Proceedings of the Royal Society B: Biological Sciences
You have accessResearch articles

Nutrients and warming interact to force mountain lakes into unprecedented ecological states

Isabella A. Oleksy

Isabella A. Oleksy

Natural Resource Ecology Laboratory, Colorado State University, Fort Collins, CO 80523, USA

Cary Institute of Ecosystem Studies, Millbrook, NY 12545, USA

[email protected]

Google Scholar

Find this author on PubMed

,
Jill S. Baron

Jill S. Baron

U.S. Geological Survey, Fort Collins, CO 80526, USA

Natural Resource Ecology Laboratory, Colorado State University, Fort Collins, CO 80523, USA

Google Scholar

Find this author on PubMed

,
Peter R. Leavitt

Peter R. Leavitt

Institute of Environmental Change and Society, University of Regina, Regina, Saskatchewan, S4S 0A2 Canada

Institute for Global Food Security, Queen's University Belfast, Belfast, Antrim BT9 5DL, UK

Google Scholar

Find this author on PubMed

and
Sarah A. Spaulding

Sarah A. Spaulding

U.S. Geological Survey/INSTAAR, University of Colorado, Boulder, CO, USA

Google Scholar

Find this author on PubMed

    Abstract

    While deposition of reactive nitrogen (N) in the twentieth century has been strongly linked to changes in diatom assemblages in high-elevation lakes, pronounced and contemporaneous changes in other algal groups suggest additional drivers. We explored the origin and magnitude of changes in two mountain lakes from the end of the Little Ice Age at ca 1850, to ca 2010, using lake sediments. We found dramatic changes in algal community abundance and composition. While diatoms remain the most abundant photosynthetic organisms, concentrations of diatom pigments decreased while pigments representing chlorophytes increased 200–300% since ca 1950 and total algal biomass more than doubled. Some algal changes began ca 1900 but shifts in most sedimentary proxies accelerated ca 1950 commensurate with many human-caused changes to the Earth System. In addition to N deposition, aeolian dust deposition may have contributed phosphorus. Strong increases in summer air and surface water temperatures since 1983 have direct and indirect consequences for high-elevation ecosystems. Such warming could have directly enhanced nutrient use and primary production. Indirect consequences of warming include enhanced leaching of nutrients from geologic and cryosphere sources, particularly as glaciers ablate. While we infer causal mechanisms, changes in primary producer communities appear to be without historical precedent and are commensurate with the post-1950 acceleration of global change.

    1. Introduction

    Over the past 70 years, human activities have modified the physical and biogeochemical drivers that control primary productivity in remote mountain lakes worldwide [13]. Mountain lakes are not alone: socio-economic and Earth System changes in response to human activities are causing fundamental shifts in the state of the planet and have been defined as the Anthropocene [4]. Expansion of industrial fertilizer manufacturing and application, intensive livestock production, and fossil fuel combustion have increased atmospheric nitrogen (N) deposition to remote watersheds by more than an order of magnitude beyond background levels [57]. Aeolian dust from distant sources subsidizes alpine ecosystems with phosphorus (P), and recent evidence suggests that dust deposition is increasing [811]. Additional stimulation of physical and biogeochemical processes comes from warming, which has increased lake temperatures worldwide [12], with higher elevation systems exposed to more rapid change [13]. Warming can additionally alter lake productivity by altering the timing and rate of snowmelt, duration of ice cover, strength and duration of lake stratification, and inputs of minerals and nutrients from the catchment [1419]. The combination of added nutrients and warming is a well-known recipe for increased primary productivity [20,21].

    Lakes that have not historically been exposed to separate or coincident increases in nutrients and energy can be forced into novel ecological states [22,23]. In this paper, we illustrate the interactive effects of these changes on two lakes in the Colorado Front Range, emphasizing responses for which there appears to be no historical precedent. We build on original palaeolimnological work in Rocky Mountain National Park that first demonstrated a relationship between atmospheric N deposition and ecosystem response to remote lakes [2426]. We use these results to frame research presented here and describe changes that have occurred in the 20 years since that work was published [27]. Interacting forces of ecological change are now evident in other regions [28,29], and we posit that similar processes may be at work throughout the southern Rocky Mountains where a warming climate and increasing P inputs interact with a legacy of N deposition to alter lake autotrophic structure and function.

