Tree-ring-based summer temperature minimum reconstruction for Taboshar , Sogd Province , Tajikistan , since AD 1840 : linkages to the Oceans

Despite the ecological importance of forest tree species, our study was motivated by scarce data on dendrochronology studies limited to the northwestern part of Tajikistan due to lack of access to such a part of the country which is difficult to access. Current studies on Junipers in the northwestern part of Tajikistan identified that tree ring-based climate reconstruction is possible. Cores of Juniperus Turkestanica from Taboshar area, Sogd province were developed into 176year tree-ring width chronology. In the current study a summer (June-September) minimum temperature reconstruction spanning AD 1840–2016 was developed, which can explain 30.0% of the instrumental variableness for the period of 1940–2015. Five warmest (1856-1866, 18691871, 1907-1916, 1923-1935, 1988-2016) and four coldest time periods (1840-1855, 18721906, 1917-1922, 1940-1984) were identified. The developed reconstruction coincides with cold and warm periods with prior investigations from tree-rings in China, Mongolia, Kyrgyzstan, Kazakhstan and Tajikistan. Spatial correlation analysis with the gridded minimum temperature data shows that the temperature reconstruction covers geographical representation over High Asia. Significant correlation was found between temperature reconstruction with summer SST, which suggests strong linkages of regional temperature variability with IndianOcean climate system. The obtained spectral peaks from the spectral analysis of the summer temperature reconstruction were significantly at 84.7-year (90%), 2.9-year (99%), 2.6-year (99%), 2.5-year (99%), 2.4-year (99%) and 2.2-year (99%).


Introduction
The climatic information in tree-ring chronologies could give a view of the climate condition in the past and thus with the integration of the present climate condition information could help to predict the climate for the future [1].The population history of many tree species in Europe and North America compared to other continents such as Central Asia is relatively well studied [2,3].Focusing on long-term climate changes in the world [4,5], Central Asia [6] and Tajikistan with suggestions for a rise in a temperature based on a meteorological dataset were presented [7].As it is mentioned [8], tree-rings are playing an important role in reconstructing past climates in the world.
The Tian Shan mountain system located in Central Asia and extends across five countries including Tajikistan [9].In Tajikistan, there are a wide variety of climatic conditions associated with high-altitude zone, geography, orography, that is of great interest from the point of view of the study and modeling of climate change at the local and regional scale [10].Tajikistan is highly vulnerable to global climate change.This is a fact that has been confirmed in recent donor reports [11].The Qurama range of the Tian Shan mountains situating within Taboshar area is one of the driest regions in the Tian Shan.This range is also rich of Junipers forests [12][13][14].Species of this plant family such as Juniperus sibirica, J. turkestanica, J. seravschanica, J. semiglobosa, and J. schugnanica are endemic to Tajikistan [15].There are numerous studies on reconstructions temperature, precipitation, drought related by Juniper tree-ring width [16][17][18][19][20][21][22][23][24][25][26][27], but in Tajikistan, a few studies are conducted [28][29][30][31] and the areas in the Qurama range of northern Tajik Tian Shan mountains are extensively missing.
To fill this gap in the knowledge, we developed tree-ring width chronologies using cores of J. turkestanica from Taboshar, Sogd Province, northern Tajikistan.This is believed that the current study is going to be the first tree-ring based 1840-2016 reconstruction in the targeted study area.Moreover, the study found out cool and warm periods for north Sogd over the 176 years.

