(B. SOC SCI (HONS.), NUS) - COnnecting REpositoriespotential global area that may support seagrass growth is estimated at 4,320,000 km2, based on environmental drivers, specifically

TRACING SOURCES AND SPATIAL DISTRIBUTION OF

SEAGRASS SEDIMENTS, YAO YAI ISLAND, THAILAND

QUAK SONG YUN, MICHELLE

(B. SOC. SCI (HONS.), NUS)

A THESIS SUBMITTED

FOR THE DEGREE OF MASTER OF SOCIAL SCIENCE

DEPARTMENT OF GEOGRAPHY

NATIONAL UNIVERSITY OF SINGAPORE

2014

i

DECLARATION

I hereby declare that this thesis is my original work and it has been written by me in its

entirety. I have duly acknowledged all the sources of information which have been used

in this thesis.

This thesis has also not been submitted for any degree in any university previously.

Quak Song Yun Michelle

24 January 2014

ii

ACKNOWLEDGEMENT

Thank you Professor Alan D. Ziegler, you have played an integral role in the birth of this

thesis – from advising me not to take up huge ideas that are larger than what I can handle, to

initiating the (still fascinating) concept of sediment tracing of which this thesis is built upon.

You have given me the liberty to pick up new skills, explore science (haphazardly) and learn

independently. Once again, I am grateful for the surreal experience of working on a seagrass

bed and freedom to discover my surroundings. From you, I have learnt a multitude of lessons,

all of which I am grateful. Here’s one from you, and now, to you: “Stop and smell the roses.”

Thank you A/P Shawn Benner and Dr Sam Evans from Boise State University for the help

rendered to me in processing the sediment samples. Thank you Dr Joy Matthews, Sylvia

Duncan and Emily Ngo Schick from UC Davis Stable Isotope Lab for advising on the sample

preparation for isotope tests and handling payments.

Thank you Nick Jachowski for the valuable seagrass data and useful remote sensing / GIS

techniques that you have shared with me. You have been one of my sources of inspiration for

picking up some programming (R), and it has certainly made things much easier!

Thank you Mr Tow and Mr Yong for your assistance in the labs. Your timely help is

immensely appreciated. Thank you for reminding us to learn as much as we can.

Thank you piiya for your care and companionship while on Ko Yao Yai or whenever I’m in

Thailand. I look forward to meeting you each time back in Thailand. If not for your capable

help, the smooth running of the project on Yao Yai would be impossible. khun tham aahaan

Thai aroy maak khcp khun maak. duu lee na ka.

A huge thank you to my friends who have ceaselessly lent a listening ear and put up with my

hilarious moments. You have undeniably been a blessing in my life. For all the conversations,

debates and uncertainties we have had about research and life, nothing compares to the

assurance of a thesis deadline (finality!) (I kid, of course). Thank you for your constant

encouragement and clockwork countdown to the successful submission of this thesis.

Deepest thanks to my family and Weizheng who have stood by me through the years and

showered me with unconditional love and care. Thank you for bringing me laughter, joy, and

ridiculous antics to keep me going. I thank God everyday for placing all of you in my life.

Here’s to a beautiful world and better environment.

iii

TABLE OF CONTENTS

Declaration .................................................................................................................................. i

Acknowledgement ..................................................................................................................... ii

Table of Contents ...................................................................................................................... iii

Abstract ...................................................................................................................................... v

List of Tables ............................................................................................................................ vi

List of Figures .......................................................................................................................... vii

I. Introduction ............................................................................................................................ 1

1.1 Seagrass and water quality changes ........................................................................ 2

1.2 Interconnectivity of ecosystems .............................................................................. 3

1.3 Sediment source tracing ......................................................................................... 7

1.3.1 Sediment fingerprinting method ............................................................. 9

1.4 Aims and Objective .............................................................................................. 10

1.5 Thesis Outline ....................................................................................................... 11

II. Methods ............................................................................................................................... 12

2.1 Study area ............................................................................................................. 12

2.2 Sampling procedure ............................................................................................. 15

2.3 Selection of stable isotopes tracers for coastal ecosystems .................................. 16

2.4 Carbon and Nitrogen isotope analysis .................................................................. 20

2.4.1 Removal of inorganic carbon ............................................................... 20

2.4.2 Sample materials and cleaning protocol ............................................... 22

2.4.3 Acid fumigation method ....................................................................... 22

2.4.4 Acid wash method ................................................................................ 24

2.5 Mixing polygon diagrams ..................................................................................... 25

2.6 Mixing models ...................................................................................................... 27

2.6.1 Basic mixing models ............................................................................. 27

2.6.2 Excess number of sources ..................................................................... 29

2.6.3 Evaluation of commonly used stable isotope models ........................... 29

III. Isotope Results ................................................................................................................... 31

3.1 Stable isotope signatures ....................................................................................... 31

3.1.1 Isotope values for dead and fresh mangrove leaves .............................. 31

3.1.2 Organic matter - leaf material ............................................................... 31

3.1.3 Organic matter - adsorbed on sediment samples .................................. 32

3.2 Mixing polygon diagrams ..................................................................................... 33

3.2.1 Organic matter - leaf material as tracers ............................................... 33

3.2.2 Evaluation of acidification methods on organic matter - absorbed on

sediments ........................................................................................................ 35

3.3 Determining an appropriate model ....................................................................... 38

IV. Discussion .......................................................................................................................... 40

4.1 Spatial distribution of sediments and relative proportions of main sources ......... 40

4.2 Catchment- to-coast linkages ................................................................................ 43

4.2.1 Hydrologic connectivity ....................................................................... 43

4.2.2 Landscape connectivity ......................................................................... 45

4.3 Implications for the seagrass bay in Yao Yai ....................................................... 47

iv

4.4 SIAR and spatial mixing models .......................................................................... 49

V. Conclusion .......................................................................................................................... 51

5.1 Summary of thesis ................................................................................................ 51

5.2 Applicability of findings to other catchments ....................................................... 52

5.3 Future research possibilities .................................................................................. 53

References ................................................................................................................................ 55

v

ABSTRACT

Coastal vegetated ecosystems are recognised for their long-term carbon sequestration and

high capacity carbon storage that can mitigate associated impacts of global climate change.

However, due to the catchment-coast connectivity, coastal ecosystems are often at the

receiving end of terrestrial-derived pollution. As sedimentation is a major threat to coastal

ecosystems, tracing may allow managers to identify point sources which can be targeted for

mitigation strategies. The significance of considering ecosystem connectivity is demonstrated

by using δ13

C and δ15

N isotope tracers and the SIAR mixing model to map the composition

and distribution of deposited sediment in a seagrass bay of Yao Yai Island, Thailand. Through

mixing polygon diagrams, weak acidification (acid fuming) on organic matter adsorbed on

sediments, to remove inorganic carbon, was found to provide the most suitable source

signatures and seagrass sediment signatures for the sediment mixing model.

Kriging interpolation showed that 50-60% of sediments in close proximity to river mouths

were terrestrial- and mangrove-derived. Rivers enhance connectivity from catchments to the

coastal bay, implying that mangroves may not be effective buffers for seagrass ecosystems

when major flow pathways directly link terrestrial and coastal areas. Landscape composition

and configuration of the catchment was identified as an important factor contributing to the

extension of the channel network which improves hydrologic connectivity of the system and

delivery of sediments to the coast. Thus, a wider catchment-coastal system perspective and

approach must be adopted when managing sedimentation problems in coastal ecosystems.

Mitigation measures should not focus on adaptation response at the coast, but concentrate on

selecting suitable targeted solutions for land use management which addresses erosion and

hydrological/landscape connectivity.

Keywords: catchment-coast connectivity; sediment tracing; acidification; landscape

connectivity; SIAR mixing model.

vi

LIST OF TABLES

Table Description Page

1.1 Basic principal assumptions in fine sediment provenance studies……………….. ..8

2.1 Typical ranges of stable isotope δ13

C and δ15

N signatures for various types of

organic matter……………………………………………………………………..

18

2.2 Brief overview of type of acid treatment and tests on sediments and leaf

material……………………………………………………………………………

20

3.1 Isotopic signatures of organic matter (leaf material)…………………………….. 31

3.2 Isotopic signatures of organic matter adsorbed on sediment samples………….... 32

vii

LIST OF FIGURES

Figure Description Page

1.1 Ecosystem connectivity facilitates the cascade of both positive and

negative effects from catchment to coast…………………………………..

5

2.1 Land cover map of Yao Yai island………………………………………... 14

2.2 Mixing polygon bounded by source isotope signatures…………………… 25

2.3 Patterns and assumptions for proportion of sources to mixture…….……... 26

3.1 δ13

C and δ15

N results for leaf material using strong acidification on

seagrass sediments, seston and seagrass detritus…………………………..

34

3.2 δ13

C and δ15

N results for leaf material using weak acidification on

seagrass sediments, seston and seagrass detritus…………………………..

34

3.3 δ13

C and δ15

N results for organic matter adsorbed on sediments, using

strong acidification process………………………………………………..

36

3.4 δ13

C and δ15

N results for organic matter adsorbed on sediments, using

weak acidification process………………………………………………....

36

3.5A δ13

C values of acid fumed seagrass sediment plot against seagrass

detritus……………………………………………………………………...

37

3.5B Seagrass sediment sampling locations……………..……………………… 37

3.6 Expansion of mixing polygon (solid line) made by modifying δ13

C and

δ15

N median values of coral and seagrass detritus source groups………….

38

4.1 Spatial interpolation of sediment composition for each source.…………... 41

4.2 Seagrass detritus-derived sediments and distribution of seagrass………… 42

4.3 Ecosystem linkages framework showing the interlinked relationships

between all components………………………………………………........

43

4.4 Sediment transport pathways……………………………………………… 44

4.5 Five major components of catchment hydrological connectivity………..... 45

4.6 Land cover/use configuration of inland catchment………………………... 48

1

1. Introduction

Seagrasses are aquatic flowering plants commonly found along tropical, temperate

and subartic coastal margins (Orth et al. 2006; Short and Wyllie-Echeverria 1996; Duarte

2002). The vast worldwide distribution of seagrass, spanning a wide range of latitudinal

regions, reflects its adaptability to various environmental conditions and habitats (Orth et al.

2006). The documented global areal extent of seagrass is reported to be 177,000 km2 (Green

and Short 2003). It is considered a conservative estimate because Southeast Asian regions are

known to contain many large unmapped meadows (Waycott et al. 2009; Ooi et al. 2011). The

potential global area that may support seagrass growth is estimated at 4,320,000 km2, based

on environmental drivers, specifically benthic irradiance modelling (Gattuso et al. 2006;

ecosystem scale prediction: Grech and Coles 2010). Global coverage of seagrass is

comparable to other geographically restricted coastal ecosystems, such as mangroves and

coral reefs (137,760-152,361 km2 and 22,000-400,000 km

2 respectively) (Mcleod et al. 2011).

Coastal vegetated ecosystems such as salt marshes, mangroves and seagrasses are recognized

for their ability to sequester large amounts of carbon (C) disproportionally to their areal extent

(Hopkinson et al. 2012; Mcleod et al. 2011). The carbon stored in vegetated coastal

ecosystems, such as mangroves, seagrasses and salt marshes is referred to as ‘blue carbon’

(Mcleod et al. 2011). The high carbon burial rate of 111 Tg C y-1

in vegetated habitats allows

these coastal ecosystems to act as effective long-term organic carbon stores, exceeding burial

rates in terrestrial sinks (Mcleod et al. 2011; Duarte et al. 2005). Seagrass meadows are

recognized for their ability to sequester carbon in their rhizome biomass and more

significantly in deposited sediments (Duarte et al. 2011; Fourquean et al. 2012). With

organic-rich sediments (averaging 4.1% organic carbon concentration; Kennedy et al. 2010),

seagrass meadows have the capacity to sequester up to 27.4 Tg C y-1

, which is 11.2% of the

yearly global organic carbon ocean burial (210-244 Tg C y-1

), despite covering less than 0.2%

of the ocean surface (Fourqurean et al. 2012; Duarte et al. 2005). Conservative estimates of

organic carbon stored in seagrass biomass and top metre of seagrass sediments is 2.52 ± 0.48

2

Mg C ha-1

and 139.7 Mg C ha-1

, respectively (Fourquean et al. 2012). To remain an effective

coastal blue carbon store, there must be a continuous increase in absolute rate of sequestration

and an expansion of its areal extent over time (Hopkinson et al. 2012).

However, seagrass meadows are facing rapid degradation, conversion and health deterioration

as a result of multiple stressors (Waycott et al. 2009; Orth et al. 2006). The accelerated

estimated mean decline in seagrass area from 0.9% yr-1

before 1940 to 7.0% yr-1

since 1990

reflects the devastating effect from a broad spectrum of anthropogenic and natural stressors

(Waycott et al. 2009). Seagrasses generally recover from natural disturbances that involve

pulses of sediment redistribution (e.g. inlet migration and hurricanes); however, human-

induced disturbances causes long-lasting changes in the sedimentary environment, often

resulting in permanent seagrass loss (Cabaco et al. 2008). Although the distribution and health

of seagrass meadows were dominantly controlled by gradual changes in environmental

conditions due to natural drivers, for example climate change and geological events, much of

the damage afflicted on seagrass meadows has been from various anthropogenic activities

concentrated at the coasts (Salomons et al. 2005; Orth et al. 2006, Short and Wyllie-

Echeverria 1996, Duarte 2002; Elliott and Whitfield 2011).

1.1 Seagrass and water quality changes

Seagrasses grow in shallow, protected waters that usually receive catchment nutrients

and sediment inputs (Orth et al. 2006). Seagrass biomass and the nutrient content of seagrass

plants and sediments usually reflect elevated nutrient concentrations in the water column or

contributing sediment (Orth et al. 2006; Mellors et al. 2005; Freeman et al. 2008; Miller and

Sluka 1999). Although seemingly contradictory, seagrasses are highly sensitive to stress-

induced changes in the water quality, yet they are also resilient to such short-term stresses

(Lapointe and Clark 1992). It is this resilience and high sensitivity towards water quality,

water irradiance and clarity that render seagrasses as excellent biological sentinels or "coastal

canaries" of harmful environmental stresses (Orth et al. 2006). For example, sediments in

3

seagrass meadows had enriched phosphorous due to chronic input of organic fishing waste

from adjacent fishing villages on Laamu Atoll, Maldives (Miller and Sluka 1999). The

nutrient enrichment was beneficial to seagrass growth – seagrass cover was higher at fishing

villages than non-fishing villages and uninhabited islands, indicating the effects of land use

on adjacent ecosystems (Miller and Sluka 1999).

