Carbon Cycling in FloodplainEcosystems: Out-Gassing and

Photosynthesis Transmit Soil d13CGradient Through Stream Food Webs

Duncan P. Gray,1* Jon S. Harding,1 Bo Elberling,2 Travis Horton,3

Tim J. Clough,4 and Mike J. Winterbourn1

1School of Biological Sciences, University of Canterbury, Private Bag 4800, Christchurch, New Zealand; 2Department of Geographyand Geology, University of Copenhagen, Øster Voldgade 10, 1350 Copenhagen K, Denmark; 3Department of Geological Sciences,

University of Canterbury, Private Bag 4800, Christchurch, New Zealand; 4Agriculture and Life Sciences Division, Lincoln University,

PO Box 84, Canterbury, New Zealand

ABSTRACT

Natural braided river floodplains typically possess

high groundwater–surface water exchange, which

is vital to the overall function and structure of these

complex ecosystems. Spring-fed streams on the

floodplain are also hotspots of benthic invertebrate

diversity and productivity. The sources of carbon

that drive these productive spring-fed systems are

not well-known. We conducted field assessments

and a manipulation, modeling, and a laboratory

experiment to address this issue. Initially d13C

values of both dissolved inorganic carbon (DIC)

and food-web components of five springs were

used to assess the sources of carbon to spring food

webs. Partial pressures of CO2 in upwelling water

ranged from 2 to 7 times atmospheric pressure, but

rapidly approached equilibrium with the atmo-

sphere downstream commensurate with 13C

enrichment of DIC. Speciation modeling and a

laboratory out-gassing experiment suggested that

downstream changes in pH could be explained so-

lely by CO2 out-gassing. However, field results

indicated that both out-gassing and photosynthetic

drawdown by aquatic plants controlled the net flux

of CO2. A whole stream manipulation indicated

out-gassing was the primary effect at the spring

source, which was confirmed by invariant diel pH.

At 1296 m downstream from the spring source a

large diel shift in pH indicated a plant effect on CO2

concentration which would contribute to the

overall downstream gradient in d13C DIC. Within

the first 1296 m the gradient in d13 DIC was

transmitted through three trophic levels of the

spring food web. These findings indicate depen-

dency on groundwater inorganic carbon by spring

stream food webs and strong hydrologically medi-

ated linkages connecting terrestrial, subsurface,

and aquatic components of the floodplain.

Key words: biogeochemistry; connectivity; car-

bon; braided river; food web; stable isotopes; New

Zealand.

Received 13 July 2010; accepted 25 February 2011;

published online 10 March 2011

Author Contributions: DPG conceived the project, designed, and car-

ried out the survey and manipulation, analyzed data, and drafted the

manuscript. JSH contributed to the design of the study, interpretation of

results, and drafts of the manuscript. BE contributed the degassing

experiment, speciation modeling, and contributed to the manuscript. TH

performed the carbon isotope analysis of organic samples and certified

reference materials, contributed to the improvement of the original

manuscript, and assisted in the interpretation of the stable isotopic data.

TJC contributed to the degassing experiment, speciation modeling, and

the manuscript. MJW contributed to the design of the study and inter-

pretation of results, identified aquatic invertebrates and helped prepare

material for isotopic analysis.

*Corresponding author; e-mail: duncan.p.gray@gmail.com

Ecosystems (2011) 14: 583–597DOI: 10.1007/s10021-011-9430-1

� 2011 Springer Science+Business Media, LLC

583

INTRODUCTION

Many natural rivers have extensive flood-plains

that contain diverse habitats. In braided systems,

floodplain habitats are in a state of continual

change (Arscott and others 2000; van der Nat and

others 2003). Floodplains are connected hydrolog-

ically to the river and wider catchment (Brunke

and Gonser 1997; Ward and others 1999; Woessner

2000) with both spatial and temporal variability in

groundwater–surface water exchange (Stanford

and Ward 1993; Brunke and Gonser 1997; Poole

2002). Feedback loops and interactions between

habitats and the wider catchment, have been in-

ferred (Hynes 1975; Stanford 1998; Wiens 2002)

and links between aquatic and terrestrial systems,

hydrology and ecology, and among organisms

across habitat boundaries have been demonstrated

(Fisher and others 2004; Paetzold and others 2005;

Malard and others 2006; Tockner and others 2006).

As a result, Hynes’ (1975) assertion that the

‘‘stream is ruled by its Valley’’ has received re-

newed attention.

Early biogeochemical studies of streams focused

on small, forested headwater ecosystems and con-

sequently on the sources and dynamics of organic

carbon in heterotrophic food webs (Fisher and

Likens 1973). Dissolved inorganic carbon (DIC)

provides the basis of autotrophic aquatic food webs

in streams with adequate sunlight for photosyn-

thesis (Raven and others 1985; Sand-Jensen and

Frost-Christensen 1998). From an ecosystem per-

spective the distinction between organic and inor-

ganic carbon may be misleading as the two are

interchangeable through the metabolic processes of

respiration and photosynthesis (Shibata and others

2001). Although hydrologists and geochemists are

aware of the biological controls acting upon DIC in

streams (Clark and Fritz 1997; Jones and Mulhol-

land 1998; Telmer and Veizer 1999; Doctor and

others 2008), there is limited biological research on

the connections between DIC and in-stream eco-

logical patterns. That DIC may be the primary

source of carbon to a stream food web and that the

sources of that DIC are varied and complex has

been inferred (Finlay 2003; Hawke and Polaschek

2005). However, few studies have identified and

traced the signature of d13C DIC through a stream

food web (Rounick and James 1984; Finlay 2004;

Jepsen and Winemiller 2007).

In streams, the amount of inorganic carbon

available to primary producers is dictated by the

balance between atmospheric diffusion, in-stream

metabolism, and groundwater inorganic carbon

inputs when present (Allan and Castillo 2007).

Diffusion to the atmosphere results in a shift in the

CO2(aq) concentration of water towards that of the

atmosphere (ca. 387 ll l-1), whereas in-stream

metabolism will either increase or decrease CO2(aq)

concentration depending on the balance between

photosynthesis and respiration (Fisher and Likens

1973). Groundwater is generally supersaturated

with CO2 due to soil respiration within the catch-

ment (Dawson and others 1995; Jones and Mul-

holland 1998) and in the Sleepers River watershed,

Vermont, USA, partial pressure of carbon dioxide

(PCO2) was 4–74 times that of the atmosphere

(Doctor and others 2008).

The isotopic evolution of DIC begins with disso-

lution of atmospheric CO2 (d13C � -8& V-PDB)

by meteoric water. DIC is generally depleted in 13C

relative to the atmosphere due, in part, to the

addition of CO2 from soil respiration. The isotopic

composition of soil derived CO2, in turn depends

on the photosynthetic pathway of the predominant

surface vegetation. The C3 metabolic pathway for

carbon fixation in photosynthesis operates in about

85% of plant species and predominates in most

terrestrial ecosystems (Clark and Fritz 1997). Most

C3 plants have d13C values that range from -24 to

-30& with an average value of about -27&

(Clark and Fritz 1997). Photosynthetically fixed

carbon in the form of vegetative matter accumu-

lates within soils where aerobic bacteria convert

the majority back to CO2, resulting in soil CO2

concentrations 10–100 times greater than in the

atmosphere (Elberling and Ladegaard-Pedersen

2005). Microbial and plant root respired CO2 have

the same d13C value as the source vegetation,

although out-gassing from soils may result in iso-

topic enrichment in excess of 4& (Cerling and

others 1991; Aravena and others 1992, but see

Davidson 1995). Therefore, the d13C of aerobic soils

and soil water DIC in C3 dominated landscapes,

such as New Zealand, can be expected to be

approximately -23& ± 3. A further addition to

soil and groundwater DIC comes from bicarbonate

(HCO3-) ions derived from bedrock weathering.

