Quaternary Research 81 (2014) 251–259

Contents lists available at ScienceDirect

Quaternary Research

j ourna l homepage: www.e lsev ie r .com/ locate /yqres

Consensus among multiple trophic levels during high- and low-waterstands over the last two millennia in a northwest Ontario lake

Moumita Karmakar a,⁎, Joshua Kurek a, Heather Haig b, Brian F. Cumming a

a Paleoecological Environmental Assessment and Research Laboratory (PEARL), Department of Biology, Queen's University, Kingston, Ontario K7L3N6, Canadab Department of Biology, University of Regina, Laboratory Building, Saskatchewan S4S0A2, Canada

⁎ Corresponding author.E-mail addresses: 9mk65@queensu.ca (M. Karmakar),

(J. Kurek), haig.a.heather@gmail.com (H. Haig), brian.cum(B.F. Cumming).

0033-5894/$ – see front matter © 2014 University of Washttp://dx.doi.org/10.1016/j.yqres.2013.12.006

a b s t r a c t

a r t i c l e i n f o

Article history:Received 13 March 2013Available online 17 January 2014

Keywords:Boreal lakesMedieval Climate AnomalyLake levelsChironomidsDiatoms

We investigated themodern distribution of fossil midges within a dimictic lake and explored downcore patternsof inferred lake depths over the last 2000 years from previously published proxies. Modern midge distributionwithin Gall Lake showed a consistent and predictable pattern related to the lake-depth gradient with recogniz-able assemblages characteristic of shallow-water, mid-depth and profundal environments. Interpretations ofdowncore changes in midge assemblages, in conjunction with quantitative lake-depth inferences across a prioridefined (based on diatom data) ~500-yr wet and dry periods, demonstrated that both invertebrate and algalassemblages exhibited similar timing and nature of ecological responses.Midgeswere quantifiedby their relativeabundance, concentrations and an index of Chaoborus to chironomids, and all showed the greatest differencesbetween the wet and dry periods. During the low lake-level period of the Medieval Climate Anomaly (MCA:AD 900 to 1400), profundal chironomids declined, shallow-water and mid-depth chironomids increased,chironomid-inferred lake level declined and the Chaoborus-to-chironomid index decreased. The coherencebetween multiple trophic levels provides strong evidence of lower lake levels in Gall Lake during the MCA.

© 2014 University of Washington. Published by Elsevier Inc. All rights reserved.

Introduction

To expand our knowledge of climate change in northwest Ontarioover the past 2000 years, Laird et al. (2012) assessed the drought signalfrom near-shore sediment cores taken from a network of six smalloligotrophic to slightly mesotrophic lakes. These lakes spanned adistance of ~250 km across the boreal forest region of the WinnipegRiver Drainage Basin (WRDB). The coreswere retrieved from a sensitivelocation in each lake, just deeper than the modern transition fromplanktonic diatoms to assemblages with increasing contributions frombenthic taxa (Laird et al., 2011; Kingsbury et al., 2012). At this ecologi-cally relevant location, a small change in either light or lake depth canresult in a shift in diatom assemblages, thereby providing the criticallocation for detecting drought through declines in water level and asso-ciated variables (details in Laird et al., 2011). This spatial network ofsites provided clear evidence that a prolonged period of aridity occurredduring AD 900–1400 (Medieval Climate Anomaly), and that conditionsover the last 100 years were indicative of relatively high effectivemoisture (Laird et al., 2012), in agreement with regional hydrologicalmonitoring (St. George, 2006; Parker et al., 2009).

The strength of quantitative (and by extension qualitative) recon-structions of lake level from biological proxies (diatoms) has been

joshua.kurek@gmail.comming@queensu.ca

hington. Published by Elsevier Inc. A

challenged (Juggins, 2013). The main criticism is based on the inabilityof modern calibration data sets to produce a consistent pattern of dia-tom optima to lake depth across regions. It was concluded that depthhas limited or no direct effect on biota but acts as a surrogate variablefor many underlying environmental factors. We agree that lake depthis a ‘composite’ and potentially complex environmental variable, but itis inextricably coupled to a range of biologically important variables in-cluding temperature, substrate, wave action, light, food availability, andchanging predator–prey interactions. Nonetheless, many studies showrepeatable and well-defined relationships to water depth for diatom(Kingsbury et al., 2012) and chironomid assemblages (e.g., Kurek andCwynar, 2009a; Engels and Cwynar, 2011; Luoto, 2012).

To further evaluate the sensitivity of biotic assemblages to historiclake-level shifts over the last 2000 years, we compared midge assem-blages to independent inferences of changes in lake depth that werederived from diatom assemblages in Gall Lake (Laird et al., 2012; Haiget al., 2013). The strength of the relationship between midge assem-blages and lake depth was assessed across a depth gradient in GallLake, thereby maximizing the possibility of detecting changes inmidge assemblages with lake depth. Midge assemblages from three apriori chosen ~500-yr periods of diatom-inferred water level (onelow-water level and two high-water levels) were analyzed. Specifically,we predict that if midges are sensitive to changes in depth then midgeassemblages during the arid period should be represented by moreshallow-water taxa, and the abundance of Chaoborus and deep-waterchironomid taxa should decrease. Alternatively, if midges don't showa significant and consistent change, then eithermidges are not sensitive

ll rights reserved.

