Muhammad Usama Zafar

WATER ANALYSIS AND CLIMATIC HISTORY OF GILGIT AND HUNZA VALLEYS (A DENDROCLIMATIC APPROACH)

A THESIS SUBMITTED IN PARTIAL FULFILLMENT OF THE REQUIREMENTS FOR THE DEGREE OF DOCTOR OF PHILOSOPHY 2013

Department of Environmental Science Federal Urdu University of Arts, Science and Technology Gulshan-e- Iqbal Campus, Karachi, Pakistan

CERTIFICATE

Certified that the candidate has completed the thesis under my supervision

Prof. Dr. Moinuddin Ahmed (Foreign Professor)

Laboratory of Dendrochronolgy and Plant Ecology of Pakistan, Department of Environmental Science Federal Urdu University of Arts, Science and Technology, Gulshan-e-Iqbal Karachi

Research title: Water analysis and climatic history of Gilgit and Hunza Valleys (A Dendroclimatic Approach)

Submitted by Muhammad Usama Zafar

M. Phil/PhD Scholar

Research supervisor Prof. Dr. Moinuddin Ahmed Foreign Professor Laboratory of Dendrochronology and plant ecology of Pakistan Department of Botany FUUAST, Karachi

Co-supervisor Dr. Muhammad Afzal Farooq Chairman Department of Environmental Science FUUAST, Karachi

Graduate Research Management Council

Department of Environmental Science Federal Urdu University of Arts, Science and Technology Gulshan-e-Iqbal Campus Karachi

iii

i-Note

In Pakistan, family of dendrochronology is very small and still it is emerging as a new science. It has been introduced in three or four universities throughout the country but not at professional level. Therefore, I also included water analysis of the same area from where I worked over dendroclimatology. Although there is no correlation between water analysis and dendroclimatology but I analyzed both. It was very difficult for me to study the two different fields and to make comprehensive results. My basic concern was towards dendroclimatology and for seeking jobs in Pakistan; I included water analysis as a separate part of my thesis.

ii- Abstract

Twenty-nine samples from different locations of Upper Indus Basin from Gilgit and Hunza valleys were selected for the investigation of physico-chemical characteristics. Sampling was performed during July in the year 2012. Eleven parameters were chosen for water analysis to assess water quality and to observe the variations among different sites. Physical factors were analyzed at site while chemical analysis was carried out in the laboratory using standard techniques of water analysis developed by (American Public Health Association, APHA) and spectrophotometeric techniques. Multivariate statistical techniques including principal component analysis (PCA) and cluster analysis (CA) were employed to interpret the data and to unravel the causes of water pollution. Results of physico-chemical properties showed that values of all parameters were in accordance with the permissible limits proposed by World Health Organization (WHO) but the high values of total alkalinity shows that water is of bicarbonate type.

Knowledge of past climate variability is necessary for understanding present and future climate tendencies. This study used three species (Picea smithiana, Juniperu sexcelsa and Pinus gerardiana) ring-width chronologies to investigate palaeo-temperature history in Gilgit and Hunza valleys Northern Pakistan. The resultant reconstruction is among the first palaeo-series from Picea smithina produced for Pakistan to date. It is in good agreement with other tree-ring based records, and with instrumental (both local and grid) data. Ten pine chronologies including three species were developed. Ring-width measurements were detrended using the standardization method to preserve as much climatic signals as possible. Crossdating exposed the presence of a strong common signal among trees. Inter-site comparison showed that a common control mechanism affected tree growth not only within sites, but also across sites. To determine whether climate was the main factor that controlled the growth of three species from Gilgit and Hunza, correlation and response functions were analyzed. Temperature and precipitation were tested for their relationship with tree growth. Mean monthly temperature and total monthly precipitation were observed as the primary growth-limiting factor. Chronologies were negatively correlated with temperature and precipitation of spring season, and climate correlation modeling showed that temperature and precipitation explained 39-63% variance in the tree-ring data. Tree-

v ring data from Picea smithiana Jutial contained the strong temperature signal, was picked for reconstruction. The Jutial chronology was then used to reconstruct March-June temperatures back to A.D. 1523. The calibration model explained 38.16% of the variance in temperature, and all calibration and verification tests were passed at good levels of significance. The reconstructed temperature was tested over decadal and century time-scale. The coolest decadal time scale period revealed that 17th century experienced lowest degree of temperature and ensuing the period of “Little Ice Age” (LIA). The temperatures reached their maximum in 19th century over century time-scale. As Pinus gerardiana Chaprot chronology exhibited strongest temperature signal among all chronologies therefore, separate exercise was performed where Jutial chronology reconstruction was compared with Chaprot reconstruction. Two species demonstrated the common pattern in spring temperatures. However, the temperature reconstruction from Chaprot was insufficient to produce a long term proxy temperature. This research has strengthened the Pakistan network of chronology sites, and confirmed that Picea smthiana, Juniperus excelsa and Pinus gerardiana have great dendro-climatic value. The last more than 450 years of temperature fluctuations were reconstructed with a high degree of fidelity. The current reconstruction added similar trend of temperature in comparison with the other studies throughout central Asia.

iii-DEDICATION

I would like to dedicate my thesis to my beloved mother

vii iv-Table of Contents

Note...... i

Abstract ...... ii

Dedication ...... iii

Table of Contents ...... viii

List of Tables ...... xi

List of Figures ...... xiii

List of Symbols, Abbreviations or Other (Optional) ...... xiv

Acknowledgements ...... ixx

Part One: Water Analysis of Gilgit and Hunza Valleys

Chapter 1 Water analysis ...... 4 1.1 Introduction ...... 4 1.2 Research objectives ...... 5 1.3 Review of literature ...... 8 1.4 Materials and methods ...... 10 1.4.1 Sampling and on-site evaluation ...... 10 1.4.2 Methods for the detection of chemical parameters ...... 12 1.4.2.1 Chloride ...... 12 1.4.2.2 Carbonate alkalinity ...... 12 1.4.2.3 Bicarbonate alkalinity ...... 12 1.4.2.4 Total hardness ...... 13 1.4.3 Statistical analysis ...... 13

Chapter 2 Results ...... 14 2.1 Temperature ...... 14 2.2 pH ...... 15 2.3 Dissolved oxygen ...... 16 2.4 Total dissolved solids ...... 17 2.5 Electrical conductivity ...... 18 2.6 Salinity ...... 19

viii

2.7 Chloride ...... 20 2.8 Total alkalinity ...... 21 2.9 Total hardness ...... 22 2.10 Sulphate ...... 23 2.11 Nitrate ...... 24 2.12 Pearson correlation matrix of all parameters ...... 29 2.13 Discussion ...... 34

Part two: Climatic history of Gilgit and Hunza Valleys (A Dendroclimatic Approach)

Chapter 3 General introduction ...... 37 3.1 Introduction to dendrochronology ...... 37 3.1.1 Brief history of dendrochronology ...... 37 3.2 Climate of Pakistan ...... 38 3.3 About the study sites ...... 39 3.3.1 Gilgit ...... 39 3.3.2 Hunza ...... 40 3.4 Purpose of study ...... 43 3.5 Review of Literature ...... 45 3.5.1 Dendrochronology in Pakistan ...... 45 3.5.2 Dendrochronology in China ...... 46 3.5.3 Dendrochronology in Nepal ...... 47 3.5.4 Dendrochronology in India ...... 47

Chapter 4 Chronology development ...... 49 4.1 Introduction ...... 49 4.2 Materials and methods ...... 49 4.3 Field methods ...... 49 4.4 Laboratory preparation ...... 53 4.1.1 Surfacing and crossdating ...... 53 4.1.2 Measurement using Velmex ...... 53 4.5 Software's used in the analysis ...... 53 4.5.1 COFECHA ...... 54 4.5.2 Chronology development ...... 55 4.5.3 ARSTAN ...... 55 4.5.4 Chronology Statistics ...... 56

4.6 Results ...... 58 4.6.1 Crossdating of all sites ...... 58 4.7 Chronology development ...... 64 4.7.1 EPS and Rbar ...... 76 4.7.2 Autocorrelation and partial autocorrelation ...... 82 4.8 Chronology comparison ...... 91 4.9 Multivariate analysis...... 94 4.10 Discussion ...... 96 4.11 Conclusion ...... 99

Chapter 5 Growth-climate response ...... 100 5.1 Materials and methods ...... 101 5.2 Climate data ...... 101 5.2.1 Temperature ...... 102 5.2.2 Precipitation ...... 103 5.3 Results ...... 105 5.4 Correlation among tree ring chronologies and temperature ...... 105 5.5 Correlation among tree ring chronolgies and precipitation ...... 106 5.6 Discussion ...... 119 5.7 Conclusion ...... 122

Chapter 6 Temperature reconstruction ...... 123 6.1 Introduction ...... 123 6.2 Material and methods ...... 123 6.3 Results ...... 125 6.4 Comparison with Pinus gerardiana reconstruction ...... 133 6.5 Discussion ...... 135 6.6 Conclusion ...... 138

Bibliography or References ...... 140

x vi- List of Tables

Table 1.1 Nearest town, elevation and map location of Gilgit and Hunza valleys...... 7 Table 1.2 Analysis parameters and their analytical procedures ...... 11 Table 2.1 Values of all sampling sites with eleven parameters from Gilgit and Hunza ...... 26 Table 2.1 Correlation matrix among all parameters ...... 28 Table 2.2 Characteristics of three groups derived from Ward's clustering of the water quality variables of the samples collected from 29 locations ...... 30 Table 3.1 Ecological characteristics of forest from sampling sites ...... 42 Table 4.1 Summary statistics of species from eleven sites collected from COFECHA ...... 59 Table 4.2 Negative (narrow) pointer years from ten sites of Gilgit and Hunza valleys ...... 62 Table 4.3 Positive (wide) pointer years from ten sites of Gilgit and Hunza valleys ...... 63 Table 4.4 Summary of COFECHA statistics ...... 88 Table 4.5 Summary of Arstan statistics ...... 89 Table 4.6 Correlation matrix of all chronologies values from ten sites ...... 91 Table 5.1 Summary of correlation function between tree ring chronologies and monthly temperature and precipitation data from Gilgit station ...... 114 Table 5.2 Summary of correlation function between tree ring chronologies and monthly temperature and precipitation data from the relevant 0.5o grid climate database (Mitchell and Jones, 2005) ...... 115 Table 5.3 Summary of response function between tree ring chronologies and monthly temperature and precipitation data from Gilgit station ...... 116 Table 5.4 Summary of response function between tree ring chronologies and monthly temperature and precipitation data from the relevant 0.5o grid climate database (Mitchell and Jones, 2005) ...... 117 Table 5.5 Summary of four tables (5.1-5.4) including only significant signs of postive and negative corrrlation and response analysis ...... 118 Table 6.1 Regression analysis of ten chronologies with Gilgit temperature data from different sites of Gilgit and Hunza valleys ...... 126 Table 6.2a Early calibration ...... 127 Table 6.2b Late calibration ...... 127

Table 6.3 Statistics for March-June actual and reconstructed (1955-2008) temperature data ...... …129 Table 6.4a Warm periods ...... 130 Table 6.4b Cold periods ...... 130

xii vii- List of Figures Figure 1.1 Map representing twenty nine sampling sites from Gilgit and Hunza rivers ...... 6 Figure 2.1 Graph shows the box and whisker plot of temperature from all sites ...... 14 Figure 2.2 Graph shows the box and whisker plot of pH from all sites ...... 15 Figure 2.3 Graph shows the box and whisker plot of dissolved oxygen from all sites ...... 16 Figure 2.4 Graph shows the box and whisker plot of total dissiolved solids from all sites .. 17 Figure 2.5 Graph shows the box and whisker plot of electrical conductivity from all sites . 18 Figure 2.6 Graph shows the box and whisker plot of salanity from all sites ...... 19 Figure 2.7 Graph shows the box and whisker plot of chloride from all sites ...... 20 Figure 2.8 Graph shows the box and whisker plot of chloride from all sites ...... 21 Figure 2.9 Graph shows the box and whisker plot of total hardness from all sites ...... 22 Figure 2.10 Graph shows the box and whisker plot of chloride from all sites ...... 23 Figure 2.11 Graph shows the box and whisker plot of nitrate from all sites ...... 24 Figure 2.12 Dendrogram resulting from Ward's cluster analysis 29 samples collected from Gilgit and Hunza Rivers...... 30 Figure 2.13 Principal Component analysis (PCA) based on eleven parameters of water samples collected from Gilgit and Hunza valleys ...... 32 Figure 2.14 Scree plot of 29 water samples with eleven parameters ...... 33 Figure 3.1 Average monthly temperature in Co and rainfall in millimeter of Gilgit station based on the data period from 1955-2009 ...... 40 Figure 3.1 Average monthly temperature in Co and rainfall in millimeter of Gilgit station based on the data period from 1955-2009 ...... 40 Figure 3.2 Map 1 showing the study sites from Gilgit and Hunza valleys. Yellow boxes are the sites from where samples are collected. The arrow from the second figure (Map 2) highlights the selected area from Northern Pakistan...... 41 Figure 4.1 Dependence of series intercorrelation with site slope.The red circle shows the slope from 25o to 35o and the blue circle represents the slope ranged 40o to 55o ...... 61 Figure 4.1a Picea smithiana Kargah chronology plots. five figures representing raw, standard, residual, arstan chronologies and sample depth respectively...... 66 Figure 4.2a Picea smithiana Jutial chronology plots. five figures representing raw, standard, residual, arstan chronologies and sample depth respectively...... 67

xiii

Figure 4.3a Picea smithiana Haramosh chronology plots. five figures representing raw, standard, residual, arstan chronologies and sample depth respectively...... 68 Figure 4.4a Picea smithiana Bagrot chronology plots. five figures representing raw, standard, residual, arstan chronologies and sample depth respectively...... 69 Figure 4.5a Picea smithiana Nalter chronology plots. five figures representing raw, standard, residual, arstan chronologies and sample depth respectively...... 70 Figure 4.6a Picea smithiana Chera chronology plots. five figures representing raw, standard, residual, arstan chronologies and sample depth respectively...... 71 Figure 4.7a Picea smithiana Chaprot chronology plots. five figures representing raw, standard, residual, arstan chronologies and sample depth respectively...... 72 Figure 4.8a Juniperus excelsa Chaprot chronology plotes. five figures representing raw, standard, residual , arstan and sample depth respectively …………………………………….73 Figure 4.9a Juniperus excelsa Nalter Kargah chronology plots. five figures representing raw, standard, residual, arstan chronologies and sample depth respectively...... 74 Figure 4.10a Pinus gerardiana Chaprot chronology plots. five figures representing raw, standard, residual, arstan chronologies and sample depth respectively...... 75 Figure 4.1b Running rbar and EPS graph of Picea smithiana from Kargah……………..77

Figure 4.2b Running rbar and EPS graph of Picea smithiana from Jutial. ………………77

Figure 4.3b Running rbar and EPS graph of Picea smithiana from Haramosh………….. 78

Figure 4.4b Running rbar and EPS graph of Picea smithiana from Bagrot……………...78

Figure 4.5b Running rbar and EPS graph of Picea smithiana from Nalter………………79

Figure 4.6b Running rbar and EPS graph of Picea smithiana from Chera.………………79

Figure 4.7b Running rbar and EPS graph of Picea smithiana from Chaprot…………….80

Figure 4.8b Running rbar and EPS graph of Juniperus excelsa from Chaprot…………..80

Figure 4.9b Running rbar and EPS graph of Juniperus excelsa from Nalter……………..81

Figure 4.10b Running rbar and EPS graph of Pinus gerardiana from Chaprot………….81

xiv

Figure 4.11 The autocorrelation coefficients (AC) and partial autocorrelation coefficients (PAC) values were calculated out to 10 lags. Red lines indicate 95% confidence interval….83-86

Figure 4.12 Two graphs show 200 years chronologies among ten sites. Arrow indicate pointer years all sites……………………………………………………………………………...... 93

Figure 4.13 Dendrogram resulting from Ward's cluster analysis of 200 years (1800-2000) among ten sites……………………………………………………………………………...... 95

Figure 4.13 Principal component analysis of ten sites using the common period of 200 years (1800-2000)……………………………………………………………………………………95

Figure 5.1 Box-plot of mean monthly temperature of Gilgit station based on the period (1955- 2009)……………………………………………………………………………………………102

Figure 5.2 Box-plot of mean monthly precipitation of Gilgit station based on the period (1955- 2009)…………………………………………………………………………………………...103

Figure 5.3 Hierarchy of method which is followed for correlation and response analysis…105

Figure 5.4 Graph representing correlation coefficients between residual chronologies and temperature of Gilgit meteorological data from ten sites respectively for 13 months span.107-109

Figure 5.5 Graph representing correlation coefficients between residual chronologies and precipitation of Gilgit meteorological data from ten sites respectively for 13 months span110-112

Figure 6.1 Actual (red) and reconstructed (dashed) March-June temperature during common period 1955-2008…………………………………………………………………………….128

Figure 6.2 Scatter plot of the observed and reconstructed temperature of the data that were used for early calibration period (1955-1985)…………………………………………………….128

Figure 6.3 The Gilgit March-June reconstruction over the entire period of 1523-2008….132

Figure 6.4 Ten year running mean window describes the trend of warm and cold years. Upper line of graph represents warm years and lower line signifies cold years during 25 year's intervals………………………………………………………………………………………132

Figure 6.5 Hundred year running mean window describes the trend of warm and cold years. Red line of the graph represents 100 mean running window during 1620-2000……………132

Figure 6.6 Comparison between the two temperature reconstructions based on 10 years moving average. Blue line indicates Pinus gerardiana Chaprot while red line shows Picea smithiana Jutial reconstruction………………………………………………………………………….133

Figure 6.7 Scatter plot of the Jutial and Chaprot temperature reconstruction based on 168 years of data (1840-2008). Jutial reconstruction is on Y-xis and Chaprot reconstruction is represented on X-xis………………………………………………………………………………………134

xvi viii- List of Symbols, Abbreviations

WWF…………………………………………………………………World Wildlife Fund

BOD………………………………………………………... Biochemical Oxygen Demand

COD………………………………………………………... Chemical Oxygen Demand

TOC………………………………………………………... Total Organic Carbon

PCA………………………………………………………… Principal Component analysis

CA………………………………………………………………….. Cluster analysis

GPS………………………………………………………………… Global Positioning System

GPE………………………………………………………………… Gold Panning Extraction

TDS………………………………………………………………… Total Dissolved Solids

DO………………………………………………………………….. Dissolved Oxygen

WHO……………………………………………………………….. World Health Organization

EC………………………………………………………………….. Electrical conductivity mg/L…………………………………………………………………milli gram per Litre

T. Alkalinity…………………………………………………………Total Alkalinity

T. Hardness………………………………………………………….Total Hardness

PC……………………………………………………………………Principal Component

PMD…………………………………………………………Pakistan Meteorological Department

Ft……………………………………………………………………..Feet mm…………………………………………………………………...millimeter oC…………………………………………………………………….Degree Centigrade

NW…………………………………………………………………. Northwest

N……………………………………………………………………. North

ES……………………………………………………………………Eastsouth

E……………………………………………………………………..East

xvii

W…………………………………………………………………….West

ITRDB………………………………………………………International Tree Ring Data Bank

SNR………………………………………………………….Signal-to-Noise ratio

EPS…………………………………………………………..Expressed population signal

DBH…………………………………………………………Diameter at breast height

ID……………………………………………………………Identity

DPL………………………………………………………….Dendrochronology Program Library

PSKAR………………………………………………………Picea smithiana Kargah

PSJUT……………………………………………………….Picea smithiana Jutial

PSHAR………………………………………………………Picea smithiana Haramosh

PSBAG………………………………………………………Picea smithiana Bagrot

PSNAL………………………………………………………Picea smithiana Nalter

PSCHR………………………………………………………Picea smithiana Chera

PSCHP………………………………………………………Picea smithiana Chaprot

JECHP ……………………………………………………….Juniperus excelsa Chaprot

JENAL………………………………………………………Juniperus excelsa Nalter

JEMOR………………………………………………………Juniperus excelsa Morkhun

PGCHP………………………………………………………Pinus gerardiana Chaprot

ACF………………………………………………………….Autocorrelation Function

PACF………………………………………………………..Partial Autocorrelation Function m…………………………………………………………….meters

PCReg……………………………………………………….Principal Component Regression

CE……………………………………………………………Coefficient of efficiency

RE……………………………………………………………Reduction of error

RP……………………………………………………………Pearson Product moment correlation

xviii

RR…………………………………………………………...Robust Correlation coefficient

RS……………………………………………………Spearman coefficient of rank correlation

RSQ………………………………………………………….Variance explained

xix xi- Acknowledgements

The author wishes to express his profound gratitude and sincere appreciation to Professor Dr. Moinuddin Ahmed for his assiduous guidance, critical review, and kind help; which never failed to inspire thoughtful approaches to the subject.

My sincere thanks are due to Dr. Jonathan Palmer, Director of Gondwana Tree ring Laboratory, New Zealand who taught me to compute dendrochronolgial data and how to interpret these. I also thanks to members of Lamont Doherty Earth Observatory Colombia University USA for providing me the relevant softwares especially to Prof. Brendon M. Buckley for his useful discussions on many aspects.

I offer my enduring gratitude to my seniors and pioneers of dendrochronology in Pakistan Dr. Muhammad Wahab and Dr. Nasrullah Khan, and my fellow students Muhammad Akbar and Alamdar Hussain at Federal Urdu University of Arts Science and Technology, who have helped me in the collection of my samples. I also offer my appreciation to Azhar Kazmi for his support in arrangement of the thesis. I owe particular thanks to Dr. Syed Shahid Shaukat, who trained me in statistical analysis and in particular provided me insights into multivariate analysis. I thank to my co-supervisor Dr. Muhammad Afzal Farooq for enlarging my vision of science and providing coherent answers to my endless questions. I cannot forget Professor Dr. Arif Zubair Dean faculty of science whose penetrating pertinent questions forced me to think more deeply.

Last but not the least, my special thanks are due to my parents, whose have supported me throughout my years of education, both morally and financially, and I owe a special debt of gratitude to my brothers, sister and my wife for her understanding, support and most of all patience during the completion of the thesis.

General introduction

This thesis comprises of two portions. Water analysis of Gilgit and Hunza valleys; which includes investigation of physico-chemical properties of surface water. It is being considered that drinking water of these areas is like mineral water and it should be supplied to all over Pakistan.

However, during last few decades increased pollution even in watershed areas create a great concern. To check whether this statement is true or false, water analysis of this area was performed. As far as water analysis is concerned, little work has carried out in these areas.

Mercury in Pan Amalgamation was found in river in high concentration in 37 samples collected from 24 different sites of Gilgit and Hunza (Biber et al. 2011). Physico-chemical properties along with some heavy metals (arsenic, chromium, copper, mercury and lead) of Handrap Lake and nullah in Ghizer district were tested and no trace metals were found (WWF). A few samples have been analyzed for only physical and chemical analysis but still not published. In present study, water samples were not restricted just only to rivers but also few water samples were collected from nallahs and tap waters (used for the drinking purpose) to check out the differences among these three types. Water samples were selected from 29 locations followed by standard sample collection techniques. Physical parameters such as pH, Electrical conductivity, temperature, total dissolved solids and salinity were identified at site. Further analysis was performed in laboratories for chemical properties including chloride, total hardness, total alkalinity, sulphate and nitrate.

The second portion of this thesis describes the climatic history of Gilgit and Hunza valleys using tree rings. These areas are temperature and moisture dependent which are the limiting factor for tree growth. Global climate is continuously and rapidly changing and not autonomous. Recent floods are examples of this global environmental change and glacier melting (Muhammad,

1 2010). The fluctuation of climate cannot be understood from few years of data. Available 40 to

50 years meteorological department data is scarce therefore we need a tool to provide a long term variations of climate. Dendrochronological work was carried out in northern areas of Pakistan since 1987. Response function analysis of temperature and precipitation in some of northern areas including Afghanistan (Khan et al. 2008) Hunza (Esper et al. 2000), Astore and Ayubia

(Ahmed et al. 2010a) were carried out. Here, we concentrated on Picea smithiana, Juniperus excelsa and Pinus gerardiana of eight locations from Gilgit and Hunza valleys to find out the history of climate. The samples collected from different locations were analyzed to develop a network of tree rings in comparison with meteorological and grid data to find out growth-climate correlation and response. For this purpose, different softwares i.e. COFECHA, DPL

(Dendrochronology Program Library), ARSTAN, Minitab, Correlation and Response Function analysis (Fritts, 1976), and principal component regression analysis have been used. The sites which showed highest climatic signals were forwarded for further analysis while those sites showing no climatic signals were rejected.

Present investigation includes following objectives.

Objective 1

To identify physico-chemical properties and extent of pollution and its concentration in water of rivers, nullahs, springs and municipal pipe lines at Gilgit and Hunza valleys.

Objective 2

To check out concentration of Chloride, total hardness, total alkalinity, nitrate and sulphate whether they are in permissible limits in drinking water, nullahs and river water.

Objective 3

To develop a network of tree ring chronologies using different species at different sites of Gilgit and Hunza valleys.

Objective 4

To explore growth-climate (temperature and precipitation) response of various tree species growing in Gilgit and Hunza valleys.