Carbonate dissolution by carbonic acid produces a

d13C DIC value intermediate between that of soil

CO2 and the carbonate source, whereas HCO3-

derived from silicate weathering will be 7–10&

greater (more enriched) than the d13C value of soil

CO2 (Kendall and others 1992; Pawellek and Veizer

1994). The d13C value of most carbonates is

approximately 0& V-PBD (Craig 1953; Taylor and

Fox 1996). d13C DIC values of groundwater inor-

ganic carbon more negative than those of surface

water have been reported in numerous studies (for

584 D. P. Gray and others

example, Finlay 2003; Doctor and others 2008) and

suggest that groundwater CO2 is likely to have an

influence on surface streams in many systems with

high groundwater–surface water exchange.

Because spring-fed streams on floodplains are

hotspots of benthic invertebrate diversity and pro-

ductivity (Gray and Harding 2009) the question

arises as to what energy pathways support them?

The primary objective of our study was to investi-

gate the importance of groundwater carbon to

floodplain spring food webs. The study was con-

ducted in three parts. First, an investigation of

multiple sites along five spring streams was made to

determine the nature and consistency of ground-

water carbon contributions to spring stream food

webs. This was followed by a laboratory based out-

gassing experiment and speciation modeling to in-

fer the degree to which downstream trends in pH

could be explained by the flux of CO2. Finally,

continuous diel measurements of pH and a whole

stream manipulation of five additional springs were

performed to investigate the relative effects of out-

gassing and photosynthesis on the longitudinal

gradient identified in part one.

METHODS

Site Selection

All spring streams (except one on the West Coast)

were located within the floodplains of braided rivers

in Canterbury, New Zealand, at altitudes between

515 and 1000 m a.s.l. (Figure 1). These catchments

were within the Torlesse Terrane that comprises

quartzofeldspathic sandstones and mudstones with

subsidiary conglomerates (Anekant and others

2004). The area occupies an orographic rain ‘‘sha-

dow’’ produced by the prevailing ‘‘westerly’’ air

flow. Consequently, whilst study sites and sur-

rounding terrestrial habitats may experience con-

siderable aridity, the alpine sourced rivers of which

they are a part are subject to frequent and extreme

flooding (Winterbourn 1997).

Initially, five floodplain spring streams were se-

lected as replicate sites to assess the response of

stream ecosystems to groundwater carbon inputs.

The streams were located in the Hawdon Valley,

2 km upstream of the Sudden Valley Stream con-

fluence, at 600 m a.s.l.; at Manuka Point in the

Rakaia Valley, at 600 m a.s.l.; at Mount Potts in the

Rangitata Valley, 550 m a.s.l.; below Mt Joseph in

the Fork Stream, a tributary of the Tekapo River,

1000 m a.s.l., and below Mt Glen Lyon in the

Hopkins Valley, 600 m a.s.l. Five sampling sites

were established along each stream at distances of

0, 6, 36, 216, and 1296 m from the source. The

streams flowed for at least 1.5 km through tussock

grassland and no forest patches were present. In an

attempt to avoid the influence of methanogenesis

or sulfur bacteria, which might have depleted d13C

values of biota and DIC, no streams were in close

proximity to swamps or standing water (Jones and

Mulholland 1998). However, we cannot com-

pletely discount chemoautotrophic bacterial path-

ways which have been identified in predominantly

aerobic freshwater systems (Trimmer and others

2009; Doi and others 2006). Streams were sampled

in September 2008.

Another five spring streams fulfilling the same

criteria as those described above were chosen for

the manipulation of out-gassing and photosynthe-

sis. We decided to use five novel springs to further

test the commonality of the pattern. A 36 m long

reach starting at the spring source was used.

Figure 1. Location of the five spring streams included in

the initial survey in September 2008 (closed circles) and

springs used in the whole stream manipulation of out-

gassing and photosynthesis in April 2009 (closed triangles).

Symbols are superimposed on outlines of major river

floodplains. Inset: location of the study area within New

Zealand.

Carbon Cycling in Floodplain Ecosystems 585

Reaches had no inflowing tributaries and were

located within the Rakaia, Waimakariri, and Tar-

amakau river floodplains (Figure 1). The manipu-

lation was carried out in April 2009.

Sample Collection and ExperimentalDesign

For the longitudinal study, spot water chemistry

samples, and data were collected at each of the five

sites along each of the five streams. Water samples

for the analysis of DIC were collected in autoclaved

250 ml, Nalgene bottles from well-mixed areas of

the stream during mid-morning through mid-

afternoon. Bottles were rinsed three times in

stream water, filled and capped underwater to en-

sure that no head space remained in the bottle.

Samples were immediately put on ice, in the dark,

and returned to the laboratory within 24 h and

refrigerated. Samples were couriered to the Geo-

logical and Nuclear Sciences Stable Isotope Labo-

ratory, Wellington and stored in a refrigerator for

5 days at 4�C, before analysis. Water samples were

reacted with 85% phosphoric acid at room tem-

perature, and CO2 was extracted and purified to be

run on a Europa Geo 20–20 mass spectrometer in

dual inlet mode. All results are reported with re-

spect to VPDB, normalized to a single internal so-

dium carbonate water standard with reported

values of -5.6& for d13C. The external precision

for these measurements is 0.1& for d13C. The car-

bon concentration in mg C kg-1 of water was cal-

culated from the measured CO2 pressures. DOC

concentrations were estimated using the spectro-

photometric method of Collier (1987). Tempera-

ture (�C) and conductivity (lS25 cm-1) were

measured with a calibrated Oakton conductivity/

temperature 10 meter and dissolved oxygen with a

YSI 550 meter. Measurements of water pH were

made with a Solstat FET pH meter and glass elec-

trode with re-calibration before each measure-

ment. Stream discharge was assessed across three

transects by measuring wetted width, depth, and

velocity determined with a Marsh-McBirney flow

meter. Percentage macrophyte/filamentous algae

cover and proportions of pools, riffles, and runs in

the reach upstream of each site were estimated,

visually. Altitude was measured using a Trimble

Recon GPS XC unit and used to calculate average

slope between sites. Finally, air temperature was

recorded at each site using a Center 300 Type K

thermometer. Subsequent continuous measure-

ments of pH were made in the Hawdon Valley

spring using an in situ YSI 6600 multi-parameter

water quality monitor.

Samples of submerged bryophytes, macrophytes,

and filamentous algae were collected mid-stream

for isotopic analysis to reduce the likelihood that

plants had been exposed to the atmosphere during

low flow. Emergent macrophytes and bryophytes

were not collected. Biofilms were scrubbed from

more than four stones using a tooth brush and

aquatic invertebrates were collected with a kick-

net. Where present the leptophlebiid mayfly Dele-

atidium and the hydrobiid snail Potamopyrgus anti-

podarum (primary consumers), predatory caddisfly

larvae, Hydrobiosis and Psilochorema (Hydrobiosi-

dae), and the predatory stonefly Stenoperla were

collected. When these taxa were not available

other members of the appropriate consumer guild

were collected: primary consumers included a

mayfly Nesameletus, two stoneflies, Austroperla and

Zelandoperla and the caddisflies Pycnocentria and

Olinga; predatory invertebrates were a stonefly

Megaleptoperla and three caddisflies Philoreithrus,

Hudsonema, and Polyplectropus. Predatory fish were

captured with a Kainga EFM 300 backpack elec-

trofishing machine (NIWA Instrument Systems,

N.Z.). The native galaxiids, Galaxias paucispondylus

and G. vulgaris were obtained where possible but, at

some sites salmonids (Oncorhynchus mykiss, Salmo

trutta,) or bullies (Gobiomorphus breviceps) were

collected. Lastly, riparian fishing spiders (Dolomedes

aquaticus), which have been shown to derive

greater than 90% of their diet from aquatic insects

(Williams 1979), were collected from each site.

