Depth-Wise Distribution of Soil-Carbon Stock Under Shaded- Perennial Agroforestry Systems: Case Studies from India and Costa Rica

Home , Chikmagalur

DEPTH-WISE DISTRIBUTION OF SOIL-CARBON STOCK UNDER SHADED- PERENNIAL AGROFORESTRY SYSTEMS: CASE STUDIES FROM INDIA AND COSTA RICA

NILOVNA CHATTERJEE

A DISSERTATION PRESENTED TO THE GRADUATE SCHOOL OF THE UNIVERSITY OF FLORIDA IN PARTIAL FULFILLMENT OF THE REQUIREMENTS FOR THE DEGREE OF DOCTOR OF PHILOSOPHY

UNIVERSITY OF FLORIDA

2018

To my late grandfather R. K. Sen Sharma

ACKNOWLEDGEMENTS

At the outset, I am indebted to my parents, Biman and Gouri Chatterjee for their relentless support of my decision to endeavor on this career path. While the decision did not come without its consequences— moving 8,800 miles away from the familiar East to the foreign West only added to the challenge. In this spirit, I would not have been able to dream of an education abroad had it not been the grace and kindness of my advisor

Dr. P.K. Nair and the School of Forest Resources and Conservation who offered me a

Graduate Research Fellowship to pursue this program. Even though, I lacked formal training in forestry and soil science, Dr. P.K. Nair believed in my enthusiasm and abilities to pursue a doctoral degree in soil carbon sequestration. His excellent mentorship weaved in with critical comments, patience, and suggestions have helped me throughout my doctoral program and I consider him as the most influential mentor in my academic life.

I am thankful to Dr. Vimala Nair, who showed her concern for me through these four years. Dr. Vimala Nair facilitated the laboratory work in the Soil and Environmental

Chemistry Lab at the Soil and Water Sciences department. Immense gratitude towards

Dr. Stefan Gerber, a truly genial and compassionate individual in every sense, the go-to person for all things quantitative. He not only brought the best out of me in terms of research method, but also encouraged me every step of the way with his kindest words.

With gratitude, I acknowledge the indispensable contribution of Dr Timothy Martin, for his enthusiasm and his forever-welcoming attitude. I am grateful to Dr. Syam Viswanath for facilitating the field work studies in India and for his support over the last four years

This project was executed in conjunction with the Institute of Wood Science

Technology, India and Centro Agronómico Tropical de Investigación y Enseñanza

(CATIE), Costa Rica. I have been fortunate enough to come across several personnel who went above and beyond to help me conduct my field work in India and Costa Rica:

Mr. Philip Jacob provided logistics support, Dr. K.M. Nair, Parvathy Sivakumar helped me with soil analyses in India, Dr. Rheinhold Muschler, Dr. Elias De Melo Virginio Filho,

Patricia Leandro and Rebeca Madriz Diaz helped me facilitate the field work studies in

Costa Rica, Dr. Martin Noponen from the Rainforest Alliance provided previous dataset from the experimental site. I thank all the field assistants in India and Costa Rica who worked relentlessly with me to collect soil samples. My heartfelt gratitude goes out to

Dr. Willie Harris, Dr. George. A. O’Connor, Dr. Amy Abernethy, Dr. Jason Curtis for their help during my doctoral program.

This journey would have been very hard to negotiate without the company of kind colleagues and friends of which I don't have enough space to list them all, but here are some I would like to give a heartfelt shout-out to: Davis George Thomas (my closest confidant who always taught me to be wise and kind), Dr. Biswanath Dari, Dr. Debjani

Sihi, Natasha Chatterjee (my sister),Tony Varghese, Dr. Claudia Romero, Dr. Rafael

Tonucci, Saptarshi Chakraborty, Dr. Gregory Toth, Dr. Sumpam Tangjang, Felipe M.

Pinheiro, Neha Chitlangia, and Minyuan Tie for meaningful and frivolous conversations, and for the great times and research ideas shared. Special thanks to my colleagues Dr.

Anna Evangeline Normand, Milton Diaz and Yasmin Quintana for helping me with the paperwork required to obtain Costa Rican soil export permit.

Finally, I thank Dr. Thales A.P. West for his unwavering support and love. He always encouraged me to persevere despite hardship.

TABLE OF CONTENTS

page

ACKNOWLEDGEMENTS ...... 4

LIST OF TABLES ...... 9

LIST OF FIGURES ...... 11

ABSTRACT ...... 14

CHAPTER

1 GENERAL INTRODUCTION ...... 16

Objectives ...... 22 Dissertation Outline ...... 22

2 LITERATURE REVIEW ...... 23

Carbon Sequestration ...... 23 Carbon Sequestration Potential of Agroforestry Systems ...... 24 Available Literature on Soil C Sequestration in Agroforestry Systems ...... 26 Soil Carbon in Aggregates and its Stability ...... 28 Shaded Perennial Agroforestry Systems and Carbon Storage ...... 29 Management Practices in Shaded Perennial AFS and C Storage ...... 34 Selection of Species ...... 34 Inorganic vs. Organic Fertilizer: ...... 35 Plant Residue ...... 35 Synthesis ...... 36

3 META-ANALYSIS ...... 37

Introduction ...... 37 Materials and Methods...... 41 Agroforestry Systems and Agroecological Regions ...... 41 Soil-Depth Classes and SOC Stock Calculations ...... 43 Meta-Analytical Approach ...... 44 Meta-Regression, ANOVA and Linear Mixed Effect Model ...... 46 Results ...... 48 AFS vs. Agriculture ...... 49 AFS vs. Forest ...... 50 AFS vs. Pasture ...... 51 SOC Stocks under AFS with Varying Age of System ...... 51 Age group: 0–5 years ...... 51 Age group: 5–10 years ...... 52 Age group: 10–20 years ...... 52

Age group: >20 years ...... 53 Mixed Effect Models, Meta-regression and ANOVA ...... 54 Discussion ...... 54 SOC Stock Changes under Agroforestry Systems across Agroecological Regions ...... 55 AFS vs. Agriculture ...... 55 AFS vs. Forest ...... 56 AFS vs. Pasture ...... 58 AFS vs. Uncultivated Land ...... 59 SOC Stocks in Relation to Age of Trees in AFS Management Practices ...... 59 SOC Stock Changes and the Forestry – Agroforestry – Agriculture/Pasture Continuum ...... 61 Data Quality ...... 62 Methodological Challenges ...... 63 Conclusions ...... 64

4 DEPTH-WISE DISTRIBUTION OF SOIL-CARBON STOCK IN AGGREGATE- SIZED FRACTIONS UNDER SHADED-PERENNIAL AGROFORESTRY SYSTEMS IN KARNATAKA, INDIA ...... 77

Introduction ...... 77 Materials and Methods...... 80 Study Location ...... 80 Land-use Systems ...... 81 Soil Sampling ...... 82 Soil Preparation and Analysis ...... 83 Statistical Analyses ...... 85 Results ...... 86 Soil Organic Carbon Stock up to 1m Depth ...... 86 SOC Stock Distribution in Whole Soil under Various Land-Use Systems and Depth Classes ...... 86 Soil Organic Carbon in Macro-Sized Fraction (>250 μm) ...... 87 Soil Organic Carbon in Micro-Sized Fraction (250–53 μm) ...... 88 Soil Organic Carbon in Silt and Clay Fraction (<53 μm) ...... 88 Interaction Effects and Analysis of Variance (ANOVA) ...... 88 Discussion ...... 89 Land-Use System – Soil Depth Class – Aggregate-Sized Interactions in SOC Storage ...... 89 SOC Stocks in Whole Soil under Different Land-Use Systems up to 1m Depth ...... 90 Soil Organic Carbon in Aggregate-Size Fractions ...... 91 Macroaggregates (>250 μm) ...... 91 Microaggregates (250–53µm) ...... 93 Silt and clay fraction (<53 μm) ...... 94 SOC Distribution in Different Soil Aggregate Classes Studies ...... 97 Rhizodeposition in Shaded Perennial Systems ...... 98 Conclusions ...... 101

5 AGGREGATE-SIZE FRACTIONS AND SOIL-CARBON STOCKS UNDER ORGANIC AND CONVENTIONAL COFFEE AGROFORESTRY SYSTEMS IN COSTA RICA ...... 114

Introduction ...... 114 Materials and Methods...... 117 Study Location ...... 117 Land-Use Systems and Management Practices ...... 118 Soil Sampling ...... 119 Soil Preparation and Analysis ...... 120 Statistical Analyses ...... 122 Results ...... 123 Soil Organic Carbon Stock up to 1m Depth ...... 123 SOC Stock Distribution in Whole Soil under Various Land-Use Systems and Depth Classes ...... 124 Soil Organic Carbon in Macro-Sized Fraction (>250 μm) ...... 124 Soil Organic Carbon in Micro-Sized Fraction (250 μm–53 μm) ...... 125 Soil Organic Carbon in Silt and Clay Fraction (<53 μm) ...... 125 Carbon Sequestration Potential (CSP) ...... 126 Interaction Effects and Analysis of Variance (ANOVA) ...... 126 Discussion ...... 127 Land-Use System – Soil Depth Class – Aggregate Size Interactions in SOC Storage...... 127 SOC Stocks in Whole Soil under Different Land-Use Systems up to 1m Depth ...... 128 Soil Organic Carbon in Aggregate-Size Fractions ...... 131 Macroaggregates (>250 μm) ...... 131 Microaggregates (250–53µm) ...... 132 Silt and clay fraction (<53 μm) ...... 133 Rhizodeposition and Management under Shaded Perennial Systems ...... 135 Timber species vs. N2 fixing species ...... 135 Organic vs. Conventional management ...... 137 Conclusions ...... 139

6 SUMMARY AND CONCLUSION ...... 153

APPENDIX: ADDITIONAL ASSESSMENTS ...... 159

LIST OF REFERENCES ...... 204

BIOGRAPHICAL SKETCH ...... 224

LIST OF TABLES

Table page

3-1 Major agroforestry systems covered in the meta-analysis...... 66

3-2 Number of data points across different agroecological regions and agroforestry systems included in the meta-analysis on soil organic carbon stocks...... 67

3-3 Agroecological features and system characteristics of various agroforestry systems included in the meta-analysis ...... 68

3-4 Summary of the meta-analysis on percentage changes in soil organic carbon stock up to 40 cm depth in different agroecological regions around the world ... 69

3-5 Summary of the meta-analysis on percentage changes in soil organic carbon stock up to 60 cm depth n different agroecological regions around the world .... 70

3-6 Summary Results of Statistical Analyses: Mixed Effect Models, Meta- regression ANOVA and Tests for Symmetry ...... 71

4-1 Soil characteristics (bulk density, pH, and particle-size distribution) at different depths in five land-use systems in Koppa, Chikmagalur, Karnataka, India ...... 103

4-2 Depth-wise distribution of soil-fraction-sized classes under five land-use systems in Koppa, Karnataka, India ...... 104

4-3 Analysis of variance (ANOVA), (factor analysis) with interaction effects of Treatments, Depth and Fraction size on SOC stocks without individual effect of each site level ...... 105

4-4 Analysis of variance (ANOVA), (factor analysis) with interaction effects of Treatments, Depth and Fraction size on SOC stocks showing individual effect of each site level ...... 106

5-1 Variations within management practices in coffee AFS in Costa Rica across conventional and organic treatments...... 140

5-2 Soil characteristics (bulk density, pH, and particle-size distribution) at different depths in five land-use systems in Turrialba, Cartago, Costa Rica ..... 141

5-3 Depth-wise distribution of different soil-fraction-size classes under six land- use systems in Turrialba, Cartago, Costa Rica ...... 142

5-4 Analysis of variance (ANOVA), (factor analysis) without the specific level of site ...... 143

5-5 Analysis of variance (ANOVA), (factor analysis) showing individual effect of each site level ...... 144

5-6 Change in SOC stocks (2001-2017) and carbon sequestration potential (CSP) under various AFS up to a depth of 40cm ...... 145

A-1 Mixed Effects model, comparing effect size of SOC stock for land-use change from Forest to Agroforest (Control: Forest) ...... 198

A-2 Heterogeneity Estimator of SOC stocks for Forest to Agroforest conversion ... 198

A-3 ANOVA, comparing effect size of SOC stock for land-use change from Forest to Agroforest (Control: Forest) ...... 199

A-4 Rosenthal and Orwin’s test for publication bias (control: Forest) ...... 199

A-5 Mixed Effects model, comparing effect size of SOC stock for land-use change from Agriculture to Agroforest (Control: Forest) ...... 199

A-6 Heterogeneity Estimator of SOC stocks for Agriculture to Agroforest conversion ...... 200

A-7 ANOVA, comparing effect size of SOC stock for land-use change from Agriculture to Agroforest (Control: Agriculture) ...... 200

A-8 Rosenthal and Orwin’s test for publication bias (control: Forest) ...... 201

A-9 Mixed Effects model, comparing effect size of SOC stock for land-use change from Pasture to Agroforest (Control: Pasture) ...... 201

A-10 Heterogeneity Estimator of SOC stocks for Pasture to Agroforest conversion . 201

A-11 ANOVA, comparing effect size of SOC stock for land-use change from Pasture to Agroforest (Control: Pasture) ...... 202

A-12 Rosenthal and Orwin’s test for publication bias (control: Pasture)...... 202

A-13 Mixed Effects model, comparing effect size of SOC stock for land-use change from Uncultivated Land to Agroforest (Control: Uncultivated Land). .... 202

A-14 Heterogeneity Estimator of SOC stocks for Pasture to Agroforest conversion . 203

A-15 ANOVA, comparing effect size of SOC stock for land-use change from Pasture to Agroforest (Control: Uncultivated Land) ...... 203

A-16 Rosenthal and Orwin’s test for publication bias (control: Uncultivated Land) ... 203

LIST OF FIGURES

Figure page

3-1 Percentage changes in soil organic carbon stock (0–20cm) between agroforestry systems in the arid and semiarid (ASA) region ...... 72

3-2 Percentage changes in soil organic carbon stock (0–20cm) between agroforestry systems in the lowland humid tropics and subtropics (LHT) region ...... 73

3-3 Percentage changes in soil organic carbon stock (0–20cm) between agroforestry systems in the Mediterranean (MED) region...... 74

3-4 Percentage changes in soil organic carbon stock (0–20cm) between agroforestry systems in the temperate (TEM) region...... 75

3-5 Summary of results of a global meta-analysis showing percentage changes in soil organic carbon stock between Agroforestry systems (AFS) ...... 76

4-1 Location of the study in Koppa, Chikmagalur, Karnataka, India ...... 107

4-2 Plot selection, location: Devon Plantations, Koppa, Chikmagalur, Karnataka, India ...... 108

4-3 Total soil organic carbon (SOC) content in the whole soil up to 1 m depth in five different land-use systems in Koppa, Chikmagalur, Karnataka, India...... 109

4-4 Depth-wise mean soil organic carbon (SOC) stock in the whole soil up to 1 m depth in five different land-use systems in Koppa, Karnataka, India...... 110

4-5 Depth-wise mean soil organic carbon (SOC) stock in macroaggregates (>250µm) soil up to 1 m depth in five different land-use systems in Koppa, Karnataka, India...... 111

4-6 Depth-wise mean soil organic carbon (SOC) stock in microaggregates (250– 53 µm) soil up to 1 m depth in five different land-use systems in Koppa, Karnataka, India...... 112

4-7 Depth-wise mean soil organic carbon (SOC) in stock in silt + clay fraction (<53µm) soil up to 1 m depth in five different land-use systems in Koppa, Karnataka, India...... 113

5-1 Location of the study in Turrialba, Cartago province, Costa Rica ...... 146

5-2 Management practices selected for the study in Turrialba, Costa Rica ...... 146

5-3 Total soil organic carbon (SOC) stock in the whole soil up to 1 m depth in six different land-use systems in Turrialba, Costa Rica...... 147

5-4 Depth-wise mean soil organic carbon (SOC) stock in the whole soil up to 1 m depth in six different land-use systems in Turrialba, Costa Rica...... 148

5-5 Depth-wise mean soil organic carbon (SOC) stock in macroaggregates (>250µm) soil up to 1 m depth in six different land-use systems in Turrialba, Costa Rica...... 149

5-6 Depth-wise mean soil organic carbon stock (SOC) in microaggregates (250µm–53 µm) soil up to 1 m depth in six different land-use systems in Turrialba, Costa Rica...... 150

5-7 Depth-wise mean soil organic carbon (SOC) in stock in silt + clay fraction (<53 µm) soil up to 1 m depth in six different land-use systems in Turrialba, Costa Rica...... 151

5-8 Soil organic carbon stocks in whole soil under various AFS treatments up to a depth of 40cm for the years 2001 and 2017...... 152

A-1 PRISMA flow diagram detailing screening process of articles included in meta-analysis...... 159

A-2 Percentage changes in soil organic carbon stock (0–40cm) between agroforestry systems in various agroecological regions ...... 160

A-3 Percentage changes in soil organic carbon stock (0–60cm) between agroforestry systems in various agroecological regions ...... 164

A-4 Percentage changes in soil organic carbon stock (0–100cm) between agroforestry systems in various agroecological regions ...... 168

A-5 Percentage changes in soil organic carbon stock (0–200cm) between agroforestry systems in various agroecological regions ...... 172

A-6 Percentage changes in soil organic carbon stock (60–100cm) between agroforestry systems in various agroecological regions ...... 175

A-7 Percentage changes in soil organic carbon stock (0–20cm) between agroforestry systems of varying age in various agroecological regions ...... 178

A-8 Percentage changes in soil organic carbon stock (0–40cm) between agroforestry systems of varying age in various agroecological regions...... 182

A-9 Percentage changes in soil organic carbon stock (0–60cm) between agroforestry systems of varying age in various agroecological regions ...... 186

A-10 Percentage changes in soil organic carbon stock (0–100cm) between agroforestry systems of varying age in various agroecological regions ...... 190

A-11 Funnel Plot and Normal Q-Q plot of effect sizes of SOC stock changes for AFS vs. Agriculture...... 196

A-12 Funnel Plot and Normal Q-Q plot of effect sizes of SOC stock changes for AFS vs. Forest...... 197

A-13 Funnel Plot and Normal Q-Q plot of effect sizes of SOC stock changes for AFS vs. Pasture ...... 197

A-14 Funnel Plot and Normal Q-Q plot of effect sizes of SOC stock changes for AFS vs. Uncultivated Land...... 198

Abstract of Dissertation Presented to the Graduate School of the University of Florida in Partial Fulfillment of the Requirements for the Degree of Doctor of Philosophy

DEPTH-WISE DISTRIBUTION OF SOIL-CARBON STOCK UNDER SHADED- PERENNIAL AGROFORESTRY SYSTEMS: CASE STUDIES FROM INDIA AND COSTA RICA

Nilovna Chatterjee

August 2018

Chair: P.K. Ramachandran Nair Cochair: Vimala D. Nair Major: Forest Resources and Conservation

Storing carbon (C) in soil is one of the strategies accepted by the United Nations for mitigating atmospheric concentrations of carbon dioxide and other greenhouse gases that cause global warming. Agroforestry systems are believed to have high potential for storing C in soil compared to treeless agricultural systems. This hypothesis was tested for shaded perennial agroforestry systems in Koppa, Karnataka, India and

Turrialba, Cartago, Costa Rica. Growing shade-tolerant perennial crops under trees is an economically attractive land-use activity in the tropics; but the importance of these systems in facilitating ecosystem services, such as soil carbon sequestration, is seldom recognized. The major objective of the study was to estimate soil carbon stock in shaded perennial coffee and tea agroforestry systems in comparison with common land-use systems such as forest and homegarden in India and monoculture coffee in

Costa Rica. Soil samples collected from four depths (0 – 10, 10 – 30, 30 – 60, 60 – 100 cm) were fractionated to three fraction-size classes (250 – 2000 µm, 53 – 250 µm, <53

µm) and their total C content was determined. Within 1 m soil profile, the total C stock was higher under forest in both the study sites. Within the shaded perennial agroforestry

systems, the highest soil SOC stocks were in coffee under Grevillea shade (142 Mg C ha-1) in India and coffee under Terminalia shade (125 Mg C ha-1) in Costa Rica. In terms of C stock in three soil fraction-size classes, differences among land-use systems were most pronounced for macro-size (250 – 2000 µm), followed by micro-size (53 – 250 µm) and silt + clay (< 53 µm) fractions. Coffee grown under non-N2 fixing timber tree species significantly improved SOC stocks compared to that under N2 fixing shade trees. The study presented shows that SOC stocks under perennial AFS are higher than under agricultural systems (homegarden), and comparable to those under natural forest. That by itself is an important finding in the context of climate-change mitigation. However, since agroforestry systems are extremely site-specific, more investigations of this nature are needed from a variety of situations before broadly applicable policy recommendations can be made.

CHAPTER 1 GENERAL INTRODUCTION

Globally, there is a consensus on the rising concentrations of carbon dioxide

(CO2) and other greenhouse gases (GHG) in the atmosphere and their effect on climate change. Tropical deforestation has long been listed as one of the major causes of CO2 emissions, and several reports are available on the extent of such emissions (IPCC,

2014). Annual GHG emissions from agricultural production during 2000–2012 were

-1 estimated at 5.0–5.8 Pg CO2 equivalent yr while annual GHG flux from land use and

-1 land-use change activities accounted for 4.3–5.5 Pg CO2 equivalent yr (IPCC, 2014).

Baccini et al. (2012) estimated that tropical deforestation and resultant land-use

−1 changes contributed about 1.0 Pg C yr (about 15% of total anthropogenic CO2 emissions) during 2000–2010. Typical farming practices like land clearing, excessive use of fertilizers and other agrochemicals also make agricultural practices a significant contributor to greenhouse gas emissions to the atmosphere. Although the extent of anthropogenic contributions to global warming is not universally accepted, the projected value of 1.2 Pg C yr-1 (van der Werf et al., 2009) from deforestation and land degradation is a matter of serious concern.

Consequently there is an increasing demand for adopting land management practices to mitigate greenhouse gas emissions from agricultural and forest lands by enhancing soil carbon (C) sequestration (Lal et al., 2015). We have been facing one of the greatest challenges of the twenty-first century in meeting society’s growing food needs while simultaneously reducing agriculture’s environmental impacts (Foley et al.,

2011). There lies a dire need to establish techniques that promote C sequestration, thereby providing social, environmental, and economic benefits. Two key approaches

can be visualized: i) reducing anthropogenic emissions of CO2, ii) creating and promoting C sinks that are capable of securely storing atmospheric C in the soil. It is logical to assume that systems such as agroforestry consisting of multiple plant species associations on the same unit of land have high potential to act as a C sink and promote

C storage (Montagnini and Nair, 2004; Morgan et al., 2010). The tree component of agroforestry systems (AFS) act as soil C sinks (Lorenz and Lal, 2014; Nair and Nair,

2014) due to their high above- and belowground biomass, fast growth and productivity, and extensive root system. Higher aboveground biomass leads to higher soil organic carbon (SOC) accumulation via litterfall while higher belowground biomass enhances root activity and rhizodeposition (Asbjornsen et al., 2013; Dietzel et al., 2017). Trees with proliferating root systems, enhance rhizodeposition in soil even up to 1.5 m depth

(Pierret et al., 2016a). Additionally, root exudates from trees promote encapsulation of C within aggregates and reduces losses of C via mineralization (Chenu and Plante, 2006).

This ensures secure storage of C in soil (Šimanský and Bajcan, 2014). Thus, adopting tree-based land management practices such as agroforestry can promote “secure” storage of C even in deep soil. Furthermore, management practices such as agroforestry improve the amount of plant residue on soil surface, thus promoting soil aggregation, aggregate stability and SOC within aggregates (Chen et al., 2017; Elliott,

1986; Six et al., 2002). Improved soil aggregation can increase SOC storage by preventing losses from mineralization and enhancing C sequestration.

Soil is an important part of the biosphere and forms an effective C sink. It has a higher potential to store C compared to vegetation and atmosphere (Lal et al., 2015).

The soil C pool stands at 2300 Pg which is thrice the size of the atmospheric C pool of

770 Pg and almost four times the size of biotic pools (610 Pg) (Lal, 2005). Global estimates of C stocks indicated that, 1.1 to 2.2 Pg C can be removed from the atmosphere within a time frame of 50 years by implementing AFS (Albrecht and Kandji,

2003). Research also indicates that by adding trees in grassland or pasture systems, the SOC content can be increased considerably (Haile et al., 2010; David Scott Howlett et al., 2011; Moreira et al., 2017; Yelenik et al., 2004). Considering these beneficial effects of AFS on SOC stocks, these systems merit recognition and serious consideration.

Globally, an estimated 1600 Mha of land are managed under various agroforestry systems (Nair, 2012). In general, AFS on fertile humid sites have higher C sequestration rates than those on arid, semiarid, and degraded sites, and tropical AFS have higher C sequestration rates than temperate AFS (Feliciano et al., 2018).

Wright et al. (2001) estimated that the goal of assimilating 3.3 Pg C year-1 would require 670−760 Mha area of improved maize (Zea mays) cultivation, whereas this goal could be achieved by adoption of 460 Mha of agroforestry in the temperate region. It has even been suggested that agroforestry is the only system that could realistically be implemented to mitigate the atmospheric CO2 through terrestrial C sequestration

(Albrecht and Kandji, 2003).

Depending on the land-use type and rooting depth of vegetation, considerable amount of C can be accumulated at lower depths of soil (Dietzel et al., 2017; Fontaine et al., 2007; Pierret et al., 2016). The C at lower soil depths, being better protected from physical disturbances lead to longer mean residence time (MRT) in soil (Six et al.,

1998). However, most of the soil C studies investigate only top 30 cm or less, underestimating the potential of subsoil C sequestration.

The extent of C retention in soils depends on, among other things, the nature of soil aggregation (Carter and Gregorich, 2010; Haile et al., 2010a; Saha et al., 2010;

Takimoto et al., 2009). Carbon storage in soils can be short-term storage as in macroaggregates (>250 µm diameter), intermediate storage as in microaggregates

(250–53 µm diameter), long-term storage as in silt-and-clay size fraction (<53 µm).

Some studies have been initiated in AFS to investigate the allocation of C within soil aggregates (Chen et al., 2017; Gama-Rodrigues et al., 2010; Monroe et al., 2016;

Saha et al., 2010; Tonucci et al., 2011). However, all C sequestration studies that have been conducted on AFS so far have been site-specific; i.e., each one focused on the specific site where the study was conducted. Although there have been similarities between studies in terms of the type of systems and type of tree and understory components, each study has been unique in its system specificity. This is particularly so in the case of shaded perennial AFS. The number of studies on such systems has been very few, and each study has been specific to its location.

Shaded perennial agroforestry systems are managed, vertically stratified plant associations involving shade-tolerant and/or shade-adapted crops under tall growing trees (Nair, 2017). The overstory species of these combinations include those that are either deliberately planted as shade trees as in plantations of cacao (Theobroma cacao), coffee (Coffea spp.), and tea (Camellia sinensis). These systems contain mainly two woody species. The shade trees reduce the stress of cacao, coffee and tea by ameliorating adverse climatic conditions and nutritional imbalances (Tscharntke et al.,

2011). Beer et al. (1998) stated that shade trees buffer high and low temperature extremes by as much as 5°C and can produce up to 14 Mg ha–1yr–1 of litterfall and pruning residues, containing up to 340 kg N ha–1yr–1. These systems are considered as an important category of agroforestry system.

Coffee, a native of Ethiopia, and Tea, a native of China, are understory species in their habitat, and are cultivated almost exclusively under the shade of a variety of trees. While cacao agroforestry systems are mostly grown in lowland humid tropics (up to 500 m above sea level), coffee and tea are grown on medium elevations of 500 –

1500 m above sea level. These shaded perennial AFS render ecosystem services with high value for supporting human livelihoods including carbon storage, biodiversity conservation, and maintenance of soil fertility (Nair, 2017). With their plant diversity and spatial arrangements, these systems often mimic the structure and composition of forest

(Hombegowda et al., 2015). Among many different AFS focused on agronomical production, shaded perennial systems can be considered as systems with highest effectiveness in sequestering carbon (Ehrenbergerová et al., 2015). Assemblages of dense floristic species in shaded perennials enhance optimum resource utilization compared to mono-species systems (Nair, 2017) and may aid in promoting higher net primary productivity and C sequestration (Kirby and Potvin, 2007). Because of these rather unique characteristics, it is likely that a positive relationship exists between shade trees, shade tolerant species and C sequestration under these systems. As in the case of all land-use systems, the extent of C sequestration under shaded perennial AFS could be site specific based on the overall effect of the different interactions between shade trees and the understory shade-tolerant species. These interactions are

dependent upon site conditions (soil/climate), component selection

(species/varieties/provenances), previous land use type, belowground and aboveground characteristics of the over-story and understory species, and management practices

(Haggar et al., 2011; Soto-Pinto and Aguirre-Dávila, 2014). Little research has been done on these aspects of agroforestry in general, and no study has been reported comparing the extent of soil C sequestration under shaded perennial systems of the same type from two different continents.

The southern state of Karnataka in India produces about 70% of India’s coffee as per 2016–2017 statistics generated by the Coffee Board of India. Specifically, the district Chikamagalur district in Karnataka produces 71010 Mg of Coffee (Coffee Board of India). Coffee in Karnataka is mostly grown under the shade of tall trees. This coffee growing belt in southern India has a long-standing history of diverse, multistrata agroforestry based land-use practices which are even known to mimic forest like ecosystems (Hombegowda et al., 2015). Thus, coffee growing practices in Karnataka offer a good opportunity to study the trends of shaded perennial agroforests and other multistrata systems. In terms of management practices and environmental consciousness, the state of Karnataka is comparable to the coffee growing Central

American country of Costa Rica. In Costa Rica, coffee is usually grown under shade trees although the annual production is much lower than in Karnataka at 8082 Mg as per reports generated in the year 2017 (USDA-FAS, 2017). The idea of comparing the soil organic carbon stocks in two different countries guided by similar land management principles for growing coffee seemed important.

Objectives

1. To present a quantitative “state of the art” on relative soil C contribution from trees in agroforestry systems in comparison to other land-use systems (Agriculture, Forestry, Pasture, or Uncultivated Land) in four major agroecological regions (arid and semiarid, ASA; lowland humid tropics, LHT; Mediterranean, MED; and temperate, TEM) around the world through a rigorous meta-analysis of available literature sources.

2. To investigate the soil C storage in whole soil as well as fractionated size-classes of soil in a gradient of tree-based land used systems, including forests, shaded perennial coffee and tea agroforestry systems, and a farmer’s homegarden, up to one-meter depth in Karnataka, India.

3. To investigate the soil C storage in whole soil as well as fractionated size-classes of soil in a gradient of tree-based land use systems including forests, shaded perennial coffee agroforests under varying management practices (organic vs. conventional), in comparison with shade devoid, monoculture coffee in Turrialba, Costa Rica.

4. To compare the results on soil carbon sequestration under shaded coffee systems in the two continents mentioned above.

Dissertation Outline

The dissertation contains six chapters including this one (Chapter 1, General

Introduction), Chapter 2 presents a literature review, highlighting the importance of agroforestry systems in SOC storage and carbon sequestration potential. Chapter 3 provides a quantitative analysis in the form of statistical meta-analysis of available literature across the globe on soil organic carbon stock improvements under AFS.

Chapter 4 and Chapter 5 present the methodologies, land-use practices selected and results of SOC investigations under shaded perennial agroforestry systems in

Karnataka, India and Turrialba, Costa Rica respectively. Finally, Chapter 6 compares the results from shaded coffee systems in the two continents and provides the summary and conclusions with suggestions for future research along these lines and potential management implications.

CHAPTER 2 LITERATURE REVIEW

Carbon sequestration and land-use practices that aid in storing carbon have been a hotly debated research topic of this century. Shaded perennial agroforestry systems (AFS) which involves growing of special commodity crops like cacao, coffee and tea under the shade of tall trees are widely distributed in the tropics but has not been widely researched, especially regarding their role in soil carbon sequestration.

This chapter presents a review of relevant literature on soil C sequestration in agroforestry. In order to evaluate and quantify the improvement in SOC stocks under

AFS, a rigorous, quantitative assessment (meta-analysis) of available literature is presented in Chapter 3. This chapter gives a qualitative overview of carbon sequestration in general, soil carbon sequestration cascading into relationship between carbon sequestration and AFS. Additionally, literature on shaded perennial systems and their role on carbon sequestration highlighting the scope and limitations are also presented.

Carbon Sequestration

Adoption of climate-change mitigation strategies by sequestering C has been acknowledged and extensively researched in the recent past. The United Nations

Framework Convention on Climate Change (UNFCCC) during the recently concluded

Paris Agreement in December 2015 recognized that climate change represents an urgent and potentially irreversible threat to human societies and the planet. It also recognized that reductions in global emissions will be required in order to achieve the ultimate objective of the Convention. The Convention also emphasized the urgency in addressing climate change mitigation strategies (Paris Agreement 2015). The universal

agreement approved by 195 countries seek to keep global temperature rise of this century well below 2°C and to drive efforts to limit the temperature increase even further to 1.5 °C above pre-industrial levels. Realizing the threat of global warming, the

UNFCCC suggested two key activities to mitigate this threat:

 To reduce the anthropogenic emission of CO2  To create or promote carbon (C) sinks in the biosphere.

The second activity proposes storing atmospheric C in the soil, and in that context, land- use systems such as agroforestry have considerable importance. Due to the high carbon sequestration potential, agroforestry has become an attractive choice for such projects under the umbrella of afforestation and reforestation.

Carbon Sequestration Potential of Agroforestry Systems

Agroforestry involves the deliberate growing of trees and shrubs on the same unit of land as agricultural crops or animals, either in some form of spatial mixture or temporal sequence (Nair, 1993, 2011). Agroforestry has been suggested as one of the most efficient land-management systems for mitigating atmospheric CO2 (Albrecht and

Kandji, 2003; Dixon, 1995). These systems are believed to have a higher potential to sequester C than monoculture systems (Kirby and Potvin, 2007; Sharrow and Ismail,

2004); aboveground as well as belowground (Haile et al., 2008; Montagnini and Nair,

2004). The large volume of aboveground biomass and deep root systems of trees in

AFS have received increased attention for climate change adaption and mitigation (Nair,

2012). Further, between 30 and 300 Mg C ha−1 may be stored in agroforestry soils up to

1 m depth (Nair et al. 2010). It has been estimated that in the tropics one hectare of agroforestry land-use practices may offset 5–20 hectares of deforestation (Dixon, 1995).

Global estimates for the C sequestration potential of agroforestry systems over a 50-

year period range between 1.1 and 2.2 Pg C yr−1 but the estimates of global land area under agroforestry systems are highly uncertain (Dixon, 1995). This gap in estimates have been filled over the years with more intensive research. Some reports suggested that agroforestry is estimated to be practiced on 1000–1023 Mha globally and to sequester from 30 to 322 Pg C yr−1 (Jose and Bardhan, 2012). An additional 12,000 Mg of C per year could be sequestered, increasing to 17,000 Mg C per year by 2040, simply through improving tree management practices (Negash and Starr, 2015). If the current 630 Mha of unproductive cropland and grassland were converted to agroforestry, a further 586,000 Mg C yr−1 could be added by 2040 (Smith et al., 2008).

In general, agroforestry systems on fertile, low humid tropics have higher C sequestration rates than those on arid, semiarid, and degraded sites (Feliciano et al.,

2018). The tropical agroforestry systems have higher vegetation C sequestration rates than temperate agroforestry systems. The soil organic carbon stocks in agroforestry systems may persist for millennia indicating that terrestrial sequestration for climate change mitigation occurs particularly by reduction in net SOC losses followed by accumulation of the most stable SOC pool (Lorenz and Lal, 2014; Mbow et al., 2014;

Schmidt et al., 2011). However, there is lack of consensus over the period for which C has to be immobilized in soil before it is considered to be sequestered as a useful contribution to climate change mitigation (Mackey et al., 2013).

Higher SOC pools in agroforestry systems can be particularly achieved by increasing the amount of biomass C returned to the soil, by strengthening soil organic matter (SOM) stabilization and by decreasing the rate of biomass decomposition (Lal,

2005; Sollins et al., 2007). Compared to monocrop systems, AFS are more efficient in

capturing the resources available at the site. Agroforestry systems are also characterized to improve direct C inputs to the soil. These include systems which have a potential to return pruning to the soil as mulch and abundant tree litter from shade species could be left to decompose on site. Such systems also allow woody species to grow and add litter on soil surface and below-ground during crop fallow phases (Lal et al., 2015).

While most studies report aboveground C sequestration, belowground C are often not reported from agroforestry systems (Jose and Bardhan, 2012). The dynamics of belowground C, mainly consisting of soil organic matter and root systems, are complex and poorly understood, and thus create a gap in our understanding of the C cycle in terrestrial ecosystems.