    Using algal and biogeochemical proxies from lake sediment cores, we examined patterns in biogeochemistry (elemental composition, stable isotopes), lake primary production, and composition of primary producers (pigments of algae and cyanobacteria, diatom subfossils) from the end of the Little Ice Age ca 1850 to present in an alpine lake, Sky Pond, and a subalpine lake, The Loch. Sky Pond was cored previously to explore changes in diatom and N isotope proxies caused by atmospheric N deposition [26,27]. Additional proxies, described below, build on earlier work and address changes since the initial core in 1997. The Loch had not previously been cored, but information gained by the addition of this second proximal lake was anticipated to either further support conclusions if results for the two lakes were similar, or provide data to determine why they may differ. In particular, we were interested in whether contemporary assemblages of primary producers differ from those in the past, if similar changes occurred in two lakes located in the same watershed, and whether changes were synchronous with other major shifts in the Earth System since 1950. The period around 1950 has been suggested as the start date for the Anthropocene and the beginning of rapid environmental and socio-economic changes have been termed the Great Acceleration [4].

    2. Methods

    Sky Pond and The Loch are located in the Loch Vale watershed (LVWS), Rocky Mountain National Park, Colorado, USA (electronic supplementary material, figure S1A). The northeast-facing watershed covers 6.6 km2 with an elevation range from 3048 to 4120 m (electronic supplementary material, table S1). Lakes have received elevated wet N deposition over background levels since measurements began in 1984, and the 5-year rolling average of wet N deposition since 1995 has been stable at 3.0–3.5 kg N ha−1 yr−1 [30]. Total atmospheric P deposition during 2002–2010 was estimated as 0.015 kg P ha−1 yr−1 at a nearby research site at Niwot Ridge [31].

    Sky Pond is an alpine headwater lake at the base of Taylor Rock Glacier. Sky Pond is small (3 ha, 1.2 × 105 m3), with a maximum and mean depth of 7.2 m and 4.5 m, respectively (electronic supplementary material, figure S1B and table S1). The Loch is a 5.0 ha subalpine lake fed by two streams, one of which flows from Sky Pond (electronic supplementary material, figure S1A). The distance between the two lakes is 1 km. The Loch has a maximum and mean depth of 4.7 m and 1.5 m, respectively. Both lakes have low specific conductance, nutrient concentrations, dissolved organic carbon (DOC) levels, and phytoplankton abundance as chlorophyll a. Dissolved inorganic nitrogen : total phosphorus (DIN : TP) ratios are 69 and 56 for The Loch and Sky Pond, respectively (electronic supplementary material, table S1). Benthic blooms of chlorophytes and charophytes (Spirogyra spp., Zygnema spp., and Nitella spp.) have been present in the bottom of The Loch and in the littoral zones of Sky Pond since 2010 (electronic supplementary material, figure S1D-F).

    Sediments were collected from each lake using gravity-coring systems (electronic supplementary material). Core chronologies were established from 12 sections for Sky Pond and 14 sections for The Loch by analysis of 210Pb activities and application of a constant rate of supply model (CRS model; [32]). Methods for evaluating 137Cs follow Davis et al. [33].

    Stable isotope ratios and elemental composition of N and C were determined at the Institute of Environmental Change and Society, University of Regina, on whole dried sediment samples using a ThermoQuest DeltaPLUS XL isotope ratio mass spectrometer equipped with a continuous flow unit (Con Flo II), an automated Carlo Erba elemental analyzer as an inlet device, and following standard procedures [34]. Stable N (δ15N) and C (δ13C) isotopic compositions were expressed in the conventional notation relative to standards. Sample reproducibility ranged from <0.10‰ (δ13C) to less than 0.25‰ (δ15N).

