Biogeochemical properties of fine particulate organic matter as an indicator of local and catchment impacts on forested streams


  • Takashi Sakamaki,

    Corresponding author
    1. Department of Forest Sciences, University of British Columbia, 3041-2424 Main Mall, Vancouver, BC V6T 1Z4, Canada
    2. Transdisciplinary Research Organization for Subtropics and Island Studies, University of the Ryukyus, 1 Senbaru, Nishihara, Okinawa 903-0213, Japan
    Search for more papers by this author
  • John S. Richardson

    1. Department of Forest Sciences, University of British Columbia, 3041-2424 Main Mall, Vancouver, BC V6T 1Z4, Canada
    Search for more papers by this author

Correspondence author. E-mail:


1. The magnitude and spatial scales of human effects on stream habitats need to be correctly measured to achieve sound ecosystem management. We investigated whether the biogeochemical properties of fine particulate organic matter (FPOM) might be indicative of reach-scale vs. catchment-scale effects on forested stream ecosystems.

2. Along each of three forested streams, we established 4–5 sampling stations at 0·6–2·0-km intervals, which represent a range of local riparian forest conditions (e.g. vegetation types and forestry practices). At each station, rock biofilm (considered as a representative of stream-origin POM), FPOM in sediment (FPOMS) and suspended in water (FPOMW) and three species of benthic macroinvertebrates were collected during a summer low-flow period.

3. Measures of δ13C, C:N and chlorophyll a:C for FPOMW, FPOMS and biofilm were longitudinally heterogeneous, reflecting the reach-scale, local environment. δ13C, C:N and chlorophyll a:C of FPOMS were significantly related to irradiance and streambed coarse particulate organic matter (CPOM) abundance, suggesting that the relative contribution of in-stream primary production and CPOM breakdown are dominant controls of FPOM properties.

4. In redundancy analysis, variations in FPOM properties were principally explained by local environmental factors correlated with stream size and/or longitudinal position of the sampling stations (irradiance, streambed CPOM, stream gradient, discharge, riparian vegetation). No significant effect of riparian forestry activities (logged, riparian reserve or no harvest in the past 75 years) was found on FPOM properties.

5. Resource use by primary consumers was species dependent. δ13C of larval Lepidostoma roafi depended on the relative abundance of stream-origin POM in pools of stored FPOM. Selective intake/assimilation of stream-origin POM by this species was probably enhanced when stream-origin POM was abundant. δ13C of larval Despaxia augusta or larval Paraleptophlebia spp. were not related to FPOMSδ13C, suggesting no linkage with FPOM composition.

6.Synthesis and applications. The tight linkages between biogeochemical properties of FPOM, resource use by primary consumers, and reach-scale environment indicate that local effects were greater than signals transmitted from upstream in this study. The use of in-stream FPOM properties can help managers to measure reach-scale effects of environmental changes on forested stream habitats.


Human impacts on the structures and processes of forested stream ecosystems have been reported from local to catchment scales. Modifications of riparian vegetation by logging alter organic matter resource inputs and the local physical environment (i.e. light and temperature) in adjacent streams. This further affects in-stream autotrophic and heterotrophic processes, and community structures (e.g. Wallace et al. 1999; Kiffney, Richardson & Bull 2003). The effects of forestry practices on in-stream invertebrate communities have also been detected at the catchment scale (Kreutzweiser et al. 2008; Zhang, Richardson & Pinto 2009). Catchment-scale impacts can be considered as cumulative effects transmitted from upstream reaches. To achieve sound ecosystem management for forested streams, the heterogeneity and continuity of stream habitats and the spatial scales of ecological responses to natural or human-caused environmental variations need to be well understood. Practical methods are needed to assess the responses of forested stream ecosystems to environmental changes over different spatial scales.

Terrestrially derived coarse particulate organic matter (CPOM; >1 mm), such as leaves and wood, is the predominant organic matter source to forested streams. It is further processed into fine particulate organic matter (FPOM; <1 mm) through physical and biological breakdown (e.g. Wallace et al. 1991; Webster et al. 1999). In forested streams, the advective transport flux and standing stock of FPOM have been reported to be higher than, or comparable with those of CPOM (e.g. Kiffney, Richardson & Feller 2000; Thompson & Townsend 2005). In addition, stream-origin POM (e.g. algal, bacterial and fungal biomass) is also included in pools of stored FPOM. Overall, FPOM quality and quantity potentially reflect various environmental factors linking with terrestrial organic matter input and in-stream production. However, longitudinal variations of biogeochemical properties of FPOM in forested streams have not been examined in terms of their spatial scale or linkage with local environments.

