- In stream ecology, we know pretty well about how flow regime can structure
ecological communities. The effects of flow variation on ecosystem functioning have
been widely investigated in streams which are drought-prone or affected by water
abstraction in environmental flow studies – but not so much in perennial streams like
this.
E.g. we know qualitatively how discharge litter breakdown (shredder feeding,
microbial activity, physical fragmentation); but when there are more peak flows -
uncertain
- Given LB as the choice of ecological functioning that reflects stream integrity in
many studies and biomonitoring work. We ask in streams with high variability in
discharge and perhaps LB, is it still a reliable, robust bioindicator of human
disturbances? We tend to rely on a popular bioassessment framework originally
developed for Swedish streams, the range of natural variation of k within a given
stream is about 75-133% of the mean. Should we assume that range to remain
largely similar in other temperate streams, where temporal flow variations are
considerably larger?
It is important because when a disturbance occurs, we examine whether k at the
greatest impact of disturbance till a certain time after is inside or outside this range
in orange. Conclude the effects of disturbance to be significant.
- Our study is motivated by the concern that in regions with high variability in flows
across time, we might be less able to detect human influences on ecosystem
functioning. We empirically quantified the variability of flow-k relationships and also
the natural range of k, both within streams and across streams in nearby catchments,
which can help adjust reference conditions in bioassessment
- LB, k, we measured refers to the exponential decay coefficient of litter mass loss. It
is temperature and day corrected to account for the differences in thermal
regime/study duration among sites. We measured total k using coarse-mesh bags,
and microbial-mediated k using fine-mesh bags which excluded most
macroinvertebrates.
- We pick small forested streams in 3 geographic regions in temperate Canada. One
near Vancouver, not so far from Victoria, and two regions in central Ontario. We
repeated our sampling for 3-4 years with substantial year-to-year discharge
differences. The litterbags were incubated for about a month during early autumn,
coinciding with the litterfall period. The timing is therefore ecologically relevant to
stream consumers relying on litter inputs as basal resources. We computed
hydrologic indices based on discharge data, reflecting for instance the frequency and
duration of flow events, flashiness and magnitude. We also measured variables
related to shredder assemblages, fungal biomass, and water chemistry (focus on
hydrology and k).
Before getting to our hypotheses, we used total k and microbial-mediated k to derive
fragmentation rate FR, and dissolution and microbial decomposition rate DMDR,
based on a 2017 modelling study. FR refers to the sum of shredder feeding and
physical abrasion rate; FR and DMDR add up to the total k. We first hypothesize that
FR is positively associated with hydraulic force, which increases with say the
frequency of high flows. Conversely, DMDR changes little along this hydraulic
gradient. We further hypothesized that, given the expected hydraulic influences on
FR, the variability of FR across years would exceed that of DMDR within individual
streams.
To represent the hydraulic force affecting k, we establish a composite hydrologic
index, which integrates numerous hydrologic variables strongly associated separately
with the LB rates, total k, FR, and DMDR. We achieved this using forward selection
and RDA.
We then tested the significance of relationships between the litter breakdown rates
and composite hydrologic index for each study region, controlling for the random
effects of sites (LME). We also examined the extent of variation of k, including its
inter-year variation within each particular stream, and inter-site variation among
streams within a region.
During the study period, we found discharge to vary substantially among years. This
natural variation driven by weather did not involve streambed drying, nor signs of
catastrophic debris flows. The hydrograph here displayed flow variation of a typical
stream. We can notice the difference in the magnitude of high flows, their duration
and timing, and flow variability, etc.
For the results of flow-k relationships, considering all regions together, hydrologic
variables both before and during litterbag incubation could significantly affect total k
and DMDR, but surprisingly they had no effects on FR. However, when we examine
individual regions, hydrologic variability became unrelated to total k and DMDR. We
think that the effects of flow variability on k were likely contributed by regional
differences, rather than temporal differences within regions. Here, plot log k against
composite hydrologic index for one study region. Data points from each year are
represented by a different colour. You can see the index of Y4 different from the rest,
and Y1 in black also different from Y3 in blue. Despite this hydrologic variation, we
see no clear pattern of changes in k from results of LME.
Natural range of k: for individual streams, we found that the year-to-year variability
of k was similar among LB types, and also similar among study regions overall. The
magnitude of inter-site variability of FR, or its sd is 75% of the mean, much greater
than that of total k and DMDR (around 20-30% of the mean). This means it is harder
for the effects of human disturbance to exceed the reference condtion for FR to be
considered significant. Total k and DMDR are therefore more sensitive indicators.
Inter-site variability seemed to be different between regions, benchmarks developed
for one region may not necessarily work well in another.
Referring to the hypotheses on flow-ecology relationships, hydrologic conditions had
minimal effects on k across years, therefore rejecting the first one. The extent of
variability of FR across years approximated DMDR, rejecting the second one.
Hydrologic variability might have interacted with some fine-scale habitat attributes
which we didn’t measure, to produce complex hydraulic effects on shredder feeding,
physical abrasion. We may say hydrologic indices on the site level are poor surrogates
of the hydraulic conditions these leaf bags experience. How does this add to the
literature of flow effects on k? while severe flow reduction and droughts can highly
suppress k, our study highlights the unclear effects of natural flow variation in
perennial streams on k in the intermediate range of discharge.
Nonetheless, the natural range of k we found could help to refine site and region-
specific benchmarks for stream bioassessment. For instance, the inter-year variation
of k is useful when we have data before and after disturbance in a particular stream.
The range of k is expressed by the % of the inter-year mean, so no signs of
disturbance when k is reduced less than 40%, increases less than 35% for BC1. This
range is wider than the one recommended in the framework developed for Swedish
streams. And if we sample multiple impacted and control streams after a
disturbance, we can use the inter-site mean of the region. Similarly, we see that the
Swedish framework is overconservative and can lead to inaccurate conclusions of
human disturbance effects.
In summary, this study shows that k could vary substantially in natural streams across
years in temperate zone, but its variation is not well predicted by that of discharge.
Developing the relationships between flow regime and k may require finer-scale
measurements of hydraulic conditions. In terms of bioassessment, total k may not be
sensitive enough as the effect size of some disturbances are similar to its range of
natural variation. We may not assume a one-size-fit-all framework and need to
regionally adjust reference conditions of total k. Last but not least, the range of
reference conditions I show are not going to be static when patterns of hydrologic
extremes and leaf input timing are shifting. In a changing climate, I may come back a
few years later to present a very different set of numbers.
