Altitudinal gradients in Magellanic sub-Antarctic lagoons: the effect of elevation on freshwater macroinvertebrate diversity and distribution

Study area

The Magellanic sub-Antarctic ecoregion, located in southern Chile, is characterized by watersheds with acute environmental clines, diverse vegetation profiles, and the presence of a wide variety of habitats and microhabitats over a short elevational range (Pisano, 1977; Contador et al., 2015). This ecoregion is part of the South American forest biome, harboring the largest forest and wetland areas of the Southern Hemisphere. Embedded within the ecoregion is the Cape Horn Biosphere Reserve (CHBR; Fig. 1), an area designated to protect the ecoregion from the pressures of global change (Rozzi et al., 2006). Furthermore, this region contains some of the world’s cleanest rainwater, as it is located to the south of the typical tracks of industrial-polluted wind currents (Hedin, Armesto & Johnson, 1995; Weathers et al., 2000; Mach et al., 2016). Navarino Island lies in the southern part of this ecoregion. The mountain ranges in Navarino Island are characterized by short and steep altitudinal gradients (rising to ∼1,000 m a.s.l.), with associated changes in air and water temperatures with elevation (Contador et al., 2015). Annual mean air temperature is 5.7 °C close to sea level, decreasing to 0 °C at 728 m a.s.l. (Méndez, Rozzi & Cavieres, 2013). This study was conducted in lagoons (lakes and ponds) located along the altitudinal gradient of the Róbalo River watershed, which provides drinking water to Puerto Williams, the world’s southernmost town. The annual average water temperature in the running waters of the watershed is 5.7 °C at 120 m and only 1 °C at 586 m a.s.l. (Contador et al., 2015).

Study site selection, habitat characterization and physico-chemical parameters

This study was conducted in lagoons located throughout the altitudinal gradient of the Róbalo river watershed (Table 1, Fig. 1). These lagoons are surrounded by evergreen forests and Sphagnum peat bogs, mixed forests and shrub patches, deciduous forests near the tree line (Krummholz), and high Andean tundra composed of cushion plants, mosses and lichens. Four lagoons were selected, located at different altitudinal levels, thus allowing for comparison between the different zones found within the gradient. Along with the physical description of each lagoon, the surrounding habitats were characterized. Within each lagoon, four different habitat types were identified based on substrate composition, and the presence/absence and type of vegetation (vascular or non-vascular): (i) rock bottoms (gravels from 1 to 10 cm), (ii) submerged vegetation (aquatic or submerged vascular plants), (iii) debris (leaf litter, roots, or logs), and (iv) aquatic mosses (completely or partially submerged). Temperature, pH and conductivity were measured three times in each habitat type on each sampling occasion (n = 3), using a multiparameter sensor (Conductivity pH TDS Hanna Tester HI98130). Additionally, HOBO® data loggers (model U22 Water Temperature Pro version 2) were permanently installed in each lagoon (anchored at 20 cm depth at a randomly selected location) from February 2015 until March 2016, recording water temperature every 4 h. Mean, maximum and minimum daily and monthly temperatures (°C) were calculated using the data obtained.

Biological sampling and macroinvertebrate identification

During the 2015 austral summer (January, February) freshwater macroinvertebrates were collected. From each lagoon three samples were taken in each habitat with a D-frame net (150 µm mesh). Each sample consisted of 1 min of net sweeping (standard sampling), and a complementary collection to obtain a comparable and representative sample of each habitat (Table 2). The latter samples were analyzed with each standard sample. Macroinvertebrates were stored in flasks in 70% ethanol, and returned to the laboratory for sorting, identification to the lowest taxonomical level possible and functional feeding group assignment according to the available literature (Domínguez & Fernández, 2009; Flint, Holzenthal & Harris, 1999; Libonatti, Michat & Torres, 2011; McLellan & Swick, 2007; Moorman et al., 2006; Siddall & Borda, 2004; Von Ellenrieder, 2003). Field and sampling permits were provided by Omora Ethnobotanical Park, and field collection of specimens were approved by Universidad de Magallanes (certificate number: no 80/CEC/2018).

