Climate change is poised to alter mountain stream ecosystem processes via organismal phenological shifts

Significance In mountain regions globally, climate change is reducing snowpack, advancing snowmelt, and altering environmental regimes of rivers born in these elevations. Here, we conducted an experiment simulating end-of-century vs. current flow regimes in Sierra Nevada mountain streams to examine impending shifts in biodiversity and ecosystem processes. Early snowmelt destabilized stream epilithic biofilm metabolism and altered key ecosystem functions such as insect production and emergence, via shifts in community composition, structure, and phenology (i.e., timing of development). Notably, some processes showed sensitivity to climate change on fine timescales, with implications for predator–prey synchrony. As climate continues to change quickly in high-altitude mountain ecosystems, the resilience of stream ecosystem functions may hinge on the presence of diverse ecological communities.


Figure S1. Diagram and photos of the outdoor channel array at the Sierra Nevada Aquatic
Research Laboratory (SNARL) in Mammoth Lakes, California, USA.(A) Channels were assigned to one of three treatments with three replicate channels each in a block design.Channel number is shown to the left of the channel inlet and colored according to treatment.The treatments were: (1) current hydrologic conditions based on the historic (long-term) hydrograph at Convict Creek with a flow regime that reaches baseflow conditions around August 3rd (i.e., Current treatment); (2) hydrologic conditions under a mitigated climate change scenario, where the stream would return to baseflow conditions three weeks earlier than it currently does (i.e., 3week treatment); and (3) hydrologic conditions under unmitigated climate change, where the stream would return to baseflow six weeks earlier than it currently does (i.e., 6-week treatment).(B) Aerial photo of the artificial channels, courtesy of Google Maps.(C) Photos of high flow (left; ~15 L/s) and low flow (i.e., baseflow) conditions (right; ~1.5 L/s).Photo credits: Carol Blanchette (B), Guillermo de Mendoza (C1), and Ludmila Sromek (C2). Figure S2.Historical flow regime of Convict Creek.Long-term seasonal trends in mean daily discharge in Convict Creek from March to November each year from 1960 to 1974 (in dashed green lines), as measured by the available data from U.S. Geological Survey streamgage 10265200.We averaged the 15 years of streamflow data to create a daily mean discharge value (black line).We then used this average flow regime as the basis for the "Current" flow regime and advanced the start of summer low flows for the 3-week and 6-week treatments from this baseline, in agreement with downscaled climate change projections from Reich et al. 2018 (1; Figure S3).The first-order streams that our experimental channels are simulating are smaller than Convict Creek and located at a higher altitude.They thus experience faster return to summer baseflow conditions.The boxplots display channel-specific medians (bold lines), interquartile ranges (boxes), and interquartile range*1.5 extents (whiskers).Hypoxia, defined as dissolved oxygen concentrations dropping below 2 mg/L (2), was never detected in the experiment, and almost all readings (99.6% of them) were over 5 mg/L, a threshold often used to assess long-term performance of sensitive, coldwater mayflies and stoneflies (3 and references therein).The "Feeder" label refers to measurements taken at the two inlets connecting Convict Creek to the experimental channels.Figure S7.Seasonal and time-varying estimates of epilithic biofilm production (GPP) and respiration (ER).(A) No differences were observed between treatments for cumulative seasonal GPP.(B) Cumulative seasonal ER was significantly higher in magnitude for the 6-week treatment relative to the other two treatments (indicated by *).(C) No predicted responses were observed for GPP.(D) An immediate increase in the magnitude of ER occurred in the 6-week treatment based on comparisons between the middle and start of the experiment.Shaded error ribbons and error bars represent ±1 standard error.Each breakpoint in the time series plots denotes a sampling event.The three potential response types (immediate treatment effect, delayed treatment effect, and seasonal effect) are listed, and colored in black when supported (see conceptual framework in Fig. 1, and Table S1 for how statistical tests connect with each response type).Earlier, extended low flows had a delayed effect on benthic stream invertebrate community composition and an immediate effect on emergent insect community composition.Black arrows represent significant post-hoc differences due to predicted responses, where arrow end points indicate ellipse centroids.(A) The benthic stream invertebrate community experienced a delayed response, based on differences between the 6-week treatment at the end and the middle of the experiment.(B) The community composition of emergent stream invertebrates changed immediately in the middle period between the 6-week treatment and the Current treatment.There was a delayed change in the community in the 3-week treatment.The middle period in this study begins ten days preceding the onset of summer low flow for the 6-week treatment, to reflect that falling discharge can cause effects before minimum summer low flow is reached.The three potential response types (immediate treatment effect, delayed treatment effect, seasonal effect) are listed, and colored black when supported (see conceptual framework in Fig. 1, and Table S1 for how statistical tests connect with each response type).We measured response diversity using dissimilarity in how species abundance responded to change in discharge, via the method described in Ross et al. 2023 (4).We note dissimilarity has a minimum value of 1, and higher values indicate higher response diversity.The 15 most abundant benthic taxa were included in our analyses, and response diversity was calculated for each channel independently.The boxplot displays the median (bold red line), interquartile range (box), and interquartile range*1.5 (whiskers) of all channel response diversity values.Samples from the first sampling date were excluded for this analysis to focus on the experimental response.The dashed horizontal lines are median dissimilarity values used as benchmarks, reported to have low and high dissimilarity in Ross et al. 2023.S1.Description of how statistical tests support evidence for each of the potential ecological responses described in Figure 1 (no effect, seasonal effect, immediate treatment effect, delayed treatment effect).Responses reflect a change in magnitude or phenology of the response variable.A seasonal effect is inferred when an advanced low flow treatment and the Current conditions treatment exhibit a shift in the same direction and magnitude over time.An example of a seasonal effect is Chironominae emergence (Figure 3; Table S13), where all three treatments increase in emergence in the middle period compared to the start period.An immediate effect occurs when there is a difference in how treatments change from the start to the middle period, or when a difference occurs between treatments within the middle period (i.e., at the onset of treatment differences).An immediate effect is seen in Chironominae, where emergence under the 6-week treatment is significantly different from emergence under the Current treatment (middle period).Lastly, a delayed effect occurs when there is a difference in how treatments change from the middle to end period, or between treatments within the end period.An example of a delayed effect occurred in benthic Chironomini (Figure 3), where abundance increased significantly between the middle and end periods for the 6-week treatment, but not for the Current treatment.Immediate and delayed effects could theoretically occur sequentially in the same response variable, but we did not expect it a priori, and it did not occur in our study.

