Chlorophyll a-specific 1 14 C , δ 13 C and δ 15 N values in stream periphyton : implications for aquatic food web studies

Periphytic algae attached to a streambed substrate (periphyton) are an important primary producer in stream ecosystems. We determined the isotopic composition of chlorophyll a in periphyton collected from a stream flowing on limestone bedrock in the Seri River, central Japan. Stable isotope ratios of carbon (δC) and nitrogen (δN) and natural radiocarbon abundances (1C) were measured in chlorophyll a (δCchl, δ Nchl and 1 Cchl) and bulk (δCbulk, δ Nbulk and 1 Cbulk) for periphyton, a pure aquatic primary producer (Cladophora sp.) and a terrestrial primary producer (Quercus glauca). Periphyton δCbulk and δCchl values did not necessarily correspond to δ Cbulk for an algal-grazing specialist (Epeorus latifolium). Periphyton 1Cchl values (−258 ‰ in April and −190 ‰ in October) were slightly lower than 1Cbulk values (−228 ‰ in April and−179 ‰ in October) but were close to the1C value for dissolved inorganic carbon (DIC; −217± 31 ‰), which is a mixture of weathered carbonates (1C=−1000 ‰), CO2 derived from aquatic and terrestrial organic matters (variable 1C) and dissolved atmospheric CO2 (1 C approximately +30 ‰ in 2013).1Cchl values were also close to1 Cbulk for E. latifolium (−215 ‰ in April and −199 ‰ in October) and Cladophora sp. (−210 ‰), whereas the 1Cbulk value for Q. glauca (+27 ‰) was closer to 1C for atmospheric CO2. Although the bulk isotopic composition of periphyton is recognised as a surrogate for the photosynthetic algal community, natural periphyton is a mixture of aquatic and terrestrial organic materials. Our results indicate that the bulk periphyton matrix at the study site consists of 89 to 95 % algal carbon (derived from C-depleted DIC) and 5 to 11 % terrestrial organic carbon (derived from C-enriched atmospheric CO2).


Introduction
The bioavailable energy in a natural ecosystem often originates from not only in situ photoautotrophs but also resources produced in other ecosystems.In most freshwater ecosystems (e.g.streams), periphytic algae attached to a substrate (periphyton) play an important role as benthic primary producers (Allan and Castillo, 2007).Terrestrial material (e.g.leaf detritus) is another resource for animals, especially in small headwater streams (Vannote et al., 1980).Although the relative importance of aquatic and terrestrial resources for food webs is a major concern in stream ecology ( Vannote et al., 1980;Junk et al., 1989;Thorp and Delong, 1994), the energy flow from periphyton to animal consumers has not yet been adequately assessed, because few studies have traced algal signatures through trophic pathways.In stream food webs, macroinvertebrates are the dominant animal consumers, and observation of their gut contents is a direct measure that can be used to trace energy flow (Winemiller, 1990;Hall et al., 2000).However, the diets of stream macroinvertebrates are sometimes too diverse to identify, and are not necessarily identical to what they actually assimilate (Whitledge and Rabeni, 1997;Finlay, 2001).
The stable isotope ratios of carbon (δ 13 C) and nitrogen (δ 15 N) have contributed to food web research over the last 40 years (after DeNiro and Epstein, 1978;Minagawa and Wada, 1984).In stream ecosystems, environmental heterogeneity within a small area (e.g.habitat variability in terms of light or flow regimes) is reflected in variations in periphyton δ 13 C (Ishikawa et al., 2012a), which often makes it difficult to estimate the relative importance of aquatic (e.g.periphyton) and terrestrial (e.g.leaf detritus) resources for Published by Copernicus Publications on behalf of the European Geosciences Union.N. F. Ishikawa et al.: Chlorophyll a-specific 14 C, δ 13 C and δ 15 N values in stream periphyton macroinvertebrates (Finlay et al., 1999;Zah et al., 2001;Doi et al., 2007;Dekar et al., 2009).