Objective 5

To reconstruct temperature more than past 400 years from Gilgit and Hunza valleys

PART ONE

WATER ANALYSIS OF GILGIT AND HUNZA VALLEYS

1 Chapter No. 1

Water analysis

This chapter describes the importance of water in terms of quality parameters. The problem of statement is well discussed followed by site description. An over view of water analysis carried out throughout Pakistan, is given. Sample collection techniques, field and laboratory methods are discussed. Eleven parameters are detected in which first five are physical (temperature, pH, electrical conductivity, total dissolved solids, and salinity) and remaining are the chemical parameters consisting of dissolved oxygen, chloride, total hardness, total alkalinity, sulphate and nitrate.

1.1-Introduction

Water is a natural source and basic human need but sometimes its quality is deteriorated by anthropogenic activities. Fresh water shortage is ever-increasing in the water-starved regions due to increasing population (Seckler et al. 1998). Over extractions of underground water not only depletes water table but also makes good quality aquifer vulnerable to be contaminated by unfavorable substances (Shah et al. 2002). Over population and heavy industrialization have produced a critical situation for water resources (Chaudhary et al. 2001). Most of developed countries adopted alternative supplies for their domestic use, while in other parts of the world particularly in developing countries like in Pakistan; alternative supplies are not available to handle the whole urban population (Farooq, 2008). For the evaluation and water resources and to reduce the threats of pollution, quality plays an important role rather than quantity. There are many parameters which represents the water quality and composition in specific localities and time (Praus, 2005). Certain indicators of surface water quality have been familiarized to measure the fitness of the water and assumed to be the gauge of quality of water whether water is for the use of drinking or other industrial or agricultural purposes.

The water quality is measured by its physical, chemical and biological parameters while the other parameters are heavy metals, pesticides, organic matter including BOD, COD and TOC. Physico-chemical changes such as pH, alkalinity, hardness, nutrients and other heavy metals cause sensitiveness to aquatic organisms (Khan et al. 1999). The problem in the assessment of

1 water quality is the complexity of analysis and large number of data sets which contain much information regarding the behavior of the water. As it is difficult to treat all the parameters in combination, many researchers interpret the water quality parameters individually by describing the seasonal variability and their causes.

Indus River is the biggest source of water in Pakistan covering the area of 1,140,000 sq. kms and has social and economic value. The main source of Indus is in Tibet, it begins in the convergence of Sengge River and Gar River that drains the Ngangoing Kangri and Gangelise Shah ranges. The Indus then winds itself from north to south through Gilgit Baltistan just south to the Karakorum Range then bends to the south, coming out of the hills between Peshawar and Rawalpindi, plains of Punjab and Sindh and then routes to lower Sindh where it finally falls into Arabian Sea.

The Indus is nourished by glaciers and snows of the Himalayas, the Karakorum and the Hindu- Kush that originate in the Indian State of Jammu and Kashmir and The Northern Areas of Pakistan. The Indus consists of two basins i.e. Upper Basin and Lower Basin. The parts of the HinduKush and the Karakorum ranges in the northern territory of Pakistan are drained by Gilgit River (that is my study area) which is bordered with Afghanistan and China in the north. The Gilgit River combines with the network of different types of rivers including Ghizer, Yasin, Ishkuman and Hunza River which then finally joins the Indus River near Juglot. The upper parts of the basin are generally glaciated and covered with permanent snow.

The Hunza River basin is also a part of my study area, actually the sub basin of the Gilgit River but owing to its substantial size and significance, it is considered as a separate basin. Like Gilgit River, Hunza River also drains the Karakorum Mountains consisting of large glaciated area situated in the north. The Karakorum Highway that links Pakistan to China passes across this basin. Karimabad is the capital of Hunza valley, extended over miles and miles of terraced field and orchards. To check out quality of water from study areas following objective will be covered.

1.2-Research objectives

1. To obtain water quality data 2. To examine physical and chemical contamination

3. To learn trends of dissolved solid concentrations and its load at different sites 4. To identify whether the concentration of dissolved solids are within the permissible limits of World Health Organization (WHO, 1993). 5. To recognize the water quality and ecological status through the use of multivariate statistical techniques 6. To classify possible factors which are responsible for the variation in water quality of Gilgit and Hunza Rivers 7. To relate multivariate statistical techniques to study homogeneity and heterogeneity among sampling stations and to differentiate water quality variable for temporal variation of Gilgit and Hunza Rivers

Figure 1.1: Map representing twenty nine sampling sites from Gilgit and Hunza Rivers

Table 1.1: Nearest town, elevation and map location of water collection of Gilgit and Hunza valleys.

S. No. Locations Elevation in meters Co-ordinates 1 Baseenpur 1683 35o50N, 74o15E 2 Baseenpur (spring) 1700 35o50N, 74o15E 3 Kargah 1674 35o50N, 74o15E 4 Gilgit city 1574 35o54N, 74o21E 5 Gilgit tap water 1574 35o54N, 74o21E 6 Jutial 1748 35o53N, 74o20E 7 Nomal 2507 36o08N, 74o12E 8 Nalter (spring) 2968 36o07N, 74o10E 9 Nalter (Lake) 2968 36o07N, 74o10E 10 Danyore 1580 36o08N, 74o51E 11 Juglot Gah Nala 1610 36o09N, 74o51E 12 Haramosh Nala 1600 35o07N, 74o08E 13 Aliabad Nala 1700 36o09N, 74o52E 14 Aliabad tapwater 1700 36o09N, 74o52E 15 Atabad 2400 36o20N, 74o52E 16 Gulmit 2412 36o20N, 74o52E 17 Hussaini 2433 36o20N, 74o52E 18 Ghalapur Nala 2500 36o36N, 74o51E 19 Khyber Nala 2678 36o34N, 74o48E 20 Passu 2700 36o46N, 74o90E 21 Gulkin Nala 2403 36o24N, 74o52E 22 Batura Glacier 2540 36o30N, 74o52E 23 Batura Lake 2540 36o30N, 74o52E 24 Shimshal River 2850 36o20N, 75o01E 25 Shimshal 2850 36o20N, 75o01E 26 Morkhun 2780 36o40N, 74o52E 27 Boiber tributary 3075 36o40N, 74o52E 28 Boiber Nala 3075 36o40N, 74o52E 29 Sost River 3075 36o41N, 74o52E

1.3-Review of Literature

The works regarding the analysis of water quality of Gilgit-Baltistan area are scant. Water of Gilgit and Hunza rivers was analyzed to check concentration of mercury from water samples collected from 37 sites. The main source of mercury was Pan Amalgamation in the small scale gold panning and Extraction (GPE). Samples were tested in terms of dissolved and suspended mercury in water and the main purpose of research work was to create a hydrological modeling to recognize the source, fate and transport of mercury and to build up scenarios to reduce mercury concentration to permissible limits (Biber, 2011).

Physical and microbial analysis was carried out from Nomal valley (Ahmed et al. 2007) by collecting water samples throughout the year except January and February. They found the highest fecal contamination of water at source in the months of May-August.

Physico-chemical quality of Jhelum River water for irrigation and drinking purposes at District Muzaffarabad, Azad Kashmir were pointed out by Sarwar et al. (2007). Various physico- chemicals were checked like pH, Electrical Conductivity, total dissolved solids and suspended solids. They suggested that Jhelum River water is suitable for drinking and irrigation purpose.

Kabul River and its tributaries were assessed for its organic and faecal coliform strength starting from Warsak Reservoir to the confluence point of Kabul and Indus Rivers (Khan et al. 1999). Thirty eight samples offered high concentration of fecal contamination rendering the water unfit for irrigation and human consumption. Khan et al. (1999) claimed that organic and fecal contaminations were caused due to the effluent discharged from Khazana Sugar Mills, Colony Sarhad Textile Mills, Amerjee Papers and Paper Board Mills and from different tannery industries. Faecal contamination, using most probable number technique, was reported in one of the residential sector of Islamabad city (Azhar, 1996) and from Risalpur, Pubi and Tarnab (Ihsan-Ullah et al. 1999).

The whole Lahore city was investigated for its bacteriological quality of drinking water and 530 water samples were collected from different localities during the months of April and May (Anwar et al. 2010). Among 530 samples, 197 samples were positive for bacteriological contamination. Anwar et al. (2010) concluded that bacterial contamination is a significant problem in Lahore.

Besides northern areas in Pakistan, surface water of Lower Indus Basin from Kashmor to Keti- Bander was assessed for its physical, chemical, trace metals and microbiological analysis (Farooq. 2012). Multivariate analysis using cluster analysis and factor analysis explained that surface water is of acceptable quality in terms of its physico-chemical properties and the level of coliform bacteria; however the levels of some heavy metals like lead, mercury and cadmium exceeded the WHO (1993) permissible limits.

Limnological studies of Keenjhar Lake were conducted by Lashari et al. (2009) which dealt with the physico-chemical properties of water including temperature, pH, alkalinity, chloride, conductivity, TDS, turbidity, DO, calcium and magnesium. The outcome of the study demonstrated that all the parameters of Keenjhar Lake are in accordance with the permissible limits for aquatic quality characteristics. Physico-chemical analysis for the potable water of Khairpur city was conducted to investigate the quality of water for the city population (Pirzada et al. 2011). The concentration of pH, electrical conductivity, total dissolved solid, hardness, alkalinity, chloride and sulphate were measured. They showed that estimated limits of cations and anions were safe according to the limits proposed by WHO (1993).

Drinking water for bacteriological contamination was tested from ground and surface water samples in Rohri city (Shar et al. 2010). They agreed that water was contaminated with total coliform, including Escherichia coli and other Heterotrophic plate count bacteria in pre- and post-storage. Fecal contamination was also identified by Shar et al. (2009) in main reservoir; distribution line and consumer tap in Sukkur city. The reason was in fact the occurrence of animal excreta in nearby water sources which may increase the number of coliform and Escherichia coli in the drinking water of Sukkur.

Zubair et al. (2009) used factor analysis for the determination of trace metals from both open and bore wells and found elevated concentrations of Pb and Zn in well water in both pre- and post- monsoon periods.

Little work has been carried out from Gilgit and Hunza Rivers in terms of its physico-chemical characteristics whether surface water is suitable for human consumption. So bearing this in mind, we focus to estimate homogeneity and heterogeneity among sampling stations and to distinguish water quality variables for temporal variations in Gilgit and Hunza Rivers.

1.4-Materials and methods

1.4.1-Sampling and on site evaluation

Twenty nine locations were chosen from the valleys of Gilgit and Hunza for the estimation of physico-chemical parameters. Samples were collected in the month of July 2012. The collection was carried out in such a way that samples did not get contaminated with other substances. At each site, surface water was collected and kept in polythene plastic bottles formerly washed in 10% nitric acid for 24 hours and rinsed with distilled water. These bottles finally swamped with sample water also for two to three times. 500 ml water was collected in each bottle and six parameters were noted at the spot with the assistance of Sension 156 HACH potable multi- parameter, USA. The parameters were temperature, pH, electrical conductivity, total dissolved solids, salinity and dissolved oxygen. Two-three drops of nitric acid were introduced in the bottles so that chemical characteristics of the water could not be deteriorated.

Table 1.2: Analysis parameters and their analytical procedures

Serial Variables Abbreviation Analytical method Units No. Physical Parameters 1 Temperature Temp Sension 156 oC HACH, 2 pH pH Sension 156 No HACH, 3 Electrical Conductivity EC Sension 156 µS cm-1 HACH, 4 Total Dissolved Solids TDS Sension 156 mg L-1 HACH, 5 Salinity Sal Sension 156 % HACH, Chemical Parameters 6 Dissolved Oxygen DO Sension 156 mg L-1 HACH, 7 Chloride Cl-1 Titration (Silver mg L-1 Nitrate) 8 Total Hardness Ca+Mg Titration (EDTA) mg L-1

-1 9 Total Alkalinity CO3+HCO3 Titration (H2SO4) mg L

-1 10 Sulphate SO4 Spectrophotometer mg L

-1 11 Nitrate NO3 Spectrophotometer mg L

The procedures followed were those described by APHA (2003).

1.4.2-Methods for the detection of chemical parameters

1.4.2.1-Chloride

Titration by silver nitrate

25 ml of sample was taken in titration flask and few drops of potassium chromate were added to it as an indicator, and then titrated with 0.0141N solution of AgNO3 until slight reddish color attained. The final reading of Chloride was obtained by putting the value AgNO3 consumed in the burette.

Cl in mg/L = (ml of AgNO3 used in titration * 0.0141(N) * 35,450) / Vol. of sample

1.4.2.2-Carbonate alkalinity

Titration by sulphuric acid

25 ml of sample was taken in titration flask with a few drops of phenolphthalein indicator were added to it. The solution became pink. The solution was titrated with 0.1N H2SO4 until the pink color was disappeared.

Alkalinity in mg/L = (ml of sulphuric acid used in titration * 0.1N *50,000) / Vol. of sample

1.4.2.3-Bicarbonate alkalinity

Titration by sulphuric acid

A few drops of methyl orange as an indicator were added to the sample of 25 ml of water. The sample turned yellow by the addition of indicator. The solution was titrated with 0.1N H2SO4 standard acid which changed the sample from yellow to orange.

Alkalinity in mg/L = (ml of sulphuric acid used in titration * 0.1N *50,000) / Vol. of sample

1.4.2.4-Total Hardness

Titration by EDTA

Measured volume of 25 ml of sample was taken in titration flask and few drops of Eriochrome Black T were added as indicator and titrated with standard solution of 0.01M EDTA until sample turned blue.

Total Hardness in mg/L = (ml of EDTA used in titration * 1000) / Vol. of sample

1.4.3-Statistical analysis

For statistical analysis, I used statistical software “Minitab” version 11.12. Box and whisker plots are produced for every site which describe the minimum, maximum and mean values, quartile 1, quartile 3, and outliers of the data. Correlation matrix among all parameters are created to check whether the concentration of one parameter affect the concentration of other. Multivariate techniques (cluster analysis and principal component analysis) are applied to the datasets to yield comprehensive results.

Chapter No. 2

Results

This chapter details the Analysis of water collected from 29 locations of Gilgit and Hunza valleys. Results of Physico-chemical properties of water are discussed. Finally, these physico- chemical properties are compared by means of multivariate analysis including cluster and Principal component analysis.

2.1-Temperature

20 Temperature

Figure 2.1: Graph shows the box and whisker plot of temperature from all sites. Central line in the box is the mean, upper line represents 3rd quartile and lower line expresses 1st quartile. Asterisks in the graph show outliers.

Box and whisker plot shows that most of the temperatures of the sites fell from 11oC to 13oC. The lowest temperature was 7.4oC which was observed from Nalter (spring). Both samples from Gilgit showed highest temperatures among all samples (more than 25oC). Although there are no permissible limits of temperature set by WHO (1993), yet temperature may be helpful in the growth of some microorganisms in water.

2.2-pH

8 pH

Figure 2.2: Graph shows the box and whisker plot of pH from all sites. Central line in the box is the mean, upper line represents 3rd quartile and lower line expresses 1st quartile. Asterisks in the graph show outliers.

Most of the samples were found within the permissible limits described by WHO (1993) whereas seven samples did not expose the same results. The Morkhun valley showed the highest value (pH=8.75) and the lowest values were observed from three sites i.e. Gilgit tap water, Nalter Lake and Shimshal River (pH=7.0). Apparently, most of the samples collected from Hunza rivers are towards the basic side having pH more than 8 (Table 2.1). Lower value of standard deviation (0.59) defines that pH of all sites is similar.

2.3-Dissolved Oxygen

2.0

1.5

Dissolved Oxygen

1.0

0.5

Figure 2.3: Graph shows the box and whisker plot of dissolved oxygen from all sites. Central line in the box is the mean, upper line represents 3rd quartile and lower line expresses 1st quartile. Asterisks in the graph show outliers.

Only one outlier was observed showing the lowest D.O 0.49 mg/L from Nalter Lake. Most samples expressed D.O ranged 1.4-1.9 mg/L. The mean value was obtained 6.26 mg/L whereas the highest value was seen from Nalter spring (2.2 mg/L). There is no guideline of D.O described by WHO (1993). The value of standard deviation of dissolved oxygen (24.9) explains that a little bit difference in the D.O of all sites occurred.

2.4-Total dissolved solids

300

200

100 Total dissolved solids Total

Figure 2.4: Graph shows the box and whisker plot of total dissolved solids from all sites. Central line in the box is the mean, upper line represents 3rd quartile and lower line expresses 1st quartile. Asterisks in the graph show outliers.

No outlier was found in the box and whisker analysis. The highest value of total dissolved solids touched nearly to 280 mg/L which was far from the WHO (1993) allowable limits. Maximum value was seen from Boiber tributary and the minimum value was reported from Gulkin Nala (TDS = 19.3 mg/L). All values of TDS were beneath the permissible limits designed by WHO (1993). Total dissolved solids have the standard deviation 67.4 which is high showing a considerable difference among the samples.

2.5-Electrical conductivity

500

400

300

200 Conductivity

100

Figure 2.5: Graph shows the box and whisker plot of electrical conductivity from all sites. Central line in the box is the mean, upper line represents 3rd quartile and lower line expresses 1st quartile. Asterisks in the graph show outliers.

Two sites (Juglot Gah Nala and Boiber tributary) exceeded the allowable limits of drinking water quality. These two sites represented higher values of conductivity with more than 500 mg/L. The lowest value was observed from Gulmit site (26.3 mg/L).The other samples have the conductivity values within the tolerable limits. A high value of standard deviation was obtained (144.6) indicating a vast difference in the values of conductivity exists among the samples.

2.6-Salinity

0.3

0.2 Salinity 0.1

0.0

Figure 2.6: Graph shows the box and whisker plot of salinity from all sites. Central line in the box is the mean, upper line represents 3rd quartile and lower line expresses 1st quartile. Asterisks in the graph show outliers.

Only one outlier was seen representing higher value of 0.3 from Boiber tributery. This value was not observed from any other sample. All other samples attained values from 0.0 to 0.1. Out of 29 water samples, 14 showed 0.0 while the rest 14 showed 0.1 value of salinity. This parameter also has no guideline of WHO (1993) acceptable limit of drinking water quality. Lowest value of standard deviation was obtained among samples i.e. 0.068.

2.7-Chloride

10 Chloride 9

Figure 2.7: Graph shows the box and whisker plot of chloride from all sites. Central line in the box is the mean, upper line represents 3rd quartile and lower line expresses 1st quartile. Asterisks in the graph show outliers.

The lowest value was 6.0 mg/L obtained from water sample of Hussaini also being represented in outlier of box and whisker plot. The highest value was less than 15 mg/L reported from Shimshal site. Chloride was experimented much lesser than the WHO (1993) tolerable limits. Occurrence of most samples in terms of chloride was 10-12 mg/L.

2.8-Total alkalinity

1200

1100

1000

900

800

700

Total Alkalinity Total 600

500

400

Figure 2.8: Graph shows the box and whisker plot of total alkalinity from all sites. Central line in the box is the mean, upper line represents 3rd quartile and lower line expresses 1st quartile. Asterisks in the graph show outliers.

Range of the samples was found to be 600-900 mg/L with no outlier. Total alkalinity was seen higher in all samples beyond the WHO (1993) drinking water acceptable limits. The range of values from lower to higher side was 360 to 1200 mg/L. Highest value was inspected from Gilgit tap water and lowest value from Nalter spring. Value of 207 of standard deviation was reported from all samples which is highest among all parameters explaining a large difference among the values of alkalinity.

2.9-Total hardness

250

200

150

100 Total Hardness Total

Figure 2.9: Graph shows the box and whisker plot of total alkalinity from all sites. Central line in the box is the mean, upper line represents 3rd quartile and lower line expresses 1st quartile. Asterisks in the graph show outliers.

Six samples were scrutinized to be higher with the elevated value observed from Shimshal (240 mg/L). All the six higher values were from Hunza Rivers. The lowest value was experienced from Jutial (28 mg/L). Most of the samples existed within the range of 50 to 150 mg/L with no outlier. The mean value was 100 mg/L as apparent from the graph.

2.10-Sulphate

120

100

60 Sulphate 40

Figure 2.10: Graph shows the box and whisker plot of sulphate from all sites. Central line in the box is the mean, upper line represents 3rd quartile and lower line expresses 1st quartile. Asterisks in the graph show outliers.

Only one outlier was seen touching the highest value of sulphate (119 mg/L) from Aliabad tap water whereas the water sample collected from Nala of the same site did not have the same value (only 18 mg/L). The difference might be the clarity of the two samples. Aliabad tap water sample was more turbid than Nala that might increase the sulphate concentration. The lowest value was reported from Gulkin Nala and most existence of sample value was 18-44 mg/L. All samples expressed permissible WHO (1993) limits.

2.11-Nitrate

Nitrate 20

Figure 2.11: Graph shows the box and whisker plot of nitrate from all sites. Central line in the box is the mean, upper line represents 3rd quartile and lower line expresses 1st quartile. Asterisks in the graph show outliers.

Mean value of nitrate was just exceeding the value of 20 mg/L. Box in the graph covered the area of 13 to 33 mg/L representing the incidence of most samples. The highest value was above 40 mg/L from Boiber Nala and lowest value from Nomal i.e. 5 mg/L. Importantly; all samples are within the permissible limits of WHO (1993). No outlier was observed in the analysis.

First 12 locations were from Gilgit River and the last 17 locations were selected from Hunza River (Table 2.1). We took averages of all parameters from two rivers and then compared these averages. Temperature, pH, DO, salinity, chloride and nitrate got the same values. The other parameters TDS, conductivity, total alkalinity, total hardness and sulphate showed dissimilarities in results. The higher values were obtained from Hunza River explaining Hunza River was more disturbed than Gilgit River.

The values of twenty nine water samples with eleven parameters are presented in Table 2.1. World Health Organization limits for safe drinking water are shown on the bottom of the table.

Above discussion shows that Boiber tributary crossed the WHO (1993) approved limits of pH, conductivity, total alkalinity and total hardness and same was the case with Boiber Nala. The values of all parameters were satisfying the corresponding limits (except in few cases) whereas total alkalinity was found high even in minimum value column. It means that all samples have high values of alkalinity. Total dissolved solids (TDS), Dissolved Oxygen (D.O), salinity, chloride, sulphate and nitrate met the acceptable limits designed by WHO (1993).

Table 2.1: Values of all sampling sites with eleven parameters from Gilgit and Hunza Rivers

Total Total Locations Temperature PH D.O TDS Conductivity Salinity Chloride Alkalinity Hardness Sulphate Nitrate Baseenpur 13.4 8.54 0.98 23 40.2 0 14.00 420 40 10 11.1 Baseenpur (Spring) 12.31 7.57 0.91 22.3 35.7 0.1 10.00 460 32 10 27.4 Kargah 13 7.6 1.8 24.3 38.2 0.1 10.00 700 32 11 10.9 Gilgit City 27.6 8.1 1.7 31 69.5 0 10.00 480 48 18 37.4 Gilgit tap water 25.2 7 1.65 35 83.6 0.1 9.93 1200 40 11 32.9 Jutial 12.1 7.48 1.89 27.6 43.6 0.1 12.00 560 28 19 12.4 Nomal 12.9 8.65 1.29 110 68.6 0.1 8.00 1020 120 22 5 Nalter (Spring) 7.4 8.2 2.2 24.8 38.5 0.1 10.00 360 132 11 36.7 Nalter lake 9.4 7 0.49 61.4 145.7 0 7.94 960 80 21 9.5 Danyore 16.5 7.5 1.73 86.7 204 0 9.93 1020 100 37 21 Juglot Gah 12.4 7.77 1.5 211 505 0.1 Nala 12.00 860 80 23 9.7 Haramosh Nala 14.2 7.4 1.74 102.2 240 0.1 11.91 900 140 43 26.3 Aliabad Nala 12.3 7.1 1.63 36.5 86.1 0 9.93 840 40 18 22.5 Aliabad Tap 13.2 7.42 1.48 47.7 115.4 0 water 12.00 920 220 119 18.5 Atabad 10.2 7.4 2.01 69.7 164 0 7.94 700 100 21 29.4 Gulmit 12.1 7.86 1.47 85.7 20.3 0 12.00 980 72 32 18.5 Hussaini 10.2 7.75 1.41 71.21 168.9 0 6.00 820 100 24 16.4 Ghalapur Nala 11.8 8.49 1.36 166.9 70.4 0 12.00 700 116 9 13.7 Khyber Nala 16.7 8.54 1.85 171.8 316 0.1 12.00 740 172 78 10.8 Passu 11.3 7.2 1.7 52.6 124.4 0 11.90 880 80 34 33.2 Gulkin Nala 10.8 8.37 2.07 19.3 36.3 0 12.00 540 36 9 17.6 Batura Glacier 11.1 8.55 2.02 52.3 96.2 0 12.00 600 48 19 33.8 Batura Lake 11.1 8.3 1.98 119.9 219 0.1 10.00 1000 168 46 31.5 Shimshal River 10.9 7 2.01 78.3 184.8 0 14.89 800 120 48 36.5 Shimshal 11 7.2 1.92 181 423 0.1 9.93 860 240 91 39.9 Morkhun 13.3 8.75 1.26 182 426 0.1 8.00 840 136 79 36

Boiber 14.5 8.55 1.46 277 513 0.3 Tributery 10.00 820 176 86 12.1 Boiber Nala 13.2 8.73 1.7 155 290 0.1 8.00 560 196 60 42.7 Sost River 13.5 8.26 1.7 109.5 187.5 0.1 14.00 900 140 50 12.4

WHO No 500 No No guideline 6.5-8.5 400 mg/L permissible guideline mg/L guideline 250 150 500 50 limits (1993) mg/L 250 mg/L mg/L mg/L mg/L

Table 2.2: Correlation matrix among all parameters

Total Total Temperature pH D.O TDS Conductivity Salinity Chloride Alkalinity Hardness Sulphate pH -0.006

D.O -0.067 0.2

TDS -0.039 0.379 0.216

Conductivity -0.01 0.149 -0.134 0.877

Salinity 0.088 0.288 -0.165 0.592 0.561

Chloride 0.014 -0.023 0.131 -0.093 -0.137 -0.125 Total Alkalinity 0.155 -0.358 -0.071 0.289 0.273 0.067 -0.133 Total Hardness -0.166 0.189 0.039 0.638 0.603 0.354 -0.099 0.253

Sulphate 0.001 0.083 -0.18 0.565 0.631 0.349 0.032 0.301 0.843

Nitrate 0.126 -0.122 -0.152 -0.089 0.084 -0.09 -0.131 -0.156 0.205 0.142

2.12- The Pearson correlation matrix of all parameters

Correlations of all parameters from 29 samples of Gilgit and Hunza Rivers are shown in Table 2.2. Eleven parameters from 29 locations from two rivers were analyzed for correlation. The significance level at degrees of freedom 27 is checked using r Table in the following order.

