Carbon and Nitrogen Dynamics in Fluvial Systems Across Biomes

University of New Hampshire University of New Hampshire Scholars' Repository

Doctoral Dissertations Student Scholarship

Spring 2020

Carbon and nitrogen dynamics in fluvial systems across biomes

Bianca Rodriguez-Cardona University of New Hampshire, Durham

Follow this and additional works at: https://scholars.unh.edu/dissertation

Recommended Citation Rodriguez-Cardona, Bianca, "Carbon and nitrogen dynamics in fluvial systems across biomes" (2020). Doctoral Dissertations. 2515. https://scholars.unh.edu/dissertation/2515

This Dissertation is brought to you for free and open access by the Student Scholarship at University of New Hampshire Scholars' Repository. It has been accepted for inclusion in Doctoral Dissertations by an authorized administrator of University of New Hampshire Scholars' Repository. For more information, please contact [email protected].

CARBON AND NITROGEN DYNAMICS IN FLUVIAL SYSTEMS ACROSS BIOMES

BIANCA M. RODRÍGUEZ-CARDONA BS, University of Puerto Rico, 2011 MS, University of New Hampshire, 2015

DISSERTATION

Submitted to the University of New Hampshire

in Partial Fulfillment of

the Requirements of the Degree of

Doctor of Philosophy

Earth and Environmental Sciences

May, 2020

This dissertation was examined and approved in partial fulfillment of the requirements for the degree of Doctor of Philosophy in Earth and Environmental Sciences by:

Dissertation Director, Dr. William H. McDowell Professor Natural Resources and the Environment Department of Natural Resources and the Environment

Dr. Adam S. Wymore Research Assistant Professor Department of Natural Resources and the Environment

Dr. Serita Frey Professor Department of Natural Resources and the Environment

Dr. Anne Lightbody Associate Professor Department of Earth Science

Dr. Robert G.M. Spencer Associate Professor Department of Earth Ocean and Atmospheric Science Florida State University

On [December 5, 2019]

Approval signature are on file with the University of New Hampshire Graduate School.

ii DEDICATION

Para mis amados Padres que me dijeron que soñara y gracias a sus consejos he logrado mas de lo pensando, y lo que falta por llegar…

iii ACKNOWLEDGEMENTS

I would first like to thank my advisor, Bill McDowell, for giving me the opportunity to join the lab group to pursue a masters with him and allowing me to stay and complete a doctoral degree. You have facilitated many opportunities to pursue science in multiple locations and collaborations with numerous people all the while allowing the flexibility to pursue the research topics of my interest. I would also like to thank Adam Wymore for working with me side by side since my masters and encouraging me to push forward at all times. I would also like to thank

Anne Lightbody, Serita Frey, and Rob Spencer for their help in developing my research, analyzing, and approach it from different perspectives.

A special thanks to Carla López-Lloreda, Albertyadir de Jesus Román, Valeria Bonano,

Katherine Pérez Rivera, Jody Potter and other members of the Water Quality Analysis Lab for their assistance in field work, analyzing samples and data. Special thanks to collaborators in

Siberia, Anatoly Prokushkin and Roman Kolosov for their help at the Evenkia Field Station and associated work at the site. Thank you to the working group, collaborators, and staff at National

Center for Ecological Analysis and Synthesis (NCEAS) for helping to analyze and manage all the long-term data. I would like to thank my Nesmith officemates and graduate students in

NREN and NRESS that have come and gone for their support, friendship, and brainstorming sessions throughout the whole process of graduate school. And a special thanks to my parents and family for their unconditional support in everything I have set my mind to do.

The research presented here was supported by the National Science Foundation Award

No. ICER 14-45246, Crossing the boundaries of Critical Zone science with a virtual institute

(SAVI); NSF DEB-1556603, Deciphering the role of dissolved organic nitrogen in stream nutrient cycling; Russian Foundation for Basic Research (RFBR) #14-05-00420, Small

iv catchments within the continuous permafrost zone of Central Siberia: the role of wildfire and forest succession in stream biogeochemistry; RFBR #18-05-60203, Landscape and hydrobiological controls on the transport of terrigenic carbon to the Arctic Ocean; NSF DEB-

1546686, Luquillo Long-Term Ecological Research Program; NSF EAR-1331841, Luquillo

Critical Zone Observatory; and NSF DMR-1644779 and the State of Florida. Part of the work here was conducted as a part of the Stream Elemental Cycling Synthesis Group funded by NSF

DEB #1545288, through the Long-Term Ecological Research Network Communications Office

(LNCO), NCEAS, and University of California-Santa Barbara. Data was also obtained from the

National Atmospheric Deposition Program (NADP) website (National Atmospheric Deposition

Program (NRSP-3).

v TABLE OF CONTENTS

Acknowledgements ...... iv List of tables ...... vii List of figures ...... x Abstract ...... xiv

CHAPTER PAGE Introduction ...... 1 1. Long-term trends in dissolved organic matter from fluvial systems across biomes ...... 5 Abstract ...... 5 Introduction ...... 7 Methods ...... 10 Results ...... 13 Discussion ...... 17 Tables ...... 25 Figures ...... 27 Supplementary information: ...... 37 2. Wildfires lead to decreased carbon and increased nitrogen concentrations in upland arctic streams ...... 52 Abstract ...... 52 Introduction ...... 53 Methods ...... 55 Results and discussion ...... 60 Tables ...... 67 Figures ...... 72 Supplementary information ...... 76 3. Nitrate uptake enhanced by the availability of dissolved organic matter in tropical montane streams ...... 78 Abstract ...... 78 Introduction ...... 80 Methods ...... 81 Results ...... 85 Discussion ...... 86 Tables ...... 92 Figures ...... 94 Supplementary information ...... 98 Conclusion ...... 101 References ...... 104

vi LIST OF TABLES

Chapter One: Long-term trends in dissolved organic matter from fluvial systems across biomes

Table 1.1 Sites from which DOC and DON time series were obtained with the number of individual streams used from each site. DOC and DON median concentrations with minimum and maximum values for each site. and - no data available. Table 1.2 Counts (#) and percentages (%) of statistically significant positive and negative and no trend seasonal Sen slopes for DOC, DON, and DOC:DON molar ratios across all sites and total of all sites, with the total across all sites given at the bottom. There is no DON data available for KNZ. Table S1.1. Individual stream (with abbreviated stream name) characteristics such as latitude (Lat) and longitude (Long), mean annual temperature (MAT), mean annual precipitation (MAP), watershed elevation (Elev) and area along with mean stream pH, temperature (Temp), and Conductivity (Cond). Reference describe sites that are specifically labeled reference watersheds, or no management or manipulations have been documented for these watersheds. Forested in Finland sites are less impacted than those described as suburban. – denote data is not available for that stream.

Table S1.2. Mean solute concentrations for every stream. – denote data is not available for that stream.

Table S1.4 Sampling rates and analytical methods for analysis of dissolved organic carbon (DOC) or total organic carbon (TOC), total dissolved nitrogen (TDN), total nitrogen (TN) or - + dissolved organic nitrogen (DON), nitrate (NO3 ), and ammonium (NH4 ) concentrations for each site. – corresponds to no analysis performed.

Table S1.6. Tau values for DOC:DON molar ratios calculated using Seasonal Mann–Kendall. Red denotes statistically significant increasing trend, blue denotes statistically significant decreasing trends, and gray are no significant trends for the given period of time. The total column are Tau values for the total record of available data (see Table S1.3 for details on record length) for each stream. – denote data is not available for that stream.

vii Chapter Two: Wildfires lead to decreased carbon and increased nitrogen concentrations in upland arctic streams

Table 2.1 Watersheds sampled in 2011, 2013, 2016, and 2018 with their respective burn year and watershed size. N11 is currently the only site with two recorded burns, in 1899 and 2013.

+ 3- Table 2.2 Mean NH4 and PO4 concentrations across the burn gradient for streams sampled June 2013, 2016, and 2018. 10 represent sites burned between 3 and 7 yrs, 25 watersheds burned between 18 and 25 yrs, 50 watersheds between 51 and 57 yrs, 60 watersheds burned between 66 and 71 yrs, and >100 watersheds burned up to 122 years ago. Parentheses represent standard error and superscript represent statistical difference between groups (α=0.05). Respective n- + 3- values: NH4 June:16,16,17,4,10, July:5,5,3,10; PO4 June:15,15,16,4,10, July:4,5,3,7.

SUVA254 June:13,10,13,4,8, July:1,1,1,3; FI June:13,8,13,4,7, July:1,1,1,3; HIX June: 7,6,7,2,3,

July: 1,1,1,3; S275-295 June: 11,9,12,4,7, July: 1,1,1,3; S350-400 June:11,9,12,4,7, July: 1,1,1,3; SR June:11,9,12,4,7, July: 1,1,1,3. Statistical analyses were excluded from July data due to low n. – represents no data available.

Table 2.4 A four component PARAFAC model for samples from 2016 only. Table 2.5 Summary of the final backward step wise regression model for DOC, DON, and DIN concentrations, DOC:DIN molar ratios, relative abundance of aliphatics (with N-containing aliphatics), and slope ratio (SR) as the dependent variables. The initial model for all included watershed area (km2), elevation (m), watershed slope, percent of watershed facing north, percent of watershed facing south, and years since the last fire (YSF) as independent variables. Bold values highlight statistically significant relationships (α < 0.05). * denote statistically significant models and – represents variables that were dropped from the final model.

Table S2.1 Details on individual nutrient pulse additions per site from 2016 through 2018 which includes additions where NH4 was added with PO4, NH4 only, and NO3 only but here we only present uptake metrics for NH4 and NO3 only. The experimental reach length, average stream width and depth, discharge (Q), added mass of sodium chloride (NaCl), sodium nitrate (NaNO3), and ammonium sulfate ((NH4)2SO4), and uptake metrics for each addition as uptake length (Sw), uptake velocity (Vf), areal uptake (U). * represent undetectable uptake, # represent dates where stream depth measurements could not be taken, and – represent that the given solute was not added for that pulse addition.

viii Chapter Three: Nitrate uptake enhanced by availability of dissolved organic matter in tropical montane streams

Table 3.1. Watershed area and mean DOC, DON, and nutrient concentrations for Quebrada Prieta (QPB) and Quebrada Toronja (QT) with minimum and maximum values in parenthesis.

Table 3.2. Uptake metrics for all NO3 only additions, NO3 added with acetate and NO3 added with urea at QPB and QT along with discharge from each date. Values with * are uptake metrics - determined as estimates of the minimum values could have been detected in the NO3 -only - - experiment. – denotes undetectable uptake for additions of NO3 + acetate or NO3 + urea. Table S3.1. Discharge (Q), experimental stream reach length, mean stream width and depth and the mass (g) and percent recovered (% Rec.) of added solutes for each individual addition in Quebrada Toronja (QT) and Quebrada Prieta (QPB).

ix LIST OF FIGURES

Chapter One: Long-term trends in dissolved organic matter from fluvial systems across biomes

Figure 1.1 Location of all sites across the Northern Hemisphere. See Table 1 for complete site names.

Figure 1.2 Time series of monthly SO4 (Yellow) and NO3 (blue) atmospheric deposition from National Atmospheric Deposition Program collection sites at (a) Bonanza Creek, (b) Hubbard Brook Forest, (c) H.J. Andrews Forest, (d) Konza Prairie, (e) and Luquillo Forest. The dark grey line corresponds to the length of records for stream water DOC and DON concentrations analyzed for trends in this study. Figure 1.3 Seasonal Sen slopes for DOC for each individual stream in (a) Bonanza Creek, (b) Lamprey River Basin, (c) Konza prairie, (d) Hubbard Brook Forest, (e) H.J. Andrews Forest, (f) Luquillo Experimental Forest, and (g) Finland. Stars denote statistically significant Sen slopes. Blue bars represent sites not affected by acid deposition and grey bars are sites historically affected by acid deposition. Bonanza, Lamprey, Hubbard Brook, and Andrews streams are ordered by watershed size; Luquillo streams are ordered by size within their respective major watersheds; Finland streams are ordered by latitude from North to South. Figure 1.4 Seasonal Sen slopes for DON for each individual stream in (a) Bonanza Creek, (b) Lamprey River Basin, (c) Hubbard Brook Forest, (d) H.J. Andrews Forest, (e) Luquillo Experimental Forest, and (f) Finland. Stars denote statistically significant Sen slopes. Note no Konza slopes or Finish stream (YLIJOKI 1) as there is no DON data available for these sites. Blue bars represent sites not affected by acid deposition and grey bars are sites historically affected by acid deposition. Bonanza, Lamprey, Hubbard Brook, and Andrews streams are ordered by watershed size; Luquillo streams are ordered by size within their respective major watersheds; Finland streams are ordered by latitude from North to South. Figure 1.5 Seasonal Sen slopes for DOC:DON molar ratios for each individual stream in (a) Bonanza Creek, (b) Lamprey River Basin, (c) Hubbard Brook Forest, (d) H.J. Andrews Forest, (e) Luquillo Experimental Forest, and (f) Finland. Stars denote statistically significant Sen slopes. Note no Konza slopes or Finish stream (YLIJOKI 1) as there is no DON data available for these sites. Blue bars represent sites not affected by acid deposition and grey bars are sites historically affected by acid deposition. Bonanza, Lamprey, Hubbard Brook, and Andrews streams are ordered by watershed size; Luquillo streams are ordered by size within their respective major watersheds; Finland streams are ordered by latitude from North to South. Figure 1.6. DOM Sen slopes grouped by history of acid deposition where sites affected by acid deposition are in grey (FIN,HBF,LMP,KNZ) and sites not affects by acid deposition (BNZ,AND,LUQ) in blue. Letter denote statistically significant differences. Figure 1.7. Correlation plot matrix of individual stream (A) DOC seasonal Sen slopes from, (B) DON seasonal Sen Slopes, and (C) DOC:DON molar ratios seasonal Sen slopes for each site related to the median values of stream DOC, DON, NO3, NH4, PO4, Si, Na, K, Ca, Mg, SO4, and

x Cl concentrations, stream conductance, temperature, watershed area, and stream elevation. Width of the ellipsis describe variability in the relationships where wider ellipses pertain to more variability. The strength of the correlation is described by colors where colder colors represent positive strong correlations and warmer colors strong negative correlations. Blank boxes denote no data available or gaps within that predictor variables. KNZ was excluded here due to low n in sites. Figure 1.8. Correlation plot matrix of individual streams in the Lamprey River Basin DOC, DON and DOC:DON molar ratio seasonal Sen slopes from correlated to the population, housing per area, houses on septic tanks, percent of impervious surface, percent developed, percent agriculture cover, percent forest cover, percent shrub cover, percent wetland cover, and percent bare land at the sub-watershed level. Width of the ellipsis describe variability in the relationships where wider ellipses pertain to more variability. The strength of the correlation is described by colors where colder colors represent positive strong correlations and warmer colors strong negative correlations. Figure 1.9. Sen slopes for DOC, DON, and DOC:DON molar ratios grouped by forested sites (green) and disturbed sites (purple) across all streams regardless of biome; DOC Sen slopes p=0.23, DON Sen slopes p=0.33, and DOC:DON Sen slope p=0.71. Figure 1.10. DOC and DON tau values from the seasonal Mann-Kendall test where each point is an individual stream color coded by their respective site. Values that lie on the 0-lines are values that are not statistically significant for DOC or DON; points within quadrants are statistically significant for both DOC and DON. Roman numerals denote the quadrant number. KNZ and one Finish stream (YLIJOKI 1) are not included here because they did not have DON data. ). The different quadrants describe trends in DOM where QI describes increases in DOC and DON, QII are streams with declines in DOC and increases in DOC, QIII shows declined in DOC and DON, and QIV described increases in DOC with declines in DON.

Chapter Two: Wildfires lead to decreased carbon and increased nitrogen concentrations in upland arctic streams

Figure 2.1 Map of sub-watersheds sampled in the Central Siberian Plateau. Each watershed with their respective fire history between 3 and >100 years since the last fire (YSF), draining into the Kochechum River and Nizhnyaya Tunguska River which are part of the greater Yenisei River Basin. Here, 10 YSF represents watersheds burned between 3 and 7 yrs, 25 YSF watersheds burned between 18 and 25 yrs, 50 YSF watersheds between 51 and 57 yrs, 60 YSF watersheds burned between 66 and 71yrs, and >100 YSF watersheds burned up to 122 years ago. See supplementary table 1 for further details on individual watersheds. Figure 2.2 DOM, nutrient concentrations, and stoichiometric ratios across the burn gradient. - Boxplot panels represent (A) DOC, (B) DON, (C) NO3 concentrations, and (D) molar DOC:DIN ratios, across 17 streams sampled 2011, 2013, and 2016 - 2017 during June and July across the burn gradient. The paired boxplots correspond to June (dark shade) and July (lighter shade). Boxes represent interquartile range with the median value as the bold line, whiskers represent 1.5 interquartile range, and points are possible outliers. Letters denote significant

xi differences (α=0.05) where uppercase correspond to June and lowercase to July. Note that 60 YSF in July was excluded from statistical analysis due to low n. Statistics for DOC p=1.8x10-8, F=14.8, df=4 June and p=8.3x10-4, F=8.56, df=4 July); DON p=0.0002, F=6.53, df=4 June and p=0.005, F=5.81, df=4 July; DIN p=0.008 June, p=0.04 July; DOC:DIN p=0.0008 June and p=0.06 July. Respective n-values across the burn gradient for June: 16, 16,19, 4, 10; July: 5, 5, 3, 1, 10. Figure 2.3 DOM composition of streams across the burn gradient. Boxplots represent (A) slope ratios (SR) from June 2013 and 2016, relative abundance of (B) aliphatic compounds including N-containing aliphatics, (C) polyphenolic compounds, and (D) condensed aromatic compounds. Streams burned 60 years ago for FT-ICR MS were excluded from statistical analyses due to low n. Boxes represent interquartile range with the median value as the bold line, whiskers represent 1.5 interquartile range, and points are possible outliers. Uppercase letters denote significant differences (α=0.05). Only data from June 2016 for FT-ICR MS compound groups and June 2013 and 2016 for SR. Respective n-values across the burn gradient SR: 11, 9, 12, 4, 7; FT ICR: 11, 8, 9, 2, 6.

+ - Figure 2.4 Drivers of DIN uptake velocity. Uptake velocity (Vf) of NH4 (circles) and NO3 (triangles) grouped as DIN Vf from streams across the burn gradient (10 YSF red, 25 YSF orange, 50 YSF blue, 60 YSF yellow, and >100 YSF green) related to (A) ambient DIN concentration, (B) molar DOC:DIN ratios, and (C) relative abundance of polyphenolics. Vf values here are from 2016 through 2018. The n-values for NH4 Vf is 18 and NO3 Vf is 10.

Chapter Three: Nitrate uptake enhanced by availability of dissolved organic matter in tropical montane streams

Figure 3.1. Map of headwaters of the Río Espiritu Santo Watershed (RES) with outlined watersheds of study sites Qda. Toronja (QT) in purple and Qda. Prieta-B (QPB) in grey near the El Verde Field Station (EVFS) in the Luquillo Experimental Forest in northeast Puerto Rico.

- Figure 3.2. Combined mean NO3 uptake metrics from QPB and QT for (A) uptake length (Sw), - - (B) uptake velocity (Vf), and (C) and areal uptake (U) during NO3 only additions (blue), NO3 - added with acetate (orange), and NO3 added with urea (green). Bars represent standard error and letters denote statistically significant differences. Figure 3.3. Combined mean DOM uptake metrics from QPB and QT for acetate (orange) during - - NO3 and acetate additions and urea (calculated as bulk DON) (green) during NO3 and urea additions for (A) uptake length (Sw), (B) uptake velocity (Vf), and (C) areal uptake (U). Bars represent standard error.

- Figure 4. Relationships between ambient DOC and DON concentrations during the NO3 -only - - additions. (A) DOC and NO3 concentrations on January 4 in QT, (B) DON and NO3 - concentrations on January 4 in QT, (C) DOC and NO3 concentrations on January 13 in QT, (D) - - DON and NO3 concentrations on January 13 in QT, (E) DOC and NO3 concentrations on - January 12 in QPB, and (F) DON and NO3 on January 12 in QPB. Purple points represent QT and grey points are from QPB. Individual points correspond to a sample collected throughout the

xii BTC and background samples are not shown. See Supplementary Figure1 for figures from all - seven NO3 -only additions.

- Figure S3.1. Ambient concentrations of DOC and DON with manipulated NO3 concentrations during the NO3 only additions at QT in (A) Jan. 4, (B) Jan. 5, (C) Jan. 13 and (D) Jan. 15. Each individual point corresponds to a sample collected throughout the BTC and background samples are not shown.

- Figure S3.2. Ambient concentrations of DOC and DON with manipulated NO3 concentrations during the NO3 only additions at QPB in (A) Jan. 7, (B) Jan. 12, and (C) Jan. 14. Each individual point corresponds to a sample collected throughout the BTC and background samples are not shown.

xiii ABSTRACT

CARBON AND NITROGEN DYNAMICS IN FLUVIAL SYSTEMS ACROSS BIOMES

Bianca M. Rodríguez-Cardona

University of New Hampshire

May, 2020

Carbon (C) and nitrogen (N) cycles are tightly coupled in ecosystems as many N transformations rely on carbon as the energy source. In aquatic ecosystems this coupling between

C and N has been studied via assessment of dissolve organic carbon (DOC) and inorganic-N concentration or whole stream manipulations of C and/or N to determine the influence of C on N processing. However, the majority of these studies have been performed in temperate systems and whether these relationships hold across other biomes, like artic and tropics, is unclear. In addition, how the more N-rich organic matter or dissolved organic nitrogen (DON) interacts with inorganic-N has also received little attention. Overall, the research presented here aims to better understand the how C and N relate in streams and what is the role of C in N transformations across different biomes.

In the chapters presented here, I use a variety of approaches to better understand C and N coupling in streams such as assessing long-term trends in DOC and DON concentrations across biomes to whole stream manipulations in boreal forests of the Central Siberian Plateau (CSP) and the Luquillo Mountains of Puerto Rico. Time series analyses on dissolve organic matter

(DOM, composed of DOC and DON) have demonstrated that DOC and DON concentrations are

xiv changing (positive and negative trends) in streams regardless of biome. The potential drivers of changes in DOM are watershed specific suggesting that DOM is controlled by local and regional-scale factors. Experimental work has focused on the coupled biogeochemistry of DOM and nitrate, including how variation in DOC concentration and DOM composition affect nitrate

and ammonium uptake. I performed whole stream manipulations of NO3 and NH4 in Siberia to study watershed resiliency and recovery from major disturbances including permafrost thaw and wildfires. DOC concentrations returned to pre-fire conditions after 50 years while recovery for

NO3 was 10 years. Uptake rates of NO3 and NH4 declined with increasing NO3 and NH4 concentrations, DOC:DIN molar ratios, and relative abundance polyphenols; as the frequency of fire or the rate of permafrost thaw continue to increase in high latitude regions, streams could become major exporters of inorganic nutrients to receiving bodies including the major Arctic rivers and the Arctic Ocean. I performed similar stream manipulations but of NO3 and C sources

in Luquillo which revealed that the demand for NO3 is energy-limited and is enhanced with the availability of DOM. In addition, these streams are carbon limited and have high demand of added C sources. Collectively, these dissertation chapters demonstrate that across biomes DOM availability can be an important factor in nitrogen processing and the quantity and composition of DOM is tightly linked to the landscape.

xv INTRODUCTION

Carbon (C) plays a major role in nitrogen (N) processing across ecosystems as major transformations in the N-cycle require C as an energy source. Organic C and N make up the dissolved organic matter (DOM) pool and found in the forms of dissolved organic carbon (DOC) and dissolved organic nitrogen (DON), and the bulk DOM pool is a major energy source for biogeochemical processes (Meyer & Wallace, 1998). Various studies have reported that DOC concentrations have been increasing in aquatic systems due to declines in acid deposition that are now making DOC more soluble and mobile from soils to bodies of water (Evans et al., 2005;

Evans et al., 2006; Hruška et al., 2009; Monteith et al., 2007; Sawicka et al., 2017; Worrall et al.,

2004). But these studies mostly focused on northern Europe and it is unclear if these increasing trends in DOC are a global or regional patter. In addition, very little attention has been given to trends in DON , which is another major component of DOM and is tightly related to inorganic nitrogen (Wymore et al., 2015). Changes to both DOC and DON concentrations can have major implications to the C:N stoichiometry of the bulk DOM pool and influence nutrient processing in streams (Rodríguez-Cardona et al., 2016; Wymore et al., 2016).

The coupling of DOM and inorganic nitrogen has been a central topic in various studies that have demonstrated that DOC concentrations drive N transformations, especially nitrate

(NO3) removal (Bernhardt & McDowell, 2008; Helton et al., 2015; Rodríguez-Cardona et al.,

2016; Taylor & TownSend, 2010; Wymore et al., 2016). Numerous of studies have also focused on N cycling instreams, most prominently through the Lotic Intersite Nitrogen eXperiments

(LINX) I and II, which consisted of whole stream manipulations of labeled inorganic N in

various streams across the contiguous US to understand NO3 and ammonium (NH4) uptake dynamics (Hall et al., 2009; Mulholland et al., 2009, 2008; Peterson et al., 2001). However,

1 interactions between C and N and our understanding of N cycling have mostly been conducted in temperate and North American biomes and thus our global understanding of how energy and nutrient cycling are linked within stream ecosystems is limited.

Globally, tropical rainforest and permafrost dominated landscapes store a disproportionate amount of the global terrestrial carbon (Bunker et al., 2005; Drake et al., 2019;

Schuur et al., 2008; Zimov et al., 2006); yet, in-stream the C dynamics vary drastically. For example, DOC is maintained at relatively low concentrations in many tropical montane streams like Puerto Rico (~1 mg C/L) driven by tight drainage pathways through low carbon soils

(McDowell & Asbury, 1994; Silver et al., 1994) while in central Siberia, DOC reaches significantly higher concentrations (~25-35 mg C/L) linked to permafrost that provides an ephemeral source of OM as many small streams dry and or freeze after the spring freshet (Vonk et al., 2015). While the magnitude of DOC reaching streams in these contrasting systems varies, so does DOM composition. In artic streams and rivers DOM is most bioavailable during the spring freshet (i.e. fresh litter leachate) (Abbott et al., 2014; Spencer et al., 2008) while streams in Luquillo are characterized by aromatic and lignin rich DOM (Hernes et al., 2017; Miller et al.,

2016) but little is known about the role of DOM in N.