Response type
Relevant  S3.Pairwise comparisons of dissolved oxygen among period-treatment groups.We examined dissolved oxygen over time via post-hoc pairwise comparisons (P-values) corrected by the Benjamini-Hochberg method.A bolded P-value indicates that the pairwise comparison was significant for that response type.If the Current and 3-week or 6-week treatments are significant for the same comparison across periods (e.g., start vs. middle) then P-values are bolded for seasonal effects but not for immediate or delayed treatment effects, as that would indicate similar change over time regardless of treatment.See conceptual framework on response types in Fig. 1 S5.Pairwise comparisons of biofilm respiration among period-treatment groups.We examined biofilm respiration over time via post-hoc pairwise comparisons (P-values) corrected by the Benjamini-Hochberg method.A bolded P-value indicates the pairwise comparison was significant for that response type.If the Current and 3-week or 6-week treatments are significant for the same comparison across periods (e.g., start vs. middle) then P-values are bolded for seasonal effects but not for immediate or delayed treatment effects, as that would indicate similar change over time regardless of treatment.See conceptual framework on response types in Fig. 1, and Table S1 S6.Effects of period-treatment on the abundance of benthic taxa that significantly explained community dissimilarity in Figure S8.Significant responses are assigned to seasonal and to treatment effects (immediate or delayed).Asterisks indicate P < 0.05 for periodtreatment.Short horizontal dashes (~) indicate P < 0.10.A positive sign (+) or negative sign (-) indicates if advanced low flow treatment had a positive or negative effect, respectively, on taxa abundance, provided that treatment differences explained post-hoc differences.A combination of positive and negative signs (±) indicates that low flow treatment had positive and negative effects at different times.See conceptual framework on response types in Fig. 1, and Table S1 for how statistical tests connect with each response type.S8.Pairwise comparisons of benthic Hydroptila abundance among period-treatment groups.Benthic Hydroptila abundance post-hoc pairwise comparisons (P-values) corrected by the Benjamini-Hochberg method.A bolded P-value indicates the pairwise comparison was significant for that response type.If the Current and 3-week or 6-week treatments are significant for the same comparison across periods (e.g., start vs. middle) then P-values are bolded for seasonal effects but not for immediate or delayed treatment effects, as that would indicate similar change over time regardless of treatment.See conceptual framework on response types in Fig. 1, and Table S1 S9.Pairwise comparisons of benthic Simuliidae abundance among period-treatment groups.Benthic Simuliidae abundance post-hoc pairwise comparisons (P-values) corrected by the Benjamini-Hochberg method.A bolded P-value indicates the pairwise comparison was significant for that response type.If the Current and 3-week or 6-week treatments are significant for the same comparison across periods (e.g., start vs. middle) then P-values are bolded for seasonal effects but not for immediate or delayed treatment effects, as that would indicate similar change over time regardless of treatment.See conceptual framework on response types in Fig. 1, and Table S1 S11.Pairwise comparisons of benthic stream invertebrate NMDS community composition among period-treatment groups.A PERMANOVA test was first conducted with period-treatment as a fixed effect and it was found to significantly explain benthic community composition (pseudo-F5,35 = 2.571, P < 0.001).Benthic stream invertebrate NMDS community post-hoc pairwise comparisons (P-values) corrected by the Benjamini-Hochberg method.A bolded P-value indicates the pairwise comparison was significant for that response type.If the Current and 3-week or 6-week treatments are significant for the same comparison across periods (e.g., start vs. middle) then P-values are bolded for seasonal effects but not for immediate or delayed treatment effects, as that would indicate similar change over time regardless of treatment.See conceptual framework on response types in Fig. 1, and Table S1 S12.The effects of period-treatment on the abundance of emergent taxa that significantly explain community dissimilarity (P ≤ 0.002) in Figure S8.Significant responses are determined to be due to seasonal or treatment effects.Asterisks indicate P < 0.05 for periodtreatment.A positive sign (+) indicates that low flow treatment had a positive effect on taxa abundance, provided that treatment differences explained post-hoc differences.See conceptual framework on response types in Fig. 1, and Table S1 S14.Pairwise comparisons of emergent stream invertebrate NMDS community composition among period-treatment groups.A PERMANOVA test was first conducted with period-treatment as a fixed effect and it was found to significantly explain emergent community composition (pseudo-F8,89 = 5.7277, P < 0.001).Emergent stream invertebrate NMDS community post-hoc pairwise comparisons (P-values) corrected by the Benjamini-Hochberg method.A bolded P-value indicates the pairwise comparison was significant for that response type.If the Current and 3-week or 6-week treatments are significant for the same comparison across periods (e.g., start vs. middle) then P-values are bolded for seasonal effects but not for immediate or delayed treatment effects, as that would indicate similar change over time regardless of treatment.See conceptual framework on response types in Fig. 1, and Table S1 S15 Assessment of potential Brewer's Blackbird effects on benthic and emergent macroinvertebrate communities, using a Before-After-Control-Impact (BACI) design.We tested if benthic community composition and structure changed immediately after blackbird arrival in channels that were visited relative to channels that were not visited.The interaction term of treatment*time (time being binary: pre vs post-bird presence) was not statistically significant when examining either benthic or emergent invertebrate abundance, richness, or composition, denoting no significant bird effects.Degrees of freedom (df) lists the df for the treatment*time interaction first, then the df of residuals.

