SPATIO-TEMPORAL VARIATION IN FISH SPECIES RICHNESS OF THE UPPER
MISSISSIPPI RIVER SYSTEM BY RAREFACTION.

Todd M. Koel1 and Michelle M. Cripps1,2.  
1Minnesota Department of Natural Resources, Mississippi Monitoring Station, Lake City, MN
55041  2River Studies Center, University of Wisconsin-La Crosse, La Crosse, WI 54601.

The Long Term Resource Monitoring Program (LTRMP) has documented spatial and temporal
trends in fish communities of the Upper Mississippi River System (UMRS) for over a decade.  A
metric based on these communities that can be useful for understanding the status of this system's
health is species richness.  A problem typically encountered when comparing species richness
among sampling units is that values are greatly affected by both sample size (total number of
individuals collected) and spatial area covered.  We used rarefaction methodology to calculate
expected species richness values at standardized sample sizes E(Sn) because for most gear types
the number of LTRMP samples taken and individuals collected has varied among aquatic habitat
areas, trend areas (river pools), seasonal time periods, and years.  Spatially-randomized fish
collections, stratified by natural aquatic/geomorphic habitat, were completed in all six LTRMP
trend areas.  A standardized total of 12 electrofishing, 18 fyke-netting, and 24 hoop-netting
collections were randomly selected for rarefaction from each trend area, habitat, and year,
1994-1999.  

At the pool scale, no difference was noted in mean E(S460) by electrofishing among contiguous
backwater shorelines, main channel borders, and side channel borders within pools 4, 13, 26, or
the open (non-pooled) Mississippi River (P>0.2).  However, mean E(S460) values of main
channel borders of pool 8 and La Grange Reach of the Illinois River were significantly lower
than that of other habitat types within each trend area (P<0.05).  At the system scale, mean
E(S460) values by electrofishing were significantly higher in contiguous backwater shorelines,
main channel borders, and side channel borders of pools 4, 8 and 13 than of pool 26, the open
river, or La Grange Reach (P<0.0001).  Mean E(S460) by electrofishing in contiguous
backwaters ranged from 31 in pool 8 to 21 in pool 26; in main channel borders it ranged from 28
in pool 13 to 14 in La Grange Reach; and in side channel borders it ranged from 32 in pool 8 to
20 in La Grange Reach.  Similar trends were noted by fyke-netting in contiguous backwaters and
by hoop-netting in main channel and side channel borders, although total numbers of fish
collected and mean E(Sn) were lower.  Systemically, the LTRMP collections indicated that pools
4, 8 and 13 consistently had the highest expected fish species richness.  This trend may be
attributed to low habitat heterogeneity and predictability of pool 26, the open river, and La
Grange Reach.

Keywords: large river fishes, standardized diversity metric, aquatic habitat comparisons