1) Greater than 0.349 but less than 0.449 (p<0.05) 2) Greater than 0.449 but less than 0.554 (p<0.01) 3) Greater than 0.554 (p<0.001)

Temperature did not illustrate any significant correlation with all parameters. It was negatively correlated with pH, DO, TDS, conductivity and total hardness and positively correlated with salinity, chloride, total alkalinity, sulphate and nitrate. pH was found significantly correlated with total dissolved solids and total alkalinity at (p<0.05) and positively correlated with all parameters except chloride and nitrate.

No significant correlation was found among the dissolved oxygen and all parameters. It was positively correlated with three parameters and negatively correlated with other parameters. TDS showed significant correlations with four parameters i.e. conductivity, salinity, total hardness and sulphate at (p<0.001). Conductivity also expressed strong significant positive correlations with salinity, hardness and sulphate and weak positive correlation with total alkalinity.

Positive significant correlations were observed among salinity, total hardness and sulphate at (p<0.05) level and chloride did not show any significant correlation with any other parameter. Total alkalinity and total hardness indicated significant positive correlation with sulphate at (p<0.05) and (p<0.001) respectively. There was no significant correlation between sulphate and nitrate.

The above discussion explains that total hardness and sulphate are the two parameters which are highly correlated with most of the parameters. It indicates that if the concentration of these variables is disturbed, it is expected to alter the concentration of most variables thereby creating problem with the quality of water.

Figure 2.12: Dendrogram resulting from Ward‟s clustering of 29 samples collected from Gilgit and Hunza Rivers

Table 2.3: Characteristics of three groups derived from Ward‟s clustering of the water quality variables of the samples collected from 29 locations

Water Cluster I Cluster II Cluster III quality (1,2,4,8,3,18,6,21,22) (5,7,16,9,10,23,12,29,14,13,20,15,17,24) (11,27,25,2619,28) variables Temperature 13.28 13.07 13.51 pH 8.10 7.56 8.25 D.O 1.66 1.59 1.61 T.D.S 43.50 76.17 196.3 Conductivity 52.07 143.73 412.16 Salinity 0.04 0.035 0.13 Chloride 11.33 10.45 9.98 T. Alkalinity 535.56 924.2 780 T. Hardness 56.89 108.5 166.66 Sulphate 12.89 37.5 69.5 Nitrate 22.33 22.4 25.2

The characteristics of the three groups derived from agglomerative cluster analysis are presented in the sequel (Table 2.3, Fig. 2.12). Temperature was quite similar in all clusters. The pH of the water was found slightly more alkaline for cluster I and III. Dissolved oxygen was almost similar in all groups whereas total dissolved solids were remarkably higher in group III. Conductivity of the samples showed greater values for groups II and III. Salinity was found to be higher in cluster III while chloride was little bit higher in group I. Total alkalinity was found to be low in cluster I as compared to other two clusters. Total hardness of groups III was seems to be higher than the other two groups. Sulphate and nitrate showed greater values for group III.

Figure 2.13: Principal Component analysis (PCA) based on eleven parameters of water samples collected from Gilgit and Hunza valleys

Principal Component analysis (PCA) was applied on normalized data sets (11 variables) separately for 29 locations (Fig. 2.13) to find similarities or dissimilarities among variables. PCA of the data sets produced first five PCs with eigen values > 1 explaining 80.8 percent of total variance with respect to quality water data sets. Eigen value measures the significance of the factor and values greater than one are considered as significant (Shrestha and Kazama, 2007). Six parameters formed a close cluster in combination with sulphate. Hardness and TDS occurred as a group in the nearby area. The conductivity and alkalinity were found to be located quite apart from the rest indicating they have least correlation with the other variables.

The scree plot (2.14) was used to explain the number of PCs to be retained in order to understand the fundamental data structure (Vega et al. 1998). The scree plot of the present study showed that first five PCs have eigen values greater than one (Fig. 2.14).together the first four components explains 70.4% of the total variance inherent in the data set.

Figure 2.14: Scree plot of 29 water samples with eleven parameters.

Among five PCs, PC1 explaining 33% of total variance has weak positive loadings on TDS, conductivity, salinity, total hardness and sulphate. PC2 (14.5% of total variance) has moderate positive loadings on DO and nitrate. PC3 with 13.4% of total variance caused moderate positive loading over pH and alkalinity. PC4 having 10.2% of total variance has moderate loadings on temperature and chloride and weak loadings over nitrate. PC5 has also moderate positive loadings on temperature and chloride

2.13-Discussion

Northern areas of Pakistan are mountainous rural region with a population of 900,000 living in villages typically encompass 50-200 households (Nanan et al. 2003). In this region, the main source of water supply is from melting snows. It comes through channels (river lets) and small streams to the mouth of village and considered as the only supply of water for the villages as there are no wells and hand pumps which can substitute this water. The water from melting snows runs down with the collection of various materials on its way and converted to turbulent mountainous stream (McCarrison, 1906).

The utilization of such water for domestic purpose may cause harmful diseases. Access to safe drinking water is the basic human right of every citizen. Safe water is the water complying with National Drinking Water Quality Standards and meeting the quality in accordance with WHO (1993) and UNICEF joint report; Access means the availability of water at least 20 liters per person per day from an improved source within one kilometer. Present study is a first comprehensive study in which we investigated water resources (river, stream, lakes and nullahs) of Gilgit and Hunza valleys.

Mean conductivity was found satisfactory among all samples but water from Juglot Gah Nala exhibited high amount of conductivity. This sample also has high amount of TDS, perhaps this might be due to the presence and amount of minerals in water added from rocks and glaciers. Besides all parameters, total alkalinity crossed the tolerable limit of drinking water. As the rocks of these valleys contain high amount of carbonate and bicarbonate which mixes with the water passing through it, causing the high alkalinity in water.

Correlation analysis results explained that there are specific relationship pattern among pH, TDS, conductivity, total hardness and sulphate and these results are also confirmed by principal component analysis. Correlation analysis results showed the direct significant positive relationship (P<0.001) between electrical conductivity with total hardness and sulphate. It means that total hardness (Calcium+Magnessium) and sulphate control the conductance of surface water of the Rivers as reported in ground water of urban areas of Karachi (Farooq, 2008).

The cluster analysis of overall data set showed three major groups while the discriminating variable for the groups were six variables (temperature, pH, DO, salinity, chloride and nitrate) which showed homogeneity within groups but heterogeneity between the three

34 groups. TDS and hardness also exhibited considerable differences between the groups. Likewise, vast differences in the mean values were observed between the clusters for conductivity and total alkalinity. Cluster analysis highlights that at present, most of the samples collected from Hunza River have high values of TDS, conductivity, total alkalinity and hardness which may indicate the high concentration of salts in Hunza River as compared to Gilgit River.

In present study, we found the lowest temperature (7.4oC and 9.4 oC) from Nalter site (lake and spring respectively). The lowest temperature of water is due to the fact that the water which we collected from Nalter was coming from glaciers therefore found to be the lowest among all sites. Similar temperature (8 oC) was also observed by Islamuddin (2011) who worked over Nalter Lake. Water temperature from Nomal valley, located at lower height (2507 m) than Nalter valley (2968 m) was 13 oC in current study whereas (Ahmed and Shah, 2007) described the temperature of the same valley about 25 oC. The difference in results may be the difference in collection season.

All the physico-chemical properties from present study and Islamuddin (2011) study are concurrent with each other and are in accordance with the limits of WHO (1993) except total alkalinity (Carbonate and bicarbonate) were relatively higher in both studies. Our physico- chemical characteristics also match with the findings of Jhelum River, District Muzzafarabad Azad Kashmir study (Sarwar et al. 2007) and with that of some studies of lower Indus Basin (Farooq, 2012; Pirzada et al. 2011; Lashari et al. 2003) but total alkalinity was lower in their studies as compared to current studies.

There are no hard and fast rules for the permissible limits of physical properties (temperature, pH, turbidity, dissolved oxygen, TDS, conductivity and salinity) developed by WHO (1993). The concentrations of all chemical parameters were found within permissible limits in samples with the exception of total alkalinity in the samples.

Total alkalinity in terms of drinking water contaminant is not a primary or secondary source. However, alkaline water has a bitter taste and slippery feel. High alkaline water like in current study can cause drying of skin. Alkalinity is important for fish and aquatic life because it acts as a buffer against rapid pH changes (Benjamin, 2002; Hemond, 2000). High alkalinity is also important in agricultural activity. Use of high alkaline water affects the plant growth by excessive salts raising osmotic pressure in soil solution and it causes the reduction of water availability. This high alkalinity results in lower leaf-area index (Tyagi, 2003).

Finally it is concluded that physico-chemical properties of surface water of study areas are within drinking permissible limits of WHO (1993) but present water is considered as bicarbonate type. However Nano-filteration techniques should be installed at the mouth of water supply to reduce the total alkalinity of Gilgit and Hunza Rivers.

PART TWO

DENDROCLIMATIC HISTORY OF GILGIT AND HUNZA VALLEYS

Chapter No. 3

General introduction

The science of dendrochronology, its brief history and importance in the study of past climate variations are presented in this chapter. Climate of Pakistan including five provinces is also described. Brief introduction of Gilgit and Hunza valleys and its climate are discussed. Review of literature from Pakistan, China, Nepal and India are also presented.

3.1-Introduction to dendrochronology

The systematic study of tree rings pattern designated to a particular event with the passage of years is known as dendrochronology (Cook and Kariukstis, 1992).

3.1.1-Brief history of dendrochronology

According to Heizer (1956), the early Greeks were considered first to note the annual tree layers. They also knew that the widths of these layers were dependent on environmental conditions. Duhemel and Buffon in 1737, two French naturalists examined the frost damaged layers of 20 rings occurred in the bark of several felled trees. Other investigators confirmed their observations. Twining (1827) and Charles Babbage (1838) in England recognized crossdating based on relative ring widths. In 1892, a Russian worker, F. N. Shevedov was the first person to crossdate the annual rings and determined that structure of these episodes was because of past climate changes.

Andrew Ellicot Douglas is recognized as the father of dendrochronology. Douglass founded laboratory of tree ring research at the University of Arizona, Tucson (United States) in 1937 (the first institution specialized only for tree ring studies). Originally, the dendrochronological techniques were created to date archaeological structures; but later on tree ring analysis was used in various disciplines. These include plant ecology, geomorphology, hydrology, glaciology, seismology, entomology and importantly climatology. An important facility in the area of Dendrochronolgy is the establishment of the international tree-ring data bank (ITRDB) in 1974, which contribute to the global scientific community and free access to the tree-ring data (Grissino-Mayer and Fritts, 1997).

In Pakistan, dendrochronological work began in the late 80s when Ahmed (1988) presented a paper on problematic issues related to tree age estimate. Later a complete dendro-

37 chronological laboratory was developed by Prof. Dr. Moinuddin Ahmed in 2005 namely "laboratory for Dendrochronology and Plant Ecology of Pakistan in the Department of Botany at Federal Urdu University of Arts, Science and Technology Karachi. This is also my institution where I conducted my research.

3.2-Climate of Pakistan

Pakistan lies in the temperate zone bordering India to the east, Afghanistan to the west, China to the north and Iran to the south west. Northern Pakistan spans an area of 72,500 square kilometers between latitude 34o-37o and longitude 72o-78o. The silk route also known as the Karakoram Highway makes a link to China through Khunjerab Pass.

Pakistan has four seasons; cool to cold winters from December to February, dry and hot spring from March to May, southwest monsoon rainy summer season from June to August and the last autumn season that starts in September and ends in November. The coastal area along Arabian Sea is usually warm while temperature reaches even in negative in some regions of Gilgit Baltistan and some part of Northern Areas of Pakistan.

The temperature of the capital city Islamabad ranges from 2oC in January to 40oC in June. Average precipitation for July and August is about 255 millimeters and comprises the majority of the annual total 1140 mm approximately.

Baluchistan occupies 44% of Pakistan‟s land area but less population density due to scarcity of water. Again it has the hot summers usually as high as 50oC and the record-breaking temperature was 53oC in Sibi on 26th May 2010. Some cities have temperature below 0oC on average like in Ziarat and Quetta in winter.

Khyber Pakhtun Khwa (KPK) is another province of Pakistan formerly known as North West Frontier Province (NWFP) located in North West of the country and borders with Afghanistan. Naran (Kaghan) valleys, Swat valley, Kalam and Upper Dir are the areas famous for its tourism. The climate of KPK varies immensely as it mainly mountainous region. Most of the northern areas are extremely cold in winter with temperature regularly below zero. The summer is pleasant with heavy rainfall in some areas like in Swat (1200 mm approximately) and with low humidity. One of the hottest places of Asia is situated here i.e. Jacobabad while on the other hand, the northern mountains have temperate weather in the summer and intensely cold in winter.

Azad Jammu and Kashmir meet the lower area of Himalayas including Hari Parbat peak and Jamgarh peak etc. Azad Kashmir is one of the most beautiful regions of the subcontinent which receives rainfall in both summer and winter and average rainfall exceeds to 1400mm and Muzzafarabad and Pattan are considered as the wettest areas in Pakistan.

3.3-About the study sites

Ecological characteristics of sampling sites are presented in Table 3.1.

3.3.1-Gilgit

Gilgit Baltistan is one of the five provinces of Pakistan and is important for its tourism and water resources. Gilgit valley is situated at the elevation of 1,454 meters (4770 ft) with latitude and longitude 35o 55‟N and 74o 20‟E respectively Pakistan Meteorological Department (PMD), (1961-1990). The region is also famous because three Asian mountains i.e. The Himalayas, The Karakorum and The Hindu Kush ranges meet here. To the Northwest, place of interest of Gilgit valley is Kargah which lies 10km from Gilgit town. The summer season of Gilgit is brief and hot and the summer temperature may rise up to 40oC in July. The temperature in winter falls below zero. Rainfall is scanty in Gilgit averaging from 120 to 240 millimeters.

Figure 3.1shows average monthly temperature and total rainfall of nearby Gilgit station. The climatic data is short approximately extending over 50 years. Another problem with the meteorological station data is that it is away in terms of elevation from tree ring sample collection sites. Our station data suggests that highest rainfall occurs in late spring (April- May) which is also known as pre-monsoon period. Minimum rainfall occurs in November, high temperature in summer (June-August, monsoon period) and lowest temperature occurs in January.

30.0 25.0 20.0 15.0 10.0 5.0 0.0 JAN FEB MAR APR MAY JUN JUL AUG SEP OCT NOV DEC

Temperature (C) Precipitation (mm)

Fig. 3.1: Average monthly temperature in Co and rainfall in millimeter of Gilgit station based on the data period from 1955 to 2009. 3.3.2-Hunza

The territory of Hunza spans about 7900 square kilometers and borders the Gilgit river basin in the west, Afghanistan and China in north and the Shigar and Indus River Basin in the south. The major valleys in Hunza are Nilt, Nagar, Shimshal, Morkhun, Chapursan and Hanging Glacier in high Karakurum Range. Karimabad is the main town which is also a popular place for tourism with surrounding mountains of Ultar Sar, Rakaposhi, Hunza peak, Passu peak, Diran peak and Lady finger peak, all 6000 meters or higher. Maximum temperature is 27oC in May and minimum sometimes reaches up to -10OC in January.

Hunza is one of the primary destinations in Pakistan and is the centre piece of tourism in the northern region. Hunza can be divided into two regions, Lower Hunza and Upper Hunza. The former of which is also called as Central Hunza and the latter as Gojal. The Central Hunza starts from Sikanderabad leading up to Karimabad whereas Upper Hunza leads from Karimabad to Khunjerab extending all the way up to international border with China.

The Hunza valley is also mountainous in the Gilgit Baltistan and is situated to the north of Hunza River at an elevation of 2438 meters (7998 ft) with latitude and longitude of 36o 16‟N and 74o 44‟E respectively. The Hunza River joins with tributaries like Chapursan, Khunjerab, Ghujerab, Shimshal and Hisper Rivers. Nagar is the large Kingdom across the Hunza River. Passu is the important village for farmers and 15 km away from Gulmit.

Fig. 3.2: Map 1 showing the study sites from Gilgit and Hunza valleys. Yellow boxes are the sites from where samples are collected. The arrow from the second figure (Map 2) highlights the selected area from Northern Pakistan.

Table 3.1: Ecological characteristics of forest from sampling sites

S. Species Site Latitude Longitude Elevation Slope Aspect No. in meters 1 Picea Kargah 35o53 74o11 2989 34o NW smithiana 2 Picea Jutial 35o50 74o20 3250 40o N smithiana 3 Picea Haramosh 35o53 74o53 3296 53o E/S smithiana 4 Picea Bagrot 36o01 74o36 3130 45o E smithiana 5 Picea Nalter 36o02 74o35 3100 25o SW smithiana 6 Picea Chera 36o9 74o11 2900 36o N smithiana 7 Picea Chaprot 36o14 74o16 3000 35o N smithiana 8 Juniperus Chaprot 36o14 74o16 3130 45o N excelsa 9 Juniperus Nalter 36o90 74o11 2900 36o N excelsa 10 Juniperus Morkhum 36o37 74o56 3475 40o W excelsa 11 Pinus Chaprot 36o14 74o16 2850 26o N gerardiana

3.4-Purpose of the study

The formation of annual layers of wood by trees in response to the conditions within their growing season provides an alternative measure of climate on an accurately dated timescale (i.e. dendroclimatology). Its use has been crucial to the development of the now famous Northern Hemisphere “hockey stick” temperature reconstruction for the past millennium (Mann et al. 1998). With careful site and species selection, targeted areas can be recognized where tree growth is highly sensitive to available soil moisture and hence can be used to reconstruct quantified measures of climate beyond the range of instrumental and historical records. This methodology is highly robust and has been used to develop tree-ring-based drought reconstructions across the USA for the past millennium (Cook et al. 2004), and similarly for monsoon Asia (Cook et al. 2010).

In Pakistan, variation in climate significantly impact people and the economy, largely through restrictions in hydroelectric power generation and accessibility of water for human population, industry and agriculture. Understanding how past climatic variability developed and persisted is a timely scientific problem. The key northern region of Gilgit-Baltistan is particularly fragile as the Indus River passes through the region and many of its tributaries are sourced here and contribute to our national water resources. The harsh winters and a short growing season mean local agricultural productivity is vulnerable to the fluctuation in climate. The minor value of export trade from this region (Rs13.3 million, 2009) does not capture the importance of subsistence agricultural production for the 2 million people living there, or feeding the significant numbers of tourist.

The sustainable development of the region depends on knowing the full range of natural climate variability. Monitored climate records are simply too short (often <60 years) to capture the range of past conditions – a situation that is even more critical when climate predictions are attempted. This is a widespread problem often faced throughout the world and the solution adopted in several other countries, including the USA, has been to use a substitute or a proxy-climate indicator to provide the long record. Some would argue that speleotherms provide a similar high resolution proxy but clearly without the spatial resolution and high sample depth. The only suitable proxy that has been proven to be sensitive to changes in moisture supply, able to provide broad spatial coverage, has clearly-resolved annual radial growth, can be exactly dated and provide long enough records, are tree-rings. This understanding is not new (e.g. Fritts, 1976), but only really over the past decade has the

43 power of tree-ring analysis (dendroclimatology) and its well-developed statistical methods been brought to bear on the reconstruction of the joint space-time properties of past climate.

Terrible floods have recently devastated Pakistan, and for many other years the variability in climate is of grave concern and has a significant, widespread, negative impact on the economy and livelihoods. It is hard to predict the future on the basis of past 50 years climatic variations however it would be easy to predict or construct a reliable model of future climatic variation if the past climatic data is 500 years. Therefore, we can be better prepared ourselves if we know how likely climate fluctuates. This field of science (dendroclimatology) has been widely and successfully applied overseas to reconstruct past climatic history for hundreds of years longer than any instrumental records. This enabled them a much clearer understanding of the climatic pattern and helps with planning anticipated events in the future. We are using the same technique in Gilgit and Hunza valley of Northern Areas of Pakistan.

We know the selected areas have old conifer trees though being rapidly cut illegally, so this research opportunity is disappearing and some other limited studies have already proven the conifers to be sensitive to climate conditions. Therefore, here we propose to develop a network of tree-ring chronologies to investigate out the climatic history for the Gilgit and Hunza Valley areas which is the key agricultural regions in the Gilgit-Baltistan Region.

3.5-Review of Literature

3.5.1-Dendrochronology in Pakistan

The Himalayan conifer might be the oldest in the region and one Juniper marcopoda tree was reported by Bilham et al. (1983) with 1200 rings in the Hunza valley Karakurum but crossdating was not possible.

Dendrochronological work started in 1987 in Pakistan for the first time when Ahmed explained dendrochronology and its scope in Pakistan. Using tree rings, he (1988a) described population structure of some planted tree species in Quetta. Within the same year (1988b), problems which encountered in estimation of age in different tree species were also identified by him. Tree ring chronologies of Abies pindrow were presented from moist temperate Himalayan Region of Pakistan by Ahmed et al. (1989). Ahmed with Sarangzai (1991, 1992) used Juniperus excelsa and Chilghoza (Pinus gerardiana) to estimate the age and growth rates using standard dendrochronological techniques.

Juniperus excelsa from six different sites of Hunza Karakurum (Chaprot, Morkhun1, Morkhun2, Morkhun 3, Morkhun4 and Hunza) was used reconstructing modes of regional climate over the past 500 years (Esper, 2000). He observed more than thousand years (1450) year‟s old juniper trees. Inter-regional pointer years reflecting common years within Karakorum and Tien Shan (China) were also observed by Esper et al. (2001). Twenty sites were analyzed for this purpose out of which 15 from Karakurum and 5 from Tien Shan China and most concentrated species from these sites was juniper. 1300 years climatic history was analyzed for Western Central Asia in which 20 individual sites were used in the Northwest Karakorum of Pakistan by Esper et al. (2002).

Ahmed and Naqvi (2005) constructed tree ring chronologies of Picea smithiana from Himalayan Range of Pakistan. Khan et al. (2008) identified the dendroclimatic investigations of Picea smithiana from Afghanistan. In (2009) Ahmed showed some preliminary results for dendroclimatic investigation using Picea smithiana of Chera and Nalter and presented 600 years chronology. Abies pindrow from Astore and Ayubia was analysed for growth-climatic response function by Ahmed et al. (2010a). Tree ring chronologies from seven sites of Karakorum Range were constructed by Ahmed et al. (2010b). Zafar et al. (2010) described for its chronological work comparison with other sites. The species was same between two sites i.e. Picea smithiana. A collection of 28 tree ring sites from six different species i.e.

Picea smithiana, Juniperus excelsa, Abies pindrow, Pinus gerardiana, Cedrus deodara and Pinus wallichiana were shown to having dendroclimatic potential and he also explained that these species are suitable for long term climatic reconstruction (Ahmed et al. 2011) but I will describe only a few. Climate/growth correlation of the Karakorum Range was described by Ahmed et al. (2012). Dendroclimatic and dendrohydrological response of the tree species from Gilgit valleys were also presented by Ahmed et al. (in press). Recently Cook et al. (2013) reconstructed five hundred years river flow of Indus River by using tree rings.

3.5.2-Dendrochronology in China

Lot of work has been carried out in China in terms of dendroclimatology and dendrohydrology. Xiang et al. (2000) worked on ten species from three Gorges reservoirs in which five species do not show distinct ring boundaries. The other species included Cathaya argyrophylla, Cinnamomum camphora, Gordonia acuminate, Pinus massoniana and Schefflera delavayi were only 38 to 138 years long showed double and missing rings. These species were used for preliminary climate modeling and river flow. The climate modeling expressed significant correlation with current summer rainfall and summer river flow.