Spiders were found beneath overhanging tussocks

directly adjacent to the stream. All organic samples

were held on ice in the dark, returned to the lab-

oratory within 24 h and frozen at -20�C.

The out-gassing and photosynthesis stream

manipulation used a different method for the

measurement of DIC. Vacuum sealed 12 ml Exe-

tainers� were opened beneath the surface of the

water to avoid ambient air contamination. Filled

vials were kept on ice before saturated sodium

azide was added prior to shipping to the University

of California Davis (USA) stable isotope laboratory.

Samples were analyzed using a Sercon Trace Gas

system interfaced to a PDZ Europa 20–20 IRMS.

Raw data were normalized using a single-point

additive correction based on replicate analysis of

lithium carbonate (LSVEC) with a d13C of

-46.7& ± 0.07; n = 8.

To test the relative effects of out-gassing and

photosynthetic uptake of CO2(aq) on d13C DIC, a

fully factorial manipulation in five springs was

performed. Initially, water chemistry measure-

ments and water samples were collected from the

source of each stream and at 6, 16, and 36 m

586 D. P. Gray and others

downstream. To inhibit out-gassing, but not pho-

tosynthesis, a sheet of transparent polythene was

laid over the first 36 m of each stream. Once a

satisfactory seal had been achieved and all gas

pockets had been minimized, the sheet was left in

place for 1 h to ensure that all pre-manipulation

water had left the reach. Water sampling was then

repeated at each point. The polythene sheet was

then removed and all macrophytes, bryophytes,

and algal mats were removed from the reach;

substrates with obvious diatom layers were either

removed or inverted. The streams were then left for

a further 1 h before re-sampling. The polythene

sheet was then put in place again and a final set of

water samples taken after 1 h. After each sample

collection the residence time of water in each reach

was estimated using a fluorescein dye tracer.

Water samples for chemical speciation modeling

and controlled out-gassing were collected from the

Hawdon Valley spring because of access issues at

other sites during high summer flows. Samples

were collected at mid-depth using gas-tight 130 ml

glass bottles, and transported to Lincoln University

stable isotope laboratory, New Zealand, for carbon

and isotope analysis, and to the Soil–Water

Chemical Laboratory, Department of Geography

and Geology, University of Copenhagen, Denmark

for metal analysis. Total dissolved Fe2+, Fe3+, Al3+,

Mn2+, Mg2+, Ca2+, K+, and Na+ were determined on

acidified water samples (10% HNO3) by atomic

absorption spectrophotometry (AAS). Concentra-

tions of Cl-, NO3-

, and SO42- were measured using

an ion chromatograph in one determination

(Dionex Ionpac AS14 4 mm column and CD20

conductivity detector). Replicate samples were

interspersed with standards and indicated a preci-

sion better than 4%.

In the laboratory a controlled degassing experi-

ment was performed to quantify the importance of

CO2 out-gassing on changes in pH and d13C DIC.

Initially, a 2 ml water sample was taken from a

sealed bottle via a septum. This water was injected

into a pre-evacuated 12 ml Exetainer� pre-purged

with helium. The dissolved CO2 was allowed to

equilibrate with the He headspace for 24 h, prior to

d13C analysis. After the first water sample had been

taken the remaining water was decanted into a

200 ml beaker and stirred constantly whilst at a

temperature of 18 ± 1�C. Water pH was measured

on 12 occasions during the 1200 min degassing

experiment, with further water samples taken from

the beaker at the time of every 2nd pH measurement

for d13C DIC analysis. Isotopic analysis was per-

formed using a PDZ Europa 20/20 TG11-CF/IRMS.

All results were corrected for linearity and normal-

ized to the V-PDB scale using two in-house gas

standards (-8.13& ± 0.03 and -38.57& ± 0.07)

supplied by the Scripps Institution of Oceanography

(La Jolla, CA 92093-0244, USA) and calibrated

against the international standard NBS-19. All

measurements were replicated three times.

Organic Sample Preparationand Analysis

Bryophytes, macrophytes, and filamentous algae

were rinsed to remove contaminant material. Guts

of all aquatic invertebrates and riparian spiders

were removed where possible. Although the ceph-

alothorax and leg coxae of Dolomedes aquaticus may

contain stored food materials, Collier and others

(2002) found no significant difference in d13C val-

ues between them and legs minus coxae; therefore,

we combined all lower leg and cephalothorax

material in samples. Snails were removed from their

shells prior to isotope analysis. This can be achieved

either through manual extraction of soft tissues or

by dissolving the shell in HCl. The two techniques

were compared and no effect of HCl was found on

d13C values (P = 0.993). However, d15N values for

material treated with HCl were significantly lower

(P = 0.01) and manual extraction was used there-

after. All organic material was dried at 45�C for at

least 48 h before grinding with a mortar and pestle.

Because endogenous lipids have more depleted

d13C values than other major compounds, their

presence can affect conclusions drawn from d13C

analyses in food web studies (Logan and others

2008). Therefore, lipids were extracted from all

samples as described by Logan and others (2008).

Samples were then re-dried at 45�C for at least 48 h

and frozen at -80�C prior to isotope analysis. Val-

ues of d13C were determined using continuous flow

IRMS in the Stable Isotope Laboratory, University of

Canterbury. Samples were combusted at 1050�C in

a Costech ECS 4010 elemental analyzer and the

CO2 was injected into a Thermo Delta V Plus IRMS

via a Conflo III gas interface. Values of d13C were

normalized to the V-PDB scale based on replicate

analyses of certified primary reference materials

(IAEA-CH-3; NBS-22; USGS24) for each analytical

sequence. Analytical precision for all samples was

better than 0.15& V-PDB.

Estimations and Statistical Analyses

Excess PCO2

The partial pressure of carbon dioxide (PCO2) in

water samples was estimated using the relationship

given by Doctor and others (2008):

Carbon Cycling in Floodplain Ecosystems 587

CTa0

KH¼ PCO2 ð1Þ

where CT is the total DIC concentration (mol C l-1)

of the sample, a0 is the ionization fraction between

CO2 and H2CO3 (Stumm and Morgan 1981), and KH

is the Henry’s law equilibrium constant for CO2 in

water. Because KH is temperature dependent, the

value was estimated from the temperature of the

stream at the time of sample collection using the

relationships provided in Telmer and Veizer (1999).

Excess PCO2 (ePCO2) is the ratio of the calculated

value of PCO2 in the sample to that of the atmo-

sphere, which was assumed to be 387 ll l-1.

Food Web Components and Water Chemis-try Relationships between water chemistry vari-

ables, d13C DIC, d13C of food web components and

distance downstream from the source were assessed

using mixed effects linear models. Mixed models

were used to remove variation contributed by ran-

dom, higher level effects (stream), before assessing

patterns due to fixed variables. To test for homo-

geneity of slopes between significant food web

components and DIC we used a mixed effects linear

model ANCOVA. The d13C value of filamentous

algae and distance were log10 transformed to im-

prove normality and homogeneity of variances.

Quantification of Carbon Utilization Methods to

partition the sources of carbon in stable isotope

studies center around the use of mixing models

(Rasmussen 2010). To separate contributions of

terrestrial organic carbon from groundwater inor-

ganic carbon we have used a gradient based model,

which produces estimates of the proportion of ter-

restrial consumption (pT) from signature slopes of

consumers (Rasmussen 2010). The d13C value of

inorganic carbon and most autochthonous pro-

duction exhibit pronounced gradients along rivers

relative to terrestrial signatures, which exhibit no

gradient. Terrestrial food consumption therefore

reduces the downstream d13C signature slope of

consumers in proportion to the amount of terres-

trial food consumed.

