Background

In aquatic systems, biological invasions can result in adverse ecological effects [1, 2]. Invasive species threaten biodiversity [3–5] and impose considerable economic costs [6], placing increased demands on policy-makers, resource managers, and scientists [7]. The introduction and spread of invasive species can occur by natural or human pathways and can include: shipping networks and canals [8, 9], escapes from aquaculture, aquaria and ornamental trade [10], stocking (including authorized and unauthorized attempts), recreational boating, live food trade, as well as sport fish and baitfish introductions, which may be deliberate or accidental. Additionally, the secondary spread of introduced species poses challenges for resource managers [11–13].

Options for managing non-native fish species can include no action, control and/or containment, population extirpation, and/or species eradication [14]. Containment, such as implementing barriers, is typically the most desirable tactic to prevent the spread of non-native species into novel environments [15–18]. However, where containment is not possible or has not been successful, eradication has been proposed as a valid option for managing biological invasions [19, 20]. Eradication is the elimination of whole fish populations or fish species from distinct habitats or bodies of water [21]. Eradication approaches tend to be targeted, for example, by exploiting vulnerable periods in the life cycle [22, 23] or by focusing on areas of high abundance [24].

The types of fish management techniques available to resource managers to implement eradication programs can vary widely. They can include chemicals, harvest regimes, physical removal, or biological control [25]. The effectiveness of chemical eradication (e.g., rotenone, Fintrol) depends on environmental conditions (e.g., water temperature, depth, pH, discharge, target fish species, hydrology, substrate composition, areas of groundwater recharge; [26]); there are also concerns of collateral damage when non-target species are affected by chemical treatments [27]. Harvest regimes can include intentional over-fishing (e.g., gill netting, angling) of target species [28–30] or modification of angling regulations (e.g., favour overharvest of target species). Physical removal techniques can include traps, electrofishing, and/or netting programs while biological controls can include the introduction of predators, intraspecific manipulation, or targeted pathological reactions [31]. When implementing fish management programs, risk analysis is required to help decide when management strategies should be utilized, what strategy should be chosen, and what the likelihood of success of different strategies are [18]. The risk analysis includes identification and assessments of hazards, including predicting the likelihood and severity of adverse effects [32].

The success of non-native fish management approaches can vary greatly depending on the management objectives for the project: whether control, eradication, removal or containment (amongst others) was the ultimate goal of the project. As can be expected given the complexities of the natural environment, success can be difficult to quantify and some approaches can be unsuccessful despite best efforts [7, 19, 25]. Failure of non-native fish eradication techniques can occur due to a number of factors, including (but not limited to) ineffective capture techniques (e.g., size-specific efficiencies), habitat complexity (e.g., areas of refuge, plant density), species-specific factors (e.g., size, habitat preferences), and physical water properties (e.g., water chemistry, temperature, water depth; [33]). Determining the outcomes of management interventions, especially when restoration of freshwater ecosystems is a goal (i.e., to eradicate non-native target fish species from a specific waterbody), requires long-term evaluation and assessment in relation to meeting the objectives [19, 25, 34]. Post-program evaluation and assessment is required not only to determine the effectiveness of techniques but also to explore the cost-effectiveness and cost/benefit of each strategy. Narrative syntheses, meta-analyses and systematic reviews can be valuable approaches to determine broad-scale effectiveness of how management interventions can be effective and to identify future research needs.

Objective of the review

The objective of the systematic review is to evaluate the existing literature base to assess the effectiveness of different non-native fish eradication methods in global freshwater environments. The purpose of the review is not to question whether or not the eradication of a target species is an appropriate objective, but simply to evaluate the effectiveness of each technique for eradicating the target species in a desired freshwater body (including both lakes and rivers).

Primary question

What is the effectiveness of non-native fish eradication techniques in freshwater ecosystems and what effects do the various sources of potential heterogeneity have on the outcome?

Components of the primary question

The primary study question can be broken down into the study components:

Subject (population):

Non-native freshwater fish

Intervention:

Fish eradication method

Comparator:

No intervention or alternative method

Outcomes:

Magnitude of decreased abundance relative to control or eradication of target fish species.