In general, food webs of small, forested streams are based mainly on terrestrial-origin POM (e.g. Wallace et al. 1999), while those of larger streams depend more on algae (e.g. Thorp et al. 1998). However, even in small streams, organisms may preferentially utilise algae despite its relatively low abundance (e.g. McCutchan & Lewis 2002; England & Rosemond 2004). In addition, basal resource composition affects food web structure; algal-based food webs may have longer food chains compared with detritus-based food webs (Tavares-Cromar & Williams 1996; Thompson & Townsend 2005). The coexistence of detritus- and algal-based energy flows theoretically stabilizes the food web (Moore et al. 2004). Thus, the origin and composition of POM is an important consideration for the conservation and management of aquatic ecosystems (e.g. McCutchan & Lewis 2002; Sakamaki, Shum & Richardson 2010).

We investigated biogeochemical properties of in-stream FPOM (i.e. C and N stable isotope ratios, C:N and chlorophyll a:C) in three small streams located in a forest-dominated area of southwestern British Columbia, Canada. We examined the applicability of the FPOM properties as an indicator to assess responses of forested stream ecosystems to natural and forestry-related variations of environmental conditions. Specifically, our study objectives were to (i) measure spatial heterogeneity of FPOM properties along the streams, (ii) explore which local environmental factors significantly affect FPOM properties, (iii) test whether resource use by benthic primary consumers is linked with the FPOM properties, and (iv) examine whether FPOM properties can be a useful indicator to assess local- and catchment-scale impacts of natural- and forestry-related environmental changes in forested stream ecosystems.

Materials and methods

Study Sites

Field sampling was conducted in three streams (C Creek, Donegani Creek and Spring Creek) in the UBC Malcolm Knapp Research Forest (49°16′N, 122°34′W), which is located about 60 km east of Vancouver, British Columbia, Canada. All the streams flow through a mountainous, forested area which is dominated by three conifer species, Tsuga heterophylla (western hemlock), Thuja plicata (western red cedar) and Pseudotsuga menziesii (Douglas-fir). The major riparian species are Alnus rubra (red alder), Acer circinatum (vine maple) and Rubus spectabilis (salmonberry). The forest was logged in the early 1900s, and the current second-growth stand was fire-initiated in 1931 and was c. 75 years old at the time of our study.

Each stream has heterogeneous riparian conditions because of experimental forestry activities (clear-cut, thinned, 10-m riparian reserves, 30-m riparian reserves or no forest harvest in the past 75 years) as well as natural variation in vegetation (conifer-dominant or conifer-deciduous mixed). The thinning site had 50% tree removal based on basal area. The experimental logging treatments were conducted in winter 1998. In each of the three streams, 4–5 stations for sampling and measurements were set at 0·6–2·0-km intervals within the first- to third-order reaches (Table 1). Reaches directly below major tributaries were avoided. The sampling stations were chosen to represent different types of riparian condition based on aerial photographs (taken in September 2006; Google Earth), maps of primary vegetation and stand age (surveyed in 1989; UBC research forest) and preliminary visual observations. Each station was located within a relatively homogeneous reach of at least a few hundred metres with a specific type of stream and riparian condition. This design enabled us to test for the responses of biogeochemical properties of in-stream FPOM to local environments.

Table 1.   Physical properties and riparian conditions of sampling stations
Stream/stationDistance (km)Altitude (m)Overstorey vegetationUnderstorey vegetationAge class (years)Riparian forestry activityWetted width (m)Discharge (L s−1)Relative light intensity (%)
  1. Distance is shown as the distance from the uppermost station of each stream. Means ± SE are shown for wetted width (n = 4–6), discharge (n = 3), and relative light intensity (n = 6). For riparian overstorey vegetation, forest types and 1st- and 2nd- dominant tree species are shown. For riparian understorey vegetation, the density is shown as H, high or L, low. The forest age class is based on a 1989 survey by UBC Malcolm Knapp Research Forest. The number of the station name is a rank based on discharge; the smaller number indicates a smaller discharge.

  2. Con, coniferous; Mix, coniferous-deciduous mixed; WH, western hemlock; RC, western red cedar; RA, red alder; BP, balsam-poplar; DF, Douglas-fir.