2326-years ring width chronology was prepared by Zhang (2003) to check out the climatic variability on the north eastern Qinghai-Tibetan Plateau using Sabina przewalskii. The average length of samples was found to be 574 years while only six samples were exceeding 1000 years. Using 13 months window from previous September to current September with one year lag effect, ring width indices indicated strong positive correlation with temperature in October of previous growth year and during May and June, it showed positive correlation with precipitation and negative with temperature.

Yu et al. (2004) reconstructed May-July precipitation in the north Helan Mountain since AD 1726. He used standard chronologies of five tree ring sites, early wood ring width, latewood ring width, total ring width, minimum early wood density, maximum latewood density and their climatic response relationship. The rainfall from May to July was reconstructed using transfer function. His precipitation showed six reconstruction periods with precipitation lower than mean and eight periods with the precipitation higher than mean and three wet intervals.

Climate response variations between male and female dioecious Fraxinus mandshurica trees were analyzed by Lushuang et al. (2010). The results obtained were of the evidence that growth pattern in two genders were similar from 1950 to 1970 but different from 1931 to

1940. The climate growth response between male and female was also different as female trees showed significance relation in November to precipitation while male trees showed significance relation to temperature in November of the previous year. The final results suggested that climatic sensitivity was different in male and female and female represented high climatic signals in comparison with male as female can bear more stress of the environment.

3.5.3-Dendrochronology in Nepal

Tree ring chronologies from Nepal were discussed by Bhattacharya et al. (1992). Twenty five sampling sites were used but only ten tree ring chronologies were crossdated. Eight species were discussed in which two species i.e. Juniper at some sites and Pinus roxburghii created problem of dating. Various chronology statistics were discussed including subsample signal strength (sss), signal to noise ratio (SNR) and expressed population signal (EPS).

Cook et al. (2003) developed 32 tree ring chronologies network from the Himalayas of Nepal and found suitable for reconstruction of temperature over the past few hundred years. He represented six indigenous tree species which were fir (Pseudotsuga), spruce (Picea mariana), hemlock (Tsuga), juniper (Juniperus excelsa), pine (Pinus) and elm (Ulmus). The result showed strongest increase in temperature of October-February season over the past 400 years.

Dendroclimatic (temperature and precipitation) investigation was held to detect climate perceptions in Langtang Central Park Nepal comprising 250 years tree ring chronological data using Abies pindrow from two sites. The result illustrated that Abies pindrow can be used for reconstruction of monthly temperature (Chhetri, 2008).

120 cores from 60 trees of Abies spectabilis from two sites i.e. Chandan bari and Cholangpati Langtang National Park were crossdated to obtain mean tree ring width, series intercorrelation and mean sensitivity. Chrononlogies were only 100-300 years old and negatively correlated with minimum monthly temperature and positively correlated with total monthly precipitation (Chettri et al. 2010).

3.5.4-Dendrochronolgy in India

Yadav and Singh (2002) demonstrated the dendroclimatic potential of Taxus buccata from Western Himalayan, India. They developed 345 years ring width chronology and described

47 the indirect correlation of tree growth with pre-monsoon temperature. They also found out a significant correlation of yew (Taxus buccata) with Abies pindrow chronology.

Cedrus deodara from two sites of Western Himalaya by Pant et al. (2000) were subjected to densitometric and Response function analysis. Data was obtained from densitometric analysis for earlywood, latewood, minimum, maximum and mean densities and total ring width. Response Function analysis was used that indicated significant relationships between pre- monsoon (March, April, May) and also pre-monsoon climate reconstruction was established using these two species.

Borgaonkar et al. (2009) presented 458-year tree ring chronology of Himalayan cedar from three high elevation sites of Western Himalaya (India) in relation to climate and glacier fluctuations. Dendroclimatic investigations showed significant positive relationship of tree ring index with winter i.e. December-February and summer precipitation and indirect relationship with summer temperature while in case of past glacial fluctuation records, suppressed and released growth pattern in tree ring chronology was noticed, explaining the rapid retreat of Himalayan glaciers.

A long term rainfall reconstruction of 694-years was established using Pinus gerardiana and Cedrus deodara from Himachal Pardesh, India by Singh et al. (2009). He developed a correlation of January-February precipitation of (AD 1310-2005) years concluding that these months have direct relationship with growth of these species. He also developed reconstruction of March-July precipitation and explains 46% of variance showing 20th century was the wettest and 18th century was the driest period.

In the same way, Cedrus deodara from 11 moisture stressed sites from monsoon shadow zone of the Western Himalaya, India were used to develop chronology back to AD1353 and showed direct relationship with March, April, May, June (MAMJ) precipitation. The data obtained from reconstruction explained drought in fifteen and sixteen centuries and MAMJ precipitation over monsoon shadow zone was directly related to El Nino Southern Oscillation (ENSO) (Yadav, 2011).

Chapter 4

Chronology development

4.1-Introduction

This chapter describes the development of new tree-ring chronologies of Picea smithiana, Juniperus excelsa and Pinus gerardiana in the study area. I present a brief site description of standard field methods and laboratory techniques that were applied to collect and prepare the samples. Tree ring sequences are identified and crossdated by a combination of visual and computer-aided techniques. Standardization pursues to remove non-climatic signals from raw tree ring data followed by the selection of suitable standardization method, which plays an important role in dendroclimatic research. Standardization techniques using negative exponential curve or linear curve are investigated and their statistics are discussed. Finally, chronologies are compared by means of correlation analysis and multivariate analysis including cluster analysis and Principal component analysis.

4.2-Materials and methods

Sampling was carried out from eight sites from which eleven chronologies were produced. Field sampling was conducted in the month of June and July as in these months, the sites are easily accessible.

4.3-Field Methods

For the selection of sites, high elevations were targeted because rings of trees were expected to be most sensitive to (i.e., limited by) climate at such locations. The Swedish increment borer was used for the collection of all samples from living trees. On average, 15 trees from each site were sampled from the following species:

1. Picea smithiana from Kargah, Jutial, Haramosh, Bagrot, Nalter, Chera and Chaprot, 2. Juniperus excelsa from Chaprot, Nalter and Morkhun 3. Pinus gerardiana from Chaprot

Those trees were selected having high DBH (diameter at the breast height) with the assumption of direct relationship between DBH and age (sensu Fritts, 1976, Schweingruber et al. 1990)., Two cores from each tree were collected from opposite sides of the trees but in the

49 case of Juniperus excelsa, two to three radii were taken. DBH of the trees were measured using DBH tape. Injuries and branches were avoided following the methods of Stokes and Smiley (1968).

360 cores from living trees were collected from eleven sites with the lowest elevation at 2850m and highest elevation from 3475m. Pinus gerardiana was existent at lower elevation whereas Juniperus excelsa species were cored at the highest elevation among all sites (Table 3.1). Most of the sites situated at northern aspect with the minimum and maximum slope of 25o to 53o respectively (Table 3.1).

Some snaps taken from the forest of the study area

Picea smithiana forest from Kargah

Picea smithiana from Bagrot

Picea smithiana from Haramosh

Picea smithiana from Nalter

Juniperus excelsa from Nalter

4.4-Laboratory preparation

In laboratory; the cores were air dried for two days for further processing. These dried cores were mounted on wooden groove with the help of water soluble glue and were fixed with masking tape and again left for drying for 48 hours. Each core at the time of mounting was given an ID, date of collection, species and site name. Sanding machine with papers of different grits was used for surfacing the rings following Orvis and Grissino-Mayer (2002).

4.4.1-Surfacing and crossdating

After mounting, the next step was to count the rings from bark to pith and to assign calendar years under powerful microscope using skeleton plot method followed by Stokes and Smiley (1968). First those cores were selected whose outside ring was known means that year of collection were dated from outside to pith. Narrow and wide rings were marked on the skeleton plot and most narrow rings in the whole stand were circled as pointer years. One dot was marked after every ten years, two dots after every fifty years and three dots after every century using lead pencil. Pattern of narrow and wide rings of one core was matched by the other core of the site. This way, visual crossdating was achieved.

4.4.2-Measurement using Velmex

The ring‟s widths of crossdated cores were measured in millimeter using measure J2X. The identity was given using the criteria SSSSTTC. The first two SS stands for species, the next two SS stands for sites, TT stands for tree number and C represents the core number like in case of Juniperus excelsa from Nalter (JENL101) JE describes species name: Juniper excelsa, NL shows site Nalter: 10 represents that this is the 10th tree of the stand and 1 quantifies that this is core number 1 of the two or three. Black mark on the monitor screen was used to calculate the values by measuring the distance travelled between two successive rings. The values were stored numerically by the program itself. The program started counting from bark to pith one by one. Ring widths were measured to 0.001 mm accuracy selected from the program menu.

4.5-Software’s used in the analysis

Following software were used in overall analysis

 Velmex Measure J2X

 COFECHA Richard Holmes (1934-2003).  DPL (Dendrochronological Program Library) is the package program containing 36 different programs written by Richard Holmes (1934-2003).  ARSTAN (Cook et al. 1986)  Minitab (version 11.12)

4.5.1-COFECHA

The raw ring width measurement taken in millimeter was subjected to COFECHA (Holmes et al. 1994; Grissino-Mayer 2001) to check the quality of crossdating. Default commands were followed with 32 year cubic spline 50% wavelength cutoff for filtering; 50 year segment length with 25 year lagged and 99% confidence interval with 0.3281 critical level of correlation value to incorporate the results. COFECHA embodies seven parts; part one describes title page, options selected, summary and absent rings; part two tells graphical representation in the form of Histogram; part three shows master series with samples depth and absent rings by year; part four demonstrates Bar plots of master dating series; part five illustrates correlation of each series by master series; part six represents potential problems including low correlation, divergent year to year changes absent rings and outliers. The second part of COFECHA is of much importance and signifies the following results;

 Number of dated series which tells how many samples in a stand is crossdated.  Master series which tells the longest crossdated core in the whole series.  Total rings and total dated ring in the whole stand.  Series intercorrelation which shows how much pattern of rings is similar or dissimilar to one another.  Average mean sensitivity is the measure of relative differences in widths between two adjacent rings.  Flags which are the source of problems in crossdating.

Here, we adopted an approach where we concentrated first, second and fifth parts to explain our results.

4.5.2-Chronology development

Dendrochronologists don‟t use ring width measurement to find past climatic variations as climatic signals in tree ring widths are small so these signals must be enhanced by indexing procedure. Mean chronology of a given site can be obtained by averaging the indices of many trees. The random non climatic noise caused by any measurement errors cancel one another and signal to noise ratio is enhanced. If there is greater climatic variations among the sample ring width we require small number of cores to extract signal-to-noise ratio.

4.5.3-ARSTAN

Software ARSTAN was used to transfer cross-dated raw data to develop standardized chronology. We developed master chronology through first deterending method include Standard, Residual and Arstan chronologies. The residual chronology, with mean index value of 1 removes autocorrelation and strengthens exogenous signals. Sample depth is included in graphs to note where sample size begins to decrease substantially.

ARSTAN stands for Auto Regressive STANdardization. Trends (systematic changes) in the trees were removed from the software ARSTAN followed by Cook (1985). Deterending with a cubic spline of 32 years was adopted for standardization to minimize the loss of low frequency signal in the series using Arstan code (Holmes, 1992; Cook et al. 1986) and ring width index of each sample was obtained by dividing raw ring width value with corresponding smoothed value. The standardization minimizes the unwanted information known as noise and maximizes the required variation explained by Cook and Holmes (1986) and three chronology values were obtained i.e. raw chronology, standard chronology and residual chronology. The raw ring width chronology is just averaged non standardized raw data. The residual version of chronology is produced by autoregressive modeling of the detrended measurement series and chronology is averaged to standardized value (mean index=1). Robust mean value function produces chronology with strong common signal and without persistence. If there is no autoregressive modeling, standard chronology is produced. The pooled autoregression is reincorporated into residual version to produce Arstan chronology (Holmes, 1994).

Chronology development

Residual Standard Arstan chronology chronology chronology

-lag +lag +- lag

Hierarchy shows chronologies are divided into three i.e. residual, standard and arstan chronologies.

 Residual chronology with no lag (previous) year effect  Standard chronology with lag year effect  Arstan chronology with some lag year effect

4.5.4-Chronology statistics

ARSTAN describes the following statistics; statistics of raw tree-ring measurement, statistics of standard tree ring measurements, statistics of residual tree ring measurement and statistics of arstan tree ring measurements. First year, last year, total years, mean index, standard deviation, skewness coefficient, kurtosis coefficient, mean sensitivity and series correlation are common results in these four chronology statistics. We also used the auto and partial autocorrelation up to back ten years using 95% confidence interval (t-1 to t-10). The minimum common year period among all samples is mentioned (from starting year to ending year). Besides all other statistics, four main statistics are much important including Rbar, SNR, EPS, and SSS (Cook and Kariukstis, 1990) which are also further discussed.

According to Briffa and Jones (1990), Rbar is the average correlation between all possible series and was calculated for 50 years windows lagged by 25 years. It is an indication of common variance and is independent of sample size. The EPS partly depends on sample size, measures how well the finite chronology compares with a theoretical infinite population (Wigley et al. 1984). Its values range from zero to one with no test for a threshold level of the statistics however Wigley et al. (1984) recommended the value of 0.85 for a threshold. Subsample Signal Strength is the measure of a subset of index time series which describes the chronology of a larger set of index time series and is measured as the quotient of EPS values of a subset and reference sample.

Whereas Rtotal is the average correlation between corresponding time interval of index time series, Rwithin is the average correlation between corresponding time interval of index time series from different cores taken from the same tree, Rbetween is the average correlation between corresponding time interval of index time series from different cores taken from different trees and Reffective is the effective correlation coefficient describes between and within cross section signal.

4.6-Results

4.6.1-Crossdating of all sites

On average, 67.7% of all collected samples were crossdated. Picea smithiana from all seven sites showed good crossdating as compared to Juniperus excelsa from three sites. Crossdating was found least successful with the samples of Juniperus excelsa from Nalter, where only 30% of cores were forwarded for the chronology construction. On average, high rejection rate (30%) was caused due to lack of tree ring pattern which mean to say that there were too many missing or false rings that confound our ability to identify the pattern of narrow and wide rings that allows for crossdating. However, relatively poor physical quality of the samples could not be observed. Several cores were too short that they were rejected from the stand or some of tree samples, rings were too narrow to be measured. Trees from few sites like Picea smithiana from Jutial and Nalter, nearly all cores were crossdated.

Crossdating success was accomplished with high replication i.e. every site consisted of 23 radii from 15 trees. The minimum 12 radii from 20 trees were observed in Juniperus excelsa from Nalter. Rest of the (uncrossdated) samples from this site is preserved to crossdate in the future due to shortage of time. Another factor that gives the additional support to the better achievement of crossdating was the low occurrence of missing rings i.e. only 0.99% of all crossdated rings were absent in the radii.

Portions of the time series with two or more series were found to be 445 years on average with the highest portions presented in Picea smithiana from Nalter, Chera and Haramosh respectively. Picea smithiana from other three sites (Jutial, Bagrot and Chaprot) were of more than 400 years and least year portion was observed in case of Picea smithiana from Kargah. Juniperus excelsa from three sites have the portion more than 300 years. The shortest series (212 years) was found in Pinus gerardiana Chaprot. Most of the missing rings were narrow in the years 1971 and 1917, and were nearly absent in all species.

Series intercorrelation and mean sensitivity are the projections of year to year variability in the chronology (Fritts, 1976). Individual series correlation occurred in Picea smithiana Jutial and Haramosh which means every core of the site showed good correlation with the other samples. The lowest individual series correlation was obtained in Juniperus excelsa and Picea smithiana from Chaprot which indicates that Chaprot site has the minimum correlation. But Pinus geradiana from the same site reversed the results i.e. good individual series

58 correlation. The results showed that different species from the same site expressed different correlation among all the cores. Picea smithiana from the other sites stated good correlation ranging from 0.479-0.875.

Table 4.1: Summary statistics of species from eleven sites collected from COFECHA. 1 = percentage of core samples that were crossdated; 2 = the portion of the time series with two or more series; 3= individual series correlation; 4 = individual mean sensitivity; 5 = highest correlation with 50 years dated segment; 6= lowest correlation with 50 years dated segment; 7 = mean measurement of rings; 8 = percentage of missing rings

Site 1 2 3 4 5 6 7 8 PSKAR 70% 367 0.47-0.79 0.19-0.30 0.75 0.20 0.96 0.053% PSJUT 90% 479 0.72-0.96 0.24-0.40 0.98 0.89 0.90 0.177% PSHAR 67% 520 0.74-0.88 0.27-0.42 0.94 0.75 0.60 0.010% PSBAG 67% 460 0.51-0.81 0.22-0.39 0.95 0.62 0.74 0.037% PSNLT 90% 601 0.49-0.77 0.15-0.32 0.75 0.36 0.83 0.076% PSCHR 60% 596 0.38-0.87 0.15-0.43 0.80 0.52 0.96 0.231% PSCHP 57% 496 0.27-0.58 0.20-0.27 0.67 0.34 0.96 0.021% JECHP 73% 339 0.11-0.85 0.21-0.35 0.71 0.35 0.93 0.213% JENLT 30% 378 0.37-0.65 0.19-0.32 0.73 0.50 0.68 0.021% JEMOR NIL 495 0.01-0.16 0.37-0.55 0.31 -0.9 0.53 0.023% PGCHP 73% 212 0.61-0.83 0.22-0.44 0.67 0.34 0.97 0.228%

PSKAR= Picea smithiana Kargah, PSJUT=Picea smithiana Jutial, PSHAR=Picea smithiana Haramosh, PSBAG= Picea smithiana Bagrot, PSNLT= Picea smithiana Nalter, PSCHR= Picea smithiana Chera, PSCHP= Picea smithiana Chaprot, JECHP= Juniperus excelsa Chaprot, JENLT= Juniperus excelsa Nalter, JEMOR=Juniperus excelsa Morkhun, PGCHP= Pinus gerardiana Chaprot, Results suggest that climate does not significantly limit tree growth at some sites. An indication of how climate is the limiting factor for the growth of trees is obtained from mean sensitivity, which is a measure of year to year variability in ring-width (Fritts, 1976). Values range from zero (no change from one ring to the next) to a value of two indicating highest sensitivity exists in tree samples that might be climate or other factors cause year to year changes (Fritts, 1976). Here in this study, the individual mean sensitivity values reveal low variability. The highest individual mean sensitivity values occurred in Picea smithiana Haramosh (0.27-0.42) while the least was in Juniperus excelsa Nalter (0.19 to 0.32). It means

59 individual core of Juniperus excelsa from Nalter has the less sensitivity to climate. Picea smithiana from first four sites exhibited good individual sensitivity.

The mean ring width was 0.82 mm (millimeter) with the fastest growing trees unsurprisingly from the lowest altitudinal site (Pinus gerardiana from Chaprot) where the mean measurement for all growth rings was 0.97 mm (Table 4.1).

On average, the highest correlation of 50-years dated segments with 25 years lagged was 0.79 while the lowest was 0.48 calculated from all sites (Table 4.1). The highest correlation among segments occurred in 17th and 20th century whereas the lowest correlation among segments happened in 19th century. The results indicated that 17th and 20th century among all sites and all species crossdated well while on the other hand 19th century proved low occurrence of crossdating among all sites and all species. The present study stated that except for Juniperus excelsa from two sites (Nalter and Morkhun), good crossdating occurs between two cores collected from the same tree, among the same species growing at the different sites, between the different species growing at the same sites and even different species growing at different sites. Juniperus excelsa from Morkhun did not show reliable results for further investigation and was therefore rejected.

0.95 0.9

0.85 0.8 0.75 0.7 0.65

SeriesIntercorrelation 0.6 0.55 0.5 20 25 30 35 40 45 50 55 Slope Rank

Fig. 4.1: Dependence of series intercorrelation on site slope. The red circle shows the slope from 25o to 35o and the blue circle represents the slope ranged 40o to 55o.

Six sites were situated at the lower steep slope whereas four sites were sampled from higher steep slope (Fig. 4.1). The lower values of series intercorrelation occurred from 25o to 35o while the higher values happened from 40o to 55o (Fig. 4.1). It is also apparent from the Figure that series intercorrelation values at lower steep slope range 0.55-0.75 and in case of higher steep slope, it starts from 0.65 and ends at 0.92. It means series intercorrelation is also affected by steepness of site (Fenwick, 2003). Trees growing on steep slopes produced ring- width pattern common to more individuals within a site which are designated by higher series intercorrelation values.

Table 4.2: Negative (narrow) pointer years from ten sites of Gilgit and Hunza valleys. Plus sign indicates the presence of pointer year in the site.

Year PSKAR PSJUT PSHAR PSBAG PSNAL PSCHR PSCHP JECHP JENLT PGCHP Similar sites 2001 + + + + + + + + + 9 1985 + + + + + 5 1974 + + + + + 5 1971 + + + + + + 6 1961 + + + + + 5 1947 + + + + + + + 7 1917 + + + + + + + 7 1877 + + + 3 1865 + + + + 4 1810 + + + + + 5 1802 + + + + + 5 1785 + + + + 4 1742 + + + + + 5 1717 + + + + 4 1707 + + + + + 5 1701 + + + + 4 1626 + + + + + + 6 1603 + + + + + 5 1602 + + + + + 5 1574 + + + + 4 1573 + + + + 4 1572 + + + + 4 1492 + + + 3

Table 4.3: Positive (wide) pointer years from ten sites of Gilgit and Hunza valleys. Plus sign indicates the presence of pointer year in the site.

Year PSKAR PSJUT PSHAR PSBAG PSNAL PSCHR PSCHP JECHP JENLT PGCHP Similar sites 2005 + + + + + 5 1994 + + + + + 5 1981 + + + + + 5 1973 + + + + + 5 1960 + + + + 4 1958 + + + + + + 6 1945 + + + + + 5 1924 + + + + + 5 1906 + + + + 4 1883 + + + + + + + 7 1834 + + + + 4 1826 + + + + 4 1804 + + + + + 5 1766 + + + + + 5 1748 + + + + + 5 1747 + + + + + 5 1661 + + + + 4 1621 + + + + + 5 1596 + + + + 4

Negative pointer years (narrow rings) from all ten sites are described in Table 4.2. Most of the pointer years are similar in case of Picea smithiana with the higher degrees in Jutial, Haramosh, Bagrot and Chera. Juniperus excelsa and Pinus gerardiana from Chaprot showed the least percentage of pointer years. Juniperus excelsa from Nalter explained the lowest resemblance of pointer years among all sites. The most evident year was experimented 2001 which was seen in almost all chronologies. 1971, 1947, 1917 and 1626 years also revealed strong occurrence too.

In the same way, positive pointer years (wide rings) were also found similar in case of Picea smithana as compared to other species (Table 4.3). The highest common positive pointer years were witnessed in Picea smithiana from Chera. Picea smithiana from Jutial, Haramosh and Chaprot presented close resemblance. The most evident years among all chronologies were 1883 which was seen in seven sites followed by 1958.

Overall, similar pointer years were experimented in current study, due to similar climatic conditions that limit the tree growth. Out of three species, Picea smithiana from every site and from different exposures showed close resemblance. So it appears that ring-width pattern was affected by variation in temperature because narrow rings were formed during the below average temperature.

4.7-Chronology development

Chronologies from all ten sites are presented in Figs. 4.1a – 4.10a. Each Figure is made up of five plots: the raw chronology (which is the biweight robust mean of crossdated ring-width measurements); standard, residual and arstan chronologies respectively (i.e. based on single detrended data) and sample depth (which explains how number of series changes with respect to time). The most important feature of the chronology plots is the close resemblance among all the raw chronologies.

In most of the chronologies, noticeable period of above average growth occurred in 16th century (i.e. up to 1600). The period of below average encountered during the last century (1900-2000) in almost all chronologies. It indicates that many sites show increased growth in 16th century or it may point to the fact that growth of trees were better during early growing season where as sites show decreased growth in 20th century where the growth was declining. Another factor may be the climate because surprisingly all the chronologies exhibited same above and below average growth during 16th and 20th century respectively so it points to favorable conditions for growth of trees during 16th century and unfavorable conditions during 20th century.

In the present study it was analyzed that how many cores attained the age more than 300 by making the mean among all chronologies. On average, it was observed that every site contained 10 cores attaining the age more than 300 years. So every species at every site, large number of crossdated cores was found to be more than 300 years. It means that all species forests were as old as 300 years and can tell the history of past 300 years climate. However,

Picea smithiana from some sites got the age more than 500 years. These results were also confirmed by taking average of sample size that covers the area of 300 years (Fig. 4.1a-4.10a) where we got thirteen cores from all chronologies more than 300 years on average.

