Kuhn Et Al. - 2011 - Spatial Variability in Macroinvertebrate Assemblages Along and Among Neighbouring Equatorial Glacier-Fed Streams
Kuhn Et Al. - 2011 - Spatial Variability in Macroinvertebrate Assemblages Along and Among Neighbouring Equatorial Glacier-Fed Streams
Kuhn Et Al. - 2011 - Spatial Variability in Macroinvertebrate Assemblages Along and Among Neighbouring Equatorial Glacier-Fed Streams
SUMMARY
1. During the past two decades, understanding of the structure and function of glacier-fed
stream ecosystems at temperate latitudes has increased substantially. In contrast,
information on their tropical counterparts is very limited. We studied three neighbouring
glacier-fed streams in the tropical Andes of Ecuador. Our main goals were (i) to determine
overall longitudinal patterns in density, taxon richness and the composition of macroin-
vertebrate assemblages and driving factors in equatorial glacial streams and (ii) to examine
variability among replicate streams in faunal metrics and assemblages, and stream-specific
effects of supposed environmental key factors.
2. We measured four geographical and 17 environmental factors and collected five Surber
samples (500 cm2) of macroinvertebrates at each of nine sites, three sites along three
streams. The streams were located 1–5 km apart. In each stream, the three sites were
placed at comparable distances from the glacier and were grouped as ‘upper’ (50–200 m),
‘middle’ (1.5 km) and ‘lower’ sites (3.5–5.6 km).
3. In total, 2200 individuals (64% chironomids) were collected and 47 taxa (30 dipterans, 18
of these Chironomidae) identified. Density ranged from 176 to 372 ind. m)2, and the
number of taxa ranged from 2 to 6 at the upper sites and 868–3044 ind. m)2 and 21–27 taxa
at the lower sites. Density, number of taxa, rarefied richness and axis-1 coordinates from a
MDS ordination increased logarithmically with distance from the glacier. These faunal
metrics were equally related to altitude and glacier per cent of catchment and correlated
with maximum conductivity, mean temperature, mean daily maximum temperature and a
channel stability index. As expected, the mean difference in distance decay in similarity
was higher at the upper (47% km)1) than at the lower reaches (20% km)1) of the streams.
4. The number of taxa varied among sites within the upper and middle groups, but not
among the lower sites. In contrast, but in accordance with our expectation, assemblage
composition did not differ among upper sites but did so at middle and lower sites,
following a supposed decrease in environmental harshness along the streams. Relation-
ships between faunal metrics and the four environmental variables mean temperature, the
Correspondence: Dean Jacobsen, Freshwater Biological Section, Department of Biology, University of Copenhagen, Helsingørsgade 51,
DK-3400 Hillerød, Denmark. E-mail: Djacobsen@bio.ku.dk
stability index, chlorophyll a and coarse particulate organic matter also varied among the
three streams. Generalised linear model analyses revealed that temperature interacted
with stream on macroinvertebrate density, while chlorophyll a had a significant effect on
the number of taxa in interaction with stream and stability.
5. The basic predictions of the Milner et al. (2001a), model regarding longitudinal faunal
patterns and temperature and stability as main driving factors were met by our three
replicate equatorial glacial streams. Qualitative departures from the model were mainly
because of zoogeographical differences. We demonstrated that variability in assemblages
between comparable sites in closely situated streams was considerable, and the effect of
key environmental factors varied among streams and interacted with other factors.
Quantifying spatial variation in benthic assemblages may help us foresee possible
consequences for biodiversity as a result of glacial retreat.
2
1
2
1
3 2 1
3
in environmental conditions. The field work was mean daily minimum temperature, coefficient of
carried out during April–May 2009. variation in temperature (=CV temp.) and CV water
depth.
Instead of the commonly used Pfankuch index
Environmental variables
(Pfankuch, 1975), we developed and applied a new
In addition to four geographical variables (altitude, stability index based on four measurable parameters:
catchment extension, per cent of catchment glacier- CV temp., CV water depth, skewness of a conductiv-
covered and distance from glacier), we measured 17 ity versus time curve from a dilution gauging and
environmental site characteristics. However, referring tractive force divided by mean substrate particle size
to the conceptual model for the zonation of macroin- (from transects). The values of each of these four
vertebrates in glacier-fed streams (Milner et al., parameters were ranked from 1 to 9, one being given
2001a), we paid special attention to the effect of to the highest values of CV temp., CV depth and
temperature and physical channel stability as driving tractive force ⁄ mean substrate particle size and to the
forces and to two measures of available food sources, lowest value of skewness, and then the four ranks
benthic periphyton biomass and coarse particulate were summed for each site (the stability index was
organic matter, as additional explicative factors. therefore ranked from 4 to 36). We believe this index
At each site, electronic temperature and water-level captures the instability caused by glacial melt water
loggers (HOBO U20) were placed inside a plastic tube contribution to the study region. Glacial melt water is
(part of a drainpipe, length: 60 cm and diameter: close to 0 C, and the daily melting and discharge
8 cm) and fastened to the bank with string and metal regime determines the variability in temperature and
tubes. The exact position in reference to the bottom of depth. The components of the stability index and
the streambed and water surface was noted and used some additional variables are explained later.
