|
' he Instream Flow Incremental Methodology (IFlh11 was
<br />developed by the U.S. Fish and Wildlife Service to pro-
<br />vide a standard analytical technique for recommending
<br />flows for a stream. Originally, IFiM was developed for
<br />application in small simple cold-water stream systems from
<br />v: hick water was to be diverted for off stream consumptive uses
<br />or to be allocated for other development projects. One of the
<br />initial objectives of the method was to assess with relative ease
<br />changes in fish standing crop and species composition due to
<br />changes in stream flow (Bovee 1978)..Several assumptions are
<br />made in utilizing the method in a given stream. The purpose of
<br />Our note is to critique some of these assumptions.
<br />Only recently has the applicability of IFIM been investigated
<br />in a relatively small wannwater stream (Orth and Maughan
<br />1982). These authors provided sufficient data in the published
<br />literature for our reanalysis. We will therefore use these data and
<br />other published examples for the evaluation of the underlying
<br />asssumptions. Undoubtedly. results of many other IFINI studies
<br />exist in unpublished reports. nonrefereed papers, proceedings,
<br />etc.
<br />Theorn• and Afechanic•s of Calculations
<br />An important integral component of IFIM is PIIABSIM
<br />(Physical Habitat Simulation). The PHABSIM procedure con-
<br />tains four primary components: (1) physical measurements of
<br />depth, velocity, substrate., and cover within the stream reach,
<br />(2) computer simulation of the stream hydraulics, (3) deter-
<br />mination of a composite probability of use from the suitability
<br />value for each combination of depth, velocity, and substrate
<br />found within the stream reach, for each species and life history
<br />phase, and (4) the calculation of weighted usable area (\1'UA)
<br />for each stream flow, species, and life history phase for each
<br />season. Other factors such as water quality and food can be
<br />also incorporated in the calculation of WUA but the level of
<br />complexity in application and interpretation increases substan-
<br />tially. Biological interactions (competition, predation, etc.) are
<br />recognized as important factors but at present cannot be
<br />included in the application of the method.
<br />The outputs of computer simulations of physical habitat
<br />variables (based on measurements of depth, velocity, substrate,
<br />cover, etc.) in a stream reach and species life stage "probability
<br />of use" or "preference" or "suitability" curves (based on instan-
<br />I,meous fish measurements in the field, expert opinion, or from
<br />literature sources) are integrated into a potential available NVUA
<br />for each flow. A relationship is then established between the
<br />availability of potential WUA for each life history phase of a
<br />species and stream flow in each season or time period.
<br />Four basic assumptions are as follows: (1) depth, velocity,
<br />and substrate are the most important physical habitat variables
<br />affecting the distribution and abundance of fishes; b0im ioral
<br />preference of a life stage of a species for each physical varinhie
<br />can be established from instantaneous fish mcasurcnrcnrs in the
<br />field and probability of use curves or suitability or utilii.alion
<br />indices constructed; (2) depth, velocity, and substrate indepen-
<br />dently influence habitat selection by fishes; (3) preference
<br />factors for depth, velocity, and substrate, etc., can be combined
<br />through multiplication to create WUA index; and (4) a positive
<br />linear relationship exists between size of WUA and the biomass
<br />of fishes: a slope of I is assumed to relate biomass and \VUA
<br />(Bovee 1978). Presumably, an increase in the WUA will result
<br />in increased fish populations because populations are implicitly
<br />assumed to always be habitat limited.
<br />926
<br />The WUA (in optimum habitat equi%alents). based on the
<br />composite suitability of three principal habitat varia5les, is
<br />derived for each stream reach from instantcureous measurement,
<br />as follows:
<br />WUA = C; A;
<br />where Ci = f'(V,) x f(Di) . f; (S,), f,:(Vi) = nritabilitc
<br />weighting factor for the selocit,. in cell i, f,i(D;) = suitabilil`
<br />weighting factor for the depth in cell i. f (S,) = suitabilitd
<br />weighting factor for the substrate type in cell i, )t = tile nunibe"
<br />of cells, and A; = the area of cell i. Weighting factors are de
<br />rived from "utilization" or "preference" or "suitability." "elec-
<br />tivity," or "probability of use" curses. The methocl assumes thai
<br />behavioral characteristics of a life stage of a species can by
<br />defined by these tunes. The made or the peak of the curve i,
<br />interpreted as the optimum value of a variable for fish usage [Intl
<br />is given a weighting factor of 1. i-lie tails of the tune represent
<br />zero weighting factor. Generall•:, weighting values bet•,veen t!
<br />and I are empirically determined (in equid alent optimum habitat
<br />units) from analysis of instantaneous observations of tish dis
<br />tribution over the range of each variable (Bovee 1982). A.-
<br />example of velocity and depth preference curve fur adult small
<br />mouth bass..Sulnto&e.s dalcnniew. is given in Fig. I.
<br />SuitabilitA- Irrdrx awl Prefrrcuc•t Cur vt.%
<br />The original application of IFI(\i , treatment of "suitability"
<br />"preference" cures as probability functions, led to the calculation of a joint probability function by multiplication of univari
<br />ate preference factors as simple conditional probabilities (Bovee
<br />1978, 1952). '1 his procedure is crrTect only when probabilitir
<br />are statistically independent. A transformation of unitiariat,
<br />preference factors into simple probabilities is erroneous. Fir,•
<br />the mode or peak of curves 0i(mil in Fig. I only have
<br />subjective rating of 1.0, chich i,, not cgtrivalent lo 'l I w+ahili,
<br />of 1.0. That is, the curves shculd no! nugget ch,ct d ere i
<br />100% chance (a certainty) of locating a species population
<br />specified segment of a population. A rating of 1.0 -implj mean.
<br />that most organisms were obsei vcd or captured at that deptl
<br />and/or velocity at the time of collection. The cure does m?,
<br />have any statistical distribution and cannot be considered as .
<br />probability function. The ordinate values between 0 and I.u
<br />(calculated from proportional scaling offish catches) hake beer
<br />incorrectly interpreted a-z actual probahili!ics in i IIABSINI
<br />Probability is an area under the cure and not a value of th,
<br />ordinate. We are not aware of any published study that ha
<br />addressed these statistical or mathematical distincti-ms. and }et
<br />the suitability function in the form of the joint roil iihihl
<br />function continues to be used (Bovee 1982).
<br />The ratings of (lie "preference" or "suitabilit)' curves ar:
<br />ratios. However, these ratios are haled upon a shifting denomin
<br />ator. For example, if largest number, 10, were obtained a:
<br />a particular depth and.or velocity , these variables ss ill be gis,
<br />a rating (if 1.0. If in anolhcr sa!npling the )_rcatc,?t nun!her w„
<br />100 organisms the same yariahlc. would also he given a ratio;
<br />of 1.0. Obviously. there is a ditfcrence in (he biomass of 10 all
<br />100 organisms. In our view the development of the "preference'
<br />curve as we described will lead one to expect loss correlati(
<br />between "suitability" and fish standing stock.
<br />Because fishes may respond to a multitude of factors in th
<br />field, thus manifesting daily changes in their distribution
<br />different tunes may be obtained on different sampling, dates o,
<br />times within a season. Forexample, fishes change po.ition fr(ir
<br />C.:v .1. f'r,A A.111111. S, i, ".1 . 42. 1Q
<br />
|