Fig. 4.1a: Picea smithiana Kargah chronology plots. Five figures representing raw, standard, residual, arstan chronologies and sample depth respectively

Fig. 4.2a: Picea smithiana Jutial chronology plots. Five figures representing raw, standard, residual, arstan chronologies and sample depth respectively

Fig. 4.3a: Picea smithiana Haramosh chronology plots. Five Figures representing raw, standard, residual, arstan chronologies and sample depth respectively

Fig. 4.4a: Picea smithiana Bagrot chronology plots. Five Figures representing raw, standard, residual, arstan chronologies and sample depth respectively

Fig. 4.5a: Picea smithiana Nalter chronology plots. Five figures representing raw, standard, residual, arstan chronologies and sample depth respectively

Fig. 4.6a: Picea smithiana Chera chronology plots. Five figures representing raw, standard, residual, arstan chronologies and sample depth respectively

Fig. 4.7a: Picea smithiana Chaprot chronology plots. Five figures representing raw, standard, residual, arstan chronologies and sample depth respectively

Fig. 4.8a: Juniperus excelsa Chaprot chronology plots. Five figures representing raw, standard, residual, arstan chronologies and sample depth respectively

Fig. 4.9a: Juniperus excelsa Nalter chronology plots. Five figures representing raw, standard, residual, arstan chronologies and sample depth respectively

Fig. 4.10a: Pinus gerardiana Chaprot chronology plots. Five figures representing raw, standard, residual, arstan chronologies and sample depth respectively

4.7.1-EPS and Rbar

Positive autocorrelation was observed within the chronologies. It was also clearly obtained between different chronologies while on the other hand; negative autocorrelation was seen among chronologies but less than positive autocorrelation.

The output file obtained from program ARSTAN (Fig.4.1b-4.10b) describes the running EPS and Rbar as an overview of the reliability of chronology with respect to time for Picea smithiana, Juniperus excelsa and Pinus gerardiana. All chronologies were found reliable for reconstruction up to 300 years on average. However, Picea smithiana from four sites Jutial, Haramosh, Bagrot and Nalter showed suitability of samples up to 400 years. The lowest reliability among all sites was observed in Pinus gerardiana Chaprot i.e. only 160 years. Picea smithiana from Kargah, Chaprot and Juniperus excelsa from Chaprot satisfied the threshold limit of EPS.

EPS and Rbar looks weak in the case of Picea smithiana from Chera before late 1600s (Fig. 4.6b) clearly showing that crossdating is not correct up to 1600. Juniperus excelsa from Nalter too looks weak with clear dating issue (Fig. 4.9b). It is suggested that collection of more samples from these two sites will be helpful to produce better chronology.

Fig. 4.1b: Running Rbar and EPS graph of Picea smithiana from Kargah

Fig. 4.2b: Running Rbar and EPS graph of Picea smithiana from Jutial

Fig. 4.3b: Running Rbar and EPS graph of Picea smithiana from Haramosh

Fig. 4.4b: Running Rbar and EPS graph of Picea smithiana from Bagrot

Fig. 4.5b: Running Rbar and EPS graph of Picea smithiana from Nalter

Fig. 4.6b: Running Rbar and EPS graph of Picea smithiana from Chera

Fig. 4.7b: Running Rbar and EPS graph of Picea smithiana from Chaprot

Fig. 4.8b: Running Rbar and EPS graph of Juniperus excelsa from Chaprot

Fig. 4.9b: Running Rbar and EPS graph of Juniperus excelsa from Nalter

Fig. 4.10b: Running Rbar and EPS graph of Pinus gerardiana from Chaprot

4.7.2-Autocorrelation and partial autocorrelation

Autocorrelation is the serial dependence of the observations in time series and is calculated by autocorrelation function (ACF) and partial autocorrelation function (PACF). Autocorrelation function tells measures the correlations between observations at different times a part (lags) while the partial autocorrelation describes the autocorrelation at different lags by allowing the effects of autocorrelation at intermediate lags. According to Brown and Rothery (1993), Both ACF and PACF are used to select Autoregressive models that best describes the time series.

Autocorrelation properties of all chronologies from ten sites were investigated by calculating the autocorrelation coefficients (ACs) and partial autocorrelation coefficients (PACs) for the first ten lags i.e. one lag being equal to one year (Figs. 4.11). The structure of autocorrelation is quite similar across all sites. In all chronologies there was continuous drop from lag 1 to 10. All autocorrelations were found to be positive except it was pronounced to be negative in 4 lag in case of Pinus gerardiana Chaprot. The PACFs for the chronologies showed the predominance of AR (1) which is in agreement with the models selected by the ARSTAN program. The first two lags were found to be high in all chronologies for PACFs.

Picea smithiana from Kargah

Picea smithiana from Jutial

Picea smithiana from Haramosh

Figs. 4.11: The autocorrelation coefficients (AC) and partial autocorrelation coefficients (PAC) values were calculated out to 10 lags. Red lines indicate 95% confidence interval. (To be continued.,)

Picea smithiana from Bagrot

Picea smithiana from Nalter

Picea smithiana from Chera

Figs. 4.11: The autocorrelation coefficients (AC) and partial autocorrelation coefficients (PAC) values were calculated out to 10 lags. Red lines indicate 95% confidence interval. (To be continued.,)

Picea smithiana from Chaprot

Juniperus excelsa from Chaprot

Juniperus excelsa from Nalter

Figs. 4.11: The autocorrelation coefficients (AC) and partial autocorrelation coefficients (PAC) values were calculated out to 10 lags. Red lines indicate 95% confidence interval. (To be continued.,)

Pinus gerardiana from Chaprot

Figs. 4.11: The autocorrelation coefficients (AC) and partial autocorrelation coefficients (PAC) values were calculated out to 10 lags. Red lines indicate 95% confidence interval. Summary of COFECHA and ARSTAN statistics is shown in the Tables (4.4, 4.5) respectively. The mean segment length produced by COFECHA program was 296 years (Table 4.4) with values ranging from 167 years (Pinus gerardiana Chaprot) to 375 years (Picea smithiana Nalter). Most of the crossdated cores were established from Picea smithiana of Jutial i.e. 36 cores from 20 trees. The oldest living tree (619 years) belonged to Picea smithiana from Nalter. The series intercorrelation (the mean correlation with master chronology produced by COFECHA program) was 0.693 on average.

Mean EPS value from all sites was obtained 0.94 and was much higher from the threshold value (EPS>0.85). There was a remarkable increase in SNR value in Picea smithiana Jutial indicating that the agreement between time series (i.e. the strength of crossdating) is better which may or may not be related to climate. Rbar within the trees was higher for the single detrended data (mean Rbar within trees= 0.70) whereas mean Rbar values and mean Rbar between the trees remained nearly equal (mean Rbar= 0.50; mean Rbar between the trees= 0.48) (Table 4.5).

Much strongest common signal between trees was found in Picea smithiana of Jutial. This chronology also has the highest value of mean sensitivity with the smallest first order autocorrelation (Table 4.4). The next strongest common signal was observed in Juniperus excelsa Chaprot and Picea smithiana at Haramosh. Both these two chronologies have also strong mean sensitivity and low autocorrelation. This may be due to occurrence of all three sites on a very steep slope. The weakest mean sensitivity and highest first order autocorrelation was seen in Picea smithiana Chaprot chronology (Table 4.4). It is clear from

86 the Tables (4.4 and 4.5) Picea smithiana from Jutial has highest values of mean sensitivity, EPS, SNR, mean correlation among all radii (Rbar), mean correlation within trees, mean correlation between trees and percentage variance in first eigen value. These all values define the climatic signal in a chronology. The higher the values, the greater will be the climatic signals.

Table 4.4: Summary of COFECHA statistics

S. Species & Cores∕trees 1st order Series Mean Max period Mean Common Cross No site name autocorrelation Intercorrelation sensitivity In years length of years dated series cores 1 PCSM 30∕15 0.55 0.671 0.241 1475-2008 269 1908-2008 21 Kargah (534) (101) 2 PCSM 40∕20 0.50 0.913 0.358 1523-2008 360 1837-2008 36 Jutial (486) (172) 3 PCSM 30∕15 0.51 0.849 0.319 1467-2009 339 1760-2009 20 Haramosh (543) (250) 4 PCSM 30∕15 0.63 0.735 0.304 1480-2009 273 1870-2004 20 Bagrot (530) (135) 5 PCSM 30∕15 0.39 0.720 0.278 1394-2005 360 1797-2005 18 Chera (612) (209) 6 PCSM 40∕20 0.61 0.636 0.215 1387-1986 375 1800-1986 35 Nalter (619) (186) 7 PCSM 30∕15 0.71 0.562 0.229 1520-2008 276 1870-2008 17 Chaprot (489) (139) 8 JUEX 30∕15 0.54 0.549 0.310 1670-2008 249 1887-2008 22 Chaprot (338) (121) 9 JUEX 40∕20 0.67 0.563 0.236 1676-2009 287 1828-2009 12

Nalter (332) (181) 10 PIGE 30∕15 0.69 0.735 0.289 1737-2008 167 1895-2009 22 Chaprot (271) (114) PCSM= Picea smithiana, JUEX= Juniperus excelsa, PIGE= Pinus gerardiana, SD= standard deviation

Table 4.5: Summary of Arstan statistics

S. Species & (EPS) (SNR) Rbar Within Between Common Eigen Percent Commulative No site name trees rbar trees rbar period value#1 variance variance for first five PCs 1 PCSM 0.954 20.681 0.508 0.776 0.497 100 yrs 9.543 47.7% 83.1% Kargah 2 PCSM 0.993 148.329 0.805 0.893 0.802 100 yrs 28.560 79.3% 91.2% Jutial 3 PCSM 0.973 36.134 0.707 0.845 0.698 100 yrs 10.470 69.8% 94.1% Haramosh 4 PCSM 0.913 10.522 0.429 0.897 0.413 100 yrs 5.684 40.6% 86.2% Bagrot 5 PCSM 0.945 17.13 0.488 0.670 0.476 100 yrs 9.959 55.3% 80.9% Chera 6 PCSM 0.951 19.37 0.564 0.717 0.554 100 yrs 8.951 59.7% 81.2% Nalter 7 PCSM 0.889 8.05 0.335 0.371 0.333 100 yrs 6.42 40.8% 89.0%

Chaprot 8 JUEX 0.929 13.033 0.275 0.331 0.270 100 yrs 10.142 40.8% 75.4% Chaprot 9 JUEX 0.914 10.578 0.281 0.715 0.266 100 yrs 8.517 31.5% 58.9% Nalter 10 PIGE 0.963 26.132 0.543 0.775 0.535 100 yrs 12.430 56.5% 85.9% Chaprot

4.8-Chronologies comparison

Fig. 4.12 shows the chronologies comparison from ten sites. Only the outer two hundred years that all chronologies shared were extracted for comparison (PGCHP; EPS >0.85 just only 160 years). From this Figure it is clear that 1802, 1810, 1865, 1917, 1944, 1971, 1985 and 2001 years were narrow in all chronologies. The results were also confirmed by making inter site comparison of tree-ring index chronologies, developing a correlation matrix over the same common two hundred years as shown in Table 4.6. It is checked whether all these chronologies were inter-correlated with one another or not. All the species were tested to find out the positive or negative correlation.

Table 4.6: Correlation matrix of all chronologies values from ten sites

PSKAR PSJUT PSHAR PSBAG PSNAL PSCHR PSCHP JECHP JENAL PSJUT 0.614 *** PSHAR 0.595 0.641 *** *** PSBAG 0.478 0.737 0.56 *** *** *** PSNAL 0.402 0.288 0.232 0.294 *** *** ** *** PSCHR 0.487 0.602 0.606 0.678 0.462 *** *** *** *** *** PSCHP 0.344 0.295 0.203 0.398 0.247 0.342 *** *** ** *** *** *** JECHP 0.383 0.293 0.459 0.189 0.009 0.155 0.067 *** *** *** * JENAL 0.206 -0.01 0.385 -0.033 -0.107 -0.052 -0.108 0.469 ** *** *** PGCHP 0.383 0.539 0.289 0.456 0.012 0.399 0.248 0.222 -0.146 *** *** *** *** *** ** **

The period of analysis is from 1800 to 2000 while the asterisks below each value indicate its level of significance: (i.e. ***P<0.001; **P<0.01; *P<0.05). All sites are positively correlated and significant at (p<0.05, 0.01and 0.001) with the exception of Juniperus excelsa from Nalter which expressed a negative relationship with some Picea and Pinus chronologies. The highest correlation (0.737) was recorded between PSBAG and PSJUT followed by PSBAG and PSCHR (0.678). The Juniperus excelsa Nalter site showed no relationship with five sites of Picea

91 smithiana (Jutial, Bagrot, Nalter, Chera and Chaprot) however, it was found significantly correlated with PSKAR (P<0.01) and PSHAR (P<0.001). A good correlation was also observed between the two Junipers (P<0.001).

1600 1400 1200 1000 800 600 400 200 0 1800 1810 1820 1830 1840 1850 1860 1870 1880 1890 1900 PSKR PSJL PSHR PSBG PSNL PSCH PSCR JECH JENL PGCH

2000 1800 1600 1400 1200 1000 800 600 400 200 0 1900 1910 1920 1930 1940 1950 1960 1970 1980 1990 2000 2010

Fig. 4.12: Two graphs show 200 years chronology similarities among ten sites. Arrows indicates the pointer years among all sites.

4.9-Multivariate analysis

Multivariate statistical analyses were employed to better understand the similarities and ecological status of the studied system (May et al. 2006; Hayal et al. 2009; Pejman et al. 2009). Cluster analysis (CA) and principal component analysis (PCA) were performed among the ten chronologies and the characteristics of three groups derived from these analyses are presented in Fig. 4.13.The highest similarity level was observed among first three sites, and the lowest was exhibited in Juniperus excelsa Nalter (not shown), which seems to be an outlier. Perhaps this might be due to Nalter is dry temperate site and Juniperus excelsa and Picea smithiana are the characteristics species of this area however, Juniperus excelsa in this valley are growing on a slope where soil moisture is much better due to the permanent snow on the top therefore, this site is different from others. Picea smithiana from Jutial and Chaprot formed a separate cluster. Second cluster was formed among four chronologies in which Picea smithiana from Haramosh and Bagrot, and Juniperus excelsa from both sites showed similarities. A third cluster was apparent among four chronologies Picea smithiana from Kargah, Nalter and Chera while the fourth cluster was comprised of Pinus gerardiana from Chaprot.

The PCA confirmed the results of the cluster analysis as shown in Fig. 4.14, where the first three PCs (principal components) exceed values of 1.0, and therefore show the reliability of PCA (Shrestha and Kazama, 2007). According to the eigen value criterion, only PC‟s that exceed 1.0 are considered as significant, and the Kaiser criterion explains that only first five factor groups could be used because successive eigen values are less than one. PC1 resulted positive with the highest value of Picea smithiana Jutial (0.41) whereas the lowest value was seen in Juniperus excelsa Nalter (0.061) further indicating its outlier status. Four clusters were observed (Fig.4.14): the first grouping; Picea smithiana from Jutial, Chaprot, Bagrot and Haramosh; a second grouping Picea smithiana from Kargah and Nalter. Although being different species, Picea smithiana Chera and Pinus gerardiana from Chaprot formed a third, while Juniperus excelsa evidenced similar results in this analysis as observed in cluster analysis by forming separate clusters. Hence, the final conclusion from multivariate analysis (CA and PCA), is the construction of four groups among ten chronologies (Figs. 4.13; 4.14) having close resemblances i.e. 1- Picea smithiana (Jutial and Chaprot), 2- Picea smithiana (Haramosh and Bagrot), 3- Picea

94 smithiana (Nalter and Chera) and 4- Juniperus excelsa (Chaprot and Nalter) with the correlation values of 0.295, 0.560, 0.462 and 0.469 respectively.

Fig. 4.13: Dendrogram resulting from Ward‟s cluster analysis of 200 years (1800-2000) among ten sites.

Fig. 4.14: Principal component analysis of ten sites using the common period of 200 years (1800-2000).

4.10-Discussion

All forests studied for research were disturbed due to human cutting, however most sampled trees were presented in forested areas that were not easily accessible, and therefore suffered minimal human disturbance. Conifers were more representative than broadleaf species and Picea smithiana was the most common genus at eleven sites. Picea smithiana from seven sites, Juniperus excelsa from three sites and Pinus gerardiana from one site were selected. Each site and species provided good crossdating under the microscope. Every tree species presented similar narrow rings not only among the species but also among the sites, pointer years or rings of one species and site overlap with other species and site. Therefore, it is suggested that the area and species fell under similar climatic conditions.

COFECHA statistics indicated that the highest correlation occurred in Picea smithiana from Jutial and lowest correlation is shown by Juniperus excelsa from Chaprot. It is suggested due to highest crossdated cores from Picea smithiana of Jutial, highest correlation occurred. However Picea smithiana from Haramosh and Bagrot, and Pinus gerardiana from Chaprot showed good correlation too. The percentage of missing rings was found in all chronologies with high percentage in Pinus gerardiana from Chaprot and Picea smithiana from Chera (nearly 0.23%). These two species are at lower elevation among all sites. Hence the tendency to exhibit locally absent rings was more predominant at lower elevations, as would be expected due to greater moisture stress (Fritts, 1976). The similar tendency of missing rings was also observed at lower elevation in Nepal (Bhattacharya et al. 1992).

Longest mean length of series was found in Picea smithiana from Nalter (375 years) and smallest mean length series was found in Juniperus excelsa from Chaprot (249 years). Picea smithiana from Haramosh had highest common years i.e. 250 years and Picea smithiana from Kargah had lowest common years i.e. only 101 years. It means that more cores of Picea smithiana from Haramosh spanned at least the past 250 years. The forest of Haramosh trees spanned the age of 250 years while forest from Picea smithiana Kargah had the minimum age of just only 100 years. The forest of Haramosh is not easily accessible compared with Kargah, and therefore intensive cutting of trees was minimal.

Picea smithiana from Jutial, Haramosh and Bagrot had attained the mean sensitivity values of 0.3 while Picea smithiana from other two sites achieved mean sensitivity values of 0.2. According to Speer (2010), a series with mean sensitivity values around 0.1 is so complacent that it is difficult to crossdate, and series with mean sensitivity values more than 0.4 is sensitively tricky to crossdate. Juniperus excelsa from Morkhun attained mean sensitivity value more than 0.4, meaning that the rings were extremely narrow as to render crossdating too difficult, and therefore these trees were not included for further analysis. Mean sensitivity around 0.2 is generally accepted (Fritts, 1976; Speer, 2010) because they are sensitive enough to crossdate and reconstruct past climate. In the present study, mean sensitivity values for all series were around 0.2. The chronology statistics are consistent with the result of Ahmed et al. (2010, 2011) and Borgaonkar et al. (2009).

The values of inter-series correlation (Rbar) and signal-to-noise ratio (SNR) from most of the sites showed that Rbar and SNR increased with increasing growth-rates. In other words, fast growing sites have better scenarios for climatic studies. However, for Picea smithiana from Haramosh and Pinus gerardiana from Chaprot, the reverse was the case. Here, slow growth rates showed the best prospects. Ahmed et al. (2011) explained that Pinus gerardiana and Pinus wallichiana from Chitral Gol National Park and Astore, respectively, expressed best prospects with faster growth rates. In the present study we have inverse results as compared to the findings made by Ahmed et al. (2011). The differences in results may be the differences in the ecological tolerances of the species situated at different elevations.

We identified 22 negative pointer years that were found among all chronologies. The most consistent of these are present in maximum number for the years 2001, 1971, 1947, 1917, 1877, 1802, 1742, 1626, 1603, 1572 and 1492. The year 1877 (observed in current analysis) was the biggest ENSO drought of instrumental times across much of Asia (Aceituno et al. 2009) and it was the first of the great Victorian droughts that led to millions of deaths around the globe. It shows up very strongly in Vietnam too. Nineteen positive pointer years were witnessed common in all chronologies having consistent positive pointer years in 1958, 1883 and 1804. Our results are quite similar with the findings of Esper et al. (2001) who described 429 tree ring-width values from twelve Juniperus excelsa sites and three mixed sites (Juniper, Picea and Pinus) from northwest Karakorum of Pakistan and southern Tien Shan in Kirghizia for extreme years since

1427 (Esper et al. 2001). A comparison between Karakorum and Tien Shan showed similar 17 negative inter-regional pointer years (1917, 1877, 1871, 1833, 1806, 1802, 1790, 1742, 1669, 1653, 1611, 1605, 1591, 1572, 1495, 1492, and 1483 AD) and eight positive inter-regional pointer years (1916, 1804, 1766, 1703, 1577, 1555, 1514, 1431 AD). Esper et al. (2001) demonstrated that these regional pointer years from Northern Karakorum of Pakistan and Tien Shan Kirgizia were due to extreme climatic conditions which limited the tree growth on large scale independent of site ecology, from the lower, arid to the upper humid timberlines and in different exposures. From our analyses, the authors determined that the main limiting factor for tree growth was temperature variations.

In the current study we note the incidence of three successive negative growth rings (1572- 1574). The year 1572 was also evidenced in the study of Esper et al. (2001), indicating that climate was harsh during these three consecutive years across a broad region. Another noticeable point is the occurrence of similar trend after every two hundred years; 1602 and 1802, 1717 and 1917, 1785 and 1985, 1802 and 2001. It may suggest that cycle of extreme climate year happened after every two hundred years. The existence of 2005 positive pointer year probably hints to the occurrence of extensive rainfall that might be the reason of favorable growth of ring experienced in most of the northern parts of Pakistan including Gilgit and Hunza valleys.

The principal component analysis and correlation analysis exposed a high degree of resemblances among most of the chronologies. The reason is that all chronologies were developed from the same time period and sites were located in a region that is affected by similar climatic and weather pattern.

In general, the chronologies from Gilgit and Hunza valleys have produced some helpful results that should be extended by more intensive sub-regional sampling. The chronologies from Picea smithiana proved to be excellent in internal dating and hence strong common signal has resulted. It suggests that this species should be studied further. One chronology of Juniperus excelsa and one of Pinus gerardiana from Chaprot indicate moderate potential in common signal evaluation. Ring-width chronologies of high elevation Juniperus excelsa (Morkhun) provided no utility for dendroclimatological signals.

The strong correlation among the ten chronologies is observed because many pairs of the sites were situated at small distances from each other. Therefore, variety of species has quite similar results due to similar ecological conditions. In addition, likely ecological conditions and close resemblance in site histories, common climatic influences occurred in the whole region. With the help of systematic sampling program, ring-width of Picea smithiana, Juniperus excelsa and Pinus gerardiana in Gilgit and Hunza valleys focused on the potential of past climatic records.

4.11-Conclusion

Ten chronologies were developed from Gilgit and Hunza valleys, Pakistan. The chronologies spanned 271-619 years. The resemblance of pointer years among sites exhibited a common growth pattern. These facts highlight the status of careful site selection for dendroclimatic research. Juniperus excelsa from Morkhun exhibited extreme sensitivity, such that crossdating was not achievable for this study. An inter-site comparison shows that common signal occurs not only between individuals of a single site, but also in different species on larger (regional) scales. A similar pattern of pointer years, high correlation, high values of mean sensitivity, EPS and SNR values demonstrates the high potential for further investigations for dendroclimatic, dendrohydrological and drought years‟ reconstruction for the past 500 years or more. However for better understanding of ring pattern before 1700 AD, the sample size should be increased.

Chapter No. 5

Growth-climate response

The next step after chronology development is to check the relationship between tree growth and climate, to investigate the conditions that most affected the rate of cambial cell divisions during the annual growth period. Trees that grow in dry temperate sites, such as those used for this study may have a very short growing period of a few weeks. According to Fritts (1976) and Blasing et al. (1984), growth-climate relationships can be examined by calculating correlations between tree-ring index chronologies and climatic data. The response function is a form of regression equation, in which the climate is used as the independent variable and tree ring data as dependent variable. Response function analysis is a multivariate technique that is used to determine the tree ring response to climatic factors, so that these results can be used to deduce the climatic conditions into the past. This multiple regression technique first employs Principle Component analysis (PCA) on the monthly climatic data in order to develop a new set of uncorrelated (i.e., orthogonal) variables for regression. These uncorrelated variables are then compared against tree growth through correlation statistics to produce a set of correlation coefficient (i.e. response function) that tells us about the relationship between tree growth and climate (Fritts et al. 1971; 1974).

(Ahmed et al. 2011, 2012) have demonstrated that climate influences the growth of several Pakistan conifers by examining the strength of temperature and precipitation signal in tree ring chronologies from Picea smithiana, Juniperus excelsa, Pinus gerardiana, Cedrus deodara, Abies pindrow, and Pinus wallichiana. Several other tree species have been shown to be useful for dendroclimatic investigations in the Himalayan and Karakorum regions (e.g., Ahmed, 1987; Bhattacharya and Yadav, 1999; Yadav and Park, 2000; Singh and Yadav, 2007; Singh et al. 2009; Ahmed et al. 2009; Ahmed et al. 2011, Esper et al. 2002). Many Himalayan conifers and broad-leaved tree species have been successfully used for dendroclimatic studies, but most of these studies were limited to the eastern Himalaya (Ahmed et al. 2011). The western Himalaya are home to widespread forests with a number of different species divided between the moist and dry temperate areas of the country (Champion et al. 1965; Ahmed et al. 2006).

100

Here we investigated the growth climate correlation and response in order to identify the climate variables that have significant effect on some species of the genus Pinus. Tree ring data from ten chronologies were correlated against mean monthly temperature and total monthly precipitation. The results are presented below.