Secondary questions

Following the development of the primary question, secondary questions were developed to expand on relevant areas of interest to project stakeholders and user groups, including Canadian federal natural resources government agencies (e.g., Parks Canada) and members of the international scientific community specializing in invasive fish eradication science. The secondary questions are meant to help guide the overall goals of the systematic review and to ensure that areas of interest are encompassed in the methods. The secondary questions for this systematic review are:

  1. (a)

    To what extent does effectiveness vary with eradication technique (e.g., electrofishing, piscicides, unlimited recreational catch limits)?

  2. (b)

    What other strategies/techniques are being employed but are under-represented in the evidence and literature base?

  3. (c)

    What factors (e.g., type and size of water body, species, elevation, time since invasion/introduction, ecoregion) influence the effectiveness of each type of eradication method and in what context is each technique most effective?

Methods

Details related to each step in the systematic review protocol are outlined in Fig. 1.

Fig. 1
figure 1

Overview of steps in the proposed systematic review protocol

Full size image

Search strategy

Search terms

Similar to the way that the Review Team formulated the primary and secondary questions, the Review Team collaborated with project stakeholders (e.g., Parks Canada) and members of the international scientific community specializing in invasive fish eradication science to generate a list of relevant search terms (Table 1). The terms were broken into three components: population, intervention and outcome.

Table 1 Proposed search string for the execution of the search strategy
Full size table

The terms in each of the categories of population, intervention and outcome can be combined using the Boolean operators “OR” and/or “AND”, as suggested in the search string. The asterisk (*) is a wildcard and represents any characters (e.g., remov* includes remove, removal, removing, removed) while the dollar sign ($) includes zero or one character (e.g., rod$ and $reel includes rod-and-reel, rod and reel). The terms are combined in the following general format:

$$\left( {\text{Population}} \right){\text{ AND }}\left[ {\left( {\text{Intervention term}} \right){\text{ AND }}\left( {\text{Outcome term}} \right)} \right]$$

[Fish* AND (Invasive OR Non$Native OR Alien OR Exotic OR introduced OR non$indigenous OR IAS OR Invasive$species OR Alien$invasive) AND (Fresh$water OR Stream* OR Water* OR River* OR Lake* OR Reservoir* OR Pond* OR Canal* OR Harbor* OR Harbour* OR Port* OR Wetland*)] AND [(Hyraulic OR Screen* OR Weir* OR Net* OR Gill OR Trammel OR Hoop OR Trap OR Cast OR Lift OR Seine* OR Trawl* OR Electrofish* OR Electric OR Cull OR Piscicide* OR Rotenone OR Antimycin OR Fintrol OR Explosive* OR Primacord OR Biocide OR Angl* OR Trotline* OR “Rod-and-reel” OR Limb$line* OR De$water OR Drawn&down OR Pump*) AND (Restor* OR Rehabilitat* OR Remov* OR Eradicat* OR Control* or Suppress* OR Reduc* OR Renovat* OR Exclusion or Exclude*)].

Abbreviated search

When a complex search string is not accepted by the search engine, the help menu will be consulted and the search terms will be modified. The search terms will be recorded in the article databases in order to preserve all metadata associated with the search.

Article type

The search will include a variety of article types, including primary literature in peer-reviewed journals and grey literature. The search strategy will strive to minimize publication biases by focussing efforts equally on each article type and putting equal weight on the information provided in each article type.

Document/file formats

The search will not have any document type restrictions (e.g., PDF vs. MS-PowerPoint vs. MS-Word). All formats will be acquired and if specialized software is required, alternative formats will be requested for ease of file transferability. Where books are identified, digital copies will be sought (either through internet searches for availability or requests to authors) in order to ensure that all obtainable records are made available as an output from this review.

Computer settings

The browsing history and cookies will be disabled on all computers used to conduct the search. The members of the Review Team will not access any electronic accounts (e.g., email, website) during the search period and will use “private mode” (Safari) for web browsers to reduce the possibility of user-specific search results.

Language

English search terms will be used to conduct all searches in all databases. All references that are returned will be included in the database. When articles in other languages are returned using the search strategy, those records will be reported in the database.

Publication databases

  1. 1.

    Waves (Fisheries and Oceans Canada)—Canadian government books, reports, government documents, theses, conference proceedings and journal titles

  2. 2.

    Index to Theses Online—Dissertations and theses from the UK & Ireland

  3. 3.

    Science.gov—U.S. Federal Science

  4. 4.

    ISI Web of Science core collection—Multidisciplinary research topics including journals, books, proceedings, published data sets, and patents

  5. 5.