Physical Settings
The mountains comprising Tajikistan belong to the giant mountain systems of Central Asia -Tian Shan and Pamir-Alay.The Tian Shan mountain system just partly covers the country including the Qurama Range (Tian Shan orogen) and cross Uzbekistan, Kyrgyzstan.Its length is about 170 km from North-West to South-East.The highest point is the Boboiob (In translation means "the father of waters or the grandfather of waters".From the Taj.Bobo -Grandfather and Tajik Ob -Water) which rises up to 3,769 m above sea level [32].In this part of Tian Shan Mountains middle relief prevails.It is composed of metamorphic schists, sandstone and granitic rocks.At an altitude of 1800 m above sea level, there are coniferous-deciduous forests, Juniper forests, on the northern slopes of the walnut, alpine meadows and fescue steppes, and down the slopes there are steppes and xerophilous bushes [33].
Nonferrous metal in the Qurama Range Ridge is available for the big amount metal mines resources in this area that were founded in the early 1950's.The area under investigation lies in a mountainous valley with substantial varieties in both the occasional and day by day air temperature because of a mainland atmosphere with sweltering and dry summers and short, icy winters and described as parched mainland [34].
The targeted study area (TTT 40°37' N/69°45' E) is situated northeast of the town of Taboshar at an elevation rising from 1485 to 1585 m above sea level (als).(Figure 1).It is located in the northern part of Gafurov district in Sogd Province, Tajikistan with temperate climate zone [35][36][37].Since in Taboshar the uranium mining industry was set up by USSR in the late 1940s and early 1950s as a part of the nuclear weapon program, the research area under investigation is probably impacted by remnants of nuclear weapons production.[38][39][40][41].It can be seen that rainfall in this area decreases with rising temperature and is almost absent in the summer.Mean annual precipitation calculated during 1890-2016 is 165 mm in Khujand, with 73 mm in spring March-May period (Figure 2a).Highest temperature period is June-July with mean long-term calculations during 1932-2016 year is 35.1 ºC with highest mean air maximum temperature existent in July at 36.0ºC and coldest month is January with a mean air minimum temperature --3.1ºC (Figure 2b). the mean annual temperatures minimum (green line), maximum (pink line), mean (red line) and the total annual precipitation (blue line) for their common periods 1932-2016 (Table 1).
Juniperus turkestanica Komarov, var J. pseudosabina Fischer and Meyer C. A. (1842) as [42] report could be found throughout of Xinjiang, Mongolia, Kazakhstan, Kyrgyzstan, Tajikistan, Pakistan and Afganistan in various proportion.This range would include the Tian Shan and Altai, Pamir and Tarbagatay Mountains [43].In addition, photographer Vladimir Dinets adds: "Var.J. turkestanica and J. semiglobosa form extensive woodlands or even forests in some areas, from Transcaucasia to Xinjiang.They are logged to extinction for firewood in many areas, but under strict protection in others".Based upon his report, J. turkestanica can survive at a lower elevation and on drier sites than the other less adapted plants.As the author [44] emphasizes Junipers stands grow on the Qurama Range from 1300 m above sea level (als).
Endemics occur at the ecotone zone at elevation of 1400-4000 m above sea level (als), mainly in Juniper forests [45].Junipers occupy about 150 thousand ha.Juniper forests are good regulators of surface runoff preventing soils from erosion in mountains and valleys, as well as CO2 sinks.Junipers can be as old as more than five centuries [10].

Sampling and data from ring growth
In October 2016, during conducted field investigation, 40 Juniper trees in the open stands in general were chosen for sampling (Figure 1b) and two cores were extracted from cross-slope sides of each tree.In general, at chest height but from some trees whose crown growth was too low to the ground it had to be taken at the height of 0.2-0.5 m using the 5-mm diameter increment borers.Minimizing non-climatic effects on radial growth, only trees with no harm or malady were extracted.The height of the sampled trees ranged from 7 to 12 m depending on site conditions and their diameter ranged from 92 to 320 cm.Since we found false rings connected with dating errors, missing rings and some wood anomalies we kept out the collected samples for further analysis from about 20 trees.However, the amount of sampling trees was 40.
Further, the collected samples were air dried, mounted on a wooden base and the cores were cut and polished with sandpaper (400-800 grit cm -2 ) for the purpose of examination in laboratory.Before the cross-dating procedure, visual measurement of the samples was done and the tree-ring widths were pre-marked.This technique is proved to be very helpful for quickly finding the right ring on the sample and verification of correctness of recorded and measured rings [46].The cores were cross-dated using standard dendrochronology methods [47].Width measurements of rings were performed using LINTAB VI system with 0.01 mm accuracy.The individual sequences positions, correlation and decision were done with use of the TSAP-Win [48] and COFECHA software [49].
For detrending, indexing and standard chronology, the series produced using the ARSTAN software were chosen for further analysis.The exponential or linear regression curve was used to standardized ring-width [50].A bi-weighted robust estimation was applied to remove the effects of endogenous stand disturbances.The reliability of chronology period was estimated in the 50-year window with a 25-year lag by threshold of expressed population signal (EPS≥0.85).[51].

Meteorological data and climate response analysis
To measure the effectiveness and guidance of linear relationships, we run Pearson Correlation produces a sample correlation coefficient between the climatic data and the ring chronology, for this, we used the SPSS program.Monthly mean/maximum/minimum temperature, total precipitation data were computed from daily records for the stations which vicinity of the sampling site and were obtained from the Hydrometeorological climate records station.We also used sources such as [52,53], to complete breaks in the meteorological data (Table 1).To find a good correlation between tree-ring width chronology and climate data we use grid temperature dataset, included precipitation, temperature minimum/maximum/mean covering the period of 1940-2015 at grid point (38º75-41º75' N / 66º75'-72º25' E) and interpolated to our sampling site (Figure 1).Received from the Climatic Research Unit (CRU TS 4.00), East England, UK database, [54,55], which was calculated by using an automated interface program developed in MATLAB [56].Such kind of testing statistics were calculated to demonstrate the stability of our reconstruction, like Reduction of error (RE), coefficient of efficiency (CE) statistics, the Sign Test, the Durbin-Watson value and Pearson's correlation coefficient [47].In order to show the geographical representation of the reconstructed summer temperature, a spatial correlations performance between reconstruction of temperature and the grid-based data set CRU TS 4.0 [57] for the period from 1940 to 2015 was done using the KNMI climate researcher web-page [58].
The spectral multitaper method was applied for analysis of reconstruction signal to detect characteristics of local climate variability in the frequency domain [59].The analysis of reconstruction over the period of AD 1840-2016 was executed.Our analysis used 5×3π in a red noise background which provides more strong and significant signals.