Recently, periphyton and terrestrial leaf detritus have been distinguished using natural radiocarbon abundances ( 14 C).Periphyton 14 C is often derived from aged carbon reservoirs, such as bedrocks and soils, and is relatively low compared to terrestrial leaf detritus that reflects the 14 C value for modern atmospheric CO 2 .Macroinvertebrate and fish 14 C values lie between those for periphyton and leaf detritus, indicating that 14 C can be used to estimate the energy flow in stream food webs (Ishikawa et al., 2014b).Although bulk δ 13 C, δ 15 N and 14 C values imply that the periphyton is isotopically identical to periphytic algae, it is actually a mixture of algae, heterotrophic fungi and bacteria, together with the exopolymeric substances exuded by these organisms, protozoa, small metazoa and other non-living particulate organic materials (Cross et al., 2005).All of these components may originate from different sources and have unique δ 13 C, δ 15 N and 14 C values (Hladyz et al., 2011;Ishikawa et al., 2012b;Imberger et al., 2014;Fellman et al., 2015).Therefore, the algal and non-algal taxonomic compositions of the periphyton community potentially influence its bulk isotopic composition.
Because the densities of living algae and non-algal materials (e.g.leaf detritus or animal remains) usually differ, algae and other materials in periphyton are sometimes separated by centrifuging slurry washed from stream cobbles or rocks (Hamilton and Lewis, 1992;Small et al., 2011).However, the density-separation method does not often work well when the non-algal fraction contains large amounts of dead algae, and these two components are barely distinguishable even under a microscope (Finlay, 2004).The δ 13 C and 14 C values for bulk periphyton and its potential carbon sources (e.g.particulate organic carbon: POC; dissolved organic carbon: DOC; and dissolved inorganic carbon: DIC) can be used to separate the algal carbon fraction from the non-algal carbon fraction (Fellman et al., 2015), although it is still difficult to quantitatively and directly estimate the relative abundances of the aquatic (i.e.algae) and terrestrial (i.e.leaf detritus) carbon fractions in periphyton based on their bulk isotopic compositions.
To assess the accuracy of using the bulk isotopic composition of periphyton to represent that of aquatic primary producers, we used an algal biomarker found in the periphyton matrix.Chlorophylls are the ubiquitous antenna pigments of the photoautotrophs, and the chlorophyll a concentration, in particular, has been used as an indicator of in situ primary production because it is immediately degraded in the inactive state (Carpenter et al., 1986;Amir-Shapira et al., 1987).Several previous studies have successfully used the δ 13 C, δ 15 N and 14 C values for chlorophyll a and its derivatives to understand modern environments or reconstruct palaeoenvironments (e.g.Hayes et al., 1987;Sachs et al., 1999;Ohkouchi et al., 2005Ohkouchi et al., , 2008;;Kusch et al., 2010;Tyler et al., 2010;Higgins et al., 2012).
In this study, differences in the δ 13 C, δ 15 N and 14 C values of chlorophyll a (δ 13 C chl , δ 15 N chl and 14 C chl ) and bulk (δ 13 C bulk , δ 15 N bulk and 14 C bulk ) for periphyton were compared to distinguish aquatic (i.e.algae) and terrestrial (i.e.leaf detritus) carbon fractions in the periphyton community.Because the 14 C value is internally corrected by its δ 13 C (Stuiver and Polach, 1977), 14 C chl does not depend on the isotopic fractionation that occurs during algal photosynthesis and chlorophyll a biosynthesis.Therefore, the 14 C chl value for periphyton should reflect that for photosynthetic autotrophs (i.e.primary producers), and can be used as a proxy for aquatic carbon for animals at higher trophic levels of the food web.The 14 C chl values for periphyton, DIC and an algal-grazing specialist were compared to identify trophic transfers of carbon.Pure primary producers (i.e.aquatic algae and terrestrial plants) were used to assess the potential differences in δ 13 C, δ 15 N and 14 C values between chlorophyll a and bulk cells.

