WSP12173 - Laserfiche WebLink

Home Browse Search

o '-, ~.... -..J ~ ,... Platania. Dudley, and Maruca. 2000. Drift of Fishes in the San Juan River 1991.1997. FINAL (as previously described). However, flow was not measured consistently throughout the seven year sampling period and there were also occasional problems with individual flow-meter readings. Thc meters would sometimes appear to be operating "sluggishly" prcsumably because sand or silt had infiltrated the internal mechanism and hindered the movement of gears. Procedures for detecting faulty or inaccurate flow-meter readings and for estimating flow were devised to remedy these individual situations. For each year and site where flow was measured, flow-meter data were examined for missing or unusually low mcter readings. Low readings were confirmed as defective if the original field sheets and notes contained corroborative information such as the collectors' observations on flow- meter condition or tile state of the river. Drift-net sets with faulty or missing flow readings that also lacked information on the duration of the sampling period were removed (e.g., 1-10 July 1992 at the downstream Four Comers location). Samples that persisted for more than four hours were also removed from the CPUE dalaSet because the accumulation of debris in the net negatively affected filtering efficiency. This included six samples from the downstream location and one set from the upstream location for Four Corners in 1993. New flow-meter valnes were estimated, whenever possible, from temporally proximate sets condncted at the same sampling location under the same flow conditions. From these adjacent samples, a per-minute average of flow-meter revolutions was calculatcd and subsequently multiplied by the number of minutes (for the set in question) to obtain an estimated value of flow. On rare occasions, flow-meter readings were faulty for a week or more (c.g.. 31 July-t4 August 1993 at the upstream Four Comcrs location). The estimates of the volume ofwatcr sampled during those periods are less reliable than estimates for individual samples We attempted to develop general empirical relationships among volume of water sampled, length of time sampled and mean daily discharge. This exercise was performed with the goal of developing an equation for estimating daily flow readings for years/sites lacking flow measures (1991 Four Comers; 1993 Mexican Hat; 1994 four Corners and Mexican Hat). Regression analysis was used to evaluate the possibility of employing two models, ltnear and logarithmic (e.g., see Johnston et aI., 1995) to determine relationships between the proportion of flow sampled (PF) and the proportion oftime (out of 24 hours, PT) sampled daily. The value PF is equal to the volume of water sampled divided by total volnme of water available and was determined from mean discharge. A second independent variable, mean daily discharge, Q (or, for the logarithmic model, In[Q]), was added to these models after noting that an increase in discharge may decrease the proportion of flow sampled per unit time. During periods of high flow, the cross-sectional area of the river generally increased but drift-net size remained the same. Furthermore, collectors frequently relocated drift-nets to rnesohabitats with a lower-water velocity in an effort to maximize net efficiency. This relocation usually resulted ill the drift-nets set closer to the shore and away from the thalweg. We tested four linear models, combining data for all years and sites where volume of water sampled had been measured. All four relationships were significant (i.e. p < 0.05); however, significance was due primarily to large sample size (n = 379) and not strength of relationship. In fact, R' values did not exceed 0.16, except for the logarithmic relationship that included both proportion time sampled and average discharge (R' = 0.70). This R' value was higher than expected and suggested that a eonsiderable amount of variability in the amount of discharge being sampled can be explained by measured parameters. This resulted in the decision not to estimate flow on a daily basis for the five sampling periods that Jacked flow measuremellts, but rather to estimate volume of water sampled during the entire individual sampling periods (n=5). To do this, we tested the same four statistical models described above, using year/site totals for volume of water sampled and time sampled. In addition, we also examined the relationship between volume of water sampled and time sampled (i.e., not proportions). tl