Data analyses

Abiotic data were tested for normality and found to be non-normally distributed (Shapiro–Wilks test). The non-parametric Kruskal–Wallis (K–W) one way analysis of variance was therefore used. When significance was achieved, Wilcoxon post hoc pairwise comparisons were made. Diversity metrics were calculated for each lagoon along the altitudinal gradient. We used absolute and mean richness to describe taxonomic richness at each elevation, Shannon–Wiener diversity index (H′), Peilou’s evenness (J′) and N1 of Hill’s number series (Shannon–Wiener equivalent for comparison) to assess community structure (Hill, 1973; Magurran, 1988). Community structure indices were calculated using Primer 5.0 (Clarke, 1993). Changes in macroinvertebrate assemblage richness, abundance, diversity metrics and composition were tested with the non-parametric permutation test PERMANOVA (Anderson, 2001). Prior to PERMANOVA analyses, the PERMDISP test was performed to assess homogeneity within and between groups (Anderson, 2005). Euclidean distance between pairs of observations was calculated for the univariate analyses (Claudet et al., 2006). For multivariate analyses, Bray–Curtis dissimilarity matrices were calculated between pairs of observations and data were log(x + 1) transformed. All PERMANOVA analyses were performed using FORTRAN software (Anderson, 2005). For univariate and multivariate analyses we used a factorial design, considering elevation and habitat type. To determine specific importance of macroinvertebrates at each site, and their associated functional feeding group, we used the Similarity Percentages multivariate analysis SIMPER (Primer 5.0, Clarke, 1993), with a 90% contribution cutoff.

Altitudinal abiotic parameters

The sampled lagoons remained surface frozen for between four and seven months, with water temperature dropping to around 0 °C in the first weeks of May (austral autumn) and subsequently, depending on elevation, beginning to thaw in mid-September (austral spring) to mid-December (austral summer). The maximum average temperature was recorded in Castor lagoon (20 m a.s.l.), reaching 16.99 °C in December, while the minimum average temperature recorded was −0.19 °C in June in Róbalo lake (Fig. 2).

Conductivity showed significant differences along the altitudinal gradient (K–W, P < 0.0001, Table S1) associated with the increase of elevation, ranging from 147.3 to 49.5 µS (Table 3). Significant differences in pH were detected between Castor lagoon and the remaining three sampled lagoons along the gradient (K–W, P < 0.0001). This low elevation lagoon showed acidic values, while the others were close to neutral pH. Water temperature recorded in the sampling events showed significant variation between elevations (K–W, P < 0.0001, Table S1), decreasing with increasing altitude.

Altitudinal diversity metrics

Taxonomic richness

A total of 9384 macroinvertebrates were collected in the study, from which we identified 28 taxa belonging to 13 orders and 17 families. Of the 28 taxa, 16 were identified to genus/species level, and the remaining to order, family or subfamily level, catalogued as morpho-species (Table S2). Sites at lower elevations showed the highest absolute richness, S = 18 at 20 m and S = 17 at 250 m a.s.l., while in sites at the higher elevations we recorded fewer taxa, S = 11 at 480 m and S = 6 at 700 m a.s.l.

