The mangrove rivulus fish (Kryptolebias marmoratus) is an emerging model for research in the biological sciences. The species is androdioecious, with populations consisting predominantly of self-fertilizing hermaphrodites and a low abundance of males, but no females. Males arise either through environmental sex determination at the embryonic stage or environmentally driven sex change from hermaphrodite at the adult stage. Marked sexual dimorphisms in external morphological characters accompany gonadal transitions, including the loss of a caudal ocellus ('eyespot') and the development of orange color in males. How well external morphological features predict internal gonadal morphology, however, has not been systematically examined across the geographical range of this species. Our goal was to identify reliable external morphological predictors of animals possessing an ovotestis (hermaphrodite) or a testis (male) and to see if the strength of prediction is consistent throughout rivulus' geographical range. We demonstrate that the state of the ocellus and, to a greater extent, orange coloration are key predictors of internal gonadal state. However, the degree to which these characteristics are predictive of internal gonadal state was region-specific, with the ocellus and orange status of animals from Bahamas and Belize, respectively, having a lower probability of forecasting internal gonadal state than those from other regions. These results will help to streamline laboratory studies by allowing unambiguous assignment of sex, and will assist field studies by providing greater opportunity to determine sex ratios in wild populations.

This study explored whether convict cichlid fish mount a hormonal response to aggressive encounters where dominance status remains unresolved. Hormone samples were collected at two time points before an aggressive interaction to obtain confinement-induced and baseline measures, and at one time point following a contest across a clear partition (experimental) or exposure to an opaque partition with an opponent on the opposite side (control). There was no overall significant effect of treatment (control vs. experimental) on hormone release rates but there were trends for cortisol and testosterone (T). A priori linear contrasts showed that individuals that engaged in aggressive interactions had lower postfight cortisol and T release rates than controls, suggesting that aggression, in this context, might attenuate the synthesis of both hormones. Cortisol decreased significantly between initial confinement and baseline, indicating that individuals habituate to the water-borne hormone collection procedure. Contrary to expectation, individuals with higher baseline T and 11-ketotestosterone (KT) release rates took longer to initiate conflict. None of the other measures of behavior were predicted by baseline hormone release rates, and contest behavior did not predict postfight hormone release rates. There was a significant positive relationship between KT and T at all time points. As with studies that employ mirror image stimulation, we found no hormonal response to unresolved contests despite high levels of aggressive behavior. Our study is unique because we demonstrate that animals engaged in conflict with live opponents also do not mount a significant hormonal response when clear dominance relationships are not established.