No abstract

No abstract

The barnacle Megabalanus coccopoma is native to shorelines from Baja California to Peru and has been introduced to a number of other locations including the Atlantic US SE coast, where it was first recorded in 2006. In 2009, the range of M. coccopoma in the SE US extended from Ft. Pierce, FL north to Cape Hatteras, NC with seasonal populations found as far north as Kitty Hawk, NC. During the exceptionally cold winter of 2009/2010, the range of M. coccopoma shifted dramatically due to the dieback of all monitored populations north of Florida. We examined body size, distribution, and density of M. coccopoma during the summers of 2010, 2011, and 2012 to describe the extent of the range retraction and the rate of range re-expansion. In 2010, recruits were found as far north as Tybee Island, Ga, but no established populations were found north of Florida. In 2011 recruits were found at Rodanthe, NC but established populations were still limited to Florida. By 2012 populations were established in Rodanthe, NC, slightly north of its previously known range limit. Estimated rates of range re-expansion were 255.8 km/yr in 2010 and 794.1 km/yr in 2011. Rates of re-expansion to the north in 2010 and 2011 were faster than have previously been reported for any marine species, and are one of the few rates published for any tropical marine invertebrate.

A major goal of invasion biology is to predict range shifts and potential range limits of nonnative species. Species distribution models (SDMs) are commonly used to achieve these goals, but the predictive accuracy of these models is rarely tested using natural range shifts. The barnacle Megabalanus coccopoma is native to the eastern tropical Pacific and has been introduced to a number of locations globally including the southeastern United States. During the unusually cold winter months of 2010, the range of M. coccopoma within the USA SE retracted 825 km. The ability of the SDM MaxEnt to accurately predict the range retraction and M. coccopoma's range within the USA SE was tested using distributional data from before and after the range retraction. Three MaxEnt models were trained using data from the global range, the native range, and the USA SE introduced range.Only the model trained on data from the USA SE was able to accurately predict the entire extent of the range retraction and most known populations prior to the range retraction. Globally trained models may provide the most conservative estimates of potential distributions; however, niche shifts may limit the ability of these models to accurately predict range shifts.