Comparing Upstream and Downstream sites based on stream, fish, and invertebrate variables
Mundahl & Mundahl (2022), Ecological Processes
Compare upstream vs downstream sites using NMDS ordination
Focus:
Stream habitat
Fish species abundance
Invertebrates species abundance
Stream: 18 variables (% , averages)
Fish: 19 species counts
Invertebrates: 39 species counts
# A tibble: 6 × 6
Site `Stream Width` `Mean Depth` `Mean CV` `% Fines` Embeddedness
<dbl> <dbl> <dbl> <dbl> <dbl> <dbl>
1 1 2.6 26 14 46 5
2 2 2.3 45 6 67 5
3 3 4 41 7 58 4
4 4 5.2 48 2 68 5
5 5 4.3 47 14 87 5
6 6 4.1 24 29 56 5
Data Setup
Grouped sites into Upstream (1–12) and Downstream (13–22)
Split data into
Ordination
Ran NMDS for each domain using appropriate distance metrics:
Gower for stream (% and averages)
Bray-Curtis for fish and invertebrate abundance
All stress values < 0.2 → valid ordinations
Used envfit to regress variables onto NMDS space.
Filtered meaningful variables(p_val< 0.05).
Filtered arrows for plotting based on R² thresholds:
Stream > 0.5
Fish > 0.4
Invertebrate > 0.3
R² thresholds were somewhat arbitrary but kept to retain variables with ecological meaning and interpretive value.
Upstream and Downstream sites are clearly separated.
Ellipses show minimal overlap, confirming strong differentiation.
Key drivers: Stream Width, Buffer Width, % Fines, Embeddedness.
Distinct clustering of Upstream vs Downstream sites.
No ellipse overlap highlights strong ecological separation.
Influential species include Brown Trout, Blacknose Dace, Creek Chub, and Slimy Sculpin.
Upstream and Downstream sites show limited clustering.
Significant ellipse overlap suggests subtle ecological separation.
Key indicator taxa include Baetis, Simulium, Hydropsyche, and Caenis.
Focuses on NMDS-important variables only
Use Wilcoxon tests to find key variables that show significant differences between Upstream and Downstream sites.
| Variable | p_value |
|---|---|
| Stream Width | 0.0000865 |
| % Fines | 0.0013709 |
| Embeddedness | 0.0097236 |
| Buffer Width | 0.0035998 |
| Variable | p_value |
|---|---|
| Creek Chub | 0.0000770 |
| Central Stoneroller | 0.0001224 |
| Brown Trout | 0.0001764 |
| Slimy Sculpin | 0.0075564 |
| Variable | p_value |
|---|---|
| Caenis | 0.0024967 |
| Baetis | 0.0008543 |
| Stenonema | 0.0370432 |
| Hydropsyche | 0.0091010 |
| Cheumatopsyche | 0.0040332 |
| Simulium | 0.0001354 |
% Fines and Embeddedness are high at Upstream sites and lower at Downstream.
Downstream sites feature wider buffers and stream channels.
Brown Trout and Slimy Sculpin are more abundant at Downstream sites.
Creek Chub is more abundant at Upstream sites.
Baetis, Simulium, and Hydropsyche are more abundant at Downstream sites.
Stenonema, Cheumatopsyche, and Caenis are more abundant at Upstream sites.
Yes, based on NMDS biplots upstream and downstream sites differ with respect to the variables measured.
Fish variables show the least overlap between upstream and downstream.
Downstream: Higher levels of % Fines and Embeddedness (stream), Brown Trout and Slimy Sculpin (fish), and Baetis, Simulium, Hydropsyche (invertebrates).
Upstream: Wider buffers and stream channels, more Creek Chub and Stenonema (fish), and more Cheumatopsyche and Caenis (invertebrates).