xml version 1.0 encoding UTF-8
REPORT xmlns http:www.fcla.edudlsmddaitss xmlns:xsi http:www.w3.org2001XMLSchema-instance xsi:schemaLocation http:www.fcla.edudlsmddaitssdaitssReport.xsd
INGEST IEID E5D96ROT0_KZ8Z4X INGEST_TIME 2014-10-03T21:44:52Z PACKAGE UFE0046763_00001
AGREEMENT_INFO ACCOUNT UF PROJECT UFDC
1 INTERRELATIONSHIPS BETWEEN FI RE, SOIL ORGANIC MATTER, AND NU TRIENT BIOAVAILABILITY IN AMAZONIAN SMALLHOLDINGS By DAMION J. GRAVES A THESIS PRESENTED TO THE GRADUATE SCHOOL OF THE UNIVERSITY OF FLORIDA IN PARTIAL FULFILLM ENT OF THE REQUIREMENTS FOR TH E DEGREE OF MASTER OF SCIENCE UNIVERSITY OF FLORIDA 2014
2 2014 Damion J. Graves
3 To all my friends and family, both present and departed
4 ACKNOWLEDGMENTS First, o thank my advisor, Dr. Michelle Mack for making this work possible and guiding me to its completion Dr. Jack Putz and Dr. And rew Zimmerman Dr. Putz for his intellectual and moral support, and Dr. Zimmerman f or his generous material, intellectual, and technical support in completing the black carbon analysis. Dr. Nick Comerford and the late Dr. Hugh Popenoe provided invaluable inspiration and intellectual support, and both also served on an earlier version of my committee. Numerous other faculty, students, and laboratory staff at the University of Florida Soil and Water Science Department provided technical advice and encouragement. My colleagues in the Ethnoecology Society and the Plant and Ecosystem Ecology R esearch Group provided ongoing advice, encouragement, criticism and material support In Loreto, I am indebted to Dr. Jim Penn for providing excellent advice on fieldwork, as well as generally paving the way for my work, and Yully Rojas Rategui for her he lp in planning and conducting the field work. Dr. Mike Gilmore, Dr. Lorgio Verdi, and various other faculty, staff, and students at Universidad Nacional de Amazonia Peruana also assisted in arranging field work and preparing samples for transport. T he Work ing Forests in The Tropics program provided funding for a preliminary visit to the region in 2006. Finally, I wish to thank the people from the communities of Padre Cocha, San Martin Tipishka, El Chino, San Pedro, Diamante, 18 Febrero, and Tamshiyacu for their hospitality and generosity with their time. Back in the Ecosystem Ecology L ab at the University of Florida Department of Biology, I received help from many people in processing my samples and completing my laboratory analyses. Julia Reiskind provided tireless professional support and guidance in the completion of the phosphorus analysis and other laboratory procedures.
5 Grace Crummer provided essential assistance in developing protocols for the black carbon analysis and conducted all of the carbon and nitrogen measurements on the elemental analyzer. Megan Lipscey and Corey Hanlon volunteered many hours of their time and assisted me with grinding soil samples and conducting the phosphorus and black carbon analyses. Finally, Dr. Camila Pizano helped me d evelop and run the statistical analyses.
6 TABLE OF CONTENTS page ACKNOWLEDGMENTS ................................ ................................ ................................ .. 4 LIST OF TABLES ................................ ................................ ................................ ............ 8 LIST OF FIGURES ................................ ................................ ................................ .......... 9 LIST OF ABBREVIATIONS ................................ ................................ ........................... 11 ABSTRACT ................................ ................................ ................................ ................... 12 CHAPTER 1 INTRODUCTION ................................ ................................ ................................ .... 14 2 METHODS ................................ ................................ ................................ .............. 17 Study Sites ................................ ................................ ................................ .............. 17 Field Data and Sample Collection ................................ ................................ ........... 20 Laboratory Analysis ................................ ................................ ................................ 22 Statistical Analysis ................................ ................................ ................................ .. 24 3 RESULTS ................................ ................................ ................................ ............... 27 Patch level Data from Field Measurements ................................ ............................ 27 Soil Chemistry ................................ ................................ ................................ ......... 30 Interrelationships Between Variables ................................ ................................ ...... 33 Effects Across All Sites ................................ ................................ ..................... 33 Multivariate Analyses by Landform ................................ ................................ ... 34 Upland patches ................................ ................................ .......................... 34 Lowland patches ................................ ................................ ........................ 38 4 DISCUSSION ................................ ................................ ................................ ......... 55 APPENDIX A DATA FROM CHARCOAL F URNA CES ................................ ................................ 63 B DATA FROM ORILLA SIT E ................................ ................................ .................... 65 C EXCHANGEABLE CATION DATA ................................ ................................ ......... 67 D PATCH LEVEL FIELD DATA ................................ ................................ .................. 68 E CORRELATION MATRICES AND INTER DEPTH CORRELATIONS .................... 72
7 F FULL SET OF EIGENVAL UES AND COMPONENT LO ADI NGS FOR PRINCIPAL COMPONENT ANALYSES ................................ ................................ 74 G KEYS TO SYMBOLS USED IN SCATTERPLOTS GROU PED BY FIELD ............. 78 H PRINCIPAL COMPONENT SCATTER PLOTS AND REGRESSION SCATTERPLOTS ................................ ................................ ................................ ... 79 I SONDEO QUESTIONNAIRE ................................ ................................ ................ 103 J INFORMATION FROM FARMER INTERVIEWS ................................ .................. 105 LIST OF REFERENCES ................................ ................................ ............................. 111 BIOGRAPHICAL SKETCH ................................ ................................ .......................... 113
8 LIST OF TABLES Table page 2 1 Basic data for communities included in the study ................................ ............... 25 2 2 Basic data for fields sampled in the study ................................ .......................... 26 2 3 List of transformations used for statistical analyses ................................ ............ 26 3 1 Descriptive statistics for upland patches* ................................ ........................... 45 3 2 Desc riptive statistics for lowland patches* ................................ .......................... 45 3 3 Summary of ANCOVA on 0 5 cm depth ................................ ............................. 51 3 4 Summary of ANOVA on 5 10 cm depth ................................ .............................. 51 3 5 Variate regressions on upland patches ................................ .............................. 53 3 6 Variate regressions on lowland patches ................................ ............................. 54 A 1 Mean analytical C, %N, and P sol measurements of material from a pit furnace .. 63 B 1 Field data for orilla site ................................ ................................ ....................... 65 B 2 Laboratory data for orilla site ................................ ................................ .............. 66 D 1 Field data for upland patches ................................ ................................ ............. 68 D 2 Field data for lowland patches ................................ ................................ ............ 70 E 1 Correlation matrices and inter depth correlations for upland patches* ............... 72 E 2 Correlation matrices and inter depth correlation s for lowland patches* .............. 73 F 1 Eigenvalues and component loadings for upland patches, 0 5 cm depth ........... 74 F 2 Eigenvalues and compone nt loadings for upland patches, 5 10 cm depth ......... 75 F 3 Eigenvalues and component loadings for lowland patches, 0 5 cm depth ......... 76 F 4 Eigenvalues and component loadings for lowland patches, 5 10 cm depth ....... 77 G 1 Community abbreviation codes ................................ ................................ .......... 78
9 LIST OF FIGURES Figure page 3 1 Months since last burn by field type ................................ ................................ .... 43 3 2 Patch percent slope by field type ................................ ................................ ........ 44 3 3 pHw in 0 5 cm soil (x) vs 5 10 cm soil (y) by patch ................................ ............. 46 3 4 Soluble P in 0 5 cm soil (x) vs 5 10 cm soil (y) by patch ................................ .... 46 3 5 Total C (mg/g) in 0 5 cm soil (x) vs 5 10 cm soil (y) by patch ............................. 47 3 6 % N in 0 5 cm soil (x) vs 5 10 cm soil (y) by patch ................................ ............. 47 3 7 Bulk density (g/cm3) in 0 5 c m soil (x) vs 5 10 cm soil (y) by patch ................... 48 3 8 Clay sized particle mass percent by patch ................................ ......................... 48 3 9 ECEC (cmol/kg) by patch ................................ ................................ ................... 49 3 10 Total C, acid Insoluble C, and BC content of soils by patch ............................... 50 3 11 PCA summary plots ................................ ................................ ............................ 52 C 1 Frequency histograms of exchangeable cation measurements by patch ........... 67 G 1 Key to upland fields ................................ ................................ ............................ 78 G 2 Key to lo wland fields ................................ ................................ ........................... 78 H 1 Principal component groupings by field* ................................ ............................. 79 H 2 Scatterplots with grouping by field for PC #1, upland 0 5 cm depth ................... 80 H 3 Scatterplots with grouping by field for PC #2, upland 0 5 cm depth ................... 81 H 4 Scatterplots with grouping by field for PC #3, upland 0 5 cm depth ................... 82 H 5 Scatterplots with grouping by field for PC #4, upland 0 5 cm depth ................... 83 H 6 Scatterplot s with grouping by field for PC # 1, upland 5 10 cm depth ............... 84 H 7 Scatterplots with grouping by field for PC #2, u pland 5 10 cm depth ................. 85 H 8 Scatterplots with grouping by field for PC #3, upland 5 10 cm depth ................. 85 H 9 Regression and scatter plots for upland patches, 0 to 5 cm depth ..................... 86
10 H 10 Regression and scatter plots for upland patches, 5 to 10 cm depth ................... 89 H 11 Scatterplots with grouping by field for PC #1, lowland 0 5 cm depth .................. 91 H 12 Scatterplots with grouping by field for PC #2, lowland 0 5 cm depth .................. 93 H 13 Scatterplots with grouping by field for PC #3, lowland 0 5 cm depth .................. 93 H 14 Scatte rplots with grouping by field for PC #1, lowland 5 10 cm depth ................ 94 H 15 Scatterplots with grouping by field for PC #2, low land 5 10 cm .......................... 95 H 16 Regression and scatter plots for lowland patches, 0 to 5 cm depth .................... 96 H 17 Regression and scatter plots for lowland patches, 5 to 10 cm depth ................ 102
11 LIST OF ABBREVIATIONS BC Analytical black carbon pool created by applying the final combustion procedure of the Kuhlbusch method to acid insoluble C samples. Measured on an elemental analyzer. C ACID INS Analytical r ecalcitrant carbon pool created by subjecting samples to the first portion of the Kuhlbusch method, consisting of repeated alternate washing with strong acids and bases. Measured on an elemental analyzer. % C LAY Cla y sized particle mass percent as determin ed by the simplified hydrometer method of particle size analysis (Gee and Bauder 1986). dbh Diameter at breast height: Diameter measurement taken on a plantain stem at approximately 1.4 m. ECEC Effective Cation Exchange Capacity, calculated by the summatio n of cations method. Lowland Fields located on periodically inundated river terraces. PCA Principal Component Analysis, as conducted on the set of transformed variates. pH w pH of sample prepared in water solution, as measured by microelectrode. P SOL Water soluble phosphorus concentration, as determined by Mehlich 1 extraction. SOM Soil organic matter. TC Total carbon content, as measured on an elemental analyzer. Upland Fields located on well drained interfluves.
12 Abstract of Thesis Presented to the Graduate School of the University of Florida in Partial Fulfillment of the Requirements for th e Degree of Master of Science FIRE, SOIL ORGANIC MATTER, AND NUTRIENT BIOAVAILABILITY IN AMAZONIAN SMALLHOLDINGS By Damion J. Graves May 2014 Chair: Michelle Mack Major: Botany Fire residues including black carbon compounds and unburned organic matter, are an under examined component of many ecosystems. In highly weathered humid tropical soi ls of the humid tropics, charcoal and related residua l organic matter that remain after b urning may provide a stable reservo ir of nutrients and increase soil nutrien t retention capacity This study examined the impact of burning and other land management practices on soil organic matter and soil chemistry ac ross a range of recently burned agricultural fields in the Peruvian Amazon I recorded management histories, observed farming practices, and collected and analyzed soil samples from both upland and floodplain s ites. Upland sites were characterized by longe r and more variable burn intervals, and more active management of charcoal as a soil amendment. Floodplain sites tended to be manage d with annual or bi annual burn s, but less active manipulation of charcoal. To determine the nature and quantity of organic residues, soil samples were analyzed for total soil carbon (TC) and black carbon (BC) content. These data were then compared with indicators of soil nutrient availability, which included extractable phosphorus (Psol), cation exchange capacity (ECEC), and pH. TC in the surface 5cm of
13 upland fields ranged 1.6 8.2%, while BC ranged 0.4 17.5 mg/g. Psol concentrations in the surface 5 cm of upland fields ranged 0.6 31.1 mg/kg, and declined rapidly with depth. TC in the surface 5 cm of floodplain sites ranged 0. 7 5.6%, while BC ranged 1.2 12.8 mg/g. Psol in the surface 5 cm of floodplain fields ranged 1.5 90.4 mg/kg, and declined gradually with depth. Mult ivariate statistical analyses o f upland site soil data show tight links among TC, BD, soluble P, and to tal N Similar analyses on floodplain data indicated that clay particle size content drove much of the variation with tight linkages among other variables. The se results su ggest that the burns le ft enough charcoal and unburned material to form a subs tantial pool of plant available nutrients, aid in the retention of nutrients via expanded exchange capacity, and increase phosphorus bioavailab ility by raising the pH In the case of upland fields, these residues may be the prime source of soil nutrients
14 CHAPTER 1 I NTRODUCTION Regardless of the cause or purpose of a fire, its most dramatic impact on the land scape is usually the conversion of aboveground biomass to CO 2 Nutrients that are not volatilized by combustion are often quickly removed from the soil as ash is washed or blown from the surface. Thus, it is appropriate that most studies of nutrient flux after fires in slash and burn agricultural systems emphasize d the loss of nutrients from the system via volatili zation, erosion, and leaching. The broad consensus that emerges from the m any studies on the effect of cutting and burni ng on nutrient dynamics is that burning releases phosphorus, nitrogen, and other nutrients to the soil in a large pulse, followed by a long decline in plant available nutrient levels (e.g ., Nye and Greenland 1960, Palm et al 1996, Juo and Manu 1996). Nevertheless, com plete combustion of the organic matter in a vegetation fir e is virtually impossible due to lack of oxygen in the interior of larger fuel particles coupled with patches of fue l with high fuel moisture content This heterogeneity in fuel properties as well as fuel quantities typically results in highly heterogeneous burn dynamics. Thus, in addition to ash, charcoal and unburned necromass are also commonly left af ter a fire ofte n associated with the same particles of organic matter. Charcoal, or more pro perly black carbon (BC), is produced when organic material is pyrolized under oxygen limiting conditions (Glaser 2002). BC may persist for centuries or more and aid in the retenti on of soil nutrients (G laser 2002). This nutrient retent ion is posited to be the result of exchange sites on the carbon skeletons of some types of BC that provide both nutrient sorption and pH buffering capacity. For these reasons, the production and use o f charcoal as a soil amendment has been proposed
15 for tropical regions with highly weath ered soils, such as the Amazon B asin (Glaser 2007). Despite this interest in BC, most experimental studies on fire ecology in this region utilized complete burns in thei r plots, burning repeatedly if necessary to reduce all available plant material to ash (e.g., Jordan 1989, Zarin 1998). Furthermore, apparently no fie ld studies of slash and burn agricultural systems have directly quantified BC produced over a range of tra ditional farming settings or investigated the impact of charcoal and slightly charred or ganic residues on soil chemistry. T his study examined the r elationship between BC soil organic matter (SOM) and nutrients across a range of recently burned fields in the lowland Peruvian Amazon. Farmers in this region employ a variety of slash and burn techniques, which usually involve clearing and burning a patch of secondary forest, planting field crops for several years, and then allowing the patch to regenerate for several decades ( Denevan and Padoc h 1987 Coomes et al 2000). I n some areas, the fallow period includes t he development of orchards that may mimic natur al forest structure (Denevan and Padoch 1987 ). As few nutrients are typically available in tropical min eral soil s and chemical fertilizers are prohibitively expensive for most farmers (Fearnside 1987), management of SOM is an impo rtant component of all of these systems (Szott et al 1999, Tiessen et al 1994). This study involved interviewing farmers, colle cting field data, and analyzing soil samples from five different communities in the P eruvian D epartment of Loreto. It sought to address the following research questions: 1. Which cropping patterns, burn regimes, and landforms are associated with the highest q uantities of charcoal and soil organic matter? 2. Where and how does the presence of charcoal influence bulk soil chemistry and nutrient concentrations?
