Soil Organic Carbon Studies in Semi-Arid Tropical India

MISSING IMAGE

Material Information

Title:
Soil Organic Carbon Studies in Semi-Arid Tropical India Sources and Stabilization
Physical Description:
1 online resource (141 p.)
Language:
english
Creator:
Winans, Kiara Sage
Publisher:
University of Florida
Place of Publication:
Gainesville, Fla.
Publication Date:

Thesis/Dissertation Information

Degree:
Doctorate ( Ph.D.)
Degree Grantor:
University of Florida
Degree Disciplines:
Soil and Water Science
Committee Chair:
Reddy, Konda R
Committee Co-Chair:
Hanlon, Edward A
Committee Members:
Silveira, Maria L
Jones, James W
Wani, Suhas

Subjects

Subjects / Keywords:
autocorrelation -- carbon -- fractions -- isotopes -- kriging -- management -- modeling -- quality -- soil -- uncertainty
Soil and Water Science -- Dissertations, Academic -- UF
Genre:
Soil and Water Science thesis, Ph.D.
bibliography   ( marcgt )
theses   ( marcgt )
government publication (state, provincial, terriorial, dependent)   ( marcgt )
born-digital   ( sobekcm )
Electronic Thesis or Dissertation

Notes

Abstract:
Soil organic carbon (SOC) is one of the key constituents of soil organic matter (SOM), and thus is used as an indicator of SOM.  The SOM includes floral and faunal residues at all stages of decomposition (SSSA, 1979).  As organic material decays, it stabilizes in soil and then plays a critical role in holding water and nutrients available for plant growth and in supporting soil structure.  The SOM management is essential in the semi-arid tropics areas characterized by savannah vegetation and soils with low SOM and low nutrient reserves (El-Swaify et al., 1985).  Properly managing SOM may increase SOC sequestration, nutrient availability, soil moisture retention, and increased crop productivity (Lal, 2006).  Further, an accurate estimation of SOC stocks on a per hectare basis at unsampled sites requires an efficient sampling protocol.  Several studies (e.g., Wu et al., 2009) demonstrated how geostatistics can increase the efficacy of sampling protocols used to measure SOC, thereby more readily identifying management practices that improve soil quality and reduce carbon emissions.  Finally, for use of SOC data in a predictive mode to test various scenarios (e.g., increased temperature) is possible in a modeling environment.  However, uncertainty or variability is presented by both measured and modeled SOC pools.  Narrowing the gap between modeled and measureable SOC pools will increase the opportunity to validate the model simulations and subsequently increase the current understanding of SOC dynamics.  The overall objectives of the research are described in the following 1) To examine the relationships between SOC and selected C fractions (i.e., microbial biomass carbon (MBC) and stable isotope d13 C) using sorghum-based cropping systems and conservation agriculture practices on SOC accumulation and vertical distribution in semi-arid tropical soils. 2) To assess the short-term (16-months) soil C, and C fraction responses to conservation agricultural management practices in semi-arid tropical Vertisols: tillage, residue, and crops. 3) To employ spatial autocorrelation to determine the sampling density required for achieving a given accuracy in the estimation of the mean concentration of SOC, and to evaluate the benefits of different covariates in regression kriging. 4) To quantify uncertainty (or variability) in observed vs. predicted modeled and measured SOC pool values
General Note:
In the series University of Florida Digital Collections.
General Note:
Includes vita.
Bibliography:
Includes bibliographical references.
Source of Description:
Description based on online resource; title from PDF title page.
Source of Description:
This bibliographic record is available under the Creative Commons CC0 public domain dedication. The University of Florida Libraries, as creator of this bibliographic record, has waived all rights to it worldwide under copyright law, including all related and neighboring rights, to the extent allowed by law.
Statement of Responsibility:
by Kiara Sage Winans.
Thesis:
Thesis (Ph.D.)--University of Florida, 2012.
Local:
Adviser: Reddy, Konda R.
Local:
Co-adviser: Hanlon, Edward A.
Electronic Access:
RESTRICTED TO UF STUDENTS, STAFF, FACULTY, AND ON-CAMPUS USE UNTIL 2013-02-28

Record Information

Source Institution:
UFRGP
Rights Management:
Applicable rights reserved.
Classification:
lcc - LD1780 2012
System ID:
UFE0044388:00001


This item is only available as the following downloads:


Full Text

PAGE 1

1 SOIL ORGANIC CARBON STU DIES IN SEMI ARID TROPICAL INDIA: SOURCES AND STABILIZATION By KIARA S. WINANS A DISSERTATION PRESENTED TO THE GRADUATE SCHOOL OF THE UNIVERSITY OF FLORIDA IN PARTIAL FULFILLMENT OF THE REQUIREMENTS FO R THE DEGREE OF DOCTOR OF PHILOSOPHY UNIVERSITY OF FLORIDA 2012

PAGE 2

2 2012 Kiara S. Winans

PAGE 3

3 I dedicate this dissertation t o all who have supported me throughout the process

PAGE 4

4 ACKNOWLEDGMENTS To my advisor s Dr. K. Ramesh Reddy and Dr. Ed Hanlon I would like to express my thanks for their patience and encouragement as my mentor s during my doctoral degree. To the members of my committee, Drs. Suhas P. Wani, Jim Jones, and Maria Silveira I would like to give my sincere thanks for their contributi on to the research and writing of this work. I would especially like to thank Dr. Suhas P. Wani for his mentorship and support in many aspects of this research conducted at the International Cro ps Research Institute (ICRISAT) in Patancheru, India. I thank Dr. Sahrawat, Dr. Goovaerts, and my many friends and family members for their assistance and continued support. This research was also made possible in part through a Graduate Research Assistantship provided by UF a Research Scholarship provided by ICRIS AT and the Quantitative Spatial Ecology, Evolution, and Environment (QSE 3 ) IGERT, NSF funded program.

PAGE 5

5 TABLE OF CONTENTS page ACKNOWLEDGMENTS ................................ ................................ ................................ .. 4 LIST OF TABLES ................................ ................................ ................................ ............ 8 LIST OF FIGURES ................................ ................................ ................................ .......... 9 LIST OF ABBREVIATIONS ................................ ................................ ........................... 12 ABSTRACT ................................ ................................ ................................ ................... 14 CHAPTER 1 INT RODUCTION ................................ ................................ ................................ .... 16 Management Practices Affecting C Storage ................................ ........................... 17 Dissertation Overview ................................ ................................ ............................. 21 Dissertation Objectives ................................ ................................ ........................... 21 Dissertation Organization ................................ ................................ ........................ 22 2 SOIL ORGANIC CARBON ACCUMULATION AND VERTICAL DISTRIBUTION IN CULTIVATED AND UNCULTIVATED VERTISOLS ................................ ........... 24 Background ................................ ................................ ................................ ............. 24 Materials and Methods ................................ ................................ ............................ 27 Site Description: The International Crops Research Institute for the Semi Arid Tropics (ICRISAT) ................................ ................................ .................. 27 History of t he Land Use ................................ ................................ .................... 28 Experimental Design ................................ ................................ ........................ 28 Cropping System A ................................ ................................ .......................... 29 Cropping System B ................................ ................................ .......................... 29 Inorganic Fertilizer and Organic Manure Application ................................ ........ 29 Tillage ................................ ................................ ................................ ............... 29 Soil Sampling: ICRISAT: Cultivated and Uncult ivated Soils ............................. 30 Soil Analysis ................................ ................................ ................................ ..... 30 Calculation ................................ ................................ ................................ ........ 31 Statistical Analysis ................................ ................................ ............................ 32 Results and Discussion ................................ ................................ ........................... 32 Soil Properties ................................ ................................ ................................ .. 32 Vertical Distribution of Total Soil Carbon in Cultivated and Uncultivated Soils ................................ ................................ ................................ .............. 32 Stable Carbon and Nitrogen Isotope Depth Profiles in Cultivated and Uncultivated Soils ................................ ................................ .......................... 35 Soil Microbial Biomass C and N in Cultivated and Uncultivated Soils .............. 38 Summary ................................ ................................ ................................ ................ 38

PAGE 6

6 3 MANAGEMENT STRATEGIES FOR VERTISOLS: SOIL ORGANIC CARBON AND AGGREGATE SIZE CLASSES IN CEREAL/LEGUMINOU S CROPPING SYSTEMS ................................ ................................ ................................ ............... 49 Background ................................ ................................ ................................ ............. 49 Materials and Methods ................................ ................................ ............................ 50 Site Description ................................ ................................ ................................ 50 Historical Land Use ................................ ................................ .......................... 51 Experimental Design ................................ ................................ ........................ 51 Soil Sampling and Analysis ................................ ................................ .............. 52 Modified Wet Sieving Procedure ................................ ................................ ...... 53 Microbial and Total Soi l Organic Carbon Fractions and Permanganate Oxidation of C (POXC) ................................ ................................ .................. 53 Calculations ................................ ................................ ................................ ...... 53 Statistical Analysis ................................ ................................ ............................ 53 Results and Discussion ................................ ................................ ........................... 54 Selected Soil Properties ................................ ................................ ................... 54 Effects of Management Practices on SOC Storage ................................ .......... 54 Effects of Management Practices on SOC Storage in C Associated with Aggregate Size Classes ................................ ................................ ................ 54 Effects of Management Practices on Soil Microbial Biomass Carbon .............. 56 Comparison between Permanganate Oxidation of C (POXC) and C Associated with Aggregate Size Classes ................................ ...................... 56 Summary and Discu ssion ................................ ................................ ....................... 56 4 UTILIZING GEOSTATISTICS TO INCREASE THE EFFICACY OF SAMPLING PROTOCOLS FOR MEASURING SOIL ORGANIC CARBON ............................... 64 Background ................................ ................................ ................................ ............. 64 Materials and Methods ................................ ................................ ............................ 66 Site Description ................................ ................................ ................................ 66 Soil Sampling and Soil Analysis ................................ ................................ ....... 67 Geostatistical Analysis ................................ ................................ ...................... 68 Results and Discussion ................................ ................................ ........................... 70 Exploratory Data Analysis ................................ ................................ ................ 70 Variography ................................ ................................ ................................ ...... 70 Sampling Design ................................ ................................ .............................. 71 Benefit of Secondary Information ................................ ................................ ..... 72 Summary ................................ ................................ ................................ ................ 73 5 QUANTIFICATION OF UNCERTAINTY IN MODELABLE AND MEASUREABLE SOIL ORGANIC CARBON POOLS ................................ ................................ ...... 100 Background ................................ ................................ ................................ ........... 100 Description of Measured Soil Organic Matter Fractions ................................ 100 Physical Characteristics ................................ ................................ ................. 100 Biological Characteristics ................................ ................................ ............... 102

PAGE 7

7 Chemical Characteri stics ................................ ................................ ................ 103 Description of Modeled Soil Organic Matter Pools ................................ ......... 103 Biological Characteristics ................................ ................................ ............... 104 Chemical ................................ ................................ ................................ ........ 104 Physical ................................ ................................ ................................ .......... 105 Descrip tion of Different Model Initialization Procedures ................................ 105 Gijsman et al. (2002); Porter et al. (2009) Method ................................ ........ 105 B asso et al. (2011) ................................ ................................ ......................... 105 Linear Regression Approach Developed by Adiku (published in Porter et al., 2009) ................................ ................................ ................................ ........... 106 Materials and Methods ................................ ................................ .......................... 106 Assumptions Used for the Comparison of Measured and Modeled Pools ...... 106 Descriptions of Initialization Approaches: Investigatio ns using my data ......... 107 Description of Experimental Unit ................................ ................................ .... 107 Historical Land Use ................................ ................................ ........................ 108 Land Use: Data as Considered within the Model ................................ ............ 108 Measured SOC Fractions: Soil sampling and analysis ................................ ... 109 Modified Wet Sieving Procedure ................................ ................................ .... 109 Microbial and Total Soil Organic Carbon Fractio n ................................ .......... 110 Statistical Comparison of Measured and Modeled Values ............................. 110 Results ................................ ................................ ................................ .................. 110 Discussion ................................ ................................ ................................ ............ 112 6 SYNTHESIS AND CONCLUSIONS ................................ ................................ ...... 124 Objec tive 1 ................................ ................................ ................................ ..... 125 Objective 2 ................................ ................................ ................................ ..... 126 Objective 3 ................................ ................................ ................................ ..... 127 Objective 4 ................................ ................................ ................................ ..... 127 Synthesis and Future Research ................................ ................................ ..... 128 LIST OF REFERENCES ................................ ................................ ............................. 131 BIOGRAPHICAL SKETCH ................................ ................................ .......................... 140

PAGE 8

8 LIST OF TABLES Table page 2 1 Site 1 Cropping systems, fertilizer and manure treatments, ICRISAT, Patancheru, India, 1997 2011. ................................ ................................ ........... 40 2 2 Bulk density values used for calculations for soil carbon stocks in BW1 and the black reserve grassland (GL), 2009, ICRISAT, Patancheru. ........................ 41 2 3 Nitrogen (N), carbon (C) concentrations (mg g 1 soil) and SOC stocks (g m 2 ) in January 2011 for 0 30, 30 60, and 60 90 cm for cultivated, cropping system A and cropping system B ................................ ................................ ....... 42 2 4 13 C (2011), for 4 soil depths (0 15, 15 30, 30 60, 60 90 cm) in cropping systems A and B ................................ ................................ .................. 43 2 5 15 N (2011), for 4 soil depths (0 15, 15 30, 30 60, 60 90 cm) in cropping systems A and B ................................ ................................ .................. 44 3 1 Average difference in mg SOC 100g 1 soil for June 2009 and November 2010, 0 30 cm soil depth, under conventional tillage and minimum tillage ......... 58 4 1 Kriging Variance (KV), Sampling density, Stnd error, and 95%CI. ..................... 74 4 2 SOC and TKN. ................................ ................................ ................................ .... 75 4 3 SOC and SM. ................................ ................................ ................................ ..... 76 4 4 SOC and Silt (%). ................................ ................................ ............................... 77 5 1 Measured data for simulations, 2009 10, ICRISAT, India ................................ 115 5 2 Output data for simulations for Basso et al. (2011) initialization procedure, 2009 10, ICRISAT, India ................................ ................................ .................. 116 5 3 Output data for simulations for Gijsman Porter (Gijsman et al., 2002; Porter et al., 2009) initialization procedure, 2009 10, ICRISAT, India ......................... 117 5 4 Output data for simulations for Adiku (in Porter et al., 2009) initialization procedure, 2009 10, ICRISAT, India ................................ ................................ 118

PAGE 9

9 LIST OF FIGURES Figure page 1 1 Conceptual diagram of carbon cycle for semi arid tropical (SAT) soils with indication of which chapter(s) of the dissertation the topic is addressed ............ 23 2 1 Average SOC stocks (g m 2 ) at 0 30 soil depth, for two cultivated Cropping system A (Crop A) and Cropping system B (Crop B) ................................ .......... 45 2 2 Soil stable carbon isotope depth profiles for two cultivated Croppin g system A (Crop A) and Cropping system B (Crop B) ................................ ...................... 46 2 3 Soil stable nitrogen isotope depth profiles for two cultivated Cropping system A (Crop A) and Cropping system B (Crop B) ................................ ...................... 47 3 1 Soil C fractionation methods. ................................ ................................ .............. 59 3 2 Scatterplot of SOC (mg/kg soil) associated with SOC size fraction (2000 212m) and POXC (mg/kg soil) ................................ ................................ ......... 60 3 3 Scatterplot of SOC (mg/kg soil) associated with SOC size fractions (212 53m) and POXC (mg/kg soil) ................................ ................................ ........... 61 3 4 Sca tterplot of SOC (mg/kg soil) associated with SOC size fractions (<53m) and POXC (mg/kg soil) ................................ ................................ ...................... 62 3 5 Soil respiration (y axis) and soil microbial biomass carbon (SMBC) (x axis) ...... 63 4 1 (a) Location map of soil samples collected within the Kothapally microwatershed on February 2010 ................................ ................................ ..... 78 4 2 Histogram of Soil Organic C arbon (SOC) %. Cumulative distribution is indicated by the blue line. ................................ ................................ ................... 79 4 3 Histogram of TKN and summary statistics. Cumulative distribution is indicated by the blue line. ................................ ................................ ................... 80 4 4 Scatterplot of SOC and Total Kjeldahl Nitrogen (TKN), with summary statistics. ................................ ................................ ................................ ............. 81 4 5 Scatterplot of SOC and Soil Moisture (SM) normal scores, with summary statistics. ................................ ................................ ................................ ............. 82 4 6 Scatterplot of SOC and Silt% normal scores, with summary statistics. .............. 83 4 7 Map of lo cal correlation coefficients estimated by geographically weighted regression for three pairs of variables. ................................ ............................... 84

PAGE 10

10 4 8 Soil Moisture (SM) (%) sampled values, Kothapally watershed, Andhra Pradesh, India, February 2010. ................................ ................................ .......... 85 4 9 Silt content (%) sampled values, Kothapally watershed, Andhra Pradesh, India, February 2010. ................................ ................................ ......................... 86 4 10 Experimental variogram of soil organic carbon content (%) with the exponential model fitted ................................ ................................ ...................... 87 4 11 Experimental variogram of carbon content (%) regression residuals when using TKN (mg/k g) as secondary variable. ................................ ......................... 88 4 12 Experimental variogram of soil organic carbon content (%) regression residuals when using soil moisture (%) normal scores as secondary variable.. 89 4 13 Experimental variogram of soil organic carbon content (%) regression residuals when using silt (%) normal scores as secondary variable. .................. 90 4 14 Map of ordinary kriging results: a) estimated values, and b) kriging variance (KV) for SOC, Kothapally watershed, Andhra Pradesh, India, February 2010. ... 91 4 15 Increase in the prediction variance of soil organic carbon (SOC) as the sampling grid increases, Kothapally watershed, Andhra Pradesh, India, February 2010. ................................ ................................ ................................ ... 92 4 16 Power for the detection of a given differ ence in soil organic carbon (SOC) for sampling grids with a spacing ranging from 50 to 500 meters ............................ 93 4 17 Increase in the prediction variance of soil organic carbon (SOC) with secondary varia ble total kjedahl nitrogen (TKN) ................................ ................. 94 4 18 Power for the detection of a given difference in Soil organic carbon (SOC) using secondary variable TKN. ................................ ................................ ........... 95 4 19 Increase in the prediction variance of soil organic carbon (SOC) with secondary variable soil moisture. ................................ ................................ ....... 96 4 20 Soil organic carbon (SOC) with secondary varia ble s oil moisture ...................... 97 4 21 Increase in the prediction variance of soil organic carbon (SOC) with secondary variable silt content ................................ ................................ ........... 98 4 22 Soil organic carbon (SOC) for sampling grids with secondary variable silt content ................................ ................................ ................................ ................ 99 5 1 Basso et al. (2011) iterative initialization approach, DSSAT, 2012. .................. 119 5 2 Comparison between Basso et al. (2011) and observed values (n=6). ............ 120

PAGE 11

11 5 3 Comparison between Gijsman Porter approach and observed values (n=6). 121 5 4 Comparison between Adiku (published in Porter et al., 2009) and observed values (n=6). ................................ ................................ ................................ ..... 122 5 5 Comparison between each tr eatment using each approach (n=12) ................. 123

PAGE 12

12 LIST OF ABBREVIATION S AR All crop residues BBF Broad bed furrow BD Bulk density C Carbon CaCO 3 Calcium carbonate CO 2 Carbon dioxide CT Conventional tillage DAP Di ammonium phos phate DSSAT Decision Support System for Agrotechnology Transfer FOM Fresh organic matter FYM Farm yard manure GWR Geographically weighted regression HCl Hydrochloric acid ICRISAT International Crops Research Institute for the Semi Arid Tropics KP Koth apally KV Kriging variance MBC Microbial biomass carbon MT Minimum tillage M+CP M aize and chick pea sequential M/PP Maize and pigeon pea intercrop N Nitrogen NR No crop residues OK Ordinary kriging

PAGE 13

13 POM Particle organic matter POXC Permanganate oxidation carbon RK Regression kriging RK Residual kriging SAT Semi arid tropical SOC Soil organic carbon SOM Soil organic matter SR Solar radiation TKN Total Kjeldahl Nitrogen U Urea VPDB Vienna Peedee Belemnite

PAGE 14

14 Abstract of Dissertation Presented to the Graduate Schoo l of the University of Florida in Partial Fulfillment of the Requirements for the Degree of Doctor of Philosophy SOIL ORGANIC CARBON STU DIES IN SEMI ARID TROPICAL INDIA: SOURCES AND STABILIZATION By Kiara S. Winans August 2012 Chair: K Ramesh Reddy Co Chair: Ed Hanlon Major: Soil and Water Science As organic material decays, it stabilizes in soil and then plays a critical role in holding water and nutrients available for plant growth and in supporting soil structure. Management of soil organic matter (SOM) is essential in the semi arid tropics (SATs) characterized by savannah vegetation and soils with low SOM and low nutrient reserves. Properly managing SOM may increase soil organic carbon (SOC) sequestration, nutrient availability, soil moisture ret ention, and increased crop productivity. This dissertation research (1) investigated the effects of diverse sorghum (Sorghum biocolor L. Moench) based cropping systems and of conservation agriculture practices on SOC accumulation and its vertical distribu tion (Study 1); (2) assessed the short term (16 months) soil C, and C fraction responses to conservation agricultural management practices in semi arid tropical Vertisols (Study 2); (3) applied and evaluated selected aspects of geostatistics (Study 3); and (4) investigated and quantified the uncertainty presented by both measured and modeled SOC pools (Study 4).