Study site and sample collection
In April and October 2013, field sampling was undertaken at Kawachi in the upland of the Seri River (watershed area = 30 km 2 ; 35 • 15 N, 136 • 20 E in Shiga Prefecture, central Japan), which flows into Lake Biwa, the largest lake in Japan.The reach of the river studied flows over limestonebasalt bedrock (dominated by cobbles) and contains different light and flow environments.It has a slope of 1 to 2 % and was 10 to 15 m wide, 10 to 40 cm deep and 250 m in altitude.The dominant riparian trees are from the family Fagaceae and Taxodiaceae (higher plants with C 3 photosynthesis).Further details of this site and the DIC δ 13 C and 14 C values have been reported in Ishikawa et al. (2012b).
We randomly collected several submerged cobbles from various habitats (e.g.open/shaded and riffle/pool), which were rinsed gently with distilled water before the periphyton was removed from the cobble surface with a brush and distilled water.The resulting slurry was placed in a 100 mL polypropylene bottle, which was frozen until further processing.As reference samples of pure aquatic and terrestrial primary producers, a filamentous green alga, Cladophora sp., and several fresh leaves from the Japanese blue oak, Quercus glauca, were collected in April.Several individuals of the mayfly larva, Epeorus latifolium, were collected by hand in both April and October.The larvae of E. latifolium have highly specialised mouths for grazing (Takemon, 2005), and their amino acid δ 15 N values indicate that they are algalgrazing specialists (Ishikawa et al., 2014a).

Laboratory sample processing
All samples were lyophilised with a freeze drier (FDU-1200, Eyela, Tokyo, Japan) in the dark.The gut contents of E. latifolium larvae were removed prior to lyophilisation.The periphyton samples were ground to a fine powder with a mortar and pestle, after all large invertebrates (e.g.chironomids) had been manually removed.Cladophora sp. and Q. glauca were ground with a vibrating mill (TI-100, CMT, Fukushima, Japan).The periphyton, Cladophora sp. and Q. glauca samples were split into two vials for bulk and compound-specific isotope analyses.The vials for the bulk periphyton and Cladophora sp. were treated overnight with 1 M HCl solution to remove any carbonate; they were then washed and lyophilised again.The algal community in periphyton previously collected from the same site (November 2008) and the gut contents of E. latifolium were observed under a microscope.
Chlorophyll a was extracted using the modification of the method of Chikaraishi et al. (2005Chikaraishi et al. ( , 2007)).Briefly, the powdered periphyton, Cladophora sp. and Q. glauca were sonicated in 100 % acetone at 0 • C for 15 min, followed by liquid-liquid (water : n-hexane = 3 : 1, v/v) extraction, with NaCl salting out to remove the lipids.The n-hexane layer was extracted and dried with a stream of argon, and the precipitate (i.e.pigments) was dissolved in N ,Ndimethylformamide (DMF) after filtration using a syringe (0.50 mm × 25 mm; Terumo, Tokyo, Japan) equipped with a filter (4 mm × 0.2 µm PTFE, 100 pk; Grace Dawson Discovery Science, Maryland, USA) to remove any remaining particles.The laboratory standard for chlorophyll a was bought commercially (lot DCL2671; Wako Pure Chemical Industries, Osaka, Japan) and the standard for phaeophytin a was made by adding 1 M HCl solution to the chlorophyll a standard.Absorption spectra of our laboratory standards were consistent with those reported in the literature (Chikaraishi et al., 2007;Tyler et al., 2010).
The pigments in DMF were introduced into a highperformance liquid chromatography (HPLC) apparatus (1260 series; Agilent Technologies, California, USA), comprising a G4225A degasser, a G1312B binary pump, a G1367E autosampler, a G1316C column oven, a G1315D diode-array detector and a G1364C fraction collector.All solvents were better than HPLC grade (Wako Pure Chemical Industries).A Zorbax XDB C18 column (5 µm/4.6 × 250 mm; Agilent Technologies) and an XDB C18 guard column (5 µm/4.6 × 12.5 mm) were used in the first purification step.In the first step, the solvent gradient programme was as follows: acetonitrile : ethyl acetate : pyridine = 75 : 25 : 0.5 (v/v/v) held for 5 min, then gradually changed to 50 : 50 : 0.5 (v/v/v) in 55 min.The flow rate of the mobile phase was 1.00 mL min −1 .The column oven was set at 30 • C. We identified chlorophyll a and phaeophytin a based on their retention times and UV-visible spectral patterns, compared with those of laboratory standards (Fig. S3).