C Creek
 C10280Con (WH, RC)L79–98No1·0 ± 0·10·9 ± 0·53·7 ± 1·4
 C20·7200Con (RC, WH)L79–98Reserves (30 m)1·1 ± 0·23·2 ± 1·810·8 ± 6·5
 C31·3140Mix (WH, RA)H79–98Reserves (10 m)2·1 ± 0·314·1 ± 5·147·0 ± 22·8
 C71·8110Mix (WH, BP)H79–98No1·4 ± 0·221·3 ± 5·922·8 ± 4·3
Donegani Creek
 D40310Con (WH, DF)L139–158No4·0 ± 0·615·4 ± 1·36·0 ± 3·2
 D50·7220Con (WH, DF)H139–158No2·6 ± 0·116·9 ± 7·57·0 ± 1·7
 D81·5110Mix (RA, DF)H39–58No1·6 ± 0·344·3 ± 5·28·2 ± 4·6
 D92·540HLogged (clear-cut)3·1 ± 0·5109·4 ± 12·139·6 ± 13·5
Spring Creek
 S60420Con (RC, WH)L59–78Logged (thinning)1·8 ± 0·318·9 ± 14·74·3 ± 1·7
 S102300Mix (RA, WH)H39–58Reserves (30 m)3·4 ± 0·5164 ± 11413·6 ± 11·1
 S113·3200Mix (RC, RA)L119–138No3·1 ± 0·6218 ± 1500·8 ± 0·5
 S124130Mix (RC, RA)H79–98No4·5 ± 0·4225 ± 9732·5 ± 29·2
 S134·980Con (DF, WH)L119–138No3·2 ± 0·8455 ± 30243·6 ± 21·1

Sampling and Sample Processing

All field sampling was conducted during a summer low-flow period (between 15 June and 30 July 2007). Sixteen-litre stream water samples for FPOM were collected from each sampling station three times, i.e. once in each of the periods 15–30 June, 1–15 July, and 15–30 July. All stations within one stream were sampled on the same day. Water samples were not collected for several days after rain. Known volumes of water samples were filtered through pre-combusted GF/F filters (Whatman, Kent, UK) and FPOM suspended in stream water (FPOMW) was collected for analyses of C, N and chlorophyll a (chl-a) concentrations, and stable isotope ratios of C and N.

FPOM in sediment (FPOMS), biofilm, benthic macroinvertebrates, CPOM and riparian soil were also collected. Each of those sample types was collected once per site (riparian soil: 15–30 June, FPOMS: 1–15 July, invertebrates, CPOM and biofilm: 15–30 July), and all stations within one stream were sampled on the same day, to minimise possible confounding effects of temporal variation. At each station, fine sediment from the streambed was collected from five pools within a 50–100-m reach at 10–20-m intervals. The 0–3-cm layer of sediment was collected using a core sampler with a 4 cm diameter. In the laboratory, each sediment sample was dispersed in a plastic container filled with 500 mL of deionised water, and shaken for 20–30 s. The water in the container was screened through a 1-mm sieve, and known volumes of water were filtered through pre-combusted GF/F filters for analyses of stable isotope ratios of C and N, and also the relative amount of C, N and chl-a.

For biofilm sampling, several cobbles (c. 10–20 cm in diameter) were collected at each station from three pools and three riffles along the 50–100 m reach. In the field the samples from each pool/riffle were separately brushed, and the biofilm was rinsed into a plastic container filled with 300–500 mL deionised water. In the laboratory, the samples were shaken for 20–30 s, and known volumes of water were filtered through pre-combusted GF/F filters for analyses of stable isotope ratios of C and N, and the relative amount of C, N and chl-a.

At each station, benthic macroinvertebrates and streambed CPOM were collected using a Surber sampler (30 × 30 cm frame with 250 μm mesh net; BioQuip, Rancho Dominguez, CA, USA) from three pools and three riffles along the 50–100-m reach. In the laboratory, the sample was washed on a 500-μm sieve, and all invertebrates that were visible to the naked eye were picked with tweezers. Then larvae of three species found most commonly in this study, i.e. Despaxia augusta (Plecoptera, Leuctridae), Lepidostoma roafi (Trichoptera, Lepidostomatidae) and Paraleptophlebia spp. (Ephemeroptera, Leptophlebiidae) were treated further for stable isotope analysis. D. augusta and L. roafi are categorised as shredders, i.e. leaf detritus consumers, while Paraleptophlebia spp. is generally considered to be a collector (Richardson 1991; Merritt & Cummins 1996). All remnants of the Surber sample were washed on a 1-mm sieve with tap water and retained to determine the mass of CPOM.

To estimate the biogeochemical properties of terrestrial POM entering streams from adjacent riparian areas, riparian soil was collected from three plots along the 50–100-m reach of each station. In each plot, six core samples of the 0–4-cm layer (4 cm diameter) were randomly collected from an area of c. 20 m2, and then mixed to make a composite sample. When the ground in the riparian zone was covered with large-size debris (e.g. leaves and wood debris), the debris was removed before soil samples were collected. In the riparian zone, leaves of plants/trees and the bark of trees were also collected to determine stable isotopic signatures.