Literature on Soil C Sequestration in Agroforestry Systems

Sharrow and Ismail (2004) reported C sequestration to be higher in silvopastoral systems compared to forests and pastures. They concluded that agroforestry systems had both forest and grassland nutrient cycling patterns and could produce more belowground biomass. Haile et al. (2008) observed higher soil C in silvopastoral systems (slash pine (Pinus elliottii) + bahiagrass (Paspalum notatum) of Florida compared to the open pasture of bahiagrass up to a depth of 100 cm. The results indicated that total soil organic carbon (SOC) content was higher by 33% in silvopastoral systems near trees and by 28% in the alleys between tree rows than in adjacent open pastures. In a study conducted in Brazil, Schroth et al. (2002) observed that multi-strata systems had an aboveground biomass of 13.2–42.3 Mg ha–1 and a belowground biomass of 4.3–12.9 Mg ha–1 up to a depth of 20 cm compared to those of monoculture at 7.7–56.7 Mg ha–1 and 3.2–17.1 Mg ha–1, respectively. In Kerala, India

Saha et al. (2009) found higher amount of SOC in multispecies homegarden (HG) agroforestry systems compared to monoculture paddy fields. They reported that C stock was highest in forests (176.6 Mg ha-1), followed by HG (119.3 Mg ha-1), and least C stocks were found rice-paddy field (55.6 Mg ha-1) up to a depth of 100 cm. In West

African Sahel, Takimoto et al. (2008) found higher amount of SOC (aboveground + belowground) in parkland agroforestry systems (Faidherbia albida and Vitellaria paradoxa trees as the dominant species), compared to live fence and fodder bank. In a

Dehesa system in Spain, Howlett et al. (2011) reported SOC to be 40 Mg C ha-1 up to a depth of 30 cm.

Sonwa et al., (2007) estimated potential sequestration rates of 5.9 Mg C ha−1 yr−1 for cacao agroforests of Cameroon, 6.3 Mg C ha−1 yr−1 for shaded coffee in Togo

(Dossa et al., 2008) and between 0.3 and 1.1 Mg C ha−1 yr−1 for agroforestry in the

Sahel (Takimoto et al., 2009). The C sequestering potential of indigenous “Ensete” agroforestry systems in the Ethiopian rift valley with understory plant combinations of

Brassica oleracea L. Erythrina spp., Dioscorea alata L., Musa spp, etc, were explored by Negash and Starr (2015), who reported soil organic carbon ranging from 109 to 253

Mg ha-1 up to a depth of 60 cm. Interestingly, the study disclosed that the SOC reported was a function of best management practices like improved shading, increased plant species density adopted in traditional agroforestry systems and varying altitude had no effect on SOC. However, these claims have not been validated and dependable quantitative data on C sequestration potential of different agroforestry systems is not available. These variations in results across studies under AFS indicate the need to expand these studies across various AFS systems.

Soil Carbon in Aggregates and its Stability

Soil aggregates and soil fraction size play a critical role in SOC retention (Six et al., 2002). Soil aggregates are formed by the admixture of mineral particles with organic and inorganic substances (Bronick and Lal, 2005). They are classified depending on their particle size as macroaggregates (>250 µm), microaggregates (250–53 µm) and silt and clay (<53 µm). These aggregates provide physical protection of soil organic matter by the forming a barrier of microorganisms, microbial enzymes, and their substrates (Six et al., 2002). The inclusion of organic material within aggregates lowers the rate of decomposition (Adu and Oades, 1978). The macroaggregates demonstrate higher SOC concentration than microaggregates and silt and clay fractions; at the same time, macroaggregates are more sensitive to the effects of land-use changes (Bronick and Lal, 2005; Tisdall and Oades, 1982) while SOC stored in the silt and clay (<53 µm) are securely held and are physically protected from the effects of land-use changes

(Bronick and Lal 2005; Lal, 2004). Macroaggregates are formed with newly incorporated organic matter, and smaller sized aggregates, and bound with fine roots, plant and microbial residues. Carbon tracer studies have shown the redistribution of C from macroaggregates to microaggregates over time, as microaggregates are formed within macroaggregates (Six et al., 2002). The binding agents that form microaggregates are microbial polymers, root exudates, and polyvalent cations. Silt

+clay sized aggregates contain little occluded OM, and mineral sources become important binding agents (organo-mineral complexes). Mean residence times for macroaggregates, microaggregates, and silt+ clay sized aggregates vary from 1–10, 25, and 100 – 1000 years (Parton et al. 1987). As one considers the three aggregates sizes in descending order, C: N ratios and occluded OM decrease mainly because the

binding agents shift from biotic (in macroaggregates) to more mineral sources (in silt + clay); (Qiming et al. 2003). When soil organic matter is the main binding agent, a hierarchy of protection has been demonstrated as smaller aggregates are stored in increasingly larger size. This is attributed to the fact that silt + clay sized aggregates are protected within macroaggregates (Six et al., 2000; Tisdall and Oades, 1982). Since mineral complexes are generally very stable in the soil, the protection of this hierarchical stabilization mechanism will promote long term C storage in the soil. The protection of macroaggregates, in particular, helps to protect C stored in smaller aggregate sizes.

Given the importance of size fractions in SOC storage, the potential for C sequestration in soils under AFS must determine the extent of C storage in different aggregate classes at deeper soil depth classes. Available literature has shown that AFS the storage of recalcitrant C in the smallest fraction size in deep soil (Chen et al., 2017;

Haile et al., 2008; Howlett et al., 2011; Saha et al., 2010; Tonucci et al., 2011). None of these studies have explored the extent of C storage in aggregate under shaded perennial AFS.

Shaded Perennial Agroforestry Systems and Carbon Storage

Cacao (Theobroma cacao L.) and coffee (Coffea arabica, Coffea canephora and other species within the genus Coffea), tea (Camellia sinensis) are cultivated typically in agroforestry systems (AFS) in close association with a rich list of tree species as an overstory on the same plot (Beer et al., 1998; Montagnini and Nair, 2004; Nair, 2017).

These systems are commonly termed as shaded perennial agroforestry systems and are credited for stocking significant amounts of carbon, bolstered with enormous potential to mitigate climate change. Carbon stocks in shaded agroforestry systems with perennial crops may vary between 12 and 228 Mg ha−1 (Somarriba et al., 2014, 2013).

A shade based AFS plantation has two components: shade tolerant species like cacao, tea or coffee and shade providing canopy (overstory). The canopy component usually includes plants taller than cacao, tea and coffee trees (De Beenhouwer et al., 2013).

Together, coffee, tea and cacao represent the second largest export products from developing countries and these crops cover a substantial amount of the world’s agroforest area (Hombegowda et al., 2015). Additionally, these crops provide income for over 30 million smallholders, mostly in developing countries (Donald, 2004). In traditional AFS, the crops are grown under a more or less dense canopy of various indigenous shade tree species (Perfecto and Vandermeer, 2008) while commercial production of these crops are usually devoid of shade canopies. Traditional cacao and coffee management is handled by community based farmers and it combines sustainable yields with some degree of biodiversity conservation (Moguel and Toledo,

1999). However, fluctuating prices and increasing demand for cacao and coffee on the world market, and increasing local human population pressure, push farmers to intensify the traditional agroforestry management and/or expand the cultivated land area

(DeFries et al., 2010). Intensification practices include removal of slow growing tree species that are suboptimal for the provision of shade, and the thinning of shade trees

(Aerts et al., 2011). Even though moderate shade levels have little effect on cacao or coffee yield (Perfecto et al., 2005), farmers in many parts of the world are converting shaded cacao and coffee systems into unshaded monocultures to increase short-term income. The ultimate consequence of such transformation of the natural forest into a plantation with no canopy is loss of biodiversity, faunal habitat loss, soil carbon loss is concerning (Tscharntke et al., 2011).

Shaded AFS are considered to improve soil organic carbon (SOC) storage due to their high plant diversity and forest-like structure and composition; however, no authentic data is available to validate this conjecture, particularly in stocking C in the soil within soil fraction sizes. The SOC stock in cacao + Gliricidia (Gliricidia sepium) AFS in

Indonesia amounted to 155 Mg C ha−1(0–100 cm soil depth) (Smiley and Kroschel,

2008) and SOC reserves in cacao + Erythrina (Erythrina poeppigiana) in Costa Rica was 240 Mg ha−1 (0 – 45 cm depth) (Fassbender, 1993).

A cacao alley cropping system in Costa Rica was reported to contain 162 Mg C ha−1 in the 0 – 40 cm soil depth (Oelbermann et al., 2006), a West African cacao AFS had 18.2 Mg C ha−1 in the 0 – 15 cm soil depth (Isaac et al., 2005) and a homegarden system in Kerala was reported to contain 119.3 Mg ha-1 in the 0 – 100 cm depth while an adjacent paddy field from the same study location was reported to contain 55.6 Mg ha-1 only by Saha et al. (2010). These studies on soil C storage have not examined the extent of C-storage variation under different land-use systems involving various plant forms (trees and crops alone and in association), various shade density under similar ecological conditions. The above-mentioned studies were limited to humid climatic conditions, indicating that climatic factors are important enough to be considered while taking up C stock studies in such systems. This is mainly attributed to the decomposition of litter and biomass; the rate of which could be directly dependent on climatic conditions including precipitation, humidity, soil moisture etc.

In Indonesia, the aboveground plant biomass was significantly lower in agroforestry with reduced canopy cover, primarily due to the removal of large trees

(Steffan-Dewenter et al., 2007). Shaded perennial agroforests also have considerable

potential to sequester carbon in soils. In a Chrono-sequence study in Indonesia, maize monocultures lost considerable amounts of SOC with time. Interestingly, conversion of maize monocultures into shaded agroforests increased SOC stocks (Dechert et al.,

2004). Secondary forests on formerly degraded lands can have high SOC indicating a significant soil carbon sequestration capacity of shaded perennial AFSs (Somarriba et al., 2013). Further bolstering the potential of shaded perennial AFSs, soil carbon stocks in cacao and coffee agroforests in Indonesia and India were reported to differ only slightly from those of natural forests (Hertel et al., 2009; Hombegowda et al., 2015) .

The annual leaf litter C input to the soil is much lower in shaded agroforests than in natural forest, while the importance of root litter C flux to the soil is particularly high in shaded AFS. This is due to a fine root production and turnover in of a similar magnitude to natural forests (Hertel et al., 2009).

The relative importance and overall effect of the different interactions between shade trees and coffee/cacao are dependent upon site conditions (soil/climate), component selection (species/varieties), belowground and aboveground characteristics of the trees and crops, and management practices. By increasing the shade cover from

50% to 93% from the monoculture to the multispecies shade tolerant cacao AFS system, Abou Rajab et al. (2016) demonstrated that carbon sequestration rates were significantly highest in cacao-multi plots with 18 Mg C ha-1 yr-1 compared to the less productive cacao-gliricidia and cacao-mono stands with 13 Mg C ha-1 yr-1 and 9 Mg C ha-1 yr-1. Shade based AFSs positively influence the SOM content in soil. Beer et al.

(1990) reported SOM increased by 21 % under pruned leguminous Erythrina poeppigiana and by 9% under unpruned non-leguminous Cordia alliodora over a 10

year period following conversion of sugar cane fields to shade-based cacao plantations.

Ehrenbergerová et al. (2015) investigated coffee plantations located in the foothills of the Peruvian Andes and evaluated the aboveground and soil carbon storage of agroforestry coffee plantations with different dominant shading trees, including Inga spp., Pinus spp. (both 15 years old) and Eucalyptus spp. (7 years old). The total carbon stock at the Inga site was 119.9 ± 19.5 Mg ha-1, of which 69 % was located in the soil and 29 % in the trees. The sun coffee site contained 99.7 ± 17.2 Mg ha-1 which was mainly located in the soil (99%) while litter and coffee shrubs represented the remaining fraction, estimated at 1 and 0.2 %, respectively. The total carbon stock at the Pinus site was 177.5 ± 14.1 Mg ha-1, where the majority of carbon was fixed in the soil (57%) and trees (40%). Tumwebaze and Byakagaba (2016) and compared the SOC stocks among

Coffea arabica L. (Arabica coffee), Coffea canephora (Robusta coffee) agroforestry systems and coffee monoculture in Uganda up to a depth of 30 cm. They reported higher SOC under shaded AFS than coffee mono crop system. Under cacao AFS in southeast Ghana, Asase and Tetteh, (2016) reported similar SOC stocks under AFS as that of forest (15 Mg C ha-1 in AFS and 16 Mg C ha-1 under forest). In a coffee AFS planted with Inga densiflora in Costa Rica, Hergoualc’h et al., 2012 reported that SOC stocks under AFS remained unchanged while the monoculture coffee plots showed significant decrease up to a depth of 10 cm, 9 years after the establishment. Several other studies on shaded perennial AFS reported improved SOC stocks under these systems even up to a depth of 100 cm (Cardinael et al., 2017; Hertel et al., 2009; Kim et al., 2016; Monroe et al., 2016; Norgrove and Hauser, 2013).

The veritable explosion on available literature indicate that the efficiency of shaded AFS is site specific and depends on management practices. Even though, it is apparent that shaded perennial AFS are efficient in promoting soil carbon, the extent of the stability of carbon will depend of how these systems allocate C in aggregate fractions.

Management Practices in Shaded Perennial AFS and C Storage

Agricultural practices such as tillage, organic management, fertilizer application, species of shade tree, crop species and plant residue input have been identified to affect carbon sequestration (FAO, 2015). Some of the farming practices that might affect C sequestration are discussed below.

Selection of Species

The carbon sequestration potential of a system is often driven by the species of plants included (Liu et al., 2016; Post and Kwon, 2000). Improved aboveground biomass is believed to increase C storage. With enhanced biomass, faster growth and improved rhizodeposition, the trees and perennial crops have better C sequestration potential than annual herbs (Montagnini and Nair, 2004). There is not much of a consensus whether trees species diversity improves C storage. Saha et al. (2009) observed that carbon sequestration increased with trees species diversity, however, Liu et al., (2016) argued that tree species composition rather than species diversity impacts soil organic carbon. Different tree species influence soil carbon sequestration through the variation in litter production and root activity (Jones et al., 2009). Rooting patterns, spatial variation and rooting depth vary among tree species which affect soil carbon differently (Pierret et al., 2016).

Inorganic vs. Organic Fertilizer:

The optimization of fertilization strategies is an urgent need for sustainable agricultural practices and in order to improve plant production and subsequent soil carbon storage (Rong et al., 2016). The application of organic fertilizers are considered to improve SOC over conventional fertilizers (Velmourougane, 2016) but many scientists argue that the application of organic fertilizer leads to GHG emission in the form of N2O. Most shaded perennial coffee farms use inorganic fertilizers with the exception of small holder farmers. Studies have not simultaneously compared the use of inorganic fertilizer vs. organic fertilizer in coffee farms and how they alter soil carbon storage which remains an unexplored research area.

Plant Residue

SOC mineralization is a function of plant residue inputs and quality. The continual addition of decaying plant residues to the soil surface contributes to the biological activity and the carbon cycling process in the soil. Breakdown of soil organic matter and root growth and decay also contribute to these processes. Plant residue in soil is also a driver of carbon sequestration. The effect of plant residue on SOC varies between studies and there is a need to improve our understanding on how they affect C dynamics. About 12% of plant residues get converted to SOC (Mbow et al., 2014).

Ghimire et al. (2017) reported that addition of legume plant residues is easily decomposable compared to oat plant residues which has a slower decomposition rate.

Thus, the microbial carbon use efficiency varies under different plant residue type and soils with greater microbial carbon use efficiency have greater SOC sequestration. The results of this study suggested that the choice of the plant residue affects SOC mineralization and nutrient cycling in agroecosystems.

Synthesis

The available literature indicates the importance of agroforestry systems such as shaded perennials in enhancing soil carbon sequestration. Limited information is available of soil aggregation and carbon storage within these agroforestry systems.

Several studies have pointed out that shaded perennial agroforestry systems with its improved aboveground biomass and proliferating root systems often mimic forest like ecosystems and can ensure high C storage compared to monoculture systems and

Homegarden. Due to differences in management, shade tree species and type of plant input, the carbon sequestration potential within similar systems is variable and highly site specific. The potential of carbon sequestration of these systems can be beneficial for the environment and help stake holder in designing sustainable land-use practices.

The popularity of these special commodity crops like coffee, tea and cacao across the tropics encourage the study by extending the applicability of the results to policy makers and carbon auditors. Based on the current state of literature and knowledge gaps, the hypotheses formulated for this study are:

 Shaded perennial AFS contain more C in the deeper soil layers compared to treeless monoculture systems and sparsely rooted Homegarden systems.

 Shaded perennial AFS compared to treeless systems store more recalcitrant C in deeper soil layers, indicating their more efficient C sequestration potential.

 The SOC stocks in whole soil and aggregates under shaded perennial AFS varies on practices like use of shade trees, species of shade trees and type of management practices

CHAPTER 3 META-ANALYSIS

This chapter presents a statistical assessment of the available literature on soil organic carbon stock improvement upon adopting various agroforestry systems. To quantify the relative soil C contribution from trees in agroforestry systems (AFS), the study assessed the reported differences in SOC stocks under agroforestry systems in comparison with other land-use systems (Agriculture, Forestry, Pasture, or Uncultivated

Land) in various soil-depth classes in four major agroecological regions (arid and semiarid, ASA; lowland humid tropics, LHT; Mediterranean, MED; and temperate, TEM) around the world.

Introduction

Ecosystems can act as carbon (C) sinks by capturing and storing significant amounts of atmospheric carbon dioxide (CO2) in biomass and in the soil. Tree-based land-use systems such as agroforestry are reported to have the potential to enhance C sequestration both above- and below ground (Nair et al., 2009; 2010). The importance of managing such systems through appropriate agroecological (carbon) farming methods is increasingly being recognized as a promising strategy for enhancing soil carbon sequestration. Indeed, agroforestry has figured prominently in the deliberations and action plans related to climate change adaptation and mitigation and environmental protection at various international, national, regional forums. The success of all such efforts depends on the abundance and availability of rigorously proven scientific facts concerning the underlying assumptions on the role and potential of agroforestry systems (AFS) to sequester C. Although several studies on C sequestration under AFS are reported in the literature, they are highly variable in the study procedures as well as

the nature of systems and study locations such that the results lack the needed quantitative rigor. This makes it difficult to extrapolate them to broader contexts of systems and locations outside the specific locations of the individual studies. The objective of this study was to undertake a statistically rigorous, quantitative assessment of the scattered results on soil organic carbon (SOC) stocks reported under various

AFS, in comparison with those in agricultural, forestry, and pasture systems in different agroecological regions around the world through a meta-analytical approach.

Given that intentional integration of trees in agricultural (crop and livestock) systems is the fundamental attribute of agroforestry systems (Nair, 1993), AFS offer greater opportunities than monocultural (single-component) agricultural systems for capture and storage of atmospheric CO2 in biomass and soils. This has been attributed to several reasons including efficient C (and nutrient) cycling within the soil–plant system, increased return of biomass C to soil, decreased biomass decomposition and soil organic matter (SOM) destabilization in the tropics, and sequestration of soil C in deeper layers of soil (Montagnini and Nair, 2004; Nair, 2012; Nair and Nair, 2014;

Oelbermann and Voroney, 2011; Saha et al., 2010; Tonucci et al., 2011).

Since the beginning of realization of the importance of carbon sequestration (i.e., capture and storage of atmospheric CO2 in long lived pools: UNFCCC, 2006), in climate-change mitigation and adaptation, a large body of scientific literature has accumulated on C sequestration under land-use systems; AFS are no exception. Nair et al. (2010) referenced most of such publications in AFS; several others have appeared since; e.g., Abou Rajab et al., 2016; Cardinael et al., 2017; Jose & Bardhan, 2012;

Lorenz & Lal, 2014; Post & Kwon, 2000; Upson et al., 2016; Kim et al., 2016; De

Stefano and Jacobson, 2017 . While many of these are based on biomass and soil sampling undertaken at a specified period, some are based on field measurements and simulation models based, naturally, on various assumptions and reports; and, almost all of them report the amount of SOC stocks under specific systems at (during) the time of study. While these reports are extremely valuable to understand the existing conditions, they represent only one point in the continuum of land-use changes. They are inadequate to predict the rate of change in SOC with alterations in land management practices from agriculture to AFS (introducing trees into agricultural fields) or forest to

AFS (converting forestlands into AFS by outright forest clearing and AFS establishment or thinning existing forest and underplanting shade-tolerant specialty species). Such information is needed for properly assessing the benefits associated with the adoption of AFS in terms of economic advantages and incentives for local farmers. A meta- analysis, which is a statistical procedure for comparing and synthesizing results from different studies for finding common patterns, discrepancies, or other interesting relationships that may not be detectable from individual studies (Borenstein et al.,

2009), provides a more comprehensive approach to attaining such information. When the treatment (or effect size) is consistent among studies, meta-analysis can be used to identify this common effect.

The effects of adopting alternative land management practices on SOC stocks and other ecosystem services in several land-use systems have been summarized through several meta-analyses (Berthrong et al., 2012; De Beenhouwer et al., 2013;

Don et al., 2011; Guo & Gifford, 2002). Berthrong et al. (2009) evaluated how afforestation affected SOC along with other soil properties like N, pH, CEC across pine

plantations; Guo & Gifford (2002) found SOC stocks declined after the conversion from pasture to plantation, native forest to crops, and pasture to crops. Reporting the results from a meta-analysis on the effect of cover crops on C sequestration in agricultural soils, Poeplau & Don (2015) suggested that cover crops significantly improved SOC stock to an average of 16.7 Mg C ha-1 than the reference croplands. The SOC storage in tropical soil post afforestation and reforestation was studied by Marín-Spiotta &

Sharma (2013) who reported that SOC stocks differed with soil depth and environmental factors, highlighting the importance of comparing C across consistent depths and age of management practices. A global meta-analysis conducted by Don et al. (2011) found that SOC losses could partly be reversible under tropical weathered soils. According to the authors, if agricultural land was afforested, SOC improved by

29%, and when cropland was converted to grassland, the increase was 26% for global land-use change in soil. In addition to such “general” meta-analyses that are relevant to

AFS, two studies that are specific to AFS have also been reported recently (Kim et al.,

2016; De Stefano & Jacobson, 2017). The analysis presented here entails a substantial improvement over those two AF-specific analyses in a number of ways including use of larger number of data points, coverage of broader and more diverse range of ecological regions, and adoption of more elaborate and rigorous statistical procedures.

Recognizing the extreme location-specificity of AFS practiced around the world depending on biogeographic, ecological, and socioeconomic conditions, this study focuses on understanding the trend of SOC stock changes (∆SOC) across Forest-

Agroforest-Agriculture/Pasture continuum in different agroecological regions around the world where agroforestry is practiced.

Materials and Methods

Agroforestry Systems and Agroecological Regions

The AFS case studies used for the analysis were selected from peer-reviewed publications found in ISI Web of Science and Google Scholar, based on a search with keywords (alone or in combinations): agroforestry/ specific agroforestry system, soil organic carbon (SOC) stock, C sequestration, soil C stock, and soil C pool. Only those reports that included data on soil C stocks were included. The selected case studies consisted of all three major groups of AFS (Agrisilviculture, Agrosilvopasture, and

Silvopasture) according to Nair’s (1985) classification; they were further organized into subgroups of AFS (Nair 2014); thus, the dataset represented 15 AFS (Table 3-3-1).

Non-agroforestry land-use practices (control) were grouped into four categories based on the information provided by the authors: (1) Agriculture, (2) Forest, (3) Pasture, and

(4) Uncultivated Land. Uncultivated land was the term used in some publications to refer to barren/fallow land that was not under any distinguishably managed land-use systems as agriculture, forestry, or pasture.

The study included different types of AFS, from 30 countries across Asia, Africa,

North America, Latin America, and Europe (Table 3-3), with a total of 858 data points from 78 case studies. The data points were categorized under four agroecological regions (Table 3-3) according to the Köppen climate classification (The Köppen

Classification System):

 Arid and semiarid (ASA): Köppen Group B (BWh, BSh, BWk, BSk, BWn, BKn);

 Lowland humid tropics (LHT): Köppen group A, tropics (Aw, Af, Am) and part of Group C, sub-tropical climate (Cfa and Cwa);

 Mediterranean (MED): Köppen Csa, Csb and Csc;

 Temperate (TEM): Köppen Group C (Cwb, Cwc, Cfb, Cfc) and Group D (Ds, Dw, Df).

Each region had four types of AFS, except ASA that had three:

 ASA: Agrisilviculture, Agrosilvopasture, Silvopasture;  LHT: Multistrata Systems, Agrisilviculture, Silvopasture, Agrosilvopasture;  MED: Agrisilviculture, Agrosilvopasture, Silvoarable, Silvopasture;  TEM: Agrisilviculture, Silvoarable, Silvopasture, Protective Systems.

The identified case studies were successively scrutinized by analysis of their abstracts following the Preferred Reporting Items for Systematic Reviews and Meta-

Analyses (PRISMA) (Appendix A, Figure A-1), an evidence-based reporting tool used for systematic reviews and meta-analyses (Moher et al., 2009). The publications were then included in a preliminary list, which were further screened to select only those that had reported soil C concentration or stock per unit land area (Mg C ha-1 or equivalent), for land-use comparisons (Agroforestry vs. Agriculture/Forestry/Pasture/Uncultivated land) with means and number of replicates. If SD (standard deviation) values were missing from a study, other variance measures like standard error (SE) of means were used to calculate the SD, or error bars of the figures were used for SD estimation. The overall coefficient of variance (CV), without the missing SD, was generated and then multiplied with the means of missing SD to obtain the imputed SD. Additionally, the data sets were categorized according to the age of the AFS into four groups: 0–5, 5–10, 10–

20, and >20 years, implying the time since agroforestry practices were initiated; for example, a study that reported SOC stocks from an AFS site of 15 years was placed in the 10–20 years age group.

Soil-Depth Classes and SOC Stock Calculations

Studies selected for the analysis had reported varying depth classes from 10 to

250 cm. Results of an initial analysis for the depth class below 100 cm (100–200 cm) showed the results were skewed because of the low number of data points. Therefore, soil depths up to 100 cm only were included in further analysis, and these were grouped into four soil-depth classes (all in cm): 0–20, 0–40, 0–60, 60–100, and 0–100. All the included case studies, however, did not have all these depth classes, and had not reported SOC stocks in those distinct depth-classes. Nevertheless, the studies were categorized according to the above-mentioned depth classes as best as possible. For example, if a study reported SOC stocks or concentration up to 10cm, the derived data set was added to 0–20 cm depth class for this analysis.

The SOC stock comparisons of soils with varying bulk densities and soil types to a depth class (0–100 cm) would imply unequal soil mass comparison whereas SOC stock investigation required the analysis of equivalent soil masses (Ellert and Bettany,

1995; Poeplau and Don, 2013). Mass correction, however, was not performed in our study because it would have involved comparison of paired sites and not chronosequence land-use change over the same site. The remaining studies reported C

-1 concentration as percentage (SOC concentration). In such cases, 푆푂퐶푠푡표푐푘 (Mg C ha ) was calculated as:

푆푂퐶푠푡표푐푘 = 푆푂퐶푐표푛푐 × 퐵퐷 × 퐷 (3-1)

-1 -3 where 푆푂퐶푐표푛푐 is the C concentration (g 100 g soil), 퐵퐷 is the bulk density (g cm ) and

퐷 is the depth of sampling or the depth increment (cm).

Meta-Analytical Approach

We evaluated the effect-size based on the Response Ratio (RR) of SOC stocks.

The response ratio, suitable for variables measured on a true ratio scale with a natural scale-unit and a natural zero point, quantifies the proportionate difference between the treatment mean (훸푡) and the control mean (훸푐) (Borenstein et al., 2009). To analyze changes in SOC stocks up to 100 cm depth for Agroforest vs. Agriculture or Agroforest vs. Forest or Agroforest vs. Pasture or Agroforest vs. Uncultivated Land, we calculated the response ratios as:

푅푅 = 훸푡/훸푐 (3-2)

To assess the relative impact of adopting AFS (either AFS vs.

Agriculture/Forest/Pasture/Uncultivated Land) on SOC stock, we considered AFS as the treatment group and Agriculture/Forest/Pasture/Uncultivated Land as the control group.

The calculated response ratio thus quantified the percentage decrease or increase of soil carbon (Mg C ha-1) between the two land management systems (AFS vs. Forest or

Agriculture vs. AFS or AFS vs. Pasture). Since the RR must be log-transformed for statistical validation, all reported SOC stock values were log-transformed to achieve normality before conducting these analyses. The natural-log response ratio (ln RR) was calculated as:

푙푛(푅푅) = [푙푛( 훸푡) − 푙푛 (훸푐)] (3-3)

The corresponding sampling variance for each ln(푅푅) was calculated as:

2 1 1 (3-4) 푉푙푛(푅푅) = 푆푝표표푙푒푑 ( 2 + 2) 푛1(푋푡) 푛2(푋푐)

Weighted response ratio for each category was calculated from the RR of individual

pairwise comparison between each AFS (treatment) and control

(Agriculture/Forest/Pasture/Uncultivated Land) as:

푚 푘 ∑푖=1 ∑푗=1 푤푖푗푙푛(푅푅)푖푗 (3-5) 푙푛(푅푅)푤 = 푚 푘 ∑푖=1 ∑푗=1 푤푖푗 where, RRij is the jth comparison (j = 1, 2, …, k) in the ith group of AFS type (i = 1, 2, …,

m). The standard error, (푆퐸) of the 푅푅 weighed was calculated as:

1 (3-6) 푆퐸(푙푛(푅푅)푤) = √ 푚 푘 ∑푖=1 ∑푗=1 푤푖푗

Where the weighting factor (푤푖푗) was estimated as:

푤푖푗 = 1/푉푎푟 (푙푛 푅푅 ) (3-7)

Mean weighted effect size for each observation was calculated with 95% confidence

intervals(퐶퐼). In order to evaluate if adopting AFS (treatment) had a significant effect,

The 95% confidence interval (95% CI) was calculated as:

퐶퐼 = 푙푛 (푅푅푤) ± 1.96 푠푒 (ln (푅푅푤)) (3-8)

To facilitate interpretation, along with response ratio, percentage changes in soil organic

carbon (∆SOC%) stock was also considered, which was calculated as:

∆푆푂퐶% 푠푡표푐푘 = (exp(푙푛 푅푅푤) − 1) × 100 (3-9)

Corresponding variance in percentage changes in soil organic carbon stock,

푉푎푟(∆푆푂퐶%)푠푡표푐푘푎푝푝푟표푥 was calculated as:

2 푉푎푟(∆푆푂퐶%)푠푡표푐푘푎푝푝푟표푥 = 100 × 푉ln(푅푅) × exp (2 × 푙푛(푅푅)) (3-10)

Analyses were performed with the packages metafor (Viechtbauer, 2010) and ggplot2

(Wickham, 2017) and R version 3.4.0 (R Core Team, 2017). Random effect models are considered where the heterogeneity of true effect size among studies arises from random variation around the mean effect sizes of population studies. This allows different study specific effect sizes (Borenstein et al., 2009).

Meta-Regression, ANOVA and Linear Mixed Effect Model

In this study, we performed meta-regression (ANOVA), to assess the relationship between the moderator variables or covariates (Soil Depth, Agroecological region, Type of AFS nested within Agroecological region, and Age of management practices) and a dependent variable (SOC stocks in Mg C ha-1). The Linear Mixed Effects model with four categorical factors (moderators): Depth, Agroecological region, Type of AFS nested within Agroecological region, and Age of management practices were considered separately for each control. Under the random/mixed effects model, for any predicted value there is a distribution of effect sizes, i.e. the distribution is centered over the predicted value, but the population effect size can fall to the left or right of center).

Thus, the hypothesized models were of the form:

푦푖 = 훽0 + 훽1푥푖1 + ⋯ + 훽푝푥푖푝 + 휇푖 + 휀 푖 (3-11)

where, th 푦푖 = 푙푛(푅푅) of the i study.

p+1 = the total number of moderators,

th th 푥푖푗 = value of the j moderator variable for the i study,

2 휀푖 = sampling error, 휀푖 ~ N (0, vi ) and

휇푖 = random error to account for the heterogeneity among sites,

2 2 휇푖 ~ N (0, τ ), where τ denotes the residual heterogeneity among the

true effects.

The goal of the analysis was then to examine to what extent the moderators included in the model influence the size of the average true effect. In addition to testing the impact of moderators for statistical significance, it is important to quantify the magnitude of their relationship with effect size (Borenstein et al., 2009). For this purpose, index based on the percent reduction in true variance, analogous to the R2 index was used with the primary studies The R2 index, which is defined as the ratio of true variance (between studies variance) explained to total true variance was used to quantify the proportion of variance explained by the covariates. In meta-analysis, the total variance includes both variance within studies and between studies. The covariates were considered into the analysis using a pre-defined sequence and assess the impact of it on SOC stocks. Only categorical variables were considered as covariates – a process known as dummy coding. Dummy coding helps in assessing the impact of covariates and predicts effect size in the studies included for this meta- analysis. The R2 index in meta-analysis is computed as:

τ2 (3-12) 2 푒푥푝푙푎푖푛푒푑 푅 = 2 τ푡표푡푎푙

2 2 where τ푒푥푝푙푎푖푛푒푑 is the true variance explained and τ푡표푡푎푙 is the total variance.

The intercept in the full mixed effect model is the combined effect of level 0 (reference category) of each factor. The reference categories are the categories that come first in alphabetic order. Standardized effect sizes were considered, giving greater weight to

2 2 studies with low variance measures. A weight factor (1/variance = 1/ (τ + vi )) was used in the analysis (usual in meta-analyses). Restricted Maximum Likelihood based inferential techniques was used to estimate the amount of residual heterogeneity (τ2) and standard maximum likelihood were used for other parameters. KN-HA adjustments

(Knapp and Hartung, 2003) were used, to account for the heterogeneity while estimating the other model parameters among the true effects in the mixture model.

Other detailed statistical treatment of the datasets included Logarithmic

Transformation of data and Test for Asymmetry, Rosenthal’s and Orwin’s Fail Safe

Number. Details of these analyses are given in Appendix A.

Results

The meta-analysis yielded results on changes in (∆SOC) stock in different soil depth classes across Forest-Agroforest-Agriculture/Pasture/Uncultivated Land continuum in various agroecological regions around the world. Statistically, when AFS were compared to other land-use systems, the Kendall rank correlation test for publication bias and the regression test with standard errors as predictors in a meta- analytic model did not indicate any significant bias (Appendix A1, Figure A-10–A-13).

Furthermore, the funnel plot was symmetric and the QQ plot did not show any significant departure from normality, i.e. the data points were mostly distributed along the x = y line. In comparisons of AFS with adjacent controls, neither the Kendall rank correlation test nor the regression test for publication bias indicated any bias (Appendix

A).

A summary of (∆SOC) stock up to 40 cm soil depth for all land-use systems combined in different regions is presented in Table 3-4, and a similar summary (∆SOC)

stock up to 60 cm depth across Forest-Agroforest-Agriculture/Pasture/Uncultivated

Land continuum with respect to major agroforestry systems is presented in Table 3-5.

Further, a summary of percentage changes in soil organic carbon (∆SOC%) stock between AFS vs. Agriculture/Forestry/Pasture in four soil depth classes (0–20, 0–40, 0–

60, and 0–100 cm) in the three major agroecological regions is presented as Figure 2.

Other results are presented in Appendix and as supplementary information. Salient aspects of the results of comparison of percentage changes in soil organic carbon

(∆SOC%) stock for AFS vs. Agriculture/ Forest/Pasture are summarized below. The percentage changes in soil organic carbon (∆SOC%) stock for AFS vs. Uncultivated lands comparison had low numbers of data points and did not show any consistent trend; those results are therefore not reported in the text; however, they are included in the summaries (Table 3-4, Figure 2).

AFS vs. Agriculture

Comparison of AFS vs. Agriculture showed that, in general, percentage in soil organic carbon (SOC%) stock improved under AFS across all agroecological regions and depth classes. Outliers or exceptions to this general trend were noted only in a few cases with smaller sample sizes. Within the 0–20 cm depth class, the maximum SOC% stock “gain” (i.e., increase) was noted under Silvopastoral systems in LHT region

(151.6%) (Figure 3-2), followed by Silvoarable Systems in the MED region (52.5%:

Figure 3-3). In the 0–40 cm depth profile, AFS vs. Agriculture showed substantial improvements in SOC% stock with the highest percentage increases under

Agrisilviculture and Multistrata Systems in the LHT region, 33.6% and 24.1% respectively (Table 3-4). The LHT region showed SOC% stock gains under all depth classes up to 100 cm (Figure 2). The Silvoarable systems in the MED region showed

the least increase in soil organic carbon percentage (SOC%) stock (1.4%: Table 3-4).