16 3. Does the inf luence of charcoal differ from that of any other soil organic matter? Addressing these questio ns will help to bridge the knowledge gap between experimental studies of charcoal and biochars as a soil amendment and the current role of charcoal in low input tropical farming systems. By determining which factors influence the yield of BC and othe r fire residues that affect bulk soil chemistry, this study may inform future efforts to employ BC to increase the productivity of these systems.
17 CHAPTER 2 METHODS Study Si tes I n July and August of 2007, interviews were conducted and soil samples were colle cted from 19 cultivated fields in and near five different agrarian communities 50 100 km from the C None were connected by road but all were accessible by boat. Mean annual temperature for this region is 26 C, with a mean annual precipitation 308 7 mm. The months of May August e ach receive approximately 200 mm of prec ipitation which is low for this region (Kalliola et al, 1993) and this is the time of lowest water levels in rivers, s treams, lakes, and water tables The communities ranged in p opulation from 3000 to around 50, and ranged in time since establishment fro m 17 to over 100 years. E ach community was center ed on a river bluff, with cultivated fields dispersed across wide area s on a variety of landforms In smaller, more recently established communities, fields were < 5 km from farmer home s, whereas in the lar gest and oldest community, Tamshiyacu, fields were up to 15 km away. Nine fields were on well drained uplands that had not experienced flooding in recent memory, bro fields were situated in periodically in eld was on ferralsols and acrisols, while the lowland fields were on gleyic cambisols and a lisols ( FAO 2007) Most of the communities had access to all of the above landforms, which allow ed exploitation of a v ariety of habitats ( Tables 2 1 and 2 2 ) The residents of the five communities studied were of mixed European and Indigenous Amazonian an cestry, and most spoke Spanish as their first language. The
18 largest and oldest of t he communities, Tamshiyacu, was migrants from the nei ghboring department of San Mart n, and was an important regional population center during the r ubber boom of the early 20th C entury (Hern ndez 1946). The smallest community 18 F ebrero, consisted primarily of families who had moved several kilometers upriver from Tamshiyacu after the construction of a concrete track through the area in the early 199 The other three communities a re located in the T ahuayo R iver watershed. El Chino i s a small trade and transportation center w ith residents who maintain commercial and family ties with villages in a more remote hinterland. San Pedro and Diamante were m ore recently established villages consisting primar ily of familie s from El Chino, who had moved both to avoid the flooding problems frequently encountered there and to gain access to more land. Many families indicated that they had moved to their current c ommunity from elsewhere in the local area within the past ten years. Movement to and from Tamshiyacu was especially common among the farmers interviewed (Personal Communication, I nterviews with F armers, July and August 2007) The cycle of clearing and burn ing typically followed by farmers in this region, described by several previous authors (e.g., Denevan and Padoch 1987, Zarin et al 1998, Coomes et al 2000 ), usually involves relatively short crop cycles, with burns taking place every 1 to 5 years. In upla nd interfluves, burn cycles tend to be at the longer end of this spectrum and many fields are converted to orchards or abandoned after a few years of cultivation. In annually flooded areas, farmers tend to follow a more tightly constrained pattern of annu al burning, and continuous croppin g is sometimes practiced indefinitely.
19 Farmers were recruited for the study with a snowball sampling method (Goodman 1961). All of the farmers approached were wil ling to participate The majority of fields held by any g iven farmer had generally not been cleared or burned for some time, and had either been abandoned o r converted to orchards. I n t he smaller communities, samples and data were gathered from virtually every eligible household, but most fields were outside the scope of the study and thus omitted In Tamshiyacu, th e largest community, only a small fraction of all eligible fields were visited. The s tudy began with three selection criteria for fields; they had to have been burned within the last 20 years, they ha d to contain manioc, plantain, or both, and they had to have been under at least intermittent cultivation for at least 50 years. The presence of manioc or plantain was used to limit the study to fields under current management for field crop production In fact, all of the fields encountered where manioc or plantain were present had been burned within four years. In addition to this, it was generally impossible to ascertain field histories past 20 30 years, and there were not enough old fields under current cul tivation for sampling purposes As a result, the third criterion was eliminated from consideration. At each field, interviews were conducted based on a modification of the Sondeo (Hildebrand 1986) a pproach. Questionnaires sought to elicit explanations of a farmer's overall management plan, and the factors and decisions that influenced that plan. S pecific information was also gathered on management techniques, including weeding, crop spacing, and harvest. F armers were explicitly asked abou t the history o f burning, cropping, and clearing of each field Through discussion of visual indica tors in each field, qualitative information was elicited from each farmer about the characteristics of
20 the last burn, soils, and local hydrology, as well as management unit s within each field. This information was used to inform sampling design F ield sizes ranged 0.02 3.3 ha, with an average field size of approximately 0.5 ha. Time since last burn varied between 2 weeks and 3 years, and time since initial clearing v aried fr om < 1 year to > 40 years. The older fields tended to be in transition to orchard crops, while farmers cultivating fields that they had not recently burned (i.e., within the past two years) often reported that they soon planned to abandon them. All of the fa rmers interviewed engaged in both market oriented and subsist ence activities and all grew manioc or plantain, most often both. Plantain was present in all of the lowland fields, and absent in only two of the upland fields. Manioc was absent from two upland fields and one lowland field, and present in all others. Pineapple was also cultivated in specific well drained locations on six of the upland sites, almost always interc ropped with manioc and plantain, while papaya was planted extensively on six of the l owland sites. Other field crops in cultivation in some areas included maize and sugar cane. The most common orchard crops in cultivation were brazil nut ( Bertholletia excelsa ), umar ( Poraqueiba paraensis ), aguaje ( Mauritia flexuosa ), and caimito ( Chrysoph yllum cainito ). Field Data a nd Sample Collection Within each field, 2 5 patches were est ablished for sampling purposes. Patches were delineated on the basis of information from farmers about cropping history and differences in soil characteristics, as well as field indicators that suggested potential differences in burn dynamics and so il chemistry. Within patch variation in slope, aspect, crop complexes planted, size of crop plants, and the completeness of the last burn were all minimized As patch delineat ions were based on a variety of edaphic conditions and
21 management regimes, estimated patch sizes v aried widely, from 50 to 1100 m 2 While all patches within a field shared a common history of clearing and burning, this commonality is likely overwhelmed by the many differences between patches as a result of landscape posit ion and fire dynamics; thus, the patch was deemed to be the sampling unit. Within each patch, the major crop plants present were recorded along with other cultivated plants. For the most p revalent crops, i.e., manioc, plant ain, and pineapple, the average planting distance s w ere recorded. T he height range of manioc stems and the diameter (dbh) range of plantain stems within each patch were also recorded. The slope and aspect of each patch we re measured and recorded, as was percent ground cover. Transects were established within each patch, with soil cores taken randomly along each transect. Three cores were taken in each patch with a 4 cm diamet er corer, at depths of 0 5 cm and 5 10 cm. Th e c ores from each depth from a patch were combined. The type and depth of organic matter above the soil surface varied widely. Where a layer of charcoal substantial enough to sample was present, it was collected for further analysis. Man y farmers harvested f elled wood for charcoal production in simple soil furnaces constructed on site. Only the largest chunks of charcoal were packaged for transport and sale, while the finer charcoal particles, mixed with ash and soil, were left on site, and often used as soil amendments. S amples of this material were collected from each site w ith c harcoal furnaces
22 Laboratory Analysis All samples were weighed, air dried, and transported to the University o f Florida for laboratory analysis. Extractable phosphorus (P sol ) was de termined using the Mehlich 2 double acid extraction method (Mehlich 1978) with extractant P concentrations then read colorimetrically on a microplate reader, using the ascorbate method (Murphy and Riley 1962). Bulk density (BD) was determined by oven dr ying a sub sample of each set of combined cores at 110 o C back calculating the oven dried equivalent mass, and dividing by total core volume. Water solution pH was measured with a micro electrode in a 1:1 ratio of deionized water:air dried soil A rapid as sessment of clay siz ed particle w percent (hereafter referred to as % clay or clay content) was conducted on samples from the surface 5 cm using the simplified hydrometer method of particle size analysis (Gee and Bauder 1979). Samples from the uppermost 5 cm were extracted using the unbuffered salt extraction method (Grove et al 1982), and the extractant was analyzed for exchangeable potassium, calcium, magnesium, aluminum, and sodium. Effective cation exchange capacity (ECEC) was calculated by summation of cations. Total C and N content s were determined on an elemental autoanalyzer after drying at 60 o C and grinding to a fine powder with a mortar and pestle. Samples from the surface 5 c m were analyzed for BC content using the Kuhlbusch (1995) method. This i nvolved a series of chemical treatments with NaOH, HCl, and HNO 3 followed by combustion at 340 in a pure stream of O 2 For the chemical treatments, s amples were first treated with two consecutive washes of 1 M NaOH, then a wash of 70% HNO 3 then
23 5 more c onsecutive washes of 1 M NaOH, followed by a wash with 1% HCl, then two consecutive washes with deionized water. After the addition of treatment solution at each step, samples were placed in a sonic bath for 20 minutes to ensure complete reaction, then cen trifuged to separate the sample from the supernatant, with the supernatant discarded after each treatment. For the thermal treatment, oven dried samples which had already been chemically treated were placed in a pure stream of O 2 and kept at a temperature of 340 for 1 hour. Samples were oven dried and re weighed after each type of treatment, an d C content determined with an elemental analyzer. The following calculations were used to determine two different analytical C pools: Acid insoluble C ( mgC/g soil ) = (%C from EA) (post chemical treatment sample mass/pre chemical treatment sample mass)(10) B C ( mgC/g soil ) = (%C from EA) (post chemical treatment sample mass/pre chemical treatment sample mass)(post thermal treatment sample mass/pre thermal treatment sam ple mass)(10) A homogenized analytical standard was prepared using a mixture of upland mineral soils collected from a depth of 20 30 cm and a mixture of charred material collected from a local charcoal furnace. Specifically, 117.7 g of material from a cha rco al furnace was mixed with 352.5 g of mineral soil from a local field, creating a mixture of approximately 25% char and 75% mineral soil by weight. The average measurement of a cid insoluble C (C acid insol ) for this standard was 9.77 mg/g (SE= 0.22 % ) whi le the ave rage measurement of BC was 7.42 mg/g (S.E.= 0.2 % ) Randomly selected analytical replicates of unknown samples yielded coefficients o f variation ranging 1.6
24 10.8 % for C acid insol measurements, and 2.3 17.4 % for BC measurements. Analysis of a c hernozem soil, a stand ard used in the 2007 ring trials of seven different BC determination methods (Hammes et al 2007 ), yielded measurements of 12.5 mg/g C acid insol and 10.29 mg /g BC which were closest to the mean measurements yielded by the acid dichrom ate oxidation method used in the ring trials, and below the mean measurements for five of the methods tested. Missing and erroneous measurements from individual analyse s were replaced with depth wide, landform wide averages for each analysis. Statistical Analysis T o address the r esearch questions, several different statistical analyses were con ducted In all cases, separate analyses were conducted on the 0 5 cm and 5 10 cm depth due to large differences in their nutrient concentrations To examine the in fluence of charcoal and burn regime on bulk soil chemistry and nutrient concentrations (Questio n #2 ) a nested analysis of covariance ( ANCOVA ) was run o n the 0 5 cm depth data T ime since fire was divided into bins (recent 0.5 5 months interm ediate 6 10 m onths, and old >10 months ), and included as a fixed factor. To account for site facto rs, fire treatment was nested within a double nested random factor of comm unity nested within landform. B C (mg/g soil) and % clay were included as covariates An a nalysis of variance ( ANOVA ) was conducted on data from the 5 10 cm depth, using the same scheme of factors from the 0 5 cm depth. As there were no BC or % clay measurements from this depth, the goal of this analysis was simply to determine whether time since fire had a ny influence on soil chemistry.