In order to understand nutrient dynamics in streams whole stream manipulations of a given solute facilitate the quantification of its demand by the in-stream biota. This approach also normalizes uptake metrics making it easier to compare across streams and studies. Nutrient pulse additions consist of adding a conservative tracer (e.g. NaCl) along with the solute of interest

(DOC, NO3, or NH4) at the top of an experimental stream reach and samples are collected at a fixed point at the bottom of the experimental reach. Samples are collected as the solution moves through a fixed point at the bottom of the reach and the stream returns to ambient conditions

2 creating a breakthrough curve (BTC) which described the change in concentration over time

(Tank et al., 2008). The different uptake metrics describe different components of nutrient spiraling such as uptake length (Sw) which is the distance a molecule travels before being taken up by the biota, uptake velocity (Vf) that describes the efficiency of the microbial community to take up nutrients, and areal uptake (U) which is a gross nutrient uptake rate (Newbold et al.,

1981; Workshops, 1990). These uptake metrics can be compared to ambient stream concentrations or DOM composition to gain insights on controls of N processing rates across streams in different biomes.

Understanding how DOM and nitrogen are linked in streams across globally distinct biomes has important implications for our understanding of how energy and nutrients are retained and exported from watersheds. Here I examine how sources, quantity, and composition of organic matter influence nitrogen processing in streams across biomes. In this dissertation I address three specific questions to better understand the trends and role of DOM on nitrogen processing.

1. How are DOC and DON concentrations changing in fluvial systems across different

biomes of the Northern Hemisphere?

2. What is the resiliency of arctic streams after wildfires? And how do wildfires influence

nitrogen processing in these streams?

- 3. How does the availability of DOC influence NO3 uptake in tropical montane streams?

In chapter one I analyze long-term trends in DOC and DON concentrations in watersheds ranging from permafrost regions like Alaska and boreal forests in Finland to tropical forest of

Luquillo Mountains. Time series analyses from 74 different streams were used to obtain seasonal

3 Sen slopes for DOC and DON to assess direction and magnitude of these trends and relate them to ambient stream concentrations, stream physical chemical parameters (pH, temperature, conductivity), and watershed characteristics (area and elevation) to determine potential drivers in changes of DOM. I also used to evaluate how the stoichiometry of organic matter pool is changing given the different patterns in DOC and DON trends. In chapter two, I determined the resiliency of watersheds post wildfires using a space for time approach in the Central Siberian

Plateau. I also determined uptake rates for NH4 and nitrate NO3 by whole stream manipulations in a subset of these watersheds to study how watersheds process inorganic N after wildfires and how these relate to DOM composition and quantity. In chapter 3, I assessed the relationship between DOC on NO3 processing in streams of the Luquillo Experimental Forest. Here, I

performed NO3 additions with and without a C source to determine the effects of the availability of carbon on NO3 uptake. Collectively, my dissertation will help us understand C and N dynamics and how DOM availability influences nitrogen processing across biomes.

Each chapter has been formatted as individual research articles for publication in peer- reviewed journals and Chapter 2 is currently in-review in Scientific Reports and Chapter 3 is currently in-review in Freshwater Science.

4 CHAPTER ONE

LONG-TERM TRENDS IN DISSOLVED ORGANIC MATTER FROM FLUVIAL SYSTEMS ACROSS BIOMES

Abstract

Dissolved organic carbon (DOC) concentrations have increased in the past 50 years in many streams and rivers draining temperate watersheds in Europe and North America. Less is known about changes in DOC in other biomes, or if changes in dissolved organic nitrogen

(DON) track those of DOC. Here we present long-term DOC and DON data from multiple streams (n=74) across biomes ranging from the tropics to northern boreal forests and the Arctic.

For each individual stream within a given site, we obtained seasonal Sen slopes for time series of

DOC, DON, and DOC:DON molar ratios and related these slopes to stream physico-chemical parameters, watershed descriptors, and stream chemistry to determine global and regional drivers. Sen slopes varied in direction and magnitude across sites and streams, from -0.23 to 0.22 mg C/L per year for DOC, -0.02 to 0.01 mg N/L per year for DON, and -0.87 to 1.03 per year for

DOC:DON molar ratios and the magnitude of change in DOC concentrations was greater than that of DON across all sites. There were no consistent global or regional trends in DOC and

DON concentrations, but some sites displayed consistent positive or negative responses across a set of streams. Sen slopes for DOC:DON were mostly positive with the exception of sites within the suburban sites New Hampshire, which had mostly declining Sen slopes. There were no global predictor variables, but regional relationships between Sen slopes and water temperature, concentrations of major ions, and watershed area were observed. Decoupled trends in DOC and

DON occurred in multiple sites, with associated changes in dissolved organic matter stoichiometry. Because increasing or decreasing trends in dissolved organic matter (DOM) can

5 have major implications for stream metabolism, food webs, and drinking water quality, understanding the both regional and global trends in DOC and DON concentrations is an important for realistic models and watershed management protocols.

6 Introduction

Aquatic ecosystems reflect their surrounding landscape where vegetation, soils, and geology can all influence surface water chemistry (Hynes, 1975). Soils can regulate the levels of

DOM in aquatic systems, as they regulate both the production of DOM and the losses of DOM along hydrologic flow paths (McDowell & Likens, 1988). Given the tight connection between terrestrial and aquatic ecosystems, changes in concentrations of DOM in aquatic ecosystems can be due to either changes in net production of DOM within the aquatic system or increased leaching losses from the terrestrial to aquatic ecosystem (Roulet & Moore, 2006). These influences can also be tied to external drivers such as temperature and rainfall, and atmospheric deposition that impacts the entire landscape (Monteith et al., 2007).

Numerous studies have documented increases in dissolved organic carbon (DOC) concentrations associated with recovery from acidic deposition after the implementation of the

Clean Air Act in the USA and similar legislations in Europe. DOC concentrations have increased between 50-91% in streams and lakes of northern and central Europe, United Kingdom (UK), and eastern North America (Driscoll et al., 2003; Evans et al., 2005; Gavin et al., 2018;

Lawrence et al., 2011; Monteith et al., 2007; Worrall et al., 2004). As acidic deposition decreased due to reduced atmospheric emissions, the ionic strength of soil water and streams decreased with concurrent increases in solubility and inputs of DOC to adjacent bodies of water

(Borken et al., 2011; De Wit et al., 2007; Evans et al., 2005; Hruška et al., 2009). In addition to declines in atmospheric deposition, a suite of different hypotheses have been put forward to explain the increasing trends in DOC concentrations such as increasing precipitation and runoff

(De Wit et al., 2007; Strååt et al., 2018; Worrall et al., 2004), rising CO2 and increased primary productivity (Freeman et al., 2004), enhanced microbial OM decomposition due to increased

7 temperatures (De Wit et al., 2007; Finlay et al., 2006; Worrall et al., 2004), permafrost thaw for higher latitude ecosystems (Frey & McClelland, 2009; Frey et al., 2007; Frey & Smith, 2005;

Larouche et al., 2015), and changes in land-use (Kritzberg, 2017). Many of the proposed hypotheses can better explain short-term or seasonal patterns in DOC increases (i.e. temperature, hydrology, precipitation).

A few studies in various catchments and countries have described declines in DOC concentrations over time as well as no changes. In several northern latitude/boreal ecosystems, declining DOC has been correlated to acid deposition, decreasing soil organic matter solubility and ultimately decreasing DOC transport to streams (Clair et al., 2008). Other studies have found declines in streamflow inputs from upstream and acidification of the lakes (Schindler et al.,

1997). DOC concentrations have also declined in permafrost-dominated regions, especially in the summer and autumn when the active layer (upper most part of permafrost that thaws and freezes annually) is most in transition and there is greater adsorption to the mineral layer and infiltration of DOM deeper into permafrost soil (Striegl et al., 2005). Soil pore waters in Sweden

have shown decreasing trends in DOC and SO4 concentrations that could be attributed to increases in the soil aluminum pool (Löfgren et al., 2010) even though acid deposition has decreased. In several of the previously mentioned studies, a smaller fraction of the systems studied demonstrated no temporal changes in DOC concentrations (Clair et al., 2008; Monteith et al., 2007; Worrall et al., 2004). In Finland, long-term records (since 1975) show DOC concentrations have been unchanged (Räike et al. 2012). While these different trends (increasing, decreasing, no change) suggest that the controls on DOC concentrations vary regionally, a broad multi-biome investigation on global DOC trends is lacking. In addition, temporal trends in dissolved organic nitrogen (DON), a small but biologically important portion of the DOM pool

8 (Brookshire et al., 2005; Wymore et al., 2015), are very limited. Changes in DOM concentrations can have effects on ecosystem services, biological processes, and downstream transport of nutrients and organic matter. Increasing DOC and DON concentrations are an issue for drinking water treatment processes in the creation of disinfection by products (Chow et al., 2013; Lee et al., 2006; Leenheer & Croue, 2003) and influence primary productivity by altering light attenuation and UV entering aquatic systems (Cory et al., 2014; Schindler et al., 1997). Changes to DOM concentrations will also influence ecosystem biogeochemistry as inorganic nitrogen can be influenced by DOC (Bernhardt & McDowell, 2008; Helton et al., 2015; Rodríguez-Cardona et al., 2016; Taylor & Townsend, 2010) and DON concentrations (Brookshire et al., 2005; Wymore et al., 2015) where promoting heterotrophic uptake of N can lead to greater release of gaseous

forms of N2O and could become a bigger concern than CO2 emissions from streams. Therefore, understanding the magnitude of change in DOM across streams will better elucidate the ultimate fate of DOM and nutrients in aquatic ecosystems.

In this study we determined trends in concentrations of DOC, DON and DOM stoichiometry in streams across biomes of the Northern Hemisphere. We also assessed the relationships of potential predictor variables to changes in concentrations of DOC, DON and

DOC:DON molar ratios. Given the rich literature on increasing trends in DOC due to decreases in atmospheric deposition, we expect to see predominantly increases in DOC concentrations in those sites historically affected by acidic deposition and no changes in those sites that were unaffected. We also expect that changes in DOM will be closely related to solutes altered by

atmospheric deposition (e.g. pH, SO4, NO3). Lastly, we hypothesize that changes in concentrations of DON should track those of the broader OM pool as measured by DOC and reflect a stoichiometrically neutral organic matter pool.

9 Methods

Site selection and data set compilation

Long-term data on DOC and DON concentrations were compiled for 74 individual streams across seven different sites (Fig 1.1) in the Northern Hemisphere spanning a large gradient in biome, climate, and DOM quantity (Table 1.1). Of the seven sites, five are members of the U.S. Long-Term Ecological Research (LTER) program and data were obtained from each

LTER site’s publicly available data repository. Data sets from sites in Finland were contributed by the Finnish Environmental Institute (SKYE); and the data set from the Lamprey River Basin in New Hampshire, USA was provided by the Water Quality Analysis Laboratory at the

University of New Hampshire (Table 1.1). See Table S1.1 and S1.2 for more details for each site, individual streams, and published papers that generated parts of the data set.

We only included streams with both DOC and DON data that had data sets of at least eight years duration with continuous weekly or monthly data (Table S1.3), with the exception of

Konza Prairie (KNZ) that has no DON data. Most of the streams analyzed here are considered reference since they are not directly influenced by anthropogenic activity with the exception of streams in LMP and Finland that are all suburban sites and W1 in Hubbard Brook Experimental

Forest which was amended with wollastonite (See Table S1.1).

For consistency across sites we set minimum detection limits (MDL) for each solute:

- - DOC (0.1 mg C/L), TDN (0.05 mg N/L), DON (0.01 mg N/L), NO3 (0.005 mg NO3 -N/L), and

+ + NH4 (0.004 mg NH4 -N/L). In addition, we only used DON values that were 5% or more of the

TDN pool. For data points that were below the MDL for any solute, values were replaced with half the MDL. To estimate DOC and TDN from the Finnish data set we multiplied TOC and TN by 0.95 (Kortelainen et al., 2006; Mattsson et al., 2005). These calculated TDN values for the

10 - + Finnish data were then used to determine DON for these sites as: DON = TDN – (NO3 + NH4 ).

For more details on analytical methods see Table S1.4.

Sites affected by acid deposition

Atmospheric deposition data was obtained from the National Atmospheric Deposition

Program (NADP) website (National Atmospheric Deposition Program (NRSP-3) for LTER sites included here that have nearby NADP sampling locations.

Lamprey River Basin Land-Use

The Lamprey River Basin (LMP) was used as a case study to determine if trends in DOM were related to land-use classifications. Land use characteristics for sub-watersheds of the LMP were obtained from NOAA’s Coastal Change Analysis Program (2006).

Time series and trend analyses

Time series were created from mean monthly DOC and DON values for each individual stream. Time series were restricted to streams that had consecutive years of data available where no more than three months’ worth of data were missing. In the case of gaps in the time series, gap filling was achieved by predictive mean matching using the mice package (van Buuren &

Groothuis-Oudshoorn, 2011). The exception to this was Bonanza Creek (BNZ) as data were only available during the freshet period and late summer so we restricted data from May to August since those were the months that consistently had data across years. No gap filling was performed for BNZ as the yearly time series were short. Time series were created using the stats package (R-Core-Team, 2016) and a frequency of 12 was set, with the exception of Bonanza

11 Creek where the frequency was set to 4, where frequency equals the number of months in each annual time series. DOC and DON time series objects for each individual stream were then used to calculate trends in DOM using the seasonal Mann-Kendall tau (Hirsch et al., 1982) and p- values using the Kendall package (McLeod, 2011). Statistically significant tau values describe a strong monotonic trend. The magnitude of that trend was determined with the seasonal Sen slope

(Hirsch et al., 1982) obtained from the trend package (Pohlert, 2018) for each stream for the length of time available for DOC, DON, and DOC:DON ratios (8-45 years).

Statistical Analyses

Tau values and p-values were compared simultaneously in increasing five-year increments starting from 2015 until the end of the time-series across all streams to test whether the length of time analyzed would change the direction of any DOM trends. Sen slopes were related to ambient stream concentrations of nutrients and DOM, watershed characteristics

(watershed area, elevation) and stream physico-chemical parameters (temperature and conductivity reported as specific conductance or conductivity) where available using a linear correlation matrix with the corrplot package (Wei & Simko, 2017). Tau values were used to evaluate the direction of change in the stoichiometry of the DOM pool. Non statistically significant tau values for DOC or DON were replaced with zeros. Differences between acid deposition-affected and unaffected site and between forested and disturbed sites were determined by a Kruskal-Wallis rank sum test using the stats package (R-Core-Team, 2016) due to uneven sample size between groups. For all statistical analyses α=0.05 and analyses were conducted in

RStudio (version 1.2.1335, RStudio, Inc. Team, Boston, MA 2016).

12 Results

DOM and DOC:DON Sen slopes

Sen slopes showed no consistent increasing or decreasing trends for DOC (Fig 1.3) or

DON (Fig 1.4), but predominantly increasing trends for DOC:DON molar ratios (Fig 1.5).

Collectively across all streams, there are more streams increasing in DOC (32%) and more decreasing in DON (24%) but a large portion of the streams studied here (57-65%) demonstrate no change in DOC or DON (Table 1.2). We also found that length of data record for DOC and

DON time series did not affect the direction of the trends (Table S1.4).

Several sites showed consistent declining trends in DOC such as BNZ (Fig 1.3a) while

KNZ and HBF (Fig 1.3c and d) showed positive trends. Streams within sites (e.g. LMP, AND,

LUQ, and FIN; Fig 1.3b, e, f, g) exhibited large variability in DOC Sen slopes including both positive, negative, and non-significant patterns. Trends for DON were also inconsistent. BNZ streams showed both increasing and decreasing trends but only two of the declines were statistically significant (Fig 1.4a). In LMP only one stream had a statistically significant increasing trend in DON (Fig 3b). Streams within HBF and AND (Fig 1.4c) show consistent positive and negative trends, respectively, however trends at HBF were not statistically significant. Almost all Sen slopes at AND and LUQ were declining in DON and were statistically significant (Fig 1.4d and e). Streams across Finland showed both increases and declines in DON slopes with some streams not showing any trends (Fig 1.4f).

Sen slopes for DOC:DON ratios show a more consistent pattern across sites. There was a predominance of increasing trends (Fig 1.5) where 30% (Table 1.2) of streams had statistically significant increasing trends. Only 10% of streams showed statistically significant declining trends in DOC:DON ratios. The majority of these streams were found in Finland (Fig 1.5f) and

13 the Lamprey River Basin (Fig 1.5b). A large portion of streams (60%) did have statistically significant trends in DOC:DON ratios (Table 1.2). The length of data record for DOC:DON time series did not affect the direction of the DOC: DON ratio trends (Table S1.6). Inconsistencies in direction and magnitude of trends in DOC, DON, and DOC:DON ratios occurred regardless of land-use history (forested or disturbed) (Fig S1.2) with no statistically significant differences.

Acid Deposition History

Sites historically affected by acid deposition (FIN, HBF, LMP, KNZ) had more positive

DOC (Fig 1.6a) and DON (Fig 1.6b) Sen slopes than those not affected by acid deposition (BNZ,

AND, LUQ) and this difference was statistically significant (DOC p=0.0003 and DON p=0.009).

Sen Slopes for DOC:DON ratios were greater in sites not affected by atmospheric deposition

(p=0.0002) than those affected by deposition (Fig 1.6c).

Potential predictors of DOM trends

At the global scale, there were no consistent drivers of relationships between DOC Sen slopes and predictor variables (Fig 1.6). Instead, site- to watershed-level controls seem to be driving the time series of DOC and DON concentrations and the DOC:DON ratios. DOC Sen slopes from BNZ streams had strong positive correlations with stream temperature (r= 0.56) (Fig

1.6a). AND DOC slopes correlated negatively with mean concentrations of DON (r= -0.60) and

NH4 (r= -0.53) and positively with mean DOC:DON ratios (r= 0.55), and elevation (r= 0.66) (Fig

1.6a). For HBF, DOC Sen slopes corelated positively with mean concentrations of NO3 (r= 0.99),

Ca (r= 0.83), Cl (r= 0.81), and stream temperature (r= 0.90) and negatively with Na (r= -0.57), K

(-0.95), Mg (r= -0.70), SO4 (r= -0.96), Cl (r= -0.81), watershed area (r= -0.65), and elevation (r=

14 -0.77) (Fig 1.6a); however most of these relationships are driven by the largest DOC Sen slope of

W1. LUQ streams had the greatest number of positive strong correlations between DOC Sen

slopes and NO3 (r= 0.53), Si (r= 0.80), Na (r= 0.89), K (r= 0.68), Ca (r= 0.69), Mg (r= 0.87), SO4

(r= 0.76), Cl (r= 0.86), conductance (r= 0.91), and stream temperature (r= 0.89), and negatively with DOC:DON ratios (r= -0.93) and elevation (r= -0.84) (Fig 1.6a). DOC Sen slopes for Finland

(Fig 1.6a) and LMP (Fig 1.6a) did not exhibit any strong correlations with any potential predictor

variables. LMP does demonstrate strong positive correlations with NH4 and PO4 but these are due to the site CB, which has higher than usual NH4 and PO4 concentrations, relative to the

Lamprey River Basin. Relationships between KNZ DOC Sen lopes and predictor variables were not determined due to the small number of streams.

DON Sen slopes do not correlate strongly with predictor variables at the global scale (Fig

1.6b). At BNZ, the strongest positive correlations with DON Sen slopes were DOC (r= 0.55),

DON (r= 0.64), Na (r= 0.63), K (r= 0.72), watershed area (r= 0.56), and stream temperature (r=

0.70) while negatively with NO3 (r= -0.57), NH4 (r= -0.57), and elevation (r= -0.52) (Fig 1.6b).

DON slopes in streams of Finland (Fig 1.6b) do not seem to correlate with any predictor variables. AND DON slopes correlate positively with DOC:DON ratios (r= 0.71) and negatively

with NH4 (-0.56) (Fig 1.5k). At HBF, DON Sen slopes correlate positively with DOC (r= 0.61),

K (r= 0.90), SO4 (r= 0.76), conductance (r= 0.61), watershed area (r= 0.89), and elevation (r=

0.96) and negatively with NO3 (r= -0.87), Ca (r= -0.80), Cl (r= -0.76), and stream temperature

(r= -0.98) but these relationships are driven by the larger DON Sen slope at W9, except with Ca,

Cl, and elevation (Fig1. 6b). DON Sen slopes for the LMP do not demonstrate any correlations with predictor variables (Fig 1.6b). The strongest correlations between DON Sen slopes and predictor variables in LUQ were mostly negative between slopes and Na (r= -0.59), K (r= -0.70),

15 Ca (r= -0.53), SO4 (r= -0.68), conductance (r= -0.53), and stream temperature (r= -0.64) and positively with DOC (r= 0.52), DOC:DON (r= 0.75), and elevation (r= 0.55) (Fig 1.6b).

Relationships between KNZ DON Sen slopes and predictor variables were not determined due to the small number of streams.

The Sen slopes of DOC:DON molar ratios did not correlate strongly with any predictor variables at the global scale (Fig 1.6 c). Streams in BNZ correlated positively with NO3 concentrations (r=0.69) and negatively with concentrations of DOC (r=-0.73), DON (r=-0.56),

SO4 (r=-0.54), Cl (r=-0.56), and DOC:DON molar ratios (r=-0.75) (Fig 1.6 c). Finland nor AND streams presented strong correlations with predictor variables. HBF streams show strong positive correlations with NO3 (r=0.89), Ca (r=0.67), Cl (r=0.92), and stream temperature (r=0.91) while strong negative correlation with K (r= -0.89), Mg (r=-0.75), SO4 (r=-0.86), watershed area (r= -

0.84), and watershed elevation (r= -0.85). In the LMP correlation between DOC:DON Sen slopes and predictor variables were mostly positive but not as strong such as with NO3 (r= 0.52), Mg

(r= 0.51), Conductance (r =0.51) and negatively with DOC (r= -0.56) and DOC:DON ratios (r= -

0.67). Lastly, LUQ streams had strong negative correlations with Ca (r= -0.62) and watershed area (r= -0.78).

Land-use influence on DOM Sen Slopes

In the LMP, DOC Sen slopes correlated negatively with percent of houses on septic tanks

(r= -0.64), grassland (r= -0.57), shrubland (r=-0.68), and wetlands (r=-0.57) and positively with developed land (0.57) (Fig 1.8). DON correlated negatively with percent of houses on septic tanks (r= -0.68), and percent agriculture (r= 0-.60). DOC:DON only presented strong correlations in the positive direction with population in the watershed (r= 0.50) and negatively

16 with wetland (r= -0.54) (Fig 1.8). Across all 74 streams evaluated here, historic land-use (forest vs disturbed) did not yield statistically significant differences for DOC, DON, and DO:DON Sen

Slopes (Fig 1.9).

DOM pool dynamics

Directional trends for DOC and DON were both synchronous and asynchronous and varied by site (Fig 1.10). Only 16 streams out of the 71 (KNZ and one stream in Finland were excluded due to no DON data) streams with DOC and DON trends showed statistically significant trends in DOC and DON (either both positive or both negative), 33 had no changes in both DOC and DON , while all other streams had an increase in DOC or DON and no change in the other. The only sites with statistically significant tau values for both DOC and DON were

Finland, Bonanza, Andrews, and Luquillo, describing statistically significant coupled or decoupled changes in the DOM pool.

Discussion

Contrary to expectations from earlier studies in New England and Europe, concentrations of DOC and DON did not vary over time in any consistent way with a much larger portion of sites exhibiting DOC and DON Sen slopes with no change. This is one of the first studies where time series trends have been simultaneously determined for DOC and DON to capture the inherent heterogeneity of DOM across numerous streams and biomes. Using trends for

DOC:DON molar ratios across these same streams we find that the organic matter pool across multiple biomes is becoming predominantly C-rich and relatively depleted in N. Relationships between DOC and DON Sens slopes and their potential predictor variables were also not

17 consistent across sites. Rather, stronger relationships were found at the site level suggesting that watershed-scale controls on DOM are driving long-term patterns. Trends in DON did not always track those of DOC, with decoupled Sens slopes 47% of the time. Variability in the changes of

DOC and DON over time are likely the result of diverse sources of DOM among sites and multiple pathways by which DOM is transported from hillslopes to the stream.

Global patterns in DOC and DON trends

The global patterns presented here do not conform to our expectations of increasing DOC concentrations. Many studies examining the response of DOC over time are reported from regions that were exposed to significant levels of acidic deposition (Driscoll et al., 2003;

Monteith et al., 2007; Worrall et al., 2004). While these studies have informed the notion that

DOC concentrations are increasing globally, we present evidence that the direction of change in concentrations of DOC and DON is highly variable. Changes in DOM will likely impact low-

DOM streams such as Konza, Hubbard Brook, or Luquillo the greatest, as they are streams where small changes in DOC and DON can reflect a large percentage change and have meaningful ramifications in nutrient processing and metabolic regimes (Bernhardt et al., 2018).

Our DOC slopes ranged between -0.23 and 0.22 mg C/L per year which is within the range of variability reported from other studies ranging between -0.25 and 0.51 mg C/L per year in streams primarily from northern-latitudes (Clair et al., 2008; De Wit et al., 2007; Driscoll et al.,

2003; Evans et al., 2005). Our DON slopes ranged between -0.02 and 0.01 mg N/L per year which are an order of magnitude greater than those reported earlier (0.003 mg N/L per year in total N for 3 streams in Canada; (Clair et al., 2008). The variation found in DOC and DON

trends in our data set is similar to the trends and magnitudes that were found for NO3 and NH4

18 concentrations in various streams throughout USA where drivers also varied by watershed

(Argerich et al., 2013).

Each watershed, regardless of latitude and biome, has its own drivers of changes in DOM which suggests that the controls on DOM dynamics are specific to each site and even to the sub- watershed level. Lack of relationships between DOC and DON Sen slopes and predictor variables is probably due to the large variability in directions and magnitude of DOM trends within a site and heterogeneity among watersheds at the cross-site scale. The lack of a global driver of changes in DOM is not a surprise as others have not found consistent global or even regional predictor variables for gross primary productivity in streams, which is closely tied to

DOM supplies (Bernhardt et al., 2018), changes in concentrations of inorganic nitrogen

- (Argerich et al., 2013) and NO3 uptake (Wymore et al., 2016). This is contrary to what others have proposed of consistent underlying factors that have led to increase in DOC concentrations across multiple sites in the UK (Evans et al., 2005; Monteith et al., 2007). Declines in acid deposition and changes in ionic strength can explain some of the increases in DOC shown in this study, as has been found in most of the studies in northern Europe (Hruška et al., 2009; Monteith et al., 2007; Worrall et al., 2004) but not all regions presented here are significantly affected by atmospheric deposition (i.e. LUQ, AND, Fig 1.2) and yet DOM has been changing in these sites.

In addition, even sites that have been affected by acid deposition do not always show increasing trends in DOC (i.e. HBF, LMP, and FIN).

Understanding DOM stoichiometry in a global context

Dissolved organic carbon and dissolved organic nitrogen represent two ways to measure the DOM pool, and yet there has been surprisingly little effort in the literature to use both bulk

19 elemental analyses as a way to describe the heterogenous DOM pool (McDowell et al., 2019).

