Abundance and patterns of transparent exopolymer particles (TEP) in Arctic floodplain lakes of the Mackenzie River Delta


  • C. Adam Chateauvert,

    1. Departments of Geography and Biological Sciences, Simon Fraser University, Burnaby, British Columbia, Canada
    Search for more papers by this author
  • Lance F. W. Lesack,

    Corresponding author
    1. Departments of Geography and Biological Sciences, Simon Fraser University, Burnaby, British Columbia, Canada
      Corresponding author: L. F. W. Lesack, Departments of Geography and Biological Sciences, Simon Fraser University, Burnaby, British Columbia V5A 1S6, Canada. (llesack@sfu.ca)
    Search for more papers by this author
  • Max L. Bothwell

    1. Pacific Biological Station, Environment Canada, Nanaimo, British Columbia, Canada
    Search for more papers by this author

Corresponding author: L. F. W. Lesack, Departments of Geography and Biological Sciences, Simon Fraser University, Burnaby, British Columbia V5A 1S6, Canada. (llesack@sfu.ca)


[1] The Mackenzie River Delta is a lake-rich arctic floodplain that receives high inputs of dissolved organic matter (DOM) and suspended particulates from allochthonous and autochthonous sources, and may transfer carbon from dissolved to particulate phase via in situ formation of transparent exopolymer particles (TEP). TEP provides food for grazers, surfaces for bacteria, and increased potential for aggregation and sedimentation of organic matter. During open water 2006, we tracked TEP abundances in three Delta lakes representing gradients that include declining river-to-lake connection times, increasing levels of dissolved organic carbon (DOC), and declining chromophoric-DOM (CDOM). Unexpectedly, TEP abundances were highest immediately after the flood, when autochthonous autotrophic production was at a seasonal low and CDOM a seasonal high. Moreover, the lake with the strongest riverine influence and lowest levels of autochthonous autotrophic production had the highest mean TEP-carbon (TEP-C) concentrations among the lakes. The mean proportion of particulate organic carbon (POC) represented by TEP-C increased with increasing river connection time, and appears to represent a substantial proportion of POC in Mackenzie Delta Lakes. Unexpectedly, the TEP gradient was most strongly related to CDOM (river water source) rather than overall DOC. Variations in CDOM accounted for 53% of TEP-C variation among the lakes, indicating allochthonous matter was the most important source of TEP. DOC release from in situ macrophytes during periods of high photosynthesis may contribute to TEP formation in the lake with lowest riverine influence, but pH levels >9.5 driven by the high photosynthetic rates complicate the interpretation of results from this lake.

1. Introduction

[2] The discovery of new classes of particles, such as transparent exopolymer particles [Alldredge et al., 1993], Coomassie blue particles [Long and Azam, 1996], and DAPI yellow particles [Mostajir et al., 1995] represents an important advance toward understanding the nutritional quality of particles and cycling of carbon through marine and freshwater ecosystems. Transparent exopolymer particles (TEP) are carbon-rich and seem to be ubiquitous in aquatic ecosystems. TEP form via progressive aggregation of colloidal dissolved organic matter (DOM) to particulate organic matter (i.e., TEP-class POM) [Passow, 2002a]. Specifically, polysaccharide fibrils form TEP precursors which are then stabilized into aggregates by entanglement of the fibrils, and chemical crosslinks from divalent cationic (Ca2+ and Mg2+) [Kloareg and Quatrano, 1988] and hydrogen bonds [Chin et al., 1998]. Chin et al. [1998] estimated that ∼10% of DOM may form spontaneously assembled particles such as TEP. The DOM necessary for TEP formation is mucilaginous material excreted by a variety of organisms [Decho, 1990; Leppard, 1995; Wotton, 2005]. Whereas this material is biotic in origin, the aggregation is abiotic, changing the traditional view of bacteria as the primary means of transforming DOM to POM. Phytoplankton have been identified as a strong source of fibrillar exudates [Myklestad, 1995]. Laboratory based studies show bacteria also have the potential to release TEP forming precursors [Stoderegger and Herndl, 1999; Passow, 2002a; Radic et al., 2003; Radic et al., 2006], though there is little evidence of in situ bacterial communities doing so. However, bacteria are important for TEP production through exudate modification or TEP precursor scavenging [Grossart and Simon, 2007; Sugimoto et al., 2007].

[3] TEP formation in inland waters and how it may differ from marine environments has thus far received little attention, particularly in systems with complex mixtures of DOM. The role of suspended sediments is also not understood, but it may enhance removal of TEP via enhanced ballasting of aggregates and sedimentation [Passow, 2002a; Burd and Jackson, 2009]. A general class of lakes, which are particle-rich and remain poorly understood, comprise those associated with circumpolar river deltas [Lesack and Marsh, 2007; Lesack and Marsh, 2010]. The sheer abundance of these lakes [Emmerton et al., 2007] coupled with primary productivity and biodiversity levels exceeding what is expected of such high latitudes make these ecosystems ecologically important [Squires et al., 2009]. Lakes are positioned in a gradient of elevations relative to the river channel, with seasonal changes in river levels leading to shorter river-to-lake connection times in higher elevation lakes and longer connection times in lower elevation lakes. This in turn drives gradients in water renewal rates (short connection time = slower renewal, long connection time = faster renewal), creating chemical and biological gradients among the lakes [Lesack et al., 1998, Lesack and Marsh, 2010] and diverse DOM compositions that depend on lake position [Gareis, 2007; Chateauvert, 2008; Tank, 2009]. Declining river connection times generally lead to an increasing supply of total DOM within the lakes because substantial amounts of nonchromophoric DOM are derived from the high macrophyte biomasses in lakes with short connection times, whereas chromophoric-DOM (CDOM) generally decreases along this gradient because it is mainly delivered via river water during flooding events [Tank et al., 2011]. These naturally occurring gradients provided an opportunity to examine variations of TEP with naturally varying levels and compositions of DOM.

