Eyes on the Eel

Eyes on the Eel: Synoptic summer surveys of food web biomass UCB participants: Gabe Rossi, Phil Georgakakos, Suzanne Kelson, Keith Bouma-Gregson, Mary Power, Stephanie Carlson, David Dralle, George Greer, Sally Thompson, Collin Bode, Sarah Kupferberg

Collaborators and advisors: Peggy Wilzbach, Bill Trush, Alison O’Dowd (Humboldt State University), Bret Harvey (USFS); Pat Higgins, David Sopjes, (Eel River Recovery Project); Darren Mireau (CalTrout)


We carry out surveys of selected tributary-mainstem confluences during summer baseflow to document how key species and functional groups are responding to hydrologic conditions during the previous winter and the current summer. We hope to complement and combine data with other groups with similar goal in the Eel and in other rivers of Northern California. We are making use of bathymetric LiDar flown by the Eel River CZO and NCALM in July 2014, and building on 25 years of food web research carried out within the Angelo Reserve, located in the upper South Fork Eel watershed. By estimating the biomass of key photosynthetic primary producers, invertebrates, and fish and other aquatic vertebrates at various tributary-mainstem confluence sites during the biologically active summer season, we address several hypotheses and questions listed below. Collaborators should feel free to suggest additional questions and improved field methods by emailing us at the Contact site, and using the subject header “Eyes on the Eel”.

Background hypothesis I: Effect of annual hydrology. During the summer low flow season, the algal-based food web of the Eel River may assemble into one of three alternative states (Power et al. 2014). In ‘salmon’ state, rock substrates and early summer blooms of macroalgal (Cladophora glomerata) become covered with highly edible, fat-rich diatoms, which in turn support high primary consumer (invertebrates and tadpoles that eat living or dead plant tissue) productivity. Diatom energy, nutrients, and lipids including PUFAs are transferred up the food chain to support rearing juvenile salmonids and other aquatic predators. In Dicosmoecus state, large macroalgal blooms do not proliferate, and epilithic diatoms are sequestered by predator resistant (armored or sessile) grazers. Salmon state generally occurs after winters with one or more scouring floods, which reduce defended, slower-growing grazers; Dicosmoecus state when scouring winter floods have not occurred. In recent years, a third summer food web state, Cyanobacteria state, has emerged. Some potentially neurotoxic cyanobacteria (often dominated by Anabaena or its relatives) occur where low summer base flow has disconnected and greatly warmed sunlit mainstem or side pools. These taxa overgrow the edible diatom epiphytes and their green macroalgal hosts. Other neurotoxic cyanobacteria (Phormidium spp, Bouma-Gregson unpublished) have been observed attached to rocks in mainstem riffles where flows remain relatively strong. Learning more about how environmental conditions, biotic and abiotic, steer the river food web to one of these state or another over the course of summer algal succession would guide our attempts to forecast and manage summer food web structure in the Eel.

Background hypothesis II: Food web assembly is controlled less by energy loading and more by disturbance regime as one moves downstream into habitats with increasing sunlight and bedload mobility (decreasing substrate stability). See Finlay et al. 2010; Power, Kupferberg, Cooper and Deas, 2015.

Background hypothesis III: Summertime water extraction can potentially tip the Eel food web from salmon-supportin to cyanobacterialy degraded . See Bauer et al. 2015; Carah et al., 2015, Power et al. 2015.


Surveys should occur roughly synchronously, within the same week. We target mid-June, when production of algae, invertebrates, and growth of fish is typically peaking, and two stressful summer base flow extremes–July, when temperatures are warmest, and September, when discharge is lowest.

All surveys that document, even roughly, the abundance of algae, invertebrates, and fish and other vertebrates at sites, are useful. We call our quickest, simplest approach “Swat Team”. A slightly higher level of effort (“Quantitative Transect (QT)” approach) is what we have put into practice with teams of Berkeley faculty and students, and requires more equipment, time and effort. A third level (“QT with drift, emergence, and key invertebrate, amphibian, and fish counts”) would require two day visits to each site, and a level of effort commensurate with graduate student research. We surveyed Quantitative Transects at four tributary-mainstem confluences in summer 2015, and plan to supplement it with the more intensive third survey approach in a small number of sites in 2016.