    Carotenoids, chlorophylls, and their derivatives were quantified as metrics of algal and cyanobacterial abundance over time [35] using methods described in depth in the electronic supplementary information. We restricted our analysis to chemically-stable biomarkers of total algae and cyanobacteria (pheophytin a, β-carotene), chlorophytes (pheophytin b), chlorophytes and cyanobacteria (lutein and zeaxanthin), filamentous and colonial cyanobacteria (myxoxanthophyll), cyanobacteria within the Nostocales (canthaxanthin), total cyanobacteria (echinenone), cryptophytes (alloxanthin), mainly diatoms (diatoxanthin), and a combination of diatoms, chrysophytes, and some dinoflagellates (fucoxanthin). Lutein (chlorophytes) and zeaxanthin (cyanobacteria) were not separated on our HPLC system and are presented together. Diatom subfossils were analysed from Sky Pond following Pite et al. [36]. Identifications and habitat preference (planktonic, benthic) follow the Diatoms of North America project [37].

    Since preliminary screening of the data revealed nonlinearities, our sediment cores lack annual laminations, and compression of sediments leads to uneven sampling through time, this precluded us from using simple linear regression or non-parametric approaches to assess trends in our data due to violation of model assumptions [38,39]. Instead, we modelled historical trends in C and N content and algal pigments using generalized additive models (GAMS) [40] with the gamm() function [41] in the mgcv package [42] within the R statistical environment [43] following Simpson and Anderson [44] (full description in the electronic supplemental material). Trends in seasonal air and water temperature from the LVWS long-term record were assessed using simple linear regression. All data are available at https://doi.org/10.5066/P9MVU3CX.

    3. Results

    Analysis of 210Pb and 137Cs activities provided a reliable sediment chronology for Sky Pond (electronic supplementary material, figure S2) but not for The Loch. In Sky Pond 210Pb declined regularly between 2 and 8 cm depth, with some suggestion of sediment mixing in the surface layer and a slight change in sediment accumulation rates between ca 1930 and ca 1950. Overall, the 137Cs profile showed a clear peak ca 1963, based on 210Pb-established chronology, reflecting influx from open-air atomic blasts (1963), resulting in nearly linear depth-age relationships.

    There were no obvious changes in measured activity of either radioisotope in the sediments of The Loch, with very low activities of 210Pb and 137Cs in all samples (data not shown). Instead, sediment chronology was approximated by assuming both lakes received similar atmospheric influx of reactive N, and that changes in standardized (z-transformed) δ15N values should be temporally coherent in the two basins, as seen in much of the southern Rocky Mountains (electronic supplementary material, figure S3A,B; [27]) and remote lakes where atmospheric N influx dominates N budgets [7]. Both lakes showed parallel changes in δ15N with depth, with highly correlated changes in the uppermost 6 cm (Pearson's R = 0.9, p < 0.001) that suggest similar sedimentation rates (electronic supplementary material, figure S3). Consequently, we applied the Sky Pond chronology to The Loch sediment core. Our approach allows us to evaluate qualitative differences in the response of independent geochemical and pigment proxies between The Loch and Sky Pond.

    Bulk sediment δ13C values became progressively and significantly depleted mid-twentieth century in Sky Pond, with declines of approximately 1‰ to −27‰ (figure 1a). Bulk sediment δ13C values in The Loch were slightly more enriched than the alpine Sky Pond. Bulk δ15N declined in both lakes, though they declined to lower values in Sky Pond compared to The Loch (figure 1b). In Sky Pond, the %N was variable in the full sediment record and increased modestly beginning ca 1950, whereas values in The Loch were stable (0.51 ± 0.02%) until the mid-twentieth century, then increased steadily to 0.7 ± 0.04% in the most recent sediments (figure 1d). The Sky Pond %C was 2–5%, and The Loch %C was 7–8% (figure 1c). The higher C:N ratio in sediments from The Loch reflects the higher %C in the subalpine lake, and until ca 1950, C : N ratios averaged 16.5 ± 0.4 in The Loch and 9.8 ± 0.4 in Sky Pond (figure 1e). After ca 1950 the C : N ratio in The Loch decreased linearly to its current value of about 12 in the most recent sediments, primarily because of the increase in %N in recently deposited sediments. In Sky Pond, both %C and %N increased concurrently since the mid-twentieth century (figure 1c,d).

    Figure 1.