25 T o compare the influence of charcoal to that of other SOM (question 2), principal component analyse s and subsequent regressions were conducted on data from 0 5 cm using all of the variables listed in Table 2 3 PCA was used both to id entify closely associated variabl es and to reduce the dataset for subsequent regression analyses. S eparate analyses were conducted on upland and lowland fields. PCA s and regressions were also conducted on data from the 5 10 cm depth; these were necessarily constrained to examining only the relationship between time since last burn, TC pH, and nutrients. Principal components with Chi squared values significant at th e .05 level were retained for fur ther investigation. Within each component, v ariabl es with component loading s were selected for further investigation. The relation ships within each set of variable s were examined with a series of regression analyses; a priori assumptions about soil dynamics were used to determine dependent an d independent variables. Due to the larg e number of comparisons, the cutoff for s tatistical significance was set at 0.05. Several transformations were required to ensure normality and colinearity in the dataset. ( Table 2 3 ) Table 2 1. Basic data for commu nities i ncluded in t he s tudy Community Community Age (Y ears) Mean Field S ize (H a) Standard D eviation Mean Field Age (Y ears) Standard Deviation Altura Sites S ampled Restinga Sites S ampled Orilla Sites S ampled San Pedro 19 0.41 0.30 11.5 12.3 6 0 0 Diamant e 17 0.96 0 7 0 1 0 0 El Chino 45 0.56 0.23 20 19.80 1 0 1 Tamshiyacu >100 0.54 0.80 11.00 11.92 3 12 0 18 Febrero Unknown 0.6 0 8 0 1 0 0
26 Table 2 2. Basic data for fields s ampled in t he s tudy Community Field Size (H a) Land F orm Plantain P resent Mani oc P resent Pineapple P resent Time Since Last Burn (M onths) Length of Last Fallow Period (Y ears) Month of Last B urn 18 Febrero 0.6 Altura X X 2 >30 November Diamante 0.96 Altura X X X 9 >30 December El Chino 0.4 Altura X X X 9 20 January El Chino 0.72 Orilla X X 36 >30 September San Pedro 0.17 Altura X X 5 8 September San Pedro 0.25 Altura X 7 20 April San Pedro 0.35 Altura X X X 6 >30 July San Pedro 0.64 Altura X 34 9 September Tamshiyacu 0.02 Restinga X X 38 >10 July Tamshiyacu 0.13 Rest inga X X 8 5 April Tamshiyacu 0.14 Restinga X X 30 8 April Tamshiyacu 0.2 Restinga X X 2 4 July Tamshiyacu 0.24 Altura X 23 4 July Tamshiyacu 0.25 Altura X X X 3 6 May Tamshiyacu 0.25 Restinga X X 12 >30 August Tamshiyacu 0.25 Restinga X X 8 > 30 September Tamshiyacu 0.28 Restinga X X 3 >10 August Tamshiyacu 0.5 Restinga X X 22 5 November Tamshiyacu 3.3 Restinga X X 0.5 Unknown May Table 2 3. List of transformations u sed f or statistical a nalyse s Variate Unit of Expression Transformation Soluble P hosphorus (P sol ) Mg/kg S oil Column R ank Total Carbon C ontent (TC) % Carbon by Dry W eight Arcsine of Square Root of Proportion of C Total Nitrogen C ontent (TN) % Nitrogen by Dry W eight Arcsine of Square Root of Proportion of N pH in Water S olut ion (pH water ) Antilog of pH Column R ank Bulk Density (BD) Grams of Soil per Cubic C entimeter None Effective Cation Exchange Capacity (ECEC) Centimole s per Kilogram of S oil Base 10 L ogarithm Black Carbon C ontent (BC) Milligrams per Gram of S oil Square R oot Acid Insoluble Carbon C ontent (C acid insol ) Milligrams per Gram of S oil Natural L ogarithm Clay C ontent (% Clay) % Clay by Dry W eight None Time Since Last B urn Months None
27 CHAPTER 3 RESULTS Patch level D ata from Field Measurements While t he distri bution of time since burn across all fields between upland and lowl and fields includes fields at the extreme ends of the sampling range, the vast majority of upland fields were recently burned, and only one upland field fell into the range of 10 to 35 mont hs since last burn (Figure 3 1) Conversely, several lowland fields fell in the middle of the sampling range. This reflects the fact that lowland fields tend to be managed under shorter burn cycles. Thus, while more upland fields had been very recently bur ned, they were also reported to be on much longer average burn cycles. Over half of all patches had a slope of <10% (Figure 3 2) with slopes of <5% characteristic of lowland patches out the flattest portions of hillsides and hilltops for cultivation. The upland patches with steeper slopes often comprised the sloping edges of a hilltop field Upland and lowland fields were generally both planted in the same variable matrix of field crops (Appendix D). Twenty one out of 45 upland patches contained manioc, 22 contained plantai n, and 10 contained pineapple. Four upland patches contained very small stands of sugar cane, and one included a small planting of maize. Thirty five out of 40 lowland patches cont ained manioc, while 37 contained plantain. Twelve lowland patches included rice, 6 were extensively planted in papaya, and 6 contained small plantings of maize. One lowland patch consisted of a small area planted extensively in sugar cane. Manioc was typic ally planted evenly across entire field s and was thus often interplanted with other field crops. Spacing between planted manioc stems ranged
28 approximately 0 .5 2.5 m. Spacing in mo st patches range d 1 1.5 m. Plantain was often planted in a manner similar t o manioc in lowland fields, but at upland sites, it was often intentionally planted in locations with abundant charcoal and ash, such as areas around charred chunks of timber. The spacing between planted corms r anged approximately 1.5 7 m, with most 3 5 m; average spacing was slightly wider i n upland fields The s pacing of these field crops was the result of many different factors. Manioc and plantain are both vegetatively propagated and in some cases, the spacing adopted by the farmer was reportedly simply the result of dividing a set amount of planting material over a particula r area. In other instances, spacing s were reportedly dictated by physical barriers such as logs or other debris. While the location and spacing of plantain may have been influenced b y the presence of ash and organic matter in some upland f ields, this reportedly was not a factor in planting decisions in lowland fields. In almost all fields, farmers reported ly used the same sequence of prac tices in clearing and burning field s First, th ey cleared the understory with machetes. Then they felled the large trees with axes, sometimes leaving trees which were too difficult or dangerous to fell. These t asks were often completed in communal work parties. The cut material was then almost always l eft to dry out before burning, anywher e fro m two weeks to three months. T he only locations where active manipulation of cut material was observed were sites where farmers had constructed charcoal furnaces; leftover material from these charcoal furnaces was applied to homegardens on occasion, but not broadcast over fields. In upland fields, material tended to be left longer before burning, especially in areas where large trees had been fe lled. In one lowland field that had been on a short burn cycle for some time, the farmer indicated that there had been
29 no delay between clearing and burning. Farmers typically reported igniting several points in their fields with small torches and then leaving the area, returning at a later time to observe the effects of the burn and begin planting. In every field surveyed, the visual e ffects of the last burn were highly heterogeneous. This is to be expec ted, both because of the way that burns we re conducted, and because of inevitably high and variable fuel moisture contents. All fields were at least occasionally weeded with machetes, so ruderal weeds almost always took the form of low lying ground cover. Percent weed cover r anged from 0 to 100%, with most patches either mostly covered with vegetation or mostly devoid of livin g ground cover (Median 25.0%, mean 36.9%, standard deviation 35.5%) In some cases, percent weed cover may have been the result of edaphic factors, but in others, it merely reflected the f requency of weeding in a patch. As was previously mentioned, many f armers reported having moved around the local area, many participated in a system of communal work parties, and most had access to both upland and floodplain sites. It was also common to hear from re cent arrivals to a village that an elder member of the co mmunity had given them one of their old fallows to work. The practices witnessed all appeared to fall under a single broad land management system, and aside from the broad differences between practices on different landforms, variation in their implementat ion most often appeared to arise from unexpected weather events, local field conditions, or extreme home. While each field was owned and managed by one person, it was likely worked by many other people at the same time This was true throughout the cycle of clearing and burning: a young couple report ed that a re cently felled upland field that they had
30 been having trouble getting out to visit had been burned by an unknown passerby, aside from those differences observed between landforms, there do not appear to be any large overall difference in farming practices from one field to the next or one community to the next. The crop assemblages farmers planted were, to some extent, a function of broad landscape scale constraints, but the two most common field crops, manioc and plantain, were grown almost everywhere. Furthermore, the limiting factor in planting these crops was most often access to p ropagules. For these reasons, it was impossib le to thoroughly address t he question of which cropping patterns contributed to the highest quantities of charcoal and soil organic matter, the scope of the study was restricted to examining the relationships be tween soil chemistry, land form s and fire management practices. Soil Chemistry M easurements of total N, total C, % clay, bulk density, BC, and acid insoluble C did not reveal extreme differences in distribution between upland and lowland fields (Tables 3 1 and 3 2) Conversely, the measurements of soluble P and ECEC showed differences approaching an order of magnitude between the two landforms. Measurements of pH from the surface 5 cm in upland patches ranged 3.62 5.01, and measurements from the 5 10 cm layer ranged 3.52 5.01 (Table 3 1 and Figure 3 3) While the measurement ranges for the two depth interval s were almost identical pH measurements from the lower 5 cm were skewed slightly higher than those from the surface. M easureme nts from lowland patches ranged 4.7 7.4 in the surface 5 cm, and 4.4 7. 3 in the lower 5 cm. In contrast to upland sites, pH measurements from lowland sites were slightly lower in the b ottom 5 cm than on the surface. Nonetheless
31 comparison of the two depths for each patch shows that pH in most patches on both landforms tended to decrease slightly wi th depth. P sol in the uppermost 5 c m in upland patches ranged 0.6 31.1 mg/kg, and measurements from the 5 10 cm layer ranged from undetectable concentrations to 5.3 mg/kg (Fig 3 4). Despite this extreme difference in range, most measurements from both dept hs fell between 0 and 10 mg/kg. Soluble P measurements f rom lowland patches ranged 1.5 90.4 mg/kg in the surface 5 cm, and from unde tectable concentrations to 81.4 mg/kg in the 5 10 cm layer. While measurements from the bottom 5 cm were generally lower tha n those from the surface, measurements from both depths showed similar bimodal distributions. The difference between depths on both landforms was most often between 0 and 10 mg/kg. T C in the surface 5 cm o f upland patches ranged 1 6 8 2 mg/g and 1 3 28 mg/g in the 5 10 cm layer (Fig ure 3 5) Most measurements from the surface 5 cm were between 2 0 and 4 0 mg/g with the lower 5 cm in most patches showing a decline of approximately 1 0 2 0 mg/g from the surface. Surface TC in lowland patches rang ed 7 5 6 mg/g and also most often declined approximately 1 0 to 2 0 mg/g in the lower 5 cm. Mean N content was almost identical i n upland and lowland patches, 0.27 and 0.28% respectively. On both landforms, m ost measurements ranged 0.1 5 0 .4% at the surface, and 0. 1 0.25% i n the lower 5 cm (Fig ure 3 6) S urface bulk density ranged 0.47 1.07 g/ cm 3 in upland patches, and 0.46 1.17 g/ cm 3 in lowland patches. B ulk density increased fairly consistently with depth on both landforms. Measurements from the 5 10 cm layer o f upland pat ches ranged 0.64 1.46
32 g/ cm 3 while those for the 5 10 cm layer of lowland patches ranged 0.3 1.28 g/ cm 3 (Figure 3 7). Measurements of clay sized particle percentages by weight from th e top 5 cm showed no s ubstantial difference between upland and lowland pa tches (Figure 3 8) However, it is important to note that the min e r ology o f clay s encountered on the two landforms were substantially differe nt and that clay minerolog y was not explicitly investigated Thus, similar clay contents observed in sa mples from different landforms should not be presum ed to signify similar effects on bulk soil chemistry. ECEC in the surface 5 cm of upland patches ranged 89 514 cmol/kg, while surface ECEC in lowland patches ranged 184 1247 cmol/kg (Figure 3 9 and Appendix C). The larger average total charge in lowland patches was primarily driven by concentrations of Ca 2+ and Mg 2+ which were not found in abundance in upland areas. Conversely, Al 3+ concentrations comprised most of the charge in upland patches, and were very low in lowland patches in both absolute and relative terms. Thus, the measurements from upland patches not only suggest fewer cation exchange sites in these locations, but also indicate that the majority of the exchange sites are occupied by Al 3+ BC measurements from the top 5 cm were 0.43 17.52 mg/g in u pland patches, and 1.22 12.80 mg/g in lowland patches (Figure 3 10) C acid insol measurements ranged 7.86 42.33 m g/g in upland patches, and 3.22 21.69 mg/g in lowland patches. BC typ ically comprised 10 40% of tot al C in both upland and lowland patches. C acid insol typically comprised 30 60% of total C in upland patches, and 20 70% of total C in lowland patches
33 Interrelationships B etween Variables Effects A cross All S ites ANOVA and ANCOVA Time since last burn had no significant effect on any of the 0 5 cm depth soil variables, although some inter group difference in least squares means could be seen in TC N and ECEC (Table 3 3) Patches in the i ntermediate treatment group (time since last burn 6 10 months) had th e highest mean TC and N measurements while those in the recent group (time since last burn 0.5 5 months) had the lowest. This finding is most likely the result of fields from the intermediate group occurring primarily on lowland sites with higher nutrient concentrations, and relatively nutrient poor upland patches forming a substantial portion of the recent category. This is also the most likely explanation for the apparent increase in ECEC with time since burn. B C showed a significa nt positive correlation with TC and negative correlation with BD. This most li kely indicates that BC production was directly tied to the total amount of organic matter in eac h patch, and that SOM was the strongest determinant of bulk density. Clay content was a statistically sig nificant covariate for each variate. This suggest s that the presence of certain clay types, such as those found in recent alluvial deposits, has a substantial relative influence on soil chemical parameters. It could also suggest that slope and landscape po sition, which partially determine clay content, are important determinants of bulk soil chemistry. For 5 10 cm depth measures, time since last burn had no significant effect on any of the variates, although some inter group difference in least squar es mean s could be seen in C and N (Table 3 4) Again, these differences are likely the result of patches
34 in the intermediate time since last burn group being located in lowland fields, rather than any particular post burn dynamics. The ANOVA and ANCOVA analyses o n all patches from both landforms indicate that few generalizations can be made across all of the sites sampled, and that time since burn had no significant effect across all of the fields surveyed This is, in part, because differences due to time since b urn are confounded by the fact that the two landforms are not equally represented in each group. Furthermore, the variation found across the sites studied overwhelms the differences one might observe if it wer e possible to make a series of measurements in each location over the timeframe examined. Detection of statistically significant differences due to time since last burn would likely require a much longer timescale. Multivariate Analyses by Landform Upland p atches T he first 4 principal components from t he PCA conducted on the surface 5 cm of upland sites were retained for further analysis These 4 components explained 75.9% of the variation i n the data (Figure 3 11 Table F 1 ) The first principal component (eigenvalue 3.10, 31 % of variation explained) included significant component loadings for BD TC N and P sol The grouping of these varia bles together suggests that this component primarily represents SOM along with the P and N either adsorbed to or included in SOM particles The association of C aci d insol with this component may indicate that C acid insol affected P and N concentrations, or simply that C acid insol is strongly tied to TC Time since last burn, with the highest component loading at 0.70, was likely the driver of variation for the secon d principal component (eigenvalue 1,85, 18.5% of