Methods are suggested here
Level 1. Casual Team assessments
Level 2.Quantitative Transects (QT)
Level 3. Quantitative Transects with drift, emergence, and key invertebrate, amphibian, and fish counts

Philosophical background: How can we get meaningful data at appropriate scales? First, we want biomass because this is the currency of trophic transfer, but for some organisms, we also should track numbers, the currency of population change. In fish, for example, size distributions will be important indicators of population dynamics. So we’ll need estimates of lengths, and length/mass regressions, for countable organisms.

We try to scale our spatial sampling to natural scales of local habitats and organisms. Ideally, we would sample a reach that is 8 times the length of the estimated local bankfull width (Wb) of the channel, benchmarking reaches that are 8 x (Wb) long. Particles sampled for biofilms and invertebrates should be sampled in a stratified fashion to proportionally represent the relative area of the facies covered by particular substrate types within a survey reach. Within each facies, particles could be selected haphazardly or systematically to represent the D25, D50, and D75 of a site. We should also set drift nets for the mobile, edible invertebrates that are key food organisms for fish, but unlikely to be sampled with netted cobbles. Fish estimates (with snorkeling counts, bankside counts, or camera methods) will require calibration with other methods like electroshocking that will be done for other projects.  Over time, as we refine methods and extend our collaborations, we will need to revisit goals and methods and cross-train and cross-calibrate.  We should try for at least semi-quantitative biomass estimates from our collaborative surveys, and hopefully also keep track of abundance changes when feasible.

For Level 2 Quantitative Surveys, we have selected four tributary mainstem confluences, one in the mainstem Eel at Jordan Creek, and three in the South Fork.  We hope to have some established near the Eel  estuary. While it would be great to have more frequent seasonal sampling, our 2015 focus was on the warmest time of year (mid July?) and the lowest flow (September).  In summer 2016, we will add to these dates a June survey, documenting the summer baseflow food web when it may be most vibrant.

We have found that with a team of 8-10, Level 2 surveys take one day per site, or 4 days per June, July or September sampling effort.

Physical description

Initial set up:

At a given site (mainstem-tributary confluence), four ‘hydraulic units’ (a riffle and a receiving pool on upstream and downstream reaches of the tributary, and above and below the confluence on the mainstem) are benchmarked on the first visit.  For each riffle-pool unit, three permanent x-stream transects are benchmarked at upstream (riffle), downstream (pool), and ~midpoints (pool) of each of the four reaches. We will sketch maps of substrate facies, categorized by substrate type on plan view maps from Google Earth and bathymetric LiDar. In addition, sketches of microhabitat flow estimates in broad intervals (see Appendix A) and overhead (forest, canyon wall) canopy cover (with cameras or spherical densiometer measurements) will be added as overlays on these maps. Laminated copies will be made and used during subsequent surveys so that important changes (e.g. in big wood, algal cover, etc.) can be documented (on transparent film laid over each laminated map).

Using inconspicuous markers (e.g. nails driven into adjacent woody vegetation or into the bed and marked with tree tags), we will mark pre-measured areas in the bed for later bankside visual counts of vertebrates (fish, amphibians, reptiles, mussels, crayfish), as described below. We might also leave markers of known length through the observation quadrats to serve as scales for fish length estimates. Two cameras, iButtons, and drift nets will be installed at appropriate times, and the cameras and nets moved from site to site. (We need to check to see if there are any threats of transferring pathogens or invasive species from downstream to upstream sites, on our equipment or our boots/shoes.)

Then, on each survey of a site:

Transects for documenting physical conditions and macroalgal/macrophyte abundance (Appendix A).

We will run a meter-tape across the three benchmarked 3 x-stream transects.

At ~ 15 or more fixed points along the meter tape (we should discuss whether to make these evenly spaced across the channel as I’ve transitions between substrates or to map boundaries between patches or depth contours as Nancy Grimm’s group has done in the past in AZ), we will record: stream depth, substrate texture (estimated by eye according to Wentworth scale, see Appendix A), stream velocity (possibly just surface, in broad categories that can be estimated by eye, Appendix A), and record attached algae and its condition (details in Appendix A) or deposited litter or detritus; conspicuous macrobiota, and notes on other important observations at the point.