in calculating the exact water level and water-level Stream slope at each site was measured using a
variations. The loggers were left in the field to gather transparent plastic tube carrying flowing water from
data and retrieved 1–2 months later. From these the upstream to the downstream end of the reach. The
recordings, several variables were obtained: mean slope was calculated as the difference between the
temperature, mean daily maximum temperature, water level inside the tube when raised until flow
2011 Blackwell Publishing Ltd, Freshwater Biology, doi:10.1111/j.1365-2427.2011.02648.x
Spatial variability in equatorial glacier-fed streams 5
stopped and that of the surface of the stream water at Stones were collected for the quantification of
the downstream end, divided by the distance between chlorophyll a as a measure of benthic algal biomass.
the upstream and downstream end of the tube Five samples, each containing three stones (size: large
(c. 25 m). At each of the nine study sites, 10 transects gravel), were collected at random, covered with 96%
with three-metre intervals were laid across the stream ethanol and placed in the dark until processing in the
and used to record width, depth and mineral sedi- laboratory. There, the containers were given a 10-min
ment type. Depth and bottom material were recorded ultrasonic bath to increase the extraction efficiency.
at five points along each transect, giving a total of 50 After settlement for a few hours, a sample was
measurements per stream. The sediment was transferred to a spectrophotometer and absorption
additionally grouped according to grain size: silt measured at 665 and 750 nm. Stone surface area was
(0.004–0.062 mm), sand (0.063–2 mm), small gravel estimated using the formula: A = ((LW) + (LH) +
(2–30 mm), large gravel (30–64 mm), small cobble (WH)) · 1.15 proposed by Graham, McCaughan &
(64–128 mm), large cobble (128–256 mm) and boulder McKee (1988), where A is the area, L is length, W is
(256 mm<) following the Wentworth Scale (Giller & width, H is height of the stone and 1.15 is a fixed
Malmqvist, 1998). Based on these 50 points, the factor to corrugate for the irregular shape of the
substrate type diversity was calculated (Shannon– stones. Chlorophyll a was calculated using this
Wiener index). formula: Chlorophyll a (mg m)2) = Abs(665–750) ·
Tractive force was calculated as (TG = qgRS), where V ml · 10 000 cm2 m)2 ⁄ 83.4 ml mg)1 · A cm2, where
TG is the tractive force (N m)2), q is the density of V is the volume of ethanol used, 83.4 is the absorption
water (1000 kg m)3), g is the gravitational acceleration coefficient for chlorophyll a in 96% ethanol and A is
(9.81 m s)1), R is the hydraulic radius (m) (estimated the summarised area of the stones.
as width + depth) and S is the slope of the energy line.
The tractive force is a measure of the power per-
Macrobenthos sampling
formed by the water on the streambed; with higher
values, larger particles will be moved by the water At each site, five quantitative Surber samples
and the requirement for the macroinvertebrates to (500 cm2; mesh size 200 lm) were collected randomly
hold on increases. In periods with high discharge, the from pebble–cobble substratum in riffle ⁄ run habitats.
level of tractive force increases. All samples were collected during daytime and
Current velocity and discharge were measured preserved in the field in 70% ethanol. In the labora-
twice at each site at the end of our field work tory, samples were rinsed through a 200-lm sieve and
during morning ‘base flow’ by means of dilution sorted without the use of magnification. No subsam-
gauging (White, 1978). A bucket of a known amount pling was applied. Benthic coarse particulate organic
of dissolved salt (volume and conductivity) was matter (CPOM) was obtained by filtering once more
added at the upstream end of the 15- to 25-m the remaining material from Surber samples through
stream reach, and we measured the conductivity a 200-lm sieve. CPOM was oven-dried (80 C, 24 h),
every 5 or 10 s at the downstream end of the reach. weighed, burned in a mufle furnace (for 4 h at 550 C)
Large skewness of the conductivity versus time and reweighed to calculate ash-free dry mass
curve from the dilution gauging is considered a (AFDM).
measure of hydraulically dead space and thus of the Identification of the Ecuadorian stream fauna to
extent of low-stress refugia for macroinvertebrates. species level is not possible. Invertebrates other than
Mean current velocity was calculated as the time non-insects were identified mostly to genus according
elapsed for half of the salt to pass the stream reach to Roldán (1996), Merritt & Cummins (1996), Fernán-
divided by the length of the reach. Discharge (Q) dez & Domı́nguez (2001) and separated into morpho-
was calculated from these same measurements by species. Larvae and pupae of Chironomidae were
the formula: Q = V · C ⁄ A, where V is the volume of sorted under a stereoscopic microscope at 10 · mag-
salt water in the bucket, C is the conductivity of the nification, dehydrated in 96 and 99% ethanol and
salt water in the bucket and A is the integrated area mounted in Euparal. Larvae were identified to genus
beneath the curve from the conductivity versus time and morphospecies under a compound microscope at
plot. maximum 400 · magnification using current taxonomic
2011 Blackwell Publishing Ltd, Freshwater Biology, doi:10.1111/j.1365-2427.2011.02648.x
6 J. Kuhn et al.
literature (Wiederholm, 1983; Epler, 2001). Larvae of ANOSIM generates a statistical parameter R that is
the subfamily Diamesinae were identified according indicative of the degree of separation between groups:
to Ruiz-Moreno et al. (2000). a score of 1 indicates complete separation and a score
of 0 indicates no separation (Gucht et al., 2005).
