Multiple stressors disrupt sex hormone and fitness outcomes: Effects of turbidity and hypoxia on an African cichlid fish

Freshwater organisms are faced with a complex array of environmental stressors that can negatively affect endocrine function and subsequent fitness outcomes. Hypoxia and turbidity are examples of two environmental stressors that are increasing due to human activities that could lead to endocrine disruption and reduced reproductive output. Our research addresses how hypoxia and elevated turbidity affect traits related to reproductive success, specifically sex hormone concentrations, investment in reproductive tissues, and body size. We used wild fish from two populations (a river and a swamp) of an African cichlid, Pseudocrenilabrus multicolor, to produce offspring that were reared in a full factorial split brood rearing experiment  (hypoxic/normoxic x clear/turbid). River and swamp populations represent divergent habitat types with respect to the stressors of interest, being well-oxygenated but turbid or hypoxic and clear, respectively. Overall, we found evidence for plastic responses to both stressors. Specifically, we found that there was an interactive effect of oxygen and turbidity on testosterone in males from both populations. Additionally, males of both populations reared under hypoxic conditions were significantly smaller in both mass and standard length than those raised under normoxic conditions, and males reared under hypoxic conditions invested less in reproductive tissues (quantified as gonadosomatic index). Hypoxia and turbidity are experienced naturally by this species and these environmental stressors did not affect the number of eggs laid by females when experienced in the absence of another stressor (i.e., normoxic/turbid or hypoxic/clear). However, there was an interactive effect of hypoxia and turbidity as females reared and maintained in this treatment combination laid fewer eggs. This research underscores the importance of considering the possibility of stressor interactions when determining how anthropogenic stressors affect fitness outcomes.

Methods
Rearing Experiment

Research was conducted under approval from The Ohio State University Institutional Animal Care and Use Committee (2014A00000055-R1). To assess the effects of hypoxia and turbidity on hormone levels and fitness metrics, we measured testosterone and estradiol in F1 P. multicolor males derived from wild-caught fish reared in a split brood rearing experiment (full factorial, hypoxic/normoxic x clear/turbid treatments). Fish used as parents in this rearing study were collected from one swamp site (Lwamunda) and one river site (Ndyabusole) in the Lake Nabugabo region of Uganda in May 2018. The swamp site is typically hypoxic (0.79 ± 0.1 mg/L, mean ± SE point in time measurements) and clear, albeit tannin-stained (2.02 ± 0.3 Nephelometric turbidity units; NTU) while the river site is typically normoxic (6.84 ± 0.2 mg/L) and moderately turbid (18.81 ± 1.1 NTU); measurements were taken between June and August in 2015, 2016, and 2017 from Oldham (2018). Ten broods from each of two populations (one river and one swamp population) were reared under each combination of rearing conditions. Once a female released her brood, the young were housed for one week under normoxic, clear conditions before being split randomly into the four different treatment groups. In the hypoxic treatment group, the oxygen was reduced gradually over a period of one week by placing bubble wrap over the surface of the water. In the turbid tanks, turbidity was gradually increased over a one week period by adding approximately 1.5 mL of bentonite clay solution (100g clay/ L water) Dissolved oxygen concentrations were measured throughout the rearing experiment (DO: hypoxic mean ± SE = 2.27 ± 0.01 mg/L O2 or normoxic mean = 7.52 ± 0.001 mg/L O2) using a YSI Pro2030 multimeter probe 3-7 days per week and was adjusted as needed by adjusting bubble wrap, and bubbling water with ambient air or nitrogen gas. Turbidity (clear: 1.31 ± 0.03 NTU or turbid: 16.6 ± 0.14 NTU) was measured 1-2 days per week using a Hach2100Q portable turbidimeter and was adjusted as needed during weekly water changes or through the addition of the bentonite clay solution. Fish were fed Tetramin Tropical crisps once a day, ad libitum for five minutes. The treatments and populations were dispersed randomly across 80 aquaria to minimize small differences in light and temperature (mean ± SE: 24.9 ± 0.02 ℃ )(Williams, 2023). We maintained fish in a 12L:12D photoperiod. All fish were mature at the time of sampling. To distinguish between fish within a tank we used white visible implant elastomer tags. We tagged 348 fish (156 males and 192 females) when they were between 345 to 386 days old.

The results reported here are part of a larger study where male and female reproductive behaviors (Williams et al., 2024) were examined using the fish described here. For additional details of the rearing experiment components see (Williams, 2023, Williams, et al., 2024) and (Tiarks, 2024). We used subsets for measures of hormones (n = 77; when males were between 466 to 514 days old) and egg count (n = 37 when females were between 615 to 794 days old). After hormone and fecundity and reproductive behavior (Williams, et al., 2024) data were collected, we sampled fish for morphological metrics (n = 255) and GSI (n = 234). The age of fish ranged from 571 to 827 at the time of final sampling due to restrictions to laboratory access during the COVID-19 pandemic.

