A Limnological Examination of the Southwestern Amazon, Madre de Dios,

by

Alana Urnesha Belcon

Earth and Ocean Sciences Duke University

Date:______Approved:

______Paul Baker, Supervisor

______Bruce Corliss

______Gary Dwyer

______Sherilyn Fritz

______Jennifer Swenson

Dissertation submitted in partial fulfillment of the requirements for the degree of Doctor of Philosophy in Earth and Ocean Sciences in the Graduate School of Duke University

2012

ABSTRACT

A Limnological Examination of the Southwestern Amazon, Madre de Dios, Peru

by

Alana Urnesha Belcon

Earth and Ocean Sciences Duke University

Date:______Approved:

______Paul Baker, Supervisor

______Bruce Corliss

______Gary Dwyer

______Sherilyn Fritz

______Jennifer Swenson

An abstract of a dissertation submitted in partial fulfillment of the requirements for the degree of Doctor of Philosophy in Earth and Ocean Sciences in the Graduate School of Duke University

2012

Copyright by Alana Urnesha Belcon 2012

Abstract

This dissertation investigates the limnology of the southwestern Peruvian

Amazon centered on the Madre de Dios department by examining first the geomorphology and then the ecology and biogeochemistry of the region’s fluvial systems.

Madre de Dios, Peru is world renowned for its prolific biodiversity and its location within the biodiversity hotspot. It is also a site of study regarding the development of the Fitzcarrald Arch and that feature’s geomorphological importance as the drainage center for the headwaters of the - the Amazon’s largest tributary and as well as its role as a physical divider of genetic evolution in the Amazon.

Though each of these has been studied by a variety of prominent researchers, the ability to investigate all the aspects of this unique region is hampered by the lack of a regional geomorphological map. This study aims to fill that gap by using remote sensing techniques on digital elevation models, satellite imagery and soil, geology and geoecological maps already in publication to create a geomorphological map. The resulting map contains ten distinct landform types that exemplify the dominance of fluvial processes in shaping this landscape. The river terraces of the are delineated in their entirety as well as the various dissected relief units and previously undefined units. The demarcation of the boundaries of these geomorphic units will provide invaluable assistance to the selection of field sites by future

iv

researchers as well as insights into the origin of the high biodiversity indices of this region and aid in planning for biodiversity conservation.

Secondly this study examines 25 tropical floodplain lakes along 300 km of the

Manu River within the Manu National Park in the Madre de Dios department.

Alternative stable state and regime shifts in shallow lakes typically have been examined in lakes in temperate and boreal regions and within anthropogenically disturbed basins but have rarely been studied in tropical or in undisturbed regions. In contrast this study focuses on a tropical region of virtually no human disturbance and evaluates the effects of hydrological variability on ecosystem structure and dynamics. Using satellite imagery a 23 yr timeline of ecological regime shifts in Amazon oxbow lakes or “cochas” is reconstructed. The study shows that almost 25% of the river’s floodplain lakes experience periodic abrupt vegetative changes with an average 3.4% existing in an alternative stable state in any given year. State changes typically occur from a stable phytoplankton-dominated state to a short lived, <3 yr, floating macrophytic state and often occur independent of regional flooding. We theorize that multiple dynamics, both internal and external, drive vegetative regime shifts in the Manu but insufficient data yet exists in this remote region to identify the key processes.

To complete the investigation of tropical limnology the third study compares and contrasts the nutrient-productivity ration of floodplain and non-floodplain lakes globally and regionally. For over 70 years a strong positive relationship between sestonic chlorophyll-a (Chl-a) and total phosphorus (TP) has been established with phosphorus

v

generally viewed as the most limiting factor to productivity. Most of these studies, however, have focused on northern, temperate regions where the lakes are typically postglacial, isolated and fed by small streams. Relatively little work has been done on floodplain lakes which are semi or permanently connected to the river. This study examines the relationship between nutrients and productivity in floodplain lakes globally through an extensive literature synthesis. Values for total phosphorus, total nitrogen and chlorophyll-a were collected for 523 floodplain lakes, represented by 288 data points while 551 data points were collected for 5444 non-floodplain lakes. Analysis revealed that globally, floodplain lakes do not show any significant difference in the total phosphorus/chlorophyll-a relationship from that found in non-floodplain lakes but significant differences are seen between tropical and temperate lakes. We propose that the term ‘floodplain’ lake should serve as purely a geographical descriptor and that it is lacking as an ecological indicator. Instead factors such as precipitation seasonality, hydrological connectivity and regional flooding regimes are better indicators of high or low productivity in floodplain lakes.

vi

Contents

Abstract ...... iv

List of Tables ...... x

List of Figures ...... xi

Acknowledgements ...... xv

1. Introduction ...... 1

2. The Geomorphological Units of the Madre de Dios Basin, Peru ...... 6

2.1 Introduction ...... 6

2.2 Data and methodology ...... 7

2.2.1 Study Area ...... 7

2.2.2 Methods ...... 9

2.2.2.1 Elevation analysis ...... 9

2.2.2.2 Landsat TM ...... 10

2.2.2.3 Additional Maps & Resources ...... 16

2.3 Results ...... 20

2.3.1 Fitzcarrald Arch ...... 20

2.3.2 Dissected lowland relief ...... 24

2.3.3 River Floodplain and Terraces ...... 24

2.3.4 Alluvial fans ...... 25

2.3.5 Lake District ...... 27

2.3.6 Madre de Dios River Aggradation Plains ...... 27

2.4 Discussion ...... 27

vii

2.4.1 Modern Depositional Landscapes ...... 29

2.4.1.1 Modern Meandering Floodplain ...... 29

2.4.2 Sub-modern Depositional Relief ...... 29

2.4.2.1 Pleistocene Alluvial Terraces ...... 29

2.4.2.2 Flood Basins ...... 30

2.4.3 Dissected Relief ...... 30

2.4.3.1 Fitzcarrald Arch ...... 30

2.4.3.2 Dissected Lowland Relief ...... 31

2.4.3.3 Dissected Alluvial Deposits ...... 31

2.4.4 Ancient Aggradation Plains ...... 32

2.4.4.1 Madre de Dios River Aggradation Plains ...... 32

2.4.4.2 Lake District ...... 32

2.4.5 Bamboo Forests ...... 34

2.5 Conclusion ...... 35

3. Twenty-three Year Timeline of Ecological Stable States and Regime Shifts in Amazon Oxbow Lakes ...... 36

3.1 Introduction ...... 36

3.2 Methods ...... 45

3.3 Results ...... 48

3.3.1 Ecological Regime Shifts ...... 48

3.4 Discussion ...... 54

3.4.1 Ecological Regime Shifts ...... 54

3.4.2 Lake Connectivity ...... 59

viii

3.5 Conclusion ...... 64

4. Global Phosphorus and Chlorophyll-a Relationships in Floodplain Lakes ...... 65

4.1 Introduction ...... 65

4.2 Methods ...... 67

4.3 Results ...... 69

4.4 Discussion ...... 88

4.5 Conclusion ...... 94

5. Conclusions and Future Research ...... 95

Appendix A - Non-Floodplain Lakes ...... 99

Appendix B - Floodplain Lakes ...... 117

Biography ...... 145

References ...... 129

ix

List of Tables

Table 1: Ten Principal Rivers of the World (Gupta 2008) ...... 2

Table 2: Name, Area and Percentage of Study Region of Each Geomorphic Unit...... 20

Table 3: Distribution of Landsat 5 images collected for the Manu National Park (1986- 2008) ...... 46

Table 4: Nutrient comparisons for various lake types with p-values for slope and intercepts. Green boxes represent statistically significant values...... 78

x

List of Figures

Figure 1: Madre de Dios study region located southeast Peru and ranging from the foothills of the eastern Andean Cordillera to the Amazonian lowlands...... 10

Figure 2: Comparisons of terrain texture were used to differentiate between dissected relief units...... 10

Figure 3: Example of abrupt elevation changes used to distinguish between geomorphic units...... 11

Figure 4: Mean slope calculated from 90m SRTM using a 10*10 cell moving window. Flatter (darker) areas include the alluvial megafan and the aggradation plains of the Madre de Dios River while lighter areas indicated more steeply sloped topography. .... 12

Figure 5: Locations of transects shown in Figs 6 and 8. White lines indicate transverse and longitudinal megafan transects while the black line shows the river terraces transect...... 13

Figure 6: Longitudinal and transverse elevation transects of an alluvial megafan in the Madre de Dios department. The location of the transects are shown in Fig 4. The concave shape of the transverse transect helps identify the unit as a megafan and not simply an alluvial deposit...... 14

Figure 7: 3D rendering of elevation model using ArcScene 10. Geomorphological features such as the alluvial megafans, dissected forests, rivers and river floodplains are all distinguishable...... 15

Figure 8: Investigation of river terraces using a) DEM images. Black line represents location of transect. b) Landsat TM band 4 satellite image and c) elevation transect created by the black line shown in a and b. A mean of 30m is subtracted by each terrace height to account for the 30m canopy cover...... 17

Figure 9: Elevation map (above) of the study region with the corresponding hypsometric curve (below)...... 21

Figure 10: Geomorphological map of the Madre de Dios region, Peru with an underlying elevation hillshade layer...... 21

Figure 11: Elevation transect of dissected alluvial deposits along the southwestern flank of the Andean foothills...... 24

xi

Figure 12: Landsat TM band 4 image showing typical lakes found within the ancient aggradation plains of the Madre de Dios River...... 24

Figure 13: SOTERLAC Soil map of the Madre de Dios region with region of round lakes outlined in white...... 28

Figure 14: Manu Biosphere Reserve comprising of the Manu National Park, the Manu Reserve Zone and the Cultural Zone...... 38

Figure 15: Annual Precipitation and Temperatures at Cocha Cashu Biological Station between 1984 and 1989...... 39

Figure 16: (a) Dominant phytoplankton state. (b) Submerged macrophytes – 2003. (c) Floating vegetation – 2006...... 42

Figure 17: Manu oxbow lakes or “cochas” used in study. Lakes 1-8 are lie within uncontacted tribe territory so their local names are unknown...... 44

Figure 18: NDVI images of Cocha Cashu before, during and after the 2006-2007 regime shift from phytoplankton to floating macrophytes (Pistia stratiotes). By September 2008, the floating vegetation mat had collapsed the dominant vegetation was once again phytoplankton. Surrounding forest vegetation NDVI values fluctuate depending on moisture content...... 49

Figure 19: NDVI timeline from 1986-2008 of Cocha Cashu (Lake 13). The 2006-2007 surface vegetation produces significantly higher values, whereas the submerged 2003 vegetation outbreak does not affect the NDVI value of the lake...... 50

Figure 20: Scatterplot of the NDVI values for Lake 14 – Cocha Totora. The red line represents 2δ above the mean while the green line shows the 0.3 NDVI minimum criteria for a regime shift to be identified...... 52

Figure 21: NDVI matrix for 25 Manu lakes from 1986 to 2008. Darker green represent increased surface vegetation...... 53

Figure 22: Percentage of the 25 studied Manu lakes that experienced regime shifts from phytoplankton-dominated to floating vegetation in any given year...... 57

Figure 23: NDVI timeline for Lake 9 – Cocha Gamarota showing the continuous maintenance of floating vegetation as the stable state on this lake...... 58

xii

Figure 24: Depiction of the effects on the vegetative state of Cocha Cashu by the mega- flood of 2003 that precipitated a switch from phytoplankton dominated to submerged macrophytes...... 61

Figure 25: Locations of lakes used in study. Orange circles represent floodplain lakes while blue circles show non-floodplain lakes. Each circle shows the location of one study unless the study sampled a large geographical location or multiple countries...... 70

Figure 26: Distribution diagrams of global total phosphorus, total nitrogen and chlorophyll values. The top panel shows raw values while the bottom panel shows log 10 values...... 71

Figure 27: Global Floodplain and Non-floodplain Mean Total Phosphorus, Total Nitrogen and Chlorophyll-a values...... 73

Figure 28: Global total phosphorus/chlorophyll-a regression...... 74

Figure 29: Global total nitrogen/total phosphorus regression...... 75

Figure 30: Global total nitrogen/chlorophyll-a regression...... 76

Figure 31: Total phosphorus/chlorophyll-a correlations for floodplain (orange) and non- floodplain lakes (blue)...... 77

Figure 32: Total nitrogen/chlorophyll-a correlations for floodplain (orange) and non- floodplain lakes (blue)...... 80

Figure 33: Total nitrogen/total phosphorus correlations for floodplain (orange) and non- floodplain lakes (blue)...... 81

Figure 34: Total phosphorus/chlorophyll-a correlations for floodplain (orange) and shallow (<3m) non-floodplain lakes (blue)...... 82

Figure 35: Total phosphorus/chlorophyll-a correlations for floodplain (orange) and deep (3-10m) non-floodplain lakes (blue)...... 83

Figure 36: Total phosphorus/chlorophyll-a correlations for temperate floodplain (orange) and temperate non-floodplain lakes (blue)...... 85

Figure 37: Total phosphorus/chlorophyll-a correlations for temperate floodplain (grey) and tropical floodplain lakes (red)...... 86

Figure 38: Total phosphorus/chlorophyll-a correlations for temperate lakes (purple) and tropical lakes (green)...... 87

xiii

Figure 39: Total phosphorus/chlorophyll-a regression lines for various studies...... 90

xiv

Acknowledgements

Thank you to my dissertation committee, Drs. Paul Baker, Bruce Corliss, Gary

Dwyer, Jennifer Swenson and Sherilyn Fritz, without whose assistance this dissertation would not have been completed. I would like to especially thank my advisor Paul Baker, whose passion for all science allowed me to craft a unique dissertation topic with his blessing and Bruce Corliss, who has been a mentor to me from day one and whose advice has guided me through varied life decisions. Jennifer Swenson and Sheri Fritz went above and beyond the call of duty in assisting me with developing my ideas and I must extend my gratitude to Edgardo Latrubesse at the University of Texas – Austin for dedicating weeks of his time to teaching me the art of geomorphological mapping. I’m also grateful to John Terborgh and Lisa Davenport whose in-depth knowledge of the

Manu is unparalleled and who provided the seeds for this body of work and finally, I would also like to thank Laura Johnson, Andy Nunnery and Trevor Nace for keeping me sane as we all went through this process simultaneously.

This dissertation would not have been possible without the funding support provided by the Center for Latin American and Caribbean Studies, the department of

Earth and Ocean Sciences and the National Geographic Society.

xv

1. Introduction

Any description of the largest rivers of the world, whether sorted by discharge or drainage area, results in a list dominated by tropical and sub-tropical locations (Table 1).

The Amazon River of South America discharges quadruple the volume of water into the ocean than the Congo River, the 2nd largest river by discharge. Together with the Nile

River, these three tropical rivers drain over 14,000,000 km2 of land, twice the area of the contiguous United States. If the classification is expanded to include the sub-tropics, this results in the addition of the Yangtze River in China, the world’s third longest river with one of the most populated river basins with over 400 million persons residing within its watershed (Varis and Vakkilainen 2001). In addition to the food, energy and transportation resources they provide, these rivers have spawned thousands of floodplain lakes which themselves serve as vital resources for local populations (Scheffer et al. 2006; Zeug et al. 2005). Yet, traditionally, limnological research has focused on rivers and large, deep lakes located in temperate latitudes. Even as studies have expanded to include the examination of shallow lakes, particularly in terms of water quality, nutrient content and ecological regime shifts between stable alternate states, still the focus has remained on temperate lakes. The study of tropical limnology is limited by both funding and accessibility but as global population growth places disproportionately greater pressure on tropical water resources it is increasingly important that we understand how tropical systems function and how they differ from

1

Table 1: Ten Principal Rivers of the World (Gupta 2008)

River Average Drainage Area Region discharge (km2) (m3/s) Amazon 175 000 6 915 000 South America

Congo 41 200 3 822 000 Africa

Yangtze 35 000 1 940 000 China (Changjiang) Brahmaputra 33 600 580 000 Bhutan Yenisey 18 040 2 580 000 Russia Zambezi 17 600 1 331 000 Africa Lena 16 200 2 490 000 Russia

Mississippi 15 500 3 230 000 North America Ganga 15 000 952 000 India Mekong 14 800 811 000 Southwest Asia

temperate systems. This dissertation investigates the field of tropical limnology, in particular the Madre de Dios region located in the southwestern Amazon, from three varying perspectives.

The Madre de Dios department is the third largest department in Peru with an area of 85,300 km2 and a population of only 109,600 (2007 census). The region lies within the Andean biodiversity hotspot (Myers et al. 2000) and contains the Manu National

Park where over 20,000 flora, 1,000 bird and more than 200 mammal species have been identified. The forest, rivers and lakes of the Madre de Dios are home to several endangered species such as the giant river otter (Pteronura brasiliensis), the black caiman

(Melanosuchus niger) and the jaguar (Panthera onca). The Madre de Dios River for which

2

the region is named has its headwaters in the foothills of the Andes and the Manu River, one of the tributaries of the Madre de Dios, is one of the few remaining tropical rivers with its entire watershed contained with protected land. In this case, the Manu National

Park which is 15,328 km2 and whose population is limited to a small number of indigenous tribes and research scientists. This presents a unique opportunity to study tropical limnology within a region that is limited to humans and therefore protected as much as possible from anthropogenic influences. This dissertation used this exceptional opportunity to examine three aspects of Amazon limnology.

In the first chapter, the geomorphology of the region is examined and mapped using satellite remote sensing. An understanding of the unique biodiversity of the

Amazon basin must be based on an understanding of the physical development of the landscape. The Madre de Dios basin is renowned for both its high biodiversity and its interesting geomorphology. The Fitzcarrald Arch, a 40,000 km2 highland feature with a mean elevation between 400 to 600 m begins at the base of the Andes, stretches northeast across the Madre de Dios region into Brazil. It acts as a drainage center for the southwestern Amazon and an evolutionary divider between the north and south

Amazon basins. Due to the extreme remoteness of the region, developing a detailed understanding of the geomorphology is nearly impossible and the current theories are based on river outcrops and temporary cut banks. Using a variety of remote sensing methods, this study aims to delineate the geomorphic units of the region and increase of understanding of the development through time of this unique terrain.

3

The second chapter continues the examination of the limnological features of the

Madre de Dios region by examining 25 floodplain lakes along the Manu River, within the Manu National Park. Specifically the study analyzes vegetative regime shifts within these lakes and the influence of hydrological connectivity on the initiation and duration of the ecological states. The examination of ecological regime shifts in lakes in such a pristine location is an extremely rare opportunity. Our understanding of ecological regime shifts in lakes is dominated by anthropogenic impacts such as over fishing, non- source and point source pollution and invasive species introduction. This investigation that builds a 25 yr timeline of naturally-occurring ecological regime shifts in pristine, tropical floodplain lakes is the first time such a study has been conducted and any knowledge gleaned will assist in the understanding of how these systems work and aid in the development of better lake management practices globally.

The third chapter widens the scope of the research and compares and contrasts floodplain lakes to non-floodplain lakes globally and regionally with regards to nutrients and productivity. The relationship between nutrients, in particular, phosphorus and nitrogen to algal production as measured by chlorophyll-a, has been established for many decades in the literature. However, no systematic examination of this relationship has been conducted for floodplain lakes. Due to their location on the floodplain of a river, these lakes are frequently inundated by water and can be flooded for weeks or months at a time. How does this hydrological connection to the parent river

4

affect the productivity of these lakes? Is there a global signal that differs from non- floodplain lakes? These are the questions that this third chapter addresses.

Collectively there three analyses coalesce into a body of work that contributes in multiple ways to out understand of tropical limnology. As populations in developing countries, largely located within the tropics, continue to rapidly increase the demands for food and water will continue to exert pressure on the rivers and lakes of these regions. Conservation and management of these resources are dependent on an improved understanding of the natural processes that govern these systems. It is therefore imperative that work such as this, which employs a multifaceted approach to tropical limnology, be encouraged and fast tracked while pristine, tropical regions still exist to be studied.

5

2. The Geomorphological Units of the Madre de Dios Basin, Peru

2.1 Introduction

The Madre de Dios region which is centered around the Peruvian department of the same name, lies within the Tropical Andes hotspot (Myers et al. 2000), is renowned for some of the highest biodiversity in the world (Brooks et al. 2006), and contains the

Manu National Park which is biological rich in both flora and fauna (Terborgh 1999).

The region includes the Fitzcarrald Arch, a 400m-600m a.s.l. physiographic feature that extends over 400,000 km2 from the base of the Andes in southern Peru into western

Brazil. The arch separates two major Amazonian tributary basins (Hermoza et al. 2005) and as such, may have served as a vicariant boundary (Daly and Mitchell 2000). There continues to be significant debate on the geodynamic origins of the Fitzcarrald Arch with possible mechanisms including (1) the subduction of the subduction of the buoyant

Nazca ridge (Espurt et al. 2007; Regard et al. 2009) (2) reactivation of Paleozoic structures (Jacques 2003) or simply a megafan (Picard et al. 2008).

However, little is known about the geological history of the lowlands of the

Madre de Dios region. Some work has been done on outcrops along rivers where stratigraphy can be observed and materials for dating can be collected. The river terraces along the Madre de Dios River have been examined to try to understand the origin of the modern Amazon drainage (Campbell Jr et al. 2006), the development of the lowland

6

relief (Räsänen et al. 1990) and to establish the paleoclimatological record for the southwestern Amazon (Rigsby et al. 2009). Apart from the river terrace outcrops, the inaccessibility of the interior and its dense vegetation cover make geological study without a drill core nearly impossible. For this reason we turn to producing maps based mostly on remote sensing.

The high biodiversity of the Madre de Dios region and by extension the Andean hot spot can be, in part, attributed to the effect of fluvial processes on the geomorphology of the landscape (Hamilton et al. 2007). Geomorphological maps are important representations of the surface and sub-surface processes that have shaped a landscape, thus they provide valuable information to geologists, engineers, conservationists, urban planners and foresters etc. Some work has been done on identifying the geoecological units of the Peruvian lowland forest (Räsänen et al. 1993) but there has been no previous geomorphological mapping of the Madre de Dios region as a whole. This study presents and interprets such a map, constructed using satellite imagery, high resolution elevation data, a few age dates from previous geological expeditions and other supplemental material.