Although this stoichiometric approach to understanding nutrient and carbon dynamics has a rich history (Elser et al., 2000), the principles have seldom been applied to understanding changes in

DOM composition over time. Here we present the first assessment of DOC:DON time series.

Across sites there is a predominant increase in DOC:DON ratios, a pattern that seems to hold regardless of atmospheric deposition history or land use. The exception to this general pattern is

LMP, a region in which streams are mostly declining in DOC:DON ratios as DOM becomes more N-rich.

Multiple streams showed synchronous trends in DOC and DON, similar to what others have also found with declines in both DOC and DON (Dillon & Molot, 2005). These cases describe what is expected where DON tracks changes in DOC and the same biogeochemical factors influence the C-rich and N-rich fractions of the DOM pool. Even though increases may be synchronous, the magnitude of change of DOC is greater than that of DON which will lead to increases in C:N ratios. Various studies have determined that increasing C:N ratios can enhance

N removal from streams, at the local and regional scale (Rodríguez-Cardona et al., 2016;

Wymore et al., 2016), inhibit certain N transformations like nitrification (Starry et al., 2005;

Strauss & Lamberti, 2002), and increase retention of N species (Dodds et al., 2000, 2004).

Changes to C:N ratios will thus influence nutrient exports downstream that can affect trophic assemblages of receiving bodies of water (Schade et al., 2005).

Changes in the DOM pool was not always synchronous demonstrating that the DOM pool as a whole is very dynamic and the different constituents of DOM do not always track each other or have the same controls (McDowell et al., 2004). For example, we found streams declining in DOC but increasing in DON (quadrant IV Fig 1.9) as well as sites that changed in

20 either DOC or DON but not in the other constituent (e.g. where points exist on one of the zero- lines Fig 1.9). This suggests a biogeochemical decoupling of the C-rich and N-rich fractions of the DOM pool where DON can cycle relatively independently of DOC. Changing DON while

DOC is stable can be due to DON that is often found in low concentrations as it is such a small portion of the DOM pool (Pagano et al., 2014) and can be more mobile and reactive than DOC due to its hydrophilic nature (Aiken et al., 1992; Hood et al., 2003). In addition, different source and reactivity of DOM along hydrologic pathways can enhance the decoupling of DOC and

DON (Inamdar et al., 2012). Changes at the source could make DOM reaching the stream much more refractory; no changes in DOC but declines in DON could occur as microbes mine more readily accessible organic molecules for energy. The composition of the organic matter will also influence these dynamics as DON is considered more hydrophilic (Aiken et al., 1992; Hood et al., 2003) and can be lost more easily along the way. Decoupling in DOC:DON ratios could lead to a major shift in inorganic-N dominated systems as carbon availability declines. Therefore, if ambient DOM stoichiometry is changing, nutrient dynamics and exports can change. There do not seem to be instances where DOC is declining with increasing DON (quadrant II Fig 1.9) as

DON is tightly linked to its C-rich sources and autochthonous DON contributions are small compared to heterogenous terrestrial inputs.

Drivers of DOM trends

Although declines in atmospheric deposition are one of the major drivers for increases in

DOC, there are other potential explanations for these trends. In Finland, increases in DOC could be due to increasing stream water temperature and high percent of peatlands in the watersheds

(Räike et al., 2012; Sarkkola et al., 2009). For prairie ecosystems like KNZ, intermittent weather

21 patterns (wet/dry periods) of these streams that can lead to increased inputs of soil OM (Rüegg et al., 2015). In other sites historically affected by acid deposition such as HBF and LMP, DOC trends are less clear with greater variability in direction and magnitude despite acid deposition having greatly influenced the northeast region of the US (Goodale et al., 2005; Likens &

Bormann, 1974; Likens et al., 1996). Others have also found declining trends in DOC for the

Lamprey River mainstem (LMP73) with no clear drivers (Coble et al., 2018).

Sites not affected by acid deposition also show variability in DOC trends. Streams in

LUQ exhibit mostly increasing trends in DOC that could be related to the high frequency of hurricanes in the last several decades (McDowell et al. 2013). On the other hand, streams in BNZ had mostly declining trends in DOC attributed to increased permafrost thaw leading to adsorption of DOC to the newly exposed mineral layer (Striegl et al., 2005). In the AND the predominant declines could be due to growing forests that are taking up C and N in soils, especially in the younger forests, and decreasing inputs to streams (Lajtha & Jones, 2018).

Trends in DON were equally as variable in direction and magnitude as DOC with potentially very similar drivers as DOC across the different sites. However, there are far more declining trends in sites not affected by atmospheric deposition and numerous streams with no detectable change in DON in sites affected by atmospheric deposition. DON is often found in low concentrations as it is such a small portion of the DOM pool (Pagano et al., 2014) and can be more mobile and reactive than DOC as it is more hydrophilic (Aiken et al., 1992; Hood et al.,

2003). Few studies have addressed trends in DON but those that have found increasing trends related to changes in stream temperature (Clair et al., 2008), nutrient inputs from anthropogenic activity (Goolsby & Battaglin, 2001), and hydrology (Lepistö et al., 2008) while others have

22 found declines in DON due to changing precipitation patterns (Dillon & Molot, 2005) or no trends in DON (Coble et al., 2018).

Influence of land-use on DOM trends

Broad landscape classification of the different streams did not yield any distinguishable patterns between historically forested or disturbed streams as Sen slope values overlapped (Fig.

1.9) where the controls on changes to DOM are at a much finer scale within a sub-watershed. In the LMP, increasing agricultural cover and percent of the population on septic tanks correlated negatively with DOC and DON trends suggesting that increased inputs of nutrients from agriculture can lead to declines in DOC and DON. The history of the surrounding landscape may be less important when compare to the proximal drivers of nutrient inputs that can alter the in- situ biogeochemical processes.

Conclusion

This study has added to the growing body of literature on long-term trends of DOC, but we have been able to expand the scope of these studies by including underrepresented biomes like the tropics, arctic, and prairie. In addition, this study provides some of the first large-scale changes in DON concentrations and DOM stoichiometry in streams. Although recovery from atmospheric deposition and changes in ionic strength are very plausible explanations and drivers of trends in DOC, it is clear that other drives may be at work and increases in DOC concentrations are not universal. We have shown that changes in DOM are occurring regardless of the deposition history of the site demonstrating how dynamic the aquatic DOM pool is across biomes and that there is no universal response. Drivers of biogeochemical processes will be at

23 the catchment scale and will differ within one region and most certainly across different regions and biomes. These changes in DOC and DON will have implications for in-stream biogeochemical processes as well as DOM exports to receiving bodies of water, especially those systems where changes in DOM are decoupled. Continuing to monitor these long-term trends in

DOM is important to better understand the ultimate fate of DOM and nutrients in streams.

24 Tables

Site Individual DOC DON Site Biome Abbreviation streams (mg/L) (mg/L) 11.40 0.44 Finland FIN Boreal 32 (1.9 – 78.85) (0.02 – 12.40) 3.63 0.26 Bonanza Creek, AK BNZ * Tundra 9 (0.27 – 29.44) (0.02 – 1.91) 2.29 0.08 Hubbard Brook Experimental Forest, NH HBF * Temperate 5 (0.18 – 24.49) (0.01 – 0.42) 4.80 0.19 Lamprey River Basin, NH LMP Temperate 9 (0.04 – 18.43) (0.02 – 0.89) 0.99 0.03 H.J. Andrews Experimental Forest, OR AND * Temperate 9 (0.19 – 4.27) (0.01 – 0.15) 0.88 Konza Prairie, KS KNZ * Prairie 2 - (0.10 – 11.98) 1.05 0.05 Luquillo Experimental Forest, PR LUQ * Tropical 8 (0.05 – 15.16) (0.01 – 1.12)

Table 1.2 Counts (#) and percentages (%) of statistically significant positive and negative and no trend seasonal Sen slopes for DOC, DON, and DOC:DON molar ratios across all sites and total of all sites, with the total across all sites given at the bottom. There is no DON data available for KNZ.

DOC DON DOC:DON

Total Positive Negative No Change Positive Negative No Change Total Positive Negative No Change Total Site Stream Streams Streams s # % # % # % # % # % # % # % # % # %

AND 9 2 22 3 33 4 44 9 0 0 5 56 4 44 9 4 44 0 0 5 56

BNZ 9 0 0 2 22 7 78 9 0 0 2 22 7 78 9 1 11 0 0 8 89

FIN 32 14 44 3 9 15 47 31 7 23 4 13 20 65 30 9 30 4 13 17 57

HBF 5 1 20 0 0 4 80 5 0 0 0 0 5 100 5 2 40 0 0 3 60

LMP 9 0 0 1 11 8 89 9 1 11 0 0 8 89 9 0 0 3 33 6 67

KNZ 2 2 100 0 0 0 0 ------

LUQ 8 4 50 0 0 4 50 8 0 0 6 75 2 25 8 5 63 0 0 3 38

Total 74 23 32 9 12 42 57 71 8 11 17 24 46 65 70 21 30 7 10 42 60

2 6

Figures

Figure 1.1 Location of all sites across the Northern Hemisphere. See Table 1 for complete site names.

Figure 1.3 Seasonal Sen slopes for DOC for each individual stream in (a) Bonanza Creek, (b) Lamprey River Basin, (c) Konza prairie, (d) Hubbard Brook Forest, (e) H.J. Andrews Forest, (f) Luquillo Experimental Forest, and (g) Finland. Stars denote statistically significant Sen slopes. Blue bars represent sites not affected by acid deposition and grey bars are sites historically affected by acid deposition. Bonanza, Lamprey, Hubbard Brook, and Andrews streams are ordered by watershed size; Luquillo streams are ordered by size within their respective major watersheds; Finland streams are ordered by latitude from North to South.

Figure 1.4 Seasonal Sen slopes for DON for each individual stream in (a) Bonanza Creek, (b) Lamprey River Basin, (c) Hubbard Brook Forest, (d) H.J. Andrews Forest, (e) Luquillo Experimental Forest, and (f) Finland. Stars denote statistically significant Sen slopes. Note no Konza slopes or Finish stream (YLIJOKI 1) as there is no DON data available for these sites. Blue bars represent sites not affected by acid deposition and grey bars are sites historically affected by acid deposition. Bonanza, Lamprey, Hubbard Brook, and Andrews streams are ordered by watershed size; Luquillo streams are ordered by size within their respective major watersheds; Finland streams are ordered by latitude from North to South.

Figure 1.5 Seasonal Sen slopes for DOC:DON molar ratios for each individual stream in (a) Bonanza Creek, (b) Lamprey River Basin, (c) Hubbard Brook Forest, (d) H.J. Andrews Forest, (e) Luquillo Experimental Forest, and (f) Finland. Stars denote statistically significant Sen slopes. Note no Konza slopes or Finish stream (YLIJOKI 1) as there is no DON data available for these sites. Blue bars represent sites not affected by acid deposition and grey bars are sites historically affected by acid deposition. Bonanza, Lamprey, Hubbard Brook, and Andrews streams are ordered by watershed size; Luquillo streams are ordered by size within their respective major watersheds; Finland streams are ordered by latitude from North to South.

Figure 1.6. DOM Sen slopes grouped by history of acid deposition where sites affected by acid deposition are in grey (FIN,HBF,LMP,KNZ) and sites not affects by acid deposition (BNZ,AND,LUQ) in blue. Letter denote statistically significant differences.

Figure 1.7. Correlation plot matrix of individual stream (A) DOC seasonal Sen slopes from, (B) DON seasonal Sen Slopes, and (C) DOC:DON molar ratios seasonal Sen slopes for each site related to the median values of stream DOC, DON, NO3, NH4, PO4, Si, Na, K, Ca, Mg, SO4, and Cl concentrations, stream conductance, temperature, watershed area, and stream elevation. Width of the ellipsis describe variability in the relationships where wider ellipses pertain to more variability. The strength of the correlation is described by colors where colder colors represent positive strong correlations and warmer colors strong negative correlations. Blank boxes denote no data available or gaps within that predictor variables. KNZ was excluded here due to low n in sites.

Figure 1.8. Correlation plot matrix of individual streams in the Lamprey River Basin DOC, DON and DOC:DON molar ratio seasonal Sen slopes from correlated to the population, housing per area, houses on septic tanks, percent of impervious surface, percent developed, percent agriculture cover, percent forest cover, percent shrub cover, percent wetland cover, and percent bare land at the sub-watershed level. Width of the ellipsis describe variability in the relationships where wider ellipses pertain to more variability. The strength of the correlation is described by colors where colder colors represent positive strong correlations and warmer colors strong negative correlations.

Figure 1.9. Sen slopes for DOC, DON, and DOC:DON molar ratios grouped by forested sites (green) and disturbed sites (purple) across all streams regardless of biome; DOC Sen slopes p=0.23, DON Sen slopes p=0.33, and DOC:DON Sen slope p=0.71.

Figure 1.10. DOC and DON tau values from the seasonal Mann-Kendall test where each point is an individual stream color coded by their respective site. Values that lie on the 0-lines are values that are not statistically significant for DOC or DON; points within quadrants are statistically significant for both DOC and DON. Roman numerals denote the quadrant number. KNZ and one Finish stream (YLIJOKI 1) are not included here because they did not have DON data. ). The different quadrants describe trends in DOM where QI describes increases in DOC and DON, QII are streams with declines in DOC and increases in DOC, QIII shows declined in DOC and DON, and QIV described increases in DOC with declines in DON.

Supplementary Information:

Table S1.1. Individual stream (with abbreviated stream name) characteristics such as latitude (Lat) and longitude (Long), mean annual temperature (MAT), mean annual precipitation (MAP), watershed elevation (Elev) and area along with mean stream pH, temperature (Temp), and Conductivity (Cond). Reference describe sites that are specifically labeled reference watersheds, or no management or manipulations have been documented for these watersheds. Forested in Finland sites are less impacted than those described as suburban. – denote data is not available for that stream. MAT MAP Elev Area Temp Cond Anthropogenic Site Stream Name Stream Lat Long pH Land-use History °C mm m km2 °C µs/cm Activity Aura 54 ohikulku FIN Aur 60.47 22.36 5.1 660 10 874 - - - Suburban Disturbed va6401 Eura 42 Pori-Rma FIN Eur 21.73 4.3 622 9 1336 - - - Suburban Disturbed va6900 Iijoki Raasakan FIN Iij 65.33 25.42 1.6 474 17 14191 - - - Forested Forested voimal FIN Kalajoki 11000 Kal 61.95 27.22 2.3 524 110 13 - - - Suburban Disturbed FIN Kelopuro 28 Kel 63.16 30.69 3.2 554 166 0.76 - - - Forested Forested Kemijoki Isohaara FIN Kem 65.79 24.55 1.3 516 13 51127 - - - Forested Forested 14000 Kiiminkij 13010 4- FIN Kii 65.18 25.36 1.7 468 0 3814 - - - Forested Forested tien s Kisko 14 Vanhak FIN Kis 60.13 23.16 5.1 612 12 1047 - - - Suburban Disturbed va6111 FIN Kivipuro 39 Kiv 63.87 28.65 3.2 554 186 0.54 - - - Forested Forested FIN Kojo 35 Pori-Tre Koj 63.88 28.67 4.2 615 210 1.20 - - - Forested Forested FIN Koskenkylanjoki 6030 Kos 60.50 25.95 3.2 554 9 895.00 - - - Suburban Disturbed FIN Kotioja 1 Kot 66.14 26.15 3.2 554 162 18.00 - - - Forested Forested Kymij Huruksela 033 Forested FIN K.Hur 60.49 26.45 4.7 608 20 37159 - - - Forested 5600 Kymijoki Ahvenkoski Forested FIN K.Ahv 60.49 26.45 3.2 554 0 - - - - Forested 001 Kymijoki Kokonkoski Disturbed FIN K.Kok 60.53 26.91 3.2 554 12 - - - - Suburban 014 FIN Lapuanjoki 9900 Lap 63.53 22.53 2.9 536 9 4122 - - - Suburban Disturbed Lestijoki 10800 8-tien FIN s Les 64.06 23.66 2.5 528 5 1373 - - - Suburban Disturbed

FIN Murtopuro 42 Mur 63.76 28.49 3.2 554 179 4.90 - - - Forested Forested FIN Oulujoki 13000 Oul 65.02 25.47 1.9 460 0 22845 - - - Forested Forested Pajo 44 Isosilta FIN Paj 60.46 22.68 4.9 640 10 1088 - - - Forested Forested va6301 FIN Perhonjoki 10600 Per 63.85 23.22 2.6 534 5 2524 - - - Suburban Disturbed Pyhajoki Hourunk Disturbed FIN Pyh 64.46 24.27 3.2 554 10 3712 - - - Suburban 11400 FIN Savi 12 mittapato Sav 62.51 26.03 3.2 554 94 5.39 - - - Suburban Disturbed Siikajoki 8-tien s Forested FIN Sii 60.60 22.67 2.1 486 51 15.40 - - - Forested 11600 FIN Simojoki As. 13500 Sim 65.66 25.08 1.5 490 9 4218 - - - Forested Forested FIN Skatila vp 9600 Ska 63.09 21.89 3.4 525 3 3160 - - - Forested Forested Tornionj Kukkola Forested FIN Tor 65.96 24.05 1.1 540 21 4923 - - - Forested 14310 Uske 16 Salon yp Disturbed FIN Usk 60.39 23.13 4.8 618 3 566 - - - Suburban va6101 FIN Valipuro 38 Val 63.87 28.66 3.2 554 186 0.86 - - - Forested Forested FIN Virojoki 006 3020 Vir 60.58 27.71 4.6 624 9 357 - - - Suburban Disturbed Vuoksi Vastuupuomi Forested FIN Vuo 61.20 28.78 3.8 671 70 61466 - - - Forested 061 FIN Ylijoki 1 Ylij 66.15 26.16 3.2 554 164 56 - - - Forested Forested BNZ C1 C1 65.15 -147.65 -4.3 312 537 6.70 7.42 4.00 53.3 Reference Forested BNZ C2 C2 65.16 -147.60 -4.3 312 375 5.20 7.66 4.90 78.7 Reference Forested BNZ C3 C3 65.14 -147.57 -4.3 312 357 5.70 7.54 3.80 72.7 Reference Forested BNZ C4 C4 65.16 -147.50 -4.7 309 267 10.00 7.82 5.09 103 Burned 1999 Forested BNZ CB CB 65.10 -147.40 -5.4 296 250 16.70 7.55 5.00 67.4 Reference Forested BNZ CJ CJ 65.15 -147.49 -4.7 309 224 41.70 7.60 6.50 82 Reference Forested BNZ P6 P6 65.18 -147.39 -5.6 294 465 7 7.82 4.18 123 Burned 1999 Forested BNZ PC PC 65.15 -147.48 -4.7 309 217 101.50 7.81 8.00 102.5 Burned 1999 Forested BNZ PJ PJ 65.15 -147.48 -4.7 309 222 59.80 7.85 8.49 111 Reference Forested HBF Watershed 1 W1 43.95 -71.73 4.8 1160 482 0.12 5.53 8.00 13 Wallastonite Additions Disturbed HBF Watershed 6 W6 43.95 -71.74 4.8 1160 546 0.13 5.21 5.50 12.5 Reference Forested HBF Watershed 7 W7 43.93 -71.77 4.8 1160 614 0.77 5.99 4.80 12.6 Reference Forested 38

HBF Watershed 8 W8 43.93 -71.76 4.8 1160 602 0.59 5.66 4.60 12.9 Reference Forested HBF Watershed 9 W9 43.93 -71.75 4.8 1160 696 0.68 4.64 3.40 17.9 Reference Forested LMP College Brook CB 43.14 -70.94 8.9 1169 17 2.27 7.16 10.74 524 Urban Disturbed LMP Lamprey River LMP73 43.10 -70.95 8.9 1169 14 479.20 6.70 11.93 87 Suburban Forest LMP Little River LTR 43.11 -71.01 8.9 1169 26 51.70 6.69 9.55 62.9 Suburban Forested LMP Moonlight Brook MLB 43.08 -70.94 8.9 1169 4 0.90 6.58 9.48 455 Urban Disturbed LMP North Branch NBR 43.06 -71.24 8.9 1169 64 41.50 6.46 10.86 74 Suburban Forested LMP North River NOR 43.08 -71.04 8.9 1169 30 128.90 6.62 11.21 61 Suburban Forested LMP Pawtuckaway PWT 43.10 -71.20 8.9 1169 136 2.60 5.88 12.16 25 Forest Forested LMP Rum Brook RMB 43.05 -71.03 8.9 1169 30 4.90 6.68 9.84 113 Suburban Forested Wednesday Hill LMP WHB 43.12 -71.00 8.9 1169 32 1.00 7.09 9.46 221 Suburban Disturbed Brook AND Look Creek LOOK 44.28 -122.10 7.5 1950 422 62.42 7.40 - 31.3 Partially Harvested Forested AND Mack Creek MACK 44.22 -122.17 7.5 1950 755 5.81 7.30 - 26.75 Partially Harvested Forested AND Watershed 1 WS01 44.21 -122.26 7.5 1950 439 0.96 7.50 - 43.1 Clear-cut Forested AND Watershed 2 WS02 44.21 -122.23 7.5 1950 545 0.60 7.50 - 36.45 Reference Forested AND Watershed 6 WS06 44.27 -122.18 7.5 1950 878 0.13 7.50 - 38.6 Clear-cut Forested Partial Overstory Forested AND Watershed 7 WS07 44.27 -122.17 7.5 1950 918 0.15 7.50 - 42.9 Harvest AND Watershed 8 WS08 44.27 -122.17 7.5 1950 962 0.21 7.50 - 38.3 Reference Forested AND Watershed 9 WS09 44.20 -122.25 7.5 1950 426 0.09 7.50 - 48.45 Reference Forested AND Watershed 10 WS10 44.22 -122.25 7.5 1950 461 0.10 7.50 - 43.6 Clear-cut Forested KNZ Elder Spring edlr 39.10 -96.61 12.1 866 337 - - - - Reference Forested KNZ Hiking Trail hikx 39.11 -96.61 12.1 866 324 - - - - Agriculture Forested LUQ Biesley 1 Q1 18.32 -65.75 22.8 1768 377 0.07 7.26 23.00 97.4 Reference Forested LUQ Bisley 2 Q2 18.31 -65.75 22.8 1768 218 0.06 7.29 22.90 92.6 Reference Forested LUQ Bisley 3 Q3 18.31 -65.75 22.8 1768 198 0.35 7.32 22.63 85.6 Reference Forested LUQ Qda. Guaba QG 18.28 -65.79 21.4 3229 642 0.13 6.75 20.80 47 Reference Forested LUQ Qda. Prieta QP 18.32 -65.82 22 4587 431 0.31 7.02 22.00 79.6 Reference Forested LUQ Qda. Sonadora QS 18.32 -65.82 22 3495 740 2.60 6.90 21.60 49.8 Reference Forested 39

LUQ Río Icacos RI 18.28 -65.79 21.4 2254 686 3.30 6.76 21.00 55.5 Reference Forested LUQ Mameyes Puente Roto MPR 18.33 -65.75 23 2843 498 17.7 7.44 23.3 94.5 Reference Forested

Table S1.2. Mean solute concentrations for every stream. – denote data is not available for that stream.

