Holeck and colleagues (2004) rightly draw attention to transoceanic ships as a source of nonindigenous species (NIS) in the North American Great Lakes. We agree that “management strategies aimed at preventing new invasions must consider the linkages between NIS and vectors” (p. 927), and that invasive species from anthropogenic sources will be a growing problem as global trade increases. We are concerned however, that the inference that midocean ballast water exchange (BWE) has been ineffective is incorrect and might mislead further policy development.
At issue is their plot of the cumulative number of recorded NIS against the year of their discovery, from 1955 to 2000. This interval is divided into two periods: before a voluntary policy of BWE was instituted in 1989, and afterward. Linear regressions were fit to the data in each period independently, so that the estimated slopes represent the pre- and post-1989 discovery rates, respectively. On the basis of these results, the authors report that “the annual rate from 1989 to 2000 was more than double that observed between 1959 and 1988” (p. 923), from which they conclude that transoceanic shipping “remains the largest source of NIS in the Great Lakes” (p. 927). We believe the authors intend to imply that BWE has been ineffective or is possibly even related to an increase in the rate of invasion.
The authors acknowledge that investigation bias and time lags between species establishment and discovery obscure the true rate of species invasion. However, they do not acknowledge that these biases alone could be sufficient to cause the rate of discoveries to increase even when the introduction rate is constant or zero (Costello and Solow 2003). Statistical methods are available for estimating and correcting these biases (Solow and Costello 2004), but the authors did not report such an analysis.
Moreover, even under the assumption that there is no lag between introduction and discovery, the conclusion that BWE is associated with accelerated introductions is not supported. Visual inspection of the discovery record suggests that if there is a break point separating two periods that differ in discovery rate, it predates the 1989 policy. Using break-point regression (Muggeo 2003), we confirmed that the true division is the year 1982 (± 0.955 standard error), well before the institution of BWE. Indeed, this date even precedes the discovery of zebra mussels (Hebert et al. 1989)—the immediate impetus for the recent increase in research on aquatic species invasions—which suggests that it is unlikely that this acceleration of discoveries is the result of investigation bias.
Using the statistically determined break point, we estimated that the discovery rate after 1982 was 0.92 per year (± 0.065, 95 percent confidence interval), approximately triple the rate during the period before 1982, which was 0.31 per year (± 0.069). This break-point model explains the vast majority of the variation in cumulative discoveries (adjusted R2 = 0.9902). Thus we infer that the discovery rate (though not necessarily the introduction rate) of invasive species has probably accelerated, but that this is most likely unrelated to BWE.