PAGE 15

15 In general, the key findings of the studies showed that diversified sorghum based cropping system with minimum tillage and farm yard manure (FYM) treatment contributed to increased C storage within the soil profile. The addition of FYM had the most positive effect on C storage as a result of several factors as indicated by stable C and N isotopes and C values. Within the 16 month study, a slow incr ease of soil C was evident in increased C associated with the 2000 212 m size class. Utilizing geostatistics, information about the pattern of spatial autocorrelation can be used to determine the sampling density required for achieving a given accuracy of the estimation of the mean concentration of SOC. The effect of the sampling density was quantified using power functions computed for different detection limits. Finally, the investigation of measured and modeled SOC pools showed that the measured SOC po ols did not have the properties in the tested soils and land use combinations assumed by any of the models. These findings demonstrate the importance of biogeochemical C cycling at various depths within the soil profile in semi arid tropical Vertisols us ing conservation agricultural management practices; and management options that enhance soil C accumulation in nutrient limited soils. Further, this research increases our understanding of applications of geostatistics to soil properties for purposes of i ncreased efficacy of sampling protocols for C in semi arid tropical Vertisols. The research also showed that there were real differences regarding assumptions in the models compared to the measurement methods in SOC pools. No decisions could be made using the models, but the measurements confirmed likely SOM, conservation agricultural management strategies for semi arid tropical Vertisols.

PAGE 16

16 CHAPTER 1 INTRODUCTION Soil organic carbon (C) is one of the key constituents of soil organic matter (SOM), and thus is used as an indicator of SOM and related soil functions. As organic material decays, it stabilizes in soil and then plays a critical role in holding nutrients and water available for plant growth and in supporting soil structure. In semi arid tropical (SAT) soils, SOM is rapidly decomposed due to high temperatures and SAT soils are typically characterized by low soil fertility (El Swaify et al., 1985). Management of crop residues has been identified as a means of increasing soil fertility in SATs (Vinee la et a l., 2008). The rate of C accumulation and loss are critical determinants of how much C is stored in the soil Ultimately, the amount of C stored in the soil is dependent on th e difference between the amount of CO 2 vegetation fixes during photosynt hesis and that which vegetation releases during respiration, or the net primary productivity (Reddy and DeLaune, 2008). Plant leaves and roots that are incorporated into the soil continue to breakdown and decay into SOM. The SOM as indicated by C, stored in soils depends highly on the physical, chemical, and biological properties of the soil, all of which are influenced by climate. Climatic differences, for example between an arid and a cold climate have a major impact on soil C storage. Lal (2004) exemp lifies the extremes of soil C storage due to climatic differences, reporting 30 t/ha soil organic C stored in arid regions compared to 800 t/ha in cold regions (per 1 m soil depth). Overall, the importance of C as an indicator of soil fertility and the pr oblems of low nutrient and low water availability in SAT ecosystems have provided the need for research regarding the effects of management practices on organic matter. The rate of

PAGE 17

17 C accumulation and loss are critical determinants of how SAT soils functio n. To reiterate, C storage is essential in SAT soils, especially those that are used to maintain food security. On a broader scale, although SAT soils are low in organic C, if appropriate production systems and management practices are identified, then t here is potential for tropical soils to store more C and thus to impact large scale processes such as global warming (Lal, 2004). Management Practices Affecting C Storage The effects of management (e.g., tillage, residue, and fertilizer) on C loss, accum ulation, and storage show the potential for increased C storage due to improved management practices (e.g., Srinivasarao et al., 2009; Vineela et al., 2008; Porter et al., 2009). However, a review of literature shows that there are gaps in our understandi ng of the effect of diverse cropping systems and conservation management practices on C storage and stabilization, in the surface soils and at various depths within the soil profile. Also, there is a lack of consensus on methods to measure C, for purposes of short and long term studies and for modeling efforts. Overall, studies of soil C demonstrate that crop type, residue management, tillage practice, and soil type contribute the C accumulation in the soil. However, increased C accumulation in the soil does not neces sitate increased soil C storage. S tudies on soil cultivation (e.g., tillage) and C accumulation suggest that increased cultivation intensity leads to a loss of C rich materials (Six et al., 2004). Studies have shown that t illage practice pr edominantly effects the placement of the crop residue in the soil profile (Hauert and Liniger, 2008). Once incorporated into the soil profile, soil environment factors that contribute to crop residue conversion to SOM include soil moisture, soil temperatu re, and soil microorganisms. Researchers also showed that conservation

PAGE 18

18 tillage and no tillage practices tend to allow for increases in SOM accumulation, as compared to conventional tillage practices(Wander and Bidart, 2000; Franzluebbers 2004 ). In gener physical condition, to prepare seedbeds, to control for weeds, or to incorporate fertilizers and plant residues. The type of tillage is dependent on the management objective(s). For example, conservation tillage is employed to incorporate and conserve crop residues (i.e., plant material including leaves, stalks, and roots left after a crop is harvested) to improve soil quality (e.g., soil structure) and/or to avert soil erosion ( FAO 20 10 ). Overall, information on the effects of conservation agricultural practices on SOM and the many interactions and feedbacks between microbes, roots, soil type (e.g., soil structure) and environmental factors is still limited. I n the SATs, the soil t ype plays a key role in C storage. For example, Vertisols (highly clayey soils) have been shown to have the potential to accumulate high levels of C, in particular in legume based management systems, e.g., pigeon pea (Manna et al., 2008). Several studies define the need to understand C storage in Vertisols in relation to diversified cropping systems and conservation agriculture production given the critical role SOM plays as a source for plant available nutrients and in the retention of soil water (Adiku et al, 2008). Further, s tudies (e.g., Lal, 2004) highlight the potential of semi arid soils with respect to global C cycling. T he lack of consensus on methods to measure C poses additional research challenges for purposes of short and long term studie s and for modeling efforts. M ethods used to measure C are related but not synonymous to conceptual C pools. Conceptual C pools are fairly dynamic; each is composed of one or more biological,

PAGE 19

19 chemical, and or physical C constituents. Figure 1 1 shows a si mplified schematic of soil C pools in relation to turnover rate, wherein three conceptual C pools are defined. The soil organic matter rapidly breaking down or being decomposed by microorganisms within days to years is considered the active pool (soil org anic matter pool 1, SOM1) T he organic matter that breaks down slowly, within years to a decade, or is partially decomposed floral and faunal is considered the intermediate pool (soil organic matter pool 2, SOM2) T he stabilized organic matter or humic c arbon breaks down in decades to hundreds of years and is considered the passive pool (soil organic matter pool 3, SOM3). During the past decade, quantitative methods have been developed to mea sure C fractions in relation to estimation of the short term im pacts of management practices on C accumulation in soils (Wang and Hsieh, 2002; Paul et al., 2008). Many of these studies were based on the investigation of SOM aggregate conceptual models (Emerson, 1959), microaggregate stabilization theory (Edwards and B Oades (1982), formation of aggregates occurs when stable microaggregates (<212m) are bound together by macroaggregates (>212 m). In later studies, macro and micro aggreg ate formation was determined to be directly linked to SOM, except in oxide rich soils (e.g., Oxisols) (Oades and Waters, 1991). Cambardella and Elliott (1992) devised a size based fractionation method that can be used to measure particle organic matter (P OM) in relation to SOM influenced by management practices, which Six et al (2004) advanced. In addition, quantification of soil C pools with temporal variability or spatial variability can be accounted for based on physical characteristics of the landscap e such

PAGE 20

20 as soil texture (e.g., clay content). Several studies (e.g., Wu et al., 2009) demonstrated how geostatistics can increase the efficacy of sampling protocols used to measure C, thereby providing incentives for the management practices that improve s oil quality (e.g., soil C storage). Accuracy of measured C pools is essential if measured values are used for modeling purposes. From a modeling perspective, the simplification of the C pools allows for extrapolation from current knowledge on C turnover to test environmental (e.g., temperature) and or management (e.g., tillage) effects on C stability. The conceptual C pools do not correspond directly with experimentally measured C fractions, however as noted earlier in text U ncertainty or variability is presented by both measured and modeled C pools. Importantly, improvement of methods to measure and to model C pools may aid decision making processes with respect to SOM management. For example, a modeling approach can be used to generate information about the potential effect of SOM management on the passive C pool and or long term soil functions, e.g., to support plant growth or store C. The model used in this dissertation research was the Decision Support System for Agrotechnology Transfer (DSSAT) CENTURY model developed by Gijsman et al. (2002), incorporating the CENT URY model (Parton et al., 1988a; Parton and Rasmussen, 1994; Metherell et al., 1993) with the DSSAT model (Hoogenboom et al., 2004). The DSSAT CENTURY module provides a flexible frame work for SOM pool initialization, using one of three approaches ( 1) Gijsman et al. (2002), ( 2) Basso et al. (2011), and ( 3) Adiku in Porter et al. (2009). In the model, the conceptual C pools are defined based on decomposition and or physically defined C fractions. Overall,

PAGE 21

21 measureable C fractions can be compared to modelable, conceptual C pools. However, the uncertainty of both measured and modeled C pools requires extensive investigation. Dissertation Overview The overall goal of this dissertation i s to improve the understanding of how diversified cropping systems and conservation agriculture a ffect SOM accumulation in semi arid tropical soils. Chapters 2 through 5 address more specific objectives: Figure 1 1 also provides a conceptual diagram of th e C cycle and indicates which chapter(s) of the dissertation the topic is addressed. Dissertation Objectives Objective 1: Determine the effects of diverse sorghum ( Sorghum biocolor L. Moench) based cropping systems and of conservation agriculture practice s on soil organic carbon (C) accumulation and its vertical distribution in the Indian semi arid tropics. Hypothesis: Comp arison between cultivated soils that received either farm yard manure or inorganic fertilizer in a factorial design was expected to r espond differently than SOC accumulation in uncultivated soils (1) at specific depths and (2) as a function of depth. Objective 2: Assess the short term (16 months) soil C, and C fraction responses to conservation agricultural management practices in semi arid tropical Vertisols: tillage, residue, and crops. Hypothesis: The short term changes in SOC storage due to selected management practices should not be indicated in the whole soil, but should be indicated in C associated with size fractions Objective 3 : Use information about the pattern of spatial autocorrelation to determine the sampling density required for achieving a given accuracy in the estimation of the

PAGE 22

22 mean concent ration of soil organic carbon; and compare and evaluate the benefits of different covariates in regression kriging (RK). Hypothes i s: Information about the pattern of spatial autocorrelation will allow for proper selection of SOC sampling density with control of the confidence level and avoid unnecessary sampling Objective 4: Compare 4 methods used to estimate the SOC pools in the CENTURY based Decision Support System for Agrotechnology Transfer (DSSAT) model (Tsuji et al., 1994) ; and q uantify the uncertainty in observed vs. predicted modeled and measured SOC pool values Hypothesis: Pre dicted values will have a low correlation with observed values due to current model components (explanatory variables, parameters, equations). Dissertation O rganization CHAPTER 1: Introduction to management effects on C storage in semi arid tropical soils; measured and modeled C pools; and an overview of the objectives and format of the dissertation. CHAPTER 2: A study investigating C accumulation and vertical distribution in diversified sorghum based cropping systems. CHAPTER 3: A study investigating the i mpacts of conservation agricultural practices in C association with physical soil size fractions. CHAPTER 4: A study showing the use of autocorrelation of C and use of secondary soil property data to predict C values with a fixed accuracy and detection lim it. CHAPTER 5: A study investigating and quantifying measured and modeled C pools.

PAGE 23

23 CHAPTER 6: A synthesis of conclusions drawn from all the dissertation studies and a discussion on the implications of SOM management on C in SAT soils. Figure 1 1 C onceptu al diagram of c arbon cycle for semi arid tropical (SAT) soils with indication of which chapter(s) of the dissertation the topic is addressed

PAGE 24

24 CHAPTER 2 SOIL ORGANIC CARBON ACCUMULATION AND VERTICAL DISTRIBUTION IN CULTIVATED AND UNCULTIVATED VERTISOLS Bac kground Soil organic matter (SOM) management is esse ntial in the semi arid tropical areas characterized by savannah vegetation and soils with low SOM and low nutrient reserves ( El Swaify et al., 1985) Properly managing SOM may increase C sequestration, n utrient availability, soil moisture retention, and crop productivity (Lal, 2006) Typically, fresh organic matter is added to the soil as plant residues, of which a large portion may be lost because of high decomposition rates given the relatively high te mperatures in tropical areas ( Wani et al., 2004) A high rate of decomposition combined with bare soil and cultivation further increase the loss of fresh organic matter and soil organic carbon content (Chandran et al., 2009). In agricultural systems, cons ervation tillage or no tillage practices tend to increase SOM and thus C accumulates compared to under conventional tillage practices, while carbon to nitrogen ratio ( C:N ratio)) influence the composition of SOM (Franzluebbers et al., 2008). A c onservation tillage practice developed by scientists at the ICRISAT research farm (Hyderabad, India) is defined as the broad bed furrow (BBF) landform system, which consists of a broad bed (1.05 m) and a 0.45 meter furrow. The BBF system reduces runoff and soil loss, increases retention of nutrients, and helps to facilitate precise fertilizer and seed placement, and sowing of the crops on the beds (Pathak et al., 2005). Soil lo ss in the BBF system averaged 1.5 t ha 1 yr 1 where soil loss in conventional tillage systems averaged 6.0 t ha 1 yr 1 (Wani et al., 2003). Moreover,

PAGE 25

25 importantly, the tillage practice predominantly affects the potential loss and placement of crop residue i n the soil profile (Hauert and Liniger, 2008). R esearchers have also demonstrated the potential for significant increases in C and crop productivity in sorghum ( Sorghum biocolor L. Moench) based cropping systems with a diverse set of crops and inputs of inorganic fertilizers and MT in the semi arid tropical agroeco systems (Wani et al., 2007; Srinivasarao et al., 2009). Fu r ther, studies indicated that legume based systems support higher C sequestration than cereal crops, and grasslands sequestered more C than annual crop systems (e.g., Bhattacharyya et al., 2007). In a study of 28 benchmark sites covering arid, semi arid, and moist humid tropical locations in India, Ramesh et al. (2007) showed that the inclusion of legumes in rotation or as an intercrop, such as cotton ( Gossypium L.) plus sorghum and pigeon pea intercrop system, positively influenced the soil quality as indicated by C content. A number of hypotheses and various methods have been developed to help explain the C accumulation in the surface soil (0 30cm) and distribution with depth in the soil profile (e.g., Wynn et al., 2006). Methods to assess C accumulation and distribution within the soil profile include but are not l imited to stable carbon ( 13 C/ 12 C) and nitrogen ( 15 N/ 14 N) isotope ratios (Wynn et al., 2006), radiocarbon dating (Trumbore et al., 2000), and carbon fractionation (e.g., permanganate oxidation carbon (POXC)) (Culman et al., 2012). As is often noted, these methods provide operationally defined C fractions (e.g., Silveira et al., 2009). Often researchers identify functional fractions of the soil C within the representative soil depth (e.g., Six et al., 2004). It is often implied that the subfractions of SOM undergo similar physical, chemical, and biological reactions within

PAGE 26

26 the representative soil depth (Yoo et al., 2011). Few studies have investigated processes related to C distribution with depth as a continuous process within the soil profile (Wynn et al ., 2006; Yoo et al., 2011). Wynn et al. (2006) provided a review of 13 C C data, to which they applied a continuous depth function. Wynn et al (2006) explained two hypotheses related to C variation with depth as follows: The C dis tribution within the soil pro file varies as an effect of ( 1) t he input of SOM with differing isotopic composition (e.g., root derived versus leaf derived SOM), and or ( 2) t he preferential decomposition of SOM components (e.g., cellulose ) that results in the stabilization of more or le ss labile components of (e.g., Trumbore et al., 2000) results supported hypothesis (2) (Wynn et al. 2006), but arrived at these conclusions using different methods. In general, there is consensus amongst the studies that the decomposition rates of C tends to decrease with depth, and recalcitrant carbon increases with depth (e.g., Trumbore et al., 2000). In a global study, Jobbagy and Jackson (2000) reported that the C consistently varied spatially and with depth, and was positively correlated with clay content and moisture, and negatively correlated with temperature. Studies have also described the effects of plant and root distribution (e.g., Daly, 2000) and soil mixing (Yoo et al., 201 1) on C accumulation and vertical distribution, all of which were noted to be driven in some way by climatic conditions (e.g., precipitation, temperature, solar radiation). Further, studies have shown that the C accumulation in the surface soil (0 30 cm s oil depth) may not indicate C storage in deeper layers of the soil (Baker et al., 2007) yet the primary effects of soil

PAGE 27

27 disturbance (e.g., tillage) on the whole soil and C fractions are primarily evident within the top 40 cm of soil (Yoo et al., 2011; Job bagy and Jackson, 2000). Because of the potential benefits of increased soil carbon such as improved soil conditions for crop growth, there is a critical need to study soil carbon accumulation and vertical distribution research in the naturally nutrient li mited soils of the semi arid tropics where this data is still limited In addition, comparison between different methods to assess C fractions to quantify C accumulation and distribution within the soil profile are needed We address ed these needs by ex amining the relationships between C and selected 13 C SOC ) using sorghum based cropping systems and conservation agriculture practices on C accumulation and vertical distribution in semi arid tropical soil s The primary objective of this study was to determine the effects of diverse sorghum (( Sorghum biocolor L. Moench) based cropping systems and of conservation agriculture practice s on soil organic carbon (C) accumulation and its vertical distribution in the Ind ian semi arid tropics. We hypothesized that comparison between C and selected C fractions would consistently indicate the effects of divers ified sorghum based cropping systems on C accumulation (1) at specific depths and (2) as a function of depth. Mater ials and Methods Site Description: The International Crops Research Institute for the Semi Arid Tropics (ICRISAT) The International Crops Research Institute for the Semi Arid Tropics (ICRISAT) established in 1972, is located in Patancheru, Andhra Pradesh, India (17.5 N, 78.5 E; 545 m elevation). ICRISAT serves as a research farm for the study of various crops on

PAGE 28

28 Alfisols (red soils) and Vertisols (black soils). The current research was conducted on Vertisols at the ICRISAT farm in Patancheru (Hyderabad India) In this study, crops were grown without irrigation and solely dependent on rainfall. Depending on the treatments, one or two crops were grown per year: the winter season crop (the post rainy season crops were grown during November to February ) and the summer season crop (the rainy season crops were grown during June to October). The climate is semi arid tropical, with a rainy season that lasts for 2 to 4.5 months from June to October (S W monsoon), post rainy season (October mid February), an d summer season (March to May) Mean monthly air temperatures range between 18 and 42C, with average precipitation during the rainy months of 50 mm where potential evapotranspiration exceeds precipitation for 3 or more months of the year. History of t he Land Use The cultivated cropping system A included sorghum intercropped with pigeon pea alternated every other year with greengram grown in rotation with sunflower. The cultivated cropping system B included greengram winter sorghum that was alternated every year with soybean/pigeon pea intercrop. However, before 2003, these sampling locations were planted in sorghum/pigeon pea intercrop every year. After 2003, the cropping system was further diversified to the current cropping systems (A and B). Exp erimental Design The experimental site for this study is referred to as the black soil watershed experimental unit on e (BW1). The main treatment was the cropping system, and the sub treatments were application of (1) organic farm yard manure (FYM) and (2) inorganic fertilizer. The study site was cultivated with minimum tillage (MT) (Table 2 1). The two cropping systems were as follows:

