1. Bycatch of non‐target species in commercial fishing nets can have adverse impacts on their populations. Freshwater turtle populations are particularly susceptible to increases in adult mortality, and freshwater turtles are among the most threatened vertebrates.
2. As a case study, the population‐level impacts of bycatch mortality on freshwater turtles were evaluated in Lake Opinicon, Ontario, Canada, a lake that supports a small‐scale commercial fishery. Using population viability analyses, the impacts of bycatch on common snapping turtles (Chelydra serpentina), eastern musk turtles (Sternotherus odoratus), northern map turtles (Graptemys geographica), and painted turtles (Chrysemys picta) were evaluated.
3. In all four species, even low levels of additional annual female mortality as a result of bycatch were sufficient either to reduce population size or to cause extirpation of the local population within 500 years. Bycatch reduction programmes, such as seasonal closures and implementation of bycatch reduction devices, can help alleviate the risk of extirpation. Changes to fishing season length could help reduce the number of snapping turtles and musk turtles captured. Installation of simple bycatch reduction devices can exclude between 95% and 100% of snapping turtles and between 0% and 97% of the other three species, depending on the width of the exclusion device. If combined, these two bycatch reduction methods would help prevent adult female mortality and help maintain turtle populations in Lake Opinicon.
4. Although these findings are specific to the study area, the same principles apply to other areas where similar simple bycatch reduction strategies can be employed to prevent the extirpation of other freshwater turtle species. Considering the consequences of bycatch and of bycatch reduction programmes on populations provides managers with important information to support development of risk‐averse conservation strategies.
In eastern Ontario, Canada there is a small‐scale fishery that uses passive impoundment style nets (e.g. fyke nets) to collect a variety of panfish (Burns, 2007; Larocque et al. , 2012a). This study was conducted on Lake Opinicon, located 100 km south west of Ottawa, Ontario, Canada. Lake Opinicon is a mesotrophic, shallow lake with a mean depth of 2.8 m and a surface area of ~780 ha (Agbeti et al., 1997). This lake supports a small‐scale freshwater commercial fishery, characteristic of other lakes in the region, with a single fisher licensed to deploy 80 nets simultaneously. Although turtle bycatch occurs in this type of small‐scale commercial fishery (M.A. Carrière pers. comm. Oct. 18, 2013), fishers are not required to report the number of turtles captured or killed (Larocque et al. , 2012b; OMNR, 2013). Therefore, catch rates from fishing nets deployed as part of a continuing study of turtle bycatch in Lake Opinicon were used to estimate bycatch in commercial fishing nets (see below).
Population viability analysis
Population viability was modelled in the software Vortex 9.99 (Lacy, 1993, 2000) using species‐specific parameters collected from the literature (Supporting information, Table S1). Population viability analysis has previously been used to evaluate the population consequences of bycatch issues in marine mammals (Harwood, 2000; Majluf et al. , 2002; Goldsworthy and Page, 2007) and in seabirds (Majluf et al. , 2002). In the PVA models, rates of mortality, reproductive age of males and females, and maximum reproductive age were set as fixed variables. However, stochasticity was built into the models through variation in annual clutch sizes and variation in annual number of clutches (see Lacy, 1993, 2000 for a detailed discussion of model assumptions and application in Vortex). Given the lack of a density‐dependent reproductive response in turtles (Brooks et al. , 1991), this parameter was not incorporated into the PVA models. Similarly, to isolate the effect of bycatch mortality and simplify the models, inbreeding was not included.
Lake‐specific life‐history parameters of each species could not be calculated owing to the unavailability of long‐term demographic data; therefore, an effort was made to use literature on northern populations that should have life histories that are close approximations of the life histories of the study populations (Galbraith, 1986; Iverson, 1991, 1992). The viability of the populations and influence of adult mortality were modelled over 500 years using 1000 iterations. This time horizon was chosen because of the long generation times that typify most freshwater turtles. Carrying capacity (K) was set to the initial population size and the population was considered extinct when there was only one individual remaining (Supporting information, Table S1). A detailed description of how population sizes were estimated can be found in the Supporting information, Methods S1.