Background information and definitions
Predatory fish can have negative impacts on amphibian populations, often through direct predation on embryos and larvae. This is particularly the case if the fish are invasive species, often introduced for fishing. For example, a systematic review found that evidence indicates that newts, salamanders and some frog species are less likely to be found in water bodies stocked with salmonids, such as salmon and trout than those with no stocking (Stewart et al. 2007).
There is a large amount of literature that is not included here examining the success of controlling fish by catching, which may be undertaken specifically for the conservation of amphibian species (e.g. Knapp et al. 2004).
Knapp R.A. & Matthews K.R. (2004) Eradication of nonnative fish by gill netting from a small mountain lake in California. Restoration Ecology, 6, 207–213.
Stewart G.B., Bayliss H.R., Showler D.A., Sutherland W.J. & Pullin A.S. (2007) What are the effects of salmonid stocking in lakes on native fish populations and other fauna and flora? Part A: Effects on native biota. Systematic Review No.13. Report.
Supporting evidence from individual studies
A replicated, before-and-after study in 1986–1993 of ponds on the island of Samsø, Sweden (Amtkjær 1995) found that fish and eel Anguilla anguilla control was short-term and did not tend to increase breeding success by green toads Bufo viridis. Breeding was successful in two and failed in two of six ponds with just fish removal. One of the ponds was colonized by adults two years after fish and eels were removed, but breeding was not recorded. Only one male was seen in one of the ponds that was enlarged and had fish removed. Fish or eels were reintroduced to ponds within 1–2 years. In winter (1986–1993), fish were removed from six ponds (three twice). Seven ponds had fish removed and were enlarged. Ponds were monitored by call and torch surveys and by counting tadpoles and metamorphs during 4–6 visits in April–September.Study and other actions tested
A before-and-after study in 1992–2000 at two sites in England, UK (Watson 2002) found that fish control by catching and treatment with rotenone had no significant effect on great crested newt Triturus cristatus populations. At one site, there was no significant increase in great crested newt numbers in the three years following fish removal, which the authors considered to have been only partially effective. At the second site, although great crested newt adults and eggs were recorded following fish control, no larvae were seen. Over 2,000 sticklebacks were removed from the pond, but they were observed again a few years after treatment. Electro-fishing and treatment with rotenone were undertaken at a forest pond in 1996. At the other site, a pond (600 m2) was netted twice to remove trout in autumn 1997. Great crested newts were surveyed at that site in 1992–2000.Study and other actions tested
A before-and-after, site comparison study in 1993–2003 of two lakes in a National Park in Washington, USA (Hoffman, Larson & Samora 2004) found that northwestern salamanders Ambystoma gracile increased significantly following elimination of non-native brook trout Salvelinus fontinalis. Day surveys showed that numbers of egg masses increased from 11 to 25–107/150 m and larvae from 5 to 18–90/150 m. Numbers increased to similar to those in the existing fishless lake (egg masses 65–165/150 m; larvae: 57–114/150 m). Night surveys showed a similar pattern with larvae increasing from 72 to 172/150 m and becoming similar to the fishless lake (50–145/150 m). Trout were removed from June to September 1993–2002 using gill nets (42 m long, 2 m tall). One to four nets were set once to several times during a field season. Salamanders were monitored using snorkel surveys along 25 m transects (four nearshore and two offshore) once or twice annually from July to September. Five night and 17–18 day larvae/neotene surveys and 10 egg mass surveys were completed per lake.Study and other actions tested
A replicated, controlled, before-and-after study in 1996–2005 of 21 lakes in California, USA (Vredenburg 2004) found that mountain yellow-legged frogs Rana muscosa increased following fish removal. One year after removal, numbers had increased for frogs (0.1 to 1.0/10 m) and tadpoles (0.1 to 8.1). Following removal, numbers were significantly greater than in lakes with fish (frogs: 0.1; tadpoles: 0.1/10 m). Within three years there was no significant difference between numbers within removal lakes and fishless control lakes (frogs: 7 vs 5; tadpoles: 10 vs 30/10 m). Trout Oncorhynchus mykiss, Salvelinus fontinalis were eliminated from three, and greatly reduced in two, removal lakes. Fish were removed by gill-netting starting in 1997–2001. Frog visual encounter surveys along shorelines and snorkelling surveys were undertaken in trout removal lakes (n = 5), fish-containing lakes (n = 8) and fishless lakes (n = 8) each two weeks in 1997–2001 and 2–3 times in 2002–2003.Study and other actions tested
A replicated, before-and-after study in 1996–2005 in six lakes in California, USA (Knapp, Boiano & Vredenburg 2007) found that mountain yellow-legged frog Rana muscosa densities increased significantly following predatory fish removal. In three lakes, densities increased significantly from the first five (1996–2002) to last five surveys (2004–2005) for tadpoles (0–12 to 4–91/10 m) and frogs (1–2 to 24–29). Increases were significantly greater than in fishless control lakes for tadpoles (+35 vs +2) and frogs (+25 vs +1). Within 1–3 years of starting fish control, frogs were detected in three lakes where they were previously absent (frogs: 3–67; tadpoles: 0). Complete eradication of fish was achieved from three lakes within 3–4 years, in the other three small numbers remained because of connecting streams. Non-native trout (Oncorhynchus sp., Salmo sp., Salvelinus sp.) were removed using 3–13 sinking gill nets (36 m long x 1.8 m high) set continuously in each lake. Netting was continued until catch rates fell to zero for an entire summer. Fish were eliminated from connecting streams when they dried out, using gill nets and electro-fishing. Frogs and tadpoles were recorded using visual surveys of lake perimeters before and 1–6 times after fish eradication started, up until 2005.Study and other actions tested
A replicated, controlled study in 2003–2006 of 16 lakes in northern California, USA (Pope 2008) found that cascades frog Rana cascadae density, survival, recruitment and population growth rate increased following elimination of fish. Initially, frog densities were similar in the 12 treatment lakes (2 frogs/100 m). However, following fish elimination, densities were significantly higher in removal lakes (frogs: 5–20/100 m; larvae: 12–40/100 m) than in fish stocked and stocking-suspended lakes (frogs: 2; larvae: 1–2). By 2006, there was no significant difference in frog densities in removal lakes and four existing fishless lakes. By 2006, survival estimates of frogs at removal lakes (94%) were higher than those in fishless (64%) and fish-containing lakes (75%). The same was true for population growth rates (removal: 1.7–3.0; fishless: 1.2–1.4; with fish: 0.9–1.2) and recruitment rates (removal: 0.8–1.8; fishless: 0.4–0.6; fish: 0.2–0.5). Twelve lakes were randomly assigned as fish-removal, stocking-suspended or continually stocked lakes. An additional four lakes were fishless. Trout were removed from autumn 2003 to spring 2004 with multiple, repeated sets of sinking gill nets. Frogs were surveyed in 2003 and every two weeks from June to September in 2004–2006. Visual encounter surveys of the shoreline and capture-mark-recapture surveys were undertaken.Study and other actions tested