Response
Figure S3.Average recorded and targeted daily discharge for each treatment throughout the experiment.(A) The recorded discharge was calculated using high frequency water depth sensors and rating curves in each channel.Discharge was manually controlled by opening or closing the sluice gate at the inlet of each channel to match the target flow regime.(B) Target discharge in the Current treatment follows the historic average flow regime timing of snowmelt recession in the water source, Convict Creek (see Fig. S3).The other two treatments were designed to experience summer low flow three and six weeks earlier in the year, respectively, following downscaled climate change projections (see methods for details).The designed onset of summer low-flow conditions is indicated for each treatment by a dashed vertical line colored by treatment.The shaded area represents ±1 standard error.
Figure S5.Mean, minimum, and diel range of water temperature for each treatment throughout the experiment.Advanced low flow treatments affected all water temperature metrics immediately.The designed onset of summer low-flow conditions is indicated for each treatment by a dashed vertical line colored by treatment.The means of each water temperature metric (± standard error) are shown by treatment.
Figure S8.Low-flow treatment effects on stream invertebrate community structure.Earlier, extended low flows had a delayed effect on benthic stream invertebrate community composition and an immediate effect on emergent insect community composition.Black arrows represent significant post-hoc differences due to predicted responses, where arrow end points indicate ellipse centroids.(A) The benthic stream invertebrate community experienced a delayed response, based on differences between the 6-week treatment at the end and the middle of the experiment.(B) The community composition of emergent stream invertebrates changed immediately in the middle period between the 6-week treatment and the Current treatment.There was a delayed change in the community in the 3-week treatment.The middle period in this study begins ten days preceding the onset of summer low flow for the 6-week treatment, to reflect that falling discharge can cause effects before minimum summer low flow is reached.The three potential response types (immediate treatment effect, delayed treatment effect, seasonal effect) are listed, and colored black when supported (see conceptual framework in Fig.1, and TableS1for how statistical tests connect with each response type).
Figure S9.Response diversity in our study relative to values in the literature.We measured response diversity using dissimilarity in how species abundance responded to change in discharge, via the method described in Ross et al. 2023(4).We note dissimilarity has a minimum value of 1, and higher values indicate higher response diversity.The 15 most abundant benthic taxa were included in our analyses, and response diversity was calculated for each channel independently.The boxplot displays the median (bold red line), interquartile range (box), and interquartile range*1.5 (whiskers) of all channel response diversity values.Samples from the first sampling date were excluded for this analysis to focus on the experimental response.The dashed horizontal lines are median dissimilarity values used as benchmarks, reported to have low and high dissimilarity in Ross et al. 2023.
Figure S10.Taxa driving community dissimilarity in Figure S8.The NMDS space is the same here as that in Figure S8.(A) Benthic taxa that significantly explained community dissimilarity among samples are shown (P ≤ 0.002).The position of the taxa label indicates where the taxa is in the NMDS space.The label position for Drunella flavilinea is moved to prevent overlap and the line connected to it indicates the actual space it corresponds to in the NMDS.(B) Emergent taxa that significantly (P ≤ 0.002) explained community dissimilarity among samples.