Monte-Carlo randomisation of the group labels was
Data treatment
used to generate null distributions to test the hypoth-
We applied four different univariate metrics to cha- esis that within-group similarities are higher than
racterise the fauna: density of individuals, number of would be expected by chance. These analyses were
taxa encountered in the five samples (also denoted performed using PAST (Paleontological statistics,
taxon density), rarefied richness and Pielou’s index of version 1.79, Oslo).
evenness. To correct for the effect of varying number Generalised linear model analysis (GLM) was used
of individuals in samples on the number of taxa to build Poisson log-linear models to test the effect of
recorded, we included a finite, individual-based mean temperature, stability, benthic chlorophyll a and
rarefied richness (performed without replacement on CPOM on the faunal metrics density, number of taxa,
44 individuals; the lowest number in the five pooled rarefied richness and Pielou’s evenness. We focussed
Surber samples of any site). These were calculated on the effect of temperature and stability as explan-
using the software ‘Species diversity and richness’ atory variables because they are the two basic factors
version 3.02, Pisces Conservation Ltd, Hants, UK. in the conceptual model by Milner et al. (2001a,b) and
One-way A N O V A s were performed to analyse for because an earlier study along stream G12 confirmed
differences in faunal density and taxon richness these to be the factors primarily related to the
between the three sites within groups, using individ- distribution of the fauna (Jacobsen et al., 2010). We
ual Surber samples. A N O V A s and Pearson’s product also included a ‘stream’ variable to allow for inter-
moment correlation analyses between log (x + 1)- stream comparisons of the effect of these variables.
transformed environmental variables and faunal met- Stream was treated as a fixed effect, but we obtained
rics were performed in Excel. similar results by means of the generalised linear
Non-metric multidimensional scaling (NMDS) based mixed model function (glmmPQL; MASS library for
on Bray–Curtis similarities calculated on square-root- R), with stream as a random effect. Change in density,
transformed assemblage data was used to examine richness and evenness owing to each factor was
spatial patterns in macroinvertebrate assemblage com- modelled considering each factor independently and
position among sites. This ordination technique has no in combination with other factors, including biologi-
assumptions with repect to distributions and repre- cally reasonable two-way interactions, and squared
sents samples as points in low-dimensional space, such variables. Indeed, the relationships between density
that the relative distances among points reflect the and taxon richness, and environmental variables are
relative similarities of the samples with respect to both often curvilinear (Austin, 1980). The more parsimoni-
abundance and composition (Gucht, Vandekerckhove ous model was identified using the corrected Akaike’s
& Vloemans, 2005). The NMDS goodness of fit was Information Criterion (AICc), which is recommended
estimated with a stress function (which ranges from 0 to for small sample sizes (AIC, see Venables & Ripley,
1), with values close to zero indicating a good fit. The 2002) in likelihood ratio tests to find the difference
Bray–Curtis index was also used to illustrate the between the initial model and the reduced model,
similarities between faunal assemblages at different dropping an ‘effect’ term. All analyses were per-
sites: 0% indicating no species in common and 100% formed on log-transformed data using the MASS
indicating complete species match occurring in the library for R (R Development Core Team, 2008).
same abundances. These procedures were performed
using Primer 5 Ltd. version 5.2.4, 2001, Plymouth, UK.
Differences in the composition of the macroinver- Results
tebrate assemblages among site groups were further
Environmental variables
tested using an analysis of similarities (ANOSIM). The
ANOSIM tested the null hypothesis that within-group The nine stream sites could all be characterised as
similarity was equal to between-group similarity. small (width 0.39–2.04 m; depth 0.03–0.19 m) with
2011 Blackwell Publishing Ltd, Freshwater Biology, doi:10.1111/j.1365-2427.2011.02648.x
Spatial variability in equatorial glacier-fed streams 7
Table 1 Environmental characteristics measured at the nine sites along the Antisana streams G15, G14 and G12
Site Variable G15.1 G15.2 G15.3 G14.1 G14.2 G14.3 G12.1 G12.2 G12.3
Altitude (m.a.s.l.) 4835 4521 4335 4789 4535 4196 4728 4496 4225
Catchment (km2) 0.46 1.02 1.47 0.65 1.12 14.27 2.13 3.21 4.33
Glacier cover in catchm. (%) 94 45 31 92 67 10 100 66 49
Distance from glacier (m) 100 1500 3500 200 1500 5600 50 1500 4300
Min. turbidity (NTU) 812 211 65 181 111 58 285 442 411
Max. pH 7.98 7.07 6.44 6.31 7.04 7.96 7.47 6.36 6.39
Max. Conduc. (lS cm)1) 3 8 24 11 10 47 5 10 14
Mean temp. (ºC) 1.56 4.20 5.05 2.48 4.31 7.42 1.59 2.86 5.38
CV temp. (%) 111 102 63 126 80 34 180 46 51
Mean daily max. temp. (ºC) 4.38 12.35 10.65 9.47 10.70 11.26 5.44 5.21 10.94
Mean daily min. temp. (ºC) 0.47 0.30 1.56 0.00 0.60 4.32 0.00 1.36 2.62
Mean depth (cm) 5 6 10 6 10 18 3 13 19
CV depth (%) 52 84 69 18 20 30 110 35 27
Substrate types (nr) 6 8 6 5 6 8 5 6 10
Substrate type div. (H¢) 1.63 1.88 1.52 1.60 1.70 1.95 1.05 1.60 2.11
Mean current (m s)1) 0.24 0.11 0.13 0.25 0.24 0.19 0.22 0.38 0.21
Mean width (cm) 67 59 63 72 133 133 39 204 158
Slope (cm m)1) 11.7 9.7 7.3 24.0 10.8 2.2 1.4 8.7 9.0
Stability index 9 18 25 15 21 31 16 24 21
Chlorophyll (mg m)2) 0.14 3.92 2.63 0.55 6.17 1.85 0.27 3.15 0.99
CPOM (g AFDM m)2) 7.98 13.26 145.50 8.74 21.18 22.04 8.68 43.55 33.26
AFDM, ash-free dry mass; CPOM, coarse particulate organic matter; CV, coefficient of variation.