252 M. Karmakar et al. / Quaternary Research 81 (2014) 251–259

to subtle changes in lake depth or the quantitative diatom inferencesmay not accurately represent lake-level changes in Gall Lake.

Methods

Study site

Gall Lake (50°14′N, 91°27′W) is located in northwest Ontario,Canada, within the eastern region of the English River Watershed(Fig. 1). The study site is ~20 km northwest of Sioux Lookout and islocated within a forested region with minimal human disturbance(i.e., mainly forested, no cottages, and minimal human activity in thecatchment). Average summer temperature in Sioux Lookout is 17.1°Cand average annual precipitation is ~820 mm (Environment Canada,http://ec.gc.co/dccha-ahccd). The catchment is characterized as borealforest and is dominated by black spruce (Picea mariana), jack pine(Pinus banksiana), and poplar (Populus spp.), along with white birch(Betula papyrifera), balsam fir (Abies balsamea), and larch (Larix laricinaspp.). Gall Lake is a small, headwater lake with a surface area of 19 haand a maximum water depth of 18 m. The eastern basin has a gentlysloping bottom from which surface-sediment samples were collected(Fig. 1b). Water chemistry indicates that Gall Lake (July, 2008,Kingsbury et al., 2012) has a low specific conductance (~25 μS/cm),low-to-moderate concentrations of total phosphorus (TP = 12.3 μg/L)and a relatively high concentration of dissolved organic carbon(DOC = 15.8 mg/L). During June 2010 and 2011, the metalimnionextended between 4 and 6 m at Gall Lake.

Surface sample collection and midge processing

To quantify the distribution of modern-day midge assemblages,subfossil remains were identified in surface-sediment samples (top0–1 cm) taken from Gall Lake. Samples were collected along a transectat ~1-m depth increments to the center of the lake (Fig. 1b) using agravity corer (Glew, 1989). Samples from N15-mdepthwere absent be-cause sufficient sediment was no longer available. Standard procedureswere followed to isolate midge remains from the surface sediments(Walker, 2001). Briefly, 15 g of wet sediment was deflocculated in 5%KOH for ~10 min at 200°C, and then rinsed with running tap waterthrough a 100-μm mesh sieve. Using a Bogorov counting tray and fine

Figure 1. a) Map showing the location of theWinnipeg River Drainage Basin, comprising the EnLaird et al., 2011). b) Bathymetry of Gall Lake with sample locations indicated. Contour lines rethat was taken from a depth of ~7.5 m.

forceps, chironomid head capsules and Chaoborus mandibles werepicked by hand under a dissection microscope at 20× magnification.Specimens were then placed onto cover slips and mounted on slidesusing Entellan®. Chironomid head capsules were identified usingWiederholm (1983) and the fossil key of Brooks et al. (2007). Chaoborusmandibles were identified using the fossil key of Uutala (1990). A min-imum equivalent of 50 whole chironomid head capsules were countedper sample (Heiri and Lotter, 2001; Larocque, 2001; Quinlan and Smol,2001).

Assessment of changes in midges over the last 2000 years

Three a priori 500-yr periods that represent primarily moremesic orarid conditions based on quantitative inferences from diatom assem-blages (Fig. 2, Haig et al., 2013) were chosen for analysis of midges.These periods include two relatively mesic periods (from ~AD 100 to600, and from ~AD 1400 to 1900; henceforth termed Wet 1 and Wet2, respectively), and a period of aridity during the MCA (~AD 900 to1400; henceforth termed Dry 1). Diatom-inferred (DI) depths duringtheWet 1 andWet 2 generally ranged between 10 and12 m, and duringDry 1 ranged between 8 and 10 m.Wet 1 is generally represented by anabundance of between ~65 and 85% planktonic diatom taxa. Wet 2 wasmore variable and represents an initial transition from arid conditionsto a period of more mesic conditions, with planktonic diatom taxarepresenting ~60% of the relative abundance during the 1400s, to ~75to 80% relative abundance between AD 1600 and 1900 (Haig et al.,2013). During the MCA (Dry 1) the percentage of planktonic diatomswas typically b70% (Laird et al., 2012; Haig et al., 2013). The subtlebut consistent changes among these periods, sets the stage for thisstudy where we assess whether midge assemblages show a responseconsistent with the diatom inferences.

Chironomid head capsules and Chaoborusmandibles were analyzedin the top 50 cm of a near-shore piston core taken in 2010 from awaterdepth of ~7.5 m.We sectioned the core into 0.25-cm intervals and sam-ples were analyzed every other interval, resulting in bi-decadal tempo-ral resolution during each of the 500-yr periods. The chronology of thecore is well established and based on both 210Pb and AMS 14C dates(Laird et al., 2012; Haig et al., 2013). Total 210Pb was measured on alow-background gamma counter (Schelske et al., 1994) and sedimentage was determined using a constant rate of supply model (Binford,

glish River Watershed and the Lake of theWoods/Rainy River Watershed (modified frompresent depth intervals of 2 m. The black square represents the location of the piston core

Figure 2.Diatom-inferredwater depths from the Gall Lake near-shore core (Haig et al., 2013). Changes in the diatom assemblages and depth-based inferences were used to define three apriori 500-yr periods that provide the framework for the present study (Wet 1, AD 100–600; Dry 1, AD 900–1400; and Wet 2, AD 1400–1900).