2.2 Data and methodology

2.2.1 Study Area

The Madre de Dios region of the Peruvian Amazon contains some of the most highly endemic regions of the world (Swenson et al. 2012) with more than 20,000 flora species (Myers 1988), 1000 bird species and 200 mammal species . The elevation ranges 7

from over 1500m in the foothills of Andean eastern Cordillera to 300m in the Amazon lowland forest. The climate is largely humid tropical and the region experiences seasonal climate with a dry period from April to September and an average annual precipitation of 2300 mm measured at (Osher and Buol 1998). The annual temperature fluctuation is small but diurnal temperatures experience larger ranges with the mean monthly minimum, 16.8°C, occurring in July and the mean monthly maximum, 32°C, occurring in October.

Several large rivers exist within the Madre de Dios department. The Manu River originates in the foothills of the Andes and flows eastward and slightly southward to its intersection with the Alto Madre de Dios at Boca Manu where both rivers become the

Madre de Dios. The Inambari and Tambopata rivers flow northward down the flank of the eastern Andes. They merge with the Madre de Dios as it continues flowing eastward until it crosses into Bolivia and veers sharply northward. The Madre de Dios River is a major tributary for the Madeira River which itself is the largest tributary of the Amazon

River in the area of its watershed (1,380,000 km2), the magnitude of its discharge, averaging about 30,000 m3 s-1, and its total suspended sediment discharge (280 106 t year-

1) (Filizola and Guyot 2009) and is itself one of the largest rivers in the world. The area under study is 239,186 km2, comparable to the area of the United Kingdom and lies between 9°24’ S to 13°23’S and 67°43’W to 73°12’W. The study region has the Madre de

Dios political department at its center but extends beyond the boundaries of the department to include the depositional and erosional features of the Madre de Dios

8

River. It does not include the Andes Mountains and lies below 770 m (Fig 1) which allows for the entire Fitzcarrald Arch and Andean foothills to be mapped.

2.2.2 Methods

Determination and mapping of the geomorphological units of the Madre de Dios region was done using a variety of geospatial analyses. The primary process involved overlaying and hand digitizing the units in ArcGIS 10 (2010) using a combination of the following information: (1) elevation data from the Shuttle Radar Topography Mission

(SRTM) 90 meter resolution Digital Elevation Model (DEM), (2) Landsat Thematic

Mapper (TM), (3) USGS South American Geological map, (4) USGS South American Soil map, (5) the Soil and Terrain Database for Latin America and the Caribbean

(SOTERLAC) and (6) Räsänen’s Peruvian Geoecological map (Räsänen et al. 1993).

2.2.2.1 Elevation analysis

The SRTM elevation model was selected over the higher resolution Advanced Spaceborn

Thermal Emission and Reflection (ASTER) 30 meter Global DEM due to the decreased vertical accuracy with ASTER (Hirano et al. 2003). The reduced resolution did not affect the efficiency of the analysis as the geomorphological units of this region are measured in km2 so SRTM’s 90m resolution was more than sufficient. SRTM data was downloaded from SRTM CGIAR website in 1° tiles and mosaicked together using ESRI ArcMap 10.0 to cover the study area. Using ArcMap’s 3D Analyst extension, elevation transects were created to help identify units and the boundaries between units. Of particular

9

9°24’15”S

10

13°23’46”S

73°12’03”W 67°43’58”W

Figure 1: Madre de Dios study region located southeast Peru and ranging from the foothills of the eastern Andean Cordillera to the Amazonian lowlands.

importance was the differentiation of surface terrain texture (Fig 2) because changes in terrain ruggedness often indicate unit boundaries. These, along with abrupt changes in elevation along a transect (Fig 3), were used to identify changes different landscape units and boundaries. Additional surface and focal statistics were calculated in ArcGIS

10 to assist in the quantitative analysis of the landscape such as the slope and variance in slope within each unit (Fig 4). Figure 5 shows the location of alluvial megafans and river terraces where landform shapes were particularly important. For example, alluvial megafans have a distinctive convex cross-section (Fig 6) derived from their quasi- isochronous formation (Regard et al. 2009). ArcScene 10 enabled 3D rendering of the elevation model to further highlight unit structure and enable differentiation between units. Figure 7 shows a portion of the study area in 3D where the floodplain and terraces of the Manu River can be seen in the center of the scene with the dissected terra firme forest to the left and the alluvial megafans to the right.

2.2.2.2 Landsat TM

Landsat TM images consist of 7 bands, blue, green and red as well as near, mid, and far infrared, and one thermal band. Each band represents the spectral reflectance of the surface in different wavelengths. Images were downloaded from the United States

Geological Survey (USGS)’s and the Brazilian National Institute for Space Research

(INPE) archives. Composite images of different bands were used for analysis but

11

550

500 450 400

Elevation(m) 350 300 0 3 7 10 13 17 20 23 27 30 33

Distance (km) 12

Smooth terrain Rough terrain

Figure 2: Comparisons of terrain texture were used to differentiate between dissected relief units.

Abrupt elevation change

13

Figure 3: Example of abrupt elevation changes used to distinguish between geomorphic units.

14

Figure 4: Mean slope calculated from 90m SRTM using a 10*10 cell moving window. Flatter (darker) areas include the alluvial megafan and the aggradation plains of the Madre de Dios River while lighter areas indicated more steeply sloped topography.

15 c’

c

a’ b’ a b

Figure 5: Locations of transects shown in Figs 6 and 8. White lines indicate transverse and longitudinal megafan transects while the black line shows the river terraces transect.

280 270 Longitudinal Transect 260 250 240

230 Elevation (m) Elevation 220

210 a a’ 0 2 3 5 7 8 10 12 13 15 17 18 20 22 23 25 27 28 30 Distance (km)

280 b b’

270 16

260 250

240

Elevation (m) Elevation Transverse Transect 230 220 0 2 4 6 7 9 11 13 15 17 18 20 22 24 26 28 29 31 33 35 37 39 40 42 44 46 Distance (km)

Figure 6: Longitudinal and transverse elevation transects of an alluvial megafan in the Madre de Dios department. The location of the transects are shown in Fig 4. The concave shape of the transverse transect helps identify the unit as a megafan and not simply an alluvial deposit.

Madre de Dios aggradation Plains Alluvial Megafans

Dissected terra

firme forests

17

Figure 7: 3D rendering of elevation model using ArcScene 10. Geomorphological features such as the alluvial megafans, dissected forests, rivers and river floodplains are all distinguishable.

primarily band 4, the near infrared (NIR) (.76-.9 µm) band which is particularly sensitive to water and vegetation, worked well to highlight features in this highly fluvial forest environment. This band was especially useful in highlighting the river terraces where vertical distance from the river decreases soil water content and, therefore, changes the floral species assemblages (Hamilton et al. 2007). Different assemblages reflect light differently enabling demarcation of each terrace (Fig 8). Each Landsat image covers

31,820 km2 requiring a mosaic of 12 images to cover the entire study area. Each image was radiometrically and atmospherically corrected using the revised calibration procedures (Chander et al. 2009) and dark object subtraction (Song et al. 2001). Due to the high mean cloud cover of this region and the necessity of using completely cloud free images, there is variance in the date and time of the images used spanning from July

1996 to September 2009. This resulted in large differences in sun angle and atmospheric haze and therefore, radiometric and atmospheric correction were unable to bring all the images to the same reflectance. In addition to Landsat imagery, Google Earth was also used to visually analyze surface morphology.

2.2.2.3 Additional Maps & Resources

Several previously published maps assisted in the construction of this geomorphological map. Räsänen’s geoecological map of the Peruvian Amazon (Räsänen et al. 1993) was particularly useful in this study. Additionally, the geology, soil and political maps from

18

a c’ b c’

c c

c c c’

19

Figure 8: Investigation of river terraces using a) DEM images. Black line represents location of transect. b) Landsat TM band 4 satellite image and c) elevation transect created by the black line shown in a and b. A mean of 30m is subtracted by each terrace height to account for the 30m canopy cover.

the USGS (USGS 2000) and Soil and Terrain Database for Latin America and the

Caribbean (SOTERLAC) from the World Data Center for Soils (WDC-Soil) (Dijkshoorn et al. 2005) were overlain in ArcGIS and used as guides in describing the individual units. Due to study region being heavily forested, the geology and soil maps lack detail and artificial boundaries often arise at political boundaries because different countries use different classification schemes. Ages for some units from radiocarbon and radiometric dating were used from the literature.

2.3 Results

With the exclusion of the Andean Cordillera, the elevation of the study region ranges from 771 m to 116 m and is distributed in a non-symmetrically concentric pattern cut by a radial drainage pattern (Fig 9). Ten unique geomorphic units were identified in the

Madre de Dios region (Table 2, Fig 10) and can be divided into the following major categories:

2.3.1 Fitzcarrald Arch

The most distinctive topographic feature of the Madre de Dios region is the Fitzcarrald

Arch, which is defined as the region with elevation values greater than 400m. The Arch occupies 14% of the study area and can be divided into 2 sections – a major unit of

20

33,492 km2 area that stretches eastward from the base of the Andes and a minor unit of

1,881 km2 that separated from the major unit by dissected lowland relief.

Figure 9: Elevation map (above) of the study region with the corresponding hypsometric curve (below). 21

Table 2: Name, Area and Percentage of Study Region of Each Geomorphic Unit.

Geomorphic Unit Area (km2) Percentage of Elevation Variance in Slope Study Area Range (m) (STD) (°) Fitzcarrald Arch 33,492 14.01 262-771 2.88 Dissected Relief-Madre de 19,251 8.05 191-485 2.07

Dios Dissected Relief – Lowland 111,508 46.65 116-662 2.22

Forest Lake District 16,801 7.03 162-389 1.58

22 Aggradation Plains – Madre 15,538 6.50 138-286 1.37

de Dios Forested Swamp 15,870 6.64 144-260 0.94 Modern Floodplain 7,905 3.31 147-512 1.98 Alluvial Terraces 1454 0.61 177-409 1.74

Megafans 13,643 6.52 178-448 1.47 Dissected Alluvial Deposits 3,427 1.43 299-689 2.56

23

Figure 10: Geomorphological map of the Madre de Dios region, Peru with an underlying elevation hillshade layer.

2.3.2 Dissected lowland relief

There are two major units of dissected lowland relief that are distinguished from each other by their different elevations. The first encompasses the Manu/Madre de Dios and

Las Piedras rivers, sums to 31,734 km2 which is 8% of the study area and consists of higher mean elevation, ~380 m, but more gently dissected terrain. The second unit covers the entire northern portion of the study area and exhibits lower elevations, but more defined dissected relief. This latter unit covers 110,961 km2 which is 46% of the total study area; however, this unit extends beyond the study area and into Brazil and

Bolivia.

2.3.3 River Floodplain and Terraces

The floodplains of the two major rivers within the study area, Madre de Dios and

Las Piedras Rivers, were digitized and both the modern and older river terraces of the

Madre de Dios River were identified. Collectively, the meander plains and river terraces of the two rivers comprised 4% of the study area. T3, the highest terrace at 70-75 m above the Madre de Dios River (Fig 8) covers 472 km2. These relative elevation values are calculated with a 25m adjustment for the height of the forest vegetation because

ASTER records canopy height rather than ground values and the average canopy height of the Madre de Dios forest is estimated to be 25 m (Gentry 1990). T2, the middle terrace

24

is 25-30 m high and 982 km2 are while T1, the youngest terrace is closest to the river and lies only 5 m above it.

2.3.4 Alluvial fans

The eastern fringe of the Andes Mountains in the Madre de Dios basin consists of a series of 4 northeast downward sloping alluvial fans ranging from 30 km to 80 km in length. Their concave shapes are similar to those of the megafans of the Central Andes

(Dade and Verdeyen 2007; Horton and DeCelles 2001). Their mean size, 3410.75 km2, is smaller than those in Bolivia, though their total area, 13,643 km2 falls within the typical

Andean range. Northwest of the megafans are two smaller alluvial depositional wedges that sum to 3427 km2. These features differ from the larger megafans in two major ways:

(1) shape - they do not exhibit the concave shape and instead generally slope down in an eastward direction and (2) relief – unlike the megafans which exhibit ‘smooth’ terrain, the alluvial depositional wedges are heavily dissected (Fig 11). The magnitude of dissection decreases as the further southeast the fan are with the largest, southern most fan exhibiting the smoothest, least dissected terrain. The depositional wedges and the megafans are all transected by straight, often braided rivers such as the Alto Madre de

Dios that flow down the eastern flank of the Andes into the Madre de Dios Basin.

25

Elevation Elevation (m)

Distance (m)

26 Figure 11: Elevation transect of dissected alluvial deposits along the southwestern flank of the Andean foothills.

2.3.5 Lake District

One distinct unit, located on the eastern part of the study zone that straddles the border between Peru and Bolivia, is distinguished by numerous small, circular or near circular lakes (Fig 12), covers 16801 km2 and occupies 7% of the total study zone. These lakes are noted by Räsänen (1993), and have been sampled by (Bush et al. 2007). There are approximately 50 lakes in the region which are typically less than 1 km in diameter and are concentrated in a T formation running north-south to the Peru/Brazilian border.

2.3.6 Madre de Dios River Aggradation Plains

Along the Peru/Bolivia boundary, the old aggradation floodplains of the Madre de Dios

River can be seen. Former meaner channels, scrollbars and oxbow lakes can be observed parallel to the modern channel course. The average elevation of this floodplain region is

300 m and it extends for 15,538 km2, some of which also extends beyond the study area.

2.4 Discussion

The evolution of the Madre de Dios landscape has been shaped by two dominant forces – the climate of the region and the geodynamics of the Andes Mountains. The high precipitation characteristic of the region ensures that fluvial processes dominate the landscape and adjust ancient foreland basins. The most noteworthy physiographic feature of the study area is the Fitzcarrald Arch. Indeed, the fluvial processes of this

27

28

Figure 12: Landsat TM band 4 image showing typical lakes found within the ancient aggradation plains of the Madre de Dios River.

region are directed by the Fitzcarrald Arch. The various geomorphic units of the study area can be characterized as modern or ancient and as either erosional or depositional landscapes.

2.4.1 Modern Depositional Landscapes

2.4.1.1 Modern Meandering Floodplain

The modern floodplains of the study area were created by the downcutting and subsequent infilling by the Madre de Dios and Las Piedras Rivers. These are white water

Amazonian rivers with their sources in the Andean Mountains (Madre de Dios) and the

Fitzcarrald Arch (Las Piedras). High sediment loads result in rapid channel erosion with meander migration rates as high as 25 m/yr (Terborgh and Petren 1991). The successional vegetation prevalent along the river margins has a species composition differing greatly from the lowland forest. As much as 12% of the Peruvian lowland forest exists in successional stages along river channels (Salo et al. 1986).

2.4.2 Sub-modern Depositional Relief

2.4.2.1 Pleistocene Alluvial Terraces

Three alluvial terraces were identified on the northern boundary of the Madre de

Dios River (Fig 8). T3, the highest terrace is composed of the Miocene Inururi Formation, unconformably overlain by the late Miocene-Pleistocene Madre de Dios Formation. T2 is intermittently present along the Madre de Dios and radiocarbon dating of wood from the T2 terrace yields dates ranging from 25,040 ±130 cal yrs BP to 29,850 ±100 cal yrs BP 29

(Rigsby et al. 2009). T1, the most recent terrace has been dated from 3,780 ±50 to 11,970

±100 cal yrs BP. According to Rigsby et al., the Madre de Dios Formation was downcut before 29,850 cal yrs B.P and then infilled by fluvial strata creating the T3 surface. These aggradated strata were then downcut forming T2 before 11,970 cal yrs B.P. which was then infilled and downcut again after 3,780 to form T3. Vegetation type and composition is strongly affected by terrace elevation and its effect on soil saturation through distancing from the river.

2.4.2.2 Flood Basins

Found just beyond the meander plains of the modern rivers, flood basins are regions of the lowest alluvial terrace that are seasonally or permanently flooded and are often characterized by swamp vegetation (Räsänen et al. 1993). Though they comprise less than 1% of the Madre de Dios study region, they are much more prevalent in the

Maronon and Ucayali basins of the Pastaza fan in Central Peru and account for as much as 6.5% of all Peruvian lowland forest (Räsänen et al. 1993).

2.4.3 Dissected Relief

2.4.3.1 Fitzcarrald Arch

Dissected forests form the major elevated regions of the study area. The highest elevations are dominated by the Fitzcarrald Arch which acts as a source region for many of the tributary rivers of the region. The Fitzcarrald Arch is considered to be the elevated

30

landform between 450-600 m. Although the arch actually extends into Brazil, the highest elevations of the Arch are within the Madre de Dios basin.

2.4.3.2 Dissected Lowland Relief

Surface erosion by the numerous small streams flowing off of the Fitzcarrald

Arch has created highly dissected relief that constitute a large part of the Madre de Dios region. Differentiation between units is based on degrees on dissection and mean elevation. The SOTERLAC soil database lists 3 varieties of cambisols, soils with weak horizon differentiation, for the dissected plains for the Madre de Dios.

2.4.3.3 Dissected Alluvial Deposits

Large fluvial deposits, categorized as megafans and alluvial deposits and dated to the Pliocene and Pleistocene, are found along the flank of the Andean Cordillera

(Räsänen et al. 1990; Regard et al. 2009). Though little work has been done on these small

Madre de Dios megafans, research conducted on their much larger Bolivian counterparts indicate that they are deposited through frequent avulsions and might necessitate that a drainage basin reach a threshold size, 104km2, before they are formed

(Horton and DeCelles 2001).

31

2.4.4 Ancient Aggradation Plains

2.4.4.1 Madre de Dios River Aggradation Plains

The region lies on the transition from lowland forest to savannah vegetation as the climate becomes drier from west to east with mean annual precipitation values falling to 2000 mm yr-1 (Bush et al. 2007). At the southernmost extent of the Peru/Bolivia border, within the aggradation plains, the dissected forest gradually changes to swamp/savannah vegetation that is characterized by poorly-drained depressions and vast swamp areas similar to the savannah region of the Beni of Bolivia (Räsänen et al.

1993). This zone continues for a significant portion of the northern Bolivian lowlands and only the westernmost boundary coincides with our study area.

2.4.4.2 Lake District

The Lake District lies within the old aggradation plains of Madre de Dios River. Lake sediment cores sampled from three of the lakes between 1990-2001 (Bush et al. 2007) gave dates between 6,000-8,000 yrs B.P. and showed evidence of evidence of settlement

(charcoal) at multiple lakes and agriculture at one. The lakes are found within two soil zones classified as acrisols (Fig 13). An acrisol is defined as having clay-enriched subsoil

(FAO 1998). This region also lies within the transitional climate zone between the wetter forests of the western Madre de Dios and the drier savannahs of the Beni region. The precipitation seasonality results in cycles of soil saturation that first leach iron from the soil during flooding and then subsequently precipitates iron oxides during dry periods

32

33

Figure 13: SOTERLAC Soil map of the Madre de Dios region with region of round lakes outlined in white.

(Fritsch et al. 2007; Osher and Buol 1998). We theorize that the formation of laterite duricrust has created the impermeable layers that allow the lakes to form but field studies are needed to investigate this suggestion.

2.4.5 Bamboo Forests

Investigation of the Landsat satellite images revealed large areas of light colored vegetation. Smith and Nelson have identified that as much as 165,000 km2 in the southwestern Amazon is covered by forest dominated by two woody bamboo species,

Guadua sarcocarpa and Guadua weberbaueri (Smith and Nelson 2011). Numerous studies have investigated the mystery of the southwestern Amazon bamboo forests (Griscom and Ashton 2003; Nogueira et al. 2008; Torezan and Silveira 2000) with no correlation found between the topography and the vegetation distribution nor any obvious relationship between soil type and vegetation. One hypothesis links fire disturbance to the growth of bamboo forests, particularly those found in the Southwestern Amazon that flower and die synchronously, hence providing abundant fuel for forest fires (Jon and Bond 1999). Though wide-spread in the Madre de Dios region, the bamboo forests were not placed on this geomorphological map as there are no obvious links between geomorphology and these distinctive forests.

34

2.5 Conclusion

Investigation of digital elevation models, soil and geological maps, geoecological classification and radiometric date provided in the literature have enabled the differentiation of the geomorphological units of the Madre de Dios region stretching from the foothills of the Eastern Andean Cordillera to the Peru/Brazil border. This region contains the unusual Fitzcarrald Arch which acts as a hydrological drainage center for the region and generates the headwaters for many rivers in the region and forms a hydrological divide between N-flowing and S-flowing Amazon tributaries.

Further field work is deemed necessary in order to validate the boundaries of the defined units as well as to gain a better understanding of the origins and history of the identified units. Despite its limitations, this geomorphological map provides useful insight into the landforms of this biologically diverse region and will enable better informed decisions regarding future geological and biological field work and conservation practices.

35

3. Twenty-three Year Timeline of Ecological Stable States and Regime Shifts in Amazon Oxbow Lakes

3.1 Introduction

The study of alternative stable states in shallow lakes and the triggers for regime shifts have been topics of growing concern over the past few decades (Neubert and

Caswell 1997; Scheffer et al. 2001). The stable state of an ecosystem is described as that set of biotic interactions and abiotic conditions under which the system resides. These states are resistant to change because of feedback mechanisms that maintain the system within a given range of conditions, but a variety of perturbations can trigger a rapid switch, or regime shift, from one stable state to another. Most studies of limnological regime shifts studies have been undertaken in temperate and boreal regions, typically within anthropogenically impacted basins. The introduction of excess nutrients, particularly phosphorus and nitrogen, into streams and lakes has led to eutrophication of lakes and rivers, with resultant water-quality degradation, fish kills and disruption of trophic structure (Carpenter 2003). Few studies have evaluated the mechanisms behind natural regime shifts in shallow lakes (Scheffer and Jeppesen 2007), and such studies are difficult because of the pervasive impact of human activities. To address this lack of knowledge, this study investigates the occurrence of ecological regime shifts in oxbow lakes in a protected national park along the 300 km long Manu River, Peru.

