ABSTRACT
Near-term extirpations of macroinvertebrates are predicted for mountain streams worldwide as a warming climate drives the recession of high-elevation ice and snow. However, hydrological sources likely vary in their resistance to climate change. Streams fed by more resistant sources could persist as climate refugia for imperiled biota. In 2015-2016, we measured habitat characteristics and quantified macroinvertebrate community structure along six alpine streams in the Teton Range, Wyoming, USA. Strong differences in habitat characteristics (e.g., temperature, bed stability, conductivity) confirmed three major alpine stream sources: surface glaciers, perennial snowfields, and subterranean ice. Subterranean ice-fed streams – termed “icy seeps” – are common but globally understudied. Macroinvertebrate communities in glacier- and snowmelt-fed streams differed significantly in multivariate space, with icy-seep communities intermediate, incorporating components of both glacier- and snowmelt-fed assemblages. Because the thermal environment of subterranean ice, including rock glaciers, is decoupled from large-scale climatic conditions, we predict that icy seeps may persist longer on the landscape than streams fed by surface ice and snow sources. Our results suggest that icy seeps are suitable habitat for many macroinvertebrates occupying streams fed by vulnerable hydrological sources. Thus, icy seeps may act as key climate refugia for mountain stream biodiversity, an idea in need of further investigation.
INTRODUCTION
The highest rates of climate change are occurring above the permanent treeline in alpine and arctic ecosystems (Bradley et al. 2006). In the Rocky Mountains, warming is proceeding two to three times more quickly than the global average (Hansen et al. 2005, Pederson et al. 2010), resulting in extensive loss of glaciers and long-term snowpack (Hall & Fagre 2003). Streams fed by permanent ice may exhibit short-term increases in flow as air temperature rises and source ice melts more quickly, but eventually they will shift to reduced flow with the potential for intermittency or drying permanently (Hotaling et al. 2017a). As climate change proceeds, invertebrate diversity at the mountain range scale is predicted to decrease due to both overall loss of habitat and summit traps, where the highest-altitude species and communities have nowhere left to disperse as warmer conditions and lower elevation communities shift upward. Biodiversity loss will be compounded by the loss of specific aquatic habitat types, particularly the unique conditions associated with meltwater from once-permanent sources like glaciers, snowfields, or subterranean ice (Brown et al. 2007, Milner et al. 2009, Jacobsen et al. 2012, Finn et al. 2013a. 2013b, Hotaling et al. 2017a). Given that many alpine stream communities appear adapted to unique, meltwater-associated regimes, they are likely to be highly vulnerable to climate change (Giersch et al. 2017, Lencioni 2018). However, because alpine streams are heterogeneous with respect to hydrological source, the potential also exists to identify stream types that may be locally buffered from broad-scale climate patterns and therefore represent potential climate refugia for alpine stream biota (Morelli et al. 2016).
A major focus of alpine stream biology is understanding the links between hydrological sources, the in-stream conditions they promote, and resident biotic communities (Ward 1994, Hotaling et al. 2017a). Historically, according to primary hydrological source, three types of alpine streams have been recognized, following Ward (1994): surface glacier-fed, snowmelt-fed, and groundwater-fed streams. A fourth, understudied stream type exists – icy seeps – which are fed by subterranean ice (Hotaling et al. 2017a, Hotaling et al. 2019a). The most common form of subterranean ice in alpine zones is rock glaciers, masses of debris-covered ice that act as conveyor belts moving fallen rock and other debris slowly downhill (Anderson et al. 2018). There may be more than 10,000 rock glaciers in the western United States (Johnson 2018) and they also appear common in mountainous regions worldwide (e.g., Lilleøren et al. 2011, Scotti et al. 2013, Charbonneau & Smith. 2018). In contrast, there are ∼1,250 surface glaciers and ∼3,750 perennial snowfields in the western United States (Fountain et al. 2017). Thermal conditions in rock glaciers are decoupled from ambient climate due to insulating debris cover (Clark et al. 1994, Anderson et al. 2018, Knight et al. 2019) and local-scale air circulation patterns (Millar et al. 2013). In a climate change context, an intriguing aspect of rock glaciers and other subterranean ice is their potential to persist on the landscape longer than surface glaciers, snowfields, and other more vulnerable sources of alpine streams. Icy seeps therefore may act as climate refugia for cold-adapted stream organisms (Hotaling et al. 2019a).