Downstream Patterns of DIC Mixed model anal-

ysis was also used to test the importance of different

physico-chemical variables on downstream

enrichment of d13C DIC. The following transfor-

mations were made; slope (degrees) and ePCO2

were log10 transformed, whereas upstream per-

centage macrophyte cover and percentage riffle

were arcsine square root transformed to improve

patterns of normality and heteroscedasticity.

Stream Out-Gassing and Photosynthesis Manipu-lation The effect of whole stream manipulations

on d13C of DIC was assessed using mixed effects

linear models. Values of d13C DIC, standardized to

the value at source of each stream, were used as the

response to control for variable stream source val-

ues. The change in residence time of water along

each reach as a result of macrophyte removal was

used as the stream level random factor. Significant

differences during mixed effects linear modeling

were identified using ANOVA and P values. All

analyses were conducted in R using the package

‘‘lme4’’ (R Development Core Team 2007).

Experimental Out-Gassing and Speciation Model-ing Modeling CO2 equilibrium requires calcula-

tions of ion activities from concentrations and mass

balance equations. The computer program PHRE-

EQE was used to calculate geochemical equilibrium

based on thermodynamic data given for tempera-

tures between 1 and 100�C (Parkhurst and others

1990). Input data for simulations included con-

centrations of dissolved chemical elements; Fe2+,

Fe3+, Mg2+, Al3+, Mn2+, Ca2+, K+, Na+, Cl-, NO3-,

SO42-, total inorganic carbon, H+, and tempera-

ture. Activity coefficient calculations were per-

formed with the Davies (1962) approximation,

valid for ionic strengths below 0.1 M. The maxi-

mum ionic strength for these water samples was

below 0.05 M and the Davies approach is therefore

appropriate. PHREEQE has been used in a similar

way by Serrano and others (2000) and by Elberling

and Matthiesen (2007) for evaluating CO2 degas-

sing. In our study the speciation model was not

calibrated but validated against observed pH values.

RESULTS

Longitudinal Study

The five spring streams were characterized by

broadly similar physico-chemical conditions. Dis-

charge increased rapidly from the initial spring

source, but generally plateaued approximately

200 m downstream. An exception was Mount Potts

stream, which joined a large spring at 800 m. The

rapid increases in discharge indicate that ground-

water inputs continued, at least in the upper 200 m

of the streams.

Water temperature at spring sources (6.4–9�C)

was similar to the annual average air temperature

of the sites (Gray and others 2006). pH was circum-

neutral and conductivity was low and relatively

constant (range 39–97 lS25 cm-1) along the length

588 D. P. Gray and others

of each stream (Table 1). Water at the spring

sources was not saturated with oxygen (range 76–

94%) but became saturated downstream (101–

108%). DOC concentrations were low, and at some

spring sources were below the detection level of

1.6 g m-3. All streams flowed across river flats and

therefore slopes were low. Spring sources were

typically pools with a high percentage of macro-

phyte cover (mean 56% SD 15.2), which declined

to a mean of 0.7% (SD 0.12) at the downstream

sites as current velocity increased. Both tempera-

ture and pH increased logarithmically downstream

(Figure 2). DIC concentration (mmol kg-1) de-

clined logarithmically with distance downstream,

except at the lowest site on the Hopkins Valley

stream (Figure 2). Between this site and the sam-

pling site immediately upstream, discharge more

than doubled in the absence of tributaries sug-

gesting considerable input of up-welling ground-

water. Groundwater may therefore have provided

the high quantities of DIC measured at the lower

site. Before exclusion of this data point no signifi-

cant relationship was found between DIC concen-

tration and distance downstream (P = 0.28), but

when the point was removed a significant decline

in DIC concentration was found (P = 0.003). The

ePCO2 at spring sources ranged from 2 to 7 times

atmospheric pressure, but was invariably less than

1.5 at the downstream sites. DIC became more

enriched in 13C downstream (Figure 2). Although

the same general pattern was observed in all

streams, actual d13C values ranged from -15.6 to

-12.2& at sources and from -12.5 to -9.0&

1296 m downstream.

The d13C of biofilms, moss, primary consumers,

invertebrate predators, and fish all showed signifi-

cant positive relationships with d13C DIC (Fig-

ure 3). Thus, multiple trophic levels of the spring

food web tracked the rapid downstream enrich-

ment of d13C DIC. However, a non-significant

relationship was found for filamentous algae when

an outlier, that may have been exposed to atmo-

spheric CO2 was included (P = 0.4), although a

significant positive relationship was obtained when

the outlier was excluded (P < 0.001). The only

food web components not related to d13C DIC were

macrophytes and riparian spiders.

A recent study in forested temperate systems has

shown that fractionation effects on algal d13C were

equal to or more substantial than the effect of

variation in d13C DIC suggesting that patterns

transmitted through the food web might in fact be

driven by CO2 concentration, rather than the iso-

topic value of source inorganic carbon (Finlay

2004). However, regressions between 1/CO2 and

fractionation (as in Finlay 2004, equ. 2) between

d13C of DIC and primary producers in our streams

were all non-significant. CO2 concentration did not

appear to have a primary influence on the extent of

isotopic fractionation during aquatic photosynthe-

sis in the streams studied. However, we did not

analyze individual algal species, thus we cannot

exclude an algal fractionation effect.

Gradient based mixing models allowed for the

partitioning of autochthonous and allochthonous

contributions to the d13C signatures of stream biota

(Table 2). The value of pT calculated for biofilm was

0.41 indicating approximately 41% allochthonous

content. Invertebrate predators and fish utilized a

slightly greater proportion of terrestrial carbon in

their diets, potentially indicative of the consump-

tion of terrestrially derived prey. Fishing spiders

derived 70% of their carbon from terrestrial sour-

ces. Overall, the results indicated a considerable

influence of inorganic groundwater carbon on the

food webs of these spring streams.

Table 1. Physico-Chemical Characteristics Along Five Spring Streams Sampled in September 2008

All sites Source Lowest site

Mean Range Range Range

pH 7.0–8.0 7.0–7.5 7.6–8.0

Temperature (�C) 7.9 6.4–11.2 6.4–9.0 7.3–11.2

Conductivity (lS25 cm-1) 57 39–97 41–65 39–80

Dissolved oxygen (% saturation) 95 76–112 76–94 101–108

DOC (g m-3) 2.18 0–3.2 0–2.6 1.9–3.2

Average slope (degrees) 1.6 0–13 0 0.37–1

Up-stream macrophyte cover (%) 31 0.5–90 15–90 0.5–1

Up-stream riffle (%) 37.0 0–90 0–5 30–90

Air temperature (�C) 15.2 9.1–20.3 na na

Hawdon Valley sites are included in this data set. pH values are only shown as a range. Up-stream macrophyte and riffle values are for the stream reach directly above each site.

Carbon Cycling in Floodplain Ecosystems 589

To identify downstream drivers of d13C DIC we

compared gradients to physico-chemical descriptors

of the stream. Both macrophyte cover (P = 0.005)

and ePCO2 (P < 0.001) showed a significant rela-

tionship with d13C DIC. In plots of ePCO2 and

up-stream macrophyte cover versus d13C DIC

(Figure 4) regression lines for ePCO2 and d13C DIC

were similar for all streams except Fork stream.

Isotopic enrichment of DIC along Fork stream was

comparable to that at the other streams, but ePCO2

at the source was only twice that of the atmo-

sphere, and decreased by only one unit over

1296 m. Regressions lines for macrophyte cover

versus DIC d13C were all broadly similar suggesting

that macrophyte cover may be the controlling

variable. Collinearity between predictor variables

Figure 2. Downstream

gradients of pH,

temperature, DIC

concentration, excess

partial pressure of CO2

(ePCO2), and d13C DIC in

five spring streams

(mean ± 1SE). Means

and error bars were not

calculated for pH. Water

samples from Hawdon

Valley springs were

partially frozen during

storage resulting in gas

head formation.