253M. Karmakar et al. / Quaternary Research 81 (2014) 251–259

1990). Four AMS 14C dates were determined for the deeper intervalsbased on AMS dates of pollen isolated from the sediments. Sedimentage was based on the midpoint of the 2-sigma range with the highestrelative area under the probability distribution using CALIB 6.01(Reimer et al., 2009). Complete details of the age–depth model are inHaig et al. (2013).

Numerical analyses of surface assemblages

Chironomid countswere expressed as percent relative abundance ofall identifiable chironomids. Chaoborusmandibles were expressed as anindex of Chaoborus relative to chironomids [(# of Chaoborusmandibles/2)/((# of Chaoborusmandibles/2) + (# of chironomid head capsules))](Quinlan and Smol, 2010). This index is henceforth referred to as theChaoborus-to-chironomid index. The dominant pattern of variation inthe modern chironomid assemblages in Gall Lake was summarized byprincipal components analysis (PCA) using the computer programCANOCO v. 5 (Ter Braak and Šmilauer, 1998), based on square-roottransformed relative abundances. A depth-constrained cluster analysiswas used to identify groups of similar chironomid assemblages in themodern surface sediments, with squared-chord distance as a measureof dissimilarity (Grimm, 1987). Weighted-averaging (WA) was usedto provide an estimate of the depth optimum for each chironomidtaxon, with the weight being the relative abundance of a taxon at eachdepth. These weighted averages were then used to assign individualchironomid taxa to one of the three groups identified in the cluster anal-ysis: a shallow-water, amid-depth, and a profundal assemblage.WA re-gression and calibration with inverse deshrinking were then used todevelop and assess a predictive model for lake depth from chironomidassemblages. This model was developed using the program C2 1.6.5(Juggins, 2003).

Numerical analysis of downcore assemblages

The same taxonomic references and counting strategies describedabove were used to analyze the midge assemblages in the sedimentcore over the last 2000 years. However, midge assemblages wereexpressed in two ways: i) as relative abundances (and/or indices), de-scribed above; and ii) as concentrations of chironomids and Chaoborusper gram dry weight of sediment. Individual chironomid taxa werealso assigned to the groups defined above (shallow-water, mid-depthand profundal assemblages) to summarize the assemblage shifts. Corre-lations between the relative abundance data and the concentration datawere run on the three groups to determine the degree of similaritybetween these metrics.

To assess if the chironomid assemblages were significantly differentbetween the two 500-yr mesic periods (Wet 1 andWet 2) and the aridMCA (Dry 1), an Analysis of Similarity (ANOSIM) was performed usingPRIMER 6.0 (Clarke, 1993). ANOSIMwas runon relative abundancedataafter removing rare taxa (those taxa b3% in at least one sample), andsamples within the 500-yr periods were treated as replicates. ANOSIMis a non-parametric test that can differentiate between a priori definedgroups. In the ANOSIM, a test-statistic (R) is generated based on ranksimilarities calculated within and between groups (Clarke, 1993). IfR = 1 then all similarities within groups are greater than the similari-ties between groups. To identify which groups differed from eachother, multiple pairwise ANOSIMs were run (i.e., Wet 1 versus Dry 1,Wet 1 versus Wet 2, and Dry 1 versus Wet 2). We hypothesized that,if the chironomid assemblages follow similar directional change, thenWet 1 and Wet 2 would be more similar, in comparison with eitherWet 1 and Dry 1, orWet 2 and Dry 1. To aid in the comparison of chang-es in diatom andmidge assemblages over the last 2000 years, a summa-ry of the percent profundal chironomids and planktonic diatoms werecompared, as were diatom and chironomid inferences of lake depth.

254 M. Karmakar et al. / Quaternary Research 81 (2014) 251–259

The chironomid-inferred depth reconstruction was assessed followingprocedures outlined by Telford and Birks (2011). If our reconstructionexplains more variation than 95% of reconstructions (n = 499) gener-ated frommodels trained on random data, the reconstruction was con-sidered statistically significant.

Results

Midge assemblages from the modern surface sediments of Gall Lake

Chironomid assemblages showed a distinct pattern of variationacross the gradient of lake depth, and three depth-related zones wereidentified based on cluster analysis (Fig. 3). The profundal or deep-water assemblage was characterized by the highest abundances ofChironomus anthracinus-type and Sergentia coracina-type (10% and 50%,respectively). The mid-depth assemblage was characterized by a declin-ing abundance of S. coracina-type and increased relative abundances ofMicropsectra contracta-type, and Micropsectra insignilobus-type from~10 to 6.25 m water depth (Fig. 4). The shallow-water assemblage wasgenerally characterized by higher abundances of taxa including:Cladopelma, Microtendipes pedellus-type, Tanytarsus, Polypedilum,Tanytarsus mendax-type, Pagastiella, Tanytarsus chinyensis-type, andZalutschia. Several taxa were relatively ubiquitous across the depthgradient such as Procladius, Psectrocladius sordidellus-type, whereasPsectrocladius (Monopsectrocladius), Heterotrissocladius marcidus-type,and Tanytarsini (which likely include several taxa that are difficult toidentify given the fragmentation of head capsules) showed relativelyhigh abundances in the shallower to mid-depth region (Fig. 3). Thethree zones identified by the cluster analysis were also clearly supportedby the first two PCA axes of the modern chironomid assemblages.The first axis differentiated the profundal taxa, C. anthracinus andS. coracina-types from the more diverse shallow-water assemblage,whereas the mid-depth taxa were represented by negative loadings onthe second axis, primarily from M. contracta and M. insignilobus-types.Chaoborus were represented by two taxa, Chaoborus trivittatus andChaoborus flavicans-types. Similar to the transition in chironomid assem-blages, the Chaoborus-to-chironomid index was much higher in the