medium flow velocity (0.11–0.38 m s)1), diverse sub- tendency in CPOM pools, but with large local devi-
stratum composed mainly of gravel, pebble and ations, and even at the upper sites close to the glaciers,
cobbles, high turbidity (58–812 NTU), low conductiv- biomasses of c. 10 g AFDM m)2 were found.
ity (3–47 lS s)1) and circum-neutral pH (6.31–7.96)
(Table 1). Other variables differed more among
General longitudinal patterns and driving factors
sites, e.g. slope (1.4–24.0%), benthic chlorophyll a
(0.14–6.17 mg m)2) and benthic CPOM (8–145 g In the 45 Surber samples from the nine sites, we
AFDM m)2). collected a total of 2200 individuals, 64% of them
The geographical variables such as altitude, glacier chironomids. A total of 47 taxa, of which 30 were
% of catchment and distance from the glacier were all dipterans, 18 of these Chironomidae, were identified
highly correlated (Appendix 1). All environmental (Appendix 2). Density varied between 176 and 3044
variables, except pH, CV depth, mean current, mean ind. m)2. The number of taxa ranged from 2 to 6 at the
width and slope, were significantly correlated with upper sites and from 21 to 27 at the lower sites.
one or more of these geographical variables. In Chironomidae was the most abundant group at all
general, many of the environmental variables were sites, except at site G15.2 that had a high abundance of
intercorrelated (Appendix 1). Simulium. In particular, Podonominae was completely
All temperature parameters increased downstream dominant at the upper sites (>98% of ind.). Single
from the glacier (temperature variation decreased), individuals of a few other groups were found at the
but mean daily maximum temperature showed an upper sites (Muscidae and Staphylinidae). In all three
irregular longitudinal pattern. Overall, the proposed streams, Podonominae became less dominant and
stability index increased with the distance from the Orthocladiinae more abundant with increasing
glaciers, even though G12.3, the lowermost site, distance from the glaciers. The two genera Atopsyche
scored the lowest stability index. Benthic algal bio- and Cailloma (Hydrobiosidae and Trichoptera) were
mass (chlorophyll a) was quite similar in the three found in all three streams at middle and lower sites,
streams and peaked at middle sites in all three while the ephemeropteran Andesiops (Baetidae) did
streams. There was an overall downstream increasing not appear until the lowest sites. The plecopteran
2011 Blackwell Publishing Ltd, Freshwater Biology, doi:10.1111/j.1365-2427.2011.02648.x
8 J. Kuhn et al.
Claudioperla (Gripopterygidae) was rare and appeared G15 G14 G12
in low numbers only at site G14.3. Elmidae (Coleop-
(a) 3500
tera) were common at two of the three lower sites.
3000
Overall, with increasing distance from the glacier,
Number of taxa
a, and only rarefied richness with CPOM. In addition, 20
the number of taxa and rarefied richness were 15
significantly correlated with CV temp., mean depth
10
and the number of substrate types. Pielou’s evenness
did not correlate with any of the environmental 5
variables (Table 2). 0
The NMDS ordination had a low stress (0.02), 0 1000 2000 3000 4000 5000 6000
indicating a good degree of representation (Fig. 3).
(c) 14
The diagram revealed a clear longitudinal pattern in
the composition of the macroinvertebrate assem- 12
Rarefied richness
0.17 )0.05 0.03 )0.23 )0.32 )0.33 )0.01 )0.17 )0.05 )0.07 1.00
Pielou’s evenness
revealed in the three streams (Fig. 2). Following
Rarefied richness
logarithmic relationships between faunal metrics and
0.94** 1.00
three middle sites (all situated at 1500 m from the
Density ðind: m2 Þ glaciers), this variability among streams equalled
1.00
2300 m (200–2500 m) for density, 1700 m (400–
CPOM ðg AFDM m2 Þ 2100 m) for taxon richness, 800 m (700–1500 m) for
0.38
0.63
rarefied richness and 1800 m (800–2600 m) for assem-
Chlorophyll a ðmg m2 Þ blage composition expressed as axis-1 coordinates
)0.24 0.62 0.58 )0.43 0.39 )0.30 0.81** 0.63
)0.12 0.68* 0.59 )0.50 0.38 )0.26 0.84** 0.62
0.81** 0.80** )0.15 0.67* 0.55 )0.37 0.47 )0.28 0.89** 0.57
from the NMDS ordination (Fig. 5). A N O V A s con-
Stability index
firmed that density varied significantly among the
Table 2 Pearson’s correlation coefficients between fauna metrics and environmental variables. Bonferroni correction of P not applied
Substrate types
according to their location as upper, middle or lower
0.27
CV depth (%)
sites) were significantly different (R = 0.951,
P = 0.003, ANOSIM). Further, ANOSIM analyses
0.72*
0.08
)0.41 0.36
CV temp. (%)
similarity between replicate sites which decreased
Mean temp. (C) slightly with increasing distance from the glaciers
(Fig. 4).
