16 April 2009

AOP Effectiveness monitoring biology discussion

Questions we addressed

  • What kinds of scenarios will we encounter in attempting to implement protocols to evaluate the effectiveness of passage restoration or impairment?
  • How do these scenarios dictate the design (comparative, response)?
  • How do these scenarios dictate the analysis and interpretation of outcomes?

Variables influencing scenarios

  1. Position of barrier within a stream network with respect to the following:
  2. Habitat changes – lack of comparability between above-below locations or to reference sites
  3. Proximity to confluence, e.g.,
  4. Barrier on a tributary just upstream of confluence with larger mainstem
  5. Major tributary confluence just above or below barrier
  6. Other major change in habitat above or below, or presence of nonnative species
  7. Presence of other movement barriers above or below
  8. Outcomes determined by above (examples)
  9. Above-below comparison
  10. Above – below barrier
  11. Before-after comparison
  12. Before – after barrier removed
  13. Between-stream comparison
  14. Barrier – no barrier (all else equal)
  15. Barrier removed – no barrier (all else equal)
  1. Presence of fish relative to the barrier or restored location
  2. Possible states
  3. Presence known before barrier restored
  4. Presence known after barrier restored
  5. Presence known before and after
  6. Presence not known
  7. Outcomes determined by above
  8. Choice of response to measure
  9. Occupancy (not needed if presence known)
  10. Genetic diversity
  11. Movement
  12. Retrospective study (single sampling)
  13. Before-after (more than one sampling)
  1. Variables influencing responses (covariates)
  2. Habitat size
  3. Degree of passage impairment
  4. Time since restoration or barrier
  5. Type of movement/season sampled
  6. Presence of nonnatives, source connectivity, stocking records
  7. Etc.

Table 1. Four different study designs based on a simple classification of scenarios involving only pairwise comparisons. Other designs are possible, including before-after, control-impact, or multiple treatments (e.g., natural stream, restored passage, barrier in place, degree of passage impairment, natural barrier, etc.).

Location / Timing
Above vs. reference / Before-after
Above vs. reference / Retrospective
Above vs. below / Before-after
Above vs. below / Retrospective

Each different scenario will have a different analysis and interpretation. We agreed that it is not possible to combine data from different scenarios into a common analysis. Aggregating paired comparisons for data collected under a common scenario could be useful for a “case-control” type of statistical analysis (e.g., logistic regression) or other suitable statistical approach. This, in addition to adjusting for the effects of covariates (see above) would allow for assessing differences in responses between sites or before and after imposition of treatments (e.g., installation or removal of a fish passage barrier).

In terms of responses, the same protocols could be used to measure different responses (e.g., genetics, movement, occupancy) and covariates (see above, but list not complete) for each scenario.