    Scopus—Abstract and citation database of peer-reviewed literature including journals, books, and conference proceedings.

Search engines

The first 100 hits (based on relevance) will be examined for the appropriate fit for the review questions.

  1. 6.

    Google Scholar.

Specialist websites

The first 50 documents from each search will be included in the reference database and checked for relevance. Reference lists of included material will be searched and any relevant documents will be included and added to the reference database. Where links to other organisations are included on the websites, the links will be followed to try to capture any organisations that were not initially included in the website searches. All articles will be exported into EndNote prior to assessment of relevance.

  1. 7.

    Atlantic Salmon Federation

  2. 8.

    Centre for Ecology and Hydrology

  3. 9.

    Centre for Environment, Fisheries and Aquaculture Science

  4. 10.

    Commonwealth Scientific and Industrial Research Organisation

  5. 11.

    Convention on Biological Diversity

  6. 12.

    Department of the Environment, Food and Rural Affairs

  7. 13.

    Desert Fishes Council

  8. 14.

    Fisheries and Oceans Canada

  9. 15.

    Fisheries Research Service

  10. 16.

    Food and Agriculture Organization of the United Nations

  11. 17.

    Joint Nature Conservation Committee

  12. 18.

    National Institute of Water and Atmospheric Research

  13. 19.

    National Park Service

  14. 20.

    Natural England

  15. 21.

    Natural Resources Canada

  16. 22.

    Natural Resources Wales

  17. 23.

    Northern Ireland Environment Agency

  18. 24.

    Pacific Salmon Foundation

  19. 25.

    Parks Canada

  20. 26.

    Trout Unlimited

  21. 27.

    The Nature Conservancy

  22. 28.

    United Nations Environment Programme

  23. 29.

    US Forest Service

  24. 30.

    US Fish and Wildlife Service

  25. 31.

    Western Native Trout Initiative

  26. 32.

    World Wide Fund for Nature

  27. 33.

    World Wildlife Fund.

Other literature searches

Reference sections of accepted articles will be hand searched to evaluate relevant titles, symposium papers, and other articles that have not been found using the search strategy. Authors of any unpublished references will be contacted to request access to the full article. Stakeholders will be consulted for insight and advice for new sources of information.

The Review Team will contact authors of unobtainable articles in an attempt to gain access to the full article. We will also use social media to alert the community of this systematic review and to reach out to area experts for research articles that are difficult to obtain, or for suggestions of articles to include. Any article provided will also be used to test the comprehensiveness of our search strategy and, where appropriate, adjustments will be made to the search strategy to make sure it is comprehensive and inclusive. Any changes made to the search strategy will be justified and documented in the final review document.

Search record database

All articles generated by each of the search strategies will be exported into separate EndNote databases. After all searches have been completed and references found using each different strategy have been compiled, the individual databases will be merged into an overall EndNote database library. Duplicates will be identified. All references regardless of their perceived relevance to this systematic review will be included in the database. This database will act as the archive and will remain unchanged throughout the review process, since it is the direct product of the search strategy and will be useful in the future when updating the systematic review archive (general updating timeframe is currently every 5 years).

Article screening and study inclusion criteria

Screening process and inclusion criteria

Articles found using the search criteria will be screened in three distinct stages; title, abstract and full text.

Before the screening process begins, two reviewers using a subset of 10 % of all articles or 100 abstracts (whichever is bigger) will undertake consistency checks to ensure consistent and repeatable decisions are being made in regards to which articles get screened out and which go on in the process to be further reviewed. The two reviewers will use a Kappa test to determine consistencies in screening decisions. A Kappa score of ≥0.6 indicates substantial agreement between reviewers and will be required to be achieved before any further screening is conducted for the review. The results from the consistency check will be discussed and discrepancies will be reviewed by both reviewers to understand why the choice was made to include/exclude the article.

All article screening decisions will be included in the database, so it will be clear at what level any article was excluded. If the decision to include or exclude a specific article is unclear, that article will be retained and will go on to the next level of screening. If there is further doubt, the Review Team will discuss those articles as a group to come up with a decision. Any articles that do not have abstracts (as is the case for some grey literature), those articles will automatically be screened at the full text level. Justification of the reason for inclusion or exclusion of an article will be explained and recorded in the article database, and all articles excluded at the full text level will be included with the review, in compliance with CEE guidelines.