The Manu River is a white-water, upper tributary of the Amazon River and has its headwaters in the foothills of the Andes Mountains, which are its source of nutrients.

36

It is one of the few remaining large rivers whose entire watershed lies within protected forest before joining with the Madre de Dios River. The Manu Biosphere Reserve (12⁰S,

71⁰W) contains the Manu National Park, with 1.5 million ha of protected tropical forest, as well as the Reserved Zone with an additional 257,000 ha, and the restricted Cultural

Zone with an area of 91,400 ha (Fig 14). Elevation in the National Park ranges from 300 m in the lowlands to 1500 m in the Andes. This rapid transition in elevation gives rises to diverse habitat and some of the highest biodiversity and species endemism in the continent and world (Brooks et al. 2006). The Manu is home to over 20,000 species of plants, 1000 bird species, more than 200 mammal species and many more. It serves as protected habitat to several endangered Amazonian keystone species, including the

Black Caiman (Melanosuchus niger) and the Giant River Otter (Pteronura brasiliensis), which both use the Manu’s oxbow lakes for feeding, mating and raising their young.

The region has a seasonal climate with a dry period from April to September (Fig 15) and an average annual precipitation of 2300 mm measured at Puerto Maldonado (Osher and Buol 1998).

Human habitation in the park is limited to small communities of Amazonian tribal groups, predominantly the Machiguenga. Fewer than 2,000 persons are estimated to live within the National Park (similar in size to the state of Connecticut) including estimates of several communities uncontacted by modern civilization (Levi et al. 2009).

Tourism is allowed in the Cultural and Reserved Zones but not within the National Park where the only visitors are researchers at the Cocha Cashu Biological Station.

37

Figure 14: Manu Biosphere Reserve comprising of the Manu National Park, the Manu Reserve Zone and the Cultural Zone.

38

39

Figure 15: Annual Precipitation and Temperatures at Cocha Cashu Biological Station between 1984 and 1989.

There are no roads within the park, ensuring that access occurs only along the river. This limited accessibility has ensured the preservation of the park and severely limits anthropogenic influences. Though observed data and modeled predictions show reduced precipitation in the Central Andean Mountains that feed the Manu River

(Urrutia and Vuille 2009; Vuille et al. 2003) and the record Amazonia drought of 2005 did affect the Madre de Dios department (Marengo et al. 2008), we do not yet have any evidence that the flooding regime of the Manu River has been affected by global warming.

Many oxbow lakes lie along the Manu River within a 7-km-wide floodplain, resulting from a meander migration rate of up to 25 m/yr (Terborgh and Petren 1991).

The lakes can be categorized as ‘connected’ or ‘unconnected’, where connected lakes have permanent or semi-permanent streams that connect them to the river, whereas unconnected lakes have no connection with the river expect during periods of high flood

(Davenport 2008; Tejerina-Garro et al. 1998). Connected lakes are frequently inundated with sediment-bearing water from the Manu River and therefore quickly fill up often lasting just a few decades. Unconnected lakes fill much more slowly due to their reduced connection with the river and can last centuries. Manu lakes exhibit three vegetative states - i) phytoplankton (algae), ii) submerged macrophytes and iii) floating macrophytes. Phytoplankton domination is the most common state of the observed

40

lakes though fringe littoral communities of submerged and floating macrophytes are not uncommon (LD pers. obs.).

The Cocha Cashu Biological Station was established in the 1960’s on the banks of one of the river’s oxbow lakes, Cocha Cashu, from which it takes its name. ‘Cocha’ is the

Quechua word for ‘lake’. The lake is 1.4 km long and 150 m at its widest. The maximum depth is 4.0 m with an average depth of just 1.2 m. The lake is unconnected from the river, with no perennial streams connecting it to the river, although floods can inundate it. For the first 30 years of observation, Cocha Cashu existed in a phytoplankton dominated state with little or no floating or submerged macrophytes (Fig 16a).

Floods in the Manu are placed in two categories – ‘normal’ floods and ‘mega- floods’. Normal floods occur several times during each rainy season and can be localized, with some tributaries and lakes being flooded while others are not. Mega- floods, however, occur once every few years, typically after days of extremely heavy rainfall, and the resulting deluge sweeps over the entire floodplain, scouring lake sediments and clearing the forest floor of debris. Three mega-floods have been observed at Cocha Cashu – 1983, 1999 and 2003. The first observed regime shift occurred immediately after the 2003 mega-flood, which swept over the entire flood plain. As the water receded, the lake rapidly clarified, and a thick layer of submerged macrophytes, primarily Najas aguta, began to grow along the bottom of the lake (Fig 16b). This was the first observed occurrence of highly transparent water and submerged vegetation at

41

a b

c

42

Figure 16: (a) Dominant phytoplankton state. (b) Submerged macrophytes – 2003. (c) Floating vegetation – 2006. Photos courtesy of Davenport & Terborgh.

Cocha Cashu. Simultaneously at Cocha Salvador, a larger and slightly deeper (max depth - 4.4 m, mean depth – 2.3 m) lake that lies 50 km downstream of Cocha Cashu (Fig

17), the same flood resulted in a shift from phytoplankton prevalence to a one-time occurrence of floating vegetation Nuphar spp. as well as submerged Najas spp. in the shallower regions of the lake. The deepest parts of the lake remained phytoplankton dominated and the lake has remained in a phytoplankton state ever since. The submerged vegetation that flourished at Cocha Cashu after the 2003 mega-flood, persisted until January 2006, when a normal flood inundated the lake. The ecological state switched rapidly from submerged Najas to a thick, mat of floating vegetation –

Pistia stratioles (water cabbage) (Fig 16c). The mat lasted until 2007, when it was colonized by islands of sedge, Scirpus cubensis with high water clarity (Secchi depth

>2m). Over the dry season the mat of combined vegetation died, and by September 2008 submerged vegetation again flourished throughout the deeper parts of the lake. This regime shift from floating vegetation to submerged vegetation was not accompanied by any floods.

Due to the protected nature of the national park and the limited anthropogenic impacts, the observed regime shifts are assumed to be natural and not anthropogenically driven. Given the available data, we have structured our study to address the following questions: Do all the lakes of the Manu experience alternative states and abrupt shifts between them? How often do these shifts happen? What are the possible mechanisms

43

44

Lake 1 Lake 6 Lake 11 – Cocha Secreta Lake 16 – Cocha Gallareta Lake 21– Cocha Sacarita Lake 2 Lake 7 Lake 12 – Cocha Nueva Lake 17 – Cocha Salvadorcillo Lake 22 – Cocha Juarez Lake 3 Lake 8 Lake 13 – Cocha Cashu Lake 18 – Cocha Salvador Lake 23 – Cocha Garza Lake 4 Lake 9 – Cocha Gamarota Lake 14 – Cocha Totora Lake 19 – Cocha Otorongo(a) Lake 24 – Cocha Largarto Lake 5 Lake 10 – Cocha Maisal Lake 15 – Cocha Panawea Lake 20 – Cocha Otorongo(b) Lake 25 – Cocha Tipisco

Figure 17: Manu oxbow lakes or “cochas” used in study. Lakes 1-8 are lie within uncontacted tribe territory so their local names are unknown.

that both drive the shifts and maintain the states? Given the range of difficulties - remote access, no historical data collection – we turned to the use of historical satellite imagery to answer these questions.

We approached the problem by constructing an ecological timeline of the Manu lakes reaching as far back as possible with available satellite imagery. In February 2009, the entire Landsat image archive was made freely available for download, which enabled us to investigate the history of ecological states of the Manu oxbow lakes from

1986 to 2008 and examine the possibility that the 2003 and 2006 regime shifts observed at

Cocha Cashu might be natural, floodplain-wide phenomena.

3.2 Methods

The Manu National Park and most of the Reserved Zone lie within a complete

Landsat satellite image (Row 4, Path 68; Landsat 5 and 7). Although oxbow lakes exist along the entirety of the Manu River, we limited our study to those upstream of the

Pinguen River confluence and well within the restricted zone of the Bioreserve to exclude possible impacts from the town of Boca Manu, located on the immediate outskirts of the Cultural Zone. To detect changes in lake vegetation, all available images with less than 50% cloud cover over the floodplain were collected from the United States

Geological Survey (USGS)’s and the Brazilian National Institute for Space Research

(INPE)’s archives. This resulted in a compiled dataset of 27 Landsat-5 and 1 Landsat-7 images spanning 23 years from 1986 to 2008, with all but one image captured during the 45

dry season between April and October (Table 3). The distribution of satellite images allowed for the investigation of annual and seasonal variability among the lakes, as several years had multiple images.

All images were radiometrically and atmospherically corrected using the revised calibration procedures (Chander et al. 2009) and Dark Object Subtraction (Song et al.

2001). To detect lake surface vegetation, Bands 3 (Red) and 4 (Near Infrared) of the corrected images were then used to create a Normalized Difference Vegetation Index

(NDVI) image, (NIR-Red) / (NIR+Red) (Peñuelas et al. 1993).

Table 3: Distribution of Landsat 5 images collected for the Manu National Park (1986-2008)

Image Date 1986 Aug 04 1998 Jun 18 1987 Aug 23 1998 Aug 05 1989 Aug 12 1998 Sep 06 1989 Oct 07 1999 Jul 07 1990 May 27 2000 Sep 27 1990 Jul 14 2001 Aug 13 1990 Sep 16 2003 Apr 04 - Landsat 7 1991 Jul 17 2004 Aug 05 1992 Jul 11 2004 Sep 22 1995 Apr 23 2005 Jan 12 (rainy season) 1996 Jun 12 2006 Jun 24 1996 Jul 14 2007 May 26 1997 Aug 18 2007 Aug 14 1997 Sep 03 2008 Sep 01

Previous image processing by Swenson and Bowers (unpublished) of Cocha Cashu indicated that NDVI was more successful than Tasseled Cap transformations or other change-detection methods to identify surface vegetation on the Manu lakes.

46

To investigate detection of sub-surface or suspended vegetation, we explored a range of indices along with NDVI, such as the Normalized Difference Water Index

(NDWI), which uses Landsat bands 4-Near Infrared and 5-Shortwave Infrared, (NIR-

SWIR)/(NIR+SWIR), and reflects changes in vegetation water content. A variety of focal statistics, mean, mode, variance, range and standard deviation were calculated spatially using a 3-cell moving window in an attempt to highlight textural changes and possibly identify differences in surface vegetation. However, the spatial resolution proved to be too low to successfully differentiate between floating vegetation species so these results were not included in this study.

NDVI images were evaluated both visually and statistically; visual interpretation was particularly important in the identification of partial lake surface coverage. Twenty- five lakes were identified and used for the study, starting with the most upstream lake on the Manu River. The first eight lakes lie within uncontacted tribe territories and therefore have not been given names by the Machiguenga tribes. They are identified in this body of work simply as Lakes 1 to 8. The remaining 17 lakes are labeled by number, as well as by the local names. The 25 lakes were individually hand digitized for all 28

NDVI images rather than creating one polygon per lake for all time steps. This procedure, though lengthy, was done to decrease inaccuracies, as lake size and boundary location fluctuate each year depending on season, recent precipitation etc.

47

Within each lake, 300 random points were sampled to calculate NDVI mean and standard deviation.

3.3 Results

3.3.1 Ecological Regime Shifts

NDVI values mathematically range from -1 to +1 but all vegetation produce positive values. Manu forest ranges between 0.6 to 0.7 NDVI, whereas the lake surfaces varied between 0.0 and 0.53. The observed floating vegetation coverage at Cocha Cashu in 2006-2007 is readily visible in the satellite images (Fig 18). However, the regime shift from phytoplankton to submerged macrophytes that occurred in 2003 and was documented in the field could not be distinguished due to the interference of the overlying water. The 2006-2007 floating vegetation registers with values as high as 0.53

NDVI, while the 2003 event appears as only water, with NDVI values at 0 (Fig 19). The same results were gained with the water index, NDWI, with no changes observed for

Cocha Cashu in the 2003 image. Our focal statistics readily identified the boundaries between vegetation and water, but Landsat’s 30m resolution was too coarse for vegetation species mapping. Given these results we concluded that our methods could only be used confidently for the detection of surface vegetation and not submerged vegetation.

48

July 1999 January 2005 June 2006

49

August 2007 September 2008

Figure 18: NDVI images of Cocha Cashu before, during and after the 2006-2007 regime shift from phytoplankton to floating macrophytes (Pistia stratiotes). By September 2008, the floating vegetation mat had collapsed the dominant vegetation was once again phytoplankton. Surrounding forest vegetation NDVI values fluctuate depending on moisture content.

50

Figure 19: NDVI timeline from 1986-2008 of Cocha Cashu (Lake 13). The 2006-2007

surface vegetation produces significantly higher values, whereas the submerged 2003 vegetation outbreak does not affect the NDVI value of the lake.

Surface vegetation regime shifts were identified using a combination of visual and statistical methods. A regime shift from phytoplankton-dominated water to surface vegetation dominance was determined to have occurred if the NDVI value was more than 2σ (standard deviations) away from the mean NDVI for that lake and greater than

0.3 (Fig 20). As Landsat’s pixel size is 30m and the oxbow lakes are typically 100m wide

(3-4 pixels), edge vegetation can impact the mean NDVI value for a lake. Requiring a minimum NDVI value of 0.3 reduces the margin of error caused by lake boundaries.

Additionally, we required that visual analysis of the lake verified surface vegetation coverage ≥ 75% of the lake surface. Partial vegetative states were identified if NDVI values fell below 0.3 but surface vegetation coverage was ≥ 30% of the lake surface or conversely, if NDVI values were above 0.3 but the vegetation coverage was ≤ 30% indicating a small but very dense proliferation.

This analysis identified floating vegetation shifts in 9 lakes (Fig 21 ) which is 36% of all lakes studied with the transitions occurring with some level of synchronization.

1990 showing shifts in 3 lakes, 2004 had 4 lakes and 2006/2007 having shifts in 7 lakes.

Upper watershed lakes experienced many more shifts than lower watershed lakes with only 3 lakes in the lower 10 having regime shifts. Lakes 4 and 9-Cocha Gamarota were the only lakes with floating vegetation as the dominant vegetation type though it should

51

52

Figure 20: Scatterplot of the NDVI values for Lake 14 – Cocha Totora. The red line represents 2δ above the mean while the green line shows the 0.3 NDVI minimum criteria for a regime shift to be identified.

53

Figure 21: NDVI matrix for 25 Manu lakes from 1986 to 2008. Darker green represent increased surface vegetation.

be noted that Lake 4 was a dying lake and was rapidly losing area in each subsequent image.

A pattern of rapid floating plant growth, persistence for one to three years followed by a rapid return to a phytoplankton-dominated state, appears to be a recurrent pattern for regime shifts in these lakes. Of the shifts identified only one lake,

Lake 11, had floating vegetation for 4 years. All other shifts lasted 1-2 years.

Additionally, the transformations are very rapid, with the majority of lakes transitioning from NDVI values of close to 0 (indicating bare water) to values 3.0 or higher in less than a year, indicating fairly lush, thick growth.

3.4 Discussion

3.4.1 Ecological Regime Shifts

Despite the inability of our remote sensing method to identify periods of submerged vegetation dominance, the technique was successful in identifying shifts from phytoplankton to floating vegetation based on significant changes in NDVI (0-0.3).

Additionally, the availability of the Landsat database has enabled the construction of an ecological timeline for a region where fieldwork is difficult, expensive and potentially dangerous. One drawback to our approach, however, is the limited number of available satellite images due in part to the extremely high cloud cover in this region as well as to 54

the reduced storage of scanned imagery in low demand locations. Therefore, although the Landsat 5 satellite was launched in 1984 and scans the globe every 16 days, only 27 images were considered adequate for our study. A second limitation is that given the average brief duration of floating vegetation regimes shown though our remote sensing analysis, we speculate that regime shifts that occurred between satellite images might have been missed as annual flooding washes away the floating vegetation. Thirdly,

Landsat satellite imagery is unable to identify submerged vegetation in these lakes, such as the growth that occurred at Cocha Cashu in 2003. The analysis done in this project, along with field observations in the Manu (Terborgh and Davenport, unpublished), indicates that the natural primary state of the Manu’s oxbow lakes is one dominated by phytoplankton with fringes of submerged or floating macrophytes. Therefore while it is possible that our method missed regime shifts from phytoplankton to submerged vegetation, this transition has a much lower likelihood than that from phytoplankton to floating vegetation and we are confident that we have documented the majority of regime shifts that have occurred on the dates for which we have images.

Although we have only one rainy season satellite image, early dry season images indicate that shifts from phytoplankton to floating vegetation typically establish at the end of the rainy/beginning of the dry season. Thick floating mats rapidly develop during the dry season with the highest NDVI values recorded between May and July. In the

55

majority of the shifts, these mats are typically gone or largely reduced by the next year’s dry season, likely because floods washed out the floating plants during the rainy season.

In any given year the majority of Manu’s lakes are in a phytoplankton dominated state regardless of connectivity, with floating vegetation covering less than 12% of the lakes

(Fig 22) with a mean of 3% and standard deviation of 4%. Only one lake, Lake 9 –Cocha

Gamarota exhibited floating vegetation as the primary state with an average NDVI value ≥ 0.3 for 20 of the 28 images (Fig 23). Communication with the local Machiguenga tribe indicates that the lake’s name - “Gamarota” is their word for the Pistia stratiotes plant, further substantiating that this lake’s stable state is that of floating vegetation.

The prevalence of regime shifts in upper watershed lakes is likely directly related to lake depth with upper watershed lakes being shallower than their lower counterparts.

Floating macrophytes are dependent on high water nutrient content and are therefore favored by shallow lakes where sediment nutrient sequestration is less likely due to wind-driven turbulence (Scheffer et al. 2003). However, that the primary stable state of

Manu lakes is phytoplankton-dominated is in direct contrast to most shallow lakes where the pristine state is one of submerged macrophytes (Scheffer et al. 2001) and eutrophication is typically the result of anthropogenic influences resulting in the addition of nutrients or the effect of catastrophic events such as a hurricane

56

20

18

16

14

12

10

8 Percentage (%) Percentage

6 57

4

2

0

Year

Figure 22: Percentage of the 25 studied Manu lakes that experienced regime shifts from phytoplankton-dominated to floating vegetation in any given year.

0.7

0.6

0.5

0.4

0.3

0.2 58

0.1

0.0

90 Jul90 Jul91 Jul92 Jul96 Jul99

05 Jan05

98 Jun98 96 Jun96 Jun06

89 Oct89

95 Apr95

90Sep 97Sep 98Sep 00Sep 04Sep 08Sep

97 Aug 97 87 Aug87 Aug89 Aug98 Aug01 Aug04 Aug07

Aug86

90 May90 May07

Figure 23: NDVI timeline for Lake 9 – Cocha Gamarota showing the continuous maintenance of floating

vegetation as the stable state on this lake.

(Carpenter 2003; Dent et al. 2002). Even among studies, phytoplankton- dominated lakes are unusual, especially in the lower Amazonia where most studies have been conducted. These investigations described lakes partially or completely covered with floating vegetation regardless of season (Camargo and Esteves 1995;

Tundisi 1983). Oxbow lakes in the lower Parana River, the second largest river in South

America, display similar features with Central Amazon lakes where annual flooding lasts for months and lakes are permanently covered with floating vegetation (Camargo and Esteves 1995; Izaguirre et al. 2004).

3.4.2 Lake Connectivity

One major difference between lower Amazonian lakes and the Manu watershed is the flooding regime. Unlike the lower Amazon where annual flooding arrives in a wave that inundates the entire floodplain for months (Richey et al. 1989), flooding in the

Manu is quite flashy in nature, occurring rapidly after periods of heavy rain and characteristically lasting only a few weeks at a time during the rainy season

(unpublished river level data). These flash floods occur several times a year, creating hydrological conditions different to those experienced downstream. Although regime shifts from one ecological state to another can occur without a flood, as seen in Cocha

Cashu in 2008, the flooding regime of the Manu can be used as one possible explanation why phytoplankton domination is the primary stable state for this region’s lakes. With 59

such frequent flooding and resultant high turbidity, photosynthesis would be limited for submerged macrophytes. Additionally frequent inundation would likely wash floating macrophytes out of the lake making it difficult for them to persist over many years.

The mega-flood of 2003 that precipitated the shift from phytoplankton to submerged vegetation at Cocha Cashu was the result of several days of extremely heavy rain producing flood waters that were much more dilute in nutrients than the river during normal flow. This impoverished water swept through the lake, replaced the previous nutrient-rich basin and as the flood subsided, this nutrient-poor water was unable to support a large biomass of phytoplankton. The lack of phytoplankton resulted in the clarification of the lake for the first time in 30 years of observation and allowed sunlight to penetrate to the bottom, initiating the growth of a rich layer of benthic vegetation (Fig 24). This same flood resulted in all three vegetative phases, submerged, floating and phytoplankton at Cocha Salvador for the same reasons except that

Salvador’s greater depth likely prevented submerged vegetation growth in the deeper parts of the lakes. It is noteworthy that the 2003 mega-flood was the third such flood observed at Cocha Cashu since the 1970’s, yet the first two did not result in regime shifts nor did the 2003 flood trigger changes in any lakes other than Cocha Cashu and Cocha

Salvador. In systems with alternative stable states internal feedback mechanisms maintain the lake in a given state until a critical threshold is reached and then the system

60

61

Figure 24: Depiction of the effects on the vegetative state of Cocha Cashu by the mega-flood of 2003 that precipitated a switch from phytoplankton dominated to submerged macrophytes. 61

rapidly switches from one state to another (Scheffer et al. 2001). It is therefore likely that for these nutrient-rich lakes significant dilution is necessary to eradicate the lake’s ability to sustain algae. While there are no historical records of lake nutrient content before and after the 2003 mega-flood and how it compared to the previous observed mega-flood in

1983, we postulate that the 2003 flood either contained lower nutrient levels. We theorize that the phytoplankton biomass was therefore reduced to such low levels that light- limitation was no longer a hindrance to the growth of submerged macrophytes and the vegetation underwent a regime shift.

Manu’s numerous, erratic floods may also explain the short residence times for floating vegetation mats in both connected and unconnected lakes, as the mats establish themselves during the dry season but are subsequently washed away during the rainy season floods. Floating vegetation has a higher chance of survival in an unconnected lake as floods are typically less frequent and of lower magnitude.