DOC DON TDN NO3 NH4 PO4 Si TDP Na K Ca Mg SO4 Cl Site Stream mg/L mg/L mg/L mg/L µg/L µg/L mg/L mg/L mg/L mg/L mg/L mg/L mg/L mg/L

FIN Aur 13.30 0.64 2.09 1.40 58 81 - 0.04 8.00 3.20 10.80 6.90 - - FIN Eur 9.31 0.51 1.52 0.84 210 17 - 0.01 7.40 3.20 13.80 5.60 - - FIN Iij 9.69 0.32 0.37 0.04 8 8 - 0.01 1.70 0.60 2.90 1.00 - - FIN Kal 19.95 0.69 1.33 0.60 85 40 - 0.04 4.60 2.30 6.70 4.00 - - FIN Kel 9.50 0.22 0.25 0.01 9 2 - 0.00 1.00 0.30 1.10 0.20 - - FIN Kem 7.60 0.28 0.33 0.04 12 6 - 0.01 1.70 0.60 4.40 1.30 - - FIN Kii 14.06 0.42 0.51 0.06 8.50 11 - 0.02 2.20 0.70 3.30 1.30 - - FIN Kis 9.50 0.53 0.95 0.44 25 21 - 0.02 3.80 1.60 6.70 2.80 - - FIN Kiv 27.08 0.47 0.48 0.01 6 5 - 0.01 1.50 0.60 1.30 0.80 - - FIN Koj 9.50 0.44 1.05 0.53 64 16 - 0.01 6.80 2.00 7.70 2.70 - - FIN Kos 9.41 0.50 1.43 0.96 43 42 - 0.02 6.30 3.20 9.70 5.10 - - FIN Kot 15.20 0.44 0.52 0.07 10 10 - - 1.20 0.50 5.40 2.00 - - FIN K.Hur 7.41 0.31 0.54 0.22 22 3 - 0.01 6.23 1.50 5.30 1.50 - - FIN K.Ahv 7.32 0.33 0.57 0.24 22 4 - 0.01 6.20 1.60 5.50 1.60 - - FIN K.Kok 7.13 0.31 0.55 0.23 21 3 - 0.01 6.40 1.60 5.40 1.50 - - FIN Lap 19.00 0.63 1.53 0.70 230 34 - 0.02 6.50 3.10 8.50 4.00 - - FIN Les 19.00 0.58 0.88 0.27 54 30 - 0.03 2.70 1.40 3.90 1.70 - - FIN Mur 23.75 0.45 0.51 0.01 7.50 11 - 0.01 1.50 0.60 1.70 0.80 - - FIN Oul 8.93 0.29 0.36 0.05 14 6 - 0.01 1.70 0.70 2.90 1.00 - - FIN Paj 11.40 0.63 2.14 1.40 56 98 - 0.04 7.90 4.30 9.90 7.10 - - FIN Per 19.00 0.58 1.05 0.36 110 30 - 0.02 2.70 1.50 4.60 1.70 - - FIN Pyh 17.10 0.56 0.89 0.30 44 23 - 0.03 4.60 1.90 9.90 2.70 - - FIN Sav 10.45 0.55 1.73 1.10 80 65 - 0.03 6.60 2.80 9.22 6.70 - - FIN Sii 18.05 0.54 0.82 0.22 55 36 - 0.03 3.10 1.40 4.45 2.05 - - FIN Sim 11.40 0.37 0.45 0.04 13 5 - 0.01 1.50 0.60 3.30 1.40 - - FIN Ska 19.00 0.64 1.90 0.98 190 43 - 0.02 7.10 3.10 9.30 4.84 - - 41

FIN Tor 5.42 0.21 0.27 0.01 7 5 - 0.01 1.40 0.70 3.80 1.00 - - FIN Usk 10.45 0.58 1.90 1.20 79.50 73 - 0.03 10.00 3.80 12.00 8.00 - - FIN Val 30.35 0.48 0.52 0.01 7 3.25 - 0.01 1.10 0.30 0.80 0.50 - - FIN Vir 14.25 0.57 0.92 0.34 30 16 - 0.02 3.30 1.60 5.80 1.70 - - FIN Vuo 6.75 0.22 0.42 0.15 8 2 - 0.00 5.20 1.30 5.10 1.30 - - FIN Ylij 13.11 0.44 0.57 0.06 19 12 - - 1.30 0.60 5.25 2.00 - - BNZ C1 3.66 0.24 0.53 0.27 27.17 - 15.11 - 0.97 0.33 7.65 1.57 3.18 0.28 BNZ C2 2.63 0.25 0.82 0.56 26.61 - 21.76 - 1.04 0.37 10.81 3.18 6.41 0.33 BNZ C3 5.01 0.30 0.71 0.43 26.61 - 16.16 - 1.18 0.38 11.07 1.82 7.17 0.27 BNZ C4 2.28 0.24 0.87 0.65 25.21 - 21.31 - 1.30 0.57 15.71 3.01 7.08 0.32 BNZ CB 3.28 0.23 0.54 0.33 16.81 - 20.80 - 1.14 0.41 10.18 2.32 4.89 0.34 BNZ CJ 5.06 0.31 0.62 0.32 16.11 - 22.68 - 1.27 0.52 13.71 2.51 6.36 0.28 BNZ P6 4.36 0.27 0.72 0.46 24.16 - 20.93 - 1.32 0.51 18.80 3.17 15.26 0.38 BNZ PC 4.74 0.28 0.54 0.27 15.97 - 23.50 - 1.27 0.58 15.92 3.16 8.40 0.39 BNZ PJ 5.10 0.33 0.60 0.26 21.01 - 22.47 - 1.38 0.66 17.18 3.44 9.27 0.41 HBF W1 2.28 0.08 0.28 0.17 4 0.3 5.29 - 0.69 0.11 1.18 0.16 0.98 0.43 HBF W6 1.94 0.07 0.12 0.01 4 0.3 4.26 - 0.77 0.15 0.63 0.21 1.21 0.41 HBF W7 1.68 0.06 0.10 0.02 4 0.3 5.20 - 0.77 0.15 0.91 0.30 1.19 0.37 HBF W8 2.63 0.07 0.13 0.04 4 0.7 5.80 - 0.83 0.16 0.76 0.30 1.18 0.38 HBF W9 7.52 0.14 0.21 0.02 4 0.3 6.10 - 0.71 0.16 0.62 0.21 1.15 0.39 LMP CB 4.16 0.23 0.83 0.60 38.87 16.51 10.31 - 80.79 5.01 23.84 6.56 4.55 140.46 LMP LMP73 5.25 0.21 0.37 0.12 0.02 0.00 5.18 - 14.54 1.15 6.37 1.32 1.87 24.25 LMP LTR 5.45 0.17 0.30 0.10 0.01 0.00 7.17 - 11.93 0.79 5.04 1.16 1.38 20.16 LMP MLB 1.64 0.09 0.79 0.70 0.04 0.00 10.85 - 70.66 3.86 37.53 6.56 4.47 163.60 LMP NBR 5.03 0.18 0.25 0.05 0.01 0.00 6.39 - 14.36 0.73 4.98 1.05 1.62 24.55 LMP NOR 5.59 0.20 0.28 0.05 0.01 0.00 6.42 - 10.53 0.95 5.44 1.03 1.82 16.68 LMP PWT 5.48 0.19 0.24 0.01 0.01 0.00 7.87 - 2.06 0.76 3.98 1.01 1.00 2.49 LMP RMB 5.56 0.21 0.38 0.11 0.02 0.01 10.67 - 20.59 2.14 10.80 1.67 2.16 36.84 LMP WHB 3.07 0.14 0.81 0.65 0.01 0.00 9.05 - 32.23 1.39 16.18 6.70 3.59 50.46 41

AND LOOK 0.90 0.03 0.04 0.00 6.00 11.00 7.78 0.01 2.07 0.40 3.00 0.88 0.08 0.67 AND MACK 0.84 0.03 0.09 0.05 8.00 7.00 6.80 0.02 1.49 0.34 2.46 0.87 0.10 0.65 AND WS01 1.14 0.05 0.08 0.01 13 21.00 9.10 0.03 2.64 0.20 4.64 1.02 0.18 0.74 AND WS02 1.47 0.03 0.03 0.00 6 22.00 9.00 0.04 2.54 0.38 3.47 0.82 0.12 0.90 AND WS06 0.52 0.02 0.03 0.00 5 13.00 8.20 0.03 2.11 0.38 3.94 0.99 0.07 0.61 AND WS07 0.53 0.02 0.02 0.00 4 22.00 9.32 0.03 2.32 0.58 4.07 1.12 0.05 0.60 AND WS08 1.14 0.02 0.03 0.00 5 23.00 8.04 0.04 3.05 0.47 3.46 0.67 0.11 0.71 AND WS09 1.77 0.04 0.05 0.00 6 17.00 8.90 0.03 2.67 0.17 4.69 1.35 0.19 1.31 AND WS10 0.89 0.03 0.04 0.00 6 32.00 8.88 0.05 2.49 0.23 4.30 1.09 0.15 1.26 KNZ edlr 0.80 - - 0.06 16 ------KNZ hikx 0.94 - - 0.29 26 ------LUQ Q1 0.72 0.05 0.20 0.14 5.34 1.00 34.41 - 8.58 0.91 5.52 3.11 1.28 8.35 LUQ Q2 0.85 0.05 0.19 0.13 2.50 1.00 29.48 - 7.43 0.93 5.98 2.85 0.94 7.39 LUQ Q3 0.80 0.05 0.15 0.11 2.50 1.00 28.42 - 7.56 0.72 4.68 3.07 0.64 8.36 LUQ QG 0.90 0.04 0.11 0.07 2.50 1.00 15.25 - 5.27 0.45 2.76 1.07 0.41 6.30 LUQ QP 1.30 0.06 0.12 0.06 2.50 1.00 19.11 - 6.24 0.33 4.35 3.27 0.48 8.18 LUQ QS 1.58 0.07 0.14 0.08 2.50 1.00 11.10 - 4.72 0.28 2.49 1.47 0.57 7.17 LUQ RI 0.99 0.04 0.15 0.09 7.29 1.00 19.21 - 5.40 0.55 3.57 1.24 0.43 6.17

LUQ MPR 1.12 0.06 0.13 0.06 2.50 1.00 20.84 - 6.57 0.68 7.70 1.99 1.17 7.78

Table S1.3. Start and end date of time series for each stream along with the length of the data record. All records started on January and ended in December for every year analyzed for DOC and DON except BNZ which started in May and ended in August for every year analyzed. – correspond to no data available. Total record Start Year End Year Site Stream length DOC DON DOC DON DOC DON LUQ RI 2000 2015 15 LUQ MPR 1998 2015 17 LUQ QP 2000 2015 15 LUQ Q1 1998 2015 17 LUQ Q2 1998 2015 17 LUQ Q3 1998 2015 17 LUQ QG 2000 2015 15 LUQ QS 2000 2015 15 AND MACK 2003 2015 12 AND WS02 2003 2015 12 AND WS08 2004 2015 11 AND WS09 2004 2015 11 AND WS01 2004 2015 11 AND WS06 2003 2015 12 AND WS07 2003 2015 12 AND WS10 2004 2015 11 AND LOOK 2006 2015 9 LMP CB 2005 2016 11 LMP LMP73 2000 2016 16 LMP LTR 2004 2016 12 LMP MLB 2008 2016 8 LMP NBR 2004 2016 12 LMP NOR 2004 2016 12 LMP PWT 2004 2016 12 LMP RMB 2004 2016 12 LMP WHB 2003 2016 13 FIN Aur 1988 2015 27 FIN Eur 1991 2015 24 FIN Iij 1995 2015 20 FIN Kal 1975 1984 2012 37 31 FIN Kel 1989 1975 2015 26 40 FIN Kem 1991 1988 2015 24 27 FIN Kii 1995 1975 2015 2012 20 37 FIN Kis 1996 1989 2015 19 26

FIN Kiv 1978 1991 1994 2015 16 24 FIN Koj 1975 1995 2015 40 20 FIN Kos 1994 1996 2015 21 19 FIN Kot 1981 1978 2001 1994 20 16 FIN K.Hur 1975 2015 40 FIN K.Ahv 1984 1994 2015 31 21 FIN K.Kok 1988 1981 2015 2001 27 20 FIN Lap 1996 2012 2015 16 19 FIN Les 1998 1995 2015 17 20 FIN Mur 1978 1994 16 FIN Oul 1975 2015 40 FIN Paj 1985 2015 30 FIN Per 1982 2015 33 FIN Pyh 1996 2015 19 FIN Sav 2006 2015 9 FIN Sii 1983 2015 32 FIN Sim 1991 2015 24 FIN Ska 1975 2015 40 FIN Tor 1991 2015 24 FIN Usk 1988 2015 27 FIN Val 1978 1994 16 FIN Vir 2003 2015 12 FIN Vuo 1989 2015 26 FIN Ylij 1981 2001 20 - HBF W1 2006 2015 9 HBF W6 1996 2015 19 HBF W7 1996 2015 19 HBF W8 1996 2015 19 HBF W9 1996 2015 19 KNZ edlr 1994 2016 22 - KNZ hikx 1994 2016 22 - BNZ C1 2002 2010 8 BNZ C2 2002 2010 8 BNZ C3 2002 2010 8 BNZ C4 2002 2010 8 BNZ CB 2002 2010 8 BNZ CJ 2002 2010 8 BNZ P6 2005 2010 5 BNZ PC 2002 2010 8 BNZ PJ 2002 2010 8

Table S1.4 Sampling rates and analytical methods for analysis of dissolved organic carbon (DOC) or total organic carbon (TOC), total - + dissolved nitrogen (TDN), total nitrogen (TN) or dissolved organic nitrogen (DON), nitrate (NO3 ), and ammonium (NH4 ) concentrations for each site. – corresponds to no analysis performed.

Sampling - + Site Sampling and storage TOC or DOC TDN TN or DON NO3 NH4 Reference Frequency Samples composite over a Persulfate digestion and 3-week period and DON - mathematically Colorimetric, analysis by automated Colorimetric, Martin 3-week refrigerated each week; determined by automated AND DOC - combustion colorimetric, Technicon and Harr composites filtered after arriving at subtracting NH3-N and cadmium Technicon Auto- Auto-analyzer II 1998 the lab and analyzed NO3-N from TDN reduction analyzer II within 2 days Daily samples were Shimadzu TOC-5000 collected with DON - mathematically DOC - combustion plumbed to an Antek Daily to Bi- autosamplers as a determined by Ion Ion Petrone et BNZ Shimadzu TOC- 7050 nitric oxide weekly composite of samples subtracting NH3-N and chromatography chromatography al. 2006 5000 detector to quantify collected at 6 pm and 6 am NO3-N from TDN total dissolved nitrogen the following day TOC - oxidized to TN - analyzed carbon dioxide by calorimetrically after Mattsson TOC - deep-frozen and combustion and oxidation with et al. analyzed within 1–3 Monthly to determined by peroxodisulfade and 2005; FIN months – – – Weekly infrared reduction with Cd–Cu Kortelaine TN - analyzed on the day spectrometry; column; n et al. after sampling DOC - determined DON - determined as 2006 as 95% of TOC 95% of TN Since June 1, 2013 samples are brought back DOC - high- DON - mathematically Colorimetric, to the Pierce Lab, and High-temperature temperature determined by Ion automated Buso et al. HBF Weekly immediately filtered; catalytic oxidation catalytic oxidation subtracting NH3-N and chromatography indophenol- 2000 Prior, samples were not technique technique NO3-N from TDN blue method filtered or frozen before analysis. DOC - high- DON - mathematically Colorimetric, High-temperature Filtered during collection temperature determined by Ion SmartChem 200 Coble et LMP Weekly catalytic oxidation and frozen until analyzed catalytic oxidation subtracting NH4-N and chromatography using the al. 2018 technique technique NO3-N from TDN alkaline phenate Colorimetric Colorimetric determination determination DOC - high Kemp and Every other on a flow on a flow KNZ Frozen until analyzed temperature – – Dodds day solution solution combustion 1998 analyzer analyzer

DOC - high- High-temperature DON - mathematically Colorimetric, Filtered during collection temperature catalytic oxidation determined by Ion SmartChem 200 Merriam LUQ Weekly and frozen until analyzed catalytic oxidation technique subtracting NH4-N and chromatography using the et al. 2002 technique NO3-N from TDN alkaline phenate

Table S1.5. Tau values for DOC and DON calculated using Seasonal Mann–Kendall. Red denotes statistically significant increasing trend, blue denotes statistically significant decreasing trends, and gray are no significant trends for the given period of time. The total column are Tau values for the total record of available data (see Table S1.3 for details on record length) for each stream. – denote data is not available for that stream. DOC DON 2015- 2015- 2015- 2015- 2015- 2015- 2015- 2015- 2015- 2015- 2015- 2015- 2015- 2015- 2005 2000 1995 1990 1985 1980 1975 2005 2000 1995 1990 1985 1980 1975 Site Stream Total Total 10 15 20 25 30 35 40 10 15 20 25 30 35 40 Years Years Years Years Years Years Years Years Years Years Years Years Years Years FIN Aur 0.199 0.268 0.203 0.234 0.200 -0.126 -0.147 -0.127 -0.121 -0.141 FIN Eur 0.127 -0.134 FIN Iij 0.216 0.142 0.240 0.216 0.138 FIN Kal 0.109 -0.083 0.120 0.145 0.173 0.074 FIN Kel 0.152 -0.164 0.102 0.165 0.326 0.176 0.173 0.184 0.075 FIN Kem FIN Kii 0.226 0.323 0.226 0.125 0.108 0.189 0.238 0.132 FIN Kis 0.203 0.235 0.369 0.147 0.229 0.226 0.263 FIN Kiv -0.166 0.215 0.162 FIN Koj -0.255 -0.301 -0.422 -0.365 -0.376 -0.352 -0.304 -0.255 FIN Kos 0.312 0.209 0.304 0.348 0.209 0.122 0.171 FIN Kot FIN K.Hur -0.259 -0.293 -0.333 -0.506 -0.493 -0.405 -0.259 -0.268 -0.233 -0.261 -0.249 -0.291 -0.268 FIN K.Ahv 0.136 -0.455 -0.498 -0.372 -0.109 0.120 0.332 0.220 FIN K.Kok 0.350 -0.151 -0.172 0.321 -0.128 -0.401 -0.246 -0.128 FIN Lap 0.213 0.136 0.285 -0.112 FIN Les FIN Mur -0.165 -0.134 0.170 FIN Oul -0.238 -0.257 -0.158 -0.176 -0.093 -0.152 -0.136 -0.086 FIN Paj 0.112 0.155 0.112 0.120 0.130 0.099 0.120 FIN Per 0.181 -0.175 -0.187 -0.195 -0.260 -0.168 -0.116 FIN Pyh 0.142 0.165 0.109 0.125 0.167 0.092 47

FIN Sav 0.154 FIN Sii 0.213 0.151 0.268 0.134 0.145 0.239 0.204 FIN Sim 0.135 0.153 0.131 0.163 FIN Ska -0.140 -0.135 -0.096 -0.105 -0.123 -0.150 -0.105 FIN Tor 0.122 0.112 FIN Usk 0.084 0.226 0.179 0.174 0.177 -0.148 FIN Val -0.277 0.149 0.178 0.131 FIN Vir FIN Vuo 0.116 -0.245 0.085 FIN Ylij ------BNZ C1 BNZ C2 -0.333 -0.221 -0.422 BNZ C3 -0.267 BNZ C4 BNZ CB BNZ CJ BNZ P6 -0.381 BNZ PC BNZ PJ HBF W1 0.243 0.183 HBF W6 -0.143 HBF W7 HBF W8 HBF W9 LMP CB 0.170 LMP LMP73 -0.135 -0.300 -0.132 -0.288 LMP LTR LMP MLB LMP NBR 48

LMP NOR LMP PWT 0.161 0.143 LMP RMB LMP WHB -0.142 AND LOOK -0.185 -0.197 AND MACK -0.229 -0.269 AND WS01 -0.308 -0.346 -0.180 -0.276 AND WS02 -0.154 AND WS06 0.321 0.223 -0.164 AND WS07 -0.150 -0.287 -0.246 -0.251 AND WS08 0.312 0.278 -0.166 AND WS09 -0.142 AND WS10 -0.177 -0.217 -0.231 -0.268 KNZ edlr 0.287 0.200 0.358 ------KNZ hikx 0.359 -0.191 0.162 0.366 ------LUQ MPR 0.149 0.168 -0.206 -0.321 -0.244 LUQ Q1 0.176 0.178 -0.194 -0.252 -0.206 LUQ Q2 0.159 0.162 -0.178 -0.412 -0.264 LUQ Q3 0.155 0.172 -0.218 -0.373 -0.251 LUQ QG -0.136 -0.306 -0.136 LUQ QP LUQ QS -0.107 -0.107 LUQ RI

2015- 2015- 2015- 2015- 2015- 2015- 2015- 2005 2000 1995 1990 1985 1980 1975 Site Stream Total 10 15 20 25 Years 30 Years 35 Years 40 Years Years Years Years FIN Aur 0.158 0.205 0.156 0.185 0.162 FIN Eur 0.110 0.212 0.189 FIN Iij -0.145 FIN Kal -0.174 -0.211 -0.126 -0.039 FIN Kel -0.147 0.083 FIN Kem 0.252 FIN Kii -0.165 FIN Kis -0.223 FIN Kiv 0.211 0.233 0.084 0.067 FIN Koj 0.090 0.177 0.198 0.090 0.071 0.035 FIN Kos 0.111 FIN Kot -0.098 FIN K.Hur -0.134 -0.122 -0.189 -0.174 0.032 FIN K.Ahv 0.291 -0.442 0.237 0.234 FIN K.Kok 0.235 0.110 0.205 0.268 FIN Lap 0.142 0.279 0.154 FIN Les 0.109 FIN Mur -0.303 -0.276 -0.267 -0.252 -0.167 FIN Oul -0.075 -0.112 -0.123 -0.147 -0.075 FIN Paj -0.108 -0.213 -0.108 FIN Per 0.173 0.147 0.212 0.241 0.113 FIN Pyh 0.096 -0.214 FIN Sav -0.241 FIN Sii -0.278 -0.209 -0.298 -0.277 -0.129 FIN Sim -0.030 FIN Ska 0.146 0.167 0.022 0.122 0.211 0.146 FIN Tor -0.047 FIN Usk 0.087 0.200 0.255 0.144 FIN Val -0.249 0.166 0.168 FIN Vir 0.167 BNZ C1 BNZ C2 0.244 BNZ C3 BNZ C4 BNZ CB BNZ CJ BNZ P6 BNZ PC BNZ PJ HBF W1 0.250 0.232 HBF W6 0.162 HBF W7 HBF W8 0.160 0.178 HBF W9

LMP CB LMP LMP73 0.239 LMP LTR -0.178 -0.241 LMP MLBNBR LMP NBR LMP NOR -0.184 LMP PWT LMP RMB -0.176 -0.136 LMP WHB AND MACK 0.174 0.193 AND WS02 AND WS08 0.186 0.211 AND WS09 0.183 AND WS01 0.058 AND WS06 0.224 0.204 AND WS07 AND WS10 0.149 0.149 AND GOOK LUQ RI LUQ MPR 0.265 0.355 0.358 LUQ QP 0.019 0.052 0.019 LUQ Q1 0.304 0.321 0.357 LUQ Q2 0.268 0.415 0.401 LUQ Q3 0.286 0.376 0.369 LUQ QG 0.131 0.391 0.131 LUQ QS 0.035 0.064 0.035

CHAPTER TWO

WILDFIRES LEAD TO DECREASED CARBON AND INCREASED NITROGEN

CONCENTRATIONS IN UPLAND ARCTIC STREAMS

Abstract

The Central Siberian Plateau (CSP) is undergoing rapid climate change that has resulted in increased frequency of forest fires and subsequent alteration of watershed carbon and nutrient dynamics. Across a watershed chronosequence (3 to >100 years since wildfire) we quantified the effects of fire on quantity and composition of dissolved organic matter (DOM) composition, stream water nutrients concentrations, as well as in-stream nutrient uptake rates. Wildfires increased concentrations of nitrate for a decade, while decreasing concentrations of dissolved organic carbon and nitrogen (DOC and DON) and aliphatic DOM contribution for five decades.

These post-wildfire changes in stream DOM result in lower uptake efficiency of in-stream nitrate in recently burned watersheds. Nitrate uptake (as uptake velocity) is strongly dependent on DOM quality (e.g. polyphenolics), ambient dissolved inorganic nitrogen (DIN), and DOC:DIN ratios.

Our observations and experiments suggest that a decade-long pulse of inorganic nitrogen and a reduction of DOC export occur following wildfires in streams draining the CSP. Increased fire frequency in the region is thus likely to both decrease DOM and increase nitrate delivery to the main stem Yenisei River, and ultimately the Arctic Ocean, in the coming decades.

Introduction

Arctic soils store large amounts of organic matter and nutrients in permafrost (Schuur et al., 2008; Zimov et al., 2006). When permafrost is degraded, interactions between nutrient availability and organic matter composition will play an important role in determining the net effects on delivery of dissolved organic carbon (DOC) and inorganic nutrients from upland landscapes to major arctic rivers and ultimately the Arctic Ocean (Frey & McClelland, 2009;

Schuur et al., 2008). Predicted and recently observed changes in the concentration and export of organic matter (OM) and nutrients vary geographically across the Arctic and as a function of permafrost structure, underlying parent material, vegetation, and disturbance (Frey &

McClelland, 2009; McClelland et al., 2007). Boreal forests have been burning with greater frequency due to longer growing seasons, warmer temperatures, and changing weather patterns

(Kawahigashi et al., 2011; Kharuk et al., 2016; Shvidenko & Schepaschenko, 2013), thereby adding additional uncertainty to projections of how these ecosystems will respond to climate change. The response of arctic watersheds varies dramatically among regions, with both increases and decreases in concentrations of DOC and inorganic nutrients following wildfire, and differing return intervals to baseline conditions (Betts & Jones, 2009; Burd et al., 2018; Diemer et al., 2015; Larouche et al., 2015; Parham et al., 2013; Petrone et al., 2007). Although various studies have documented the effects of fire on stream chemistry, few have evaluated how these changes will impact the processing and export of nutrients from arctic fluvial systems. Arctic rivers mobilize large quantities of organic matter and nutrients to the Arctic Ocean (Holmes et al., 2012) which may be altered due to increasing permafrost thaw and wildfire frequency.

Watersheds in the Central Siberian Plateau (CSP), a major component (2.5 million km2) of the

Yenisei River Basin, provide a platform for understanding the long-term resilience and recovery

of stream chemistry following fire in taiga landscapes. Streams of the CSP drain watersheds with extensive larch forest (Larix gmelinii) growing on continuous permafrost (300-500m deep) underlain by unglaciated basalt deposits (Prokushkin et al., 2007, 2011).

Uptake and transformation of nutrients and organic matter by stream channel processes can result in losses of C and N through mineralization and denitrification, as well as alteration of

- the timing of delivery of DOC and nitrate (NO3 ) to downstream ecosystems (Dodds et al., 2002;

Mulholland et al., 2009; Rodríguez-Cardona et al., 2016; Wymore et al., 2016). Uptake rates

(usually expressed as uptake velocity, Vf) are commonly used to quantify nutrient demand and the efficiency with which in-stream biota remove and process nutrients at a given concentration

(Newbold et al., 1981; Workshops, 1990). Uptake velocity is typically determined via in-situ nutrient manipulations since uptake is difficult to infer from downstream changes in the ambient pool of nutrients. Generally, as nutrient concentrations increase, uptake efficiency of the microbial community declines, resulting in a larger fraction of nutrients exported downstream

(Dodds et al., 2002; Peterson et al., 2001). Uptake dynamics have not been evaluated extensively in arctic fluvial systems affected by wildfires (Diemer et al., 2015).

Here we examine how years since the last fire (YSF) affects nutrient concentrations and the quantity and composition of DOM entering fluvial networks using a space-for-time approach spanning a 100-year well-constrained wildfire chronosequence across 17 watersheds (Fig 2.1)

(Table 2.1) in the CSP (Prokushkin et al., 2011). We also determined uptake rates of inorganic N

- in a subset of these watersheds with individual nutrient pulse additions of nitrate (NO3 ) and

+ ammonium (NH4 ) to determine how fire influences the processing and export of nutrients in arctic fluvial systems. We assess changes to the composition of DOM via optical properties and

Fourier transformed ion cyclotron resonance mass spectrometry (FT-ICR MS) to provide an

ultra-high resolution assessment of molecular formulae for thousands of individual DOM molecules (>10,000) (Stubbins et al., 2014) in multiple water samples from each stream.

Samples were taken in multiple years during June (freshet) and July (summer low flow).

Methods

Grab samples across chronosequence:

We collected stream water samples in June and July from 2016 to 2018. Samples were filtered through pre-combusted Whatman GF/F glass fiber filters and collected in acid washed high density poly-ethylene bottles (DOC, TDN, and nutrients), pre-combusted amber glass vials

(optical DOM metrics), and polycarbonate bottles (FT-ICR MS) and rinsed 3 times in the field.

Samples were kept frozen except for DOM optical samples which were kept at 4°C; all samples were transported and analyzed at the Water Quality Analysis Laboratory at the University of

New Hampshire, except samples collected for FT-ICR MS which were analyzed at the National

High Magnetic Field Laboratory at Florida State University.

Nutrient Pulse Additions:

Short term nutrient pulse additions were performed at several sites across the fire chronosequence in June 2016, June and July 2017, and July 2018, following the methods for

- break through curve (BTC) integration (Tank et al., 2008). Amendments consisted of NO3 as

+ 3+ + NaNO3, and NH4 as (NH4)2SO4, with and without PO4 as KH2PO4 but we report NH4 uptake

3+ + from both as PO4 revealed no influence on NH4 uptake metrics. Amendments were added along with NaCl as a conservative tracer mixed with stream water where we increased background nutrient and Cl concentrations 2-3X above background. Experimental reach lengths

ranged between 15 m to 200 m (Table S2.1) and were selected to exclude large pools, tributaries, or other overland flows. The amendments were released at the top of the experimental reach and samples were collected at a fixed point at the bottom of the experimental reach throughout the break through curve (BTC), where between 20 and 40 samples were collected per experiment, depending on flow conditions and changes in specific conductance monitored with a HANNA HI

991300 (HANNA Instruments, Woonsocket, RI, US). Background samples were collected as described above in duplicates before every experiment at the top and bottom of the experimental reach.