PAGE 29

29 Cropping S ystem A S orghum and pigeon pea ( Cajanus L) intercrop alternated every other year with mungbean ( Vigna radiat a L ) in the summer season in rotation with sunflower ( Helianthus annuus L ) in the winter season; and Cropping S ystem B M ungbean ( Vigna radiata L ) and winter sorghum intercrop alternated every year with soybean ( Glycine L) and pigeon pea ( Cajanus L) interc rop. Each cropping system was grown on a Vertisols within the ICRISAT research farm. The uncultivated grassland (GL) was located approximately 5 00m from the cultivated systems in uncultivated reserve Vertisols at the ICRISAT research farm Inorganic Fert ilizer and Organic Manure Application Each cropping system had a sub treatment of farm yard manure (FYM) and inorganic fertilizer as described in the following. The FYM was composed of cow manure, collected from cattle at the research farm and stored in s haded, enclosed areas open to air flow. FYM was applied at 10 t ha 1 (2t ha 1 C, 0.1t ha 1 N, 0.01t ha 1 P) every other year by broadcasting on the soil surface. The inorganic fertilizer was applied as 60 kg N ha 1 and 13 kg P ha 1 cereals, and 20 kg N h a 1 and 17 kg P ha 1 for legumes, as urea (U) and di ammonium phosphate (DAP), respectively, in rows with a seed and fertilizer drill. Extractable potassium (K) levels were determined for all sites were found to be high (>125 ppm) (ICRISAT, 2011), ranging from 171 to 270 mg kg 1 Therefore, no additional K was applied to the soil. Tillage The MT was defined by the BBF landform system, which consisted of a broad bed (1 .05 meter) and a 0. 4 5 meter furrow. The BBF system was employed to reduce runoff

PAGE 30

30 and so il loss, to increase retention of nutrients and to avoid waterlogging, as well as to facilitate precise fertilizer and seed placement, and sowing of the crops on the beds (Pathak et al., 2005). Soil Sampling: ICRISAT : Cultivated and Uncultivated Soils Soil samples were collected from the two sites from cultivated and uncultivated treatments considering elevation, slope, and land use. Soil samples were collected from 0 15, 1 6 30, 3 1 60, and 6 1 90 cm soil depths in January 20 11 before harvest of the post ra iny season crop. Three cores were collected randomly from each plot from each depth and mixed to create a homogenous composite sample. Soil Analysis The bulk density of the samples was determined using core samples (diameter 50 mm) of known volume (100 c ubic cm), wet weight minus oven dried weight. Soil moisture was estimated for 10 g of soil, wet weight minus oven dry soil. Samples were dried at 105 C for 24 h. The soil pH was measured with a glass electrode using soil to water ratio of 1:2 and partic le size by hydrometer method (Gee and Bauder, 1986). All soil samples were analyzed in the laboratory for physical, chemical, and biological characteristics. A coding system was used so that the soil information was not kno wn to the operator at the time of analysis. Soil samples were air dried and sieved through a 2 mm sieve. A subsample of the 2 mm sieved soil was further ground and sieved through a 53 m sieve for analysis of total soil carbon ( C and SIC) using the Thermo Finnigan MAT Delta Plus XL IRM S interfaced via a Costech ECS 4010 elemental analyzer for continuous flow measurement of the stable C isotope ratios. Soil inorganic carbon was estimated using a pressure calcimeter method (Loeppert and Suarez, 1996). Soil organic carbon was calculated as the difference between total C and

PAGE 31

31 inorganic C. Because inorganic C was present in the soil samples, the soil samples that were analyzed for stable C isotope ratios were acidified. Soil samples were weighed in tin caps and moistened with a drop of dis tilled water, and then placed in an incubator with an amount of calcium carbonate (CaCO 3 ) equal to the largest mass of soil weighted (~60 g) in the tins and 40 ml of 1 M of hydrochloric acid (HCl) until the reaction was complete. Once the reaction was co mplete, samples were dried in a desiccator. (( R sample / R standard ) standard. R sample and R standard are the fr actions of heavy to light isotopes in the sample and standard, respectively. For carbon, the international standard used is the Vienna Peedee Belemnite (VPDB). The standard for nitrogen is atmospheric nitrogen (Fry, 2006). The decomposing pool of SOM in volves kinetic fractionation whereby through microbial respiration, 12C is preferentially respired (Huntington et al., 1998). As such, the alteration of the isotopic ratio may exhibit trends in the preferential removal of the labile fraction of SOM. Tren 13 15 N were investigated to determine the effects of the natural abundance of stable C and N isotope composition and the effect of cultivation on these natural depth trends. Calculation The C and SIC stocks were calculated using Eq. (2 1) (2 1) Where BD is bulk density (g cm 3 ) and C is soil organic carbon (mg g 1 ).

PAGE 32

32 Statistical Analysis The data were statistically analyzed using the JMP statistical package (8) for analysis of variance (SAS Institute Cary, NC). The crop and treatment (FYM or DAP+U) and depth were considered as fixed effects. Significant differences were analyzed p < 0.05. Results and Discussion Soil Properties Mean soil pH was 7.8. Soil BD averaged 1.3 g cm 3 in cultivate d soils (0 30 cm soil depth) with MT and 1.4 g cm 3 in uncultivated soils (0 30 cm soil depth) at the ICRISAT research farm (Table 2 2). The bulk density values observed in the current study were comparable to results from previous studies on black soils within the same area ( Bhattacharyya et al., 2007). Vertical Distribution of Total Soil Carbon in Cultivated and Uncultivated Soils The uncultivated GL soils had significantly higher C stored in the 0 30 cm soil depth compared to that of the cropping system A an d B (DAP+U and FYM) (Table 2 3) (p <0.05). There was no significant difference in C stored within or between the cropping systems and the uncultivated GL soil in the 31 60 and 61 90 cm soil depths (Table 2 3) (p <0.05). The comparison between cultivat ed soils and uncultivated GLs in the 0 30 cm soil depth suggests that the MT with diverse sorghum based cropping systems do not sustain soil carbon levels (Fig ure 2 1). The C storage in the surface soils was not an indication of the C storage wi thin the s ubsurface soil layers with cultivation practice employed. The FYM treatment in cropping system B showed higher C in the total soil profile, 0 90 cm compared to the DAP+U treatments (in cropping systems A and B) and

PAGE 33

33 the uncultivated GL (Table 2 3). The re sults suggest that the cropping system B with FYM and MT may support C storage within the soil profile (0 90cm) comparable to the uncultivated GL system in Vertisols. The sorghum based systems were all diversified (4 species), but there were some distinct ive differences between the cropping syste ms that may have affected C retention and distribution within the soil. Studies have shown that winter sorghum can provide increased yields in integrated nutrient management systems ( Srinivasarao et al., 200 9 ) and soil moisture retention in rainfed agroecosystems (Pawar et al., 2005). Although, in the current study, the overall yields in the winter sorghum cropping system were lower compared to the post rainy sorghum system, added benefit of increased soil moistur e retention may have supported the decomposition of SOM and nutrient availability for plants and microbes. The below ground contribution of the roots and related microbial activity may be ~50% higher compared to the above ground, plant derived contributio n to the C in the soil ( Brugnoli and Farquhar, 2000 ; Walker et al., 2003). As such, in a diversified cropping system, such as in cropping system B (with mungbean+winter sorghum), there may be multiple benefits to both the soil and the crop in maintaining equilibrium or increased levels of C with time. Further, legumes were an integrated component of each cropping system; we tested only the combined effect of the cropping system. We did not quantify the legume contribution to C accumulation and distributi on within the soil in this study, just the overall cropping system effects. Legumes with high N fixing capacity in the cropping system have shown to contribute to increased C accumulation (Wani et al. 2003 ; Ramesh et al., 2007).

PAGE 34

34 Between cropping system and fertilization combinations, cropping system B (FYM) had significantly higher C compared to cropping system A (DAP+U and FYM) (Fig ure 2 1) (p <0.01). The addition of FYM with a C: N ratio of 24:1 may be readily decomposed and or incorporated into slow C turnover pools within the soil ( Francioso et al., 200 7 ). Once C is incorporated into the slow soil C turnover pools it is likely less available to biological decomposition. Studies have shown that manure is an active component of what is typically cons idered to be a recalcitra nt carbon pool within the soil ( Francioso et al., 200 7 ). Further, manure may also be less biodegragded at deeper depths within the soil profile due to decreased microbial activity below plant rooting depths. The effect of fertil izers to enhance crop growth and subsequently support increased net primary productivity may also result in increased C (Alvarez, 2005). However, in the current study, the overall effect of enhanced crop production was not observed. The reason for this r esult was most likely because the crop residues were harvested and removed from the system and thus only the remaining root system and leaf fall throughout the growing season directly contributed to C accumulation in the soil. Overall, the average accumu lation of C within the entire measured soil profile was in the order of cropping system B (FYM)>uncultivated (GL)>cropping system B (DAP+U)>cropping system A (DAP+U)> cropping system A (FYM), and C storage was highest in the surface soils (0 30 cm) in the uncultivated GL soils. The higher carbon levels in the cropping system B, compared to cropping system A may be attributed to the combination of crops, in particular the benefits of winter sorghum and soybean pigeon pea intercropping system. For example, S ing h et al (200 7 ) found that soybean

PAGE 35

35 pigeon pea intercrop ping system stored more carbon compared to other intercropping systems. No significant difference in depth effect was observed in the amount of SIC stored (g m 2 ) between the cultivated cropping systems and the uncultivated system (data not shown). Deposition and precipitation of CaCO 3 from the surface soils (0 30 cm) to lower depths within the soil profile may explain the statistically equivalent SIC values in the top soil layers of the uncultiv ated and the cultivated systems. Lower SIC storage in the root zone (15 30 cm) of the soil may be the product of the reaction of acidic root exudates and carbonic acid formed by carbon dioxide from root respiration that react to solubilize the CaCO 3 (Sahr awat et al., 2005). Stable Carbon and Nitrogen Isotope Depth Profiles in Cultivated and Uncultivated Soils 13 C SOC values in the cropping system B (DAP+U and FYM) generally increased and in cropping system A (DAP+U and FYM) decreased by 0.5 0 / 00 within the first 30 60 cm ( Table 2 4, Fig ure 2 2 13 C SOC values in the 30 60 cm soil depth may be attributed to the surface vs. root derived SOM. Hobbie and Werner (2004) observed that the plant bi omass roots are generally mor e 13 C 13 C SOC values of different compounds vary where lignin are 13 C depleted compared to the bulk plant materia l, and sugars, amino acids, and hemicelluloses are 13 C enriched (Bowling et al., 2008). Therefore, the enriche d 13 C compounds like sucrose tend to lead to an enrichment of 13 C in the roots (Hobbie and Werner, 2004). Further, t 13 C SOC values above 60 cm soil depth in the cultivated systems with FYM and DAP+U treatments may be the ef fect of SOM

PAGE 36

36 13 C SOC values throughout the soil profile. The preferential removal of 13 C enriched cellulose during microbial degradation of plant residues may account for a 1 decrease in the 13 C value, leaving a relatively 13 C depleted lignin substrate (Benner et al ., 1987; Santruckova et al., 2000). Thus, it is implied that when decomposition of organic matter increases, the more labile SOM may be removed, thereby red ucing t he 13 C SOC values. As such, the increase in 13 C depleted materials in the surface horizon compared to the relatively non decomposed, non enriched material in the subsurface 13 C SOC values throughout the soil profile. Additionally, studies showed that due to the chemical stability of 13 C depleted lipids and lignin, they can accumulate during SOM decomposition and their concentration in some cases may increase with depth and SOM age (Wynn and Bird, 200 7). I 13 C SOC values in the surface (0 30 cm) soil depth were slightly enriched ( 15 N values were significantly (p<0.05) higher in the cultivated cropping sys tem A with inorganic fertilizer inputs, at 31 60 and 61 90 cm soil depths compared to the uncultivated system, at 0 30 cm soil depth (Table 2 5). In general, 15 N values were higher in cultivated cropping systems (cropping system A and B) compared to unc ultivated soils (Table 2 15 N values in cultivated systems and non 15 N values in uncultivated soil may be the effect of conditions where production was high ; therefore the N may not have been preferentially decompos ed but instead more completely utilized in the uncultivated soils (Kendell et al., 200 7 ). Other

PAGE 37

37 studies show similar conclusions -N consumed by microbes was mineralized resulting in a preferential loss of the light isotope (Dijkstra et al., 2008). The re lationships between the C and N isotopes with depth are expressed in Figures 2 3 and 2 4, respectively. In cultivated soil, the 13 C and 15 N showed weak linear relationships with depth. The uncultivated GL soils showed slightly stronger linear relationshi ps for 13 C (RMSE=18.81) and 15 N with depth (RMSE=11.99). These re sult s may indicate that the C in the cultivated soils is fairly unstable. Also, C was a poorly correlated with 13 C and 15 N values in cultivated systems (R 2 < 0.2) (Fig ure 2 4). However, in un cultivated GL systems, C provided an indication of 15 N (R 2 =0.85 p=<0.001 ) and 13 C (R 2 =0.68 p=0.001 ) values as a function of depth within the soil profile. The 13 C/ 12 C and 15 N/ 14 N isotope ratios observed in cultivation practices did not show distinctive isotopic signatures, but the comparison between the cultivated sites 15 N 13 C values in the cultivated soils, we may infer that the N supply was not limited and t he decomposition of organic matter increased, leaving depleted C material at all depths within the soil profile (0 90 cm), except in the cropping system B (FYM). In the cropping system B (FYM), we may infer that the N supply was not limited and the decomp osition of organic matter was limited, leaving enriched C material at all depths within the soil profile (0 90 cm). Other long term studies highlighted the positive effects of manure application on C storage, increased root yields, and higher humification rate constant ( Francioso et al., 2007 ). In particular the researchers measured the humic

PAGE 38

38 fraction and 13 C for C 3 C and C 4 C species and found that the mechanism involved was mainly increased preservation of C 3 C from degradation in the humic C pool. So il Microbial Biomass C and N in Cultivated and Uncultivated Soils No significant differences were observed with soil depths in SMBC and SMBN between cultivated and uncultivated (p<0.05). In most cases, SMBC and SMBN showed little to no correlation with C values with depth (R 2 < 0.2). The SMBC averaged 385 mg kg 1 + 160 mg kg 1 The SMBN averaged 40 mg kg 1 + 12 mg kg 1 The C: N ratio averaged 10 + 3. Summary Overall, the treatment effects were observed within the surface soil layers (0 30 cm) (p=<0.05). Ba sed on the comparison of cultivated and uncultivated GL soils, we conclude that a diversified sorghum based cropping system with MT and FYM treatment contributed to increased C storage within the soil profile. The diversity of crops grown in the soils may provide the improved soil conditions (e.g., increased soil moisture) necessary for increased C storage. The addition of FYM had the most positive effect on C storage as a result of several factors as indicated by stable C and N is o topes and C values. The comparison of results in the cultivated and uncultivated on stable C and N isotope depth profiles showed that in the uncultivated soils and in crop B, FYM, the lighter fraction of the C remained in the soil, indicated by higher soil C fractions (C and SMB 13 C values, and thus possibly less decomposed SOM. Increased C storage may be mainly due to increased preservation of C 3 C from degradation, suggesting that the bi annual supply of FYM can significantly increase the preservation

PAGE 39

39 of old C in the humic pool ( Francioso et al., 2007), especially in light of the practice of removing all above ground plant material after harvest. Comparison between C and 13 C and 15 N indicated the effects of diversified sorghum based cropping systems at specific depths (0 30, 31 60, and 61 90 cm). However, in cultivated soils the effects of the cropping system were not consistently ind i cated as a continuous function with depth and processes related to C accumulation and distribution within the soil profile. In u ncultivated GL soils, C and 13 C and 15 N consistently indicated C accumulation and distribution within the soil profile. The effects of cropping systems showed distinct shifts in stable C and N isotope values most probably due to root vs. shoot derived or ganic material and SOM age. Further investigation of soil and plant lignin content compared to stable isotopes may provide information needed to understand the patterns in isotope values with depth within the soil profile. Additionally, the nonlinearity a nd variability of C, 15 N, and 13 C within the upper 60 cm of the soil suggests that instability of C and disturbance due to cropping systems may limit the potential use of linear functions to assess C as a continuous function with depth within the soil prof ile. The conclusions of this study have implications for land use and research in semiarid tropical soils and for modelers who would like to apply continuous depth functions to analyze C constituents with soil depth within agroecosystems.

PAGE 40

40 Table 2 1 Si te 1 Cropping systems, fertilizer and manure treatments, ICRISAT, Patancheru, India, 1997 2011 Site Crop(Year 1) Crop (Year 2) FYM Inorganic Fertilizer cropping system A sorghum/pigeon pea mungbean+sunflower 10 t ha 1 ; broadcasting method Cereals: DAP : 12 kg ha 1 N (1000kg ha 1 C; 100kg ha 1 N; 10kg ha 1 P) DAP: 13 kg ha 1 P Urea: 48 kg ha 1 N Legumes: DAP: 16 kg ha 1 N DAP: 18 kg ha 1 P Urea: 4 kg ha 1 N basal application cropping system B greengram/ winter sorghum soybean/pigeon pea 10 t ha 1 ; broadcasting method Cereals: DAP: 12 kg ha 1 N (1000kg ha 1 C; 100kg ha 1 N; 10kg ha 1 P) DAP: 13 kg ha 1 P Urea: 48 kg ha 1 N Legumes: DAP: 16 kg ha 1 N DAP: 18 kg ha 1 P Urea: 4 kg ha 1 N basal application Uncultivated GL NA NA / represents intercrop; + rotation cropping system; and NA = not applicable. All sites were cultivated using minimum tillage, except the uncultivated GL that was subjected to no tillage treatment.

PAGE 41

41 Table 2 2 Bulk density values used for calculations for soil carbon stocks in BW1 and the black reserve grassland (GL), 2009, ICRISAT, Patancheru BD (g cm 3 ) BW1 Depth (cm) 2009/10 0 15 1.3 16 30 1.3 31 45 1.4 46 60 1.4 61 90 1.4 91 120 1.4 GL 0 30 1.4

PAGE 42

42 Table 2 3 Nitrogen (N ), carbon (C) concentrations (mg g 1 soil) and SOC stocks (g m 2 ) in January 2011 for 0 30, 30 60, and 60 90 cm for cultivated, cropping system A and cropping system B systems, with farm yard manure (FYM) and inorganic fertilizer (DAP+U) treatments and unc ultivated GL soils, ICRISAT, Patancheru, India Crop:Fert Depth (cm) mg N g soil 1 mg C g soil 1 SOC (g m 2 ) Crop A:FYM 30 0.95 b 12.6 c 2316c* 60 0.45 ab 5.05 a 1851a 90 0.30 b 4.60 a 1622a Total 5789 Crop A:DAP+U 30 1.00 b 12.1 c 2216c 60 0.40 ab 5.20 a 1900a 90 0.30 b 5.45 a 1908a Total 6024 Crop B:FYM 30 1.05 b 15.5 b 2906b 60 0.35 b 7.20 a 2693a 90 0.40 a 6.80 a 2463a Total 8062 Crop B:DAP+U 30 1.17 ab 13.8 bc 2594bc 60 0.47 a 5.73 a 2111a 90 0.40 a 5.50 a 2033a Total 6738 Uncultivated GL 30 1.43 a 18.6 a 3500a 60 0.40 ab 6.10 a 2394a 90 0.37 ab 4.87 a 1694a Total 7588 *Sums for different treatments (farm yard manure (FYM) and inorganic fertilizer (DAP+U)) for a given cropping system followed by different letters designate statistical s ignificance at the 0.05 probability level. Statistical significance is indicated per depth (0 30, 30 60, 60 90cm). Crop A represents s orghum pigeon pea intercrop alternated every other year with the greengram and sunflower sequential cropping system Crop B represents greengram winter sorghum alternated every other year with soybean pigeon pea intercrop

PAGE 43

43 Table 2 4 13 C (2011), for 4 soil depths (0 15, 15 30, 30 60, 60 90 cm) in cropping systems A and B with farm yard manure (FYM) and inorganic fer tilizer (DAP+U) treatments, and uncultivated GL, ICRISAT, Patancheru, India *Means for different cropping systems (cropping systems A and B) and treatments (FYM and DAP+U), and uncultivated GL at a given depth followed by different letters designate statistical significance at the 0.05 probability leve l (n=60). Statistical significance is indicated per depth (0 15 15 30, 30 60, 60 90cm). Depth (cm) Uncultivated GL Cropping system A Cropping system B FYM DAP+U FYM DAP+U 15 19a* 23b 23b 23b 24b 30 20a 24b 24b 23ab 23ab 60 21a 24b 22ab 22ab 23b 90 23a 23a 23a 23a 24a

PAGE 44

44 Table 2 5 15 N (2011), for 4 soil depths (0 15, 15 30, 30 60, 60 90 cm) in cropping systems A and B with farm yard manure (FYM) and inorganic fertilizer (DAP+U) treatments, and uncultivated GL, ICRISAT, Patancheru, India Depth (cm) Uncultivated GL Cropping system A Cropping system B FYM DAP+U FYM DAP+U 15 1.79b* 6.75ab 12.3a 10.9a 7.54ab 30 2.09b 9.04ab 12.6a 11.0a 7.54ab 60 3.39b 5.03b 16.0a 11.1ab 8.84ab 90 3.61b 14.2a 16. 0a 14.0a 10.2ab *Means for different cropping systems (cropping systems A and B) and treatments (FYM and DAP+U), and uncultivated GL at a given depth followed by different letters designate statistical significance at the 0.05 probability level (n=60).