The results showed an increase in SOC% stock (40.6%) in the 0–60cm depth class under Multistrata Systems in the LHT region (Table 3-5), and 48% increase in the 60–

100 cm depth class (Appendix A, Figure A-6). However, SOC stocks under ASA and

LHT Agrisilviculture systems declined, compared with Agriculture, in the 0–60 cm depth class by 9.9% and 9.3% respectively (Table 3-5). In the 0–100 cm soil depth, improved

SOC% stock was noted under AFS vs. Agriculture systems (Multistrata Systems:

+17.9%; Agrisilviculture system: +36.8% LHT; and Silvopasture; +8.8% MED; (Appendix

A, Figure A-4).

AFS vs. Forest

The comparison of AFS vs. Forest showed a general trend line of reduced

SOC% stock under AFS across most agroecological regions and depth classes (Tables

3-4 and 3-5, Figure 3-5). Multistrata Systems vs. Forest had the least decline in SOC% stock (0.2%) in the top soil (Figure 3-2). Similar trend in SOC% stock declines were observed in the 0–40 cm depth with Agrisilviculture systems (27.6%, –6.9% and –

14.7%) with ASA, LHT and TEM regions respectively (Table 3-4). The only positive

SOC% trend was noted under Multistrata Systems vs. Forest (+12.6%: Table 3-4), indicating improvement in SOC stock over the “control” (Forest). While SOC% for

Agrisilviculture vs. Forest continued to decrease within 0–60 cm (–26.8%: Appendix A,

Figure A-3), Multistrata Systems vs. Forest had increasing SOC% (+9.3%) within the

LHT region. In the 0–100 cm depth, AFS vs. Forest improved SOC% stock under

Multistrata systems and Agrosilvopasture, while Agrisilviculture had a decline (–26.8%;

LHT: Appendix A, Figure S4b). In the MED region, Silvopasture showed an improved

SOC% by 56.1% over Forest within 0–100 cm (Appendix A, Figure A-4).

AFS vs. Pasture

Comparing AFS vs. Pasture showed varying trends in SOC% stock improvements across different agroecological regions and depth classes, especially in the tropics. In the top soil (0–20 cm), the ASA region showed SOC% stock improvement across Agrisilviculture (+15%), Agrosilvopasture (+12.8%) and a substantial increase of

+89.4% in the Silvopasture systems (Figure 3-1). The improvement was only marginal

(+2.2%) under Silvopasture in the TEM region. In the LHT region, SOC stocks declined under all AFS practices; the Silvopasture systems within MED region too reported declined SOC% stock (–9.2%) in the top soil (Figure 3-3). The trend was, however, reversed, down the soil profile (0–40 and 0–60 cm), where Agrisilviculture and

Silvopasture systems continued to show improvement in SOC% stock over Pasture. On the other hand, in the 0–100 cm depth, the percentage changes in soil organic carbon

(∆SOC%) stock were higher under Silvopasture systems within ASA and MED regions by 27.1% and 22.7% respectively (Appendix A, Figure A-4). Overall, the results for AFS vs. Pasture varied across agroecological regions, with ASA being the region with the highest SOC% stock improvements.

SOC Stocks under AFS with Varying Age of System

For assessing the effect of the duration of management practices on SOC stock,

AFS under each agroecological region and specific soil depth classes were grouped into four “age” groups: 0–5, 5–10, 10–20, and >20 years.

Age group: 0–5 years

Overall, SOC% stocks increased in the top soil (0–20 cm) under AFS compared with other land-use systems during the first five years of establishment of AFS in all agroecological regions. The increases in soil organic carbon under AFS compared with

Pasture were 15.7% in ASA, 3.5% in LHT, and 5.6% in the MED region; compared with

Agriculture, AFS had 9.3% SOC% stock increase in LHT (Appendix A, Figure A-7) and

6% in TEM region (Appendix A, Figure A-7). On the other hand, comparing AFS with

Forest, there was an overall reduction in SOC% stock in AFS in the 0–40 cm depth:

21.1% decline in ASA, 24.3% in LHT, and 12.8% in TEM. In the 0–60 cm soil-depth class, overall SOC stock under AFS in the ASA region increased by an average of

8.55% compared to agricultural systems during the 0–5 years. In LHT, overall, AFS between 0–5 years increased SOC% stocks by 9.7%. In TEM region, AFS vs. Forest comparisons showed SOC% decline in AFS by 10.8% (Appendix A, Figure A-9). In the

0–100cm soil layer, SOC% increased by 19.1% under AFS vs. Agriculture in LHT

Age group: 5–10 years

In the ASA region, AFS aged between 5–10 years, compared with other land-use systems, improved SOC% by 18.1% overall in the soil depth class 0–20 cm. Typically,

AFS vs. Pasture comparisons demonstrated an increase in SOC by 60.5% under AFS while compared with Forests, the SOC% stock under AFS decreased by 26.5%

(Appendix A, Figure A-7). In the LHT, overall, AFS improved SOC% stocks by 27.5%. In other regions, the percentage changes in soil organic carbon (∆SOC%) were less noticeable. In 0– 40cm depth, overall, in the ASA and LHT region, SOC% under AFS improved by 3.4% and 12.19% respectively during 5–10 years. In the 0–60cm depth class, SOC% stock under AFS compared with Pasture increased by 28.2%, and by

27.8% against Agriculture in the ASA (Appendix A, Figure A-8).

Age group: 10–20 years

Comparatively, SOC% stock increased under AFS with increasing age of systems. Overall, under the 10–20 years old AFS, SOC% stocks were higher by 73.3%

over Pasture land in the top soil 0–20 cm. In LHT region, SOC% stocks under AFS was

31.6% higher than under Agriculture. Compared with Forest, however, SOC% was

27.5% lower under AFS (Appendix A, Figure A-7). In the TEM region, SOC% stock under AFS was 8.6% higher compared to Pasture. In MED region, AFS had 40.1% increase in SOC% stock over Agriculture, while compared with Forest there was a decline of 47.9% (Appendix A, Figure A-7). In the 0–40cm depth, the ASA region reported a steep overall SOC% increase (71.03%) when AFS was compared to Pasture land (Appendix A, Figure A-8). In the 0–60cm depth, the comparison of AFS vs. Pasture showed an increase in SOC% stock by 28.21% and 19.6% in the ASA and TEM region respectively. AFS vs. Agriculture practices in the LHT and MED regions improved

SOC% stock by 40.1% and 7.7% respectively (Appendix A, Figure A-9). In the soil profile 0–100cm, AFS over Pasture had SOC% stock increase of 27.1% and 53.1%, for

ASA and LHT regions, respectively. AFS over Agriculture improved SOC% stocks by

55%, while compared to Forest, AFS improved had an SOC stock increase of 3% in the

0–100 cm depth.

Age group: >20 years

In the top soil profile, within the LHT region, SOC% stocks improved by 4.3%

(Appendix A, Figure A-7) overall. In the TEM region AFS had a 27.8% decline in SOC% stocks in when compared with Uncultivated Land after >20 years within the 0–40cm depth in the LHT region, SOC% stocks increased overall by 7.2%. Comparisons of AFS to Agricultural land and Forests showed SOC stock increment by 9.1% and 19.1% respectively. Comparing AFS vs Pasture in the TEM region, SOC% increased by 22.2% under AFS, and in the MED region, AFS compared with single species Agriculture had

9.6% increase of SOC stocks (Appendix A, Figure A-7).

In summary, despite some deviations, there was an overall increase in SOC under AFS compared with Pasture and Agriculture but decrease when compared with Forests across various agroecological regions.

Mixed Effect Models, Meta-regression and ANOVA

Linear mixed effect models with four categorical covariates (moderators) (i)

Depth, (ii) Agroecological region, (iii) Type of AFS nested within agroecological region

(iv) Age of AFS management system, were analyzed individually for comparisons of

Forest vs. Agroforest, Agriculture vs. Agroforest, Pasture vs. Agroforest, and

Uncultivated land vs. Agroforest. (Appendix A, Tables A-3, A-7, A-11, A-14). The results of analyses of Mixed Effects Model and meta-regression individually with

Agriculture, Forest, Pasture, and Uncultivated Land, showing detailed results of tests for heterogeneity are summarized in Table 3-6.

Discussion

This meta-analysis consisting of 858 data points extracted out of 78 peer- reviewed publications from around the world and covering comparisons of Agroforestry vs. Agricultural, Forestry, Pasture or Uncultivated Land represents a detailed effort in quantitatively assessing the impact of agroforestry systems on SOC% stocks. Ideally, to comprehend the influence of different land-use systems on SOC, it is important to consider both SOC stocks and concentrations; using only one of these might cause a bias in assessing the treatment effect. However, it is widely accepted that if only one measure is to be considered, then SOC stocks give a more accurate estimation, and in our dataset, about, 95% of the studies reported SOC stocks (Mg C ha-1). Additionally,

SOC stocks can be used to estimate the SOC sequestration potential (Mg C ha-1 yr-1) of

AFS. Therefore, SOC stock was the chosen measure for this meta-analysis.

SOC Stock Changes under Agroforestry Systems across Agroecological Regions

AFS vs. Agriculture

Overall, SOC stocks were higher under AFS compared with Agricultural systems across agroecological regions, which indicates that AFS, in general, have distinct advantage over monocrop systems in terms of soil C storage. The Mixed Effects model, run as a meta-regression model, showed that the effect sizes of SOC stocks for the

AFS vs. Agriculture comparison were significantly differed from zero. This means that the results are rigorous, devoid of publication biases. Furthermore, the agroecological region turned out to be a significant covariate that influenced SOC stocks (Appendix A), confirming that SOC stock increase varies across regions. For example, SOC stocks improved by 26 % up to 100 cm in the LHT region while the increase in SOC stocks under AFS over Agriculture in the MED region was 5.8% in the same soil-depth class

(Figure 3-2, Figure 3-5, Appendix A, Figure A-3). Repeated input of hedgerow pruning from woody hedgerow species to the alleys in tropical alley cropping, and leaf-litter addition form overstory woody species are the main source of biomass addition to soils in Tropical Agrisilviculture compared to Agricultural systems under similar conditions.

Minimum (or no) tillage could be another possible reason for the high improvement in

SOC stocks under Tropical Agrisilviculture (Oelbermann et al., 2006; Schroth & Zech,

1995).

Multistrata Systems, which include Shaded Perennial AFS and tropical

Homegardens in the LHT region store substantially high amount of SOC stocks even in deeper layers of soil (Abou Rajab et al., 2016; Gama-Rodrigues et al., 2010; Monroe et al., 2016; Saha et al., 2010). These multistrata AFS render ecosystem services with high value for supporting human livelihoods including but not limited to soil carbon

storage (Millennium Ecosystem Assessment, (MEA, 2005; Nair, 2017). Our results showed that overall, the multistrata systems stored 48% higher SOC stocks compared to monoculture Agriculture (Figure 3-2, Appendix A, Figure A-2, A-3, A-4), and were efficient and the most consistent AFS in storing SOC stocks at all soil depth classes up to 100 cm depth, especially in the 60–100 cm depth. It is worth noting that SOC% stocks were substantially higher under Agrosilvoarable systems in the Mediterranean region, ranging from 52.5% in the top soil to 8.5% in the deep soil. Similar results were also reported by (Burgess et al., 2005; Cardinael et al., 2015, 2017; Upson & Burgess,

2013; Upson et al., 2016). In the TEM region, our analysis showed that SOC stocks of

Agrisilviculture (alley cropping systems) increased by 18.4% over Agriculture, which is in accordance with results reported by Cardinael et al. 2012, 2015, 2017; Pardon et al.,

2017; and Wotherspoon et al., 2014.

AFS vs. Forest

Comparing AFS vs. Forest, the SOC% stocks were lower in AFS across almost all depth classes and all agroecological regions, the maximum decline being in the MED region (−47.9%) in the upper 0–20 cm, and the least in the LHT region (−2.1%). It is a well-accepted that deforestation leads to loss in SOC stocks. This was evident in our study as well. The forests retain major proportion of SOC in the topsoil, thus land-use change causes losses in SOC% stocks especially in the top soil (Guo and Gifford, 2002;

Leuschner et al., 2013); similar trends were also reported in the analysis by De Stefano

& Jacobson (2017). This observation was further strengthened by the Mixed Effects

Models with Forest as control, where Depth of soil (p= 0.025*) was a significant covariate to SOC stock changes upon deforestation.

Additionally, the type of AFS nested within agroecological region was also a significant covariate to SOC stock changes upon deforestation (p= 0.022*). This can be explained with the example of SOC stocks under Multistrata Systems in LHT region.

Multistrata systems, like Shaded perennial AFS, have been recognized for substantially higher soil-carbon storage potential than conventional monoculture systems (Abou

Rajab et al., 2016; Gama-Rodrigues et al., 2010; Norgrove and Hauser, 2013; Soto-

Pinto et al., 2010) and are rated as comparable to secondary forests as they often

“mimic” forest-like environment (Beer et al., 1998; Tscharntke et al., 2011). This analysis showed that between Forest and Multistrata systems, the difference in SOC between the two, up to 100 cm depth, was practically very little (3.4% lower in

Multistrata system). The litterfall is the main source of C addition to soil in such systems, the extent of which depends on the type of the shade tree as well as the shaded species. Among the three-major Multistrata Systems, perennial species of coffee

(Coffea spp. ), cacao (Theobroma cacao), and tea (Camellia sinensis), the litterfall is much lesser for tea, estimated at 3.28 Mg ha-1yr-1 (Paudel et al., 2015); for coffee, it could range from 3.3 to 6.3 Mg ha-1yr-1 (Celentano et al., 2011); and for cacao as high as 10 Mg ha-1yr-1 (Beer, 1998). But given that these Multistrata systems have similar fine-root production and turnover rates as for forests (Hertel et al., 2009; Tscharntke et al., 2011), the combined root litter C flux to the soil from the understory (shaded) species and the overstory shade tree could be comparable to or even higher than that under forest, as reported from studies in Brazil and Indonesia (Gama-Rodrigues et al.,

2010; Hertel et al., 2009; Monroe et al., 2016); this is corroborated by our analysis,

which reported increase in SOC stocks by 34.4% compared to Forest between 60–100 cm (Appendix A, Figure A3).

AFS vs. Pasture

The SOC stocks under AFS vs. Pasture varied significantly across Depth (p=

0.0025), Agroecological region (p=0.038), and AFS nested within the agroecological region (p<.0001), (Appendix A, Table A-11). Establishment of AFS over Pasture requires site preparation for planting trees and introducing livestock. These land management practices could potentially reduce SOC stocks in some cases (Guo and

Gifford, 2002). Comparing AFS vs. Pasture, SOC stocks under AFS improved in the

ASA region from 66.2% in the top 20 cm and 27% in deep soils (up to100 cm) (Figure 3-

1, Appendix A, Figure A-2). However, in TEM and LHT regions, SOC stocks declined with increased depth by 5.3% and 21% respectively. Thus, silvopastoral systems might be more efficient in improving SOC stocks in ASA than TEM and LHT region.

Reduced SOC stock in silvopastoral systems compared to adjacent pasture have been reported before (Seddaiu et al., 2013). In pastures, the annual turnover from rhizodeposition, which increases the SOC stocks, is much higher than under trees

(Jobbágy and Jackson, 2000). Introducing leguminous plants often increases soil nitrogen, enhances fertility and improves belowground productivity, and consequently helps build up higher SOC stocks (Boddey et al., 2010; Conant, 2010). Thus, planting trees within a pasture might not substantially improve SOC stocks due to “soil carbon saturation” of pastures which is already at a threshold SOC level (Hassink et al., 1994,

1997) in certain agroecological region like TEM.

AFS vs. Uncultivated land

In majority of the cases, the category of Uncultivated Land included Fallow land, and land-use types which were not defined as Forest/Pasture/Agriculture. In LHT region, conversion of Uncultivated Land to AFS decreased SOC stocks (−3.8%) in the top soil but increased SOC stocks (6.4%) in soil layers up to 60cm (Figure 3-2,

Appendix A, Figure A-2, A-3). However, in TEM region, alteration of management practices from fallow land to AFS led to a major decline (−43.8%) in SOC stocks up to

100 cm. Agrisilviculture systems like Improved Fallows were primarily “designed” for and promoted in temperate, tropical and arid region (Buresh and Cooper, 1999) where degraded soils are a major constraint to improving traditional agricultural practices

(Sanchez, 1999; Sanchez, 2002). Improved fallows have yet to prove their worth in the temperate region because the long dry season in those regions limits the growth and nitrogen fixation potential of the fallow species. Fallows also do not perform well in shallow soils, poorly drained ones, or frost-prone areas (Sanchez, 2002). These AFS are effective for improving nutrients but fall short to significantly increase SOC stocks

(Buresh and Cooper, 1999; Kaonga and Coleman, 2008). Additionally, phosphorus (P) in soil is also a limiting factor in N2 fixation in poor soil which inhibits SOC stock improvement by reducing belowground productivity (Boddey et al., 2010; Conant et al.,

2001).

SOC Stocks in Relation to Age of Trees in AFS Management Practices

Studies on changes in SOC content following change in land management practices and age of trees have reported conflicting results, some indicating increase in

SOC content with stand age (Berthrong et al., 2009), while others had contrary results, especially in early duration after afforestation (Laganière et al., 2010). Ideally, land-use

changes from Agriculture to AFS should affect the surface soil layer consistently across age of stand/trees; however, a majority of studies have found that the SOC stocks to decline initially, and then increase with increase in stand age (Chen et al., 2005). A cross section of these contrasting results was found in our analyses too that age of trees in AFS influenced SOC stocks changes differently under different agroecological regions (Appendix A). The discussion here is limited to influence of tree age on SOC stocks in comparisons between AFS vs. Agriculture or Pasture because transformation of Forest to any other land use (including AFS) is known to cause SOC stocks declines and age of trees per se may not have a substantial role in it.

In our analysis, in ASA region, older AFS improved SOC% stocks better in the top soil. In deeper soil layers (up to 100cm), AFS aged between 10–20years aided in improved SOC stocks (27.3%), (Appendix A, Figure A-7). We did not have sufficient data points under managed AFS >20 years for cumulative analysis. In the LHT region, maximum change in %SOC stocks (range: 14.9% in top soil to 55% in deep soil) was in systems aged between 10–20years. Systems aged >20 years improved SOC% stocks ranging from 4.4% in the top soil to 14.5% in deep soil. Other results did not show any consistent trend. Establishment of AFS involves the introduction of varying species and structural arrangement which could affect SOC stock accumulation differently, i.e. management practices of similar age group can affect SOC stocks differently as trees species interacting with land-use change alters SOC differently (Bárcena et al., 2014;

Nair et al., 2011). The available results from this analysis seem to be inadequate to draw rigorous (convincing) conclusion on the effect of tree management and age on

SOC stock changes under AFS across various agroecological regions. There is a need

to monitor SOC stock changes in greater detail within and beyond the first decades after land-use change and determine the time required for AFS to improve SOC stocks significantly.

SOC Stock Changes and the Forestry – Agroforestry – Agriculture/Pasture Continuum

It is a common perception that conversion of Forest to any other land management practices will result in land degradation in general and decrease in SOC stocks in particular. Nair et al. (2009) proposed “forest>agroforest>monoculture and arable crops” as the general sequence in terms of SOC stock decline. This trend was observed across all AFS in our analysis. The only exception was in the case of AFS vs.

Pasture management practices. Even though AFS vs. Pasture positively affected SOC stocks under AFS in the ASA region, AFS had lower SOC stocks in TEM and LHT regions than in Pasture. It is not unusual to find higher SOC stocks in pasture than AFS or even forests. Guo and Gifford (2002) and Post and Kwon (2000) reported that when forests were replaced by pastures, the considerable loss of aboveground biomass did not necessarily imply loss or decline in SOC. Indeed, pasture established under previously forested areas could have greater potential to store carbon in soil. Further, there could be “soil carbon saturation” in pasture as explained by Hassink (1997).

According to our analysis, AFS significantly affected SOC stocks even up to 100 cm soil depth (0–100 cm). Thus, this study provides a rigorous statistical, global validation of individual results reported by numerous studies that incorporating deep rooted trees on land leads to increased soil carbon stocks (Haile et al., 2010; Howlett et al., 2011;

Montagnini & Nair, 2004; Nair et al., 2009; Saha et al., 2010; Nair et al., 2010). The extent of such AFS-induced SOC storage, however, varies significantly across

agroecological region, age of management practices, and depth of soil, in addition, naturally, to obvious effects of local climatic and soil conditions.

Data Quality

The reliability of the results of any analysis of this nature will depend inevitably on the quality and rigor of the data upon which it is based. It is important to recognize the deficiencies of the available datasets on SOC stocks in AFS, upon which the analysis was based. As detailed by Nair (2012), AFS, by their very nature, are extremely site- specific, and are characterized by high variability in site-specific features such as soil type, plant species and density, and management practices. The difficulties arising from this diversity and complexity are confounded by the heterogeneity caused by lack of standardized methodologies, and sometimes failure to follow even available ones.

Equally disturbing is that some studies do not report the important parameters such as soil bulk density, and the methods followed in sampling and analysis of soils and computation of data. Common problems associated with soil sampling include (lack of) statistical rigor in adherence to replication and randomization, definition of sampling depths, lack of uniformity between soil depth and soil-horizons, method of sample preparation, non-randomized nature of repeated measures from same sites while sampling different vertical and horizontal intervals, and so on. Furthermore, differences in SOC stock improvement on adopting AFS are also attributable to the intrinsic properties of soil such as soil type, mineralogy, clay content, texture, and land-use history. Thus, adaptation of a specific AFS over Agriculture does not necessarily guarantee improvement of SOC stocks and could be regarded as case specific. The extent of seriousness of these deficiencies of the study and their impact on the results presented cannot be gauged at this stage.

Methodological Challenges

During the past three decades, studies investigating C stocks under agroforestry pointed out methodological challenges related to sampling, analysis, computations, and interpretation of results (Nair, 2011). Agroforestry systems are characterized by high variability in site-system features, soil type, tree species and management practices pertaining to geographical locations, which produced an inconsistent estimate of agroforestry C stocks (Montagnini & Nair, 2004; Nair, 2012, 2011). Lack of standardized methodologies may increase the heterogeneity of dataset, and assessment of the consistency of effects across studies is an essential part of meta-analysis. Nair (2012,

2011) pointed out several methodological issues related to SOC measurement and assessment in agroforestry; these included: sampling depth (sampling is usually limited to surface soils, leaving out more stable subsoil C); pseudo-replication (use of pre- existing field plots of the same contiguous experimental unit without true replicates of treatments: we addressed this by log-transformation of data to improve precision of interpretation); repeated measures (when time and space are considered as treatments, they are not usually assigned randomly, which bypasses the assumption of univariate analysis. Furthermore, unavailability of bulk density (BD) values is another major issue.

Soil C stock is typically expressed in mass per area (Mg C ha-1), which is obtained by multiplying mass per unit mass of soil (g C 100 g-1 soil), with the BD expressed in mass per volume of soil (g cm-3 or Mg m-3), and with soil sampling depth (cm). Some reports on soil C stock in AFS either do not report BD or do not follow any uniform depth classification.

All the above-mentioned challenges were encountered during the meta-analysis data collection and analyses, especially issues related bulk density and missing SD, SE

and variances. For the studies that we used in the analysis that had not reported BD, the information was obtained by contacting the authors who graciously obliged to provide the data. The lack of clear and complete data about the processes associated with C dynamics and storage resulted in unrestrictive criteria for data inclusion, leaving out some potentially important variables, such as soil clay content. As suggested by

Nair (2011, 2012), more accuracy in description of how the data were collected, analyzed, and managed is needed. A unified methodological approach will surely benefit future agroforestry research.

Conclusions

Our analysis showed evident trends of differences between AFS and other comparable land-use systems (Agriculture, Forestry, and Pasture) in terms of SOC stocks across various agroecological regions. Overall, while SOC stocks were higher under AFS compared with sole-crop agricultural systems under comparable conditions, the stocks were generally higher under Forestry. Comparing AFS vs. Agriculture,

Silvoarable systems are highly effective in improving SOC stocks in the Mediterranean region, while practices like Multistrata systems in the lowland humid region are highly effective in increasing SOC stocks in soils even up to 100 cm depth; indeed, these

Multistrata systems were comparable to Forests in terms of their SOC stocks. The AFS vs. Pasture comparisons demonstrated that AFS like Silvopasture and Agrosilvopasture are effective in improving SOC stocks in the arid and semi-arid regions, but not so in the temperate regions. The SOC stock improvements on adopting AFS depends also on the age of management practices, systems aged between 10–20 years being considerably more effective in improving SOC stocks than the relatively juvenile systems of less than

10 years of age.

The advantages of agroforestry as a sustainable land management approach and the importance of soil carbon sequestration in mitigating climate change as well as enhancing soil productivity have been well established. The important role of AFS in soil

C sequestration, though widely recognized, quantitatively convincing results based on rigorous studies undertaken under a variety of land-management scenarios on this aspect are still limited. In that context, the impetus this study gives for more rigorous, precise, and comprehensive studies of this nature in the future is as significant a contribution as the results generated.

Table 3-1. Major agroforestry systems covered in the meta-analysis. Agroforestry systems Description (AFS) Agrisilviculture Alley Cropping: Fast-growing, preferably leguminous, woody species grown in crop fields; the woody species pruned periodically at low height (<1.0 m) and the biomass added to the alleys between shrub rows.

Improved Fallow: Fast-growing, preferably leguminous, woody species planted and left to grow during fallow phases between cropping years for site improvement; woody species may yield economic products Multipurpose Trees (MPTs): MPT are scattered haphazardly or planted in some systematic arrangements in crop or animal production fields; trees provide fruits, fuelwood, fodder, timber, etc.

Multistrata Systems Homegardens: Multistory combinations of various trees and crops in homesteads; livestock may or may not be present. Shaded Perennial Systems: Growing shade-tolerant species such as cacao and coffee under or in between overstory shade, timber, or other commercial tree crops. Typically found in the LHT region. Silvopasture Integrating trees in animal production systems, cattle grazing on pasture under widely spaced or scattered trees. Found in all regions.

Agrosilvopasture Tree-livestock-crop admixed, crop-animal-wood integrated production, woody hedgerows, green manure. Parkland AFS: Parkland AFSs include scattered MPTs on farmlands. Found in the ASA, TEM and MED region. Silvoarable Systems Traditional European AFS, practiced in the TEM and TEM+ MED region. Widely spaced trees are intercropped with arable crops in these systems.

Source: Nair (1993).

Table 3-2. Number of data points across different agroecological regions and agroforestry systems included in the meta-analysis on soil organic carbon stocks. Agroforestry Systems ASA LHT MED TEM Total (n) / Agroecological Regions Agrisilviculture 30 274 15 43 362 Agrosilvopasture 9 59 4 22 94 Silvopasture 67 65 25 45 202 Multistrata Systems n.a 168 n.a n.a 168 Silvoarable Systems n.a n.a 27 1 28 Protective Systems n.a n.a n.a 4 4 Overall 106 566 71 115 858 Note: ASA=Arid and Semiarid; LHT=Lowland Humid Tropics and subtropics; MED=Mediterranean; TEM=Temperate agroecological regions respectively; n.a = not available. Total number of peer-reviewed publications selected: 78 Total number of data points (studies) used: 858 Variable reported: SOC (Mg C ha-1)

Table 3-3. Agroecological features and system characteristics of various agroforestry systems included in the meta-analysis Agroecological AFS practiced Major soil types Mean Mean annual Region (Terms used in annual precipitation FAO system in temp. (mm) italics) + (°C) Arid and Agrisilviculture, Aridisols. 16–34 50–180 semiarid (ASA) Agrosilvopasture, Regosols, Silvopasture Ferrasols, Lixisols

Lowland Humid Multistrata Oxisols, Tropics and Systems, Ultisols, subtropics Agrisilviculture, Alfisols, 18–30 ≥3600 (LHT) Agrosilvopasture, Vertisols, Silvopasture Cambisols

Mediterranean Agrisilviculture, Umbrisols, 0–18 350–900 (MED) Agrosilvopasture, Vertosols, Silvopasture, Luvisols Silvoarable Systems

Temperate Agrisilviculture, Umbrisols, 0–30 300–850 (TEM) Agrosilvopasture, Luvisols, Silvopasture, Vertosols, Protective Systems Spodosols, Andosols

Table 3-4. Summary of the meta-analysis on percentage changes in soil organic carbon (∆SOC%) stock up to 40 cm depth across Forest-Agroforest-Agriculture/Pasture/Uncultivated Land continuum in different agroecological regions around the world Agroecological Region Agroforestry System Percentage changes in (ref. Table 3-3) (AFS) soil organic carbon stock AFS AFS AFS AFS vs. # data points in vs. vs. vs. Uncultivated meta-analysis Agriculture Forest Pasture land ASA (Arid and semiarid) Agrisilviculture –7.69 n.a. 72.5 n.a. 6

Agrosilvopasture 20.2 –27.6 n.a. n.a. 4 Silvopasture n.a. 28.1 n.a. 10 LHT (Lowland Humid Multistrata syst. 24.1 12.6 n.a. n.a. 25 Tropics and subtropics) Agrisilviculture 33.6 –6.9 13.9 n.a. 39 Agrosilvopasture 5.9 –6.8 –18.1 n.a. 16 Silvopasture n.a. n.a. 40.01 22.6 11 MED (Mediterranean) Agrisilviculture 17.1 n.a. n.a. n.a. 2 Silvoarable 1.4 n.a. n.a. n.a. 12 Silvopasture n.a. n.a. –1.6 n.a. 2 TEM (Temperate) Agrisilviculture 18.45 –14.7 n.a. n.a. 19 Agrosilvopasture n.a. n.a. n.a. – 45 4 Silvopasture n.a. –15.5 n.a. 7.34 7 Note: Positive values indicate increase in SOC% stocks under AFS, and negative values indicate decrease in SOC% stocks under AFS, compared to the system concerned; n.a = not available

Table 3-5. Summary of the meta-analysis on percentage changes in soil organic carbon (∆SOC%) stock up to 60 cm depth across Forest-Agroforest- Agriculture/Pasture/Uncultivated Land continuum with respect to major agroforestry systems in different agroecological regions around the world Agroecological Agroforestry Percentage changes in Region System (AFS) soil organic carbon (ref. Table 1) (%∆SOC) stock AFS vs. AFS vs. AFS vs. # data Agriculture Forest Pasture points ASA (Arid and Agrisilviculture –9.9 n.a. 29.5 13 semiarid) Silvopasture n.a. n.a. 16.3 8 LHT (Lowland Multistrata 40.6 20.4 n.a. 35 Humid Tropics syst. and subtropics) Agrisilviculture –9.3 n.a. 3.8 20 Silvopasture n.a. n.a. –3.0 8 MED Silvoarable 8.5 n.a. n.a. 7 (Mediterranean) TEM Agrisilviculture n.a. –10.8 n.a. 3 (Temperate) Silvopasture n.a. n.a. –2.7 4 Note: Positive values indicate increase in SOC% stocks under AFS, and negative values indicate decrease in SOC% stocks under AFS, compared to the system concerned; n.a = not available

Table 3-6. Summary Results of Statistical Analyses: Mixed Effect Models, Meta- regression ANOVA and Tests for Symmetry Parameters/ Control for Mixed Effect Models: AFS vs. Descriptors Agriculture Forest Pasture Uncultivated land Depth p=0.36 p = 0.02 p =0.002 p =0.06 Region p = 0.0005 p =0.8 p <0.038 p =0.0006 Region: Type of p =0.0065; p =0.02 p <0.001; p =0.03; AFS Age of p =0.010 p =0.4 p =0.59 p =0.017 Management Funnel Plot Symmetric Symmetric Symmetric Symmetric Visual Assessment Rosenthal’s Fail- N= 1034204 N= 1769 N=204125 N=157473 Safe Number Orwin’s fail-Safe N = 399 N= 131 N=252 N=77 Number

Figure 3-1. Percentage changes in soil organic carbon (∆SOC%) stock (0–20cm) between agroforestry systems when compared to forest or agricultural land- use or pastures in the arid and semiarid (ASA) region (Agroforest vs. Agriculture, Forest vs. Agroforest, Agroforest vs. Pasture).

¶The numbers in parentheses indicate the number of observations

¶¶ The square in the center indicate the mean. The whiskers indicate the 95% confidence interval of ∆SOC % values

Figure 3-2. Percentage changes in soil organic carbon (∆SOC%) stock (0–20cm) between agroforestry systems when compared to forest or agricultural land- use or pastures in the lowland humid tropics and subtropics (LHT) region (Agroforest vs. Agriculture, Forest vs. Agroforest, Agroforest vs. Pasture).

¶The numbers in parentheses indicate the number of observations

¶¶ The square in the center indicate the mean. The whiskers indicate the 95% confidence interval of ∆SOC % values

Figure 3-3. Percentage changes in soil organic carbon (∆SOC%) stock (0–100cm) between agroforestry systems when compared to forest or agricultural land- use or pastures in the Mediterranean (MED) region (Agroforest vs. Agriculture, Forest vs. Agroforest, Agroforest vs. Pasture).

¶The numbers in parentheses indicate the number of observations

¶¶ The square in the center indicate the mean. The whiskers indicate the 95% confidence interval of ∆SOC % values

Figure 3-4. Percentage changes in soil organic carbon (∆SOC%) stock (0–100cm) between agroforestry systems when compared to forest or agricultural land- use or pastures in the temperate (TEM) region (Agroforest vs. Agriculture, Forest vs. Agroforest, Agroforest vs. Pasture).

¶The numbers in parentheses indicate the number of observations

¶¶ The square in the center indicate the mean. The whiskers indicate the 95% confidence interval of ∆SOC % values

Figure 3-5. Summary of results of a global meta-analysis showing percentage changes in soil organic carbon (∆SOC%) stock between Agroforestry systems (AFS) vs. Agriculture/Forestry/Pasture in four soil depth classes (0–20, 0– 40, 0–60, and 0–100 cm).

Note: Positive values represent increase in SOC % under AFS, and negative values represent reduction in SOC under AFS, compared to the system concerned.

CHAPTER 4 DEPTH-WISE DISTRIBUTION OF SOIL-CARBON STOCK IN AGGREGATE-SIZED FRACTIONS UNDER SHADED-PERENNIAL AGROFORESTRY SYSTEMS IN KARNATAKA, INDIA

Introduction

Globally, land-use changes associated with forestry and agriculture, such as deforestation and subsequent conversion to agriculture are estimated to be responsible for 25% of the accumulated GHG (greenhouse gas) emissions (Ciais et al., 2013).

Adoption of tree-based land-use systems has been recognized as a strategy for addressing global terrestrial carbon stock.

The potential of agroforestry systems (AFS) involving purposeful integration of trees within agricultural practices to sequester more soil carbon than conventional, monoculture systems has been well demonstrated (Montagnini and Nair, 2004; Nair et al., 2009; Nair et al., 2010). A major source of C input to AFS is the substantial amounts of tree litter and tree roots added to the soil often on a continuing basis. Agroforestry systems also facilitate better use of land resources, space and nutrient uptake by niche differentiation and effective resource partitioning. This further cascades into improved carbon inputs within the ecosystem (Thakur et al., 2015). Soil organic carbon (SOC) turnover rates in AFS vary significantly depending on a number of factors. These include plant species composition, spatial arrangements through effects of microclimate and litter-quality variations (Cao et al., 2016; Liu et al., 2011), land-use and resource management (Cadotte, 2013; Tilman et al., 2012), and soil texture and aggregation

(Cao et al., 2016; Six et al., 2002).

Among the various AFS, the shaded perennial systems, in which coffee (Coffea arabica, Coffea canephora), tea (Camelia sinensis), cacao (Theobroma cacao) and

other shade-tolerant perennial species are cultivated under or among tall-growing shade trees are reported to have a high potential for sequestering C in soil (Monroe et al.,

2016). This is based on the notion that the accumulation and subsequent turnover of leaf litter, roots, and woody material from the shade species as well as the lower-story species provide a continuous supply of organic materials to the soil (Beer, 1988;

Oelbermann and Voroney, 2007; Oelbermann et al., 2006). On an average, shaded

AFS receive a continuous deposition of plant litter as high as 10 Mg ha-1 yr-1 in some cases (Beer et al., 1998) and these systems encounter very little removal of C from the system via harvested products. The SOC stocks under shaded perennial AFS are generally high (≈150 Mg C ha-1) in most coffee and tea growing regions of the world

(Abou Rajab et al., 2016; Valencia et al., 2014).

The macroaggregates demonstrate higher SOC concentration than microaggregates and silt and clay fractions; at the same time, macroaggregates are more sensitive to the effects of land-use changes (Bronick and Lal, 2005; Tisdall and

Oades, 1982) while SOC stored in the silt and clay (<53 µm) are securely held and are

physically protected from the effects of land-use changes (Bronick and Lal 2005; Lal,

2004). Given the importance of size fractions in SOC storage, several studies on the potential for C sequestration in soils under AFS have emphasized the importance of determining the extent of C storage in different aggregate classes at deeper soil depth up to 1m (Haile et al., 2010; Haile et al., 2008; Howlett et al., 2011; Saha et al., 2010;

Tonucci et al., 2011).