Calculations of uptake metrics follow the break through curve integration method (Tank et al., 2008) where uptake length (Sw) was determined as the negative inverse of the ratio between the natural log of the fraction of background corrected nutrient and Cl retrieved and the reach length distance. Uptake velocity (Vf) was determined as:

! " !! = (1) ""

Where Q is discharge, w stream width, and Sw the uptake. Uptake velocity is the best metric to compare across sites and dates as it is normalized for stream depth and velocity

(Peterson et al., 2001; Tank et al., 2008; Workshops, 1990). Uptake velocity is reported in terms

- + of DIN (combining NO3 and NH4 Vf) due to low statistical power if reported individually.

Discharge was determined using the dilution gaging method (Kilpatrick & Cobb, 1985) using

NaCl dissolved in stream water and added to the stream as an instantaneous slug addition and conductivity was logged every 5 seconds using a HOBO conductivity data logger (Onset,

Bourne, MA).

Water Chemistry:

All samples were analyzed for DOC and total dissolved nitrogen (TDN) using high

- temperature catalytic oxidation with a Shimadzu TOC-V CPH/TNM. Nitrate (NO3 ) was analyzed with ion chromatography using an Anion/Cations Dionex ICS-1000 with AS40

+ autosampler. Ammonium (NH4 ) was measured using a SmartChem 200 discrete automated

3- colorimetric analyzer using the alkaline phenate standard method. Phosphate (PO4 ) was analyzed by discrete colorimetric analysis on a Seal AQ2 (Seal Analytical) by the ascorbic molybdenum blue method (EPA 365.1). DON was determined arithmetically by subtracting DIN

- + (NO3 + NH4 ) concentrations from TDN. Samples below detection limit were substituted with

- + 3- half the detection limit; method detection limit for NO3 (4 µg/L), NH4 (5 µg/L), PO4 (1 µg/L),

DOC (0.05 mg/L), TDN (0.035 mg/L). Stream chemistry data was also obtained from the same streams as this study during June 2011(Parham et al., 2013) and 2013 (Diemer et al., 2015).

Those samples were also collected and analyzed as described here.

Optical Properties:

All grab samples across chronosequence from July 2013, June 2016, and July 2018 and background nutrient pulse samples from June 2016 and July 2018 were analyzed for DOM optical properties. UV absorbance was measured with a Shimadzu photo diode array detector with HPLC at 1 nm intervals between 200 to 700 nm. Specific UV absorbance at 254 nm

(SUVA254) was determined by dividing the UV absorbance at 254 nm by DOC concentration, where SUVA is an index of DOM aromaticity (Weishaar et al., 2003). Fluorescence index (FI) was determined as the ratio of 470 and 520 nm fluorescence intensities at 370 nm excitation and

FI is used to identify sources of DOM (terrestrial vs microbial) (Cory & Mcknight, 2005).

Humification index (HIX) was determined as the ratio between the area under the emission spectra 435–480 nm and the sum of the peak area 330-345 nm and 435-490 nm and describes the degree of humification of DOM (Ohno, 2002). Spectral slope (S) was determined using a non- linear fit of an exponential function to the log transformed absorption spectrum in the ranges of

275–295 nm and 350–400 nm (Helms et al., 2008). Slope ratio (SR) was determined as the ratio of the slope (275-295 nm) and slope (350-400 nm) which have been negatively correlated to

DOM molecular weight and aromaticity (Helms et al., 2008).

Excitation emission matrices (EEM) were conducted on room temperature water at excitations ranging from 240 to 450 nm at 1 nm intervals and emission ranging from 350 to 550 nm at 2.5 nm intervals using a Jobin Yvon Horiba Fluoromax-3 fluorometer. For 2016 samples

EEM emission range was extended from 300 to 600 nm, but for purposes of Parallel Factor

Analysis (PARAFAC) modeling these were clipped to match samples from previous years.

PARAFAC was performed on 132 samples of stream water and soil pore water collected in 2011, 2013, and 2016. PARAFAC is a multivariate technique that decomposes fluorescence spectra into individual fluorescence components (Stedmon & Bro, 2008). Modeling of

PARAFAC components followed established procedures and was normalized to reduce collinearity with reversal of the normalization prior to exporting the model (Murphy et al., 2013).

Prior to PARAFAC modeling, data in the region of first-order and second-order scatter was removed.

FT-ICR MS Analysis:

Samples analyzed for FT-ICR MS were collected in June 2016 and July 2018 only. DOM was obtained for FT-ICR MS by a solid phase extraction (SPE) method (Dittmar et al., 2008).

Briefly, each sample was acidified to pH 2 and passed through a Bond Elute PPL (Agilent

Technologies) cartridge. Residual salts were rinsed from the cartridge with acidified (pH 2)

Milli-Q water. The stationary phase was then dried with a stream of N2 and the DOM was eluted with 100% MeOH (JT Baker) to a final concentration of 50 µg/mL C and stored at 4°C prior to analysis. Negatively charged ions were produced via direct infusion electrospray ionization at a flow rate of 700 nL/min and analyzed with a custom-built FT-ICR MS equipped with a 21-tesla superconducting magnet (Smith et al., 2018).

Each signal >6σ RMS baseline noise was assigned a molecular formula with

EnviroOrg©™ software (Corilo, 2015) developed at the National High Magnetic Field

Laboratory. The mass measurement accuracy did not exceed 200 ppb. Each molecular formula was classified based on stoichiometry; condensed aromatic (modified aromaticity index (AImod)

≥ 0.67), polyphenolic (0.67 > AImod > 0.5), unsaturated, low oxygen (AImod < 0.5, H/C < 1.5, O/C

< 0.5), unsaturated, high oxygen (AImod < 0.5, H/C < 1.5, O/C ≥ 0.5), aliphatic (H/C ≥ 1.5, N =

0), peptide-like (H/C ≥ 1.5, N > 0) (Koch & Dittmar, 2006; Spencer et al., 2014). Although FT-

ICR MS allows for the precise assignment of molecular formulae to peaks that may represent multiple isomers, they describe the underlying molecular compounds comprising DOM, therefore the term compound may be used when describing the peaks detected by FT-ICR MS.

Signal magnitude was converted to relative abundances by dividing the signal magnitude of an individual peak by the sum of all assigned signals. Subsequently, the percent contribution of each compound category was calculated as the percent that the relative abundance in each compound category contributed to the summed abundance of all assigned formulae.

Statistical Analyses:

Normality was tested for all parameters using the Shapiro-Wilk normality test (α≤0.05).

Differences across parameters that follow the assumptions of normality were determined using a one-way ANOVA and nonparametric Kruskal-Wallis rank sum test for parameters that violate the assumptions of normality. We used simple linear regressions to determine relationships between DIN Vf and DIN concentrations, stoichiometric ratios, and DOM compound groups.

DIN concentrations and DOC:DIN ratios were log transformed to normalize data. To determine if watershed characteristics were potential predictors of patterns in stream chemistry across the fire chronosequence, we used stepwise multiple regression analysis with backward selection.

Watershed characteristics (area, elevation, slope, aspect) and years since the last fire were loaded as independent variables and DOC, DON, DIN, DOC:DIN, combined aliphatics and N- containing aliphatics relative abundance, and slope ratio as the dependent variables. Final models were used as those with lowest AIC scores. For all statistical analyses α=0.05 and were conducted in RStudio (version 1.2.1335, RStudio, Inc. Team, Boston, MA 2016) using the stats

(R-Core-Team, 2016), rcompanion (Mangiafico, 2017), and MASS (Venables & Ripley, 2002) packages. The PARAFAC model was validated using split-half analysis using the drEEM toolbox (Murphy et al., 2013) in MATLAB (MATLAB R2015b; The Mathworks, Natick, USA,

2008).

Results and Discussion

Fires altered concentrations of DOC, dissolved organic nitrogen (DON), and nutrients in streams of the CSP. Concentrations of DOC (Fig 2.2A; p=5.87x10-9, F=14.33, df=4) and DON

(Fig 2.2B; p=2.78x10-5, F=7.61, df=4) both increased with longer YSF, with patterns consistent

in June and July. Similar responses between June and July reflect consistency across changing flow regimes shifting from run-off dominated spring freshet in early June and flows in July that are more representative of low-flow conditions typical of summer. In recently burned sites DOM inputs to the stream were likely reduced due to the combustion of the organic layer, which reduces soil standing stocks of C and production of DOM from soil OM (Kawahigashi et al.,

2011). Post-fire declines in stream water DOC and DON concentrations can also be attributed to deepening of the active layer, which exposes mineral soil (Frey & McClelland, 2009; Schuur et al., 2008) leading to increased adsorption of DOM (Frey & McClelland, 2009; Prokushkin et al.,

2007; Striegl et al., 2005). Changes to concentrations of DOC and DON across the chronosequence indicate an ecosystem recovery time of approximately 50 years post wildfire disturbance, during which time the vegetation (moss, lichens, trees) undergoes regrowth reestablishing OM inputs into the soil. DOC concentrations in central Siberia exhibit a positive relationship with forest productivity and can serve as a proxy for forest C stocks (Prokushkin et al., 2007, 2011). The stabilization of DOC and DON concentrations after 50 YSF suggests the reestablishment of an equilibrium as described by resilience theory and the ecosystem succession and nutrient retention hypothesis (Holling, 1973; Vitousek & Reiners, 1975).

Stream dissolved inorganic nitrogen (DIN) responded differently to forest fires than

- DOM. NO3 concentrations were highest in the most recently burned sites and decreased with increasing YSF (Fig 2.2C, p<<0.05); consistent responses to fire were observed in both June and

- July. Multiple mechanisms have been proposed to explain the elevated NO3 concentrations often observed in recently burned watersheds on permafrost, such as an increase in soil aeration from improved drainage after fire leading to declines in soil denitrification (Betts & Jones, 2009), increases in mineralization-nitrification of organic nitrogenous compounds (Mroz et al., 1980)

and decreased nutrient uptake due to the reduction in biomass (Bayley et al., 1992; Jiang et al.,

- 2015). In contrast to DOC and DON, concentrations of NO3 demonstrate a shorter recovery time, returning to their pre-disturbance concentrations approximately at 10 years. In larch forests post-fire, growth of the moss layer and OM storage is very slow (Prokushkin et al., 2006) but there is a greater demand for inorganic N from growing shrubs and larch seedlings, which

- require fire to propagate (Prokushkin et al., 2018, 2006). This relatively short recovery of NO3 in the CSP is consistent with earlier studies, which have found recovery intervals of 5-6 years after fire for inorganic N in other boreal streams (Bayley et al., 1992). Since the recovery of

- DOC and DON concentrations lags behind that of NO3 , it appears that the processes which produce, and export DOM serve as the rate-limiting step in determining the resilience and stability of heterotrophic processes in these watersheds. As a result, these CSP streams remain in a state of relatively higher energy limitation compared to their pre-disturbance steady-state as described by molar DOC:DIN ratios (Fig 2.2D; p=0.9.78x10-5). In the study streams, DIN is

- + mostly composed of NO3 where ammonium (NH4 ) is often at or below detection limit (~5

µg/L). During the post-fire interval when DIN is elevated and DOC is reduced, this balance between energy and nutrient demand (as measured by stoichiometry) is re-established after approximately ten years reflecting a system that returns to a state of inorganic-N limitation and high DOM availability. During this initial recovery period when DON concentrations are relatively low, the predominate in-stream nutrient source may shift from organic to inorganic forms of N. There was no statistically significant difference in ammonium and phosphate concentrations across the burn gradient (Table 2.2).

The effects of fire on landscape recovery are also reflected in DOM composition. As surface flow paths change with landscape recovery, exported OM shifts in quantity, composition,

and bioavailability from microbial-derived low molecular weight DOM from deep soils to less processed and more aromatic DOM derived from surface flow paths (Mann et al., 2012). This is reflected in the spectral slope ratios (SR) shifting from low molecular weight to higher molecular weight values with increasing time since fire (Fig 2.3A, p=0.002). Aliphatic compounds

(including N-containing aliphatics) increased with increasing YSF (Fig 2.3B; p=0.04, F=2.96, df=3) while polyphenolic and condensed aromatic compounds decreased after 60 YSF although the change was not statistically significant (Fig 2.3C p=0.07, F=2.49, df=3; Fig 2.3D p=0.18,

F=1.72, df=3). With the prolonged absence of fire in boreal forests, decomposition rates decline thus promoting greater storage of OM in soils (Wardle et al., 2003), a change in watershed C dynamics that is reflected in the increase of aliphatic compounds. Condensed aromatic compounds which are related to black carbon derived from combustion and charred material

(Stubbins et al., 2014) were relatively low across our study watersheds, similar to others that have also found no clear relationship between black carbon in streams and fire history (Ding et al., 2015). DOM composition as assessed by ultrahigh-resolution mass spectrometry reflects the recovery of the adjacent landscape, patterns that are not evident using DOM optical properties only (Table 2.3) or the four component PARAFAC model that describes mostly humic-like

DOM (Table 2.4).

Biogeochemical regimes in stream ecosystems reflect multiple physical, chemical, and biological processes occurring at the watershed scale as well as underlying geology and climate.

Wildfire disturbance interacts with each of these drivers to determine stream chemistry and microbial biogeochemical processes. To test hypotheses regarding the potential role of other landscape scale drivers we used a backward stepwise regression to relate stream chemistry and

DOM composition (parameters that demonstrated significant changes across the burn gradient)

to watershed characteristics (area, slope, elevation, aspect, and YSF). Across stream chemistry models YSF consistently was one of the best predictors of DOC, DON, and DIN concentrations, as well as DOC:DIN ratios. The best predictor of SR was watershed slope, but the relative abundance of aliphatic compounds was not predictable with landscape characteristics (Table

2.5). Time since wildfire disturbance often emerges as a strong predictor variable of stream chemistry even when additional landscape variables are included (Santos et al., 2019).

The response in stream chemistry due to fires influences the processing and export of

DIN. Across all watersheds, DIN uptake velocities (Vf) correlated negatively with DIN concentrations (Fig 2.4A, r2=0.26, p=0.005). This pattern has been observed in numerous studies in a variety of biomes, suggesting that the efficiency with which the microbial community removes DIN decreases with increasing DIN concentrations (Dodds et al., 2002; Ensign &

Doyle, 2006; Peterson et al., 2001). The changes in concentrations of DOC due to fires also influence DIN processing where streams with greater DOC:DIN molar ratios supported the greatest DIN removal (Fig 2.4B r2=0.39, p=0.0004). Relationships between uptake and molar ratios (Rodríguez-Cardona et al., 2016; Wymore et al., 2016) demonstrate the important role of

DOM in DIN processing and indicate that the balance and interactions between DOM and DIN can influence the resilience of heterotrophic processes. Moreover, DOM compound groups such as polyphenolics, can enhance DIN removal (Fig 2.4D, r2=0.36, p=0.004); increasing relative abundance of polyphenolics leads to higher DIN uptake velocities. Alternatively, greater availability of DIN could have enhanced DOM decomposition by relieving any DIN limitations that inhibit the breakdown of aromatic molecules like polyphenolics. Although nutrient pulse additions do not allow us to determine what specific processes are responsible for DIN removal from the water column, the data demonstrate a close relationship between DOM and DIN uptake,

supporting the idea that these patterns are due to heterotrophic or assimilative processes that require an energy source such as DOM. With shorter fire return intervals, elevated DIN concentrations, and declines in DOM, and DOC:DIN ratios we can expect greater export of DIN as the strength of watersheds as sinks is greatly reduced leading to larger amounts of nutrients reaching larger rivers and ultimately the Arctic Ocean. DOC:DIN ratios provide a valuable integrator of energy versus nutrient limitation in stream ecosystems, in which energy limitation

(low DOC:DIN ratios) or nutrient limitation (high DOC:DIN ratios) can be important in determining the delivery of nitrogen and OM to downstream ecosystems (Rodríguez-Cardona et al., 2016; Wymore et al., 2016). Downstream exports of inorganic N will likely persist for several years to a decade post fire, until DOM inputs from the watershed are re-established.

Across arctic watersheds, the response of stream chemistry to biomass burning is not consistent. Watershed recovery following wildfire across arctic watersheds appears to be a function of many variables, including composition of vegetation and rates of regrowth, nature and extent of OM consumed by fire, and the effects of both of these on active layer depth (Betts

& Jones, 2009; Burd et al., 2018; Frey & McClelland, 2009; Kawahigashi et al., 2011;

Prokushkin et al., 2018). The extent to which our findings on the CSP can be applied across the

Arctic is unknown. The majority of previous studies include only the assessment of stream chemistry before and after a fire (Betts & Jones, 2009; Larouche et al., 2015; Petrone et al.,

2007), making predictions of long-term recovery and resiliency that we have developed using a chronosequence approach difficult due to a paucity of data. Differences in the response to disturbance are likely driven by the role that permafrost structure (i.e. mineral soil layer vs. deep soil OM) and its continuity, hydrology, and vegetation communities (i.e. tundra vs. taiga) play in regulating elemental fluxes and the balance of energy and nutrient demand in arctic fluvial

networks. Predictive models of how arctic watersheds respond to fire disturbance are unlikely to be universally applicable across the pan-Arctic, but changes such as those that we have documented are likely relevant at many other similar arctic sites. A better understanding of the effects of large-scale perturbations like fires is important to develop a greater mechanistic understanding of how site conditions drive variability in responses, exports, and resiliency to increasing alteration of arctic landscapes.

Tables

Table 2.1 Watersheds sampled in 2011, 2013, 2016, and 2018 with their respective burn year and watershed size N11 is currently the only site with two recorded burns, in 1899 and 2013.

Watershed area Site Year Burned Years sampled (km2) N1 1947 68.7 2013, 2016 N2 1947 254.3 2013, 2016, 2018

N3 1960 67.6 2013, 2016

N4 1960 3.7 2011, 2013 N5 1960 91.68 2016 N6 1960 15.4 2011, 2013, 2016 N7 1960 15.6 2011 N8 1993* 89.1 2011, 2015 N9 1993 20.8 2011, 2013, 2016, 2018 N10 1993 70.5 2011, 2016 N11 1899, 2013 4.9 2013, 2016, 2018 N13 2009 6.9 2013 N14 2009 9.6 2013, 2016 N15 2009* 36 2016 N19 1896 52.2 2016, 2018 N20 1896 3.8 2016, 2018 N10A 1816 40.38 2016

Years + -3 since NH4 PO4 the last (µg/L) (µg/L) fire June July June July 5.23 a 5.35 13.15 a 7.13 10 (1.10 (1.90) (2.19) (2.73) ) 7.71 a 2.93 8.35 a 4.25 25 (2.20 (0.40) (1.42) (1.43) ) 8.87 a 2.78 6.30 a 2.96 50 (1.89 (0.28) (1.09) (0.45) ) 11.52 a 11.00 a 60 - - (5.81 (1.29) ) 10.26 a 5.98 12.33 a 9.02 >100 (6.02 (1.13) (3.66) (3.41) ) p-value 0.52 0.38 0.12 0.20

df 4 3 4 3 – represents no data available

Table 2.3 Mean DOM optical properties across the burn gradient for streams sampled June 2013, 2016, and 2018. July values are only from 2018, with low n. HIX values are from June 2016 and July 2018 only. 10 represent sites burned between 3 and 7 yrs, 25 watersheds burned between 18 and 25 yrs, 50 watersheds between 51 and 57 yrs, 60 watersheds burned between 66 and 71 yrs, and >100 watersheds burned up to 122 years ago. Parentheses represent standard error and superscript represent statistical difference between groups (α=0.05). Respective n-values: SUVA254 June:13,10,13,4,8, July:1,1,1,3; FI June:13,8,13,4,7, July:1,1,1,3; HIX June: 7,6,7,2,3, July: 1,1,1,3; S275-295 June: 11,9,12,4,7, July: 1,1,1,3; S350-400 June:11,9,12,4,7, July: 1,1,1,3; SR June:11,9,12,4,7, July: 1,1,1,3. Statistical analyses were excluded from July data due to low n. – represents no data available.

Years since SUVA254 S275-295 S350-400 FI HIX SR the last fire (L mg C-1 m-1) (nm-1) (nm-1)

June July June July June July June July June July June July

3.88 a 2.69 1.31 a 1.35 0.963 a 0.0137 a 0.0152 0.0173 a 0.0184 0.79 a 0.82 10 0.82 (0.11) (0.007) (0.002) (1.74x10-4) (1.15x10-4) (0.006) 3.94 a 2.91 1.29 a 1.32 0.964 a 0.0134 a 0.0157 0.0173 a 0.0182 0.78 ab 0.86 25 0.86 (0.10) (0.011) (0.002) (2.15x10-4) (9.96x10-5) (0.010) 4.03 a 1.29 a 0.964 a 0.0132 a 0.0174 a 0.75 b 50 ------(0.10) (0.008) (0.002) (1.50x10-4) (7.22x10-5) (0.005) 4.44 a 2.66 1.30 a 1.34 0.960 a 0.0129 a 0.0170 0.0170 a 0.0184 0.76 ab 0.92 60 0.92 (0.29) (0.016) (0.000) (8.63x10-5) (5.29x10-5) (0.008) 3.96 a 3.45 1.30 a 1.31 0.964 a 0.0131 a 0.0144 0.0173 a 0.0189 0.79 b 0.76 >100 0.78 (0.08) (0.012) (0.001) (1.59x10-4) (1.17x10-4) (0.007)

p-value 0.12 - 0.27 - 0.61 - 0.04 - 0.17 - 0.002 -

F-value 1.95 - - - 0.67 - - - 1.70 - - -

df 4 - - - 4 - - - 4 - - -

Table 2.4. A four component PARAFAC model for samples from 2016 only.

Component Excitation Emission Description Literature reference 1a Stedmon & Markager, 1 <250 (345) 462 UVC humic-like 2005 FH1; Singh et al., 2013

UVA humic-like: expected to be C5; Kothawala et al., 2 <250 478 photoresistant & smaller molecular 2013 size compounds C1 Ishi and Boyer 2012

UVC humic-like and UVA marine humic-like; expected to be 3; Stedmon & Markager, 3 <250 (307.5) 404 photodegradaded by UVA but not 2005 as much as C4 and moderate C3; Ishi and Boyer 2012 molecular size between C2 and C4

UVC humic-like and UVA humic- like; expected to be photodegraded C3 Kothawala et al., 2013 4 280 510 by UVA and large molecular size 2 Ishi and Boyer 2012 hydrophobic compounds

Table 2. 5 Summary of the final backward step wise regression model for DOC, DON, and DIN concentrations, DOC:DIN molar ratios, relative abundance of aliphatics (with N-containing aliphatics), and slope ratio (SR) as the dependent variables. The initial model for all included watershed area (km2), elevation (m), watershed slope, percent of watershed facing north, percent of watershed facing south, and years since the last fire (YSF) as independent variables. Bold values highlight statistically significant relationships (α < 0.05). * denote statistically significant models and – represents variables that were dropped from the final model.

Initial Final Area Elevation Slope % North % South YSF Model Model (km2) (m) AIC R2 AIC R2 β p-value β p-value β p-value β p-value β p-value β p-value DOC 259.27 0.33* 257.35 0.33* -0.03 0.011 0.03 0.059 - - 0.16 0.015 0.23 9.54x10-4 0.06 4.84x10-7 DON -414.55 0.22* -416.33 0.23* -0.0007 0.031 0.0008 0.034 - - 0.002 0.111 0.003 0.015 0.001 1.29x10-4 DIN -575.94 0.22* -577.88 0.23* -1.83x10-4 0.054 - - 1.54x10-2 3.60x10-4 -1.12x10-3 0.047 -1.231x10-3 0.045 -2.23x10-4 0.037 DOC:DIN 1277.06 0.07 1273.08 0.07* - - - -272 0.055 - - - - 8.90 0.018 Aliphatics 105.51 -0.03 98.32 -0.01 - - - 0.37 0.458 ------

SR -244.48 0.07 -251.12 0.11* - - - -0.001 0.009 ------

Figures

Figure 2.2 DOM, nutrient concentrations, and stoichiometric ratios across the burn gradient. - Boxplot panels represent (a) DOC, (b DON, (c NO3 concentrations, and (d) molar DOC:DIN ratios, across 17 streams sampled 2011, 2013, and 2016 - 2017 during June and July across the burn gradient. The paired boxplots correspond to June (dark shade) and July (lighter shade). Boxes represent interquartile range with the median value as the bold line, whiskers represent 1.5 interquartile range, and points are possible outliers. Letters denote significant differences (α=0.05) where uppercase correspond to June and lowercase to July. Note that 60 YSF in July was excluded from statistical analysis due to low n. Statistics for DOC p=1.8x10-8, F=14.8, df=4 June and p=8.3x10-4, F=8.56, df=4 July); DON p=0.0002, F=6.53, df=4 June and p=0.005, F=5.81, df=4 July; DIN p=0.008 June, p=0.04 July; DOC:DIN p=0.0008 June and p=0.06 July. Respective n-values across the burn gradient for June: 16, 16,19, 4, 10; July: 5, 5, 3, 1, 10.

Figure 2.3 DOM composition of streams across the burn gradient. Boxplots represent (a) slope ratios (SR) from June 2013 and 2016, relative abundance of (b) aliphatic compounds including N- containing aliphatics, (c) polyphenolic compounds, and (d) condensed aromatic compounds. Streams burned 60 years ago for FT-ICR MS were excluded from statistical analyses due to low n. Boxes represent interquartile range with the median value as the bold line, whiskers represent 1.5 interquartile range, and points are possible outliers. Uppercase letters denote significant differences (α=0.05). Only data from June 2016 for FT-ICR MS compound groups and June 2013 and 2016 for SR. Respective n-values across the burn gradient SR: 11, 9, 12, 4, 7; FT ICR: 11, 8, 9, 2, 6.

Supplementary Information

Supplementary Table 2.1 Details on individual nutrient pulse additions per site from 2016 through 2018 which includes additions where NH4 was added with PO4, NH4 only, and NO3 only but here we only present uptake metrics for NH4 and NO3 only. The experimental reach length, average stream width and depth, discharge (Q), added mass of sodium chloride (NaCl), sodium nitrate

(NaNO3), and ammonium sulfate ((NH4)2SO4), and uptake metrics for each addition as uptake length (Sw), uptake velocity (Vf), areal uptake (U). * represent undetectable uptake, # represent dates where stream depth measurements could not be taken, and – represent that the given solute was not added for that pulse addition.