Land use changes in upper catchments also result in high erosion yields that contribute to

siltation and deterioration of sediment conditions in coastal waters (Duarte 2002; Salomons

2005; Lee et al. 2006). Prolonged reduction of underwater irradiance inhibits photosynthesis

processes and seagrass growth, leading to large-scale seagrass die-off (Burkholder et al. 2007;

Lee et al. 2007). Common causes of light reduction are the overgrowth of phytoplankton,

epiphytes and macroalgae due to nutrient over-enrichment, resuspension of meadow bed

sediments and increased sediment runoff from upper catchments (Burkholder et al. 2007; Lee

et al. 2007). The exchange of material and nutrients between ecosystems is facilitated by

transport pathways such as rivers and surface runoff. Such cascading effects from upland

catchments to coastal zones reflect the connectivity between the catchment and coastal zones

(Sheaves 2009; Mitchell et al. 2013; Russell et al. 2013).

1.2 Interconnectivity of ecosystems

Estuaries are often considered as 'open' and multi-interfaced systems with coupled

major influences and ecosystems (Elliott and Whitfield 2011; Alvarez-Romero et al. 2011).

Coastal ecosystems such as seagrass, mangrove and coral reefs are located at the boundaries

of terrestrial and offshore marine ecosystems. They form a crucial connection between the

two different environments (Sheaves 2009; Mitchell et al. 2013; Alvarez-Romero et al. 2011).

Hemminga et al. (1994) effectively illustrated carbon flux exchange between mangroves and

seagrass ecosystems, highlighting the buffering effect of seagrass meadows situated between

mangroves and coral reefs in Gazi Bay, Kenya. This interconnectivity of ecosystems not only

enhances the exchange and transfer of nutrients, energy, and materials between coastal

4

ecosystems (Orth et al. 2006; Salomon 2005; Mitchell et al. 2013), but also increases the

susceptibility of coastal habitats that are at the 'receiving end' of the cascade of environmental

effects originating from terrestrial ecosystems (Lee et al. 2006; Elliott and Whitfield 2011).

Conversely, ecosystem linkages allow for the cascade of positive effects that habitats can

create throughout ecosystems (Russell et al 2013). For example, healthy coastal ecosystems

are able to reduce adverse effects originating from the catchment by acting as 'buffers' and

sinks for harmful substances (Figure 1.1). Lee et al. (2006) highlights the impact and stressors

of urbanization on coastal ecosystem structures, and identified sedimentation as a major

pollutant and problem. Besides exacerbating poor water clarity, suspended fine sediments also

have the ability to absorb and adsorb nutrients and pollutants (Lee et al. 2006; Owens 2007).

Therefore, sediments, along with water, act as a link and medium of nutrients/pollutants

transfer between terrestrial, fluvial, estuarine and marine environments, thus connecting river

catchments to coastal ecosystems (Salomons 2005). This dynamic exchange and transport of

material is termed as the 'catchment-coast continuum' (Owens 2007; Salomons 2005).

Depending on the catchment connectivity, modifications in the sediment or water source and

fluxes upstream, for example by land cover/use change, will drive changes in downstream and

coastal areas (Salomons 2005). The magnitude of impacts at the coastal zone is confounded

by a few inherent complexities in catchment-coast sedimentary systems (Owens 2007). Non-

linear and unpredictable natural events such as extreme storms introduce uncertainty and

feedbacks in management response (Slob and Gerrits 2007). Furthermore, the buffering

capacity of catchment soils and sediments alters the quantity and quality of sediment fluxes to

coastal ecosystems (Fryirs 2012). The dynamics of sediment fluxes are usually influenced by

delayed and non-linear responses to changes in the upstream sources (Salomans 2005; Fryirs

2012). The interconnectivity of ecosystems, shown through the example of sediment transport

and exchanges, illustrates that these ecosystems are not mutually exclusive (Sheaves 2009;

Figure 1.1). Each ecosystem cannot be considered in isolation, but as part of a larger system

of the catchment-coast continuum (Salomons 2005; Sheaves 2009).

5

Figure 1.1. Ecosystem connectivity facilitates the cascade of both positive and negative effects from catchment to coast. The ability of coastal ecosystems to function as

‘buffers’ or ‘sinks’ is dependent on the adaptability threshold and response rate of the ecosystem, and the magnitude and frequency of pollution events.

(Source: Author’s own.)

6

Increasing emphasis has been placed on research in coastal-catchment linkages that highlight

the adverse impact of human modifications in upland catchment areas on coastal ecosystems

(Salomons 2005). Understanding the connectivity between ecosystems, and the processes that

affect them, is crucial for proper seagrass ecosystem management for ecology and coastal

protection. Some of the management concerns include inter-ecosystem exchange facilitation

and blue carbon sequestration in seagrass sediments. The latter has resulted in many studies

relating to the trapping ability of seagrass beds (e.g. Mellors et al. 2002; Cabaco et al. 2008;

van Katwijk et al. 2010; van der Heide et al. 2011) and quantifying the amount of carbon

stored in seagrass meadows (e.g. Duarte et al. 2011; Fourqurean et al. 2012; Duarte et al.

2005; Kennedy et al. 2010; McLeod et al. 2011). However, little is known about how seagrass

ecosystems are simultaneously or solely affected by natural variations in the environment and

anthropogenic activity across different scales and regions (Orth et al. 2006). Aggravation of

seagrass health could occur in varying degrees concurrently, making it difficult for

pinpointing the main source of stress. Duarte et al. (2004) acknowledge the important effects

that direct or indirect human interventions exert on seagrass ecosystems at regional and global

scales. The challenge in coastal ecosystem management lies in separating these effects from

ecosystem responses to background natural environmental changes that are common to highly

dynamic coastal ecosystems. Thus, to ensure effective management strategies for coastal

ecosystems, anthropogenic source of disturbances, which tend to manifest at a local scale,

have to be distinguished from indirect effects, which usually occur at a larger spatial scale

(Duarte et al. 2004).

Despite the increased attention on coastal ecosystems, and seagrass ecology in particular,

public awareness of seagrass matters remains lacking, possibly due to the ineffective

dissemination of scientific understanding (Orth et al. 2006; Duarte et al. 2008). Furthermore,

Orth et al. (2006) suggests that the inherently 'obscure' nature of submerged seagrass

ecosystems and elusiveness of its fauna, unlike the more attractive coral reefs, renders

seagrass meadows less appealing to the public. Moreover, unlike mangroves, seagrass

7

ecosystems may not provide sufficient coastal protection extreme storm events, creating a

perception of its structural unimportance. However, seagrasses do provide valuable ecosystem

services which help to sustain neighbouring ecosystems such as mangroves and coral reefs

(e.g. Unsworth et al. 2012). The misconception of the unimportance of seagrass ecosystem is

exacerbated by the lack of media attention on seagrass ecosystems, which feeds back to the

lack of awareness and the undue imbalance charisma of seagrass ecosystems (Duarte et al.

2008). Public awareness of the potential goods and services that seagrass ecosystems can

provide to adjacent coastal habitats is crucial to effective conservation and management of

such ecologically important and valuable ecosystems (Duarte et al. 2008). As such, research

into the ecological connectivity and linkages between seagrass ecosystems and other

terrestrial and coastal ecosystems is beneficial to the understanding of the importance of such

contributors to estuarine ecosystem function, and to assess management decisions.

1.3 Sediment source tracing

Sediment tracing is a tool for studying sediment linkages between and within

ecosystems. Sediment fingerprinting provides a direct approach of identifying erosion sources

in a catchment and apportioning the amount of sediment contributed from these sources. This

is carried out through a combination of field data collection, laboratory analyses, and

statistical modelling methods (Davis and Fox 2009; Collins and Walling 2004). Aside from

using sediment fingerprinting as a research technique, Mukundan et al. (2012) highlights its

potential to serve as a management tool to identify major sources of fine particulate sediment,

and sediment-associated nutrients and contaminants, for erosion management, sediment

budgets and pollution mitigation strategies (Foster and Lees 2000; Walling 2005).

Fingerprinting studies are based on comparing the composition of soil properties (natural

tracers) of accumulated sediment at a sink, with soil properties from different areas or erosion

sources around the catchment (Guzman et al. 2013). Such natural tracer properties include

physical, chemical and biological aspects of soil/sediments. Biogeochemical tracers are more

8

commonly used compared with physical tracers (e.g. particle size, colour and sediment-

discharge relationships). Biogeochemical tracers (in descending preference of usage) include,

inorganic tracers (e.g. mineral magnetism and mineral elements such as Fe, Al, Ni),

radionuclide tracers (e.g. 137

Cs, 210

Pb, 7Be) and organic tracers (e.g. plant pollen, total organic

carbon/phosphorus/nitrogen, stable isotopes δ13

C, δ15

N, etc.) (Davis and Fox 2009; Guzman et

al. 2013). A range of assumptions are involved at each stage of analysis and for each type of

tracers (e.g. Foster and Lees 2000; Davis and Fox 2009; Collins and Walling 2004). Most

importantly, tracers should fulfil the fundamental assumption of sediment tracing technique:

that it can differentiate between erosion sources and maintain its tracer properties between

sediment generation (erosion), transport (delivery), deposition and analysis (Guzman et al.

2013; Mukundan et al. 2012) (Table 1.1).

Table 1.1. Basic principal assumptions in fine sediment provenance studies.

(Source: Foster and Lees 2000).

Applicability Assumption

1 All tracer

studies

Tracer must distinguish between erosion sources

2 Tracer is transported and deposited the same way as medium of interest (i.e. in

association with fine sediment)

3 Tracer properties are not affected by selective erosion and transport (e.g.

particle size or density)

4 Studies on

deposited

sediments

Tracer properties of each source sediments remain chemically different over the

period of sediment deposition (conservativeness of tracer properties)

5 Tracer signatures of source sediments remain chemically unchanged (no

transformation by enrichment, dilution or depletion) from the point of

deposition to analysis

6 Mixing

models

Mixing models used to reconstruct sediment provenances are able to deal with

inherent variability in source signatures and provide estimates of source

contributions within acceptable known or predictable tolerances

Multiple tracers should be utilised to distinguish between sediment sources based on several

characteristics, and to provide a more reliable and accurate representation of mixtures

comprised of catchment-derived material (Foster and Lees 2000; Davis and Fox 2009; Collins

and Walling 2004). Composite fingerprinting, used in multivariate tracer suites, combine

individual tracer properties that are influenced by various contrasting environmental controls

or watershed characteristics such as land use, rock type and soil depth (Davis and Fox 2009;

Collins and Walling 2004). As such, usage of multiple tracers allow for more possibilities in

terms of research purposes, while ensuring that accuracy is not undermined.

9

1.3.1 Sediment fingerprinting method

The sediment fingerprinting methodology can be generalized into five steps (Foster

and Lees 2000; Davis and Fox 2009; Mukundan et al. 2012; Small et al. 2002):

1) Identify and classify sediment sources;

2) Sample collection and laboratory analysis of sediment at sources and sinks;

3) Select unique tracers representative of each sediment source (by statistical

test/literature);

4) Utilize a multivariate mixing model for sediment source apportionment;

5) Explanation and environmental management conclusions.

Without any guidelines on the optimal number of samples to effectively represent sediment

sources (Collins and Walling 2004), there could be an infinite number of possibilities of

sediment sources. Phillips et al. (2005) identified two methods of classifying sources - a

priori and a posteriori aggregation approaches. The a priori method combines sources based

on similarity of isotopic signatures and logical association between sources (Phillips et al.

2005). As a result of combining sources, the variability of isotopic signatures of the

aggregated source increases, translating into greater uncertainty in source contribution

estimates from mixing models (Phillip et al. 2005; Foster and Lees 2000). Therefore, it is

important to define the variability in source properties and select tracer fingerprints that have

low variability to minimize errors in mixing model results and source properties (Foster and

Lees 2000; Small et al. 2002). The a posteriori approach of aggregating sediment sources is

used when an a priori combination of related similar isotopic signature sources is insufficient

for mixing models to find a unique solution (Phillips et al. 2005).

Following the collection of sediment samples at sources and sinks or time-integrated

sampling of suspended sediments (Phillips et al. 2000), relevant laboratory analysis would be

carried out (Section 2.4). In multivariate tracer studies involving a suite of tracers (e.g.

major/minor/trace/rare earth elements), an extra step of statistical analysis is included for

10

selecting suitable tracers based on the ability to discriminate between sources (e.g. Mann-

Whitney U-test, Kruskal-Wallis H-test, Wilcoxon rank-sum test and the Tukey test)

(Mukundan et al. 2012; Davis and Fox 2009). This is followed by the use of classification

techniques to further narrow down the selection to an optimal combination of tracers (e.g.

linear methods: stepwise multivariate discriminating function; nonlinear methods: logistic

regression, artificial neural network, cluster analysis, PCA, factor analysis, etc.) (Mukundan

et al. 2012; Foster and Lees 2000; Davis and Fox 2009). However, this classification

procedure is only necessary if a large suite of tracers is used.

The selected signatures and representative source materials are applied to mixing models to

estimate relative contribution of each sediment source, using a variety of approaches

including bivariate regression models, multiple regression or linear programming techniques

for solving simultaneous equations (Foster and Lees 2000). More recently, the development

of Bayesian approaches have allowed for the inclusion of uncertainty and variability in source

signatures into mixing model outcomes, thus producing more robust results (Parnell et al.

2010, 2013; Small et al. 2002; Davis and Fox 2009) (Section 2.6). From the mixing model

results, environmental problems can be identified and appropriate management measures and

mitigation efforts can be devised.

1.4 Aims and hypotheses

To illustrate the role of sediment in catchment-coast connectivity, this thesis aims to

use sediment tracing methods to identify primary sources of sediment and the spatial

distribution of deposited sediment in a seagrass meadow of Yao Yai island, Thailand. The

following hypotheses are tested:

1) Spatial distribution of sediment deposited in the bay would have a large terrestrial

signature, related to major pathways of sediment transfer and transport, and

proximity to sources.

11

2) Organic matter adsorbed on sediments would serve as a better tracer host (for

source identification) than organic matter as leaf material from the various plant

types in the different ecosystems.

The selection of an appropriate tracer is crucial in producing meaningful mixing results.

Stable isotope δ13

C and δ15

N signatures of two forms of organic matter (plant material versus

mixture of organic matter) are tested and evaluated using mixing polygons to determine the

appropriate tracer. Furthermore, the method of inorganic carbon removal will be assessed.

The δ13

C and δ15

N data obtained from the selected tracer material and appropriate

acidification process will be modelled to form a deposited sediment spatial distribution map

of the bay. Various reasons for the spatial distribution patterns will be explored. The

implications on catchment-coast connectivity and system approaches in land use management

will be discussed in relation to the first hypothesis.