Invertebrate and algal-microbial biofilm samples

Purpose: we want to sample, in a haphazard but stratified fashion, the surficial organisms that would be available, if edible, to epibenthic primary consumers that feed predatory invertebrates or vertebrate consumers like fish. We will not use these surveys to look at infaunal organisms that fish cannot access, for example. (Phil’s point about emergence of infauna feeding fish later needs discussion—if this is a “snap shot” of food web biomass, maybe we should actually measure real time emergence and input to pan traps during the survey…we need to agree on the time scale we intend to represent with our survey visit. The emergence and pan trapping wouldn’t add too much more work, I think, samples are clean.) Similarly, we will look at surface standing crops of epilithic diatoms, cyanobacteria or green algae. The key here is to try to collect samples so that biomass density on substrate area projected to the water surface (of producers or invertebrates) can be estimated, and that larger biomasses can be extrapolated to reach scales using the maps that give proportional areas of particular substrate facies or other microhabitats (riffles vs runs vs pools).

Algae and cyanobacteria will be coarsely estimated in the transects as in the Swat and QT methods above, but we will also sample quantitatively by scraping a defined area (either the sunlit surface of a stone, or adefined area within the stone) with a razor blade (trying to minimize damage to cells). A plastic frame for a 35 mm slide works well as a sampling template. (This samples a square that is 35 x 25 = 875 mm2.) Small flexible rubber gaskets also work. Collected material can be stored in vials or Eppendorf flip top plastic vials, preserved with a drop of 30% formaldehyde if it can’t be refrigerated and microscopically examined within a few days. After 24 h, the formalin can be rinsed (under a hood, or with the person up wind from the chemical, and wearing gloves) from the sample (store and dispose as a toxic waste, or filter and re-use), and the sample inspected under 200x or 400x.

Epifaunal invertebrates. We will use large 1-mm mesh aquarium nets to sample epifaunal invertebrates on coarse (large-pebble-small boulder) substrates, quickly and carefully rolling the stone into the net, carefully brushing off all attached fauna into a white tray of water streamside. then using Peggy’s rapid assessment method to count numbers (and estimated lengths) of the collected individuals, keeping track of taxa at least to order (see Appendix C for identification guide). If wood or leaves are important substrate components, these can be sampled in a similar fashion. Mystery beasts, if common, could be photographed and sent to Peggy Wilzbach, Mike Bogan, or other invertebrate experts. We should sample at least 12? bed particles per site this way, choosing sizes in a stratified fashion according to the proportional cover of the substrate type in the study reach (appendix D gives an example). Area of cobble must be sampled or estimated, perhaps with foil, although I’d prefer a simpler procedure—estimate dimensions (major and minor axes when rock was on the bed) and shape, note these, then estimate habitable area on xl spreadsheet later from surface area formulae for shapes. If bedrock substrate must be sampled in the river, core samplers (old plastic bottles or yoghurt containers work) of known area with gaskets to fit them tightly to the rock can be used to confine attached fauna, which can then be scraped off with fingernails or a small spatula. Appendix D also gives an example of how to adapt a quart yoghurt container for this purpose. ) At dusk (~5 pm-7pm) we can set drift nets. We should discuss how much of the water column we should sample at a reach, particularly at deeper ones. Might be best to sample from the bed to the surface, even if this requires stacking the nets vertically.

It would be good to also set emergence traps to sample both pop-up and crawling emergence, and immersion (pan) traps to catch loading from outside the river.

Fish (and other vertebrate) counts.

We envision using at least four different methods, hopefully getting data to cross calibrate two or three at any given site.

Bankside observations: At shallow sites with sufficient visibility, bankside counts of the numbers of vertebrates in marked quadrats may serve as a local density estimate. After sitting quietly on the shore watching the quadrat for 10-15 minutes (giving fish a chance to recover from the disturbance of human movements), count the numbers and estimate the lengths of vertebrates within the demarked areas. These ‘scan samples’ (Altmann 1974) can be made at 10-15 minute intervals until an adequate averaged count is obtained, bearing in mind that the time of day will affect the spatial distribution of the vertebrates.

Snorkeling surveys: For wider, deeper reaches, we may need need snorkeling counts, for which underwater slates or the white, spiral bound waterproof Rite in the Rain notebooks are handy.

Electroshocking: In our discussion at Berkeley, Stephanie Carlson stressed the need for some cross calibration of these methods (bankside and snorkeling), and offered to let us calibrate bankside observations with the electroshocking counts that she and Suzanne plan for Fox and Elder Creeks within the Angelo Reserve. Calibrating snorkeling with electroshocking would be challenging, because of restriction of our backpack shocking to smaller habitats.