Site Characteristics

At every water-sampling event, stream discharge and light intensity were measured. The discharge measurement was based on the salt dilution method. Following the salt injections, the temporal variation of salt concentration in stream water was recorded by a conductivity meter (Cond340i, WTW, Weilheim Germany). At three pools and three riffles along the 50–100-m reach of each site, the irradiance was measured using a quantum meter (BQM-S; Apogee Electronics, Santa Monica, CA, USA), which detects photosynthetically active radiation (i.e. 400–700 nm). The irradiance was also measured at unshaded reference sites <10 min prior to light measurement at each sampling station. The percentage of relative light intensity, defined as (the irradiance of station/the irradiance of unshaded reference site) × 100 (%), was used for data analyses. At each station, the wetted width of the stream was also measured at 4–6 cross-sections along the 50–100-m reach.

The first and second most dominant, overstorey tree species were determined based on visual estimation of proportional coverage along a 15-m strip of the streamside area of each station. If both dominant species were coniferous, the station was considered coniferous. If either dominant species were deciduous, the station was categorised as coniferous-deciduous mixed (at no site were both dominant trees deciduous). In addition, based on visual observation, the density of riparian understorey vegetation was categorised into two levels, high or low.

Chemical Analysis

In the preparation for analyses of C and N stable isotopes and C and N contents, FPOMW, FPOMS, biofilm, riparian soil and terrestrial POM samples were treated with 10% HCl to remove carbonate, rinsed with deionised water and dried at 60 °C. Invertebrate samples for C and N stable isotopes were not acidified, as the effect of carbonate for freshwater invertebrate analysis is generally expected to be minor. The head and thorax of invertebrates were dissected and analysed to avoid contamination of gut contents. The C and N stable isotopic compositions of these samples were determined by a continuous flow, isotope ratio, mass spectrometer system (Hydra 20/20; Europa Scientific, Crewe UK) at the Stable Isotope Facility, University of California Davis. The isotopic compositions were reported as parts per thousand deviations (‰) from Pee Dee Belemnite for carbon (δ13C) and from air for nitrogen (δ15N). In this analysis, C and N contents of the samples were also obtained as a percentage of dry mass.

To determine chl-a amount in FPOMW, FPOMS and biofilm, pigments on GF/F filters were extracted in 12 mL of 90% acetone for 24 h, and chl-a fluorescence was measured following the Welschmeyer non-acidification method (Welschmeyer 1994; TD-700; Turner Designs, Sunnyvale, CA, USA). CPOM was quantified by determining dry mass loss by combustion at 550 °C for 2 h.

Data Analysis

The variances of the biogeochemical properties of FPOMS because of the effects of stream (n = 3), station (n = 4–5 in each stream) and sampling point (n = 5 at each station) were compared by anova. The stations were nested within the streams, and the points were nested within stations. This anova design enabled us to compare the variance of the biogeochemical properties of FPOMS at the three different spatial scales. For FPOMW, the effect of three sampling occasions was tested instead of the within-station difference (i.e. no comparison between points). Where the factor was sampling occasion, the repeated-measures design was not applied, because the degrees of freedom were not sufficient for that design and the main focus of this analysis was on comparisons of variances at the different spatial scales. For biofilm, the effect of pool/riffle was also added to the three different spatial scales tested for FPOMS.

Two types of redundancy analyses (RDA), a full-factor RDA and six single-factor RDAs, were conducted. Full-factor RDA was used to understand relationships between biogeochemical characteristics of in-stream FPOM and local environmental factors of sampling stations; δ13C, δ15N, C:N and chl-a:C of FPOMS and FPOMW were dependent variables, and discharge, stream gradient, streambed CPOM abundance, irradiance, riparian vegetation and riparian forestry activities were explanatory variables. Categorical data were used for riparian vegetation (coniferous and coniferous-deciduous mixed) and riparian forestry activities (logging including clear-cut and thinning, buffer reserve and no forestry activity). None of the six explanatory variables were removed in backward elimination based on Akaike’s Information Criterion (AIC). Variable inflation factors (VIF) were low (<5·1) for all environmental variables used. Single-factor RDAs examined variances of FPOM properties explained solely by each environmental variable without covariables.

Multiple regression analyses were used to examine the relative importance of potential mechanisms directly controlling each of the specific biogeochemical properties of FPOMS and FPOMW. These regression analyses included the following explanatory variables: irradiance, streambed CPOM abundance and biogeochemical signatures of biofilm and riparian soil. Irradiance and streambed CPOM abundance were selected to represent the effects of the quantity of algal production and CPOM breakdown on bulk FPOM properties. The biogeochemical signatures of biofilm and riparian soil FPOM were selected to represent the effects of the quality of POM available for in-stream FPOM. The relative importance of those explanatory variables was examined by model selection based on the response of AIC to the addition or removal of the explanatory variables. VIFs were low (<2·7) for the explanatory variables in all the cases of regression analyses. In addition, univariate regression analyses were used to test whether the spatial variations of invertebrate δ13C were explained by FPOMSδ13C and/or biofilm δ13C.