35 variation explained) pH and BC both showed negative correlations with PC 2, suggesting a decrease in both with time sinc e burn. Conversely, TC showed a positive c orrelation with PC 2, suggesting an increase w ith time since burn. However, the sc atterplot of T C versus PC 2 (Figure H 3 ) suggests that the patches within each field, which necessarily share the same time since burn, are also often positively correlated with P C 2, so this could also simply reflect the relationsh ip between TC and pH. The contribution of time since last burn and clay content to the third principal component (eigenvalue 1.48, 14.9 % of variation explained) suggest that they are the factors driving the described variation. The presence of BC suggests either a relationship between BC and one of these two variabl es, or a contribution from BC to CEC While all three variabl es with significant loadings on the 4th principal component (eigenvalue 1.15, 11.5 % of variation explained) appear in pre vious components, this is the first component where all three of these variabl es are significant. The presence of clay in this set may indicate a mechanism which is only present in certain types of clay, but the presence of pH and ECEC may also simply sugg est the effects of pH on exchange capacity, with clay content influencing both to a smaller degree. T he first 3 principal components from the 5 10 cm depth of upland sites, which explained 76.4% of the variation in the data, were retained for further analy sis (Figure 3 11, Table F 2) The first (eigenvalue 2.16, 36.1 % of variation explained) mirrors that of the surface, suggesting that it is driven by the same mechanisms. PC 2 (eigenvalue 1.39, 15.5% of variation explained) indicates a strong inverse relat ionship between time and pH in the lower 5 cm. PC 3 (eigenvalue 1.03, 13.0 % of variation explained )
36 suggests an inverse relationship between time since last burn and both BD and P sol with time since last burn only explaining a portion of the variation in the other 2 varia bl es. The strongest overall regression relationship between vari ables a ssociated with PC1 from the surface 5 cm was a positive correlation between TC and N which suggests that the m ajority of the measured N was likely organic (Table 3 5 ) C acid insol had a strong positive corre lation with P sol Th e remaining relationship, a positive correlation between TC and P sol had a n ear significant p value of 0.08 The correlation between C acid insol and P sol was made even mor e relevant by th e weak r elationship between TC and P sol Thus, it is likely that P sol levels in upland fields were related to properties uni que to the C acid insol pool. This may reflect that P was part of partially burned mat erial, or it may reflect P that was adsorbed onto excha nge sites in this material. Finally, a strong negative relationship was also observ ed between TC and BD This suggests that variation in SOM was the most important determinant of bulk density, and that mineralogical differences between site s were relativel y unimportant. Regression analyses between variates associated with the second principal component from the surface 5 cm yielded two statistically significant relationships. The strongest regression relationship between variables associated with PC 2 from t he surface 5 cm was a negative correlation between time since la st burn and BC showing strong evidence of BC loss over the time scale While it is possible that this most recalcitrant C pool would degrade over this period BC loss with time since burn ma y instead be the result of erosion and leaching of BC particles (translocation). A negative
37 correlation was also detected between time since last burn and pH, which could be associated with BC loss ; a decline in BC content may lead to loss of buffering cap acity. None of the regression analyses conducted on variates associated with the third principal component from the surface 5 cm yielded statistically significant res ults. This indicates that no single factor examined controlled ECEC. That said, the result s of these regressions do clearly indicate that clay content had no effect on BC content, and that BC content had no effect on ECEC. Of the two new rel ationships suggested by the 4 th principal component from the surface 5 cm the positive correlation betw een clay content and pH was significant, while the relationship between ECEC and pH was not. These results are consistent with the idea that this component describes the presence of certain clay types which influenced ECEC, but were not common enough to ha ve an impact across all patches. While the relationship between pH and clay content was statistically significant across all upland patches, the fact that pH and clay content were only associated in the 4 th principal component suggests that these same clay types were the main drivers of this relationship. As with the surface 5 cm, there was a very strong correlation between total C and total N in the 5 10 cm depth of upland fields (Table 3 5) The relationship between total C and soluble P was considerably stronger than at the surface. This, coupled with substantially lower soluble P measurements from the 5 to 10 cm depth, suggests that, like N most of the labile P in the lower 5 cm was associat ed with SOM Time since last burn and pH, which were associated in the 2 nd principal component, showed relatively strong positive relationship s Coupled with the negative relationship between time and
38 pH at the surface, this likely describes the movement of basic compounds associated with ash as they were leached lowe r into the soil profile. No statistically significant relationships were detected between the three variates associated with the 3 rd principal component. This suggests that no real functional relationships existed between time, bulk density, and soluble P, and that PC #3 may simply have been des cribing some residual variatio n. Lowland p atches The f irst 4 principal components from the PCA conducted on the surface 5 cm of lowland sites which explained 73.5% of the variation in the data, were retained for fur ther analysis (Figure 3 11, Table F 3 ). T he loading cutoff of 0.4 excluded only two varia bl es fro m the first principal component (eigenvalue 3.63, 36.3% of variation explained ), P sol and ECEC. Thus, in the case of surface data in lowland fields, PCA achiev ed little reduction in the data set. All three analytical C pools had significant loadings on this component but the fact that TC had the highest loading suggests that the contribution of the other two pools may simply be a function of the ir inclusion in t he TC pool. This, coupled with t he high loading for % clay sugg ests that C and clay content were the drivers of variation in this component. The second component (eigenvalue 1.55, 15.5% of variation explained) suggests a positive relationship between time since last burn and ECEC, which is unrelated to the dynamics driving the first component. This may indicate that exchange sites were freed up by leaching or created by decomposition in the time that passed after each burn, or it may be the result of diffe re nces in the type of clay and SOM found on sites that had been in cultivation longer.
39 All of the variabl es wi th significant loadings for PC3 (eigenvalue 1.30, 13.0% of variation explained ) were also associated with PC 1. This component appears to describe i ncreases in bulk density, decreases in BC content, and lower pH associated with time since last burn. As with upland fields, the decrease in BC is unlikely to be due to degradation over the time span evaluated It is more likely the result of smaller, ligh ter BC particles being transported offsite by rill erosion, or during occasional periods of inundation. Another possibility would be that the decrease in BC is simply a function of TC on those sites which happened to be burned earlier or later, and is not directly related to time since fire. However, the relatively homogeneous grouping of fields in the scatterplot of BC versus Time since last burn (figure H 13) suggests that this is not the case. PC 4 (eigenvalue 0.88, 8. 8 % of variation explained ) is the fir st component in which P sol had a significant component loading. Its gro uping with C acid insol alone in this component suggest s that the C acid insol pool had unique properti es that were associated with P sol One possibility is that the C acid insol poo l incl uded some pyrolized C st ructures that provide exchange sites for la bile P and anot her is that the C acid insol pool reflected parti ally burned organic matter that itself contained substantial quantities of labile P. It is also possible that the source of t he organic matter measured i n the C acid insol pool was actually riverine deposits, rather than vegetation from the sites themselves, and that the increase in P sol associated with this measurement was also connected to riverine sources. All five of the poss ible principal components from the 5 10 cm depth on lowland sites e xplained s ubstantial portions of the variability, and all six variates had component
40 loadings above 0.4 on the first principal component. Thus, PCA failed to achie ve a reduction in the data set for this depth ( Figure 3 11, Table F 4 ) . Out of the 22 regression a nalyses conducted between variabl es associated with the first principal component from the surface 5 cm on lowland sites 14 revealed statistic ally significant relationships (Table 3 6 ) Time since last burn affected both BC content and bulk density. T his coupled with the strong correlation between BC and bulk density suggests that the increase in bulk density over time was linked to loss of BC. As previously mentioned, t he most likely mechanism for this loss is washing of BC from the soil surface during rai nfall and periodic inundations. T ime since last burn also had a near significant effect on N perhaps reflecting some leaching of free N from ash and decomp osing organic matter over time. However, the str ong correlation between N and TC coupled with the lack of r elationship between TC and time, suggests that most N was not directly involved with this process. Clay content was clearly a major driver of variation in lowland patches; co rrelations were found between % clay and every var iabl e examined except BC The s trong relationship between % clay and TC suggests that higher clay content either created or was associated with conditions that both fa vored the preservation of C compounds a nd also prevented their loss due to erosion or leaching Alternatively, higher clay content in certa i n patches might also reflect increased deposition from riverine sources, which may also have included deposition of organic and inorganic C Similar relat ionships we re detected between BD and both TC and BC, while the relationsh ip between C acid insol and BD was not statistically significant. This suggests that
41 both charcoal and unburned organic matter substantially decrease bulk density while partially bur ned material had little overall effect. Both total and acid insoluble C decreased with pH, while BC had no relationship This suggests that increases in pH were linked to overall organic content, either pre or post burn. Post burn SOM may have had a substa ntial influence on pH, or, alternatively, sites with relatively large amounts of surface slash and litter may simply have had larger post burn inputs of basic compounds associated with the portion of vegetation which was reduced to fine ash. In either even t, given the relatively small amount of BC present, the only way it c ould affect pH would be by buffering via adso rption of basic compounds into its honeycomb structures. The fact that BC content had no effect on pH indicates that this mechanism was not in effect on these sites, or that it was overwhelmed by other factors in the soil matrix. All three analytical C pools decreased with N but the strongest correlation o bserved was between TC and N This finding s uggests that the type of C was not particularl y important, but that unburned organic matter still had the la rgest overall effect on N content. Once again, this indicates that most of the measured N was integrated into the structure of the SOM rather than adsorbed onto surfaces due to particular chemi cal properties. The positive correlation between pH and N was likely the result o relationship with both of these varia bl es, rather than a ny direct mechanism Although the second principal component suggested a relationship between time s ince last burn an d ECEC, these two variates were not significantly correlated. The p value of 0.11 suggests that some relationship likely exists between the two, but that it is
42 not strong enough to overcome edaphic and other influences on ECEC. Likewise, w hile the fourth principal component indicated a potential relationsh ip between C acid insol and P sol this relationship was not significant. Out of the 9 regression analyses conducted on variates from the 5 10 cm depth on lowland fields only 2 revealed sta tistically significant relationships (Table 3 6 ) The positive relationship between time since last burn and bulk density suggests SOM loss over time, but there was no relationsh ip between TC and BD Thus, this relationship might instead be the result of s oi l compaction due to the collap se of soil crumb structure during field clearing operations As with analyses conducted on t he other three groups, a strong positive correlation w as detected between TC and N Again, this suggests that most N at this depth was found within SOM molecules.
43 Figure 3 1 Months since last burn by field t ype
44 Figure 3 2 Patch percent slope by field t ype
45 Table 3 1. Descriptive statistics for upland p atches Variable Mean Std Dev Min Max Soluble P (mg/kg), 0 5 cm Depth 6. 9 6.3 0.6 31.1 Soluble P (mg/kg), 5 10 cm Depth 2.8 1.3 0.0 5.3 T otal C (mg/g) 0 5 cm Depth 3 6 1 4 1 6 8 2 T otal C (mg/g) 5 10 cm Depth 1 8 4 1 3 2 8 % N, 0 5 cm Depth 0.27 0.06 0.17 0.45 % N, 5 10 cm Depth 0.18 0.03 0.13 0.24 Bulk Dens ity (g/cm3), 0 5 cm Depth 0.79 0.13 0.47 1.07 Bulk Density (g/cm3), 5 10 cm Depth 0.98 0.16 0.64 1.46 ECEC (cmol/kg), 0 5 cm Depth 255.7 121.8 89.4 514.7 Black Carbon (mg/g), 0 5 cm Depth 8.0 3.9 0.4 17.5 Acid Insoluble C (mg/g), 0 5 cm Depth 16.6 5.9 7.9 42.3 % Clay, 0 5 cm Depth 40.8 9.5 25.1 70.2 *N=45, DF=44 Table 3 2. Descriptive statistics for lowland p atches* Variable Mean Std Dev Min Max Soluble P (mg/kg), 0 5 cm Depth 45.7 25.8 1.5 90.4 Soluble P (mg/kg), 5 10 cm Dep th 33.4 27.2 0.0 81.4 T otal C (mg/g) 0 5 cm Depth 2 9 1 0 7 5 6 T otal C (mg/g) 5 10 cm Depth 1 3 3 4 2 0 % N, 0 5 cm Depth 0.28 0.08 0.07 0.42 % N, 5 10 cm Depth 0.16 0.03 0.08 0.24 Bulk Density (g/cm 3 ), 0 5 cm Depth 0.89 0.15 0.46 1.17 Bulk Density (g/cm 3 ), 5 10 cm Depth 1.04 0.17 0.30 1.28 ECEC (cmol/kg), 0 5 cm Depth 652.9 206.2 184.9 1247.0 Black Carbon (mg/g), 0 5 cm Depth 7.1 2.6 1.2 12.8 Acid Insoluble C (mg/g), 0 5 cm Depth 12.2 3.8 3.2 21.7 % Clay, 0 5 cm Depth 40.0 10.5 18.1 69.0 *N=40, DF=39
46 Figure 3 3 pHw in 0 5 cm s oil (x) vs 5 10 cm s oil (y) by p atch Figure 3 4 Soluble P in 0 5 cm s oil (x) vs 5 10 cm s oil (y) by p atch Upland Sites Lowland sites Upland Sites Lowland sites
47 Figure 3 5 Total C (mg/g) in 0 5 cm soil (x) vs 5 10 cm soil (y) by p atch Figure 3 6 % N in 0 5 cm soil (x) vs 5 10 cm soil (y) by p atch Upland Sites Lowla nd sites Upland Sites Lowland sites
48 Figure 3 7. Bulk density ( g/cm3) in 0 5 cm soil (x) vs 5 10 cm s o il (y) by p atch Figure 3 8 Clay sized particle mass percent by p atch Upland Sites Lowland sites
49 Figure 3 9 ECEC (cmol/kg) by p atch
50 Y axe s: Black Carbon and Acid insoluble Carbon, mg/g: x Black Carbon (mg/g) Acid insoluble C (mg /g) Figure 3 10. Total C, a cid Insoluble C, and BC content of soils by patch
51 Tabl e 3 3. Summary of ANCOVA on 0 5 cm d epth Time S inc e F ire BC Content Clay Content Time Since Fire Treatment Group Recent Intermediate Old Variabl e F V alue P V alue F V alue P V alue F V alue P V alue Least Squares Mean Standard Error Least Squares Mean Standard Error Least Squares Mean Standard Error Bulk Density (g/cm 3 ) 0.30 0.75 5.78 0.0186 4.67 0.0337 0.81 0.05 0.81 0.04 0.86 0.06 Total C (%) 0.62 0.57 13.11 0.0005 8.2 8 0.0052 2.906 0.024 3.649 0.015 3.610 0.036 Total N (%) 0.96 0.44 2.96 0.0894 7.06 0.0096 0.256 0.001 0.299 0.000 0.268 0.000 Soluble P (mg/kg) 0.10 0.91 1.03 0.3138 4.87 0.0304 6.883 3.860 8.019 3.224 9.913 4.361 ECEC (cmol/kg) 0.18 0.84 0.11 0.7446 3 .71 0.0580 297.00 1.40 299.76 1.30 394.90 1.51 pH water 0.04 0.96 0.19 0.6658 10.51 0.0018 Table 3 4 Sum mary of ANOVA on 5 10 cm d epth Treatment Time Since Fire Treatment Group Recent Intermediate Old Variate F Value P Value Least Squares Mean St andard Error Least Squares Mean Standard Error Least Squares Mean Standard Error Bulk Density (g/cm 3 ) 0.33 0.74 1.00 0.03 0.98 0.02 1.02 0.03 Total C (%) 2.80 0.14 1.56 0.003 1.827 0.002 1.386 0.004 Total N (%) 2.80 0.16 0.170 0.000 0.18 2 0.000 0.153 0. 000 Soluble P (mg/kg) 0.20 0.82 3.212 1.957 3.616 1.667 3.971 2.067 pH water 0.08 0.93
52 Figure 3 11. PCA summary p lot s