Population genetic studies of alpine-endemic species of the Rocky Mountains have revealed substantial genetic structure across relatively short geographic distances (Finn et al. 2006, Hotaling et al. 2018, 2019b), suggesting that mountaintop populations have evolved in considerable isolation. This process contributes to the elevated beta (among-stream) diversity of alpine streams (Brown et al. 2007, Jacobsen et al. 2012, Finn et al. 2013a), but genetic isolation and small population sizes can lead to bottlenecking and/or inbreeding depression in alpine-endemic species that are undergoing range contraction (e.g., Jordan et al. 2016). Like many mountain ranges worldwide, virtually nothing is known of alpine stream ecology and biodiversity in the Teton Range, a granite-dominated, young subrange of the Rocky Mountains. Previous studies of montane (but not alpine) streams revealed high diversity in streams in the range, particularly when compared to nearby lentic habitats (Tronstad et al. 2016, Hotaling et al. 2017b). Generally speaking, groundwater aquifers are rare in the Teton Range massif, and as such, groundwater-fed streams are also virtually non-existent (L.M.T., personal observation). Because groundwater-fed streams are traditionally considered to be the alpine stream type most resistant to warming (e.g., Milner et al. 2009, Jacobsen et al. 2012), their rarity in the Teton Range suggests alpine stream biodiversity in the region may be especially vulnerable to climate change.
In this study, we addressed two major objectives. First, we sought to provide the first assessment of alpine stream macroinvertebrate communities in the Teton Range. Second, we explored associations between primary hydrological sources (surface glaciers, snowfields, and subterranean ice) and community structure. Specifically, we asked if benthic communities associated with icy seeps have substantial taxonomic overlap with the more vulnerable stream types (glacier- and snowmelt-fed streams). Our study provides a preliminary look into whether an underappreciated but globally common alpine stream type – icy seeps – may act as refugia for ecological patterns and processes threatened by global change in mountain ecosystems. Furthermore, our results provide new insight into the biodiversity in one of North America’s flagship protected areas, Grand Teton National Park, and neighboring wilderness areas.
MATERIALS AND METHODS
Study area
During the summers of 2015 and 2016 (26 July-10 August), we sampled six streams in the Teton Range of Grand Teton National Park and the adjacent Jedediah Smith Wilderness in northwestern Wyoming, USA (Figure 1; Table 1). Study streams were selected to span the breadth of alpine hydrological sources in the range and included two streams fed by surface glaciers (‘glacier-fed’ hereafter), two fed by subterranean ice (‘icy seep’ hereafter) and two fed by permanent snowpack (‘snowmelt-fed’). In 2015, we sampled both upstream (near the source) and downstream sites on each stream (Figure 1). On average, upper sites were 111 m higher in elevation and 690 m stream distance from lower sites (Table 1). In 2016, we re-sampled the upper sites with the same methods to assess inter-annual variability. We focused on upper sites because they were as ‘true’ to primary hydrological source as possible while lower sites inherently reflected various degrees of mixing among sources. In both years, snow depth in the range was lower than average (152 cm in May, 1981-2010) with 2015 and 2016 at 63.3% and 80% of normal, respectively (Teton Pass, USDA SNOTEL).
Environmental data
At each site, we measured several environmental variables to characterize local habitat and evaluate whether instream environmental conditions varied among stream types. We measured water temperature for a full year (2015-2016) with in situ loggers (HOBO Pro v2, Onset Computer Corporation) that recorded temperature hourly. We measured specific conductivity (SPC), oxidation-reduction potential (ORP), pH, and dissolved oxygen (DO) with a Yellow Springs Instrument (YSI) Professional Plus multiparameter sonde calibrated at the trailhead (SPC, ORP, and pH) or at each site (DO). We estimated streambed stability with a modified version of the Pfankuch Index (PI) following Peckarsky et al. (2014). Total suspended sediments (TSS) were measured by filtering known volumes of streamwater through pre-weighed filters (PALL Type A/E glass fiber filters) and measuring dry mass to the 10−5 grams.