    Figure 1. Summary of temporal trends in chemistry. Year versus δ13C, δ15N, C, N, C : N, are fitted with GAM-smoothing trends and 95% confidence intervals (light grey bands) for The Loch and Sky Pond. (a) δ13C of bulk sediment; (b) δ15N of bulk sediment; (c) carbon content as a percentage of dry mass; (d) nitrogen content as a percentage of dry mass; and (e) C : N ratio. Depths prior to ca 1850 are extrapolated from 210Pb data and should be interpreted with caution. Electronic Supplemental Information for details. (Online version in colour.)

    Pheophytin a concentrations and temporal patterns were similar between the two lakes, with slightly higher values in Sky Pond. Fitted GAMs indicated a significant increase in total algal biomass in the mid-twentieth century and that the rapid changes continued to the present (figure 2a and table 1). By contrast, concentrations of ubiquitous β-carotene increased gradually in Sky Pond over the record (table 1 and figure 2b). Concentrations of chlorophyte biomarkers exhibited a marked increase in Sky Pond in the twentieth century, with a similar but smaller increase in The Loch (figure 2c,d and table 1). In Sky Pond, significant changes in the concentrations of both pheophytin b (chlorophytes only) and lutein-zeaxanthin (chlorophytes + cyanobacteria) began ca 1900 and accelerated through ca 1975, although the rate of increase slowed thereafter. Modern concentrations are three to four times higher than those early in the twentieth century. Present-day values of lutein-zeaxanthin are similar between the two lakes, but historical concentrations in The Loch were twice as high as those recorded in Sky Pond. Pheophytin b concentrations in recent sediments of Sky Pond are approximately twice as high as those in The Loch.

    Figure 2.

    Figure 2. Temporal trends in major algal functional groups inferred by pigment analyses. GAM-smoothing trends fitted are depicted with 95% confidence intervals for all major algal pigments in The Loch and Sky Pond time series. Pheophytin a (a) and β-carotene (b) are proxies for total algal biomass. Pheophytin b (c) is a proxy for total chlorophyte biomass. Lutein and zeaxanthin (d) are indicative of both chlorophytes and cyanobacteria. Diatoxanthin (e) is a proxy for total diatom biomass. Echinenone (f), canthaxanthin (g), and myxoxanthophyll (h) are proxies for various cyanobacteria (total, Nostocales, and filamentous and colonial, respectively). Alloxanthin (i) is proxy for total cryptophytes. (Online version in colour.)

    Table 1. Model summary of GAMs for all geochemical and pigment biomarkers. We used generalized additive models for the detection of trends on all sedimentary pigment and elemental biomarkers (see electronic supplemental material for details). For each predictor, effective degrees of freedom (EDF), F-statistic, p-value, and adjusted R2, and periods of significant change are shown. Note, periods of significant change were only identified for Sky Pond because of dating uncertainties with The Loch (see text for details).

    The Loch EDF F p-value Adj. R2
    δ13C 1.429 1.447 0.156 0.16
    δ15N 4.312 27.59 <0.0001 0.91
    % C 1 7.786 0.0091 0.21
    % N 6.061 31.93 <0.0001 0.90
    C : N 6.263 35.3 <0.0001 0.88
    lutein 1.584 2.076 0.266 0.19
    pheophytin b 2.203 8.291 <0.0001 0.65
    pheophytin a 4.749 37.97 <0.0001 0.90
    β-carotene 2.793 7.77 0.0004 0.47
    diatoxanthin 1 1.73 0.199 0.16
    echinenone 2.205 3.105 0.1 0.22
    canthaxanthin 2.789 24.53 <0.0001 0.68
    myxoxanthophyll 3.421 21.65 <0.0001 0.76
    alloxanthin 2.557 4.401 0.0413 0.47
    Sky Pond EDF F p-value Adj. R2 Period of change (direction)
    δ13C 5.733 27.3 <0.0001 0.81 1925–2017 (−)
    δ15N 2.586 17.32 <0.0001 0.74 1895–2000 (−)
    % C 6.19 20.56 <0.0001 0.84 1830–1920 (−) 1955–1990 (+)
    % N 4.864 23.18 <0.0001 0.78 1820–1915 (−) 1940–2017 (+)
    C : N 5.971 11.21 <0.0001 0.69 1880–1930 (+) 1965–2017 (−)
    lutein 5.198 40.7 <0.0001 0.84 1880–1975 (+)
    pheophytin b 5.862 64.77 <0.0001 0.91 1900–1975 (+)
    pheophytin a 3.426 25.49 <0.0001 0.79 1895–2017 (+)
    β-carotene 1 27.37 <0.0001 0.61 1950–2017 (+)
    diatoxanthin 6.243 22.28 <0.0001 0.75 1850–1900 (+) 1915–2017 (−)
    echinenone 3.131 7.693 <0.0001 0.26 1925–2017 (−)
    canthaxanthin 5.045 18.78 <0.0001 0.80 1890–1990 (+)
    myxoxanthophyll Not detected
    alloxanthin 3.855 13.34 <0.0001 0.65 1850–1905 (−) 1930–2017 (+)