Altitudinal variation in richness, abundance and diversity metrics

The permutation analysis detected significant differences in mean macroinvertebrate richness between elevations and habitats (PERMANOVA, Elev F = 39.695, P = 0.0002; Hab F = 5.2073, P = 0.0056, Fig. 3, Table S3). Castor lagoon, at 20 m a.s.l., had a significantly higher mean richness than the other elevations. Pairwise comparisons detected differences between all elevations, with the exception of 250 and 480 m a.s.l. (pairwise comparison, P = 0.5779, Table S4). Significant differences were also detected in macroinvertebrate mean abundance along the elevation gradient (Fig. 3), between the habitats and in interactions between factors (PERMANOVA, Elev F = 56.544, P = 0.0002; Hab F = 12.54, P = 0.0002; Elev × Hab F = 10.515, P = 0.0002, Table S3). Mean abundance was significantly different between all elevations, between rock bottom habitats and the other three habitat types (pairwise comparisons in Table S4). Differences in diversity (H′, N1) and evenness (J′) indexes were also found with elevation, showing a similar pattern with values decreasing as elevation increased (PERMANOVA, Shannon index, F = 31.566, P = 0.0002; Hill’s N1 F = 25.478, P = 0.0002, Pielou’s evenness F = 18.156, P = 0.0002, Table S3), and indicating a more uniform community at 700 m a.s.l. in Bandera lagoon. However, significant differences in mean abundance between habitats were found for the Shannon diversity index and Pielou’s evenness (PERMANOVA, Shannon index F = 3.4004, P = 0.0294; Pielou’s evenness F = 5.9582, P = 0.0026; pairwise comparisons in Table S4).

Altitudinal community structure

The freshwater macroinvertebrate assemblage showed variability along the gradient, although there were common elements through different elevations. In Bandera lagoon, 700 m a.s.l., 78.22% of the assemblage was represented by the freshwater amphipod genus Hyalella, with oligochaetes contributing 14.8% (Table 4, Fig. 4). In a similar fashion, in El Salto lagoon, Hyalella amphipods and oligochaetes were the most frequent taxa (49.67% and 24.68%, respectively; Table 4, Fig. 4). In Róbalo lake, 250 m a.s.l., amphipods represented almost half of the community (46.72%), while oligochaetes, leptophlebiid mayflies, biting and non-biting midges, leeches, and lymnaeid snails comprised the remaining representative fauna (Table 4, Fig. 4). The abundant Hyalella amphipods (30.59%), non-biting midges, lechees, diving beetles and water boatmen were the most abundant taxa in Castor lagoon, 20 m a.s.l. (Table 4, Fig. 4). Significant differences in macroinvertebrate community composition were apparent along the elevation gradient, between the habitats and in interactions between factors (PERMANOVA, Elev F = 39.695, P = 0.0002; Hab F = 5.2073, P = 0.0002; Elev × Hab F = 1.4874, P = 0.0002, Table S5). Pairwise comparisons detected differences between all elevations and habitats, with the exception between submerged vegetation and two other habitats (pairwise comparison, rock bottoms P = 0.1376, aquatic mosses P = 0.0852, Table S6).

Functional feeding groups (FFG) comprised six main categories: collector-filterers, collector-gatherers, collector-predators, herbivores (grazers, shredders and/or piercers), predators, and scrapers. Collector-gatherers and predators where present in every lagoon, although with varying representativeness. Collector-gatherers were the dominant group at all elevations, with representativeness increasing with elevation, ranging from 48.7% at 20 m a.s.l. to 99.7% at 700 m a.s.l. (Table 5, Fig. 5). Predators comprised almost a quarter of the FFG assemblage at each elevation, excepting a marked decrease at 700 m a.s.l. (to 0.3%). Herbivores, collector-filterers, collector-predators and scrapers were present at almost every elevation, but rarely represented more than 10% of the total assemblage (Table 5, Fig. 5). Similar to community composition, significant differences were found in FFG composition with elevation, between habitats and in their interaction (PERMANOVA, Elev F = 25.168, P = 0.0002; Hab F = 4.6319, P = 0.0002; Elev x Hab F = 2.1298, P = 0.0002, Table S7). Differences between all elevations and habitats were detected by pairwise comparisons, excepting between submerged vegetation and two other habitats (pairwise comparison, rock bottoms P = 0.0544, aquatic mosses P = 0.3382, Table S8).