However, given the frequency of floods in the Manu but the rarity of vegetative state changes, the dynamics of the regime shifts cannot be explained using only the regional flooding regime.

Brought together, all the transitions observed in the Manu coalesce into a model with phytoplankton as the primary stable state for the region’s oxbow lakes, while floating vegetation states are rare, and short-lived states. Although observations have

62

not yet recorded a transition from submerged macrophytes to phytoplankton dominated, we theorize that this is a possibility in unconnected lakes. Such a switch would likely require conditions that allow the buildup of nutrients in the lake through leaf fall and ground water flow and a prolonged absence of major floods.

Some discussion can be held on the validity of labeling the submerged and floating vegetative conditions as ‘stable states’. Alternative stable states are defined as the biotic and abiotic conditions that maintain themselves through positive feedbacks and resist change unless there’s a sufficiently large perturbation (Beisner et al. 2003).

However, can a floating vegetation episode be considered a stable state? The average lifetime of surface vegetation cover, based on our geospatial analysis, is just 1 year. In most stable state studies, lakes typically switch vegetation types due to anthropogenic influences, such as non-point source pollution or over fishing (Carpenter and Brock

2006; Daskalov et al. 2007; Scheffer et al. 2001). The conditions that brought about the regime shift are typically long lasting with resultant stability of the new vegetative state.

Additionally shallow lakes in most well-studied regions exist in one of two ‘natural’ states – phytoplankton and submerged vegetation (Gunderson 2000). Dominance by floating vegetation is less common in these temperate lakes and usually is a product of invasive species (Scheffer et al. 2003). However in tropical lakes floating vegetation has been documented as a stable state (Gopal 1987; Scheffer et al. 2003) where for example,

63

fluctuating water level in Lake Kariba, Africa causes regime shifts from submerged to floating vegetation. Our data suggests floating plant dominance may be a recurrent albeit short-lied state in regions with frequent, short-lived, highly turbid floods.

3.5 Conclusion

We reconstructed a 23 yr timeline of the ecological stable states and regime shifts of the oxbow lakes of the Manu National Park and showed that the shifts are natural, floodplain wide phenomena. Our work suggests that although the prevailing understanding of ecological regime shifts in shallow lakes is heavily biased towards anthropogenically impacted ecosystems, in fact natural regime shifts between multiple stable states occur in the floodplain lakes in the upper reaches of the Amazon Basin.

Further research in protected regions with limited anthropogenic impacts, such as the

Manu, is necessary to fully understand the mechanisms behind natural regime shifts in freshwater systems.

64

4. Global Phosphorus and Chlorophyll-a Relationships in Floodplain Lakes

4.1 Introduction

Phosphorus is the major nutrient that limits primary productivity in many freshwater systems, and its strong linear relationship with sestonic chlorophyll has been studied for many decades (Dillon and Rigler 1974a; Phillips et al. 2008). The majority of lakes included in studies of the phosphorus – chlorophyll relationship are deep, temperate lakes that experience seasonal stratification, with resulting nutrient depletion during summer months. A stratified lake can have up to 50% of its total phosphorus content during the summer in the hypolimnion and sediments (Guy et al. 1994). This phosphorus is therefore unavailable for algal growth during periods of stratification.

Nutrient cycling in shallow lakes differs from that of deep lakes as most shallow lakes experience no or only diurnal stratification such that dissolved nutrients are cycled into the photic zone each day. Although some phosphorus is lost to the sediment in shallow lakes, resuspension can occur rapidly due to wave action and/or the foraging of fish and other benthic organisms (Søndergaard et al. 2003). Like deep lakes, most studies of shallow lakes are from sites in temperate zones and emphasize lakes that are post glacial in origin. The shallow nature of these lakes lends itself to the proliferation of submerged macrophytes across much of the lake floor, which reduces the effect of waves on the sediment layer and traps suspended particles, resulting in decreased turbidity and increased clarity (Jacques 2003). Shallow lakes are a model system of

65

alternative stable states based on observed state changes from clear waters with submerged macrophytes to a eutrophic, phytoplankton dominated state (Scheffer et al.

1993; Van Geest et al. 2003).

One type of shallow system that may not fit the paradigm of submerged macrophytes vs phytoplankton domination is floodplain lakes. Floodplain lakes, as the name implies, are shallow lakes that lay within the floodplain of a river. Many are remnant river meanders, although other types of floodplain lakes do exist. These lakes perform important ecological and economical roles in both developed and developing countries (Scheffer et al. 2006; Zeug et al. 2005). In tropical regions where heavy precipitation and numerous rivers have led to the formation of thousands of floodplain lakes, they often are important sources of food for local communities and are also sources of regional biodiversity in that they increase landscape heterogeneity

(Agostinho et al. 2004; Ward et al. 1999). Throughout the Amazon, thousands of floodplain lakes provide shelter for juvenile fish, reptiles and mammals, as well as food sources for adult organisms. They serve as vital habitat for endangered animals, such as the black caiman (Melanosuchus niger) and the giant river otter (Pteronura brasiliensis)

(Carter and Rosas 1997), and communities depend on these lakes for significant portions of their sustenance (Almeida et al. 2003; de Castro and McGrath 2003).

Many floodplain lakes are within the flooding range of their parent river and are connected to the river via small streams. Depending on a lake’s connectivity to the river, inundation may occur frequently each time the river level rises or infrequently if the lake

66

is isolated from the river. The frequent disturbance regime of many floodplain lakes and its impact on ecosystem structure has the potential to affect patterns of nutrient cycling and phytoplankton biomass. Frequent disturbance may limit submerged macrophytic growth and instead favour floating vegetation and phytoplankton (Engle and Melack

1993). Thus, we hypothesize that the phosphorus-chlorophyll relationship derived in previous studies (references) may not adequately characterize the pattern characteristic of floodplain lakes. Yet a systematic evaluation of the relationship between nutrients and productivity for floodplain lakes does not exist, nor does a comparative study between floodplain and non-floodplain lakes. This study conducts a global analysis of the nutrient/productivity relationship for shallow lakes. We compare floodplain and non-floodplain lakes in an effort to investigate what, if any, differences exist between the two lake types.

4.2 Methods

A systematic search of two well-established limnological journals – Freshwater

Biology and Limnology & Oceanography was used for data collection. Every article published over the past ten years (2001-2011) was reviewed for values of total phosphorus (TP), total nitrogen (TN), chlorophyll-a (Chl-a), total suspended solids

(TSS), mean lake depth (z), secchi depth (SD) and for floodplain lakes, connectivity to the river. A floodplain lake may have permanent or seasonal streams that connect it to the main river and is considered a ‘connected’ lake or it may not and is then considered

67

an ‘isolated’ lake. In addition, key word searches were also performed in Web of Science and Google Scholar. Search terms included a variety of combinations of “lakes, shallow, nutrients, phosphorus, chlorophyll and floodplain”. No date limitations were used for key word searches. Non-floodplain lakes deeper than 10m were rejected from the database to allow for better comparisons with floodplain lakes, nearly all of which are shallow. Additionally, any studies with nutrient manipulation as an intrinsic part of the methods were rejected, as were as any lakes identified as saline or as seasonal lakes.

Studies that reported only soluble phosphorus or orthophosphate or nitrate values rather than total phosphorus or total nitrogen were analyzed separately.

Descriptive, regression and interaction analyses of the data were conducted using JMP Pro 9 (2010) Statistical Analysis System (SAS) software. Tests of interactions evaluated whether the slope and intercept of the regression lines of each category were statistically different from each other, which is a test of whether the relationship between nutrients and chlorophyll differs depending on lake type. Analyses of the nutrients - chlorophyll relationship were done for all floodplain lakes versus all non- floodplain lakes and then followed by analyses according to latitude, as well as lake depth. Individual tests for outliers for each lake and nutrient combination and eliminated before analysis. All lakes located between 30⁰N and 30⁰S were grouped into a tropical/subtropical group, as the number of tropical lakes alone was insufficient for a separate analysis group. All lakes in latitudes greater than 30⁰ N and S were grouped as temperate lakes. Regressions and tests of interactions were made between temperate

68

floodplain and non-floodplain lakes and between tropical floodplain to temperate floodplain lakes. Aside from analysis by latitude, non-floodplain lakes were divided into shallow lakes (<3m) and deep lakes (3-10m). All the floodplain lakes were characterized as shallow lakes. Regressions and test of interactions were then made between floodplain lakes and shallow and deep non-floodplain lakes.

4.3 Results

Methodical searching of Freshwater Biology and Limnology and Oceanography produced 52 studies from those journals with the requisite data while key word searches produced 24 usable studies. Some studies reported individual data points for each lake sampled while others described only mean values resulting in 287 data points for 523 floodplain lakes and 505 data points for 5,444 non-floodplain lakes. Twenty-five countries were represented (Fig 25), and the majority of the studies were located in Europe (46%). Most of the studies reported total phosphorus, 99%, while 58% gave total nitrogen, 89%

Chlorophyll-a, 40% secchi depth, 33% mean depth, and only 1% reported total suspended solids or connectivity of floodplain lakes to the river. Given the limited number of values for secchi depth, total suspended solids and connectivity, these variables were not analysed further. Additionally orthophosphate and nitrate values were reported in too few studies for separate analysis and were therefore discarded from the database. Distribution histograms of the total phosphorus, total nitrogen and chlorophyll-a data revealed extreme one-tailed skew in all instances (Fig 26). To

69

30°N

30°S

70

Figure 25: Locations of lakes used in study. Orange circles represent floodplain lakes while blue circles show non-floodplain lakes. Each circle shows the location of one study unless the study sampled a large geographical location or multiple countries.

180 700 700 160 600

600 140

500 500 120 400 100 400

300 80 300 Frequency Frequency 60 Frequency 200 200 40 100 20 100 0 0 0

Total Phosphorus (µg/l) Total Nitrogen (µg/l) Chlorophyll-a (µg/l)

250 120 140 100 120

200

71

100 80 150 80 60 60

100 Frequency

Frequency 40

Frequency 40 50 20 20 0 0 0

Log Total Phosphorus (log µg/l) Log Total Nitrogen (log µg/l) Log Chlorophyll (µg/l)

Figure 26: Distribution diagrams of global total phosphorus, total nitrogen and chlorophyll values. The top panel shows raw values while the bottom panel shows log 10 values.

compensate, the natural logs of total phosphorus, total nitrogen and chlorophyll-a were calculated and used for all analysis. Comparisons of total phosphorus, total nitrogen and chlorophyll-a between floodplain and non-floodplain lakes consistently show narrower ranges and higher means for floodplain lakes than for their non-floodplain counterparts

(Fig 27).

Total nitrogen was the only measurement with more reported values in floodplain lake studies than in non-floodplain studies. Even so, the values show greater ranges and lower means for the non-floodplain lakes. As expected, globally, total phosphorus and chlorophyll-a have a statistically significant correlation, r2=0.5889, p <

.0001 and show a linear relationship of LogChl=0.9217 LogTP-0.4438 (Fig 28).

Phosphorus and nitrogen themselves are significantly correlated, r2=0.6577, p<.0001 and the relationship is linear: LogTN=0.6034 LogTP+1.8311 (Fig 29). The total nitrogen/chlorophyll-a correlation exhibits an r2 = 0.567, p<.0001 and a linear relationship

LogChl=1.0683LogTN-1.9503 (Fig 30). When separated into floodplain and non- floodplain lakes, significant differences are seen in the total phosphorus/chlorophyll intercept but not between the slopes between lake types (Fig 31). Indeed all the intercepts had statistically significant differences between all comparisons (Table 4). The linear regression for floodplain lakes is LogChl=0.871LogTP-0.3356, r2=0.6714 and the regression for non-floodplain lakes is LogChl=0.8548LogTP-0.3686, r2=0.5084. Floodplain lakes exhibit a tighter relationship, r2 = 0.6714 compared to r2 = 0.513 for non-floodplain lakes.

72

120 120 150

100 100

80 80 100

60 60

Frequency Frequency 50 Frequency 40 40 20 20 0 0 0

Total Phosphorus (log µg/l) Log Total Nitrogen (µg/l) Log Chlorophyll-a (µg/l) 75

n Mean Log TP (µg/l) Mean Log TN (µg/l) Mean Log Chl (µg/l)

Floodplain 287 1.78 2.94 1.17 73

Tropical 91 1.95 3.03 1.20 Temperate 196 1.62 2.84 1.13 Non-Floodplain 505 1.41 2.50 0.88 Tropical 11 1.46 2.48 0.90 Temperate 494 1.35 2.51 0.85 Figure 27: Global Floodplain and Non-floodplain Mean Total Phosphorus, Total Nitrogen and Chlorophyll-a values.

3

2.5

2

Chl = 0.9217TP - 0.4438

1.5 R² = 0.5889

g/l) µ 1

a a (log -

74 0.5

Chlorophyll 0 0 0.5 1 1.5 2 2.5 3 3.5

-0.5

-1

-1.5 Total Phosphorus (log µg/l)

Figure 28: Global total phosphorus/chlorophyll-a regression.

4

3.5

3

2.5 TN = 0.6034TP + 1.8311 R² = 0.6577

2

1.5

TotalNitrogen (log µg/l) 75

1

0.5

0 0 0.5 1 1.5 2 2.5 3 3.5 Total Phosphorus (log µg/l)

Figure 29: Global total nitrogen/total phosphorus regression.

3

Chl = 1.0683TN - 1.9503 2.5 R² = 0.567

2

1.5

1

0.5

TotalNitrogen (log µg/l)

0 76

0 0.5 1 1.5 2 2.5 3 3.5 4

-0.5

-1

-1.5 Chlorophyll-a (log µg/l)

Figure 30: Global total nitrogen/chlorophyll-a regression.

3 Chl = 0.871TP - 0.3356 R² = 0.6714

2.5

2

1.5

a (µg/l) a - 1 Chl = 0.8824TP - 0.4119 R² = 0.5131

0.5

Log Chlorophyll Log 0 -0.5 0 0.5 1 1.5 2 2.5 3 3.5 4

77 -0.5

-1

-1.5 Total Phosphorus (µg/l) Non-Floodplain Lakes Floodplain Lakes Linear (Non-Floodplain Lakes) Linear (Floodplain Lakes)

Figure 31: Total phosphorus/chlorophyll-a correlations for floodplain (orange) and non-floodplain lakes (blue).

Table 4: Nutrient comparisons for various lake types with p-values for slope and intercepts. Green boxes represent statistically significant values.

Lake Comparisons Nutrient relationship Slope Intercept

Floodplain vs Non-floodplain TP:Chl 0.1972 0.0005 TN:Chl 0.2644 <0.0001 TP:TN 0.0014 <0.0001 Floodplain vs Shallow Non- TP:Chl 0.1831 0.0020 floodplain Lakes Floodplain vs Deep Non-floodplain TP:Chl 0.7505 <0.0001 Lakes

78 Temperate Floodplain vs TP:Chl 0.6446 <0.0001 Temperate Non-floodplain Lakes Temperate Floodplain Lakes vs TP:Chl 0.9361 <0.0001 Tropical Floodplain Lakes Temperate Lakes vs Tropical Lakes TP:Chl 0.2801 <0.0355

Investigation of the total nitrogen/chlorophyll-a relationship by lake type resulted in statistically significant differences among lake types (Fig 32). Floodplain lakes exhibited an r2 = 0.4802 and a linear regression line of LogChl=1.107 LogTN-2.0317 while non- floodplain lakes had an r2 = 0.5127 and a linear regression of LogChl=0.9436LogTN-

1.6961. The intercepts were different (p<.005) from each other but not the slopes were not (p=0.2644). The relationship of total nitrogen/total phosphorus was the only one to experience different slopes and intercepts between floodplain and non-floodplain lakes

(Fig 33). Floodplain lakes have a TN/TP regression line LogTN=0.5211LogTP+1.9887 with an r2=0.6138 while for non-floodplain lakes the relationship is

LogTN=0.8066LogTP+1.755, r2=0.5811. The intercepts were statistically different (p

<.0001), whereas the slopes were not significantly different from one another (p=0.0014).

All other comparisons showed significantly different intercepts but no difference in slopes. In the comparison of the global data set of shallow non-floodplain lakes, <3 m, to floodplain lakes, the floodplain lakes have a regression line of LogChl =0.8653LogTP-

0.3273 and an r2=0.6669. Shallow non-floodplain lakes have a linear relationship expressed as LogChl=0.7622LogTP-0.3554 with an r2=0.443 (Fig 34). Interestingly, similar results were gained when comparing deep, 3-10m, non-floodplain lakes to floodplain lakes (Fig 35). Although deep non-floodplain lakes display lower mean total phosphorus and chlorophyll values and substantially more scatter, r2=0.3485, than floodplain lakes, r2=0.6714, the regression lines for both types of lakes are very similar: non-floodplain lakes: LogChl=0.9026LogTP-0.5882 and floodplain lakes:

79

3

2.5 Chl = 1.107TN - 2.0317 R² = 0.4802 2

1.5

g/l)

µ a ( a - 1 Chl = 0.9436TN - 1.6961

0.5 R² = 0.5127 80

LogChlorophyll

0

0 0.5 1 1.5 2 2.5 3 3.5 4

-0.5

-1

-1.5 Log Total Nitrogen (µg/l) Floodplain Lakes Non-floodplain Lakes Linear (Floodplain Lakes) Linear (Non-floodplain Lakes)

Figure 32: Total nitrogen/chlorophyll-a correlations for floodplain (orange) and non-floodplain lakes (blue).

4 TN = 0.5211TP + 1.9887 R² = 0.6138

3.5

TN = 0.6443TP + 1.7595 3 R² = 0.5811

2.5

LogTotal Nitrogen (ug/l) 2

81

1.5

1 0 0.5 1 1.5 2 2.5 3 3.5 Log Total Phosphorus (ug/l)

Floodplain Non-Floodplain Lakes Linear (Floodplain) Linear (Non-Floodplain Lakes)

Figure 33: Total nitrogen/total phosphorus correlations for floodplain (orange) and non-floodplain lakes (blue).

3

2.5 Chl = 0.8653TP - 0.3273 R² = 0.6669 2 Chl = 0.7622TP - 0.3554 R² = 0.443

1.5 a a (ug/l) - 1

82

0.5

LogChlorophyll

0 0 0.5 1 1.5 2 2.5 3 3.5 4

-0.5

-1 Log Total Phosphorus (ug/l)

Floodplain Lakes Shallow Non-floodplain Lakes Linear (Floodplain Lakes) Linear (Shallow Non-floodplain Lakes)

Figure 34: Total phosphorus/chlorophyll-a correlations for floodplain (orange) and shallow (<3m) non-

floodplain lakes (blue).

3

2.5 Chl = 0.871TP - 0.3356 R² = 0.6714 2

1.5

a a (ug/l) - 1

Chl = 0.9026TP - 0.5882 0.5 R² = 0.3485

83

LogChlorophyll 0 0 0.5 1 1.5 2 2.5 3 -0.5

-1

-1.5 Log Total Phosphorus (ug/l)

Deep Non-floodplain Lakes Floodplain Lakes Linear (Deep Non-floodplain Lakes) Linear (Floodplain Lakes)

Figure 35: Total phosphorus/chlorophyll-a correlations for floodplain (orange) and deep (3-10m) non- floodplain lakes (blue).

LogChl=0.871LogTP-0.3356. Investigating differences based on latitude led to similar results. There were not enough tropical non-floodplain lakes to compare lake types in this region. However temperate floodplain and non-floodplain lakes have remarkably similar total phosphorus/chlorophyll relationships (Fig 36). Temperate floodplain lakes showed a linear regression LogChl=0.9246LogTP-0.373 and an r2=0.7646. Temperate non- floodplain lake’s regression was LogChl=0.9593LogTP-0.5133 with an r2=0.5513.

The final analysis looked solely at floodplain lakes divided by latitude to investigate any effects climate might have on the nutrient/productivity relationship in floodplain lakes. All temperate lakes were compared to all tropical lakes with tropical lakes having similar slopes to temperate lakes but a lowered regression line (Fig 37). The temperate lakes had a regression line LogChl=0.9667TP-0.4976 and an r2=0.6036 while tropical lakes were Chl=0.8463TP-0.451 with an r2=0.5172. Tropical floodplain lakes were compared to temperate floodplain lakes (Fig 38). Tropical floodplain lakes had higher mean total phosphorus levels and therefore a significantly different intercept but they did not show any significant difference in the phosphorus/chlorophyll slope. The linear regression for tropical floodplain lakes was LogChl=0.9316LogTP-0.6311 with an r2=0.6006. The temperate floodplain lakes displayed less scatter with an r2=0.7646 and a regression equation Log[Chl]=0.9246Log[P]-0.373.

84

3

Chl = 0.9246TP - 0.373 2.5 R² = 0.7646

2

Chl = 0.9593TP - 0.5133

1.5 R² = 0.5513

1

a a ug/l) (log -

0.5 Chlorophyll

0 85

0 0.5 1 1.5 2 2.5 3 3.5

-0.5

-1

-1.5 Total Phosphorus (log ug/l)

Temperate Non-floodplain Lakes Temperate Floodplain Lakes Linear (Temperate Non-floodplain Lakes) Linear (Temperate Floodplain Lakes)

Figure 36: Total phosphorus/chlorophyll-a correlations for temperate floodplain (orange) and temperate non-floodplain lakes (blue).

3

2.5 Chl = 0.9246TP - 0.373 R² = 0.7646

2

Chl = 0.9316TP - 0.6311 1.5 R² = 0.6006

a a ug/l) (log -

1

Chlorophyll

0.5 86

0 0 0.5 1 1.5 2 2.5 3

-0.5 Total Phosphorus (log ug/l)

Temperate Floodplain Lakes Tropical Floodplain Lakes Linear (Temperate Floodplain Lakes) Linear (Tropical Floodplain Lakes)

Figure 37: Total phosphorus/chlorophyll-a correlations for temperate floodplain (grey) and tropical floodplain lakes (red).