Consequently, these

samples were omitted

from DIC analyses. The

excluded outlier of DIC

concentration from the

lower Hopkins Valley site

is shown as asterisk.

Significance of the

relationship between log

distance and water

chemistry variables are

shown after testing with

mixed effects linear

models with stream as the

random effect.

590 D. P. Gray and others

Figure 3. Relationships

between d13C of food web

components sampled

along five spring streams

and d13C DIC. Plots are

real data, but trend lines

and P values are derived

from mixed effects linear

models with stream as

random factor. The

outlier in the plot of

filamentous algae versus

d13C DIC is shown as

asterisk. Mixed effects

ANCOVA indicated no

significant difference

between the slopes of

each significant line (df

11, X2 = 5.779, P = 0.2).

Table 2. Linear Regression Parameters and Associated Mixing Model Estimates of the Proportion of Ter-restrial Carbon Consumption (pT) by Components of the Spring Biota

Consumer/resource Slope (±SE) Intercept (±SE) r2 RMA slope pT

Invert. consumer/biofilm 0.49 (±0.08) -17.61 (±2.21) 0.69 0.59 0.41

Invert. predator/biofilm 0.34 (±0.11) -20.63 (±2.99) 0.39 0.56 0.44

Fish/biofilm 0.75 (±0.13) -21.14 (±2.89) 0.29 0.52 0.48

Spiders/biofilm 0.17 (±0.06) -21.77 (±1.75) 0.34 0.3 0.7

After Rasmussen (2010). pT is estimated from the slope of the consumer–resource signature relationship.

Carbon Cycling in Floodplain Ecosystems 591

was moderate (VIF = 1.5–3.2) (Quinn and Keough

2002). Percentage upstream cover of macrophytes

was significantly related to ePCO2 (r = 0.621) and

the percentage of riffle habitat (r = -0.654). Due to

this collinearity these analyses were not able to

disentangle the effects of macrophyte cover and

ePCO2 on downstream patterns of DIC d13C.

Stream Out-Gassing and PhotosynthesisManipulation

After controlling for between-stream effects (stan-

dardization) and within-stream changes in water

residence time due to the manipulation (random

factor), a negative shift in d13C DIC due to both

out-gassing and photosynthetic uptake of CO2 was

observed (Table 3). However, only out-gassing had

a significant effect on d13C DIC along the initial

36 m of the streams.

Experimental Out-Gassing andSpeciation Modeling

We observed an increase in pH from 6.9 to 7.6 and

a shift in d13C DIC values from -21.7 to -15.1 as

water de-gassed over 1200 min (Figure 5). The

steady state pH value based on speciation modeling

was 7.68. This value indicated that steady state had

almost been reached after 1200 min, and that the

shift in pH along the spring streams could be ex-

plained solely by degassing of CO2. The large dis-

crepancy between d13C DIC of spring source water

collected during this out-gassing experiment and

that taken during the initial survey is, we believe, a

seasonal effect. The initial survey was performed

during early spring when soils were cold and pre-

cipitation levels low, whereas the de-gassing

experiment was carried out during summer when

soils were warm and precipitation higher. Elevated

summer temperatures increase levels of respiration

and CO2 production in soils. Subsequent precipi-

tation events flush this CO2 into the groundwater

system (Keifer and Amey 1992; Davidson and

Janssens 2006). We have insufficient data to fur-

ther investigate this phenomenon.

Diel Patterns of pH

Speciation modeling indicated that pH was an

appropriate proxy variable for CO2 content of water

and consequently d13C DIC in these streams. Con-

tinuous monitoring at the spring source over a

3 day period revealed an almost constant pH with a

maximum value of 6.94 and an amplitude of 0.03

(Figure 6). Thus, it seems that plants were not

having a significant effect on the quantity or d13C

of DIC at the spring source. However, 1296 m

downstream from the source maximum pH was

8.26 and amplitude was 0.93. During daylight the

increase in stream pH above that dictated by out-

gassing was the result of uptake of CO2 by plants

for photosynthesis. At night, photosynthesis ceased

and there was a net addition of CO2 to the stream

from respiration, potentially offsetting the effects of

out-gassing. Our finding confirms the predomi-

nance of out-gassing close to the spring source, but

indicates a strong effect of photosynthesis and res-

piration by aquatic plants in downstream reaches.

DISCUSSION

Our study showed that the d13C gradient trans-

mitted through the spring stream food web was the

result of both physical out-gassing and photosyn-

thetic uptake of CO2 from the water. The average

d13C value of DIC at spring sources was -12.7&

suggesting that DIC was a combination of C3

derived CO2 and HCO3- derived from mineral

Figure 4. Relationships between d13C DIC, excess partial pressure of carbon dioxide (ePCO2) and percentage upstream

cover of macrophytes (arcsine square root transformed) in individual streams (real data). The r2 values for individual

regressions are shown below each plot. Hawdon Valley data are excluded.

592 D. P. Gray and others

dissolution by carbonic acid. The spring stream food

webs appeared to be autochthonous-based. The

unusual isotopic depletion of carbon fixed within

spring streams permitted the inference that bio-

logical communities were dependent upon CO2

derived from the respiration of soil organisms in the

catchment of the spring.

Drivers of Downstream d13C Enrichment

A number of studies have suggested that carbon

(isotope) exchange with atmospheric CO2

(d13C � -8&) can be partially responsible for

downstream enrichment of d13C DIC values of

water (Taylor and Fox 1996; Yang and others 1996;

Atekwana and Krishnamurthy 1998; Karim and

Veizer 2000; Helie and others 2002; Kanduc and

others 2007). These studies compared headwaters

with lower reaches of large rivers and showed that

headwaters were relatively depleted in d13C DIC.

Doctor and others (2008) pointed out that where

CO2(aq) out-gassing occurs, isotopic equilibrium

between CO2(aq) and atmospheric CO2 can only be

attained after chemical equilibrium has been

reached (that is, an ePCO2 of 1). Therefore, when a

chemical potential for CO2(aq) evasion exists

(ePCO2 > 1) the isotopic signature of DIC will be

regulated predominantly by out-gassing. Isotopic

enrichment of 13C DIC occurs as HCO3- is con-

verted to CO2 to maintain the ionic balance as H+

declines (Hendy 1971; Doctor and others 2008). In

the groundwater dominated streams of our study,

partial pressures of CO2 were in excess of atmo-

spheric pressure at the majority of sites. Only the

lower sites on the Hopkins Valley and Mt Potts

streams had reached chemical equilibrium and

might therefore have begun to be influenced,

Table 3. Linear Mixed Effects Model Tested with Maximum Likelihood Estimation to Investigate the Mainand Interactive Effects of the Prevention of Out-Gassing and Removal of Macrophytes (Fixed CategoricalVariables) on the Downstream Shift in d13C of DIC

Predictors Type var. X2 df P value

Stream Random 0.22463

Gas Treatment 6.7641 2 0.03

Macrophyte Treatment 2.0492 2 0.35

Distance Covariate 13.143 1 <0.001

Gas:macrophyte Treatment interaction 1.1533 3 0.76

This analysis accounted for stream specific change in residence time of water due to the removal of macrophytes (random effect) and the stream specific variation in d13C DIC atsource (response = standardized d13C DIC). We used model simplification to estimate the Chi square (X2) statistic and its significance level (P value) for each fixed effect and theirinteraction. Var. = variance explained by the random block effect. P values < 0.05 are indicated in bold.

Figure 5. Gradients in Hawdon spring water chemistry

during the laboratory de-gassing experiment. A Changes

in d13C DIC over time (error bars equal 1SD), B changes

in pH over time (the horizontal dashed line refers to the

steady state pH calculated by PHREEQE).

Figure 6. Diel measurements of pH taken at 5 min

intervals at 1296 m downstream and at the source of the

Hawdon Valley spring between 15 and 18 January and

15 and 18 February 2010, respectively.