1.5 Thesis outline

Chapter 2 will describe the study site and go into detail about the research methods

utilized in this thesis. This chapter also focuses on the use of stable isotopes for tracing in

coastal ecosystems. It presents literature findings on typical isotope values for sources, and

the evaluation of different mixing models. Chapter 3 presents and discusses the results

attained through relevant laboratory methods and statistical analysis. Chapter 4 suggests

plausible reasons for the results obtained from kriging interpolation, and discusses the

implications on coastal-catchment management, focusing on the importance of linkages

between ecosystems and the coastal-catchment continuum. Chapter 5 concludes with

evaluating the applicability of the findings to other coastal catchments and suggestions on

future research possibilities.

12

2. Methods

2.1 Study area

The study site is located at the southern bay of Yao Yai island (98 o

35’E, 7o55’N),

which is situated in Pangnga Bay, between Phuket and Krabi, Thailand. The bay is about 1.2

km wide and 3 km long, and has nine sub-watersheds but dominated by one (Figure 2.1). The

dry season occurs during November–April (Northeast monsoon) and wet season during May–

October (Southwest monsoon) (mean annual precipitation is 2266 mm) (Chansang 1984). The

study area is subjected to semidiurnal tides with a tidal range of about 2.5 m during spring

tide (Chansang 1984). The sheltered bay protects the seagrass and coral reef ecosystems from

strong open sea waves allowing for deposition of sediments in the bay (Figure 2.1). While

tidal flow and wave currents are not intensively studied here, it is logical to assume that the

geomorphology of the bay regulates and allows for some circulation, but does not allow

extreme mixing with the open sea.

The geology of the island is relatively homogeneous, mostly sandstone, with some outcrops.

The dominant land cover is now rubber and coconut plantations (Figure 2.1), with increasing

amounts of natural forests converted for agriculture purposes. Quarries are found around the

island and road cuts are frequently found in the catchment, especially along the two ‘claws’

that shelters the seagrass bay. Some settlements can be found with minor drain networks

which lead to the bay. The likely primary livelihood of the community was once fishing, but

it is now heavily involved in rubber plantations.

Two rivers are located within the catchment. The main channel originates upland where

agriculture land and settlements are located, and flows along the west side of the mangrove

area. The secondary channel flows along the east edge of the mangrove and stretches only

half the forest length (Chansang 1984). Anthropogenic pollution from boat diesel, fertilizers

and sewage probably contributes to nutrient loads into this river via direct runoff, or indirectly

through runoff erosion, sediment transport, and deposition.

13

The riverine mangrove forest is a narrow strip, about 1 km wide and stretches about 3 km

inland (Figure 2.1). There are at least 10 species of mangroves found in their natural condition

with minimal cutting; Rhizophora apiculata is the dominant species (Changsang 1984). Four

to five species of seagrasses were recorded, along the eastern portion of the bay: Halodule

pinifolia, Cymodocea rotundata, Thalassia hemprichii, Enhalus acoroides and Halophila

ovalis (Chansang 1984; Chansang and Poovachiranon, 1994). The seagrasses grow on

shallow sandy and muddy substrates that are typically N-limited environments (Burkholder et

al. 2007). Predictions of seagrass cover was based on algorithmic modelling of various

geographical (distance from river/mangrove/land), biophysical (phosphate, salinity, pH,

secchi disk depth, turbidity and temperature) and depth parameters with 85% maximum

accuracy in a prior study (Jachowski, n.d.; Figure 4.2). Coral reefs can be found farther south

of the bay. The presence of hard corals may indicate a higher total inorganic carbon signature

of sediments due to carbonates found in dead hard corals.

14

Figure 2.1. Land cover map of Yao Yai island. Rubber and coconut plantations are the dominant land

covers of the island. The 1.2 km by 3 km bay is fed by nine sub-watersheds (delineated with the black

solid line). Landcover analysis was carried out with a 2 m resolution DigitalGlobe satellite image (Date

of image: July 2012), using a supervised classification technique in ArcGIS v10.1 program. Ground-

truthing was done during the two visits to the study site. (Source: Author’s own)

Yao Yai island,

Phang Nga Bay

15

2.2 Sampling procedure

Samples were collected in February and October 2012. Four major end-members (or

sources) were chosen to represent probable sources of deposited sediment in the seagrass bay:

terrestrial erosion sources, mangrove and coral sediments and seston. Detritus material from

terrestrial plants, mangrove trees, seagrass vegetation and seston were collected as sources of

sediment organic matter. The method of selection and aggregation of sediment sources

follows the a priori method suggested by Phillips et al. (2005) (Section 1.3.1).

Terrestrial erosion sources such as quarries, road/slope cuts, plantations, dry creeks and

possible channel heads were sampled (n=32). As suggested by many studies, channel banks

may be an important source of sediments. Therefore, samples were collected within

mangroves and along the river banks at the edges of the mangroves (n=31). Some samples

were collected at coral areas (n=15). Sampling at the seagrass bed was stratified spatially and

distributed across the bay to ensure good representation of spatial deposition (n=27).

The top 15 cm of the substrate/deposited material was collected using plastic PVC scoops and

stored in ziplocks to avoid contamination. About 0.5-1.2 kg of sediments were collected at

each sampling point to ensure that sufficient amount of fines (<63 µm) were obtained for

chemical analysis.

Detritus material, in the form of whole brown plant leaves, was collected from paddies,

natural forests, rubber and coconut plantations, to determine the contribution of organic

matter from terrestrial plants. The leaves were stored in pre-combusted glass vials during

sample collection. Terrestrial plants (n=14), mangrove (n=12) and seagrass plants (n=12)

were sampled to determine if the organic matter in the mangrove and seagrass bed originated

primarily from the internal nutrient or decomposition cycling of the ecosystems, or from

terrestrial sources. Although Bouillon et al. (2008) found no difference in isotopic values for

floating mangrove leaves (collected in creeks or offshore) and fresh leaves, both fresh and

16

senescent leaves were picked to represent mangrove leaves and to verify this finding. Using

isotopic methods, Kennedy et al. (2010) found that non-seagrass organic matter had a stronger

contribution to accumulated carbon in seagrass sediments. Thus, it is important to examine

the degree to which sediments in mangrove and seagrass beds are affected by mixing from

terrestrial organic matter. Seston samples (suspended particulate matter in the water column:

organic matter, suspended sediments, zooplankton and phytoplankton) were collected from

the top 0.5 m of the water surface at the mouth of the bay using a plankton net. Samples were

kept refrigerated before further processing.

2.3 Selection of stable isotope tracers for coastal ecosystems

All stable isotope data results are reported as per mille (‰) deviations from a

standard Vienna-Pee Dee Belemnite (PDB) for carbon (δ13

C) and atmospheric air for nitrogen

(δ15

N):

where R values represent either 13

C/12

C (for C isotopes) or 15

N/14

N (for N isotopes). δ13

C

values in Section 2.3 are reported as deviations from PDB standard. It is similar to the newer

Vienna-PDB standard (Coplen 1994).

Carbon isotopes for organic matter

Coastal ecosystems contain a broad spectrum of vegetation types, from terrestrial

plants in the catchment to mangroves and seagrasses towards the coast. These plants have

different photosynthetic types and biochemical pathways, with a unique carbon isotope

fractionation pattern that fixates carbon differently (Boutton 1991; Schimel 1993). For

example, plants with a C3 pathway of photosynthesis incorporate CO2 into a 3-carbon

compound, whereas the more efficient plants with C4 pathways incorporates it into a 4-

carbon compound (Boutton 1991). As a result, C3 plants have distinctively lower δ13

C values

ranging from -32 to -20‰ (mean = -28 to -27‰), compared with -17 to -9‰ (mean = -14 to -

(1)

17

13‰) for C4 plants (Boutton 1991; O'Leary 1988) (Table 2.1). Hobbie and Werner (2004)

and O'Leary (1981; 1988) provide several explanations to the mechanisms that result in

isotopic differences in C3 and C4 plants. Most terrestrial plant species are C3 plants, with the

exception of corn, tropical grasses, salt marsh grasses and plants living in dry regions or in

high salinity areas (Boutton 1991). In general, mangrove trees are C3 plants that have δ13

C

values ranging from -30 to -24‰ (Hemminga and Mateo 1996) (Table 2.1)

Aquatic plants, such as seagrasses, show significantly more positive δ13

C values than

terrestrial C3 plants (O'Leary 1981), ranging from -15 to -3‰ (mean = -10 to -11‰) (Boutton

1991; Hemminga and Mateo 1996) (Table 2.1). Despite having isotope signatures that lie

typically within the range of C4 plants, seagrasses still have the C3 type photosynthetic

metabolism (Hemminga and Mateo 1996). Phytoplankton or seston have δ13

C values ranging

from -30 to -18‰ (usually near -22‰) (Boutton 1991). However, these results are

compounded by effects from diffusion of CO2 dissolved in water, salinity, temperature, CO2

availability and mixing flow dynamics of the environment (O'Leary 1988; Boutton 1991).

Nitrogen isotopes for organic matter

The δ15

N values of seagrass leaves vary from -2‰ to 12.3‰ with the most frequent values

occurring between 0‰ to 8‰ (Lepoint et al. 2004) (Table 2.1). The reasons owing to large

variations in δ15

N signatures of plant types are poorly understood, but are generally due to

inorganic N uptake by seagrasses, simultaneously occurring denitrification and nitrification

processes that affect sediment and water geochemistry, and N2 fixation by seagrass organisms

(Lepoint et al. 2004). Large variations that are also observed in terrestrial and mangrove plant

δ15

N values (Bouillon et al. 2008), may be attributed to complex biogeochemical processes by

microbial organisms which mobilizes and fixes nitrogen in the sediments or soil due to

microbial activities. Nitrogen fixing which occurs more prevalently in tropical areas would

drive δ15

N values of terrestrial plant tissue towards 0‰ (Ometto et al. 2006; Lepoint et al.

2004). On the contrary, Muzuka and Shunula (2006) found δ15

N values of mangrove

18

vegetation ranging from -1.5‰ to 3.2‰ in Tanzania, while values of -0.4‰ to 6.3‰ were

found in Trang, Thailand (Kuramoto and Minagawa 2001) (Table 2.1). Gonneea et al. (2004)

reported δ15

N values of 6.79 ± 3.43‰ for fresh leaves and significantly heavier 9.75 ± 3.21‰

for senescent leaves. The varying ranges for δ15

N values of mangrove vegetation across

different sites illustrate the site-specificity of nitrogen isotope readings.

Table 2.1. Typical ranges of stable isotope δ13

C and δ15

N signatures for various types of organic matter.

Organic

matter

δ13

C Source δ15

N Source

C4 plants -17 to -9‰

(mean = -14 to -13‰)

Boutton (1991);

O'Leary (1988)

Large variations Bouillon et al.

(2008)

C3 plants -32 to -20‰

(mean = -28 to -27‰)

Mangrove -30 to -24‰ Hemminga and

Mateo (1996)

-0.4 to 6.3‰ (Trang) Kuramoto and

Minagawa (2001)

-1.5‰ to 3.2‰

(Tanzania)

Muzuka and

Shunula (2006)

6.79 ± 3.43‰

(fresh leaves)

9.75 ± 3.21‰

(senescent leaves)

Gonneea et al.

(2004)

Aquatic

plants

-15 to -3‰

(mean = -10 to -11‰)

Boutton (1991);

Hemminga and

Mateo (1996)

-2‰ to 12.3‰

(usually 0‰ to 8‰)

Lepoint et al.

(2004)

Seston -30 to -18‰

(usually near -22‰)

Boutton (1991) -1.2‰ to 10.6‰

(mean = 4.5‰, but

varies, depending on

sampling sites)

Cloern et al.

(2002)

Stable isotopes for soil/sediments

Soil organic carbon retains the isotopic signature of the vegetation that degrades to form soil;

thus, most tropical forests have soil carbon with δ13

C values of about -24 to -29‰ (Schimel

1993), which lie within the range of terrestrial C3 plants. Likewise, δ13

C values of mangrove

and seagrass sediments are expected to display similar carbon isotope signatures to the

vegetation it supports. Unlike carbon isotopes, nitrogen isotopes are more volatile to changes

in environmental conditions. Anaerobic environmental conditions and high microbial activity

that causes gaseous nitrogen losses, favour the denitrification process involving isotopic

fractionation. It causes δ15

N of residual nitrate to increase as nitrate concentration decreases

19

(Mukundan et al. 2012). High N fertilization in agricultural areas would also leave behind a

substrate enriched in 15

N (Ometto et al. 2006). This enrichment in soil 15

N-isotope results in

higher δ15

N values. Therefore, sediments in anaerobic mangrove environments may display a

higher δ15

N value compared to terrestrial soils where aerobic conditions suppress

denitrification processes. However, such assumptions cannot be readily applied on all sites,

especially if agriculture (use of N based fertilisers) and wetlands (similar anaerobic processes)

are common land use features in the catchment.

The δ15

N readings may be complicated by nitrogen fixing processes; however, when used

together with δ13

C data, the uncertainty of δ15

N results may be minimalized. Furthermore,

single tracers alone may not differentiate between certain sediment sources; but a second

tracer may produce distinct signatures between the two sources (e.g. δ13

C does not

differentiate well between mangrove and terrestrial sediments, however δ15

N allows for

discrimination between the sources). This demonstrates the importance of using multiple

tracers instead of a single tracer. Stable isotopes δ13

C and δ15

N remain as one of the most used

sediment source tracers in coastal ecosystems due to the distinct isotopic signatures of

different types of plants, especially between terrestrial and aquatic plants. These stable

isotopes have also shown greater sensitivity than total elemental composition (Davis and Fox

2009). Although elemental tracers are time and cost efficient (multiple elements can be tested

simultaneously) (Mukundan et al. 2012), existing literature supports the use of stable isotopes

like δ13

C and δ15

N that are effective tracers for coastal ecosystems.

20

2.4 Carbon and Nitrogen isotope analysis

2.4.1 Removal of inorganic carbon

Soils and sediments contain both inorganic carbon (IC) in the form of carbonates and

organic carbon (OC) that is derived from microorganisms, animal or plant matter in various

states of decomposition (Bisutti et al. 2004). The two forms of carbon have distinct δ13

C

signatures and must be separated prior to isotope analysis to prevent influence of δ13

C

concentrations in sediment organic matter (Harris et al. 2001; Komada et al. 2008; Brodie et

al. 2011). Complete removal of carbonates without compromising OC would improve the

accuracy of C/N analysis, and isotopic composition for identifying organic matter sources

(Kennedy et al. 2005; Komada et al. 2008). Bisutti et al. (2004) summaries the advantages

and problems of current available methods for determining total OC from solid samples. Both

strong acidification (acid washing) and weak acidification (acid fumigation) were carried out

on the sediment samples (Table 2.2). Assuming that seagrass detritus and seston would

contain high IC content due to their presence in coastal waters and coral reefs, these samples

were also acid treated.

