Abundance Trends and Status of the Little Colorado River Population of Humpback Chub: An Update Considering Data From 1989-2008

Abstract

Mark-recapture methods have been used for the past two decades to assess trends in adult abundance and recruitment of the Little Colorado River (LCR) population of humpback chub. These methods indicate that the adult population declined through the 1980s and early 1990s but has been increasing for the past decade. Recruitment appears also to have increased, particularly in the 2003-4 period. Considering a range of assumed natural mortality-rates and magnitude of ageing error, it is unlikely that there are currently less than 6,000 adults or more than 10,000 adults. Our best estimate of the current adult (age 4 years or more) population is approximately 7,650 fish. Recent humpback chub assessments using the Age-Structured Mark-Recapture model (ASMR) and reported in 2006 (Melis and others, 2006) and 2008 (Coggins, 2008a,b) have provided abundance and recruitment trend estimates that have changed progressively over time as more data are considered by the model. The general pattern of change implies a less severe decline in adult abundance during the late 1980s through early 1990s, with attendant changes in recruitment supporting this demographic pattern. We have been concerned that these changes are not indicative of the true population and may be associated with a 'retrospective' bias as additional data are included in the ASMR model. To investigate this possibility, we developed a realistic individual-based simulation model (IBM) to generate replicate artificial data sets with similar characteristics to the true humpback chub data. The artificial data have known abundance trends and we analyzed these data with ASMR. On the basis of these simulations, we believe that errors in assigning age (and therefore brood-year) to fish based on their length are likely to have caused the retrospective bias pattern seen in the assessments and to have caused both less severe trends in the adult abundance estimates and progressively more severe downward bias in estimates of adult mortality-rates. This 'smearing', or assignment of fish from a single brood-year into multiple incorrect brood-years, is a result of variation in growth rates. The IBM simulations indicate that as a result of this error source, the best estimates of abundance and recruitment for any calendar year are those obtained from data collected previous to and within a year or two after each calendar year.