PAGE 45

45 Figure 2 1 Average SOC stocks (g m 2 ) at 0 30 soil depth, for two cultivated Cropping system A (Crop A) and Cropping system B (Crop B) and 2 treatments for each (farm yard manure (FYM) and inorganic fertilizer (DAP+U)) sites and the control uncultivated grassland (GL) site. Different letters designate statistical significance at the 0.05 probability level b a bc c c

PAGE 46

46 Figure 2 2 Soil stable carbon isotop e depth profiles for two cultivated Cropping system A (Crop A) and Cropping system B (Crop B) and 2 treatments for each (farm yard manure (FYM) and inorganic fertilizer (DAP+U)) sites and the control uncultivated grassland (GL) site 1 3C

PAGE 47

47 Figure 2 3 Soil stable nitrogen isotope depth profiles for two cultivated Cropping system A (Crop A) and Cropping system B (Crop B) and 2 treatments for each (farm yard manure (FYM) and inorganic fertilizer (DAP+U)) sites and the control uncultivated grassland (GL) s ite 15N

PAGE 48

48 FIGURE 2 4 CORRELATION GRAPHS F OR 13C, 15N, AND SOC (MG G 1 ) FOR CROPPING SYSTEMS A AND B: (A) 13C CROP A, DAP+ U (R 2 =0.09) (B) 15N CROP A, DA P+U (R 2 =0.16) (C) 13C CROP A, FYM (R 2 =0.09) (D) 15N CROP A, FY M (R 2 =0.15) (E) 13C CROP B, DAP+ U (R 2 =0.01) (F) 15N CROP B, DA P+U (R 2 =0.02) (G) 13C CROP B, FY M (R 2 =0.05) (H) 15N CROP B, FYM (R 2 =0.36) (I) 13C UNCULT. GL (R 2 =0.68) (J) 15N UNCULT GL (R 2 =0.85) ( a ) R 2 = 09 ( b ) R 2 = 16 ( c ) R 2 = 0.09 ( d ) R 2 = 0.15 ( e ) R 2 = 0.01 ( f ) R 2 = 0.02 ( g ) R 2 = 0.05 ( h ) R 2 = 0.36 ( I ) R 2 = 0.68 ( J ) R 2 = 0.85

PAGE 49

49 CHAPTER 3 MANAGEMENT STRATEGIE S FOR VERTISOLS: SOI L ORGANIC CARBON A ND AGGREGATE SIZE CLASSES IN CERE AL/LEGUMINOUS CROPPI NG SYSTEMS Background During the past decade, quantitative methods have been developed to understand the short term impacts of management practices on C long term storage and stability and the mechanism s governing C accumulation in soils (Wang and Hsieh, 2002; Paul et al., 2008). Studies have shown that C fractions are sensitive to management changes and act as potential indicators of change in soil ecosystems ( Cambardella and Elliott, 1992; Six et al., 1998; Wander et al., 2004). Many of these studies were based on the investigation of SOM aggregate conceptual models (Emerson, 1959), microaggregate stabilization theory (Edwards and Bremner, 1967), and the Tisdall and Oades (1982). In later studies, macro and micro aggregate formation was determined to be directly linked to SOM, except in oxide rich soils (e.g., Oxisols) (Oades and Waters, 1991). In addition, researchers (e.g., Six et al., 1998) develope d c onceptual models to assess the C associated with size fractions related to C pools and showed that the short term effects of tillage, crop, and crop residue may be due to the effects these practices have on the mechanisms by which C inputs (e.g., crop r esidue) stabilize in the soil. Soil C storage and stabilization mechanisms include aggregation, biochemical recalcitrance, and protective capacity of adsorbing organic materials (Baldock and Skjemstad, 2000). The assessment of C fractions is based on bio logical, physical, and or chemical characteristics of the organic constituents of the soil that are then related to C pools designated based on soil C decomposition and stabilization. The methods to estimate C fractions remain varied (Figure 3 1) and time consuming, however. Characterization of key constituents, i.e., measureable biological, chemical properties provides useful information regarding the sensitivity of these particular C constituents to

PAGE 50

50 management or environmental factors. Further, consensus among researchers on methodologies used to characterize soil C may facilitate data comparison across geographic regions to fill knowledge gaps regarding soil C dynamics related to environmentally challenging issues such as global climate change and carbon cycling. The m ajority of previous studies were conducted in temperate regions and little information is available on C dynamics in semiarid tropical regions. Organic matter management is critical in the semiarid tropics and plays a critical role in regul ating crop productivity Nutrient availability to crops is strongly dependent on the decomposition of the organic matter and plant residues. Our understanding relative to C responses in tropical soils managed within a range of conservation practices is s till limited. Due to high temperatures and rapid decomposition rates, SOM may be rapidly decomposed and lost from the system (Wani et al., 20 03 ). However, previous studies provide evidence that improved SOM management practices may reduce C losses and im prove C retention within the soil (Lal, 2004; Bhattacharyya et al., 2007; Srinivasarao et al., 2009). I n the current study, we assessed the short term (16 months) soil C responses to conservation agricultural management practices in semi arid tropical Vert isols: tillage, residue, and crop. W e hypothesized that the C in the whole soil would not and that the C associated with size fractions would serve as sensitive indicators of changes due to agricultural management factors. We used a range of methods to c haracterize the soil C for purposes of comparison of successful methods in broader contexts both within the current study area (ICRISAT, Hyderabad, India) as well in other regions. Materials and M ethods Site Description The study was conducted at the Int ernational Crops Research Institute for Semi arid Tropics (ICRISAT), Patancheru, India (17.5 N, 78.5 E, ~545 m elevation). Soils are classified as a

PAGE 51

51 Vertisols, a montmorillonite, Typic Pellustert developed on basaltic alluvium (El Swaify et al., 1985). The climate is semi arid with summer months between March to June and winter months between November to February. The rainy season occurs during the months of June to October. Temperatures range between 18 to 42C and annual average precipitation of 90 0 mm. The highest temperatures occur between March June and the lowest rainfall occurs from January to May. Historical L and U se The original vegetation was predominately grassland. Conversion to cultivated crops started in the 1920s, and more intensive cultivation occurred in the 1970s. During the four years before the initiation of the study, the study site was cultivated with maize ( Zea mays L.) alternated by chickpea ( Cicer arietinum L.) every other year. The land use history was variable previous to that time. The crops were not irrigated. Two or more crops were grown per year: the winter season crop, sown from October to December and harvested from March to June and the summer crop, sown at the beginning of the rainy season in June or July and h arvested in September October. Experimental Design The experimental design was split split plot with main plot tillage treatments; sub plot crop residue treatments and sub sub plot cropping systems. Tillage treatments included minimum tillage ( MT ) and con ventional tillage (CT). In CT the field was ploughed at the end of the winter season and the BBF landform was implemented before the start of the summer season. In the MT system, the BBF was implemented in the summer season and maintained throughout the winter season. The landform was maintained at ~ 1 meter width, with ~0.5 meter width furrow between each bed. Residue treatments included all crop residues (AR) and no crop residues (NR) added. In NR all crop residues at the harvest of each crop were re moved from the plots. In

PAGE 52

52 AR all crop residues at the harvest of each crop were redistributed as evenly as possible on the soil surface. Each sub treatment was divided into two plots and two cropping systems. Cropping systems were m aize and chick pea sequ ential (M+CP), and maize and pigeon pea intercrop (M/PP) systems In the M+CP rotation, maize was planted in June and harvested in November, followed by chickpea planting, which was harvested in March. In the M/PP intercrop system, pigeon pea was planted at the same time as maize, in June, in the summer season. Maize was harvested in November, while pigeon pea was harvested in March. Nitrogen and P were applied at rates of 60 and 13 kg ha 1 yr 1 respectively, for cereals, and 20 kg ha 1 yr 1 of N and 17 kg ha 1 yr 1 of P for legumes. Nitrogen was applied as urea and P as di ammonium phosphate in rows with a seed and fertilizer drill. Nitrogen in cereals was applied as 50% N basal application and the remaining 50% was applied in two equal split applicat ions at a one month interval. Soil test extractable potassium (K) levels were high (>125 mg kg 1 ) (ICRISAT, 2011), and thus no additional K was applied to the soil. All plots received gypsum at 200 kg ha 1 yr 1 and boron at 2.5 kg ha 1 yr 1 Zinc sulpha te at 50 kg ha 1 was applied alternate years to maize crop. All fertilizers were applied before planting of the summer season crop, within 1 2 weeks before tillage, in June 2009. No additional fertilizer was applied throughout the growing season. Soil S ampling and Analysis Soil samples were collected in June 2009, before the start of the experiment for soil characterization, at 0 15, 15 30, 30 60, and 60 90 cm soil depths. Soil samples were collected in November 2010, with in the second year of the exper iment, post harvest of maize, at 0 30 cm soil depth For each plot, three sub samples were randomly collected and mixed together to create one composite soil sample. All samples were air dried and sieved through a 2 mm sieve.

PAGE 53

53 Modified W et S ieving P roce dure The soil was separated into three size classes: macro (2000 and a silt and clay (1986). 100 g of air dried soil was submerged in a 500 mL beaker of dei onizer water and thus subjected to the disruptive forces of slaking for about 5 min before placing it on top of a 212 sieve. The sieving was done manually. An attempt was made to use comparable energy input by moving the sieve up and down approximatel y 50 times in 2 min. The fraction remaining on the top of a 212 sieve was collected in a hard plastic pan and allowed to oven dry at 65C and weighed. Water plus soil <212 was poured through a 53 sieve, and the same sieving procedure was repeated. The average recovery mass percentage of soil fractions after the wet sieving procedure was 97% of the initial soil mass. Microbial and Total Soil Organic Carbon Fractions and P ermanganate Oxidation of C (P OXC ) The SOC concentration of soil size fraction s and the whole soil was determined using the Walkley Black method (Nelson and Sommers, 1996). The SMBC was determined using chloroform fumigation (Anderson and Domsoh, 1978 and Brookes et al., 1985). The permanganate oxidation of carbon (POXC) analysi s w as based on Weil et al. (2003). Calculations The SOC s tocks were calculated using Equation 3 1. SOC s tock ( g m 2 ) = BD ( g cm 3 ) soil depth ( cm ) SOC ( mg g 1 ) *10 ( 3 1) The final values for SOC stocks were reported in g m 2 Statistical Analysis Stat istical analysis was performed with the software package JMP 8.02 (SAS Institute Cary, NC). Two and three

PAGE 54

54 T test were used to examine the effects of tillage, crop, and residue treatmen ts on SOC. Statistical significance was determined at the p < 0.05 level except if indicated differently. Results and Discussion Selected S oil P roperties In the top (0 15 cm) depth, the percent of coarse sand, fine sand, silt, and clay was 13, 15, 23, an d 50, respectively. Soil texture varied little with soil depth (data not shown). The soil was slightly alkaline with an average pH of 7.9. On average, soil bulk density was 1.33 g cm 3 which is consistent with results from previous studies on Vertisols within the same area (Bhattacharyya et al., 2007). Effects of Management Practices on SOC Storage Soil C stocks at the initiation of the study (June 2009) were not significantly different acros s the study site ( p <0.05) (Figure 3 2 ). Similarly, 16 months (approximately 1 year) after the treatments were imposed (November 2010), no significant differences in SOC stocks in the whole soil were observed d ue to treatments ( p <0.05) (Figure 3 2 ). The change in SOC stocks in the whole soil may not be apparent in t he current study due to the variability associated with this parameter and the short term duration of the study. Other studies have demonstrated that significant changes in SOC in tropical areas may be apparent after only 4 years, whereas similar changes may be apparent after 100 years in a temperate region (Adiku et al., 200 8 ). Although we expect more rapid change in SOC in tropical areas, we did not expect change in the SOC in the whole soil after approximately 1 year. Further, because of the variabili ty within the data, the differences between treatments were not statistically different (alpha=0.05). Effects of Management Practices on SOC Storage in C Associated with Aggregate Size Classes In June 2009, the mass of the soil particles in each aggregat e size class, on average for the experimental unit accounted for 46%, 15%, and 31%, respectively, for macro micro and clay +

PAGE 55

55 silt size class of the total soil subsample (data not shown). In November 2010, the mass of the soil particles, on average fo r the experimental unit accounted for 57%, 21%, and 19%, respectively, for macro micro and clay + silt aggregate size class of the total soil subsample (data not shown). The decrease of particles in the clay and silt size class correlated with those in the macroaggregate size class. These results suggest that free particles and silt and clay sized aggregates ( < 53m) formed into microaggregates and continued aggregating into macroaggregates (e.g., Tisdall and Oades, 1982; Oades, 1984; Six et al., 2004 ). Averaged across all treatment, t he percentage of SOC associated with the soil aggregate size classes in 2009 were 54%, 15%, and 30% of the total SOC (mg SOC 100 g soil 1 ) for the macro micro and clay + silt size fractions, respectively (Table 3 1). In November 2010, approximately 63%, 16%, and 21% of the total SOC (mg SOC 100 g soil 1 ) were associated with the macro micro and clay + silt size fractions, respectively (Table 3 1). Significant changes of SOC associated with aggregate size classes were observed primarily as an effect of the cropping system and tillage. In the macroaggregate size class (2000 212 m), the SOC was significantly higher in CT compared to MT with all residue added treatments ( p =0.04) (Table 3 1). However, crop treatmen ts did not affect SOC in that same size class ( p >0.05). As such, when all residues were added, SOC was higher in the CT, indicating that the CT treatment incorporated the residue into the soil making it available for microbial breakdown more so than leav ing all residues on the surface as in the MT (essentially acting as a mulch layer). No significant effects ( p >0.05) were observed in the other size classes, which averaged 74 + 20 g 100g soil 1 and 38 + 11 g 100g soil 1 for the macroaggregate and clay and sil t classes.

PAGE 56

56 Effects of Management Practices on Soil Microbial Biomass Carbon No significant effect of management was observed in the MBC ( p >0.05). In the current study, the MBC averaged 241 mg C kg soil 1 (ranging from 160 to 338 mg C kg soil 1 ). Th ere was no correlation between MBC and SOC associated with the silt and clay (<53 m) nor the 53 212 m aggregate size classes. The percentages of MBC to SOC in the whole soil ranged from 4 8%. Many studies suggest that the MBC is approximately 1 5% of the tota l organic matter (e.g., Glaser et al., 20 06); however, other studies state that soil MBC can be much higher than 1 5% (Denef et al., 2009). Comparison between Permanganate Oxidation of C (POXC) and C Associated with Aggregate Size Classes The POXC measur ed and compared with C associated with aggregate size classes showed a low correlation with the macro aggregate size class (2000 212 m) (R 2 = 0.36 p=0.12 ) (Figure 3 2 ), compared to the micro aggregate size class (212 53 m) (R 2 =0.55 p=0.03 ) (Figure 3 3 ) There was no correlation between the C associated with the silt and clay size fraction (<53m) and POXC (R 2 =0.0002 p=0.97 ) (Figure 3 4 ). Overall, the POXC was more closely related with the 212 53 m size fraction. There were no non linear relationsh ips across treatments, which suggest that POXC may be a sensitive indicator for the 212 53 m size fraction. Summary and Discussion The short term (16 months) effects of conservation agricultural practices (tillage, residue, and crop rotation) on SOC store d in the whole soil and in soil size classes were assessed. Within the 16 months of this study, C did slowly increase and was evident in increased C associated with the macro aggregate size class. However, the consistency of the MBC suggests that there i s a relationship between a stable soil microbial population and the breakdown of crop residues incorporated into the soil Further investigation of the data showed that as MBC increase s the soil respiration or C loss also increase s (R 2 =0.87 p=<0.001 ) (Fig ure 3 5 )

PAGE 57

57 Use of comparative methods for further characterization of C showed that POXC detects early changes in soil C due to agricultural management practices in short term (16 months) studies. However, additional research is required to assess the POX C measure throughout longer periods of time (e.g., 4 years). It is also noteworthy that the size fractionation method employed in this study has limitations. Further refinement of the method using density fractionation was first articulated by Six et al (2000). The density fractionation procedure essentially isolates a light and a heavy size fraction in addition to the physically separated fractions using sieving. By isolating the density fraction, researchers are able to distinguish between fine and coarse organic matter. Although the refined method is used, general reference to the macro and micro size fractions is still used to assess agricultural management practices such as tillage, applied to a range of minerologies (e.g., 2:1, 1:1 clay types) ( e.g., Six et al., 2004; Denef et al., 2004). However, it is still unknown as to how well these fractionation procedures capture the various interactions between microbes, roots, binding agents, and structure within the soil environment.

PAGE 58

58 Table 3 1 Aver age difference in mg SOC 100g 1 soil for June 2009 and November 2010 0 30 cm soil depth, under con ventional tillage and minimum tillage with all residue, size fraction 2000 212m ICRISAT, Patancheru, India 2000 212m Tillage *Residue Avg diff SOC (mg 100gsoil 1 ) Conventional tillage *All residue 235a Minimum tillage *All residue 112b Statistical analysis was conducted per size fraction. Different letters designate statistical significance at the 0.05 proba bility level per size fraction, using LSM St udent T test.

PAGE 59

59 Figure 3 1 Soil C fractionation methods

PAGE 60

60 The line indicates the linear regression for predicted points based on observed points (indicated by black diamonds ). Figure 3 2. Scatterplot of SOC (mg/kg soil) associated with SOC size fraction (2000 212m) and POXC (mg/kg soil), ICRISAT, Patancheru, India (R 2 = 0.36 p=0.12 ). Data represent the average across all treatments (n = 1).

PAGE 61

61 T he line indicates the linear regre ssion for predicted points based on observed points (indicated by black diamonds ). Figure 3 3 Scatterplot of SOC (mg/kg soil) associated with SOC size fractions (212 53m) and POXC (mg/kg soil), ICRISAT, Patancheru, India (R 2 =0.55 p=0.03 ) Data represen t the average across all treatments (n = 1).

PAGE 62

62 T he line indicates the linear regression for predicted points based on observed points (indicated by black diamonds ). Figure 3 4 Scatterplot of SOC (mg/kg soil) associated with SOC s ize fractions (<53m) and POXC (mg/kg soil), ICRISAT, Patancheru, India (R 2 =0.0002 p=0.97 ) Data represent the average across all treatments (n = 1).