Table 2.2. Brief overview of type of acid treatment and tests on sediments and leaf material.

Weak acidification (Acid fumed) Strong acidification (Acid washed)

δ13

C δ15

N Type δ13

C δ15

N Type

Sediments

Terrestrial Acidified Non-acidified Single Acidified Acidified Dual2

Mangrove Acidified Non-acidified Single Acidified Acidified Dual

Corals Acidified Non-acidified Single Acidified Acidified Dual

Seagrass Acidified Non-acidified Single Acidified Acidified Dual

Leaves

Terrestrial Non-acidified Non-acidified Dual Non-acidified Non-acidified Dual

Mangrove Non-acidified Non-acidified Dual Non-acidified Non-acidified Dual

Seagrass Acidified Non-acidified Single Non-acidified Non-acidified Dual

Seston Both1 Both Dual Both Both Dual

1 ‘Dual’ refers to the simultaneous testing of δ

13C and δ

15N isotopes on a single sample.

2 ‘Both’ refers to samples that undergo both acidification and non-acidification for δ

13C and δ

15N tests.

Generally, carbonate removal by acidification is uncomplicated, but there is risk in losing

volatile organic carbon (VOC) (Bisutti et al. 2004). There is a lack of consensus as to which

acidification technique, aqueous or vaporous method, is the most accurate and reproducible.

Brodie et al. (2011) warns against using fumigation or the vaporous method as it is too

unreliable and cannot be easily replicated. Furthermore, organic matter in soils may be

21

affected by acid fumes and release CO2 (Bisutti et al. 2004). However, this finding was not

supported by experiments carried out by Walthert et al. (2010). The loss of OC is also

associated with the aqueous method of removing carbonates, especially when filtration is

involved, acid wash is discarded and samples are heated (Walthert et al. 2010). A significant

decrease in OC may lead to a change in δ13

C signature (Walthert et al. 2010); yet, Komada et

al. (2008) did not find any correlation between %OC and δ13

C. Using effervescence as a

visual indicator of a completed reaction between acid solution and samples is extremely

difficult and subjective (Walthert et al. 2010).

Although Komada et al. (2008) recommended the vaporous method (acid fumigation), caution

has to be taken to prevent overexposure to acid fumes; 6-8 hours of fumigation is sufficient

(Walthert et al. 2010; Harris et al. 2001; Komada et al. 2008). In addition, vapor acidification

method should be avoided when samples contain high IC such as calcium carbonate (CaCO3

>50wt %) or dolomite (Hedges and Stern 1984; Walthert et al. 2010; Shubert and Nielsen

2000). Alternatively, Walthert et al. (2010) suggested a modified fumigation method that

includes an initial procedure of adding 1% HCl acid to high IC samples to prevent loss of

samples through bubbling.

Acid treatment was not used prior to δ15

N and %N analysis on solid samples, as acidifying

solid samples lead to artificial enrichment in N (e.g. Komada et al. 2008; Harris et al. 2001;

Walthert et al. 2010; Kennedy et al. 2005; Brodie et al. 2011) and decrease in N content in

other experiments (e.g. Walthert et al. 2010). Ryba and Burgess (2002) were not able to trace

the source of nitrogen despite various experiments to account for the additional nitrogen.

Hence, to avoid potential contamination and artificial enrichment of δ15

N values, tests for any

nitrogen related analysis was carried out on unacidified whole sediments (Table 2.2).

22

2.4.2 Sample materials and cleaning protocol

Sediments were dried at 105oC for at least 48 hours. Leaf samples of mangrove,

terrestrial plants, and seagrass leaves and detritus were dried at 60oC for at least 24 hours,

before they were ball-milled (Retsch Planetary Ball Mill PM400, Zirconium Oxide Jars and

Balls) and separated into size fractions with aluminum sieves to obtain the <63 microns

fraction. The grounded plant material and <63 microns sediments were further pulverized

manually to obtain a homogeneous mix for a more accurate chemical analysis.

To prevent contamination, gloves were worn when handling samples and equipment. All

glassware, including the dessicator, was rinsed with acetone and deionized (DI) water to

remove contaminants (Shubert and Nielsen 2000). Equipment used to manage samples were

thoroughly washed with DI water prior to each sample.

2.4.3 Acid fumigation method

Sample preparation for isotope analysis was adapted from Harris et al. (2001) and

recommended by UC Davis Stable Isotope Laboratory (UC Davis SIF), where the samples

were sent for analysis.

The dessicator was leached with 12M HCl acid for at least 4 hours prior to fumigation

(Brodie et al. 2011). For δ13

C analysis, 30-35 mg of sediment material was placed into silver

(Ag) foil boats and wetted with deionized (DI) water using a pipette to enhance acid

permeation (Brodie et al. 2011; Komada et al. 2008). Samples were placed in the dessicator

with 12M HCl for 6 hours before overnight oven drying at 60oC. Ag boats were encapsulated

with tin (Sn) foil boats to prevent leakage of samples if Ag boats had become too brittle from

the acid fumes. Approximately 40 mg of non-fumigated dried, grounded sample material was

placed into Sn foil boats and crimped close for δ15

N tests.

23

Non-fumigated mangrove and terrestrial plant leaf samples (2-3 mg) were placed into Sn foil

boats for dual isotope analysis. Separate sets of seagrass leaves and detritus samples were

prepared for standalone C and N isotopes tests. The same procedure for removing carbonates

was performed on seagrass samples that were sent for δ13

C testing. Approximately 6-7 mg of

non-fumigated dried seagrass material was placed into Sn foil boats and crimped close for

δ15

N analysis (2-3 mg of fumigated seagrass material for δ13

C analysis).

The water samples were filtered with pre-combusted 0.7 µm GF/F Whatman Glass Filter

Fibres (450oC, 4 h) to isolate seston. The filter papers containing residue seston were acid

fumigated, following the same procedures as sediment samples for single δ13

C and δ15

N

isotope analysis.

Soils, sediments and glass filter samples are analyzed for δ13

C and δ15

N isotopes using an

Elementar Vario EL Cube or Micro Cube elemental analyzer (Elementar Analysensysteme

GmbH, Hanau, Germany) interfaced to a PDZ Europa 20-20 isotope ratio mass spectrometer

(Sercon Ltd., Cheshire, UK) (UC Davis SIF, n.d.). Replicates of an isotopic Certified

Reference Material (CRM; B2151, Cert no. 162517, δ13

C = -26.27 ± 0.15‰, δ15

N = 4.42 ±

0.29‰) were included in the samples that were sent for isotope analysis to check for

reproducibility of analysis during the period of measurement across a range of sample masses

(mean δ13

C: -26.53 ± 0.12‰ and δ15

N: 4.93 ± 0.27‰ , with n=25). The standard deviations

(SD) lie within the SD range of those given by the standard, and the long-term SD of 0.2‰

for δ13

C and 0.3‰ for δ15

N (UC Davis SIF, n.d.). Furthermore, during isotope analysis,

samples were interspersed with several replicates of at least two different laboratory standards

that are compositionally similar to the type of samples to determine accuracy of absolute

values. These standards have been calibrated against NIST Standard Reference Materials

(IAEA-N1, IAEA-N2, IAEA-N3, USGS-40, and USGS-41). Triplicates of each leaf and

seagrass detritus sample were tested. The absolute difference between repeated

determinations on the same sample was usually less than 0.2‰ for δ13

C and 1.0‰ for δ15

N.

24

2.4.4 Acid wash method

Samples were sent to the Boise State University Stable Isotope Laboratory for

isotopic analysis using the acid wash method to remove IC. Sediment samples were acid

washed with 10% HCl acid. One hundred mg of each sample was soaked in 30 ml of HCl acid

for 24 hours prior to decanting and a second 24-hour HCl acid soak. Samples were then rinsed

four times with 30 ml deionized water and dried in an oven at 70oC. Samples were re-

powdered, and 15-40 mg of sample (depending on organic material concentration) were

crimped in Sn capsules and analyzed on a Thermo DeltaV Isotope Ratio Mass Spectrometer

coupled with a Costech EA 4010. Acid washed sediments were tested for both δ13

C and

δ15

N. Plant samples were not acid treated. Seston samples were put through the same acid

fume treatment described before. Data were standardized using two IAEA reference materials

for both isotopes. To assess reference material correction consistency and instrument

function, each sediment run included up to six internal glycine standards (δ13

C: -43.25 ±

0.13‰; δ15

N: 3.78‰ ± 0.11‰) and three Montana Soil standards, and peach leaf standards

were used for the plant runs.

25

2.5 Mixing polygon diagrams

Prior to the use of mixing models, a mixing diagram can be constructed to assess the

possible sources applicable for use in the mixing model and the likely range and distribution

of provenance from each source to the mixture (Phillips and Gregg 2003). This preliminary

method is useful when no unique solution is available due to an excess of sources (Section

2.6.2).

Figure 2.2. Mixing polygon bounded by source isotope signatures. Mixtures lying within the mixing

polygon have a set of source proportions.

A mixing polygon is constructed by connecting previously known isotopic signatures or the

median (50th percentile) isotopic values of each source group (Figure 2.2). A basic principle

of reading a mixing polygon is that mixture samples lying within the region have contributing

material from sources which geometrically bound the polygon (Figure 2.2). There are a few

patterns and assumptions involved when using the standard linear mixing model and mixing

polygon to analyze source provenance (Phillips and Gregg 2003):

26

A) When a mixture is outside a mixing

polygon bounded by all sources, no

possible source proportion combination is

possible (Figure 2.3A).

B) When a source is outside of a mixing

polygon bounded by all other sources, it

must contribute (cannot be 0) to a mixture,

if the mixture is also outside the mixing

polygon (Figure 2.3B).

C) When a source is inside of a mixing

polygon bounded by all other sources, it

need not contribute (may be 0), if the

mixture is also inside the mixing polygon

(Figure 2.3C).

D) When a mixture is near the edge the

mixing polygon, it has well constrained

ranges of solution, showing a higher

proportion of materials from sources that

form that boundary (Figure 2.3D).

E) When a mixture is near the center of the mixing polygon (Figure 2.3E), or if the mixing

polygon is small and compact with small differences between sources (Figure 2.3F), the

spread and range of possible solutions is wider. In the event that too many sources have been

identified, mixing diagrams allow one to narrow down the number of sources to only the

relevant and practical sources to be used for the mixing model.

Geometric procedures have been successfully used to quantify proportions of three food

sources to a diet using two isotope tracers (Phillip 2001 references herein). This method

utilises Euclidean distances for line segments between mixture and sources to compute source

Figure 2.3. Patterns and assumptions for proportion

of sources (a, b, c, d, e) to mixture (M) using the

standard linear mixing model, shown in a mixing

polygon. (Adapted from Phillips and Gregg 2003).

27

contributions. Although the geometric diagrams are intuitive tools, based on the logical

argument that mixtures in closer proximity to sources would reflect a higher proportion of

that source, this method may not provide correct solutions to the three end-member sources

mixing situation (Phillips 2001). The Euclidean distance based method either overestimates or

underestimates sources; therefore with such flaws, the model should not be used to apportion

mixtures (Phillips 2001). To address this problem, a linear mixing model can be created based

on isotopic mass balance (Section 2.6.1).

2.6 Mixing models

The usefulness of each type of source tracer (organic matter as leaf material versus

organic matter adsorbed on sediments) was assessed using mixing polygons, formed by the

source isotope signatures. The criterion was that seagrass sediment samples should lie within

the boundary of the mixing polygon. Stable isotope mixing models were created to estimate

proportion of sources contributing to a mixture. Although more commonly used for ecological

and biological studies involving reconstructions of animal diets, its application has expanded

across disciplines to include sediment source tracing, pollution source identification and water

source tracing (Phillips et al. 2005; Phillips 2012).

2.6.1 Basic mixing models

Mixing models are made up of mathematical equations, based on multivariate mixing

model using isotopic mass balance (Davis and Fox 2009; Parnell et al. 2010; Guzman et al.

2013):

where Mi is the concentration of tracer i in the mixture sample, xis is the concentration of

tracer i in source s, and Ps is the relative contribution of source s. The mixing model assumes

(2)

(3)

28

that all the relative proportions of source contributors would sum to 1 (Equation 3). In the

simplest scenario, one tracer (e.g. δ13

C) is used to distinguish between two source

contributions to the mixture (δ13

Cmix) with a constraint that sums the contributing proportions

(Mx) to 1:

(4)

(5)

Mathematically, a system with two equations (Equations 4 and 5) and two unknowns (M1 and

M2) provide a unique solution. The two equations can be solved algebraically by rearranging

and substitution. This model assumes that the isotope value of the mixture is derived solely

from the two sources, and has to fall between the two end-member values to be meaningfully

explained as a mixture of them (Phillip 2012). If the mixture falls outside this range, the

model can be mathematically solved, but will produce a negative proportion for one source,

which is not possible (Phillip 2012). As such, another constraint to the mixing model limits

all source contributions to positive values between 0 and 1.

In tracer studies, a second tracer is usually used to more accurately differentiate between

source groups. With two tracers, the mixing model can handle up to three source groups, by

solving the following equations via a matrix framework (Phillips et al. 2005; Phillips 2012):

(6)

(7)

(8)

Here, the system has three equations (Equations 6 to 8) and three unknowns (M1, M2, M3)

which also provides a unique solution. Likewise, similar assumptions are enforced in the two

isotope tracer with three sources mixing model. The mixture isotope values (δ13

Cmix and

δ15

Nmix) should lie within the mixing space defined by the sources (e.g. Figure 2.2) (Phillips

2012). In both scenarios, the model implicitly assumes that the partitioning of sources is the

same for both tracers (i.e. δ13

C and δ15

N) (Phillips 2001; 2012).

29

2.6.2 Excess number of sources

The mixing model is functional with a determined solution for n isotope tracers with

n+1 sources (Phillips et al. 2005; Phillips and Gregg 2003). The lack of multiple distinctive

tracers in a relative source contribution study restricts the number of sources to two or three

sources. The small number of sources is sufficient in diet reconstruction studies as animals

usually have less than five major sources of food. However, in reality, natural systems are

inherently complex, consisting of a large number of (inexhaustible) sources (Phillips and

Gregg 2003). The usefulness of the mixing model (Equation 1 and 2) diminishes when the

number of sources exceeds n+1, resulting in a mathematically undetermined model where

there are more unknowns than equations (Phillips et al. 2005). The over-parameterized model

and over-determined matrix can still be solved iteratively while maintaining the same

assumptions as before. However, instead of a definite solution, a range of feasible solutions

will be produced via an optimization procedure based on the minimization of an objective

function (Phillips and Gregg 2003; Phillips et al. 2005; Mukundan et al. 2012; Guzman et al.

