With the sun high, fly rod in hand, and wet wading a river, winter steelhead feel like a distant memory. As a scientist, however, this is when redd count data from the prior winter steelhead season are being analyzed, and in some cases, preliminary run-sizes are being estimated.
Most anglers understand the general process for estimating the annual escapement of steelhead.
Managers walk or float the streams to enumerate redds. Sometimes they complement those surveys with helicopter flights. Usually, a portion of the river is surveyed and counts from those areas are expanded into habitats that were not surveyed.
It's a fairly simple approach, but, as we discuss here, there are important nuances that can have profound effects on estimates generated from redd counts.
Spawning steelhead on the Elwha. Redd counts are an indirect measure — the real goal is to estimate how many fish like these returned to a given river in a given year. Every nuance below is about how cleanly that translation can be made.
Nuance #1: Test Redds
Most anglers are aware of the term "test redd." It refers to areas where a female steelhead tested the gravel but didn't deposit eggs. Generally, test redds are considered to be significantly smaller than full redds that contain eggs. Unfortunately, the proportion of redds that don't contain eggs is rarely tested and consequently, it's difficult to determine whether test redds are consistently smaller than full redds. This is problematic because surveyors must arbitrarily make a decision based purely on a visual guess.
A single isolated redd in clear, shallow water. From the bank, this is the call a surveyor has to make on every dig they walk past: full redd, or test? On size alone, it's often impossible to tell.
A 2011 study in Idaho on adfluvial rainbow trout sought to answer those two questions by excavating redds to determine which had eggs and then comparing the size of test redds to full redds. Adfluvial rainbow trout migrate to a lake and then back to their natal stream, just like "steelhead" in the Great Lakes.
The authors identified 103 likely redds and excavated and measured 51 redds to determine the presence of eggs and compare the size. Only 34 of those redds contained eggs. The remaining 17 were barren. In other words, 33% of the redds were test digs.
Redds with eggs were larger than test digs, and some test digs were quite small, but the range of measurements for full redds and test digs overlapped enough that they could not be reliably distinguished based solely on size. In other words, a surveyor would need to verify the presence of eggs to accurately distinguish a redd.
Nuance #2: Multiple Redds per Female
Not only do steelhead dig test redds, they may also place their eggs in multiple redds. This is a great strategy to spread risk across multiple sites but further complicates estimating escapement from redd counts alone.
Nuance #3: Skewed Sex Ratios
Enumerating redds only accounts for females, but many steelhead populations — if not all — have skewed sex ratios that are biased towards females. Some populations, such as those in the interior Columbia River may consist of 60%–70% females, depending on the year and river. The sex ratio is typically less biased in winter runs near the coast, but the issue remains: if we are counting redds to estimate females, how do we account for males?
Of course, it's the females that deposit eggs and produce the next generation of steelhead, but the variability in sex ratios is still important. For instance, creel surveys on the Olympic Peninsula are used to estimate the total number of steelhead caught and released by anglers. That number is then compared to the escapement estimate of males and females. If the sex ratio varies by year and population, it could bias estimates about encounter rates.
Nuance #4: Surveyor Bias
Some redd surveyors are highly experienced, while others are not. This can result in surveyor bias, where each person has slightly (or large) differences in interpretation of what is and is not a redd. This can happen even with experienced surveyors.
Differences in surveyors' interpretation of redds can lead to substantial "noise" in the data, particularly when it comes to distinguishing test digs from full redds. This is an unavoidable challenge that is almost impossible to fully resolve.
Nonetheless, it can be accounted for by occasionally repeating surveys with different surveyors. That allows scientists to quantify the variation that exists between surveyors.
There is a common saying in science: if you do something wrong, do it consistently wrong.
Implications
So how well do state agencies account for these shortcomings in the most fundamental aspects of management?
In western Washington streams, managers have used Snow Creek — a small stream with a weir to count adult steelhead — to estimate the proportion of redds to fish. They have generally found a ratio of approximately 1.6–1.8 fish per redd. That helps account for females digging multiple redds and sex ratio. It does not, however, account for distinguishing between test and full redds.
For unknown reasons the ratio from Snow Creek is not consistently used. In the Skagit River, for example, managers used to use a ratio of 1.6–1.7 fish/redd. Strangely, at some point that changed and now they use 2 fish/redd. It's entirely unclear when and why they changed from one ratio to another.
In contrast, on the Olympic Peninsula they use the ratio from Snow Creek.
None of these methods account for the complexity in accounting for test digs. There is a common saying in science that if you do something wrong, do it consistently wrong. This implies your data will be comparable across a time series, and can even be corrected, since it was collected in a consistent way. Unfortunately, this type of work is rarely done for steelhead and it remains unknown what proportion of counted redds truly contain eggs — and raises the question: are we overestimating steelhead abundance? And if so, to what degree?
Answering this question is not just academic — it is imperative to effectively managing steelhead, especially when estimates fall near critical abundance thresholds. There are established methods to improve the quality of redd counts, but new technologies such as sonar also hold promise for improving our ability to estimate wild steelhead abundance. The first step in effectively managing wild steelhead is to collect accurate biological data to inform management decisions, otherwise it may be difficult to identify trends and limiting factors.
Support the Work
Better Science. Better Decisions. More Wild Fish.
Every donation helps us put more anglers in the field as scientists — gathering the data that protects wild fish across the Pacific Rim.
Make a Donation