Articles will be excluded based on the following pre-defined inclusion criteria (Table 2).

Table 2 Description of inclusion criteria used by the Review Team when screening articles at the title and abstract level
Full size table

Study quality assessment

Each of the studies that make it to the full text screening level will be classified and coded in the article database using a number of parameters including (but not limited to):

  • Study setting—Lab or field

  • Study design (BA/CI/BACI/RCT)

  • Temporal extent of study

  • Replication—Replicated or unreplicated

  • Confounding factors—Present, not present, unclear, and whether they were accounted for in the study

  • Clarity of objectives in relation to methods used (e.g., is the ultimate type of management intervention objective clearly identified, including eradication, controlling, containing etc.)

  • Use of (and number of) control and reference sites

  • Effort devoted to eradication techniques (e.g., press versus pulse, proportional area treated)

  • Statistical methods used in assessment of success (e.g., were results analysed statistically?)

  • Accounting for and/or identifying potential effect modifiers (see list in following sections).

Bilotta et al. [35] have outlined criteria for the assessment of the internal validity of a study. Their assessment criteria have been adapted from the Cochrane Collaboration’s Risk of Bias Tool [36] for use in the field of environmental science. The assessment criteria include assessing selection bias, performance bias, attrition bias, reporting bias, as well as those biases that may only be relevant in unique situations (e.g., contamination). The criteria outlined in Bilotta et al. will be used by the Review Team for this review and included in the reference database. The information for each article retrieved using the search strategy will be uniquely coded based on the criteria (generally categorised as “low risk”, “high risk”, or “unclear risk”) to help assess the quality of each article, and to provide insight into any potential risk of bias present in each of the studies. This information will be instrumental in helping to determine reliability of the evidence base available for potentially conducting a meta-analysis on the effectiveness of each eradication method.

Data extraction strategy

Metadata will be extracted from the included studies by the Review Team and will be recorded in a MS-Excel database that will be made available with the published systematic review article, as an additional supporting file. The extracted information will be used to assess the overall effectiveness of each intervention strategy, and when sufficient, good quality data exists, the information will be used in a meta-analysis. Some of the outcome data that will be recorded will include: outcome means, measures of variation (e.g., standard deviation, standard error, confidence intervals), and sample sizes. When data is presented in tables or graphs, all information will be extracted and recorded. If it is not possible to decipher information from graphs, the main contact author for the article will be contacted (via email or phone) by the Review Team to request the information. During that request, the Review Team will also solicit the author to suggest any grey literature that they may know of related to the systematic review topic. Where only raw data is provided in the article, the Review Team will calculate summary statistics. In those instances, it will be recorded in the MS-Excel database how the calculations were done and with what information. To ensure that data is being extracted in a consistent and repeatable manner, two reviewers will extract information from 10 of the same articles. Afterwards, the information will be compared. Any inconsistencies will be discussed amongst the Review Team members, and if any disagreement occurs, they will be discussed with the entire Review Team to ensure all reviewers are extracting and interpreting data in the same manner.

Potential effect modifiers and reasons for heterogeneity

The Review Team will extract data on potential effect modifiers (see Table 3) from articles that are included at the full-text level of screening. All information will be recorded in the MS-Excel database.

Table 3 presents the criteria that will be considered as potential effect modifiers and/or reasons for heterogeneity and will be extracted. Further factors may be identified, defined and included throughout the process, through consultation with external experts.

Table 3 Data extraction table
Full size table

Data synthesis and presentation

A narrative synthesis of data from all articles included in the systematic review will be generated. The synthesis will aim to be as visual as possible, summarizing information in tables and figures. The ultimate goal of this review is to assess the effectiveness of each different eradication technique and to identify the factors that influence the overall success rate of each type of method, in order to better inform management agencies who routinely have to decide when, where and how non-native fish eradication programs should be implemented. All efforts will be made to provide quantitative assessments and meta-analysis of the articles included in this review, when the study designs and evidence-base allow. The review team has conducted some scoping exercises, particularly to help develop an efficient search strategy and to get a sense of the existing literature base. When doing so, the review team got the sense that sufficient evidence may exist to allow them to conduct a meta-analysis on some intervention types, but it is unlikely that a meta-analysis may be possible for all intervention types. It will also depend on the literature base for each target species. Sufficient evidence may exist for some more common target (problem) species but not for more regionally-relevant species.