Soil depth is another key aspect of SOC stock studies as the main source of belowground carbon comes from roots (Monroe et al., 2016). The shaded perennial root systems and the associated shade trees penetrate deep into soil (Monroe et al., 2016).

It has been well recognized that there is substantial input of organic matter to the subsoil horizons via plant roots and its exudates, dissolved organic matter and bioturbation processes.(Rumpel and Kögel-Knabner, 2011). Moreover, subsoil C is often considered more important than topsoil C in terms of source vs. sink relations for

CO2 because carbon accumulated in subsoil is less impacted by disturbances induced by land use changes (Batjes, 1996; Lorenz and Lal, 2005).

Some research results are available on depth-wise distribution of SOC stock in soil aggregates under AFS in different ecological regions (Chen et al., 2017; Gama-

Rodrigues et al., 2010; Haile et al., 2008; Haile et al., 2010; Saha et al., 2010; Tonucci et al., 2011). These results indicate that the extent of soil C sequestration will depend on several site-specific factors including soil type, land-use system, and soil-depth class, with strong interactions among these factors. It is important, however, to extend such studies to a wide spectrum of land-use systems in order to arrive at widely applicable conclusions on the role of AFS in soil C sequestration vis-à-vis other

ecosystem services. The underlying hypothesis of this study is that deep rooted, tree- based shaded perennial AFS, could help improve soil aggregation and C storage relative to Homegarden systems which have less proliferating root systems. The study reported here was undertaken in this context to assess the management effects of shaded perennial AFS on water stable soil aggregates, soil aggregate associated carbon and its depth-wise storage within different soil aggregate classes in the Western

Ghats region of India, a Biodiversity Hot Spot and World Heritage Centre.

Materials and Methods

Study Location

The study was conducted in Devon Tea and Coffee plantations, a privately- owned enterprise in Chikmagalur district, Karnataka, in southern India (Figure 4-1), nestled within the Biodiversity Hotspot of the Western Ghats (Myers et al., 2000) at an altitude of 1040 meters above sea level (12°54’42’’– 13°53’53’’N; 75°04’46’’–

76°21’50’’E). The district receives a mean average annual precipitation of 2400 mm; it has a mean average temperature of 26°C and an average humidity of 92%. The rainy season spans between May and October, with a distinct dry season from November to

April. The district has about 2,100 km2 (30% of land area) under native forests. The soils of the study site, classified as Ultisols and Alfisols, and are well drained mixed alluvial and clay to silty clay in texture. One aspect of the biodiversity of the region is the vast diversity of trees and the predominance of shade-grown coffee. As of the year

2017, about 75% of India’s coffee production takes place in this region (Coffee Board of

India), all of which is grown under a wide variety of shade/timber trees, often in association with other shade-loving spices such as cardamom (Elettaria cardamomum), and black pepper (Piper nigrum). The common shade trees include fast-growing and

nitrogen-fixing shade trees like Erythrina spp, Gliricidia sepium and introduced fast growing trees like Grevillea robusta, Toona ciliata besides native timber species like

Acrocarpus spp, Albizia lebbeck, Melia dubia, Terminalia paniculata. Tea is also grown under the shade of G. robusta in the region, though to a lesser extent than coffee, accounting for only 2.5% of India’s total production as per statistics generated in the year 2016 (Tea Board of India).

Land-use Systems

Five different land-use systems (treatments) were selected for the study (Figure 4-2):

1. Coffee + Grevillea: Coffee bushes under Grevillea robusta trees; >50% tree cover; age of plantation ~70 years; area per plot per replicate ~ 1.5 ha.

2. Coffee + Mixed Shade: Coffee bushes under a mixture of shade trees including Acrocarpus spp, Albizia lebbeck, Artocarpus hirsutus, Erythrina sp., Ficus racemosa, Melia dubia, Terminalia paniculata, Toona ciliata; <50% tree cover; age of plantation ~55 years; area per plot per replicate ~ 1.5 ha.

3. Tea + Grevillea: Tea bushes under Grevillea robusta trees; >50% tree cover; age of plantation ~85 years; area per plot per replicate ~ 1.5 ha.

4. Homegarden: Farmer’s crop field with a mixture of annual crops and various fruit trees of uneven age; total area ~ 0.5 ha.

5. Forest: Native, moist, deciduous forest from an adjoining site.

The Grevillea trees had been established at 6 x 6 m spacing. The forest stand was mostly fragmented natural reserve area under Koppa Forest division with Albizia amara, Artocarpus integrfoli, Bombax ceiba, Careya arborea, Chukrasia tabularis,

Pterocarpus marsupium, as some of the dominant tree species. The soils in coffee AFS are typically hand tilled once every 2–3 years. The Homegarden in the study site contained agricultural crops such as cassava (Manihot esculenta), ginger (Zingiber officinale), and various vegetables under the mixed canopies of overstory species including banana (Musa spp.), papaya (Carica papaya), fruit tress (e.g., guava, Psidium

guajava), and areca palms (Areca catechu). Prior to the establishment of Homegarden, the area was under natural forests, which was cleared for agriculture but not under

‘slash and burn’ system. The forest site for the study was relatively undisturbed and can be classified as moist deciduous type reserve forests under the sub-humid climate zone.

Soil Sampling

Soil samples were collected following the pattern for a randomized complete block design (RCBD) from selected plots of all five land-use systems. Each sampling site measuring 1m x 1m area was dug up to collect representative soil samples from four depth classes: 0–10, 10–30, 30–60, and 60–100 cm. Samples from four randomly selected sampling sites per plot were composited to form one replicate. Four such replicates were collected from four plots for each treatment (land-use system), giving a total of 80 samples (5 treatments x 4 depth classes x 4 replicates per plot). At each sampling location, the selected 1 m x 1m area was dug to more than one-meter depth to collect samples form the four depth classes. For determination of bulk density of soil at each depth class, a steel cylinder of known volume was inserted horizontally on the wall of the pits at the center of each depth class; the soil inside the cylinder was collected, weighed, dried, and weighed again. All samples were air-dried at room temperature and transported to the laboratory of the National Bureau of Soil Science and Land Use

Planning (NBSS-LUP), Indian Council of Agricultural Research, Bengaluru, India, for further processing. The soil samples were sieved (2 mm sieve, i.e., #10 U.S. Standard

Testing Sieve) at the NBSS-LUP soils laboratory. The portion of soil that did not pass through 2 mm sieve was separated and discarded; the 2 mm-sieved soil samples,

hereafter referred to as the whole soil, were bagged, and shipped to the University of

Florida, Gainesville, FL, USA, for further analyses.

Soil Preparation and Analysis

Sub samples of the whole soil were fractionated manually into three aggregate size classes >250 μm , 250–53 μm and <53 μm at the Environmental Soil Chemistry

Lab, Soil and Water Sciences Department, University of Florida, according to the methods described by (Elliott, 1986), further modified by (Six et al., 2002) that had been adapted and followed by previous researchers (Gama-Rodrigues et al., 2010; Haile et al., 2008; Haile et al., 2010; Saha et al., 2010; Tonucci et al., 2011). The method consisted of wet sieving using disruptive forces of slaking followed by sieving through a series of two sieve sizes (250 and 53 µm) to obtain the three fraction size classes: macro (>250 µm), micro (250–53 μm), and silt- and clay- sized fraction (<53 µm).

A 100 g sample of 2 mm sieved air-dried soil was placed in a 500 mL beaker.

Distilled water (250 mL), enough to completely cover the soil, was poured into the beaker to initiate the process of slaking for disintegrating the water unstable aggregates in soil, leaving only water stable aggregates to proceed with further analysis. The soil solution was wet-sieved manually by moving the sieve up and down about 5 cm each,

100 times in two minutes. The soil fraction that did not pass the 250 μm sieve was backwashed with a distilled water-filled wash bottle into a glass beaker. The remaining soil solution was next poured over the 53 μm sieve (#270 U.S. Standard Testing Sieve), and the procedure was repeated. The three soil fractions >250 μm, 250–53 μm, and

<53 μm, were dried at 65°C, weighed, ground for homogenization using a QM-3A High

Speed Vibration Ball Mill for 10 min, and stored in individually sealed and labeled plastic

bags for C analysis. The average recovery of initial soil mass was 98%. The weight of oven-dried samples (dry weight) for each aggregate size was noted.

The whole soil, not treated with the slaking or fractionation procedure, was dried and ground for homogenization. Soil samples were analyzed in a Carlo Erba NA1500

CNHS elemental analyzer for carbon and nitrogen percentages. The machine standard used was a low organic content soil standard (OAS). To ensure accuracy of results, the analyzer was calibrated using standards of a known C concentration, and determinations of C content for 5% of samples were repeated. Soil pH was determined in a 1:10 soil: water suspension. The total C stored to a meter depth is the sum of the C stored at each of the depths within the soil profile. The soil samples were shipped to the

Waters Agricultural Lab, Georgia, USA for textural analyses by the hydrometer method

(Gee and Bauder, 1979). Soil characteristics are elaborated in Table 4-4.

The C storage was calculated as:

퐶푠푡표푐푘 = 퐶푐표푛푐푒푛푡푟푎푡푖표푛 × 퐵퐷 × 퐷푒푝푡ℎ × 푊푒푖𝑔ℎ푡 퐹푟푎푐푡푖표푛 (4-1) where,

−1 Cstock = C storage expressed in Mg ha (per cm soil thickness unless specified otherwise) in each fraction for a given depth

−1 Cconcentration = C concentration in size fraction (g 100 g ) of that fraction size

BD = Bulk density (Mg m-3)

Depth = Depth of soil profile (cm)

Weight Fraction = weight of the fraction in the whole soil as a ratio (dimensionless).

The depth-wise distribution of different soil-fraction-size classes is given in Table 4-2.

Statistical Analyses

To test the effects of Treatment, Fraction size and Depth on SOC stocks, full fixed effect model, which is a linear regression model was fitted using R studio version

3.4.2 with multicompview package (Hothorn et al., 2008). Full fixed effect models were run with all three interaction terms: Treatments, Depth and Fraction size with reference categories Forest for treatments, the top soil profile 0–10 cm for depth and aggregate size >250µm as fraction size, where the response variable was SOC stocks (Mg C ha-1).

To elucidate specific interaction effects of Treatment, Fraction size and Depth on changes in SOC stock, individual first order (two factor) and second order (three factor) interaction effects analysis of variance (ANOVA) were also performed.

The following equation illustrates the first and second order interactions:

푌푖푗푘푙 = 휇 + 훼푖 + 훽푗 + 훾푘 + (훼훽)푖푗 + (훽훾)푗푘 + (훼훾)푖푘 + (훼훽훾)푖푗푘 + 휀푖푗푘푙 (4-2) where,

푌푖푗푘푙 = SOC stock (in log scale), for the observation corresponding to the 푖th treatments and the 푗th depth level and 푘th fraction size in the 푙th replicate,

푖 = 1, ⋯ , 5, 푗 = 1, 2, 3, 푘 = 1, ⋯ , 4, 푙 = 1, ⋯ , 4.

휇 = overall effect,

훼푖 = effect due to the 푖th treatments (site)

훽푗 = effect due to the 푗th depth level,

훾푘= effect of the 푘th Fraction size.

(훼훽)푖푘 = interaction effect of the 푖th treatments and the 푗th depth level

(훽훾)푗푘= interaction effect of the 푗th depth level and the 푘th fraction size

(훼훾)푖푘 = interaction effect of the 푖th treatments and the 푘th fraction size

(훼훽훾)푖푗푘= interaction effect of the 푖th treatments and the 푗th depth level and 푘th fraction size

휀푖푗푘푙 = normal random error

Tukey’s HSD was performed to compare mean differences between land-management practices on SOC in whole soil, macro sized, micro sized and silt- and clay sized fractions for all sites. All statistical tests were performed in R studio version 3.4.2. The R packages, ggplot2 ( Wickham 2017), lsmeans (Lenth, 2016), reshape2 (Wickham,

2007) and multicompview (Hothorn et al., 2008) were used for additional analyses.

Results

Soil Organic Carbon Stock up to 1m Depth

The total SOC stocks in whole soil (for all depth classes and fraction sizes combined) down to 1 m depth varied among different treatments (land-use practices)

(Figure 4-3), with the highest under Forest (172.3 Mg C ha−1) and lowest under

Homegarden (89.3 Mg C ha−1). The overall results for SOC stock were in the following order: Forest > Coffee + Grevillea > Tea + Grevillea > Coffee + Mixed Shade >

Homegarden (Figure 4-3). The Coffee + Mixed shade site had 64.4% and 35.8% less carbon stock than Forest and Coffee + Grevillea sites respectively, whereas the AFS sites, Coffee + Grevillea and Tea + Grevillea had 21% and 49% lesser SOC stocks than

Forest. Compared to sparsely shaded Homegarden, Coffee + Grevillea site had 59.5% higher SOC stock.

SOC Stock Distribution in Whole Soil under Various Land-Use Systems and Depth Classes

The SOC stocks in the 0–10 cm soil depth under different land-use systems were in the following order, with corresponding quantities (Mg C ha-1): Forest (57.5) ≥ Coffee

+ Grevillea (50.3) > Tea + Grevillea (41.0) > Coffee + Mixed Shade (38.4) >

Homegarden (19.1) (Figure 4-4).

In the 10–30 cm soil depth, the C stocks were in the order (Mg C ha-1): Forest

(49.7) ≥ Coffee + Grevillea (43) (Figure 4-4), whereas in the 30–60 cm soil depth, the order was (Mg C ha-1): Forest (37.8) ≥ Coffee + Grevillea (29.7) > Tea + Grevillea (23.9)

> Homegarden (20.1) (Figure 4). Overall, SOC stocks decreased with increasing soil depth. Beyond 60 cm, the highest SOC stocks were under Forest (27.2 Mg C ha-1) and the lowest under Coffee + Mixed Shade (14.4 Mg C ha-1) (Figure 4-4); Tea + Grevillea had significantly higher SOC stocks (22.4 Mg C ha-1) than the other land use types in this depth class.

Soil Organic Carbon in Macro-Aggregates (>250 μm)

The total SOC contained in the macroaggregates in the entire 1 m depth profile were 119.1, 76.8, 71.6, 57.6 and 45.5 (Mg C ha-1) in Forest, Coffee + Grevillea, Tea +

Grevillea, Coffee + Mixed Shade and Homegarden treatments, respectively (Figure 4-

5), with the highest amount under the Forest. In the uppermost depth (0–10 cm) and second upper depth (10–30 cm), Homegarden had the least SOC content, while Coffee

+ Grevillea had the least SOC stocks within macroaggregates in the lower depth classes (30–60 cm) and (60–100 cm), at 10.5 Mg C ha-1 and 6.5 Mg C ha-1 respectively.

Among the three aggregate-size classes, the highest proportion of total C was retained in the macroaggregates. In the second lower depth class of 30–60 cm, both the coffee

AFS (Coffee + Grevillea, Coffee + Mixed Shade) and Forest had statistically similar mean SOC stocks. In the lowest depth class (60–100 cm), Forest and Tea + Grevillea contained the highest SOC stocks with overlapping mean SOC stocks.

Soil Organic Carbon in Micro-Aggregates (250–53 μm)

The total SOC contained in the microaggregates in the entire 1 m depth profile was highest under Coffee + Grevillea system (Figure 4-6). Depth-wise, in the 0–10 cm,

Coffee + Grevillea, Coffee + Mixed Shade had higher SOC stocks (8.3 Mg C ha-1and 7

Mg C ha-1) while Forest had 5 Mg C ha-1. In 10–30 cm depth class, the mean SOC stocks were not significantly different among the treatments. At the lowest depth class

(60–100 cm) too, the treatments did not differ significantly except Forest which had

7.2 Mg C ha-1.

Soil Organic Carbon in Silt and Clay Fraction (<53 μm)

The SOC contents in the silt + clay fraction was significantly higher under Coffee

+ Grevillea, especially in the 0–10 cm depth (Figure 4-7). While Coffee + Grevillea had significantly higher SOC stocks in the top two soil layers (0–10 cm and 10–30 cm), there was no difference between Coffee + Grevillea, Coffee + Mixed shade and Forest, in terms of SOC content in the 30–60 cm layer. In the lowest soil depth (60–100 cm), however, SOC held in the smallest fraction (<53 μm) was highest under Forest (Figure

4-7).

Interaction Effects and Analysis of Variance (ANOVA)

The Treatments (p=0.016), Soil Depth (p<0.0001), and Fraction size (p<0.0001), and all interactions has significant effect on total SOC stocks (Table 4-3, Table 4-4).

Individual effect of treatments (land-use systems) on SOC stocks, compared to Forest as reference, was statistically similar for the two grevillea systems (Coffee + Grevillea and Tea + Grevillea). On the other hand, the other two treatments, Coffee + Mixed

Shade (p=0.03) and Homegarden (0.006) had significantly different effect on SOC stocks compared to the reference treatment, Forest (Table 4-4). Compared to the

reference depth class 0–10cm, the effect of all other depth classes on SOC stocks was significantly different (p<0.0001). Similarly, compared to macroaggregates (>250μm),

SOC stocks were significantly different in the other two fraction size classes, microaggregates, (250–53 µm) and, silt + clay (<53 μm), in improving (Table 4-4). The

Interaction effects of Treatment × Fraction size was also significant (p<0.0001).

Additionally, the effect of Treatment × Fraction size interaction on SOC stocks was significant for individual treatments (Treatment × Fraction size: Coffee+ Grevillea,

Coffee+ Mixed Shade, Homegarden and Tea + Grevillea) (Table 4-4).

Discussion

Land-Use System – Soil Depth Class – Aggregate-Sized Interactions in SOC Storage

It is imperative that analyses and interpretations of all interactions are included while reporting research results (Vargas et al., 2015). For this study, we adopted a full fixed effect model and performed an analysis of variance (ANOVA) to elucidate the effects of the factors Treatments, Depth and Fraction size on SOC stocks and all possible interaction effects at all levels between/among the variables (Table 4-3 and

Table 4-4). The significant Treatment × Fraction size interaction (Table 4-4) indicates that the effect of Treatments on SOC stocks is influenced by the type of Fraction sizes.

Considering the macroaggregate (>250 µm) fraction size as the reference, the other two fraction sizes showed significant different main effect on SOC stock, which indicates that there were differences between fraction sizes in the amount of SOC stocks stored within them (Table 4-4), or, the amount of SOC stock is influenced by land-use system and fraction size. Therefore, SOC stock variations at different depth classes cannot be attributed to the relative levels of various fraction sizes; for example, preponderance of

a specific fraction size such as macroaggregates or silt + clay is no indication of higher or lower levels of total SOC stock at any depth class. The interaction effects of

Treatment × Fraction size was statistically significant for all treatment categories indicating the stability of the effect of Treatment on SOC stocks as Fraction size varied

(Table 4-4). Thus, the amount of SOC stock at any depth class was influenced by land use system and soil fraction size class, whereas effect of fraction size class was not different at different depth classes.

SOC Stocks in Whole Soil under Different Land-Use Systems up to 1m Depth

The SOC stock in a defined soil volume at any time represents the net difference between input and output of organic carbon. Both inputs and outputs are strongly influenced by vegetation type, climate, ecosystem productivity, soil aggregates, texture, and management practices (Six et al., 2002). Although our study did not quantify the organic matter input, a strong correlation between SOC stocks and higher organic matter input is expected in tree based, shaded perennial AFS, which often mimic forest- like ecosystem (Hombegowda et al., 2015).

Among the land-use systems investigated, SOC stocks were highest in Forest, lowest in Homegarden and intermediate for the other three shaded perennial AFS. The total amount of SOC in whole soil up to a depth of 1 m varied with land-use types

(Figure 4-3). The SOC stock under the Coffee + Grevillea was comparable to the values

(156 Mg C ha-1) reported by Hombegowda et al. (2015) under coffee AFS in Karnataka, where our study was located. All three land-use systems/treatments under shaded perennial AFS had significantly higher SOC stocks than the Homegarden systems

(89.3 Mg C ha-1). Total SOC stock decreased with soil depth in all land-use systems in this study (Figure 4-4), and has been observed for shaded coffee in other coffee

growing countries (Tumwebaze and Byakagaba, 2016; Tumwebaze et al., 2012, in

Uganda; Dossa et al. 2008, in Togo; Ehrenbergerová et al., 2015, in Peru; Soto-Pinto et al., 2009, in Chiapas, Mexico; and, Hombegowda et al., 2015 in Karnataka, India).

Our results are comparable to SOC stock values reported under shaded AFS by

Dossa et al. (2008), who reported that shaded perennial coffee systems yield higher belowground carbon than open-grown (sun) coffee up to a depth of 25 cm. In Uganda,

Tumwebaze and Byakagaba (2016) estimated SOC stocks at 54 Mg C ha−1(0–30 cm) in

(Coffee + Fruit trees) while Ehrenbergerová et al. (2015) reported SOC stocks of 82.6

Mg C ha−1 (Coffee + Inga), 101.8 Mg C ha−1 (Coffee + Pinus), 96.6 Mg C ha−1 (Coffee +

Eucalyptus) up to 30 cm depth in Villa Inca, Peru. Soto-Pinto et al. (2009) found SOC stocks up to 121 Mg C ha−1 (0–30 cm) in shaded coffee AFS in Chiapas, Mexico. In shaded AFS, improved litterfall from shade trees compared to Homegarden and monoculture practices contributed to higher soil C in the top soil while these deep- rooted shade trees and perennials contributed to improved C stocks in deeper soil layers. Thus, shaded perennial coffee AFS have a significant potential to improve soil organic carbon compared to other systems like Homegarden and monoculture practices as noted in this study and other above-mentioned studies on similar AFS conducted globally.

Soil Organic Carbon in Aggregate-Size Fractions

Macroaggregates (>250 μm)

The macroaggregates (>250 μm) represented greater soil weight recovery (post wet sieving and oven drying) across all treatments and depth classes (Table 4-2), ranging from 67.3% under Forest to 47.6% under Homegarden. Six et al. (2002) also demonstrated greater (>250 μm) fraction dry weight recovery in forested and afforested

sites compared with tilled agricultural fields in Ohio. The SOC stock within macroaggregates in 0–10 cm was the highest under Forest (Figure 4-5). Even though the weight of soil after oven drying under Homegarden systems was 52.4% within this size fraction at 0–10 cm, SOC content was significantly low at 10.1 Mg C ha-1. At the

60–100 cm depth, within macroaggregates, no significant differences were noted in

SOC content between Forest and Tea + Grevillea (Figure 4-5). Greater mass recovery of macroaggregates in soil fractions lead to significantly greater C storage in this fraction, as these are known to have greater C per unit mass. Thus, increases in recovery of this fraction will lead to greater C storage (Six et al., 2000).

Increased storage of C in the macroaggregates is indicative of change in management activities, such as increased litter inputs from over-story shade and cessation of tillage, especially in surface horizons (Six et al., 2000). The macroaggregates are usually comprised of fresh residues of organic input. These fresh residues turn into intra- aggregate particulate organic matter and aid in formation of microbial-derived binding agents. Aggregate stability is strongly correlated with soil organic matter content

(Chaney and Swift, 1984). The organic matter inputs by litterfall from shade trees and perennials like coffee and tea are the source of macroaggregates formation in the soil

(Carter and Gregorich, 2010)..

The dry weight percentage of macroaggregates in the top soil was significantly higher in Forest (Table 4-2) as these are devoid of any management practices. All other treatments had similar dry weight percentage of macroaggregates in the top soil.

However, in the lowest depth (0–100 cm), AFS systems seemed to improve the weight of macroaggregates distribution across all treatments, the highest being Tea + Grevillea

at 51.3% which was very close to that of Forest (58.5%), (Table 4-2). Similarly, the SOC content under Tea + Grevillea was statistically similar to that of Forest (Figure 4-5). This indicates that Tea + Grevillea AFS improves macroaggregate formation with increasing depth, comparable to forest. Additionally, Tea + Grevillea also improves SOC storage

(similar to Forest) in the macroaggregates with increasing depth class compared to other systems (Figure 4-5).

The SOC contents in macroaggregates varies with the land-use changes within relatively “short” time spans (<100 years) especially in the top soil. Thus, the differences in SOC contents of macroaggregates among soils under forest in comparison to other treatments point to this. The forest has been in place for a long period of time (>100 years) and the Tea + Grevillea has been in place for over 85 years. Thus, the SOC content in Forest and Tea + Grevillea within macroaggregates was significantly higher than in the other treatments which have been in place under varying time scale ranging from 70 years to 40 years. Similar trends were reported in cacao and rubber AFS by

Gama-Rodrigues et al. (2010) and Chen et al. (2017). The C stock values can be considered as characteristic of each system. Thus, the differences in SOC stock in macroaggregates could also be a reflection of the “tree effect” in time.

Microaggregates (250–53µm)

Microaggregates are composed of diverse mineral, organic and biotic materials that are bound together by various physical, chemical and biological processes

(Totsche et al., 2017). Consequently, microaggregates can withstand strong mechanical and physicochemical stresses and survive slaking in water, allowing them to persist in soils for several decades (Totsche et al., 2017). The three-dimensional structure of soil microaggregates defines a rather stable and complex system of interconnected and

dead-end voids and pores of various size, shape and geometry. They provide an extremely heterogeneous, complex internal and external biogeochemical interface

(Chenu et al., 2000; Totsche et al., 2017). Organic C inside this class has lower decomposition rate than C stored within macroaggregates and can store C for a longer time than in larger-size fractions (Six et al., 2002). In other words, the micro-sized class would be in between the macro-sized class and the silt-and-clay-sized class in terms of

SOC stability.

Within microaggregates, in 0–10 cm depth class, Coffee + Grevillea and Tea +

Grevillea had the highest SOC at 13 and 11.5 Mg C ha-1 respectively. The SOC stock in microaggregates under the Coffee + Grevillea system was statistically similar to that of

Forest even up to 60 cm depth (Figure 4-6). The results showed that shaded AFS had relatively higher SOC storage within microaggregates even up to 100 cm (Table 4-2,

Figure 4-6). The carbon stored within microaggregates is on average older than the carbon stored in macroaggregates where the mean residence time (MRT) in macroaggregates ranged between 1–10 years while that in microaggregates ranged between 10 and 100 years (Fontaine et al., 2007; Totsche et al., 2017). Considering that MRT of organic carbon is intimately associated with stability of microaggregates, the C stored in microaggregates are more robust than that in macroaggregates, improved SOC storage within this fraction size under shaded AFS suggests relatively more secure and stable storage of C under AFS compared to monocrop systems and

Homegardens.

Silt and clay fraction (<53 μm)

The percentage distribution of soil fraction size classes under different land-use systems demonstrated abundance of silt-and-clay fraction in the soil under Coffee +

Grevillea at 25.6% on an average under all depth classes. However, the percentage distribution of soil fractions under Forest seemed to be skewed more towards macroaggregates at an average of 67.3% and least towards silt-and-clay fraction at an average of 10.9% (Table 4-2). The percentage distribution of fraction size classes to the silt-and-clay fraction increased with depth under all land-use types similar to Chen et al.,

2017; Gama-Rodrigues et al., 2010; Saha et al., 2010. Moreover, in all land-use systems, the percentage distribution of soil fraction was more in the macro-sized fractions at the upper depth classes with increasing proportion of the silt-and-clay fraction at the lower depth classes, and micro-sized fractions distributed more or less equally in the different depth classes (Table 4-2). Post wet sieving, the distribution of silt-and-clay fraction under Forest within 60–100 cm accounted to only 13.9 g while the

Coffee + Grevillea AFS reported 32.2 g (Table 4-2). Additionally, textural analyses revealed that the clay content within Forest was lower (36.4 g) compared to Tea +

Grevillea (56 g) and Coffee + Grevillea (44 g), (Table 4-1) which indicates that Grevillea based AFS promotes aggregate distribution especially improving the allocation within the smallest aggregate size classes. The impacts of shaded perennial, grevillea based

AFS on soil aggregates within silt-and-clay are also likely attributable to mechanisms, such as altered community composition and biomass, changes in organic matter inputs, and shifts in microclimate or soil construction (Chen et al., 2017).

The C stored in the silt-and-clay sized fraction is considered as the most stable form of C stored in soil. Entrapment and adsorption of organic carbon to mineral surfaces within this fraction class ensures its close contact to mineral surface and physical protection against enzymatic attacks (Lehmann, 2007; Totsche et al., 2017). It

was pointed out by Chenu and Plante (2006) that root exudates have similar effect.

Interaction between clay minerals and humic substances provide the stability of C encapsulated within the silt-and-clay sized fraction class. Clay particles create small pores (<1 μm), where organic matter can remain unreachable to microorganism. In addition, organic matter may also get combined with layer silicate clays

(e.g.,vermiculite) in their interlayers and form compounds highly resistant to decomposition.

Across all depth classes, Coffee + Grevillea showed significantly higher SOC stocks than Forest (Figure 4-7). The dry weight distribution of silt and clay fractions were in the order:

Coffee+Grevillea>Coffee+MixedShade>Tea+Grevillea>Homegarden>Forest.

As pointed out by Chenu and Plante (2006), the pressure exerted by roots of these deep rooted system as in shaded AFS lead to a reorientation of the mineral material, which also could have promoted aggregation in clay microstructures formation. The carbon stored in this fraction are more stable in the soil, associated with recalcitrant plant/microbial residue and clay minerals, and the C contained in it is relatively old. In our study, the oven dry weight percentage after wet sieving under this size fraction increased with increasing depth especially under Coffee + Grevillea (Table 4-2), further supporting the notion that these deep rooted shaded perennial AFS improved aggregation of silt-and-clay-sized fraction which are securely stored in soil sinks and less likely to be decomposed by microorganisms. This supports the conjecture that deep rooted, tree-based land-use systems like shaded AFS promote sequestration of C in silt-and-clay-sized fractions throughout all soil-depth classes (Figure 4-7).

SOC Distribution in Different Soil Aggregate Classes Studies

Results from our study indicate that AFS impact the distribution of soil C among different aggregate size fractions differently, and therefore the associated stability of soil

C. Similar results were also reported by Gama-Rodrigues et al. (2010), Haile et al.

(2008), Howlett et al. (2011), Saha et al. (2010), and Tonucci et al. (2011). The percentage distribution of silt-and-clay fraction in the soil under coffee AFS was the highest at 55.5 Mg C ha-1 which accounted for 30% of the total SOC stocks up to 1m depth. Coherent to this study, Gama-Rodrigues et al. (2010) and Saha et al. (2010) also reported higher SOC stocks in the silt and clay fraction at lower depth in shaded perennial cacao and Homegarden AFS, which further strengthens the claim that AFS promotes SOC stocks in the smallest aggregate fraction.

Studies conducted in the tropics by Gama-Rodrigues et al. (2010), Saha et al.

(2010) and Tonucci et al. (2011) reported that about 70% to 56% of SOC in the whole soil was located inside the macroaggregates at all soil depths studied (to 1 m), compared to 52.4% under coffee AFS in our study. Interestingly, Haile et al. (2008) and

Howlett et al. (2011) reported from subtropical Florida and Mediterranean conditions in

Spain, respectively, the predominance of macroaggregates under AFS (61% to 49.3%) and consequently higher SOC stock within macroaggregates. Continuous addition of organic matter through litter fall combined with limited or no tillage in the AFS treatments can be expected to promote macroaggregate formation and maintenance under shaded perennial AFS. The presence of high concentrations of fine roots in the surface soil up to 15 cm depth in these relatively undisturbed (no-till) soils and coarse roots at the subsurface soils up to 100 cm can also be expected to have contributed substantially to belowground C stocks in these shaded AFS (Cuenca et al., 1983) .

The model of aggregate hierarchy (HM), first proposed by Tisdall and Oades

(1982) stated that the addition of organic matter to soils results in the formation of clay and silt particles first. The formation of microaggregates and macroaggregates begin if the soil organic matter (SOM) binding capacity of the silt and clay fractions are saturated. This model demonstrates the tendency of an increased concentration with increasing aggregate-size classes as macro and microaggregates are composed of smaller silt-and-clay aggregates with organic binding agent (Eliot, 1986; Chen et al.,

2017). Our results showed similar trends as the HM model theory and Chen et al., 2017, with increase in aggregates both under Forest and other shaded perennial AFS (Figure

4-5). Thus, SOC stocks were higher within macroaggregates than microaggregates and silt-and-clay fractions (Figure 4-6 and Figure 4-7).

Rhizodeposition in Shaded Perennial Systems

In general, SOC stocks are greater in sites with diverse and intense vegetation communities. AFS, such as shaded perennial-crop systems, have higher potential for sequestering carbon in the soil compared to treeless systems (Montagnini and Nair,

2004 ; Nair et al., 2010; Nair, 2017). The variation in SOC stocks under shaded perennial systems may be attributed to continuous input of leaves, foliage and rhizodeposition by roots. The structure, proliferation of roots, and rooting depths also play a crucial role (Hombegowda et al., 2015). Rooting depth is very much a function of the soil type (Pierret et al., 2016). The roots of the coffee shrubs can extend up to 2 m in total length (Cuenca et al., 1983). Cannell and Kimeu (1971) have suggested that the greatest root concentration is in the 30–70 cm depth (in Kenya), which corroborates the view that higher SOC distribution in 0–60 cm of soil profile in our study was possibly due to higher rhizodeposition. In contrast, Grevillea trees have less number of feeder (fine)

roots and an abundance of coarse roots near the soil surface. The bulk of root mass for

Grevillea has been reported to be concentrated in the depth 45–70 cm depth-zone. In

Tea, the majority of the fine roots are usually found in the 0–45 cm depth (Niranjana and

Viswanath, 2008). The roots of Grevillea also penetrate deeper into the soil (2.4 m versus 1.5 m for tea roots and 1.7 m coffee roots, Niranjana and Viswanath 2008).

Our results also demonstrated high SOC stocks even in deeper soil layers >30 cm in Coffee + Grevillea and Tea + Grevillea systems (Figure 4-4). Substantial evidence of rhizodeposition in deeper soil layers from Grevillea is found in the literature. For example, Niranjana and Viswanath (2008) reported that bulk of root mass for Grevillea

(67 % of <2 mm, 66% of 2 to 5 mm roots and 94% of 5 to 15 mm roots) was located at a depth of 45–67.5 cm. In contrast, majority of the fine roots of tea were found in the 0–45 cm depth. The large SOC stock increases in the subsoil (Figure 4-7) indicate that belowground carbon inputs from roots and/or leaching of organic acids and soluble humus fractions to deeper soil layers are important processes for SOC accumulation

(Chenu and Plante 2006; Angst et al., 2016). Considering that trees like Grevillea in shaded AFS have deeper rooting systems than the overstory species of Homegarden,

(dominated by species such as papaya and banana that have sparse root systems and only relatively few large trees) and that often more than half of the carbon assimilated by trees is transported belowground for the proliferation of roots (Don et al., 2011;

Poeplau and Don, 2015; Angst et al., 2016), it is no surprise that SOC stocks were substantially high even at lower depth under these systems.

Acknowledging the ongoing controversy surrounding the use and role of

Grevillea robusta as a shade tree for coffee in the Biodiversity Hotspot of Western

Ghats and the purported interaction between Biodiversity and C sequestration, it behooves us to refer those issues in this study. It seems that there is a general expectation that all studies involving Grevillea robusta in the region, even if not designed to address those issues, should relate the results at least parenthetically to the “hot button” issues of biodiversity conservation and exotic vs. indigenous controversy. Among the treatments selected for this study, the two Grevillea-based

AFS (Coffee + Grevillea and Tea + Grevillea) had higher SOC stocks than Coffee +

Mixed Shade and Homegarden up to a depth of 1m (Figure 4-4). Although there is a prevailing pre-conceived notion about positive correlation between C sequestration and species diversity, the relationship between shade tree C stock and species diversity is not always significant (Richards and Mendez, 2014). The existing literature on this relationship is also quite nebulous, where some studies reported a positive relationship between SOC stocks and species diversity (Häger, 2012; Saha et al., 2009) while no such relationship was reported by others (Kirby and Potvin, 2007; Méndez et al., 2009).

In our study, even though Coffee + Mixed Shade and Homegarden had higher tree species diversity, these systems had lesser SOC stocks than Coffee and Tea AFS under Grevillea. The practice of planting Grevillea as a shade tree by relatively large- scale coffee growers in Karnataka has been severely criticized by conservationists who perceive a reduction of biodiversity of native species due to preferential planting of exotic trees (Garcia et al. 2010; Nath et al., 2016). Grevillea, although introduced more than 150 years ago into the region and having acclimatized well, is still denigrated as an exotic species, while coffee is also an exotic species sensu stricto, and no such exotic stigma is attached to other exotic species that have been introduced into the region

100

relatively recently. Thus, such controversies appear to be fueled more by factors other than scientific evidence.