22

3

Chl = 0.9667TP - 0.4976 2.5 R² = 0.6036

2

Chl = 0.8463TP - 0.451

1.5 R² = 0.5172

1

a a ug/l) (log -

0.5 Chlorophyll

87 0 0 0.5 1 1.5 2 2.5 3 3.5

-0.5

-1

-1.5 Total Phosphorus (log ug/l)

Tropical Lakes Temperate Lakes Linear (Tropical Lakes) Linear (Temperate Lakes)

Figure 38: Total phosphorus/chlorophyll-a correlations for temperate lakes (purple) and tropical lakes (green).

4.4 Discussion

This limnological analysis highlighted several of the biases and challenges that exist in the field. Although 25 countries are represented in the database, 42% of the studies are located in Europe, 30% in North America and only 15% in the tropics. Measurements of sestonic nutrients were common, although there was considerable variation in the nutrient form that was measured. While most studies reported total phosphorus and total nitrogen a variety of phosphate and nitrate species were reported. Additionally only 40% of the identified studies reported secchi depth, an easily measurable representation of lake water clarity and only 33% reported lake depth. Less than 1% of all the publications described watershed land use and, of the floodplain lakes, only three studies described the connectivity of the lakes to the river. These inconsistencies in field measurements limit global analysis to just phosphorus, nitrogen and chlorophyll-a values.

Distribution histograms of total phosphorus, total nitrogen and chlorophyll all show greater ranges and lower means for non-floodplain lakes than for floodplain lakes.

The upper values for both lake types are very similar, but non-floodplain lakes consistently had lower extreme values, leading to decreased means and greater ranges.

It is not unexpected that floodplain lakes on a whole, maintain higher nutrient

88

and primary productivity values, because frequent incursions from the river can increase available nutrients. Some rivers, such as the lower Amazon, bring nutrient -rich water into its floodplain lakes on a recurrent basis, boosting productivity (Kern et al.

2002). However even relatively nutrient-poor rivers can increase suspended nutrient content, as floodwaters disturb the sediment layer. Flooding can also prevent submerged macrophyte growth, which in the long term decreases resuspension rates

(Scheffer 1998).

Consolidating all lakes regardless of type produces a linear TP/Chl-a regression equation LogChl=0.9217Log[TP]-0.4438. This equation has a shallower slope than many of the established TP/Chl regression lines (Fig 39) (Nicholls and Dillon 1978; Phillips et al. 2008), but these were typically generated from a relatively small number of temperate lakes. Even global regression lines, such as (Jones and Bachmann 1976), used only 143 lakes. In comparison this study analysed over 5,000 lakes from 25 countries, giving a much broader representation of the world’s lakes. Separating the data into temperate and tropical lakes sheds some light on the difference between this analysis and prior studies. Temperate lakes generate a regression equation with a notably steeper slope,

LogChl=0.9667LogTP-0.4976, than tropical lakes, LogChl=0.8463LogTP-0.451, which suggests that tropical lakes have lower TP:Chl ratios than their temperate counterparts with the same level of phosphorus producing lower values of chlorophyll. This might be

89

2.5

2.0

1.5

Sakamoto 1966 1.0

Dillon 1974 NES 1974 0.5

Jones 1976

a a µg/l) (log - Carlson 1977 0.0 Phillips 2008 0 0.2 0.4 0.6 0.8 1 1.2 1.4 1.6 1.8 2

Chlorophyll Belcon Global 90

-0.5 Belcon Temperate Belcon Tropical -1.0

-1.5

-2.0 Total Phosphorus (log µg/l)

Figure 39: Total phosphorus/chlorophyll -a regression lines for various studies.

due to a combination of factors. The majority of tropical lakes are riverine lakes (Lewis

1996), and tropical floodplain lakes exhibit higher means but lower TP:Chl ratio than their temperate counterparts. This suggests that although nutrient values are higher in tropical lakes, the efficiency of their use is lower. This might be influenced by higher turbidity in floodplain lakes preventing phytoplankton growth by light attenuation or by the presence of chemically bound, non-available forms of phosphorus (Engle and

Melack 1993). Limited work on tropical floodplain lakes prevents any conclusive statements as to the cause of this latitudinal difference but undoubtedly the answer lies in a synergy of the various factors. The global TN/CHl relationship exhibits slightly more scatter than the TP/Chl relationship, r2=0.567 TN/Chl and r2=0.6227 TP/Chl, but the correlation between TN/Chl is still highly significant, p<.0001. Alhough less studied that phosphorus, nitrogen limitation of algal growth has been documented in freshwater lakes with joint nitrogen and phosphorus enrichment boosting productivity more than either nutrient solely (Fritsch et al. 2007).

Given the aforementioned differences between temperate and tropical lakes, it is somewhat surprising that analysis of global floodplain versus non-floodplain lakes produces such similar regression lines for the relationship between TP and Chl, although their intercepts are statistically different from each one other. The variation in intercept of the TP-Chl relationship in the comparison of floodplain with non-floodplain

91

lakes may result because of the inclusion of non- floodplain temperate lakes with exceedingly low nutrients and algal production. These clear oligotrophic lakes are rare in tropical regions, except for the few large, deep tectonic lakes, which were excluded from this study. However, given the expected increased turbidity and associated nutrient recycling that floodplain lakes experience, as well as the potential for nutrient replenishment from the main river, the question remains as to why floodplain lakes are so similar in their nutrients/productivity ratio and distribution to non-floodplain lakes.

The answer may be that the nutrient loading of floodplain lakes spans a greater range of values than commonly assumed and thus the nutrient:chlorophyll relationship is also variable. For example, in the well-studied floodplain lakes of the lower Amazon (Junk et al. 1989; Mertes 1994), the river floods its banks for several months, inundating its floodplain with nutrient rich water and increasing productivity (Bayley 1995). A floodplain lake that is therefore close to or highly connected to the river is fertilized and productivity increases. In contrast, in the headwaters of the Amazon at the foothills of the Andes Mountains storms are often highly localized and intermittent during the rainy season. The flood wave is multimodal rather than the monomodal wave experienced in the lower Amazon, and floods are flashy in nature, rising and falling in a matter of days

(Davenport, unpublished data). Limited work has been done on the limnology of these upper watershed lakes but preliminary work indicated that lakes that are highly

92

connected to the river have lower nutrient levels and significantly lower productivity than lakes that are isolated from the river (Davenport 2008). This indicates a relatively nutrient poor river where frequent dilution from heavy precipitation along with the short residence time within the watershed prevents the accumulation of nutrients. It is therefore possible for floodplain lakes with the same system and with similar degrees of connectivity exhibit diverse ecological states. Given the complexity of determining floodplain productivity states a hydrogeomorphic classification of floodplain lakes might be useful in a similar way that wetlands are categorized based on surface flow, ground water and precipitation (Brinson 1993). Such a classification would be dependent on lake morphology, how connected the lake is to the river during both the low and high flood periods; lake size, deeper lakes typically are less turbid than shallow lakes; nutrient level of the river as inundation maybe fertilize or dilute the lake depending on the concentration of nutrients in the incoming water as well as the precipitation and flooding regime. Such a classification however would be dependent on future improvement in the study of floodplain lakes.

93

4.5 Conclusion

Global analyses of over 5000 lakes within 25 countries show strong significant relationships between total phosphorus and chlorophyll-a. The global linear regression model exhibits a lower slope than in prior studies, but this was expected given the relatively small sample size and limited geographic distributions of older models.

Comparisons between floodplain and non-floodplain lakes showed no significant differences between regression models, although floodplain lakes consistently have higher mean nutrient levels than their non-floodplain counterparts. Temperate floodplain lakes do have higher phosphorus:chlorophyll ratios than tropical lakes, a difference that is likely due to the higher turbidity and suspended solids typical of tropical floodplain lakes. Our analysis suggests that more detailed characterization and classification is necessary for floodplain lakes, because morphology, connectivity and hydrology likely play large roles in the ecological status of any given system. The ability to construct such classification is currently hampered by the lack of widespread basic limnological data for floodplain lakes, such as lake depth, turbidity, TSS and connectivity. Future work in floodplain limnology, particularly in the tropics should ensure that such values are reported.

94

5. Conclusions and Future Research

This dissertation examined tropical limnology in southwestern Amazonia from three distinct perspectives with the aim to increase our understanding of the dynamics that drive tropical fluvial environments and how they differ from traditionally studied temperate systems. The first study built the foundation for the dissertation with the construction of a geomorphological map of the Madre de Dios region which delineates the landscape units and documents the impact of fluvial processes on this geologically and biologically notable region. The definition of the geomorphic units of the Madre de

Dios highlights the role of the Fitzcarrald Arch as a drainage center for the headwaters of the Amazon River and the subsequent creation of the highly dissected topography of this uplifted geomorphological feature. The map also presents an original detailing of the boundaries of a lake unit characterised by small, round, non-floodplain lakes.

Further research, particularly field examination, is required to verify the origin, age and morphology of these lakes but remote sensing analysis suggests a correlation between the presence of acrisols – leached soils with clay-enriched subsoil, in a transitional climate that experiences cyclic water-saturation and drying periods and the location of these lakes. We theorize that the combination of soil type and climate seasonality resulted in the formation of impermeable laterite duricrusts that enabled lake formation.

The mapping of the geomorphic units of the Madre de Dios facilitates the future 95

selection of field sites for a variety of researchers and updates our understanding of the development of the Amazon River. It also provides a guide in the consideration and selection of conservation units and management practices.

The second study focused on the ecology of floodplain lakes within the Madre de

Dios watershed, specifically along 300 km of the Manu River. Using Landsat TM images spanning 25 yrs (1986-2010), a timeline of ecological regime shifts for 25 oxbow lakes was created. The analysis, the first of its kind for multiple tropical lakes, showed that unlike lower Amazon lakes whose stable vegetative state is typically floating macrophytes, the lakes of the Manu exist in a phytoplankton dominated state with unsynchronized shifts to floating vegetation that are typically ephemeral, <3 yrs when compared to traditional regime shifts. Though field observations linked flooding events to vegetation shifts at two lakes, analysis of the extended floodplain did not support the hypothesis of flooding initiated regime shits. However, due to the limited cloud free satellite images sourced for this region and limited number of water level loggers used in the field, further research is necessary to assess the proposed theory. Long-term, basin-wide monitoring of lake ecology and hydrological connectivity in this pristine location is needed to gain a comprehensive understanding of the flooding and ecological regimes of these biologically diverse and ecologically vital lakes.

96

The third study extended the analysis of tropical floodplain lakes and compared and contrasted global floodplain lakes to non-floodplain lakes in terms of nutrient content and productivity. Given the hydrological differences between the two lake types, there is the potential for large differences in lake productivity however analysis of over 5000 lakes showed no signification distinctions between floodplain and non- floodplain lakes on a global scale. In contrast, substantial differences were found between temperate and tropical floodplain lakes with tropical lakes exhibiting lower nutrient efficiency than their temperate counterparts. This may be due to a variety of possible dissimilarities such as turbidity regimes, total suspended solids in the water column or differing nutrient sequestration process. However, with no global standards regarding limnological data collection insufficient data are available to investigate the mechanisms behind this difference and additional field work is necessary.

Collectively these three studies represent preliminary work in tropical limnology that aids in the development of our understanding of how these systems work. Tropical climates, particularly the consistent high temperatures and typically high precipitation, affect all aspects of their fluvial environment: the ecology, biogeochemistry and geomorphology of the landscape. It is imperative that these be studied in their own right and not be considered as the hotter, wetter analogies of the more frequently studied temperate systems. With global population growth focused in the tropics with the

97

resulting increased demands placed on tropical environments, it is imperative that our actions and decisions are based on a thorough understanding of these ecosystems.

98

Appendix A - Non-Floodplain Lakes

Citation # of Location Lake ID Area Mean TP (µg/L) TN CHL lakes (km2) Depth Mean (µg/L) (µg/L) (m) Mean Mean (Jeppesen et al. 15 Denmark 3.34 2.2 200 2100 107 2003) (Medina- 1 Spain La Caldera 0.02 4.3 5.79 0.71 Sánchez et al. 2004) (Smith and 15 Quebec, Aylmer 32.4 7 16.5 613.5 Prairie 2004) Canada Brome 14.7 4 13.7 473

Denison 0.3 1.6 47.1 690.7 7.4 Libby 0.4 1 20.8 396.6 8.2 Lovering 4.9 9.8 16.5 436.9 1.8

Magog 11.5 7.5 25.1 476.7 3.6 Montjoie 3.3 8 13.1 377.5 1.6 Parker 0.2 3.6 18.5 623.8

Simoneau 0.5 9.3 12.9 388.9 1.5 St. Georges 0.5 1.8 45.1 957.1 33.1 Tomod 0.8 0.9 58.5 834.4

Truite 2.4 3.3 31.7 448.9 2.9 Waterloo 1.5 3 45.5 698.5 20.3 Webster 0.1 1.3 26.1 525.1 7.8

(Lottig et al. 16 Wiscosin 15 550 2011) (Zhang et al. 38 China Oligotrophic 12 230 1.04 2010) sites (O) Mesotrophic 27 560 5.69 sites (M) Eutrophic 118 1770 37.5 sites (E) (Reche et al. 2 Spain Caldera 0.021 0.14 27.5 0.7 2009) Rio Seco 0.004 0.52 29.8 1.5

(Stets and Cotner 2 Minnesota Christmas 1.12 0.7 2008) Lake Lake Owasso 1.55 1.3 (Bergström et al. 3907 Sweden Region 1 14 728 8.1 2005) (n=171) 99

Citation # of Location Lake ID Area Mean TP (µg/L) TN CHL lakes (km2) Depth Mean (µg/L) (µg/L) (m) Mean Mean Region 2 9 441 4.9 (n=312) Region 3 13 524 6.2 (n=208) Region 4 12 606 4.6 (n=213) Region 5 12 501 4.4 (n=118) Region 6 9 352 3.6 (n=170) Region 7 12 504 3.2 (n=249) Region 8 15 722 7.2 (n=258) Region 9 10 420 2.8 (n=498) Region 10 10 433 3.5 (n=259) Region 11 8 312 3 (n=473) Region 12 7 290 2.3 (n=344) Region 13 7 286 1.1 (n=634) (Trigal et al. 10 Sweden High biomass 2.8 13.7 522 8180 2011) Low biomass 4 11.3 524 3460

(Kosten et al. 83 South Cool Lakes 1.6 910 3000 277 2011) America Intermediate 1.4 280 2800 93 Lakes Warm Lakes 2.3 70 500 10 (Lauridsen et al. 5 Denmark Lake Holm 0.22 0.8 10 400 3.8 2011) Lake Kvie 0.29 1.2 48 690 6.8

Lake Magle 0.15 3.6 13 760 4.1 Lake Nors 3.9 3.6 23 540 11.3 Lake Soby 0.73 2.8 18 390 8.3

(Harrison and 2 Ontario, Blue Chalk 0.52 8.5 5.42 1.49 Smith 2011) Canada Plastic 0.32 7.9 3.05 1.29 (Vanni et al. 3 Ohio Burr Oak Lake 2.78 4.3 9.73 79 2011) 100

Citation # of Location Lake ID Area Mean TP (µg/L) TN CHL lakes (km2) Depth Mean (µg/L) (µg/L) (m) Mean Mean Pleasant Hill 3.17 4.8 42.5 1030.6 Lake Acton Lake 2.4 3.9 47.4 3003.7 (Liboriussen et 16 Denmark 1 27 393 6 al. 2011) (Drakare and 2 Sweden Hemsjon 3.7 20 340 Liess 2010) Stora Hensjon 3 8.9 490 (Quinlan and 80 Ontario, 2.485 11.6 9.1 236.4 Smol 2010) Canada (Lund et al. 2010) 1 Denmark Lake 0.54 5 158.5 820 Nordborg (SØNdergaard et 18 Denmark 1.1 2.6 167 1650 55 al. 2010) (Kruk et al. 2009) 18 Uruguay Aguada 0.011 3 43 975 3.9 Barro 0.135 3 32.8 884 4.9 Blanca 0.287 3 51.9 1017 3.4

Chaparral 0.012 3 47.2 598 8.7 Chica 0.022 3 90.5 1164 46.5 Cisne 1.57 3 413 1048 4.2

Clotilde 0.177 3 27.7 451 4.1 Diario 0.618 3 75.8 825 1.3 Escondida 0.108 3 24.2 489 1.1

Garcia 0.052 3 29.8 332 0 Mansa 0.176 3 184.2 1534 4.5 Moros 0.01 3 28.7 437 14

Nueva 0.005 3 60.9 1160 7 Nutrias 0.338 3 99.8 1136 3.3 Pajarera 0.005 3 179.8 2691 13.8

Ponderosa 0.006 3 86.5 888 9.4 Redonda 0.047 3 23.9 514 1.5 Techera 0.01 3 37.9 1681 13.7

(Bjerring et al. 54 Europe 7.82 1.6 107 1936 47 2009) (Bramm et al. 2 Denmark Lake 0.4 1 275 1780 112 2009) Sobygard Lake 0.21 0.8 96 2650 36

101

Citation # of Location Lake ID Area Mean TP (µg/L) TN CHL lakes (km2) Depth Mean (µg/L) (µg/L) (m) Mean Mean Stigsholm (Sayer et al. 39 UK and Bayf 0.027 0.92 262 5.1 2010) Denmark Bees 0.026 0.61 110 25.5

Bigw 0.043 0.67 137 69.9 Blic 0.101 0.95 83 34.2 Blue 0.036 1.2 42 10.9

Bond 0.019 2.13 19 8.3 Buck 0.004 1.23 92 56.8 Burf 0.047 1.18 157 85

Cosh 0.051 1 85 20.5 Crom 0.023 0.58 389 43.2 Deco 0.094 1.4 184 52.9

Dend 0.045 1.39 55 7.2 Doj 0.02 1.08 110 18.6 Enso 0.106 2.04 76 29.6

Felb 0.027 0.9 139 18.3 Gamm 0.016 1.1 157 78.5 Grep 0.016 1.47 22 2.3

Gub 0.006 0.55 225 20.6 Gunt 0.017 1.13 98 83.7 Have 0.047 0.58 80 56.6

Hoki 0.008 0.82 41 2 Hokm 0.142 1.46 74 19.2 List 0.015 0.78 27 5.2

Lopo 0.005 1.36 30 10.8 Lyli 0.081 1.05 259 5.1 Melt 0.074 1.24 243 57.1

Narf 0.225 1.05 28 4.5 Pedh 0.031 0.78 224 60.7 Saha 0.053 1.56 560 29.3

Salh 0.012 0.74 84 21 Scot 0.02 0.85 51 5.5

102

Citation # of Location Lake ID Area Mean TP (µg/L) TN CHL lakes (km2) Depth Mean (µg/L) (µg/L) (m) Mean Mean Selb 0.016 0.83 34 5.3 Sort 0.046 0.65 4056 44.6 Sten 0.01 0.54 38 2.8

Stra 0.083 1.09 283 17 Stru 0.028 0.84 151 7.6 Tran 0.03 0.67 89 37.6

Upto 0.069 0.9 33 9.7 Wolt 0.041 0.84 63 13.1 (Kruk et al. 2010) 11 South Subpolar 438 2106 20.1 America, Europe, North America 217 Temperate 144 1515 31.9

40 Subtropical 388 4286 76.7 42 Tropical 85.6 474 12.2 (Jansson et al. 15 Sweden High Alpine 1 0.27 8.6 1.9 81 2010) High Alpine 2 0.038 1.9 5.5 109 High Alpine 3 0.17 5.3 1.8 66 High Alpine 4 0.17 4.7 4.5 91

Mid Alpine 5 0.11 4.5 7.3 145 Mid Alpine 6 0.019 2.2 7.7 146 Mid Alpine 7 0.014 1.8 10.5 181

Low Alpine 8 0.043 2.8 6.1 178 Low Alpine 9 0.035 2.8 8.1 150 Low Alpine 10 0.033 1.9 8.7 194

Sub Alpine 11 0.052 1.7 3.8 224 Sub Alpine 12 0.088 5.1 4.3 148 Sub Alpine 13 0.1 0.6 9.5 414

Conif. Forest 0.025 1.5 9.4 346 14 Conif. Forest 0.11 2.3 5.5 389 15 (ÖZkan et al. 1 Turkey Lake Pedina 0.07 1.2 40 450 22 2010)

103

Citation # of Location Lake ID Area Mean TP (µg/L) TN CHL lakes (km2) Depth Mean (µg/L) (µg/L) (m) Mean Mean (Teixeira-de 10 Uraguay Temperate 0.29 1.8 81 1140 41 Mello et al. 2009) and Denmark Subtropical 0.2 1.6 220 570 30 (Trochine et al. 2 Argentina Fantasma (FA) 0.01 2 51.95 682.1 13.04 2009) Nirihuaau 0.006 1.5 23.57 339.74 1.96 Grande (NGr) (Leira et al. 2009) 45 Ireland Low 0.14 2.75 30 6.4 Alkalinity Shallow Low 0.82 7.3 7 1.8 Alkalinity Deep Moderate 3.88 2.25 64 26.4 Alkalinity Shallow Moderate 0.42 6.25 31 11.4 Alkalinity Deep High 0.96 2.7 12 2.6 Alkalinity Shallow High 0.51 7.05 20 4.1 Alkalinity Deep (Trigal et al. 28 Northern 0.038 0.89 273 80 35 2007) Iberian Plateau, Spain (Vanormelingen 28 De Maten 0.017 0.54 700 94 et al. 2008) reserve, Genk, Belgium (GÉLinas and 7 Laurentian Violon 0.38 8.9 4 0.26 Pinel-Alloul region, 2008) Canada Connelly 1.24 7.7 5.9 1.04 Morency 0.26 8.7 7.5 2.38

Purvis 0.19 7.8 8.3 0.89 Truite 0.51 9.3 6.9 1.62 Tracy 0.08 8.1 4.2 0.09

Rond 0.17 7.2 8 1.12

104

Citation # of Location Lake ID Area Mean TP (µg/L) TN CHL lakes (km2) Depth Mean (µg/L) (µg/L) (m) Mean Mean (Buchaca and 82 Pyrenean 0.057 3.27 161.53 1.18 Catalan 2007) mountain range (Hanson et al. 168 Northern 0.054 4 10 474 2007) Highland Lake District, Wiscosin (Rautio and 10 Northern S6 3E-04 0.3 4 193 Vincent 2006) Canada S7 2E-05 0.3 12 602 S8 1E-05 0.3 9 774