The purified chlorophyll a and phaeophytin a were collected using the fraction collector, and were dried with a stream of argon.Because phaeophytin a was more abun-dant than chlorophyll a in the April sample, we purified phaeophytin a together with chlorophyll a and combined them for the isotope measurements.The C and N isotopic compositions of phaeophytin a are theoretically identical to those of chlorophyll a, because phaeophytin a is an early degradation product of chlorophyll a, and no C or N atoms are replaced in this step.Each fraction was dissolved in DMF and re-introduced into the HPLC apparatus.A PAH column (5 µm/4.6 × 250 mm, Agilent Technologies) and a PAH guard column (5 µm/4.6 × 12.5 mm) were used in the second purification step.In the second step, the solvent gradient programme was as follows: acetonitrile : ethyl acetate : pyridine = 80 : 20 : 0.5 (v/v/v) held for 5 min, then gradually changed to 0 : 100 : 0.5 (v/v/v) in 35 min.The flow rate of the mobile phase was 1.00 mL min −1 .The column oven was set at 15 • C.After the second step, the fractions of chlorophyll a and phaeophytin a were dried and washed with water : n-hexane (3 : 1, v/v).The n-hexane layer was carefully extracted, dried again and frozen until the isotope measurements were made.The abundances of chlorophyll a and phaeophytin a were estimated using conversion formulae between the absorbance at 660 nm and the dry weights of the laboratory standards.The dried chlorophyll a and phaeophytin a were dissolved in dichloromethane and then transferred to tin capsules for δ 13 C and δ 15 N measurements or to quartz tubes for 14 C measurements.The tin capsules and quartz tubes were dried again prior to measurements.
30 min and at 850 • C for 2 h.The CO 2 gas was cryogenically purified in a vacuum line and reduced to graphite with hydrogen and an iron catalyst at 550 • C for 10 h.The14 C values for the bulk samples and chlorophyll a samples were measured with an accelerator mass spectrometer (AMS) at the Institute of Accelerator Analysis (Kanagawa, Japan; AMS lab code IAAA) and at the Atmosphere and Ocean Research Institute, University of Tokyo (Chiba, Japan; AMS lab code YAUT), respectively.The 14 C (‰) value was defined as follows (Stuiver and Polach, 1977): The 14 C value of the international standard (oxalic acid) takes into account radioactive decay since 1950 (Stuiver and Polach, 1977).The 1σ analytical precision of the 14 C measurements was within 3 ‰ for bulk and 8 ‰ for chlorophyll a.The HPLC procedural blank for carbon (e.g.potential contamination by column breeding), assessed with an elemental analyser, was below the detection limit (< 0.177 µg C), which represents less than 0.2 % carbon in the purified chlorophyll a molecules used for the AMS measurement.
To determine the carbon transfer pathway in this stream ecosystem, the δ 13 C and 14 C values for all samples were compared with those for DIC, DOC and POC collected at the same site in the Seri River in 2009 to 2010 (Ishikawa et al., 2012b(Ishikawa et al., , 2015)).

Sample observations
Microscopic observations show that diatoms and cyanobacteria are the dominant photoautotrophs in the periphyton community at the study site (Fig. S1).Both the periphyton and gut contents of E. latifolium consisted of not only algal cells but also amorphous and unidentified particles (Fig. S2).The exuvium of small invertebrates (approximately 500 µm) was found in the periphyton matrix (Fig. S2a), the isotopic composition of which would have differed from that of pure algae.The UV-visible spectra show different compositions of photosynthetic pigments between April and October.Chlorophyll a (Mw, 892.5) and phaeophytin a (Mw, 870.6; the Mg atom is replaced by two H atoms in the centre of the tetrapyrrole ring of the chlorophyll a molecule) were the dominant pigments in the periphyton matrix in both April and October (Fig. S3).The combined abundance of chlorophyll a and phaeophytin a per unit dry weight was greater in October than in April, indicating that the algal biomass of the periphyton community was greater in October than in April (Table S2).