[4] Based on the literature, we considered a broad range of measures that may be related to TEP formation in arctic delta lakes. We then postulated the pattern of TEP we expected to find (Figure 1), given the particular gradients present in the Mackenzie Delta (more details in Chateauvert [2008]). During open water 2006, we tracked measures of TEP abundance and environmental variables in three Mackenzie Delta lakes, representing a gradient of river-to-lake connection times and examined how well the TEP patterns fit our expectations. We assessed the hypothesis that the gradients in DOC and total suspended solids (TSS) should drive a gradient in TEP abundances, where lakes with short river connection times will have high TEP, because such lakes have high autotrophic production and high autochthonous DOC levels, whereas lakes with long river connection times will have low TEP, because such lakes have high TSS levels and low autotrophic production with DOC mostly in the form of allochthonous terrestrial DOM.

Figure 1.

Hypothesized gradient of transparent exopolymer particle (TEP) abundance, as a function of average river-to-lake connection times driven by elevations of the lakes relative to the river in the Mackenzie Delta. Differing river connection times drive gradients in physical, chemical, and biological variables potentially related to TEP formation. Arrow-heads indicate direction of increase, solid lines indicate expected enhancement of TEP formation, and broken lines indicate suppression of TEP. Lake closure-classes are based on Marsh and Hey [1989].

2. Methods

2.1. Study Area

[5] The Mackenzie River is an important representative of large north-flowing rivers in the circumpolar Arctic and the Mackenzie Delta is located where the Mackenzie River discharges into the Beaufort Sea (68–69°N 134–137°W) [Lesack and Marsh, 2010]. Within this extensive 13,000 km2 delta, there are over 45,000 lakes [Emmerton et al., 2007] with mean depths from about 0.5 m to 4.5 m, depending on time of year, whether the year is wet or dry locally, and whether delta water levels are higher or lower than average [Lesack and Marsh, 2010]. The open water period in the central delta is from June to November. Peak annual water levels in the river are partially controlled by the amount of water contained within the winter snowpack, but more commonly, they are strongly controlled by ice breakup effects [Andres and Doyle, 1984; Prowse, 1986]. Delta lakes are perched at a range of elevations above Mackenzie channels [Marsh and Hey, 1989] that govern river-to-lake connection times. During such connection, lakes are replenished in varying degrees with river water, nutrients, dissolved organic matter, and sediments [Lesack and Marsh, 2010; Marsh and Lesack, 1996]. The lakes have been classified [Marsh and Hey, 1989] as in Mackay [1963]. No-closure lakes (Figure 1) remain in connection with river channels for the entire summer. Low-closure lakes are flooded each spring but are cut off from the river for some portion of the summer. High-closure lakes are not necessarily flooded every spring and never during the summer.

[6] Three representative and well-studied lakes were chosen for the present study. Based on water level records from 1964 to 2005, Lake 129 is no-closure class, area 38 ha, mean depth 1.3 m, with continuous river connection. Lake 56 is high-closure class, 3.1 ha, mean depth 1.1 m, with an average river connection time of 9.3 d/yr (11 days in 2006). Lake 520 is high-closure class, 0.2 ha, mean depth 2.2 m, with average river connection of 6.5 d/yr (10 days in 2006). The fraction of ambient lake water mixtures consisting of “new” river water following the flood peak [Lesack and Marsh, 2010, equation (3)] is an average of 0.41 in Lake 520, 0.59 in Lake 56, and 0.95 in Lake 129. During 2006, the flood peak at Inuvik was about 1 m higher than the long-term average and the third highest level since 1973. Whereas all three lakes should have been fully flushed with river water at the beginning of our 2006 observation period, the lakes still strongly vary in their degree of riverine and autotrophic influence. Lake 129 remains connected to the river and may receive river water pulses throughout the open water period. Lakes 56 and 520 do not have substantially different river connection times, but Lake 520 is a deeper thermokarst lake (an important subclass) [Tank et al., 2009a] that intermittently receives modest inputs of dissolved organic carbon from melting permafrost [Tank et al., 2011]. The macrophyte communities in these lakes are a function of long-term average water levels, particularly the multiyear Chara communities in Lake 520, though some variation can occur in years with higher or lower than normal water levels [Squires and Lesack, 2003; Squires et al., 2002]. Locations and further information on these lakes are in Figure 1 and Table 1 of Lesack and Marsh [2010].

[7] Total ionic concentration in Mackenzie River water is around 4.2 to 5.5 milli-equivalents/L, with the lakes during open water averaging around 5.4 in Lake 129, 4.8 in Lake 56, and 5.0 in Lake 520. Ca2+ concentrations in the river are 1.4 to 1.9 milli-equivalents/L, with the lakes during open water averaging 1.8 in Lake 129, 1.0 in Lake 56, and 1.2 in Lake 520. Further information on ion concentrations is available in Lesack et al. [1998] and Lesack et al. [1991].

2.2. Sample Collection

[8] During the summer of 2006 (early June – late August), water samples were obtained with a tube sampler, integrated over depths up to 1.5 m from the lake surface (less if lake <1.5 m deep). Samples were taken from roughly the middle of each lake, in the same location at weekly intervals. Mackenzie Delta Lakes are generally steep-sided because of thermokarst effects and relatively shallow. The macrophyte communities thus tend to be distributed over the whole lake-bottom. Given that the lakes are well mixed during the open water season, our point samples were taken as representative values of the chemical and biological measurements made. Samples were stored in 1 L high-density-polyethylene (HDPE) bottles and placed on ice in a cooler. Samples were processed immediately upon return to the lab.

2.3. Chl-a, TSS, POC, DOC, CDOM, Dissolved Carbohydrates

[9] Chlorophyll-a (Chl-a) was measured via vacuum filtering lake water at no more than 175 mm Hg through Whatman GF/C filters. The filters were frozen at −20°C until subsequent analysis. Extraction of Chl-a was performed in 90% ethanol following Nusch [1980], and allowed to extract for 24 h at 4°C. Chl-a was then determined fluorometerically [Welschmeyer, 1994].

[10] Total suspended solids (TSS, mg/L) was measured gravimetrically via filtering lake water through combusted, preweighed GF/C filters and drying to stable weight [Wetzel and Likens, 2000]. These filters were then combusted at 550°C for 9 h (time to stable weight) before being reweighed. The filter's combusted weight was subtracted from its dry weight to obtain particulate organic matter (POM), then multiplied by 0.47 ± 0.01 [Dean, 1974] to estimate particulate organic carbon (POC, mg/L).