The log10- and square-root-transformations were applied for the data of streambed CPOM abundance and relative light intensity, respectively, to improve their normality in advance of all the statistical analyses. R version 2.10.1 with packages, Vegan, MASS and Car, was used for RDA and multiple regression analyses. For the other statistical analyses, sas version 9.1 (SAS Inc., Cary, NC, USA) was used. proc mixed was used for the anovas.


Measures of δ13C, δ15N, C:N, and chl-a:C for FPOMW, FPOMS, and biofilm were longitudinally heterogeneous (Fig. 1). Most of the biogeochemical properties of FPOMW, FPOMS and biofilm were significantly different among stations, and/or the largest portion of variance was explained by stations (Table 2). δ13C and chl-a:C of FPOMS were marginally different between the stations (= 0·061 and 0·081, respectively). The biogeochemical signatures or concentration of FPOMW did not differ statistically among the three sampling occasions.

Figure 1.

 Longitudinal variations of chemical properties of FPOMW, FPOMS and biofilm in the streams: C Creek, open circle; Donegani Creek, grey square; Spring Creek, solid triangle. The horizontal axis is the average discharge of three measurements at each station. The positions of sampling stations in each stream are represented by discharge rather than using linear distance; the larger discharges indicate downstream positions. The points and error bars indicate mean and SE (FPOMW: n = 3, FPOMS: n = 5, biofilm: n = 6), respectively. The concentrations of particulate organic carbon in stream water are also shown (mean and SE from three sampling occasions).

Table 2. anova results for the effects of sampling locations at different spatial scales (streams, stations nested under streams, points nested under stations), occasions (only for FPOMW) and geomorphological structures (pool or riffle; only for biofilm) on biogeochemical properties of FPOMW, FPOMS and biofilm
DependentSource of variation
  1. FPOM, fine particulate organic matter; POC, particulate organic carbon; PON, particulate organic nitrogen; Chl-a, chlorophyll a.

  2. The effects of those factors are also tested for the concentrations of FPOMW. The percentages of variance explained by the factor considered relative to the total explained variance are shown. Asterisks indicate significant effects: *< 0·05, **< 0·01, ***< 0·001. Degrees of freedom are shown in parentheses.

FPOMWStream (2)Station (stream) (10)Occasion (2)
 δ13C (24)2·195·6***2·3
 δ15N (24)8·789·41·9
 C:N (24)0·893·2***5·9
 Chl-a:C (24)25·3*64·99·8
 POC conc. (24)5·3**94·6***0·1
 PON conc. (24)7·7***92·2***0·1
 Chl-a conc. (24)4·5**94·5***1·0
FPOMSStream (2)Station (stream) (6)Point (station) (15)
 δ13C (22)24·050·725·4
 δ15N (22)4·547·2**48·4
 C:N (22)34·942·5*22·5
 Chl-a:C (20)3·424·871·8*
BiofilmStream (2)Station (stream) (6)Point (station) (10)Pool/riffle (1)
 δ13C (47)23·8***37·5**27·411·3**
 δ15N (47)11·254·9**33·90·01
 C:N (47)25·2***39·9***27·97·0*
 Chl-a:C (47)19·8*47·3**9·423·6**

The full-factor RDA model (i.e. six variables: discharge, stream gradient, streambed CPOM abundance, irradiance, riparian vegetation and riparian forestry activities) explained 72·6% of total variance of the biogeochemical properties of FPOM (Table 3). The two axes, RDA1 and RDA2, represented 51·2% (= 0·02) and 20·6% (= 0·64) of the total variance explained, respectively. The full-factor RDA showed that all the six environmental variables explained the variance of the FPOM properties comparably (12·2–20·8% of explained variance), and also that no explanatory variable alone explained a significant portion of the variance of FPOM properties (= 0·13–0·44). Meanwhile, single-factor RDAs showed that five of the six explanatory variables, other than riparian forestry activities, significantly explained the variability in the biogeochemical properties of FPOM (18·7–26·7% of total variance, = 0·01–0·03). Each of those five significant variables showed correlations with RDA1, which explained the major portion of the variance of FPOM properties. The insignificant variable, i.e. riparian forestry activities, was correlated with RDA2 which explained a minor portion of the variance of FPOM properties.

Table 3.   Summary of redundancy analyses
Full-factor RDASingle-factor RDA
Explanatory variablesVariance %PVariance %P
  1. The percentages of variance are shown as (variance explained by each variable/total variance) as well as (variance explained by each variable/total explained variance) (in parenthesis).