2013). Modellers that utilize this frequentist approach usually minimize the residual sum of

squares, which are the differences between recorded and targeted values (Walling 2005;

Davis and Fox 2009; Mukundan et al. 2012).

2.6.3 Evaluation of commonly used stable isotope models

The IsoSource model by Phillips and Gregg (2003) uses iterative procedures to

determine the upper and lower bounds for ranges of feasible source solutions. This model

does not involve optimal minimization of any objective function, but operates on the notion

that predicted mixture signatures should fall within a user specified tolerance range (~ ±0.1-

0.2%) of the targeted sample mixture (Phillips and Gregg 2003). Although the IsoSource

model does not allow for the incorporation of variation and uncertainties within sediment

fingerprinting studies, as described in Small et al. (2002), extensions such as IsoError and

concentration-weighted IsoConc, permit users to adopt source variation and report standard

errors and confidence intervals for source proportion estimates (Phillips and Gregg 2001;

30

Phillips and Koch 2002; Parnell et al. 2010). Results from IsoSource model do not represent

probabilistic distributions of solutions; the model provides a range of values which are

plausible, based on the geometry of the model system (Parnell et al. 2010).

Foster and Lees (2000) stressed the importance of ensuring that natural source variability is

not overlooked, as it often contributes to high uncertainty in modelling outcomes. However,

the mixing models created by Phillips and Gregg (2001: IsoError; and 2003: IsoSource) are

either constrained by the number of sources, or do not incorporate variation as part of the

modelling process. To assess source group and analytical uncertainties in sediment

fingerprinting models, Small et al. (2002) designed a simple but robust Bayesian Markov

Chain Monte Carlo (MCMC) based model. Improvements to the Bayesian statistical

framework allows for uncertainties to be incorporated by accounting for variability and prior

information. An example is the SIAR (Stable Isotope Analysis in R) model formulated by

Parnell et al. (2010) which uses MCMC algorithms and includes an overall residual error term.

It is frequently used for various types of source partitioning. The SIAR model, which is used

in this study, addresses some recurring issues in mixing models. Unlike the IsoSource model,

SIAR can cope with multiple sources, uncertainty, and variability, which includes external

sources of variation that is not related to isotopic uncertainty (Parnell et al. 2010). The latter is

made feasible by the Bayesian method, which provides users the flexibility of integrating

external (prior) information of potential source distribution patterns into the mixing model

(Parnell et al. 2010; Phillips 2012).

Due to its flexibility and functionality, the SIAR mixing model was chosen to perform the

mixing analysis and partitioning of the seagrass sediment sources. Default values were used

for concentration dependence (concdep=0) in the siarmcmcdirichletv4 command model for

multiple sediment mixture data points. Trophic enrichment factors (TEF) and prior

information were not utilized. Number of iterations was set at 1000000 with 100000 'burnin'

(the number of initial iterations to be discarded).

31

3. Isotope Results

3.1 Stable isotope signatures

3.1.1 Isotope values for dead and fresh mangrove leaves

The Mann-Whitney U test on dead versus fresh mangrove leaf samples showed

significant difference (p<0.05, 2-tailed) for δ13

C, δ15

N and CN ratio (Table 3.1). However, the

non-parametric Kruskal-Wallis test for differences between organic matter source groups

showed that mangrove leaf isotope signatures are collectively significantly different (p<0.05)

from all other groups of organic matter. Hence, for the purpose for this study, dead and fresh

mangrove leaves were combined as a group to minimize the number of source groups.

3.1.2 Organic matter - leaf material

Mean organic matter (leaf material) δ13

C from the two laboratories were similar

(Kruskal Wallis, p<0.05). The leaf samples were not acidified prior to stable isotope analysis.

However, one set of seagrass detritus samples was acid fumigated (weak acidification) to

remove particulate inorganic carbon attached to its leaves (Section 2.4.1). The acid fuming

inorganic carbon removal process for seagrass detritus resulted in a more negative value of

δ13

C (-12.05‰) than those not undergoing carbonate removal (-10.65‰) (Table 3.1). This

was unexpected as the removal of IC should result in a more positive δ13

C value.

Table 3.1. Isotopic signatures of organic matter (leaf material). Leaf material samples did not undergo

weak acidification, with the exception of one set of seagrass detritus (*). Weak acidification for seagrass detritus

ONLY (median ± 1 MAD*)

(triplicates of each sample)

No acidification on all samples

(median ± 1 MAD)

(single tests with a few duplicates)

δ13C (‰) δ15N (‰) δ13C (‰) δ15N (‰)

Organic

matter

(Leaf

material)

Terrestrial

leaves

-29.70 ± 1.59 (n=14) 0.33 ± 1.88 (n=14) -30.61 ± 1.04 (n=14) 1.02 ± 1.18 (n=14)

Mangrove

leaves

-29.35 ± 0.68 (n=12) 2.78 ± 0.65 (n=12) -30.20 ± 1.03 (n=12) 2.94 ± 0.57 (n=12)

Seagrass

detritus

-12.05 ± 0.98 (n=12)** 3.10 ± 0.23 (n=12)** -10.65 ± 1.10 (n=12) 3.97 ± 0.13 (n=12)

Seston -23.52 ± 0.38 (n=6)

6.21 ± 0.22 (n=6) -23.76 ± 0.28 (n=10) 4.76 ± 0.28 (n=10)

* Median Absolute Deviation

**Seagrass detritus were acid fumed (weak acidification process) prior to analysis.

32

3.1.3 Organic matter - adsorbed on sediment samples

Median sediment δ13

C values for sample groups were similar between the two

acidification processes, except for coral sediments and seagrass sediments (Table 3.2).

Slightly higher δ13

C ‰ values (0.24-0.84‰) were recorded for terrestrial, mangrove and

seston samples that underwent the weak acidification (acid fumigation) process to remove

carbonates (Table 3.2). Such trends are expected because less inorganic carbon would be

removed for the weak acidification method, leaving a higher δ13

C ‰ value. The disparity in

the results between both acidification processes was more clearly observed in the δ13

C coral

sediments. The removal of almost all inorganics by the strong acidification (acid washing)

process resulted in a lower isotopic value (-22.00‰) than weak acidification (acid

fumigation) procedure (-9.31‰), which retained part of the natural signature of high

carbonate sediments. This explanation could be extended to the results in the seagrass

sediments which may have elevated concentration of inorganics that were not completely

removed (-19.09‰). The high Median Absolute Deviation (MAD) (± 5.42‰) observed in

seagrass sediments that underwent weak acidification reflected a wider spread of δ13

C ‰

values, possibly because of a larger contribution of high-carbonate content material, such as

coral sediments in some seagrass sediment samples, that can skew the δ13

C results.

Table 3.2. Isotopic signatures of organic matter adsorbed on sediment samples. Sediment samples were

acidified following two different acidification procedures – weak (acid fume) and strong (acid wash). Weak acidification (median ± 1 MAD) Strong acidification ( median ± 1 MAD)

δ13C (‰) δ15N (‰) δ13C (‰) δ15N (‰)

Organic

Matter

(Adsorbed

on

sediments)

Terrestrial -27.33 ± 0.90 (n=32) 5.13 ± 1.47 (n=32) -28.17 ± 0.60 (n=25) 3.99 ± 0.92 (n=25)

Mangrove -26.59 ± 0.39 (n=31) 3.47 ± 0.31 (n=31) -27.36 ± 0.40 (n=23) 2.76 ± 0.36 (n=23)

Coral -9.31 ± 2.18 (n=15) 5.52 ± 0.55 (n=15) -22.00 ± 1.64 (n=6) 4.17 ± 0.74 (n=6)

Seston -23.52 ± 0.38 (n=6) 6.21 ± 0.22 (n=6) -23.76 ± 0.28 (n=10) 4.76 ± 0.28 (n=10)

Seagrass -19.09 ± 4.84 (n=27) 4.02 ± 0.35 (n=27) -23.47 ± 1.17 (n=20) 3.13 ± 0.26 (n=20)

Despite the incomplete removal of carbonates, the weak acidification method tended to

'normalize' the carbonates in samples with a range of inorganic carbon concentration. In

essence, it likely removed a near-equal mass of inorganics across all samples, leaving

originally low-carbonate content samples with close to 0% inorganic C. However, the weak

33

acidification method retained any high isotopic content of the sediment samples, therefore

reflecting the natural isotopic signature of sediments from the respective environments.

3.2 Mixing polygon diagrams

3.2.1 Organic matter - leaf material as tracers

The δ13

C and δ15

N signatures of mangrove and terrestrial leaves were not

significantly different (Mann-Whitney U test, p<0.05). As such, the two organic matter

sources were combined to form one end-member source group. Following the strong

acidification procedure on the seagrass sediment, seston and seagrass detritus samples,

roughly two-thirds of the seagrass sediment samples lay outside the mixing polygon (Figure

3.1). Thus, no solution was possible for those samples from the identified sources. There is

likely another unidentified source that contributes to these samples.

Following a weak acidification procedure on the seagrass sediment, seston and detritus

samples, seagrass sediment samples showed a high contribution of seston material and

seagrass detritus. Most of the samples lay in close proximity to the bounding triangle, formed

by the terrestrial/mangrove, seston material and seagrass detritus end members (Figure 3.2).

Thus, sediment compositions are expected to be constrained to the two sources, and have a

smaller range of possible solutions. About one-third of the samples lie outside the boundary

of the mixing polygon (Figure 3.2).

34

Figure 3.1. δ

13C and δ

15N results for leaf material using strong acidification on seagrass sediments,

seston and seagrass detritus. About two-thirds of the seagrass sediment samples lay outside the

bounding polygon. Grouping the terrestrial and mangrove leaves to form a combined source group did

NOT provide a suitable model to determine source contributions.

Figure 3.2. δ

13C and δ

15N results for leaf material using weak acidification on seagrass sediments,

seston and seagrass detritus. One-third of the seagrass sediment samples lay outside the bounding

polygon. Grouping the terrestrial and mangrove leaves to form a combined end-member group did

NOT provide a suitable model to determine source contributions.

35

3.2.2 Evaluation of acidification methods on organic matter - adsorbed on sediments

Both acidification methods applied to organic matter adsorbed on sediments showed

that individual source groups cluster together, and have distinctly different δ13

C and δ15

N

isotopic signatures (Mann-Whitney U test, p<0.05). Terrestrial and mangrove sediment

sources are differentiated by their median δ15

N signatures (Figure 3.3 and 3.4). The median

δ13

C signature of coral source sediments after strong acidification lay inside the mixing

polygon (Figure 3.3), implying that it need not contribute to the mixture samples (Phillips and

Gregg 2003) (Figure 2.3C), adding uncertainty in the representativeness of sediment

proportion as coral source sediments. On the contrary, the δ13

C signature of coral sediments

after weak acidification, was markedly different, and is considered a sediment source that

forms part of the bounding polygon (Table 3.2; Figure 3.4). Samples were taken from areas in

close proximity to coral reefs that flank the sides of the bay, hence coral fragments and

sediments clearly contributes to the sediment pool in the bay (Figure 3.5). Therefore, weak

acidification should be used, such that coral sediments are included as an end member

sediment source. Furthermore, nearly two-thirds of the seagrass mixture samples that

underwent strong acidification lay outside the mixing polygon (Figure 3.3), rendering this

carbonate removal method less effective.

36

Figure 3.3. δ

13C and δ

15N results for organic matter adsorbed on sediments, using strong acidification

process, bounded by four sources. Coral sediment source signature lies within the mixing polygon,

implying that it need not contribute to mixtures, adding uncertainty to results. Two-thirds of the

seagrass sediment mixture samples lie outside the mixing polygon.

Figure 3.4. δ

13C and δ

15N results for organic matter adsorbed on sediments using weak acidification

process. Almost all mixture samples lie within the mixing polygon and form a good model.

37

Strong acidification, which removed most inorganic carbon from samples, skewed the δ13

C

readings towards a more negative value that is reflective of organic matter. The strong

acidification procedure led to the close clustering of seagrass sediment mixture samples

between -22.0‰ to -25.0‰ (Figure 3.3). In contrast, weak acidification preserved δ13

C

signatures of samples with naturally high carbonate concentration, resulting in an apparent

clustering of the seagrass sediment mixtures into two groups (-25.0‰ to -18.0‰, and -15.0‰

to -8.0‰) (Table 3.2; Figure 3.4). Seagrass sediment samples with more positive δ13

C

signatures appears to have influence from carbonate-rich sediments and are therefore likely to

have larger contributions from coral sediments and seagrass detritus (Figure 3.4). To illustrate

this influence, δ13

C values of seagrass sediment were plotted against seagrass detritus with a

1:1 line to show expected δ13

C isotope signatures of sediment when detritus contributes 100%

(Figure 3.5A). Samples 102, 108, 109 and 110 appeared to have a lower δ13

C signature than

seagrass detritus, supporting the possibility of high-carbonate coral sediment influence due to

their close proximity to reef areas (Figure 3.5B). Samples 122 and 123, which have more

negative δ13

C signatures, were located further from reef areas (Figure 3.5B), and therefore

have less influence from these source areas.

Figure 3.5. A) δ

13C values of acid fumed seagrass sediment plot against seagrass detritus with a 1:1 to

show hypothetical situation where seagrass detritus contributes material to seagrass sediments to make

up the same δ13

C signature. Dotted isolines represent an offset of 5‰ and 10‰ (n=11). B) Seagrass

sediment sampling locations. Samples 102, 108, 109, 110 (red) are located nearest to the coral areas,

and have more positive δ13

C. Samples 122 and 123 (blue) are located further away from the corals, and

have more negative δ13

C values.

(a) (b)

Coral

38

3.3 Determining an appropriate model

The mixing polygon diagrams show that the use of organic matter adsorbed on

sediment surfaces serves as a better tracer than leaf material alone, which represents organic

matter. Organic matter is composed of organic compounds that derive from living organisms

– both plant and animal (Bisutti et al. 2004). By solely obtaining the isotopic signature of leaf

organic matter, the contribution of organic material from animals is disregarded. This direct

method of determining the isotopic signature of organic matter from its different sources is

incongruent with the composition of organic matter. Deposited seagrass sediment samples are

a combination of different sources of organic matter. As organic matter adsorbed on sediment

surfaces represents a mix of both plant and animal matter, it is therefore a more accurate

representation of the sediment sources, and should be used for sediment source tracing of

seagrass sediments.

Figure 3.6. Expansion of mixing polygon (solid line) made by modifying δ

13C and δ

15N median values

of coral and seagrass detritus source groups. New isotope values, of coral and seagrass detritus end-

members for the five samples outside the original polygon, are actual values of samples from each

group. Original isotope signatures for sources were used for samples which lay within the original

mixing polygon (dotted line).