A second method for large habitats might come from camera traps. Gabe Rossi has suggested several underwater camera techniques for counting and measuring sizes of free swimming fish, including baiting them in so they pass near cameras (his notes copied in text below). The literature of folks developing and using these methods suggest that post-processing software can allow us to capture individual frames from video and use them and to accurately measure fish lengths in those frames. There are several software packages available for this:

  • Research software called Vidsync developed by Jason Neuswanger (University of Georgia) which does basically the same thing…”synchronize the left and right videos, play them back in sync, mark them up with symbols to annotate the things you want to measure, and output synced screenshots” This software is for Mac OS X 10.5 or 10.6. These were developed for PhD projects and have been used to measure salmonids. (http://www.chenakings.org/stereo-videogrammetry-software-for-3d-video-analysis/). Mep has been in touch with Jason, and he’s willing to work with us.

If we wanted to go the stereo (or mono) video route, I think I could manufacture camera housings that fit onto rebar or a T-post… but I would need to know what size and shape of Cameras we wanted to use. We’d want it small enough to work on different scales of stream. I don’t think they need to be fancy cameras, but they would need to able to film videos underwater. Do we have two underwater cameras available? I think MP4 videos would be fine, but I would want to check with the software specs to make sure the kind of video we can get would work with the software package we would use.

There are some good papers on using stereo video for estimating fish length see attached Cappo et al. (2007). I think the biggest drawback is not being familiar with the software, but it also seems pretty doable with some basic computer skills. Hardy and Shortis (1998) estimates approximately one hour is required in the laboratory to capture the 32 images and process them into a calibration file. So, given that we’ve not done this before, I’d assume at least 2-3 total hours of post processing, per site, per measurement. Is that too much time and uncertainty?

You can also estimate fish length with a single video, using parallax, where you use the movement of a single fish, between frames of known time difference, to estimate distance from the video camera, and then you use trigonometry to estimate the fish’s length. This wouldn’t require the fancy post-processing software, just a way to freeze frames at known time intervals. One of the major problems is that fish rarely cooperate and swim by the camera at a nice 90 degree angle. There are ways to deal with this as well…some folks have developed equations between vertical parameters (e.g. the depth of a fish at a specific fin) and fish length.

Another, low tech option for a single camera is to install some sort of grid of known distance in the viewframe quadrat and use known distances between the camera and the grid, to estimate fish length. The grid could be established on the first trip, and then simple markers could be re-installed so that subsequent video frames could be rotated into the grid space, when doing analysis. This would probably be the lowest tech video method, but might introduce the most error in length estimates.


We are considering the Fox Creek, Redwood Creek, and Elder Creek confluences in the upper South Fork Eel (6 sites), and are would like suggestions from Keith and Gabe, Peggy and other Arcatans for ~3-4 more pairs of sites (tributary with adjacent mainstem) representing the lower South Fork and the mainstem Eel below the S.Fk confluence. I’m wondering if we could work with Bill Trush and Alison O’Dowd, who are doing some research for FOER, to get them to help us survey trophic level biomass at their site in the upper Mainstem, close to the Van Arsdale dam outlet into the Eel. Also, wonder if Sharon Kramer or Ken Mierzwa would be interested in doing surveys in or near the estuary….

Appendix A. Transect survey methods (MEP)

I repeatedly survey permanent cross-stream transects (“x-scns”) to follow changes in algae, environmental conditions, and associated organisms over time, documenting seasonal and interannual variation. Transects are benchmarked at both ends with nails in trees or bedrock—these can be labeled with aluminum tree tags (Forestry Supply) and should be carefully located with hand-drawn maps, GPS, triangulation (measure distance to spot from two other conspicuous landmarks), etc. Benchmarks should be on ‘permanent’ structures (e.g. large trunks) high enough so that high as well as low flows can be documented (just above bank full is ideal, higher than that and it’s hard to read the tape while walking the streambed). On each survey, stretch a meter tape tightly between the nails. Nail to nail distance should vary less than 1 cm over repeated surveys. Take enough points to obtain about 15-20 measurements for each x-scn. At 0.5 m or 1.0 m intervals along each transect:

  1. note distance along the meter tape (X-strm (m))
  2. measure water depth (cm) (a ski pole marked in decimeters is good, doesn’t flex in high flow, and saves you from falls)
  3. measure velocity with a current meter, or estimate surface current velocity (cm s-1) by timing floating objects across your measurement point1
  4. note dominant and subdominant substrate particle sizes2
  5. using a diving mask or plexiglass view box, note the dominant and subdominant macroscopic algal taxa within an estimated 10 cm x 10 cm area around each sampling point3
  6. record the modal height (cm) of attached filaments or the length from point of attachment of strands of algae floating over the site on the water surface, if these obscured view of the bed.
  7. characterize algal density (% cover) 4
  8. if possible (for Cladophora, zygnematales, perhaps Nostoc) characterize algal condition5
  9. note conspicuous animals within the same 100 cm2 observation area, bearing in mind that some will be attracted to you (minnows) and others will be hiding (mayflies) 6.

Key–Codes and categories used:

Flow velocity (cm s-1)
0 (silt settles vertically)

Wentworth Particle size (median diameter, mm)
<<< 2 mm, not gritty = silt, mud M
< 2, gritty = sand S
2-8 = small gravel Gs
8-16 = large gravel Gl
16-32 = small pebbles Ps
32-64 = large pebbles Pl
64-256 = cobbles C
> 256 but discrete = boulder Bld
continuous = bedrock Br

  1. Algal (etc.) codes

Clad Cladophora (note color as green G, yellow Y or Rust R)

Moug, Spg Mougeotia or Spirogyra (need to distinguish under scope)

Zyg Zygnema (more chatreuse than Spg or Moug, earlier in season)

diat sk diatom skin (gold, orange or yellow-brown) on rocks

bg sk dark or grey skin (often Phormidium, Tolypothrix or Scytonema)

Nos b Nostoc balls

Nos e Nostoc ears (colonized by Cricotopus midge larvae)

Riv Rivularia black freckles to larger circular spots or blotches

Anab deep turquoise Anabaena, loosely epiphytic on Clado

Phor Phormidium—dark brown epilithic cyanobacteria turf in riffle flow

Sedge dominant bunched graminoid along active channel is Carex nudata

Moss Fontinalis is long moss, other riverine mosses are mats < 3 cm high

Terr lvs Terrestrial leaf litter

  1. Algal density (~ % cover)
  1. a few filaments
  2.  < 10% basal cover
  3. 10-50% basal cover
  4. 50-99% cover
  5. can’t see substrate through growth
  1. Algal condition (can’t assess for many taxa)
  1. nearly detritus
  2. senescent, discolored, fragile and falling apart
  3. discolored but less fragile
  4. slightly discolored, but robust
  5. vibrantly healthy
  1. commonly observed Fauna (optional)

Diptera (flies)

tufts (retreats of midge, Pseudochironomus) woven into Cladophora (long woven infested Clado streamers look like dreadlocks)

tube midges-small light colored tubes attached to rocks

Lepidoptera (moths)

Pet – Petrophila, aquatic moth larvae, live under transparent retreats attached to boulders as larva, these retreats condensed to small (~1.5 x .75 cm2) white pupal cases in stage before moth emerges.

Ephemeroptera (Mayflies)

Hept – heptageniids

Baet – baetids (Baetis, Centroptilum)

Ephem –ephemerellids

Siph – siphlonurids

Leps –Leptophlebiiids

Trichoptera (Caddisflies):

Dicos – Dicosmoecus gilvipes-saddle shaped head aperture framed with two ‘saddlepack’ stones

Oncos – Oncosmoecus, similar to Dicosmoeucus, circular head aperture

Tin – Tinodes tube dwelling caddis, like spaghetti

Leuco rice grain size sessile caddis

Gum – Gumaga, slender curved sand cases

Neo – Neophylax, stone cylindrical case

Glosso – Glossosoma, turtle shell stone case


Rams horns – Helisoma

Physella spp


Ferrissia — fw limpets



R roach

Stl steelhead- Onchorhynchus mykiss

Coho-coho, Onchorhynchus

Stk stickleback

PM pike minnow

GSfish green sunfish-Lepomis cyanellus

Bass – large mouth or black bass-Micropterus

Bullh-Black bullheads—Ictalurus nebulosus


Pacific pond Turtle—

Rough skinned newt

Giant Pacific salamander

Foothills yellow-legged frog


Pacific tree frog

Glossary of terms