PAGE 63

63 Figure 3 5 Soil respiration (y axis) and soil microbial biomass carbon (SM BC) (x axis), ICRISAT, Patancheru, India ( R 2 =0.87 p=<0.001 ). Data represent the average across all treatments (n = 24)

PAGE 64

64 CHAPTER 4 UTILIZING GEOSTATISTICS TO INCREASE THE EFFICACY OF SAMPLING PROTOCOLS FOR MEASURING SOIL ORGANIC CARBON Background Managem ent practices that e nhance SOM accumulati on and thus SOC sequestration are of importance for agricultural production and global carbon cycling. Conservation agricultural practi ces such as minimum tillage ( MT ) crop rotation, and cover crops can potentiall y increase SOC accumulation in soils (Lal, 2004). These management practices may be of particular use in semi arid tropical soils that have inherently low soil fertility. Research shows that SOM management practices support increased SOC sequestration, n utrient availability, and decreased carbon emissions (Lal, 2004). Studies in semi arid tropical soils (Wani et al., 2007; Srinivasarao et al., 2009) indicate further potential for significant increases in SOC and crop productivity with improved agroecosyt ems management practice. Several studies (e.g., Wu et al., 2009) have demonstrated the use of geostatistic al methods to improve the efficacy of sampling protocols used to measure SOC, thereby more readily identifying management practices that improve soil quality and reduce carbon emissions. In general, prior information about a research site may be necessary unless the average variogram (McBratney and Pringle, 1999) of the soil property of interest (e.g., carbon) is known. Geostatistics provides a set of tools to incorporate the location of observations into their statistical processing for mapping of soil properties from a typically small set of available data. The cornerstone of the geostatistical approach is the estimation of the variogram that measure s how the spatial dependence of soil properties changes depending on the distance between observations. Spatial sampling schemes can be

PAGE 65

65 optimized utilizing information about the spatial correlation of soil properties described by their variogram (Ferreyra et al., 2002). In addition, the accuracy of local prediction can be increased using geostatistical interpolation procedures that account for significant correlation between primary and secondary variables (Simbahan et al., 2006). While secondary informa tion may prove to be useful in the estimation of a primary variable at unsampled locations, there is need to develop methodologies for specific applications e.g., assessment of SOC (Simbahan et al., 2006) or specific areas of interest (Hengl et al., 2004). Grid (systematic) sampling is an easy way to collect information for geostatistical analysis and has been shown to be the most efficient for mapping purposes (Burgess et al., 1981). Depending on the size of the prediction support (i.e., the number of sa mples used per prediction area) used in kriging, one can distinguish point kriging and block kriging (Burgess and Webster, 1980). In point kriging, observations that are typically recorded for small soil cores that can be assimilated to points are used to predict the soil property at unsampled point locations. In block kriging, the objective is to predict the average values of soil properties throughout areas or blocks that are bigger than the sampling support (e.g. a watershed in the present study), an o peration known as spatial aggregation or upscaling (Goovaerts, 1998). McBratney and Pringle (1999) demonstrated that block kriging is more useful than point kriging for the assessment and recommendation of management practices since the block averages are more amenable to management decisions than point estimates. Regardless of the size of the prediction support, kriging provides both a predicted value and the associated variance of prediction errors. The so called kriging variance

PAGE 66

66 (KV) can be used to comp ute confidence intervals for the prediction using the assumption that the prediction errors are normally distributed (Webster and Oliver, 1990). One characteristic of the KV is that its computation requires only the knowledge of the geographical coordinat es of the samples and the variogram model. The KV can thus be used to quantify the expected benefit of increasing the sampling density on the precision of the prediction (Burgess et al., 1981). This study presents an empirical application of geostatistics to semi arid tropical Vertisols in a micro watershed of Andhra Pradesh, India. The primary objectives of this study were ( 1) to show how information about the pattern of spatial autocorrelation can be used to determine the sampling density required for a chieving a given accuracy in the estimation of the mean concentration of soil organic carbon; ( 2) to compare the prediction performance of ordinary kriging (OK) and regression kriging (RK), and ( 3) to evaluate the benefits of different covariates in RK. Ma terials and Methods Site D escription Kothapally (KP) micro 640m) is located in the Ranga Reddy district of Andhra Pradesh state, (India) in the Krishna Basin, nearly 40 km from the International Crops Research Institute for the Semi Arid Tropics (ICRISAT) Center. The KP micro watershed covers 465 ha, of which 430 ha are cultivated. The KP micro watershed is characterized by an undulating topography with an average slope of 2.5%. Soils are predominantly Vertisols. The soil depth ranges from 30 to 120 cm (Sreedevi at al., 2004).

PAGE 67

67 The climate is semi arid tropical dry, with a monsoonal rainy season that lasts for two to 4.5 months of the year (June to October). Approxima tely 85% of the water supplied to the fields is generated from monsoonal rains. The summer months are from March to June, and winter months from November to February. Mean monthly air temperatures range between 18 and 42C and the monthly average precip itation during the rainy months range between 0 430 mm (annual average ~90 mm) where potential evapotranspiration exceeds precipitation. Soil S ampling and S oil A nalysis A set of 21 plots were sampled in February 2010 using a 400 m grid spa cing within a 4 65 ha area (Figure 4 1a). During sampling, geographic positioning system (GPS) coordinates were used to locate each sampling location. A few samples were eliminated from the sampling due to site specific conditions, e.g., sample point was located in a wa terway. Within each plot, three soil core samples were collected from the top 30 cm and mixed to create a homogenous, composite sample A subsample of the soil sample was air dried, ground, and sieved through a 2 mm screen for analysis of extractable nutri ents. A second subsample of the 2 mm sieved so nitrogen (TN). The soil organic carbon (SOC) concentration determined using the Walkley Black method (Nelson and Sommers, 1996). Total nitrogen was estimated using the Kj eldahl method (referred to as total Kjeldahl nitrogen (TKN) (Kjeldahl, 1883 ). Soil moisture was determined on 10 g of soil dried at 105 C for 24 hours. The bulk density of the samples was determined by the field moist method using a core sample (diameter 50 mm) of known volume (100 cubic cm). The soil pH was measured with a

PAGE 68

68 glass electrode using soil to water ratio of 1:2. Particle size analysis was conducted using the hydrometer method (Gee and Bauder, 1986). Geostatistical A nalysis A geostatistical an d exploratory spatial data analysis of physical and chemical properties was conducted using the commercial GIS software SpaceStat (BioMedware, Inc. 2011) Based on the shape of sample histograms, a few variables were transformed (normal score transform, Goovaerts, 1997) before conducting the analysis since statistics, such as variograms, are sensitive to the presence of extreme values. Scatterplots were created to assess the correlation between soil C and potential covariates. The experimental variogram of each soil property (e.g. soil texture, soil C) was computed and a model was fitted by least square regression. Ordinary kriging (OK) (i.e., a generalized least square regression algorithm) was then used to predict values of soil properties at the node s of an interpolation grid to illustrate how these soil properties vary across the watershed (Goovaerts, 1997). Simple block kriging was used to analyze the impact of sampling density on the estimation of SOC at the scale of a watershed. The KV associat ed with the prediction of average soil organic carbon throughout the Kothapally watershed was computed for sampling grids with spacing ranging from 500 m (13 nodes ) to 50 m (1169 nodes); see Figure 1. In all cases, the watershed was discretized using a 50 m spacing grid to compute the point to block covariances in the kriging system (Goovaerts, 1998). The benefit of incorporating secondary information in the prediction was explored using residual kriging (RK) whereby the prediction variance is obtained in three steps: ( 1) a linear regression is fitted between SOC observations and the covariates, ( 2) the

PAGE 69

69 variogram of regression residuals is computed and used in simple block kriging, and ( 3) the simple block kriging variance is added to the variance of the r egression estimate. The interpretation of the magnitude of the KV must be accomplished in relation to the objective of the analysis. If the objective is to detect a change or difference in average SOC of magnitude D, the KV can be used to compute the pro bability that this difference can be detected with a given significance level. The so called power, which represents the probability to reject the null hypothesis of equality of means when the means are actually different, was calculated as: Power= (1 error) (4 1) ( (D+ z ) G ( (D z ) ; z = |G 1 ( /2)|; and G = cumulative distribution function for normal distribution, level Geographically weighted regression (GWR) ( Fotheringham et al., 2002 ) was conducted at the nodes of a 50 m spacing grid to assess how the correlation betwee n variables changed across the study area. At each grid node ui, the regression model was built using the ten nearest neighbors. To attenuate the impact of distant observations in the computation of the local regression coefficients, each observation rec eived a weight that increases with its proximity to the grid node. The following bi squared weights were assigned to each observation at ui: = (1 )2 if d i < G and 0 otherwise (4 2) Where d i is the Euclidian distance b etween u and u i and G is the bandwidth that was identified with the distance between ui and the tenth most distant observation.

PAGE 70

70 The bandwidth is therefore larger in sparsely sampled areas (spatially adaptive bandwidth). Results and Discussion Explorator y Data Analysis The histograms generated along with some basic statistics are provided in Figures 4 2 and 4 3. The histograms of soil SOC and TKN data were relatively symmetrical. The distribution of soil texture parameters (sand, silt, clay) was skewed and these observations were normal score transformed (Goovaerts, 1997) before geostatistical analysis. The nature and strength of relationships for a few pairs of variables was quantified using scatterplots and correlation coefficients (Figures 4 4 throu gh 4 6). The correlation between SOC and TKN was linear with a strong correlation of 0.90. The correlation between SOC and other covariates, such as % silt and soil mo isture, was much weaker: 0.30. The maps of GWR based correlation coefficients (Figure 4 7) illustrate how the relationship between SOC and two covariates (e.g., silt % and SOC %) changed across the Kothapally watershed. The GWR for SOC and TKN showed a narrow range of correlation coefficients (0.81 0.98), where correlation was higher in are as that were more densely sampled. The spatial distribution of correlation coefficients for SOC and soil moisture (SM) followed the spatial trend in SM data v alues within the watershed (Figure 4 8). The correlation coefficient gradient (from 0.66 to 0.01 ) between SOC and silt % was consistent with the increase in s ilt % across the watershed (Figure 4 9). Variography The variograms generated to characterize the spatial variability of either SOC or the residuals of the linear regression between SOC and a s econdary variable (e.g., silt

PAGE 71

71 %) are shown in Figures 4 10 through 4 13. All variograms (Figures 4 10 through 4 13) are fairly flat with values close to the sample variance (horizontal dashed line), which indicates that most of the spatial variability acro ss the watershed occurs at distances smaller than the shortest sampling interval. The small number of observations available, combined with the relatively large separation distance between observations, makes the sample variograms hard to model. In other words, ideally the observations should have been taken closer to each other. The model was however fitted using least square regression with the constraint of a zero nugget effect for most variables. As expected, the variograms of SOC residuals display a lower sill than the SOC variogram because the residual variability (i.e. variability that is not explained by secondary variables) is smaller than the original variability. Similarly, the ranges of residual variograms tend to be shorter because the resi dual variability is typically less structured spatially after removal of the influence of covariates. The SOC variogram range was 881m. The SOC with silt % as a secondary variable showed results consistent with the silt distribution across the watershed and correlation map. Because of the distinct change in silt content % across the watershed, the variability explained by the silt and SOC values drops off at 257m. A similar result was observed in the SOC and TKN variogram but the range extends to 304m. As such, using TKN and silt content % as secondary data reduced the range of the variograms. Sampling D esign The variogram model for SOC generated to predict the value of SOC and the associated KV at the nodes of a 50 m grid (point kriging) and to predi ct the average value for the watershed (block k riging) is shown in Figure 4 14. Table 4 1 lists the results of the block KV computed for five sampling grids with 50, 100, 200, 300, and 400

PAGE 72

7 2 m spacings, corresponding to a decreasing number of samples: from 1169 to 13. Figure 4 15 illustrates how the block KV increases as the number of samples decreases. This greater KV results in wider confidence intervals that are listed in Table 5 1. Similarly, as the number of observations decreases and the prediction becomes less reliable, the ability to detect differences in SOC declines, as illustra ted by the power curves in Figure 4 16. These curves are plotted for differences of increasing magnitude and show that the power of the method increases with wider differe nces and larger number of observations (i.e. smaller spacing of the sampling grid). In other words, it is easier to detect changes in SOC content of larger magnitudes when more observations are available. For example, for a 400 m sampling grid (21 sample s) and D equal to 0.1, the power equals 0.35. As the spacing grid decreases to 300 m (35 samples), the power increases to 0.70. Benefit of S econdary I nformation The kriging procedure described for SOC in text above was repeated using RK and different cov ariates: TKN, SM, and silt. The KV decreases with the inclusion of th ese covariates (Figures 4 17, 4 19, 4 21). At the same time, the confidence intervals get narrower while the power to detect changes in SOC increases (Table 4 2 through 4 4). The powe r to detect differences in SOC ranging from 0.1 to 0.5 % was equal to 1 given KV for sampling density for 50,100,200,300 and 400 grid s pacing with secondary data (Figures 4 18, 4 20, 4 22). The power reached 1 for SOC detection as low as 0.2 0.3 % when S M is used as secondary information. Similarly, the power with silt % as a secondary data reached 1 with SOC detection of 0.3 0.4 % SOC.

PAGE 73

73 Summary This study shows how information about the pattern of spatial autocorrelation can be used to determine the sam pling density required for achieving a given accuracy of the estimation of the mean concentration of soil organic carbon. The effect of the sampling density is quantified using power functions computed for different detection limits, D. With this informa tion, a researcher can determine the mean concentration of SOC at a fixed confidence interval and related sampling density required to meet the study objectives. For example, if a researchers wishes to see an SOC difference of 0.2 %, then 35 samples are r equired using a 300 m grid. Covariate TKN, SM, and silt % decreased KV, resulting in increased prediction at a fixed confidence interval. However, the benefits of using secondary data TKN with respect to the cost of data collection and analysis are compar able to SOC. Soil moisture and silt % may provide secondary data at a lower cost and still slightly increase the predictive power of SOC.

PAGE 74

74 Table 4 1 Kriging Variance (KV), Sampling density, Stnd error, and 95%CI Spacing (m) Sampling density KV* S td** error 95%CI*** 50 1169 0.00000 0.00000 0.00000 100 309 0.00002 0.00462 0.00925 200 74 0.00019 0.01361 0.02721 300 35 0.00053 0.02298 0.04595 400 21 0.00109 0.03300 0.06601 500 13 0.00218 0.04669 0.09338 Total kriging variance (KV) is equal to the squared standard error of regression plus the residual kriging variance ** Std: standard *** CI represents Confidence Interval

PAGE 75

75 Table 4 2 SOC and TKN Spacing (m) Sampling density KV* Std** error 95%CI*** 50 1169 0.00000 0.00 000 0.00000 100 309 0.00000 0.00150 0.00301 200 74 0.00002 0.00442 0.00884 300 35 0.00006 0.00800 0.01600 400 21 0.00017 0.01300 0.02599 500 13 0.00035 0.01877 0.03755 Total kriging variance (KV) is equal to the squared std error of regression plus the residual kriging variance ** Std: standard *** CI represents Confidence Interval

PAGE 76

76 Table 4 3 SOC and SM Spacing (m) Sampling density KV* Std** error 95%CI*** 50 1169 0.00000 0.00000 0.00000 100 309 0.00001 0.00341 0.00682 200 74 0.00010 0.01024 0.02049 300 35 0.00031 0.01748 0.03497 400 21 0.00067 0.02590 0.05180 500 13 0.00143 0.03775 0.07550 Total kriging variance (KV) is equal to the squared std error of regression plus the residual kriging variance ** Std: standard *** CI represents Confidence Interval

PAGE 77

77 Table 4 4 SOC and Silt (%) Spacing (m) Sampling density KV* Std** error 95%CI*** 50 1169 0.00000 0.00000 0.00000 100 309 0.00007 0.00864 0.01728 200 74 0.00038 0.01948 0.03895 300 35 0 .00088 0.02970 0.05941 400 21 0.00157 0.03960 0.07920 500 13 0.00269 0.05189 0.10378 Total kriging variance (KV) is equal to the squared std error of regression plus the residual kriging variance ** Std: standard *** CI represents Confidence Interval

PAGE 78

78 Figure 4 1 (a) Location map of soil samples collected within the Kothapally microwatershed on February 2010, and three grids of increasing spacing intervals that were used for spatial interpolation and geographically we ighted regression: (b) 50 m grid, (c) 100 m grid, and (d) 200 m grid. (a) ( b ) (a) (b) (c) (d)

PAGE 79

79 Figure 4 2 Histogram of Soil Organic Carbon (SOC) %. Cumulative distribution is indicated by the blue line. Summary s tat istics Number of data: 21 Minimum: 0.32 Maximum: 0.95 Mean: 0.65 Std dev.: 0. 05 SOC

PAGE 80

80 Figure 4 3 Histogram of TKN and s ummary statistics. Cumulative distribution is indicated by the blue line. Summary s tat istics Number of data: 21 Minimum: 356 Maximum: 959 Mean: 628 Std dev.: 175 TKN

PAGE 81

81 Figure 4 4 Scatterplot of SOC and Total Kjeldahl Nitrogen (TKN), with summary statistics Summary s tat istics Number of data: 21 SOC mean: 0.65 SOC s td dev.: 0.20 TKN mean: 627 TKN s td dev.: 175 Coef. of var.: 0. 9 0 SOC

PAGE 82

82 Figure 4 5 Scatterplot of SOC and Soil Moisture (SM) normal scores, with summary statistics. Summary s tat istics Number of data: 21 SOC mean: 0.65 SOC s td dev.: 0.20 SM mean: 0.07 SM s td dev.: 0.05 Coef. of var.: 0. 30 SOC

PAGE 83

83 Figure 4 6 Scatterplot of SOC and Silt% normal scores, with summary statistics. SOC Summary s tat istics Number of d ata: 21 SOC mean: 0.65 SOC s td dev.: 0.20 Silt_norm scores mean: 1.15E 14 Silt_norm scores s td dev.: 0.99 Coef. of var.: 0. 30

PAGE 84

84 Figure 4 7 Map of local correlation coefficients estimated by geograph ically weighted regression for three pairs of variables: a) SOC and total Kjeldahl nitrogen, b) SOC and soil moisture_normal scores, and c) SOC and silt_normal scores. (a) (b) N 0.5 9 0.03 N 0. 98 0.90 0.81 (c) 0.6 6 0.01 N

PAGE 85

85 Figure 4 8 Soil Moisture (SM) (%) sampled values, Kothapally watershed, Andhra Pradesh, India, February 2010 N SM ( % ) N

PAGE 86

86 Figure 4 9 Silt content (%) sampled values, Kothapally watershed, Andhra Pradesh, India, February 2010 Silt ( % )

PAGE 87

87 Figure 4 10 Experimental variogram of soil organic carbon content (%) with the exponential model fitted, sill: 0.04532, nugget: 0.0, range: 495 meters. The variogram value Spacing (m)

PAGE 88

88 Figure 4 11 Experimental variogram of carbon co ntent (%) regression residuals when using TKN (mg/kg) as secondary variable. The fitted model is spherical with a sill: 0.00593, nugget: 0.0, range: 304 meters. The horizontal dashed line denotes the variance of the observations. (h) = variogram value Spacing (m)

PAGE 89

89 Figure 4 12 Experimental variogram of soil organic carbon content (%) regression residuals when using soil moisture (%) normal scores as secondary variable. The fitted model is exponential with a sill: 0.04353, nugget: 0.0, range : 881 meters. The horizontal dashed line denotes the variance of the observations. (h) = variogram value Spacing (m)

PAGE 90

90 Figure 4 13 Experimental variogram of soil organic carbon content (%) regression residuals when using silt (%) normal scores as secondary variable. The fitted model is exponential with a sill: 0.02408, nugget: 0.01465, range: 257 meters. The horizontal dashed line denotes the variance of the observations. (h) = variogram value Spacing (m)

PAGE 91

91 Figure 4 14 Map of ordinary kriging results: a) estimated values, and b) kriging var iance (KV) for SOC, Kothapally watershed, Andhra Pradesh, India, February 2010 Kriging estimate Kriging variance 0.05 0.04 0.80 0.52 (a) (b) N

PAGE 92

92 Figure 4 15 Increase in the prediction variance of soil organic carbon (SOC) as the sampling grid increases, Kothapally watershed, Andhra Pradesh, India, Februa ry 2010

PAGE 93

93 Figure 4 16 Power for the detection of a given difference in soil organic carbon (SOC) for sampling grids with a spacing ranging from 50 to 500 meters, Kothapally watershed, Andhra Pradesh, India, February 2010

PAGE 94

94 Fig ure 4 17 Increase in the prediction variance of soil organic carbon (SOC) with secondary variable total kjedahl nitrogen (TKN) normal scores as the sampling grid increases, Kothapally watershed, Andhra Pradesh, India, February 2010

PAGE 95

95 Figure 4 18 Power for the detection of a given difference in Soil organic carbon (SOC) using secondary variable TKN normal scores and sampling grids with a spacing ranging from 50 to 500 meters, Kothapally watershed, Andhra Prades h, India, February 2010

PAGE 96

96 Figure 4 19 Increase in the prediction variance of soil organic carbon (SOC) with secondary variable soil moisture normal scores as the sampling grid increases, Kothapally watershed, Andhra Pradesh, India, F ebruary 2010

PAGE 97

97 Figure 4 20 Soil organic carbon (SOC) with secondary variable soil moisture for sampling grids with a spacing ranging from 50 to 500 meters, Kothapally watershed, Andhra Pradesh, India, February 2010

PAGE 98

98 Figure 4 21 Increase in the prediction variance of soil organic carbon (SOC) with secondary variable silt content (%) normal scores as the sampling grid increases, Kothapally watershed, Andhra Pradesh, India, February 2010

PAGE 99

99 Figure 4 22 Soil organic carbon (SOC) for sampling grids with secondary variable silt content with a spacing ranging from 50 to 500 meters Kothapally watershed, Andhra Pradesh, India, February 2010

PAGE 100

100 CHAPTER 5 QUANTIFICATIO N OF UNCERTAINTY IN MODELABLE AND MEASUREABLE SOIL ORGANIC CARBON POOLS Background Uncertainty or variability is presented by both measured and modeled soil organic carbon. In an agroecosystem, the effects of driving factors external to the system such as land use or management (e.g. tillage) change this variability. Narrowing the gap between measureable soil organic matter (SOM) fractions and modeled SOM pools will increase the opportunity to validate model simulations and subsequently increase the curre nt understanding of carbon (C) dynamics. My approach to narrowing this gap and the main objective of this study is to investigate the differences between a physical fractionation method to measure C pools and three model based initialization approaches in the Century based DSSAT (the Decision Support for System for Agrotechnology Transfer) model ( Gijsman et al., 2002; Porter et al., 2009 to model soil organic carbon (SOC). Further, data from one short term (16 month) study (Study 2 Chapter 3) are used to compare the measured and the modeled approaches, to quantify the uncertainty presented by both. I hypothesized that the investigation and the quantification of the uncertainty presented by both modeled and measured soil organic carbon would improve our th e current understanding of soil C pool initialization for use in simulating carbon dynamics for decision support for SOM management. Description of Measured Soil Organic Matter Fractions Physical C haracteristics Soil texture affects the decomposition of S OM, with decomposition rates decreasing in the order of sand>clay>silt ( Chris topher et al. 2008 ). Many of the studies of decomposability and soil texture were based on the investigation of SOM aggregate