53 Table 3 5 Variate regressions on upland p atches P.C. # X Y R 2 Adjusted R 2 F ratio Prob >f Slope Intercept 0 to 5 cm Depth 1 Total C Bulk Density 0.26 0.24 14.98 0.0004 1.9295718 1.1560283 Acid Insoluble C Bulk Density 0.20 0.18 10.53 0.0023 0.1909682 1.3195435 Total C Total N 0.59 0.58 61.63 <.0001 0.1235128 0.0288088 Acid Insoluble C Total N 0.24 0.22 13.18 0.00 07 0.0088209 0.0277376 Total C Soluble P 0.07 0.05 3.16 0.0828 98.207959 4.4880233 Acid Insoluble C Soluble P 0.31 0.29 19.14 <.0001 23.590174 42.12902 2 Time Total C 0.04 0.02 1.96 0.1690 0.0007161 0.182172 Time B.C. 0.30 0.28 18.14 <. 0001 0.0418328 3.0813654 Time pH water 0.13 0.11 6.29 0.0160 0.1516338 22.713854 Total C pH water 0.01 0.02 0.31 0.5795 31.867352 16.993077 B.C. pH water 0.05 0.03 2.28 0.1382 3.760768 33.221141 3 Time ECEC 0.00 0.02 2.64 0.1128 0.000 2096 2.3636648 % Clay ECEC 0.05 0.03 2.19 0.1465 0.0048608 2.1706904 % Clay B.C. 0.00 0.02 0.08 0.7858 0.0034509 2.5709368 B.C. ECEC 0.00 0.02 0.01 0.9420 0.0028683 2.3695914 4 % Clay pH water 0.13 0.11 6.301 0.0159 0.4970884 2.7031709 pH water ECEC 0.03 0.01 5.552 0.2489 0.0026858 2.4235857 5 to 10 cm Depth 1 Total C Bulk Density 0.11 0.09 5.40 0.0249 4.0485499 1.5222223 Total C Total N 0.42 0.41 31.13 <.0001 +0.1408561 0.0233839 Total C Soluble P 0.21 0.19 11.46 0.0015 +4 42.75328 36.80904 2 Time pH water 0.11 0.09 5.21 0.0275 +0.4237678 19.256718 3 Time Bulk Density 0.00 0.02 0.18 0.6748 0.0010406 0.984519 Time Soluble P 0.00 0.02 0.18 0.6723 0.0835706 23.738207
54 Table 3 6 Variate regressio ns on lowland p atches P.C. # X Y R 2 Adjusted R 2 F ratio Prob >f Slope Intercept 0 to 5 cm Depth 1 Time Bulk Density 0.22 0.20 10.78 0.0022 0.0061911 0.7967608 Time pH water 0.02 0.00 0.83 0.3668 0.1516338 22.713854 Time Total C 0.07 0.04 2.71 0.1 078 0.0006925 0.1778588 Time C Acid Insoluble 0.03 0.00 1.12 0.2966 0.0056694 2.5232342 Time B.C. 0.12 0.10 5.25 0.0275 0.0162785 2.859246 Time Total N 0.10 0.07 4.02 0.0522 0.0002073 0.0557038 % Clay Bulk Density 0.13 0.10 5.49 0.0244 0.0050 309 1.0881982 % Clay pH water 0.12 0.10 5.13 0.0294 0.3833139 5.1819595 % Clay Total C 0.27 0.25 14.13 0.0006 0.0015011 0.107761 % Clay C Acid Insoluble 0.18 0.16 8.43 0.0061 0.0153423 1.827351 % Clay B.C. 0.06 0.04 2.52 0.1204 0.0125276 2.1209521 % Clay Total N 0.30 0.29 16.62 0.0002 0.0003974 0.0367967 Total C Bulk Density 0.16 0.14 7.41 0.0098 1.9824582 1.2197039 Total C pH water 0.11 0.08 4.58 0.0389 126.4004 0.703386 Total C Total N 0.59 0.58 61.63 <.0001 0.1235128 0.0288088 C Ac id Insoluble Bulk Density 0.01 0.01 0.44 0.5091 0.0422666 0.990301 C Acid Insoluble pH water 0.18 0.16 8.35 0.0063 13.105313 11.4833 C Acid Insoluble Total N 0.22 0.20 10.74 0.0022 0.0093935 0.0297533 B.C. Bulk Density 0.17 0.14 7.52 0.0093 0.114 5986 1.1875805 B.C. pH water 0.00 0.02 0.10 0.7498 1.1528223 17.477784 B.C. Total N 0.05 0.03 2.08 0.1572 0.0032715 0.0441014 pH Total N 0.15 0.12 6.55 0.0146 0.0002484 0.0475855 2 Time ECEC 0.07 0.04 2.64 0.1128 0.0033075 2.7441701 4 C Acid Insoluble Soluble P 0.02 0.00 0.85 0.3616 4.5750511 31.66523 5 to 10 cm Depth 1 Time Bulk Density 0.07 0.04 2.74 0.0159 0.0039037 0.9842439 Time Total C 0.03 0.00 1.15 0.2904 0.0002307 0.1166356 Time Total N 0.07 0.05 2.95 0.0938 0.0001 0.0408208 Time Soluble P 0.09 0.07 3.71 0.0617 0.3084895 15.996054 Time pH water 0.08 0.05 3.15 0.0841 0.2860959 24.677 Total C Bulk Density 0.04 0.02 1.71 0.1983 2.321753 1.3042183 Total C Total N 0.69 0.68 82.86 <.0001 0.2290332 0.01341 86 Total C Soluble P 0.01 0.01 0.51 0.4805 88.257105 30.496696 Total C pH water 0.06 0.03 2.20 0.1466 179.65301 0.1510828
55 CHAPTER 4 DISCUSSION In the study area in Amazonian Peru, there are large differences between upland and lowland agricultural fields in terms of edaphic, vegetation, and cultural factors. These differences translate into different soil nutrient dynamics for each landform. Thus, i n addressing the question s of how and where the presence of charcoal influences soil chemistry and nut rient conce ntrations, upland and lowland areas need to be considered as different systems. Upland sites can be characterized by locations which were typically high above the water table, an d thus nutrient dynamics on these sites were driven by leaching, ru noff, and other weathering processes. C onversely, lowland sites are characteristically less well drained occasionally inundated, and more frequently experience puddling due to slow infiltration. While the conditions associated with weathering were importa nt determinants of soil fertility on upland sites there also appeared to be little important variation in these factors ; thus, conditions related to the quantity and type of SOM present were more likely to determine nutrients Lowland sites, on the other hand, appeared to consist of a more heterogeneous soil matrix in which depositional processes and microsite hydrology were at least as important as organic matter inputs in their influence on bulk soil chemistry. These differences are also connected to dif ferences in cultural practices on the two landforms. The longer average fallow times in upland fields are likely due, in part, to the need to accumulate enough biomass to burn and produce a useful quantity of ash and BC The shorter burn cycles on lowland to retain organic matter, as evide nced by their higher soil C contents, along with periodic deposition of silt and SOM from riverine sources.
56 These differences certainly t ranslate i nto differences in the time spent working fields located on these two landforms Most of the lowland fields sampled were adjacent to other fields, and thus appeared to be in high demand. In some cases, families in Tamshiyacu were w illing to frequently travel se veral kilometers upriver by canoe and then walk several kilometers into the interior in order to work outlying lowland fields. Conversely, upland fields were most often surrounded by abandoned fallow s or fairly mature forest, and farmers often men tioned that they got out to visit their upland fields only sporadically. When longtime residents of a community gave one of their fields to new arrivals, these were almost always upland fields. Meanwhile, certain prime lowland sites mo stly remain ed in one family for generations. To a certain degree, however, upland fields represented an important insurance policy for farmers. However productive lowland sites migh t be, they were also at risk of flooding, which could lead to catastrophic crop loss. Plantain, in particular, was likely to be killed by floods ; the fact that it is a perennial crop would make such losses even more devastating. Productive, open, and visible lowland fields near the center of town also invited crop theft, while up land sites were generally more secure due to their remote locations. While interpreting and extrapolating the results of this study, it is important to consider the difference in sampling regime between the two landforms. Despite the fact that they were on a longer burn cycle, all but two of the upland fields examined had been burned b etween 2 and 9 months prior to sampling Meanwhile, despite their generally shorter burn cycle, the lowland fields sampled were more evenly distributed across the full range o f times since last burn. Thus, our examination of time since burn
57 on upland sites was primarily focused on changes in chemistry in the first 9 months after a burn, while our examination of lowland sites captured more of the periodicity in their cultivation cycle. The full history of burn intervals would likely be an interesting factor to consider on either landform, but would essentially require sampling from hundreds of fields before data of this nature could be used to make reliable predictions. On upland sites, SOM was clearly t he most important driver of nutrient availability. In the surface 5 cm, P sol N and BD were all significantl y correlated with TC or C acid insol ; in the case of P sol and N the correlation was positive, w hile in the case of BD it was negative Total C was likely the causative varia bl e in the first principal component in the lower 5 cm, where it was negativ ely correlated with BD and positively correlated with N and P sol However, despite the contribution of BC to the total C pool we were unable to confirm any effect of BC on measured chemical parameters in upland fields. C acid insol almost always tracked the relationships that TC had with other varia bl es, albeit with slightly lower p values. The o ne exception was C acid insol s posi tive rela tionship with P sol which was significant wh ere the relationship with T C was not. Thus, we can conclude that unburned and partially charred material s were most important in maintaining P availability in upland fields, and that BC in these fields p layed a minor role in chemical processes. It is indeed possible, perhaps likely, that a portion of the BC particles found in upland fields provided some adsorption sites and chemical buffering. However, these effects were apparently miniscule in comparison to the direct mineralization of unburned and partially burned organic matter in an extremely acidic soil matrix. Another possible contribution to the matrix of SOM is that of humic compounds and root exudates. While the BC measurements yielded by the Kuhl busch
58 (1995) method correspond with some of the most conservative estimates of BC from the ring trials (Hammes et al 2007), the acid insoluble C pool cannot be conclusively linke d to any particular known C pool. The possibility remains that a portion of th e carbon measured in the acid insoluble pool is from these sources. Nonetheless, the acid insoluble C pool certainly represents fairly recalcitrant material, regardless of the source. We can therefore conclude that unburned litter and more labile humic com pounds are excluded from this pool. On lowland sites, principal component analysis failed to separate variables into meaningful components. In the surface 5 cm, clay content was fou nd to be a driver of all other variates except B C. Meanwhi le, BC was only f ound to have a negative effect on bulk density. Th is, coupled with the positive effect that total C had on total N and pH, suggests that direct mineralization of unburned and partially charred organic matter was also an important driver of nutrient availab ility on lowland fields, but that these processes were likely modulated by differences in clay content and microsite hydrology. Despite the presumed retention of SOM in many lowland patches, mean C measurements for all three of the pools examined were slig htly lower on these sites than in upland fields. However, the standard deviation of these measurements was also smaller. This is consistent with the fact that longer burn intervals in upland fields allowed more time for accumulation of biomass; the substan tial variation in SOM was likely a function of differences in the initial quantity of cut vegetation, which was then likely further amplified by differences in burn dynamics. Given the extreme differences in soluble P measurements between upland and lowlan d sites, it is reasonable to presume that P availability is the primary reason that
59 lowla nd fields are perceived by farmers to be more fert ile than upland fields. While total P was not measured, l ocations on both landforms likely had large stocks of occlud ed P The most likely reason for the relatively large amount of available P on lowland sites was the substantially higher pH in these areas. While residues fr om fires appeared to make important contribut ions to labile P stocks in upland fields, and appeare d not to in lowland fields, the actual quantity of P released from these residues was likely about the same on both landforms. On lowland sites, this amount was miniscule compared to background levels of labile P, while on upland sites, it likely often con stituted the majority of the labile P pool. E ffective cation exchange capacity measurements also showed extreme differences between upland and lowland fiel ds, and ECEC was also likely a primary driver of fertility in lowland areas. The lack of statisticall y significant correlations between ECEC and any of the other variates suggests that it is related to multiple factors or unmeasured mineralogical factors. The fact that ECEC was associated with other variabl es in the principal component analyses suggests t hat it was sometimes influenced by burn dynamics and biological inputs, but that these influences were unimportant relative to background mineralogical conditions. While farmers indicated that inputs from ash and organic matter were important for the ferti lity of thei r fields, and that, as a result, upland fields needed to be left fal low for long periods to a llow aboveground biomass to re a c cumulate it is unlikely that management of soil fertility is the primary consideration in clearing and burning practi ces. Crops such as manioc are well suited to growing in nutrient poor acid soils; what they are not well suited to is competition for light. The primary motivation for
60 clearing and burning is simply to remove shading trees, and then clear the ground for pl anting. Likewise, farmers often reported leaving fields to fallow when weed pressure became too great. After a burn, and especially in upland areas, farmers were certainly keen to take advantage of areas which they deemed to have burned particularly well, and residue from charcoal furnaces was often used as a soil amendment. In both cases, however, these appeared to be secondary considerations. Given current interest in biochars as soil amendments and atmospheric carbon sinks, one may well wonder whether th e systems described here could be modified to expand the contri bution of BC to both As the analy tical BC fraction was only found to affect bulk density on lowland sites, it seems reasonable to presume that it o nly reached a fraction of its potential contr ibutions to pH buffering and nutrient retention on this landform. Likewise, the strong positive correlation bet ween the C acid insol fraction and P sol on upland sites, with no other apparent corre lations between pyrogenic C and soil nutrient concentrations, suggests that this important contribution to fertility on upland in terfluves could also be increased There are three principle ways in which these practices might be altered to achieve these goals: 1) increasing the quantity of BC compounds produced and left on site, 2) creation of species of BC with higher sorption capacity and 3) incorporating some form of tillage to increase the residence time of BC in the soil. Modification of burning methods to include more cutting and piling of material might achie ve higher BC yields or the production of more sorptive BC species but this would also necessitate increased labor inputs. Several farmers did spend a great deal of time constructing and managing charcoal furnaces, but the ir primary rea son for this work wa s charcoal production for sale as described by Coomes and Burt ( 2001 )
61 Charcoal is arguably a more important commodity than any crops that are currently grown in charcoal amended soils. Thus it is difficult to imagi ne farmers investing substantial time a nd energy in producing charcoal as a soil amendment without substantial changes in crop prices and market structure. As for tillage, farmers in this region generally lack the equipment needed to efficiently bury organic m atter on a large scale and there i s also currently little incentive to do so. The introducti on of small scale labor intensive horticultural cropping systems might make such activity economically viable, but this would likely also require substantial changes in the current market structure In contrast to the marketing of charcoal, which farmers reported selling to dealers who would collect it either at field sites or at the village boat landing, marketing high value crops and forest products often entailed a long and treacherous trip by boa t to Iquitos; in order to avoid outright robbery, farmers usually end up selling their wares at cut rate prices to aggressive dealers who board the riverboats as they approach the city. It is likewise difficult to evaluate the long term impact of th ese cle aring, burning, and cropping practices as the y are likely to shift in response to economic factors. For example, many farmers who had previously focused on orchard crops in their upland fallows reported abandoning these practices as a result of falling fr uit prices, while others reported that manioc, which had previously been a subsistence crop, now fetched decent prices in urban markets. Changes such as these will inevitably influence The results c learly indicate that residues from slas h fires contribute to soil fertility on both upland and lowland fields. In the uplands these residues represent the most
62 important source of nutrients and chemical buffering in the system, while on lowland sites, the ir effects are often modulated, and likely sometimes eclipsed, by background fluvial factors. In both cases, however, unburned material may well remain the most imp ortant pool of nutrients over short and medium term s Experimental field studies incorporati ng partially burned material along with examination of individual sites over several burn cycles would provide further useful insights into the chemistry and importance of these residues.
