The objectives of this study were to measure injuries and impairments directly observed from gags Mycteroperca microlepis caught and released within a large-scale recreational fishery, develop methods that may be used to rapidly assess the condition of reef fish discards, and estimate the total portion of discards in the fishery that suffer latent mortality. Fishery observers were placed on for-hire charter and headboat vessels operating in the Gulf of Mexico from June 2009 through December 2012 to directly observe reef fishes as they were caught by recreational anglers fishing with hook-and-line gear. Fish that were not retained by anglers were inspected and marked with conventional tags prior to release. Fish were released in multiple regions over a large geographic area throughout the year and over multiple years. The majority of recaptured fish were reported by recreational and commercial fishers, and fishing effort fluctuated both spatially and temporally over the course of this study in response to changes in recreational harvest restrictions and the Deepwater Horizon oil spill. Therefore, it could not be assumed that encounter probabilities were equal for all individual tagged fish in the population. Fish size and capture depth when fish were initially caught-and-released also varied among individuals in the study and potentially influenced recapture reporting probabilities. The Cox proportional hazards regression model was used to control for potential covariates on both the occurrence and timing of recapture reporting events so that relative survival among fish released in various conditions could be compared. A total of 3,954 gags were observed in this study, and the majority (77.26%) were released in good condition (condition category 1), defined as fish that immediately submerged without assistance from venting and had not suffered internal injuries from embedded hooks or visible damage to the gills. However, compared to gags caught in shallower depths, a greater proportion of gags caught and released from depths deeper than 30 meters were in fair or poor condition. Relative survival was significantly reduced (alpha (underline)<(/underline)0.05) for gags released in fair and poor condition after controlling for variable mark-recapture reporting rates for different sized discards among regions and across months and years when individual fish were initially captured, tagged and released. Gags released within the recreational fishery in fair and poor condition were 66.4% (95% C.I. 46.9 to 94.0%) and 50.6% (26.2 to 97.8%) as likely to be recaptured, respectively, as gags released in good condition. Overall discard mortality was calculated for gags released in all condition categories at ten meter depth intervals. There was a significant linear increase in estimated mortality from less than 15% (range of uncertainty, 0.1-25.2%) in shallow depths up to 30 meters, to 35.6% (5.6-55.7%) at depths greater than 70 meters (p < 0.001, R2 = 0.917). This analysis demonstrated the utility of the proportional hazards regression model for controlling for potential covariates on both the occurrence and timing of recapture events in a large-scale mark-recapture study and for detecting significant differences in the relative survival of fish released in various conditions measured under highly variable conditions within a large-scale fishery.

成比例危險迴歸模式常被用於分析存活資料，Weibull模式更是其中惟一兼具加速失敗特性者。本論文將利用兩種分析方法，以研究任意設限及截斷資料的Weibull迴歸模式。第一種方法是利用最大概似估計法求算設限及截斷資料下的參數估計值（MLE），第二種方法則是對左設限及區間設限分別以所在區間之中點代入，稱其為中點估計法，再求算模式中的參數估計值（MDE）。並對此兩種估計方法進行比較。模擬結果顯示，相當地大樣本之下，最大概似估計法在許多情況均優於中點估計法；而在樣本少、危險率為平穩或接近平穩且區間設限比率約為0.5時，中點估計法是可被推薦的。而且，本論文亦提出對設限及截斷資料的Weibull模式之適合度檢驗程序。 / The proportional hazards regression model is most commonly used model for lifetime data. The Weibull model is the only parametric model which has both a proportional hazards representation and an accelerated failure-time representation. This paper studies the use of a Weibull-based proportional hazards regression model when any censored and truncated data are observed. Two alternative methods of analysis are considered. First, the maximum likelihood estimates（MLEs） of parameters are computed for the observed censoring and truncation pattern. Second, the estimates where midpoints are substituted for left- and interval-censored data（midpoint estimation, MDE）are computed. Then, MLEs are compared with MDEs. Simulation studies indicate that for relative large samples there are many instances when the MLE is superior to the MDE. For small samples where the hazard rate is flat or nearly so, and the percentage of interval-censored data is nearly half of samples, the MDE is adequate. Also, an evaluation of the adequacy of the Weibull model for any censored and truncated data is proposed.