PAGE 101

101 conceptual models (Emerson, 1959), microaggregate st abilization theory (Edwards and Oades (1982). Elliot (1986) tested the conceptual models using fractionation methods and a series of field experiments. Using the Elliot (19 86) wet sieving size fractionation procedure the soil is separated into three size classes: macro (2000 (212 53 and clay and micro (212 53 d, silt, and clay aggregated particles, whereas the silt and clay free. In addition, researchers (e.g., Six et al., 1998) developed conceptual models to assess the C associated with each size fraction in relation to management practices in agroecosystems. Quantitative methods used to fractionat e soils include density and size based fractionation methods (Wander, 2004). As fresh organic matter (FOM) (i.e., litter incorporated into the soil from the previous crop) enters the soi l the organic matter fragments into smaller particles, referred to as particulate organic matter (POM) These POM particles are associated with the >53 size fraction (Six et al., 2004). The POM may constitute a large portion of SOM (Wander, 2004). Some literature states that the percent of POM present in the soil may range from 20 40% (Six et al., 1998; Six et al. 2000). Wander (2004) states that i n mineral soils, the POM may account for approximately 25% or less of the total SOM Other studies (e.g., Skjemstad et al. 1993) found that the amount of C associated with the silt and clay size fractions is approximately 59% of the total carbon. In gene ral, studies show that as the sand content increases, the portion of carbon in the POM increases ( Six et al., 200 4 ). In coarser textured soils, POM content tends to

PAGE 102

102 decline with increasing clay content (Wander, 2004). POM content tends to be higher in lo am compared to sandy loam or silt y loam soils ( Six et al., 2000). Biological C haracteristics Soil biological organisms mediate SOM and thus the cycling of nutrients within organic matter. As the microbial organisms consume the organic matter they produc e organic compounds that act as binding agents in soil aggregation and assist in C stabilization in soil resulting in C that may not decay for centuries (Six et al., 2004). Soil organic matter is also lost or respired due to biological microbial activit y. The organic matter and the organic compounds, produced by the microbes, i ncorporated into the soil are also often associated with some of the C in each of the aggregate size classes or size fractions (2000 212, 212 53, <53 m). Organic matter and / or compounds associated with each size fraction are biologically characterized by their turnover time or decomposition rate. Depending on the crop, the >53 m (2000 212, 212 53 m) size fractions have a potential decomposition rate comparable to that of the crop (Six et al., 2000). It is hypothesized that FOM inputs will mineralize at a similar rate as free organic matter (i.e. not occluded within aggregates) already within the soil ( Six et al., 2002 ). However, and noteworthy, if disturbance (i.e., tillage) to the system is frequent enough, then the turnover of the 2000 212 m may be fast enough to stabilize (e.g., occlude) FOM (Six et al., 2004). Also noteworthy, there is a lack of information on the tim e scales turnover in relation to inputs of FOM in agro ecosystems. Some general estimates do exist, for example, Plante et al. (2002) estimated the mean residence time of the 2000 212 m to average 27days with estimates as low as 5 days. Studies have shown that the SOM associated with the <53 m fraction is not microbial ly derived (Six et al., 2000). The chemical composition of the C compounds

PAGE 103

103 (e.g., aromatic carbon structures) associated with the <53 m suggest that these compounds are resistant to decomposition (Christiansen, 1996). Also noteworthy, densi ometric analyses of silt sized organo mineral complexes do not lead to distinct and homogeneous pools of organic matter with different turnover times and thus methods such as size fractiona tion may provide more reliable estimates (Six et al., 2000). Che mical C haracteristics Carbon compounds, carbohydrates, amino acids and proteins, nucleic acids, lipids, lignins, and humus (in that order) have different degrees of decomposability (Cadisch and Giller, 1997). The FOM added to the soil surface is not alway s considered to be part of the active SOM pool, however, the characteristics of the FOM (e.g., lignin content) are relevant to the determination of the active pool (Porter et al., 2009). Chemically recalcitrant SOM structures contribute to the recalcitrant pool if these structures are associated with fine particle size frac tion (<53 m) (Six et al., 2000 ). Research has shown that the <53 m size fraction is primarily associated with aromatic C compounds (e.g., lignin) ( Oades 1 990 ; Christensen 1996) Descr iption of Modeled Soil Organic Matter Pools T he measured SOM fractions (described above) are thought to be equivalent to model ed conceptual SOM fra ctions or pools, if and only if these fractions are unique and no t compos ed of more than one biologically C p ool, defined by decomposition rate (Smith et al., 1997 ). The simplification of the SOM dynamics in a modeling environment is a tool that allows for extrapolation from current knowledge on SOM to test environmental and management effects on SOC cycling. T o reiterate, conceptual (modeled) SOM pools do not correspond directly with experimentally measured SOM fractions The modeled conceptual pools are defined on the basis of ranges of

PAGE 104

104 decomposition rates, which may include C from different size classes as de termined by physical fractionation methods. Biological C haracteristics Similar to the measured SOM fractions, the modeled SOM pools are biologically characterized by their turnover time s or decomposition rate s Soil biological organisms or microbial bioma ss are considered as an active pool of C SOM 1, with a turnover time of days (P arton et al., 1987, 1988 ). The organic matter and the organic compounds, produced by the microbes, incorporated into the soil are often associated with an intermediate pool of C SOM 2, with a turnover time of years (Porter et al., 2009). The FOM inputs are considered to mineralize at a similar rate as the crop, in particular based on the metabolic (e.g., sugars) and structural (e.g., lignin) litter. The organic matter that is stabilized in the soil is associated with a passive pool of C with a turnover time of hundreds of years (Porter et al., 2009). To initialize the SOM pools in DSSAT FOM is subtracted from the total organic carbon pool (Porter et al., 2009). In general the decomposition rate of organic matter is computed on a daily basis, resulting in a shift of the amounts of organic matter from more active pools (FOM and SOM 1) to stable pools (SOM 2 and SOM 3) (Porter et al., 2009). Chemical Carbon compounds (e.g., proteins, sugars, lignins ) are considered in the FOM added to the soil surface Chemically recalcitrant SOM structures (e.g., lignin) are conceptually included in the SOM 2 and SOM 3 pools, but are not considered in initialization procedures for the DSSAT model.

PAGE 105

105 Physical In general the modeled effect of increasing clay content is (1) decreased decomposition of carbon in SOM 1 ; (2) decreased carbon in SOM 1 from the SOM 2 pool ; (3) decreased carbon in SOM 2 ; and (4) increased carbon in SOM 3 (Porter et al ., 2009). Further, one of the initialization procedures for DSSAT considers the SOM 3 pool in relation to the silt and clay size fraction, as described below in model initialization procedures, Adiku (published in Porter et al., 2009). Description of Diff erent Model Initialization Procedures Gijsman et al. (2002) ; Porter et al. ( 2009) Method Gijsman et al. (2002) approach uses a field history code, representing one of two management scenarios (i.e., grassland and cultivated) This method was later modifie d (Porter et al., 2009) to include the duration in years that the management scenario had been in effect before the start of simulation. The other two pools, SOM 1 and SOM 2, are then assumed to be 5% and 95% of the remaining amount of soil carbon, respec tively. Basso et al. (2011) The Basso et al. (2011) approach is similar to the Gijsman Porter approach, but uses an iterative procedural approach to estimate soil C pools using field specific management histor y and soils. An initial total C value is r un for a set timeframe iteratively until the measured target total C value is realized. At the end of the iteration the total C value should match the observed soil C for that field, and the SOM 1, SOM 2 and SOM 3 pools that are predicted in the model fo r that sequence of simulations are assumed to be the initial conditions for forward simulations. Because the decomposition of SOM 2 occurs on the order of years, this conceptual C pool may be

PAGE 106

106 estimable within the timeframe of most available data (i.e. on the timescale of 20 to 50 years). Importantly, this method allows the user to approximate the SOC pools based on observed total soil C and limited field history information. Linear Regression Approach Developed by Adiku (published in Porter et al., 2009) The linear reg ression approach developed by Adiku (personal communication, Gainesville, FL) is based on a set of linear regression equations for mineral associated carbon following con cepts developed by Duxbury ( 2006 ) Adiku (published in Porter et al., 2 009) refined the linear equations to use the silt and clay size fraction of the soil, rather than the total carbon, to estimate the recalcitrant C pools, namely SOM 3. The other two pools are then estimated based on an assumption that the remaining C afte r computing SOM 3 C was composed of 5% of microbial C (SOM 1) and 95% of intermediate C (Equation 5 2). SOM3= (0.015 (clay + silt) + 0.069) / OC (5 2) Where SO CT =0.05*(1.0 SOM3) And, SO MT =1.0 SOM3 SO CT Materials and Methods Assumptio ns U sed for the C omparison of M easured and M odeled P ools B iological physical, and chemical characterization of both measured soil organic carbon fractions and modeled soil carbon pools are described above in text. Based on these descriptions, the followin g assumptions were made for comparison of measured and of modeled soil organic carbon. The FOM is not included in the total measured soil organic carbon; it is added at the point of model initialization given the selected crops, in this case, maize (Zea m ays L.) and chickpea (Cicer arietinum L.) and respective model reference values for C compounds (e.g., sugars, proteins, lignin). Measured soil

PAGE 107

107 microbial biomass carbon (SMBC) and the active SOM 1 are compared, assuming the measured fraction is composed o f material that decomposes at the same rate. Measured silt and clay associated soil organic carbon (<53 m) and the passive, SOM 3 pool are compared. Finally, SOM 2 is considered as SOM 2 = 1 SOM 3 SOM 1 and compared to the >53 m size fraction or POM fraction defined in the introduction of physical fractions. The implications of the noted assumptions are as follows. Given previously described work by Six et al. (2000), the decomposition rate of the FOM in the soil may be the same as that of the respec tive crop grown on the soil in no tillage systems. However, due to the disturbance and mechanical breakdown of litter in tilled soils, the decomposition rate as well as the occlusion of carbon material within aggregates may increase. As such, if model sim ulations are run using different growing seasons in tilled soils, the total C and SOM pools model estimates may not be reasonable compared to measured SOC values. Note that planting may also cause soil disturbance within the DSSAT model. Descri ptions of I nitialization A pproaches : I nvestigat ions using my data Description of E xperimental U nit Study 2, CHAPTER 3 data were used for this study. The research was conducted on Vertisols at the ICRISAT farm in Patancheru (Hyderabad, India). In this study, crops w ere grown without irrigation and solely dependent on rainfall. Depending on the treatments, one or two crops were grown per year: the winter season crop (the post rainy season crops were grown during November to February) and the summer season crop (the r ainy season crops were grown during June to October).

PAGE 108

108 The climate is semi arid tropical, with a rainy season that lasts for 2 to 4.5 months from June to October (S W monsoon), post rainy season (October mid February), and summer season (March to May). Mean monthly air temperatures range between 18 and 42C, with average precipitation during the rainy months of 50 mm, where potential evapotranspiration exceeds precipitation for 3 or more months of the year. Historical L and U se The original vegetation w as predominately grassland. Conversion to cultivated crops started in the 1920s, and more intensive cultivation occurred in the 1970s. During the four years before the initiation of the study, the study site was cultivated with maize alternated by chickp ea every other year. The land use history was variable previous to that time. L and U se : Data as C onsidered within the M odel The maize based cropping system was used with a combination of two treatments: no tillage (NT), all crop residues (AR) and conven tional tillage (CT), no crop residue (NR). The measured values were also for the maize rotation with chickpea, but for minimum tillage and not no tillage, so there may be some discrepancy in the comparison due to this difference in modeled and experimenta l factors. Noteworthy, however, is that the operational component in the no tillage option in DSSAT is that planting occurs in which there is some disturbance to the soil. The total SOC is estimated in units kg/ha, partitioned into each SOM pool ( also in units kg/ha ) The Basso et al (2011) approach was initialized using a reasonable total SOC value, based on historic SOC data for the area, to spin upreach a measured SOC value or the target total SOC value (Figure 5 1). T he simulation was run for 20 yea rs. This procedure was run iteratively until the target total SOC value was realized This

PAGE 109

109 procedure allows for a spin up period and establishment of total SOC as well as SOM pool values in the model A lso, a fter the target value was realized, the simulati on was continued for an additional time period of two years with the total SOC and SOM pool values estimated at the end of the run up period. For each observation, a range of 4 20 iterations were run to meet the target measured total SOC value. The Gijsm an initialization procedure used the initial values for total carbon (i.e., observed values) and an assumed 20 year land use history The land use history assumed was poorly managed, previously grassland. The Adiku (published in Porter et al., 2009) appr oach was initialized using the initial values (i.e., observed values) for total carbon assuming only the silt+clay content was known. Measu red SOC Fractions: Soil sampling and analysis Soil samples were collected in June 2009, before the start of the expe riment for soil characterization, at 0 30 cm soil depths. For each plot, three sub samples were randomly collected and mixed together to create one composite soil sample. All samples were air dried and sieved through a 2 mm sieve. The bulk density of th e soil samples was determined using core samples (diameter 50 mm) of known volume (100 cubic cm), wet weight minus oven dried weight. Soil moisture was estimated for 10 g of soil, wet weight minus oven dry soil. Samples were dried at 105C for 24 h. The soil pH was measured with a glass electrode using soil to water ratio of 1:2 and particle size by hydrometer method (Gee and Bauder, 1986). Modified Wet Sieving Procedure The soil was separated into three size classes: macro (2000 m), and a silt and clay procedure of Elliott (1986). 100 g of air dried soil was submerged in a 500 mL beaker

PAGE 110

110 of deionizer water and thus subjected to the disruptive forces of slaking for about 5 mi n before placing it on top of a 212 was made to use comparable energy input by moving the sieve up and down was colle cted in a hard plastic pan and allowed to oven dry at 65C and weighed. procedure was repeated. The average recovery mass percentage of soil fractions after the wet sieving proc edure was 97% of the initial soil mass. Microbial and Total Soil Organic Carbon Fraction The SOC concentration of soil size fractions and the whole soil was determined using the Walkley Black method (Nelson and Sommers, 1996). The SMBC was estimated usi ng chloroform fumigation (Anderson and Domsoh, 1978 and Brookes et al., 1985). Statistical C omparison of M easured and M odeled V alues An analysis of variance (ANOVA) was run on all data, each data point was assumed to be an independent observation. Compa rison of measured and modeled values was observed using x y plots of simulated vs. observed data, R squared values and root mean square error (RMSE) values. Mean FOM values were available, but replication wise FOM values were not available, so there was n o statistical comparison between values measured and simulated values. However, using inspection, simulated values were compared between each initialization approach. Results Values used to compare with the simulated data included measured data from Jun e 2009 and November 2010 (Table 5 1). The output values for the Basso et al

PAGE 111

111 (2011) initialization procedure are presented in Table 5 2 with the start up or initialization total carbon value used to reach the measured target total carbon values (also noted in the table) (Table 5 2). The output values using the Gijsman Porter approach are shown in the total carbon and each carbon pool value (Table 5 3). Similarly, the output values using the Adiku (published in Porter 2009) approach are shown in the total carbon and each carbon pool value (Table 5 4) Comparison between the Basso et al. (2011) initialization approach and observed values showed low correlation between observed and simulated SOM 2 values (R 2 =0.23; RMSE=55.87) and total carbon values (R 2 =0.24 ; RMSE=148.11), with no correlation in SOM 1 and SOM 3 values (Figure 5 1). Observed values ranged between 500 to 2000 kg/ha and simulated values ranged between 0 to 150 kg/ha for SOM 2. Comparison between the Gijsman Porter initialization approach and observed values showed a good correlation between observed and simulated values SOM 1 (R 2 =0.80; RMSE=4.99) and total carbon (R 2 =0.80; RMSE=109.98) (Figure 5 2). Observed values ranged between 1000 to 1500 kg/ha and simulated values ranged between 65 to 65 kg/hafor SOM 1. Comparison between Adiku (published in Porter et al., 2009) initialization procedure and observed values showed good correlation between the SOM 3 (R 2 =0.47; RMSE=0.98) and total carbon (R 2 =0.64; RMSE=140.35), and low correlation for S OM 1 (R 2 =0.23; RMSE=14.18) and SOM 2 (R 2 =0.10; RMSE=208.97) (Figure 5 3). Observed values ranged between 1500 to 1000 kg/ha and simulated values ranged between 15.5 to 19.6 kg/ha. A comparison between observed and simulated values for each initializati on approach by treatment showed no or low correlation for each appraoch (Figure 5 4).

PAGE 112

112 The Basso et al. (2011) approach best estimated the SOM 2 pool. I hypothesize that the disturbance and or distribution of FOM into the SOM pools is in part accounted for using this approach, even when limited or no FOM information is available. In other words, given the appropriate season and year, reference values for FOM inputs for selected crops within the area can be used to implement the model. When the iterative a pproach is used to reach the measured target total SOC value, it will also apply the FOM to the soil system. This kindof a spin up to initialize the SOM pools and FOM values based on the measured total SOC value may reduce uncertaintly within the simulate d values. The good correlation between measured and simulated SOM 1 values using the Gijsman et al. (2002) approach (R 2 =0.80; RMSE=4.99) may indicate that the ratios assumed for the SOM 1 pool in the DSSAT model (i.e., SOM 1 equivalent to ~5 % of the tota l C) allows the model to account for 80% of the variability of the active C in simulated C dynamics. The correlation between measured and simulated SOM 3 values using the Adiku (published in Porter et al., 2009) approach (R 2 =0.47; RMSE=0.98). suggests tha t this approach best estimated the SOM 3 pool, however some improvements to this approach may further reduce the uncertainty of modeled SOM values. Discussion This study used three different methods to estimate initial soil C pools for use in the DSSAT Ce ntury model and compared those pools with those measured using soil samples from the field and a fraction method. There was considerable variability in the field sample measurements of total soil C (in 2009 and 2010), and in the fractions obtained from the laboratory procedure. Calculations of changes in soil C during a 16 month time period were made by subtracting soil C measured in 2009 from

PAGE 113

113 measurements in the same plots 16 months later. These changes varied between about + 2,500 to 1,500 kg/ha among t he different replications. Simulated changes in soil C during this same time period were mostly negative, meaning that the model predicted net losses in total soil C during that 16 month time period, but the magnitude of the simulated C losses varied consi derably among the different model based procedures that were used (varying between about + 200 to 200, 300 to 800, and 1,300 to 1,900 kg/ha for the Basso, Gijsman Porter, and Adiku procedures, respectively. This variation among procedures highlights the initialization problem discussed earlier in the chapter. The fact that the magnitudes of these simulated changes were generally lower in magnitude than the soil C changes that were observed also highlights the problem of reliable initialization of soi l C pools in the model Comparing the C pools determined from the three model based procedures indicated that there are significant correlations between fractionation pool sizes and those estimated by model procedures. For example, the Basso procedure showe d the highest correlation between the >53 m size fraction and SOM 2 C and the Adiku method was highest correlated with the <53 m size fraction that was assumed to be associated with SOM3. However, these relationships are only indicators that fractionation methods may be correlated with pools estimated using model based procedures. The relationships should not be used without additional research. Further study and assessment of, and comparison between observed and simulated SOC pools throughout a longer per iod of time are needed to evaluate the adequacy of mode led C pool predictions based on initialization using fractionation methods Further comparison of methods to measure SOC may also be required for comparison between

PAGE 114

114 measured and modeled pools. Furthe r refinement of soil C models is desirable such that the models make use of observable pools instead of conceptual pools based strictly on decomposition rate time constants.