13 C composition
The periphyton δ 13 C bulk and δ 13 C chl values were −20.7 and −20.0 ‰, respectively, in April, and −26.2 and −25.2 ‰, respectively, in October (Fig. 1).The algal-grazer E. latifolium δ 13 C bulk values were −26.6 and −26.5 ‰ in April and October (Fig. 1), respectively.In October, δ 13 C bulk and δ 13 C chl values for periphyton were close to the E. latifolium δ 13 C bulk value.In contrast, neither the periphyton δ 13 C bulk Biogeosciences, 12, 6781-6789, 2015 www.biogeosciences.net/12/6781/2015/nor δ 13 C chl value was close to the E. latifolium δ 13 C bulk value in April.This is partly because the periphyton δ 13 C bulk values vary from −32 to −16 ‰ among the stream habitats (e.g.open/shaded and riffle/pool) of this study site owing to the variable isotopic fractionation between DIC and algae (Ishikawa et al., 2012b).Such a large variation in periphyton δ 13 C bulk values on a small spatial scale may cause inconsistency between the δ 13 C values for periphyton (primary producers) and E. latifolium (primary consumers).
A mismatch between the δ 13 C bulk values for periphyton and grazers is often observed (Dekar et al., 2009), although 13 C is not enriched through the trophic levels (Vander Zanden and Rasmussen, 2001).There are four independent scenarios that explain our δ 13 C results.Firstly, E. latifolium assimilates the 13 C-depleted fraction in periphyton.Secondly, E. latifolium assimilates the terrestrial organic matter, which is more 13 C-depleted than the periphyton.Thirdly, the periphyton δ 13 C bulk and δ 13 C chl values varied by 6 ‰, whereas the E. latifolium δ 13 C bulk values did not change greatly between April and October, suggesting that primary consumers integrate temporal fluctuations in the δ 13 C values for primary producers.Finally, the δ 13 C chl value is not a reliable proxy for the δ 13 C of bulk algae, because the δ 13 C chl value is affected by the isotopic fractionation that occurs during chlorophyll a biosynthesis.To provide a more precise estimate of algal carbon, the 14 C chl signature is useful because it is corrected for isotopic fractionation by δ 13 C in Eq. ( 1) (Stuiver and Polach, 1977).
δ 13 C bulk and δ 13 C chl values were −23.0 and −24.7 ‰, respectively, for Cladophora sp. and −30.9 and −32.0 ‰, respectively, for Q. glauca (Fig. 1).The δ 13 C chl value for primary producers is controlled by the δ 13 C value for their carbon source (i.e.DIC for Cladophora sp. and atmospheric CO 2 for Q. glauca) and by internal isotopic fractionation between bulk cells and chlorophyll a molecules.Sachs et al. (1999) reported that δ 13 C chl values for a cultivated green alga Dunaliella tertiolecta were 0.5 to 4.0 ‰ lower than those for their bulk cells, which is consistent with our Cladophora sp.data.Chikaraishi et al. (2005) reported the same δ 13 C bulk value (−30.9 ‰) for the fresh leaves of the Mongolian oak Q. mongolica as for our Q.glauca data.In contrast, in this study, the Q. glauca δ 13 C chl value (−32.0 ‰) was lower than that for Q. mongolica (−29.2 ‰) reported in Chikaraishi et al. (2005).