63 APPENDIX A DATA FROM CHARCOAL F URNACES Active c harcoal furnaces w ere encountered in two upland fields and two lowland fields. One upland field near the town of 18 Febrero contained both a pit style rectangular charcoal furnace, located under a thatch roof, a nd an above ground charcoal kiln located in the open. One uplan d and two lowland fields near Tamshiyacu ea ch contained open air charcoal kilns Material was collected from the surface at each of the three open air kilns, and soil cores were taken at 0 5 cm and 5 10 cm from all 4 locations. As detailed in the methods section, material from the pit furnace and two of the open air kilns was used to formulate an analytical standard for the BC analysis. The full suite of soil analyses was conducted on samples from the 3 open air kilns, and a reduced suite of analyses condu cted on samples from the pit furnace (Table A 1, A 2). Table A 1 Mean a nalytical C, %N, and P sol measurements of material from a pit f urnace Measurement Value Soluble P (mg/kg) 7.92 % N 0.63 Total C (mg/g) 381. 0 Acid Insoluble C (mg/g) 135. 4 BC (mg/ g) 70. 6
64 Table A 2 Labor atory m easurements of m aterial f rom c harcoal k ilns Community Landform Depth Bulk Density pH w Soluble P (mg/kg) T otal C (mg/g) % N Acid Insoluble C (mg/g) BC (mg/g) ECEC (cmol/kg) K + (cmol/kg) Ca +2 (cmol/kg) Mg +2 (cmol/kg) Al +3 ( cmol/kg) NA + (CMOL/KG ) % CLAY 18 Febrero Upland Surface 67.60 31 2 3 0.95 19 5 0 5 6 0 5 cm 0.68 4.91 2.49 19. 1 0.21 1 2 7 6 2 49.07 42.90 0.00 0.00 0.00 26.16 49.07 5 10 cm 0.96 4.47 2.25 1 5 9 0.18 Tamshiyacu Upland Surface 21. 88 9 4 7 0.42 3 2 5 3 0 7 0 5 cm 1.00 5.46 3.60 1 9 6 0.19 1 1 1 6 6 32.04 21.67 129.73 0.00 0.00 24.74 32.04 5 10 cm 1.14 4.68 0.80 1 2 8 0.14 Lowland Surface 4 2 9 0.43 3 6 5 2 8 2 0 5 cm 0.63 7.16 75.20 4 5 5 0.46 1 4 3 8 6 52.95 82.14 104 1.41 215.15 0.00 33.93 52.95 5 10 cm 0.91 5.38 5.97 1 1 9 0.19 0 5 cm 0.70 6.55 25.60 4 7 3 0.40 2 2 3 1 0 7 34.18 59.03 411.67 176.94 0.00 26.31 34.18 5 10 cm 1.13 6.46 78.22 1 7 9 0.20 19 5 0 5 6
65 APPENDIX B DATA FROM ORILLA SIT E We colle cted samples and field data from four patches in one field in a seasonally inundated area, locally identified as an Orilla, near the town of El Chino. The entire field compr ising approximately 0.7 ha had last been burned 3 years (36 months) earlier, and had been in continuous cultivation since that time. Prior to that, it had been in fallow for at l east 30 years. Portions of the field, including 2 of the sampling patches, had also subsequently been burned 1 year (12 months) previously. In addition to plan tain and manioc, the first patch sampled also contained sugar cane, watermelon, and pumpkin (Table B 1) T h is was an interesting site that afforded an opportunity to collect exploratory data on active floodplain soils under a faster cycling burn regime (Ta ble B 2) However, these data were omitted from the main analysis due to the small sample size. Table B 1 Field data for orilla s ite Patch # Community Time Since Last Burn (months) Length of last fallow (years) % Slope Aspect (degrees) Manioc Present Pl antain Present Patch size (m 2) 1 El Chino 36 >30 0 X X 3000 2 12 >30 0 X X 400 3 12 >30 5 230 X X 1100 4 36 >30 0 X X 750
66 Table B 2 Laboratory data for orilla s ite Patch # 1 2 3 4 Soluble P (mg/kg) 0 5 cm Depth 10.06 8.13 8.51 4.36 Sol uble P (mg/kg) 5 10 cm Depth 1.37 4.37 2.96 1.42 T otal C (mg/g) 0 5 cm Depth 7 4 4 2 3 7 3 3 1 2 0 2 T otal C (mg/g) 5 10 cm Depth 8 9 1 7 2 1 6 3 1 1 4 % N 0 5 cm Depth 0.72 0.26 0.38 0.23 % N 5 10 cm Depth 0.12 0.22 0.23 0.18 Bulk density ( g/cm 3 ), 0 5 cm Depth 0.99 0.58 0.98 1.00 Bulk density ( g/cm 3 ), 5 10 cm Depth 1.31 1.04 1.02 0.96 ECEC (cmol/kg) 0 5 cm Depth 837.26 342.55 562.79 401.29 BC ( mg/g ), 0 5 cm Depth 6.14 7.69 9.51 4.14 Acid insoluble C ( mg/g ), 0 5 cm Depth 19.98 12.51 13.05 8.28 % Cl ay 48.67 53.29 46.26 43.94
67 APPENDIX C EXCHANGEABLE CATION DATA Figure C 1. Frequency histograms of exchangeable cation measurements by p atch
68 APPENDIX D PATCH LEVEL FIELD DATA Table D 1 Field d a ta for upland p atches Patch # Com munity Time Since Last Burn (Months) Length of Last Fallow (Years)* Delay Before Burning (Months) % Slope Aspect (Degrees) % Ground Cover Pineapple Present Manioc Present Plantain Present Manioc Spacing (m) Manioc Height (m) Plantain Spacing (m) Maximum Pl antain dbh (cm) Patch Size (m2) Field Size (Ha) 1 El Chino 6 20 0.5 0 10 X 1.5 3 4.5 6 250 0.4 2 6 20 0.5 0 10 X 1.5 3 4.5 10 500 0.4 3 6 20 0.5 0 10 X X X 1.5 3 4.5 4 450 0.4 4 6 20 0.5 0 10 X X 1.5 3 4.5 10.2 500 0.4 5 6 20 0.5 0 10 X X 1.5 3 3 200 0.4 6 18 Febrero 8 >30 1 0 80 X 3.5 14 2100 0.6 7 8 >30 1 5 324 70 X 1.5 1.5 300 0.6 8 8 >30 1 0 0 X 2.75 13.6 300 0.6 9 8 >30 1 3 20 0 500 0.6 10 8 >30 1 13 20 90 2700 0.6 11 Diamante 7 >30 1.5 0 5 X X X 2 2.25 4 10.8 700 0.96 12 7 >30 1.5 5 40 40 2 3000 0.96 13 7 >30 1.5 0 0 X X X 2 2.75 4 12.9 800 0.96 14 7 >30 1.5 0 5 X X X 2 3 4 17 500 0.96 15 7 >30 1.5 5 90 70 X X X 2 2.75 4 7.7 500 0.96 16 San Pedro 2 20 1 16 80 0 X 1 0.6 200 0.25 17 2 20 1 4 80 0 X 1 0.6 800 0.25 18 2 20 1 31 80 0 X 1 0.6 400 0.25 19 2 20 1 5 220 0 X 1 0.6 650 0.25 20 2 20 1 37 270 0 X 1 0.6 450 0.25 21 San Pedro 9 8 1 12 284 0 X 3.25 17 500 0.17 22 9 8 1 9 315 30 X 3.5 8 400 0.17 23 9 8 1 17 250 0 X X 3 13.5 200 0.17 24 9 8 1 6 170 10 X X 3.5 12.5 200 0.17 25 9 8 1 7 160 50 X 4 12 300 0.17
69 Table D 1 Continued Patch # Community Time Sin ce Last Burn (M onths) Length of Last Fallow (Y ears)* Delay Before Burning (M onths) % Slope Aspect (D egrees) % Ground Cover Pineapple Present Manioc Present Plantain Present Manioc Spacing (m) Manioc Height (m) Plantain Spacing (m) Maximum Plantain dbh (cm) Patch S ize (m 2 ) Field Size (Ha) 26 San Pedro 9 9 1 17 146 70 X 3.25 17 1500 0.64 27 9 9 1 15 146 20 X 2.5 15.4 100 0.64 28 9 9 1 25 282 80 X 3.75 16.1 50 0.64 29 9 9 1 50 160 90 X 4 15.2 1500 0.64 30 9 9 1 5 232 80 X 3.25 17 2 025 0.64 31 San Pedro 36 >30 3 24 240 80 X X 1.25 3.5 400 0.35 32 36 >30 3 35 264 80 X X 1.25 3.5 400 0.35 33 36 >30 3 14 140 80 X X 1.25 3.5 1500 0.35 34 36 >30 3 35 205 80 X 1.25 3.5 700 0.35 35 36 >30 3 56 210 80 X X 1.25 3.5 4.5 15.5 350 0.35 36 Tamshiyacu 2 6 0.5 35 0 80 X 1 0.85 250 0.25 37 2 6 0.5 4 170 80 X 1 1.25 700 0.25 38 2 6 0.5 35 270 80 X 1 1 150 0.25 39 2 6 0.5 35 184 80 X 1 1.25 150 0.25 40 2 6 0.5 11 30 70 X 1 1 150 0.25 41 Tamshiyac u 0.5 4 1.5 0 0.75 2.5 1200 0.24 42 0.5 4 1.5 15 300 0.75 2.5 300 0.24 43 0.5 4 1.5 36 270 0.75 2.5 200 0.24 44 0.5 4 1.5 17 300 0.75 2.5 100 0.24 45 0.5 4 1.5 35 260 10 X 0.75 2.5 200 0.24
70 Table D 2 Field data for lowland p atches Patch # Time Since Last Burn (M onths) Length of last fallow (Y ears)* Delay Before Burning (M onths) % Slope Aspect (D egrees) % Ground Cover Manioc Present Plantain Present Rice Present Maize Present Papaya Present Manioc Spacing (m) Manioc Height (m) Plantain Spacing (m) Ma ximum Plantain dbh (cm) Patch size (m2) Field Size (Ha) 1 36 Unknown 1.5 0 X X 0.35 1.5 16.2 4 1500 3.3 2 36 1.5 0 X X 1.5 1.5 16.5 4 1400 3.3 3 36 1.5 0 X X 1.5 1.5 18.5 4 2500 3.3 4 36 1.5 0 X 4 3000 3.3 5 36 1.5 0 X 4 800 3.3 6 3 8 2 0 0 X X X X 500 0.14 7 3 8 2 29 230 0 X X X X 175 0.14 8 3 8 2 6 336 0 X X X X 175 0.14 9 3 8 2 25 var 0 X X X X 350 0.14 1 0 3 8 2 20 60 0 X X X X 175 0.14 11 24 <10 2 0 90 X X 3.25 1.5 18.3 3 600 0.28 12 24 <10 2 0 5 X X 17.2 3 900 0.28 13 24 <10 2 0 5 X X 16.5 3 400 0.28 14 24 <10 2 0 95 X X 0.35 1.5 22.5 3 500 0.28 15 0.5 <10 2 15 5 X X 1 7.2 3 25 0.02 16 0.5 <10 2 5 70 5 X X X 25 0.02 17 12 4 0.5 0 80 X X 1.25 1.5 12.3 3 150 0.2 18 12 4 0.5 0 30 X X 0.55 1.5 16.1 3 150 0.2 19 12 4 0.5 0 90 X X 0.8 1.5 8.2 3 400 0.2 20 12 4 0.5 0 30 X X 0.825 1.5 13.2 3 100 0.2 2 1 12 4 0.5 0 95 X X 1.1 1.5 17.9 3 1000 0.2 22 8 5 Unknown 0 80 X X 1.1 1.5 17.7 3 500 0.5 23 8 5 0 5 X 13.5 3 400 0.5 24 8 5 0 35 X X 0.3 1.5 16.3 3 400 0.5 25 8 5 0 30 X X 1.25 1.5 7.8 3 600 0.5 26 8 5 0 25 X 11 3 3 00 0.5 27 3 5 0 10 X X X 0.15 6.7 100 0.13 28 3 5 0 30 X X X X 0.225 7.4 200 0.13 29 3 5 0 30 X X X X 0.25 4.9 500 0.13 30 3 5 0 10 X X X 1 6 500 0.13
71 Table D 2 Continued Patch # Time Since Last Burn (M onths) Length of last fa llow (Y ears)* Delay Before Burning (M onths) % Slope Aspect (D egrees) % Ground Cover Manioc Plantain Rice Maize Papaya Manioc Spacing (m) Manioc Height (m) Plantain Spacing (m) Ma ximum Plantain dbh (cm) Patch S ize (m2) Field Size (Ha) 31 12 >30 3 0 0 X X X 1 0.75 19.3 4.5 500 0.25 32 12 >30 3 0 90 X X 16.4 4.5 100 0.25 33 12 >30 3 0 70 X X X 3 0.75 15 4.5 400 0.25 34 12 >30 3 0 20 X X 3 0.75 14.2 4.5 250 0.25 35 12 >30 3 0 90 X X X 1.5 0.75 15.8 4.5 150 0.25 36 24 >30 3 0 20 X X 1 0.75 17.2 4 300 0.25 37 24 >30 3 7 340 5 X X 1.25 1 25.2 5.5 2800 0.25 38 24 >30 3 0 80 X X 1 0.75 20.8 4.5 250 0.25 39 24 >30 3 0 50 X X 2 1.25 14.5 6.5 50 0.25 40 24 >30 3 0 2 X X 0.25 1 22.8 4.5 500 0.25
72 APPENDIX E CORRELATION M ATRICES AND INTER DEPTH CORRELATIONS Table E 1 Correlation matrices and inter d epth correlations for upland p atches* Soluble P %C %N pH Bulk D ensity ECEC BC Acid Insoluble C % Clay 0 5 cm D epth Soluble P 1 %C 0.26 1 %N 0.35 0.77 1 pH 0.24 0.08 0.14 1 Bulk D ensity 0.30 0.51 0.64 0.07 1 ECEC 0.10 0.30 0.24 0.18 0.11 1 BC 0.08 0.17 0.11 0.22 0.19 0.01 1 Acid Insoluble C 0.55 0.38 0.48 0.19 0.44 0.23 0.27 1 % Clay 0.29 0.04 0.14 0.36 0.06 0.2 2 0.04 0.22 1 Time Since Last B urn 0.13 0.21 0.02 0.36 0.17 0.01 0.54 0.14 0.01 5 10 cm D epth Soluble P 1 %C 0.46 1 %N 0.29 0.65 1 pH 0.23 0.12 0.09 1 Bulk D ensity 0.07 0.33 0.39 0.02 1 Time Since Last B urn 0.06 0.04 0.1 4 0.33 0.06 Inter Depth C orrelations 0 5 cm vs. 5 10 cm 0 5 vs 5 10 0.42 0.40 0.44 0.77 0.29 See Table 2 3 for units of expression and transformations used.
