconstruction of this study showed that in the last glacial period, the air temperature in this region was 4 to 5 degrees colder and the amount of precipitation was 20% less than the current period.