Abstract.-Understanding the influences of local and regional processes on the dynamics of self-sustaining trout populations would help fishery biologists better manage trout populations and protect rivers supporting trout. We explored hypotheses behind long-term temporal variation in density, growth, and survival of brown trout Salmo trutta and brook trout Salvelinus fontinalis using data collected over several decades on Michigan's Au Sable River. Regression models developed for these species emphasized the influence of yearclass strength on older age-classes, year-class strength being positively related to spawner abundance for both species and negatively related to high spring streamflow conditions for brown trout. Age-class density was also positively associated with high levels of large woody debris (LWD) in streams. Annual growth increments of brown trout and brook trout were often negatively related to increased age-class density and LWD and positively affected by elevated total phosphorus levels, cool summers, and warm winters. Annual survival of trout from age 0 to age 4 was negatively related to intra-and interspecific age-class density, and in three of seven models, positively associated with levels of LWD. Our findings emphasize the importance of year-class strength to trout population dynamics as well as the need to include collection of regional-and local-scale habitat data in studies of trout population dynamics.

Abstract.-The ability to describe regional patterns in trout density would be useful for biologists concerned with population status across large regions as well as managers of rivers at the local scale. Noting the importance of flow conditions at the time of emergence to trout year-class strength in Michigan streams and the influence of age-0 trout abundance on subsequent abundance of older age-classes, we assessed the potential for regional synchrony in the population dynamics of brown trout Salmo trutta and brook trout Salvelinus fontinalis among Michigan rivers. We used correlation analyses to look for regional synchrony in May stream discharge (approximating the time of brown trout fry emergence) and fall trout density among many Michigan trout streams. We found a high degree of synchrony in average May discharge among streams, particularly those in the northern portion of Michigan's Lower Peninsula. There were significant correlations in the long-term densities of brown trout and brook trout year-classes among several rivers in this area, including sites up to 140 km apart and rivers draining into different Great Lakes. Predicted numbers of days to 50% swim-up of brown trout fry were similar among four streams and synchronous, further supporting the hypothesis of synchrony in trout population dynamics in Michigan streams at the regional scale. Long-term trout population estimates and streamflow data collected from a network of long-term index (fixed) sites throughout Michigan will aid in further description of the spatial extent of synchrony in trout populations.

Summary 1. Fisheries models generally are based on the concept that strong density dependence exists in fish populations. Nonetheless, there are few examples of long‐term density dependence in fish populations. 2. Using an information theoretical approach (AIC) with regression analyses, we examined the explanatory power of density dependence, flow and water temperature on the per capita rate of change and growth (annual mean total length) for the whole population, adults, 1+ and young‐of‐the‐year (YOY) brook trout (Salvelinus fontinalis) in Hunt Creek, Michigan, USA, between 1951 and 2001. This time series represents one of the longest quantitative population data sets for fishes. 3. Our analysis included four data sets: (i) Pooled (1951–2001), (ii) Fished (1951–65), (iii) Unfished (1966–2001) and (iv) Temperature (1982–2001). 4. Principle component analyses of winter flow data identified a gradient between years with high mean daily winter flows, high daily maximum and minimum flows and frequent high flow events, and years with an opposite set of flow characteristics. Flows were lower during the Fished Period than during the Unfished Period. Winter temperature analyses elucidated a gradient between warm mean, warm minimum and maximum daily stream temperatures and a high number of minimum daily temperatures >6.1 °C, and years with the opposite characteristics. Summer temperature analyses contrasted years with warm summer stream temperatures vs years with cool summer stream temperatures. 5. Both YOY and adult densities varied several‐fold during the study. Regression analysis did not detect a significant linear or nonlinear stock–recruitment relationship. AIC analysis indicated that density dependence was present in 15 of 16 cases (four population segments × four data sets) for both per capita rate of increase (wi values 0.46–1.00) and growth data (wi values 0.28–0.99). The almost ubiquitous presence of density dependence in both population and growth data is concordant with results from other trout populations and other studies in Michigan.

The average hooking mortality per capture event for 630 trophy‐sized wild brook trout Salvelinus fontinalis (mean total length, 33.9 cm) caught on five hardware lures was 4.3% during the first 48 h after capture, Mortality was 8.3% for brook trout caught on Mepps spinners and Cleo spoons equipped with a treble‐pointed hook, whereas mortality was significantly lower (2.4% per hooking event; P < 0.05) for fish caught on the same lures with a single‐pointed hook. The 10.9% mortality caused by treble‐hook Mepps spinners was significantly higher than mortality caused by single‐hook Cleo spoons (1.6%). Mortality for brook trout caught on single‐hook Cleo spoons and single‐hook Mepps spinners combined (2.4%) was also significantly lower (P < 0.05) than mortality offish caught on Mepps spinners with treble hooks. There was no mortality among 126 brook trout caught with Rapala lures rigged with two treble hooks. We believe that the differences in mortality of brook trout caught with different lures are primarily attributable to differences in the frequency and extent of damage to the gill arches and esophagus area. Certain lures were more likely to be engulfed deeply, particularly by larger fish, and thus were more likely to cause death. Lures that exhibit vigorous wobbling action when retrieved appear less likely to be deeply engulfed and consequently cause less mortality. Hooking mortality estimates for brook trout caught on Mepps and Cleo lures combined were positively and significantly correlated with size of fish (P < 0.003). The probability of death within 48 h of capture for profusely bleeding brook trout that were hooked in the gills or throat increased rapidly with increasing water temperature. Fish that did not bleed profusely after capture with treble‐hooked Mepps and Cleo lures that did not penetrate the gill or throat region were unlikely to die as a result of temperature effects, unless temperatures were above approximately 14°C. The probability ofdeath was not significantly associated with temperatures ranging from 5.6 to 17.8°C when brook trout were hooked with single‐pointed hooks at anatomical sites other than the gills or throat and did not bleed heavily. Present regulations on Michigan's trophy trout lakes, which restrict lures to single‐pointed hooks and forbid harvest offish less than 38.1 cm in total length, appear quite adequate to minimize losses due to hooking mortality.

We evaluated the response of benthic macroinvertebrates in a Michigan trout stream to flow reduction by diverting water from a 602 m reach of Hunt Creek from June through August of 1994, 1997 and 1998. We also assessed the utility of the Physical Habitat Simulation system (PHABSIM) in predicting the response of benthic insects to water withdrawals by testing the assumption of a positive linear relationship between modelled habitat (weighted usable area, WUA) and the density of 13 benthic insect families. Our findings showed that the density of filter feeding and grazing insect taxa, as well as insects classified as obligate erosional zone taxa, declined significantly in the dewatered (treatment) zone (TZ) when 90% of flow was diverted. Density of Ephemeroptera, Plecoptera and Trichoptera (EPT) taxa in the TZ was significantly lower when 90% of water was diverted as compared to density at baseflow or when flow was reduced by 50%. The density of all insects in an upstream reference zone riffle (RZ), where flow was not altered, did not change among experimental periods. Although overall reductions in the density of benthic insects at 90% flow reduction coincided with lower PHABSIM predictions of WUA, we found poor linear correlation between WUA at different flows and the density of the 13 benthic insect families for which WUA was modelled. The low proportion of variation explained by WUA for all families modelled suggests that WUA alone is not an accurate predictor of benthic insect density. Resource managers should consider the potential consequences of water withdrawals to all components of stream communities, including benthic macroinvertebrates. However, caution should be applied when using the PHABSIM technique in groundwater-fed streams such as Hunt Creek, because most relationships between WUA and benthic insect density were insignificant.