[11] Dissolved organic carbon (DOC) was measured on lake water filtered through prerinsed Whatman GF/F filters. Filtered samples were stored at 4°C for up to 4 months in new, prewashed HDPE bottles until analysis [Tank et al., 2011]. DOC was analyzed [Gareis et al., 2010] as nonpurgeable organic carbon, based on the high temperature catalytic oxidation method, via a Shimadzu, Total Organic Carbon Analyzer (TOC-VCPH).

[12] Chromophoric dissolved organic matter (CDOM) was measured on lake water filtered through a 0.22 μm Durapore (PVDF) membrane filter. Samples were stored in new, prewashed HDPE bottles at 4°C for 2 to 4 weeks until analysis [Tank et al., 2011]. Spectrophotometric absorbance at 330 nm in a 5 cm path length, quartz cuvette was used as a relative measure of CDOM in the water [Kirk, 1994; Whitehead et al., 2000]. Comparisons with other measures of DOM aromaticity, such as specific UV absorbance at 245 nm (SUVA254), during other work in this study system [Gareis et al., 2010; Tank et al., 2011] indicates that absorbance at 330 nm is a reasonable measure for routine tracking of CDOM.

[13] Total dissolved monosaccharides (MONO) was measured on GF/F filtered lake water via the TPTZ method [Myklestad et al., 1997]. Total dissolved carbohydrates (TDCHO) was measured via the same method, after a hydrolysis treatment with HCl to a final concentration of 0.09 N and incubation in sealed glass ampoules at 150°C for 1 h. Total dissolved polysaccharides (POLY) was calculated as the difference between TDCHO and MONO.

2.4. TEP Analysis

2.4.1. Slide Preparation

[14] Semipermanent slides were prepared in duplicate based on the method of Alldredge et al. [1993] with few modifications. 1–4 ml samples were filtered on to 25 mm diameter, 0.4 μm polycarbonate membrane filters at a constant filter pressure of no more than 150 mm Hg. Samples were stained with 0.5 ml of 0.2 μm filtered alcian blue. The 0.03% alcian blue in 0.06% acetic acid was drawn through the filter immediately. The filter-transfer-freeze method [Hewes and Holm-Hansen, 1983] was used to transfer filtered material to a glass slide and all transfers were done by the same analyst. A loop of gel (0.035 g/ml gelatine, 25% glycerine in distilled water) was placed over the filtered material while still frozen and left to solidify. Prepared slides were stored at −20°C in sealed bags.

2.4.2. TEP Enumeration and Calculation of Size Distributions

[15] A Moticam™ 1300 color camera connected to a computer was used to capture digital images of 10 fields per slide at 250× magnification on a Leitz, Aristoplan microscope. Using Motic Images Advanced 3.0™ image analysis software, individual TEP were manually delineated to obtain the cross-sectional area of TEP particles. From the cross-sectional area of TEP particles, their equivalent spherical diameter (ESD) and volume (ESV) were calculated in order to place them into size categories and to calculate a volumetric-based abundance (i.e., TEP-v = ppm-v) and size distribution.

[16] The size distribution of TEP is described by the power law

equation image

Where dN is the number of particles in size interval d(dp) (interval described by mean maximal ESD) and k and b are constants, where b describes the shape of the size distribution.

2.4.3. TEP Carbon

[17] The carbon content of a given TEP particle was estimated using the equation of Mari [1999]

equation image

where TEPcarbon is in μg C and R is the equivalent spherical radius (μm). TEP carbon concentration (TEP-C) was then obtained via summing the size distribution of particles multiplied by their size dependent C content (more details in Chateauvert [2008] and Mari [1999]).

2.5. Bacteria

[18] Our model for the bacterial community is:

equation image

where free-living bacteria are not attached to any particle type, TEP-attached bacteria include all bacteria associated with TEP, and other-attached bacteria encompasses bacteria attached to inorganic detritus or any non-TEP organic particle. This paper addresses only Total Suspended Bacteria, whereas the behavior of free-living versus attached bacteria in relation to TEP is the subject of a separate paper [see Chateauvert, 2008].

2.5.1. Total Suspended Bacteria

[19] Total suspended bacteria, based on samples of bulk water with no prefiltration, were preserved in 1% glutaraldehyde, and stored in the dark at 4°C until slide preparation. Replicate slides were prepared based on the methods of Porter and Feig [1980] and Yoon and Rosson [1990]. A surfactant (Tween 80) was added to a final concentration of ∼1%, vortexed and allowed to penetrate aggregates for at least 2 h before being sonicated below 10 W for 30 s [Yoon and Rosson, 1990]. DAPI was then added to a final concentration of 10 μg/ml and let stand for at least 10 min. 1–2 ml sample aliquots were filtered on to phosphobuffered saline (PBS) conditioned, 0.2 μm, nucleopore filters, at a filter pressure of 178 mm Hg and was followed by a rinsing with PBS. Filtered bacteria were transferred to a glass slide using the filter-transfer-freeze method [Hewes and Holm-Hansen, 1983] and covered with a loop of hot gel. This technique was used to facilitate staining for TEP for examination of TEP-attached bacteria [Chateauvert et al., 2012]. Testing with this method showed that the transfer efficiency for free-living bacteria was generally >90% [Chateauvert, 2008]. Because TEP particles are larger than bacteria, transfer efficiency for TEP should be even higher. To be conservative, we did not attempt to correct abundances of TEP or bacteria. Enumeration of bacteria was performed at 1000x magnification on a Leitz, Arisitocrat epifluorescence microscope. Bacteria were counted in 10 fields on each slide.

2.6. Data Analysis

[20] To test for differences in TEP abundance among lakes and for a seasonal time effect within lakes, a randomized block design was performed using lake as a random (block) effect and date as a fixed effect. This analysis gave the same result as a repeated-measures analysis of variance (ANOVA). Relationships between relevant variables were assessed by linear regression. The strength of the relationships were reported via coefficients of determination (r2) and significance was tested using ANOVA. Variables included in multiple linear regression analyses were chosen based on biological relevance and the strength of the relation between the variable and TEP. Model performance was based on the r2 and the inclusion of additional variables was based on significant increases to the r2 value.

