**Background:** The Delta Smelt (Hypomesus transpacificus) is an endangered osmerid fish endemic to the San Francisco Estuary, with wild populations at or below detection limits of monitoring programs. A captive breeding program has existed since 1996, but it was unknown whether captive-reared individuals could survive, feed, and maintain condition outside hatchery conditions. This study evaluated the use of in situ enclosures as a tool to test captive-reared Delta Smelt under semi-natural conditions and to inform future supplementation efforts.
**Methods:** In winter/spring 2019, captive-reared adult Delta Smelt (206–278 days post hatch) were deployed for four weeks at two sites: the Sacramento River near Rio Vista (RV) and the Sacramento River Deepwater Shipping Channel (DWSC). Three enclosure designs were tested: Design A (41% open perforated steel), Design B (63% open), and Design C (63% open with external woven mesh wrap on lower half). At RV, 64 fish were placed in each of 6 enclosures (2 replicates per design); at DWSC, 60 fish per enclosure. Fish were tagged with Visible Implant Alphanumeric tags, weighed, and measured before and after deployment. Survival was monitored daily. Zooplankton samples were collected weekly adjacent to enclosures. At study end, stomach contents from 10 fish per enclosure (60 per site) were analyzed. Statistical analyses included t-tests for pre/post comparisons, linear mixed-effects models for growth by enclosure type, and ANOVA for instantaneous ration and zooplankton biomass.
**Key Results:** Survival was high across all enclosure types and sites: at RV, mean survival was 96.4% ± 0.8% (range 87.5–100%); at DWSC, 98.6% ± 0.3% (range 96.7–100%). At RV, fish gained weight (mean change +0.197 ± 0.006 g, +15.9%; t = -34.791, df = 597.72, p < 2.2e-16), increased fork length (+0.164 ± 0.011 cm, +2.9%; t = -4.5123, df = 684.31, p = 7.547e-06), and increased condition factor K (+0.040 ± 0.004 g; t = -6.8555, df = 687.21, p = 1.6e-11). At DWSC, fish lost weight (mean change -0.248 ± 0.009 g, -15.9%; t = -14.814, df = 493.77, p < 2.2e-16) and decreased in K (-0.111 ± 0.004; t = 11.802, df = 366.76, p < 2.2e-16), with no change in fork length. There was no significant effect of enclosure type on weight change at either site (RV: ANOVA F = 0.2232, df = 2, p = 0.8122; DWSC: ANOVA F = 0.5411, df = 2, p = 0.583). Instantaneous ration differed across enclosure types at RV (ANOVA, F₂,₅₇ = 4.25, p = 0.019) but not at DWSC (ANOVA, F₂,₅₇ = 2.53, p = 0.089). Diet analysis showed fish consumed wild zooplankton: at RV, diet was dominated by cladocerans (79.1%, primarily Daphnia) and cyclopoids (16.6%); at DWSC, fish consumed cladocerans (47.8%, primarily Chydorus sp.), cyclopoids (39.1%, primarily Acanthocyclops sp.), and calanoids (10.0%). Zooplankton biomass was significantly higher at RV than DWSC (ANOVA, F₁,₆ = 14.67, p = 0.009). Water velocity reduction by enclosures averaged 70.68% for Design A, 43.98% for Design B, and 50.47% for Design C.
**Clinical Implications:** This study provides the first evidence that captive-reared Delta Smelt can survive and forage successfully under semi-natural conditions in the wild, with survival rates far exceeding the authors' initial expectations of low survival. The enclosures represent a new tool for testing habitat management actions and for acclimating fish prior to release (soft release strategy) for supplementation programs. The contrasting results between sites (weight gain at RV vs. weight loss at DWSC) highlight the importance of site selection, with floodplain-influenced locations like Rio Vista (downstream of Yolo Bypass) providing better food resources. The lack of significant differences among enclosure designs suggests that a range of mesh sizes can support Delta Smelt, though the larger mesh (Design B) showed modest foraging benefits at one site. These findings support ongoing supplementation efforts for Delta Smelt and provide a methodological framework applicable to other sensitive endangered fish species.