One of the most popular models for simulating the effect of regulated streamflow on fish habitat is the Instream Flow Incremental Methodology (IFIM). This model was developed by the Instream Flow Group (IFG) and is composed of components which simulate water temperature, water quality, and physical habitat. The physical habitat component (PHABSIM), however, is so frequently the only p art of the method used that PHABSIM and IFIM are often confused.
PHABSIM (and therefore IFIM) is based on several assumptions which are not always met. Tests relating fish biomass to amounts of habitat based on these assumptions have given inconsistent results. Because IFIM has become so widely used for water management decisions (it is now a legal requirement in Idaho, California, Colorado, Washington, and Oregon and is receiving increasing use in Canada and New Zealand) there is a danger of extensive damage to natural resources if its predictions are inaccurate. The purpose of our paper is to examine the validity of PHABSIM's assumptions, to review the accuracy of PHABSIM's predictions, and to discuss other methods that have been used for determining instream flow needs in New Zealand.

The extensive ecological degradation and loss of biological diversity resulting from river exploitation is eliciting widespread concern for conservation and restoration of healthy river ecosystems. Current management approaches often fail to recognize the fundamental scientific principle that the integrity of flowing water systems depends largely on their natural dynamic character, i.e. the natural flow regime. This natural flow regime varies on a wide range of time scales and includes five critical components: magnitude, frequency, duration, timing, and rate of change of hydrologic conditions. Together, these components can characterize the entire range of flows and specific hydrologic phenomena. Human alteration of the natural hydrologic processes disrupts the dynamic equilibrium of a river. Dams are among the most obvious and prevalent disruptions. Effects of this disruption are not only physical, but ecological in nature. Recent management approaches have taken into account the flow needs of only a few specific economically or recreationally important species. Furthermore, these approaches have focused solely on "minimum flow" requirements for a given system. The authors argue that implementation of a more sophisticated assessment of the needs of the river system, in the form of the Instream Flow Incremental Methodology (IFIM), will allow for management plans to move towards a natural flow regime that is beneficial for the entire ecosystem, not simply high profile species. Examples and suggestions as to implementation of these types of management strategies are given.

A review and reanalysis of the published literature show that several assumptions are violated in the application of the Instream Flow Incremental Methodology (IFIM) without consideration of the implications of so doing. The fundamental assumption of a positive linear relationship between "potential available habitat" (WUA) and biomass of fish has neither been documented nor validated, particularly in warmwater streams. Absence of correlation precludes prediction of changes in fish populations. In some studies the test of this assumption appears to be equivalent to a calibration operation. The other assumption violated includes independent selection of habitat variables by fish. The presence of significant interaction among habitat variables can affect the stream flow recommendations. Another problem exists in application of Physical Habitat Simulation (PHABSIM): one WUA unit should not be interpreted as being equal to another in biological production or habitat value unless shown to be an exact replica. Several combinations of physical variables could give rise to the same amount of WUA, none of which may be correlated to the biomass of fish. The utilization, suitability, or preference curves should not be treated as probability functions; a rating of 1.0 is not equivalent to probability of 1.0. Care should be taken to distinguish between real behavioral preferences of fishes based on distributional occurrence from abundance (relative or absolute size) in a stream.