5.1-Materials and methods

To determine which chronology best suited our study; we developed a preliminary correlation among climatic data from the Gilgit station and gridded data (Mitchell and Jones, 2005) with three chronologies using point-by-point regression. The percent variance for correlation was obtained at 95 percent confidence interval using software known as CORRELATION AND RESPONSE FUNCTION with packaged software DPL from the LDEO (Lamont Doherty Earth Observatory) TRL website, as described by Fritts (1976). The Pearson correlation coefficients were calculated among tree ring chronologies and the monthly series of temperature and precipitation from both station and grid data for the 13 months period ranging from previous October to current October (Figs. 5.1; 5.5). The residual version of each chronology was used to estimate the growth-climate correlation coefficients because the residuals version is pre-whitened in order to remove the low order persistence due to autocorrelation (Cook, 1985). PCA was performed on the climate variables by employing a variance maximizing rotation of original variable space (Richman, 1986). The first principal component explains the common variance of the chronology set and hence point out to the regional growth signal.

5.2-Climate data

Like other parts of the world, the scarcity of long meteorological records for statistical calibration to the tree rings in the Himalayas is a problem (Cook et al. 2003; Bhattacharya et al. 1992). One solution of this difficulty has been to introduce seasonal data into a 0.5o latitude- longitude grid data set that can be useful for its proximity to the tree ring location (Cook et al. 2003). Here we used gridded climate data from the CRU TS 2.1(http:/www.cru.uea.ac.uk/) for the region over northern Pakistan for comparison with the Gilgit meteorological data (temperature and precipitation). The Gilgit local station sits at an elevation of 1460 m, whereas Hunza has no local climatic station. Therefore, Gilgit‟s climatic data is considered to be the most representative of the local climate for all of the ten sites. The records for the Gilgit local station

101 span from 1955-2009, and are significantly shorter than the gridded product that extends from 1901-2002 (Mitchell and Jones, 2005). For many of the tree ring samples sites, no local records exist, and the ones that do are from much lower elevation. Therefore, each tree ring chronology was estimated with Gilgit local station data and also with CRU data.

5.2.1-Temperature

The Gilgit meteorological station is located in Gilgit (35o55N, 74.20E) in close vicinity to most of the tree ring sites. Although short (just greater than 50 years), its instrumental record is the longest in Northern Pakistan. Box and whisker plot of Gilgit monthly temperature (oC) is presented in Fig. 5.1. Maximum temperature peaks in July while the lowest temperature is recorded in December and January. The mean annual temperature in Gilgit is 15.93oC obtained by averaging fifty years data. The warmest year on record was 1961 when the average annual temperature reached 17.53oC, while 1989 was the coldest year on record with annual mean temperature of 14.53oC.

Figure 5.1: Box plot of mean monthly temperature of Gilgit station based on the period (1955- 2009). Asterisks in the figure show the outliers from the data.

102

5.2.2-Precipitation

Like temperature, the precipitation data used in the current study were recorded at Gilgit. The observation period was the same 1955-2009. Box and whisker plots of Gilgit monthly precipitation in millimeters are presented in Fig. 5.2. As expected, there is far more variability with precipitation than with temperature, as seen by the outliers above the mean. Maximum precipitation is generally seen for the months of April and May, with November being on average the driest. The mean total precipitation from the station is approximately 131.4 mm annually, nearly similar in both halves of the record. 1996 was the wettest year on the record where total precipitation reached nearly 251.7 mm per annum. The driest year was 1977 where the annual precipitation accounted for just 40.7 mm. The precipitation data is scattered evenly throughout the year with no consistent change. The minimum sum of 141.9 mm (November) and maximum sum of 1384 mm (May) throughout the entire period of (1955-2009).

Figure 5.2: Box plot of mean monthly precipitation of Gilgit station based on the period (1955- 2009). Asterisks in the figure show the outliers from the data.

103

From the above discussion, it is expected that annual radial tree growth is sensitive to temperature and precipitation, because in this high and arid region tree ring growth can be restricted by extreme variations of both variables. If precipitation is sufficient, then temperature is the expected principal factor to restrict tree growth, and vice versa. Gilgit experiences high temperature in summer and relatively little rainfall throughout the year, hence, temperature and precipitation both are likely to limit growth. The climate of the region is dry temperate (Ahmed et al. 2006) and is characterized by maximum temperature during summer months (June- August). The rainfall, maximum occurs during late spring that is in the months of March and April. The mean temperature at Gilgit station is between 20-25oC in summer months and mean precipitation during late spring lies between 20 to 30 mm (see section 3 Fig. 3.1).

The hierarchy (Fig. 5.3) shows the method that is adopted for further analysis. Out of three chronology versions, the pre-whitened residual chronology (RES) was used for all correlation and response function analyses because the AR modeling of the detrended series (Cook, 1985) increases the confidence of the climate growth relationship without the possible spurious correlation of trend in data. The RES series were compared with climate data to identify significant months as shown below.

104

CHRONOLOGY

STANDARD ARSTAN RESIDUAL

CORRELATION RESPONSE FUNCTION FUNCTION

LOCAL GRIDDED LOCAL GRIDDED CLIMATE CLIMATE CLIMATE CLIMATE

Figure 5.3: Hierarchy of method which is followed for correlation and response analysis

5.3-Results

5.4-Correlation among tree-ring chronologies and temperature

The Growth-temperature relationships for all ten chronologies from ten sites against the Gilgit station data are shown in figure 5.2. Low positive correlation coefficients (54) were found in comparison with negative coefficients (76), which imply that high temperature has a negative impact on tree-growth during the growing seasons (March-May). All tree ring indices showed similar positive trend in temperature correlation with respect to the winter season (previous December to January) including previous November also. On the other hand, three species exhibited different response in terms of spring and summer seasons. Picea smithiana from all

105 sites showed negative correlation in spring season (March-May). Juniperus excelsa from two sites exhibited negative relationship with temperature from May to July, and this might be considered as summer season whereas Pinus gerardiana presented a long term negative correlation including five months (March-July).

5.5-Correlation among tree-ring chronologies and precipitation

Like temperature, the relationship between tree growth and precipitation was seen from Gilgit precipitation data (Fig. 5.3). In contrast with temperature, correlation comparison explains more positive correlation (92) and less negative correlation coefficients (38) pointing to the fact that more rainfall is good for tree growth during its growing season.

Synchronization occurred across all chronologies when comparison was made in correlation analysis. The results showed that tree ring indices were positively correlated with precipitation for the spring season, in particular the four months of February to May, with the exception of the two species from Nalter, having slightly different results for these months (Picea smithiana showed negative correlation in the month of May and Juniperus excelsa showed low positive correlation in the months of February-May. Most probably it happened due to the occurrence of these two species as Nalter is wet site, receives extensive amount of permanent snow on the top.

The overall results indicate that different species have different response to climate even when situated at the same site locations (as in case of temperature, the two species from Nalter showed different response), or different species may exhibit a similar response to climate even when situated at different area (like in case of precipitation, all species exhibited similar response).

106

0.4

0.2

0 pO pN pD J F M A M J J A S O -0.2

Correlationcoefficients -0.4 Picea smithiana Kargah

0.4 0.2 0 pO pN pD J F M A M J J A S O -0.2 -0.4

-0.6 Correlationcoefficients Picea smithiana Jutial

0.4 0.2 0 pO pN pD J F M A M J J A S O -0.2 -0.4

Correlationcoefficients -0.6 Picea smithiana Haramosh

0.4

0.2

0 pO pN pD J F M A M J J A S O -0.2

Correlationcoefficinets -0.4 Picea smithiana Bagrot

Figure 5.2: Graphs representing correlation coefficients between residual chronologies and temperature of Gilgit meteorological data from ten sites respectively for a 13 month span. Shaded areas indicate the significant months over 13 month period. (To be continued.,)

107

0.6

0.4

0.2

0 pO pN pD J F M A M J J A S O

Correlationcoefficients -0.2 Picea smithiana Nalter

0.4

0.2

0 pO pN pD J F M A M J J A S O -0.2

Correlationcoefficient -0.4 Picea smithiana Chera

0.4

0.2

0 pO pN pD J F M A M J J A S O -0.2

Correlationcoefficients -0.4 Picea smithiana Chaprot

0.4 0.2 0 pO pN pD J F M A M J J A S O -0.2 -0.4

Correlationcoeffcients -0.6 Juniperus excelsa Chaprot

108

0.3 0.2 0.1 0 -0.1 pO pN pD J F M A M J J A S O -0.2 -0.3 Correlationcoeffiicents -0.4 Juniperus excelsa Nalter

0.4 0.2 0 pO pN pD J F M A M J J A S O -0.2 -0.4

Correlationcoefficients -0.6 Pinus gerardiana Chaprot

109

0.6

0.4

0.2

0 pO pN pD J F M A M J J A S O Correlationcoefficients -0.2 Picea smithiana Kargah

0.6

0.4

0.2

0 pO pN pD J F M A M J J A S O

Correlationcoefficients -0.2 Picea smithiana Jutial

0.6

0.4

0.2

0 pO pN pD J F M A M J J A S O

Correlationcoefficients -0.2 Picea smithiana Haramosh

0.4

0.2

0 pO pN pD J F M A M J J A S O -0.2

Correlationcoefficients -0.4 Picea smithiana Bagrot

Figure 5.3: Graphs representing correlation coefficients between residual chronologies and precipitation of Gilgit meteorological data from ten sites for a 13 month span. Shaded areas indicate the significant months over 13 month period. (To be continued.,)

110

0.4

0.2

0 pO pN pD J F M A M J J A S O -0.2

Correlationcoefficients -0.4 Picea smithiana Nalter

0.4 0.3 0.2 0.1 0 -0.1 pO pN pD J F M A M J J A S O

Correlationcoefficients -0.2 Picea smithiana Chera

0.5 0.4 0.3 0.2 0.1 0 -0.1 pO pN pD J F M A M J J A S O

Correlationcoefficients -0.2 Picea smithiana Chaprot

0.6

0.4

0.2

0 pO pN pD J F M A M J J A S O

Correlationcoefficients -0.2 Juniperus excelsa Chaprot

111

0.3 0.2 0.1 0 -0.1 pO pN pD J F M A M J J A S O -0.2

Correlationcoefficients -0.3 Juniperus excelsa Nalter

0.4 0.3 0.2 0.1 0 -0.1 pO pN pD J F M A M J J A S O

Correlationcoefficients -0.2 Pinus gerardiana Chaprot

112

Correlation and response function analyses were performed to find out the consistency among the results. The four Tables (5.1-5.4) explain the comparison among residual chronologies and Gilgit meteorological and grid data. Only significant correlation signs are inserted in the Tables. High variance (chronologies versus Gilgit meteorological data) was obtained in correlation analysis (Table 5.1) as compared to the variance obtained (chronologies versus grid data) in correlation (Table 5.2). Besides this, the highest variance was seen for the Chaprot site (Pinus geradiana and Juniperus excelsa) for both local and grid comparison, respectively (Tables 5.1, 5.2). Juniperus excelsa from Nalter exhibited the lowest variance in both local and grid comparison (Tables 5.1, 5.2).

We then summarized these Tables by counting the positive and negative significant signs (Table 5.5) to pick the best seasons that had the most influence on tree growth. In the case of temperature, almost all chronologies showed significant positive response to winter season, particularly so for previous December and current January (Table 5.5). The spring season (March-June), had a strong negative effect on tree growth. While in the case of precipitation, significant positive response of spring season (February-May) to tree growth was identified. Hence in both cases (temperature and precipitation), the spring season response of March-May is dominant across all the chronologies.

113

Table 5.1: Summary of correlation function between tree-ring chronologies and monthly temperature and precipitation data from Gilgit station Temperature Precipitation

Site pO pN pD J F M A M J J A S O pO pN pD J F M A M J J A S O Total Variance

PSKAR + + + + 43.35%

PSJUT + - - - - + 62.20%

PSHAR + - - - + + + 55.34%

PSBAG + - - - + + + + 59.18%

PSNAL + + + + + + - 60.38%

PSCHR + + + + + 57.73%

PSCHP + - + + 39.55%

JECHP + - - - + + 59.31%

JENAL - - - 36.84%

PGCHP + - - - - - + + 62.96%

PSKAR= Picea smithiana from Kargah, PSJUT= Picea smithiana from Jutial, PSHAR= Picea smithiana from Haramosh, PSBAG= Picea smithiana from Bagrot, PSCHP= Picea smithiana from Chaprot, JECHP= Juniperus excelsa from Chaprot, JENAL= Juniperus excelsa from Nalter, PGCHP= Pinus gerardiana from Chaprot

114

Table 5.2: Summary of correlation functions calculated from tree-ring chronologies and monthly temperature and precipitation data from the relevant 0.5o grid climate database (Mitchel and Jones, 2005) Temperature Precipitation

Site pO pN pD J F M A M J J A S O pO pN pD J F M A M J J A S O Total Variance

PSKAR + + - 21.86%

PSJUT - - - - + + + + 33.73%

PSHAR - - - + + + 34.59%

PSBAG - - + + + + 38.94%

PSCHR - + + + 31.95%

PSNAL + + + - 37.68%

PSCHP + + + 33.10%

JECHP - - + + + + + 44.35%

JENAL - - + + + 30.82%

PGCHP - - - - + + + + + 40.12%

115

Table 5.3: Summary of response function between tree-ring chronologies and monthly temperature and precipitation data from Gilgit station

Temperature Precipitation

Site pO pN pD J F M A M J J A S O pO pN pD J F M A M J J A S O

PSKAR + + - - - - + + + - - -

PSJUT + + - - - + + +

PSHAR + + + - - - + +

PSBAG + + - - - - + + + + + + -

PSNAL + + - + + + + + - -

PSCHR + + + - - - + + - + +

PSCHP + + - - + + +

JECHP - + + - -

JENAL - + - - +

PGCHP + + + - - - - - + + +

116

Table 5.4: Summary of Response functioncalculated from tree-ring chronologies and monthly temperature and precipitation data fromrelevant Grid data (Mitchell and Jones, 2005)

Temperature Precipitation

Site pO pN pD J F M A M J J A S O pO pN pD J F M A M J J A S O

PSKAR + - - + - - - -

PSJUT + - - + +

PSHAR + - - + + + -

PSBAG + - - + - + + + + +

PSNAL + + + + + + + -

PSCHR + - - + + + + +

PSCHP + - + + + -

JECHP - + + + + + +

JENAL + + - - - - + + + +

PGCHP + - - + + + - +

117

Table 5.5: Summary of four tables (5.1-5.4) including only significant signs of positive and negative correlation and reponse analysis

Temperature Precipitation

pO pN pD J F M A M J J A S O pO pN pD J F M A M J J A S O

Positive 0 0 13 2 1 0 0 0 1 1 0 0 0 0 0 1 0 5 8 6 1 0 0 0 0 0 Table 5.1 Negative 0 0 0 0 0 3 4 6 6 3 0 0 0 0 0 0 0 0 0 0 0 1 0 0 0 0

Positive 0 0 0 0 0 0 0 0 1 1 0 0 0 0 2 2 2 8 6 5 7 1 0 0 0 0 Table 5.2 Negative 2 0 0 0 0 2 2 5 2 3 2 0 0 0 0 0 0 0 0 0 0 2 0 0 0 0

Positive 0 1 9 7 4 1 0 0 1 1 0 0 0 4 4 0 0 3 8 6 3 0 0 0 0 0 Table 5.3 Negative 2 0 0 0 1 5 6 2 3 2 1 2 2 1 0 0 1 0 0 0 0 2 2 2 0 0

Positive 2 1 1 5 1 0 1 0 3 2 0 0 0 2 3 0 1 3 5 7 6 1 5 0 0 1 Table 5.4 Negative 0 0 0 0 1 6 4 4 1 1 1 0 0 0 0 0 0 0 0 0 0 4 0 1 0 1

Cumulative 2 2 23 14 6 1 1 0 6 5 0 0 0 6 9 3 3 19 27 24 17 2 5 0 0 1 Positive Cumulative 4 0 0 0 2 16 16 17 12 9 4 2 2 1 0 0 1 0 0 0 0 9 2 3 0 1 Negative

118

5.6-Discussion

Correlation analysis was performed to assess the relationship between tree growth and climate, using nearby station and gridded data, in conjunction with traditional Response function analysis (Fritts, 1976). A window of 13 months from previous October to current October was used because tree growth season in this area is considered to initiate around March and terminate by the end of September, with some carry-over effects from the prior year.

The important consideration for dendroclimatic studies is the potential age of trees (Ahmed et al. 2011). For example, Picea smithiana from some sites exceed 500 years (see section 4) and older ages are obtained using the same species from India (Singh et al., 2004; Singh and Yadav, 2007). Juniperus excelsa by Esper et al. (1995) from Morkhun were found to be the oldest with some trees significantly greater than 1000 years and no other species were older than junipers. We have collected samples from the same place but were unable to crossdate due to difficulties in accounting for locally absent and/or false rings.

The correlation between site chronologies declined with increasing separation distance (Ahmed et al. 2011). This was observed both among sites of the same species and among sites composed of different species. A much stronger correlation was sometimes found between two different species growing at the same site than between sites of the same species with a little separation of 0.5 kilometers. These findings sustain the practice of dense multi-species tree-ring network for better spatial and temporal coverage to account for the effects of local topography. The best vision for this in the Karakoram Range appears in terms of Cedrus deodara and Pinus gerardiana and matches with the reported studies made by neighboring India by Borgaonkar et al. (2009). However, in another study based on seven sites from Karakorum Range of Northern Pakistan, this trend is not seen (Ahmed et al. 2012). Our results match with the findings of Ahmed et al. (2012) where we observed Juniperus excelsa and Picea smithiana from the same site (Nalter) but found no significant correlation. It may therefore not be necessary that correlation of site chronologies declined with increasing distance.

The heterogeneous nature of the tree-ring network with respect to distance between chronologies is identical to the findings made by Archer and Blenkinship (2010), who established high

119 heterogeneity among different climate station data in the Karakoram, and claimed that the heterogeneity would make for a successful reconstruction of hydroclimate over the Karakoram.

A summarized result of significant monthly climate correlations and percentage of variance explained were chosen. Low correlation occurred in the comparison of tree ring indices and grid climate as compared to correlation between tree ring indices and local climate, possibly due to the fact that grid climate has been interpolated to cover a large spatial area, whereas local meteorological data reflects the climate of a specific site.

The same pattern across all the species can be seen for temperature in this study, owing to temperature‟s greater spatial homogeneity than for rainfall. March-June temperature revealed the most significant negative correlation in our analyses, suggesting that these months have a negative influence on tree growth if the temperature surges. Higher temperature increase evapotranspiration, and results in the decrement of soil moisture leading to low tree growth. There is a strong positive correlation to previous December in almost all species, which indicates that all species require higher than average temperature in this month for better growth.

A strong positive correlation was also observed during correlation and response analysis in spring (February-April) rainfall suggesting that spring rainfall enhances tree growth. The rainfall response was positive with the greatest correlations for February-May. No months were found to be significant during summer and autumn months (June-September), regarded as the monsoon period, because our sites are located in a dry temperate area where the monsoon winds have very little influence. A similar broad outline was also seen in response function analysis, which supported the outcomes of the correlation function.

Treydte et al. (2006) used Juniperus excelsa from Bagrot site (Hunza). They detected the highest correlated month for precipitation was July. Their study was based on oxygen isotope concentrations. We have Picea smithiana from the same site but correlated months are February- May. The different rainfall response between Treydte et al. (2006) and present study might be the elevation. Treydte et al. (2006) selected low and high elevations Juniperus excelsa. We have Picea smithiana from the same site at the elevation of 3130 m but it did not show the similar response. The difference in results may be the difference in elevation in species or amount of moisture in soil. Juniperus excelsa used in Treydte et al. (2006) reconstruction grow on the drier

120 site of the dry temperate area whereas Picea smithiana grow on better sites where the moisture is available. It might also be suggested that Treydte et al. (2006) and Yadav et al. (2002) studies used RCS which created an artificial positive trend bias in their results – purely an artifact of the procedure.

Ahmed et al. (2011) found a poor correlation to gridded climate data by the high elevation site Bagrot 5 compared to the low elevation site Bagrot 1. The Morkhun site was also used by Treydte et al. (2006) for rainfall reconstruction, and was shown to have a third type of response – cold and dry. These sites were more than 1000 years old from high elevation (3900 m). The explanation was given by Ahmed et al. (2011) that residual or weakened bands of cloud from summer monsoon only reach the highest zones of forest, driven by the strong orographic effects of mountain ranges. As our sampling sites are located either below or away from the zone influenced by the summer monsoon there is a difference in summer rainfall correlations.

Cedrus deodara and Pinus gerardiana from district Chitral were analyzed for its dendroclimatic potential which disclosed significant negative correlation with temperature and significant positive correlation with precipitation in the spring season (Khan, 2011). The results of Khan (2011) support the current study, which agrees with the findings of Wahab (2011) who worked over district Dir by using Cedrus deodara and Picea smithiana.

Cook et al. (2003) presented a network of 32 sites from Nepal, including Picea smithiana and Pinus wallichiana to produce temperature reconstruction from two species that share a positive correlation with summer (June-July) temperatures. Here we have positive correlation to June- July temperature only in the case of Picea smithiana from Nalter, while Picea smithiana from other sites did not respond the same. Perhaps this is because Nalter is a cold place surrounded by snow covered peaks and covered with clouds throughout the year with plenty of soil moisture. Therefore hot summers might be expected to enhance tree growth of tree as in the case of Nepal (Cook et al. 2003).

The current study sites are located at higher elevations and receive extensive winter snowfall. This winter snowpack results in more available soil moisture than at other dry temperate sites, so increased moisture loss may not considerably affect the moisture available later in the year. High temperature in winter can favor rapid net photosynthesis and increased physiological activity that

121 can lead to the early initiation of cambial activity, rapid growth or formation of wide rings (Tranquillini, 1964; Fritts, 1976). This clarifies the positive response of winter temperature, especially in December, to the variations between chronologies. Borgaonkar et al. (2009) showed a similar pattern of relationship for Cedrus deodara of neighboring India from the Western Himalaya. Singh et al. (2009) also detected a positive relationship of tree-ring index series with winter (December-February) temperature and summer precipitation. Our data explains that most of the sites expressed a positive relationship with December.

In the spring (March-May), the situation is different. The temperature rises gradually above the average annual value while the amount of precipitation is very small, which leads to reduced ring-width formation due to higher evapotranspiration. Therefore, more rainfall in these months (March-April) is conductive to better growth. Several other studies also indicated similar response of pre-monsoon (spring season) on Western Himalayan conifers (Borgaonkar et al. 1994, 1996; Yadav et al. 1999; Yadav and Singh, 2002).

5.7-Conclusion

It was concluded that growth for all three species was directly affected by a combination of temperature and precipitation. The tree-ring data were positively correlated with previous winter temperatures with highest correlation for previous December to current January, and negatively correlated with temperature over the entire period of the spring season with the strongest and most consistent relationships for March-June. Chronologies were also positively correlated with spring season precipitation with the highest relationship with February to May.

122

Chapter No. 6

Temperature reconstruction

6.1-Introduction

Several studies highlight the importance of reconstructing past temperature variability before the instrumental period for comparing the natural and anthropogenic climatic changes (Briffa and Osborn, 1997; Jones et al. 1996; Man et al. 1998). These studies have been conducted primarily with conifers, like Abies pindrow, Cedrus deodara, Picea smithiana, Pinus geradiana and Pinus wallichiana from the Himalayan region (Ahmed et al. 2011; Bhattacharya and Yadav, 1999). Conifers including Juniperus excelsa were used to reconstruct past climate back to AD 600 from the Karakorum Range (Esper et al. 2002). Tree-ring chronologies from the Himalayas of Nepal were used to reconstruct the past 400 years of temperature (Cook et al. 2003). Recently, Cook et al. (2013) successfully reconstructed past 500 years of Indus river flow by using tree ring chronology in comparison with Partab flow data. In the present study, we reconstruct past temperature more than 400 years by using Picea smithiana, Juniperus excelsa and Pinus gerardiana chronologies from sites at Gilgit as these substitute records, will provide valuable data for climate change studies with regional and global perspective.

One current dilemma is the conflicting reports that some Himalayan glaciers are rapid retreating when they are in fact advancing (e.g. Owen, 2009; Cook et al. 2013). Here, we intend to predict the temperature influences and investigate the similarity and synchronicity of past responses to known major cooling events such as the “Little Ice Age” (LIA) (Luckman &Villalba, 2001). This will provide supporting information to WAPDA that will contribute efforts attempting to model future outcomes.

123

6.2-Materials and methods

For temperature reconstruction we followed the calibration and verification techniques of Fritts, (1976) and Cook and Kairiukstis, (1990). Precipitation reconstruction model was performed separately but we got poor results and it is still important to explore the past precipitation in future. The four months which showed significance in correlation analysis were used to make a season (March-June) as these months had the most influence on tree growth and relationship was stable in time. Tree ring indices were used as the predictors and local temperature was used as the predictand. The point by point regression option was chosen with a predictor and predictand common period from 1955-2008, and this time span was split into two equal calibration and verification period.

