Cost-effective monitoring of lakes newly infested with zebra mussels

This project will develop recommendations for underwater survey methods to estimate zebra mussel population abundance and distribution. Having better survey methods will guide and improve treatment options and pre- and post-treatment monitoring.

Currently, when a lake is newly infested with zebra mussels, it’s difficult to know the extent of the population. The earliest stages of a lake colonization are difficult to monitor because abundance is low, mussels are sparsely distributed, and their small size makes them hard to find and count. These survey methods are needed by agencies, watershed districts, and lake managers who are confronted with a new infestation.

SCUBA divers will use line-transect sampling to create, test, and standardize sampling methods to quantify zebra mussel populations.  They will work in ten to twenty newly infested lakes throughout Minnesota over two years. Simulation modeling will also be used to evaluate the efficiency of alternative survey designs and to provide recommendations regarding appropriate sampling effort.

Implementing these recommendations will result in standardized data which will help guide zebra mussel treatments and evaluate their effectiveness statewide and beyond. 

Click here for additional details on this project, including specific survey designs, presentations, and a simulation study.

Progress:

As of January 2019, field surveys were complete and three different survey techniques were implemented: double-observer surveys with distance sampling, double-observer surveys without distance sampling, and quadrat counts. Two approaches for analyzing the data were developed: a straightforward approach that can be implemented with existing open-source software and a more refined approach that can be used to explore the effect of covariates (such as plant presence) on detection probabilities and zebra mussel density. Both methods produced density estimates that were 3 times larger than the observed densities (uncorrected for detection). These results demonstrate the importance of estimating and adjusting for detection probabilities <1 rather than relying on observed counts when comparing densities over time or space.

In summer 2018, the field team has visited 30 lakes and sampled roughly half. Sampling was split up into phases: the first two phases were used to quickly assess relative abundance and spatial distribution of mussels in a set of candidate lakes without attempting to estimate detection probabilities or correct for imperfect detection. The third phase was used to more rigorously compare alternative survey methods useful for estimating abundance (i.e., correcting, as necessary, for mussels not observed in the surveyed area) in a small number of low density lakes.

During the 2017 field season, students from Carleton College completed a simulation study to explore the efficiency of different survey designs using simulations. Their results support the use of distance sampling for estimating density of zebra mussels in lakes, but point to the need for increased sampling effort to reduce uncertainty associated with density estimates.

Final outcomes:

We evaluated 5 different survey designs for estimating zebra mussel density (2 designs in 2017 and 3 designs in 2018), employing methods that utilize counts by two divers to estimate the probability of detecting mussels in the surveyed area. We also compared survey designs in terms of their density estimates, associated measures of uncertainty, and sampling efficiencies (time required to complete a survey), using data collected in 3 lakes of varying density and using a simulation study and analytical framework informed by our data. In 2017 in Lake Burgan, we estimated that a diver could detect between 5% and 41% of the mussels present in the surveyed area, depending on the specific diver and on whether the lake bottom was vegetated, with vegetation having the larger effect on detection. Accounting for low detectability of zebra mussels led to an estimate of density over three times higher than the observed density. Thus, for every zebra mussel detected by our divers, approximately two were missed. Using the data collected in 2018 and further simulation and analytical work, we found that double-observer survey designs that allow for imperfect detection are optimal when surveying lakes at low density, whereas quadrat counts that assume perfect detection are optimal at higher densities. We developed a training video, data collection worksheets, and an analysis tutorial so that others may implement our proposed survey designs in newly infested lakes (all available here). These tools benefits Minnesotans by providing better ways to monitor lakes infested with zebra mussels and to evaluate the effects of treatment options on zebra mussel density.

Zebra mussel survey design

Watch this video to learn more about survey design, data entry, and observer methods.

Project manager: John Fieberg

Funded by: Environment and Natural Resources Trust Fund as recommended by the Legislative-Citizen Commission on Minnesota Resources

Project start date: 2017

Project end date: 2019