73 Table E 2 Correla tion matrices and inter depth correlations for lowland p atc hes* Soluble P %C %N pH Bulk density ECEC BC Acid Insoluble C % Clay 0 5 cm Depth Soluble P 1 %C 0.25 1 %N 0.05 0.81 1 pH 0.35 0.33 0.38 1 Bulk Density 0.20 0.40 0.43 0.20 1 ECEC 0.17 0.16 0.18 0.36 0.12 1 BC 0.10 0.23 0.23 0.05 0.41 0.06 1 Acid Insoluble C 0.15 0.58 0.47 0.42 0.11 0.004 0.37 1 % Clay 0.23 0.52 0.55 0.34 0.36 0.17 0.25 0.43 1 Time Since Last Burn 0.08 0.26 0.30 0.15 0.47 0.25 0.35 0.17 0.16 5 10 cm Depth Soluble P 1 %C 0.11 1 %N 0.24 0.83 1 pH 0.29 0.23 0.49 1 Bulk density 0.25 0.21 0.36 0.27 1 Time Since Last Burn 0.30 0.17 0.27 0.28 0.26 Inter Depth Correlations 0 5 cm vs. 5 10 cm 0 5 vs 5 10 0.51 0.60 0.50 0.82 0.41 Se e Table 2 3 for units of expr ession and transformations used.
74 APPENDIX F FULL SET OF EIGENVAL UES AND COMPONENT LO ADINGS FOR P RINCIPAL COMPONENT ANALYSES Table F 1 Eigenvalues and component l oading s for u plan d patches, 0 5 cm d epth Principal Component Num ber 1 2 3 4 5 6 7 8 9 10 Eigenvalue 3.10 1.85 1.48 1.15 0.68 .57 0.45 .31 0.25 0.14 % of Variation Explained 30.97 18.54 14.85 11.54 6.85 5.73 4.48 3.08 2.53 1.44 Cumulative % of Variation 30.97 49.51 64.36 75.91 82.76 88.48 92.96 96.03 98.56 100.0 C hiSquare 161.12 108.00 79.45 52.95 28.93 19.93 11.68 5.25 2.78 0.00 DF 44.91 40.24 33.80 27.29 20.98 15.02 9.80 5.52 2.22 Prob>ChiSq <.0001 <.0001 <.0001 0.002 0.115 0.176 0.291 0.452 0.287 Component Loadings by Variate + Soluble P 0.63* 0.21 0. 38 0.02 0.44* 0.38 0.22 0.18 0.03 0.02 Total C 0.72* 0.54* 0.06 0.02 0.29 0.17 0.04 0.03 0.08 0.26 %N 0.84* 0.36 0.02 0.11 0.09 0.14 0.02 0.08 0.28 0.21 pH 0.32 0.58* 0.11 0.61* 0.07 0.10 0.36 0.14 0.10 0.07 Bulk Density 0.74* 0.21 0. 17 0.19 0.19 0.47* 0.05 0.20 0.21 0.02 ECEC 0.12 0.36 0.60* 0.60* 0.13 0.20 0.08 0.14 0.21 0.05 BC 0.38 0.46* 0.50* 0.31 0.41* 0.26 0.04 0.18 0.09 0.11 Acid Insoluble C 0.75* 0.18 0.29 0.28 0.07 0.10 0.31 0.32 0.16 0.01 % Clay 0.25 0.37 0 .59* 0.43* 0.39 0.13 0.26 0.04 0.15 0.02 Time Since Last Burn 0.23 0.70* 0.50* 0.12 0.12 0.13 0.30 0.22 0.12 0.10
75 Table F 2 Eigenvalues and component l oadin gs for upland p atches, 5 10 c m d epth Principal Component Number 1 2 3 4 5 6 Eigenvalue 2.16 1.39 1.03 0.58 0.53 0.31 % of Variation Explained 36.08 23.18 17.16 9.74 8.76 5.10 Cumulative % of Variation 36.08 59.25 76.41 86.15 94.91 100 ChiSquare 52.08 28.29 15.03 4.52 2.86 <.0001 DF 14.69 12.10 8.60 5.04 1.87 Pro b>ChiSq <.0001 0.0053 0.0766 0.4824 0.2182 Component Loadings by Variate Soluble P 0.61* 0.26 0.56* 0.08 0.47* 0.10 Total C 0.88* 0.06 0.06 0.20 0.15 0.40* %N 0.83* 0.23 0.11 0.25 0.23 0.36 pH 0.11 0.85* 0.16 0.31 0.38 0.07 Bulk Dens ity 0.56* 0.07 0.66* 0.46* 0.18 0.02 Time Since Last Burn 0. 10 0.74* 0.48* 0.41* 0.22 0.03
76 Table F 3 Eigenvalues and component l oadin gs for lowland patches, 0 5 cm d epth Principal Component Number 1 2 3 4 5 6 7 8 9 10 Eigenvalue 3.63 1.55 1.30 0.88 0.84 0.57 0.48 0.35 0.29 0.12 % of Variation Explained 36.26 15.51 12.96 8.78 8.38 5.72 4.77 3.49 2.89 1.23 Cumulative % of Variation 36.26 51.78 64.75 73.52 81.90 87.63 92.39 95.89 98.77 100 ChiSquare 143.70 77.11 58.21 40.47 31.79 19.80 13.98 8. 57 5.39 0.0 DF 44.42 40.71 33.67 26.85 20.21 14.48 9.42 5.28 2.1667 Prob>ChiSq <.0001 0.0005 0.0054 0.0444 0.0486 0.1568 0.1433 0.1463 0.0783 Component Loadings by Variate Soluble P 0.35 0.35 0.43 0.58* 0.39 0.28 0.03 0.01 0.03 0.07 Total C 0.83* 0.30 0.07 0.04 0.12 0.28 0.18 0.17 0.078 0.23 %N 0.82* 0.33 0.04 0.03 0.32 0.05 0.13 0.05 0.23 0.22 pH 0.56* 0.33 0.58* 0.14 0.19 0.21 0.04 0.37 0.01 0.08 Bulk Density 0.62* 0.26 0.44* 0.36 0.14 0.22 0.27 0.02 0.29 0.02 ECEC 0.03 0.87* 0.17 0.24 0.15 0.08 0.09 0.33 0.12 0.03 BC 0.49* 0.18 0.48* 0.21 0.60* 0.13 0.13 0.12 0.20 0.03 Acid Insoluble C 0.68* 0.14 0.24 0.51* 0.28 0.15 0.03 0.09 0.30 0.08 % Clay 0.71* 0.26 0.11 0.18 0.08 0.43* 0.39 0.19 0.05 0.02 Time S ince Last Burn 0.49* 0.43* 0.49* 0.05 0.23 0.30 0.41* 0.11 0.02 0.02
77 Table F 4 Eigenvalues and component l oading s for lowland patches, 5 10 cm d epth Principal Component Number 1 2 3 4 5 6 Eigenvalue 2.58 1.16 0.75 0.72 0.67 0 .12 % of Variation Explained 43.08 19.25 12.54 11.10 11.20 1.97 Cumulative % of Variation 43.08 62.34 74.88 86.83 98.03 100 ChiSquare 76.4396 40.1567 31.0584 28.1914 23.3409 <.0001 DF 14.95 12.90 9.12 5.29 2.22 Prob>ChiSq <.0001 0.0001 0.0003 <.00 01 <.0001 Component Loadings by Variate Soluble P 0.49* 0.57* 0.28 0.22 0.55* 0.01 Total C 0.72* 0.60* 0.05 0.15 0.21 0.22 %N 0.88* 0.40* 0.04 0.01 0.03 0.26 pH 0.66* 0.17 0.27 0.41* 0.53* 0.08 Bulk Density 0.57* 0.25 0.77* 0.14 0.01 0.03 Time Since Last Burn 0.53* 0.47* 0.08 0.68* 0.19 <0.01 *Statistically significant at 0.05 + Loadings with an absolute value >0.4 in bold, loadings with an absolute value <0.1 in italics.
78 APPENDIX G KEYS TO SYMBOLS USED IN SCATTERPLOTS GR OUPED BY FIELD Figure G 1 Key to upland f ields Figure G 2 Key to lowland f ields Each field is identified by a three digit code. The first two digits are lett ers which form an abbreviation of the community where the field was located. The last di git is simply a sequential number reflecting the order in which fields were sampled in each community. The following chart shows the abbreviations used for each community: Table G 1 Community a bbreviati on c odes Code Community CH El Chino DF 18 Febrero DI Diamante SP San Pedro TS Tamshiyacu
79 APPENDIX H PRINCIPAL COMPONENT SCATTERPLOTS AND REG RESSION SCATTERPLOTS Upland Fields, 0 5 cm D epth Upland Field s, 5 10 cm D epth Lowland Fields, 0 5 cm D epth Lowland Fields, 5 10 cm D epth Figure H 1 Principal component groupings by f ield* A key to the symbols used in all scatterplots grouped by field is located in Appendix G.
80 Figure H 2 Scatterplots with grouping by f ield for PC #1, u pland 0 5 cm depth
81 Figure H 3. Scatterplots with grouping by field for PC #2, u pland 0 5 cm depth
82 Figure H 4. Scatterplots with grouping by field for PC #3, u pland 0 5 cm depth
83 Figure H 5 Scatterplots with grouping by field for PC #4, u pland 0 5 cm depth
84 Figure H 6. Scatterplots with grouping by field for PC # 1, u pland 5 10 cm depth
85 Figure H 7 Scatterplots with grouping by field for PC #2, u pland 5 10 cm depth Figure H 8 Scatterplo ts with grouping by field for PC #3, u pland 5 10 cm depth
86 Figure H 9 Regression and scatter plots for upland patches, 0 to 5 cm d epth
87 Figure H 9 Continued
88 Figure H 9 Continued
89 Figure H 10 Regression and scatter pl ots for upland p atche s, 5 to 10 cm d epth
90 Figure H 10 Continued
91 Figure H 11 Scatterplots with grouping by field for PC #1, l owland 0 5 cm depth
92 Figure H 11 Continued
93 Figure H 12 Scatterplots with grouping by field for PC #2, l owland 0 5 cm depth Figure H 1 3. Scatterplots with grouping by field for PC #3, l owland 0 5 cm depth
94 Figure H 14 Scatterplots with grouping by field for PC #1, l owland 5 10 cm depth
95 Figure H 15 Scatterplots with grouping by field for PC #2, l owland 5 10 cm
96 Figure H 16 Reg ression and scatter plots for lowland patches, 0 to 5 cm d epth
97 Figure H 16 Continued
98 Figure H 16 Continued
99 Figure H 16 Continued
100 Figure H 16 Cont inued
101 Figure H 16 Continued
102 Figure H 16 Continued Figure H 17 Regression and scatter plots for lowland patches, 5 to 10 cm d epth
103 APPENDIX I SONDEO QUESTIONNAIRE 1. Cuando lleg a este region? (When did you arrive in this area?) 2. C uando Tumb y quem su chacra la ltima vez? (When did you last clear and burn your field?) Haba tumbado y quemado antes? (Had you cleared and burned it previously?) Cuantos veces? (How many times?) 3. Por favor, describa la ltima tumba y quema. (Please des cribe the last clearing and burning. ) 4. Cuanto tiempo pas entre tumba y quema? (How much time passed between clearing anf burning?) 5. Porque? (Why?) 6. Por favor, describa como sembr las plantas. (Please explain how you planted your crops.) 7. Cuanto tiempo p as entre quema y la dia que comienz plantar? (How much time passed between burning and the day you started planting?) 8. Porque? (Why?) 9. Como decide donde plantar las siembras en el campo? (How do you decide where to plant crops in the field?) 10. Porque sel eccion este sitio? (Why did you select this site?) 11. Que fue los raznes mas importantes? (What were the most important reasons?) 12. Cuando seleccion? (When did you select it?) 13. Cual siembras plantaba aqui en los aos pasados? (What crops did you grow here in years past?) 14. Las cosechas eran buenas? (Were the yields good?) 15. Cual plantas cultiva aqui actualmente? (What plants do you cultivate here currently?) 16. Que tipo de cosecha espera este ao? (What sort of harvest do you expect this year?) 17. Cual siembras planea plantar o cosechar en el futuro? (What crops do you plan to plant here in the future?) Que tipos de cosechas espera en el futuro? (What sorts of yields do you expect in the future?)