    Strong differences in historical patterns of abundance were seen in other algal groups. Concentrations of diatoxanthin decreased over the period of record in both lakes (figure 2e and table 1). Overall, the rate of change was fastest from ca 1915 to the present in Sky Pond. By contrast, the concentrations of cyanobacterial biomarkers were an order of magnitude lower than the concentrations of green algae and diatoms, and the historical composition differed both within and between Sky Pond and The Loch. Total cyanobacteria (as echinenone; figure 2f) had contrasting nonlinear patterns between lakes over the record, but both lakes showed a marked increase in abundance of Nostocales forms beginning mid-twentieth century (canthaxanthin; figure 2g). Colonial cyanobacteria (as myxoxanthophyll; figure 2h) increased sixfold in The Loch but were undetectable in Sky Pond. Alloxanthin from cryptophytes increased in Sky Pond sediments commensurate with changes in Nostocales and total cyanobacteria, but this pattern was not found in The Loch (figure 2i). Overall, low concentrations for echinenone, canthaxanthin, myxoxanthophyll, and alloxanthin compared to green algal pigment proxies suggest that these taxonomic groups comprise a relatively minor proportion of the algal community both historically and at present.

    Analysis of diatom subfossils in Sky Pond sediments showed a switch from benthic to planktonic species ca 1950, resulting in a change in the planktonic:benthic ratio from 0.25 prior to 1950 to greater than 0.50 in uppermost sediments (figure 3; electronic supplementary material, figure S4). Specifically, the relative abundance of planktonic A. formosa and Fragilaria crotonensis increased in the mid-twentieth century while the abundance of dominant benthic species, Stauroneis spp., declined over the same period (electronic supplementary material, figure S4).

    Figure 3.

    Figure 3. Ratio of Planktonic to Benthic (P : B) diatom valve counts in Sky Pond sediments since ca 1850. There was a significant positive trend (F(1,20) = 22.69, R2 = 0.53, p = 0.0001) in P:B with a shift towards more planktonic than benthic species, with an inflection point ca 1950.

    4. Discussion

    (a) Patterns and processes

    Alpine Sky Pond and subalpine The Loch differed in their C and N values and isotopic signatures, possibly reflecting a different origin of C from autochthonous and allochthonous sources. The Loch sediments had more than double the C content of Sky Pond, but the %C for both lakes falls within ranges reported from both alpine and subalpine lakes elsewhere [45,46].

    Carbon isotope values of −23‰ to −24‰ suggest terrestrial vegetation contributed C to the subalpine lake, whereas the more depleted −26‰ to −27.5‰ values in Sky Pond suggest algal or respiratory C sources. In addition to terrestrial input, enriched δ13C values in The Loch could reflect higher contributions from benthic algae (−20‰ to −24‰) which typically predominate in shallow clearwater lakes [46,47]. Benthic algae in Sky Pond could also be a source of enriched δ13C; in a previous study of Sky Pond winter samples had depleted values of −30‰ δ13C or less, while summer values ranged −23 to −27‰ [48].

    The δ13C signal in The Loch remained within the historical range of variability (–23 ± 0.2‰) over the sediment record, whereas values declined from −26.0‰ to −27.5‰ in Sky Pond between ca 1950 and the present (figure 1a). Sky Pond, the larger and deeper of the two lakes, consistently has greater planktonic productivity, with chlorophyll a concentrations two to five times as high as in The Loch [48,49]. Depletion of δ13C coincided with an increase in planktonic diatoms in the mid-twentieth century as well as an increase in the planktonic:benthic ratios. In Sky Pond and other remote mountain lakes, this change resulted from a replacement of oligotrophic diatoms to more mesotrophic and planktonic A. formosa, and, to a lesser degree, F. crotonensis (figure 3; electronic supplementary material figure S4; [27,50,51]). Others have noted similar responses and suggest changes in primary production and the source of CO2 (e.g. enriched atmospheric, depleted respiratory) might be responsible [52,53].