Max. conductivity ðlS cm1 Þ Relationships between the four univariate faunal
metrics, the NMDS axis-1 coordinates and the focal
environmental variables mean temperature, the sta-
0.60
Max. pH
bility index, chlorophyll a. and CPOM also varied
)0.98** 0.68* )0.82** 0.94** )0.64
Min. turbidity (NTU) among the three streams (Fig. 6). Recognising that
)0.03 0.40
0.13
Catchment ðkm2 Þ
Altitude (m.a.s.l.)
Rarefied
of taxa
Pielou’s
Density
0.8
0.6
12.3
0.4
15.1 14.1
0.2
Fig. 3 Bray–Curtis similarity-based Non-
0 metric multidimensional scaling ordina-
15.3 tion showing square-root-transformed
–0.2 12.1 faunal data from the three Antisana
streams. Results are based upon sum-
–0.4
marised data from five Surber samples
15.2
from each site. Site groups 1, 2 and 3 are
–0.6 14.2
12.2 marked. Stress value is shown. Environ-
–0.8 mental variables that show significant
Pearson’s correlations with the x-coordi-
–1.6 –1.4 –1.2 –1.0 –0.8 –0.6 –0.4 –0.2 0 0.2 0.4 0.6 0.8 1.0 1.2 nates in the ordinations diagram are
Distance to glacier Mean temperature Stability index included.
70
60
Mean similarity (%)
50
40
30
20
10
0 1800 m
1 versus 1 2 versus 2 3 versus 3 1 versus 2 2 versus 3 1 versus 3
Sites
500
(a) 500
(b) 500
(c) 500
(d)
0 0 0 0
0 1 2 3 4 5 6 7 8 0 5 10 15 20 25 30 35 0 1 2 3 4 5 6 7 0 20 40 60 80 100 120 140
30 30
30 30
25 25
Number of taxa
25 25
20 20 20 20
15 15 15 15
10 10 10 10
5 5 5 5
(e) (f) (g) (h)
0 0 0 0
0 1 2 3 4 5 6 7 8 0 5 10 15 20 25 30 35 0 1 2 3 4 5 6 7 0 20 40 60 80 100 120 140
14 14 14 14
Rarefied richness
12 12 12 12
10 10 10 10
8 8 8 8
6 6 6 6
4 4 4 4
2 (i) 2 (j) 2 (k) 2 (l)
0 0 0 0
0 1 2 3 4 5 6 7 8 0 5 10 15 20 25 30 35 0 1 2 3 4 5 6 7 0 20 40 60 80 100 120 140
1 1 1 1
0.9 0.9
Pielou evenness
1 1 1 1
0 0 0 0
–1
–1
(q) (r) –1
(s) –1
(t)
–1.5 –1.5 –1.5 –1.5
0 1 2 3 4 5 6 7 8 0 5 10 15 20 25 30 35 0 1 2 3 4 5 6 7 0 20 40 60 80 100 120 140
Mean temp. (ºC) Stability index Chlorophyll a (mg m–2) CPOM (g AFDM m–2)
Fig. 6 Density of individuals (a,b,c,d), number of taxa (e,f,g,h), rarefied richness (i,j,k,l), Pielou’s evenness (m,n,o,p) and Non-metric
multidimensional scaling x-coordinates (q,r,s,t) of macroinvertebrate assemblages as a function of mean temperature, stability
index, chlorophyll a and coarse particulate organic matter in each of the three Antisana glacial streams.
were probably due to sporadic immigrant individuals Patagonian streams (Miserendino & Pizzolon, 2000).
from lower sites, rather than representing perma- Even though the family is relatively common in the
nently reproducing populations. high Andes, it has not been found above 3900 m a.s.l.
Other Diamesinae genera (probably Paraheptagyia) in Ecuador (Jacobsen, 2004). The model of Milner et al.
appeared lower down the streams, but they never (2001a) further predicts that chloroperlid and
became numerically important. In New Zealand, the nemourid stoneflies should appear as soon as Tmax
Diamesinae were largely replaced by other taxa such exceeds 4 C. These two families do not occur in the
as the ephemeropteran Deleatidium (Leptophlebiidae), Ecuadorian highlands, where Gripopterygidae
at the sites nearest to the glacier, even though these (Claudioperla) is the only family of stoneflies and was
sites were the coldest (Milner et al., 2001b; Winterbo- found at the lowest sites. This Austral family is also
urn et al., 2008; Cadbury et al., 2010). Leptophlebiidae common in the streams of glacial origin in Bolivia
have been collected in a glacial stream at 4400 m a.s.l. (Molina, 2004), Patagonia (Weis & Bonetto, 1988) and
in Bolivia (Molina, 2004) and are also common in cold New Zealand (Milner et al., 2001b). Hydrobiosidae
2011 Blackwell Publishing Ltd, Freshwater Biology, doi:10.1111/j.1365-2427.2011.02648.x
12 J. Kuhn et al.
Table 3 Results of the generalized linear model deviance anal- expectations and further supported this logarithmic
ysis on on the fauna metrics density, number of taxa, rarefied development of macroinvertebrate communities
richness and Pielou’s evenness
along glacier-fed streams. These relationships were
Effect AICc D AICc LRT P-value basically the same with other geographical variables
such as altitude and per cent of glacier cover in the
Density
Stream · Temperature 147.2 6.1 4.321 <0.001 catchment.