3. Results

3.1. TEP Abundances

[21] Volumetric abundances of TEP varied seasonally by over 2 orders of magnitude from 296 ppm immediately after flood-water replenishment in the lakes (early June) to 5 ppm in late August (Table 1, Figure 2a), with TEP-C mass concentrations (Figure 2b) following a similar pattern from 3.24 to 0.17 μg C/mL. Abundances dropped rapidly in all lakes after the initial sampling date, then tended to slowly decline over the remainder of the open water period. Differences in abundance over time were significant, both with and without the first sampling date included (randomized block ANOVA, F = 9.37, p < 0.0001, df = 10, and F = 2.88, p = 0.027, df = 9, respectively).

Table 1. Descriptive Statistics for TEP Abundances (ppm-v) During the Summer of 2006
First Sample Date Included
All Lakes3333.354.64.81296
First Sample Date Excluded
All Lakes3018.913.64.8160.1
Figure 2.

Seasonal patterns of (a) volumetric TEP abundance and (b) TEP-C concentration in Lakes 520, 56, and 129 during open water 2006. Both plots have been scaled to exclude very high values on 12 June (137, 97, and 296 in Figure 2a, respectively, for Lakes 520, 56, and 129; 52 in Figure 2b for Lake 520).

[22] Lakes 129, 56, and 520 had mean volumetric TEP abundances respectively of 48.8, 22.5, and 28.5 ppm with the first sampling date included, and 24.1, 15.1, and 17.5 ppm with the first sampling date excluded (Table 1). TEP-C concentrations respectively were 1.03, 0.79, and 5.24 μg C/mL with the first sampling date included, and 0.871, 0.545, and 0.559 μg C/mL with the first sampling date excluded. Differences in TEP volumetric abundance among the lakes were not significant, but TEP-C concentrations were significantly different if the very high TEP values observed on the first sampling date were excluded (F = 7.06, p = 0.005, df = 2).

3.2. TEP Mass and Particle Size Distribution

[23] TEP-C versus volumetric TEP abundance (Figure 3) were strongly related, except for the high values on the earliest sampling date. In principle, TEP-C should be a better measure of overall TEP mass because it accounts for the fractal relation between volume-specific carbon content versus TEP particle size [Mari, 1999].

Figure 3.

Relation between TEP-C concentration versus volumetric TEP abundance. (a) A log plot that includes the high values on 12 June. (b) With 12 June excluded, this is shown as a linear plot.

[24] The TEP size distribution constant, b (spectral slope), ranged from 2.01 to 2.97 in Lake 129, 1.99 to 3.44 in Lake 56, and 2.02 to 3.42 in Lake 520 (Figure 4), with means of 2.57 (SD = 0.31), 2.84 (SD = 0.41), and 2.64 (SD = 0.36), respectively. This indicates a modest average bias (3.0 indicating equal contributions from all classes) toward larger particles contributing proportionally more TEP mass than the smaller particles. The lower limit of b was essentially equal in the three lakes, while the maximum value for Lake 129 was lower than that of Lake 56 and 520. There was no evidence of a consistent seasonal trend in TEP size distributions among these lakes.

Figure 4.

Seasonal changes in the particle size-distribution constant (b = spectral slope) of TEP in Lakes 520, 56, and 129 during open water 2006. At slope = 3 (line on plot), all particle sizes contribute equally to the total volumetric abundance. Above this line, smaller particles dominate the total volume; below the line, larger particles dominate.

[25] Our measures of TEP mass appear weakly and inversely related to b (Figure 5) overall, and in Lakes 129 and 56, but not in Lake 520. Log linear relations for volumetric TEP abundance versus b are statistically significant in the 3 lakes overall (r2 = 0.12, p = 0.053, df = 32), and in Lake 129 (r2 = 0.40, p = 0.035, df = 10) and Lake 56 (r2 = 0.37, p = 0.047, df = 10). Log linear relations for TEP-C versus b are similar but weaker and are not statistically significant.

Figure 5.

Plots of volumetric (a) TEP abundance and (b) TEP-C concentration versus the particle size-distribution constant (b = spectral slope) of TEP in Lakes 520, 56, and 129 during open water 2006. Data points are coded to separate relations among Lakes 520, 56, and 129. Significant direct (+) and inverse (−) relations are annotated on each panel for individual and all lakes (more detail in Table 2).

[26] The overall mean proportion of POC in the lakes represented by TEP-C was 83.7%, with mean values of 121, 69.0, and 69.9% respectively in Lakes 129, 56, and 520 (Figure 6). Of the 29 observations among the 3 lakes, 7 exceeded 100%, with 5 of those occurring in Lake 129 and one each in Lakes 56 and 520. The method for estimating TEP-C concentration assumes there is little variation in TEP C-content after accounting for its particle size distribution [Mari, 1999]. Our result here suggests there is some variability in C-content that requires further investigation.

Figure 6.

Seasonal changes in the % contribution of TEP-C to POC concentrations in Lakes 520, 56, and 129 during open water 2006. The y axis has been scaled to exclude one very high value (313) on 13 August in Lake 129.

3.3. Relations With Environmental Gradients

[27] Considering the three lakes together, TEP-C was related in varying degrees to 7 of the 9 variables assessed, with the 4 strongest being direct relations with CDOM, Suspended Bacteria, POC, and TSS (Figure 7, Table 2). Because the relations between TEP volumetric abundance and environmental variables are very similar to those of TEP-C, only the plots for TEP-C are shown. Within the lakes, the sets of variables most strongly related to TEP-C shifted from TSS, Chl-a, and CDOM in Lake 129 (strongest river influence), to POC, DOC, and CDOM in Lake 56, and finally to CDOM, Suspended Bacteria, and an inverse relation with Polysacchardes in Lake 520 (weakest river influence).

Figure 7.

Plots of TEP-C concentration versus measures of (a) DOC, (b) POC, (c) CDOM, (d) TSS, (e) Chl-a, (f) monosaccharides, (g) TDCHO, (h) polysaccharides, (i) total Suspended Bacteria, and (j) pH during open water 2006. Data points are coded to separate relations among Lakes 520, 56, and 129. Significant direct (+) and inverse (−) relations are annotated on each panel for individual and all lakes (more detail in Table 2). Statistically significant relations for individual lakes are shown as color-coded broken lines, with lumped relations for all lakes combined shown as a solid black line. The special case of pH > 9.5 is shown as a solid gray line.