Figure 3. Distribution of the relative abundance of the dominant (N3%) chironomid taxa, the Cacross a gradient of depth in Gall Lake. Chironomid taxa are arranged based on their weighted

deeper waters and declined in an approximately linear fashion to 0 by~3 m water depth (Fig. 3). A strong predictive model for lake depth,based on theWA optimum of the 29 chironomid taxa present in the sur-face sediments of Gall Lake, was developed (bootstrapped r2 = 0.80;RMSEP = 1.95 m, n = 31, Supplemental 1).

Downcore midge and diatom assemblages

Generally, the abundances (Fig. 5) and concentrations (Fig. 6) ofprofundal taxa (e.g., S. coracina) are high during the wet periods andlow during the dry period from AD 900 to 1400. The dry period wasalso characterized by higher abundances of several mid-depth (e.g.,Tanytarsus pallidicornis-type, Polypedilum nubeculosum-type) andshallow-water taxa (e.g., Polypedilum and Pagastiella) (Fig. 5). Therewere two shallow-water taxa, M. pedellus-type and Tanytarsus, thatwere only present in Wet 1, and not Wet 2. The differentiation between500-yr periods was visually more apparent when the chironomid taxawere grouped into depth categories based on their abundance in themodern-day surface samples (Fig. 6). In Wet 1 and Wet 2, the relativeabundance of profundal taxa ranged from ~20 to 50%, whereas in Dry1, the abundance of profundal taxa was consistently b20% (Fig. 6).Trends in the relative abundance data were similar to the patterns ob-served in the concentration data, which were significantly correlatedfor the profundal (r = 0.85, n = 73, p b 0.05), mid-depth (r = 0.69,n = 73, p b 0.05), and shallow-water (r = 0.80, n = 73, p b 0.05)groupings (Fig. 6). The Chaoborus-to-chironomid index was higher dur-ing the wet periods in comparison to Dry 1 (Fig. 6). During Dry 1, theChaoborus-to-chironomid index reached the lowest sustained values ofall three periods. The Chaoborus-to-chironomid index was significantlycorrelated to the concentration of Chaoborus (r = 0.93, n = 73,p b 0.05).

The ANOSIM results confirmed that the chironomid assemblageswere significantly different between all three 500-yr periods (GlobalR = 0.49, p b 0.001). However, the pairwise comparison between theWet 1 and Dry 1 and Wet 2 and Dry 1 were higher (R = 0.71, 0.61,respectively, p b 0.001 for both) than between Wet 1 and Wet 2(R = 0.31, p b 0.001). Chironomid taxa that exhibited the greatest

haoborus-to-chironomid index, and the PCA axis-1 scores of the chironomid assemblages-average optima. Three zones were defined by cluster analysis.

Figure 4.PCAbiplot of chironomid assemblages. Thefirst axis separates theprofundal taxa (open triangles) C. anthracinus and S. coracina-types fromthemore diverse shallow-water (openrectangles) and mid-depth assemblages (open circles), which themselves are defined based on their depth optima.

255M. Karmakar et al. / Quaternary Research 81 (2014) 251–259

mean differences between the wet and dry periods were profundal taxasuch as S. coracina-type and C. anthracinus-type, and shallow-water taxasuch as Polypedilum and Pagastiella, and mid-depth taxa such as

Figure 5. The relative abundance of the dominant chironomid taxa (N3%) in the 500-yr time pchironomid taxa are arranged according to their weighted-average depth optima.

P. nubeculosum-type and T. pallidicornis-type. Using untransformed rela-tive abundance data, the chironomid-based reconstruction explainedsignificantly (p = 0.03) more variation than 95% of reconstructions

eriods. Because of insufficient sediment availability, ~AD 600–900 was not included. The

Figure 6. Summary diagram showing relative abundances and concentrations of the three chironomid assemblage groups (shallow, mid-depth, and profundal), in the three a prioridefined 500-yr periods (Wet 1, Dry 1 and Wet 2). Chaoborus-to-chironomid index and the concentration of Chaoborus taxa are also shown.

256 M. Karmakar et al. / Quaternary Research 81 (2014) 251–259

generated from models developed with random environmental data(Telford and Birks, 2011).

Comparison of the response of chironomid assemblages within thezones previously described by the diatom assemblages were similarand support a conclusion of overall lower water levels in Gall Lake dur-ing the MCA (Dry 1). Dry 1 is defined by a decline in planktonic diatomtaxa from 70 to 80% consistently less than 70% between AD 900 and1400 (Fig. 7A), and diatom-inferred decreases in lake level of just over1 m (Fig. 7B). This trend during Dry 1 is mirrored in the chironomid as-semblages with lower abundances of profundal chironomids (Fig. 7A)and sustained lower chironomid-inferred water levels of ~1–2 m(Fig. 7). Similarly, during the two wet periods, the abundance ofprofundal taxa doubles, and inferred water levels are ~1–2 m higherthan the average over the last 2000 years (Fig. 7).