Number of taxa Temperature and stability were the two main
Stream · Chlorophyll a 56.9 2.2 3.631 0.023 variables structuring faunal patterns (Table 2, Fig. 3).
Rarefied richness
Temperature parameters and the stability index were
None
Pielou evenness the only environmental variables that correlated
Stream · Stability 74.3 2.3 3.865 0.013 significantly with the density of individuals, the
Stream · Stability 79.7 7.7 4.633 0.002 number of taxa and geographical variables. Like
· Chlorophyll a
Jacobsen et al. (2010), we found that mean daily
Only significant effects are shown. AICc is the corrected maximum temperature showed an irregular longitu-
Akaike’s Information Criterion for the initial model after dinal pattern, and no truly kryal zone (Tmax < 4 C
removal of the ‘effect’ term. DAICc corresponds to the difference
sensu Steffan, 1971; Ward, 1994; Milner et al., 2001a)
between the AICc of the initial model and that of the reduced
model. Likelihood ratio test (LRT) and associated P-value test was identified, as even mean daily maximum tem-
the hypothesis that the suppression of the ‘effect’ term provides perature exceeded 4 C at all localities. This could be
no better fit than the initial model. explained by the high insolation at the equator,
heating up the small amount of melt water in the
morning faster than the discharge of cold melt water
(Trichoptera) were found in all three streams, and the increases (Jacobsen et al., 2010). High water tempera-
family is abundant and widespread in high Andean ture in proglacial reaches has also been reported from
(Jacobsen, 2004, 2008) as well as in New Zealand streams with low discharge from small glaciers in the
alpine streams (Milner et al., 2001b; Cadbury et al., European Alps (Maiolini & Lencioni, 2001). In con-
2010). This family of predatory free-living larvae trast, mean temperature increased systematically with
seems to occupy the equivalent ecological niche to increased distance from the glaciers, and this param-
Rhyacophilidae in European glacial streams (Castella eter was also more closely related to the faunal
et al., 2001). Coleoptera are usually not present in metrics than was maximum temperature. This sug-
alpine streams at temperate latitudes, and even gests that mean temperature might be an ecologically
Elmidae are rarely abundant (Ward, 1994). Elmidae more meaningful parameter in setting physiological
(Coleoptera) were found in two of the three lower limits for the distribution of species.
sites on Antisana, and this family is very common in The proposed stability index correlated significantly
high Andean streams (Jacobsen, 2008). Thus, qualita- with density, the number of taxa and rarefied
tive departures from the Milner et al. (2001a) model richness. Following the intermediate disturbance
were mainly attributable to zoogeographical differ- hypothesis, both taxon richness (Townsend, Scars-
ences between the Neotropical and the north-temper- brook & Dolédec, 1997) and assemblage evenness
ate (and New Zealand) stream fauna. (Death & Winterbourn, 1995) are expected to peak at
intermediate levels of disturbance (or stability)
because a certain level of disturbance should impede
General longitudinal patterns and driving factors
dominance from competitively superior species while
Overall, our data support the general conceptual allowing the existence of species able to cope with
model of logarithmic increase in density and richness such environmental conditions in a space with low
of fauna with increasing distance from a glacier interspecific competition (McAuliffe, 1984). Positive
(Milner et al., 2001a) and confirm the longitudinal rather than peaked relationships between faunal
patterns reported earlier by Jacobsen et al. (2010) from metrics and stability are generally found in glacier-
one of the streams. The greater distance decay in fed streams, probably because glacial streams only
similarities between sites close to the glaciers than cover the lower end of a stability gradient. Surpris-
further downstream was in accordance with our ingly, we found no relationship between stability and
2011 Blackwell Publishing Ltd, Freshwater Biology, doi:10.1111/j.1365-2427.2011.02648.x
Spatial variability in equatorial glacier-fed streams 13
evenness. Further studies are needed to evaluate the such as temperature and stability also varied among
wider usefulness of our proposed stability index, but streams (Table 3, Fig. 6). If we discard the possibility
here it seemed to work at least as well as the that at some sites variables were not correctly
commonly used, but nevertheless subjective, Pfan- measured, we may conclude that their regulating
kuch index. effect is stream specific and depends on other,
Benthic algal biomass (chlorophyll a) varies greatly perhaps unknown, factors. The three streams differed
within and among glacial streams; in some cases, somewhat with respect to exposure, insulation and
biomass increases steadily with increasing distance morphometry, and similar hydrological events could
from the glacier (Lods-Crozet et al., 2001), but this is therefore cause different effects in the three streams,
not always so (Gı́slason et al., 2001; Cadbury et al., perhaps implying unsynchronised temporal variabil-
2010). In our three streams, algal biomasses were quite ity in macroinvertebrate assemblages. We suspect that
similar and peaked at middle sites in all three streams, these circumstances are responsible for the varying
a result confirming the findings of Jacobsen et al. relationships between fauna and environmental fac-
(2010). This was probably due to high insolation at tors such as temperature and stability among the
these sites compared with the generally more incised streams.