    The N content in Sky Pond ranged from 0.2 to 0.5% over time and exhibited a slight increase in the twentieth century. In The Loch, N accounted for approximately 0.5% of dry mass until the mid-twentieth century, when it nearly doubled to 0.8% N. This increase occurred during the same period as the documented increase in atmospheric N deposition [7,54], the decline in δ15N, and the increase in primary production observed in our study lakes. Presently, we do not know why the trend is so much more dramatic in The Loch, but speculate that it reflects a subsidy of N-rich autochthonous organic matter from Sky Pond during spring snowmelt flushing [55]. Beginning ca 1950, the C : N of The Loch sediments decreased in response to the increased N content of the sediments. With the decline in ratios in The Loch ca 1950, both lakes currently display C:N ratios indicative of in situ primary production [56].

    Nitrogen isotopes displayed similar patterns over time in both lakes. The δ15N signal was relatively stable at 4–5‰ until the twentieth century. The most rapid decline in isotopic signature occurred mid-twentieth century, similar to patterns in many other temperate and Arctic lakes [7,29,50]. The decline in δ15N has become even more depleted in the 20 years since Sky Pond was last cored, and the most recent sediments from our 2017 core show values of 1.5–2.0‰.

    (b) Benthic and planktonic dynamics in Sky Pond

    Benthic diatoms often predominate in undisturbed shallow, oligotrophic lakes, particularly at high elevations, due to high transparency and sedimentary nutrient sources [46,47,57]. However, algal communities can shift from benthic to planktonic assemblages in response to warming [58] or nutrient influx [7,27,59]. The shift to dominance by A. formosa was noted by Wolfe et al. [27,50] from the previous study of Sky Pond, and was attributed by them to the regional rise in atmospheric N deposition. A recent study by Sivarajah et al. [51] concludes warming can also enhance growth of A. formosa. We suggest increased N deposition in the mid-twentieth century was the initial stimulus for A. formosa, and rapid warming since 1970 may now contribute to the continued dominance of this diatom in the Sky Pond water column.

    Changes in the abundance of Sky Pond diatoms were concurrent with an increase in the abundance of benthic chlorophytes. Whereas diatoms were mostly benthic before ca 1950, thereafter species were evenly distributed between benthic and planktonic genera. Together, this change in benthic algae probably reflects altered nutrient and climatic regimes (as above) that allow chlorophytes to exploit benthic habitats. Because the pigment proxies for diatoms declined through the sediment cores, even though the proportion of benthic to pelagic diatoms decreased, it suggests a loss of diatoms in the benthos. Chlorophytes are thought to be more sensitive to climatic variation than are diatoms and are often the predominant algal taxa during warm periods such as the Holocene Thermal Maxima [60,61]. Dense benthic chlorophyte mats might have shaded attached diatoms, reducing their growth [62], or chlorophytes may simply be more competitive for nutrients than benthic diatoms [63]. Furthermore, in warmer climates, preferential warming in shallow zones where light penetrates to dark sediments may occur, benefitting algal groups with higher thermal optima, such as chlorophytes [46,64,65].

    Multiple factors may have influenced the rise of benthic chlorophytes. Previous experiments in Sky Pond, The Loch, and other study lakes found that biomass of benthic chlorophytes and cyanophytes increased with N and P fertilization, even when ambient N levels were high [63,66]. Other traits may favour green algae over diatoms, including a higher tolerance to UV-B radiation [67], more efficient nutrient uptake [68,69], and an increase in nutrient availability in lakes stocked with trout (see below). Although we cannot easily distinguish among these possibilities from our results, the shift in dominance from benthic to planktonic diatoms, combined with expansion of benthic green algae, is without historical precedent in the LVWS. Fish introduction was previously ruled out as a contributing factor to diatom assemblage shifts in Sky Pond [50], although trout were stocked and established in both Sky Pond (1931–1939), and The Loch (1914) [70]. Introduced fish can increase nutrient availability by harvesting aquatic instars of terrestrial insects and excreting nutrient-rich faecal pellets that can sink to the lake bottom [71,72]. Several pigments representing different types of algae increased in Sky Pond roughly commensurate with trout introductions, including benthic pheophytin b, lutein-zeaxanthin (chlorophytes), and canthaxanthin (Nostocales). Canthaxanthin increased in the early part of the twentieth century in The Loch. Based on these observations, we cannot rule out an influence of fish introductions on the nutrient status of the lakes, and trout may have been a contributing driver to current algal condition.