S9 2E-05 0.4 33 612 S10 1E-04 0.5 13 568 A1 0.04 1.5 5 226

A2 0.262 9 5 146 A3 5E-06 0.2 21 412 A4 0.06 0.3 5 172

A5 0.02 1 4 194 (Liboriussen and 13 Denmark 10.4 60 1330 22.5 Jeppesen 2006) (Irfanullah and 1 Cheshire, Delamere 0.018 1.7 200 2200 290 Moss 2005) England Lake (PÅLsson et al. 5 Faroe Eystara 0.03 3 16 1 2005) Islands Mjaavatn Saksunarvatn 0.08 6.5 6 1.3

Leynavatn 0.18 13.7 3 1.4 Sorvagsvatn 3.56 27.5 5 0.8 Toftavatn 0.52 5.8 11 1.6

(Stephen et al. 6 Finland, Vesijarvi 1 50 4 2004) Sweden, England, The Netherlands, Spain Little Mere 1 184 3 Naardermeer 1 30 5 Sentiz 1 100 0

Xeresa 1 17 3

105

Citation # of Location Lake ID Area Mean TP (µg/L) TN CHL lakes (km2) Depth Mean (µg/L) (µg/L) (m) Mean Mean (Fernández- 1 North-west Lake Sentiz 0.047 0.8 34.27 9.1 Aláez et al. 2004) Spain (Drakare et al. 7 Lappland, 2.4 20.6 428 1.5 2003) Sweden (Castro et al. 1 Portugal Lake Vela 0.7 0.9 140 59.5 2007) (Gutseit et al. 10 Southern Low DOC 9.4 5.3 2007) Sweden High DOC 24 27

(Hessen and Leu 6 Svalbard Solvatn 0.005 76 1.57 2006) Gasedammen 0.001 32 0.99 Brandallaguna 0.127 11 2.63

Ovrevatn 0.007 11 2.41 0.08 5 1.71 Akselpytten 0.003 4 0.9

(Amsinck et al. 29 Faroe 0.06 26 250 1.2 2006) Islands (Brunberg et al. 2 Coastal Lake 0.024 0.9 13 1215 2002) Sweden Hallefjard Lake 0.23 1.5 16 1480 Eckarfjarden (Brodersen and 21 Southern Lake-01 0.353 16.33 747 Anderson 2002) West Greenland Lake-02 0.368 10.5 653

Lake-06 0.215 12 956 Lake-07 0.038 34 1218 Lake-09 0.034 8 613

Lake-11 0.086 17 625 Lake-16 0.033 11.67 450 Lake-19 0.17 8.33 450

Lake-22 0.28 4.33 262 Lake-41 0.5 4 360 Lake-42 0.067 7.5 750

Lake-43 0.095 7 683 Lake-45 0.032 5.33 277 Lake-46 0.074 4 180

106

Citation # of Location Lake ID Area Mean TP (µg/L) TN CHL lakes (km2) Depth Mean (µg/L) (µg/L) (m) Mean Mean Lake-47 15 3 110

Lake-48 0.026 4.25 143 Lake-49 0.207 3.67 143 Lake-50 0.06 6.5 210

Lake-52 0.013 6 90 Lake-53 0.073 7.5 175 Lake-71 0.135 19 1860

(Kalff and 2 Kenya Lake 147 7 44 27 Watson 1986) Naivasha Lake Oloidien 5.5 7 44 23 (de Souza 1 Brazil Itapeva Lake 234.1 1825 490 20.25 Cardoso and da Motta Marques 2009) (Ostrofsky and 49 Yellowknife, lake 3 0.13 25.95 3.4 Rigler 1987) Canada Lake 6 0.15 22.25 2.3 Lake 5 0.09 28.15 1.8 Fiddlers L.S. 0.46 38.15 13.1

Fiddlers L.N. 0.33 14.85 3.2 Fox 0.58 30.2 5.1 Long 1.15 15.5 2.6

Stock 0.51 28.6 7.2 frame 0.93 21.4 6.9 Grace 0.63 18.3 5.7

lake 1 0.22 26.25 3.3 Daigle 0.13 15.7 4.1 Finger 0.04 11.9 2.2

Shadow 0.09 21.1 4.2 X 0.07 20.55 2.1 Rater 0.23 40.5 4.1

Walsh 8.85 12.75 1.6 Vee 0.71 27.85 5.6 David 0.14 27.35 5.9

Gold 0.04 34.25 8.9 107

Citation # of Location Lake ID Area Mean TP (µg/L) TN CHL lakes (km2) Depth Mean (µg/L) (µg/L) (m) Mean Mean Gar 0.34 22.65 5.2

Trapper 0.28 29.15 1.8 Tibbit L.N. 1.49 13.65 2 Tibbit L.S. 2.02 9.35 1.9

lake 50 0.1 39.8 11.3 Peninsula 0.86 7.65 2.3 lake 49 0.24 17.05 4.1

Lake 48 0.09 17.45 3.2 Reid 5.63 7.7 1.5 Pickerel 1.61 20.15 8.1

lake 47 0.14 25.4 5.5 lake 46 0.16 21.95 7.3 lake 45 0.65 22.1 6.4

lake 44 0.06 25.45 4.4 lake 43 0.08 9.4 3.2 Lake 42 0.06 34.85 8.6

Lake 41 0.19 36.55 11.5 lake 40 0.08 20.85 3.4 Lake 39 0.36 18.6 3.2

lake 38 0.09 57.2 16.9 lake 36 0.08 14.75 2.2 lake 9 0.08 41.45 12.7

lake 8 0.85 48 26.8 Pontoon 3.38 12.25 2.6 lake 10 N. 0.16 21.85 5.1

Lake 10 S. 0.07 18.5 4.1 Madeline 0.98 18.1 3.5 Tom 0.15 20.45 2.7

lake 7 0.3 27.75 7.2 (Masson et al. 10 Southern Beaulac 0.87 9.3 8.14 0.62 2000) Quebec, Canada Bonny 0.07 10.7 10.85 1.48

108

Citation # of Location Lake ID Area Mean TP (µg/L) TN CHL lakes (km2) Depth Mean (µg/L) (µg/L) (m) Mean Mean Brassard 0.04 6.1 22.74 6.11 Cromwell 0.11 10 12.34 2.02 Desjardins 0.06 4.6 16.9 2.16

echo 1.76 9.1 15.52 1.59 Pontbriand 1.96 9.3 12.72 0.79 Triton 0.02 3.2 8.14 1.04

Desmarais 0.24 7 24.31 2.19 Waterloo 1.47 4.9 33.91 7.55 (Dillon and 4 Southern Limestone 10.4 3.75 Rigler 1974b) Ontario, Regn Canada W Highlands 13 5.8 Lowlands 31 18.4 E Highlands 10.8 3.5

(Smith 1982) 52 NC and 1.226 10.943 Minnesota 8.705 2.934 14.903 7.353 6.188 13.379

14.702 7.353 16.363 11.017 13.726 12.089

12.827 13.098 18.468 13.492 17.492 14.819

24.471 30.722 9.895 48.197 21.358 47.839

44.867 50.098 40.801 62.083 82.804 5.617

62.366 5.605 54.5 5.236 36.363 4.624 109

Citation # of Location Lake ID Area Mean TP (µg/L) TN CHL lakes (km2) Depth Mean (µg/L) (µg/L) (m) Mean Mean 36.866 4.209

26.301 4.612 24.588 4.369 12.335 7.749

43.823 12.198 44.324 29.246 44.305 34.837

72.04 32.257 241.513 265.529 109.13 116.158

68.148 60.677 54.892 70.236 53.44 64.776

49.964 58.137 44.851 58.088 54.198 50.851

49.317 48.804 40.267 54.264 35.182 54.206

34.234 61.17 35.182 54.206 14.954 28.807

24.819 23.164 20.835 19.161 20.023 14.248

27.132 13.534 25.548 10.69 20.88 8.046

24.908 5.492 21.762 5.56 12.3 1.563

17.444 2.93

110

Citation # of Location Lake ID Area Mean TP (µg/L) TN CHL lakes (km2) Depth Mean (µg/L) (µg/L) (m) Mean Mean 13.488 3.905

11.94 4.615 9.687 4.425 8.874 4.305

9.361 5.599 7.907 5.554 (Smith and 14 Minnesota Shagawa 40.6625 20.0375 Shapiro 1981) Oregon Cline's Pond 104 78.38 Ohio Twin 36.775 12.875 Seattle Green 38.25 18.95

Ontario Gravenhurst 35.142857 7.7857143 Bay Ontario Little Otter 49.65 19.25 Scotland Loch Leven 85.75 82.25

Sweden Ekoln 55.25 14.575 Sweden Boren 26.75 7.225 Sweden Norrviken 167.9 75.6

Sweden Edssjon 291.66667 100.66667 Sweden Oxundasjon 141 42.166667 Sweden Ramsjon 417.33333 117.66667

Sweden Ryssbysjon 310 31.5 (Schindler 1977) 2 Canada 7.4971 6.4438 9.6025 5.5716

11.1839 3.8275 8.9163 10.0291 14.0267 8.9632

17.7951 3.4399 20.7941 5.9593 19.1245 13.5173

13.73 7.3159 16.5114 6.3469 19.8898 7.219

111

Citation # of Location Lake ID Area Mean TP (µg/L) TN CHL lakes (km2) Depth Mean (µg/L) (µg/L) (m) Mean Mean 18.0053 10.126 16.1999 11.2888 14.3973 11.1919

28.0597 15.1647 27.5249 19.1376 28.1966 21.0756

32.0272 21.2694 37.2123 20.3973 29.6044 29.6996

33.9514 34.0601 37.0299 34.6415 38.8856 44.5252

21.6852 10.5135 (Bergmann and 30 Canada 55.7 31.2 Peters 1980) 48.6 34

43.2 11.8 25.6 17.6 39.8 10

29 1.95 22.1 3.23 20.3 3.1

23.9 5.15 20.8 4.6 18.8 4.51

18 5.02 16 4.25 13.1 2.59

11.9 2.33 11.1 1.8 20.679 7.835

1.1 0.416 2.9 0.672

112

Citation # of Location Lake ID Area Mean TP (µg/L) TN CHL lakes (km2) Depth Mean (µg/L) (µg/L) (m) Mean Mean 3.4 1.055 4.2 1.44 5.2 1.44

4.2 1.95 4.7 2.079 7.56 1.694

7.822 2.079 10.908 3.48 12.698 3.742

13.988 4.51 11.441 5.021 (Jun et al. 2009) 1 Yunnan- lake Dianchi 4.4 175.25 2877.5 84 Guizhou plateau, China (Diaz et al. 2007) 39 Andean Alicura 110 7.2 56.3 0.7 Patagonia Argentino 500 6.5 - 0 Arroyito 15 13.6 90.8 1.9 Bayley Willis 10 46.2 223.6 1.9

Buenos Aires 550 2.9 - 0 Carrilaufquen 5 190.4 588.5 10.8 Chica Carrilaufquen 7 195 1520.9 8.6 grande Caviahue 95 321.6 80 0.8

Ceferino 9 18.6 99.6 1.3 Correntoso 50 4.3 98.9 0.1 Escondido 10 5 - 0

Espejo 245 8 96 0.3 Espejo Chico 68 8 - 0.6 Ezquerra 4 9.6 - 5.1

Fonk 85 12.9 - 1.4 Futalaufquen 168 6.2 - 0.5 Guillelmo 100 8.7 - 0.6

113

Citation # of Location Lake ID Area Mean TP (µg/L) TN CHL lakes (km2) Depth Mean (µg/L) (µg/L) (m) Mean Mean Gutierrez 111 3.4 - 0.4

Hess 80 6.1 48.4 0.3 Ingenieros 3 18 - 0.2 Lacar 277 4 97.5 1.5

Lolog 200 3.8 36 0.1 Los barreales 120 15.2 85.2 1 Mari Menuco 140 10.8 71.9 0.6

Mascardi 218 9.9 1.6 0.2 Moreno 180 8.8 50.2 0.4 Nahuel Huapi 464 9 71 1.4

Ne-Luan 6 199.5 881.5 10.1 Pellegrini 18 24 - 14.1 Piedra del 120 8.2 73.6 1.2 Aguilla Ramos Mexia 60 12.6 84.6 2 Roca 80 8 - 0.8 Schmoll 5 5.8 - 0.1

Tonchek 12 7 - 0.6 Traful 200 8 48.7 0.4 Trebol 12 6.2 151 0.9

Verde (Chall 6 15.7 40 9.7 Huaco) Verde VLA 9 10 143 2.3 Vintter 300 1.8 - 0.3 (Jones and 24 Iowa Blackhawk 3.66 305 63.7 Bachmann 1978) Center 1.14 98 80.8 Clear 14.43 40 18.4 Cornelia 0.85 65 59.3

Crystal 1.08 302 146.1 East Okoboji 7.64 178 169 East Twin 0.72 372 347

Five Islands 4.12 91 27.8 High 2.18 319 11.2

114

Citation # of Location Lake ID Area Mean TP (µg/L) TN CHL lakes (km2) Depth Mean (µg/L) (µg/L) (m) Mean Mean Ingham 1.44 184 15.5 Iowa 3.02 210 90 Little Wall 1.01 107 71.3

Lost Island 4.65 146 123.8 Lower Gar 0.98 131 175 Minnewashta 0.47 132 53.3

North Twin 1.77 81 199 Silver 4.32 191 50.6 Spirit 21.68 42 23.6

Storm 12.46 81 86.4 Swan 4.2 94 52.2 Trumbell 3.29 116 97.2

Tuttle 9.06 188 126.8 Twelve Mile 0.83 444 93.7 Upper Gar 0.14 123 161.1

(Elser et al. 2009) 34 , Bolvatnet 10.221341 164.0185 3.39 Colorado, Sweden Djupen 7.123965 123.6792 5.77 Flaksjoen 6.5044898 146.5101 4.85

Hornsjoen 14.557668 189.0905 8.73 Kroksjoen 16.725831 187.4096 16.82 Langrumpa 10.221341 169.4811 4.42

Ljosvatnet 24.159534 181.947 76.3 mellsjoen 11.770029 142.168 6.93 Musvoltjonna 7.123965 141.3276 6.29

Muvatnet 6.1947522 119.8974 4.02 Ner-Asta 8.9823907 141.4677 3.61 Reinsvatnet 9.2921283 116.1155 3.71

Valsjoen 16.416093 279.4337 16.92 Revsvatnet 6.8142274 581.8383 9.65 A6 6.1947522 63.45035 1.02

A7 8.9823907 92.44422 1.13

115

Citation # of Location Lake ID Area Mean TP (µg/L) TN CHL lakes (km2) Depth Mean (µg/L) (µg/L) (m) Mean Mean V3 7.7434403 103.0893 1.94

V8 8.0531779 165.8393 3.21 B3 9.9116035 195.8137 2.96 B5 5.8850146 169.2009 1.89

D13 7.123965 180.8265 2.83 Fuller 3.4071137 49.30358 2.94 Horseshoe 3.4071137 308.5676 1.42

island 4.6460642 202.3968 1.13 Little Molas 4.3363265 58.54801 4.44 Twin (large) 6.5044898 111.3533 0.25

Twin (small) 6.5044898 68.07256 0.39 Yule (large) 2.4779009 112.1797 2.34 Yule (small) 1.5486881 31.37501 0.13

Haiyaha 4.9558018 329.2975 24.13 Lake of Glass 4.3363265 181.6669 10.82 Sky 4.0265889 251.4203 11.7

The Loch 6.8142274 262.0654 21.8 Emerald 4.6460642 256.6027 4.8

116

Appendix B - Floodplain Lakes

Citation # of Location Lake ID Area Mean TP TN CHL lakes (km2) Depth (µg/L) (µg/L) (µg/L) (m) Mean Mean Mean (Wu et al. 30 China RL South 917.1 2.5 20 650 2006) Dongting West Dongting 520.5 4.8 100 740 East Dongting 1478.2 6.1 50 790 Poyang 3583.7 2.5 20 1210

SL Honghu 348.4 1.9 80 1690 Laojianghe 18.4 4.9 80 1070 Tianezhou 14.5 7.6 20 700

Shijiu 210.4 2.3 70 690 Yangcheng 119.1 1.6 90 630 Dianshan 63.7 2.2 250 1520

Longgan 316.2 2.1 30 520 Junshan 192.5 3.7 50 620 NC Hongxing 1 2.7 250 3390

Sanliqi 2.7 2 180 2570 Nanhu 8 2.4 330 3830 Houguan 10.1 2.8 60 990

Longyang 1.4 1.5 960 8670 Moshui 2.7 2.1 1450 5260 Sanjiao 2.5 1.5 330 8230

NE Huama 4.7 2.2 60 1220 Qiaodun 8 2.6 40 590 Bao'an 3.6 2.6 90 920

Qihu 1 1.3 80 1310 Taojiada 1.5 1.7 70 790 Zhangdu 41.1 1.7 50 1000

Niushan 38.2 3.9 40 660 Qingling 2.1 1.3 230 2510 Wuchang 100.5 2.1 60 480

117

Citation # of Location Lake ID Area Mean TP TN CHL lakes (km2) Depth (µg/L) (µg/L) (µg/L) (m) Mean Mean Mean Daoshui n.d. 2.9 70 850

Gehu 146.5 1.3 250 1220 (Yang et al. 49 China Yuandang 12.9 1.7 150 2441 24.08 2008) Cheng 45 2.2 292 2837 57.62 Jinji 6.5 2.3 230 3532 77.02

Dushu 9.6 2.6 290 3540 72.05 Meiliang 129.3 2.6 172 2654 17.45 Taihu 2425 2.8 108 1771 7.89

Changdang 85 1.6 116 2275 25.77 Tianmo 12 11 73 885 11.32 Nanyi 148.4 2.4 86 2339 9.83

Gucheng 24.5 3 48 617 4.29 Xuanwu 3.7 1.3 189.8 2097.3 77.02 Mochou 0.33 1.5 515.4 3236.4 71.78

Chaohu 769.6 3.5 192.5 2035 15.67 Shitang 23.3 2.1 71.1 1003 10.4 Caizi 172.1 4.7 93.5 671.1 13.81

Matang 3.4 3.5 47.5 416.6 4.37 Huang 118.6 3 49.3 736.6 2.26 Daguan 180.6 3 56.3 741.4 4.05

Bohu 180.4 4.3 66.8 598 3.45 Hualiangting 72 34 36 317.4 7.52 Gujiao 2.4 8.1 70 520 11.07

Gantang 0.8 2.7 548.9 2864.9 24.64 bali 16.2 3.4 113.6 1209.6 11.66 Kaotian 3.7 2.5 80 470 12.7

Taibai 25.1 2.5 125.5 1429.3 4.72 jingzhu 3.5 8.9 80 620 12.79 Xianrenba 2.7 6.4 70 740 10.67

Wushan 16.1 2.1 207.4 2358.9 35.8 Chidong 26.8 5.3 68 764 16.72 Cehu 11.8 3 38 997 3.19

118

Citation # of Location Lake ID Area Mean TP TN CHL lakes (km2) Depth (µg/L) (µg/L) (µg/L) (m) Mean Mean Mean Cihu 8.15 3.2 141 1872 59.36

daye 68.7 4.5 72 1705 7.22 Baoan 48 2.6 36 348 5.58 Liangzi 304.3 4 38 491 5.53

Donghu 33.7 4.3 172 1865 48.98 Tangsun 36.6 2.9 43 1452 5.95 luhu 40.2 3.2 51 600 9.89

Futou 114.7 2.2 35 532 4.27 Xiliang 72.1 2.9 30 439 2.14 Huanggai 70 5.8 50 1139 14.6

Hong 344.4 2.3 62 1295 31 (Torremorell 1 Argentina Laguna 30.1 1.9 500 1000 257 et al. 2009) Chascomus (Knowlton 12 Missouri Cooley 239 2000 48 and Jones 1997) Dalton 235 1600 37 Sunshine 282 1700 67

Teteseau 190 1200 64 RK 512 61 700 25 Rk 528 64 700 25

RK 330 94 800 29 RK 304 120 1300 22 RK 398 129 1200 31

RK 346 197 2100 21 RK 351 241 2000 20 Rk 386 225 1900 23

(Sampaio 1 Brazil Lake Massacara 21.93 83.82 9.9 and López 2000) (Jones et al. 166 Missouri Austin (Texas) 20 555 6.8 2008) Bismark (St. 21 380 6.2 Francois) Capri (St. 6 285 1.3 Francois) 119

Citation # of Location Lake ID Area Mean TP TN CHL lakes (km2) Depth (µg/L) (µg/L) (µg/L) (m) Mean Mean Mean Carmel (St. 8 320 1.6 Francois) Clearwater 13 225 4.6 (Wayne) Council Bluff 7 225 2 (iron) Crane (Iron) 13 240 4.1 Fellows (Greene) 12 325 4.1 Fourche Creek 9 235 2.1 (Ripley) Indian Hills 32 600 13.4 (Crawford) Killarney (Iron) 56 565 25.3

Little Prairie 26 470 7.6 (Phelps)

Loggers 9 195 2.6 (Shannon) Lower Taum 12 210 3.6 Sauk (Reynolds) Marseilles (St. 9 330 1.6 Francois) McDaniel 32 460 15.1 (Greene) Miller 20 500 6.9 Community (Carter) Monsanto (St. 9 375 2 Francois) Noblet 11 210 2 (Douglas) Peaceful Valley 30 730 16.1 (Gasconade) Pomme de Terre 24 505 12.9 (Hickory) Ripley (Ripley) 24 600 10.9 Roby (Texas) 14 425 3 Shayne 6 255 1.1 (Washington) Sims Valley 24 475 11.5 (Howell) Springfield 59 950 10.5 (Greene)

120

Citation # of Location Lake ID Area Mean TP TN CHL lakes (km2) Depth (µg/L) (µg/L) (µg/L) (m) Mean Mean Mean Stockton (Cedar) 11 385 5.7 Sunnen 12 275 2.9 (Washington) Table Rock 9 365 4.1 (Taney) Taneycomo 21 705 2.5 (Taney) Turner (Dent) 17 485 9.2 Wappapello 36 490 19.7 (Wayne) Ziske (Dent) 25 580 13.8