Reach U Added Uptake Width Depth Q NaCl NaNO3 (NH4)2SO4 Sw Vf Site Date length (ug m2 Solutes Nutrient (m) (cm) (L/s) (kg) (g) (g) (m) (mm/min) (m) min)

N11 6/5/16 NH4+PO4 NH4 100 2.69 16.78 148.09 3.90 NA 249.6 557.76 5.92 14.81

N11 6/5/16 NO3 NO3 100 2.69 16.78 148.09 4.12 17.2 - * * *

N9 6/6/16 NH4+PO4 NH4 190 4.13 34.2 1041.13 10.00 NA 183 501.73 30.15 75.37

N9 6/6/16 NO3 NO3 190 4.13 34.2 1041.13 12.00 328 - 2182.55 6.93 21.14

N6 6/7/16 NH4 NH4 40 3.6 # 494.92 10.00 - 45 384.58 21.45 53.62

N6 6/7/16 NO3 NO3 132 3.6 # 494.92 11.00 41 - 201.60 40.92 81.83

N20 6/8/16 NH4+PO4 NH4 100 2.06 18.25 110.4 3.42 - 70.2 386.44 8.32 20.80

N20 6/8/16 NO3 NO3 100 2.06 18.25 110.4 6.09 56.1 - 973.30 3.30 23.62

N11 6/11/16 NH4+PO4 NH4 100 1.78 15 25.2 3.02 - 267.6 221.14 3.84 9.60

N11 6/11/16 NO3 NO3 100 1.78 15 25.2 3.00 13.6 - 252.61 3.36 393.76

N9 6/12/16 NH4+PO4 NH4 90 3.24 19.16 172 8.40 NA 178.3 * * *

N9 6/12/16 NO3 NO3 190 3.24 19.16 172 9.04 342.3 - * * *

N6 6/13/16 NH4+PO4 NH4 40 3.85 40 181.17 2.65 NA 17.2 165.68 17.04 42.60

N6 6/13/16 NO3 NO3 132 3.85 40 181.17 2.89 7.5 - * * *

N20 6/14/16 NH4+PO4 NH4 100 1.64 17.39 34.07 1.52 NA 20 217.36 5.73 14.34

N20 6/14/16 NO3 NO3 100 1.64 17.39 34.07 1.57 18.1 - * * *

N11 6/17/16 NO3 NO3 100 1.4 9.97 4.11 0.45 3.9 - * * *

N20 6/19/16 NO3 NO3 65 1.06 14.83 8.09 0.62 9.5 - * * *

N11 6/4/17 NH4+PO4 NH4 100 1.66 0.09 29.45 2.00 - 169.7 518.58 2.05 19.58 76

N11 6/4/17 NO3 NO3 100 1.66 0.09 29.45 2.00 73 - * * *

N4 6/6/17 NH4+PO4 NH4 100 1.60 0.17 33.70 2.00 169.7 221.35 5.71 49.42

N4 6/6/17 NO3 NO3 100 1.60 0.17 33.70 2.00 72.9 - * * *

N11 6/19/17 NH4+PO4 NH4 100 1.50 0.13 19.16 0.75 - 106.2 222.37 3.45 32.87

N11 6/19/17 NO3 NO3 100 1.50 0.13 19.16 0.75 27.3 - * * *

N11 7/19/17 NH4+PO4 NH4 38 0.46 0.09 0.79 0.05 - 6.4 34.69 2.97 35.59

N11 7/19/17 NO3 NO3 38 0.46 0.09 0.79 0.05 4.1 - * * *

N9 7/22/17 NH4+PO4 NH4 40 1.43 0.06 3.143 0.06 - 27 17.85 7.38 31.36

N9 7/22/17 NO3 NO3 40 1.43 0.06 3.143 0.06 10.9 - 392.50 0.34 2.45

N20 7/23/17 NH4+PO4 NH4 29 0.58 0.04 0.866 0.06 - 12.7 130.48 0.70 1.76

N20 7/23/17 NO3 NO3 29 0.58 0.04 0.866 0.06 16.4 - * * *

N11 7/11/18 NO3 NO3 40 1.51 # 3.00 0.02 2 - 28.64 4.16 440.96

N11 7/12/18 NH4 NH4 30 1.52 # 2.00 0.03 - 0.5 * * *

N20 7/15/18 NH4 NH4 100 1.69 16.52 51.18 0.38 - 3.8 130.15 14.04 75.39

N20 7/15/18 NO3 NO3 100 1.69 16.52 51.18 0.30 12.2 - 350.85 5.21 13.42

N20 7/15/18 NO3 NO3 100 1.69 16.52 51.18 0.22 9.5 - 187.63 9.74 24.84

N9 7/16/18 NO3 NO3 21.4 1.32 4.29 0.869 0.05 2.8 - 62.05 0.64 19.61

N9 7/17/18 NH4 NH4 15 1.48 3.08 0.457 0.05 - 3.4 5.96 3.12 10.64

N2 7/19/18 NO3 NO3 77.2 8.07 9.95 37.56 0.23 10.9 - 27.66 10.11 129.14

N2 7/20/18 NH4 NH4 77.2 8.57 9.69 34.15 0.41 - 77.5 30.21 7.92 26.55

N19 7/24/18 NH4 NH4 60 2.45 14.86 19.94 0.40 - 3.8 161.69 3.02 7.55

N19 7/24/18 NO3 NO3 60 2.45 14.86 18.35 0.40 16.5 - * * *

N20 7/25/18 NH4 NH4 86.5 1.54 12.11 13.97 0.20 - 1.9 70.10 7.33 16.13

N20 7/29/18 NO3 NO3 86.5 1.59 10.79 9.48 0.20 8.7 - * * *

CHAPTER THREE

NITRATE UPTAKE ENHANCED BY THE AVAILABILITY OF DISSOLVED ORGANIC

MATTER IN TROPICAL MONTANE STREAMS

Abstract

Tropical forests store large amounts of Earth’s terrestrial carbon (C), but many tropical montane streams have low concentrations of dissolved organic carbon (DOC). With high rates of

- nitrification and often abundant nitrate (NO3 ), energy availability likely limits certain pathways of inorganic nitrogen (N) uptake. To test this hypothesis, we conducted a series of experiments to explore the influence of DOC availability on tropical stream N cycling. Nutrient pulse additions

- of NO3 with or without an added C source (acetate or urea) were conducted in streams of the

- Luquillo Experimental Forest, Puerto Rico. In the absence of added DOC, NO3 uptake was either undetectable or had very long (>3,000 m) uptake lengths (Sw). When DOC was added with

- - NO3 , NO3 Sw were much shorter (80 to 1,200 m), with the shortest lengths resulting from additions of acetate. Comparing uptake metrics of the added C sources, there was greater demand

- for acetate compared to urea. During additions of NO3 only, ambient concentrations of DOC and dissolved organic nitrogen (DON) often decreased, suggesting increased metabolic demand for energy from the ambient organic matter pool under elevated nutrient levels. Collectively, these results demonstrate that pathways of inorganic nitrogen cycling are energy limited at this tropical

- site and that availability of DOC enhances NO3 removal. The response of ambient DOC and

- DON to increases in NO3 points to important feedbacks between inorganic nitrogen and

- dissolved organic matter including organic nitrogen. Understanding the controls on NO3

processing in these streams is important to predict network scale exports of nitrogen from tropical ecosystems.

Introduction

Tropical forests store large amounts of Earth’s terrestrial carbon (C) in vegetation

(Bunker et al., 2005) and soil (Drake et al., 2019), but concentrations of dissolved organic carbon

(DOC) are often low in montane tropical streams (McDowell & Asbury, 1994). Inputs of DOC to streams are controlled by the magnitude of source terms in soil solution and the extent to which hydrologic flow paths route water through mineral soils that retain DOC through adsorption (McDowell & Likens, 1988). In the Luquillo Mountains of Puerto Rico, DOC produced by upper soil horizons is retained very effectively by mineral soils leading to low delivery of DOC to streams (McDowell, 1998). These hillslope and riparian dynamics combined with low rates of in-stream primary production (Ortiz-Zayas et al., 2005; Vannote et al., 1980) can result in energy-limited biogeochemical reactions in streams.

Carbon and nitrogen cycles are interrelated as many N transformations require an energy source. Various studies have demonstrated this coupling between C and N in aquatic ecosystems.

- For example, rates of nitrate (NO3 ) uptake can be significantly increased with the addition of

DOC (Bernhardt & Likens, 2002), a general relationship that holds across multiple study systems

(Bernhardt & McDowell, 2008; Rodríguez-Cardona et al., 2016; Sobczak & Findlay, 2002;

Thouin et al., 2009). There are exceptions, however, with some experimental work finding no relationship between C availability and N uptake (Richey et al., 1985) suggesting that other factors can also limit inorganic N uptake. The relationship between the composition of DOC and

N cycling has also received little attention. Organic matter C:N ratios can be predictive of rates of N transformations, as higher C:N ratios favor heterotrophic uptake of N by inhibiting nitrification (Starry et al., 2005; Strauss & Lamberti, 2000, 2002). It is unclear whether these dynamics, which have been primarily studied in temperate systems, hold in other biomes,

especially tropical ecosystems. In the Luquillo Experimental Forest (LEF), for example, previous work has demonstrated that many biogeochemical transformations of N appear to be driven more by energy availability than nutrient limitation (Koenig et al., 2017; Merriam et al., 2002; Potter

- et al., 2010). Stream NO3 concentrations correlate negatively with peak leaf fall (McDowell &

Asbury, 1994) and nitrification rates are relatively high compared with temperate systems

(Koenig et al., 2017; Peterson et al., 2001) suggesting a link between C and N cycling but direct experimental evidence is lacking.

The objective of this study was to understand the influence of DOC in N processing in

- tropical montane streams. We conducted a series of paired nutrient pulse additions of NO3 with or without an added source of C in streams of the Luquillo Mountains in Puerto Rico. To explore

- the role of organic N content in the interactions between NO3 uptake and added DOC, we also varied the carbon source by providing either acetate (C2H3O2) or urea (CH4N2O). We

- hypothesized that 1) the addition of DOC would increase the uptake rates of NO3 as measured in nutrient addition experiments with urea having a greater influence than acetate; 2) uptake rates of added N-rich DOM would be greater than those for acetate as it can meet both energy and nutrient demands (Wymore et al., 2015); and 3) ambient concentrations of DOC and DON would

- decline with added NO3 as these organic compounds are broken down for their energetic content

- to fuel NO3 uptake from the water column.

Methods

Study Site

Nutrient pulse additions were conducted in two headwater streams of the Río Espiritu

Santo Watershed in the Luquillo Experimental Forest (LEF) located in the northeast corner of

Puerto Rico. We focused on Quebrada Prieta-B (QPB) and Quebrada Toronja (QT) (Fig. 1) which are underlain by volcaniclastic lithology with andesitic igneous materials, mudstone, and clay deposits (McDowell & Asbury, 1994) and dominated by tabonuco (Dacryodes excelsa) and sierra palm (Prestoea montana). Mean annual temperature range between 21-25 ºC and annual precipitation between 250 to greater than 500 cm per year (McDowell & Asbury, 1994; Merriam et al., 2002). Both streams are small, steep, and highly shaded 1st and 2nd order streams with large rocks and boulders. Background concentrations of DOC and DON are around 1mg/L and 50

- µg/L, respectively, while the inorganic N pool is dominated by NO3 with concentrations

+ between 65 -100 µg/L, and NH4 often at or below detection limit (Table 1).

Nutrient pulse additions and uptake metrics

We performed nutrient pulse additions at both sites throughout January 2016. Additions

- of NO3 with and without an added source of DOM were conducted at each site. Nitrate was added in the form of sodium nitrate (NaNO3) with DOC as either potassium acetate, CH3COOK

or urea, CH4N2O. These solutes were added along with sodium chloride (NaCl) as a conservative

- tracer. Additions targeted an increase of Cl, NO3 , DOC, and DON concentrations 2-3x above ambient concentrations.

Experimental reaches ranged between 50 and 75 m and were chosen to avoid large pools with excessively long retention times as well as any tributaries. The solutions were released at the top of the experimental reach and samples were collected at a fixed point at the bottom of the experimental reach throughout the breakthrough curve (BTC). Between 20 and 40 samples were collected per experiment, depending on flow conditions and changes in specific conductance determined with a multiparameter sonde (YSI, Yellow Springs, OH). Duplicate background

samples were collected at the upstream addition site and downstream collection site, for a total of

- four background samples. We performed four NO3 only additions in QT and three in QPB and at

- - both sites three additions of NO3 with acetate and one NO3 with urea (Table 3.2). For additions that were performed back to back on the same date, there was at least one hour between additions once the stream returned to background conditions. Details for individual additions can be found in Table S3.1.

Calculations of uptake metrics follow methods from the Stream Solute Workshop (1990) and the break through curve (BTC) integration method (Tank et al., 2008) where uptake length

(Sw) was determined as the negative inverse of the ratio between the natural logarithm of the fraction of background-corrected solute and chloride (Cl) retrieved and the reach length distance.

Uptake length describes the average distance a molecule travels before being taken up by the biota where longer distances reflect less demand (Stream Solute Workshop, 1990). The mass

transfer coefficient or uptake velocity (Vf) was determined as:

! " !! = (1) ""

where Q is discharge, w stream width, and Sw the uptake length. Uptake velocity is normalized for stream velocity and depth and describes the efficiency with which the benthic community takes up nutrients; higher rates of Vf denote greater demand (Stream Solute

Workshop, 1990). Areal uptake (U) was determined as:

U = Vf * C (2) where C is background solute concentration. Areal uptake describes gross uptake on a per area

- per time basis (Stream Solute Workshops, 1990). Because all but one experiment with NO3

- added alone resulted in undetectable rates of NO3 uptake, the limit of detection was determined for each of these experiments and used as an estimate of Sw so that a maximum value for Vf

- could be calculated. To estimate detection limits for these experiments, NO3 concentrations were iteratively decreased across the BTC by 1-10% until a detectable Sw was obtained. This value, an estimate of the minimum length for Sw that could have been detected in the experiment, was then used to calculate the other uptake metrics for these experiments.

Discharge was determined using the dilution gaging method (Kilpatrick & Cobb, 1985) using NaCl dissolved in stream water and added to the stream as an instantaneous slug addition while conductivity was logged every 5 seconds using a HOBO conductivity data logger (Onset,

Bourne, MA). Uptake of acetate was determined by direct measurement of the acetate ion while uptake of urea was determined by bulk DON concentrations.

Water Chemistry

All water samples were filtered through pre-combusted Whatman GF/F filters and collected in acid washed high density polyethylene bottles (HDPE). Samples were frozen and transported to the Water Quality Analysis Laboratory at the University of New Hampshire for analysis. We analyzed samples for DOC and total dissolved nitrogen (TDN) using high

- temperature catalytic oxidation in a Shimadzu TOC-V CPH/TNM. Cl, NO3 , and acetate were analyzed with ion chromatography using an Anion/Cations Dionex ICS-1000 with AS40

+ autosampler. Ammonium (NH4 ) analyses were done using a SmartChem 200 discrete automated colorimetric analyzer using the alkaline phenate standard method. DON was determined

- + arithmetically by subtracting DIN (sum of NO3 and NH4 ) concentrations from TDN.

Results

- NO3 Uptake

- As expected, the addition of a DOM source enhanced NO3 uptake across both study

- streams. When added without a DOM source, NO3 uptake was generally undetectable (Fig 3.2).

- Only one of the seven NO3 -only additions (QP) had detectable uptake where Sw was 3,800 m, Vf

-1 -2 -1 - 0.05 mm min , and U 4.36 µg m min (Table 3.2). Undetectable uptake metrics during NO3 - only additions were recalculated to determine detection levels, which resulted in Sw between

-1 -2 -1 3,263 and 6,558m; Vf between 0.016 and 0.084 mm min ; and U 1.96 and 6.85 µg m min all

- - still reflecting low NO3 demand. When NO3 was added along with a DOM source Sw decreased to 80 – 1,254 m (mean = 487 m) with acetate and 148 – 1,417 m (mean = 782 m) with urea

(Table 3.2). Accordingly, uptake velocity increased to 0.16 - 1.74 mm min-1 with acetate and

0.07 - 1.72 mm min-1 with urea while U increased to 13.24 - 156.95 µg m-2 min-1, and 10.07 -

-2 -1 - 256.28 µg m min , respectively (Table 3.2). The increased rates of NO3 uptake with added

DOM were statistically significant with either acetate or urea (Sw p=0.015, Vf p=0.025, and U p=0.017).

DOM Uptake

Acetate had greater uptake relative to urea. This pattern held across experiments and

study streams. Acetate exhibited shorter Sw and higher Vf and U than urea (Fig 3.3). Uptake length for acetate varied between 44 -169 m across sites while only one of the two urea additions

had detectable uptake with a Sw of 1,546 m (Table 3.2). Uptake velocity for acetate was between

-1 -1 1.09 – 3.34 mm min while urea had a Vf of 0.16 mm min (Table 3.2). Areal uptake for acetate ranged between 30 – 162.3 µg m-2 min-1 and U for urea was 1.80 µg m-2 min-1 (Table 3.2).

Ambient DOC and DON dynamics

- We detected declines in ambient concentrations of DOC and DON with NO3 -only additions (Fig 3.4). We detected significant changes 30% of the time (2 out of 7 experiments) and concentrations of DOC and DON did not always respond in synchrony. For example, during

- th the NO3 -only addition on January 4 at QT, there was no relationship between concentrations of

- - DOC and NO3 but DON and NO3 showed a statistically significant negative relationship

2 - (r =0.41, p<<0.05) (Fig 3.4A and B). Only a week later, however, relationships between NO3

- and DOC or DON were not observed (Fig 3.4C and D). At QPB, NO3 had a statistically significant negative relationship with DOC (r2=0.86, p<<0.05) and not DON (Fig 3.4E and D).

- All relationships between DOC, DON, and NO3 can be found in Figure S3.1 and S3.2.

Discussion

- Our results indicate that the removal of NO3 from tropical montane streams is highly energy limited. Rates of nitrification are high (Koenig et al., 2017; Merriam et al., 2002) and

- rates of NO3 uptake from the water column are low, as we have demonstrated in this study and has been demonstrated previously (Hall et al., 2009; Mulholland et al., 2009). Our experimental

- - data provide strong support for an energy limitation to NO3 uptake, as NO3 uptake greatly increased with the addition of DOC (Fig 2). The higher Vf for C-rich over N-rich DOM provides further support for the conclusion that energy density, rather than N availability, drives rates of

DOM uptake in these streams.

NO3 Uptake

- Due to a lack of available energy, uptake of NO3 in streams of the LEF is very low to undetectable. The undetectable nature of NO3 uptake in both QT and QPB agrees with previous

+ + studies in the LEF where nitrate uptake was lower than NH4 uptake, where NH4 uptake is driven to a large extent by nitrification (Koenig et al., 2017; Merriam et al., 2002; Potter et al.,

- 2010). We performed 7 individual NO3 -only additions but were only able to obtain uptake metrics for one of them, with an uptake length of 3,800 m. This Sw is considerably longer than the reach length, suggesting that uptake lengths in streams of LEF are almost infinite,

- demonstrating the low demand for NO3 that makes it very difficult to determine uptake metrics

- - - in these streams. All NO3 uptake values reported here (NO3 -only and NO3 with a carbon source) fall within the range, but toward the low end, of reported values from the second Lotic

Intersite Nitrogen eXperiment (LINX II) (Hall et al., 2009). High residence time of many

- temperate streams promotes removal of NO3 from the water column (Brookshire et al., 2005;

Valett et al., 1996; Zarnetske et al., 2011), a characteristic not applicable to streams in the LEF due to their steep nature and short travel distances from headwaters to the ocean (Koenig et al.,

2017; Lloret et al., 2011; Merriam et al., 2002). And while gross primary productivity correlates

- positively with NO3 uptake (Hall & Tank, 2003; Lupon et al., 2016; Mulholland et al., 2008),

- streams within the LEF are characterized by low P:R ratios, further reducing rates of NO3

- uptake. Low NO3 demand was also found in other tropical sites including Brazil where there was

+ - greater retention of NH4 than NO3 (Finkler et al., 2018; Tromboni et al., 2018). Greater

+ retention of NH4 is not exclusive to tropical sites, however. Ammonium is considered the biologically preferred form of DIN in streams because of its lower anabolism costs relative to

- NO3 (Martí & Sabater, 1996; O’Brien & Dodds, 2010; Ribot et al., 2013).

- The relationships between NO3 uptake and DOM demonstrate the tight linkage between

C and N cycling in low-DOM tropical aquatic systems, as has been well-documented in

- temperate systems. This suggests that the controls of NO3 dynamics may be similar across biomes. Nitrate uptake was only detectable when additional DOM was available. The increase in

- NO3 uptake during the paired additions with DOM demonstrates strong energy limitations in

- these streams that inhibits the heterotrophic or assimilative removal of NO3 from the water

- column. Similarly, others have found an increase in NO3 uptake during paired stream

- manipulations of NO3 and a form of carbon (Bernhardt & Likens, 2002; Thouin et al., 2009;

- Tomasek et al., 2018) or in systems with higher ambient DOC concentrations and DOC:NO3 ratios (Rodríguez-Cardona et al., 2016; Wymore et al., 2016). The results reported here however, are some of the first to demonstrate this relationship in very low DOC tropical montane streams through manipulations of the DOM pool.

DOM Uptake

To further explore energy limitation as the primary mechanism regulating inorganic nutrient transformation, we also assessed the uptake of the added carbon sources. As expected, we identified a high demand for DOM in both experimental streams. High DOM demand has also been found in streams that are low in GPP (Robbins et al., 2017), similar to these LEF streams. The short acetate uptake lengths obtained here using nutrient pulse additions also demonstrate how quickly certain forms of DOC can be taken up, as has been found by 13C additions (Kaplan et al., 2008). The short uptake lengths of acetate compared to those of urea show a potential preference of acetate over urea. This is contrary to other studies in similarly low

DOC streams where urea or N-rich DOM was taken up much more quickly than DIN or even C-

rich DOM (Brookshire et al., 2005). Our urea measurements are likely biased as we had to assess its uptake as a change in bulk DON, and thus rates of urea uptake may be conservative.

Reach-scale DOC uptake, measured as median DOC Vf has been estimated at 2.28 mm

-1 -1 min with a range of 0.26 – 2.94 mm min (Mineau et al., 2016). Our median acetate Vf of 1.64

-1 mm min from the LEF streams falls well within the reported range. The acetate Vf values reported here are also similar to those from Cantabrian mixed forest and Mediterranean forests ranging from 0.6 to 3.99 mm/min which are similar to the LEF, with relatively low DOC and elevated DIN concentrations (Catalán et al., 2018). However, comparing uptake metrics of acetate and urea suggests a preference for C-rich over N-rich organic molecules to fuel heterotrophic processes in streams where energy density is more important than N availability.

Ambient DOC and DON dynamics

- The addition of NO3 to streams can fundamentally alter energy and nutrient demands by

- - shifting the ratio of DOC:NO3 . The decline in DOC and DON concentrations with added NO3 suggests that both the C-rich and N-rich fraction of the ambient DOM pool are being used as an energy source when additional inorganic N is available (Fig 3.4) (Kaushal & Lewis, 2005; Lutz

- et al., 2011; Wymore et al., 2015). Although the relationships between NO3 and DOC and DON were always negative, the responses of DOC and DON were not always synchronous. For

- example, there were instances where a strong negative relationship between NO3 and DON was

- detected, but not simultaneously for NO3 and DOC during the same experiment. At the same

- time, we found examples of a significant response of DOC to elevated NO3 concentration with no concurrent response in concentrations of DON. This frequent asynchrony suggests that the C- rich and N-rich fractions of the DOM pool cycle independently of each other with their own

unique set of biogeochemical controls (Wymore et al., 2018, 2015). This is contrary to the idea that DOC and DON concentrations are coupled (Campbell et al., 2000; Goodale et al., 2000;

Mann et al., 2012) and independent of DIN concentrations in streams (Brookshire et al., 2007).

DON can be an important source of N in streams that are low in DIN but in the case of streams in LEF, the ambient pool of DIN is elevated (not N-limited). Thus the reduction in concentration of N-rich DOM is likely to meet energetic demands rather than satiating nutrient demand (Kaushal & Lewis, 2005; Lutz et al., 2011; Kevin C. Petrone et al., 2009; Wymore et al.,

- 2015) which we mostly see in QT and not in QPB. The high levels of NO3 concentrations in both streams paired with low uptake rates suggest that the concentrations of DOC and DON are too low, and that DOM is not sufficiently bioavailable to meet the energetic demands required

- for the removal of NO3 from the water column. DOM within LEF streams has been previously identified to be lignin-rich (Hernes et al., 2017) and highly aromatic (Miller et al., 2016) suggesting that the DOM pool is not very bioavailable. Even though these watersheds are within the Río Espiritu Santo Watershed and in close proximity to each other, composition of the organic matter pool can often be site specific (Ledesma et al., 2018; Wymore et al., 2018). This has also been found across streams in New Hampshire (Wymore et al., 2018, 2015) and Siberia

(Diemer et al., 2015) where the dynamics of the DOM pool vary within streams of the same watershed but these controls on DOM are still not well understood.

Tropical island streams and rivers can export large amounts of DOM, nutrients, and other solutes from the mountains to the ocean due to short distances and frequent precipitation

(Gaillardet et al., 1999; Lloret et al., 2011; Matson et al., 1999; Merriam et al., 2002).

Understanding the drivers and controls on inorganic N in tropical montane systems is important to maintain ecosystem services and prevent degradation of receiving bodies of water downstream

including the estuarine ecosystems of Puerto Rico. Streams of the LEF are energetically limited and only with the increased availability of low molecular weight and labile DOM will inorganic

- N, especially NO3 , be removed from the water column. Similar to more temperate systems,

DOM has a central role in N cycling of tropical fluvial systems. This relationship between DOM and N processing is especially important in ecosystems where major disturbances can have large changes in ambient stream chemistry. For example, nitrate concentrations in LEF streams are elevated for several years post-hurricane disturbance (McDowell & Asbury, 1994; McDowell et al., 1996; Schaefer et al., 2000), while concentrations of DOC are highly responsive to changes in flow (Shanley et al., 2011; Wymore et al., 2017). With the potential for increased hurricane frequency and intensity, understanding these interactions between DOC and DIN will be critical to build predictive frameworks of the controls on the export of inorganic N.

Tables

Table 3.3. Watershed area and mean DOC, DON, and nutrient concentrations for Quebrada Prieta (QPB) and Quebrada Toronja (QT) with minimum and maximum values in parenthesis.

- + Watershed DOC DON NO3 NH4 Site Area (km2) (mg C/L) (µg N/L) (µg N/L) (µg N/L) 0.86 24.06 68.69 2.86 QPB 0.24 (0.57 - 1.29) (7.55 - 48.22) (37 - 120) (2.5 - 6.05) 0.99 66.65 115.65 5.82 QT 0.18 (0.6 - 1.59) (15.66 - 193.5) (65.4 - 200) (2.5 - 26.96)

Table 3.4. Uptake metrics for all NO3 only additions, NO3 added with acetate and NO3 added with urea at QPB and QT along with discharge from each date. Values with * are uptake metrics - determined as estimates of the minimum values could have been detected in the NO3 -only - - experiment. – denotes undetectable uptake for additions of NO3 + acetate or NO3 + urea.