Using the annual temperature data from each stream, we calculated mean temperature for the entire year (TYEAR), mean temperature between the summer solstice (21 June) and autumn equinox (22 September; TSUMMER), and maximum annual temperature range (TRANGE). We also estimated the date when seasonal snow covered (SON) and uncovered (SOFF) each site by visually inspecting thermographs to approximate the date when intraday variation ceased and in-stream temperatures became constantly ∼0°C (SON) or the opposite occurred (SOFF). Using these values, we calculated SDURATION, the total days a site was snow covered between study years (SDURATION = SON - SOFF). Finally, we used Principle Components Analysis (PCA; PC-ORD, McCune and Mefford 2006) to characterize the upper sites on each stream according to four variables (SPC, TSS, PI, TRANGE) which comprise a modified ‘glaciality index’ (Ilg & Castella 2006). The glaciality index has been useful in characterizing alpine stream hydrological sources globally (e.g., Finn et al. 2013a, Cauvy-Fraunié et al. 2015).
Macroinvertebrate sampling
We quantitatively sampled benthic macroinvertebrates at each site using a Surber sampler (Area: 0.09 m2, Mesh size: 243 μm). At each location, a composite sample of 5-10 replicates was collected depending on stream size, apparent biomass, and microhabitat diversity. Samples were elutriated in the field to reduce the amount of inorganic substrate and stored in Whirl-Pak bags with ∼80% ethanol. In the laboratory, invertebrate samples were divided into large (>2 mm mesh) and small (between 250 µm and 2 mm) fractions. For the large fraction, all invertebrates were identified. The small fraction was subsampled using the record player method when smaller invertebrates were numerous (Waters 1969). Invertebrates were sorted, identified to the lowest taxonomic level possible and counted under a dissecting microscope using keys in Merritt et al. (2008) and Thorp & Covich (2010). Insects were typically identified to genus when mature specimens were present, except Chironomidae which were classified into two groups, Tanypodinae or non-Tanypodinae. We estimated density by summing the total number of individuals for a given site and dividing by the area of streambed sampled. We calculated biomass by measuring the length of the first 20 individuals of each taxon and then using length-mass regressions to estimate individual biomass (Benke et al. 1999). We multiplied the mean individual biomass for each taxon by the total number collected to estimate total biomass.
Biological data analysis
We used analysis of variance (ANOVA) statistical tests and the R package ‘plyr’ (R Core Development Team 2017; Wickham 2011) to characterize differences in invertebrate density, biomass and richness among stream types. When stream type was significant (α = 0.05), we used Tukey’s HSD to distinguish which stream types differed from one another (P ≤ 0.05). Our use of Tukey’s HSD tests was highly conservative, given the small sample size of streams in this preliminary study. To assess the relationship between taxonomic richness or biomass with stream characteristics of interest (SDURATION, TSUMMER, Pfankuch Index), we performed both Pearson and Spearman’s rank-order correlations using the R package ‘Hmisc’ (Harrell Jr. 2013). For correlation analyses, taxonomic richness and biomass were averaged for 2015 and 2016 for each upper site.
We evaluated differences in community structure across streams, sites, and years using non-metric multidimensional scaling (NMS) with PC-ORD (McCune & Mefford 2006). We log10 (n + 1) transformed density data for all taxa, removed rare taxa (either those private to a single site in the matrix and/or representing < 1% of the total abundance), and used Sørensen’s dissimilarities to create distance matrices. We ran NMS analyses independently on two data matrices: one including each of the upper and lower sites collected in 2015 only (N = 12 sites) and the other including only the upper sites (sampled in both 2015 and 2016; N = 12 sites). Dimensionality of the final solutions was chosen as the number of axes that produced the lowest stress following 200 iterations. Following NMS we applied multi-response permutation procedures (MRPP) in PC-ORD to assess whether there were differences in either community structure and/or mean community distance within the following groups: upstream compared to downstream sites in 2015 and among the three stream types for the upper sites only (2015 + 2016).
RESULTS
Environmental variation
The upper sites of our study streams clearly separated into three groups according to the glaciality index: glacier-fed streams, snowmelt-fed streams, and icy seeps (Figure 2; Table 2). Streambed stability, measured by the Pfankuch Index (Peckarsky et al. 2014), was highest in snowmelt-fed streams and lowest in glacier-fed streams (Table 2). Annual temperature range (TRANGE) was highest in snowmelt-fed streams and lowest in icy seeps. Specific conductivity was highest in icy seeps (SPC >100 μS cm−1 at upper sites) and minimal in the other two stream types (Table 2). Total suspended solids (TSS) was highest in glacier-fed streams (mean TSS, glacier-fed = 0.157 g/L, Table 2), indicating active glacier beds were present upstream. Summer temperature (TSUMMER) was low in glacier-fed streams (mean = 1.6 °C) and icy seeps (mean = 2.2 °C), and higher in snowmelt-fed streams (mean = 6.9 °C; Table 2). TSUMMER was correlated with TRANGE and therefore excluded from the PCA. Upper sites were on average 1.2°C colder in the summer and had less stable stream beds than lower sites (mean PI, upper = 27; mean PI, lower = 20.33; Table 1). Other environmental metrics (i.e., DO, pH, ORP, days under snow) did not vary consistently among stream types or between upper and lower sites (Table 2).