Tree-ring width chronologies and climatic signals
The mean sensitivity (0.495) show high correlation by 33.3% of variance, with significant standard deviation (0.335) that reflect good responses to the variation of climate [47].Mean correlation with master series is 0.507 (Table 2).Reliable ring-width chronology spanning 1840-2016 was developed based on an EPS value ≥0.85 (Figure 3).The reason of why conditions may have very less effect a ring to be narrow (or wide) in one year on the growth of the following year we can understand from the value of first autocorrelation, which is negative value (-0.106).In general, our results on the variation of growth ring through time showed the same patterns as [60] noted: narrow one followed where broad with the width of growth rings showing periodicity; in the young ages width emerged narrow and became wider as trees grew older.

Climate response analysis
Figure 4 describes the significant coefficients for correlation functions.Coefficient correlation between standard chronology and climate reveal significant p<0.001 in the gridded temperature minimum CRU grid from the prior June-November period as well as from current January-September period.With precipitation we find correlation (p<0.05) in the prior November-December period (Khujand station) or prior December and this February (CRU grid).
Correlation positive between tree-ring and temperature maximum in the periods: prior August-  The comparison of the original-summer reconstructed minimum temperature has been shown in Figure 5a.The leave-one-out cross-validation statistics was also calculated and the results are shown in Table 3.   18ºC) shows that the second period is increased in 0.11ºC than the first period (Table 4).After 11-year simple smoothing, the temperature reconstruction showed four coldest periods (1840-1855, 1872-1906, 1917-1922, 1940-1984) and the coldest decades (1850s, 1860s, 1890s, 1900s, 1910s, 1950s, 1960s, 1970s and 1980s) with temperatures lower than the mean.

Tree-ring growth and climate
We believe that a strong correlation between the tree-ring and the summer climate is probably related to the moisture limitation occurring in the middle of the current growth season.
Although October-April is the most plentiful season of rain, we did not find any strong and significant correlations.The recorded precipitation value of Taboshar area is low and varying extremely.It was concluded [61] that in warm regions with low altitude and a large interannual change in precipitation leads to decrease in both temperature and soil moisture leading to a sharp growth of the tree during the heavy rainy season.Give such an explanation to the phenomenon, the climate and the field of study, which are close to ours, could not find its confirmation in the current study.This is possible, because we have not so thoroughly studied the cell and the development of cambium for individual periods of time.
From the results of this study, it can be concluded that optimal growth at the Taboshar site depends on cool summer months (i.e.July (Figure 2a) which are the hottest months in the year.
From the summer months, the most reiteration can be observed in July rather than June, August and September (Figure 4)) and even with the minimum temperature, the correlation with this month is lower.Based on our correlation, estimates is scarce with precipitation in both points of meteorological data.Correlation significant positive just in prior November-December (Khujand station) and prior December and this February (CRU grid) which determines drought area in this part of Tajikistan.We believe that the correlation is high in the winter due to precipitation in the form of snow, thereby retaining soil moisture, when a large amount of moisture is needed for photosynthesis at an early stages of the growing season as declare [62,63] in the Karakorum, Pakistan and in the Western Tian Shan, China respectively.As it was mentioned earlier, in the Kongtong Mountains temperature takes significant role on drought stress cause it is demonstrated by higher correlation coefficients with temperature than precipitation [64].
Deficiency of water in the growing season inhibits the expansion of tracheids.Moreover, with the increasing temperature, evaporation increases in May-September due to low precipitation, which rev up water stress (Figure 2a).When the tracheids become narrower, the dimension of the cell wall in the yearly ring growths due to a reduction in the size of the lumen [65].Our observations coincide with earlier studies that can explain the high value of early density in narrow rings (dry years) [61].In fact, this connection reflects short and acute interactions (high frequency signals) between temperature and monsoon deposition.These determinations suggest that persistence of climate variations in radial growth has increased during the latter half of the 20th century due to global warming.