PAGE 115

115 Table 5 1 Measured data for simulations, 2009 10, ICRISAT, India Rep Treatment Fraction June'09 [mg/g] June 2009 [kg/ha] Nov10 [mg/g] Nov2010 [kg/ha] R1 CTNR 1 TotC 3.80 14820 3.68 14339 R1 MTAR TotC 3.70 14430 4.34 16910 R2 MTAR TotC 3.60 14040 4.34 16924 R2 CTNR TotC 3.70 14430 3.38 13165 R3 MTAR TotC 3.80 14820 4.25 16575 R3 CTNR TotC 3.40 13260 3.79 14786 R1 CTNR 1 2.10 8185 1.91 7434 R1 CTNR 2 0.51 1984 0.99 3848 R1 CTNR 3 0.84 3260 0.79 3070 R2 CTNR 1 1.99 7777 1.88 7342 R2 CTNR 2 0.52 2031 0.72 2811 R2 CTNR 3 1.14 4465 0.77 3012 R3 CTNR 1 1.76 6872 2.04 7953 R3 CTN R 2 0.44 1726 0.91 3561 R3 CTNR 3 0.85 3308 0.84 3273 R1 MTAR 1 2.15 8386 2.50 9754 R1 MTAR 2 0.82 3206 0.89 3466 R1 MTAR 3 0.72 2813 0.95 3690 R2 MTAR 1 2.30 8969 2.49 9729 R2 MTAR 2 0.61 2361 1.06 4116 R2 MTAR 3 0.55 2163 0.80 3120 R3 MTAR 1 2.34 9125 2.44 9533 R3 MTAR 2 0.75 2909 0.95 3709 R3 MTAR 3 0.67 2602 0.85 3333 1 CT=conventional tillage, MT=minimum tillage, NR=no residue, AR=all residue, Rep=replication where R1=replication 1, etc.

PAGE 116

116 Table 5 2 Output data for simulations for Ba sso et al. (2011) initialization procedure, 2009 10, ICRISAT, India rep Treatment fraction i totC (conc) target totC (conc) i simulated (kg/ha) totC (conc) simulated (kg/ha) @ 2 yrs R1 CTNR 1 TotC 0.47 0.38 14285 0.375 14097 R2 CTNR TotC 0.46 0.37 1400 0 0.367 13817 R3 CTNR TotC 0.43 0.346 13148 0.344 12977 R1 CTNR 1 121 103 R1 CTNR 2 2175 2053 R1 CTNR 3 11916 11880 R1 CTNR FOM 74 62 R2 CTNR 1 120 102 R2 CTNR 2 2145 2026 R2 CTNR 3 11663 11628 R2 CTNR FOM 74 62 R3 CTNR 1 116 99 R3 CTNR 2 2053 1944 R3 CTNR 3 10905 10873 R3 CTNR FOM 74 62 R1 MTAR TotC 0.44 0.343 13564 0.343 13373 R2 MTAR TotC 0.45 0.36 13848 0.352 14032 R3 MTAR TotC 0.48 0.384 14697 0.38 14490 R1 MTAR 1 121 103 R1 MTAR 2 2218 2090 R1 MTAR 3 11151 11119 R1 MTAR FOM 74 62 R2 MTAR 1 122 122 R2 MTAR 2 2249 2262 R2 MTAR 3 11404 11577 R2 MTAR FOM 72 71 R3 MTAR 1 126 107 R3 MTAR 2 2338 2197 R3 MTAR 3 12161 12125 R3 MTAR FOM 72 60 1 CT=conventiona l tillage, MT=minimum tillage, NR=no residue, AR=all residue, Rep=replication where R1=replication 1, etc.

PAGE 117

117 Table 5 3 Output data for simulations for Gijsman Porter ( Gijsman et al. 2002 ; Porter et al., 2009 ) initialization procedure, 2009 10, ICRISAT, India Rep Treatment fraction i totC (conc) i (kg/ha) totC (conc) simulated (kg/ha) @ 2 yrs R1 CTNR 1 TotC 0.38 14820 0.358 140440 R2 CTNR TotC 0.37 14430 0.349 13684 R3 CTNR TotC 0.34 13260 0.321 12596 R1 CTNR 1 247 156 R1 CTNR 2 4599 3896 R1 CTNR 3 9979 9930 R1 CTNR FOM 0 62 R2 CTNR 1 236 154 R2 CTNR 2 4478 3799 R2 CTNR 3 9716 9669 R2 CTNR FOM 0 62 R3 CTNR 1 217 144 R3 CTNR 2 4115 3504 R3 CTNR 3 8928 8885 R3 CTNR FOM 0 62 R1 MTAR TotC 0.34 13260 0.321 12968 R2 MT AR TotC 0.35 13650 0.33 13330 R3 MTAR TotC 0.38 14820 0.358 14411 R1 MTAR 1 217 156 R1 MTAR 2 4115 3866 R1 MTAR 3 8928 8884 R1 MTAR FOM 0 62 R2 MTAR 1 223 159 R2 MTAR 2 4236 3964 R2 MTAR 3 9191 9146 R2 MTAR FOM 0 62 R3 MTAR 1 242 167 R3 MTAR 2 4599 4253 R3 MTAR 3 9979 9929 R3 MTAR FOM 0 62 1 CT=conventional tillage, MT=minimum tillage, NR=no residue, AR=all residue, Rep=replication where R1=replication 1, etc.

PAGE 118

118 Table 5 4 Output data for simulations for A diku (in Porter et al., 2009) initialization procedure, 2009 10, ICRISAT, India Rep treatment fraction i totC (conc) i (kg/ha) totC (conc) simulated (kg/ha) @ 2 yrs R1 CTNR 1 TotC 0.38 14820 0.32 12978 R2 CTNR TotC 0.37 14430 0.312 12643 R3 CTNR TotC 0.34 13260 0.287 11639 R1 CTNR 1 489 259 R1 CTNR 2 9292 7596 R1 CTNR 3 5039 5056 R1 CTNR FOM 0 67 R2 CTNR 1 476 253 R2 CTNR 2 9048 7400 R2 CTNR 3 4906 4923 R2 CTNR FOM 0 66 R3 CTNR 1 438 236 R3 CTNR 2 8314 6813 R3 CTNR 3 450 8 4524 R3 CTNR FOM 0 66 R1 MTAR TotC 0.38 14820 0.318 13305 R2 MTAR TotC 0.37 14430 0.31 12972 R3 MTAR TotC 0.34 13260 0.285 11974 R1 MTAR 1 489 269 R1 MTAR 2 9292 7911 R1 MTAR 3 5039 5058 R1 MTAR FOM 0 85 R2 MTAR 1 476 264 R2 MTAR 2 9048 7717 R2 MTAR 3 4906 4925 R2 MTAR FOM 0 67 R3 MTAR 1 438 0.285 247 R3 MTAR 2 8314 7136 R3 MTAR 3 4508 4526 R3 MTAR FOM 0 66 1 CT=conventional tillage, MT=minimum tillage, NR=no residue, AR=all residue, Rep=replication where R1 =replication 1, etc.

PAGE 119

119 Figure 5 1 Basso et al. (2011) iterative initialization approach, DSSAT, 2012. The star indicates the measured target total C point.

PAGE 120

120 Figure 5 2. Comparison between Basso et al. (2011) and observed values ( n=6): (a) SOM 1 (R 2 =0.05; RMSE=8.03), (b) SOM 2 (R 2 =0.23; RMSE=55.87), (c) SOM 3 (R 2 =0.19; RMSE=85.13), (d) totC (R 2 =0.24; RMSE=148.11). Units: kg/ha. (a) (b) (c ) (d)

PAGE 121

121 Figure 5 3. Comparison between Gijsman Porter approach and observed values (n=6): (a) SOM 1 (R 2 =0.80; RMSE=4.99), (b) SOM 2 (R 2 =0.17; RMSE=213.43), (c) SOM 3 (R 2 =0.02; RMSE=3.10), (d) totC (R 2 =0.80; RMSE=109.98). Units: kg/ha. (a) (b) (c) (d)

PAGE 122

122 Figure 5 4. Comparison between Adiku (published in Porter et al., 2009) and observed values (n =6): (a) SOM 1 (R 2 =0.23; RMSE=14.18), (b) SOM 2 (R 2 =0.10; RMSE=208.97), (c) SOM 3 (R 2 =0.47; RMSE=0.98), (d) totC (R 2 =0.64; RMSE=140.35). Units: kg/ha. (a) (b) (c) (d)

PAGE 123

123 Figure 5 5. Comparison between each treatment using each approach (n=12): Basso et al. (2011) (a) conventional tillage, no residue (R 2 =0.04; RMSE=70.56) and (b) minimum tillage, all residue (R 2 =0.02; RMSE=128.80); Gijsman et al. (2002) (c) conventional tillage, no residue (R 2 =0.14; RMSE=323.70) and (d) minimum tillage, all residue (R 2 =0.28; RMSE=124.60);and Adiku (published in Porter et al. (2009) (e) conventional tillage, no residue (R 2 =0.16; RMSE=802.78) and (f) minimum tillage, all residue (R 2 =0.32; RMSE=577.83).Units: kg/ha.Units: kg/ha. (a) (b) (c) (d) (e) (f)

PAGE 124

124 CHAPTER 6 SYN TH E SIS AND CONCLUSIONS This d issertation evaluates the semi arid tropical soils ability to store C under diversified conservation management practices Specifically, this research attempts to answer the question: How do diversified conservation management practices affect carbon bioge ochemistry in semiarid tropical soils? The first aim of this dissertation was to investigat e C accumulation and vertical distribution in diversified maize based cropping systems. This was accomplished by assessing C stored at different depths within the so il profile under diverse conservation agricultural practices (C hapters 2 and 3). The second aim was to investigate the impacts of conservation agricultural practices in C association with physical soil size fractions addressed in C hapter 3 The third ai m was to assess the spatial autocorrelation of C across a watershed. This was accomplished by s howing the use of autocorrelation of C and use of secondary soil property data to predict C values with a fixed accuracy and detection limit (C hapter 4) The imp lication o f the study presented in Chapter 4 is that increased accuracy of prediction of C will allow for more precise assessment of C storage at various scales (e.g., field scale and or watershed scale). Finally, the last stu dy investigat es and quantif ies measured and modeled C pools using data assessed in Chapters 2 and 3 C hapter 6 provides a synthesis of conclusions drawn from all the dissertation studies and a discussion on the implications of SOM management on C in SAT soils.

PAGE 125

125 Objective 1 In study 1 the hypothesis was that comparison between SOC and selected C fractions would consistently indicate the effects of diversified sorghum based cropping systems on SOC accumulation (1) at specific soil depths and (2) as a function of depth. The null hypothe sis was rejected based on the following findings : The nonlinearity and variability of SOC, 15N, and 13C within the upper 60 cm of the soil suggests that instability of C and disturbance due to cropping systems may limit the potential use of linear functio ns to assess C as a continuous function with depth within the soil profile. The effects of cropping systems showed distinct shifts in stable C and N isotope values most probably due to root vs. shoot derived organic material and SOM age. The conclusions of study 1 were ( 1) in cultivated soils the effects of the cropping system were not consistently indicated as a continuous function with depth and processes related to SOC accumulation and distr ibution within the soil profile, and ( 2) in uncultivated GL s oils, SOC and 13C and 15N consistently indicated SOC accumulation and distrib ution within the soil profile. Further, the comparison of cultivated and uncultivated stable C and N isotope depth profiles showed that in the uncultivated soils and in crop B wit h FYM, the lighter isotope fraction of C remained in the soil In addition h igher soil C fraction s values (SOC and MBC) and also imply possibly less decomposed SOM. Increased SOC storage in the cropping system is likely due to increased preservation of C3 C from degradation, suggesting that the bi annual supply of FYM can significantly increase the preser vation of old C in the humic pool ( Francioso et al., 2007 ), especially in light of the practice of removing all above ground plant material after harvest. Further investigation of soil and plant lignin content compared to stable isotopes may provide

PAGE 126

126 infor mation needed to understand the patterns in isotope values with depth within the soil profile. Based on the comparison of cultivated and uncultivated GL soils, we conclude that a diversified sorghum based cropping system with MT and FYM treatment contribu ted to increased SOC storage within the soil profile. Objective 2 In study 2 the stated hypothesis was that the C in the whole soil and C associated with size fractions would be sensitive and indicators of changes due to agricultural management factors. A range of methods w ere used to characterize the soil C in comparative mode, for purposes of further use and comparison of successful methods in broader contexts both within the current study area (ICRISAT, Hyderabad, India) as well in other regions. Th e null hypothesis was rejected on account of the following findings: Carbon increased slowly and only in the macro aggregate size class. T he variability of the MBC population suggests that there is a stable dynamic soil microbial population that may act to quickly breakdown incorporated residues. Further, characterization of C using POXC provided a measure of C in relation to the intermediate C pool and detect ed early changes in soil C due to agricultural management practices in the short term (16 month s). However, additional research is required to assess the POXC method in relation to physically fractionated C and throughout longer periods of time (e.g., 4 years). Finally, n o significant differences were observed in the C in the whole soil due to con servation management practices likely due to the short term timeframe of the study.

PAGE 127

127 Objective 3 In study 3 the stated hypotheses were 1) information about the pattern of spatial autocorrelation could be used to determine the sampling density required for achieving a given accuracy in the estimation of the mean concentration of soil organic carbon; and 2) regression kriging (RK) with secondary information, compared to ordinary kriging (OK) would reduce the estimated number of samples required to obtain a narrower confidence interval at a selected confidence level The null hypotheses were rejected on account of the following findings: T he effect of the sampling density was quantified using power functions computed for different detection limits, D. With this information, a researcher can determine the mean concentration of SOC at a desired confidence interval and related sampling density required to meet the study objectives. For example, if a researcher wishes to detect an SOC difference of 0.2 % with a confidence level of 0.05 then 35 samples are required using a 300 m grid. Covariate TKN, SM, and silt % decreased KV, resulting in increased prediction at a fixed confidence interval. However, the benefits of using secondary data TKN with respect to the cost of data collection and analysis are comparable to SOC. Soil moisture and silt % may provide secondary data at a lower cost and still slightly increase the predictive power of SOC. Objective 4 In study 4, the main objective was to investigate, and f urther, to quantify the uncertainty presented by both the measured and the modeled SOC pools. The hypothesis was that the q uantification of the uncertainty presented by both modeled and

PAGE 128

128 measured C pools would increase the current understanding of C dynamic s for purposes of decision support for SOM management. The null hypothesis was rejected on account of the following findings: I n conclusion, t he C pool ratios, i.e., the initial SOM 3 and SOM1 values, of the Gijsman et al. (2002) approach were similar to the observed values. The uncertainty presented by the modeled SOC pools using the Gijsman et al (2002) and Basso et al. (2009) approaches wa s primarily a result of the assumed SOM pool ratios for each respective land use. In conclusion, t he Basso et al. (2009) approach can be used to approximate total SOC and one or more SOC pool target values, however cannot be used to approximate all pools unless the observed pool ratios show a similar range of values for the SOC pool ratios. The linear regression app roach (Porter et al., 2009) showed low correlation between observed and predicted SOC3 pool values. Further study and assessment of, and comparison between observed and simulated SOC pools throughout a period of time would further establish the accuracy of modeled SOC3 and SOC1 pool predictions. Further comparison of methods to measure SOC may also be required for comparison between measured and modeled pools. Essentially, the measured SOC pools do not have the properties assumed by any of the models (e .g., SOC pool ratios). Assuming the measurements can be repeated and are representative of the total and pools, then the models require refinement, especially to account for total or pool values in such a diverse set of land uses in this experiment. Synth esis and F uture R esearch Overall, this dissertation answers the questions and objectives outlined in the introduction, and outlined above. The conclusions of these studies suggest:

PAGE 129

129 Manure and C3 plant s (e.g. legumes) contribute to C storage in the soil pr ofile of semi arid tropical Vertisols. Further, the c omparison of C stored only in 0 30 cm soil depth does not provide evidence of C storage potential due to the effects of management practices in Vertisols. Analysis of C stored at various depths and com parison between land uses is needed to extrapolate useful information on C storage. The dynamic C pools affected by management practices in semi arid tropical Vertisols are sensitive, and can be indicated using comparative methodological approaches to cha racterize physical, biological, and chemical C. For example, the comparison between stable isotopes, total organic carbon, and soil microbial biomass provided useful information on the patterns of C distribution within the soil profile that would have oth erwise not been available without the comparison of these data. The autocorrelation of C and use of s econdary information, in particular silt content and soil moisture can be used to statistically improve sampling design schemes for soil C assessments in s emi arid tropical Vertisols as shown in C hapter 4. In C hapter 5 the comparison of measured and modeled C pools provides indication of additional gaps in understanding on C dynamics in semi arid tropical soils. But this study also provided important infor mation on the potential for modeling C in soils managed using conservation agriculture practices. I believe the next logical steps for C storage research in semiarid tropical Vertiso ls managed with conservation agricultural practices is a focus on three m ain research objectives to (1) investigate the significance of FYM contribution to the preservation of old C in the humic pool (2) determine application rate of crop residues to surface soil coupled with below ground temperature and soil moisture and micr obial population

PAGE 130

130 response, and (3) investigate decomposition rates of whole soil and individual size fractions in ex situ study in relation to varying soil moisture and temperature conditions Continuing the investigations of specific processes in relatio n to the underlying response to conservation management practices is an effective way to gain more information regarding potential to store C in semiarid tropical soils.

PAGE 131

131 LIST OF REFERENCES Adiku, S.G.K., S. Narh, J.W. Jones, K.B. Laryea, G.N. Dowuona. 200 8. Short term effects of crop rotation, residue management, and soil water on carbon mineralization in a tropical cropping system. Plant Soil. 311:29 38. Alvarez, R. 2005. A review of nitrogen fertilizer and conservation tillage effects on soil organic ca rbon storage. Soil Use and Management. 21 (1): 1475 274. Anderson, J. P. E. and K.H. Domsch. 1978. A physiological method for the quantitative measurement of microbial biomass in soil. Soil Biology & Biochemistry. 10: 2 15 221. Angers, D.A., M.A. Bolinder, M.R. Carter, E.G. Gregorich, C.F. Drury, B.C. Liang, R.P. Voroney, R.R. Simard, R.G. Donald, R.P. Beyaert, and J. Martel. 1997. Impact of tillage practices on organic carbon and nitrogen storage in cool, humid soils of eastern Canada. Soil and Tillage Res earch. 41(3 4): 191 201. Basso, B., O. Gargiulo, K. Paustian, G. Robertson,C. Porter,P. Grace, and J. James 2011. Procedures for Initializing Soil Organic Carbon Pools in the DSSAT CENTURY Model for Agricultural Systems. Soil Sci.Soc.Am.J. 75(1): 69 78. Baker, J., T.E. Ochsner, R.T. Venterea, and T.J. Griffis. 2007. Tillage and soil carbon sequestration What do we really know? Agriculture, Ecosystems and Environment. 118: 1 5. Baldock, J.A. and J.O. Skjemstad. 2000. The role of the mineral matrix in prote cting natural organic materials against biological attack. Organic Geochemistry. 31: 697 710. Benner, R., M.L. Fogel, E.K. Sprague, and R.E. Hodson. 1987. Depletion of 13C in lignin and its implications for stable isotope studies. Nature. 327: 708 710. Bha ttacharyya, T., T. Srinivasarao, P. Chandran, S.K. R ay, D.K. Pal, M.V. Venugopalan, C. Mandal, and S.P. Wani. 2007. Changes in levels of carbon in soils over years of two important food production zones of India Current Science 93 ( 12 ): 1854 1863. BioMedware, Inc. 2011. SpaceStat User Manual version 2.2, 309 pages Blanco Canqui and H.; R. Lal. Corn Stover Removal for Expanded Uses Reduces Soil Fertility and Structural Stabi lity. Soil Science Society of America Journal. 73 ( 2) : 418 426 Boutton, T.W., S.R. Archer, A.J. Midwood, S.F. Zitzer, and R. Bol. 1996. Stable carbon isotope ratios of soil organic matter and their use as indicators of vegetation and climate change. In Bo utton, T.W. and S. Yamasaki (Eds.). Mass Spectrometry of Soils. Marcel Dekker Inc., New York, pp. 47 82.