Biological variation
We collected a total of 35 invertebrate taxa of which 28 were insects (Supplementary Appendix 1). Insects composed 95% of the total mean densities and 92% of the total biomass. The highest densities, biomass, and richness of invertebrate communities were collected in snowmelt-fed streams. Total macroinvertebrate density did not differ among stream types (F = 1.3, df = 2, P = 0.31; Figure 3A), but biomass was ∼7x higher in snowmelt-fed streams compared to glacier-fed streams and icy seeps (F = 7.1, df = 2, P < 0.009; Tukey’s HSD, P ≤ 0.02; Figure 3B). Additionally, invertebrate richness was ∼2x higher in snowmelt-fed streams than glacier-fed streams and icy seeps (F = 10.4, df = 2, P < 0.002, Tukey’s HSD, P < 0.008; Figure 3C). We observed generally higher and more variable densities (F = 2.7, df = 1, P = 0.13; Figure 3D) and higher biomass (F = 2.6, df = 1, P = 0.13; Figure 3E) at the lower sites. We also observed ∼50% more taxa at lower versus upper sites (F = 8.5, df = 1, P = 0.012; Figure 3F). Pearson correlations between richness or biomass and SDURATION, TSUMMER, or streambed stability indicated that only TSUMMER was significantly correlated with the biological responses (Richness: Pearson r = 0.81, P = 0.002; Biomass: Pearson r = 0.63, P = 0.038; Figure 4). Spearman’s rank-order correlations of the same relationships exhibited similar patterns, again with only TSUMMER exhibiting significant correlations with richness (Spearman’s r = 0.89, P = 0.003) and biomass (Spearman’s r = 0.67, P = 0.023).
The most stable NMS solution for comparing community structure between upper and lower sites in 2015 was three-dimensional (stress = 9.6). The first two axes explained 88% of the total variation. Communities from the six lower sites overlapped substantially in ordination space with communities from the six upper sites (MRPP A = 0.013; P = 0.27). However, the mean pairwise community distance was greater among the upper sites than the lower (0.58 vs. 0.47), a trend that is apparent in the NMS bi-plot (Figure 5A) as a constriction of the total ordination space occupied by lower sites compared to the increased space occupied by the upper sites.
The NMS analysis that included upper sites only (2015 and 2016) converged on a two-dimensional solution as the most stable result (stress = 8.9). The two axes explained 86% of the total variation, with axis-1 alone explaining 70% of the variation. In general, communities occupying glacier-fed streams had the lowest axis-1 values, communities in snowmelt-fed streams had the highest axis-1 values and icy seep communities had intermediate axis-1 values (Figure 5B). MRPP results suggested that communities occupying these three stream types were significantly different from one another (A = 0.19; P = 0.006). Pairwise differences were strong between glacier-fed and snowmelt-fed communities (A = 0.21; P = 0.005) and were weaker but significant for the two pairs that included icy seep communities (icy seep vs. glacier-fed: A = 0.11; P = 0.05; icy seep vs. snowmelt: A = 0.12; P = 0.03).
DISCUSSION
As climate change proceeds and mountain glaciers recede, there is a need to develop a clearer understanding of how patterns of extant biodiversity and habitat heterogeneity are linked in high-elevation ecosystems. In this study, we provide the first description of macroinvertebrate diversity in the high Teton Range, Wyoming, where three primary stream types exist: glacier-fed streams, snowmelt-fed streams, and icy seeps. To the best of our knowledge, no streams fed by groundwater aquifers have been documented in the alpine zone of the granitic Teton Range. From a global perspective, icy seeps, which are fed by subterranean ice (primarily by rock glaciers), rather than aquifers of liquid water, are of particular interest as they are likely to persist on the landscape longer than surface ice features (Hotaling et al. 2019a). For the Teton Range, a scarcity of groundwater-fed streams suggests that biodiversity in the region may be even more reliant on meltwater than other high-alpine regions (e.g., European ranges, Brown et al. 2007; tropical Andes, Finn et al. 2016; Glacier National Park, Giersch et al. 2017), and thus may be especially reliant on icy seep-associated climate refugia in the decades to come. In the streams sampled for this study, glacier- and snowmelt-fed streams exhibited significantly different invertebrate communities; however, icy seeps were intermediate between the two in terms of community structure and invertebrate density, biomass, and richness. These results suggest that icy seep communities share some characteristics with both glacier- and snowmelt-fed streams and might have the potential to act as climate refugia for at least a subset of the unique communities present in each of the other more vulnerable stream types. Thus, the potential for icy seeps and ice-influenced terrestrial refugia (Millar et al. 2015) to buffer climate-induced biodiversity loss has profound, global implications.