Uptake Sw Vf U Q Site Date Addition solute (m) (mm min-1) (µg m-2 min-1) (L s-1)

QPB 1/7/16 NO3 Only NO3 5742* 0.056* 3.49*

QPB 1/7/16 NO3+Acetate NO3 - - - 6.20

QPB 1/7/16 NO3+Acetate Acetate 96.9 3.34 83.5

QPB 1/8/16 NO3+Urea NO3 147 1.72 256.3 4.89 QPB 1/8/16 NO3+Urea DON 1546 0.164 1.80

QPB 1/12/16 NO3 Only NO3 3811 0.053 4.36

QPB 1/12/16 NO3+Acetate NO3 1254 0.162 13.2 3.95

QPB 1/12/16 NO3+Acetate Acetate 169 1.20 30.2

QPB 1/14/16 NO3 Only NO3 3262* 0.084* 6.85*

QPB 1/14/16 NO3+Acetate NO3 - - - 3.14

QPB 1/14/16 NO3+Acetate Acetate 146 1.53 162.3

QT 1/4/16 NO3 Only NO3 3697* 0.036* 3.15* 4.96

QT 1/5/16 NO3 Only NO3 3769* 0.037* 2.58*

QT 1/5/16 NO3+Acetate NO3 79.7 1.73 156.9 4.96

QT 1/5/16 NO3+Acetate Acetate 79.2 1.75 43.71

QT 1/11/16 NO3+Acetate NO3 127.8 1.04 193.4 3.54 QT 1/11/16 NO3+Acetate Acetate 121.1 1.09 153.3

QT 1/13/16 NO3 Only NO3 3680* 0.027* 5.28*

QT 1/13/16 NO3+Urea NO3 1417 0.0762 10.07 3.00

QT 1/13/16 NO3+Urea DON - - -

QT 1/15/16 NO3 Only NO3 6558* 0.016* 1.96*

QT 1/15/16 NO3+Acetate NO3 - - - 2.87

QT 1/15/16 NO3+Acetate Acetate 44.25 2.43 98.97

Figures

Figure 3.2. Map of headwaters of the Río Espiritu Santo Watershed (RES) with outlined watersheds of study sites Qda. Toronja (QT) in purple and Qda. Prieta-B (QPB) in grey near the El Verde Field Station (EVFS) in the Luquillo Experimental Forest in northeast Puerto Rico.

Figure 3.3. Combined mean DOM uptake metrics from QPB and QT for acetate (orange) during - - NO3 and acetate additions and urea (calculated as bulk DON) (green) during NO3 and urea additions for (A) uptake length (Sw), (B) uptake velocity (Vf), and (C) areal uptake (U). Bars represent standard error.

Supplementary Information

Table S3.1. Discharge (Q), experimental stream reach length, mean stream width and depth and the mass (g) and percent recovered (% Rec.) of added solutes for each individual addition in Quebrada Toronja (QT) and Quebrada Prieta (QPB).

Mean Mean Reach Added NaCl Added NaNO3 Added Acetate Added Urea Uptake Q stream stream Site Date Addition Length solute (L s-1) width depth % % % % (m) (g) (g) (g) (g) (m) (cm) Rec. Rec. Rec. Rec.

QPB 1/7/16 NO3 Only NO3 737 92 15.05 93 - - - -

QPB 1/7/16 NO3+Acetate NO3 6.20 55.3 1.15 6.25 88 90 NA - 737 15.03 310.76 - QPB 1/7/16 NO3+Acetate Acetate 81 NA 46.05 -

QPB 1/8/16 NO3+Urea NO3 72 49 NA 4.89 55.3 1.15 6.25 737 16.16 - - 107.55 QPB 1/8/16 NO3+Urea DON 84 NA 81

QPB 1/12/16 NO3 Only NO3 737 76 7.53 75 - - - -

QPB 1/12/16 NO3+Acetate NO3 3.95 55.3 1.16 6.25 70 67 NA - 737 7.68 145.69 - QPB 1/12/16 NO3+Acetate Acetate 75 NA 54 -

QPB 1/14/16 NO3 Only NO3 215.48 56 4.37 56 - - - -

QPB 1/14/16 NO3+Acetate NO3 3.14 55.3 0.84 8.33 74 82 NA - 201.25 4.23 126.86 - QPB 1/14/16 NO3+Acetate Acetate 68 NA 46 -

QT 1/4/16 NO3 Only NO3 4.96 75 2.21 11.23 1474 36 5.08 47 - - - -

QT 1/5/16 NO3 Only NO3 1474 65 6.88 69 - - - -

QT 1/5/16 NO3+Acetate NO3 4.96 75 2.15 9.39 83 34 NA - 1474 6.84 300.44 - QT 1/5/16 NO3+Acetate Acetate 76 NA 29 -

QT 1/11/16 NO3+Acetate NO3 72 49 NA - 3.54 50 1.6 11.32 1474 11.36 223.60 - QT 1/11/16 NO3+Acetate Acetate 86 NA 57 -

QT 1/13/16 NO3 Only NO3 737 87 11.11 95 - - - NA 50 1.67 10.05 QT 1/13/16 NO3+Urea NO3 3.00 77 74 - - NA 737 11.13 108.52 QT 1/13/16 NO3+Urea DON 80 NA - - 86

QT 1/15/16 NO3 Only NO3 267.95 79 3.35 81 - - - -

QT 1/15/16 NO3+Acetate NO3 2.87 50 1.6 9.08 81 241 NA - 246.42 3.38 126.86 - QT 1/15/16 NO3+Acetate Acetate 81 NA 26 - 98

100

CONCLUSION

In these chapters, I have demonstrated the tight link between dissolved organic matter and nitrogen in aquatic ecosystems at the global and regional scale. DOM is closely related to the surrounding landscape and large-scale changes will influence the quantity and composition of

OM coming into streams. These changes will in turn affect biogeochemical processes that alter nutrient transformations and downstream exports. I have employed a series of different approaches to demonstrate the coupling of C and N across different biomes. These relationships have been demonstrated via analysis of long-term trends in DOC and DON as well as whole stream manipulations of NH4, NO3, and NO3 with and without C sources to demonstrate the strong control of DOM has on N processing.

The DOM trends reported here add to the growing body of literature on long-term changes of DOC and DON concentrations and contribute trends in DOC:DON which have not been previously considered. In addition, I was able to expand the scope of long-term trends to biomes that are not usually addressed in these sorts of analyses such as the tropics, arctic, and prairie streams. Across biomes, DOC and DON are changing although not at the same rate or direction. DOC:DON ratios are predominantly increasing making streams more carbon rich.

These trends are also irrelevant to historic atmospheric as this was not a strong predictor of grouping variable. Fluctuating concentrations of DOC and DON, whether they are coupled or not will have implication for in-stream biogeochemical processes as well as DOM and nutrient exports to receiving bodies water. Patterns presented in Chapter 1 are an attribute of LTER sites and demonstrate the benefits of curating and maintaining long-term data sets in stream chemistry. These historical records will continue to inform us on ecosystem changes and help

101

create and inform ecosystem models to better manage and predict future trends in aquatic biogeochemical processes.

The strong control that landscape changes can have on aquatic chemistry were furthermore demonstrated in the watersheds of the Central Siberian Plateau. Across this 100-year chronosequence, I was able to show the resiliency and recovery of these watersheds is reflected in stream chemistry and different solutes return to pre-fire conditions at different times.

Watersheds affected by fires will undergo a phase of nutrient exports as uptake rates are confounded with elevated ambient concentrations that push the streams towards saturation where the microbial community declines in the efficiency of nutrient removal. Controls on this microbial efficiency were tightly correlated to ambient concentrations, C:N ratios, and composition of DOM. Although our study focuses on streams within the Central Siberian

Plateau, the results presented here can be applied to a wide array of systems in northern high- latitude and boreal regions. This will better suit predictive models of how Arctic watersheds respond to fire disturbance across the pan-Arctic. The majority of previous studies include only the assessment of stream chemistry before and after a fire, making predictions of long-term recovery and resilience difficult to evaluate and estimate. Chapter 2 addresses this challenge by taking advantage of a long chronosequence in a space-for-time substitution and combining observational and experimental data to better understand the response of watersheds to major landscape disturbances. As Pan-Arctic fire trends continue to increase, the ability to predict ecosystem responses and resiliency also grows in importance.

Nitrogen processing in tropical streams also demonstrated a strong relationship with organic matter. Streams in the Luquillo Experimental Forest are highly energy limited and with the availability of DOC the benthic biota is more efficient at N-removal. Identifying the controls

102

on nutrients in the headwaters is important as tropical montane streams and rivers can transport large amounts of DOM, nutrients, and other solutes due to short distances from the mountains to the ocean and frequent precipitation. Tropical locations like LEF are prone to increased

frequency of hurricanes and storms which have shown to increase NO3 concentrations in streams. Therefore, understanding these interactions between DOC and DIN will be critical to build predictive frameworks of the controls on the export of inorganic N.

The work carried out here suggests that the role of organic matter in nitrogen processing in streams is similar across biomes, where increasing organic matter facilitated greater removal of inorganic nitrogen from the water column (Figure 4.1). However, the rates of transformations, magnitude of concentrations, and implication of biogeochemical processes will be different at the respective sites. Future studies should take into account both and C and N as these do not cycle independently. Coupled interactions will only help to understand the ecosystem as a whole and make better prediction of the ultimate fate of biogeochemical process in our aquatic ecosystems.

Figure 4.1. Nitrate uptake rates from various sites such as boreal forests (Siberia), tundra streams (Alaska, not included in this thesis), temperate streams in NH, and tropical montane streams in Puerto Rico and Guadeloupe (not included in this thesis).

103

REFERENCES

Abbott, B. W., J. R. Larouche, J. B. Jones, W. B. Bowden, and A. W. Balser. 2014. Elevated dissolved organic carbon biodegradability from thawing and collapsing permafrost. Journal of Geophysical Research: Biogeosciences 119:2049–2063. Aiken, G. R., I. D. M. Mcknight, I. K. A. Thorn, and E. M. Thurman. 1992. Isolation of hydrophilic organic acids from water using nonionic macroporous resins Hydrophobic acids XAD - 4 ~ I NaOH ). Hydrophilic acids. Organic Geochemistry 18:567–573. Retrieved from http://www.sciencedirect.com/science/article/pii/014663809290119I Argerich, A., S. L. Johnson, S. D. Sebestyen, C. C. Rhoades, E. Greathouse, J. D. Knoepp, M. B. Adams, G. E. Likens, J. L. Campbell, W. H. McDowell, F. N. Scatena, and G. G. Ice. 2013. Trends in stream nitrogen concentrations for forested reference catchments across the USA. Environmental Research Letters 8. Bayley, S. E., D. W. Schindler, K. G. Beaty, B. R. Parker, and M. P. Stainton. 1992. Effects of Multiple Fires on Nutrient Yields from Streams Draining Boreal Forest and Fen Watersheds: Nitrogen and Phosphorus. Canadian Journal of Fisheries and Aquatic Sciences 49:584–596. Bernhardt, E. S., J. B. Heffernan, N. B. Grimm, E. H. Stanley, J. W. Harvey, M. Arroita, A. P. Appling, M. J. Cohen, W. H. McDowell, R. O. Hall, J. S. Read, B. J. Roberts, E. G. Stets, and C. B. Yackulic. 2018. The metabolic regimes of flowing waters. Limnology and Oceanography 63:S99–S118. Bernhardt, E. S., and G. E. Likens. 2002. Dissolved organic carbon enrichment alters nitrogen dynamics in a forest stream. Ecology 83:1689–1700. Bernhardt, E. S., and W. H. McDowell. 2008. Twenty years apart: Comparisons of DOM uptake during leaf leachate releases to Hubbard Brook Valley streams in 1979 versus 2000. Journal of Geophysical Research 113:G03032. Betts, E. F., and J. B. Jones. 2009. Impact of wildfire on stream nutrient chemistry and ecosystem metabolism in boreal forest catchments of interior Alaska. Arctic, Antarctic, and Alpine Research 41:407–417. Borken, W., B. Ahrens, C. Schulz, and L. Zimmermann. 2011. Site-to-site variability and temporal trends of DOC concentrations and fluxes in temperate forest soils. Global Change Biology 17:2428–2443. Brookshire, E. N.J., H. M. Valett, S. A. Thomas, and J. R. Webster. 2007. Atmospheric N deposition increases organic N loss from temperate forests. Ecosystems 10:252–262. Brookshire, E N Jack, H. M. Valett, S. A. Thomas, and J. R. Webster. 2005. Coupled cycling of dissolved organic nitrogen and carbon in a forest stream. Ecology 86:2487–2496. Bunker, D. E., F. Declerck, J. C. Bradford, R. K. Colwell, I. Perfecto, O. L. Phillips, M. Sankaran, and S. Naeem. 2005. Species loss and aboveground carbon storage in a tropical forest. Science (New York, N.Y.) 310:1029–1031. Burd, K., S. E. Tank, N. Dion, W. L. Quinton, C. Spence, A. J. Tanentzap, and D. Olefeldt. 2018. Seasonal shifts in export of DOC and nutrients from burned and unburned peatland- rich catchments, Northwest Territories, Canada. Hydrology and Earth System Sciences Discussions 1–32. Campbell, J. L., J. W. Hornbeck, W. H. McDowell, D. C. Buso, J. B. Shanley, and G. E. Likens. 2000. Dissolved Organic Nitrogen Budgets for Upland, Forested Ecosystems in New England. Biogeochemistry 49:123–142. Retrieved from

104

http://www.jstor.org/stable/1469381 Catalán, N., J. P. Casas-Ruiz, M. I. Arce, M. Abril, A. G. Bravo, R. del Campo, E. Estévez, A. Freixa, P. Giménez-Grau, A. M. González-Ferreras, L. Gómez-Gener, A. Lupon, A. Martínez, C. Palacin-Lizarbe, S. Poblador, R. Rasines-Ladero, M. Reyes, T. Rodríguez- Castillo, P. Rodríguez-Lozano, I. Sanpera-Calbet, I. Tornero, and A. Pastor. 2018. Behind the Scenes: Mechanisms Regulating Climatic Patterns of Dissolved Organic Carbon Uptake in Headwater Streams. Global Biogeochemical Cycles 32:1528–1541. Chow, A. T., F. J. Díaz, K.-H. Wong, A. T. O’Geen, R. A. Dahlgren, and P.-K. Wong. 2013. Photochemical and Bacterial Transformations of Disinfection By-Product Precursors in Water. Journal of Environmental Quality 42:1589–1595. Clair, T. A., I. F. Dennis, R. Vet, and H. Laudon. 2008. Long-term trends in catchment organic carbon and nitrogen exports from three acidified catchments in Nova Scotia, Canada. Biogeochemistry 87:83–97. Coble, A. A., A. S. Wymore, M. D. Shattuck, J. D. Potter, and W. H. McDowell. 2018. Multiyear Trends in Solute Concentrations and Fluxes From a Suburban Watershed: Evaluating Effects of 100-Year Flood Events. Journal of Geophysical Research: Biogeosciences 123:3072–3087. Corilo, Y. 2015. EnviroOrg, Florida State University: Tallahassee. USA. Cory, R. M., C. P. Ward, B. C. Crump, and G. W. Kling. 2014. Sunlight controls water column processing of carbon in arctic fresh waters. Science 345:925–928. Cory, Rose M, and D. M. Mcknight. 2005. Fluorescence Spectroscopy Reveals Ubiquitous Presence of Oxidized and Reduced Quinones in Dissolved Organic Matter Fluorescence Spectroscopy Reveals Ubiquitous Presence of Oxidized and Reduced Quinones in Dissolved Organic Matter. Environmental Science & Technology 39:8142–8149. De Wit, H. A., J. Mulder, A. Hindar, and L. Hole. 2007. Long-term increase in dissolved organic carbon in streamwaters in Norway is response to reduced acid deposition. Environmental Science and Technology 41:7706–7713. Diemer, L. A., W. H. McDowell, A. S. Wymore, and A. S. Prokushkin. 2015. Nutrient uptake along a fire gradient in boreal streams of Central Siberia. Freshwater Science 34:1443– 1456. Dillon, P. J., and L. A. Molot. 2005. Long-term trends in catchment export and lake retention of dissolved organic carbon, dissolved organic nitrogen, total iron, and total phosphorus: The Dorset, Ontario, study, 1978–1998. Journal of Geophysical Research 110:3–9. Ding, Y., Y. Yamashita, J. Jones, and R. Jaffé. 2015. Dissolved black carbon in boreal forest and glacial rivers of central Alaska: assessment of biomass burning versus anthropogenic sources. Biogeochemistry 123:15–25. Dittmar, T., B. Koch, N. Hertkorn, and G. Kattner. 2008. A simple and efficient method for the solid-phase extraction of dissolved organic matter (SPE-DOM) from seawater. Limnol. Oceanogr. Methods 6:230–235. Dodds, W. K., M. A. Evans-White, N. M. Gerlanc, L. Gray, D. A. Gudder, M. J. Kemp, A. L. López, D. Stagliano, E. A. Strauss, J. L. Tank, M. R. Whiles, and W. M. Wollheim. 2000. Quantification of the nitrogen cycle in a prairie stream. Ecosystems 3:574–589. Dodds, W. K., A. J. López, W. B. Bowden, S. Gregory, N. B. Grimm, S. K. Hamilton, A. E. Hershey, E. Martí, W. H. McDowell, J. L. Meyer, D. Morrall, P. J. Mulholland, B. J. Peterson, J. L. Tank, H. M. Valett, J. R. Webster, and W. Wollheim. 2002. N uptake as a function of concentration in streams. Journal of the North American Benthological Society

105

21:206–220. Dodds, W. K., E. Martí, J. L. Tank, J. Pontius, S. K. Hamilton, N. B. Grimm, W. B. Bowden, W. H. McDowell, B. J. Peterson, H. M. Valett, J. R. Webster, and S. Gregory. 2004. Carbon and nitrogen stoichiometry and nitrogen cycling rates in streams. Oecologia 140:458–467. Drake, T. W., K. Van Oost, M. Barthel, M. Bauters, A. M. Hoyt, D. C. Podgorski, J. Six, P. Boeckx, S. E. Trumbore, L. Cizungu Ntaboba, and R. G. M. Spencer. 2019. Mobilization of aged and biolabile soil carbon by tropical deforestation. Nature Geoscience 12:541–546. Driscoll, C. T., K. M. Driscoll, K. M. Roy, and M. J. Mitchell. 2003. Chemical response of lakes in the Adirondack Region of New York to declines in acidic deposition. Environmental Science and Technology 37:2036–2042. Elser, J. J., W. F. Fagan, R. F. Denno, D. R. Dobberfuhl, A. Folarin, A. Huberty, S. Interlandi, S. S. Kilham, E. McCauley, K. L. Schulz, E. H. Siemann, and R. W. Sterner. 2000. Nutritional constraints in terrestrial and freshwater food webs. Nature 408:578–580. Ensign, S. H., and M. W. Doyle. 2006. Nutrient spiraling in streams and river networks. Journal of Geophysical Research: Biogeosciences 111:1–13. Evans, C. D., P. J. Chapman, J. M. Clark, D. T. Monteith, and M. S. Cresser. 2006. Alternative explanations for rising dissolved organic carbon export from organic soils. Global Change Biology 12:2044–2053. Evans, C. D., D. T. Monteith, and D. M. Cooper. 2005. Long-term increases in surface water dissolved organic carbon: Observations, possible causes and environmental impacts. Environmental Pollution 137:55–71. Finkler, N. R., F. Tromboni, I. G. Boëchat, B. Gücker, and D. G. F. Cunha. 2018. Nitrogen and phosphorus uptake dynamics in tropical Cerrado woodland streams. Water (Switzerland) 10. Finlay, J., J. Neff, S. Zimov, A. Davydova, and S. Davydov. 2006. Snowmelt dominance of dissolved organic carbon in high-latitute watersheds: Implications for characterization and flux of river DOC. Geophysical Research Letters 33:2–6. Freeman, C., N. Fenner, N. J. Ostle, H. Kang, D. J. Dowrick, B. Reynolds, M. A. Lock, D. Sleep, S. Hughes, and J. Hudson. 2004. Export of dissolved organic carbon from peatlands under elevated carbon dioxide levels. Nature 430:59–61. Frey, K. E., and J. W. McClelland. 2009. Impacts of permafrost degradation on arctic river biogeochemistry. Hydrological Processes Vol. 23, pp. 169–182. Frey, K. E., J. W. McClelland, R. M. Holmes, and L. G. Smith. 2007. Impacts of climate warming and permafrost thaw on the riverine transport of nitrogen and phosphorus to the Kara Sea. Journal of Geophysical Research: Biogeosciences 112:1–10. Frey, K. E., and L. C. Smith. 2005. Amplified carbon release from vast West Siberian peatlands by 2100. Geophysical Research Letters 32:1–4. Gaillardet, J., B. Dupré, P. Louvat, and C. J. Allègre. 1999. Global silicate weathering and CO2 consumption rates deduced from the chemistry of large rivers. Chemical Geology 159:3–30. Gavin, A. L., S. J. Nelson, A. J. Klemmer, I. J. Fernandez, K. E. Strock, and W. H. McDowell. 2018. Acidification and Climate Linkages to Increased Dissolved Organic Carbon in High- Elevation Lakes. Water Resources Research 54:5376–5393. Goodale, C. L., J. D. Aber, and W. H. McDowell. 2000. The long-term effects of disturbance on organic and inorganic nitrogen export in the White Mountains, New Hampshire. Ecosystems 3:433–450. Goodale, C. L., J. D. Aber, P. M. Vitousek, and W. H. McDowell. 2005. Long-Term Decreases

106

in Stream Nitrate: Successional Causes Unlikely; Possible Links to DOC? Ecosystems 8:334–337. Retrieved from http://www.jstor.org/stable/25053830 Goolsby, D. A., and W. A. Battaglin. 2001. Long-term changes in consentrations and flux fo nitrogen in the Mississippi River Basin, USA. Hydrological Processes 15:1209–1226. Hall Jr., R. O., and J. L. Tank. 2003. Ecosystem Metabolism Controls Nitrogen Uptake in Streams in Grand Teton National Park, Wyoming. Limnology and Oceanography 48:1120– 1128. Hall, R. O., J. L. Tank, D. J. Sobota, P. J. Mulholland, J. M. O’Brien, W. K. Dodds, J. R. Webster, H. M. Valett, G. C. Poole, B. J. Peterson, J. L. Meyer, W. H. McDowell, S. L. Johnson, S. K. Hamilton, N. B. Grimm, S. V Gregory, C. N. Dahm, L. W. Cooper, L. R. Ashkenas, S. M. Thomas, R. W. Sheibley, J. D. Potter, B. R. Niederlehner, L. T. Johnson, A. M. Helton, and C. M. Crenshaw. 2009. Nitrate removal in stream ecosystems measured by 15N addition experiments: total uptake. Limnology and Oceanography 54:653–665. Helms, J. R., A. Stubbins, J. D. Ritchie, E. C. Minor, D. J. Kieber, and K. Mopper. 2008. Absorption spectral slopes and slope ratios as indicators of molecular weight, source, and photobleaching of chromophoric dissolved organic matter. Limnology and Oceanography 53:955–969. Helton, A. M., M. Ardón, and E. S. Bernhardt. 2015. Thermodynamic constraints on the utility of ecological stoichiometry for explaining global biogeochemical patterns. Ecology Letters 18:1049–1056. Hernes, P. J., R. Y. Dyda, and W. H. McDowell. 2017. Connecting tropical river DOM and POM to the landscape with lignin. Geochimica et Cosmochimica Acta. Hirsch, R. M., J. R. Slack, and R. A. Smith. 1982. Techniques of trend detection for monthly water quality data. Water Resources Research 18:107–121. Holling, C. S. 1973. Resilience and Stability of Ecological Systems. Annual Review of Ecology and Systematics 4:1–23. Holmes, R. M., J. W. McClelland, B. J. Peterson, S. E. Tank, E. Bulygina, T. I. Eglinton, V. V. Gordeev, T. Y. Gurtovaya, P. A. Raymond, D. J. Repeta, R. Staples, R. G. Striegl, A. V. Zhulidov, and S. A. Zimov. 2012. Seasonal and Annual Fluxes of Nutrients and Organic Matter from Large Rivers to the Arctic Ocean and Surrounding Seas. Estuaries and Coasts 35:369–382. Hood, E., D. M. McKnight, and M. W. Williams. 2003. Sources and chemical character of dissolved organic carbon across an alpine/subalpine ecotone, Green Lakes Valley, Colorado Front Range, United States. Water Resources Research 39:1–12. Hruška, J., P. Krám, W. H. McDowell, and F. Oulehle. 2009. Increased Dissolved Organic Carbon (DOC) in Central European Streams is Driven by Reductions in Ionic Strength Rather than Climate Change or Decreasing Acidity. Environmental Science & Technology 43:4320–4326. Hynes, H. 1975. The stream and its valley. Verh. Internat. Verein. Limnol 19:1–15. Inamdar, S., N. Finger, S. Singh, M. Mitchell, D. Levia, H. Bais, D. Scott, and P. McHale. 2012. Dissolved organic matter (DOM) concentration and quality in a forested mid-Atlantic watershed, USA. Biogeochemistry 108:55–76. Jiang, Y., E. B. Rastetter, A. V Rocha, A. R. Pearce, B. L. Kwiatkowski, and G. R. Shaver. 2015. Modeling carbon–nutrient interactions during the early recovery of tundra after fire. Ecological Applications 25:1640–1652. Kaplan, L. A., T. N. Wiegner, J. D. Newbold, P. H. Ostrom, and H. Gandhi. 2008. Untangling

107

the complex issue of dissolved organic carbon uptake: A stable isotope approach. Freshwater Biology 53:855–864. Kaushal, S. S., and W. M. Lewis. 2005. Fate and Transport of Organic Nitrogen in Minimally Disturbed Montane Streams of Colorado, USA. Biogeochemistry 74:303–321. Kawahigashi, M., A. Prokushkin, and H. Sumida. 2011. Effect of fire on solute release from organic horizons under larch forest in Central Siberian permafrost terrain. Geoderma 166:171–180. Kharuk, V. I., M. L. Dvinskaya, I. A. Petrov, S. T. Im, and K. J. Ranson. 2016. Larch forests of Middle Siberia: long-term trends in fire return intervals. Regional Environmental Change 16:2389–2397. Kilpatrick, F. A., and E. D. Cobb. 1985. Techniques of water-resources investigations of the United States Geological Survey:Measurement of discharge. Koch, B. P., and T. Dittmar. 2006. From mass to structure: An aromaticity index for high- resolution mass data of natural organic matter. Rapid Communications in Mass Spectrometry 20:926–932. Koenig, L. E., C. Song, W. M. Wollheim, J. Rüegg, and W. H. McDowell. 2017. Nitrification increases nitrogen export from a tropical river network. Freshwater Science 36:698–712. Kortelainen, P., T. Mattsson, L. Finér, M. Ahtiainen, S. Saukkonen, and T. Sallantaus. 2006. Controls on the export of C, N, P and Fe from undisturbed boreal catchments, Finland. Aquatic Sciences 68:453–468. Kothawala, D. N., K. R. Murphy, C. A. Stedmon, G. A. Weyhenmeyer, and L. J. Tranvik. 2013. Inner filter correction of dissolved organic matter fluorescence. Limnology and Oceanography: Methods 11:616–630. Kritzberg, E. S. 2017. Centennial-long trends of lake browning show major effect of afforestation. Limnology and Oceanography Letters 2:105–112. Lajtha, K., and J. Jones. 2018. Forest harvest legacies control dissolved organic carbon export in small watersheds, western Oregon. Biogeochemistry 140:299–315. Larouche, J. R., B. W. Abbott, W. B. Bowden, and J. B. Jones. 2015. The role of watershed characteristics, permafrost thaw, and wildfire on dissolved organic carbon biodegradability and water chemistry in Arctic headwater streams. Biogeosciences Discussions 12:4021– 4056. Lawrence, G. B., H. A. Simonin, B. P. Baldigo, K. M. Roy, and S. B. Capone. 2011. Changes in the chemistry of acidified Adirondack streams from the early 1980s to 2008. Environmental Pollution 159:2750–2758. Ledesma, J. L. J., D. N. Kothawala, P. Bastviken, S. Maehder, T. Grabs, and M. N. Futter. 2018. Stream Dissolved Organic Matter Composition Reflects the Riparian Zone, Not Upslope Soils in Boreal Forest Headwaters. Water Resources Research 54:3896–3912. Lee, W., P. Westerhoff, and M. Esparza-Soto. 2006. Occurrence and removal of dissolved organic nitrogen in US water treatment plants. Journal / American Water Works Association 98:102–110. Leenheer, J. A., and J.-P. Croue. 2003. Characterizing Dissolved Organic Matter. Environmental Science & Technology 37:18A-26A. Lepistö, A., P. Kortelainen, and T. Mattsson. 2008. Increased organic C and N leaching in a northern boreal river basin in Finland. Global Biogeochemical Cycles 22:1–10. Likens, Gene Bormann, F. 1974. Acid Rain : A Serious Regional Environmental Problem. Science 184:1176–1179.