39

The weak acidification (acid fumigation) method is most applicable, as its results reflect

natural signatures of each environment, as seen visually (Figure 3.4). To account for

variances in the mixture data that lie outside of the mixing polygon (i.e. five samples - 102,

109, 110, 111, 124), the δ13

C and δ15

N isotope source values of seagrass detritus and coral

source sediments would be altered (Figure 3.6, solid line). Both new isotope values of corals

and seagrass detritus end-members are actual values of a sample from each respective group.

The new isotope values were used subsequently in the mixing model for these five sample

mixtures. Original isotope signatures for sediment sources were used for seagrass sediment

samples which lay within the original mixing polygon (Figure 3.6, dotted line).

Although Walthert et al. (2010) states that an alternative procedure should be undertaken to

remove all inorganic carbon in sediments that are high in inorganic carbon (Section 2.4.1), the

selection of an appropriate procedure should be based on the experimental objectives. In this

study, a partial removal of inorganics was appropriate as it retained the natural signature of

the source type. Here, the remnants of inorganic carbon isotope signature were accepted as

legitimate and justifiable. Therefore, the data for the mixing model that was utilized is made

up of median isotopic signatures δ13

C and δ15

N from organic matter adsorbed on sediments

which had undergone weak acidification (Figure 3.5).

40

4. Discussion

The sediment distribution patterns substantiate a strong catchment-coast linkage,

especially through the river system. This finding reinforces the need for a whole-systems

approach and the consideration of the role that landscape connectivity contributes towards

curbing environmental problems such as sediment loading on coastal ecosystems that

originate at the catchment scale.

4.1 Spatial distribution of sediments and relative proportions of main sources

The Ordinary Kriging interpolation (ArcGIS v10.1) of deposited sediment

composition in the seagrass bay showed that 50-60% of the composition of sediments in close

proximity to river mouths was terrestrial- and mangrove-derived (Figure 4.1A and B). This

finding implies that material from distant sources inland can be transported throughout the

catchment via transport pathways (e.g. rivers) that connect catchment areas to the coast. The

high composition of mangrove-derived sediments at the inner bay could be related to

proximity of mangrove forests.

Seagrass detritus accumulates at an isolated location in the right flank of the bay, diagonally

across the main river, with percentage composition as high as 40-45% (Figure 4.1C). Areas

with high percentage of seagrass-derived sediments coincide with the presence of healthy

seagrass beds that are also located on the right of the bay (Figure 4.2). In contrast, very low

fractions (6-10%) of seagrass detritus sediments were found at the mouth of the main river

(Figure 4.1C). The high flow velocity of water from the river mouth could prevent the

accumulation of seagrass detritus, and contribute to a relatively higher percentage

composition from terrestrial and mangrove sediments (Figure 4.1A-C). Distribution of coral-

derived sediments exhibits a similar pattern as seagrass detritus. It appears to accumulate

away from the coast slightly further away from the coral reefs (Figure 4.1D).

41

(A)

Terrestrial (B)

Mangrove

(C)

Seagrass

Detritus

(D)

Coral

(E)

Seston

Figure 4.1. Spatial interpolation (kriging) of sediment composition for

each source.

About 50-60% of sediments at the river mouth consist of terrestrial-

and mangrove-derived sediment sources. Seagrass detritus and coral-

derived sediments reflect similar distributions – low percentage

composition (6.2-10%) at river mouth; high percentage composition at

the right side of the bay. Seagrass detritus sediments accumulates (40-

45% composition) at an isolated location at the right side of the bay.

Areas with high percentage composition of coral-derived sediments are

located near coral areas (in yellow). Seston-derived has low percentage

composition (6.2-25%), evenly distribution throughout the whole bay,

with a slightly higher concentration near the river mouth.

42

Figure 4.2. Areas with high percentage of seagrass detritus-derived sediments (Left) coincide with

presence of healthy seagrass beds (Right - red indicates presence of seagrass) (Data courtesy of

Jachowski, n.d.). Low distribution on the left side of the bay could be attributed to the high velocity

flow from river water that prevents accumulation, and contributes a relatively higher percentage

composition from terrestrial and mangrove sediments.

On the contrary, seston-derived sediments present a rather ubiquitous distribution, ranging

from 5.7-25%, with a slightly higher percentage at river mouths (Figure 4.1E). The river may

therefore act as a principal conduit where tides exchange ocean water in and out of the

mangroves. Therefore, seston may collect around the river mouth. Studies have shown strong

isotopic evidence for terrestrial sustenance of zooplankton in aquatic systems (lakes, streams,

rivers) via allochthonous organic material originating from the terrestrial watershed (e.g. Cole

et al. 2010). However, the seston composition of deposited sediment in Yao Yai seagrass bay

is inconclusive as the seston source isotope signature was obtained from collecting water

samples from the entrance of the bay away from the river mouth, instead of from a

terrestrially-derived seston source (i.e. seston samples taken upstream of rivers). The fairly

even distribution of seston-derived sediments across the bay could be attributed to a residual

effect from the distribution of other sources. Terrestrial and mangrove distribution is similar,

as with coral and seagrass detritus distribution; both groups of sources seem to display

opposite distribution patterns, leaving seston with a more uniform distribution across the bay

(Figure 4.1E).

43

4.2 Catchment- to-coast linkages

The connectivity of a catchment is an important concept which physically links water

flow paths and sediment transport pathways across spatial scales. Linkages between inland

catchments and coastal zones can be categorized into interfaces and connections (Alvarez-

Romero et al. 2011) (Figure 4.3). These linkages also affect catchment connectivity.

Interfaces are usually narrow or broad zones where at least two different ecosystems

assimilate processes (e.g. ecotones); whereas connections are linkages between distant

ecosystems that are not adjacent (e.g. well-defined and constrained pathways such as rivers,

or diffused routes such as movement of organisms) (Alvarez-Romero et al. 2011). Bracken

and Croke (2007) identified three types of connectivity related to hydrology and

geomorphology: hydrological connectivity (Section 4.2.1), landscape connectivity (Section

4.2.2) and sedimentological connectivity. These types of connectivity inter-influence each

other to create linkages in a catchment (Figure 4.3). The first two types are more relevant to

this study and will be discussed with reference to the results.

Figure 4.3. Ecosystem linkages framework showing the interlinked relationships between all

components. Arrows indicated direction of influence. (Based on concepts from Mitchell et al. (2013)

and Alvarez-Romero et al. (2011))

4.2.1 Hydrologic connectivity

Distribution patterns of terrestrial sediments indicate a relationship between major

transport pathways and the transport and deposition of such sediments (Figure 4.1A). From a

44

connectivity perspective, this implies that mangroves may not be effective sinks or buffers for

seagrass ecosystems when major flow pathways course through mangrove areas. Buffers are

physical features that disrupt lateral linkages within catchments (Fryirs 2012). On the

contrary, flow pathways (e.g. rivers) provide a direct link between inland terrestrial catchment

areas and coastal areas, thus mediating the connection between the two systems (Alvarez-

Romero et al. 2011; Salomons 2005) and providing an 'accelerated' transport of material to

the coast (Figure 4.4 (1)). Terrestrial areas can also be indirectly connected to the bay via

surface runoff processes across the mangroves to seagrasses (Figure 4.4 (2)).

Figure 4.4. Sediment transport pathways: 1) Accelerated transport from terrestrial areas via river; 2)

Indirect transport from terrestrial areas to mangrove before reaching the seagrass bay; and 3) Direct

transport into the bay between terrestrial and seagrass bay interfaces.

The network of transport pathways within catchments is comprised of rainfall-generated

runoff paths and river channels. An increase in the network density of transport pathways

should increase catchment connectivity, thereby decreasing the residence time that sediments

are stored. As a result, a higher load of material is transported in a shorter time. There are

many aspects influencing the hydrological network of a catchment: climate, runoff potential,

lateral buffering, delivery pathway and landscape position (Bracken and Croke 2007) (Figure

4.5). Climatic variables, especially rainfall (duration and intensity) and soil moisture, have a

45

major impact on the formation of surface runoff (Bracken and Croke 2007). Surface runoff

enhances hydrological connectivity, overall catchment connectivity, and subsequently,

sediment transport. One key factor contributing to surface runoff and hydrological

connectivity is landscape connectivity.

Figure 4.5. Five major components of catchment hydrological connectivity which are interlinked with

various factors within each component that influences others. (Source: Bracken and Croke 2007).

4.2.2 Landscape connectivity

In addition to hydrologic connectivity, ecosystem or landscape connectivity also

facilitates both biotic (movement of organisms) and abiotic (movement of water, nutrients and

sediments) connectivity within a catchment (Mitchell et al. 2013; Alvarez-Romero et al.

2011). Landscape connectivity comprises of two major landscape components which define

the structural connectivity of the catchment through its spatial structure and patterns:

landscape composition (area of each type of land cover/use) and landscape configuration (the

way these land cover/use are spatially arranged) (Mitchell et al. 2013) (Figure 4.3). Changes

in landscape composition and assemblage will evidently alter water and sediment delivery to

coast. For example, deforested patches of land interspersed with densely forested areas will

contribute less sediments and will not have a continuous flow path compared with a

homogeneously deforested land. However, the buffering efficiency of densely forested areas

depends on the degree of hydrologic connectivity. The spatial configuration and landscape

composition of different types of land cover may thus simultaneously contribute to the

intensification or reduction of hydrological connectivity.

46

The response and movement of organisms and materials (functional connectivity) to the

landscape arrangement (structural connectivity) may also shape landscape connectivity

(Mitchell et al. 2013) (Figure 4.3). However, the focus of this discussion will be placed on

landscape structural connectivity and its influence on ecosystem linkages which defines

interface connectivity in catchment to coast linkages. Particularly, the dominance of

landscape composition and configuration that influences sediment delivery into transport

pathways will be evaluated for highland and lowland catchment regions.

In highland areas of a catchment, land cover/use composition management is an important

control of the amount of available erodible sediment. A deforested catchment would

contribute larger amounts of sediment than would a forested catchment (Douglas 1996). Any

changes in landscape composition are especially important in small and medium catchments

as response to these inland changes is translated to coastal change in a shorter time span

(Salomons et al. 2005; Phillips and Slattery 2006). Furthermore, the magnitude of change is

usually more apparent in smaller catchments; larger catchments have larger 'buffer capacities'

in relation to their catchment size (Salomons et al. 2005), allowing the effects to be more

dissipated, distributed and absorbed over larger areas.

In lowland catchment regions, the spatial configuration of land cover/use may be

comparatively more important than its landscape composition. In the highlands, hill-slopes

typically connect directly to rivers without floodplain areas (Bracken and Croke 2007). The

steeper slopes make it difficult for effective lateral buffering to be imposed in areas adjacent

to the channels. However, the effectiveness of vegetated buffer areas is dependent on many

environmental factors, such as hill-slope length and soil infiltration rate (e.g. Ziegler et al.

2007). On the contrary, large natural bodies of water, such as lakes, floodplains that occur at

lowland catchment regions, and man-made features such as reservoirs and dams act as buffers

or sinks that disrupt connectivity and the sediment delivery to the main channel and coast

(Fryirs 2012). The location of such 'sink' areas limit the amount and rate at which sediments

47

are transported to coastal areas. For example, extensive floodplain areas that are located along

the river banks in the lowlands act as important buffers that filter sediments in surface runoff

before reaching river channels (Bracken and Croke 2007; Fryirs 2012). Alternatively,

interface linkage and lateral connectivity between floodplains and rivers allow for deposition

of sediments during flood events with overbank flow, thus reducing the suspended sediment

load of a river (Lee et al. 2006). Therefore, the lateral structural connectivity of relationship

and processes between interfaces is an important linkage that can be leveraged upon to

alleviate sedimentation from anthropogenic disturbances (Lee et al. 2006).

4.3 Implications for the seagrass bay in Yao Yai

The landscape composition of Yao Yai can be separated into the highland areas along

the right side of the catchment where most of land cover remains as natural forests, and

lowland areas have been converted to agriculture (Figure 4.6). Paddy fields or wetland

agriculture are distinctively located in valleys, forming a relatively continuous flow path to

the main river channel (Figure 4.6). Paddies are waterlogged areas that are conducive for

sediment trapping (MRC 2010; e.g. Sukristiyonubowoa et al. 2010). Paddies in the study

catchment function as buffers that remove sediments from runoff by deposition. Therefore, an

imperative management action could be to ensure that the deposition or storage capacity of

paddies upstream or adjacent to river channels is conserved to maintain the function of this

lateral buffering. However, the landscape composition of Yao Yai island is heavily influenced

by coconut and rubber plantations land use, which are located at low-lying areas (Figure 4.6).

Furthermore, paddy and wetland areas are mostly surrounded by plantations, and thus,

possibly receive a large amount of eroded sediments from surface runoff (Figure 4.6).

Although the paddies may act as sediment sinks, the intensity of sediment erosion, as a result

from landscape configuration of the remaining catchment, may still negate the effectiveness

of this function. This is evident in the results that show a substantial proportion of sediment

contribution at the river mouth derived from terrestrial sources (Figure 4.1A).

48

Figure 4.6. Land cover/use configuration of inland catchment. Land configuration of paddies or

wetland agriculture are located along valleys, creating a flow path leading to the main channel.

Coconut and rubber plantations are the main land compositions that surround the paddies and low lying

areas of the catchment.

The high contribution of terrestrial sediments at the coast could be attributed to near linear

man-made features such as unpaved roads, vehicle tracks and drainage ditches. These features

contribute to high runoff erosion because of their high impermeability (Ziegler and

Giambelluca 1997; Sidle and Ziegler 2012). The linear features alter hydrologic connectivity

of the catchment, by functioning as delivery pathways for runoff and sediment transport

during storms, thus contributing to the extension of the channel network (Bracken and Croke

2007; Sidle et al. 2004; Ziegler and Giambelluca 1997). Thus, to alleviate sedimentation in

coastal zones, catchment land management policies must not neglect the concept of roads-

river connectivity, but instead target the reduction of this connectivity through proper road

49

network planning by avoiding road construction on steep slopes and restricting road lengths

(Sidle et al. 2004; Sidle and Ziegler 2012).

The discussions on sediment management imply that it is a long-term process that extends

across different spatial scales (Salomon 2005). Both the type of land cover/use and spatial

arrangement of these land areas affect the amount of sediments entering the river before being

transported to the coast. Runoff generation and connectivity determines the way sediments are

transported across the land into the river. As such, catchment landscape connectivity and

runoff response should be collectively integrated to ensure effective sediment management.