Ann (Ste. 39 620 15.7 Genevieve) Bella Vista 23 515 8.1 (Cape Girardeau) Binder (Cole) 52 775 22.7 Boutin (Cape 21 560 6.6 Girardeau) D.C. Rogers 29 525 7 (Howard) Fayette 39 765 13.7 (Howard) Fredricktown 59 675 26.2 City (Madison) Girardeau (Cape 51 790 31.5 Girardeau) Glover 58 850 15.3 (Callaway) Goose Creek 12 385 3 (Ste Genevieve) Manito 90 960 12.8 (Moniteau) Northwoods 21 440 3.8 (Gasconade) Perry County 67 890 35.7 (Perry) Pinewoods 33 735 15.3 (Carter) Pinnacle 19 490 4.7 (Montgomery) Rocky Fork 21 540 5.7 (Boone)

121

Citation # of Location Lake ID Area Mean TP TN CHL lakes (km2) Depth (µg/L) (µg/L) (µg/L) (m) Mean Mean Mean Timberline (St. 7 275 1.5 Francois) Tishomingo 20 480 4.2 (Jefferson) TriCity (Boone) 43 795 14.6

Tywappity (Scott) Wanda Lee (Ste. 46 600 15.7 Genevieve) Wauwanoka 12 375 2.5 (Jefferson) Allaman 39 655 11.6 (Clinton) Baring Country 27 915 14.1 Club Belcher Branch 35 565 14.6 (Buchanan) Bethany 29 655 8 (Harrison) Bilby Ranch 51 1005 33.7 (Nodaway) Blind Pony 75 1150 25.1 (Saline) Bowling Green 21 500 6.2 (Pike) Brookfield 22 610 7.4 (Linn) Busch 37 (St. 26 505 8 Charles) Cameron #1 178 1445 37 (DeKalb) Cameron #2 53 785 23.5 (DeKalb) Charity 39 615 14.8 (Atchison) Deer Ridge 38 740 12.2 (Lewis) Edina City 64 1230 18.4 (Knox) Ella Ewing 79 1295 22.2 (Scotland) Elmwood 52 770 17.6 (Sullivan)

122

Citation # of Location Lake ID Area Mean TP TN CHL lakes (km2) Depth (µg/L) (µg/L) (µg/L) (m) Mean Mean Mean Forest (Adair) 22 415 4.5 Fox Valley 18 590 6.6 (Clark) Green City 71 1025 24.2 (Sullivan) Grindstone 141 2195 26.7 (DeKalb) Hamilton City 55 955 13.5 (Caldwell) Harrison 59 1090 41.2 County (harrison) Hazel Creek 27 595 7.3 (Adair) Henry Sever 44 930 14.3 (Knox) Higginsville 83 1055 20.6 (Lafayette) Hunnewell 41 775 16.4 (Shelby) Indian Creek 23 625 11.3 (Livingston) King (DeKalb) 189 1525 18.7 Kraut Run (St. 96 1090 57.7 Charles) LaBelle (Lewis) 60 1350 47.8

lake St. Louis 63 910 18.2 (St. Charles) Lancaster 70 955 32.7 (Schuyler) LaPlata (Macon) 26 785 13.4

Lawson City 33 935 22.3 (Ray) Limpp (Gentry) 105 1430 63.6

Lincoln 16 405 4.3 (Lincoln) Little Dixie 55 740 16.9 (Callaway) Long branch 45 885 11.7 (Macon) Macon (Macon) 51 865 22.2

123

Citation # of Location Lake ID Area Mean TP TN CHL lakes (km2) Depth (µg/L) (µg/L) (µg/L) (m) Mean Mean Mean Maple leaf 38 825 18.7 (Lafayette) Marie (Mercer) 14 445 3.6 Mark Twain 54 1195 14.6 (Ralls) Maysville 162 1275 38.7 (DeKalb) Memphis City 75 1205 37.4 (scotland)

Moberly 49 740 18.4 (Randolph) Monroe City 77 1105 28.1 (Monroe) Monzingo 31 855 20.5 (Nodaway) Nehai Tonkayea 14 400 2.5 (Chariton) New Marceline 83 1030 28.8 (Chariton) New Milan 38 670 11 (Sullivan) Nodaway 44 970 21.4 County (Nodaway) Old marceline 83 1115 31.4 (Linn) Paho (Mercer) 50 905 11.8 Pape (Lafayette) 67 980 21.1 Pony Express 64 965 27.4 (DeKalb) Ray County 152 1920 114.3 (Ray) Savannah City 38 765 17.3 (Andrew) Shelbina 96 1060 30.8 (Shelby) Smithville (Clay) 34 780 15.3

Spring (Adair) 26 520 6.4

Sterling Price 93 1415 53.6 (Chariton) Thomas Hill 46 760 12.8 (Randolph)

124

Citation # of Location Lake ID Area Mean TP TN CHL lakes (km2) Depth (µg/L) (µg/L) (µg/L) (m) Mean Mean Mean Thunderhead 44 915 13.3 (Putnam) Unionville 93 1185 32.7 (Putnam) Vandalia 58 905 22.4 (Audrain) Viking (Daviess) 25 500 8.1 Watkins Mill 39 610 16.4 (Clay) Waukomis 25 545 11.4 (platte) Weatherby Lake 16 385 4.1 (Platte)

Whiteside 19 640 6.1 (Lincoln) Williams (Clay) 68 880 22.3 Willow Brook 75 1095 35.8 (DeKalb) Worth County 62 1240 29.9 (Worth) Amarugia 45 670 10.2 Highlands (Cass) Atkinson (St. 69 985 31 Clair) Blue Springs 35 570 16.1 (Jackson) Butler City 68 885 35 (Bates) Cottontail 149 1050 12.6 (jackson) Gopher 94 715 14.9 (Jackson) Harmony 44 815 19.3 Mission (Bates)

Harrisonville 47 845 15.7 (Cass) Hazel Hill 49 990 33.7 (Johnson) City 40 930 15.5 (Johnson) Jacomo 30 545 16.6 (Jackson) 125

Citation # of Location Lake ID Area Mean TP TN CHL lakes (km2) Depth (µg/L) (µg/L) (µg/L) (m) Mean Mean Mean Lamar City 73 950 39.9 (Barton) Lone Jack 27 640 13.8 (jackson) Longview 32 670 10.6 (Jackson) Lotawana 31 620 15 (jackson) Montrose 152 1125 56.3 (Henry) Nell (Jackson) 78 690 12.9 North Lake 97 995 35.9 (Cass) Odessa 38 855 21.4 (lafayette) Prairie Lee 50 900 20.9 (Jackson) Raintree (Cass) 50 835 10.6 Spring Fork 139 1070 42.6 (Pettis) Tapawingo 33 740 21.8 (Jackson) Truman 37 745 14.9 (Benton) Westmoreland 21 600 5.8 (Pettis) Winnebago 44 810 17.5 (Cass) Contrary 392 3290 223.4 (Buchanan) Cooley (Clay) 260 1755 51.4

Creve Couer 9St. 148 1020 50.3 Louis) Dalton Cutoff 395 2410 97.3 (Chariton) NC-8 (Howard) 88 720 26.5

NC-16 (Ray) 61 630 25.5 NC-17 (Ray) 64 700 24.8 Sugar 332 2430 161.3 (Buchanan) Sunshine (Ray) 348 1990 99.9 Teteseau (Saline) 225 1330 78.9 126

Citation # of Location Lake ID Area Mean TP TN CHL lakes (km2) Depth (µg/L) (µg/L) (µg/L) (m) Mean Mean Mean Wakonda 81 1060 41.6 (Lewis) (Rai and Hill 8 Brazil White (8 Lakes) 246.5 28.04 1980) (Heglund 129 Yukon Flats Shallow Lakes 262 1117 11.9 and Jones National (129) 2003) park (Ibañez 1997) 1 Marchantaria Lake Camaleao 3.5km2 94.07 357.82 42.88 Island, Brazil

(Roozen et 93 River Rhine, 180 1110 56.3 al. 2003) Germany (Jones and 1 Iowa Oxbow Lake 30 24.4 Bachmann 1978) (Penczak et 2 Warta River, Lake A 132.5 23.625 al. 2004) Poland Lake C 107 18.71 (Evans et al. 12 Lower U2 6.65 14 225 5.5 2009) Mackenzie River basin, Canada 68⁰ 59' - 69⁰ 24' U1 6.65 2.5 190 2.9 L3 0.31 36.7 100 3.1 L9 3.64 40 566.7 3.8

Yaya 10.34 2.3 209 1.4 Trench 7.62 9 185 3.5 Old Trout 7.4 2.5 160 1.8

L10 2.39 20 1080 2.6 L10 1.68 40 100 2.9 Kimialuk 24.18 41.5 186.5 6.4

Denis Lagoon 0.18 9 160 3.9 Big 12.13 6.5 207.8 3 (Winemiller 10 Texas 30⁰ 37', Big Bend 0.205 154.865 385 21.5 et al. 2000) 96⁰ 37' - 95⁰ 50', 29⁰45'

Garrett Lake 0.196 250.881 5887 640

127

Citation # of Location Lake ID Area Mean TP TN CHL lakes (km2) Depth (µg/L) (µg/L) (µg/L) (m) Mean Mean Mean Korthaus 0.045 46.4595 15.5 Bottom Mexican Bend 0.056 315.925 686 73

Moelhman's 0.28 263.271 924 70 Slough PAC II Lake 0.173 353.092 616 99

Perry Lake 0.353 219.908 371 140 Rosehedge Lake 0.056 176.546 1428 26.7

Siegert's Oxbow 0.142 198.227 427 44

Stone Lake 0.061 337.606 434 88 Unpublished 7 Manu Cocha Cashu 10 1.01 National Park, Peru Cocha Gallareta 9 6.90

Cocha Maizal 15 18.53

Cocha Nueva 96 50.96

Cocha Otoronga 55 3.20

Cocha Salvador 27 6.01

Cocha Totora 10 3.01

(Sokal 2007) 1 Slave River SD30 22.05 325.01 1.2 Delta, Canada

128

References

Agostinho AA, Thomaz SM, and Gomes LC. 2004. Threats for biodiversity in the floodplain of the Upper Parana river: effects of hydrological regulation by dams. Ecohydrology & Hydrobiology 4(3):255-256.

Almeida O, Lorenzen K, and McGrath D. 2003. Commercial fishing in the Brazilian Amazon: regional differentiation in fleet characteristics and efficiency. . Fisheries Management and Ecology 10:109-115.

Amsinck SL, Strzelczak A, Bjerring R, Landkildehus F, Lauridsen TL, Christoffersen K, and Jeppesen E. 2006. Lake depth rather than fish planktivory determines cladoceran community structure in Faroese lakes – evidence from contemporary data and sediments. Freshwater Biology 51(11):2124-2142.

Bayley PB. 1995. Understanding Large River: Floodplain Ecosystems. BioScience 45(3):153-158.

Beisner BE, Haydon DT, and Cuddington K. 2003. Alternative Stable States in Ecology. Frontiers in Ecology and the Environment 1(7):376-382.

Bergmann M, and Peters RH. 1980. A Simple Reflectance Method for the Measurement of Particulate Pigment in Lake Water and its Application to Phosphorus– Chlorophyll–Seston Relationships. Canadian Journal of Fisheries and Aquatic Sciences 37(1):111-114.

Bergström A-K, Blomqvist P, and Jansson M. 2005. Effects of Atmospheric Nitrogen Deposition on Nutrient Limitation and Phytoplankton Biomass in Unproductive Swedish Lakes. Limnology and Oceanography 50(3):987-994.

Bjerring R, Becares E, Declerck S, Gross EM, Hansson L-A, Kairesalo T, NykÄNen M, Halkiewicz A, KornijÓW R, Conde-Porcuna JM et al. . 2009. Subfossil Cladocera in relation to contemporary environmental variables in 54 Pan-European lakes. Freshwater Biology 54(11):2401-2417.

Bramm ME, Lassen MK, Liboriussen L, Richardson K, Ventura M, and Jeppesen E. 2009. The role of light for fish–zooplankton–phytoplankton interactions during winter

129

in shallow lakes – a climate change perspective. Freshwater Biology 54(5):1093- 1109.

Brinson MM. 1993. A Hydrogeomorphic Classification for Wetlands. Wetlands research Program Technical Report US Army Corps of Engineers.

Brodersen KP, and Anderson NJ. 2002. Distribution of chironomids (Diptera) in low arctic West Greenland lakes: trophic conditions, temperature and environmental reconstruction. Freshwater Biology 47(6):1137-1157.

Brooks TM, Mittermeier RA, Fonseca GABd, Gerlach J, Hoffmann M, Lamoreux JF, Mittermeier CG, Pilgrim JD, and Rodrigues ASL. 2006. Global Biodiversity Conservation Priorities. Science 313(5783):58-61.

Brunberg AK, Nilsson E, and Blomqvist P. 2002. Characteristics of oligotrophic hardwater lakes in a postglacial land-rise area in mid-Sweden. Freshwater Biology 47(8):1451-1462.

Buchaca T, and Catalan J. 2007. Factors influencing the variability of pigments in the surface sediments of mountain lakes. Freshwater Biology 52(7):1365-1379.

Bush MB, Silman MR, de Toledo MB, Listopad C, Gosling WD, Williams C, de Oliveira PE, and Krisel C. 2007. Holocene fire and occupation in Amazonia: records from two lake districts. Phil Trans R Soc B 362:209-218.

Camargo AFM, and Esteves FA. 1995. Influence of water level variation on fertilization of an oxbow lake of Rio Mogi-Guaçu, state of São Paulo, Brazil. Hydrobiologia 299(3):185-193.

Campbell Jr KE, Frailey CD, and Romero-Pittman L. 2006. The Pan-Amazonian Ucayali Peneplain, late Neogene sedimentation in Amazonia, and the birth of the modern Amazon River system. Palaeogeography, Palaeoclimatology, Palaeoecology 239(1–2):166-219.

Carpenter S. 2003. Regime Shifts in Lake ecosystems: Pattern and Variation. Oldendorf/Luhe, Germany.

Carpenter SR, and Brock WA. 2006. Rising variance: a leading indicator of ecological transition. Ecology Letters 9(3):311-318.

130

Carter SK, and Rosas FCW. 1997. Biology and conservation of the Giant Otter Pteronura brasiliensis. Mammal Review 27(1):1-26.

Castro BB, Marques SM, and GonÇAlves F. 2007. Habitat selection and diel distribution of the crustacean zooplankton from a shallow Mediterranean lake during the turbid and clear water phases. Freshwater Biology 52(3):421-433.

Chander G, Markham BL, and Helder DL. 2009. Summary of current radiometric calibration coefficients for Landsat MSS, TM, ETM+, and EO-1 ALI sensors. Remote Sensing of Environment 113(5):893-903.

Dade WB, and Verdeyen ME. 2007. Tectonic and climatic controls of alluvial-fan size and source-catchment relief. Journal of the Geological Society 164(00167649):353- 358.

Daly D, and Mitchell J. 2000. Lowland Vegetation of Tropical South America-An Overview. In: Lentz D, editor. Imperfect Balance: Landscape Transformations in the pre-Columbian Americas. New York: Columbia University Press. p 391-454.

Daskalov GM, Grishin AN, Rodionov S, and Mihneva V. 2007. Trophic cascades triggered by overfishing reveal possible mechanisms of ecosystem regime shifts. Proceedings of the National Academy of Sciences 104(25):10518-10523.

Davenport LC. 2008. Behavior and ecology of the Giant Otter (Pteronura brasiliensis) in oxbow lakes of the Manu Biosphere Reserve, Peru. Chapel Hill: The University of North Carolina 244 p. de Castro F, and McGrath DG. 2003. Moving toward sustainability in the local management of floodplain lake fisheries in the Brazilian Amazon. Human Organization 62(2):123-133. de Souza Cardoso L, and da Motta Marques D. 2009. Hydrodynamics-driven plankton community in a shallow lake. Aquatic Ecology 43(1):73-84.

Dent CL, Cumming GS, and Carpenter SR. 2002. Multiple states in river and lake ecosystems. Philosophical Transactions of the Royal Society of London Series B: Biological Sciences 357(1421):635-645.

131

Diaz M, Pedrozo F, Reynolds C, and Temporetti P. 2007. Chemical composition and the nitrogen-regulated trophic state of Patagonian lakes. Limnologica - Ecology and Management of Inland Waters 37(1):17-27.

Dijkshoorn JA, Huting JRM, and Tempel P. 2005. Update of the 1:5 million Soil and Terrain Database for Latin America and the Caribbean (SOTERLAC; version 2.0). ISRIC - World Soil Information.

Dillon PJ, and Rigler FH. 1974a. The Phosphorus-Chlorophyll Relationship in Lakes. Limnology and Oceanography 19(5):767-773.

Dillon PJ, and Rigler FH. 1974b. A Test of a Simple Nutrient Budget Model Predicting the Phosphorus Concentration in Lake Water. Journal of the Fisheries Research Board of Canada 31(11):1771-1778.

Drakare S, Blomqvist P, BergstrÖm A-K, and Jansson M. 2003. Relationships between picophytoplankton and environmental variables in lakes along a gradient of water colour and nutrient content. Freshwater Biology 48(4):729-740.

Drakare S, and Liess A. 2010. Local factors control the community composition of cyanobacteria in lakes while heterotrophic bacteria follow a neutral model. Freshwater Biology 55(12):2447-2457.

Elser JJ, Andersen T, Baron JS, Bergström A-K, Jansson M, Kyle M, Nydick KR, Steger L, and Hessen DO. 2009. Shifts in Lake N:P Stoichiometry and Nutrient Limitation Driven by Atmospheric Nitrogen Deposition. Science 326(5954):835-837.

Engle DL, and Melack JM. 1993. Consequences of Riverine Flooding for Seston and the Periphyton of Floating Meadows in an Amazon Floodplain Lake. Limnology and Oceanography 38(7):1500-1520.

Espurt N, Baby P, Brusset S, Roddaz M, Hermoza W, Regard V, Antoine PO, Salas- Gismondi R, and Bolaños R. 2007. How does the Nazca Ridge subduction influence the modern Amazonian foreland basin? Geology 35(6):515-518.

Evans MS, Keating JJ, Ogbebo FE, Tumber VP, and Waiser MJ. 2009. Nutrient limitation of phytoplankton growth in Arctic lakes of the lower Mackenzie River Basin, northern Canada. Canadian Journal of Fisheries and Aquatic Sciences 66(2):247+.

132

Fernández-Aláez M, Fernández-Aláez C, Bécares E, Valentín M, Goma J, and Castrillo P. 2004. A 2-year experimental study on nutrient and predator influences on food web constituents in a shallow lake of north-west Spain. Freshwater Biology 49(12):1574-1592.

Filizola N, and Guyot JL. 2009. Suspended sediment yields in the Amazon basin: an assessment using the Brazilian national data set. Hydrological Processes 23(22):3207-3215.

Fritsch E, Herbillon AJ, Do Nascimento NR, Grimaldi M, and Melfi AJ. 2007. From Plinthic Acrisols to Plinthosols and Gleysols: iron and groundwater dynamics in the tertiary sediments of the upper Amazon basin. European Journal of Soil Science 58(5):989-1006.

GÉLinas M, and Pinel-Alloul B. 2008. Summer depth selection in crustacean zooplankton in nutrient-poor boreal lakes is affected by recent residential development. Freshwater Biology 53(12):2438-2454.

Gopal B. 1987. Water hyacinth. Amsterdam ; New York; New York, N.Y., U.S.A.: Elsevier; Distributors for the United States and Canada : Elsevier Science Pub. Co.

Griscom BW, and Ashton PMS. 2003. Bamboo control of forest succession: Guadua sarcocarpa in Southeastern Peru. Forest Ecology and Management 175(1–3):445- 454.

Gunderson LH. 2000. ECOLOGICAL RESILIENCE - IN THEORY AND APPLICATION. Annual Review of Ecology and Systematics 31(1):425-439.

Gupta A. 2008. Large Rivers: Geomorphology and Management: John Wiley & Sons.

Gutseit K, Berglund O, and GranÉLi W. 2007. Essential fatty acids and phosphorus in seston from lakes with contrasting terrestrial dissolved organic carbon content. Freshwater Biology 52(1):28-38.

Guy M, Taylor WD, and Carter JCH. 1994. Decline in Total Phosphorus in the Surface Waters of Lakes during Summer Stratification, and its Relationship to Size Distribution of Particles and Sedimentation. Canadian Journal of Fisheries and Aquatic Sciences 51(6):1330-1337. 133

Hamilton SK, Kellndorfer J, Lehner B, and Tobler M. 2007. Remote sensing of floodplain geomorphology as a surrogate for biodiversity in a tropical river system (Madre de Dios, Peru). Geomorphology 89(1–2):23-38.

Hanson PC, Carpenter SR, Cardille JA, Coe MT, and Winslow LA. 2007. Small lakes dominate a random sample of regional lake characteristics. Freshwater Biology 52(5):814-822.

Harrison JW, and Smith REH. 2011. Deep chlorophyll maxima and UVR acclimation by epilimnetic phytoplankton. Freshwater Biology 56(5):980-992.

Heglund PJ, and Jones JR. 2003. Limnology of Shallow Lakes in the Yukon Flats National Wildlife Refuge, Interior Alaska. Lake and Reservoir Management 19(2):133-140.

Hermoza W, Brusset S, Baby P, Gil W, Roddaz M, Guerrero N, and Bolaños R. 2005. The Huallaga foreland basin evolution: Thrust propagation in a deltaic environment, northern Peruvian Andes. Journal of South American Earth Sciences 19(1):21-34.

Hessen DO, and Leu EVA. 2006. Trophic transfer and trophic modification of fatty acids in high Arctic lakes. Freshwater Biology 51(11):1987-1998.

Hirano A, Welch R, and Lang H. 2003. Mapping from ASTER stereo image data: DEM validation and accuracy assessment. ISPRS Journal of Photogrammetry and Remote Sensing 57(5–6):356-370.