Table S2 . Pairwise comparisons of maximum water temperature among period-treatment groups.
We examined maximum daily temperature via post-hoc pairwise comparisons (P-values) corrected by the Benjamini-Hochberg method.A bolded P-value indicates that the pairwise comparison was significant for that response type.If the Current and 3-week or 6-week treatments are significant for the same comparison across periods (e.g., start vs. middle) then P-values are bolded for seasonal effects but not for immediate or delayed treatment effects, as that would indicate similar change over time regardless of treatment.See conceptual framework on response types in Fig.1, and TableS1for how statistical tests connect with each response type.
comparisons Rationale Seasonal effect Middle 6-week vs. Start 6-week Middle Current vs. Start Current Middle 3-week vs. Start 3-week End 6-week vs. Middle 6-week End Current vs. Middle Current End 3-week vs. Middle 3-week If the Current treatment and either the 6-week or 3-week treatment exhibit the same significant response over time, then we describe a seasonal effect.

Table S4 . Pairwise comparisons of epilithic biofilm GPP:ER ratios among period-treatment groups.
, and TableS1for how statistical tests connect with each response type.We examined GPP:ER ratios over time via post-hoc pairwise comparisons (P-values) corrected by the Benjamini-Hochberg method.A bolded P-value indicates the pairwise comparison was significant for that response type.If the Current and 3-week or 6-week treatments are significant for the same comparison across periods (e.g., start vs. middle), then P-values are bolded for seasonal effects but not for immediate or delayed treatment effects, as that would indicate similar change over time regardless of treatment.See conceptual framework on response types in Fig.1, and TableS1for how statistical tests connect with each response type.
for how statistical tests connect with each response type.

Table S7 . Pairwise comparisons of benthic Chironomini abundance among period- treatment groups.
We examined benthic Chironomini abundance via post-hoc pairwise comparisons (P-values) corrected by the Benjamini-Hochberg method.A bolded P-value indicates the pairwise comparison was significant for that response type.If the Current and 3-week or 6-week treatments are significant for the same comparison across periods (e.g., start vs. middle) then P-values are bolded for seasonal effects but not for immediate or delayed treatment effects, as that would indicate similar change over time regardless of treatment.See conceptual framework on response types in Fig.1, and TableS1for how statistical tests connect with each response type.
for how statistical tests connect with each response type.

Table S10 . Pairwise comparisons of benthic Micrasema abundance among period-treatment groups.
for how statistical tests connect with each response type.Benthic Micrasema abundance post-hoc pairwise comparisons (P-values) corrected by the Benjamini-Hochberg method.A bolded P-value indicates the pairwise comparison was significant for that response type.If the Current and 3-week or 6-week treatments are significant for the same comparison across periods (e.g., start vs. middle) then P-values are bolded for seasonal effects but not for immediate or delayed treatment effects, as that would indicate similar change over time regardless of treatment.See conceptual framework on response types in Fig.1, and TableS1for how statistical tests connect with each response type.
for how statistical tests connect with each response type.

Table S13 . Pairwise comparisons of emergent Chironominae abundance among period- treatment groups.
for how statistical tests connect with each response type.Emergent Chironominae abundance post-hoc pairwise comparisons (Pvalues) corrected by the Benjamini-Hochberg method.A bolded P-value indicates the pairwise comparison was significant for that response type.If the Current and 3-week or 6-week treatments are significant for the same comparison across periods (e.g., start vs. middle), then P-values are bolded for seasonal effects but not for immediate or delayed treatment effects, as that would indicate similar change over time regardless of treatment.See conceptual framework on response types in Fig.1, and TableS1for how statistical tests connect with each response type.
for how statistical tests connect with each response type.Start 6-week Middle Current vs. Start Current Middle 3-week vs. Start 3-week End 6-week vs. Middle 6-week End Current vs. Middle Current End 3-week vs. Middle 3-week Start 6-week Middle Current vs. Start Current Middle 3-week vs. Start 3-week Middle 6-week vs. Middle Current Middle 3-week vs. Middle Current