Table 2. Coefficient of Determination for Relations Between TEP-C Versus Pertinent Environmental Variables in Lakes 520, 56, 129, and All Lakes Combineda
VariableLake 520Lake 56Lake 129All Lakes
  • a

    The + or − sign indicates whether the relation is direct or inverse. Alpha significance levels are shown as * = 0.05, ** = 0.01, *** = 0.001.

DOC +0.71*  
TDCHO  +0.41* 
Monosaccharide  +0.51*+0.16*
Polysaccharide−0.38*  −0.18*
Chl-a +0.65*+0.64*+0.16*
TSS  +0.69***+0.24**
POC +0.72** +0.29**
Suspended Bacteria+0.66*  +0.36***
pH (<9.5, only)   −0.26**
pH (>9.5, only) +0.74*  

[28] Whereas CDOM, on its own, accounted for 53% of the variation in TEP-C among all 3 lakes combined, incorporation of additional variables via multiple regression did not account for any significant additional variation, except in the case of Suspended Bacteria (total variation accounted for =61%; Figure 8). This indicates substantial intercorrelation among the potential independent variables, which may all be linked to the gradient in river-water renewal among the lakes.

Figure 8.

Estimated versus observed concentrations of TEP-C; based on a multiple regression model incorporating CDOM and total suspended bacteria (TSB) as independent variables.

[29] TEP-C shows a striking pattern of variation versus pH (Figure 7j) that suggests a complex and unexpected underlying relation that cannot be readily incorporated into a multiple regression model. There may be differing relations with a negative versus a positive slope, depending on whether pH is <∼9.5 or >∼9.5. If the observations are partitioned in this manner, both relations with TEP-C are statistically significant (r2 = 0.26, p < 0.01, df = 23; versus r2 = 0.74, p = 0.028, df = 5). We recognize that the relation above pH 9.5 depends strongly on the observation at pH 10.6, and that further work needs to be done on the potential role of pH.

4. Discussion

4.1. Seasonal Variation in TEP

[30] Given that the high-water period associated with river flooding is not a period of significant autotrophic production, the high volumetric TEP abundances that were observed in all the lakes at this time was contrary to our hypothesis that TEP abundance would increase with increasing autotrophic production. Moreover, the subsequent gradual decline in TEP abundance over the open water period was not consistent with the general pattern of increasing seasonal autotrophic production in the lakes over a substantial portion of the same period. This seasonal pattern was also consistent with a set of follow up TEP measurements during 2007 [Chateauvert, 2008] (data not shown). Possible explanations for the high initial TEP abundances include (1) high under-ice TEP abundances in the lakes, (2) high riverine TEP abundances derived from interaction between floodwaters and the terrestrial landscape, and (3) an interaction between river water and lake water.

[31] Increases in TEP abundances could occur in lakes containing high macrophyte biomasses (shorter river connection times) due to increases in DOC from decomposition during the ice covered season [Alber and Valiela, 1994; Thornton, 2004]. Release of DOC from such decomposition may be modest because of the cold temperature. Tracer evidence in our study system [Tank et al., 2011] indicates such DOC is primarily nonchromophoric. However, DOC in lakes in the Mackenzie Delta can become concentrated into the volume of unfrozen water via solute exclusion as the ice cover grows during the winter [Lesack et al., 1990]. Most lakes in the delta do not freeze to the bottom and this effect can concentrate solutes by factors >4× [Pipke, 1996]. Such concentrated levels of DOC may be a potential source of TEP precursors or TEP itself as the winter ice breaks up and wind mixing of the water column commences. The effects of ice formation and solute exclusion have not yet been investigated for the purpose of understanding TEP behavior. However, DOC can be excluded from ice at levels that are 2 times those of inorganic solutes [Belzile et al., 2002].

[32] Mackenzie River water indeed becomes enriched in DOC, primarily as CDOM, during ice breakup and the freshet [Emmerton et al., 2008], though TEP precursors have not thus far been directly measured. Such seasonal changes in CDOM and DOC composition are documented in other large arctic rivers [Spencer et al., 2008; Neff et al., 2006] and are linked to concomitant changes in DOC lability [Holmes et al., 2008] and age [Raymond et al., 2007]. DOC enrichment during the freshet is thought to originate from snowmeltwaters percolating through terrestrial organic detritus along the surfaces of landscapes underlain by permafrost [Neff et al., 2006]. Higher flood stages during the Mackenzie freshet seem to drive higher levels of DOC and TSS than in years with lower water levels (L. Lesack, unpublished data, 2012). The spring flood during our investigation (2006) was quite high (6.56 m asl, datum as in Marsh and Hey [1989]) relative to more typical years (e.g., 2007, 5.02 m asl) and may have contributed to the high TEP abundances during our first sampling date. Thus far, we know little about the potential for terrestrially derived DOC to form TEP or other possible TEP-mimicking particles. However, the relation between CDOM versus TEP in our lakes (see following section, Factors Affecting TEP Abundance) strongly supports the likelihood of TEP forming from such DOC.

[33] An interaction between lake water and river water occurring during the flood is possible, but it seems unlikely as the water in the lowest elevation lake, where TEP abundance was highest is effectively flushed and replaced entirely by river water [Lesack and Marsh, 2010] in high flood years as was observed in 2006.

4.2. Size Distribution

[34] The TEP size distribution constant (b) provides a measure of whether the smaller or larger TEP particles are contributing disproportionally to the overall volume of TEP in the water, with b = 3 representing equal contribution from all size classes. The smaller particles usually contribute proportionately less TEP volume than the larger particles [Passow, 2002a], leading to b-values <3. Our results show that both Lake 56 and 520 experienced periods of enrichment with small TEP particles (i.e., b > 3), though the timing differed in each case (Figure 4). In Lake 520, the maximum b-value occurred on the first two sampling dates, whereas in Lake 56, the maximum b-value occurred on 17 July with a second spike above 3 on 21 August. One general possibility is that Passow's data is for salt water. The lack of strong electrolytic solution in freshwater may result in smaller TEP.