    (c) Drivers of change

    Lake sediments provide a window into past environmental change, but interpretation of causality can be difficult. Dates from 210Pb activities are reliable 100–150 years back from the present, but extrapolation beyond that is speculative [73]. While the 210Pb dates for Sky Pond mostly decreased progressively towards background levels, allowing their values to be interpreted with some confidence, 210Pb dates for The Loch were uninterpretable, as stated above. Low specific activity such as we saw in The Loch is common in alpine systems and may reflect a combination of factors, including high flushing rates that lead to low deposition of small atmospheric particles [33], localized sources of radon [74], or complicating effects of glacial flour on deposition of small suspended materials [33,72]. Because the sediment chronology in The Loch adds additional uncertainty to 210Pb dating as a proxy for time, we focus on variation in the patterns of change in geochemistry and primary producer abundance rather than timing of changes in The Loch, and speculate on the possible causes of observed change.

    Algal biomass is low in high-elevation nutrient-poor lakes, and prior analyses suggest this was the case for Rocky Mountain lakes, including Sky Pond, from the time of deglaciation (14 000 bp) until the mid-twentieth century, when productivity increased and diatom assemblages substantially changed [27,46,50]. Diagnostic pigments from our cores in Sky Pond and The Loch shared this pattern, showing slight increases in total algal abundance, diatoms, and chlorophytes between 1850 and 1950. The increases correspond with warming throughout the northern hemisphere from the end of the Little Ice Age, and stimulation by warming of algal growth in cold climates [24,75,76]. While in principle increased production after ca 1850 may also correspond to the rise of continental industrialization, including mining in the Rocky Mountains and agriculture on the adjacent plains [77], there is no evidence of increased nutrient deposition at that time [27]. Instead, we conclude that modest increases in primary production during the mid-nineteenth century were the result of warming at the end of the Little Ice Age.

    Far more pronounced changes in lake algae are evident in the twentieth century, beginning between 1940 and 1960. Our results show sharp and previously undocumented increases in overall biomass, chlorophytes, cyanophytes, and cryptophytes in both Sky Pond and The Loch. Concentrations of chlorophyte biomarkers increased 200–300% in both lakes and the increase in chlorophytes continues to the present day, in contrast to diatom pigment concentrations that decreased over the twentieth century. The shift in diatom assemblages represents a major response to atmospheric N deposition while we infer the increase in chlorophytes and overall algal productivity to indicate responses to additional drivers of P and warming.

    Several mechanisms may have combined to elevate primary production in lakes of Loch Vale during the last 70 years (figure 4). Increases in autotrophic production and shifts in algal assemblages are contemporaneous with well-documented global increases in temperature and reactive N availability since 1950 [4]. Measured summer temperatures in Loch Vale have increased since 1983 (electronic supplementary material, figure S5; [78,79]), and a longer estimate of summer temperatures using the PRISM model corroborates the recent trend (figure 4a). Atmospheric deposition of N has increased over 10-fold above background levels, influencing terrestrial and aquatic ecosystems in Loch Vale, and restructuring diatom assemblages since the 1950s [27,80]. The N deposition in Loch Vale is local evidence of a widespread phenomenon, observed even in remote regions, due to increases of reactive nitrogen released to the environment since the post-war acceleration of agriculture, combustion of fossil fuels, and other human activities (figure 4b; [81,82]). A third driver of increased productivity in remote alpine lakes such as Sky Pond and The Loch is an increase in P influx, whether from atmospheric deposition or from increased weathering of P-bearing minerals [9,83]. Phosphorus production has increased exponentially worldwide since 1950 [4], and regional mountain ecosystems in North America have received P-bearing minerals and organic compounds to North American mountains caused by dust and agricultural emissions (figure 4c; [9,11,83]). Increased weathering of P-bearing mineral apatite due to climate warming [84] and cryosphere thaw [85], might contribute P, as was suggested by Kopáček et al. [18] for granitic alpine catchments in the Tatra Mountains. Whether from atmospheric deposition or mineral weathering, an increase in P availability enhances algal productivity in most lakes [86]. Therefore, while our analyses are correlative, unprecedented increases in benthic chlorophytes and planktonic diatoms starting in the mid-twentieth century are consistent with Anthropocene's Great Acceleration [4].