104 18. Cosechar otras plantas tiles de este campo? (Will you harvest o ther useful plants from this field?) Cuales? (Which?) 19. Cultiva otras chacras? (Are you farming any other fields?) Cuanto tiempo pas hasta la ltima tumba y quema? (How much time has passed since the last clearing and burning?) 20. Cuando dejar cultivar es te campo? (When will you stop working this field?) Porque? (Why?) 21. Tumbar y quemar este campo otra vez? (Will you clear and burn this field another time?) Cuando? (When?) Porque? (Why?) 22. Puedo tomar algunas pruebas del suelo y carbn de su chacra? (Ma y I take some samples of soil and charcoal from your field?)
105 APPENDIX J INFORMATION FROM FARMER INTERVIEWS Field: DF01 Size: 0.6 Ha Landform: upland Farmer: Senor L.C. Remote field located in the jungle next to 18 febrero. Field: TS04 Size: 0.25 Ha L andform: upland Farmer: Senor T.G.T. Senor T.G.T. is from lower Amazonia, and has lived in Tamshiyacu for 10 years. This field was mature forest when he cleared, felled, and burned it the first time in 1995. He cultivated rice, manioc, and plantain for ab out 5 years, then left it fallow for 6 years. Some fruits which also persist from the first clearing and planting include pijuayo, caimito, uvilla, and umari. The field cover consists primarily of grasses. Field: TS05 Size: 0.14 Ha Landform: lowland Farm er: Senora A.F.V. Senora A.F.V. is the daughter of Senora B.V.A. from TS06. There are orillas next to this restinga site. The orillas experience inundation every year, this year the water was not as high as it usually is, but the orillas were under water in January and February. This field was in fallow for 8 years. In January, they cleared the underbrush. Right now, they are planting rice, and are going to plant manioc and plantain. Senora B.V.A. will return another time to try to plant corn. Field: CH 01 Size: 0.4 Ha Landform: upland Farmer: Don M.C. Don M.C. cleared and planted this field one year ago. Previous to that, it was in fallow just barely high enough to be considered an upland.
106 Field: CH02 Size: 0.72 Ha Landform: floodplain Farmer: Senor J.P. Field located on the far side of the river from El Chino, inundated to a depth of 1 3 m every year. The field is typically burned in August September of each year. It was mature forest or o ld fallow before Senor J.P. began working it. Field: DI01 Size: 0.96 Ha Landform: upland Farmer: Senor A.H. Sole field sampled in Diamante. Diamante was first settled in 1991. It consists of 15 houses with 25 adults. The field was mature forest before clearing and burning. It consists of a large, wide gulley sloping down toward the lowland and floodplain, along with some adjacent higher areas. Field: SP01 Size: 0.25 Ha Landform: upland Farmer: Senor M.H. Senor M.H. is 39 years old. He moved from el Chino to San Pedro when he was 16 because of problems with flooding and related crop damage at El Chino. His family has many fallows. San Pedro consists of 40 houses with 75 inhabitants. The field is planted entirely in manioc, with a few useful herbs and nothing more. It was mature forest 23 fallow for 20 years. They try to seek sites with steeper slopes because they are good for plantain the first time, they planted manioc and plantain. The field is not completely burned, with many logs on the ground. M.H. has no plans to make charcoal here. He will replant the manioc stems while harvesting the tubers another two times, then leave the field to fallow. Field: SP02 S ize: 0.17 Ha Landform: upland
107 Farmer: Senor R.H.E. which is also held by Senor R.H.E. Ten years ago, this field was an old fallow from his cousin. At that time, R.H.E. planted plantain and manioc for 2 years. The manioc did not grow well, due to the presence of leafcutter ants. The field was then left fallow for 8 years, and was just replanted 9 months ago. This site was selected because it is close to the house, and be cause it was a fallow which had been left for many years. Plantain will grow well here the first time, the second time is not so good le ave it fallow for 4 years, then return to work it another time. Field: SP03 Size: 0.64 Ha Landform: upland Farmer: Senor R.H.E. Eleven years ago, this field was mature forest or old fallow. The first time R.H.E. cleared it, he grew plantain and manioc for 2 years, then left it fallow for 9 years. He worked this field at the same time as SP02, using the same plan for burning and planting. Likewise, he will also most likely work this field for 2 more years, then leave it fallow for 4 years after that. He plants everything everywhere, because you can never tell where exactly in the field the crops will grow well. However, he also tries to put plantain in locations where there is more ash and charcoal. The field consists of a cleared hillside, with contours running east west, adjacent to SP02. Burn severity is fairly consistent across the entire field. Field: SP04 Size: 0.35 Ha Landform: upland Farmer: Senor G.U.M. Senor G.U.M. moved to San Pedro in 2003. He received this parcel as a gift from its previou s owner. The field consists of a hilltop surrounded by mature forest. The top is more level, but still sloping. It was an old fallow or mature forest when he began working it. He planted manioc first, and pineapple 4 months later. He will harvest pineapple in September, then leave the field fallow for 5 6 years. He planted plantain only on the slopes; pineapple was only planted higher on the hilltop, where the sandy soils required could be found. There are few logs on the site, with most consumed in the bur n, but a fair amount of weeds and trees coming up on the site.
108 Field: TS02 Size: 3.3 Ha Landform: lowland Farmer: Senor S.V. Senor S.V. was born here, and is about 60 years old. He primarily grows sugarcane for ve, so he only uses it for subsistence crops. The site is located alongside the river, on either side of the asphalt track to 18 Febrero. It is bordered on one side by more flood prone riverbank (barreal) planted in rice, and on the other side by the begin ning of upland fallows. This field is contiguous with other ones located both up and downriver. Weeds grow fast on this site % cover varies widely, but is reportedly almost entirely on whether or not an area has been recently weeded. Field: TS06 Size: 0.28 Ha Landform: lowland Farmer: Senora B.V.A. Senora B.V.A.. has lived here for 50 years. This area floods when the river level is high, it last flooded 7 years ago. The field is a thoroughly cleared area adjacent to other fields, fallows, and TS07. It was well burned 2 years ago. This was a 10 year old fallow when Senora B.V.A. began to work it at that time. She will continue to cultivate this site for one more year, then leave it fallow for 2 3 years. Field: TS07 Size: 0.02 Ha Landform: lowland Farm er: Senora B.V.A. This is a cleared field adjacent to TS06, and is worked by Senora B.V.A. from TS06. Background conditions and burn regime are identical to those of TS06, except that this field was burned two weeks ago. The burn was spotty and incomplete due to rain. Senora B.V.A. planted maize here, but it did not come up perhaps because of the seed used, perhaps because of leafcutter ants. The plantain clusters in this field are from the fallow period they survived the most recent burn, and are at lea st several years old. She plans to reburn again this year and work the field for 2 3 more years before leaving it to fallow again. Field: TS08
109 Size: 0.2 Ha Landform: lowland Farmer: Senor J.S.S. Senor J.S.S. has lived here for 30 years. He has three f ields we are sampling from: TS08, TS09, and TS10. This is a field next to the river, near TS02, with a similar layout. The owner of this field is unable to work it right now, and so has lent it to Senor J.S.S. This was a 4 year old fallow when he cleared and burned it 1 year ago. Before that it was a pasture (pasto chacra purma chacra). This site has been in cultivation for at least 50 years. The last burn was incomplete and spotty due to rain. Field: TS09 Size: 0.5 Ha Landform: lowland Farmer: Senor J. S.S. This s ite is located deeper in the forest a short walk from the river. This is another field lent to Senor J.S.S. by the landowner. It was previously pasture, and had been in fallow for 5 years when Senor J.S.S. burned it 8 months ago. He burned in smaller increments due to the rain, as it was not possible to wait for drier weather. He w ill work here fo r 2 more years, then leave the field fallow Field: TS10 Size: 0.13 Ha Landform: lowland Farmer: Senor J.S.S. This site is located near TS09, and shares all of its characteristics. The only difference is that it was burned 3 months ago. Field: TS11 Size: 0.25 Ha Landform: lowland Farmer: Senor R.H. This site was an old fallow or mature forest when Senor R.H. burned it 1 year ago. He will work it for 5 years, clearing weeds for the plantains, and will leave it fallow when the weed pressure becomes too great. He will then return to work it again after 5 6 years of very Field: TS12
110 Size: 0.25 Ha Landform: lowland Farmer: Senora N.H. Senora N.H. is married to Senor R.H. This is a 100m x 25m rectangular area adjacent to other fields and fallows. Seno ra R.H. has planted or tended plantain, manioc, huasai, lemon, maize, and several herbs here. Field: TS15 Size: 0.24 Ha Landform: upland Farmer: Senor R.P.M. Senor R.P.M. is 27, and has lived here his entire life. His father has lived here his entire li fe, as well. He is married to the daughter of Senora N.H. and Senor R.H. His father has lived here his entire life, as well. His family has several umari orchards deep in the jungle, and he has other fields on the river floodplains. This field was mature f orest in 2003, Senor R.P.M. cleared the umari orchard to grow manioc for 1 year. It was then left fallow for 3 years. The field was cleared about a month ago, but th ey had difficulty getting back out to burn it due to the distance from town. Two weeks ago, an unknown stranger ignited the field for them as a favor. Senor R.H. will work this site for 1 year, then leave it fallow for 4, then return to clear it once more.
111 LIST OF REFERENCES Coomes, O.T., Grimard, F., Burt, G.J., 2000. Tropical F orests and S hifting C ultivation: S econdary F orest F allow D ynamics amo ng T raditional F armers of the Peruvian Amazon. Ecological Economics 32, 109 124. Coomes, O.T., and Burt, G.J. 2001. Peasant Charcoal Production in the Peruvian Amazon : Rainforest U se and Economic R eliance. Forest Ecology and Mana gement 140, 39 50 Denevan W.M. and Padoch, C., 1987. Swidden Fallow Agroforestry in the Peruvian Amazon. New York, USA: New York Bota nical Garden Press. Fearnside, P.M., 1987. Rethinking Continuous Cultivat ion in Amazonia. BioScience 37, 209 214. Gee, G.W. and Bauder, J.W., 1979. Particle Size Analysis by Hydrometer: A Simplified Method for Routine Textural Analysis and a Sensitivity T est of Measurement Parameters. S oil Sci. Soc. Am. J. 43, 1004 1007. Goodman, L.A., 1961. Snowball Sampling. Annals of Ma thematical Statistics 32, 148 170. Glaser, B., Lehmann, J., a nd Zech, W, 2002. Ameliorating P hysical and C hemical P roperties of H ighl y W eather ed S oils in the T ropics with C harcoal a R eview. Biol Fertil Soils 35, 219 230. Glaser, B. 2007 Prehistorically M odified S oils of C entral A mazonia: a M odel for S ustainable A griculture in the T wenty F irst C ent ury. Phil. Trans. R. Soc. B 362, 18 7 196. Grove J.H., C.S. Fowler, and M.E. Sumn er. 1982. Determination of the Charge Character of Selected Acid S oils. Soil Sci. Soc. Am. J. 46, 32 38. Hammes, K., Schmidt, M.W. I., et al, 2007. Comparison of Q uantification M ethods to M easure F ire D erived ( B lack/ E lemental) C arbon in S oils and S ediments U sing R eference M aterials from S oil, W ater, S ediment, and the A tmosphere. Global Biogeochemica l Cycles, 21. Hern ndez, T., 1946. Histor a de la Fundaci n del Pue blo de Tamshiyacu. Unpublished Manuscript in t he L ibrary of Centro de Estudios Teol gicos de la Amazonia (CETA), Iquitos, Peru. Hildebr and, P.E., 1986. The Sondeo: a T eam R apid S urvey A pproach. Hildebr and, P.E., Perspectives on F arming R esearch and E xtension. Lynne Rienner Publishers, Boulder, CO, 93 98.
112 Jordan, C. F. 1989. An Amazonian R ain F orest: the S tructure and F unction of a N utrient S tressed E cosystem and the I mpact of S lash and B urn A griculture. Paris, France: UNESCO. Juo, A.S.R. and Manu, A., 1996. Chemical D ynamics in S lash and B urn A gri culture. Agricultur e, Ecosystems & Environment 58, 49 60. Kalliola, R., Puhakka, M., and Danjo y, W., 1993. Amazonia Peruana: V egetacion H umeda T ropical en el Llano S ubandino. Jyvskylv, Fin.: Gummerus Printing. Kuh lbusch, T.A., 1995. Method for D etermin ing B lack C arbon in R esidues of V egetation F ires. Environmental S cience and T echnology 29, 2695 2702. Mehlich, A., 1978 New E xtractant for S oil T est E valuation of P hosphorus, P otassium, M agnesium, C alcium, S odium, M anganese, and Z inc. Com mun. Soil Sci. P lant Anal. 9(6), 477 492. Murphy, J., and Riley, J.P., 1962. A M odified S ingle S olution M ethod for the D etermination of P hosphate in N atural W aters Anal Chim Acta 27, 31 36. Nye, P.H., and Greenland, J.D., 1960. The S oil U nder S hifting C ultivation. Farn ham Royal, Eng.: Commonwealth Agricultural Bureaux. Palm, C.A., Swift M.J. and Woomer P.L, 1996. Soil B iological D ynamics in S lash and B urn A griculture. Agriculture, Ecosystems & Environment, 58, 61 74 Szott, L.T., Palm, C.A., and Buresh, R.J., 1999 Ecosystem F ertility and F allow F unction in the H umid and S ubhumid T ropics. Agroforestry Systems 47, 163 196. Tiessen, H., Cuevas, E., and Chacon, P., 1994. The R ole of S oil O rganic M atter in S ustaining S oil F ertility. Nature 371, 783 785. United Nations Food and Agriculture Program, 2007. Digital Soil Map of The World. FAO Land Resources Digital Media Series. Zarin, D.J., Duchesne, A.L., and Hiraoka, M., 1998. Shifting C ultivation on the T idal F loodplains of Amazonia: I mpacts on S oil N utrient S tatus. Ag roforestry Systems 41, 307 311.
113 BIOGRAPHICAL SKETCH Damion Graves was born in 1973 in Nassau County, New York. After graduating from high school in northern Virginia, he attended the Evergreen State College in Olympia, Washington, where he developed a n interest in sustainability and economic justice issues After earning a Bachelor of Arts degree in 1994, he then worked as a teacher, environmental educator, and occasional far mhand and landscaper for more than a decade. He first visited Peru in 2001, and this marked the beginning of a period of independent travel and language study. Upon moving to north Florida for graduate school in 2005, Damion deepened his interest in fire e cology across a range of landscapes and fire regimes. Condu cting research in the Peruvian A mazon provided an opportunity to combine a diverse set of experiences and passions, and to vi sit a country that began to feel like a second home.