Discussion

Based on the lack of reproducible depth optima in diatom assem-blages in lake calibration data sets, Juggins (2013) concluded that com-posite variables such as lake depth need to be interpreted with greatcaution. Despite the recognized challenges with depth reconstructions(Kurek and Cwynar, 2009b; Velle et al 2012), we contend that depthis an environmental variable that represents a biologically importantvariable in both space and time. In this paper we examine thepresent-day distribution of midges within a lake and also downcorechanges, and use this information to assess if diatom-based inferencesof changes in lake depth are valid.

Modern-day distribution of midge assemblages

Gall Lake represents an excellent site fromwhich to examine the dis-tribution ofmidge assemblages across a gradient of lake depth. First, themorphometry of the lake, with a gently sloping eastern basin enabledassemblages to be sampled at a sub-meter resolution. This study with

even sample distribution allowed a robust representation of depths, afactor not common in most data sets examining the relationshipbetween midge assemblages and depth-associated gradients (Kurekand Cwynar, 2009a; Luoto, 2010; Cwynar et al., 2012; Luoto, 2012). Inaddition to the simple lake morphometry, Gall Lake generally lacksmacrophytes, except for a few emergent macrophytes at low densities,resulting in a less complicated nearshore environment. A drawbackof the transect design is that the environmental variable (depth)and chironomid species assemblages would be strongly spatiallyautocorrelated. However, our goal was to maximize the assemblagesignal to depth. Quantitative and comparative assessments of spatial au-tocorrelation fromwithin-lake chironomid distribution surveys showedlittle evidence that spatial autocorrelation is a significant concern (Velleet al., 2012; Engels et al., 2012; Luoto, 2012).

Three broad categories of chironomid assemblages were recognizedacross a depth gradient in Gall Lake and included: a shallow-water or lit-toral assemblage, a mid-depth assemblage, and a profundal assemblage.The profundal assemblage is relatively stable at water depths N10 m andis primarily composed of S. coracina-type and C. anthracinus-type. Thesetwo taxa reached peak abundances in the profundal zone and aregenerally absent at depths b5 m. S. coracina-type, themost abundantdeep-water taxon in Gall Lake, is a cold stenotherm (Brundin, 1956;Brodin, 1986) common to the profundal zone of deep lakes (Barleyet al., 2006; Kurek et al., 2012). S. coracina-type was also identifiedas a deep-water taxon in two site-specific intra-lake data sets(Austrian Alps and Finland) and one regional multi-lake data set(Luoto, 2012). In the Experimental Lakes Area of Ontario, Sergentiawas abundant in lakes of at least ~8 m depth (Quinlan et al., 2012).C. anthracinus-type was present at a relative abundance of ~10% atdepths N10 m in Gall Lake. According to Engels and Cwynar (2011),C. anthracinus-type is a deeper-water taxon. Other research hasfound C. anthracinus-type often abundant in the profundal regionand has been described as an important indicator of low-to-moderate oxygen conditions (Brodin, 1986).

Figure 7. a) Summary diagram between the relative abundance of profundal chironomid taxa in comparison to changes in the relative abundance of planktonic diatoms. b) Summary ofchanges in deviation from mean lake depth inferred for both the chironomid (CI-depth)-based and diatom (DI-depth)-based models over the last 2000 years.

257M. Karmakar et al. / Quaternary Research 81 (2014) 251–259

The high Chaoborus-to-chironomid index at deeper locations in GallLake also supports this index as another proxy of lake depth. Chaoborus,both C. flavicans and C. trivittatus-types, were important in Gall Lake rel-ative to chironomids at depths greater than 10 m, reaching abundancesof 40%of the combinedmidge assemblages. This index declines graduallywith depth fromover 40% to essentially zero by a depth of ~3 m. Howev-er, this index is complicated by the fact that it relies on both a changingnumerator (number of Chaoborus) and denominator (number ofChaoborus + number of chironomids), and can shift based on an in-crease or decrease in either chironomids or Chaoborus. The observationof higher index values at greater depths is not a surprise as Chaoborus,such as C. flavicans and C. trivittatus-types, often prefer deeper waterand can tolerate moderate levels of predation from planktivorous fish(Kurek et al., 2010, 2011). However, this trend could also arise if chiron-omidsweremore abundant innear-shore areas and at lower abundanceswith increasing depth. Quinlan and Smol (2010) suggested that thisindex could be used as a proxy of anoxia. A high index value would

arise if chironomids declined as a result of deteriorating oxygen condi-tions, as Chaoborus are tolerant of lower oxygen conditions (Quinlanand Smol, 2010), and have the ability to migrate to regions of higheroxygen, unlike profundal chironomids. Given the range of possible inter-pretations, changes in this index is problematic without additional long-term information on either the lake (e.g., water-quality, planktivorousfish populations) or other information from the core including the con-centrations of both chironomids and Chaoborus.