  2. CPOM, coarse particulate organic matter.

Irradiance13·4 (19·8)0·1626·70·01
CPOM11·4 (16·9)0·2225·00·01
Discharge14·1 (20·8)0·1318·70·02
Gradient8·3 (12·2)0·2919·70·03
Vegetation8·4 (12·4)0·3521·50·03
Forestry12·1 (17·9)0·4410·20·83

Stream gradient and streambed CPOM abundance were positively related, and both of them were negatively related to discharge (Fig. 2a). Vegetation types were also linked with discharge, i.e. high-discharge stations had more deciduous vegetation. δ13C of FPOMS and chl-a:C of FPOMW were related to those four environmental variables (i.e. stream gradient, streambed CPOM abundance, vegetation types and discharge). Based on the correlation of RDA1 with those four environmental variables including discharge, RDA1 can be interpreted to represent environmental variation associated with stream size and/or longitudinal position of the sampling stations. Irradiance was also correlated with RDA1. RDA2 was associated with the types of riparian forestry activity which was related to δ15N of FPOMS. In addition, most of the stations with relatively small discharge were in the positive direction of RDA1, and most of the stations with riparian forestry activity were in the positive direction of RDA2 (Fig. 2b).

Figure 2.

 Results of full-factor redundancy analysis. Panel (a) shows scores of constrained variables (chemical properties of FPOM; solid circles), and constraining environmental variables (arrows and bold labels): FPOMW, wat; FPOMS, sed; veg, vegetation categories (conifer-dominant or mixed); for, riparian forestry activity types (logged, reserve or no activity). Panel (b) shows site scores: con, conifer dominant vegetation; mix, conifer-deciduous mixed vegetation; res, riparian reserve; no, no forestry activities.

The multiple regression analyses showed that 43–68% of the variance of FPOM properties, except FPOMWδ15N and FPOMW C:N, was explained by the explanatory variables tested. δ13C, C:N and chl-a:C of FPOMS were significantly related to streambed CPOM and irradiance, but not related to the biogeochemical signatures of riparian soil or biofilm (Table 4). δ15N of FPOMS had a negative relationship with riparian soil δ15N. Meanwhile, significant predictors differed between FPOMW and FPOMS, even for the same properties. δ13C of FPOMW was related negatively to irradiance and positively to riparian soil δ13C. Chl-a:C of FPOMW was related positively to irradiance and biofilm chl-a:C. The regression analysis showed that biogeochemical properties of FPOMW, except δ13C, had weaker relationships with the local factors that were examined, compared with those of FPOMS.

Table 4.   Results of multivariate regression analyses for the chemical properties of FPOM vs. environmental factors: irradiance, streambed CPOM and biogeochemical properties of riparian soil (rs) and biofilm (bf)
Dependent variablesExplanatory variablesPR2
  1. CPOM, coarse particulate organic matter; FPOM, fine particulate organic matter.

  2. P-value and coefficient of determination (R2) of best models selected based on the lowest Akaike’s information criterion are shown. Explanatory variables examined are shown; bold, significant (< 0·05) and also selected; regular font, not significant (> 0·05) but selected; in parentheses, not selected. Positive/negative signs indicate trends of effects in models.

δ13CSediment+CPOM (irradiance, bf δ13C, rs δ13C)0·0010·63
Water−irradiance, +rs δ13C (CPOM, bf δ13C)0·0030·68
δ15NSediment−rs δ15N, +bf δ15N, +irradiance (CPOM)0·0240·63
Water(irradiance, rs δ15N, bf δ15N, CPOM)  
C:NSediment−irradiance (CPOM, rs C:N, bf C:N)0·0140·43
Water+CPOM (irradiance, rs C:N, bf C:N)0·190·15
Chl-a:CSediment−CPOM, +irradiance (bf chl-a:C)0·0060·64
Water+irradiance, +bf chl-a:C (CPOM)0·0410·47

δ13C of FPOMW and FPOMS had relatively high and narrow ranges compared with the signature of in-stream-origin and riparian-origin POM (i.e. biofilm and leaves) (Fig. 3). δ13C of FPOMW was slightly lower than that of FPOMS. FPOMS was higher in δ13C compared with biofilm and leaves, and similar to bark and riparian soil. Meanwhile, invertebrates’δ13C showed species-specific relationships with the signature of FPOMS and biofilm (Fig. 4, Table 5). One of the shredders, L. roafi, was lower in δ13C than FPOMS, and had similar signatures to biofilm in most of the sampling stations. δ13C of L. roafi was significantly related to that of FPOMS, and also marginally related to the signature of biofilm. The relationship of δ13C between L. roafi vs. FPOMS had a steep slope significantly larger than 1 (4·8 ± 3·4, mean ± 95%CI). δ13C of the other shredder, D. augusta, was similar to the signature of FPOMS, and was consistently higher than that of biofilm. δ13C of the collector, Paraleptophlebia spp., was lower than that of FPOMS at most of the stations. δ13C of D. augusta and Paraleptophlebia spp. did not show significant relationships with δ13C of FPOMS or biofilm.