Conclusions

Shaded perennial AFS, as compared to Homegarden, are more efficient at improving SOC stocks in deeper soil horizons and especially within silt and clay fractions in subsurface horizons. Interaction effects showed that shaded perennials like

Coffee + Grevillea and Tea + Grevillea are comparable to Forest in its contribution to

SOC stocks. These systems also improved soil aggregation and SOC within soil aggregates. Grevillea is a good choice as shade tree over coffee and tea and it significantly improved SOC stocks compared to mixed shade of native and non-native trees for our study area. The majority of SOC was present in macroaggregates throughout the soil layers up to 1 m depth, however, shaded AFS showed higher SOC than Forest in the silt and clay fractions. Although SOC inside the macroaggregates is more subjected to disturbance than that in microaggregates and silt- and-clay fractions, the extent of such disturbances is low in shaded perennial systems, therefore, C contained in this fraction can be expected to become more stabilized and sequestered over time. Thus, the coffee and tea AFS under Grevillea shade seem to play an important role in environmental protection by mitigating GHG emission through the storage of high amounts of well-protected organic carbon in the smallest soil fractions.

Additionally, the soil carbon sequestration potential of AFS can be maximized by selecting a broad range of different tree species that aide complementarity, minimizing tillage activities and leaving litter and crop residue on site. Our results also have a broader implication globally on the role of tree based, shaded perennial systems in

101

mitigating GHG emissions by improved carbon storage and subsequent sequestration in soil.

102

Table 4-1. Soil characteristics (bulk density, pH, and particle-size distribution) at different depths in five land-use systems in Koppa, Chikmagalur, Karnataka, India Land-use Depth Bulk pH Particle size distribution Types/AFS (cm) Density (g 100 g-1 soil) (Mg m-3) Sand‡ Clay Silt Coffee + Grevillea 0–10 0.96 6.3 42.4 39.6 18 10–30 0.95 5.9 36 45.6 18.4 30–60 0.81 5.6 34 49.6 16.4 60–100 0.78 5.6 37.6 44 18.4 Coffee + Mixed 0–10 1.01 6.1 38 39.6 22.4 Shade 10–30 1.06 5.5 41.6 42 16.4 30–60 0.96 5.5 39.6 46 14.4 60–100 0.98 5.2 31.2 54 14.8 Tea + Grevillea 0–10 0.87 4.8 25.2 48 26.8 10–30 0.81 5.3 22.8 50 27.2 30–60 0.78 5.1 22.9 53 24.1 60–100 0.91 5.1 16.8 56 27.2 Homegarden 0–10 1.13 6.2 42.8 28 29.2 10–30 1.09 6.0 50.8 24 25.2 30–60 1.49 5.9 46.8 28 25.2 60–100 1.07 5.8 44.4 30.4 25.2 Forest 0–10 0.67 6.3 43.2 38 18.8 10–30 0.70 6.1 28.8 46 25.2 30–60 0.89 5.9 40.8 40 19.2 60–100 0.76 5.7 44.4 36.4 19.2 ‡According to the standard classification, sand is particle between 0.05 – 2 mm in equivalent diameter, silt is particle between 0.002 – 0.05 mm in equivalent diameter, and clay is particle <0.002 mm in equivalent diameter (Weil and Brady 2014). Values reported are those obtained from composited samples within the site.

103

Table 4-2. Depth-wise distribution of soil-fraction-sized classes under five land-use systems in Koppa, Karnataka, India Percentage weight of size fraction at various depth classes Soil Size Coffee Coffee Tea Homegarden Forest Depth Fraction + + + (cm) (µm) Grevillea Mixed Grevillea shade >250 52.6 52.4 56.9 52.4 73.2 0–10 250–53 32.8 28.9 26.5 32.8 19.2 <53 23.4 21.2 18.6 16.9 8.7 >250 50.8 53.3 58.2 51.1 72.8 10–30 250–53 33.0 34.1 26.3 37.9 20.3 <53 23.9 20.2 17.6 17.1 9.2 >250 49.3 39.9 62.6 43.9 64.9 30–60 250–53 33.7 41.5 26.3 41.4 24.9 <53 22.9 21.6 15.6 17.6 11.9 >250 42.9 35.5 51.3 43.1 58.5 60–100 250–53 34.1 44.2 32.8 42.2 28.6 <53 32.2 23.9 19.6 17.1 13.9

104

Table 4-3. Analysis of variance (ANOVA), (factor analysis) with interaction effects of Treatments, Depth and Fraction size on SOC stocks (Mg C ha-1) without individual effect of each site level Category Df Sum Sq. Mean F value Pr (>F) Sq. Treatment 4 266 66.5 25.5 0.0153 Depth 3 742 247.4 11.8 < 0.0001 Fraction Size 2 5262 2360 125.1 < 0.0001 Treatment × Depth 12 269 22.4 1.06 0.39 Treatment × Fraction Size 8 1579 197.4 9.4 < 0.0001 Depth × Fraction Size 6 254 42.3 2.0 0.06 Treatment × Depth × Fraction 24 476 19.8 0.9 0.54 Size Residuals 180 3783 21 Interaction effects: Significance. codes: 0 ‘***’ 0.001 ‘**’ 0.01 ‘*’ 0.05 ‘.’ 0.1 ‘ ’1 ¶¶ Reference categories: Forest for Treatments, 0–10 cm for Depth, > 250 µm for Fraction size, Response: SOC (Mg C ha-1)

105

Table 4-4. Analysis of variance (ANOVA), (factor analysis) with interaction effects of Treatments, Depth and Fraction size on SOC stocks (Mg C ha-1) showing individual effect of each site level Category Df Sum Mean Sq F value Pr(>F) Sq Treatment 4 266 66.5 3.2 0.018 Treatment: Coffee + Grevillea 1 1 0.9 0.041 0.83 Treatment: Coffee + Mixed Shade 1 104 103.5 4.9 0.02 Treatment: Homegarden 1 159 159.3 7.5 0.006 Treatment: Tea + Grevillea 1 2 2.1 0.1 0.74 Depth 3 742 247 11.7 < 0.0001 Fraction. Size 2 5262 2630 125.1 < 0.0001 Treatment× Depth 12 269 22.4 1.0 < 0.0001 Treatment× Depth: Coffee + Grevillea 3 72 24 1.1 0.33 Treatment× Depth: Coffee + Mixed 3 64 21.3 1.0 0.4 Shade Treatment× Depth: Homegarden 3 32 10.7 0.5 0.67 Treatment× Depth: Tea + Grevillea 3 101 33.7 1.6 0.19 Treatment× Fraction Size 8 1579 197.4 9.3 < 0.0001 Treatment× Fraction Size: Coffee + 2 518 258.9 12.3 < 0.0001 Grevillea Treatment× Fraction Size: Coffee+ 2 424 212.2 10.1 < 0.0001 Mixed Shade Treatments× Fraction Size: 2 365 182.5 8.7 0.0002 Homegarden Treatment× Fraction Size: Tea + 2 272 136.1 6.4 0.002 Grevillea Depth× Fraction Size 6 254 42.3 2.0 0.06 Treatment× Depth× Fraction Size 24 476 19.8 0.9 0.54 Treatment× Depth× Fraction Size: 6 224 37.4 1.7 0.10 Coffee + Grevillea Treatment× Depth× Fraction Size: 6 67 11.1 0.52 0.78 Coffee+ Mixed Shade Treatment× Depth× Fraction Size: 6 39 6.4 0.306 0.93 Homegarden Treatment× Depth× Fraction Size: 6 146 24.3 1.1 0.33 Tea + Grevillea Residuals 180 3783 21.0 ¶Interaction effects: Significance. codes: 0 ‘***’ 0.001 ‘**’ 0.01 ‘*’ 0.05 ‘.’ 0.1 ‘ ’1 ¶¶ Reference categories: Forest for treatments, 0–10 cm for Depth, > 250 µm for Fraction size, Response: SOC (Mg C ha-1).

106

Figure 4-1. Location of the study in Koppa, Chikmagalur, Karnataka, India

107

Figure 4-2. Plot selection, location: Devon Plantations, Koppa, Chikmagalur, Karnataka, India. Photo courtesy of author

108

200

180 172.3 160

140 142.4

)

1 - 120 115.6 104.8 100 89.3

80 SOC (Mg C ha 60

0 Coffee + Grevillea Coffee + Mixed Tea + Grevillea Homegarden Forest Shade Land-use systems

Figure 4-3. Total soil organic carbon (SOC) content in the whole soil up to 1 m depth in five different land-use systems in Koppa, Chikmagalur, Karnataka, India.

109

Figure 4-4. Depth-wise mean soil organic carbon (SOC) in Mg C ha-1 stock in the whole soil up to 1 m depth in five different land-use systems in Koppa, Karnataka, India.

Note: Lower case letters indicate differences (at the 0.05 probability level) in SOC among land-use systems for soil depth within 1 m.

110

Figure 4-5. Depth-wise mean soil organic carbon (SOC) in Mg C ha-1 stock in macroaggregates (>250µm) soil up to 1 m depth in five different land-use systems in Koppa, Karnataka, India.

Note: Lower case letters indicate differences (at the 0.05 probability level) in SOC among land-use systems compared for soil depth within 1 m.

111

Figure 4-6. Depth-wise mean soil organic carbon stock (SOC) in Mg C ha-1 in microaggregates (250–53 µm) soil up to 1 m depth in five different land-use systems in Koppa, Karnataka, India.

Note: Lower case letters indicate differences (at the 0.05 probability level) in SOC among land-use systems compared for soil depth within 1 m.

112

Figure 4-7. Depth-wise mean soil organic carbon (SOC) in Mg C ha-1 stock in silt + clay fraction (<53µm) soil up to 1 m depth in five different land-use systems in Koppa, Karnataka, India.

Note: Lower case letters indicate differences (at the 0.05 probability level) in SOC among land-use systems compared for soil depth within 1 m.

113

CHAPTER 5 AGGREGATE-SIZE FRACTIONS AND SOIL-CARBON STOCKS UNDER ORGANIC AND CONVENTIONAL COFFEE AGROFORESTRY SYSTEMS IN COSTA RICA

Introduction

Soils form the greatest terrestrial carbon sinks and are estimated to hold 1462–

1548 Pg of organic carbon up to a depth of 1 m (Batjes, 1996; Lal et al., 2015). The carbon stored in surface soils (0–30 cm depth) and deep soil (30 cm and beyond), together account for three times the C stored aboveground in vegetation. The top soil C accounts to half of the soil organic carbon (SOC) and are considered to be highly susceptible to losses due to land-management practices. On the other hand, SOC stored in the deep soil are more secure to such losses. It has been widely accepted that there is a need to identify land management practices that promote the secure storage of carbon in soil (Nair et al., 2009; Powlson et al., 2011; Stout et al., 2016). Agroforestry systems (AFS) contribute to climate change mitigation by promoting C storage in soil

(Montagnini and Nair, 2004). Agroforestry systems (AFS) including tree cover on agricultural land are estimated to be practiced over one billion ha of land in the tropics (Nair et al., 2009; Zomer et al., 2016), and 1.6 billion ha globally (Nair,

2012). This indicates that AFS have a great potential to sequester both above and belowground carbon. It is commonly believed that AFS enhance SOC stocks compared to tree-less annual crop systems (Montagnini and Nair, 2004; Nair et al., 2009). Much of these claims are pertinent to SOC stock changes in the top soil and few publications are available on the effects of AFS on SOC stocks in deep soil (Abaker et al., 2016; Abou

Rajab et al., 2016; Gama-Rodrigues et al., 2010; Haile et al., 2010; Saha et al., 2010;

Tonucci et al., 2011). The evidences SOC stock improvements have often been derived

114

from extrapolation of aboveground C sequestration rates and almost all studies lack chronosequence leading to complexity of assessing long-term SOC change and nebulous C sequestration potential (CSP) estimates (Sanderman and Baldock, 2010).

As mentioned in chapter 4, soil aggregates play a pivotal role in “secure” storage of carbon. Given the importance of size fractions in SOC storage, several studies on the potential of C sequestration in soils under AFS have emphasized the importance of determining the extent of C storage in different aggregate classes at deeper soil depth up to 1 m (Haile et al., 2010; Haile et al., 2008; Howlett et al., 2011; Saha et al., 2010;

Tonucci et al., 2011).

Management practices, such as agroforestry systems promote the maintenance and accumulation of SOC by contributing more plant residues on the soil surface through improved litterfall. This further leads to improvements in soil aggregation and aggregate stability (Chen et al., 2017; Jastrow et al., 2007). Higher SOM from litterfall, pruning and composts can increase the amount of aggregates, especially macroaggregates, and promote their stability (Elliott, 1986; Six et al., 2002). Improved soil aggregation reduces the losses from microbial mineralization thereby increasing

SOC storage. Physical protection of SOM from microbial decomposition through sorption to clay minerals (Hassink, 1994), other organic molecules (Adu and Oades,

1978) and encapsulation within soil aggregates (Tisdall and Oades, 1982) lead to longer residence time of SOC within the soil aggregates. Studies on mineralization comparing whole soil vs. aggregates revealed the existence of physically protected

SOC pool in soil aggregates (Chen et al., 2017). Thus, AFS could improve soil aggregation and promote C within aggregates.

115

The main source of belowground carbon comes from roots (Pierret et al., 2016).

It has been well recognized that there is substantial input of organic matter to the subsoil horizons via plant roots and its exudates, dissolved organic matter and bioturbation processes (Rumpel and Kögel-Knabner, 2011). Moreover, subsoil C is often considered more important than topsoil C in terms of source vs. sink relations for

CO2 because carbon accumulated in subsoil is less impacted by disturbances induced by land-use changes (Batjes, 1996; Lorenz and Lal, 2005).

The available studies on depth-wise distribution of SOC stocks indicate that the extent of soil C sequestration will depend on a number of site-specific factors including soil type, land-use system, and soil-depth class, with strong interactions among these factors (Chen et al., 2017; Gama-Rodrigues et al., 2010; Haile et al., 2008; Haile et al.,

2010; Saha et al., 2010; Tonucci et al., 2011). It is important, however, to extend such studies to a wide spectrum of land-use systems in order to arrive at widely applicable conclusions on the role of AFS in soil C sequestration vis-à-vis other ecosystem services.

Costa Rica is a small Central American country where coffee is usually grown under shade trees (shaded perennial AFS). Coffee in Costa Rica is still the single most important crop in terms of land use (109,000 ha) (FAO, 2015), so the design and management of coffee AFS has major implications for C storage and policy making. The management variables in coffee AFS implemented in Costa Rica that we hypothesized would affect the SOC stocks in whole soil and aggregates are: (i) the use of shade trees with coffee versus full sun grown coffee, (ii) The species of shade trees: Timber vs. N2

116

fixing with varied patterns of pruning (the unpruned timber species vs. heavily pruned N2 fixing species); (iii) conventional chemical fertilization versus organic fertilization.

By using experimental comparison amongst the types coffee AFS we aimed to evaluate the influence of management variables under coffee AFS on SOC storage within whole soil and aggregates. Specifically, the objectives of this study were:

 Compare the differences in whole soil and aggregate-associated carbon stocks amongst coffee AFS with the above-mentioned management variables and an adjacent forest across varying depth classes up to 1m.

 Analyze whether increased organic matter inputs and aboveground biomass through varied management practices within coffee AFS helped in improving soil aggregation and SOC storage relative to sun coffee.

 Evaluate if AFS promote aggregate associated C within the smallest aggregate fraction (<53µm) across varying depth classes.

Materials and Methods

Study Location

The study was conducted in an experimental field station managed by the

‘Centro Agronómico Tropical de Investigación y Enseñanza’ (CATIE) in Costa Rica, chosen to represent low altitude coffee growing regions. Experiments were established in this site at the end of the year 2000. Geographically, the site was located in Turrialba

(9° 53´ N, 83°40´ W) at 685 m above sea level (Figure 5-1). The climate is humid tropical with no marked dry season: annual precipitation is 2600 mm yr−1 and mean annual temperature is 22°C (Haggar et al., 2011; Noponen et al., 2013). The soils of the study site have been classified as Inceptisols and Ultisols under the USDA Soil

Taxonomy classification system. The previous land-use type on these experimental plots was sugar cane, (Saccharum officinarum). Out of three experimental blocks,

117

Ultisols were present in two of the blocks and Inceptisols were present in the third experimental block.

Land-Use Systems and Management Practices

Land-use systems under varying shade and management practices were chosen for the study. The shade trees were either a timber, non N2-fixing species tree:

1. Terminalia amazonia, a Central American native species) or 2. Leguminous, N2-fixing species (Erythrina poeppigiana). The management practices selected were

1. Conventional intensive and 2. Organic intensive (Figure 5-2). Coffee, Coffea arabica

L. var. Catuura, in the experimental plot was planted at 8,000 coffee bushes per hectare in Costa Rica by planting two plants per planting hole—a common practice in Costa

Rica. Coffee planting holes were spaced 1 x 2 m apart with trees planted at 4 x 6 m.

Shade trees were planted at 417 trees per hectare. The tree management regime varied according to the shade species; the timber tree’s (Terminalia amazonia) shade was managed through periodic thinning. The leguminous, N2 fixing shade tree species were pruned regularly to provide the organic N input to the soil. Terminalia amazonia had their lower branches pruned each year to improve their form. Only the trunks of the thinned trees of the timber species Terminalia amazonia were removed, branches and leaves were left on site. In the conventional treatment with Erythrina poeppigiana shade tree, Erythrina was pruned completely (pollarded) twice a year. In organic treatment with

Erythrina poeppigiana, a minimum of three branches were left for partial shade cover after each of the two-annual pruning. In all cases, the pruned material was left on site.

Under conventional intensive management regime, the trees were pruned at a height of

1.8–2.0 m with the removal of all branches above this height (pollarding), a common practice in Costa Rica (Muschler, 2001).

118

The details of the conventional and organic management regimes are elaborated in

Table 5-1.

Six different land-use systems were selected for this study:

1. Conventional intensively managed coffee + N2 species Erythrina shade (CE); (area: 0.48 ha)

2. Conventional intensively managed coffee + timber species Terminalia shade (CT); (area: 0.4 ha)

3. Organic intensively managed coffee + N2 species Erythrina shade (OE); (area: 0.4 ha)

4. Organic intensively managed coffee + timber species Terminalia shade (OT); (area: 0.4 ha)

5. Full sun grown coffee (SC); (area: 0.4 ha)

6. Forest: Native, Talamancan montane forest (Bosque Florencia) from a nearby site (FO)

Soil Sampling

Soil samples were collected following a randomized complete block design

(RCBD) from plots of Conventional + Erythrina (CE), Conventional Terminalia (CT),

Forest (FO), Organic + Erythrina (OE), Organic + Terminalia (OT), Sun Coffee (SC).

Each sampling site measuring 1m x 1m area was dug up to collect representative soil samples from four depth classes: 0–10, 10–30, 30–60, and 60–100 cm. Samples from four randomly selected sampling sites per plot were composited to form one replicate for each depth class. Three such replicates were collected from four different, randomly selected plots of each treatment (land-use system), giving a total of 72 samples (6 treatments x 4 depth classes x 3 replicates per plot). At each sampling location, the selected 1 m x 1m site was dug down to more than one-meter depth to collect samples form the four depth classes. For determination of bulk density of soil at each depth

119

class, a steel cylinder of known volume was inserted horizontally on the wall of the pits at the center of each depth class; the soil inside the cylinder was collected, dried, and weighed. All samples were air-dried at room temperature and transported to the soil laboratory of the Centro Agronómico Tropical de Investigación y Enseñanza’ (CATIE) in

Costa Rica, for further processing. The soil samples were air-dried and sieved (2 mm sieve) at the soils laboratory in Centro Agronómico Tropical de Investigación y

Enseñanza. The portion of soil that did not pass through 2 mm sieve (#10 U.S.

Standard Testing Sieve) was separated and discarded; the 2 mm-sieved soil samples, hereafter referred to as the whole soil, were bagged, and shipped to the University of

Florida, Gainesville, FL, USA, for further analyses.

Soil Preparation and Analysis

Sub samples of the whole soil were fractionated manually into three aggregate size classes >250μm , 250–53 μm and <53 μm at the Soil and Water Sciences

Department laboratory, University of Florida, according to the methods described by

(Elliott, 1986), further modified by (Six et al., 2002) that had been adapted and followed by previous researchers of this laboratory (Haile et al. 2008; Gama-Rodrigues et al.

2010; Haile et al. 2010; Saha et al. 2010; Tonucci et al. 2011). The method consisted of wet sieving using disruptive forces of slaking followed by sieving through a series of two sieve sizes (250 and 53 µm) to obtain the three fraction size classes: macro (>250 µm), micro (53–250 μm), and silt- and clay- sized fraction (<53 µm). A 100g sample of 2 mm sieved air-dried soil was placed in a 500 mL beaker. Distilled water (250 mL), enough to completely cover the soil, was poured into the beaker to initiate the process of slaking for disintegrating the water unstable aggregates in soil, leaving only water stable aggregates to proceed with further analysis. The soil solution was wet-sieved manually

120

by moving the sieve up and down about 5 cm each, 100 times in two minutes. The soil fraction that did not pass the 250 μm sieve was backwashed with a distilled water-filled wash bottle into a glass beaker. The remaining soil solution was next poured over the

53 μm sieve (#270 U.S. Standard Testing Sieve), and the procedure was repeated. The three soil fractions >250 μm, 250–53 μm, and <53 μm (hereafter referred to as soil fractions), were dried at 65°C, weighed, ground for homogenization using a QM-3A High

Speed Vibration Ball Mill for 10 min, and stored in individually sealed and labeled plastic bags for C analysis. The average recovery of initial soil mass was 98%. The weight of oven dried samples (dry weight) under each aggregate size was noted (Table 5-2).

The whole soil, not treated with the slaking or fractionation procedure, was dried and ground for homogenization. Soil samples were analyzed in a Carlo Erba NA1500

Agricultural Lab, Georgia, USA for textural analyses by the hydrometer method (Gee and Bauder, 1979). Soil characteristics are elaborated in Table 5-2.

The total C stored to a meter depth is the sum of the C stored at each of the depths within the soil profile.

The C storage was calculated as:

퐶푠푡표푐푘 = 퐶푐표푛푐푒푛푡푟푎푡푖표푛 × 퐵퐷 × 퐷푒푝푡ℎ × 푊푒푖𝑔ℎ푡 퐹푟푎푐푡푖표푛 (5-1) where,

121

−1 Cstock = C storage expressed in Mg ha in each fraction for a given depth

−1 Cconcentration = C concentration in size fraction (g 100 g ) of that fraction size

BD = Bulk density (Mg m-3)

Depth = Depth of soil profile (cm)

Weight Fraction = weight of the fraction in the whole soil as a ratio (dimensionless).

The depth-wise distribution of different soil-fraction-size classes is given in Table 5-2.

Additionally, the C concentrations at the start of the experiment in 2001 up to a depth of

40 cm were obtained from Noponen et al., (2013) and also by personal communication with the first author. C stocks were calculated using equation (5-1). Additionally, carbon sequestration potential (CSP) was also measured.

Statistical Analyses

To test the effect of Treatment, Fraction size and Depth on SOC stocks, full fixed effect model, which is a linear regression model was fitted using R studio version 3.4.2 with multicompview package (Hothorn et al. 2008). Full fixed effect models were run with all three interaction terms: Treatments, Depth and Fraction size with reference categories Forest for treatments, the top soil profile 0–10 cm for depth and aggregate size >250µm as fraction size, where the response variable was SOC stocks in Mg C ha-

1. To elucidate specific interaction effects of Treatment, Fraction size and Depth on changes in SOC stock, individual first order (two factor) and second order (three factor) interaction effects analysis of variance (ANOVA) were also performed.

The following equation illustrates the first and second order interactions:

푌푖푗푘푙 = 휇 + 훼푖 + 훽푗 + 훾푘 + (훼훽)푖푗 + (훽훾)푗푘 + (훼훾)푖푘 + (훼훽훾)푖푗푘 + 휀푖푗푘푙 (5-2) where,

122

푌푖푗푘푙 = SOC stock (in log scale), for the observation corresponding to the 푖th treatments and the 푗th depth level and 푘th fraction size in the 푙th replicate,

푖 = 1, ⋯ , 5, 푗 = 1, 2, 3, 푘 = 1, ⋯ , 4, 푙 = 1, ⋯ , 4.

휇 = overall effect,

훼푖 = effect due to the 푖th treatments (site)

훽푗 = effect due to the 푗th depth level,

훾푘= effect of the 푘th Fraction size.

(훼훽)푖푘 = interaction effect of the 푖th treatments and the 푗th depth level

(훽훾)푗푘= interaction effect of the 푗th depth level and the 푘th fraction size

(훼훾)푖푘 = interaction effect of the 푖th treatments and the 푘th fraction size

(훼훽훾)푖푗푘= interaction effect of the 푖th treatments and the 푗th depth level and 푘th fraction size

휀푖푗푘푙 = normal random error

Tukey’s HSD was performed to compare mean differences between land-management practices on SOC in whole soil, macro sized, micro sized and silt- and- clay sized fractions for all sites. All statistical tests were performed in R studio version 3.4.2. The R packages, ggplot2 (Wickham 2017) , lsmeans (Lenth, 2016), reshape2 (Wickham,

2007) and multicompview (Hothorn et al. 2008) were used for additional analyses.

Results

Soil Organic Carbon Stock up to 1m Depth

The total SOC stocks in whole soil (in all depth classes and fraction sizes combined) down to 1 m depth varied among different treatments (land-use practices)

(Figure 5-3), with the highest under FO (146.6 Mg C ha−1) and lowest under SC (92.4

123

Mg C ha−1). The overall results for SOC stock were in the following order: FO >

OT>OE> CT> CE>SC. (Figure 5-3). The soil characteristics at different depth class are given in Table 5-2 and the depth-wise distribution of soil fractions are elaborated in

Table 5-3. The SC, CT, OE and OT treatments had 37%, 25%, 23% and 14.5% lesser

SOC stocks than the Forest. Compared to the OT and OE AFS treatments, the monoculture SC had 35%, 18.4% and 15% lesser SOC stocks in whole soil up to a depth of 1m.

SOC Stock Distribution in Whole Soil under Various Land-Use Systems and Depth Classes

The SOC stocks in the 0–10 cm soil layer under different land-use systems were in the following order, with corresponding quantities (Mg C ha-1): Forest (41.5) > OT (38)

≥ OE (37.5) > CT (36.3) >SC (28)> CE (25), (Figure 5-4). In the 10–30 cm soil layer, the systems were in the order (Mg C ha-1): OT (43) ≥ CE (42) ≥ FO (36.8) (Figure 5-4), whereas in the 30–60 cm soil layer, the order was (Mg C ha-1): Forest (43) > OT (28.2)

≥ CT (25.5) ≥ OE (24.8)>SC (21.1), (Figure 5-4). Overall, SOC stocks decreased with increasing soil depth except for FO, where SOC stocks were noted as high as 32.1 Mg

C ha-1 even up to a depth of 100cm. Beyond 60 cm, the highest SOC stocks were under

Forest (32.1 Mg C ha-1) and the lowest under CE

(14.1 Mg C ha-1), (Figure 5-4). No significant differences were noted in SOC stocks among any AFS treatments and SC beyond 30cm.

Soil Organic Carbon in Macro-Sized Fraction (>250 μm)

The total SOC contained in the macroaggregates in the entire 1 m depth profile were 116.1, 84.4, 71.6, 70.1, 68.3 and 63 (Mg C ha-1) in FO, OT, CE, CT, OE and SC treatments, respectively (Figure 5-5), with the highest amount under Forests. In the

124

uppermost depth (0–10 cm) and second upper depth (10–30 cm), CE and OE had the least SOC content respectively, while SC had the least SOC stocks within macroaggregates in the lower depth classes (30–60 cm) and (60–100 cm), at 11.5 Mg

C ha-1 and 5.5 Mg C ha-1 respectively. Among the three aggregate-size classes, the highest proportion of total C was retained in the macroaggregates. In the second depth class of 10–30 cm, both CT and FO had statistically similar mean SOC stocks. While in all other depth classes, FO contained significantly higher SOC stocks than all other treatments. Beyond 30cm, no significant differences in SOC stocks were found between shaded AFS and unshaded system (SC).

Soil Organic Carbon in Micro-Sized Fraction (250 μm–53 μm)

The total SOC contained in the microaggregates in the entire 1 m depth profile was highest under OT systems, closely followed by FO (Figure 5-6). Depth-wise, in the

0–10 cm, OE and OT had higher SOC stocks (6.4 Mg C ha-1and 6.38 Mg C ha-1) while

Forest had 3.75 Mg C ha-1. In 10–30 cm depth class, the mean SOC stocks were not significantly different among the treatments. At the second lowest depth class (30–60 cm), the treatment OT had statistically similar SOC stocks as that of FO. At the lowest depth class (60–100 cm), the FO site had significantly higher SOC stocks at 8.3 Mg C ha-1. Beyond 30cm, no significant differences in SOC stocks were found between shaded AFS and unshaded system (SC).

Soil Organic Carbon in Silt and Clay Fraction (<53 μm)

The SOC contents in the silt + clay fraction was significantly higher under the two

AFS systems with Terminalia as shade tree (CT and OT) especially in the uppermost soil layer, 0–10 cm (Figure 5-7). While CT had significantly higher SOC stocks in the top two soil layers (0–10 cm and 10–30 cm), there were significant differences noted

125

between the two organic management systems; OT and OE in the 30–60 cm but no significant differences were noted among treatments in the deepest soil profile (60 –

100 cm). However, SOC held in the smallest fraction was highest under OT (17.3 Mg C ha-1) and lowest under SC (12 Mg C ha-1) (Figure 5-7). No significant differences in

SOC stock were noted beyond a depth of 60 cm (Figure 5-7).

Carbon Sequestration Potential (CSP)

The SOC stocks calculated from the C concentrations at the start of the experiment in 2001 was compared to SOC stocks in measured in 2017 under CE, CT,

OE, OT and SC treatments (Table 5-4, Figure 5-8). Additionally, the carbon sequestration potential (Mg C ha-1 yr-1) reported following trend: OT>OE~CT>CE~SC.

The CSP was the highest under OT (1.3 Mg C ha-1 yr-1). Both SC and CE showed decrease in SOC stocks from the time of establishment of AFS in 2001 (Figure 5-8).

Interaction Effects and Analysis of Variance (ANOVA)

The Treatments (p=0.0008), Soil Depth (p<0.0001), and Fraction size

(p<0.0001), and all interactions has significant effect on total SOC stocks (Table 5-3,

Table 5-4). Individual effect of treatments (land-use systems) on SOC stocks, compared to FO as reference, was statistically different for the treatments OE (p=0.03*) and SC (p=0.0002). On the other hand, the other AFS treatments, had similar effects on

SOC stocks compared to the reference treatment, FO (Table 5-4). Compared to the reference depth class 0–10cm, the effect of all other depth classes on SOC stocks was significantly different (p<0.0001). Similarly, compared to macroaggregates (>250μm),

SOC stocks were significantly different in the other two fraction size classes, microaggregates, (250µm– 53μm) and, silt + clay (<53μm), in improving SOC stocks

(Table 5-4). The Interaction effects of Fraction size × Treatment was also significant

126

(p=0.0028). Additionally, the effect of Fraction size × Treatment interaction on SOC stocks was significantly lower for the individual treatments (Fraction size × Treatment:

SC) and (Fraction size × Treatment: OE) compared to FO (Table 5-4).

Discussion

Land-Use System – Soil Depth Class – Aggregate Size Interactions in SOC Storage

While reporting results involving multiple factors and their effects on SOC stocks, it is essential to analyze and interpret all possible interactions influencing SOC stocks

(Vargas et al., 2015). For this study, we adopted a full fixed effect model and performed an analysis of variance (ANOVA) to comprehend the effects of the factors Depth,

Fraction size and Treatments on SOC stocks and all possible interaction effects at all levels between/among the variables (Table 5-3 and Table 5-4). The significant Fraction size × Treatment interaction (Table 5-4) indicates that Fraction sizes within soil aggregates influences the effect of Treatments on SOC stocks. Considering the macroaggregate fraction size as the reference, the other two fraction sizes showed significant different main effect on SOC stock, which indicates that there were differences between fraction sizes in the amount of SOC stocks stored within them

(Table 5-4), or, the amount of SOC stock is influenced by land-use system and fraction size. The interaction effect was also significant, in Depth× Fraction size (Table 5-4), that means the fractions sizes also had significant effect on the amount of SOC stocks at different depth classes. Therefore, SOC stock variations at different depth classes can also be attributed to the relative levels of various fraction sizes; for example, predominance of a particular fraction size such as macroaggregates or silt- and-clay could also be an indication of higher or lower levels of total SOC stock at any depth

127

class. The amount of SOC stock at any depth class was not influenced by land use system (Treatments) but was influenced by soil Fraction size class. The effect of

Fraction size class was significantly different at different Depth classes. The interactions of Depth × Treatment were not significant in general and also under specific level of

Treatment interaction. Thus, SOC stocks under shaded and unshaded systems (CE,

CT, OE, OT vs. SC) did not vary significantly at any depth class. The interaction effects of Fraction size × Treatment with respect to specific treatments was significant for SC indicating that even though the effect of shade vs. unshaded treatments is not significant under whole soil, statistical differences are noted within fraction size (Table

5-4)

SOC Stocks in Whole Soil under Different Land-Use Systems up to 1m Depth

It is important to understand the trends in SOC stocks in a managed land-use system when assessing their potential environmental impacts. Factors like vegetation type, climate, ecosystem productivity, soil aggregates, soil texture, and management practices strongly influence the input and output of organic carbon within agroecosystems (Six et al., 2002). Although, a strong correlation between SOC stocks and higher organic matter input is expected in shaded perennial AFS, which often mimic forest-like ecosystem (Hombegowda et al., 2015), this study site did not demonstrate such correlations. The biomass input was higher under CE and OE treatments (Table 5-

1-2) but the SOC stocks were significantly higher in OT (Figure 5-3). On the other hand, management practices (conventional vs. organic) played a more significant role in improving SOC stocks (Table 5-1) as both OT and OE treatments reported higher SOC stocks than CT and CE treatments. Thus, the organic management could be much more important than litter input to increase the carbon significantly, in the soil layer.

128

Similar trends were also reported by Häger, (2012); Soto-Pinto and Aguirre-Dávila,

(2014).

Among the land-use systems investigated, SOC stocks were highest in FO, lowest in SC and intermediate for the other four shaded perennial AFS. The total amount of SOC in whole soil up to a depth of 1m varied with land-use types and management practices (Figure 5-3). The SOC stock under the OT (125 Mg C ha-1), OE

(114 Mg C ha-1), up to a depth of 100cm was comparable to the values of 111 Mg C ha-1 for shade grown coffee systems up to a depth of 40 cm in Costa Rica by Hergoualc’h et al., (2012) and global estimates of ~150 Mg C ha-1 under shaded perennial AFS (Abou

Rajab et al., 2016). It is interesting to note that SOC stocks at the onset of the experiment in 2001 as reported by Noponen et al., (2013) in the following order of CE,

CT, OE, OT and SC (Mg C ha-1) as: 54, 67.8, 56, 54.2 and 56.5 up to a depth of 40 cm.

In 16 years of time, the SOC stocks from this study measured in the year 2017 are being reported in the following order of CE, CT, OE, OT and SC (Mg C ha-1) as: 53.9,

73.9, 62.6, 74.5 and 52.8 (Figure 5-8), which leads us to infer that SOC stocks increased under most AFS with the exception of the treatments CE and shade devoid

SC. Additionally, the carbon sequestration potential (CSP) in the order of the treatments CE, CT, OE, OT and SC (Mg C ha-1 yr-1) were 0.3, -0.2, 0.4, 1.3 and -0.2 respectively (Table 5-4).

All AFS treatments under shaded perennial systems had higher SOC stocks than the shade devoid SC (92.3 Mg C ha-1). Total SOC stock decreased with soil depth in all land-use systems (Figure 5-4), which is universally true under most soil orders except

Spodosols, and similar observations have been reported from different coffee growing

129

regions (Tumwebaze and Byakagaba, 2016; Tumwebaze et al. 2012, in Uganda; Dossa et al. 2008, in Togo; Ehrenbergerová et al., 2015, in Peru; Soto-Pinto et al., 2009, in

Chiapas, Mexico; and, Hombegowda et al., 2015 in Karnataka, India).

