Appendix A. Description of mark–recapture methods used to estimate mortality rates in Brachyrhaphis rhabdophora populations.
To estimate mortality rates and transitional growth rates in Brachyrhpahis rhabdophora we used a serial mark-recapture design with a sampling protocol that covered a 4-wk period in the dry season of 2000 (January and February). At each location, we selected a single focal pool and, with the aid of two field assistants, attempted to collect all fish in the pool by repeated seining (typically 40–50 seine hauls). Individuals were collected early in the morning and held in a stream-side tank. Each fish was then anesthetized in MS-222, measured, sexed, and marked with two sub-cutaneous injections of latex paint (suspended in fish Ringer’s solution) in the caudal peduncle. By using combinations of six different paint colors, we were able to individually identify each marked fish. Following marking, individuals were allowed to recover in a holding tank before being released back to their natural pool at the end of the same day. Marking mortalities were extremely rare (<1%) and marked fish held under controlled conditions through the duration of the experiment did not lose their marks for any of the colors used.
To account for the potential impact of migration (in or out of the pool) or incomplete sampling on recapture rates, we used a serial-mark recapture design (Lebreton et al. 1992). This required returning to the sampling pool every 7 d over a 4-wk time span and again collecting all possible fish from the pool. Previously marked individuals were re-measured and held in the recovery pool. Unmarked fish were sexed, measured, and given a new identification mark before being placed in the recovery pool. At the end of the day, all fish were released back to the pool where they were collected. This protocol was repeated each week until the end of the experiment, resulting in an individualized recapture history for each fish in the study. In total, 311 fish were marked in Quebrada Grande and 350 fish were marked in Rio Javilla. The advantage of this approach is that recapture histories can be used to generate estimates of survival rates (by sex or size class) that account for the potential impact of migration or inefficient seining (Lebreton et al. 1992). This gives us more confidence in our survival estimates than if fish were marked, released, and recaptured just once. An additional advantage is that this protocol allowed us to track individual growth rates of fish over four weeks, providing an estimate of the proportion of fish growing into new size classes each generation.
In this study we focus on females, given the assumption that female gestation and interbrood interval constrain recruitment (Caswell 2001). We divided females into five size classes that correspond to five stages in the B. rhabdophora life cycle (see Fig. A1 below). The first stage is composed of newborn fish <18 mm in standard length (SL); the juvenile 1 stage is composed of small individuals (18 mm to <24 mm); the juvenile 2 stage consists of individuals of intermediate size (24 mm to <30 mm) that are approaching sexual maturity; adult 1 stage contains small adult females (30 mm to <36 mm); and adult 2 contains large adult females (> 36mm) with the highest fecundity. We used these size classes to estimate mortality rates, and as the basis for calculating growth rates between stages and stage-specific fecundity estimates.
Mortality rates were estimated using a model selection framework implemented in the software program MARK (White and Burnham 1999). In brief, a recapture record was created for each fish in the study, with five recapture events, equally spaced at 1-wk intervals (see above). We assigned each captured fish to one of the five life-cycle stages defined above, allowing us to estimate size-specific mortality rates for each of the two focal populations. Mortality rates were derived from model-averaged estimates taken from 15 distinct mark–recapture models (as per MARK; White and Burnham 1999). These models included both re-capture and survival parameters. The models also vary in their assumptions about whether or not survival and recapture probabilities between re-capture events were constant or time-dependent (Lebreton et al. 1992, White and Burnham 1999). Hence, the model-weighted survival estimates consider the support from each competing model, thereby providing a more robust estimate than considering any single model alone (Johnson and Omland 2004).
|FIG. A1. Life cycle graph of Brachyrhaphis rhabdophora. Nodes represent five ontogenetic stages that individuals progress through from birth to the end of life (defined in text). Solid arrows represent transitions from one stage to another or stasis in a particular stage (G = survival with progression to a larger size class, S = survival staying in the same size class). Dashed arrows represent reproduction (F = average fecundity of females in that stage). Subscripts identify size ontogenetic stages as follows: 1 = newborns; 2 = small juvenile; 3 = large juvenile; 4 = small adult; and 5 = large adult. The decimal subscripts on G1.1 and G2.1 represent the growth of newborn individuals into the small juvenile (juvenile 1) and large juvenile (juvenile 2) stages, respectively. Note that only in Grande population did newborns grow into juvenile 2 during a single iteration of the cycle.|
Caswell, H. 2001. Matrix population models. Construction, analysis, and interpretation. Sinauer Associates, Sunderland, Massachusetts, USA.
Johnson, J. B., and K. S. Omland. 2004. Model selection in ecology and evolution. Trends in Ecology and Evolution 19:101–108.
Lebreton, J. D., K. P. Burnham, J. Clobert., and D. R. Anderson. 1992. Modeling survival and testing biological hypothesis using marked animals. A unified approach with case studies. Ecological Monographs 62:67–118.
White, G. C., and K. P. Burnham. 1999. Program MARK: Survival estimation from populations of marked animals. Bird Study 46 Supplement:120–138.