PAGE 132

132 Bowling D.R., D.E. Pataki, and J.T. Randerson. 2008. Carbon isotopes in terrestrial ecosystem pools and CO2 fluxes. New Phytologist. 178 (1):24 40. B rookes, P.C., A. Landman, G. Pruden, and D.S. Jenkinson. 1985. Chloroform fumigation and the release of soil nitrogen: A rapid direct extraction method to measure microbial biomass nitrogen in soil. Soil Biology and Biochemistry. 17(6 ): 837 842. Brugnoli, E. and G.D. Farquhar. 2000. Photosynthetic fractionation of carbon isotopes. In: Leegood, S. and S. von Caemmerer (Eds.), Photosy nthesis: Physiology and Metabolism. Kluwer, Dordrecht, pp. 399 434. Burgess T.M., Webster R. 1980. Optimal interpolation and isarithmic mapping of soil properties. II. Block kriging. J. Soil Sci. 31: 333 341. Burgess, T.M., R. Webster, A.B. McBratney. 198 1. Optimal interpolation and isarithmic mapping of soil properties, IV. Sampling strategy J. Soil Sci. 32: 643 659. Burrough, P. A. 1991. Sampling designs for quantifying map unit composition, pp. 89 126. In: M. J. Mausbach & L. Wilding (eds.), Spatial var iabilities of soils and landforms. Special publication, n. 28, Madison, Soil Sci. Soc. of America. Cambardella, C.A. and E.T. Elliott.1992. Particulate soil organic matter changes across a grassland cultivation sequence. Soil Science Society of America Jou rnal. 56: 777 783. Chandran, P., S.K. Ray, S.L. Durge, P. Raja, A.M. Nimkar, T. Bhattacharyya1 and D.K. Pal. 2009. Scope of horticultural land use system in enhancing carbon sequestration in ferruginous soils of the semi arid tropics. Current Sci. 97: 1039 1046. Chivenge, P.P., H.K. Murwira, K.E. Giller, P. Mapfumo, and J. Six. 2007. Long term impact of reduced tillage and residue management on soil carbon stabilization: Implications for conservation agriculture on contrasting soils. Soil and Tillage Resear ch. 94 (2):328 337. Christopher, S., R. Lal, and U. Mishna. 2008. Regional study of no till effects on carbon sequestration in the midwestern U.S. Soil Sci. Soc. Am. J. 73:207 216. Culman, S. S.S. Snapp M.A. Freeman M.E. Schipanski J. Beniston R. Lal L.E. Drinkwater A.J. Franzluebbers J.D. Glover A.S. Grandy J. Lee J. Six J.E. Maul S.B. Mirksy J.T. Spargo and M.M. Wander. 2012. Permanganate Oxidizable Carbon Reflects a Processed Soil Fraction that is Sensitive to Management. Soil Biology and Biochemistry. SSSAJ. 76(2):494 504. Daly, C; D. Bachelet; J.M.Lenihan, R.P. Neilson, W. Parton, and D. Ojima. 2000. Dynamic simulation of tree grass interactions for global change studies. Ecological Applications. 10 ( 2 ): 449 469

PAGE 133

133 Denef K., D. Roobroeck, M .C.W. Manimel, P. Wadu, P. Lootens, and P. Boeck. 2009. Microbial community composition and rhizodeposit carbon assimilation in differently managed temperate grassland soils. Biology and Biochemistry. 41(1 ): 144 153. Dijkstra, P., C.M. LaViolette, J.S. Coyle, R.R. Doucett, E. Schwartz1, S.C. Hart, and B.A. Hungate. 2008. 15N enrichment as an integrator of the effects of C and N on m icrobial metabolism and ecosystem function. Ecological Letters. 11(4): 389 397. Duxbury JM. 2006. Develop practical methods to measure gains and losses of soil organic C over time in spatially variable soils in South Asia: development of soil C texture rel ationships. In: Soil Management Collaborative Research Support Program, Project Year 9 annual progress report. Tropical Plant & Soil Sciences Department, University of Hawaii. Edwards, A.P.and J.M. Bremner. 1967. Microaggregates in soils. J. Soil Sci. 18: 64 73. El Swaify, P. Pathak, T.J. Rego, and S.Singh. 1985. Soil Management for Optimized Productivity Under Rainfed Conditions in the Semi Arid Tropics. Advances in Soil Science. 1: 1 63. Elliot, E.T. 1986. Aggregate structure and carbon, nitrogen, and ph osphorous in native and cultivated soils. Soil Sci. Soc. Am. J. 50:627 633. Emerson, W.W. 1959. Stability of soil crumbs. Nature. 183: 538. FAO. 2010. What is Conservation Agriculture. Online http://www.fao. org/ag/ca/1a.html (retrieved on January, 01 2012). Feller, C and M.H. Beare. 1997. Physical control of soil organic matter dynamics in the tropics. Geoderma. 79 (1 4): 69 116. Ferreyra R.A., H.P Apeztegu R Sereno, and J.W Jones. 2002. Reduction of soil water spatial sampling density using scaled semiva riograms and simulated annealing. Geoderma. 110 ( 3 4 ): 265 289. Fotheringham A.S., C. Brunsdon, and M. Charlton. 2002. Geographically Weighted Regression: the analysis of spatially varying relationships, Chichester: Wiley. Fotheringham, A.S., C. Brundson, and M. Charlton. 2002. Geographically Weighted Regression: The Analysis of Spatially Varying Relationships West Sussex, England: Wiley & Sons. Francioso, O., P. Gioacchini, V. Tugnoli, D. Montecchio, and C. Ciavatta. 2007. Effect of Manure and Crop Re sidues Continuous Application on Humic C in a Cultivated Soil. Soil Ecology Research Developments. Nova Science Publishers, Inc.

PAGE 134

134 Franzluebbers,A.J. and M.A. Arshad. 1996. Soil organic matter pools with conventional and zero tillage in a cold, semiarid cli mate. Soil and Tillage Research. 39(1 2): 1 11. Fry, B. 2006. Stable Isotope Ecology ( 3 rd Ed.) Springer, New York. Fujita, K., K.G. Ofosu Budu, and S. Ogata. 1992. Biological nitrogen fixation in mixed legume cereal cropping systems Plant and Soil 141 (1 2): 155 175. Gee, G. W. and J. W. Bauder. 1986. Particle size Analysis. pp. 383 411. In A.L. Page (ed.). Methods of soil analysis, Part 1, Physical and mineralogical methods (2 ed), Agronomy. Gijsma n, A.J., G. Hoogenboom, W.J. Parton, and P.C. Kerridge. 2002. Modifying DSSAT crop modell for low input agricultural systems using a soil organic matter residue module from CENTURY. Agronomy Journal. 94:462 474. Goovaerts, P. 1997. Geostatistics for Natura l Resources Evaluation Oxford Univ. Press, NewYork. Goovaerts, P. 1998. Regional estimation of soil properties from local observations. In Proceedings of the Soil Quality Workshop, Alberta Agriculture Food, and rural development, Edmonton January 30 31, 1 998. 51 58. Goovaerts, P. 1998. Geostatistics for Natural Resources Evaluation Oxford Univ. Press, NewYork. Hati,K.M., A. Swarup, B. Mishra, M.C. Manna, R.H. Wanjari, K.G. Mandal, and A.K. Misra. 2008. Impact of long term application of fertilizer, manur e and lime under intensive cropping on physical properties and organic carbon content of an Alfisol. Geoderma. 148 (2): 173 179. Hauert, C. and H. Liniger. 2008. Assessing Soil Properties in No Till and Traditional Tillage Systems in Switzerland. In Godda rd, T. 2008. World Association of Soil and Water Conservation. No till farming systems Bangkok: World Association of Soil and Water Conservation. Hengl, T., G. Heuvelink, and A. Stein. 2004. A generic framework for spatial prediction of soil variables ba sed on regression kriging. Geoderma. 120: 75 93. Hobbie, E.A. and R.A. Werner. 2004. Intramolecular, compound specific, and bulk carbon isotope patterns in C3 and C4 plants: a review and synthesis. New Phytologist 161: 371 385. Jobbgy, E.G. and R.B. Jack son. 2000. The vertical distribution of soil organic carbon and its relation to climate and vegetation. Ecological Applications. 10:423 436.

PAGE 135

135 Kendall, C., E.M. Elliott, and S.D. Wankel. 2007. Tracing anthropogenic inputs of nitrogen to ecosystems. In: Miche ner R.H. and K. Lajtha (Eds) Stable isotopes in ecology and environmental science, (2 nd Edition) Blackwell Publishing. 375 449. Kjeldahl, J. 1883. A new method for the determination of nitrogen in organic matter. Zeitschreft fur Analytische Chemie. 22: 366 Lal, R. 2004. Agricultural activities and global carbon cycling. Nutrient Cycling in Agroecosystems 70 (2 ): 103 116 Lal, R. 2006. Enhancing crop yield in the developing countries through restoration of soil organic carbon pool in agricultural lands. Land. Degradatio n Development. 17: 197 209. Loeppert, R.H. and D.L. Suarez. 1996. Carbonate and gypsum. pp. 437 474. In D.L.Sparks (ed.) Methods of soil analysis. Part 3. Chemical methods. Soil Science Society of America, Madison, WI. Metherell, A.K., L.A. Harding, C.V. Cole, and W.J. Parton. 1993/ CENTURY manual: CENTURY soil organic matter model environment technical documentation. Agroecosystem Version 4.0. Tech. Rep. 4. USDA ARS. Great Plans Sys. Res.Unit, Fort Collins, CO. McBratney, J. and B. Pringle. 1999. Estimati ng Average and Proportional Variograms of Soil Properties and Their Potential Use in Precision Agriculture. Precision Agriculture 1 (2): 125 152. Nelson, E.W. and L.E. Sommers. 1996. Total Carbon, Organic Carbon, and Organic Matter. In Methods of Soil Analysis: Chemical Methods. Part 3. D.L Sparks, editor. Soil Sci. Soc. of Am., Madison WI. Oades, J.M. 1984. Soil organic matter and str uctural stability: mechanisms and implications for management. Plant Soil. 76:319 337. Oades, J.M. and A.G. Waters. 1991. Aggregate hierarchy in soils Australian Journal of Soil Research. 29(6): 815 828. Parton, W.J., C.V. Cole, J.W.B. Stewart, D.S. Oji ma, and D.S. Schimel. 1988. Simulating regional patterns of soil C,N,and P dynamics in the U.S. Central Grasslandsregion. Biogeochemistry. 5:109 131. Parton, W.J. and P.E. Rasmussen. 1994. Long term effects of crop management in wheat fallow: II. CENTURY m odel simulation. Soil Science Society of America Journal. 58:530 536.

PAGE 136

136 Pathak, P., Sahrawat, K.L. Rego, T.J. and Wani, S.P., 2005. Measurable biophysical indicators for impact assessment: changes in soil quality, In B. Shiferaw, H.A. Freeman and S.M. Swint on (Eds) Natural Resource Management in Agriculture: Methods for Assessing Economic and Environmental Impacts. Wallingford: CAB International. Paul, E., E. Veldkamp, and H. Flessa. 2008. Soil organic carbon in density fractions of tropical soils under for est pasture secondary forest land use changes. European Journal of Soil Science. 59: 359 371. Paustian, K., J. Six, E. T. Elliott and H. W. Hunt. 2000. Management Options for Reducing CO 2 Emissions from Agricultural Soils. Biogeochemistry 48(1): 147 163. Pawar, K.N., B.D. Biradar, J. Shamarao, and M.R. Ravikumar. 2005 Identification of germplasm sources for adaptation under receding soil moisture situations in postrainy sorghum. Agricultural Science Digest. 25 :56 58. Perfect, E., B.D. Kay, W.K.P van L oon, R.W. Sheard, and T. Pojasok. 1990a. Factors influence soil structural stability within a growing season. Soil Sci. Soc. Am. J. 54: 173 179. Perfect, E., B.D. Kay, W.K.P van Loon, R.W. Sheard, and T. Pojasok. 1990b. Rates of change in soil structural s tability under forages and corn. Soil Sci. Soc. Am. J. 54: 179 186. Plante, A.F., Y. Feng, and W.B. McGill. 2002. A modeling approach to quantifying soil macroaggregate dynamics. Can. J. Soil Sci. 82: 181 190. Porter, C.H., J. W. Jones, S. Adiku, A. J. Gij sman O. Gargiulo, and J. B. Naab. 2009. Modeling organic carbon and carbon mediated soil processes in DSSAT v4.5. Oper Res Int J. Post, W.M. and K.C. Kwon. 2000. Soil Carbon Sequestration and Land Use Change: Processes and Potential. Global Change Biolog y 6: 317 328. Ramesh, V. S.P. Wani, T.J. Rego, K.L. Sharma, T. Bhattacharyya, K.L. Sahrawat, K.V. Padmaja, R.D. Gangadhar, B. Venkateswarlu, M. Vanaja, M.C. Manna, K. Srinivas, and V. Maruthi 2007 Chemical characterization of selected benchmark spots f or C sequestration in the semi arid tropics, India Global Theme on Agroecosytems Report No. 32. International Crops Research Institute for the Semi Arid Tropics (ICRISAT); and Indian Council of Agricultural Research (ICAR), Andhra Pradesh, India; and New Delhi, India. Reddy, K.R. and R. D. DeLaune. 2008. Biogeochemistry of Wetlands: Science and Applications: Science and Applications. CRC Press, Baton Rough, Florida.

PAGE 137

137 Sahrawat, K.L., T. Bhattacharyya, S.P. Wani, P. Chandran, S.K. Ray, D.K. Pal, and K.V. P admaja. 2005. Long term lowland rice and arable cropping effects on carbon and nitrogen status of some semi arid tropical soils. Current Science. 89(12): 2159 2163. of 13C in leaf litter as related to feeding activity of soil invertebrates and microbial mineralisation. Soil Biology and Biochemistry. 32:1793 1797. Silveira, M.L., N. B. Comerford, K.R. Reddy, J. Prenger, W.F. DeBusk. 2010. Influence of military land uses on soil carbon dynamics in forest ecosystems of Georgia, USA. ECOLOGICAL INDICATORS. 10 (4): 905 909. Simbahan, G.C., A. Dobermann, P. Goovaerts, J. Ping and M.L. Hadd ix. 2006. Fine resolution mapping of soil organic carbon based on multivariate secondary data. Geoderma. 132: 471 489. Singh, P., P. Pathak, S.P. Wani K. Srinivas and L.S. Jangawad. 2007. Sustainability of Soybean based Cropping Systems on a Vertic Incept isol: 2. Simulated Yield Gaps, Soil Water Balance and Soil Organic Carbon. Indian Journal of Dryland Agriculture and Development. 22 (2):121 132. Six, J., E.T. Elliott, K. Paustian, and J.W. Doran. 1998. Aggregation and soil organic matter accumulation in cultivated and native grassland soils. Soil Sci. Soc. Am. J. 62: 1367 1377. Six, J., E.T. Elliott, and K. Paustian. 2000. Soil macroaggregate turnover and microaggregate formation: a mechanism for C sequestration under no tillage agriculture. Soil Biol. Bi ochem. 32: 2099 2103. Six, J., R.T. Conant, E.A. Paul, and K. Paustian. 2002a. Stabilization mechanisms of soil organic matter: implications for C saturation of soils. Plant Soil. 241: 155 176. Six, J., H. Bossuyt, S. Degryze, and K. Denef. 2004. A review : A history of research on the link between (micro)aggregates, soil biota, and soil organic matter dynamics. Soil & Tillage Research. 79: 7 31. Skjemstad, J.O., L.J. Janik, M.J. Head, and S.G. McClure. 1993. High energu untraviolet photo oxidization: a nov el technique for studying physically protected organic matter in clay and silt sized aggregates. J. Soil Science. 44:485 499. Smith, P., J.U. Smith, D.S. Powlson, W.B. McGill, J.R.M. Arah, O.G. Chertov, K. Coleman, U.Franko, S. Frolking, D.S. Jenkinson, L .S. Jensen, R.H. Kelly, H. Klein Gunnewiek, A.S.Komarov, C. Li, J.A.E. Molina, T. Mueller, W.J. Parton, J.H.M. Thornley, A.P.Whitmore. 1997. A comparison of the performance of nine soil organic matter models using datasets from seven long term experiments. Geoderma. 81:153 225.

PAGE 138

138 Sreedevi TK, B. Shiefaw and S.P. Wani. 2004. Adarsha watershed in Kothapally: understanding the drivers of higher impact. Global Theme on Agroecosystems Report No. 10, Patancheru 502 324, Andhra Pradesh, India: International Crops Re search Institute for the Semi Arid Tropic. Srinivasarao, C., K.P.R. Vittal, B. Venkateswarlu, S.P. Wani, K.L. Sahrawat, S. Marimuthu, and S. Kundu. 2009. Carbon Stocks in Different Soil Types under Diverse Rainfed Production Systems in Tropical India. Comm unication in Soil Science and Plant Analysis. 40: 2338 2356. Suman, A., M. Lal, A.K. Singh, and A. Gaur. 2006. Microbial Biomass Turnover in Indian Subtropical Soils under Different Sugarcane Intercropping Systems. Agronjnl. 98: 698 704. Tisdall, J.M. and J.M. Oades. 1982. Organic matter and water stable aggregates in soils. J. Soil Sci. 62: 141 163. Trumbore, S. 2000. Age of soil organic matter and soil respiration: Radiocarbon constraints on belowground C dynamics. Ecological Applications. 10(2): 399 411 Vineela, C, Wani S.P., Srinivasarao C., Padmaja B., and Vittal K.P.R. 2008. Microbial properties of soils as affected by cropping and nutrient management practices in several long term manorial experiments in the semi arid tropics of India. Applications in Soil Ecology. 40:165 173. Walker T.S., H.P. Bais, K.M. Halligan, F.R. Stermitz, J.M. Vivanco. 2003. Metabolic profiling of root exudates of Arabidopsis thaliana. Journal of Agricultural and Food Chemistry. 41:2548 2554. Wander, M.M., M.G. Bidart, and S Aref. 1998. Tillage impacts on depth distribution of total and particulate organic matter in three Illinois soils. Soil Sci. Soc. Am. J. 62:1704 1711. Wander, M. 2004. Soil Organic Matter Fractions and Their Relevance to Soil Formation In Magdoff, F. an d R. Weil (Eds.), Soil Organic Matter in Sustainable Agriculture, CRC Press, Washington, D.C. 67 102. Wang, Y. and Y.P. Hsieh. 2002. Uncertainties and novel prospects in the study of the soil carbon dynamics. Chemosphere. 49 (8 ):791 804. Wani, S.P., P. Pathak, L.S. Jangawad, H. Eswaran, and P. Singh. 2003. Improved management of Vertisols in the semi arid tropics for increased productivity and soil carbon sequestration. Soil Use and Management. 19: 217 222. Wani, S.P., K.L. Sahrawat, T.K. Sreedevi, T. Bhattacharyya, and Ch Sreenivasa Rao, 2007. Carbon sequestration in the semi arid tropics for improving livelihoods, In ternational Journal of Environmental Studies. 64(6): 719 727.

PAGE 139

139 Webster, R. and M.A. Oliver. 1990. Statistical methods in soil and land resource survey. Oxford: Oxford University Press. Weil, R.W., K.R. Islam., M. Stine, J.B. Gruver and S.E. Samson Liebig. 2003. Estimating active carbon for soil quality assessment: a simplified method for laboratory and field use. American J. Alt. Ag. 18 (1):3 17. West, T.O., and W.M. Post. 2002. Soil organic carbon sequestration rates by tillage and crop rotation: A globa l data analysis. Soil Sci. Soc. Am. J. 66:1930 1946. Wu W. D. Xie and H. Liu 2009. Spatial variability of soil heavy metals in the three gorges area: multivariate and geostatistical analyses Environmental Monitoring and Assessment. 157(1 4 ): 63 71 Wynn J.G. and M.I. Bird. 2007. C4 derived soil organic carbon decomposes faster than its C3 counterpart in mixed C3/C4 soils. Global Change Biology. 13(10): 2206 2217. Wynn, J. J.W. Harden, and T.L. Fries. 2006. Stable carbon iso tope depth profiles and soil organic carbon dynamics in the lower Mississippi Basin. Geoderma. 131(1 2): 89 109. Yoo,K, J. Ji, A. Aufdenkampe, J. Klaminder. 2011. Rates of soil mixing and associated carbon fluxes in a forest versus tilled agricultural fiel d: Implications for modeling the soil carbon cycle. Journal of Geophysical Research. 116.

PAGE 140

140 BIOGRAPHICAL SKETCH Kiara Sage Winans grew up learning subsistence farm operations, on her mother s 48 acre farm. The younger of two girls and only child on the fa rm after age 9, Kiara quickly learned the balance of hard work both at home and school. Kiara excelled in her studies and her attitude toward overcoming challenges enabled her to become a competent and well rounded student. Kiara graduated high school as a second year college student through dua l enrollment and decided to continue studies at the University of Florida in Gainesville, Florida. Kiara went on to pursue a Masters of Arts Degree in Community and Environment with a focus on Social Ecology at Ant ioch University, in Seattle, Washington. While at Antioch University, Kiara also worked with various governmental and non governmental organizations, notably the Government of Goa, India, the Wester n Ghats Environment Forum the National Geographic Societ y, and the World Wildlife Fund where she became aware of the environmental importance of diverse ecosystems, in particular in the Western Ghats of India. The work in social ecology and exposure to diverse ecosystems inspired Kiara to learn more about the principles and processes of ecosystem functions Without hesitation, Kiara began her graduate education in environmental engineering sciences. Through her m and projects related to both arid regions of Ma li, West Africa, as well as dynamic wetland systems, Kiara was keen on issues facing diverse and nutrient limited eco systems. After completing the m career in the Soil and Water Science Depart ment. During this time, Kiara dedicated herself to her work, lived and worked in Hyderabad, India for two years. International

PAGE 141

141 research and distant relationships to family and friends became challenging, but four years of determined effort resulted in th e completion of this dissertation.