[35] Another possible explanation for these high values is a rapid influx of TEP-precursors forming small particles at rates exceeding removal via aggregation and sedimentation. A strong source of TEP-precursors could be the input of DOC from flood waters at the beginning of our observation period, however, we did not observe such high b-values in Lakes 56 or 129 at that time. Alternatively, a strong source of TEP-precursors in Lake 56 around 17 July could be exudates from rapidly photosynthesizing macrophytes at that time (Figure 7j). However, b-values in Lake 520 were relatively low at that time. A second issue is that between-lake differences in aggregation rates could also contribute to differences in b-values among the lakes. Wind driven mixing of the water column enhances rates of particle collisions, potentially resulting in higher degrees of aggregation in more vigorously mixed water, with lesser aggregation and potential build-up of smaller TEP particles in calmer waters. The surface area of Lake 56 is an order of magnitude larger than Lake 520, and Lake 129 is 2 orders of magnitude larger, making Lake 56 and 129 more vigorously mixed via wind. Lake 129 indeed has the lowest average b-value (2.57), but the average value for Lake 56 (2.84) seems to be higher than in Lake 520 (2.64). More direct examination of the relationship between wind mixing, lake surface area, in situ turbulence, and TEP size distributions would be required to assess this further.

[36] Another issue is to what degree shifts from enrichment in smaller TEP particles to enrichment in larger particles may coincide with increases in volumetric TEP abundance or TEP-C. This may suggest that smaller TEP is aggregating to form larger TEP. In the two cases where b > 3 in each of Lake 520 and 56, the proportion of small particles present declined (i.e., b declines) sharply during the subsequent two weeks, and indeed resulted in increased volumetric TEP abundances (Figure 4). There was also a similar rising and falling of b-values in Lake 129, with resulting increases in volumetric TEP on Jul 11th and 31st. Moreover, there seem to be consistent inverse relations between TEP abundance versus b-values in Lakes 129 and 56 (Figure 5).

[37] On the other hand, the relation between TEP versus b in Lake 520 is not consistently inverse (Figure 5). A switch from enrichment to depletion in smaller particles could alternatively be the result of differential removal mechanisms such as rapid bacterial degradation of smaller particles or rapid removal after aggregation into larger particles. Although bacterial densities are always higher on smaller particles [Passow, 2002a], there is no evidence in our case to suggest that bacteria were rapidly degrading the TEP, as the bacterial densities on small TEP were not substantially greater than average during the weeks in question [Chateauvert, 2008]. It is possible that TEP-attached bacteria were more active during these time periods, but bacterial activity was not measured.

4.3. TEP Contributions to POC

[38] Our results suggest that the overall proportion of POC contributed by TEP-C in the study lakes is high (overall mean 83.7%) and suggests the potential for TEP-mediated carbon transfer between the dissolved and the particulate phases is also high. However, TEP-C appeared to exceed 100% of the POC in 7 of 29 our observations, indicating there must be substantial uncertainty associated with estimating the carbon content of TEP using equations from lab based studies. Some of these uncertainties are discussed at length by Mari [1999] and Engel and Passow [2001]. One uncertainty is that Mari's formula for TEP-C is based on lab samples in seawater, not freshwater. Because TEP formation depends highly on electrolyte concentrations, the formula may be inaccurate for this system. One additional uncertainty is that we used GF/C filters for TSS and POC measurements, whereas GF/F filters were used for measurements of DOC and carbohydrates, and 0.4 μm membrane filters for TEP. The somewhat larger pore size of the GF/C filters may have resulted in some underestimation of POC. In prior work, extensive testing of GF/C versus GF/F filters has generally yielded negligible differences for measurements of chlorophyll and TSS in our study system. Such testing, however, was not done with the aim of comparing TEP-C to POC and should be done in future work. We thus interpret our TEP-C data as a measure of the potential importance of TEP-mediated carbon transfer from the DOC to POC in freshwater lakes, while acknowledging that actual carbon content in TEP may be lower than our TEP-C values and likely represent a lower fraction of POC than our results indicate.

[39] The carbon content of TEP can potentially decrease as water residence time increases [Mari et al., 2007], which in our case would lead to overestimation of TEP-C as the open water season progressed. This effect would be primarily a result of declines in TEP turnover rates as water residence time increases [Mari et al., 2007], due to prolonged bacterial degradation of organic matter in the system, and a subsequent decrease in the reactivity of organic matter. Since surface active colloids are responsible for TEP formation, a decrease in colloid reactivity would result in a decrease in TEP formation and turnover. TEP turnover (ranges from 0.1 to 0.9 d−1 in marine ecosystems) is also sensitive to the concentration of non-TEP particles [Mari and Burd, 1998] via the increased number of collisions and subsequent aggregation and removal. In the context of the Mackenzie Delta, TEP turnover is thus likely to be high immediately after the flood due to higher particle concentrations and fresh sources of DOC. TEP turnover may decrease as the summer progresses and particle concentrations decrease and water residence times increase, resulting in more refractory and less reactive DOC. The “aging” of DOC in these lakes could be mitigated to some extent through macroalgal [Alber and Valiela, 1994; Thornton, 2004], or microalgal release of DOC. Further work is needed to resolve the extent that carbon content of TEP may vary in complex lake systems such as in the Mackenzie Delta.

[40] The ultimate fate of TEP-C in Mackenzie Delta lakes is thus far unknown, but it may contribute significantly to the sediments via sinking, to DIC and CO2 flux via bacterial mineralization, and may represent a food source for protists [Tranvik et al., 1993] and zooplankton [Passow and Alldredge, 1999; Ling and Alldredge, 2003].