4.4 SIAR and spatial mixing models

Despite the effectiveness of the SIAR model in generating data for the spatial

distribution of sediments in the bay, Parnell et al. (2010) posits a few caveats, which may also

be applicable to other mixing models. Those caveats relate to its use for sediment tracing:

1) As with other mixing models (e.g. IsoSource), SIAR model outputs represent

probable solutions, not exact values. The percentage ranges of sediment compositions

are not absolute and only provide a relative fraction to other sources.

2) SIAR assumes that source variability is normally distributed.

3) SIAR would always attempt to fit a model, even when sources lie outside of the

isotopic mixing polygon. Therefore, it was crucial to ensure that mixtures lie within

the isotopic mixing polygon prior to using the SIAR model.

Mathematical mixing models only consider the similarity of isotope signatures, and

sometimes may not always make logical representations according to the real world. For

example, spatial distance between sediment mixture and sources - distal versus proximate

sources - are not explicitly accounted for in the model. Miller et al. (2013) reminds modellers

about the important assumption that sediment mixing models are focused on the ultimate

source of sediments and not its proximal ones. This assumption is particularly important when

50

a catchment is large enough to contain many sediment storage areas where sediments are

deposited for long periods of time allowing for chemical changes in its isotopic signature

before being reworked, transported and deposited at the ultimate sink downstream. At present,

specialised distance weighted mixing models are not available. Although, the SIAR model can

be used to include prior knowledge about spatial information and relationship between

sources (e.g. Palmer and Douglas 2008), Davis and Fox (2009) cautions against using prior

knowledge haphazardly as it increases the potential for erroneous model design when used

incorrectly, resulting in biasness in results. More stringent criteria should be created for the

specification of prior knowledge in Bayesian model parameters (Davis and Fox 2009).

51

5. Conclusion

5.1 Summary of thesis

This study has achieved its objective of tracing sediment sources and spatial

distribution of sediment compositions of Yao Yai seagrass bay using stable isotopes δ13

C and

δ15

N. Three conclusions can be drawn:

1) Isotopic signatures of sediments at the river mouths show that 50-60% of these

sediments constitute of terrestrial and mangrove sources. This implies that

hydrological pathways such as rivers are important transport pathways in

delivering material to coastal areas.

2) Stable isotopes δ13

C and δ15

N tests have shown that organic matter adsorbed on

loose sediments were a more appropriate tracer property than organic matter in

the form of plant leaves, as it is the combination of both plant and animal matter.

When using leaf material as tracers, approximately two-thirds of seagrass

sediment samples lay outside the mixing polygon indicating that there is another

possible unaccounted source that contributes to these samples. As organic matter

attached on deposited seagrass sediments derive from both plant and animal

origins, organic material adsorbed on sediments collected from sources areas are

a more accurate representation of the sediment sources.

3) The weak acidification method used to remove inorganics from sediments was

demonstrated to be more useful in differentiating the composition of deposited

sediments as it retains the natural signature of the source type.

The catchment-to-coast continuum is supported by ecosystem connectivity which is partly

driven by the physical linkages between landscapes. Both landscape composition and

configuration affect the hydrological and sedimentological connectivity of the catchment,

which in turn strengthen or weaken the linkages between interfaces across the catchment to

the coast. As such, downstream ecosystems often receive the accumulation of cascading

52

effects from upland catchments. Coastal ecosystems, such as seagrasses, are sensitive to slight

changes in water quality and are unable to recover quickly from prolonged damaging effects.

Massive die-offs result in the loss of important ecosystem services and blue carbon storage

areas.

The results from this thesis have accentuated the importance of adopting a wider catchment-

coastal system perspective when dealing with coastal sedimentation. In doing so, a systems

approach should be considered, since there will be delays and feedbacks within the individual

ecosystems that collectively form the catchment-to-coast system. These delays and feedbacks

may complicate catchment responses to management measures. Furthermore, it draws

attention to the need for accurate sediment budgeting (Section 5.3).

5.2 Applicability of findings to other catchments

The "source-to-sink" approach to managing coastal problems is necessary in small

and medium catchments and in island-dominated regions in the South Pacific where whole

islands constitute the whole catchment (Salomons et al. 2005). With the urbanization of

catchments, a multitude of undesirable impacts resulting from such changes in land cover

would alter hydrological and sediment regimes, translating adverse impacts to coastal areas.

Urbanization of island catchments may be driven by tourism as it is often thought as a

lucrative industry that provides income for many small island communities around the South

East Asian region. Although agriculture remains the major land use and livelihood for the

community on Yao Yai, the potential for tourism expansion is a possibility that could bring

about harmful environmental effects if careful planning is not established. Building

constructions and land conversions for tourism development are expected to exacerbate the

sedimentation problem in the seagrass bay. Tourism development could also fuel the need for

a better road network. However, poorly maintained road networks are an important and often

overlooked contributor to sedimentation in the catchment. These road pathways converge to

53

form runoff networks thus expanding the channel network coverage that contributes to the

main river channel.

Regardless of the type of catchment, policy makers must therefore implement suitable land

use management policies and schemes that address the coastal environmental problems that

these landscape connectivity and linkages have inadvertently created. For example, any

inhibition on the accumulation and deposition of sediment in floodplains and catchment

sediment sinks should be avoided, as sedimentation at the coastal zones could be alleviated

through these physical features that act as lateral buffers. In addition, proper road design and

planning should be implemented to minimize sediment erosion and transport from these

potential flow pathways. Therefore, this thesis has demonstrated that the mitigation of

sedimentation at coastal zones should not be restricted to adaptation response to the damaging

effects resulting from such chronic or acute events, as this does not address the root of the

problem. These coastal management perspectives are generally applicable to most coastal

catchments, especially those facing problems which originate inland.

5.3 Future research possibilities

To further verify the spatial distribution of each sediment source, additional tracer

tests could be carried out. In addition to the use of stable isotopes δ13

C and δ15

N, Compound

Specific Isotope Analysis (CSIA), which measures the δ13

C signatures of specific organic

compounds (e.g. resin and fatty acids) associated with organic matter adsorbed to the

sediment, could be utilised (e.g. Hancock and Revill, 2011; Gibbs 2008). These sophisticated

tests have been carried out in similar estuarine environments as that of Yao Yai bay.

However, more research has to be done on their usefulness for the identified types of

sediment sources in this thesis.

Next, sediment core samples could be taken within the bay, to construct a chronological log

of the change in sediment distribution across decades, to account for land cover changes

54

through time. Furthermore, sediment deposition rate could be obtained to determine how land

cover has affected the sediment regime of the catchment and sediment spatial distribution in

this bay. From here, the influence of anthropogenic alteration of the natural environment can

be assessed.

A sediment budget of Yao Yai bay may be calculated using passive collectors at the mouth of

the river (e.g. Philips et al. 2000). The partitioning of sediment sources during each storm

event may provide an understanding of how sediment transport or source contribution

changes with different storm characteristics. These three research possibilities improve the

reliability of the results with the addition of another tracer, enhance the understanding of

historical sediment contributions of the catchment and finally, reveal how present-day

dynamics of sedimentation changes with different hydrological responses to storm events.

55

References

Alvarez-Romero J.G., Pressey R.L., Ban, N.C., Vance-Borland, K., Willer, C., Klein, C.J. and

Gaines, S.D. (2011). Integrated land-sea conservation planning: The missing links.

Annual Review of Ecology, Evolution and Systematics 42: 381-409.

Bisutti, I., Hilke, I. and Raessler, M. (2004). Determination of total organic carbon – an

overview of current methods. Trends in Analytical Chemistry 23 (10-11):716-726.

Bouillon, S., Connolly, R.M. and Lee, S.Y. (2008). Organic matter exchange and cycling in

mangrove ecosystems: Recent insights from stable isotope studies. Journal of Sea

Research 59: 44-58.

Boutton T.W. (1991) Stable Carbon Isotope Ratios of Natural Materials: II. Atmospheric,

Terrestrial, Marine and Freshwater Environments. In: Colman D.C. and Fry B. (Eds.),

Carbon Isotope Techniques. Academic Press, San Diego, California.

Bracken, L.J. and Croke, J. (2007). The concept of hydrological connectivity and its

contribution to understanding runoff-dominated geomorphic systems. Hydrological

Processes 21: 1749-1763.

Brodie, C.R., Leng, M.J., Casford, J.S.L., Kendrick, C.P., Lloyd. J.M., Zong, Y. and Bird,

M.I. (2011). Evidence for bias in C and N concentrations and δ13

C composition of

terrestrial and aquatic organic materials due to pre-analysis acid preparation methods.

Chemical Geology 282: 67-83.

Burkholder, J.M., Tomasko, D.A. and Touchette, B.W. (2007). Seagrasses and

eutrophication. Journal of Experimental Marine Biology and Ecology 350: 46-72.

Cabaco, S., Santos, R. and Duarte, C.M. (2008). The impact of sediment burial and erosion on

seagrasses: A review. Estuarine, Coastal and Shelf Science 79: 354-366.

Chansang, H. (1984). Structure of a mangrove forest at Ko Yao Yai, Southern Thailand.

Proceedings of Asian Symposium on Mangrove Environment: Research and

Management. Pp. 86-105.

Chansang, H. and Poovachiranon, S. (1994). The distribution and species composition of

seagrass beds along the Andaman sea coast of Thailand. Phuket Marine Biology

Center Research Bulletin 59: 53-64.

Cloern, J.E., Canuel, E.A. and Harris, D. (2002). Stable carbon and nitrogen isotope

composition of aquatic and terrestrial plants of the San Francisco Bay estuarine

system. Limnology and Oceanography 47 (3): 713-729.

Cole, J.J., Carpenter, S.R., Kitchell, J., Pace, M.L., Solomon, C.T. and Weidel, B. (2010).

Strong evidence for terrestrial support of zooplankton in small lakes based on stable

isotopes of carbon, nitrogen and hydrogen. PNAS 108: 1957-1980.

Collins, A.L. and Walling, D.E. (2002). Selecting fingerprint properties for discriminating

potential suspended sediment sources in river basins. Journal of Hydrology 261: 218-

244.

56

Collins, A.L. and Walling, D.E. (2004). Documenting catchment suspended sources:

problesm, approaches and prospects. Progress in Physical Geography 28 (2): 159-

196.

Coplen, T.B. (1994). Reporting of Stable Hydrogen, Carbon, and Oxygen Isotopic

Abundances (Technical Report). Pure and Applied Chemistry 55 (2): 273-276.

Davis, C.M. and Fox, J.F. (2009) Sediment fingerprinting: Review of the method and future

improvements for allocating nonpoint source pollution. Journal of Environmental

Engineering 135 (7): 490-504.

Douglas, I. (1996). The impact of land-use changes, especially logging, shifting cultivation,

mining and urbanization on sediment yields in humid tropical Southeast Asia: a

review with special reference to Borneo. In: Walling, D.E. and Webb, B. (Eds.),

Erosion and Sediment Yield: Global and Regional Perspectives (Proceedings of the

Exeter Symposium, July 1996.) IAHS Publications no. 236.

Duarte, C.M. (2002). The future of seagrass meadows. Environmental Conservation 29 (2):

192-206.

Duarte, C.M., Marba, N. and Santos, R. (2004). What may cause loss of seagrasses? In:

Borum, J., Duarte, C.M., Krause-Jensen, D. and Greve, T.M. (Eds), European

seagrasses: an introduction to monitoring and management (pp.24-32). The

Monitoring and Managing of European Seagrasses (M&MS) project.

Duarte, C.M., Middelburg, J.J. and Caraco, N. (2005). Major role of marine vegetation on the

oceanic carbon cycle. Biogeosciences 2: 1-8.

Duarte, C.M., Dennison, W.C., Orth, R.J.W. and Carruthers, T.J.B. (2008). The charisma of

coastal ecosystems: Addressing the Imbalance. Estuaries and Coasts 31: 233-238.

Duarte, C.M., Kennedy, H., Marba, N. and Hendriks, I. (2011). Assessing the capacity of

seagrass meadows for carbon burial: Current limitations and future strategies. Ocean

and Coastal Management 83: 32-38.

Elliott, M. and Whitfield, A.K. (2011). Challenging paradigms in estuarine ecology and

management. Estuarine, Coastal and Shelf Science 94: 306-314.

Foster, I.D.L. and Lees, J.A. (2000). Tracers in geomorphology: theory and applications in

tracing fine particulate sediments. In: I.D.L. Foster (Ed.), Tracers in Geomorphology,

Wiley, Chichester, UK.

Fourqurean, J.W., Duarte, C.M., Kennedy, H., Marba, N., Holmer, M., Mateo, M.A.,

Apostolaki, E.T., Kendrick, G.A., Krause-Jensen, D., McGlathery, K.J. and Serrano,

O. (2012). Seagrass ecosystems as a globally significant carbon stock. Nature

Geoscience 5: 505-509.

Freeman, A.S., Short, F.T., Isnain, I., Razak, F.A. and Coles, R.G. (2008). Seagrass on the

edge: Land-use practices threaten coastal seagrass communities in Sabah, Malaysia.

Biological Conservation 141: 2993-3005.

Fryirs, K. (2012). (Dis)Connectivity in catchment sediment cascades: a fresh look at the

sediment delivery problem. Earth Surface Processes and Landforms 38 (1): 30-46.

57

Gattuso, J.P., Gentili, B., Duarte, C.M., Kleypas, J.A., Middelburg, J.J., and Antoine, D.

(2006). Light availability in the coastal ocean: impact on the distribution of benthic

photosynthetic organisms and their contribution to primary production.

Biogeosciences 3: 489-513.

Gibbs, M.M. (2008). Identifying source soils in contemporary estuarine sediments: a new

compound-specific isotope method. Estuarine and Coasts 31: 344-359.

Gonneea, M.E., Paytan, A. and Herrea-Silveira, J.A. (2004). Tracing organic matter sources

and carbon burial in mangrove sediments over the past 160 years. Estuarine, Coastal

and Shelf Science 61: 211-227.

Grech, A. and Coles, R.G. (2010). An ecosystem-scale predictive mode of coastal seagrass

distribution. Aquatic Conservation: Marine and Freshwater Ecosystems 20:437-444.

Green, E.P. and Short, F.T. (2003). World Atlas of Seagrasses. Berkeley: University of

California Press.

Guzman, G., Quinton, J.N., Nearing, M.A., Mabit, L. and Gómez, J.A. (2013). Sediment

tracers in water erosion studies: current approaches and challenges. Journal of Soils

and Sediments 13: 816-833.

Hancock, G. and Revill, A. (2011). Land-use and erosion source discrimination of soil and

carbon sources to the Logan and Albert rivers using Compound Specific Isotope

Analysis. CSIRO Land and Water Science Report 2/11.

