A fish counter network for monitoring Scottish salmon stocks

4. Fish counting methods

There are several ways of counting adult salmon returning to a river, including whole-river trapping, resistivity counters, optical systems, hydroacoustic (sonar) cameras and the use of Passive Integrated Transponder (PIT) tagging of fish.

4.1. Whole River Trapping

Whole river trapping can only be undertaken on small rivers and, when operational, requires daily staffing to ensure the welfare of fish and other wildlife. The Marine Directorate operates two salmon traps on tributaries of the River Dee in Aberdeenshire. The Baddoch trap has intercepted emigrating smolts and returning adults on this headwater stream since 1988, whilst the Girnock trap in middle-Deeside has operated since 1966. With long time series available, the data from these traps were used to help estimate egg targets and can give information on time trends of local adult stock abundance and run timing. However, as they are located on tributaries and are far from the river mainstem outflow to sea, they cannot provide an estimate of whole-river stock abundance or trend.

4.2. Resistivity Counters

Resistivity counters were first installed on hydroelectric power schemes in the 1950s (Eatherley et al., 2005) and have developed into a trusted method of fish counting. Resistivity counters incorporate a Wheatstone bridge of three electrodes spanning a river channel or fish pass and over (or through) which fish must pass on migration. As a fish passes the electrodes it reduces the electrical resistance between them, and this change is automatically detected and measured by the counter.

The bulk resistance of the water varies depending upon its conductivity, temperature and volume. Consequently, resistivity counters cannot be used in water with high conductivity (e.g., brackish or saline waters) and the distance between a fish and the electrodes must be maintained relatively constant.

To control the height that a fish passes over the electrodes, resistivity counters can be installed on weirs with a relatively long trailing edge “Crump” profile (e.g., at river gauging stations), on fish passes with controlled water flows, or on ducts or fish lifts of fixed dimensions.

As water conductivity and temperature are not constant, resistivity counters automatically calibrate at frequent regular intervals to account for changes in these parameters (Braun et al., 2016). For water with bulk resistance of >100 Ohms, the size of the signal relates non-linearly to the size of the fish, allowing differentiation between larger salmon and smaller sea trout (where overlap in sizes of the species is insignificant or can be accounted for). The up-stream or down-stream direction of movement is determined depending on whether the fish crosses the upper or lower electrode first.

4.3. Optical Systems

Optical systems (e.g. TiVa or Vaki Riverwatcher) use underwater video cameras, infrared beams, or both to monitor fish in artificially narrow channels. Optical infrared systems automatically count fish and capture images of fish silhouettes as they pass through and break infrared beams, whereas underwater video captures images of fish that can then be viewed to identify the species and create a count of fish. Some systems now use Machine Learning algorithms to identify species or measure fish length. Optical systems can operate only over a short range (typically <50 cm) and are best suited to being placed in a fish pass or natural constriction, where target fish must pass through the infrared beams (either single detector systems or multiple parallel systems can be installed) or pass within range of the video camera. Video systems are not suitable for use in turbid water.

4.4. Hydroacoustic (sonar) cameras.

Hydroacoustic cameras illuminate targets by multi-beam sonar, producing echogram images that can identify fish regardless of lighting conditions and in poor water clarity. Whilst sonar can detect fish at a range of several metres, the operating range depends upon the size of the target – larger fish can be detected farther away from the camera. To detect and allow length measurement of salmon, the maximum working range is 20-25 m.

Unlike resistivity or optical systems, hydroacoustic systems do not require significant (and costly) infrastructure to be built within the river, offsetting the high capital cost of the cameras. However, they do require to be carefully sited in locations that provide a bed profile that is similar to the cone shape of the sonar beams. The chosen location should also avoid situations where fish “mill” back and forth through the ensonified area, or areas where coverage is affected by factors such as air bubbles from turbulence or shadowing from large, fixed objects. To meet these conditions, fish may need to be guided to swim through the ensonified area using fencing, or similar structure.

The availability of software that can efficiently identify, track and produce counts of fish passing through a location has been an issue with these systems to date but, with the development of machine learning to automate the identification and tracking of fish in sonar footage, progress is being made.

4.5. Passive Integrated Transponder

The use of Passive Integrated Transponder (PIT) tags allows tagged fish to be automatically detected and individually identified as they swim over or through a detection antenna. By tagging juvenile fish and siting antennae at different locations within a catchment, the dispersion and migration of tagged fish through the river system and the survival of fish through different life-stages can be assessed. Where antennae are placed near the mouths of rivers, this may allow juvenile productivity and adult return rates from sea to be estimated. These are additional parameters that can aid understanding of the performance of salmon stocks and some fish counters incorporate, or are located near to, PIT tag antennae to obtain maximum benefit from both types of data.