Hormone Collection

Hormone samples were collected from a subset of adult males between June 24th, 2020, and July 5th, 2020, when fish (n = 77) were between 466 to 514 days old (fish were maintained at treatment conditions from 7 days post-release). Hormones were collected between 09:30 to 13:30 to minimize diel variation in hormone concentrations (Cowan et al., 2017). We collected hormone samples using a non-invasive method that has been previously validated in P. multicolor and other species of fish (Friesen, et al., 2012, Kidd et al., 2010). Hormone samples were collected by placing individual fish in glass containers containing 200 mL of clean treated water for one hour, during which time hormones are released from the gills at a rate that is highly correlated with the concentration in plasma (Friesen, et al., 2012). The oxygen and turbidity of the sampling water matched the rearing treatment the fish was raised in. Hormone samples were extracted onto Sep-Pak Plus C18 SPE cartridges and frozen at -20 ℃ until further processing. Hormones were extracted from the cartridges using ethyl acetate and dried under a stream of nitrogen gas. Samples were then reconstituted in assay kit buffer (582701 and 582251, Cayman Chemical) according to manufacturer instructions. To quantify background levels of hormones in the water, four water samples not containing fish were also collected, and the concentrations of testosterone and estradiol were measured to establish a baseline level. The average baseline measurement of hormones (mean ± SE = 85.3 ± 12.7 pg/ml testosterone and 119.0 ± 17.0 pg/ml estradiol), was subtracted from the hormone measurements. Sample measurements lower than the baseline water measurements were replaced with zeroes for analysis (5 testosterone, 38 estradiol measurements). Because estradiol measurements were below the detection threshold for many individuals, we did not statistically analyze estradiol by itself, only the ratio of testosterone/estradiol. Estradiol averages are presented in Supplementary Figure 1.

Morphological Measurements

Standard length (cm; length from tip of snout to end of caudal peduncle) and mass (g) were collected before euthanizing (in 1:10 Eugenol: Ethanol solution) males (n = 118) and females (n = 137; see Supplementary Table 1 for treatment sample sizes) at the end of the rearing experiment.

Fitness Metrics

To understand the effects of the rearing environment and population on investment in reproduction (Brewer et al., 2008), we analyzed GSI in a subset of males (n = 107) and females (n = 127). The gonads were removed, and we calculated GSI using the following formula: GSI = (gonad mass/mass) *100.

To quantify the effects of the rearing treatments on fitness more directly, we placed a subset of sexually mature females from the rearing experiment in aquaria with non-sibling males of the same rearing treatment and population group. The males and females were not used as parents more than one time, so each brood has a distinct combination of parents. Rearing treatment conditions were maintained in these aquaria. Females were observed daily, and when they were determined to be carrying a brood, we removed the brood from her mouth, counted the number of eggs, and weighed the brood (ME104E, Mettler Toledo analytical balance) to determine brood mass (g; n = 37 broods; 2-6 broods per treatment/population combination). We then calculated the average egg mass by dividing the brood mass by the number of eggs. Average egg mass and number of eggs in the brood were used as proxies for reproductive success.

Statistics

All analyses were conducted in R, version 4.3.0 (R Core Team, 2024). Linear mixed models (LMM) were performed using Lme4, version 1.10-35.1 (Bates et al., 2015). To understand the interacting and independent effects of oxygen, turbidity, and population on sex hormones and fitness metrics, all models contained oxygen, turbidity, the interaction between oxygen and turbidity, and population as fixed factors. We assessed homogeneity of variance by examining residual plots and the assumption of normality by examining Q-Q plots. If data did not meet the model assumptions, they were transformed. Because fish were genetically related (siblings), we also included brood as a random effect in all models of hormones, GSI, and size. Averages are presented as mean ± SE. Results were considered significant at α < 0.05 (Thiese et al., 2016).

Hormones

In males, testosterone and the ratio of testosterone to estradiol (an indicator of aromatase enzyme activity (Friesen, et al., 2012, Shang et al., 2006))  were log transformed to improve model assumptions. We analyzed both testosterone and the ratio of testosterone to estradiol using LMM’s with oxygen, turbidity, the interaction between oxygen and turbidity, and population as fixed factors, log-transformed standard length as a covariate, and brood as a random effect. The time a sample was collected was initially included as a covariate in all hormone models, but it was not significant and subsequently removed from all models.

Body Size

We analyzed, log-transformed standard length and log-transformed mass in males and females separately using LMM’s with oxygen, turbidity, and the interaction between oxygen and turbidity as fixed effects and brood as a random effect. Because sampling for size data was completed over approximately eight months, age at the time of sampling was also included as a covariate in these models.

GSI

To improve model assumptions, GSI was log transformed. We analyzed GSI separately for males and females using LMM’s with brood as a random effect and oxygen, turbidity, the interaction between oxygen and turbidity, and population as fixed factors and log-transformed SL as a covariate. For females, an additional fixed factor, Brooding (Yes/No), was included to account for females that were mouth brooding at the time of sampling, as their GSI was generally much lower (0.825 ± 0.124) than non-brooding females (3.516 ± 0.321).

Fitness Metrics

Finally, we analyzed the effects of rearing treatment on egg number using a Poisson model (GLM) with oxygen, turbidity, interaction between oxygen and turbidity as fixed factors, and the mass of the mother as a covariate. One brood was determined to be influential (Cook’s distance > 1) and was removed from all models. Additionally, the sample size of the hypoxic/turbid treatment combination produced by river females was low (n = 2), so the river and swamp populations were pooled for the analyses of egg number, batch weight, and egg mass. We also analyzed the batch weight of the broods (g) and egg mass (g; batch weight/ # of eggs) using linear models with oxygen, turbidity, the interaction between oxygen and turbidity as fixed factors, and the mass of the mother as a covariate.