The recession of meltwater sources is predicted to strongly affect downstream invertebrate communities (Jacobsen et al. 2012). In the near term, rising in-stream temperatures are expected as ice melt comprises ever smaller proportions of stream flow. In alpine streams worldwide, warmer conditions have been correlated with increased species richness for microbial diversity (e.g., Wilhelm et al. 2013, Hotaling et al. 2019a), diatoms (Fell et al. 2018) and macroinvertebrates (Finn & Poff 2005, Jacobsen et al. 2012). Our study adds another line of evidence to this global pattern as we detected a positive correlation between species richness and mean summer temperature (TSUMMER) for macroinvertebrate communities of the Teton Range. We also observed greater richness at lower (mean = 15 taxa) versus upper (mean = 10 taxa) sites. This finding also aligned with, and extended, the conclusions of Tronstad et al. (2016), the only other study to investigate longitudinal patterns of macroinvertebrate richness in montane streams of the Teton Range. We observed far fewer taxa at our highest elevation sites (10 taxa; ∼3,150 m) versus the highest elevation sites included in Tronstad et al. (2016): 26 taxa at ∼2,700 m.
Although local (alpha) diversity is expected to increase with warming water temperatures, among-stream (beta) diversity may decrease as more diverse communities shift upstream, effectively homogenizing biological diversity at the regional scale (Jacobsen et al. 2012, Wilhelm et al. 2013, Hotaling et al. 2017a). In the Teton Range, we observed greater beta diversity among our upper sample sites than our lower sites, even with wide variation in landscape features (e.g., a lake separating stream reaches at one of our sites) and variable distances between upper and lower sites among our six focal streams. This spatial pattern suggests that in the Teton Range, like elsewhere in the world, heterogeneous hydrological sources bolster regional-scale alpine stream biodiversity.
Ultimately, the degree to which alpine streams will be affected by climate change in terms of flow magnitude and persistence remains largely unknown. While many studies of alpine streams operate under the assumption that perennial flow will continue for many streams in the decades to come (e.g., Jacobsen et al. 2012), this may not be the case (e.g., Haldorsen & Heim 1999). Thus, the biological ramifications of declining meltwater sources in places like the Teton Range, where groundwater aquifer-fed springs are scarce, may be even more profound than in ranges with a greater diversity of alpine stream sources (e.g., Glacier National Park, Giersch et al. 2017). Indeed, if alpine streams supported by surface glaciers and permanent snowfields transition to intermittency or dry completely, the future of biodiversity in these ecosystems may depend almost exclusively on icy seeps. Our study paired with the broader glaciological literature, however, suggests that there is room for optimism in this regard. Indeed, icy seeps have the potential to span a wide beta diversity profile, perhaps bridging the taxonomic gap between glacier- and snowmelt-fed communities (e.g., Hotaling et al. 2019a). Moreover, rock glaciers and other subterranean ice forms are common in alpine regions worldwide and likely the most resistant ice form to future warming. Clearly, research focused on subterranean ice sources and associated icy seeps, and the biological communities they support, represents a pressing need in alpine stream biology. We suggest that future studies incorporate temporal monitoring of alpine stream types, including icy seeps, to explicitly test how biodiversity and habitat characteristics may be altered across alpine stream types as climate change proceeds.
ACKNOWLEDGEMENTS
Financial support was provided by The University of Wyoming-National Park Service Research Station and the Teton Conservation District. We thank Lydia Zeglin for assistance in the field and Hanna Foster, Logan Fox, Alexis Lester, Jackson Marr, Jake Ruthven, Joe Wannemuehler, and Kara Wise who helped sort invertebrate samples. Matthew Green provided helpful comments on the manuscript. Any use of trade, firm, or product names is for descriptive purposes only and does not imply endorsement by the U.S. Government.