Get access

Rivers as islands: determinants of the distribution of Andean astroblepid catfishes

Authors

  • S. A. Schaefer,

    Corresponding author
    1. American Museum of Natural History, Division of Vertebrate Zoology, Central Park West at 79th St, New York, NY 10024, U.S.A.
      Tel.: +1 212 769 5652; email: schaefer@amnh.org
    Search for more papers by this author
  • J. Arroyave

    1. American Museum of Natural History, Division of Vertebrate Zoology, Central Park West at 79th St, New York, NY 10024, U.S.A.
    2. Department of Biology, The Graduate School and University Center, The City University of New York, 365 Fifth Avenue, New York, NY 10016, U.S.A.
    Search for more papers by this author

Tel.: +1 212 769 5652; email: schaefer@amnh.org

Abstract

The distribution of astroblepids was examined using predictive niche modelling techniques to explore the physical and environmental factors responsible for determining the limits to their geographic distribution in the tropical Andes. Astroblepids occur in streams across a wide range of elevations (100–4600 m) and ecosystems from Panama to Bolivia, with most occurrences between 500 and 2000 m and associated with a narrow range of mean temperatures (17–24° C). The Maxent-predicted distribution was 83% accurate, statistically significant (AUC = 0·965), and closely matched the known distribution, with few notable exceptions. Four environmental variables contributed a cumulative 84% to the prediction, with elevation the most important, followed by temperature seasonality, isothermality and maximum temperature of the warmest month. The greatest discrepancy between the predicted and known distributions involved areas of predicted habitat suitability where there are no associated occurrence records. A jackknife test of variable contribution to the model showed that elevation contributed to the predicted distribution in ways not simply accounted for by temperature. Contrary to expectations, land cover type and vegetation characteristics contributed relatively little to the model prediction.

Get access to the full text of this article

Ancillary