and therefore more shaded lower sites. No overall Overall, chlorophyll a did not relate significantly to
relationship between chlorophyll a and faunal metrics any of the faunal metrics, but it did show a stream-
was found. specific effect on the number of taxa and evenness in
Our levels of benthic CPOM were very high the GLM analyses. Chlorophyll a as such is not a very
compared with other studies (Lods-Crozet et al., precise measure of either quality or quantity of
2001; Cadbury et al., 2010). Even at the upper sites benthic algae as food for macroinvertebrates, as the
close to the glaciers, we found biomasses of c. 10 g taxonomic composition of the periphyton community
AFDM m)2, confirming earlier results reported by may vary greatly among streams (Biggs, 1996; Rott
Jacobsen et al. (2010). CPOM seemed to be extremely et al., 2006). For instance, both filamentous blue-green
patchy, and at sites with high values, the CPOM was algae and diatoms possess chlorophyll a, but their
primarily composed of moss and filamentous algae value as a food source for macroinvertebrates is very
detached from stones during the Surber sampling. different. Nor was it surprising that chlorophyll a
Thus, general patterns and correlations between the interacted with stability on evenness, because the
fauna and CPOM levels are difficult to demonstrate. biomass and type of algae in streams is closely
controlled by disturbance regime (Biggs, Goring &
Nikora, 1998; Hieber et al., 2001). In addition, benthic
Among-stream faunal variability and stream-specific
algal biomass is temporally variable (Cadbury et al.,
effects
2010).
In our study, spatial location and environmental Local assemblages were thus structured by the
features were obviously very strongly related, so we complex interaction of the many variables constituting
could not test for the importance of niche processes the entire environmental setting. Our results under-
versus dispersal limitation as drivers of local diversity line that assemblages distributed along spatial gradi-
and community patterns (Thompson & Townsend, ents such as glacial streams with multiple
2006; Heino & Mykrä, 2008), neither was this our goal. environmental factors varying simultaneously are
Nevertheless, as mean similarity among replicate sites not regulated independently by each environmental
decreased from upper to lower sites, following a factor in a similar and straightforward way. This
supposed decrease in environmental harshness along could explain why different studies on glacier-fed
the streams, this seems to support an expected streams have reached different conclusions on the
stronger niche selection close to the glaciers than importance of the same regulating factors, such as
further downstream (Leibold et al., 2004; Chase, 2007). chlorophyll a having a greater effect than CPOM on
Even though our three streams appeared very similar, macroinvertebrate assemblages in the Swiss Mutt
we found considerable variation in faunal metrics stream (Lods-Crozet et al., 2001), while the opposite
among replicate sites (Fig. 5), and the quantitative was true in the New Zealand Rob Roy stream
response of fauna metrics to environmental variables (Cadbury et al., 2010).
2011 Blackwell Publishing Ltd, Freshwater Biology, doi:10.1111/j.1365-2427.2011.02648.x
14 J. Kuhn et al.
The current rapid glacial retreat (Francou & Cou- linking longitudinal zonation of physical habitat and
drain, 2005; Bradley et al., 2006; Ceballos et al., 2006) macroinvertebrate communities. Ecohydrology, DOI:
is, in the near future, expected to affect natural 10.1002/eco.185.
communities and biodiversity in glacial streams Castella E., Adalsteinsson H., Brittain J.E., Gislason G.M.,
(Brown, Hannah & Milner, 2007; Milner, Brown & Lehmann A., Lencioni V. et al. (2001) Macrobenthic
invertebrate richness and composition along a latitu-
Hannah, 2009). Quantifying spatial (and temporal)
dinal gradient of European glacier-fed streams. Fresh-
variation in glacier-fed stream communities and
water Biology, 46, 1811–1831.
relationships with environmental factors will help us Ceballos J.L., Euscátegui C., Ramı́rez J., Cañon M.,
interpret future changes in macroinvertebrate assem- Huggel C., Haeberli W. et al. (2006) Fast shrinkage of
blages as sentinels of glacial retreat. tropical glaciers in Colombia. Annals of Glaciology, 43,
194–201.
Chase J.M. (2007) Drought mediates the importance of
Acknowledgments
stochastic community assembly. Proceedings of the
The funding by a WWF-Novozymes grant 2008 to DJ National Academy of Sciences, 104, 17430–17434.
and an Ecofondo grant nr. 034-ECO8-inv1 to OD is Death R.G. (2002) Predicting invertebrate diversity from
greatly appreciated. We thank Narcis Prat for help disturbance regimes in forest streams. Oikos, 97, 18–30.
with the Chironomid identifications and José Delgado Death R.G. & Winterbourn M.J. (1995) Diversity patterns
in streams benthic invertebrate communities: the
from the Hacienda Antisana for access to the study
influence of habitat stability. Ecology, 76, 1446–1460.
area.
Epler J.H. (2001) Identification manual for the larval Chiro-
nomidae (Diptera) of North and South Carolina. A guide to
the taxonomy of the midges of the southeastern United
References
States, including Florida. Special Publication SJ2001-SP13.