Carbon Cycling in Floodplain Ecosystems 593

isotopically by exchange of CO2 with the atmo-

sphere. In both streams d13C DIC was still depleted

relative to the atmosphere (-11.2 and -12.45&,

respectively) suggesting that any downstream (be-

low �1296 m) enrichment might be partly due to

mixing between-stream DIC and atmospheric CO2.

Doctor and others (2008) reported a positive shift

in d13C DIC of between 1 and 4& over tens of

meters downstream from a spring source and

attributed it to out-gassing of CO2 due to the ab-

sence of any significant photosynthetic activity. In

contrast, Hellings and others (2001) observed sea-

sonal enrichment in the order of 4–7& d13C DIC in

a highly polluted estuary and attributed it to

CO2(aq) drawdown by phytoplankton. Numerous

other studies of freshwater rivers and lakes, as well

as brackish and saline habitats, have observed a

positive correlation between photosynthetic rate

and d13C DIC (Barth and Veizer 1999; Bontes and

others 2006; Wachniew 2006; Trojanowska and

others 2008). The spring sources in our study were

dominated by the macrophyte Callitriche stagnalis,

which formed a thick mat covering the water sur-

face. Madsen (1991) showed that Callitriche copho-

carpa in Danish streams depended upon CO2(aq)

over-saturation for a sufficient supply of carbon

and a similar phenomenon might explain the

plant’s affinity for up-welling zones. In contrast,

other macrophyte taxa use morphological and

physiological adaptations (for example, the finely

dissected leaves of Myriophyllum), to increase effi-

ciency of carbon uptake, or use C4 or CAM pho-

tosynthetic pathways (Madsen 1991). Results from

Fork stream in this study suggest that macrophytes

may play an important role in regulating d13C DIC

enrichment, as isotopic enrichment equivalent to

that seen in the other streams took place in the

absence of a comparable gradient in ePCO2. Inter-

estingly, no Callitriche was seen in Fork stream,

possibly due to its low ePCO2, although a sub-

stantial cover of filamentous algae, bryophytes, and

Myriophyllum was present.

The whole stream manipulation of out-gassing

and photosynthetic uptake of CO2(aq) indicated that

close to spring sources out-gassing was the major

driver of downstream d13C enrichment. Hendy

(1971) suggested that isotopic fractionation of DIC

due to CO2(aq) out-gassing should be significant

when ePCO2 is greater than twice that of the

atmosphere. During our manipulations ePCO2

ranged from 4.5 to 8.9 and at all points in the

experimental reach there would have been a strong

chemical drive for CO2(aq) evasion, which over-

whelmed plant effects. De-gassing of Hawdon Val-

ley spring water in the laboratory suggested that all

downstream trends in pH and d13C DIC values

could be explained largely by CO2 de-gassing con-

sistent with the findings of Doctor and others

(2008). However, the incorporation of groundwa-

ter carbon into stream food webs indicates that

photosynthetic uptake of CO2(aq) does occur. At the

spring source, invariance of diel pH indicated no

change in the quantity of CO2(aq) as a result of plant

metabolism. However, 1296 m downstream in the

Hawdon spring stream we observed a large diel

shift in pH attributable to the metabolic activity of

aquatic plants. In the laboratory, in the absence of

photosynthesis, modeled and actual pH of Hawdon

spring water at steady state with the atmosphere

were 7.68 and 7.62, respectively. However, peak

daytime pH in the spring was 8.2. Thus, photo-

synthetic uptake can be expected to have an

increasing influence on CO2(aq) and d13C DIC with

increasing distance from the point of upwelling.

The downstream transition point between the

predominance of out-gassing and the increasing

influence of plant metabolism will be dynamic and

regulated by the initial ePCO2 of groundwater, the

rate of out-gassing, heterotrophic respiration of

stream biota and the biomass and metabolic activity

of aquatic plants.

Primary Producer Response

The primary food of the dominant consumers in

these springs (that is, the mayfly Deleatidium and

the snail Potamopyrgus) is biofilm (Winterbourn

2000), which showed a strong isotopic relationship

to d13C DIC and comprised 59% groundwater car-

bon. Bryophytes and filamentous algae also

showed significant relationships with d13C DIC, but

were unlikely to have been consumed directly by

any of the invertebrates used in this study (Death

2000). However, bryophytes and filamentous algae

are an integral part of the overall food web and

contribute to pools of coarse and fine particulate

organic matter (Suren and Winterbourn 1991;

Winterbourn 2000). Macrophyte d13C did not show

a significant relationship with d13C DIC. Algae vary

widely in d13C between species, water velocities,

and CO2 availability (Finlay and others 1999; Fin-

lay 2001) and similarly variable (species specific)

results have been obtained for aquatic macrophytes

(Osmond and others 1981). Furthermore, macro-

phytes in still water have been shown to have

markedly more enriched d13C values than the same

species in flowing water (France 1995). Hence,

the generally low water velocity at spring

sources compared to downstream reaches might

have contributed to the relative invariance of

594 D. P. Gray and others

macrophyte d13C values across a broad range of

d13C DIC. Some d13C enrichment of plants at spring

sources may also have occurred due to exposure to

the atmosphere during periods of low flow.

Food Web Response

An important finding of our study was that food

webs of the spring streams were tightly coupled to

the downstream trend in d13C DIC and derived at

least 50% of their carbon from groundwater. Pri-

mary consumers, predatory macroinvertebrates

and fish showed strong positive relationships with

d13C DIC. The strong relationship for fish suggests

they move around little within these streams.

However, the lack of significant downstream

enrichment of riparian spider 13C is contrary to the

assumption that Dolomedes aquaticus derives the

majority of its diet from aquatic sources (Williams

1979). Fishing spiders may be less reliant on

aquatic prey around heavily vegetated streams

such as these spring sources. Furthermore, there

may be seasonal shifts in use of aquatic versus

terrestrial prey related to the relative abundance of

each. Peak emergence times for common aquatic

insects in other streams in this part of New Zealand

are during late spring and summer (Winterbourn

and Crowe 2001), whereas our collections were

made in early spring (September).

CONCLUSION

Results of this study show that biological commu-

nities in floodplain spring streams are dependent

upon carbon derived from terrestrial vegetation in

the greater catchment. More importantly, in terms

of ecosystem processes, the study elucidates aspects

of the carbon cycle within floodplains that link

terrestrial, subterranean, and aquatic components,

a poorly understood aspect of the short-term carbon

cycle. Terrestrially fixed carbon is converted from

organic to inorganic form within soils and ground-

water, thereby becoming available to the auto-

trophic food web in the spring stream (Figure 7).

The d13C of atmospheric carbon is approximately

-8&. Photosynthesis by terrestrial C3 plants de-

pletes carbon to approximately -27& and this

carbon becomes incorporated into soils. Soil respi-

ration produces CO2, with a similar isotopic signa-

ture to its parent material, and enters groundwater.

Due to high concentrations of 13C-depleted carbon

in groundwater we observed strong biogeochemical

gradients along spring streams, gradients that re-

sulted from a combination of physical out-gassing

and metabolic activity by stream primary producers.

Upon discharge into surface streams DIC is used by

autotrophs for photosynthesis. Fractionation during

aquatic photosynthesis results in a further depletion

in 13C (reducing d13C by about 20&) prior to

incorporation into the food web. The double

depletion of d13C by photosynthesis in terrestrial

and subsequently aquatic autotrophs results in the

transmission of highly negative values through the

entire stream food web.