The comparison of regional records
Spatial correlations performed between the reconstruction and temperature minimum dataset of CRU TS 4.0 [57] reveal our record's geographical representation (Figure 7a).
Significant positive correlations are found north of approximately 30º-50º latitude with a 50º -90º east-west extension.The highest positive correlation suggested that the study area is closely linked with synoptic processes of south Asian monsoon in the High Asia.

Linkages to the Ocean climate variability
Spatial correlation between our temperature reconstruction and the sea surface temperature (SST) gridded dataset of HadlSST1 [77] for the period 1940-2015 was conducted.The significant positive correlations of the summer temperature reconstructed related to the SST and show close connections between the temperature in the western Tian Shan Mountains and the remote ocean area.As it is shown in Figure 7b, the most significant area of correlation was found in the Indian Ocean [63,78], suggesting that SST of this ocean is an important factor affecting the growth of J. Turkestanica in Taboshar.Indian Ocean is the moisture source to our study area which fall into the Indian Monsoon domain as reported by a study [79].Cycles peak fall at 2-3-years in the scope of variability of the El Niño-Southern Oscillation (ENSO) [80,81].These low frequency cycles and multidecadal cycle 84.7-yearfall inside the overall bandwidth of the winter North Atlantic Oscillation (NAO) [82] which may suggest influencing NAO on the Climate of Central Asia.However, the mechanism of influence of these circulation systems and how they control the temperature variability of Tajikistan at different times, awaits further study.

Conclusion
This study has presented the summer minimum temperature reconstructed (June-September) from J. Turkestanica chronologies.We believe that this is the first proxy record that indicate the possibility of reconstruction over 176 years and explanation the temperature over 19th -20th century.The sensibility of latest tree-growth to temperature minimum in the Taboshar area was significantly diminished under the climate warming.This record can show different ways of look onto temperature variations, despite mountain terrain and spatial temperature differences.Geographical representation of the conducted reconstruction have been confirmed by spatial correlations.A comparison with various temperature-sensitive ring series in Central and East Asia show a high correlation in the timing of warm and cold periods, i.e. warm periods (1856-1866, 1869-1871, 1907-1916, 1923-1935, 1988-2016) and cold periods (1840-1855, 1872-1906, 1917-1922, 1940-1984).The correlation between our summer temperature reconstruction with summer SST suggest strong linkages of regional temperature variability with Indian-Ocean climate system.This is the first work in the studied area and we hope to continue the scientific investigations in this area, that would help us to better understand the change in the growth of juniper trees to conditions of World Global Warming and the past climate variability in Asia.

Figure 1 .
Figure 1.Location map of the sampled sites, meteorological station, CRU and city.

Figure 2 .
Figure 2. Climate diagrams of the Khujand station: a) total monthly precipitation (filled bar), temperatures maximum (red dash line), minimum (blue dotted line) and mean (orange line); b)

Figure 3 .
Figure 3. Plot of the standard chronology with 11-year simple smoothing, its running EPS, sample depth and Rbar.

Figure 4 .
Figure 4. Correlation coefficients of tree-ring width chronologies of Juniperus Turkestanica.The horizontal lines indicate the p<0.01 and p<0.05 significance level.

Figure 5 .
Figure 5. Comparisons of summer temperature reconstructed (June-September) a) with instrumental data for common period 1940-2015; b) 11-year simple smoothed (thick line) over the period 1840-2016.Central horizontal line show the mean reconstructed values; inner round point horizontal lines show the border of ±1SD.

Figure 5b shows the original and 11 -
Figure 5b shows the original and 11-year simple smoothing summer (June-September) CRU minimum temperature since AD 1840.The mean of summer (June-September) CRU minimum temperature over the time period of 1840-2016 is 15.11ºC.The most extremes years in the summer reconstruction were 1895 (14.35ºC) and 2006 (16.11ºC).As we know in recent times, increase in temperature can be observed, that leads to call it a global warming and our reconstruction confirms it.The values inside horizontal lines (±1 SD) indicate cool and warm

Figure 6 .
Figure 6.Results of MTM spectral analysis of the summer temperature reconstruction.The dash-dotted dashed and dotted lines indicate the 90 and 95% significance levels, respectively.

Table 2 .
Summary of statistics for tree-ring width chronologies of J. turkestanica.
*AGR average growth rate; MC mean correlation with master series; MSL mean segment length; R is the all-series Rbar; MS mean sensitivity; SD standard deviation; SNR signal-to-noise ratio; First order autocorrelation; EPS expressed population signal; PC#1 is the variance explained by the first principal component.

Table 3 .
Leave-one-out cross-validation statistics for reconstruction of summer (June-September) CRU minimum temperature reconstruction in the Taboshar