We also identified a mid-depth chironomid assemblage, whichcan be viewed as a transition between habitats. Although the mid-depth zone is not as distinct as the other two zones in terms ofassemblage composition, it can be defined by several taxa and bythe declines or increases of the dominant taxa of the adjacentzones and is clearly defined on axis 2 of the PCA (Fig. 4). For example,members of the Tanytarsini group common to mid-depth include:M. contracta-type, M. insignilobus-type, and T. pallidicornis-type(Fig. 3).

258 M. Karmakar et al. / Quaternary Research 81 (2014) 251–259

The shallow-water chironomid assemblage in Gall Lake occurred atdepths generally b6 m and consisted of a diverse assemblage of manytaxa at lower % abundance in comparison to the few, but more abun-dant, chironomid taxa in the profundal zone. There is reasonable agree-ment with shallow-water taxa observed in Gall Lake and other studylakes from north-temperate regions. Tanytarsus, Microtendipes,Pagastiella, Cladopelma, and Zalutschia are known to prefer the littoralzone (e.g., Schmah, 1993; Walker and MacDonald, 1995; Kurek andCwynar, 2009a). We also observed that P. nubeculosum-type is an indic-ative of the nearshore environment similar to Kurek and Cwynar(2009a), where they found Polypedilum at shallow depth.

While depth itself is not a direct, causal factor associated with thedistribution of midges and diatoms, there are certainly a myriad offactors that co-vary in a consistent fashion with depth over time.Juggin's (2013) conclusion of a lack of consistency in diatom optima be-tween data sets may partially rest in complexities associated withbetween-lake versus within-lake calibration data sets. Generally, the‘noise’ associated with other environmental variables (i.e., nuisance orsecondary gradients) in between-lake data sets is relatively large. Thiscan obscure the relationship of biological assemblages to depth, butthis does not imply that depth-specific habitat preference is ecologicallyunimportant. Given the robust changes in chironomid assemblageswith depth, and the strong inference model based on the optima of chi-ronomid taxa, reconstruction of changes in lake depth and associatedfactors are certainly possible, and environmentally realistic. However,given the complexities of these reconstructions,multiple proxies shouldalways be examined to support inferences of changes in lake depth.

Do midges support inferences of changes in lake level over the past2000 years?

Given the difficulties associated with the complexities of a compos-ite variable such as lake depth,multiple lines of evidence should be usedto assess environmental inferences associated with historic lake-levelshifts. In present-day drainage lakes, Laird et al. (2011) summarize theevidence that supports core location as being critical for detectingchanges in diatom assemblages associated with a decline in lakedepth. Core location is also critical for detecting a signal with midgesand likely other common invertebrate indicators (e.g., Cladocera).Because chironomid assemblages at lake-water depths greater than~8–10 m exhibit little change in overall assemblage structure (Fig. 3,Supplemental 1), and associated inferences of lake-depth plateau atdepths greater than ~10 m, different coring locations within the lakewill represent different sensitivities to detect shifts in lake level. Forexample, a core at Gall Lake taken at a depth of ~8 m would likely bean ideal location to detect changes in lake level, whereas a core takenfrom a deeper location would be less sensitive (i.e., further away fromthe depth at which assemblage changes related to depth are occurring;see Haig et al., 2013; Ma et al., 2013).

To assess the utility of midges as indicators of changes in lake depthover time, we examined midge assemblages in a core taken from adepth of 7.5 m in Gall Lake. Assessment of changes in midge assem-blageswas based on a number ofmetrics including: relative abundancesand concentrations; a simplification of chironomid assemblagessummarized as shallow, mid-depth and profundal categories; depth re-constructions based on overall chironomid assemblage composition;and changes in theChaoborus-to-chironomid index, and the overall con-centration of Chaoborus taxa. Consistent with the diatom-based infer-ences, all of these metrics suggest an interpretation of a drop in waterlevel during the MCA (Dry 1) relative to the 500-yr periods precedingand following this period. A reconstruction of lake depth clearly indi-cates that lake level was ~1–2 m lower throughout theMCA, in compar-ison to the higher inferred lake levels during the Wet 1 and Wet 2periods (Fig. 7). The consistency of inferences during these 500-yr pe-riods between diatom and chironomid proxies (Fig. 7) strongly suggeststhat Gall Lake experienced reduced water levels during the MCA,

thereby adding additional support for a regional MCA signal (Lairdet al., 2012).

Conclusion

Qualitative and quantitative reconstructions of lake depth can leadto environmentally relevant and correct interpretations, as lake depthis an important but indirect composite variable. Our findings suggestthat knowledge about modern-day ecology and the distribution ofmidge assemblages are fundamental to understanding historic lakelevels at Gall Lake. Additionally, midge-based inferences of lake levelsprovide more evidence that the MCA in northwestern Ontario wasarid. Multi-proxy paleolimnological studies based on informed lakeselection and coring location, limnological-based mechanisms, andspatially-extensive networks of lakes near ecotonal boundaries needto be developed to advance our understanding of climate changeimpacts on freshwaters. Such information enables water managersand policy makers to define realistic scenarios of hydrological changesthat may occur in the near future.

Supplementary data to this article can be found online at http://dx.doi.org/10.1016/j.yqres.2013.12.006.

