Ecological Archives E096-179-A1
M. C. Jackson. 2015. Interactions among multiple invasive animals. Ecology 96:2035–2041. http://dx.doi.org/10.1890/15-0171.1
Appendix A. Information on the literature search and robustness of results.
Literature search
I searched the database Web of Science (Thomson Reuters 2013) and Google Scholar in September 2014 using the search terms invas* OR introduced OR alien OR exotic OR non-native OR non-indigenous (in title) AND interact* OR compet* OR facilit* OR meltdown (topic) AND animal OR mammal OR invertebrate OR fish (topic). I then refined this initial list by selecting the following Web of Science Categories: environmental science ecology, biodiversity conservation, zoology and marine freshwater biology. I also examined the reference list of selected articles to find relevant papers missed in the initial search. Papers were chosen if they generated data from field or experimental approaches; those which reviewed existing literature or used a theoretical modelling approach were not considered. Papers which compared the effect of invaders but did not include a treatment where the 2 species were kept together could not be used in the analysis (e.g., Gamradt and Kats 1996, Tyrrell et al. 2006). Papers also couldn’t be considered if they used a removal or enclosure method where the impact of an invader could not be separated from that of other species (e.g., Latorre et al. 2013, Kardol et al. 2014).
Robustness of results
I evaluated our data graphically with funnel plots and used Spearman rank correlation test to quantify the relationship between effect sizes and their sample sizes across studies. The funnel plots showed that the data conformed to a normal distribution and a funnel shape. This finding was supported by a non-significant result from a Spearman rank correlation test for both meta-analyses (rs = 0.06, n = 57, P = 0.66 and rs = -0.17, n = 45, P = 0.27), indicating that the results are free from publication bias. I also assessed the robustness of my results using Rosenthal’s fail-safe number (the number of non-significant, unpublished, or missing studies that would be needed to change the significance of our overall findings; Rosenthal 1979). A fail-safe number larger than 5(n) + 10 (where n is the number of studies in the meta-analysis) is considered to be robust against publication bias (Rosenthal 1979). For both meta-analysis the fall safe numbers exceeded the minimum recommended number based on our sample size (1114.5 > 295 and 1285.6 > 235), demonstrating that my results are robust. The analyses were carried out in MetaWin 2.0 (Rosenberg et al. 2000).
Literature cited
Gamradt, S. C., and L. B. Kats. 1996. Effect of Introduced Crayfish and Mosquitofish on California Newts. Conservation Biology 10:1155–1162.
Kardol, P., I. A. Dickie, M. G. St John, S. W. Husheer, K. I. Bonner, P. J. Bellingham, and D. A. Wardle. 2014. Soil-mediated effects of invasive ungulates on native tree seedlings. Journal of Ecology 102:622–631.
Latorre, L., A. R. Larrinaga, and L. Santamara. 2013. Combined impact of multiple exotic herbivores on different life stages of an endangered plant endemism, Medicago citrina. Journal of Ecology 101:107–117.
Rosenberg, M. S., D. C. Adams, and J. Gurevitch. 2000. MetaWin: Statistical software for meta-analysis: Version 2.1.
Rosenthal, R. 1979. The “file drawer problem” and tolerance for null results. Psychological Bulletin 86:638–641.
Tyrrell, M. C., P. A. Guarino, and L. G. Harris. 2006. Predatory Impacts of Two Introduced Crab Species: Inferences from Microcosms. Northeastern Naturalist 13:375–390.