108

Likens, G. E., C. T. Driscoll, and D. C. Buso. 1996. Long-term effects of acid rain: Response and recovery of a forest ecosystem. Science 272:244–246. Lloret, E., C. Dessert, J. Gaillardet, P. Albéric, O. Crispi, C. Chaduteau, and M. F. Benedetti. 2011. Comparison of dissolved inorganic and organic carbon yields and fluxes in the watersheds of tropical volcanic islands, examples from Guadeloupe (French West Indies). Chemical Geology 280:65–78. Löfgren, S., J. P. Gustafsson, and L. Bringmark. 2010. Decreasing DOC trends in soil solution along the hillslopes at two IM sites in southern Sweden - Geochemical modeling of organic matter solubility during acidification recovery. Science of the Total Environment 409:201– 210. Lupon, A., E. Martí, F. Sabater, and S. Bernal. 2016. Green light: Gross primary production influences seasonal stream N export by controlling fine-scale N dynamics. Ecology 97:133– 144. Lutz, B. D., E. S. Bernhardt, B. J. Roberts, and P. J. Mulholland. 2011. Examining the coupling of carbon and nitrogen cycles in Appalachian streams: the role of dissolved organic nitrogen. Ecology 92:720–732. Mangiafico, S. 2017. rcompanion: Functions to Support Extension Education Program Evaluation. Rpackage version 1.10.1. Mann, P. J., A. Davydova, N. Zimov, R. G. M. Spencer, S. Davydov, E. Bulygina, S. Zimov, and R. M. Holmes. 2012. Controls on the composition and lability of dissolved organic matter in Siberia’s Kolyma River basin. Journal of Geophysical Research: Biogeosciences 117:1–15. Martí, E., and F. Sabater. 1996. High Variability in Temporal and Spatial Nutrient Retention in Mediterranean Streams. Ecology 77:854–869. Matson, P. a., W. H. McDowell, A. R. Townsend, and P. M. Vitousek. 1999. The globalization of N deposition: ecosystem consequences in tropical environments. Biogeochemistry 46:67–83. Mattsson, T., P. Kortelainen, and A. Räike. 2005. Export of DOM from boreal catchments: Impacts of land use cover and climate. Biogeochemistry 76:373–394. McClelland, J. W., M. Stieglitz, F. Pan, R. M. Holmes, and B. J. Peterson. 2007. Recent changes in nitrate and dissolved organic carbon export from the upper Kuparuk River, North Slope, Alaska. Journal of Geophysical Research: Biogeosciences 112:1–13. McDowell, W. H. 1998. Internal nutrient fluxes in a Puerto Rican rain forest. Journal of Tropical Ecology 14:521–536. McDowell, W. H., and C. E. Asbury. 1994. Export of carbon, nitrogen, and major ions from three tropical montane watersheds. Limnology and Oceanography 39:111–125. McDowell, W. H., and G. E. Likens. 1988. Origin, Composition, and Flux of Dissolved Organic Carbon in the Hubbard Brook Valley. Ecological Monographs 58:177–195. McDowell, W. H., A. H. Magill, J. A. Aitkenhead-Peterson, J. D. Aber, J. L. Merriam, and S. S. Kaushal. 2004. Effects of chronic nitrogen amendment on dissolved organic matter and inorganic nitrogen in soil solution. Forest Ecology and Management 196:29–41. McDowell, W. H., W. G. McDowell, J. D. Potter, and A. Ramírez. 2019. Nutrient export and elemental stoichiometry in an urban tropical river. Ecological Applications 29:1–15. McDowell, W. H., C. P. McSwiney, and W. B. Bowden. 1996. Effects of Hurricane Disturbance on Groundwater Chemistry and Riparian Function in a Tropical Rain Forest. Biotropica 28:577. McLeod, A. I. 2011. Kendall: Kendall rank correlation and Mann-Kendall trend test. R package

109

version 2.2. Retrieved from https://cran.r-project.org/web/packages/Kendall/Kendall.pdf Merriam, J. L., W. H. McDowell, J. L. Tank, W. M. Wollheim, C. L. Crenshaw, and S. L. Johnson. 2002. Characterizing nitrogen dynamics, retention and transport in a tropical rainforest stream using an in situ15N addition. Freshwater Biology 47:143–160. Meyer, J. L., and J. B. Wallace. 1998. Leaf Litter as a Source of Dissolved Organic Carbon in Streams. Ecosystems 1:240–249. Miller, M. P., E. W. Boyer, D. M. McKnight, M. G. Brown, R. S. Gabor, C. T. Hunsaker, L. Iavorivska, S. Inamdar, D. W. Johnson, L. A. Kaplan, H. Lin, W. H. McDowell, and J. N. Perdrial. 2016. Variation of organic matter quantity and quality in streams at Critical Zone Observatory watersheds. Water Resources Research 52:8202–8216. Mineau, M. M., W. M. Wollheim, I. Buffam, S. E. G. Findlay, R. O. Hall, E. R. Hotchkiss, L. E. Koenig, W. H. McDowell, and T. B. Parr. 2016. Dissolved organic carbon uptake in streams: A review and assessment of reach-scale measurements. Journal of Geophysical Research: Biogeosciences 121:2019–2029. Monteith, D. T., J. L. Stoddard, C. D. Evans, H. A. de Wit, M. Forsius, T. Høgåsen, A. Wilander, B. L. Skjelkvåle, D. S. Jeffries, J. Vuorenmaa, B. Keller, J. Kopácek, and J. Vesely. 2007. Dissolved organic carbon trends resulting from changes in atmospheric deposition chemistry. Nature 450:537–540. Mroz, G. D., M. F. Jurgensen, A. E. Harvey, and M. J. Larsen. 1980. Effects of Fire on Nitrogen in Forest Floor Horizons. Soil Science Society of America Journal 44:395. Mulholland, P. J., R. O. Hall, D. J. Sobota, W. K. Dodds, S. E. G. Findlay, N. B. Grimm, S. K. Hamilton, W. H. McDowell, J. M. O’Brien, J. L. Tank, L. R. Ashkenas, L. W. Cooper, C. N. Dahm, S. V. Gregory, S. L. Johnson, J. L. Meyer, B. J. Peterson, G. C. Poole, H. M. Valett, J. R. Webster, C. P. Arango, J. J. Beaulieu, M. J. Bernot, A. J. Burgin, C. L. Crenshaw, A. M. Helton, L. T. Johnson, B. R. Niederlehner, J. D. Potter, R. W. Sheibley, and S. M. Thomasn. 2009. Nitrate removal in stream ecosystems measured by 15N addition experiments: Denitrification. Limnology and Oceanography 54:666–680. Mulholland, P. J., A. M. Helton, G. C. Poole, R. O. Hall, S. K. Hamilton, B. J. Peterson, J. L. Tank, L. R. Ashkenas, L. W. Cooper, C. N. Dahm, W. K. Dodds, S. E. G. Findlay, S. V Gregory, N. B. Grimm, S. L. Johnson, W. H. McDowell, J. L. Meyer, H. M. Valett, J. R. Webster, C. P. Arango, J. J. Beaulieu, M. J. Bernot, A. J. Burgin, C. L. Crenshaw, L. T. Johnson, B. R. Niederlehner, J. M. O’Brien, J. D. Potter, R. W. Sheibley, D. J. Sobota, and S. M. Thomas. 2008. Stream denitrification across biomes and its response to anthropogenic nitrate loading. Nature 452:202–205. Murphy, K. R., C. A. Stedmon, D. Graeber, and R. Bro. 2013. Fluorescence spectroscopy and multi-way techniques. PARAFAC. Analytical Methods 5:6557. National Atmospheric Deposition Program (NRSP-3). 2020. Retrieved from NADP Program Office, Wisconsin State Laboratory of Hygiene, 465 Henry Mall, Madison, WI 53706 website: http://nadp.slh.wisc.edu/ Newbold, J. D., J. W. Elwood, R. V O’Neill, and W. Van Winkle. 1981. Measuring nutrient spiralling in streams. Canadian Journal of Fisheries and Aquatic Sciences 38:860–863. O’Brien, J. M., and W. K. Dodds. 2010. Saturation of NO3− uptake in prairie streams as a function of acute and chronic N exposure. Journal of the North American Benthological Society 29:627–635. Ohno, T. 2002. Fluorescence inner-filtering correction for determining the humification index of dissolved organic matter. Environmental Science and Technology 36:742–746.

110

Ortiz-Zayas, J. R., W. M. Lewis, J. F. Saunders, J. H. McCutchan, and F. N. Scatena. 2005. Metabolism of a tropical rainforest stream. Journal of the North American Benthological Society 24:769–783. Pagano, T., M. Bida, and J. Kenny. 2014. Trends in Levels of Allochthonous Dissolved Organic Carbon in Natural Water: A Review of Potential Mechanisms under a Changing Climate. Water 6:2862–2897. Parham, L. M., A. S. Prokushkin, O. S. Pokrovsky, S. V Titov, E. Grekova, L. S. Shirokova, and W. H. McDowell. 2013. Permafrost and fire as regulators of stream chemistry in basins of the Central Siberian Plateau. Biogeochemistry 116:55–68. Peterson, B. J., W. M. Wollheim, P. J. Mulholland, J. R. Webster, J. L. Meyer, J. L. Tank, E. Marti, W. B. Bowden, H. M. Valett, A. E. Hershey, W. H. McDowell, W. K. Dodds, S. K. Hamilton, S. Gregory, and D. D. Morrall. 2001. Control of nitrogen export from watersheds by headwater streams. Science 292:86–90. Petrone, K. C., L. D. Hinzman, H. Shibata, J. B. Jones, and R. D. Boone. 2007. The influence of fire and permafrost on sub-arctic stream chemistry during storms. Hydrological Processes 21:423–434. Petrone, Kevin C., J. S. Richards, and P. F. Grierson. 2009. Bioavailability and composition of dissolved organic carbon and nitrogen in a near coastal catchment of south-western Australia. Biogeochemistry 92:27–40. Pohlert, T. 2018. Trend: Non-Parametric Trend Tests and Change-Point Detection. R package version 1.1.0. Potter, J. D., W. H. McDowell, J. L. Merriam, B. J. Peterson, and S. M. Thomas. 2010. Denitrification and total nitrate uptake in streams of a tropical landscape. Ecological Applications 20:2104–2115. Program, N. C. C. A. 2006. 2001 to 2006 Regional Land Cover Change Data – Coastal United States. Available from NOAA Data Web site: https://data.noaa.gov/dataset/noaas-coastal- change-analysis-program-c-cap-2001-to-2006-regional-land-cover-change-data-coast. Prokushkin, A. S., G. Gleixner, W. H. McDowell, S. Ruehlow, and E. D. Schulze. 2007. Source- and substrate-specific export of dissolved organic matter from permafrost-dominated forested watershed in central Siberia. Global Biogeochemical Cycles 21. Prokushkin, A. S., F. Hagedorn, O. S. Pokrovsky, J. Viers, A. V. Kirdyanov, O. V. Masyagina, M. P. Prokushkina, and W. H. McDowell. 2018. Permafrost regime affects the nutritional status and productivity of larches in Central Siberia. Forests 9:1–18. Prokushkin, A. S., A. A. Knorre, A. V Kirdyanov, and E. D. Schulze. 2006. Productivity of mosses and organic matter accumulation in the litter of sphagnum larch forest in the permafrost zone. Russian Journal of Ecology 37:225–232. Prokushkin, A. S., O. S. Pokrovsky, L. S. Shirokova, M. a Korets, J. Viers, S. G. Prokushkin, R. M. W. Amon, G. Guggenberger, and W. H. McDowell. 2011. Sources and the flux pattern of dissolved carbon in rivers of the Yenisey basin draining the Central Siberian Plateau. Environmental Research Letters 6:045212. R-Core-Team. 2016. R: A language and environment for statistical computing. R Foundation for Statistical Computing. Vienna, Austria. Räike, A., P. Kortelainen, T. Mattsson, and D. N. Thomas. 2012. 36year trends in dissolved organic carbon export from Finnish rivers to the Baltic Sea. Science of the Total Environment 435–436:188–201. Ribot, M., D. von Schiller, M. Peipoch, F. Sabater, N. B. Grimm, and E. Martí. 2013. Influence

111

of nitrate and ammonium availability on uptake kinetics of stream biofilms. Freshwater Science 32:1155–1167. Richey, J. S., W. H. McDowell, and G. E. Likens. 1985. Nitrogen transformations in a small mountain stream. Hydrobiologia 124:129–139. Robbins, C. J., R. S. King, A. D. Yeager, C. M. Walker, J. A. Back, R. D. Doyle, and D. F. Whigham. 2017. Low-level addition of dissolved organic carbon increases basal ecosystem function in a boreal headwater stream. Ecosphere 8:e01739. Rodríguez-Cardona, B., A. S. Wymore, and W. H. McDowell. 2016. DOC:NO3- ratios and NO3- uptake in forested headwater streams. Journal of Geophysical Research G: Biogeosciences 121:205–217. Roulet, N., and T. R. Moore. 2006. Browning the waters. Nature 444:283–284. Rüegg, J., J. J. Eichmiller, N. Mladenov, and W. K. Dodds. 2015. Dissolved organic carbon concentration and flux in a grassland stream: spatial and temporal patterns and processes from long-term data. Biogeochemistry 125:393–408. Santos, F., A. S. Wymore, B. K. Jackson, S. M. P. Sullivan, W. H. McDowell, and A. A. Berhe. 2019. Fire severity, time since fire, and site-level characteristics influence streamwater chemistry at baseflow conditions in catchments of the Sierra Nevada, California, USA. Fire Ecology 15:3. Sarkkola, S., H. Koivusalo, A. Laurén, P. Kortelainen, T. Mattsson, M. Palviainen, S. Piirainen, M. Starr, and L. Finér. 2009. Trends in hydrometeorological conditions and stream water organic carbon in boreal forested catchments. Science of the Total Environment 408:92– 101. Sawicka, K., E. C. Rowe, C. D. Evans, D. T. Monteith, E.I.Vanguelova, A. J. Wade, and J.M.Clark. 2017. Modelling impacts of atmospheric deposition and temperature on long- term DOC trends. Science of the Total Environment 578:323–336. Schade, J. D., J. F. Espeleta, C. A. Klausmeier, E. Megan, S. A. Thomas, and L. Zhang. 2005. A Conceptual Framework for Ecosystem Stoichiometry : Balancing Resource Supply and Demand. Oikos 109:40–51. Schaefer, D. a., W. H. McDowell, F. N. Scatena, and C. E. Asbury. 2000. Effects of hurricane disturbance on stream water concentrations and fluxes in eight tropical forest watersheds of the Luquillo Experimental Forest, Puerto Rico. Journal of Tropical Ecology 16:189–207. Schindler, D. W., P. J. Curtis, S. E. Bayley, B. R. Parker, K. G. Beaty, and M. P. Stainton. 1997. Climate-induced changes in the dissolved organic carbon budgets of boreal lakes. Biogeochemistry 36:9–28. Schuur, E. A. G., J. Bockheim, J. P. Canadell, E. Euskirchen, C. B. Field, S. V Goryachkin, P. Kuhry, P. M. Lafleur, H. Lee, G. Mazhitova, F. E. Nelson, A. Rinke, V. E. Romanovsky, N. Shiklomanov, C. Tarnokai, S. Venevsky, J. G. Vogel, and S. A. Zimov. 2008. Vulnerability of permafrost carbon to climate change: Implications for the global carbon cycle. BioScience 58:701–714. Shanley, J. B., W. H. McDowell, and R. F. Stallard. 2011. Long-term patterns and short-term dynamics of stream solutes and suspended sediment in a rapidly weathering tropical watershed. Water Resources Research 47:1–11. Shvidenko, A. Z., and D. G. Schepaschenko. 2013. Climate change and wildfires in Russia. Contemporary Problems of Ecology 6:683–692. Silver, W. L., F. N. Scatena, A. H. Johnson, T. G. Siccama, and M. J. Sanchez. 1994. Nutrient availability in a montane wet tropical forest: Spatial patterns and methodological

112

considerations. Plant and Soil 164:129–145. Singh, S., S. Inamdar, and D. Scott. 2013. Comparison of Two PARAFAC Models of Dissolved Organic Matter Fluorescence for a Mid-Atlantic Forested Watershed in the USA. Journal of Ecosystems 2013:1–16. Smith, D. F., D. C. Podgorski, R. P. Rodgers, G. T. Blakney, and C. L. Hendrickson. 2018. 21 Tesla FT-ICR Mass Spectrometer for Ultrahigh-Resolution Analysis of Complex Organic Mixtures. Analytical Chemistry acs.analchem.7b04159. Sobczak, W. V, and S. Findlay. 2002. Variation in bioavailabtlity of dissolved organic carbon among stream hyporheic flowpaths. Ecology 83:3194–3209. Spencer, R. G. M., G. R. Aiken, K. P. Wickland, R. G. Striegl, and P. J. Hernes. 2008. Seasonal and spatial variability in dissolved organic matter quantity and composition from the Yukon River basin, Alaska. Global Biogeochemical Cycles 22:1–13. Spencer, R. G. M., W. Guo, P. A. Raymond, T. Dittmar, E. Hood, J. Fellman, and A. Stubbins. 2014. Source and biolability of ancient dissolved organic matter in glacier and lake ecosystems on the tibetan plateau. Geochimica et Cosmochimica Acta 142:64–74. Starry, O. S., H. M. Valett, and M. E. Schreiber. 2005. Nitrification rates in a headwater stream: influences of seasonal variation in C and N supply. Journal of the North American Benthological Society 24:753–768. Stedmon, C. A., and R. Bro. 2008. Characterizing dissolved organic matter fluorescence with parallel factor analysis: a tutorial. Limnology and Oceanography: Methods 6:572–579. Stedmon, C. A., and S. Markager. 2005. Resolving the variability in dissolved organic matter fluorescence in a temperate estuary and its catchment using PARAFAC analysis. Limnology and Oceanography 50:686–697. Strååt, K. D., C.-M. Mörth, and E. Undeman. 2018. Future export of particulate and dissolved organic carbon from land to coastal zones of the Baltic Sea. Journal of Marine Systems 177:8–20. Strauss, E. A., and G. A. Lamberti. 2000. Regulation of Nitrification in Aquatic Sediments by Organic Carbon. Limnology and Oceanography 45:1854–1859. Retrieved from http://www.jstor.org/stable/2670646 Strauss, E. A., and G. A. Lamberti. 2002. Effect of dissolved organic carbon quality on microbial decomposition and nitrification rates in stream sediments. Freshwater Biology 47:65–74. Striegl, R. G., G. R. Aiken, M. M. Dornblaser, P. A. Raymond, and K. P. Wickland. 2005. A decrease in discharge-normalized DOC export by the Yukon River during summer through autumn. Geophysical Research Letters 32:1–4. Stubbins, A., J. F. Lapierre, M. Berggren, Y. T. Prairie, T. Dittmar, and P. A. Del Giorgio. 2014. What’s in an EEM? Molecular signatures associated with dissolved organic fluorescence in boreal Canada. Environmental Science and Technology 48:10598–10606. Tank, J. L., E. J. Rosi-Marshall, M. A. Baker, and R. O. Hall. 2008. Are rivers just big streams? A pulse method to quantify nitrogen demand in a large river. Ecology 89:2935–2945. Taylor, P. G., and A. R. Townsend. 2010. Stoichiometric control of organic carbon-nitrate relationships from soils to the sea. Nature 464:1178–1181. Thouin, J. A., W. M. Wollheim, C. J. Vorosmarty, J. M. Jacobs, and W. H. McDowell. 2009. The biogeochemical influences of NO3, dissolved O2, and dissolved organic C on stream NO3 uptake. Journal of the North American Benthological Society 28:894–907. Tomasek, A. A., T. D. Barman, P. Wang, J. L. Kozarek, C. Staley, M. J. Sadowsky, and M. Hondzo. 2018. The Effects of Turbulence and Carbon Amendments on Nitrate Uptake and

113

Microbial Gene Abundances in Stream Sediment. Journal of Geophysical Research: Biogeosciences 123:1289–1301. Tromboni, F., S. A. Thomas, B. Gücker, V. Neres-Lima, C. Lourenço-Amorim, T. P. Moulton, E. F. Silva-Junior, R. Feijó-Lima, I. G. Boëchat, and E. Zandonà. 2018. Nutrient Limitation and the Stoichiometry of Nutrient Uptake in a Tropical Rain Forest Stream. Journal of Geophysical Research: Biogeosciences 123:2154–2167. Valett, H. M., J. A. Morrice, C. N. Dahm, and M. E. Campana. 1996. Parent Lithology, Surface- Groundwater Exchange, and Nitrate Retention in Headwater Streams. Limnology and Oceanography 41:333–345. van Buuren, S., and K. Groothuis-Oudshoorn. 2011. Mice: Multivariate Imputation by Chained Equations in R. Journal of Statistical Software 45:1–67. Vannote, R. L., G. W. Minshall, K. W. Cummins, J. R. Sedell, C. E. Cushing, and R. L. Vannote. 1980. The River Continuum Concept. Canadian Journal of Fisheries and Aquatic Sciences 37:130–137. Retrieved from http://search.proquest.com.libproxy.unh.edu/docview/19210100?accountid=14612 Venables, W. N., and B. D. Ripley. 2002. Modern Applied Statistics with S (4th ed.). Retrieved from http://www.stats.ox.ac.uk/pub/MASS4 Vitousek, P. M., and W. a. Reiners. 1975. Ecosystem Succession and Nutrient Retention: A Hypothesis. BioScience 25:376–381. Vonk, J. E., S. E. Tank, P. J. Mann, R. G. M. Spencer, C. C. Treat, R. G. Striegl, B. W. Abbott, and K. P. Wickland. 2015. Biodegradability of dissolved organic carbon in permafrost soils and aquatic systems: A meta-analysis. Biogeosciences 12:6915–6930. Wardle, D. A., G. Hornberg, O. Zackrisson, M. Katela-Brundin, and D. A. Coomes. 2003. Long- Term Effects of Wildfire on Ecosystem Properties Across an Island Area Gradient. Science 300:972–975. Wei, T., and V. Simko. 2017. R package “corrplot”: Visualization of a Correlation Matrix (Version 0.84). Weishaar, J. L., G. R. Aiken, B. A. Bergamaschi, M. S. Fram, R. Fujii, and K. Mopper. 2003. Evaluation of specific ultraviolet absorbance as an indicator of the chemical composition and reactivity of dissolved organic carbon. Environmental Science and Technology 37:4702–4708. Workshops, S. S. 1990. Concepts and methods for assessing solute dynamics in stream ecosystems concepts and methods for assessing solute dynamics in stream ecosystems. Journal of the North American Benthological Society 9:95–119. Worrall, F., R. Harriman, C. D. Evans, C. D. Watts, J. Adamson, C. Neal, E. Tipping, T. Burt, I. Grieve, D. Monteith, P. S. Naden, T. Nisbet, B. Reynolds, and P. Stevens. 2004. Trends in Dissolved Organic Carbon in UK Rivers and Lakes. Biogeochemistry 70:369–402. Wymore, A. S., R. L. Brereton, D. E. Ibarra, K. Maher, and W. H. McDowell. 2017. Critical zone structure controls concentration-discharge relationships and solute generation in forested tropical montane watersheds. Water Resources Research 53:6279–6295. Wymore, A. S., A. A. Coble, B. Rodríguez-Cardona, and W. H. McDowell. 2016. Nitrate uptake across biomes and the influence of elemental stoichiometry: A new look at LINX II. Glob. Biogeochem. Cycles 30:1–9. Wymore, A. S., J. Potter, B. Rodríguez-Cardona, and W. H. McDowell. 2018. Using In Situ Optical Sensors to Understand the Biogeochemistry of Dissolved Organic Matter Across a Stream Network. Water Resources Research 1–10.

114

Wymore, A. S., B. Rodríguez-Cardona, and W. H. McDowell. 2015. Direct response of dissolved organic nitrogen to nitrate availability in headwater streams. Biogeochemistry Vol. 126, pp. 1–10. Springer International Publishing. Zarnetske, J. P., R. Haggerty, S. M. Wondzell, and M. A. Baker. 2011. Dynamics of nitrate production and removal as a function of residence time in the hyporheic zone. Journal of Geophysical Research: Biogeosciences 116:1–12. Zimov, S. A., S. P. Davydov, G. M. Zimova, A. I. Davydova, E. A. G. Schuur, K. Dutta, and I. S. Chapin. 2006. Permafrost carbon: Stock and decomposability of a globally significant carbon pool. Geophysical Research Letters 33:1–5.

115