Horton BK, and DeCelles PG. 2001. Modern and ancient fluvial megafans in the foreland basin system of the central Andes, southern Bolivia: implications for drainage network evolution in fold-thrust belts. Basin Research 13(1):43-63.

Ibañez M. 1997. Phytoplankton composition and abundance of a central Amazonian floodplain lake. Hydrobiologia 362(1):79-83.

Irfanullah HM, and Moss B. 2005. Effects of pH and predation by Chaoborus larvae on the plankton of a shallow and acidic forest lake. Freshwater Biology 50(12):1913- 1926.

Izaguirre I, O'Farrell I, Unrein F, Sinistro R, dos Santos Afonso M, and Tell G. 2004. Algal Assemblages Across a Wetland, from a Shallow Lake to Relictual Oxbow Lakes (Lower Paraná River, South America). Hydrobiologia 511(1):25-36.

134

Jacques JM. 2003. A tectonostratigraphic synthesis of the Sub-Andean basins: Implications for the geotectonic segmentation of the Andean Belt. Journal of the Geological Society 160(00167649):687-701.

Jansson M, Jonsson A, Andersson A, and Karlsson JAN. 2010. Biomass and structure of planktonic communities along an air temperature gradient in subarctic Sweden. Freshwater Biology 55(3):691-700.

Jeppesen E, Jensen JP, Søndergaard M, Kjeld Sandby H, Møller PH, Rasmussen HU, Norby V, and Larsen SE. 2003. Does Resuspension Prevent a Shift to a Clear State in Shallow Lakes during Reoligotrophication? Limnology and Oceanography 48(5):1913-1919.

Jon EK, and Bond WJ. 1999. Mast Flowering and Semelparity in Bamboos: The Bamboo Fire Cycle Hypothesis. The American Naturalist 154(3):383-391.

Jones J, and Bachmann R. 1978. Trophic status of Iowa Lakes in relation to origin and glacial geology. Hydrobiologia 57(3):267-273.

Jones JR, and Bachmann RW. 1976. Prediction of Phosphorus and Chlorophyll Levels in Lakes. Journal (Water Pollution Control Federation) 48(9):2176-2182.

Jones JR, Obrecht DV, Perkins BD, Knowlton MF, Thorpe AP, Watanabe S, and Bacon RR. 2008. Nutrients, seston, and transparency of Missouri reservoirs and oxbow lakes: An analysis of regional limnology. Lake and Reservoir Management 24(2):155-180.

Jun H, Jiantong L, and Yongding L. 2009. Phosphorus in suspended matter and sediments of a hypertrophic lake. A case study: Lake Dianchi, China. Environmental Geology 58(4):833-841.

Junk WJ, Bayley PB, and Sparks RE. 1989. The flood pulse concept in river-floodplain systems. Proceedings of the International Large River Symposium Can Spec Publ Fish Aquat Sci 106. .

Kalff J, and Watson W. 1986. Phytoplankton and its dynamics in two tropical lakes: a tropical and temperate zone comparison. Hydrobiologia 138(1):161-176.

135

Kern J, Kreibich H, and Darwich A. 2002. Nitrogen Dynamics on the Amazon Flood Plain in Relation to the flood Pulse of the Solimoes River. In: McClain ME, editor. The ecohydrology of South American rivers and wetlands. 6 ed: Wallingford: International Association of Hydrological Sciences.

Knowlton M, and Jones J. 1997. Trophic status of Missouri River floodplanin lakes in relation to basin type and connectivity. Wetlands 17(4):468-475.

Kosten S, Jeppesen E, Huszar VLM, Mazzeo N, Van Nes EH, Peeters ETHM, and Scheffer M. 2011. Ambiguous climate impacts on competition between submerged macrophytes and phytoplankton in shallow lakes. Freshwater Biology 56(8):1540-1553.

Kruk C, Huszar VLM, Peeters ETHM, Bonilla S, Costa L, LÜRling M, Reynolds CS, and Scheffer M. 2010. A morphological classification capturing functional variation in phytoplankton. Freshwater Biology 55(3):614-627.

Kruk C, RodrÍGuez-Gallego L, Meerhoff M, Quintans F, Lacerot G, Mazzeo N, Scasso F, Paggi JC, Peeters ETHM, and Marten S. 2009. Determinants of biodiversity in subtropical shallow lakes (Atlantic coast, Uruguay). Freshwater Biology 54(12):2628-2641.

Lauridsen TL, SchlÜTer L, and Johansson LS. 2011. Determining algal assemblages in oligotrophic lakes and streams: comparing information from newly developed pigment/chlorophyll a ratios with direct microscopy. Freshwater Biology 56(8):1638-1651.

Leira M, Chen G, Dalton C, Irvine K, and Taylor D. 2009. Patterns in freshwater diatom taxonomic distinctness along an eutrophication gradient. Freshwater Biology 54(1):1-14.

Levi T, Shepard Jr GH, Ohl-Schacherer J, Peres CA, and Yu DW. 2009. Modelling the long-term sustainability of indigenous hunting in Manu National Park, Peru: landscape-scale management implications for Amazonia. Journal of Applied Ecology 46(4):804-814.

Liboriussen L, and Jeppesen E. 2006. Structure, biomass, production and depth distribution of periphyton on artificial substratum in shallow lakes with contrasting nutrient concentrations. Freshwater Biology 51(1):95-109. 136

Liboriussen L, Lauridsen TL, SØNdergaard M, Landkildehus F, SØNdergaard M, Larsen SE, and Jeppesen E. 2011. Effects of warming and nutrients on sediment community respiration in shallow lakes: an outdoor mesocosm experiment. Freshwater Biology 56(3):437-447.

Lottig NR, Stanley EH, Hanson PC, and Kratz TK. 2011. Comparison of regional stream and lake chemistry: Differences, similarities, and potential drivers. Limnology and Oceanography 56(5):1551-1562.

Lund SS, Landkildehus F, SØNdergaard M, Lauridsen TL, Egemose S, Jensen HS, Andersen FØ, Johansson LS, Ventura M, and Jeppesen E. 2010. Rapid changes in fish community structure and habitat distribution following the precipitation of lake phosphorus with aluminium. Freshwater Biology 55(5):1036-1049.

Marengo JA, Nobre CA, Tomasella J, Oyama MD, Sampaio de Oliveira G, de Oliveira R, Camargo H, Alves LM, and Brown IF. 2008. The Drought of Amazonia in 2005. Journal of Climate 21(3):495-516.

Masson S, Pinel-Alloul B, and Smith VH. 2000. Total Phosphorus-Chlorophyll a Size Fraction Relationships in Southern Quebec Lakes. Limnology and Oceanography 45(3):732-740.

Medina-Sánchez JM, Villar-Argaiz M, and Carrillo P. 2004. Neither with nor without You: A Complex Algal Control on Bacterioplankton in a High Mountain Lake. Limnology and Oceanography 49(5):1722-1733.

Mertes L. 1994. Rates of flood-plain sedimentation on the central Amazon River. Geology 22(2):171-174.

Myers N. 1988. Threatened biotas: "Hot spots" in tropical forests. The Environmentalist 8(3):187-208.

Myers N, Mittermeier RA, Mittermeier CG, and Kent J. 2000. Biodiversity hotspots for conservation priorities. Nature 403(6772):853-858.

Neubert MG, and Caswell H. 1997. Alternatives to resilience for measuring the responses of ecological systems to perturbations. Ecology 78(3):653-665.

137

Nicholls KH, and Dillon PJ. 1978. An Evaluation of Phosphorus-Chlorophyll- Phytoplankton Relationships for Lakes. Internationale Revue der gesamten Hydrobiologie und Hydrographie 63(2):141-154.

Nogueira EM, Nelson BW, Fearnside PM, França MB, and Oliveira ÁCAd. 2008. Tree height in Brazil's ‘arc of deforestation’: Shorter trees in south and southwest Amazonia imply lower biomass. Forest Ecology and Management 255(7):2963- 2972.

Osher LJ, and Buol SW. 1998. Relationship of soil properties to parent material and landscape position in eastern Madre de Dios, Peru. Geoderma 83(1–2):143-166.

Ostrofsky ML, and Rigler FH. 1987. Chlorophyll–Phosphorus Relationships for Subarctic Lakes in Western Canada. Canadian Journal of Fisheries and Aquatic Sciences 44(4):775-781.

ÖZkan K, Jeppesen E, Johansson LS, and Beklioglu M. 2010. The response of periphyton and submerged macrophytes to nitrogen and phosphorus loading in shallow warm lakes: a mesocosm experiment. Freshwater Biology 55(2):463-475.

PÅLsson C, Kritzberg ES, Christoffersen K, and GranÉLi W. 2005. Net heterotrophy in Faroe Islands clear-water lakes: causes and consequences for bacterioplankton and phytoplankton. Freshwater Biology 50(12):2011-2020.

Penczak T, Galicka W, Głowacki Ł, Koszaliński H, Kruk A, Zięba G, Kostrzewa J, and Marszał L. 2004. Fish assemblage changes relative to environmental factors and time in the Warta River, Poland, and its oxbow lakes. Journal of Fish Biology 64(2):483-501.

Peñuelas J, Gamon JA, Griffin KL, and Field CB. 1993. Assessing community type, plant biomass, pigment composition, and photosynthetic efficiency of aquatic vegetation from spectral reflectance. Remote Sensing of Environment 46(2):110- 118.

Phillips G, Pietiläinen OP, Carvalho L, Solimini A, Lyche Solheim A, and Cardoso A. 2008. Chlorophyll–nutrient relationships of different lake types using a large European dataset. Aquatic Ecology 42(2):213-226.

138

Picard D, Sempere T, and Plantard O. 2008. Direction and timing of uplift propagation in the Peruvian Andes deduced from molecular phylogenetics of highland biotaxa. Earth and Planetary Science Letters 271(1–4):326-336.

Quinlan R, and Smol JP. 2010. The extant Chaoborus assemblage can be assessed using subfossil mandibles. Freshwater Biology 55(12):2458-2467.

Rai H, and Hill G. 1980. Classification of central amazon lakes on the basis of their microbiological and physico-chemical characteristics. Hydrobiologia 72(1):85-99.

Räsänen M, Kalliola R, and Puhakka M. 1993. Mapa geoecologico de la selva baja peruana: explicaciones. In: Kalliola R, Puhakka M, and Danjoy W, editors. Amazona Peruana Vegetacion humeda tropical en el llano subandino Jyväskylä: PAUT & ONERN. p 207-216.

Räsänen ME, Salo JS, Jungnert H, and Pittman LR. 1990. Evolution of the Western Amazon Lowland Relief: impact of Andean foreland dynamics. Terra Nova 2(4):320-332.

Rautio M, and Vincent WF. 2006. Benthic and pelagic food resources for zooplankton in shallow high-latitude lakes and ponds. Freshwater Biology 51(6):1038-1052.

Reche I, Ortega-Retuerta E, Romera O, Pulido-Villena E, Morales-Baquero R, and Casamayor E. 2009. Effect of Saharan dust inputs on bacterial activity and community composition in Mediterranean lakes and reservoirs. Limnology and Oceanography 54(3):869-879.

Regard V, Lagnous R, Espurt N, Darrozes J, Baby P, Roddaz M, Calderon Y, and Hermoza W. 2009. Geomorphic evidence for recent uplift of the Fitzcarrald Arch (Peru): A response to the Nazca Ridge subduction. Geomorphology 107(3–4):107- 117.

Richey JE, Mertes LAK, Dunne T, Victoria RL, Forsberg BR, Tancredi ACNS, and Oliveira E. 1989. Sources and routing of the Amazon River Flood Wave. Global Biogeochem Cycles 3(3):191-204.

Rigsby CA, Hemric EM, and Baker PA. 2009. Late Quaternary Paleohydrology of the Madre de Dios River, southwestern Amazon Basin, Peru. Geomorphology 113(3– 4):158-172. 139

Roozen FCJM, Van Geest GJ, Ibelings BW, Roijackers R, Scheffer M, and Buijse AD. 2003. Lake age and water level affect the turbidity of floodplain lakes along the lower Rhine. Freshwater Biology 48(3):519-531.

Salo J, Kalliola R, Hakkinen I, Makinen Y, Niemela P, Puhakka M, and Coley PD. 1986. River dynamics and the diversity of Amazon lowland forest. Nature 322(6076):254-258.

Sampaio EV, and López CM. 2000. Zooplankton community composition and some limnological aspects of an oxbow lake of the Paraopeba River, São Francisco River Basin, Minas Gerais, Brazil. Brazilian Archives of Biology and Technology 43:285-293.

Sayer CD, Davidson TA, and Jones JI. 2010. Seasonal dynamics of macrophytes and phytoplankton in shallow lakes: a eutrophication-driven pathway from plants to plankton? Freshwater Biology 55(3):500-513.

Scheffer M. 1998. Ecology of Shallow Lakes. London: Chapman & Hall.

Scheffer M, Carpenter S, Foley JA, Folke C, and Walker B. 2001. Catastrophic shifts in ecosystems. Nature 413(6856):591-596.

Scheffer M, Hosper SH, Meijer ML, Moss B, and Jeppesen E. 1993. Alternative equilibria in shallow lakes. Trends in Ecology & Evolution 8(8):275-279.

Scheffer M, and Jeppesen E. 2007. Regime Shifts in Shallow Lakes. Ecosystems 10(1):1-3.

Scheffer M, Szabó S, Gragnani A, van Nes EH, Rinaldi S, Kautsky N, Norberg J, Roijackers RMM, and Franken RJM. 2003. Floating plant dominance as a stable state. Proceedings of the National Academy of Sciences 100(7):4040-4045.

Scheffer M, Van Geest GJ, Zimmer K, Jeppesen E, Søndergaard M, Butler MG, Hanson MA, Declerck S, and De Meester L. 2006. Small habitat size and isolation can promote species richness: second-order effects on biodiversity in shallow lakes and ponds. Oikos 112(1):227-231.

Schindler DW. 1977. Evolution of phosphorus limitation in lakes. . Science 195:260-262.

140

Smith EM, and Prairie YT. 2004. Bacterial Metabolism and Growth Efficiency in Lakes: The Importance of Phosphorus Availability. Limnology and Oceanography 49(1):137-147.

Smith M, and Nelson BW. 2011. Fire favours expansion of bamboo-dominated forests in the south-west Amazon. Journal of Tropical Ecology 27(01):59-64.

Smith VH. 1982. The Nitrogen and Phosphorus Dependence of Algal Biomass in Lakes: An Empirical and Theoretical Analysis. Limnology and Oceanography 27(6):1101-1112.

Smith VH, and Shapiro J. 1981. Chlorophyll-phosphorus relations in individual lakes. Their importance to lake restoration strategies. Environmental Science & Technology 15(4):444-451.

Sokal MA. 2007. Assessment of hydroecological changes at the Slave River Delta, NWT, using diatoms in seasonal, inter-annual and paleolimnological experiments [NR35162]. Canada: University of Waterloo (Canada).

Søndergaard M, Jensen JP, and Jeppesen E. 2003. Role of sediment and internal loading of phosphorus in shallow lakes. Hydrobiologia 506-509(1):135-145.

SØNdergaard M, Johansson LS, Lauridsen TL, JØRgensen TB, Liboriussen L, and Jeppesen E. 2010. Submerged macrophytes as indicators of the ecological quality of lakes. Freshwater Biology 55(4):893-908.

Song C, Woodcock CE, Seto KC, Lenney MP, and Macomber SA. 2001. Classification and Change Detection Using Landsat TM Data: When and How to Correct Atmospheric Effects? Remote Sensing of Environment 75(2):230-244.

Stephen D, Balayla DM, Bécares E, Collings SE, Fernández-Aláez C, Fernández-Aláez M, Ferriol C, García P, Gomá J, Gyllström M et al. . 2004. Continental-scale patterns of nutrient and fish effects on shallow lakes: introduction to a pan-European mesocosm experiment. Freshwater Biology 49(12):1517-1524.

Stets EG, and Cotner JB. 2008. The Influence of Dissolved Organic Carbon on Bacterial Phosphorus Uptake and Bacteria-Phytoplankton Dynamics in Two Minnesota Lakes. Limnology and Oceanography 53(1):137-147.

141

Swenson J, Young B, Beck S, Comer P, Cordova J, Dyson J, Embert D, Encarnacion F, Ferreira W, Franke I et al. . 2012. Plant and animal endemism in the eastern Andean slope: challenges to conservation. BMC Ecology 12(1):1.

Teixeira-de Mello F, Meerhoff M, Pekcan-Hekim Z, and Jeppesen E. 2009. Substantial differences in littoral fish community structure and dynamics in subtropical and temperate shallow lakes. Freshwater Biology 54(6):1202-1215.

Tejerina-Garro FL, Fortin R, and Rodríguez MA. 1998. Fish community structure in relation to environmental variation in floodplain lakes of the Araguaia River, Amazon Basin. Environmental Biology of Fishes 51(4):399-410.

Terborgh J. 1999. Requiem for nature. Washington, D.C. : Island Press. 234 p.

Terborgh J, and Petren K. 1991. Development of habitat structure through succession in an Amazonian floodplain forest. In: Bell SS, McCoy ED, and Mushinsky HR, editors. In Habitat Structure. London: Chapman and Hall. p 28-46.

Torezan JM, and Silveira M. 2000. The Biomass of Bamboo (Guada weberbaueri Pilger) in Open Forest of the Southwestern Amazon. Ecotropica 6:71-76.

Torremorell A, Llames ME, PÉRez GL, Escaray R, Bustingorry J, and Zagarese H. 2009. Annual patterns of phytoplankton density and primary production in a large, shallow lake: the central role of light. Freshwater Biology 54(3):437-449.

Trigal C, GarcÍA-Criado F, and AlÁEz C-F. 2007. Macroinvertebrate communities of mediterranean ponds (North Iberian Plateau): importance of natural and human- induced variability. Freshwater Biology 52(10):2042-2055.

Trigal C, Goedkoop W, and Johnson RK. 2011. Changes in phytoplankton, benthic invertebrate and fish assemblages of boreal lakes following invasion by Gonyostomum semen. Freshwater Biology 56(10):1937-1948.

Trochine C, Modenutti BE, and Balseiro EG. 2009. Chemical signals and habitat selection by three zooplankters in Andean Patagonian ponds. Freshwater Biology 54(3):480-494.

142

Tundisi JG. 1983. A review of basic ecological processes interacting with production and standing-stock of phytoplankton in lakes and reservoirs in Brazil. Hydrobiologia 100(1):223-243.

Urrutia R, and Vuille M. 2009. Climate change projections for the tropical Andes using a regional climate model: Temperature and precipitation simulations for the end of the 21st century. J Geophys Res 114(D2):D02108.

USGS. 2000. Global GIS database. Digital atlas of Central and South America [electronic resource]. [Flagstaff, AZ];: USGS Flagstaff Field Survey;.

Van Geest GJ, Roozen FCJM, Coops H, Roijackers RMM, Buijse AD, Peeters ETHM, and Scheffer M. 2003. Vegetation abundance in lowland flood plan lakes determined by surface area, age and connectivity. Freshwater Biology 48(3):440-454.

Vanni MJ, Renwick WH, Bowling AM, Horgan MJ, and Christian AD. 2011. Nutrient stoichiometry of linked catchment-lake systems along a gradient of land use. Freshwater Biology 56(5):791-811.

Vanormelingen P, Cottenie K, Michels E, Muylaert K, Vyverman WIM, and De Meester LUC. 2008. The relative importance of dispersal and local processes in structuring phytoplankton communities in a set of highly interconnected ponds. Freshwater Biology 53(11):2170-2183.

Varis O, and Vakkilainen P. 2001. China's 8 challenges to water resources management in the first quarter of the 21st Century. Geomorphology 41(2–3):93-104.

Vuille M, Bradley RS, Werner M, and Keimig F. 2003. 20th Century climate change in the tropical Andes: Observations and model results. Climatic Change 59(1-2):75-99.

Ward JV, Tockner K, and Schiemer F. 1999. Biodiversity of floodplain river ecosystems: ecotones and connectivity1. Regulated Rivers: Research & Management 15(1- 3):125-139.

Winemiller KO, Tarim S, Shormann D, and Cotner JB. 2000. Fish Assemblage structure in Relation to Environmental Variation among Brazos River Oxbow Lakes. Transactions of the American Fisheries Society 129(2):451-468.

143

Wu SK, Xie P, Liang GD, Wang SB, and Liang XM. 2006. Relationships between microcystins and environmental parameters in 30 subtropical shallow lakes along the Yangtze River, China. Freshwater Biology 51(12):2309-2319.

Yang X, Anderson NJ, Dong X, and Shen JI. 2008. Surface sediment diatom assemblages and epilimnetic total phosphorus in large, shallow lakes of the Yangtze floodplain: their relationships and implications for assessing long-term eutrophication. Freshwater Biology 53(7):1273-1290.

Zeug SC, Winemiller KO, and Tarim S. 2005. Response of Brazos River Oxbow Fish Assemblages to Patterns of Hydrologic Connectivity and Environmental Variability. Transactions of the American Fisheries Society 134(5):1389-1399.

Zhang Y, Zhang E, Yin Y, van Dijk MA, Feng L, Shi Z, Liu M, and Qin B. 2010. Characteristics and sources of chromophoric dissolved organic matter in lakes of the Yungui Plateau, China, differing in trophic state and altitude. Limnology and Oceanography 55(6):2645-2659.

144

Biography

Alana Belcon was born in the Caribbean island of Trinidad on November 4th,

1976 in the town of Arima. She was a science teacher for 6 years before migrating to the

United States to pursue a BA in Environmental Studies at Mount Holyoke College in

South Hadley, Massachusetts which she completed in 2004. Subsequent to college, she returned to Science Education as a career path and worked for 3 years in the US and in

Trinidad before returning to US in 2007 to attend Duke University in Durham, North

Carolina. Alana was accepted to Duke as a James B. Duke Scholar and completed her

PhD in Tropical Limnology and a Certificate in Geospatial Analysis in 2012. Her tenure at Duke was funded by the Center Latin America and Caribbean Studies while her research was funded by the Department of Earth and Ocean Sciences and National

Geographic.

145