Acknowledgments

We thank Melanie Kingsbury, Brendan Wiltse, Karlee Flear, SusanMa and Donya Danesh for the fieldwork assistance. We are grateful toKate Laird and other PEARL members for the constructive commentson earlier versions of the manuscript, as well as useful comments fromthree anonymous reviewers. Water chemistry was provided throughthe assistance of Andrew Paterson at the Ontario Ministry of Environ-ment, Dorset Environmental Center. The project was funded by anNSERC Discovery Grant to Brian F. Cumming.

References

Barley, E.M., Walker, I.R., Kurek, J., Cwynar, L.C., Mathewes, R.W., Gajewski, K., 2006. Anorthwest North America training set: distribution of freshwater midges in relationto air temperature and lake depth. Journal of Paleolimnology 36, 295–314.

Binford, M.W., 1990. Calculation and uncertainty analysis of 210Pb dates for PIRLA projectlake sediment cores. Journal of Paleolimnology 3, 253–267.

Brodin, Y.W., 1986. The postglacial history of Lake Flarken, south Sweeden, interpretedfrom subfossil insect remain. Internationale Revue der Gesamten Hydrobiologie 71,371–432.

Brooks, S.J., Langdon, P.G., Heiri, O., 2007. The identification and use ofPalaearcticChironomidae larvae larvae in paleoecology. Quaternary ResearchAssociation Technical Guide 10, 1–276.

Brundin, L., 1956. ZurSystematik der Orthocladiinae (Dipt, Chironomidae). Report of theInstitute of Freshwater Research, Drottningholm, 37 5–185.

Clarke, K.R., 1993. Non-parametric multivariateanalysis of changes in communitystructure. Australian Journal of Ecology 18, 117–143.

Cwynar, L.C., Rees, A.B.H., Pedersen, C.R., 2012. Depth distribution of chironomids and anevaluation of site-specific and regional lake-depth inference models: a good modelgone bad? Journal of Paleolimnology 48, 517–533.

Engels, S., Cwynar, L.C., 2011. Changes in fossil chironomid remains along a depth gradi-ent: evidence for common faunal threshold within lakes. Hydrobiologia 665, 15–38.

Engels, S., Cwynar, L.C., Rees, A.B.H., Shuman, B.N., 2012. Chironomid-based water depthreconstruction: an independent evaluation of site-specific and local inferencemodels.Journal of Paleoecology 48, 693–709.

Glew, J.R., 1989. A new trigger mechanism for sediment samplers. Journal of Paleolimnology2, 241–243.

Grimm, E.C., 1987. CONISS — a Fortran-77 program for statistigraphically constrainedcluster analysis by the method of incremental sum of squares. Computers &Geosciences 13, 13–35.

Haig, H., Kingsbury, M.V., Laird, K.R., Cumming, B.F., 2013. A multiproxy assessment ofdrought over the past two millennia in near-shore sediment cores from a CanadianBoreal Lake. Journal of Paleolimnology 50, 175–190.

Heiri, O., Lotter, A.F., 2001. Effect of low counts on quantitative environmental reconstruc-tions: an example using subfossil chironomids. Journal of Paleolimnology 26, 343–350.

Juggins, S., 2003. C2 software for ecological and palaeoecological data analysis and visual-ization. User Guide Version 1.3.University of Newcastle, Newcastle, UK.

Juggins, S., 2013. Quantitative reconstructions in paleolimnology: new paradigm or sickscience? Quaternary Science Reviews 64, 20–32.

Kingsbury, M.V., Laird, K.R., Cumming, B.F., 2012. Consistent patterns in diatom assem-blages and diversity measures across water-depth gradients from eight Boreal lakesfrom northwestern Ontario (Canada). Freshwater Biology 57, 1151–1165.

259M. Karmakar et al. / Quaternary Research 81 (2014) 251–259

Kurek, J., Cwynar, L.C., 2009a. Effect of within-lake gradients on distribution of fossil chi-ronomids frommaar lakes in Western Alaska: implications for environmental recon-structions. Hydrobiologia 623, 37–52.

Kurek, J., Cwynar, L.C., 2009b. The potential of site-specific and local chironomid-based in-ference models for reconstructing past lake-level. Journal of Paleolimnology 42, 37–50.

Kurek, J., Cwynar, L.C., Weeber, R.C., Jefferies, D.S., Smol, J.P., 2010. Ecological distributionof Chaoborus species in small, shallow lakes from the Canadian Boreal Shield ecozone.Hydrobiologia 652, 207–211.

Kurek, J., Weeber, R.C., Smol, J.P., 2011. Environment trumps predation and spatial factorsin structuring cladoceran communities from Boreal Shield lakes. Canadian Journal ofFisheries and Aquatic Sciences 68, 1408–1419.

Kurek, J., Lawlor, L., Cumming, B.F., Smol, J.P., 2012. Long-term oxygen conditions assessedusing chironomid assemblages in brook trout lakes from Nova Scotia, Canada. Lakeand Reservoir Management 28, 177–188.

Laird, K.R., Kingsbury, M.V., Lewis, C.F.M., Cumming, B.F., 2011. Diatom-inferred depthmodels in 8 Canadian Boreal lakes: inferred changes in the benthic: planktonicdepth boundary and implications for assessment of past drought. Quaternary ScienceReviews 30, 1201–1217.