4.4. Factors Affecting TEP Abundance in Delta Lakes

[41] We expected that TEP abundances in Mackenzie Delta Lakes would increase with decreasing river connection time because of the particular gradients in chemical and biological variables found in this study system (Figure 1). Our results show that TEP-C, in particular, significantly differs among the lakes, and that the highest mean TEP-C concentration was found in the lake with the strongest riverine influence (Lake 129). This result is opposite of what we expected. Moreover, despite the fact that 7 of 9 environmental variables were weakly to moderately related to TEP-C, CDOM on its own accounted for 53% of the TEP-C among the 3 lakes and only the incorporation of suspended bacteria via multiple regression accounted for any additional variation, because of intercorrelation among the variables. Given that CDOM is unrelated to autochthonous autotrophic production within or among the lakes, this suggests a very strong overall role for allochthonous terrestrial DOC in driving TEP formation in these lakes and this role should indeed be strongest in the lake with the strongest river influence (i.e., Lake 129).

[42] Autotrophic production. Because rates of autochthonous photosynthetic production are high in Lakes 56 and 520 [Squires et al., 2009; Tank et al., 2009a], it is puzzling why such production does not seem to play a stronger overall role in driving TEP formation [e.g., Carrias et al., 2002]. Lake 56 was the only lake where total dissolved organic carbon (i.e., DOC) was a significant factor in our relations analysis, and in this case DOC was more strongly related to TEP-C than CDOM (Table 2). This is likely a result of DOC release from the in situ macrophyte community during periods of high photosynthetic production, and this seems to be reflected in the striking relation between TEP-C and pH (Figure 7j). Prior work has established that pH values reflect pCO2 levels in the lake water and are directly related to photosynthetic rates of the macrophyte communities in these lakes [Tank et al., 2009a]. Thermokarst lakes, such as Lake 520, represent an exception to this pattern because of episodic input of allochthonous DOC from their surrounding shorelines fuels metabolic release of CO2 within such lakes that counteracts CO2 drawdown by the macrophytes [Tank et al., 2009a]. Though based on only a few points with pH > 9.5, Figure 7j suggests that that the relation between TEP-C versus in situ photosynthetic production in Lake 56 switches at some threshold point from an inverse one, to one that directly increases with photosynthesis. One possible explanation suggested by prior work is that the bacteria community may be rapidly metabolizing the high quality DOC derived from the macrophytes [Tank et al., 2011] before a substantial amount of TEP is able to form, but the highest pH levels (i.e., >9.5) inhibit such bacterial metabolism [Tank et al., 2009b]. This is consistent with recent tracer analyses that show macrophyte-derived DOC represents a maximum of only 15% of DOM in these lakes despite their high macrophyte production and biomasses [Tank et al., 2011]. However, Chateauvert et al. [2012] show that the abundance of TEP-attached bacteria and its density on TEP both increase in Lake 56 with the increasing TEP-C as pH rises above 9.5. This suggests that pH inhibition of bacterial metabolism may not be strong, or alternatively, that the response of free-living bacteria that utilize DOC (specialists or generalists) may differ in their pH response from the TEP-attached bacteria.

[43] Another possible pH effect is that at values >9.5, we have observed precipitation of CaCO3 that should reduce soluble Ca2+ in the water and potentially inhibit TEP formation, yet TEP-C increased in spite of this. This suggests the possibility that in lake waters with Ca2+ near saturation levels, inhibition of TEP formation may not be an issue and even the possibility that CaCO3 precipitation events might somehow enhance TEP formation by an unknown mechanism. We are not aware of any other work that has investigated TEP in lakes where CaCO3 precipitation regularly occurs. Further investigation is required to resolve the role of autochthonous DOC in this study system.

[44] TSS. We postulated that TEP would decrease with increasing TSS because TSS would aggregate with TEP, adding ballast and causing sedimentation. Our results showed that the relationship was actually opposite, with TEP-C concentrations generally increasing with TSS concentration. While it seems fully plausible that TSS causes some TEP sedimentation, it is difficult to infer to what degree this may occur because of intercorrelation among environmental variables pertinent to TEP formation in this system. High TSS concentrations are generally associated with river flooding that also contains nutrients for autotrophs, DOM-fuel for microbial communities, and possibly allochthonous TEP.

[45] POC. We expected that the role of POC would be complex because of POC composition changing from being rich in terrestrial detritus in lakes with strong river influence to being rich in autochthonous carbon and TEP in lakes with little river influence. POC in Lake 129 did indeed have a composition similar, at least initially to river water, though with unexpectedly high TEP-C concentrations. Lake 56 fit with the expected change in POC composition associated with its higher autotrophic production, but with unexpectedly low levels of TEP. In Lake 520, the nonsignificant relation between TEP and POC was surprising given that TEP seemed to contribute an average of ∼70% of all POC in the lake. The relationship could be affected by TEP-C being overestimated and by complex nutrient recycling and carbon turnover, driven by the grazing activities of large zooplankton which are present in this lake due to the absence of fish [Riedel, 2002].

[46] Suspended Bacteria. We expected that bacteria could facilitate TEP formation via direct release of precursors [Stoderegger and Herndl, 1999; Passow, 2002b], and indirectly via algal colonization and stimulating the release of algal exopolymers and subsequent hydrolysis of mucus coatings [Smith et al., 1995], though we expected indirect facilitation would be more important in Mackenzie Delta lakes [Grossart and Simon, 2007; Sugimoto et al., 2007]. Our results show a modest but direct relation between TEP versus total suspended bacteria, and adding suspended bacteria to a regression model between TEP versus CDOM modestly improved the model over CDOM by itself. Moreover, there was a relationship between TEP and TEP-attached bacteria [Chateauvert et al., 2012] in these lakes that seems to be a simple result of bacterial colonization of TEP. However, there is also a relation between TEP and the very high abundances of other attached bacteria in these lakes. This could be because non-TEP particles are mostly entering the lake waters precolonized with bacteria derived from the flooded terrestrial landscape or from benthic macrophyte communities. This suggests bacteria could be directly facilitating TEP formation, though further work is needed to assess this possibility more directly.