15 N composition
The periphyton δ 15 N bulk and δ 15 N chl values were −5.7 and −1.5 ‰, respectively, in April, and −1.7 and +0.5 ‰, respectively, in October (Fig. 1).The algal-grazer E. latifolium δ 15 N bulk values (−3.9 ‰ in April and +1.4 ‰ in October) were 1.8 to 2.9 ‰ higher than the periphyton δ 15 N bulk values.The δ 15 N bulk and δ 15 N chl values were −4.3 and −6.0 ‰, respectively, for Cladophora sp. and −0.8 and −0.2 ‰, respectively, for Q. glauca (Fig. 1).Sachs et al. (1999) reported that the δ 15 N chl values were 2 to 9 ‰ lower than the δ 15 N bulk values for phytoplankton because of the isotopic fractionation that occurs during chlorophyll a biosynthesis.Kennicutt et al. (1992), on the other hand, reported that the δ 15 N chl values were relatively close to the δ 15 N bulk values for terrestrial C 3 plants.Therefore, the relationships between δ 15 N bulk and δ 15 N chl values for Cladophora sp. and Q. glauca are consistent with those reported in previous studies.In contrast, the periphyton δ 15 N chl values were 2.2 to 4.2 ‰ higher than their δ 15 N bulk values.This result might be attributable to the presence of cyanobacteria (e.g.Oscillatoria sp. or Homoeothrix sp., Fig. S1) in the periphyton community, because the δ 15 N bulk and δ 15 N chl values for cyanobacteria are usually different from those for algae (Beaumont et al., 2000).
The 14 C bulk and 14 C chl values were −199 ± 2.7 and −210 ± 6.8 ‰, respectively, for Cladophora sp. and +27 ± 2.3 and −10 ± 7.3 ‰, respectively, for Q. glauca (Fig. 1).The Q. glauca 14 C bulk value was not greatly different from the global mean 14 C value for atmospheric CO 2 in 2013 (approximately +30 ‰; Levin et al., 2013).However, chlorophyll a contains only 0.07 % of the carbon in bulk leaves, and Q. glauca synthesises chlorophyll a using not only atmospheric CO 2 but also aged ( 14 C-depleted) CO 2 and/or organic matter derived from other carbon sources.A candidate source is soil, as variable 14 C values for soil organic matter have been reported previously (Trumbore and Zheng, 1996;Koarashi et al., 2009).Various terrestrial plants can incorporate soil-derived carbon through their roots (Brüggemann et al., 2011;Bloemen et al., 2013).While there is no evidence that 14 C-depleted organic carbon is transferred from soil to plants, Q. glauca and probably other terrestrial plants may be able to make the chlorophyll a molecule using recycled phytol, as reported for Arabidopsis seedlings (Ischebeck et al., 2006).Chlorophyll a biosynthesis has multiple channels to acquire carbon, which does not necessarily originate from atmospheric CO 2 .The Q. glauca 14 C chl value will be different from its 14 C bulk value if Q. glauca collects phytol or its precursors from the soil.Future attention should be paid to plants' uptake of soil carbon in order to understand the carbon allocation in plants and the global carbon budget in the terrestrial biosphere.
To estimate the relative abundances of aquatic (e.g.algae) and terrestrial (e.g.leaf detritus) carbon fractions in periphyton bulk matrix, a separate two-source mixing model was applied to each of the April and October samples.We assumed that the periphyton 14 C chl value (−258 in April and −190 ‰ in October) and the Q. glauca 14 C bulk value (+27 ‰ in both April and October) represent the aquatic and terrestrial endmembers, respectively.Therefore, the periphyton 14 C bulk values in April (−228 ‰) and October (−190 ‰) were explained by both seasonal variation in the aquatic endmember and relative contributions of the aquatic and terrestrial carbon fractions to the periphyton bulk matrix.The results of the mixing model show that the periphyton bulk matrix consisted of 89 (April) to 95 % (October) aquatic carbon and 5 (October) to 11 % (April) terrestrial carbon.The E. latifolium 14 C bulk values (−215 ± 2.3 in April and −199 ± 2.2 ‰ in October) were within the range of the periphyton 14 C values (Fig. 1).The April E. latifolium 14 C bulk value was closer to the periphyton 14 C bulk value than to its 14 C chl value, suggesting that E. latifolium assimilates not only 14 C-depleted aquatic sources but also 14 C-enriched terrestrial sources in April.In contrast, the October E. latifolium 14 C bulk value was closer to the periphyton 14 C chl value than to its 14 C bulk value, suggesting that E. latifolium primarily assimilates aquatic sources in October.This seasonal variation may be attributed to the higher chlorophyll a abundance per unit dry weight in October, and/or to the higher terrestrial flux associated with the input of snowmelt in April.