Harris, D., Horwath, W.R. and Kessel C. (2001). Acid fumigation of soils to remove

carbonates prior to total organic carbon or carbon-13 isotopic analysis. Soil Science

Society of America 65: 1853-1856.

van der Heide, T., van Nes, E.H., van Katwijk, M.M. and Smolders, A.J.P. (2011). Positive

feedbacks in seagrass ecosystems – evidence from large-scale empirical data. PLoS

ONE 6 (1): e16504.

Hedges, J.I. and Stern, J.H. (1984). Carbon and nitrogen determinations of carbonate-

containing solids. Limnology and Oceanography 29 (3): 657-663.

Hemminga, M.A., Slim, F.J., Kazungu, J., Ganssen, G.M., Nieuwenhuize, J. and Kruyt, N.M.

(1994) Carbon outwelling from a mangrove forest wit hadjacent seagrass beds and

coral reefs (Gazi Bay, Kenya). Marine Ecology Progress Series 106: 291-301.

Hemminga, M.A. and Mateo, M.A. (1996). Stable carbon isotopes in seagrasses: variability in

ratios and use in ecological studies. Marine Ecology Progress Series 140: 285-298.

Hobbie, E.A. and Werner,R.A. (2004). Intramolecular, compound-specific, and bulk carbon

isotope patterns in C3 and C4 plants: a review and synthesis. New Phytologist 161

(2): 371-385.

Hopkinson, C.S., Cai, W.J. and Hu, X. (2012). Carbon sequestration in wetland dominated

coastal systems – a global sink of rapidly diminishing magnitude. Current Opinion in

Environmental Sustainability 4: 186-194.

Jachowski, N.R.A. (n.d.). Geographical approaches to marine ecosystem modeling: Spatial

prediction of mangrove, seagrass and coral ecosystems distributions in the Andaman

Sea. (PhD Thesis). National University of Singapore.

58

van Katwijk, M.M, Bos, A.R., Hermus, D.C.R. and Suykerbuyk W. (2010). Sediment

modification by seagrass beds: Muddification and Sandification induced by plant

cover and environmental conditions. Estuarine, Coastal and Shelf Science 89:175-

181.

Kennedy, P., Kennedy, H. and Papadimitriou, S.(2005). The effect of acidification on the

determination of organic carbon, total nitrogen and their stable isotopic composition

in algae and marine sediment. Rapid Communications in Mass Spectrometry. 19:

1635-1068.

Kennedy, H., Beggins, J., Duarte, C.M., Fourqueen, J.W., Holmer, M., Marba, N. and

Middelburg, J.J. (2010). Seagrass sediments as a global carbon sink: Isotopic

constraints. Global Biogeochemical Cycles 24: GB4026.

Komada, T., Anderson, M.R. and Dorfmeier C.L. (2008). Carbonate removal from coastal

sediments for the determination of organic carbon and its isotopic signatures, δ13

C

and ∆14

C: comparison of fumigation and direct acidification by hydrochloric acid.

Limnology and Oceanography: Methods 6: 254-262.

Kuramoto, T. and Minagawa, M. (2001). Stable carbon and nitrogen isotopic characterization

of organic matter in a mangrove ecosystem on the Sourthwestern Coast of Thailand.

Journal of Oceanography 57: 421-431.

Lapointe, B.E. and Clark, M.W. (1992). Sediment inputs from the watershed and coastal

eutrophication in the Florida Keys. Estuaries 15 (4): 465-476.

Lee. K.S., Park, S.R. and Kim Y.K. (2007). Effects of irradiance, temperature, and nutrients

on growth dynamics of seagrasses: A review. Journal of Experimental Marine

Biology and Ecology 350: 144-175.

Lee, S.Y., Dunn, R.J.K., Young , R.A., Connolly, R.M., Dale, P.E.R., Dehayr, R., Lemckert,

C.J., Mckinnon, S., Powell, B., Teasdale, P.R. and Welsh, D.T. (2006). Impact of

urbanization on coastal wetland structure and function. Austral Ecology 31: 149-163.

Lepoint, G., Dauby, P. and Gobert, S. (2004). Applications of C and N stable isotopes to

ecological and environmental studies in seagrass ecosystems. Marine Pollution

Bulletin 49: 887-891.

Mcleod, E., Chmura, G.L., Bouillon,E., Salm, R., Björk, M., Duarte, C.M., Lovelock, C.E.,

Schlesinger, W.H. and Silliman B.R. (2011). A blueprint for blue carbon: toward an

improved understanding of the role of vegetated coastal habitats in sequestering

CO2. Frontiers in Ecology and the Environment 9: 552–560.

Mekong River Commission (MRC) (2010). Multi-functionality of Paddy Fields over the

Lower Mekong Basin, MRC Technical Paper No. 26, Mekong River Commission,

Vientiane. 72 pp.

Mellors, J., Marsh, H., Carruthers, T.J.B. and Waycott, M. (2002). Testing the sediment-

trapping paradigm of seagrass: Do seagrasses influence nutrient status and sediment

structure in tropical intertidal environments? Bulletin of Marine Science 71 (3): 1215-

1226.

Mellors , J., Waycott, M. and March, H. (2005). Variation in biogeochemical parameters

across intertidal seagrass meadows in the central Great Barrier Reef region. Marine

Pollution Bulletin 51: 335-342.

59

Miller, M.W. and Sluka, R.D. (1999). Patterns of seagrass and sediment nutrient distribution

suggest anthropogenic enrichment in Laamu Atoll, Republic of Maldives. Marine

Pollution Bulletin 38 (12): 1152-1156.

Miller, J.R., Mackin, G., Lechler, P., Lord, M. and Lorentz, S. (2013). Influence of basin

connectivity on sediment source, transport, and storage within the Mkabela Basin,

South Africa. Hydrology and Earth System Science 17: 761-781.

Mitchell, M.G.E., Bennett, E.M. and Gonzalez, A. (2013). Linking landscape connectivity,

and ecosystem service provision: Current knowledge and research gaps. Ecosystem

16: 894-908.

Mukunda, R., Walling, D.E., Gellis, A.C., Slattery, M.C. and Radcliffe, D.E. (2012).

Sediment source fingerprinting: Transforming from a research tool to a management

tool. Journal of the American Water Resources Association 48 (6): 1241-1257.

Muzuka, A.N.N. and Shunula, J.P. (2006). Stable isotope compositions of organic carbon and

nitrogen of two mangrove stands along the Tanzanian coastal zone. Estuarine,

Coastal and Shelf Science 66: 447-458.

O'Leary, M.H. (1981). Review: Carbon Isotope Fractionation in Plants. Phytochemistry 20

(4): 553-567.

O'Leary, M.H. (1988). Carbon Isotopes in Photosynthesis. Bioscience 38 (5): 328-336.

Ometto, J. P.H.B., Ehleringer, J.R., Domingues, T.F., Berry, J.A., Ishida, F.Y., Mazzi, E.,

Higuchi, N., Flanagan, L.B., Nardoto, G.B. and Martinelli, LA.. (2006).The stable

carbon and nitrogen isotopic composition of vegetation in tropical forests of the

Amazon Basin, Brazil. Biogeochemistry 79: 251-274.

Ooi, J.L.S., Kendrick, G.A., Van Niel, K.P. and Affendi, Y.A. (2011). Knowledge gaps in

tropical Southeast Asian seagrass systems. Estuarine, Coastal and Shelf Science

92:118-131.

Orth, R.J, Carruthers, T.J.B., Dennison, W.C., Duarte, C.M., Fourqurean, J.W., Heck Jr, K.L.,

Hughes, A.R., Kendrick, G.A., Kenworthy, W.J., Olyarnik, S., Short, F.T., Waycott,

M. and Williams, S.L. (2006). A global crisis for seagrass ecosystems. BioScience 56

(12): 987-996.

Owens, P.N. (2007). Introduction: Background and summary of this issue on sediment

linkages. Journal of Soils and Sediments 7 (5): 273-276.

Palmer, M.J. and Douglas, G.B. (2008). A Bayesian statistical model for end member analysis

of sediment geochemistry, incorporating spatial dependences. Journal of the Royal

Statistical Society: Series C (Applied Statistics) 57 (3): 313-327.

Parnell, A.C., Inger, R., Bearhop, S. and Jackson, A.L. (2010). Source partitioning using

stable isotopes: Coping with too much variation. PLoS ONE 5 (3): e9672.

Parnell, A.C., Phillips, D.L., Bearhop, S., Semmens, B.X., Ward, E.J., Moore, J.W., Jackson,

A.L., Grey, J., Kelly, D.J. and Inger, R. (2013). Bayesian stable isotope mixing

models. Environmetrics 24 (6): 387-399.

60

Phillips, D.L. (2001). Mixing models in analyses of diet using multiple stable isotopes: a

critique. Oecologia 127: 166-170.

Phillips, D.L. and Gregg, J.W. (2001). Uncertainty in source partitioning using stable

isotopes. Oecologia 127: 171-179.

Phillips, D.L. and Koch, P.L.(2002). Incorporating concentration dependence in stable isotope

mixing models. Oecologia 130: 114-125.

Phillips, D.L.. and Gregg, J.W (2003). Source partitioning using stable isotopes: coping with

too many sources. Oecologia 136: 261-269.

Phillips, D.L., Newsome, S.D. and Gregg, J.W. (2005). Combining sources in stable isotope

mixing models: alternative methods. Oecologia 144: 520-527.

Phillips, D.L. (2012). Converting isotope values to diet composition: the use of mixing

models. Journal of Mammalogy 93 (2): 342-352.

Phillips, J.D. and Slattery, M.C. (2006). Sediment storage, sea level, and sediment delivery to

the ocean by coastal plain rivers. Progress in Physical Geography 30 (4): 513-530.

Phillips, J.M., Russell, M.A. and Walling, D.E. (2000). Time-integrated sampling of fluvial

suspended sediment: a simple methodology for small catchments. Hydrological

Processes 14: 2589-2602.

Russell, B.D., Connell, S.D., Uthicke, S., Muehllehner, N., Fabricius, K.E. and Hall-Spencer,

J.M. (2013). Future seagrass beds: Can increased productivity lead to increased

carbon storage? Marine Pollution Bulletin 73 (2):463-469.

Ryba, S.A. and Burgess, R.M. (2002). Effects of sample preparation on the measurement of

organic carbon, hydrogen, nitrogen, sulphur, and oxygen concentrations in marine

sediments. Chemosphere 48: 139-147.

Salomons, W. (2005). Sediments in the catchment-coast continuum. Journal of Soils and

Sediments 5 (1): 2-8.

Salomon, W., Kremer, H.H. and Turner, K. (2005). The catchment to coast continuum. In:

Crossland, C.J., Kremer, H.H., Lindeboom, H.J., Marshall Crossland, J.I., Le

Tisseier, M.D.A. (Eds.), Coastal Fluxes in the Anthropocene. Springer Verlag,

Germany, pp. 145-200.

Schimel, D.S. (1993). Theory and Application of Tracers. San Diego, CA: Academic

Sheaves, M. (2009). Consequences of ecological connectivity: the coastal ecosystem mosaic.

Marine Ecology Progress Series 391: 107-115.

Short, F.T. and Wyllie-Echeverria, S. (1996). Natural and human-induced disturbance of

seagrasses. Environmental Conservation 23 (1): 17-27.

Shubert, C.J. and Nielsen, B. (2000). Effects of decarbonation treatments on δ13

C values in

marine sediments. Marine Chemistry 72: 55-59.

Sidle, R.C., Sasaki, S., Otsuki, M., Noguchi, S. and Nik, A.R. (2004). Sediment pathways in a

tropical forest: effects of logging roads and skid trails. Hydrological Processes 18,

703-720.

61

Sidle, R.C. and Ziegler, A.D. (2012). The dilemma of mountain roads. Nature Geoscience 5:

437-438.

Slob, A. and Gerrits, L. (2007). The dynamics of sedimentary systems and the whimsicality of

policy processes. Journal of Soils Sediments 7 (5): 277-284.

Small, I.F., Rowan, J.S. and Franks, S.W. (2002). Quantitative sediment fingerprinting using

a Bayesian uncertainty estimation framework. The Structure, Function and

Management of Fluvial Sedimentary Systems (Proceedings of an international

symposium held at Alice Springs, Australia, September 2002). IAHS Publication 276:

443-450.

Stable Isotope Analysis in R (n.d.). Retrieved from:

www.tcd.ie/Zoology/research/research/theoretical/siar.php (Accessed June 2013).

Sukristiyonubowo, Gabriels, D. and Verloo, M. (2010) Sediment trapping by terraced paddy

field on slopping agricultural land. Indonesian Journal of Agricultural Science 11 (2):

57-64.

UC Davis Stable Isotope Facility (UCDavis SIF) (n.d.). Carbon (13

C) and Nitrogen (15

N)

Analysis of Solids by EA-IRMS. Department of Plant Sciences. Retrieved from:

http://stableisotopefacility.ucdavis.edu/13cand15n.html. (Assessed June 2013).

Unsworth, R.K.F., Collier, C.J., Henderson, G.M. and McKenzie, L.J. (2012). Tropical

seagrass meadows modify seawater carbon chemistry: implications for coral reefs

impacted by ocean acidification. Environmental Research Letters 7: 024026.

Walling, D.E. (2005). T racing suspended sediment sources in catchments and river systems.

Science of Total Environment 344: 159-184.

Walthert, L., Graf, U., Kammer, A., Luster, J., Pezzotta, D., Zimmermann, S. and Hagedorn,

F. (2010). Determination of organic and inorganic carbon, δ13

C, and nitrogen in soils

containing carbonates after acid fumigation with HCl. Journal of Plant Nutrition and

Soil Science 173: 208-216.

Waycott, M., Duarte, C.M., Carruthers, T.J.B., Orth, R.J, Dennison, W.C., Olyarnik, S.,

Calladine, A., Fourqurean, J.W., Heck Jr, K.L., Hughes, A.R., Kendrick, G.A.,

Kenworthy, W.J., Short, F.T., Waycott, M. and Williams, S.L. (2009). Accelerating

loss of seagrass ecosystems across the globe threatens coastal ecosystems. PNAS 106

(30): 12377-12381.

Ziegler, A.D. and Giambelluca, T.W. (1997). Importance of rural roads as source areas for

runoff in mountainous areas of northern Thailand. Journal of Hydrology 196: 204-

229

Ziegler A.D., Giambelluca, T.W. ,Plondke, D., Leisz, S., Tran, L.T., Fox, J., Nullet, M.A.,

Vogler, J.B., Troung, D.M. and Vien, T.D. (2007). Hydrological consequences of

landscape fragmentation in mountainous northern Vietnam: Buffering of Hortonian

overland flow. Journal of Hydrology 337: 52-67.