ACKNOWLEDGMENTS

Thanks to Robert MacLagan (University of Can-

terbury) for providing expertise in carbon hydro-

chemistry and Justin Kitto, Jo O’cock and Pete

McHugh for valuable assistance in the field. Several

high country lessees and the New Zealand

Department of Conservation permitted access to

land. Funding was provided by the Brian Mason

Scientific and Technical Trust and the FRST funded

Groundwater Ecosystems Program (NIWA). DPG

Figure 7. Conceptual

diagram of the major

pathways, forms, and

isotopic values of carbon

that influence spring

streams in a braided river

floodplain. Inset:

relationship between d13C

DIC and food web (solid

line) and ePCO2 (dashed

line) with distance from

spring source.

Carbon Cycling in Floodplain Ecosystems 595

was generously supported by a stipend from the

Miss E. L. Hellaby Grasslands Trust. Stephen A.

Norton provided helpful comments on an earlier

version of this manuscript.

REFERENCES

Allan JD, Castillo MM. 2007. Stream ecology: structure and

function of running waters. Dordrecht: Springer. 429.

Anekant MW, Bradshaw JD, Weaver S, Maas R, Ireland T, Eby

N. 2004. Provenance of the sedimentary Rakaia sub-terrane,

Torlesse Terrane, South Island, New Zealand: the use of

igneous clast compositions to define source. Sediment Geol

168:193–226.

Aravena R, Schiff SL, Trumbore SE, Dillon PJ, Elgood R. 1992.

Evaluating dissolved inorganic carbon cycling in a forested lake

watershed using carbon isotopes. Radiocarbon 34:636–45.

Arscott DB, Tockner K, Ward JV. 2000. Aquatic habitat diversity

along the corridor of an Alpine floodplain river (Fiume Ta-

gliamento, Italy). Archiv fur Hydrobiologie 149:679–704.

Atekwana EA, Krishnamurthy RV. 1998. Seasonal variations of

dissolved inorganic carbon and delta d13C of surface waters:

application of a modified gas evolution technique. J Hydrol

205:265–78.

Barth JAC, Veizer J. 1999. Carbon cycle in St. Lawrence aquatic

ecosystems at Cornwall (Ontario), Canada: seasonal and spa-

tial variations. Chem Geol 159:107–28.

Bontes BM, Pel R, Ibelings BW, Boschker HTS, Middelburg JJ,

Van Donk E. 2006. The effects of biomanipulation on the

biogeochemistry, carbon isotopic composition and pelagic

food web relations of a shallow lake. Biogeosciences 3:69–83.

Brunke M, Gonser T. 1997. The ecological significance of ex-

change processes between rivers and groundwater. Freshw

Biol 37:1–33.

Cerling TE, Solomon DK, Quade J, Bowman JR. 1991. On the

isotopic composition of carbon in soil carbon dioxide. Geo-

chim Cosmochim Acta 55:3403–5.

Clark ID, Fritz P. 1997. Environmental isotopes in hydrogeology.

Boca Raton, Florida: CRC Press/Lewis. p 328.

Collier KJ. 1987. Spectrophotometric determination of dissolved

organic carbon in some South Island streams and rivers. N Z J

Mar Freshw Res 21:349–51.

Collier KJ, Bury S, Gibbs M. 2002. A stable isotope study of

linkages between stream and terrestrial food webs through

spider predation. Freshw Biol 47:1651–9.

Craig H. 1953. The geochemistry of the stable carbon isotopes.

Geochim Cosmochim Acta 3:53–92.

Davidson GR. 1995. The stable isotopic composition and mea-

surement of carbon in soil CO2. Geochim Cosmochim Acta

59:2485–9.

Davidson EA, Janssens IA. 2006. Temperature sensitivity of soil

carbon decomposition and feedbacks to climate change. Nat-

ure 440:165–73.

Davies CW. 1962. Ion association. London: Butterworth. p 192.

Dawson JJC, Hope D, Cresser MS, Billett MF. 1995. Downstream

changes in free carbon dioxide in an upland catchment from

northeastern Scotland. J Environ Qual 24:699–706.

Death R. 2000. Invertebrate–substratum relationships. In: Collier

KJ, Winterbourn MJ, Eds. New Zealand stream invertebrates:

ecology and implications for management. Christchurch: New

Zealand Limnological Society. p 157–78.

Doctor DH, Kendall C, Sebestyen SD, Shanley JB, Ote N, Boyer

EW. 2008. Carbon isotope fractionation of dissolved inorganic

carbon (DIC) due to out-gassing of carbon dioxide from a

headwater stream. Hydrol Process 22:2410–23.

Doi H, Takagi A, Mizota C, Okano J, Nakano S, Kikuchi E. 2006.

Contribution of chemoautotrophic production to freshwater

macroinvertebrates in a headwater stream using multiple

stable isotopes. Int Rev Hydrobiol 91:501–8.

Elberling B, Ladegaard-Pedersen P. 2005. Subsurface CO2

dynamics in temperate beech and spruce forest stands. Bio-

geochemistry 75:479–506.

Elberling B, Matthiesen H. 2007. Methodologically controlled

variations in laboratory and field pH measurements in

waterlogged soils. Eur J Soil Sci 58:207–14.

Finlay JC. 2001. Stable carbon isotope ratios of river biota:

implications for energy flow in lotic food webs. Ecology

82:1052–64.

Finlay JC. 2003. Controls of stream water dissolved inorganic

carbon dynamics in a forested watershed. Biogeochemistry

62:231–52.

Finlay JC. 2004. Patterns and controls of lotic algal stable carbon

isotope ratios. Limnol Oceanogr 49:850–61.

Finlay JC, Power ME, Cabana G. 1999. Effects of water velocity

on algal carbon isotope ratios: implications for river food web

studies. Limnol Oceanogr 44:1198–203.

Fisher SG, Likens GE. 1973. Energy flow in Bear Brook, New

Hampshire: an integrative approach to stream ecosystem

metabolism. Ecol Monogr 43:421–39.

Fisher SG, Sponseller RA, Heffernan JB. 2004. Horizons in

stream biogeochemistry: flowpaths to progress. Ecology

85:2369–79.

France RL. 1995. 13C enrichment in benthic compared to

planktonic algae - foodweb implications. Mar Ecol Prog Ser

124:307–12.

Gray DP, Harding JS. 2009. Braided river benthic diversity at

multiple spatial scales: a hierarchical analysis of beta diversity

in complex floodplain systems. J North Am Benthol Soc

28:537–51.

Gray DP, Scarsbrook MR, Harding JS. 2006. Patterns in spatial

biodiversity in a large New Zealand braided river New Zea-

land. J Mar Freshw Res 40:631–42.

Hawke JD, Polaschek JM. 2005. 13C and 15N in periphyton in the

lower Selwyn River (Canterbury, New Zealand), a ground-

water-fed river surrounded by intensive agriculture. N Z Nat

Sci 30:1–10.

Helie JF, Hillaire-Marcel C, Rondeau B. 2002. Seasonal changes

in the sources and fluxes of dissolved inorganic carbon

through the St. Lawrence River—isotopic and chemical con-

straints. Chem Geol 186:117–38.

Hellings L, Dehairs F, Van Damme S, Baeyens W. 2001. Dis-

solved inorganic carbon in a highly polluted estuary (the

Scheldt). Limnol Oceanogr 46:1406–14.

Hendy CH. 1971. Isotopic geochemistry of speleothems. 1. Cal-

culation of effects of different modes of formation on isotopic

composition of speleothems and their applicability as palaeo-

climatic indicators. Geochim Cosmochim Acta 35:801–32.

Hynes HBN. 1975. The valley and its stream. Verhandlungen der

Internationale Vereinigung fur Theoretische und Angewandt

Limnologie 19:1–15.

Jepsen DB, Winemiller KO. 2007. Basin geochemistry and iso-

topic ratios of fishes and basal production sources in four

neotropical rivers. Ecol Freshw Fish 16:267–81.

596 D. P. Gray and others

Jones JB, Mulholland PJ. 1998. Influence of drainage basin

topography and elevation on carbon dioxide and methane

supersaturation of stream water. Biogeochemistry 40:57–72.

