History of Smelt in the Great Lakes

Figure 1. Commercial smelt harvest from Wisconsin waters of Lake Michigan through 2005.

Credit: Phil Moy, Wisconsin Sea Grant

The rainbow smelt (Osmerus mordax) is not native to the Great Lakes. The smelt is a saltwater species, though a freshwater population exists in Green Lake, Maine. Fish from this population were stocked into Crystal Lake, Michigan in 1912. Some of the fish escaped from Crystal Lake and smelt were first caught in Lake Michigan in 1926. 

Once established, the smelt population expanded rapidly in Lake Michigan becoming very abundant in the 1930s. The smelt was nearly eliminated from the lake in 1941 to 1942 by an unknown pathogen. However, by the mid 1950s and into the 1960s the fish were once again highly abundant.

Figure 1 illustrates the wide fluctuations in the Lake Michigan smelt harvest. This type of population expansion and decrease is typical of invasive populations. One factor in the population expansion of the smelt in the 1950s could be the lack of predation pressure as lake trout were essentially eliminated by sea lamprey. The smelt population declined again in the early to mid sixties, roughly coinciding with the stocking of Pacific salmon.

Smelt spend most of the year in deep water offshore. There, they feed on benthic invertebrates such as opossum shrimp and amphipods, but smelt also consume other small fish. In the spring, smelt move from deeper water offshore into shallow nearshore waters to spawn. The spawning season lasts for about two weeks in a given area, but the spawning season extends from March into May. The fish begin to move

to the mouths of tributaries when the water reaches about 40o F. The fish may swim some distance upstream to spawn or may spawn in shallow water over gravel deltas at stream mouths. Spawning generally occurs at night.

Historically they congregated in enormous numbers along shorelines and near river mouths, providing seasonal recreation and sustenance for coastal residents. So abundant were the fish that not only could they be dipped out with a net, but a pot or strainer was effective as well. Fish numbers in present-day spawning runs are lower, so it is more difficult to get a bucketful than in the past. Changes in the population abundance reflected in the harvest are likely the result of natural influences such as predation, inter-specific competition and changes in the forage base.

According to local commercial fishermen, smelt movement patterns seem to be changing. This may be related to the clearer water conditions found in the lake, which has been attributed to zebra mussels. The light-sensitive smelt seem to be seeking deeper, probably darker waters, rather than residing closer to shore.  The fish may be avoiding the deeper light penetration or may be following forage that has moved into deeper waters.

The US Geological Survey annually monitors the abundance of smelt and other Great Lakes fish. These estimates are used to monitor the fish populations in the lake for management and to establish harvest quotas. Commercial fishermen are concerned

that the assessment does not accurately reflect the abundance of smelt, and that the quotas too restrictive.

Abundance estimates are generally used to track population changes over time. These types of assessments use a fixed sampling strategy. For example, each year, the same area of the lake would be sampled with the same gear, during the same season. This is done to control as many variables as possible so that changes in the catch reflect changes in the fish population rather than being a result of the sampling methodology. The results provide an estimate of the abundance as it relates to results from previous years.

The accuracy of the estimate can be affected by changes in fish movements as well as changes in gear types. The fish could be abundant in the lake, but due to other influences are not in the sampling area when the sampling takes place. Likewise, the fish population could be depressed, but if many of them congregated in the sampling area at the right time, the population assessment could be overestimated.