Our results are comparable to SOC stock values reported under shaded AFS by

Dossa et al. (2008), who reported that shaded coffee systems yield higher belowground carbon than open-grown (sun) coffee up to a depth of 25cm. Similarly, this study found significant differences in SOC stocks between shaded AFS and SC up to a depth of

30cm. Beyond, this depth, no significant differences were noted. In Uganda,

Tumwebaze and Byakagaba (2016) estimated SOC stocks at 54 Mg C ha−1(0–30 cm) in

(Coffee + Fruit trees) while Ehrenbergerová et al. (2015) reported SOC stocks of 82.6

Mg C ha−1 (Coffee + Inga), 101.8 Mg C ha−1(Coffee + Pinus), 96.6 Mg C ha−1 (Coffee +

Eucalyptus) up to 30 cm depth in Villa Inca, Peru. Soto-Pinto et al. (2009) found SOC stocks up to 121 Mg C ha−1(0–30 cm) in shaded coffee AFS in Chiapas, Mexico. The wide range of variation within SOC stocks under shaded AFS across the globe stems ancillary factors like soil condition, climate, system age, land-use history, type of shade tree, management practices. In this study, the effects of management (conventional vs. organic) played critical role in improving SOC stocks across treatments and organic treatments under the shade of timber species Terminalia amazonia and nitrogen fixing

Erythrina poeppigiana proved better at improving SOC stocks than conventional systems in the top soil. Conventionally managed AFS under Erythrina poeppigiana was the least efficient in improving SOC stocks (Figure 5-3, Figure 5-8). However, no significant differences were noted among treatments beyond a depth of 30 cm (Figure

5-4).

130

Soil Organic Carbon in Aggregate-Size Fractions

Macroaggregates (>250 μm)

In this study, the greatest impacts of AFS on aggregate weight distribution was concentrated to the macroaggregates (Table 5-3), ranging from 70.6%, 68.5% under FO and CE to 38.5% under SC up to a depth of 100cm. Similar skewness of weight distribution towards macroaggregates within AFS were also reported by Chen et al.

(2017); Gama-Rodrigues et al. (2010); Six et al. (2000). Even though the weight of soil post oven drying was 81.5% under macroaggregates in the top soil (0–10 cm), the SOC stocks were significantly lower under SC (22 Mg C ha-1) compared to FO (37 Mg C ha-

1), OE and OT which reported 27 Mg C ha-1 and 26 Mg C ha-1 respectively. This trend within the top soil could have been due to the fact that AFS enhance organic matter accumulation in soils by including vegetation and organic inputs that increases soil microbial population which consequently increases the macroaggregates (Udawatta et al., 2008)

In the top soil, higher contents of SOC stocks were found in CT, OE and OT agroforestry than in SC; where SOC stocks decreased by 30% (Figure 5-5). This trend, however was limited to a depth of 30 cm. Beyond 30 cm, FO remained significantly higher than other treatments but the AFS treatments and SC did not show any significant differences (Figure 5-5). Increased storage of C within macroaggregates is an indicative of change in management activities including but not limited to litter inputs from over-story shade and cessation of tillage. especially in the surface horizon (Six et al., 2000). The SOC content in macroaggregates is a function of land-use changes with relatively “short” time spans (<100 years) especially in the top soil. Similar trends were also reported by Chen et al. (2017) from a study conducted in a rubber AFS site where

131

differences in SOC stocks within macroaggregates were noted only up to a depth of 30 cm. This corroborates with the trends observed in this study within macroaggregates.

The agroforestry systems considered for the study was established in the year 2000

(~18 years old), while the FO has been in place for >100 years. The SOC stock values are considered as characteristics of each system thus the significant difference noted between FO and other AFS especially in the lower depth classes (30–60 cm and 60–

100 cm) could also be a reflection of the “tree effect” in time. The lack of difference between unshaded SC and shaded AFS treatments highlights the poor correlation of above ground biomass stocks to SOC stocks. Similar trends have been reported from the same study site by Noponen et al., (2013).

Microaggregates (250–53µm)

Microaggregates are three-dimensional structure which provide a stable internal and external biogeochemical interface (Totsche et al., 2017). They have the ability to withstand external forces, physicochemical stresses and embattle slaking in water, allowing them to persist in soil for decades (Chenu et al., 2000;Chenu and Plante, 2006;

Totsche et al., 2017). The stability of C entrapped within microaggregates fall between macroaggregates and silt-and-clay sized classes (Six et al., 2000).

Within microaggregates, in the 0–10 cm depth class, the organic management systems OE and OT had significantly higher SOC stocks at 6.25 Mg C ha-1 under both treatments. In 30–60 cm depth class, OT was statistically similar to FO (Figure 5-6). The mean residence time (MRT), of microaggregates ranges between 10 and 100 years

(Fontaine et al., 2007; Totsche et al., 2017). Given the intimate association of MRT of

SOC within microaggregates and its subsequent stability, one may concur that the C stored in microaggregates are more robust to decomposition than macroaggregates and

132

improved SOC storage within this size class under AFS with organic management OE and OT are relatively more secure and stable.

Silt and clay fraction (<53 μm)

The average percentage of weight distribution of soil fraction sizes under AFS treatments did not demonstrate the abundance of silt-and-clay fraction in the top soil but weight distribution increased with depth under this class. Interestingly, the average percentage weight distribution in the top soil was highest under AFS managed under the shade of Terminalia; CT (19.6%) and OT (13.6%). The SOC stocks in the silt-and- clay fraction was significantly higher under the Terminalia shade tree treatments CT and

OT up to a depth of 60cm. In the top soil, tree species (timber vs. N2 fixing) seemed to be a significant moderator of SOC stocks. It was suggested by Garces (2011), that higher root biomass allocation in Terminalia could have contributed to the observed trend.

The percentage distribution of fraction size classes to the silt-and-clay fraction increased with depth under all land-use types similar to Chen et al., 2017; Gama-

Rodrigues et al., 2010; Saha et al., 2010. Under all treatments, post wet sieving, the distribution of silt-and-clay fraction under Forest within 60–100cm accounted to only 8.2 g while OE and OT treatments reported 18.6 g and 22.1 g respectively (Table 5-2). No significant difference was noted in SOC stocks among treatments under this depth class which indicates that treatments under Terminalia amazonia promote aggregate distribution (Garces, 2011), especially improving the allocation within the smallest aggregate size classes. In AFS, choice of shade tree becomes a critical moderator for

SOC stock improvement as seen in this study and other studies in Costa Rica reported by Montagnini, 2000; Redondo-Brenes and Montagnini, (2006). Globally, farmers have

133

many practical reasons for planting leguminous shade trees like Erythrina spp. ,

Gliricidia spp. along with coffee and cacao AFS; e.g. ease of establishment and ability to re-sprout after pollarding (Beer, 1988). However, the benefits of including Erythrina spp. In improving C stocks within intensively managed coffee plantations are only limited to the top soil (Beer, 1988; Haggar et al., 2011).

The C stored in the silt-and-clay fraction is considered to be “secure” as it is physically protected from enzymatic attacks (Barreto et al., 2011; Six et al., 1998;

Totsche et al., 2017). The clay minerals and humic substances interact forming organo- mineral complexes thus imparting stability to the C trapped in this soil size fraction. In this study, in the topsoil, the treatments CT and OT demonstrated higher SOC stocks than FO (Figure 5-7). Cascading down to 30 cm depth, the treatments CE, CT and OT reported statistically similar SOC stocks as that of FO. Even up to a depth of 60 cm, the treatment OT had significantly higher SOC stocks (Figure 5-7) as well as higher dry weight distribution within this fraction size, indicating the possibility of entrapment and adsorption of organic carbon by root exudates to mineral surfaces within this fraction class (Chenu and Plante, 2006) and promotion of silt-and-clay fraction under shaded

AFS. The average dry weight distribution of silt-and-clay fractions were in the order

CT>FO>OT>CE>SC>OE (Table 5-2) while the SOC content (in Mg C ha-1) was in the order: CT>CE>FO>OE>SC, bolstering our hypothesis that shaded perennial AFS and other tree-based systems store more recalcitrant C (i.e., C in the silt and clay fractions) in deeper soil layers, indicating their more efficient C sequestration potential. In this study, the AFS treatments CT and CE stored more recalcitrant C than FO. Similar trends were also reported by (Gama-Rodrigues et al., 2010) where shaded coffee AFS

134

and traditional cabruca and cacao AFS reported higher SOC stocks than adjacent forest in the silt-and-clay-sized fractions. This supports the conjecture that shaded AFS promote sequestration of C in silt-and-clay-sized fractions up to a depth of 60cm.

Beyond 60 cm, poor correlation was noted between the treatments which questions the commonly assumed notion that all AFS promote SOC stocks on improving aboveground biomass and introducing deep-rooted shade trees.

Rhizodeposition and Management under Shaded Perennial Systems

Timber species vs. N2 fixing species

In Costa Rica, coffee (Coffea arabica) plants are often grown under the shade of tall shade trees (Defrenet et al., 2016), as these systems are known to have higher potential for soil carbon sequestration apart from improving soil health, nutrient cycling and providing other ecosystem services (Nair, 2017). The roots of the coffee shrubs can extend up to 2 m in total length (Cuenca et al., 1983), although the fine roots are concentrated up to a depth of 40 cm, which corroborates the higher distribution of SOC stocks up to 30 cm in our study. This rooting depth is a function of soil type (Pierret et al., 2016).

In this study, we found out that aboveground biomass was not correlated to SOC stocks which means increased aboveground biomass and shade cover need not necessarily increase SOC stocks. However, within the silt-and-clay fractions, up to a depth of 60 cm, the treatments under timber species Terminalia amazonia, CT and OT were significantly higher than the treatments under than N2 fixing Erythrina spp. It is imperative to comprehend the rooting pattern, root growth and decomposition in coffee plantations under the varying shade trees (Timber vs. N2 fixing). Understanding such variations are pivotal to improvement in SOC stocks and the very existence of

135

ecosystem services like production of timber and non-timber produce, to regulation or support of various ecosystem functions such as pest control or nitrogen cycling (Meylan et al., 2017). There is still limited understanding on how the rhizodeposition from shade trees affect soil carbon sequestration (Cerda et al., 2017). Understanding the dynamics and patterns of rhizodeposition by introducing trees within agricultural crops to improve

SOC stocks are crucial in order to device management practices that will foster C accumulation in soils (Rasse and Rumpel, 2005).

Fine root length dynamics of Erythrina poeppigiana reveal that 80% of fine roots of remain distributed in the topsoil (Chesney and Chesney, 2015) which is supported by the fact that both the treatments OE and CE showed substantially high SOC stocks up to 30cm (Figure 5-4). On the other hand, the timber species, Terminalia amazonia, has a rooting depth up to 50cm (Jackson et al., 1996) which explains the significant dominance of SOC stocks up to 60 cm under OT treatment. One important difference between the two varieties of shade tree is the biomass allocation to roots. The timber species Terminalia amazonia has higher biomass allocation to lateral and main roots

(Sinacore et al., 2017) while N2 fixing species like Erythrina have lower foliage biomass allocation to roots (Chesney and Nygren, 2002). The shade tree fine roots in Erythrina spp. and Terminalia amazonia are different in its vertical distribution and more abundant coffee fine roots were noted when grown under Terminalia amazonia than under

Erythrina spp. (Garces, 2011) in the same study site as ours. The vertical distribution of fine roots in Terminalia amazonia have a desirable characteristic for shaded perennialagroforestry practices in this suboptimal region, Garces, (2011). This indicates that the timber species Terminalia is a better choice of shade tree for soil carbon

136

sequestration over N2 fixing species. Farmers have many practical reasons for choosing

N2 fixing shade tree species, such as Erythrina spp. as they are easy to establish and grows faster upon pollarding, large amount of litterfall and subsequent nutrient recycling from litterfall pruning residues (Beer, 1988). However, in our study we did not note a direct correlation between increased litterfall and SOC stocks (Table 5-1) or increased shade density (Table 5-1).

Overall, the treatments under Erythrina, CE and OE received an average biomass (including litterfall and pruning) of 17.4 Mg C ha-1 while the ones under

Terminalia, CT and OT received 8–10 Mg C ha-1, but the SOC stocks (Mg C ha-1) were in the order of: OT>OE; CT>CE. This bolsters the fact that effect of rhizodeposition is more pivotal in these systems as pointed out by Rasse and Rumpel, (2005). Similar ideas were also conveyed by Sinacore et al. (2017) highlighting differences in root biomass and architecture (rooting depth, distance, area and volume) could contribute to altered SOC stocks. Noponen et al., (2013) reported that heavily pruned CE introduced greater increase in SOC stocks than OT, however, our results showed a different trend in 0–10cm where SOC stocks in OT>CE. However, within the depth class 10–30cm, we noted the SOC stocks under these systems were similar (OT~CE).

The mixed effect models further confirmed that the treatments OE and SC were significantly lower (p=0.0008) in storing SOC stocks than the other treatments up to a depth of 100 cm (Table 5-4). Due to unavailability of enough replicates, we could not statistically assess the specific interaction effect of shade tree on SOC stocks.

Organic vs. Conventional management

In the past decade, much attention has been devoted to organic management practices for SOC stock improvements and its potential C sequestration in soil. Some

137

researchers have been proponents of organic treatments (Freibauer et al., 2004;

Scialabba and Müller-Lindenlauf, 2010), while other studies have been the opponents of organic management practices (Powlson et al., 2016, 2011; Sanderman and Baldock,

2010). These studies have warned of the short comings of organic management practices and C accounting methodologies that can over-estimate the net sequestration of C into soil (Noponen et al., 2013). The term C sequestration is often used to define an increase in SOC stocks over time following a change in land-use system. The coffee agroforestry in our study site was established in the year 2001. Over the past 17 years, the soil carbon stocks have increased under all treatments up to a depth of 40 cm except for CE and SC (Figure 5-8). The soil carbon sequestration potential was highest under OT at 1.3 Mg C ha-1yr-1 which was well within the ranges (0.65–1.54 Mg C ha-1 yr-

1) in Costa Rica reported by Noponen et al. (2013). However, scientists argue that these improvements in SOC stocks could aide in climate change mitigation only if they result in net transfer of C from atmospheric CO2 to soil which is not necessarily the case

(Noponen et al., 2013, 2012). At this point, one may question the use of added organic

C by the application of amendments such as chicken manure, coffee pulp as these applications may lead to only a transfer of C from one terrestrial pool to another and might not actually lead to “sequestration” (Powlson et al., 2016). Had alternative practices stored the C in soil for longer (e.g. through conversion of the organic input to biochar, biosolid) then it may have more net positive impact in C sequestration. In our study, we could not statistically assess the effects of management practices on SOC stocks due to lack of sufficient replicates and sub treatments within organic and conventional management practices. Analyzing such effects using mixed effects model

138

could bring about some clarity on this. Taking this discussion any further is beyond the scope of our study.

Conclusions

There lies a preconceived notion that increasing above-ground biomass by optimizing shade density, planting trees within AFS will result in an automatic proportional increase in SOC stocks. Our study from a seventeen-year-old experimental site highlight that this is not always the case. Soil organic carbon stock improvement under AFS are highly site specific. Although, overall SOC stocks in seventeen years improved under most AFS, the lack of positive relation between aboveground biomass and increased SOC stocks within treatments could be attributed to the multitude of factors influencing changes in SOC stocks like species selection, soil type, quality of litter and pruning, previous land-use and so on. The treatments CE and OE had almost double biomass and higher shade density as compared to CT and OT but the SOC stocks did not correlate to increased biomass and shade density as observed in SC treatments. Forest continued to have the highest SOC stocks especially beyond a depth of 60 cm. Adoption of AFS, improved soil C in the smallest fraction (<53 µm) which can be expected to become more stabilized and sequestered over time. Thus, the shaded coffee AFS could play an important role in environmental protection by mitigating GHG emission through the storage of high amounts of well-protected organic carbon in the smallest soil fractions. However, this is not a simplified decision. Land-use decisions designed to account GHG mitigation should carefully assess soil management practices under agroforestry that could not only improve SOC stocks but also account for emissions of GHG.

139

Table 5-1. Mean organic matter inputs (Mg ha−1 yr−1) in coffee AFS in Costa Rica across conventional and organic treatments. Treatment Herb Litterfall Coffee pruning Tree pruning Total CE 225 4104 4104 9997 17357 CT 123 3832 1659 4513 10126 OE 164 3077 7837 6352 17428 OT 1338 2199 955 4203 8696 ¶ Quantities of biomass are shown as mean values of biomass collected in 2006 and 2009 adapted from Haggar et al. (2011); Noponen et al. (2013).

140

Table 5-2. Soil characteristics (bulk density, pH, and particle-size distribution) at different depths in five land-use systems in Turrialba, Cartago, Costa Rica Land-use Types/ Depth Bulk pH Particle size distribution Treatments (cm) Density (g 100 g-1 soil) (Mg m-3) Sand‡ Clay Silt Conventional Erythrina 0–10 0.73 6.1 37.6 40.8 21.6 (CE) 10–30 0.62 5.5 45.6 36.8 17.6 30–60 0.99 5.5 43.6 36.4 20 60–100 1.03 5.3 43.6 32.4 24

Conventional Terminalia 0–10 1.13 6.0 36.8 42 21.2 (CT) 10–30 0.90 5.1 32.4 44.4 23.2 30–60 0.83 5.4 40.4 36.4 23.2 60–100 0.83 5.4 38 34.4 27.6

Forest (FO) 0–10 0.64 4.5 29.2 58.4 12.4 10–30 0.69 5.1 24.8 66.4 8.8 30–60 0.73 5.3 35.2 58.4 6.4 60–100 0.76 5.1 33.2 56.4 10.4

Organic Erythrina (OE) 0–10 0.86 6.3 37.6 40.4 22 10–30 0.92 6.3 39.6 40.4 20 30–60 1.12 5.8 41.6 36.4 22 60–100 0.98 5.9 37.2 34.4 28.4

Organic Terminalia (OT) 0–10 0.89 6.5 47.6 30.4 22 10–30 0.77 6.1 37.6 38.4 24 30–60 1.15 5.7 41.6 38.8 19.6 60–100 0.91 5.9 41.6 32.8 25.6

Sun Coffee (SC) 0–10 0.81 6.4 33.2 42.4 24.4 10–30 0.95 6.6 45.2 36.8 18 30–60 0.99 5.8 45.2 32.4 22.4 60–100 1.04 5.3 43.2 34.4 22.4 ‡According to the standard classification, sand is particle between 0.05 – 2 mm in equivalent diameter, silt is particle between 0.002 – 0.05 mm in equivalent diameter, and clay is particle <0.002 mm in equivalent diameter (Weil and Brady 2014). Values reported are those obtained from composited samples within the site.

141

Table 5-3. Depth-wise distribution of different soil-fraction-size classes under six land- use systems in Turrialba, Cartago, Costa Rica Average percentage weight (%) distribution of size fraction at various depth Soil Size CE CT FO OE OT SC Depth Fraction (cm) (µm) >250 81.0 75.8 90.5 81.7 75.9 81.5 0–10 250–53 13.9 15.7 10.3 15.6 18.0 15.5 <53 11.5 19.6 8.9 6.8 13.6 8.8 >250 75.9 71.5 74.1 74.5 84.1 73.9 10–30 250–53 20.5 22.2 15.6 16.1 15.1 17.1 <53 12.9 16.5 15.3 11.8 11.9 15.3 >250 66.1 65.0 83.7 62.2 58.4 51.9 30–60 250–53 25.8 28.5 20.0 25.0 29.7 38.3 <53 19.9 20.4 17.8 22.1 20.7 19.8 >250 68.7 53.5 70.6 55.2 53.0 38.5 60–100 250–53 23.1 35.0 29.6 33.8 37.2 38.5 <53 22.9 22.1 28.2 18.6 18.8 25.5 CE: Conventional Erythrina, CT: Conventional Terminalis, FO: Forest, OE: Organic Erythrina, OT: Organic Terminalia, SC: Sun Coffee. Note: The values presented here are average of each size class for three replicates per treatment. The sum of the three fractions should not necessarily add up to 100 due to losses during processing in the laboratory.

142

Table 5-4. Analysis of variance (ANOVA), (factor analysis) without the specific level of site Category Df Sum Sq. Mean Sq. F value Pr (>F) Depth 3 1098 366 17.3 < 0.0001 Fraction. Size 2 11062 5531 261.8 < 0.0001

Treatment 5 474 95 4.4 0.0008 Depth × Fraction Size 6 2141 357 16.8 < 0.0001 Depth × Treatment 15 323 22 1.02 0.43 Fraction Size × Treatment 10 605 60 2.8 0.0028 Depth × Fraction Size × Treatment 30 659 22 1.01 0.42 Residuals 144 3042 21 Interaction effects: Significance. code: 0 ‘***’ 0.001 ‘**’ 0.01 ‘*’ 0.05 ‘.’ 0.1 ‘ ’1 ¶¶ Reference categories: Forest for Treatments, 0–10 cm for Depth, > 250µm for Fraction size, Response: SOC (Mg C ha-1)

143

Table 5-5. Analysis of variance (ANOVA), (factor analysis) showing individual effect of each site level Category Df Sum Sq Mean Sq F value Pr(>F) Depth 3 1098 366 17.3 < 0.0001 Fraction Size 2 11062 5531 261 < 0.0001 Treatment 5 474 95 4.5 0.0008 Treatment: CE 1 41 41 1.93 0.16 Treatment: CT 1 28 28 1.31 0.25 Treatment: OE 1 97 97 4.5 0.03* Treatment: OT 1 4 4 0.21 0.64 Treatment: SC 1 304 304 14.4 0.0002 Depth × Fraction Size 6 2141 357 16.8 < 0.0001 Depth × Treatment 15 323 22 1.02 0.43 Depth × Treatment: CE 3 129 43 2.1 0.11 Depth × Treatment: CT 3 21 7 7 0.8 Depth × Treatment: OE 3 45 15 0.7 0.54 Depth × Treatment: OT 3 65 22 1.1 0.38 Depth × Treatment: SC 3 63 21 0.9 0.39 Fraction Size × Treatment 10 605 60 2.8 0.0028 Fraction Size × Treatment: CE 2 22 11 0.5 0.59 Fraction Size × Treatment: CT 2 51 26 1.2 0.3 Fraction Size × Treatment: OE 2 119 59 2.8 0.06 Fraction Size × Treatment: OT 2 9 5 0.21 0.8 Fraction Size × Treatment: SC 2 403 202 9.5 0.0001 Fraction Size × Treatment× Depth 30 659 22 1.1 0.42 Fraction Size × Treatment× Depth: CE 6 157 26 1.2 0.3 Fraction Size × Treatment× Depth: CT 6 8 1 0.06 0.9 Fraction Size × Treatment× Depth: OE 6 125 21 0.98 0.43 Fraction Size × Treatment× Depth: OT 6 174 29 1.37 0.22 Fraction Size × Treatment× Depth: SC 6 194 32 1.53 0.17 Residuals 144 3042 21 ¶Interaction effects: Significance. code: 0 ‘***’ 0.001 ‘**’ 0.01 ‘*’ 0.05 ‘.’ 0.1 ‘ ’1 ¶¶ Reference categories: Forest for treatments, 0–10 cm for Depth, > 250 µm for Fraction size,Response: SOC (Mg C ha-1) CE: Conventional Erythrina, CT: Conventional Terminalis, FO: Forest, OE: Organic Erythrina, OT: Organic Terminalia, SC: Sun Coffee.

144

Table 5-6. Change in SOC stocks (2001-2017) and carbon sequestration potential (CSP) under various AFS up to a depth of 40cm Depth (cm) Treatment SOC stocks in Avg. SOC Soil CSP (Mg C ha-1yr-1) 2001 (Mg C ha- stocks in (2001–2017) 1) 2017 (Mg C ha-1) 0–10 CE 26.4 25.3 -0.1 10–40 CE 30.5 28.5 -0.1 0–40 CE 53.9 53.9 -0.2 0–10 CT 28.4 32.9 0.3 10–40 CT 39.4 41.0 0.1 0–40 CT 67.8 73.9 0.4 0–10 OE 24.3 29.5 0.3 10–40 OE 31.8 33.1 0.1 0–40 OE 56 62.6 0.4 0–10 OT 26.4 35.5 0.6 10–40 OT 27.8 39 0.7 0–40 OT 54.2 74.5 1.3 0–10 SC 24.2 27.7 0.2 10–40 SC 32.3 25.1 -0.5 0–40 SC 56.5 52.8 -0.2

Note: The SOC for the year 2017 are up to 30 cm while the other two years are up to 40 cm. The C concentration for the year 2001 was obtained from Noponen et al., (2012) and by personal communication with the first author. SOC stocks were calculated from C concentration provided; SOC stock= Cconcentration x BD x Depth. CE: Conventional Erythrina, CT: Conventional Terminalis, FO: Forest, OE: Organic Erythrina, OT: Organic Terminalia, SC: Sun Coffee

145

Figure 5-1. Location of the study in Turrialba, Cartago province, Costa Rica

a b

Figure 5-2. Management practices selected for the study: A) Conventional intensive B) Organic intensive. Location: Turrialba, Costa Rica. Photo courtesy of author.

146

180

146.63 )

1 160 - 140 125.52 109.01 113.89 120 101.65 92.37 100 80 60

40 SOC stocks (Mg C ha C (Mg stocks SOC 20 0 CT OT CE OE SC FO Treatments

Figure 5-3. Total soil organic carbon (SOC) content in the whole soil up to 1 m depth in six different land-use systems in Turrialba, Cartago, Costa Rica.

147

Figure 5-4. Depth-wise mean soil organic carbon (SOC) in Mg C ha-1 stock in the whole soil up to 1 m depth in six different land-use systems in Turrialba, Costa Rica.

Note: Lower case letters indicate differences (at the 0.05 probability level) in SOC among land-use systems compared within 1 m

148

Figure 5-5. Depth-wise mean soil organic carbon (SOC) in Mg C ha-1 stock in macroaggregates (>250µm) soil up to 1 m depth in six different land-use systems in Turrialba, Costa Rica.

Note: Lower case letters indicate differences (at the 0.05 probability level) in SOC among land-use systems compared within 1 m.

149

Figure 5-6. Depth-wise mean soil organic carbon stock (SOC) in Mg C ha-1 in microaggregates (250µm–53 µm) soil up to 1 m depth in six different land-use systems in Turrialba, Costa Rica.

Note: Lower case letters indicate differences (at the 0.05 probability level) in SOC among land-use systems compared within 1 m.

150

Figure 5-7. Depth-wise mean soil organic carbon (SOC) in Mg C ha-1 stock in silt + clay fraction (<53 µm) soil up to 1 m depth in six different land-use systems in Turrialba, Costa Rica.

Note: Lower case letters indicate differences (at the 0.05 probability level) in SOC among land-use systems compared within 1 m.

151

Figure 5-8. Soil organic carbon stocks (Mg C ha-1) in whole soil under various AFS treatments up to a depth of 40cm for the years 2001 and 2017. The SOC for the year 2017 are up to 30 cm while that of 2001 are up to 40 cm.

Note: The C concentration for the year 2001 was obtained from Noponen et al., (2012). SOC stocks were calculated from C concentration provided; SOC stock= Cconcentration x BD x Depth.

152

CHAPTER 6 SUMMARY AND CONCLUSION

The sequestration of atmospheric CO2 by expansion of carbon sinks within agroforestry systems is a promising means to mitigate global warming. One way of accomplishing it is by expanding the carbon-sink capacity by establishing agroforestry systems (AFS) that can aid in increasing total carbon stored on land. Soil is a major carbon pool in the biosphere; however, it has received rather little attention in terms of carbon accounting than the other C pools. Storage of carbon in the soil has the potential to sequester carbon over a longer duration, in deeper soils, in association with recalcitrant chemical complexes and aggregates.

The shaded perennial AFS are a unique type of land-use system in the tropics, and they often mimic forests in canopy-architecture and possibly root configuration.

These systems can be expected to have a high potential to sequester carbon in deeper soils, and within the smallest soil aggregates. However, soil C storage in shaded perennial systems across various aggregate fractions remains largely unexplored. With this background, the study reported here was conducted based on the following premises:

 Tree-based land-use systems store relatively higher quantities of soil C, compared to treeless systems, at lower soil depths (up to 1 m from surface).

 The amount of C stored in soil under shaded perennial AFS is influenced by the species of shade tree and management practices involved.

 Under tree-based systems, the smallest-sized (silt and clay) soil fractions store higher amounts of C than larger-sized fractions even at lower soil depths.

The study consisted of three major parts: a meta-analysis, and two field-based investigations in two continents. The meta-analysis involved a rigorous, quantitative assessment of the scattered results on soil organic carbon (SOC) stocks reported from

153

various AFS, in comparison with those in agricultural, forestry, and pasture systems in different agroecological regions around the world. The field studies examined SOC storage in different aggregate-size fractions (macroaggregates: 250 – 2000 µm, microaggregates: 250–53 µm, and silt + clay <53 µm) under shaded coffee agroforestry systems as compared with traditional homegarden and natural forests in two distinct coffee growing regions of the world: southern India and Costa Rica.

The meta-analysis assessed the reported differences in SOC stocks under AFS in comparison with other land-use systems (Agriculture, Forestry, Pasture, or

Uncultivated Land) at various soil-depth classes in four major agroecological regions

(arid and semiarid, ASA; lowland humid tropics, LHT; Mediterranean, MED; and temperate, TEM) around the world. The statistically rigorous process provided a comprehensive approach to identifying the common effect of AFS on SOC stocks.

Comparing Agroforest vs. Agriculture or Agroforest vs. Pasture, SOC stocks under AFS were higher by +27% in the ASA region, +26% in LHT, and +5.8% in TEM, but –5.3% in the TEM in the 0–100 cm soil depth. Overall, while SOC stocks were higher under AFS compared with sole-crop agricultural systems under comparable conditions, the stocks were generally higher under Forestry. Comparing AFS vs. Agriculture, practices like

Multistrata systems (shaded perennial AFS) in the lowland humid region are highly effective in increasing SOC stocks in soils up to 100 cm depth; indeed, these Multistrata systems were comparable to Forests in terms of their SOC stocks. The study also showed that SOC stock improvements on adopting AFS depended on the age of management practices, systems aged between 10–20 years being considerably more

154

effective in improving SOC stocks than the relatively juvenile systems of less than 10 years of age.

The field investigations in India and Costa Rica adopted the same research protocol in collecting, processing, and analyzing soil samples to determine the amount of C stored in whole soil and three soil size fractions (250 – 2000 µm, 250–53 µm, <53

µm) to 1 m soil depth. Considering that the smallest sized fraction encapsulates C strongly and C within the smallest fraction is much older and resistant to decomposition, the systems that contain more C in this fraction size are more efficient in C sequestration. The hierarchical model (HM) for aggregate protection has identified the mechanism of protection of high C microaggregates inside of macroaggregates as a strategy to protect C in the long term. The smallest silt and clay aggregates form microaggregates inside macroaggregates, and as such, forming a hierarchical protection mechanism. This study determined the extent of C held in each of the particle size classes, and to compare storage in soils of proliferating rooting systems such as shaded perennial agroforestry systems with soils of sparsely rooted systems such as

Homegarden and monoculture cropping systems that are not influenced by the presence of trees.

The study site in India was in Devon Tea and Coffee plantations, a privately- owned enterprise in Chikmagalur district, Karnataka, in southern India at an altitude of

1040 m above sea level, mean average annual precipitation 2400 mm, mean average temperature 26°C, average humidity of 92%, soils: Ultisols and Alfisols. The Costa Rica study site was in an experimental field of CATIE (Centro Agronómico Tropical de

Investigación y Enseñanza = Tropical Agricultural Research and Training Center) at a

155

low altitude coffee growing region. The replicated field experiment had been established at the end of 2000, Turrialba in Cartago province, at an altitude of 685 m above sea level, average annual precipitation 2600 mm, mean annual temperature 22°C, soils:

Inceptisols and Ultisols. At both study locations, soil samples were collected up to a depth of 100 cm from selected treatments from four depth classes. Soils were physically fractionated into three size classes (250–2000 µm, 250–53 µm, and <53 µm), and their

C contents were determined. In the absence of a time-sequence study involving long time intervals, the SOC stock data were considered as reliable indicators of the C sequestration potential of the study site in India. For the Costa Rica study site, the SOC stock data one year post the initiation of the experiment was available (courtesy: Dr.

Martin Noponen), based on which soil C sequestration over 17 years was calculated.

Results from our studies suggest that AFS impact the distribution of soil C among different aggregate size fractions differently, and therefore the associated stability of soil

The observed differences in SOC content within soil aggregates under varying climatic region and AFS type indicate the inherent variation in soil and plant community assemblages across these geographical regions. It also reflects a myriad of direct abiotic effects including but not limited to the inherent properties of a soil based on its clay content, parent material and texture, as well as the previously mentioned variations in elevation and precipitation on plant communities. Other moderators of SOC stocks include the quantity and chemical composition of C inputs in the form of litterfall and rhizodeposition. In both study sites, the non-N2 fixing shade tree species Grevilliea robusta and Terminalia amazonia contributed to higher SOC stocks than N2 fixing

156

shaded trees. The species of shade trees was a positive driver of SOC stocks in both study sites, which in turn influenced SOC stocks positively. This is bolstered by the fact that in both study sites, the Timber species shade tree Grevillea robusta and Terminalia amazonia were better at improving SOC stocks than N2 fixing shade species.

Admittedly, the study had some limitations. In the meta-analytical study, the reliability of the results inevitably depends on the quality and rigor of the data upon which it is based. Agroforestry systems by their very nature are extremely site-specific and are characterized by high variability in site-specific features such as soil type, plant species and density, and management practices. Moreover, the datasets used for the analysis were not based on a single standard procedure of C determination, and some field experiments lacked high levels of statistical rigor. These variabilities might have influenced the quality of data used and therefore the results of the analysis.

The field study conducted in Koppa, India had some limitations as well. The study involved only a one-time estimation of SOC stock. To estimate the capacity of a system to sequester C, ideally the rate of C accumulation should be documented in a temporal scale. Additionally, being a study from a private plantation as opposed to an experimental station with long-term records, we did not have the supporting data that would have been useful in establishing the relationship between species diversity and

SOC stocks. The Costa Rica study, being on an experimental farm of a well-known research center, was relatively better in terms some of the above limitations; however, that too had limitations from the standpoint of our analysis. For example, higher number of replicates within organic and conventional practices could have helped us in carrying a statistical analysis to infer the effect of individual management practices on SOC

157

stocks. The wish-list of such conditions and ancillary data could be long, but unrealistic.

Nevertheless, the SOC stock data from similar shaded coffee AFS in India and Costa

Rica indicate that although SOC stock increase could be expected under such systems, the results could be extremely site-specific and cannot be “guaranteed.” These limitations, however, should not be misconstrued to the extent of diminishing the importance and value of the results of such a study and the potential for extrapolating them to broader contexts.

158

APPENDIX ADDITIONAL ASSESSMENTS

Additional Figures for Meta-Analysis

PRISMA 2009 Flow Diagram

Figure A-1. PRISMA flow diagram detailing screening process of articles included in meta-analysis. Procedure adapted from Moher et al., (2009)

159

Figure A-2. Percentage changes in soil organic carbon (∆SOC%) stock (0–40cm) between agroforestry systems when compared to forest or agricultural land- use or pastures in the A) ASA region B) LHT region C) MED region D) TEM region.

160

Figure A-2. Continued

161

Figure A-2. Continued

162

Figure A-2. Continued.

163

Figure A-3. Percentage changes in soil organic carbon (∆SOC%) stock (0–60cm) between agroforestry systems when compared to forest or agricultural land- use or pastures in the A) ASA region B) LHT region C) MED region D) TEM region.

164

Figure A-3. Continued

165

Figure A-3. Continued

166

Figure A-3. Continued

167

Figure A-4. Percentage changes in soil organic carbon (∆SOC%) stock (0–100cm) between agroforestry systems when compared to forest or agricultural land- use or pastures in the A) ASA region B) LHT region C) MED region D) TEM region

168

Figure A-4. Continued.

169

Figure A-4. Continued.

170

Figure A-4. Continued.

171

Figure A-5. Percentage changes in soil organic carbon (∆SOC%) stock (0–200cm) between agroforestry systems when compared to forest or agricultural land- use or pastures in the A) ASA region B) LHT region C) MED region D) TEM region.

172

Figure A-5. Continued.

173

Figure A-5. Continued.

174

Figure A-6. Percentage changes in soil organic carbon (∆SOC%) stock (60–100cm) between agroforestry systems when compared to forest or agricultural land- use or pastures in the A) ASA region B) LHT region C) MED region D) TEM region

175

Figure A-6. Continued

176

Figure A-6. Continued.

177

Figure A-7. Percentage changes in soil organic carbon (∆SOC%) stock (0–20cm) between agroforestry systems of varying age when compared to forest or agricultural land-use or pastures in the A) ASA region B) LHT region C) MED region D) TEM region (Agroforest vs. Agriculture, Forest vs. Agroforest, Agroforest vs. Pasture).

178

Figure A-7. Continued.

179

Figure A-7. Continued.

180

Figure A-7. Continued.

181

Figure A-8. Percentage changes in soil organic carbon (∆SOC%) stock (0–40cm) between agroforestry systems of varying age when compared to forest or agricultural land-use or pastures in the A) ASA region B) LHT region C) MED region D) TEM region (Agroforest vs. Agriculture, Forest vs. Agroforest, Agroforest vs. Pasture).

182

Figure A-8. Continued.

183

Figure A-8. Continued.

184

Figure A-8. Continued.