Implications of this study
Previous studies have assumed that the isotopic compositions of bulk periphyton are identical to those of periphytic algae, without direct evidence.Regarding the identification of an aquatic baseline for stream food webs, our δ 13 C chl and 14 C chl data indicate that the periphyton δ 13 C bulk and 14 C bulk values can be approximated as those for the photosynthetic algal community in periphyton (Fig. 3).However, there remain some uncertainties in our data, such as the results that the δ 15 N chl values were higher than the δ 15 N bulk values in periphyton and that the 14 C chl values were slightly lower than the 14 C bulk values.These results do not indicate that the isotopic compositions of bulk periphyton are completely consistent with those of algae.Bulk isotope analysis may underestimate the importance of aquatic production for stream food webs, especially in less productive streams where the terrestrial detritus is more abundant than the algae/cyanobacteria in the periphyton.On the other hand, chlorophyll a-specific 14 C, δ 13 C and δ 15 N values are useful tracers for precisely estimating the sources of carbon and nitrogen in stream ecosystems, in which heterogeneous resources (e.g.aquatic and terrestrial organic matters) are mixed.
Compound-specific stable isotope and radiocarbon analyses are promising tools for the precise estimation of the sources, dynamics and turnover of various organic molecules (Hayes et al., 1987;Eglinton et al., 1996;Jochmann and Schmidt, 2012;Ohkouchi et al., 2015).Chlorophyll a is a unique biomarker of in situ photoautotrophs and is more accurate than other biochemical compounds (e.g.lipids and amino acids) because it is immediately degraded in the inactive state (Carpenter et al., 1986;Amir-Shapira et al., 1987;Matile et al., 1996).However, a pitfall may exist in the chlorophyll a recycling system.Some previous studies have suggested that terrestrial plants and cyanobacteria have a salvage pathway of phytol in chlorophyll a biosynthesis (Ischebeck et al., 2006;Vavilin and Vermaas, 2007).The isotopic composition of chlorophyll a is determined by the rela-tive contributions of de novo synthesis and the recycling system to all chlorophyll a molecules.These contributions can be estimated by separate measurements of the isotopic compositions of each of chlorophyll a and its bounded phytol (e.g.Chikaraishi et al., 2005).
The isotopic composition of chlorophyll a can be used in not only stream ecosystems but also coastal ecosystems, where benthic biofilms (i.e.mixtures of algae and other heterotrophs) are important food sources for invertebrates, fish and birds (Kuwae et al., 2008(Kuwae et al., , 2012)).Furthermore, primary production in the ocean and lakes is currently estimated using the bulk isotopic composition of particulate organic matter, which is a mixture of not only phytoplankton but also heterotrophs and other organic materials derived from various sources.The use of chlorophyll a-specific isotopic compositions can avoid the "mixing effect" on the estimation of in situ primary production, and can provide more precise data for biogeochemical cycling of materials and energy.We conclude that future studies should assess the degree to which δ 13 C, δ 15 N and 14 C values differ between bulk and chlorophyll a in primary producers collected from multiple ecosystems.
The Supplement related to this article is available online at doi:10.5194/bg-12-6781-2015-supplement.
Author contributions.N. F. Ishikawa conceived the study design and conducted fieldwork.N. F. Ishikawa and H. Suga conducted pigment purification using HPLC.N. O. Ogawa conducted δ 13 C and δ 15 N analyses using EA/IRMS.M. Yamane and Y. Yokoyama conducted 14 C analysis using AMS.All authors participated in discussion.N. F. Ishikawa and N. Ohkouchi wrote the manuscript.

−Figure 3 .
Figure 3. Schematic view of the carbon cycle at the study site (Seri River) constrained by δ 13 C and 14 C.