Figure 3.

 δ13C of in-stream and terrestrial POM, and macrobenthic invertebrates. The box shows lower quartile, median and upper quartile. The whiskers show 5th and 95th percentiles. For FPOMS, FPOMW, biofilm and riparian soil, data from all the stations were pooled. For bark and leaves from riparian area and in-stream leaves, different species were pooled. The number of values pooled is in parentheses.

Figure 4.

 Relationships of invertebrate δ13C vs. FPOMSδ13C and invertebrate δ13C vs. biofilm δ13C: C Creek, open circle; Donegani Creek, grey square; Spring Creek, solid triangle. The regression line for a significant relationship (P < 0·05, pooled data for all the streams) and 1:1 lines are indicated by dotted and solid lines, respectively.

Table 5.   Results of univariate regression analysis for invertebrate δ13C vs. sediment δ13C and for invertebrate δ13C vs. biofilm δ13C
 R2PSlope (95%CI)
  1. FPOM, fine particulate organic matter.

Despaxia vs. FPOMS0·490·132·21 (−0·79–5·21)
Despaxia vs. Biofilm0·210·160·39 (−0·18–0·95)
Lepidostoma vs. FPOMS0·730·0114·74 (1·38–8·11)
Lepidostoma vs. Biofilm0·350·0530·75 (−0·01–1·51)
Paraleptophlebia vs. FPOMS0·410·182·78 (−1·54–7·10)
Paraleptophlebia vs. Biofilm0·090·350·47 (−0·59–1·52)


Linkage of FPOM Properties with Local Environment and Processes

Our results demonstrate that FPOMS properties are a good indicator of the responses of in-stream biogeochemical conditions to the local environment. The relatively weaker relationships of FPOMW properties with local factors may be attributable to advective, downstream transport of FPOMW, reducing its specificity.

The significant relationships of FPOMSδ13C and chl-a:C with streambed CPOM abundance suggest the importance of local production of FPOM through CPOM breakdown (e.g. Wallace et al. 1991; Webster et al. 1999) for the spatial variation of those FPOM properties. In addition, the significant relationships of FPOMS C:N with irradiance indicate that the algal production also exerted influence on the spatial variations of the FPOM properties (e.g. Hein et al. 2003). The regression analysis demonstrates that δ13C, C:N and chl-a:C of FPOMS were controlled mainly by quantity of local CPOM supply and primary productivity, rather than quality of POM supplied from riparian areas or biofilm.

Biogeochemical properties differ in responses to the local environment and applicability as an environmental indicator. In particular, RDA suggests that δ13C, chl-a:C and C:N of FPOMS characterise natural variations of stream environment because of upstream–downstream locations. Meanwhile, based on the regression analysis, FPOMS C:N had a relatively weaker linkage with local environmental factors compared with FPOMSδ13C and FPOMS chl-a:C. C:N of FPOM produced by CPOM breakdown tend to be lower and more homogenous compared with original CPOM materials (Yoshimura et al. 2008). Microbes decrease C:N of terrestrial-derived POM (e.g. bacteria and fungi) by efficiently absorbing dissolved nitrogen from ambient water (Hall, Peterson & Meyer 1998; Motomori, Mitsuhashi & Nakano 2001; Ashkenas et al. 2004). Such in-stream processes decreasing and homogenising C:N may weaken the relationship of FPOMS C:N with local environmental factors.

Although the RDA suggests a possible linkage of FPOMSδ15N with riparian forestry activities, mechanisms behind the variations of FPOMSδ15N could not be inferred (e.g. the negative relationship with riparian soil δ15N). Nitrogen sources (e.g. atmospheric input and dissolution from bedrock) and isotopic fractionations because of various biological processes (e.g. assimilation, mineralisation and nitrification) alter δ15N in aquatic ecosystems (Wada & Hattori 1991; Holloway et al. 1998; Barnes, Raymond & Casciotti 2008). Thus, mechanisms controlling δ15N of in-stream FPOM could be more complicated than those for the other biogeochemical properties examined.

Relationships of FPOM Properties with Resource use by Primary Consumers

The significant and marginal relationships between FPOMS/biofilm δ13C and δ13C of L. roafi indicate that the spatial heterogeneity in summer-time FPOMs persisted long enough that some consumer species reflected a local signal of their resources in our study site. Furthermore, the significant relationship between δ13C of L. roafi and FPOMSδ13C indicates that the resource use by this species varied at a spatial scale similar to the variation of FPOMS properties along the streams.