    Figure 4.

    Figure 4. Trends in climate and nutrient drivers associated with lake algal production. (a) Mean summer (JJA) modelled air temperature (PRISM Climate Group, Oregon State University, http://prism.oregonstate.edu) for Loch Vale watershed. The 120-year PRISM record showed a significant breakpoint in which summer air temperatures increased more rapidly post-1985 (Davies test p < 0.0001). (b) Global N emission rates [82]. The shaded box highlights 1940–1960 when rates accelerated (GAM analyses). (c) P accumulation rates in sediments of alpine lakes 1900–present, extracted from [11]; the shaded box is the result of a GAM analysis that shows a significant increase in the rate of P accumulation.

    (d) Beyond nitrogen deposition: lakes on a trajectory of change

    Global environmental change is altering water chemistry and algal assemblages in Sky Pond and The Loch, as well as many other remote lakes [18,27,29]. The dominant habitat for diatoms changed from benthic to pelagic, and the benthos is now host to abundant green algal mats. Of the possible drivers, emission and deposition of industrial reactive N increased first (ca 1950), followed by regional warming and P influx about 20–30 years later. Introduction of non-native trout may have stimulated a slight increase in primary productivity early in the twentieth century. We suggest that the onset of N pollution may have ‘primed the pump’ that initiated a cascade of ecological shifts, although we also recognize that the rate of change of many variables is still accelerating [2,29]. While our results are inferential only, they strongly suggest a legacy of N deposition, coupled with increasing dust (including P) deposition, and rapidly changing climatic conditions, changed the ecosystem structure of remote mountain lakes in the southern Rocky Mountains. Our analysis of sedimentary pigments and geochemistry unequivocally documents systemic changes in lake autotrophic structure that are suggestive of profound ecosystem shifts in recent decades.

    Data accessibility

    This article has no additional data.

    Authors' contributions

    I.A.O. and J.S.B. designed the study and collected the sediment cores; I.A.O., P.R.L., and S.A.S. contributed method development and laboratory analyses; I.A.O. analysed the data and I.A.O. and J.S.B. wrote the paper with contributions from P.R.L. and S.A.S.

    Competing interests

    We declare we have no competing interests.

    Funding

    This project is a product of the U.S. Geological Survey Western Mountain Initiative. IAO was also supported under NSF IGERT grant no. DGE-0966346 ‘I-WATER: Integrated Water, Atmosphere, Ecosystems Education and Research Program.’ P.R.L. was supported by Canada Research Chair, NSERC and Canada Foundation for Innovation programs, as well as the Province of Saskatchewan, University of Regina and Queen's University Belfast.

    Acknowledgements

    We acknowledge that the lakes we studied are on the traditional and ancestral homelands of the Ute, Arapaho, Cheyenne, and Comanche Nations and peoples. We are grateful for field help from Daniel Bowker, Tyler Lampard, Timothy Weinmann, John Hammond, Christa Torrens, Jason Price, Melanie Burnett, and Jessica Johnstone during winter sediment core collections. Deirdre Bateson and Heather Haig provided integral training in sedimentary pigment analyses and Jared Wolfe and Gavin Simpson provided valuable insight used to improve our statistical models. Amanda Jayo created the site map. We greatly appreciate the comments from two anonymous reviewers. Any use of trade, firm, or product names is for descriptive purposes only and does not imply endorsement by the U.S. government.

    Footnotes

    Electronic supplementary material is available online at https://doi.org/10.6084/m9.figshare.c.5036405.

    Published by the Royal Society. All rights reserved.