Simple linear regression was used to transform the tree ring data into the estimates of four month window of temperature. The data were divided into two periods (1955-1985 and 1986-2005) prior to regression, using one for calibration and other for verification, respectively, and then reversing the order for comparison of both periods. The statistics used for the calibration and verification periods are the Pearson correlation coefficient, Spearman correlation coefficient, Reduction of error (RE) and coefficient of efficiency (CE). These values are often lower in verification period as compared to calibration period. If RE and CE are negative then regression estimates is bad in verification period. Positive RE and CE is the evidence of validity of regression model (Cook et al. 1994; 1999). However, according to Brendon M. Buckley in personnel communication, the RE and particularly CE in the verification periods can be thrown way off by one or two anomalous values. Any positive value for both implies model fidelity, and negative values implies poor model fidelity, but that lack of fidelity can often be explained by a single bad value. A simple plot of the act-est data can go a long way to explaining this. “PCReg” from the LDEO (Lamont Doherty Earth Observatory) TRL website program was used to develop the reconstruction of temperature for the past 400 years. The gauge data was also seasonalized using “SEA” program from the packaged software DPL.

Reconstructions were carried out between tree-ring chronologies and Gilgit meteorological data by picking the significant months from correlation analysis (Table 5.1) to make a season as correlation analysis exhibited highest percent variance in overall correlation and response functions. Reconstruction model was performed one by one to the tree ring chronology and

124

Gilgit local data to check out which chronology best suited for reconstruction. Picea smithiana showed negative correlation with March-June (spring season) temperature, whereas Juniperus excelsa showed negative correlation May-July of the growing period (see Table 5.1).

6.3-Results

Here we present a reconstruction of March-June temperature variation for the past 400+ years using ring-width chronologies of Picea smithiana, Juniperus excelsa and Pinus gerardiana from Gilgit and Hunza valleys. Table 6.1 shows regression analysis results for the ten chronologies. Five of the sites accounted for the maximum retained variance from the model, while the other five were rejected from further analyses. The five retained sites include Pinus gerardiana from Chaprot, Picea smithiana from Jutial, Picea smithiana from Haramosh and Bagrot. Picea smithiana from Kargah, Nalter, Chera and Chaprot were rejected. A dramatic change happened in Juniperus excelsa Nalter which has the ability to make season but failed to pass regression analysis.

The highest explained variance was observed in Pinus gerardiana Chaprot, but the data are too short i.e. only more than 150 years for reconstruction. (EPS value reliable up to 1840 see section 4) whereas Picea smithiana from Jutial data is reliable up to 1530 (EPS>400) therefore selected for further reconstruction.

125

Table 6.1: Regression analysis of ten chronologies with Gilgit temperature data from different sites of Gilgit and Hunza valleys.

Chronology used Months (seasons) Variance explained

Picea smithiana Kargah No season No regression

Picea smithiana Jutial 4 months (March-June) 38.16%

Picea smithiana Haramosh 3 months (April-June) 13.94%

Picea smithiana Bagrot 3 months (April-June) 16.77%

Picea smithiana Nalter No season No regression

Picea smithiana Chera No season No regression

Picea smithiana Chaprot No season No regression

Juniperus excelsa Chaprot 3 months (May-July) 32.71%

Juniperus excelsa Nalter 3 months (May-July) No regression

Pinus geradiana Chaprot 5 months (March-July) 44.50%

The correlation analysis indicated that tree growth was affected by March to June temperatures, the season of reconstruction hence a linear regression model was developed to reconstruct past temperature for the Gilgit region. During the common period of tree ring index and station data (1955-2008), the reconstruction accounted 38.16% of the variance. Split calibration and verification were employed to assess the statistical reliability of this model. The reduction of error (RE) and coefficient of efficiency (CE) statistics were both positive in the verification period, indicating significant skill in the tree ring estimates (Fritts, 1976) (Table 6.2a and 6.2b). The model calibrated on the early period (1955-1985) explained 38.16% of the variance in March to June temperatures, while the model based on late period (1985-2008) expressed 21.9% of the variance. Both models passed all test of verification. RE>0.3, CE>0.1 demonstrated the excellent performance of both models. The tree-ring data explains 28.48% of the variance in temperature using the entire period of 1955-2008 for calibration.

126

Because of the better verification statistics and variance (Tables 6.2a and 6.2b), the early calibration model was used to reconstruct (March-June) temperature back to 1523.

Table 6.2a and 6.2b: Calibration and verification statistics for the early (a) and late (b) periods. RP= Pearson‟s product moment correlation coefficient; RR= the robust correlation coefficient; RS= the spearman‟s coefficient of rank correlation; RSQ= variance explained; RE= Reduction of error; CE= coefficient of efficiency. Table 6.2a: Early calibration

Calibration (1955-1985) Verification (1986-2008) Statistics Value Statistics Value RP 0.618 RP 0.468 RR 0.615 RR 0.438 RS 0.518 RS 0.348 RSQ 0.381 RE 0.382 CE 0.197

Table 6.2b: Late calibration

Calibration (1986-2008) Verification (1955-1985) Statistics Value Statistics Value RP 0.468 RP 0.618 RR 0.438 RR 0.615 RS 0.348 RS 0.518 RSQ 0.219 RE 0.219 CE 0.355

From the Fig. 6.1, it is clear that the reconstructed data displayed similar trends and amplitude as the observed data over most of the common period. The reconstructed temperatures were higher (by more than 1oC) than the observed data in the following years: 1957 and 1979. The reconstructed temperatures were found to be lower than the recorded data by more than 1oC in 1956 and 1980. No year represented the most serious disagreement between the two records.

127

Fig. 6.2 shows the linear relationship between instrumental record and reconstructed data for the same period. The two data sets are fairly correlated (Pearson product moment correlation; RP=0.53) by correlation equation of y=8.90+5.11x.

22 21 20 19 18 17 16 15 1950 1960 1970 1980 1990 2000 2010

reconstructed actual

Figure 6.1: Actual (red) and reconstructed (dashed) March-June temperature reconstruction during common period 1955-2008. The estimation explains 38.16% of the actual variance in this common period.

Figure 6.2: Scatter plot of the observed and reconstructed temperature of the data that were used for early calibration period (1955-1985). X-xis represented actual data while Y-xis represented reconstructed data for the common period (1955-2008). On the basis of RBar and EPS statistics (see Section 4); Picea smithiana from Jutial reconstruction is reliable back to 1523 so the proxy record is more than 400 years longer than the

128 instrumental record. The mean temperature obtained over the entire period of reconstructed data was 18.22oC just lower than the mean of March-June temperatures calculated from Gilgit actual data (actual data=18.42oC). The Table 6.3 displays that values of basic statistics are quite similar with each other.

Table 6.3: Statistics for the March-June actual and reconstructed (1955-2008) temperature data

Actual data Reconstructed data N (years) 54 54 Mean (oC) 18.42 18.32 Median (oC) 18.40 18.42 Standard deviation (oC) 0.88 0.84 Standard Error (oC) 0.12 0.11 Minimum (oC) 16.7 16.8 Maximum (oC) 20.8 20.4

The reconstruction March-June temperatures for the entire period are shown in the Fig. 6.3. Ten years running mean window described the tendency of the warming and cooling trends (Fig. 6.5) over a decadal time scale (short term period). Spring temperatures have been steadily increasing over the eleven intervals of the record considered as warm periods; 1564-1573, 1590-1608, 1615-1626, 1630-1650, 1692-1714, 1768-1787, 1794-1817, 1821-1834, 1854-1869, 1909-1922 and 1953-1979 whereas it have been decreasing in ten intervals of the record also known as cold periods; 1537-1551, 1574-1589, 1651-1671, 1675-1691, 1735-1749, 1755-1768, 1837-1853, 1870-1908, 1923-1938 and 1980-1991. The warmest anomaly occurred in 1602, 1807 and 1862 with the temperature values 18.81, 18.71 and 18.77 respectively. The coldest anomaly throughout the data was recorded in 1684 and 1901 where the temperature values were 17.47 and 17.41 respectively.

129

Tables 6.4 (a) and (b): The warmest 6.4(a) and coldest 6.4(b) non overlapping 25 year‟s periods and the warmest 6.4(a) and coldest 6.4(b) non overlapping 10 years periods of Gilgit and Hunza Valleys temperature reconstruction in oC. Dep. = departures from the mean of 18.22oC over the entire reconstruction period of 1523-2000.

Table 6.4(a): Warm periods.

Warm 25 years intervals Warm 10 years intervals Intervals Mean Departures Intervals Mean Departures 1630-1654 18.42 0.20 1537-1546 18.43 0.21 1768-1792 19.91 1.69 1574-1583 18.93 0.71 1794-1818 18.41 0.19 1615-1624 18.88 0.66 1890-1904 19.25 1.03 1630-1639 18.42 0.20 1953-1977 18.33 0.11 1675-1684 18.64 0.42 1768-1777 19.91 1.69 1794-1803 18.40 0.18 1890-1899 19.20 0.98 1900-1909 18.61 0.93 1964-1973 18.41 0.19

Table 6.4(b): Cold periods.

Cold 25 years intervals Cold 10 years intervals Intervals Mean Departures Intervals Mean Departures 1651-1675 18.07 -0.15 1564-1573 17.97 -0.25 1692-1716 17.70 -0.52 1600-1609 17.22 -1.00 1870-1894 17.80 -0.42 1661-1670 15.86 -2.36 1692-1701 16.93 -1.29 1703-1712 17.80 -0.42 1735-1744 17.85 -0.37 1779-1788 17.15 -1.07 1821-1830 17.93 -0.29 1854-1863 17.42 -0.80 1870-1879 17.84 -0.36 1910-1919 16.52 -1.70 1923-1932 17.82 -0.40 1980-1989 17.29 -0.93

130

Tables (6.4a and 6.4b) describe the coldest and warmest periods over 10 years and 25 years non overlapping intervals in the reconstruction. The warmest ten-years periods occurred during 1768- 1777 where the mean temperature was 19.91oC i.e. 1.69oC above the average 1523-2008 reconstructed mean. This warm period was also evident in the 25 years reconstructed interval. The coldest 25-year interval was seen during 1651-1675 where the lowest anomaly happened in (1661-1670) ten years interval with the mean temperature of 15.86oC lower than 2.36oC than the average reconstructed temperature.

Overall, the results reveal that the warmest periods occurred in the second half of 18th century (roughly concurrent with the large “mega-droughts” noted for Southeast Asia and beyond, by Buckley et al. (2007; 2010), Cook et al. (2010), D‟ Arrigo et al. (2012), and Sano et al. (2009) and the very last and early periods of 19th and 20th centuries, respectively. The coldest periods were observed for the 1660s, 1780s and 1910s. It is also calculated that after the long 20 years warm interval (1890-1909), cold ten years occurred later on i.e. 1910-1919.

To find out temperature anomalies over centennial time scale (long term period); we used 100 years moving average presented in Fig. 6.5.The 17 century was apparently cool in first half of the 17th century and warm during 1640-1660 which then changes to average over the entire period of 1740 in the present reconstruction. Thereafter, 1740 to 1760 experienced cooling period with the minimum cooling anomaly (1748-1788). The 19th century recorded the protracted warmth period including last two decades of 18th century and first decade of 20th century. The highest anomaly was observed during 1860-1880. The 20th century was on the average scale representing absence of any warmth period during the century.

131

C o 19 18 17

Temperature 16 15 1500 1600 1700 1800 1900 2000 year

Figure 6.3: The Gilgit March-June reconstruction over the entire period of 1523 to 2008.

Figure 6.4: 10 years running mean window describes the trend of warm and cold years. Upper line of graph represents warm years and lower line signifies cold yearsduring 25 years intervals.

20 19 18 17

Temperature(C) 16 15 1500 1550 1600 1650 1700 1750 1800 1850 1900 1950 2000 years

Figure 6.5: 100 years running mean window describes the trend of warm and cold years. Red line of the graph represents 100 mean running window during 1620-2000.

132

6.4-Comparison with Pinus gerardiana reconstruction

Pinus gerardiana Chaprot showed good correlation in correlation function (see chapter 5, Table 5.1) and also represented strong variance in transfer function (Table 6.1), therefore picked for reconstruction also. The four months similar to Jutial were chosen for reconstruction over the period of approximately 150 years (1850-2008). The Chaprot reconstruction was then compared with the Jutial reconstruction within the same period (1850-2000). Figure 6.6- shows the Picea smithiana Jutial and Pinus gerardiana Chaprot temperature reconstruction plots. While the two records exhibited a reasonable degree of similarity over most of the common periods, there are shorter intervals where the two proxies were significantly different. From the beginning up to 1870, Jutial temperatures were lower than Chaprot temperatures with the exception of a short interval that occurred where Jutial temperatures were tend to higher for the next five years and then long term decline happened from 1876 to 1920. Higher Jutial temperatures (compared to Chaprot reconstruction) occurred in the late 1950s to 1990s and then dropped in the last few years.

C 18.5 o

Temperature 17.5

17 1825 1850 1875 1900 1925 1950 1975 2000

10 per. Mov. Avg. (pgchap) 10 per. Mov. Avg. (psjutl)

Figure 6.6: Comparison between the two temperature reconstructions based on 10 years moving average. Blue line indicates Pinus gerardiana Chaprot while red line shows Picea smithiana Jutial reconstruction. The interesting periods where Jutial reconstruction caused a very sharp drop were 1860-1870 and 1910-1920. Over the entire common period (1840-2008), the mean temperature calculated in Jutial was 17.83oC, less than the Chaprot reconstruction where the mean was 18.35oC.

133

The correlation between two records was found highly significant (The Pearson‟s product moment correlation coefficients: RP = 0.60, P<0.001) over the common interval of years (1840- 2008). However, as described earlier, the strength of correlation was not stable over time. Changing correlations between 25 years sub-periods between two reconstructions are well explained (Fig. 6.6). Here, we divided the reconstructed data into 50 years interval for comparison.

C o 20

16 Jutial Jutial reconstruction in

15 15 16 17 18 19 20 21 22 Chaprot reconstructed in oC

Figure 6.7: Scatter plot of the Jutial and Chaprot temperature reconstructions based on 168 years of data (1840-2008). The three intervals were found to be highly significant (P<0.001), but there were less or more values among three periods. The highest correlation occurred during second half of the 20th century (Pearson‟s product moment correlation coefficients: RP = 0.60, P<0.001). The weakest agreement was seen in first half of the 20th century where RP = 0.57 although still significant (P<0.001). The most disagreement years were observed i.e. 1945 and 2000 where the two reconstructions have the values 16.45oC, 18.62oC (Jutial) and 18.68oC, 20.34oC (Chaprot) respectively.

134

6.5-Discussion

Picea smithiana tree-rings performed well as predictors of Gilgit temperatures. This is directed by a good amount of variance explained and also by stability of the models for both early and late periods. The calibration and verification statistics showed similar and reliable values. Pinus gerardiana explained the highest variance in the same months probably because of existence of this species at lower elevation. The negative influence of spring temperature on tree growth means that the high temperature leads to internal water deficit in the early growing season due to increased soil moisture loss by evapotranspiration.

Other temperature sensitive tree-ring records in the surrounding area offer a reference to the validation of our reconstruction in the Asian region. Published tree ring temperature records for India (e.g. Hughes, 1992; Borgaonkar et al. 1996; Yadav et al. 1997), Western Central Asia (Esper et al. 2002), Nepal (Cook et al. 2003) and Tibet (e.g. Wo, 1992; Brauning, 1994; Wo and Shao, 1995) are difficult to compare with the present study results, possibly due to the differences in reconstructed seasons or the differences in tree ring species. However, some similarities have been noted at inter-decadal and century time-scale.

Esper et al. (2002) described the past 1300 year climatic history for Western Central Asia from tree-rings. The studies were carried out from Tien Shan of Kirghizia and Northwest Karakorum of Pakistan. Most of the samples were collected from Juniper. Juniperus turkestanica chronology from Tien Shan site was found to be significantly correlated with Tien Shan summer temperature (July-August) of nearby station (Narin). The warmest decades occurred between AD 800-1000 and the coldest decades occurred in the first half of the 17th century. We found cold years in the first decade of the 17th century in decadal time scale, and in the first half of 17th century over the century time scale. The results are in agreement with the conclusions of Esper et al. (2002).The cold periods of the early 20th century were observed in the Yantze River temperature reconstruction on the Tibetan Plateau which agrees well in present study (Liang et al. 2008), however, the difference occurred in summer and spring season. Although Esper et al. (2002) and Liang et al. (2008) studies were based on summer temperature reconstruction (July-August) our study was based on spring temperature (March-June), and yet still some of the results match.

135

The reason may be the occurrence of tree-ring species. The current reconstructed study sites (Gilgit and Hunza) and Southern Tien Shan are some 500 km distant, and the Tibetan Plateau is a similar distance from the current study sites. The current reconstructed sites were present in Northwest Karakorum and Himalayan regions, and therefore correlate with spring season variability. Sites in the Northwest Karakorum are influenced by westerly winds and as well as monsoonal depression whereas sites (used in Esper et al. 2002 reconstruction) in Southern Tien Shan and Tibetan Plateau are affected by strongly continental climate without transportation of precipitation from Arabian Sea. Therefore chronologies from Tien Shan and Tibetan Plateau affected by summer temperature whereas our study sites influenced by spring temperature. Another possibility is the selection of tree species. Esper et al. (2002) reconstruction based on Junipers which was affected by summer temperature as in case of present study where Juniperus excelsa from both sites also showed negative correlation in the months of June and July. Current reconstruction was established using Picea smithiana out of three species having influence with spring temperatures.

Several studies have been carried out in Himalayan region that propose that climatic variability in this area correlate with spring temperature and El-Nino/Southern Oscillation (Dey and Bhanukumar, 1983; Douville and Royer, 1996; Li and Yanai, 1996; Overpeck et al. 1996).Various tree-ring chronologies in the Western Himalayan Region have been used in extending seasonal climatic records (e.g., Bhattacharya and Yadav, 1999). In this contest, Hughes and Davis (1987) made a pioneer effort in the analysis of Abies pindrow and Picea smithiana in the Kashmir valley which later contributed detailed reconstructions of mean temperatures for spring (April-May), late summer (August-September) precipitation since 1780 at Srinagar, Jammu and Kashmir based on width and density of annual rings of Abies pindrow (Hughes, 1992). Borganonkar et al. (1996) reconstructed pre-monsoon (March-May) temperature back to 19th century using ring-width data of Cedrus deodara from Simla and Kanasar.

Yadav et al. (1997) extended this data by April-May temperature reconstruction back to AD 1698 from Western Himalayan region of India. Cedrus deodara, Picea smithiana and Pinus wallichiana were used in overall analysis. This study showed that the first three decades of the 18th century (1700s-1720s) were warm in which warmest anomaly occurred about 1713-1722. The 19th century showed prolonged warmth in 1850s-1870s, while the period of 1810s-1830s

136 was remarkably cool. Any warming trend in recent decades of the 20th century was not evidenced in our reconstruction.

Our reconstruction has warm periods during the 19th century from 1850 to 1870 with the warmest 10 year mean anomaly for 1854-1869. The similar warmth period from 1850-1870 were also observed by Yadav et al. (1997) but according to Bradley and Jones (1993), this century has been described monotonically cool in high-altitude northern hemisphere regions. The trend of parallel cooling decades was also experimented during 1730-1750 in present reconstruction and the reconstruction of Yadav et al. (1997). The 1730s were also found to be cool in the spring temperature reconstruction valley of Kashmir in the northwestern Himalaya (Hughes, 1992, 1994). The presented reconstructed temperature for 1900s, 1920-1930 and 1980s is also in agreement with the reconstruction of Yadav et al. (1997) and also similar with the trend noted in the spring temperature reconstruction in recent decades of the 20th century in the valley of Kashmir (Hughes, 1992, 1994). Mean March-June temperature reconstruction of present study in inter-decadal time scale pointed out a weakened elevated temperature towards later parts of 20th century in inter-decadal time scale which was believed to be the result of anthropogenic activities (deforestation). The cool decades for the late 1830s were also experienced in current debate with those in the British Isles, Central Europe, Scandinavia, the polar Urals and central Korea (Hughes, 1994). So the impact of these cool decades over climate and tree growth through these larger areas appears to have been severe.

Cook et al. (2003) reconstructed the past more than 400 years of Nepali temperature by using the long term Khatmandu record for the five-month season of February-June. Our March-June reconstruction compares well with the Cook et al. (2003) Nepal reconstruction. Interestingly, there is no evidence for the abrupt extreme cold events in 1815-1822 (from the Tambora eruption in Indonesia) in present reconstruction and Karakorum tree ring data reconstruction made by Esper et al. (2002). Nor is there any support for it in Srinagar or Simla reconstructions. In contrast, these cold events have been observed in Wo and Shao (1995), Brauning (1994), Liang et al. (2008) and Cook et al. (2003) reconstructions from eastern Tibet and Nepal respectively. It is plausible that the 1815-1822 cold event had its greatest impact over eastern Nepal and Tibet and did not spread far into the western Himalayas and Karakorum. The tree ring network used

137 for the present reconstruction is geographically weighted towards the western Himalayas and Karakorum.

Distinct cold and warm intervals can be seen from the current investigation, including a period consistent with “Little Ice Age”. The 17th century appears to be cooling whereas the 19th century is markedly higher than that of other centuries. NASA (2011) describes the occurrence of “Little Ice Age” in three particularly cold intervals: 1650s, 1770s and 1850s which spread throughout Europe, North America and Asia. The above periods were marked for the expansion of mountain glaciers. These periods were also witnessed where we have coolest period with the highest departures from the mean temperature. The warmest period of the last millennium (Medieval warm period) could not be investigated as our reconstructed data is too short. Some studies show that this warm period occurred before the occurrence of LIA (NASA, 2011; Mann et al. 2009).

This all talked about variations in temperature at the level of individual countries. A comparison of our reconstruction temperature series for Gilgit and Hunza valleys agrees well against temperature reconstruction covering the past millennium in central Asia (Shi et al. 2012). Typical cooling and warming patterns appear to be better reflected in our results and Asian millennium reconstruction (Shi et al. 2012). Comparing the two reconstructions (present and Shi et al. 2012), some common features are evident. For example, the warming of the 19th and the cooling of 17th century are pronounced in both. The results indicate that a cooling trend occurred throughout central Asia in 17th century, before temperature began to rise in the 18th century before reaching a maximum by the end of 19th century.

6.6-Conclusion

Ring-width chronologies of Picea smithiana were used to reconstruct mean March-June (spring) temperatures back to A.D. 1523. The record is based on a simple linear regression technique where we calibrated our Picea smithiana chronology against March-June temperature. The calibrated model explained 38.16% of the variance in temperature and passed all calibration and verification tests at the 95% level of significance. The reconstruction exhibits a strong positive correlation with the instrumental data and is characterized by annual to multiyear variations of cool and warm periods. The 19th century experienced a prolonged warmth period over a centennial scale with the highest temperatures in the 1850s-1870s. The coldest 20 year period

138 was 1890-1910. The consistency observed on decadal scales between the present reconstruction of March-June temperatures, the mean temperatures of April-May reconstruction from the valley of Kashmir (Hughes, 1992, 1994) and the mean temperature reconstruction of April-May in the Western Himalaya of India (Yadav et al. 1997) indicates the potential for reconstructing regional-scale climatic changes using tree rings. The “Little Ice Age” can also been observed from our reconstruction, and matches with the reported cooling intervals of NASA. It is recommended that collecting more samples from other locations of the area from sensitive trees would produce handful results to extend our chronology in finding of Medieval warm anomaly.

139

Refernces

Aceituno, P., M. Prieto, M. Solari, A. Martenez, G. Poveda, M. Falvey, 2009. The 1877-1878 El- Nino episode: associated impacts in South America. Climate Change 92:389-416. Ahmed, K., and S. Shah, 2007. Physical and microbial assessment of drinking water of Nomal Valley, Northern Areas, Pakistan. Pak. J. of Zool., 39: 367-373.

Ahmed, M., 1987. Dendrochronology and its scope in Pakistan. In modern trends of Plant Science Research in Pakistan. Dept. of Botany, Peshawar University. Ahmed, M., 1988a. Population structure of some planted tree species in Quetta. J. Pure and Applied Science, 7:25-29. Ahmed, M., 1988b. Problems encountered in age estimation of forest tree Species. Pak. J. Bot., 20(1): 143-145. Ahmed, M., 1989. Tree-ring chronologies of Abies pindrow (Royle) Spach., from Himalayan region of Pakistan. Pak. J. Bot., 21(2): 347-354. Ahmed, M., and A. M. Sarangzai, 1991. Dendrochronological approach to estimate age and growth rate of various species from Himalayan region of Pakistan. Pak. J. Bot., 23(1): 78-89. Ahmed, M., and A. M. Sarangzai, 1992. Dendrochronological potential of few tree species from Himalayan region of Pakistan. A preliminary investigation. J. Pure Applied Science, 11(2):65-72. Ahmed, M., and H. Naqvi, 2005. Tree-ring chronologies of Picea smithiana and its quantitative description from Himalayan region of Pakistan. Pak. J. Bot., 37(3): 697-707. Ahmed, M., J. Palmer, N. Khan, M. Wahab, P. Fenwick and E. R. Cook, 2011. The dendroclimatic potential of Conifers from Northern Pakistan. Dendrochronologia 29: 77- 88.