Based on the relationship of FPOMSδ13C with FPOMS chl-a:C indicated in RDA, FPOMSδ13C was a good proxy for the relative abundance of stream-origin POM (mainly algae) in FPOMS. The higher FPOMSδ13C than biofilm δ13C suggests small contribution of stream-origin POM and predominance of terrestrial-origin POM in FPOMS. The steep slope of the relationship of L. roafiδ13C vs. FPOMSδ13C and the marginally significant relationship of L. roafiδ13C vs. biofilm δ13C imply that this species selectively used stream-origin POM despite its low relative abundance in the pools of stored POM. This suggests that when stream-origin POM was abundant, it was more efficiently assimilated by L. roafi. The consumption of stream-origin POM by this shredder species was possibly attributed to their assimilation of biofilm growing on leaf surfaces (Ledger & Hildrew 2001; Franken et al. 2005). Some putative shredder species have been reported to directly feed on FPOM (Carvalho & Graça 2007), so opportunistic consumption of FPOMS by this shredder species is also possible.

The degree and type of food selectivity differed between species. The collector species, Paraleptophlebia spp., may have some selectivity in their food intake and/or assimilation, because δ13C of this species deviated from FPOMSδ13C in some stations. However, factors controlling food selectivity of this species are unknown, as their food selection was independent of the relative abundance of stream-origin POM in FPOMS. Meanwhile, selective use of in-stream-produced POM by the shredder species, D. augusta, is unlikely to occur, as δ13C of this species was close to FPOMSδ13C and deviated from that of biofilm.

Our results indicate that comparing δ13C of primary consumers with δ13C of bulk FPOMS helps to analyse species-specific characteristics of basal resource use. However, it should be noted that the relationship between resource use of primary consumers and FPOM composition is probably not always because of direct feeding on FPOM. In-stream POM are diverse in their physical and chemical features (e.g. size, shape, mobility and nutrition), and the food resource selection by primary consumers depends on species and environment. The mechanisms behind the linkages between FPOM composition and resource use by primary consumers need further examination.

Implications for Stream Ecosystem Management

If the biogeochemical properties of FPOM are heterogeneous along streams and show a relationship with the local environment, local processes (e.g. input from riparian, in-stream FPOM production and decomposition, and retention) are expected to be more important than advective transport for distributions of FPOM properties. In this case, managers should focus on the local impacts of environmental alterations on the adjacent stream rather than impacts on downstream reaches. In contrast, if FPOM properties exhibit homogeneous or continuous longitudinal trends along streams and show a less close relationship with the local environment, advective transport dominates and/or the local environment has only minor effects. In this case, managers should pay more attention to human impacts at the catchment scale. In the study streams, advective transport was probably insufficient to redistribute FPOM after input/production of FPOM; reach-scale local effects were more important for in-stream biogeochemical characteristics than catchment-scale cumulative effects transmitted from upstream.

Whereas primary production and nutritional dependence on algae by primary consumers are known to increase downstream along low- to high-order reaches (e.g. Thorp et al. 1998), our results highlight that FPOM properties and resource use by primary consumers are spatially heterogeneous even within low-order forested reaches. Either natural- or human-alteration of catchment characteristics (e.g. hydrology, land-use and vegetation) could alter this spatial heterogeneity, and consequent effects on catchment-scale biodiversity are still unpredictable. However, the spatial distribution of FPOM properties along streams greatly depends on temporal variation in hydrological conditions. Temporal variation of FPOM properties still need to be examined to confirm the representativeness of our results.

Our RDA results demonstrate that the effects of riparian forestry activity on FPOM properties were insignificant. The effects of stream size and/or longitudinal position of the sampling stations more strongly influenced FPOM properties. However, vegetation and forestry activities in riparian areas affect light conditions and primary production (Kiffney, Richardson & Bull 2003) and CPOM input/abundance (Johnson & Covich 1997) in the adjacent streams, so biogeochemical properties of in-stream FPOM potentially reflect riparian forestry activities. The relative magnitude of natural and forestry-related environmental variation probably depends on the range of natural variation encompassed in assessments. The magnitude of human impact on stream habitat needs to be carefully evaluated. Spatial distributions of in-stream FPOM properties and its relationships with local environmental factors can help managers to identify spatial scales and factors which need to be focused on to achieve sound stream ecosystem management.


We thank S. Leung and J. Shum for assistance with field sampling and laboratory work. We also thank members of Stream and Riparian Ecology Laboratory of University of British Columbia for valuable comments on the manuscript. This study was financially supported by the Forest Science Program (British Columbia, Canada), the Natural Sciences and Engineering Research Council (Canada), and Sumitomo Foundation (Japan).