185

Figure A-9. Percentage changes in soil organic carbon (∆SOC%) stock (0–60cm) between agroforestry systems of varying age when compared to forest or agricultural land-use or pastures in the A) ASA region B) LHT region C) MED region D) TEM region (Agroforest vs. Agriculture, Forest vs. Agroforest, Agroforest vs. Pasture).

186

Figure A-9. Continued.

187

Figure A-9. Continued.

188

Figure A-9. Continued.

189

Figure A-10. Percentage changes in soil organic carbon (∆SOC%) stock (0–100cm) between agroforestry systems of varying age when compared to forest or agricultural land-use or pastures in the A) ASA region B) LHT region C) MED region D) TEM region (Agroforest vs. Agriculture, Forest vs. Agroforest, Agroforest vs. Pasture).

190

Figure A-10. Continued.

191

Figure A-10. Continued.

192

Figure A-10. Continued.

193

Publication Bias in Meta-analysis

In this meta-analysis, we converted the SOC stocks reported in individual studies

(Mg C ha-1) in logarithmic scale in order to ensure normal distribution of data. Meta- analysis of continuous outcomes such as C stocks traditionally uses mean difference (or standardized mean difference SMD) in pooled standard deviation (SD) units. The ratio of mean values by calculating the logarithmic ratio of mean for each study and estimating its variance is a published and well-accepted method (Borenstein et al.,

2009; Friedrich et al., 2008). We verified that the normal distribution of the dataset was better satisfied when the log transformed data was used over non-transformed SOC stock values reported in Mg C ha-1, and by Q-Q and funnel plots (Figure A-9–A-13).

Additionally, the natural origin point was 0. The variables in this study (C stocks, Mg C ha-1) were all positive and log response ratio (푙푛푅푅) assumed 0 as the natural point of origin.

To consider a meta-analysis to be valid, it is required that the usual meta-analytic assumptions are satisfied. This means that the studies under consideration should be free of publication bias (a bias that occurs when the outcome of a research influences its publication), and that the effect sizes considered should approximately conform to the normal distribution. A visual assessment for publication bias can be made through funnel plots, where asymmetric funnels indicate small effect sizes to have lower publication probabilities. In order to detect bias or systematic heterogeneity, a funnel plot of treatment effect (AFS) was generated for each control (Appendix A, Figure A-9–

A-13). Usual meta-analytic assumptions include normality of the effect sizes, which in practice is visually examined via normal Q-Q plots (Figure A-9–A-13). Here, sample quantiles are plotted against theoretical quantiles from normal distribution and checked 194

if the data show significant departure from normality. Normal Q-Q plots, which plots sample quantiles of the effect sizes against standard normal quantiles, were examined to determine if the data showed significant departure from normality (a normally distributed data should lie mostly on the x = y line), (Appendix A, Figure A-9–A-13).

It is to be noted that a meta-analysis is highly susceptible to publication bias, a bias occurring in studies where the outcome of an experiment influences the quality of publication of the research. Often, experiments with only significant results are published, while several other studies with non-significant results remain unpublished.

Because meta-analyses aim at combining results from published research, an absence of studies with non-significant results adds bias to the analysis, thereby lessening reliability of its conclusions.

Rosenthal’s and Orwin’s Fail Safe Number

In practice, failsafe numbers are often computed to heuristically assess the amount of publication bias present in the studies. A failsafe number is an estimate of the number of additional studies, all with non-significant results, that are to be included in the analysis to make the combined effect size non-significant. In this study, we computed the Rosenthal’s and Orwin’s fail safe numbers (Orwin, 1983; Rosenthal,

1979). These methods estimated the number of additional studies, all with non- significant results, that needed to be included in the analysis to make the combined effect size non-significant (Appendix A, Tables A-4, A-8, A-12, A-16). A large failsafe number reinforces the confidence on the significance of the effect (Fragkos et al., 2014),

(Appendix A, Table A-4, A-8, A-12, A-16). Additionally, we evaluated more rigorous statistical tests such as Kendall’s rank correlation test and regression test (Appendix A,

Figure A-10–A-13). In this study, Kendall’s rank correlation test was run to test 195

publication bias as described by Begg & Mazumdar, (1994) (Appendix A, Figure A-10–

A-13). Kendall’s rank correlation is used to examine whether the observed outcomes and the corresponding sampling variances are correlated. A high correlation would indicate that the funnel plot is asymmetric, which may be a result of publication bias. A test of significance for the true correlation coefficient is performed to determine the presence of significant correlation.

Additionally, a regression test was performed using SE as predictors in a meta- analytical model. Regression test examines the dependence between the observed outcomes and a chosen predictor such as standard error, and a strong relationship

(assessed via a test of significance of associated regression coefficient) asymmetry in the funnel plot, which in turn may be an indication of publication bias (Egger et al.,

1997).

Figure A-11. Funnel Plot (a) and Normal Q-Q plot (b) of effect sizes of SOC stock changes for AFS vs. Agriculture.

Regression Test for Funnel Plot Asymmetry: t = -0.6314, df = 374, p = 0.5282

196

Figure A-12. Funnel Plot (a) and Normal Q-Q plot (b) of effect sizes of SOC stock changes for AFS vs. Forest.

Rank Correlation Test for Funnel Plot Asymmetry; Kendall's tau = 0.1833, p = 0.0021, Test for funnel plot asymmetry: t = -0.6872, df = 113, p = 0.4934

Figure A-13. Funnel Plot (a) and Normal Q-Q plot (b) of effect sizes of SOC stock changes for AFS vs. Pasture

Rank Correlation Test for Funnel Plot Asymmetry: Kendall's tau = -0.2510, p < .0001; Regression Test for Funnel Plot Asymmetry: t = -2.7136, df = 232, p = 0.0072

197

Figure A-14. Funnel Plot (a) and Normal Q-Q plot (b) of effect sizes of SOC stock changes for AFS vs. Uncultivated Land.

Rank Correlation Test for Funnel Plot Asymmetry: Kendall's tau = 0.4708, p < .0001; Regression Test for Funnel Plot Asymmetry: t = -0.4734, df = 64, p = 0.6375

Table A-1. Mixed Effects model, comparing effect size of SOC stock for land-use change from Forest to Agroforest (Control: Forest) logLik deviance AIC BIC AICc -37.2 74.4 110.4 159.7 117.6

Table A-2. Heterogeneity Estimator of SOC stocks for Forest to Agroforest conversion Heterogeneity Estimator Value

τ 2 (estimated amount of residual heterogeneity) 0.07 (SE = 0.012) τ (square root of estimated tau2 value) 0.27 I2 (residual heterogeneity/ unaccounted variability) 99.6% H2 (unaccounted variability/sampling variability) 255.5

R2 (amount of heterogeneity accounted for) 14.5%

198

Table A-3. ANOVA, comparing effect size of SOC stock for land-use change from Forest to Agroforest (Control: Forest) Source F-stat df1 df2 p-value Depth 2.674 5 114 0.02* Region 0.317 3 114 0.81 Region: Type of AFS 2.74 5 114 0.02 * ASA: Arid Agrosilvopasture 0.38 1 114 0.53 LHT: Multistrata Systems 2.4 1 114 0.12 TEM: Protective Systems 0.24 1 114 0.62 TEM: Temperate Agrisilviculture 2.53 1 114 0.11 LHT: Tropical Agrisilviculture 0.85 1 114 0.35 Age Group 0.98 1 114 0.40

¶df1 and df2 are the degrees of freedom parameters of the F statistics; Effect size significantly different from zero are highlighted; * p≤ 0.05, ** p≤ 0.01. Significant codes: 0 ‘***’ 0.001 ‘**’ 0.01 ‘*’ 0.05 ‘.’ 0.1 ‘ ’ 1

Table A-4. Rosenthal and Orwin’s test for publication bias (control: Forest) Fail Safe Approach Fail-safe Number (N) Rosenthal 1769 Orwin 131

Table A-5. Mixed Effects model, comparing effect size of SOC stock for land-use change from Agriculture to Agroforest (Control: Forest) logLik deviance AIC BIC AICc -218.5 437.0 485.0 579.3 488.5

199

Table A-6. Heterogeneity Estimator of SOC stocks for Agriculture to Agroforest conversion Heterogeneity Estimator Value τ 2 (estimated amount of residual heterogeneity) 0118 τ (square root of estimated tau2 value) 0.34 I2 (residual heterogeneity/ unaccounted variability) 99.9% H2 (unaccounted variability/sampling variability) 3275.7

R2 (amount of heterogeneity accounted for) 7.65%

Test for Residual Heterogeneity: QE (df = 375) = 24614.4, p-val < .0001

Table A-7. ANOVA, comparing effect size of SOC stock for land-use change from Agriculture to Agroforest (Control: Agriculture) Source F-stat df1 df2 p-value Depth 1.08 5 375 0.36 Region 6.0 3 375 0.0005 *** Region: Type of AFS 2.41 11 375 0.006 ** ASA: Arid Agrosilvopasture 0.87 1 375 0.14 ASA: Arid Silvopasture 2.1 1 375 0.0002*** LHT: Multistrata Systems 14.1 1 375 0.65 TEM: Protective Systems 0.19 1 375 0.11 MED: Med Agrisilviculture 0.31 1 375 0.57

MED: Med Silvoarable 0.32 1 375 0.91

MED:.Med Silvopasture 0.01 1 375 0.01 TEM: Temperate Agrisilviculture 0.07 1 375 0.79 MED: Temperate Silvoarable 0.34 1 375 0.55 LHT: Tropical Agrisilviculture 10.7 1 375 0.0012** LHT: Tropical Agrosilvopasture 20.33 1 375 0.0001*** Age Group 3.8 3 375 0.01*

200

Table A-8. Rosenthal and Orwin’s test for publication bias (control: Forest) Fail Safe Approach Fail-safe N Rosenthal 1034204 Orwin 399

Table A-9. Mixed Effects model, comparing effect size of SOC stock for land-use change from Pasture to Agroforest (Control: Pasture) logLik deviance AIC BIC AICc -106.9 213 253.9 322.9 257.8

Table A-10. Heterogeneity Estimator of SOC stocks for Pasture to Agroforest conversion Heterogeneity Estimator Value τ 2 (estimated amount of residual heterogeneity) 0.06 (SE = 0.009) τ (square root of estimated tau2 value) 0.24 I2 (residual heterogeneity/ unaccounted variability) 99.5% H2 (unaccounted variability/sampling variability) 200.7

R2 (amount of heterogeneity accounted for) 52.1%

Test for Residual Heterogeneity: QE (df = 233) = 6947.2, p-val < .0001

201

Table A-11. ANOVA, comparing effect size of SOC stock for land-use change from Pasture to Agroforest (Control: Pasture) Source F-stat df1 df2 p-value Depth 3.8 5 233 0.002** Region 2.8 3 233 <0.038* Region: Type of AFS 5.57 7 233 <0.001 *** ASA: Arid Agrosilvopasture 0.006 1 233 0.93 ASA: Arid Silvopasture 19.6 1 233 <0.001 *** MED: Med Agrosilvopasture 0.23 1 233 0.62 TEM: Temperate Agrisilviculture 0.46 1 233 0.49 LHT: Tropical Agrisilviculture 4.3 1 233 0.038* LHT: Tropical Agrosilvopasture 17.4 1 233 <0.001 *** Age Group 30.62 3 233 0.59

Table A-12. Rosenthal and Orwin’s test for publication bias (control: Pasture) Fail Safe Approach Fail-safe N Rosenthal 204125 Orwin 252

Table A-13. Mixed Effects model, comparing effect size of SOC stock for land-use change from Uncultivated Land to Agroforest (Control: Uncultivated Land). logLik deviance AIC BIC AICc -44.9 89.8 115.8 144.1 122.9

202

Table A-14. Heterogeneity Estimator of SOC stocks for Pasture to Agroforest conversion Heterogeneity Estimator Value τ2 (estimated amount of residual heterogeneity) 0.2 (SE = 0.04) τ (square root of estimated tau2 value) 0.44 I2 (residual heterogeneity/ unaccounted variability) 94.7% H2 (unaccounted variability/sampling variability) 18.87

R2 (amount of heterogeneity accounted for) 19.48% Test for Residual Heterogeneity: QE (df = 65) = 823.4, p-val < .0001

Table A-15. ANOVA, comparing effect size of SOC stock for land-use change from Pasture to Agroforest (Control: Uncultivated Land) Source F-stat df1 df2 p-value Depth 0.66 5 65 0.065 Region 13.1 1 65 0.0006*** Region: Type of AFS 3.07 2 65 0.03* LHT: Tropical Agrisilviculture 6.6 1 65 0.012*

LHT: Tropical Agrosilvopasture 0.91 1 65 0.34 Age Group 3.6 3 65 0.017* Model: Mixed Effects meta-regression model Predictor: standard error df1 and df2 are the degrees of freedom parameters of the F statistics; Effect size significantly different from zero are highlighted; * p≤ 0.05, ** p≤ 0.01. Significant codes: 0 ‘***’ 0.001 ‘**’ 0.01 ‘*’ 0.05 ‘.’ 0.1 ‘ ’ 1

Table A-16. Rosenthal and Orwin’s test for publication bias (control: Uncultivated Land) Fail Safe Approach Fail-safe N Rosenthal 157473 Orwin 77

203

LIST OF REFERENCES

Abaker, W.E., Berninger, F., Saiz, G., Braojos, V., Starr, M., 2016. Contribution of Acacia senegal to biomass and soil carbon in plantations of varying age in Sudan. For. Ecol. Manage. 368, 71–80. https://doi.org/10.1016/j.foreco.2016.03.003

Abou Rajab, Y., Leuschner, C., Barus, H., Tjoa, A., Hertel, D., 2016. Cacao Cultivation under Diverse Shade Tree Cover Allows High Carbon Storage and Sequestration without Yield Losses. PLoS One 11, e0149949. https://doi.org/10.1371/journal.pone.0149949

Adu, J.K., Oades, J.M., 1978. Physical factors influencing decomposition of organic materials in soil aggregates. Soil Biol. Biochem. 10, 109–115. https://doi.org/10.1016/0038-0717(78)90080-9

Aerts, R., Hundera, K., Berecha, G., Gijbels, P., Baeten, M., Van Mechelen, M., Hermy, M., Muys, B., Honnay, O., 2011. Semi-forest coffee cultivation and the conservation of Ethiopian Afromontane rainforest fragments. For. Ecol. Manage. 261, 1034–1041. https://doi.org/10.1016/j.foreco.2010.12.025

Ajit, Dhyani, S.K., Ramnewaj, Handa, A.K., Prasad, R., Alam, B., Rizvi, R.H., Gupta, G., Pandey, K.K., Jain, A., Uma, 2013. Modeling analysis of potential carbon sequestration under existing agroforestry systems in three districts of Indo- gangetic plains in India. Agrofor. Syst. 87. https://doi.org/10.1007/s10457-013- 9625-x

Albrecht, A., Kandji, S.T., 2003. Carbon sequestration in tropical agroforestry systems. Agric. Ecosyst. Environ. 99, 15–27. https://doi.org/10.1016/S0167- 8809(03)00138-5

Asase, A., Tetteh, D.A., 2016. Tree diversity, carbon stocks, and soil nutrients in cocoa- dominated and mixed food crops agroforestry systems compared to natural forest in southeast Ghana. Agroecol. Sustain. Food Syst. 40, 96–113. https://doi.org/10.1080/21683565.2015.1110223

Asbjornsen, H., Hernandez-Santana, V., Liebman, M.Z., Bayala, J., Chen, J., Helmers, M., Ong, C.K., Schulte, L. a, 2013. Targeting perennial vegetation in agricultural landscapes for enhancing ecosystem services. Renew. Agric. Food Syst. 29, 101–125. https://doi.org/10.1017/S1742170512000385

Baccini, A., Goetz, S.J., Walker, W.S., Laporte, N.T., Sun, M., Sulla-Menashe, D., Hackler, J., Beck, P.S.A., Dubayah, R., Friedl, M.A., Samanta, S., Houghton, R.A., 2012. Estimated carbon dioxide emissions from tropical deforestation improved by carbon-density maps. Nat. Clim. Chang. 2, 182–185. https://doi.org/10.1038/nclimate1354

204

Bárcena, T.G., Gundersen, P., Vesterdal, L., 2014. Afforestation effects on SOC in former cropland: Oak and spruce chronosequences resampled after 13 years. Glob. Chang. Biol. 20, 2938–2952. https://doi.org/10.1111/gcb.12608

Barreto, P.A.B., Gama-Rodrigues, E.F., Gama-Rodrigues, A.C., Fontes, A.G., Polidoro, J.C., Moço, M.K.S., Machado, R.C.R., Baligar, V.C., 2011. Distribution of oxidizable organic C fractions in soils under cacao agroforestry systems in Southern Bahia, Brazil. Agrofor. Syst. 81, 213–220. https://doi.org/10.1007/s10457-010-9300-4

Batjes, N.H., 1996. Total carbon and nitrogen in the soils of the world. Eur. J. Soil Sci. 47, 151–163. https://doi.org/10.1111/j.1365-2389.1996.tb01386.x

Beer, J., 1988. Litter production and nutrient cycling in coffee (Coffea arabica) or cacao (Theobroma cacao) plantations with shade trees. Agrofor. Syst. 7, 103–114. https://doi.org/10.1007/BF00046846

Beer, J., Bonnemann, a, Chavez, W., Fassbender, H.W., Imbach, a C., Martel, I., 1990. Modeling Agroforestry Systems of Cacao (Theobroma cacao) With Laurel (Cordia alliodora) or Poro (Erythrina poeppigiana), Productivity Indexes, Organic Material Models and Sustainability over 10 Years in Costa Rica. Agrofor. Syst. 12, 229–249. https://doi.org/10.1007/bf00137286

Beer, J., Muschler, R., Kass, D., Somarriba, E., 1998. Shade management in coffee and cacao plantations. Agrofor. Syst. 38, 139–164. https://doi.org/10.1023/A:1005956528316

Begg, C.B., Mazumdar, M., 1994. Operating Characteristics of a Rank Correlation Test for Publication Bias. Biometrics 50, 1088. https://doi.org/10.2307/2533446

Berthrong, S.T., Jobbagy, E.G., Jackson, R.B., 2009. A global meta-analysis of soil exchangeable cations, pH, carbon, and nitrogen with afforestation. Ecol. Appl. 19, 2228–2241. https://doi.org/10.1890/08-1730.1

Berthrong, S.T., Piñeiro, G., Jobbágy, E.G., Jackson, R.B., 2012. Soil C and N changes with afforestation of grasslands across gradients of precipitation and plantation age. Ecol. Appl. 22, 76–86. https://doi.org/10.1890/10-2210.1

Boddey, R.M., Jantalia, C.P., Conceiçao, P.C., Zanatta, J.A., Bayer, C., Mielniczuk, J., Dieckow, J., Dos Santos, H.P., Denardin, J.E., Aita, C., Giacomini, S.J., Alves, B.J.R., Urquiaga, S., 2010. Carbon accumulation at depth in Ferralsols under zero-till subtropical agriculture. Glob. Chang. Biol. 16, 784–795. https://doi.org/10.1111/j.1365-2486.2009.02020.x

Borenstein, M., Hedges, L., Rothstein, H., 2009. Introduction to Meta-Analysis, in: Introduction to Meta-Analysis. John Wiley, Chichester, UK. https://doi.org/10.1016/B978-0-240-81203-8.00002-7

205

Borenstein, M., Hedges, L., Rothstein, H., 2007. Meta-Analysis Fixed effect vs .Random effects, in: Introduction to Meta-Analysis. p. 162. https://doi.org/10.1002/9780470743386

Bronick, C.J., Lal, R., 2005. Soil structure and management: A review. Geoderma 124, 3–22. https://doi.org/10.1016/j.geoderma.2004.03.005

Buresh, R.J., Cooper, P.J.M., 1999. The science and practice of short-term improved fallows : symposium synthesis and recommendations. Agrofor. Syst. 47, 345– 356. https://doi.org/10.1023/A:1006200420863

Burgess, P.J., Incoll, L.D., Corry, D.T., Beaton, A., Hart, B.J., 2005. Poplar (Populus spp) growth and crop yields in a silvoarable experiment at three lowland sites in England. Agrofor. Syst. 63, 157–169. https://doi.org/10.1007/s10457-004-7169-9

Cadotte, M.W., 2013. Experimental evidence that evolutionarily diverse assemblages result in higher productivity. Proc. Natl. Acad. Sci. 110, 8996–9000. https://doi.org/10.1073/pnas.1301685110

Cannell, M.G.R., Kimeu, B.S., 1971. Uptake and distribution of macro-nutrients in trees of Coffea arabica. Ann. appl. Biol. 68, 213–230.

Cao, Z. yuan, Wang, Y., Li, J., Zhang, J. jing, He, N. peng, 2016. Soil organic carbon contents, aggregate stability, and humic acid composition in different alpine grasslands in Qinghai-Tibet Plateau. J. Mt. Sci. 13, 2015–2027. https://doi.org/10.1007/s11629-015-3744-y

Cardinael, R., Chevallier, T., Barthès, B.G., Saby, N.P.A., Parent, T., Dupraz, C., Bernoux, M., Chenu, C., 2015. Impact of alley cropping agroforestry on stocks, forms and spatial distribution of soil organic carbon - A case study in a Mediterranean context. Geoderma 259–260, 288–299. https://doi.org/10.1016/j.geoderma.2015.06.015

Cardinael, R., Chevallier, T., Cambou, A., Béral, C., Barthès, B.G., Dupraz, C., Durand, C., Kouakoua, E., Chenu, C., 2017. Increased soil organic carbon stocks under agroforestry: A survey of six different sites in France. Agric. Ecosyst. Environ. 236, 243–255. https://doi.org/10.1016/j.agee.2016.12.011

Cardinael, R., Thevathasan, N., Gordon, A., Clinch, R., Mohammed, I., Sidders, D., 2012. Growing woody biomass for bioenergy in a tree-based intercropping system in southern Ontario, Canada. Agrofor. Syst. 86, 279–286. https://doi.org/10.1007/s10457-012-9572-y

Carter, M.R., Gregorich, E.G., 2010. Carbon and nitrogen storage by deep-rooted tall fescue (Lolium arundinaceum) in the surface and subsurface soil of a fine sandy loam in eastern Canada. Agric. Ecosyst. Environ. 136, 125–132. https://doi.org/10.1016/j.agee.2009.12.005

206

Celentano, D., Zahawi, R.A., Finegan, B., Ostertag, R., Cole, R.J., Holl, K.D., 2011. Litterfall dynamics under different tropical forest restoration strategies in Costa Rica. Biotropica 43, 279–287. https://doi.org/10.1111/j.1744-7429.2010.00688.x

Cerda, R., Allinne, C., Gary, C., Tixier, P., Harvey, C.A., Krolczyk, L., Mathiot, C., Clément, E., Aubertot, J., Avelino, J., 2017. Effects of shade , altitude and management on multiple ecosystem services in coffee agroecosystems. Eur. J. Agron. 82, 308–319. https://doi.org/10.1016/j.eja.2016.09.019

Chaney, K., Swift, R.S., 1984. The influence of organic matter on aggregate stability in some Britisch soils. J. Soil Sci. 35, 223–230.

Chen, C., Liu, W., Jiang, X., Wu, J., 2017. Effects of rubber-based agroforestry systems on soil aggregation and associated soil organic carbon : Implications for land use. Geoderma 299, 13–24. https://doi.org/10.1016/j.geoderma.2017.03.021

Chen, J.M., Govind, A., Sonnentag, O., Zhang, Y., Barr, A., Amiro, B., 2006. Leaf area index measurements at Fluxnet-Canada forest sites. Agric. For. Meteorol. 140, 257–268. https://doi.org/10.1016/j.agrformet.2006.08.005

Chen, X., Hutley, L.B., Eamus, D., 2005. Soil organic carbon content at a range of north Australian tropical savannas with contrasting site histories. Plant Soil 268, 161– 171. https://doi.org/10.1007/s11104-004-0249-9

Chenu, C., Le Bissonnais, Y., Arrouays, D., 2000. Organic Matter Influence on Clay Wettability and Soil Aggregate Stability. Soil Sci. Soc. Am. J. 64, 1479. https://doi.org/10.2136/sssaj2000.6441479x

Chenu, C., Plante, A.T., 2006. Clay-sized organo-mineral complexes in a cultivation chronosequence: Revisiting the concept of the “primary organo-mineral complex.” Eur. J. Soil Sci. 57, 596–607. https://doi.org/10.1111/j.1365- 2389.2006.00834.x

Chesney, P., Nygren, P., 2002. Fine root and nodule dynamics of Erythrina poeppigiana in an alley cropping system in Costa Rica. Agrofor. Syst. 56, 259–269.

Chesney, P.E., Chesney, P., 2008. Nitrogen and fine root length dynamics in a tropical agroforestry system with periodically pruned Erythrina poeppigiana. Agrofor. Syst. 72, 149–159 https://doi.org/10.1007/s10457-007-9064-7

Ciais, P., Sabine, C., Bala, G., Bopp, L., Brovkin, V., Canadell, J., Chhabra, A., DeFries, R., Galloway, J., Heimann, M., Jones, C., Quéré, C. Le, Myneni, R.B., Piao, S., Thornton, P., 2013. The physical science basis. Contribution of working group I to the fifth assessment report of the intergovernmental panel on climate change. Chang. IPCC Clim. 465–570. https://doi.org/10.1017/CBO9781107415324.015

207

Conant, R.T., 2010. Challenges and opportunities for carbon sequestration in grassland systems, Integrated Crop Management. 9, 1–55. https://doi.org/10.3329/jard.v7i1.4430

Conant, R.T., Paustian, K., Elliott, E.T., 2001. Grassland management and conversion into grassland: Effects on soil carbon. Ecol. Appl. 11, 343–355. https://doi.org/10.1890/1051-0761(2001)011[0343:GMACIG]2.0.CO;2

Cuenca, G., Aranguren, J., Herrera, R., 1983. Root growth and litter decomposition in a coffee plantation under shade trees. Plant Soil 71, 477–486. https://doi.org/10.1007/BF02182689

De Beenhouwer, M., Aerts, R., Honnay, O., 2013. A global meta-analysis of the biodiversity and ecosystem service benefits of coffee and cacao agroforestry. Agric. Ecosyst. Environ. 175, 1–7. https://doi.org/10.1016/j.agee.2013.05.003

De Stefano, A., Jacobson, M.G., 2017. Soil carbon sequestration in agroforestry systems: a meta-analysis. Agrofor. Syst. 1–15. https://doi.org/10.1007/s10457- 017-0147-9

Dechert, G., Veldkamp, E., Anas, I., 2004. Is soil degradation unrelated to deforestation? Examining soil parameters of land use systems in upland Central Sulawesi, Indonesia. Plant Soil 265, 197–209. https://doi.org/10.1007/s11104- 005-0885-8

Defrenet, E., Roupsard, O., Meersche, K. Van Den, Charbonnier, F., Stokes, A., Roumet, C., Rapidel, B., Melo, E. De, Filho, V., Vargas, V.J., Robelo, D., Barquero, A., Jourdan, C., 2016. Root biomass , turnover and net primary productivity of a coffee agroforestry system in Costa Rica : effects of soil depth , shade trees , distance to row and coffee age. Annals. Bot. 118, 833–851. https://doi.org/10.1093/aob/mcw153

DeFries, R.S., Rudel, T., Uriarte, M., Hansen, M., 2010. Deforestation driven by urban population growth and agricultural trade in the twenty-first century. Nat. Geosci 3, 178–181. https://doi.org/10.1038/ngeo756

Dietzel, R., Liebman, M., Archontoulis, S., 2017. A deeper look at the relationship between root carbon pools and the vertical distribution of the soil carbon pool. SOIL. 3, 139–152.

Dixon, R.K., 1995. Agroforestry systems: sources of sinks of greenhouse gases? Agrofor. Syst. 31, 99–116. https://doi.org/10.1007/bf00711719

Don, A., Schumacher, J., Freibauer, A., 2011. Impact of tropical land-use change on soil organic carbon stocks - a meta-analysis. Glob. Chang. Biol. 17, 1658–1670. https://doi.org/10.1111/j.1365-2486.2010.02336.x

208

Donald, P.F., 2004. Biodiversity Impacts of Some Agricultural Commodity Production Systems. Conserv. Biol. 18, 17–37. https://doi.org/10.1111/j.1523- 1739.2004.01803.x

Dossa, E.L., Fernandes, E.C.M., Reid, W.S., Ezui, K., 2008. Above- and belowground biomass, nutrient and carbon stocks contrasting an open-grown and a shaded coffee plantation. Agrofor. Syst. 72, 103–115. https://doi.org/10.1007/s10457- 007-9075-4

Egger, M., Smith, G.D., Schneider, M., Minder, C., 1997. Bias in meta-analysis detected by a simple, graphical test. BMJ 315, 629.

Ehrenbergerová, L., Cienciala, E., Kučera, A., Guy, L., Habrová, H., 2015. Carbon stock in agroforestry coffee plantations with different shade trees in Villa Rica, Peru. Agrofor. Syst. https://doi.org/10.1007/s10457-015-9865-z

Ellert, B.H., Bettany, J.R., 1995. Calculation of organic matter and nutrients stored in soils under contrasting management regimes. Can. J. Soil Sci. 75, 529–538. https://doi.org/10.4141/cjss95-075

Elliott, E.T., 1986. Aggregate Structure and Carbon, Nitrogen, and Phosphorus in Native and Cultivated Soils. Soil Sci .Soc.Am.J. 50, 627–633. https://doi.org/10.2136/sssaj1986.03615995005000030017x

FAO, 2015. Global Forest Resources Assessment 2015, FAO Forestry. https://doi.org/10.1002/2014GB005021

Fassbender, H., Modelos edafologicos de sistemas agroforestales en Proyecto agroforestal CATIE'CTZ. CATIE , 1993

Feliciano, D., Ledo, A., Hillier, J., Rani, D., 2018. Agriculture , Ecosystems and Environment Which agroforestry options give the greatest soil and above ground carbon bene fi ts in di ff erent world regions ? Agric. Ecosyst. Environ. 254, 117– 129. https://doi.org/10.1016/j.agee.2017.11.032

Foley, J.A., Ramankutty, N., Brauman, K.A., Cassidy, E.S., Gerber, J.S., Johnston, M., Mueller, N.D., Connell, C.O., Ray, D.K., West, P.C., Balzer, C., Bennett, E.M., Sheehan, J., Siebert, S., Carpenter, S.R., Hill, J., Monfreda, C., Polasky, S., Rockstro, J., Tilman, D., Zaks, D.P.M., 2011. Solutions for a cultivated planet.Nature. 478, 337–342. https://doi.org/10.1038/nature10452

Fontaine, S., Barot, S., Barré, P., Bdioui, N., Mary, B., Rumpel, C., 2007. Stability of organic carbon in deep soil layers controlled by fresh carbon supply. Nature 450, 277–80. https://doi.org/10.1038/nature06275

Fragkos, K.C., Tsagris, M., Frangos, C.C., 2014. Publication Bias in Meta-Analysis: Confidence Intervals for Rosenthal’s Fail-Safe Number. Int. Sch. Res. Not. 2014, 1–17. https://doi.org/10.1155/2014/825383

209

Freibauer, A., Rounsevell, M.D.A., Smith, P., Verhagen, J., 2004. Carbon sequestration in the agricultural soils of Europe. Geoderma 122, 1–23. https://doi.org/10.1016/j.geoderma.2004.01.021

Friedrich, J.O., Adhikari, N.K., Beyene, J., 2008. The ratio of means method as an alternative to mean differences for analyzing continuous outcome variables in meta-analysis: A simulation study. BMC Med. Res. Methodol. 8, 32. https://doi.org/10.1186/1471-2288-8-32

Gama-Rodrigues, E.F., Nair, P.K.R., Nair, V.D., Gama-Rodrigues, A.C., Baligar, V.C., Machado, R.C.R., 2010. Carbon storage in soil size fractions under Two cacao agroforestry systems in Bahia, Brazil. Environ. Manage. 45, 274–283. https://doi.org/10.1007/s00267-009-9420-7

Garces, A.M., 2011. Characterization of the spatial variability of soil properties and Coffee fine roots in shade tree- coffee associations under organic and conventional management practices (Doctoral dissertation). https://www.catie.ac.cr/Tesis-PhD-AMora-2011

Gee, J. W. Bauder, G. W. (1979). Particle size analysis by hydrometer: a simplified method for routine textural analysis and a sensitivity test of measurement parameters. Soil Sci. Soc of America Journal, 43, 1004–1007. https://doi.org/10.2136/sssaj1979.03615995004300050038x

Ghimire, B., Ghimire, R., VanLeeuwen, D., Mesbah, A., 2017. Cover Crop Residue Amount and Quality Effects on Soil Organic Carbon Mineralization. Sustainability 9, 2316. https://doi.org/10.3390/su9122316

Guo, L.B., Gifford, R.M., 2002. Soil carbon stocks and land use change: A meta analysis. Glob. Chang. Biol. 8, 345–360. https://doi.org/10.1046/j.1354- 1013.2002.00486.x

Gupta, N., Kukal, S.S., Bawa, S.S., Dhaliwal, G.S., 2009. Soil organic carbon and aggregation under poplar based agroforestry system in relation to tree age and soil type. Agrofor. Syst. 76, 27–35. https://doi.org/10.1007/s10457-009-9219-9

Häger, A., 2012. The effects of management and plant diversity on carbon storage in coffee agroforestry systems in Costa Rica. Agrofor. Syst. 86, 159–174. https://doi.org/10.1007/s10457-012-9545-1

Haggar, J., Barrios, M., Bolaños, M., Merlo, M., Moraga, P., Munguia, R., Ponce, A., Romero, S., Soto, G., Staver, C., de Virginio, E.M.F., 2011. Coffee agroecosystem performance under full sun, shade,conventional and organic management regimes in Central America. Agrofor. Syst. 82, 285–301. https://doi.org/10.1007/s10457-011-9392-5

210

Haile, S.G., Nair, P.K.R., Nair, V.D., 2008. Carbon Storage of Different Soil-Size Fractions in Florida Silvopastoral Systems All rights reserved. No part of this periodical may be reproduced or transmitted in any form or by any means, electronic or mechanical, including photocopying, recording, or . J. Environ. Qual. 37, 1789–1797. https://doi.org/10.2134/jeq2007.0509

Haile, S.G., Nair, V.D., Nair, P.K.R., 2010. Contribution of trees to carbon storage in soils of silvopastoral systems in Florida, USA. Glob. Chang. Biol. 16, 427–438. https://doi.org/10.1111/j.1365-2486.2009.01981.x

Hassink, J., 1997. The capacity of soils to preserve organic C and N by their association with silt and clay particles. Soil Sci. Soc. Am. J. 61, 131–139.

Hassink, J., 1994. Effects of soil texture and grassland management on soil organic C and N and rates of C and N mineralization. Soil. Biol. bBiochem. 26, 1221–1231.

Hergoualc’h, K., Blanchart, E., Skiba, U., Hénault, C., Harmand, J.M., 2012. Changes in carbon stock and greenhouse gas balance in a coffee (Coffea arabica) monoculture versus an agroforestry system with Inga densiflora, in Costa Rica. Agric. Ecosyst. Environ. 148, 102–110. https://doi.org/10.1016/j.agee.2011.11.018

Hertel, D., Harteveld, M.A., Leuschner, C., 2009. Conversion of a tropical forest into agroforest alters the fine root-related carbon flux to the soil. Soil Biol. Biochem. 41, 481–490. https://doi.org/10.1016/j.soilbio.2008.11.020

Hombegowda, H.C., van Straaten, O., Köhler, M., Hölscher, D., 2015. On the rebound: soil organic carbon stocks can bounce back to near forest levels when agroforests replace agriculture in southern India. SOIL Discuss. 2, 871–902. https://doi.org/10.5194/soild-2-871-2015

Hothorn, T., Bretz, F., Westfall, P., 2008. Simultaneous inference in general parametric models. Biometrical J. 50, 346–363. https://doi.org/10.1002/bimj.200810425

Howlett, D.S., Moreno, G., Mosquera Losada, M.R., Nair, P.K.R., Nair, V.D., 2011. Soil carbon storage as influenced by tree cover in the Dehesa cork oak silvopasture of central-western Spain. J. Environ. Monit. 13, 1897–1904. https://doi.org/10.1039/c1em10059a

Howlett, D.S., Mosquera-Losada, M.R., Nair, P.K.R., Nair, V.D., Rigueiro-Rodríguez, A., 2011. Soil carbon storage in silvopastoral systems and a treeless pasture in northwestern Spain. J. Environ. Qual. 40, 825–832. https://doi.org/10.2134/jeq2010.0145

IPCC, 2014. Summary for Policymakers, Climate Change 2014: Synthesis Report. Contribution of Working Groups I, II and III to the Fifth Assessment Report of the Intergovernmental Panel on Climate Change. https://doi.org/10.1017/CBO9781107415324

211

Isaac, M.E., Gordon, A.M., Thevathasan, N., Oppong, S.K., Quashie-Sam, J., 2005. Temporal changes in soil carbon and nitrogen in west African multistrata agroforestry systems: A chronosequence of pools and fluxes. Agrofor. Syst. 65, 23–31. https://doi.org/10.1007/s10457-004-4187-6

Jackson, R.B., Canadell, J., Ehleringer, J.R., Mooney, H. A., Sala, O.E., Schulze, E.D., 1996. A global analysis of root distributions for terrestrial biomes. Oecologia 108, 389–411. https://doi.org/10.1007/BF00333714

Jangra, R., Gupta, S.R., Kumar, R., Singh, G., 2010. Carbon sequestration in the Grevillea robusta plantation on a reclaimed sodic soil at Karnal in Northern India. Int. J. Ecol. Environ. Sci. 36, 75–86.