Ahmed, M., M. Wahab and N. Khan, 2009. Dendroclimatic investigation in Pakistan, using Picea smithiana (Wall) Boiss. Preliminary result Pak. J. Bot., 51(5): 2427-2435. Ahmed, M., M. Wahab, N. Khan, M. U. Zafar and J. Palmer, 2010b.Tree ring chronologies from upper Indus Basin of Karakorum Range, Pakistan. Pak. J. Bot., SI 42:295-307, 2010. Ahmed, M., N, Khan and M. Wahab, 2010a. Climatic response function analysis of Abies pindrow (Royle) Spach. Preliminary results. Pak. J. Bot., 42(1): 165-171. Ahmed, M., N. Khan, M. Wahab, U. Zafar and J. Palmer, 2012. Climate/Growth correlations of tree ring species of the Karakorum Range, North Pakistan. IAWA Journal 33: 51-61. Ahmed, M., T. Hussain, A. H. Shaikh, S. S. Hussain, M. F. Siddique, 2006. Phytosociology and structure of Himalayan forests from different climatic zones of Pakistan. Pak. J. Bot., 38(2): 361-363. Anwar, M. S., S. Lateef and G. M. Siddiqui, 2010. Bacteriological quality of drinking water in Lahore. Biomedica. 26:66-69.

140

Archer, D., and S. Blenkinsop, 2010. Heterogeneity and spatial representativeness of station climate and tree ring records as climate change indicators in the Upper Indus Basin. Talk presented at International Conference/Workshop on Dendrochronology held at Federal Urdu University, Karachi. November 15, 2010. Azhar, H., 1996. Determination of drinking water quality from source to consumer in Islamabad. (Dissertation) Health services academy, Faculty of Medicine, Quaid-e-Azam University Islamabad. Benjamin, M., 2002. Water Chemistry. New York: McGraw-Hill. Bhattacharya, A., C. Vadana and R. R. Yadav, 1999. Tree ring studies in the Eastern Himalayan region: Prospects and Problems. IAWA Journal 20(3): 317-324. Bhattacharyya, A., and R. R. Yadav, 1999.Climatic reconstruction using tree ring data from tropical and temperate regions of India-A Review.IAWA Journal 20(3): 311-316. Bhattacharyya, A., V. C. Lamarche and M. K. Hughes, 1992. Research Report, Tree Ring Chronologies from Nepal. Tree Ring Bulletin 52: 1992. Biber, K., 2011. Mercury fate and transport in Hunza and Gilgit rivers, Northern Areas, Pakistan- A numerical modeling approach. GSA annual meeting in Minneapolis, 241-28.

Bilham, R., G. B. Pant and J. C. Jacoby, 1983. Dendroclimatic potential of Conifer trees from the Sir Sar range in the Karakoram. Mausam 7: 45-50. Blasing, T. J., A. M. Solomon and D. N. Duvick, 1984. Response function revisited. Tree Ring Bulletin. 44: 1-15. Borgaonkar, H. P., G. B. Pant, R. Kumar, 1994. Dendroclimatic reconstruction of summer precipitation at Srinagar, Kashmir India since the last 18th century. The Holocene 4(3): 299-306.

Borgaonkar, H. P., G. P. Pant, K. R. Kumar, 1996. Ring width variations in Cedrus deodara and its climatic response over the Western Himalaya. International Journal of Climatology 16:1409-1422.

Borgaonkar, H. P., S. Ram and A. B. Sikdar, 2009. Assessment of tree-ring analysis of high elevation Cedrus deodara D. Don from Western Himalaya (India) in relation to climate and glacier fluctuations. Dendrochronologia 27: 59-69.

Bradley, R., and P. Jones, 1993. „Little Ice Age‟ summer temperature variations: their nature and relevance to recent global warming trends. The Holocene 3, 4: 367-376. Brauning, A., 1994. Dendrochronology for the last 1400 years in eastern Tibet. Geo Journal 34: 75-95. Briffa, K. R., and T. J. Osborn, 1999. Seeing the wood from trees. Science 284: 926-27. Briffa, K., and P. D. Jones, 1990. Basic chronology statistics and assessment. In E. R. Cook and L.A. Kairiukstis (Eds.), Methods of Dendrochronology: application in the environmental sciences (pp. 137-152). Dordrecht: Kluwer Academic Publishers.

141

Brown, D., and P. Rothery, 1993. Models in Biology: mathematics, statistics and computing. Chichester: John Wiley and Sons Ltd. Buckley, B. M., K. J. Anchukaitis, D. Penny, R. Fletcher, E. R. Cook, M. Sano, C. Nam, A. Wichiekeeo, T. T. Minh and T. M. Hong, 2010. Climate as a contributing factor in the demise of Angkor, Cambodia. Proceedings of the National Academy of Sciences of the United States of America. Buckley, B. M., K. Palakit, K. Duangsathaporn, P. Sanguantham and P. Prasomsin, 2007. Decadal scale droughts over northwest Thailand over the past 448 years: links to the tropical Pacific and Indian Ocean sectors. Climate Dynamics 29:63-71. Champion, G. H., S. K. Seth and G. M. Khattak, 1965. Forest types of Pakistan. Pakistan Forest Institute, Peshawar., pp-238. Chaudhary, V., K. P. Singh, G. Jacks and P. Bhattacharya, 2001.Ground water contamination at Ludhiana, Punjab, India. J. Indian Water Works Associa.,, 33: 251-261.

Chhetri, P. K., 2008. Dendrochronological analysis and climate change perceptions in Langtang National Park, Central Nepal. Climate change and disaster impact reduction. School of Applied Sciences, North Umbria University. Chhetri, P. K., and S. Thapa, 2010. Tree ring and climate change in Langtang National Park, Central Nepal. Our Nature 8:139-143. Climate Normals Of Pakistan, (1961-1990). Pakistan Meteorological Department, Karachi. Cook, E. R., 1985. A time series analysis approach to tree ring standardization. PhD dissertation, University of Arizona, Tucson, USA. Cook, E. R., and L. A. Kariukstis, 1990. Methods of Dendrochronology, Springer, New York. Cook, E. R., and R. L. Holmes, 1986. Guide for computer program ARSTAN. Adapted from user manual for program ARSTAN in Tree Ring Chronologies of Western North America: California eastern Oregon and north great Basin. Cook, E. R., C. Woodhouse, C. Eakin, D. Meko, & D. Stahle, 2004. Long term aridity changes in the Western United States. Science 306:1015-1018. Cook, E. R., D. M. Meko, D. W. Stahle and M. K. Cleaveland, 1999. Drought reconstructions for the continental United States. Journal of Climate 12: 1145-1162.

Cook, E. R., J. Palmer, M. Ahmed, C. A. House, P. Fenwick, M. U. Zafar, M. Wahab and N. Khan, 2013. Five centuries of Upper Indus Flow from tree rings. Journal of Hydrology 486: 365-375. Cook, E. R., K. J. Anchukaitis, B. M. Buckley, R. D‟Arrigo, G. C. Jacoby and W. E . Wright, 2010. Asian Monsoon Failure and Mega drought During the Last Millennium. Science 328: 486-489. Cook, E. R., K. R. Briffa and P. D. Jones, 1994. Spatial regression methods in dendroclimatology: a review and comparison of two techniques. International Journal of Climatology 14: 379-402.

142

Cook, E. R., P. J. Krusic and P. D. Jones, 2003. Dendroclimatic signals in long tree ring chronologies from the Himalayas of Nepal. International Journal of Climatology 23 (7): 707-732. Cook. E. R., and Kariukstis. L. A, 1992. Methods of Dendrochronology (Application in the Environmental Science).

D‟Arrigo, R., K. J. Anchukaitis, B. M. Buckley, E. R. Cook, R. Wilson, 2012. Regional climatic and North Atlantic Oscillation signatures in West Virginia red cedar. Global and Planetary Change 84-85: 8-13. De, A. K., 2002.Environ Chem., (4thEdn.). New Delhi, India: New Age International Publishers (232).

Dey, B., and O. S. R. U. Bhanukumar, 1983.The Himalayan winter snow cover and summer monsoon rainfall over India. J. Geophys. Res., 88: 5471-5474. Douville, H., and J. F. Royer, 1996. Sensitivity of Asian summer monsoon to an anomalous Eurasian snow cover within the Metro-France GCM. Climate Dynamics 12: 449-466. Dubey, N., 2003. A comparative status of drinking water of Bhopal city filteration plants and ground water with special reference to heavy metals and organo chemical. PhD thesis, Barkatullah University Bhopal.

Esper, J., 1994. Versucheiner Klimarekonstruktionmit Halfe der Dendrochronolgie am Beispiel des Hunza- Tales in Karakorum. Unpubl. Master thesis, University of Bonn, 128pp.

Esper, J., A. Bosshard, F. H. Schweingruber and M. Wininger, 1995. Tree rings from the upper timberline in Karakorum as climatic indicators for the last 1000 years. Dendrochronologia 13: 79-88.

Esper. J., 2000. Long term tree variations in Juniperus at the upper timber line in the Karakorum (Pakistan).The Holocene 10(2): 253-260. Esper. J., Fritz. H and M. Winiger, 2002. 1300 years of climatic history for Western Central Asia inferred from tree rings. The Holocene 12(3): 267-277. Esper. J., K. Treydte, H. Gartner and B. Neuwirth, 2001. A tree ring reconstruction of climatic extreme years since 1427AD for Western Central Asia. Palaeobotanist 50 (2001): 141- 152.

FAO., 2003. Review of World Water Resources by country water reports, Rome, Italy, 110.

Farooq. M. A., 2008. The study of groundwater pollution resulting from storm water infilteration in Urban areas of Karachi. PhD dissertation. Federal Urdu University of Arts, Science and Technology.

Farooq. M. A., 2012. Surface water quality assessment and apportionment of pollution sources of lower Indus Basin from Kashmor to Keti-Bander using multivariate statistical technique-

143

A case study. Research project, Report submitted to Higher Education Commission, Pakistan.

Fenwick. P., 2003. Reconstruction of past climates using Pink Pine tree ring chronologies. PhD dissertation. Lincoln University, New Zealand. Fowler, A., 1988. Climate reconstruction from tree rings. Weather and Climate, 8: 33-45. Fritts, H. C., 1971. Dendroclimatology and dendroecology. Quaternary Research 1(4): 419-49. Fritts, H. C., 1974. Relationships of Ring widths in arid sites conifers to variations in monthly temperature and precipitation. Ecol. Mons., 44(4): 411-440. Fritts, H. C., 1976. Tree Ring and Climate. London: Academic Press. Grissino-Mayer, H. D., 2001. FHX2 – Software for analyzing temporal and spatial patterns in fire regimes from tree rings. Tree Ring Research 57: 115-124. Grissino-Mayer, H. D., and H. C. Fritts, 1997. The International Tree-Ring Data Bank: an enhanced global database serving the global scientific community. The Holocene. 7: 235-238.

Grissino-Mayer, H. D., and H. C. Fritts, 1997. The International Tree-Ring Data Bank: an enhanced global database serving the global scientific community. The Holocene. 7: 235-238.

Hayal, B., and B. Hiilya, 2009. Detection of seasonal variations in surface water quality using discriminate anlaysis. Environ. Monit Assess., (DOI 10.1007/s10061-009-0772-5).

Heizer, R. F., 1956. The first dendrochronologist. Amer. Antiq., 22, 186-188. Hemond, H., and E. Fechner-Levy, 2000. Chemical Fate and Transport in the Environment, 2nd Edition. San Diega, California: Elsevier Science. Holmes, R. L., 1992. Program ARSTAN (Version B-1992). Laboratory of Tree Ring Research, University of Arizona, Tucson, USA. Holmes, R. L., 1994. Dendrochronology Program Library- User manual, Laboratory of Tree- Ring Research University, Tucson, Arizona USA. Holmes, R. L., R. K. Adams, H. C. Fritts, 1986. Tree ring chornologies of western North America, California, eastern Oregan and northern Great Basin.Chronology Series IV, University of Arizona, Tucson, AZ, USA. Hughes, M. K., 1992. Dendroclimatic evidence from the western Himalaya. In Bradley RS, Jones PD (eds). Routledge Press: London; Climate Science Since A.D. 1500, 415-431. Hughes, M. K., 1994. Lessons from past global change. Journal of Korea Forest Energy 14: 23- 34. Hughes, M. K., and A. C. Davis, 1987. Dendroclimatology in Kashmir using tree ring widths and densities in sub alpine conifers. In: L. Kariukstis, Z. Bednarz and E. Feliksik (eds.). Methods of Dendrochronology. East/West approaches: 163-175. Warsaw, International Institute of Applied Systems Analysis/ Polish Academy of Sciences. Ihsanullah., K. Misal, N. Tariq, Khattak and A. Sattar, 1999. Determination of different contaminants in Selective drinking water samples. The Nus., 36: 91-97.

144

Islamuddin., 2010. Hydro-chemical investigations of high altitude alpine lakes of Gilgit and Ghizer districts, Northern Areas of Pakistan. M. Phil dissertation University of Peshawar, Pakistan. Jayaraman, P. R., T. Devi and T. Nayar, 2003. Water quality studies on Karamana River Thiruvananthapuram District South Karela, India. Pollution Research.22: 89-100.

Jones. P. D., R. S. Bradley and J. Jouzal., editors, 1996. Climate variations and forcing mechanism of the last 2000 years. Berlin springer. Khan, A. R., A. Haq, W. A. Khan, M. Akif, M. Khan and M. Riaz, 1999. Quality characteristics of potable water of Mardan city (Pakistan) and surrounding areas. J. Chem. Soc. Pak., 21: 106.

Khan, N., 2011. Vegetation ecology and dendrochronology of Chitral. PhD dissertation. Laboratory of Dendrochronology and plant ecology of Pakistan, Department of Botany Federal Urdu University of Arts, Science and Technology, Karachi, Pakistan. Khan, N., M. Ahmed and M. Wahab. 2008. Dendroclimatic investigation of Picea smithiana Boiss., from Afghanistan. Pak. J. Bot., 40(3): 1063-1070. Khan. A. R., M. Akif, S. Wadud and K. Khan, 1999. Pollution studies of Kabul River and its tributeries for the assessment of organic strength and Feacal coliform. J. of Chem. Soc. Pak., 21:111-113.

Lashari. K. H., A. L. Korai, G. A. Sahoto and T. G. Kazi, 2009. Limnological studies of Keenjhar Lake, District Thatta, Sindh, Pakistan. Pak. J. Ana. Env. Chem., 10: 39-47.

Li, C., and M. Yanai, 1996. The onset and inter-annual variability of the Asian summer monsoon in relation to land sea constrast. Journal of Climate 9: 358-375. Liang, E., X. Shao and N. Qin, 2008. Tree ring based summer temperature reconstruction for the source region of the Yangtze River on the Tibetan Plateau. Science Direct 61: 313-320. Luckman, B., R. Villalba, 2001. Assessing the synchroneity of glacier fluctuations in the western cordillera of the Americas during the last millennium. In: V. Markgraf, ed., Interhemispheric Climate Linkages. Academic Press, New York: 119-140. Lushuang, G., C. Zhang, X. Zhao and K. V. Gadow, 2010. Gender-related climatic response of radial growth in Dioecious Fraxinus mandshurica trees. Tree Ring Research 66(2): 105- 112.

Mann, M. E., Z. Zhang, S. Rutherford, R. Bradley, M. Hughes, D. Shindell, C. Ammann, G. Faluvegi and F. Ni, 2009. Global Signatures and Dynamical Origins of the Little Ice Age and Medieval Climate Anomaly. Science 27: 1256-1260. Mann, M., R. Bradley, & M. Hughes, 1998. Global-scale temperature patterns and climate forcing over the past six centuries. Nature 392: 779-787.

145

May, A., E. Mutasem, D. Mark, 2006. Factor influencing development of management strategies for the Abou Ali River in Lebanon. Sci. Total Environ., 362:15-30. McCarrison, R., M. B. Captain I.M.S, 1906. Observation on endemic goiter in Chitral and Gilgit Valleys, Agency Surgeon Gilgit Kashmir.

Muhammad, N., 2010. The express Tribune October 18th 2010. Nanan, D., F. White, I. Azam, H. Afsar and S. Hozhabri, 2003. Evaluation of water, sanitation and hygiene education intervention on diarrhea in northern Pakistan. Bulletin of World Health Organization 81:3.

NASA (National Aeronautics and Space Administration) retrieved 2011-02-28. Orvis, K. H., and H. D. Grissino-Mayer, 2002. Standardizing the reporting of abrasive papers used to surface tree ring samples. Tree Ring Research 58: 47-50. Overpeck, J., D. Anderson, S. Trumborte and W. Prell, 1996. The south west Indian monsoon over the last 18000 years. Climate Dynamics 12: 213-225. Owen, L.A. 2009. Latest Pleistocene and Holocene glacier fluctuations in the Himalaya and Tibet. Quaternary Science Reviews 28: 2150–2164. Pant, G.B, K. R. Kumar, H. P. Borgaonkar, N. Okada, T. Fujiwara and K. Yamashita, 2000. Climatic Response of Cedrus deodara tree-ring parameters from two sites in the western Himalaya. Canadian Journal of Forest Research 30(7): 1127-1135.

Parashar, C., N. Verma, S. Dixit and R. Shrivastava, 2008. Multivariate analysis drinking water quality parameters in Bhopal, India. Envion Monit Assess.,140: 119-122.

Participatory Adoptive Management Plan Handrap Lake, Ghizer district Northern Areas, Pakistan. WWF for Nature – Pakistan, Northern Areas. Pejman, A., G. Bindani, A. Karbassi, N. Mehrdadi and B. Esmail, 2009. Evaluation of spatial and seasonal variation in surface water quality using multivariate statistical techniques. J. Env. Sci Tech., 6: 467-476.

Pirzada, T., M. M. A. Talpur, A.D. Rajper, M. R. Shah and S. Naseem, 2011. Quality Assessment of Potable Water of Khairpur city. J.Chem. Soc. Pak., 33: 604-611.

Praus, P., 2005. Water quality assessment using SVD- based principal component analysis of hydrological data. Water S. A., 31: 417-422.

Richman, MB., 1986. Rotation of Principal Components. Journal of Climate 6: 293-335. Sano, M., B. M. Buckley and T. Sweda, 2009. Tree-ring based hydroclimate reconstruction over northern Vietnam from Fokienia hodginsir eighteenth century mega-droughts and tropical Pacific influence. Climate Dynamics 33: 331-340.

146

Sarwar. S., F. Ahmed and J. Khan, 2007. Assessment of the quality of Jhelum River water for irrigation and drinking at District Muzaffarabad Azad Kashmir. Sarhad. J. Agric., 23: 1041-1046.

Schweingruber, F. H., L. Kairiukstis and S. Shiyatov, 1990. Sample selection. In E. R. Cook and L. A. Kairiukstis (Eds.), Methods of dendrochronology: applications in the environmental sciences (pp. 23-25). Dordrecht: Kluwer Academic Publishers. Seckler D., U. Amerasinghe, D. Molden, R. Desilva and R. Barker (1998). World water demand and supply 1990-2025: Scenerios and issues, International Water Management Institute, Colombo, Sri Lanka, 40.

Shah, T., and Deb Roy, 2002. Intensive use of ground water in India, International Water Management Institute, IWMI-TATA Water Policy Program, Anand, India, 16.

Shar, A. H., Y. F. Kazi, M. Zardari and I. H. Soomro, 2009. Bacteriological Quality of Drinking water of Sukkur City. Pak. J. Med. Res., 48:414-420. Shar, A. H., Y. F. Kazi, N. A. Kanhar, I. H. Soomro, S. M. Zia and P. B. Gumro, 2010. Drinking water quality in Rohri city, Sindh, Pakistan. African J. Biotechnol., 42: 7102-7107. Shi, F., F. Yang, T. Nakatsuka, M. Sano and Pages Asia 2k Project member, 2012. Multi- centennial temperature variability in Asia over the past 1000 years. (In press).

Shrestha, S., and F. Kazama, 2007. Assessment of surface water quality using multivariate statistical techniques: A case study of Fuji River Basin, Japan. Environmental Modeling & Software, 22: 464-475.

Singh, J. and R.R. Yadav, 2007.Dendroclimatic potential of millennium-long ring width chronology of Pinus gerardiana from Himachal Pradesh, India. Current Science 93(6): 833-837. Singh, J., R. R. Yadav and M. Wilmking, 2009. A 694-year tree ring based rainfall reconstruction from Himachal Pardesh, India. Climate Dynamics 33(7): 1149-1158.

Singh. J., R.R. Yadav, B. Dubey and R. Chaturvedi, 2004. Millenium long tree ring width chronology of Himalayan cedar from Garhwal Himalaya and its potential in climate change studies. Current Science 86(4): 590-594. Speer, J. H., 2010. Fundamental of Tree Ring Research. The University of Arizona Press Tucson. Standard methods for the examination of water and waste water treatment, American Public Health Association (APHA), American Water Works Assocation (AWWA), Water Environmental Federation, 2003. Stokes, M. A. and T. L. Smiley, 1968. An introduction to Tree Ring dating. The University of Chicago Press, Chicago Illinois (Reprinted by the University of Arizona Press, Tucson.

147

Tabachnick, B. G., and L. S. Fidell, 1983. Using multivariate statistics, Harper and Row, New York, pp. 509. Tranquillini, W., 1964. The physiology of plants at high altitude. Annual review of plant physiology 15: 345-362.

Treydte, K. S., G. H. Schlesr, G. Helle, D. C. Frank, M. Winige, G. H. Hang and J. Esper, 2006. The twentieth century was the wettest period in northern Pakistan over the past millennium. Nature 440(27): 1179-1182. Tyagi, N. K., 2003. 5 Managing saline and alkaline water for high productivity. Water productivity in agriculture: Limits and opportunities for improvement. Central Soil Salinity Research Institute, Haryana, India. Vega, M., R. Pardo, E. Barrado and L. Deban, 1998. Assessment of seasonal and polluting effects on the quality of river water by exploratory data analysis. Water Research. 32: 3581- 3592. Wahab, M., 2011. Population dynamics and dendrochronological potential of Pine tree species from District Dir. PhD dissertation. Department of Botany, Federal Urdu University of Arts, Science and Technology, Karachi, Pakistan. Welch, P. S., 1952. Limnological methods. New York: Mc-Graw Hill.

WHO‟s Guidelines for drinking water quality setup in Geneva, Switzerland, 1993. Wigley, T. M. L., K. R. Briffa and P. D. Jones, 1984. On the average value of correlated time series, with application in dendroclimatology and hydrometeorology. Journal of Climate and Applied Meteorology 23: 210-213. Wu, X., 1992. Dendroclimatic studies in China. In Climate Since A.D. 1500, Bradley RS, Jones PD (eds). Routledge Press: London. Wu, X., and X. Shao, 1995. Status and prospects of dendrochronological study in Tibetan Plateau. Dendrochronologia 13: 89-98. Xiong, L., N. Okada and T. Fujiwara, 2000. The dendrochronological Potential of ten species in three Georges Reservoir Region of China. IAWA Journal 21(2): 181-196.

Yadav, R. R., 2011. Long term hydroclimatic variability in monsoon shadow zone of western Himalaya, India. Climate Dynamics 36: 1453-1462.

Yadav, R. R., and J. Singh, 2002. Tree ring analysis of Taxus buccata from the Western Himalaya, India, and its Dendroclimatic potential. Tree Ring Research58 (1/2): 23-29.

Yadav, R. R., and J. Singh, 2002. Tree-ring based spring temperature patterns over the past four centuries in western Himalaya. Quarternary Research 57: 299-305. Yadav, R. R., and W. K. Park, 2000. Precipitation reconstruction using ring-width chronology of Himalayan Cedar from Western Himalayan, India: Prelimanary results. Proc. Indian Acad, Sci., 109(3): 339-345.

148

Yadav, R. R., W. K. Park and A. Bhattacharya, 1999. Spring-temperature variations in Western Himalaya, India as reconstructed from tree ring: AD 1390-1987. The Holocene 9 (1):85- 90.

Yadav, R. R., W. K. Park and A. Bhattacharyya, 1997. Dendroclimatic reconstruction of April- May temperature fluctuations in the western Himalaya of India since A.D. 1698. Quaternary Research 48: 187-191.

Yan, T., and M. J. Sadowsky, 2007. Determining sources of fecal bacteria in waterways. Environ. Monit. Assess., 129: 97-106.

Yu, L., S. Jiangfeng, V. Shishov, E. Vaganov, Y. Yinke, C. Qiufang, S. Junyan, W. Lei and I. Djanseitov, 2004. Reconstruction of May-July precipitation in the north Helan mountain, Inner Mangolia since AD 1726 from tree ring late-wood widths. Chinese Science Bulletin 49(4): 405-409. Zafar. M. U., M. Ahmed, M. Akbar and A. Hussain, 2010. Standard Tree Ring Chronologies of Picea smithiana from two new sites of Northern Area Pakistan. World Applied Sciences Journal 11(12):1531-1536. Zhang, Q. B., G. Cheng, T. Yao, X. Kang and J. Huang, 2003. A 2326 year tree ring record of climatic variability on the northeastern Qinghai-Tibetean Plateau. Geophysical Research Letters 30(14): 1739. Zubair, A., H. N. Abbasi and A. Hussain, 2009. Hydro chemical interpretation of storm water impact on Ground water using Factor Analysis. J. Env. P. Soc., 3: 117-123.

149