[47] CDOM. We expected the DOC gradient among these lakes would drive a similar gradient in TEP, but we did not expect the TEP gradient would be related CDOM instead of DOC. There are several possibilities to explain this. First, CDOM is considered to be a complex combination of organic molecules of terrestrial origin and some portion of this DOM pool may consist of TEP precursors. Alcian blue binds sulphated and carboxylated polysaccharides [Horobin, 1988]. While production of sulphated polysaccharides is restricted to algal species [Kloareg and Quatrano, 1988], all plants utilize the acidic polysaccharide, pectin in their cells walls [Scheller et al., 2007]. Carboxyl groups associated with pectin should stain with alcian blue. Pectin is also known to form hydrogels in the presence of Ca2+ ions [Thakur et al., 1997], however, this has not been tested under in situ, lacustrine conditions. Second, Orellana and Verdugo [2003] found that UV radiation inhibited the formation of self assembled gel particles. CDOM provides a natural sunscreen to aquatic environments and strongly absorbs UVA and UVB. As the summer proceeds, CDOM levels decline due to photobleaching [Gareis, 2007] resulting in an increase in UV penetration of the lake waters. As CDOM decreases, UV radiation could inhibit TEP formation, especially in the more transparent, higher elevation lakes of the delta. Third, it is also possible that CDOM is merely associated with the waning effect of river water input among the lakes, and it may be of relatively little importance for TEP abundance in the Mackenzie Delta. Further research is required to resolve the potential for terrestrially derived DOC (specifically pectin) to contribute TEP precursors.

[48] Carbohydrates. We expected that TEP would increase with carbohydrate concentrations. We confirmed a weak relation with monosaccharide concentrations that fit this pattern, but also a weak inverse relation with polysaccharide concentrations. Engel et al. [2004] found that dissolved polysaccharide concentration decreased as TEP increased in a lab experiment. This observation could explain our findings, but our weekly sampling provides only a crude sense of shorter time scale dynamics.

4.5. Significance and Limitations of Results

[49] Given that this was a 1 year investigation, we recognize that some caution is warranted in interpreting our results. Because this is part of a long-term investigation of the Mackenzie Delta, we have considerable understanding of its hydrologic variability [e.g., Lesack and Marsh, 2010] and a variety of knowledge about how such variability may affect other pertinent processes in this system. We feel confident that the results have been interpreted in an appropriate context. The fact that 2006 had a higher than normal flood peak, may have attenuated differences between the lakes that might have been stronger in year with a more typical flood peak. We nevertheless detected striking differences in TEP and bacterial communities [Chateauvert et al., 2012] among the lakes that were consistent with our background data on the three study lakes and with other work in our long-term investigations. Our follow-up TEP measurements during 2007 [Chateauvert, 2008], a more typical flood year, also has reinforced our confidence that the system was not behaving in a strongly atypical manner because of the high flood peak.

[50] Our findings that TEP is abundant in arctic floodplain lakes, that it may be a substantial portion of the total POC in such systems, and that much of the TEP appears to come from terrestrial sources are important from three perspectives. First, whereas TEP is now established as an important pathway for conversion of dissolved organic matter to particulate form that may remove carbon from the upper layer of the ocean, little is thus far known about the role of TEP in freshwater or how abundant such particles might be given the much lower ionic strength of freshwaters relative to the ocean. Our paper raises more questions than it answers about differences between marine and freshwater systems, such as the role CDOM or the role of CaCO3 precipitation, but such questions should lead to deeper insights about the formation and dynamics of TEP more generally. Second, whereas water and carbon fluxes from the great Arctic rivers may be driving the Arctic Ocean to behave as the world's largest estuary [McClelland et al., 2012], the role of carbon transformations during transit within these rivers and within the large complex deltas at the ocean interface is poorly understood. Our results here suggest conversion of DOC to TEP may be of analogous importance to riverine carbon fluxes, as are in situ transformations in the ocean. Third, the fundamental nature of carbon processing within lakes, and how this is affected by their surrounding terrestrial habitat, has received considerable attention [e.g., Cole et al., 2011], including the functional behavior of such ecosystems as sources and sinks of carbon and greenhouse gases [Cole et al., 2007; Prairie, 2008]. However, the role of TEP is thus far unrepresented in such models. Our results suggest the role of TEP should be assessed in carbon and foodweb models in the lake-rich environments of Arctic river floodplains, and whether TEP might have comparable importance in other types of lakes.

5. Conclusions

[51] Differing river-to-lake connection times in the Mackenzie River Delta appear to drive differing abundances of TEP, both seasonally and among lakes, though not as expected. Volumetric TEP abundances were highest immediately after the flood, when autochthonous autotrophic production is at a seasonal low point and CDOM is at a seasonal high point. Also contrary to our expectation, the lake with the strongest riverine influence and lowest levels of autochthonous autotrophic production had the highest mean TEP-C concentrations among the lakes. The mean proportion of POC represented by TEP increased with increasing river connection time, and appears to represent a large proportion (overall ∼84%) of POC in Mackenzie Delta Lakes. Whereas this value is likely an overestimate, it indicates that TEP represents a significant pathway for the transfer of carbon from dissolved to particulate phase in this type of aquatic ecosystem. We expected the DOC gradient among these lakes would drive a similar gradient in TEP, but we did not expect the TEP gradient would be related most strongly to CDOM instead of DOC. Moreover, CDOM on its own accounted for 53% of the variation in TEP-C among the 3 lakes. Because of intercorrelation among the variables, incorporation of other potentially pertinent environmental variables via multiple regression did not account for any additional variation of TEP-C, except for some modest additional variation in the case of suspended bacteria. DOC release from the in situ macrophyte community during periods of high photosynthetic production may indeed contribute significantly to TEP formation in the lake with lowest riverine and allochthonous influence, but pH levels >9.5 driven by the high photosynthetic rates complicate the interpretation of the results from this lake. Further investigation on TEP formation is needed in river influenced lakes with complex mixtures of DOC-quality and in autotrophically productive lakes with high pHs.


[52] We appreciate the assistance provided by Suzanne Tank, Jolie Gareis, Leah Honka, and Emily Hines in the lab and field. Technical and logistical assistance was provided by Sharon Katz and William Hurst and the Inuvik Research Centre/Aurora Research Institute. Financial support was received from NSERC (DGP and NRS programs to LL); the Polar Continental Shelf Project (helicopter support to LL); the Northern Scientific Training Program - Department of Indian and Northern Affairs; Science Horizons Youth Internship program - Environment Canada; and facilities use in Inuvik was subsidized by the Aurora Research Institute. Helpful thesis advice was provided by Leah Bendell-Young (SFU) and Jeff Curtis (UBC Okanagan).