Austin M.P. (1980) Searching for a model for use in North Carolina Department of Environment and Nat-
vegetation analysis. Vegetatio, 16, 11–21. ural Resources, Raleigh, NC, and St. Johns River Water
Biggs B.J.F. (1996) Patterns in benthic algae of streams. In: Management District, Palatka, Florida.
Algal Ecology (Eds R.J. Stevenson, M.L. Bothwell & R.L. Fernández H.R. & Domı́nguez E. (Eds). (2001) Guı́a para
Lowe), pp. 31–56. Academics Press, San Diego. la determinación de los artrópodos bentónicos sudamerica-
Biggs B.J.F., Goring D.G. & Nikora V.I. (1998) Subsidy nos. Editorial Universitaria de Tucumán, Tucumán.
and stress responses of stream periphyton to gradients Flory E.A. & Milner A.M. (1999) The role of competition
in water velocity as a function of community growth in invertebrate community development in a recently
form. Journal of Phycology, 34, 598–607. formed stream in Glacier Bay National Park, Alaska.
Bradley R.S., Vuille M., Diaz H.F. & Vergera W. (2006) Aquatic Ecology, 33, 175–184.
Threats to water supplies in the tropical Andes. Science, Francou B. & Coudrain A. (2005) Glacier shrinkage and
312, 1755–1756. water resources in the Andes. Eos, 86, 415.
Brittain J.E., Saltveit S.J., Castella E., Bogen J., Bønsnes Füreder L., Schütz C., Wallinger M. & Burger R. (2001)
T.E., Blakar I. et al. (2001) The macroinvertebrate Physico-chemistry and aquatic insects of a glacier-fed
communities of two contrasting Norwegian glacial and a spring-fed alpine stream. Freshwater Biology, 46,
rivers in relation to environmental variables. Freshwater 1673–1689.
Biology, 46, 1723–1736. Giller P.S. & Malmqvist B. (1998) The Biology of Streams
Brown L.E., Hannah D.M. & Milner A.M. (2007) Vulner- and Rivers. Oxford University Press, Oxford.
ability of alpine stream biodiversity to shrinking Gı́slason G., Adalsteinsson H., Hansen I., Òlafsson J.S. &
glaciers and snow packs. Global Change Biology, 13, Svavarsdóttir K. (2001) Longitudinal changes in macr-
958–966. oinvertebrate assemblages along a glacial river system
Brundin L. (1966) Transantarctic relationships and their in central Iceland. Freshwater Biology, 46, 1737–1751.
significance, as evidenced by chironomid midges with Graham A.A., McCaughan D.J. & McKee F.E. (1988)
a monograph of the subfamilies Podonominae and Measurement of surface area of stones. Hydrobiologia,
Aphroteniinae and the Austral Heptagyiae. Kungligi 157, 85–87.
Svenska Vetenskapsakademiens Hanlinger Fjärde Serien, Gucht van der K., Vandekerckhove T. & Vloemans N.
11, 472. (2005) Characterization of bacterial communities in
Cadbury S.L., Milner A.L. & Hannah D.M. (2010) four freshwater lakes differing in nutrient load and
Hydroecology of a New Zealand glacier-fed river: food web structure. Microbiological Ecology, 53, 205–220.
Max. pH
CV temp. (%)
CV depth (%)
Stability index
Substrate types
Catchment ðkm2 Þ
Altitude (m.a.s.l.)
Mean current ðm s1 Þ
Appendix 2
Complete taxon list for the five Surber samples from each of three sites in three glacial streams on Mt Antisana in Ecuador.
Stream 15 14 12
Site 1 2 3 1 2 3 1 2 3
Surber nr. 1 2 3 4 5 1 2 3 4 5 1 2 3 4 5 1 2 3 4 5 1 2 3 4 5 1 2 3 4 5 1 2 3 4 5 1 2 3 4 5 1 2 3 4 5
Family Taxa
Nematoda – sp1 1 1
– sp2 2 3 8 2 1 3 1 2 9 15 3
– sp3 2 2
Oligochaeta Naididae sp1 9 2 20 3 5 4 1 3 3 8 3 3 7 10 9 7 20 5 2 3 5 3 1
Coleoptera Elmidae Neoelmis sp1 3 2 10 3 1
Elmidae Neoelmis sp2 1 5 11 2 10 2 1 2 2
Scirtidae Prionocyphon sp1 1 1 2 2
Dytiscidae sp1 1
Staphylinidae sp1 1 1
Staphylinidae sp2 1
Ephemeroptera Baetidae Andesiops sp 1 3 10 1 26 16 2 18 10 3 12 15 4 14
Plecoptera Gripopterygidae Claudioperla sp 1 1
Trichoptera Hydroptilidae Ochrotrichia sp 1 10 3 3 2 3 1
Hydrobiosidae Atopsyche sp1 1 1 4 1 3 2 2
Hydrobiosidae Cailloma sp1 1 1 1 1 1 2 1 1 1 5 4 1 1 2
Stream 15 14 12
Site 1 2 3 1 2 3 1 2 3
Surber nr. 1 2 3 4 5 1 2 3 4 5 1 2 3 4 5 1 2 3 4 5 1 2 3 4 5 1 2 3 4 5 1 2 3 4 5 1 2 3 4 5 1 2 3 4 5
Family Taxa