Laird, K.R., Haig, H.A., Ma, Susan, Kingsbury, M.V., Brown, T.A., Lewis, M., Oglesby, R.,Cumming, B.F., 2012. Expanded spatial extent of the Medieval Climate Anomaly re-vealed in lake-sediment records across the boreal region in northwest Ontario. GlobalChange Biology 18, 2869–2881.

Larocque, I., 2001. How many chironomid head capsules are enough? A statistical ap-proach to determine sample size for paleoclimatic reconstructions. Paleogeography,Paleoclimatology, Paleoecology 172, 133–142.

Luoto, T.P., 2010. Hydrological changes in lakes inferred frommidge assemblages throughuse of an intralake calibration set. Ecological Monographs 80, 303–329.

Luoto, T.P., 2012. Spatial uniformity in depth optima of midges: evidence from sedimen-tary archives of shallow Alpine and boreal lakes. Journal of Limnology 71, 228–232.

Ma, S., Laird, K.R., Kingsbury, M.R., Lewis, C.F.M., Cumming, B.F., 2013. Diatom-inferredchanges in effective moisture during the late Holocene from nearshore cores in theSoutheastern region of the Winnipeg River Drainage Basin (Canada). The Holocene23 (4), 568–578.

Parker, B.R., Schindler, D.W., Beaty, K.G., Stainton, M.P., Kasian, S.E.M., 2009. Long-termchanges in climate, streamflow, nutrient budgets for first-order catchments at the Ex-perimental Lakes Area (Ontario, Canada). Canadian Journal of Fisheries and AquaticSciences 66, 1848–1863.

Quinlan, R., Smol, J.P., 2001. Chironomid-based inference models for estimating end-of-summer hypolimnetic oxygen from south-central Ontario shield lakes. FreshwaterBiology 46, 1529–1551.

Quinlan, R., Smol, J.P., 2010. Use of subfossil Chaoborus mandibles in models for inferringpast hypolimnetic oxygen. Journal of Paleolimnology 44, 43–50.

Quinlan, R., Paterson, J.M., Smol, J.P., 2012. Climate-mediated changes in small lakes in-ferred from midge assemblages: the influence of thermal regime and lake depth.Journal of Paleolimnology 48, 297–310.

Reimer, P.J.P., Baillie, M.G.L., Bard, E., Bayliss, A., Beck, J.W., Blackwell, P.G., Bronk Ramsey,C., Buck, C.E., Burr, G.S., Edwards, R.L., Friedrich, M., Grootes, P.M., Guilderson, T.P.,Hajdas, I., Heaton, T.J., Hogg, A.G., Hughen, K.A., Kaiser, K.F., Kromer, B., McCormac,F.G., Manning, S.W., Reimer, R.W., Richards, D.A., Southon, J.R., Talamo, S., Turney,C.S.M., van der Plicht, J., Weyhenmeyer, C.E., 2009. INTCAL09 and MARINE09 radio-carbon age calibration curves, 0–50,000 years cal BP. Radiocarbon 51, 1111–1150.

Schelske, C.L., Peplow, A., Brenner, M., Spencer, C.N., 1994. Low-background gammacounting: applications for 210Pb dating of sediments. Journal of Paleolimnology 10,115–128.

Schmah, A., 1993. Variation among fossil chironomid assemblages in surficial sedimentsof Bodensee–Untersee (SW-Germany): implications for paleolimnological interpreta-tions. Journal of Paleolimnology 9, 99–108.

St. George, S., 2006. Hydrological dynamics in the Winnipeg River basin, Manitoba.Manitoba Science, Technology, Energy and Mines. . Manitoba Geological Survey226–230.

Telford, R.J., Birks, H.J.B., 2011. A novel method for assessing the statistical significance ofquantitative reconstructions inferred from biotic assemblages. Quaternary ScienceReviews 30, 1272–1278.

Ter Braak, C.J.F., Šmilauer, P., 1998. CANOCO Reference Manual and User Guide to Canocofor Windows: Software for Canonical Community Ordination (Version 4). Center forBiometry.

Uutala, A.J., 1990. Chaoborus (Diptera: Chaoboridae) mandibles— paleolimnological indi-cators of the historical status of fish population in acid sensitive lakes. Journal ofPaleolimnology 4, 139–151.

Velle, G., Telford, R.J., Heiri, O., Kurek, J., Birks, H.J.B., 2012. Testing intra-site transfer func-tions: an example using chironomids and water depth. Journal of Paleolimnology 48,545–558.

Walker, I.R., 2001. Midges: Chironomidae and related Diptera. In: Smol, J.P., Birks, H.J.B.,Last, W.M. (Eds.), Tracking Environmental Changes Using Lake Sediments. ZoologicalIndicators, vol. 4. Kluwer Academic Publisher, Dordrecht, pp. 43–66.

Walker, I.R., MacDonald, G.M., 1995. Distribution of Chironomidae (Insecta:Diptera) andother freshwater midges with respect to tree line, Northwest Territories, Canada.Arctic and Alpine Research 3, 258–263.

Wiederholm, T. (Ed.), 1983. Chironomidae of the Holarctic region. Keys and diagnoses.Part 1. Larvae. Entomological Scandinavica Supplement, 19, pp. 1–457.