Jastrow, J.D., Amonette, J.E., Bailey, V.L., 2007. Mechanisms controlling soil carbon turnover and their potential application for enhancing carbon sequestration. Clim. Change 80, 5–23. https://doi.org/10.1007/s10584-006-9178-3

Jobbágy, E.G., Jackson, R.B., 2000. The vertical distribution of soil organic carbon and its relation to climate and vegetation. Ecol. Appl. 10, 423–436. https://doi.org/10.1890/1051-0761(2000)010[0423:TVDOSO]2.0.CO;2

Jones, D.L., Nguyen, C., Finlay, R.D., 2009. Carbon flow in the rhizosphere: Carbon trading at the soil-root interface. Plant Soil 321, 5–33. https://doi.org/10.1007/s11104-009-9925-0

Jose, S., Bardhan, S., 2012. Agroforestry for biomass production and carbon sequestration: An overview. Agrofor. Syst. 86, 105–111. https://doi.org/10.1007/s10457-012-9573-x

Kaonga, M.L., Coleman, K., 2008. Modelling soil organic carbon turnover in improved fallows in eastern Zambia using the RothC-26.3 model. For. Ecol. Manage. 256, 1160–1166. https://doi.org/10.1016/j.foreco.2008.06.017

Kim, D.G., Terefe, B., Girma, S., Kedir, H., Morkie, N., Woldie, T.M., 2016. Conversion of home garden agroforestry to crop fields reduced soil carbon and nitrogen stocks in Southern Ethiopia. Agrofor. Syst. 90, 251–264. https://doi.org/10.1007/s10457-015-9851-5

Kirby, K.R., Potvin, C., 2007. Variation in carbon storage among tree species: Implications for the management of a small-scale carbon sink project. For. Ecol. Manage. 246, 208–221. https://doi.org/10.1016/j.foreco.2007.03.072

Knapp, G., Hartung, J., 2003. Improved tests for a random effects meta-regression with a single covariate. Stat. Med. 22, 2693–2710. https://doi.org/10.1002/sim.1482

Laganière, J., Angers, D.A., Paré, D., 2010. Carbon accumulation in agricultural soils after afforestation: A meta-analysis. Glob. Chang. Biol. 16, 439–453. https://doi.org/10.1111/j.1365-2486.2009.01930.x

212

Lal, R., 2005. Soil carbon sequestration in natural and managed tropical forest ecosystems. J. Sustain. For. 21, 1–30. https://doi.org/10.1300/J091v21n01

Lal, R., 2004. Soil carbon sequestration to mitigate climate change. Geoderma 123, 1– 22. https://doi.org/10.1016/j.geoderma.2004.01.032

Lal, R., Negassa, W., Lorenz, K., 2015. Carbon sequestration in soil. Curr. Opin. Environ. Sustain. 15, 79–86. https://doi.org/10.1016/j.cosust.2015.09.002

Lehmann, J., 2007. A handful of carbon. Nature 447, 143–144. https://doi.org/10.1038/447143a

Lenth, R. V, 2016. Least-Squares Means: The R Package lsmeans. J. Stat. Software, 69(1), 1–33.

Leuschner, C., Moser, G., Hertel, D., Erasmi, S., Leitner, D., Culmsee, H., Schuldt, B., Schwendenmann, L., 2013. Conversion of tropical moist forest into cacao agroforest: Consequences for carbon pools and annual C sequestration. Agrofor. Syst. 87, 1173–1187. https://doi.org/10.1007/s10457-013-9628-7

Liu, N., Wang, Y., Wang, Y., Zhao, Z., Zhao, Y., 2016. Tree species composition rather than biodiversity impacts forest soil organic carbon of Three Gorges , southwestern China. Nat. Con.14, 7–24. https://doi.org/10.3897/natureconservation.14.6486

Liu, Z., Shao, M., Wang, Y., 2011. Effect of environmental factors on regional soil organic carbon stocks across the Loess Plateau region, China. Agric. Ecosyst. Environ. 142, 184–194. https://doi.org/10.1016/j.agee.2011.05.002

Lorenz, K., Lal, R., 2014. Soil organic carbon sequestration in agroforestry systems. A review. Agron. Sustain. Dev. 34, 443–454. https://doi.org/10.1007/s13593-014- 0212-y

Lorenz, K., Lal, R., 2005. The Depth Distribution of Soil Organic Carbon in Relation to Land Use and Management and the Potential of Carbon Sequestration in Subsoil Horizons. Adv. Agron. 88, 35–66. https://doi.org/10.1016/S0065-2113(05)88002- 2

Mackey, B., Prentice, I.C., Steffen, W., House, J.I., Lindenmayer, D., Keith, H., Berry, S., 2013. Untangling the confusion around land carbon science and climate change mitigation policy. Nat. Clim. Chang. 3, 552–557. https://doi.org/10.1038/NCLIMATE1804

Mbow, C., Smith, P., Skole, D., Duguma, L., Bustamante, M., 2014. Achieving mitigation and adaptation to climate change through sustainable agroforestry practices in africa. Curr. Opin. Environ. Sustain. 6, 8–14. https://doi.org/10.1016/j.cosust.2013.09.002

213

MEA, 2005. Ecosystems and human well-being, Millennium Ecosystem Assessment: Ecosystems and Human Well-Being: Synthesis. Washington, DC: Island Press. https://doi.org/10.1196/annals.1439.003

Méndez, V.E., Shapiro, E., Gilbert, G.S., 2009. Cooperative management and its effects on shade tree diversity , soil properties and ecosystem services of coffee plantations in western El Salvador. Agrofor. Syst. 76, 111–126. https://doi.org/10.1007/s10457-009-9220-3

Meylan, L., Gary, C., Allinne, C., Ortiz, J., Jackson, L., Rica, C., 2017. Agriculture , Ecosystems and Environment Evaluating the e ff ect of shade trees on provision of ecosystem services in intensively managed co ff ee plantations. Agric. Ecosyst. Environ. 245, 32–42. https://doi.org/10.1016/j.agee.2017.05.005

Moguel, P., Toledo, V.M., 1999. Biodiversity conservation in tradicional Coffee systems of Mexico. Conserv. Biol. 13, 11–21. https://doi.org/10.1046/j.1523- 1739.1999.97153.x

Moher, D., Liberati, A., Tetzlaff, J., Altman, D.G., Altman, D., Antes, G., Atkins, D., Barbour, V., Barrowman, N., Berlin, J.A., Clark, J., Clarke, M., Cook, D., D’Amico, R., Deeks, J.J., Devereaux, P.J., Dickersin, K., Egger, M., Ernst, E., G??tzsche, P.C., Grimshaw, J., Guyatt, G., Higgins, J., Ioannidis, J.P.A., Kleijnen, J., Lang, T., Magrini, N., McNamee, D., Moja, L., Mulrow, C., Napoli, M., Oxman, A., Pham, B., Rennie, D., Sampson, M., Schulz, K.F., Shekelle, P.G., Tovey, D., Tugwell, P., 2009. Preferred reporting items for systematic reviews and meta-analyses: The PRISMA statement (Chinese edition). J. Chinese Integr. Med. 7, 889–896. https://doi.org/10.3736/jcim20090918

Monroe, P.H.M., Gama-Rodrigues, E.F., Gama-Rodrigues, A.C., Marques, J.R.B., 2016. Soil carbon stocks and origin under different cacao agroforestry systems in Southern Bahia, Brazil. Agric. Ecosyst. Environ. 221, 99–108. https://doi.org/10.1016/j.agee.2016.01.022

Montagnini, F., 2000. Accumulation in above-ground biomass and soil storage of mineral nutrients in pure and mixed plantations in a humid tropical lowland. For. Ecol.Manag. 134, 257–270.

Montagnini, F., Nair, P.K.R., 2004. Carbon sequestration: An underexploited environmental benefit of agroforestry systems. Agrofor. Syst. 61, 281–295. https://doi.org/10.1023/B:AGFO.0000029005.92691.79

Moreira, C., Eleto, M., Antônio, L., Jacovine, G., Nolasco, S., Neto, D.O., Fraisse, C.W., Pedro, C., Soares, B., Neto, F.D.C., Ferreira, L.R., Zanuncio, J.C., 2017. Greenhouse gas emissions and carbon sequestration by agroforestry systems in southeastern Brazil. Sci. Rep. 7–16738.. https://doi.org/10.1038/s41598-017- 16821-4

214

Morgan, J. A., Follett, R.F., Allen, L.H., Del Grosso, S., Derner, J.D., Dijkstra, F., Franzluebbers, a., Fry, R., Paustian, K., Schoeneberger, M.M., 2010. Carbon sequestration in agricultural lands of the United States. J. Soil Water Conserv. 65, 6–13. https://doi.org/10.2489/jswc.65.1.6A

Mungai, N.W., Motavalli, P.P., Kremer, R.J., 2006. Soil organic carbon and nitrogen fractions in temperate alley cropping systems. Commun. Soil Sci. Plant Anal. 37, 977–992. https://doi.org/10.1080/00103620600584230

Muschler, R.G., 2001. Shade improves coffee quality in a sub-optimal coffee-zone of Costa Rica. Agrofor. Syst. 51, 131–139. https://doi.org/10.1023/A:1010603320653

Nair, P.K.R., 2017. Managed Multi-strata Tree + Crop Systems: An Agroecological Marvel. Front. Environ. Sci. 5, 1–5. https://doi.org/10.3389/fenvs.2017.00088

Nair, P.K.R., 1993. Classification of agroforesty systems, An introduction to agroforestry. https://doi.org/https://www.worldagroforestry.org/Units/Library/Books/PDFs/32_A n_introduction_to_agroforestry.pdf?n=161

Nair, P.K.R., 2012a. Carbon sequestration studies in agroforestry systems: A reality- check. Agrofor. Syst. 86, 243–253. https://doi.org/10.1007/s10457-011-9434-z

Nair, P. K. R., 2012b. Climate change mitigation and adaptation: A low hanging fruit of agroforestry. In Nair, P. K. R. and Garrity D. P. (eds), Agroforestry: The future of global land use. Springer, Dordrecht, The Netherlands. pp 31–67

Nair, P.K.R., 2011. Agroforestry systems and environmental quality: introduction. J. Environ. Qual. 40, 784–790. https://doi.org/10.2134/jeq2011.0076

Nair, P.K.R., Kumar, B.M., Nair, V.D., 2009. Agroforestry as a strategy for carbon sequestration. J. Plant Nutr. Soil Sci. 172, 10–23. https://doi.org/10.1002/jpln.200800030

Nair, P.K.R., Nair, V.D., Kumar, B.M.,Showalter, J.M., 2010. Carbon sequestration in agroforestry systems. Adv.in.Agron. 108, 237–307.

Nair, P.K.R., Saha, S.K., Stein, T. V, Andreu, M.G., Luedeling, E., Sileshi, G., Beedy, T., Dietz, J., Ruizdeoñaplaza, C., Soto-pinto, L., Paladino, S., Morales, F., Esquivel, E., Tonucci, R.G., Garcia, R., 2011. Carbon Sequestration Potential of Agroforestry Systems. Media 8, 61–83. https://doi.org/10.1007/978-94-007-1630- 8

Nair, P.K,R., Nair, V.D., 2014. Solid-fluid-gas: The state of knowledge on carbon- sequestration potential of agroforestry systems in africa. Curr. Opin. Environ. Sustain. 6, 22–27. https://doi.org/10.1016/j.cosust.2013.07.014

215

Nath, C.D., Schroth, G., Burslem, D.F.R.P., 2016. Why do farmers plant more exotic than native trees? A case study from the Western Ghats, India. Agric. Ecosyst. Environ. 230, 315–328. https://doi.org/10.1016/j.agee.2016.05.013

Negash, M., Starr, M., 2015. Biomass and soil carbon stocks of indigenous agroforestry systems on the south-eastern Rift Valley escarpment, Ethiopia. Plant Soil 393, 95–107. https://doi.org/10.1007/s11104-015-2469-6

Niranjana, K.S., Viswanath, S., 2008. Root characteristics of tea [Camellia sinensis (L.) O. Kuntze] and silver oak [Grevillea robusta (A. Cunn)] in a mixed tea plantation at Munnar, Kerala. J. Trop. Agric. 46, 13–19.

Noponen, M.R.A., Edwards-Jones, G., Haggar, J.P., Soto, G., Attarzadeh, N., Healey, J.R., 2012. Greenhouse gas emissions in coffee grown with differing input levels under conventional and organic management. Agric. Ecosyst. Environ. 151, 6– 15. https://doi.org/10.1016/j.agee.2012.01.019

Noponen, M.R.A., Healey, J.R., Soto, G., Haggar, J.P., 2013. Sink or source-The potential of coffee agroforestry systems to sequester atmospheric CO2 into soil organic carbon. Agric. Ecosyst. Environ. 175, 60–68. https://doi.org/10.1016/j.agee.2013.04.012

Norgrove, L., Hauser, S., 2013. Carbon stocks in shaded Theobroma cacao farms and adjacent secondary forests of similar age in Cameroon. Trop. Ecol. 54, 15–22.

Oelbermann, M., Voroney, R.P., 2011. An evaluation of the century model to predict soil organic carbon: Examples from Costa Rica and Canada. Agrofor. Syst. 82, 37– 50. https://doi.org/10.1007/s10457-010-9351-6

Oelbermann, M., Voroney, R.P., 2007. Carbon and nitrogen in a temperate agroforestry system: Using stable isotopes as a tool to understand soil dynamics. Ecol. Eng. 29, 342–349. https://doi.org/10.1016/j.ecoleng.2006.09.014

Oelbermann, M., Voroney, R.P., Thevathasan, N. V., Gordon, A.M., Kass, D.C.L., Schlönvoigt, A.M., 2006. Soil carbon dynamics and residue stabilization in a Costa Rican and southern Canadian alley cropping system. Agrofor. Syst. 68, 27–36. https://doi.org/10.1007/s10457-005-5963-7

Orwin, R.G., 1983. A Fail-SafeN for Effect Size in Meta-Analysis. J. Educ. Behav. Stat. 8, 157–159. https://doi.org/10.3102/10769986008002157

Pardon, P., Reubens, B., Reheul, D., Mertens, J., De Frenne, P., Coussement, T., Janssens, P., Verheyen, K., 2017. Trees increase soil organic carbon and nutrient availability in temperate agroforestry systems. Agric. Ecosyst. Environ. 247, 98–111. https://doi.org/10.1016/j.agee.2017.06.018

216

Parton, W.J., Schimel, D.S., Cole, C. V., Ojima, D.S., 1987. Analysis of factors controlling soil organic matter levels in great plains grasslands. Soil Sci. Soc. Am. jounal. 51, 1173–1179. https://doi.org/10.2136/sssaj1987.03615995005100050015x

Paudel, E., Dossa, G.G.O., Xu, J., Harrison, R.D., 2015. Litterfall and nutrient return along a disturbance gradient in a tropical montane forest. For. Ecol. Manage. 353, 97–106. https://doi.org/10.1016/j.foreco.2015.05.028

Perfecto, I., Vandermeer, J., 2008. Biodiversity conservation in tropical agroecosystems: A new conservation paradigm. Ann. N. Y. Acad. Sci. 1134, 173–200. https://doi.org/10.1196/annals.1439.011

Perfecto, I., Vandermeer, J., Mas, A., Pinto, L.S., 2005. Biodiversity, yield, and shade coffee certification. Ecol. Econ. 54, 435–446. https://doi.org/10.1016/j.ecolecon.2004.10.009

Pierret, A., Maeght, J.-L., Clément, C., Montoroi, J.-P., Hartmann, C., Gonkhamdee, S., 2016. Understanding deep roots and their functions in ecosystems: an advocacy for more unconventional research. Ann. Bot. 118, 621–635. https://doi.org/10.1093/aob/mcw130

Poeplau, C., Don, A., 2015. Carbon sequestration in agricultural soils via cultivation of cover crops - A meta-analysis. Agric. Ecosyst. Environ. 200, 33–41. https://doi.org/10.1016/j.agee.2014.10.024

Poeplau, C., Don, A., 2013. Sensitivity of soil organic carbon stocks and fractions to different land-use changes across Europe. Geoderma 192, 189–201. https://doi.org/10.1016/j.geoderma.2012.08.003

Post, W.M., Kwon, K.C., 2000. Soil carbon sequestration and land-use change: processes and potential. Glob. Chang. Biol. 6, 317–327. https://doi.org/10.1046/j.1365-2486.2000.00308.x

Powlson, D.S., Gregory, P.J., Whalley, W.R., Quinton, J.N., Hopkins, D.W., Whitmore, A.P., Hirsch, P.R., Goulding, K.W.T., 2011. Soil management in relation to sustainable agriculture and ecosystem services. Food Policy 36, 72–87. https://doi.org/10.1016/j.foodpol.2010.11.025

Powlson, D.S., Stirling, C.M., Thierfelder, C., White, R.P., Jat, M.L., 2016. Does conservation agriculture deliver climate change mitigation through soil carbon sequestration in tropical agro-ecosystems? Agric. Ecosyst. Environ. 220, 164– 174. https://doi.org/10.1016/j.agee.2016.01.005

Qiming, Liu, Shijie, Wang, Hechun, P., Ziyuan, O., 2003. The Dynamics of Organic Matter in Soil Size and Density Fractions Traced by Stable Carbon isotopes.Chin.J.Geochem. 22,10–18.

217

RStudio Team (2017). RStudio: Integrated Development for R. RStudio, Inc., Boston, MA URL http://www.rstudio.com/.

Rasse, D.P., Rumpel, C., 2005. Is soil carbon mostly root carbon ? Mechanisms for specific stabilization.Plant Soil. 269, 341–356. https://doi.org/10.1007/s11104- 004-0907-y

Redondo-Brenes, A., Montagnini, F., 2006. Growth , productivity , aboveground biomass , and carbon sequestration of pure and mixed native tree plantations in the Caribbean lowlands of Costa Rica. For. Ecol. Manag. 232, 168–178. https://doi.org/10.1016/j.foreco.2006.05.067

Richards, M.B., Mendez, V.E., 2014. Interactions between carbon sequestration and shade tree diversity in a smallholder coffee cooperative in El Salvador. Conserv. Biol. 28, 489–497. https://doi.org/10.1111/cobi.12181

Rong, Y., Yong-zhong, S.U., Tao, W., Qin, Y., 2016. Effect of chemical and organic fertilization on soil carbon and nitrogen accumulation in a newly cultivated farmland. J. Integr. Agric. 15, 658–666. https://doi.org/10.1016/S2095- 3119(15)61107-8

Rosenthal, R., 1979. The file drawer problem and tolerance for null results. Psychol. Bull. 86, 638–641. https://doi.org/10.1037/0033-2909.86.3.638

Rumpel, C., Kögel-Knabner, I., 2011. Deep soil organic matter-a key but poorly understood component of terrestrial C cycle. Plant Soil 338, 143–158. https://doi.org/10.1007/s11104-010-0391-5

Saha, S.K., Nair, P.K.R., Nair, V.D., Kumar, B.M., 2010. Carbon storage in relation to soil size-fractions under tropical tree-based land-use systems. Plant Soil 328, 433–446. https://doi.org/10.1007/s11104-009-0123-x

Saha, S.K., Nair, P.K.R., Nair, V.D., Kumar, B.M., 2009. Soil carbon stock in relation to plant diversity of homegardens in Kerala, India. Agrofor. Syst. 76, 53–65. https://doi.org/10.1007/s10457-009-9228-8

Sanchez, P. A., 1999. Improved fallows come of age in the tropics. Agrofor. Syst. 47, 3– 12. https://doi.org/10.1023/A:1006287702265

Sanchez, P.A., 2002. Soil Fertility and Hunger in Africa. Science. 295, 2019–2020. https://doi.org/10.1126/science.1065256

Sanderman, J., Baldock, J.A., 2010. Accounting for soil carbon sequestration in national inventories : a soil scientist ’ s perspective. Environ.Res.Lett. 5, 34–45 https://doi.org/10.1088/1748-9326/5/3/034003

218

Schmidt, M.W.I., Torn, M.S., Abiven, S., Dittmar, T., Guggenberger, G., Janssens, I. A., Kleber, M., Kögel-Knabner, I., Lehmann, J., Manning, D. A. C., Nannipieri, P., Rasse, D.P., Weiner, S., Trumbore, S.E., 2011. Persistence of soil organic matter as an ecosystem property. Nature 478, 49–56. https://doi.org/10.1038/nature10386

Schroth, È., Agra, S., Angelo, D., Geraldes, W., 2002. Conversion of secondary forest into agroforestry and monoculture plantations in Amazonia : consequences for biomass , litter and soil carbon stocks after 7 years. For. Ecol. Mgmt. 163, 131– 150.

Schroth, G., Zech, W., 1995. Above- and below-ground biomass dynamics in a sole cropping and an alley cropping system with Gliricidia sepium in the semi- deciduous rainforest zone of West Africa. Agrofor. Syst. 31, 181–198. https://doi.org/10.1007/BF00711725

Scialabba, N.E., Muller-Lindenlauf, M., 2010. Organic agriculture and climate change. Renw.Agr & Food Sys. 25, 158–169. https://doi.org/10.1017/S1742170510000116

Seddaiu, G., Porcu, G., Ledda, L., Roggero, P.P., Agnelli, A., Corti, G., 2013. Soil organic matter content and composition as influenced by soil management in a semi-arid Mediterranean agro-silvo-pastoral system. Agric. Ecosyst. Environ. 167, 1–11. https://doi.org/10.1016/j.agee.2013.01.002

Sharrow, S.H., Ismail, S., 2004. Carbon and nitrogen storage in agroforests, tree plantations, and pastures in western Oregon, USA. Agrofor. Syst. 60, 123–130. https://doi.org/10.1023/B:AGFO.0000013267.87896.41

Šimanský, V., Bajcan, D., 2014. Stability of soil aggregates and their ability of carbon sequestration. Soil Water Res. 9, 111–118. https://doi.org/10.1080/03650340.2015.1048683

Sinacore, K., Hall, J.S., Potvin, C., Royo, A.A., Ducey, M.J., Ashton, M.S., 2017. Unearthing the hidden world of roots : Root biomass and architecture differ among species within the same guild. PLoS One 1–22.

Six, J., Conant, R.T., Paul, E.A., Paustian, K., 2002. Stabilization mechanisms of soil organic matter: Implications for C-saturation of soils. Plant Soil 241, 155–176. https://doi.org/10.1023/A:1016125726789

Six, J., Elliott, E.T., Paustian, K., 2000. Soil macroaggregate turnover and microaggregate formation: A mechanism for C sequestration under no-tillage agriculture. Soil Biol. Biochem. 32, 2099–2103. https://doi.org/10.1016/S0038- 0717(00)00179-6

219

Six, J., Elliott, E.T., Paustian, K., Doran, J.W., 1998. Aggregation and Soil Organic Matter Accumulation in Cultivated and Native Grassland Soils. Soil Sci. Soc. Am. J. 62, 1367. https://doi.org/10.2136/sssaj1998.03615995006200050032x

Smiley, G.L., Kroschel, J., 2008. Temporal change in carbon stocks of cocoa-gliricidia agroforests in Central Sulawesi, Indonesia. Agrofor. Syst. 73, 219–231. https://doi.org/10.1007/s10457-008-9144-3

Smith, P., Martino, D., Cai, Z., Gwary, D., Janzen, H., Kumar, P., McCarl, B., Ogle, S., O’Mara, F., Rice, C., Scholes, B., Sirotenko, O., Howden, M., McAllister, T., Pan, G., Romanenkov, V., Schneider, U., Towprayoon, S., Wattenbach, M., Smith, J., 2008. Greenhouse gas mitigation in agriculture. Philos. Trans. R. Soc. Lond. B. Biol. Sci. 363, 789–813. https://doi.org/10.1098/rstb.2007.2184

Sollins, P., Swanston, C., Kramer, M., 2007. Stabilization and destabilization of soil organic matter - A new focus. Biogeochemistry 85, 1–7. https://doi.org/10.1007/s10533-007-9099-x

Somarriba, E., Cerda, R., Orozco, L., Cifuentes, M., Dávila, H., Espin, T., Mavisoy, H., Ávila, G., Alvarado, E., Poveda, V., Astorga, C., Say, E., Deheuvels, O., 2013. Carbon stocks and cocoa yields in agroforestry systems of Central America. Agric. Ecosyst. Environ. 173, 46–57. https://doi.org/10.1016/j.agee.2013.04.013

Somarriba, E., Suarez-Islas, A., Calero-Borge, W., Villota, A., Castillo, C., V??lchez, S., Deheuvels, O., Cerda, R., 2014. Cocoa–timber agroforestry systems: Theobroma cacao–Cordia alliodora in Central America. Agrofor. Syst. 88, 1001–1019. https://doi.org/10.1007/s10457-014-9692-7

Sonwa, D.J., Nkongmeneck, B.A., Weise, S.F., Tchatat, M., Adesina, A.A., Janssens, M.J.J., 2007. Diversity of plants in cocoa agroforests in the humid forest zone of Southern Cameroon. Biodivers. Conserv. 16, 2385–2400. https://doi.org/10.1007/s10531-007-9187-1

Soto-Pinto, L., Aguirre-Dávila, C.M., 2014. Carbon stocks in organic coffee systems in Chiapas, Mexico. J. Agric. Sci. 7, 117–128. https://doi.org/10.5539/jas.v7n1p117

Soto-Pinto, L., Anzueto, M., Mendoza, J., Ferrer, G.J., de Jong, B., 2010. Carbon sequestration through agroforestry in indigenous communities of Chiapas, Mexico. Agrofor. Syst. 78, 39–51. https://doi.org/10.1007/s10457-009-9247-5

Soto-Pinto, L., Anzueto, M., Mendoza, J., Ferrer, G.J., de Jong, B., 2009. Carbon sequestration through agroforestry in indigenous communities of Chiapas, Mexico. Agrofor. Syst. 78, 39–51. https://doi.org/10.1007/s10457-009-9247-5

220

Steffan-Dewenter, I., Kessler, M., Barkmann, J., Bos, M.M., Buchori, D., Erasmi, S., Faust, H., Gerold, G., Glenk, K., Gradstein, S.R., Guhardja, E., Harteveld, M., Hertel, D., Höhn, P., Kappas, M., Köhler, S., Leuschner, C., Maertens, M., Marggraf, R., Migge-Kleian, S., Mogea, J., Pitopang, R., Schaefer, M., Schwarze, S., Sporn, S.G., Steingrebe, A., Tjitrosoedirdjo, S.S., Tjitrosoemito, S., Twele, A., Weber, R., Woltmann, L., Zeller, M., Tscharntke, T., 2007. Tradeoffs between income, biodiversity, and ecosystem functioning during tropical rainforest conversion and agroforestry intensification. Proc. Natl. Acad. Sci. U. S. A. 104, 4973–8. https://doi.org/10.1073/pnas.0608409104

Stout, B., Lal, R., Monger, C., 2016. Carbon capture and sequestration : The roles of agriculture and soils. Int J Agric & Biol Eng. 9, 1–8. https://doi.org/10.3965/j.ijabe.20160901.2280

Takimoto, A., Nair, P.K.R., Nair, V.D., 2008. Carbon stock and sequestration potential of traditional and improved agroforestry systems in the West African Sahel. Agric. Ecosyst. Environ. 125, 159–166. https://doi.org/10.1016/j.agee.2007.12.010

Takimoto, A., Nair, V.D., Nair, P.K.R., 2009. Contribution of trees to soil carbon sequestration under agroforestry systems in the West African Sahel. Agrofor. Syst. 76, 11–25. https://doi.org/10.1007/s10457-008-9179-5

Thakur, M.P., Milcu, A., Manning, P., Niklaus, P.A., Roscher, C., Power, S., Reich, P.B., Scheu, S., Tilman, D., Ai, F., Guo, H., Ji, R., Pierce, S., Ramirez, N.G., Richter, A.N., Steinauer, K., Strecker, T., Vogel, A., Eisenhauer, N., 2015. Plant diversity drives soil microbial biomass carbon in grasslands irrespective of global environmental change factors. Glob. Chang. Biol. 21, 4076–4085. https://doi.org/10.1111/gcb.13011

Tisdall, J.M., Oades, J.M., 1982. Organic matter and water-stable aggregates in soils. J. soil Sci. 33, 141–163. https://doi.org/10.1111/j.1365

Tonucci, R.G., Nair, P.K.R., Nair, V.D., Garcia, R., Bernardino, F.S., 2011b. Soil carbon storage in silvopasture and related land-use systems in the brazilian cerrado. J. Environ. Qual. 40, 833–841. https://doi.org/10.2134/jeq2010.0162

Totsche, K.U., Amelung, W., Gerzabek, M.H., Guggenberger, G., Klumpp, E., Knief, C., Lehndorff, E., Mikutta, R., Peth, S., Prechtel, A., Ray, N., Kögel-Knabner, I., 2017. Microaggregates in soils. J. Plant Nutr. Soil Sci. 1–33. https://doi.org/10.1002/jpln.201600451

Tscharntke, T., Clough, Y., Bhagwat, S.A., Buchori, D., Faust, H., Hertel, D., Hölscher, D., Juhrbandt, J., Kessler, M., Perfecto, I., Scherber, C., Schroth, G., Veldkamp, E., Wanger, T.C., 2011. Multifunctional shade-tree management in tropical agroforestry landscapes - a review. J. Appl. Ecol. 48, 619–629. https://doi.org/10.1111/j.1365-2664.2010.01939.x

221

Tumwebaze, S.B., Byakagaba, P., 2016. Soil organic carbon stocks under coffee agroforestry systems and coffee monoculture in Uganda. Agric. Ecosyst. Environ. 216, 188–193. https://doi.org/10.1016/j.agee.2015.09.037

Udawatta, R.P., Kremer, R.J., Adamson, B.W., Anderson, S.H., 2008. Variations in soil aggregate stability and enzyme activities in a temperate agroforestry practice. Appl. Soil Ecol. 39, 153–160. https://doi.org/10.1016/j.apsoil.2007.12.002

UNFCC, 2006. Simplified baseline and monitoring methodologies for selected small- scale afforestation and reforestation project activities under the clean development mechanism II . General guidance III . [WWW Document]. Simpl. baseline Monit. Methodol. Sel. small-scale afforestation Refor. Proj. Act. under clean Dev. Mech. .Bonn,Germany. URL http://cdm.unfccc.int/UserManagement/FileStorage/CDMWF_AM_A3II6AX6KGW 5GBB7M6AI98UD3W59X4 (accessed 4.1.18).

Upson, M.A., Burgess, P.J., 2013. Soil organic carbon and root distribution in a temperate arable agroforestry system. Plant Soil 373, 43–58. https://doi.org/10.1007/s11104-013-1733-x

Upson, M.A., Burgess, P.J., Morison, J.I.L., 2016. Soil carbon changes after establishing woodland and agroforestry trees in a grazed pasture. Geoderma 283, 10–20. https://doi.org/10.1016/j.geoderma.2016.07.002

Valencia, V., García-Barrios, L., West, P., Sterling, E.J., Naeem, S., 2014. The role of coffee agroforestry in the conservation of tree diversity and community composition of native forests in a Biosphere Reserve. Agric. Ecosyst. Environ. 189, 154–163. https://doi.org/10.1016/j.agee.2014.03.024

Van der Werf, GR, Morton, D C, DeFries, R S, Olivier, JGJ, Kasibhatla, P S, Jackson, R B, Collatz, G J, Randerson, J.T., 2009. CO2 emissions from forest loss. Nature, 9–11.

Vargas, M., Glaz, B., Alvarado, G., Pietragalla, J., Morgounov, A., Zelenskiy, Y., Crossa, J., 2015. Analysis and interpretation of interactions in agricultural research. Agron. J. 107, 748–762. https://doi.org/10.2134/agronj13.0405

Velmourougane, K., 2016. Impact of Organic and Conventional Systems of Coffee Farming on Soil Properties and Culturable Microbial Diversity. Scientifica (Cairo). 2016, 1–9. https://doi.org/10.1155/2016/3604026

Viechtbauer, W., 2010. Conducting Meta-Analyses in R with the metafor Package. J. Stat. Softw. 36, 1–48. https://doi.org/10.1103/PhysRevB.91.121108

Weil, R., C Brady, N., 2016. The Nature and Properties of Soils. 15th edition.

Wickham, H., 2017. ggplot2 - Elegant Graphics for Data Analysis (2nd Edition). J. Stat. Softw. 77, 3–5. https://doi.org/10.18637/jss.v077.b02

222

Wickham, H., 2007. Reshaping Data with the reshape Package. J. Stat. Softw. 21, 1– 25. https://doi.org/10.18637/jss.v021.i12

Wotherspoon, A., Thevathasan, N. V., Gordon, A.M., Voroney, R.P., 2014. Carbon sequestration potential of five tree species in a 25-year-old temperate tree-based intercropping system in southern Ontario, Canada. Agrofor. Syst. 88, 631–643. https://doi.org/10.1007/s10457-014-9719-0

Wright, D.G., Mullen, R.W., Thomason, W.E., Raun, W.R., 2001. Estimated land area increase of agricultural ecosystems to sequester excess atmospheric carbon dioxide. Commun. Soil Sci. Plant Anal. 32, 1803–1812. https://doi.org/10.1081/CSS-120000251

Yelenik, S.G., Stock, W.D., Richardson, D.M., 2004. Ecosystem level impacts of invasive Acacia saligna in the South African fynbos. Restor. Ecol. 12, 44–51. https://doi.org/10.1111/j.1061-2971.2004.00289.x

Zomer, R.J., Neufeldt, H., Xu, J., Ahrends, A., Bossio, D., Trabucco, A., van Noordwijk, M., Wang, M., Bondeau, A., Feddema, J., Foley, J.A., Johnson, J.A., Runge, C.F., Senauer, B., Foley, J., Polasky, S., Alexandratos, N., Bruinsma, J., Elbehri, A., Albrecht, A., Kandji, S.T., Nair, P.K.R., Nair, V.D., Kumar, B.M., Showalter, J.M., Nair, P.K.R., Foley, J.A., Houghton, R.A., Houghton, R.A., Hall, F., Goetz, S.J., Pan, Y., Lal, R., Schramski, J.R., Dell, A.I., Grady, J.M., Sibly, R.M., Brown, J.H., Mbow, C., DiMiceli, C.M., Bartholomé, E., Belward, A.S., Dewi, S., Noordwijk, M. van, Ekadinata, A., Pfund, J.L., Liu, Y.Y., Lusiana, B., Hansen, M.C., DeFries, R.S., Townshend, J., Lal, R., Leguizamón, A., Lapegna, P., Blinn, C.E., Browder, J.O., Pedlowski, M.A., Wynne, R.H., Pompeu, G.D.S.S., Rosa, L.S., Santos, M.M., Modesto, R.S., Vieira, T.A., Peng, S.-S., Clair, S., Lynch, J.P., Li, H.L., Dakora, F.D., Keya, S.O., Mortimer, P.E., 2016. Global Tree Cover and Biomass Carbon on Agricultural Land: The contribution of agroforestry to global and national carbon budgets. Sci. Rep. 6, 29987. https://doi.org/10.1038/srep29987

223

BIOGRAPHICAL SKETCH

Nilovna Chatterjee was born in Kolkata, raised in four cities: Kolkata, Delhi, Kochi and Chennai, India. Having grown up in different states, Nilovna speaks four Indian

Languages fluently. She spent her formative years in Kochi, Kerala and considers

Kerala her home as the state had a big influence in shaping her personality, bestowed her with friends who became family and was a perfect fit culturally. Nilovna holds a

Bachelor of Engineering in bioengineering and Master of Technology in environmental engineering from Vellore Institute of Technology, Vellore, India. On completing her master’s degree, Nilovna worked at Cognizant Technologies Solution, Chennai, India as a Programmer Analyst but soon realized that she wanted to pursue a more meaningful career. Nilovna’s interest in agroforestry and climate change mitigation strategies developed during her time spent in East Africa where her family had briefly relocated to in the year 2012. In August 2014, she began doctoral study at the University of Florida,

School of Forest Resources and Conservation, with Dr. P.K. Ramachandran Nair as her major advisor. Nilovna graduated from the University of Florida with a Ph.D. in forest resources and conservation, a minor in soil and water sciences, and a concentration in agroforestry in August 2018.

224