Kanduc T, Szramek K, Ogrinc N, Walter LM. 2007. Origin and

cycling of riverine inorganic carbon in the Sava River wa-

tershed (Slovenia) inferred from major solutes and stable

carbon isotopes. Biogeochemistry 86:137–54.

Karim A, Veizer J. 2000. Weathering processes in the Indus

River Basin: implications from riverine carbon, sulfur, oxygen,

and strontium isotopes. Chem Geol 170:153–77.

Keifer RH, Amey RG. 1992. Concentrations and controls of soil

carbon dioxide in sandy soils in the North Carolina coastal

plain. Catena 19:539–59.

Kendall C, Mast M, Rice K. 1992. Tracing watershed weathering

reactions with d13C. Water–rock interaction. In: Kharaka YK,

Maest AS, Eds. Proceedings of the 7th international sympo-

sium on water rock interaction. Rotterdam, USA: Balkema.

pp. 569–72.

Logan JM, Jardine TD, Miller TJ, Bunn SE, Cunjak RA, Lutca-

vage ME. 2008. Lipid corrections in carbon and nitrogen stable

isotope analyses: comparison of chemical extraction and

modeling methods. J Anim Ecol 77:838–46.

Madsen TV. 1991. Inorganic carbon uptake kinetics of the

stream macrophyte Callitriche cophocarpa Sendt. Aquat Bot

40:321–32.

Malard F, Uehlinger U, Zah R, Tockner K. 2006. Flood-pulse and

riverscapedynamics ina braided glacial river. Ecology 87:704–16.

Osmond CB, Valaane N, Haslam SM, Uotila P, Roksandic Z.

1981. Comparisons of delta C13 values in leaves of aquatic

macrophytes from different habitats in Britain and Fin-

land—some implications for photosynthetic processes in

aquatic plants. Oecologia 50:117–24.

Paetzold A, Schubert CJ, Tockner K. 2005. Aquatic terrestrial

linkages along a braided-river: riparian arthropods feeding on

aquatic insects. Ecosystems 8:748–59.

Parkhurst DL, Thorstenson DC, Plummer LN. 1990. PHREEQE.

A computer program for geochemical calculations. US Geo-

logical Survey, Water Resources Investigations 80–96, 195.

Pawellek F, Veizer J. 1994. Carbon cycle in the upper Danube and

its tributaries: d13C DIC constraints. Isr J Earth Sci 43:187–94.

Poole G. 2002. Fluvial landscape ecology: addressing uniqueness

within the river discontinuum. Freshw Biol 47:641–60.

Quinn GP, Keough MJ. 2002. Experimental design and data

analysis for biologists. Cambridge: Cambridge University Press.

p 537.

Rasmussen JB. 2010. Estimating terrestrial contribution to

stream invertebrates and periphyton using a gradient-based

mixing model for d13C. J Anim Ecol 79:393–402.

Raven JA, Osborne BA, Johnston AM. 1985. Uptake of CO2 by

aquatic vegetation. Plant Cell Environ 8:417–25.

Rounick JS, James MR. 1984. Geothermal and cold springs

faunas—inorganic carbon sources affect isotope values. Lim-

nol Oceanogr 29:386–9.

Sand-Jensen K, Frost-Christensen H. 1998. Photosynthesis of

amphibious and obligately submerged plants in CO2-rich

lowland streams. Oecologia 117:31–9.

Serrano MJG, Sanz LFA, Nordstrom DK. 2000. REE speciation in

low-temperature acidic waters and the competitive effects of

aluminium. Chem Geol 165:167–80.

Shibata H, Mitsuhashi H, Miyake Y, Nakano S. 2001. Dissolved

and particulate carbon dynamics in a cool-temperate forested

basin in northern Japan. Hydrol Process 15:1817–28.

Stanford JA. 1998. Rivers in the landscape: introduction to the

special issue on riparian and groundwater ecology. Freshw

Biol 40:402–6.

Stanford JA, Ward JV. 1993. An ecosystem perspective of allu-

vial rivers: connectivity and the hyporheic corridor. J North

Am Benthol Soc 12:48–60.

Stumm W, Morgan JJ. 1981. Aquatic chemistry: an introduction

emphasizing chemical equilibria in natural waters. New York:

Wiley Inter-science. p 780.

Suren A, Winterbourn MJ. 1991. Consumption of aquatic bry-

ophytes by alpine stream invertebrates in New Zealand. N Z J

Mar Freshw Res 25:331–43.

Taylor CB, Fox VJ. 1996. An isotopic study of dissolved inorganic

carbon in the catchment of the Waimakariri River and deep

ground water of the North Canterbury Plains, New Zealand.

J Hydrol 186:161–90.

Telmer K, Veizer J. 1999. Carbon fluxes, pCO2 and substrate

weathering in a large northern river basin, Canada: carbon

isotope perspectives. Chem Geol 159:61–86.

Tockner K, Paetzold A, Karaus U, Claret C, Zettel J. 2006.

Ecology of braided rivers. In: Sambrook Smith G, Best JL,

Bristow CS, Petts GE, Eds. Braided rivers: process, deposits,

ecology and management. Oxford: Blackwell. p 396.

Trimmer M, Hildrew AG, Jackson MC, Pretty JL, Grey J. 2009.

Evidence for the role of methane-derived carbon in a free-

flowing, lowland river food web. Limnol Oceanogr 54:1541–7.

Trojanowska A, Lewicka-Szczebak D, Jedrysek MO, Kurasiewicz

M, Wassenaar LI, Izydorczyk K. 2008. Diurnal variations in

the photosynthesis-respiration activity of a cyanobacterial

bloom in a freshwater dam reservoir: an isotopic study. Iso-

topes Environ Health Stud 44:163–75.

van der Nat D, Tockner K, Edwards PJ, Ward JV, Gurnell AM.

2003. Habitat change in braided flood plains (Tagliamento,

NE-Italy). Freshw Biol 48:1799–812.

Wachniew P. 2006. Isotopic composition of dissolved inorganic

carbon in a large polluted river: the Vistula, Poland. Chem

Geol 233:293–308.

Ward JV, Tockner K, Schiemer F. 1999. Biodiversity of flood-

plain river ecosystems: ecotones and connectivity. Regul

Rivers Res Manage 15:125–39.

Wiens JA. 2002. Riverine landscapes: taking landscape ecology

into the water. Freshw Biol 47:501–15.

Williams DS. 1979. The feeding behavior of New Zealand Dolo-

medes species (Araneae: Pisauridae). N Z J Zool 6:95–105.

Winterbourn MJ. 1997. New Zealand mountain stream com-

munities: stable yet disturbed. In: Streit B, Stadler T, Lively

CM, Eds. Evolutionary ecology of freshwater animals: con-

cepts and case studies. Basel: Birkhauser Verlag. p 31–54.

Winterbourn MJ. 2000. Feeding ecology. In: Collier KJ, Wint-

erbourn MJ, Eds. New Zealand stream invertebrates: ecology

and implications for management. Christchurch: New Zealand

Limnological Society. p 100–24.

Winterbourn MJ, Crowe ALM. 2001. Flight activity of insects

along a mountain stream: is directional flight adaptive?

Freshw Biol 46:1479–89.

Woessner WW. 2000. Stream and fluvial plain ground water

interactions: rescaling hydrogeologic thought. Ground Water

38:423–9.

Yang C, Telmer K, Veizer J. 1996. Chemical dynamics of the ‘‘St

Lawrence’’ riverine system: delta D-H2O, delta O18 (H2O),

delta C13 (DIC), delta S34 (sulfate), and dissolved Sr87/Sr86.

Geochim Cosmochim Acta 60:851–66.

Carbon Cycling in Floodplain Ecosystems 597