Warming waters beget smaller fish: evidence for reduced size and altered morphology in a desert fish following anthropogenic temperature change

Abstract

Poikilothermic organisms are predicted to show reduced body sizes as they experience warming environments under a changing global climate. Such a shrinking of size is expected under scenarios where rising temperatures increase cellular reaction rates and basal metabolic energy demands, therein requiring limited energy to be shifted from growth. Here, we provide evidence that the ecological changes associated with warming may not only lead to shrinking body size but also trigger shifts in morphology. We documented 33.4 and 39.0% declines in body mass and 7.2 and 7.6% reductions in length for males and females, respectively, in a wild population of Amargosa pupfish, Cyprinodon nevadensis amargosae, following an abrupt anthropogenically driven temperature increase. That reduction in size was accompanied by the partial or complete loss of paired pelvic fins in approximately 34% of the population, a morphological change concomitant with altered body dimensions including head size and body depth. These observations confirm that increasing temperatures can reduce body size under some ecological scenarios and highlight how human-induced environmental warming may also trigger morphological changes with potential relevance for fitness.

1. Introduction

As temperatures increase under a changing global climate, ectothermic organisms such as fishes may experience sublethal physiological consequences, including reduced food intake and conversion efficiency [1,2], impaired mitochondrial respiratory capacity [3] and altered rates of development and growth [4–6], which together may shift fish toward smaller body sizes [7–13]. Such size reductions have been predicted under scenarios where aerobic metabolism or energy intake fails to fully compensate for increased energy demands at higher temperatures [5,7,12,14,15].

Recently, several empirical studies have documented shrinking body sizes in fishes associated with climate warming [7,10,11,16,17]. Exposure to higher temperatures during early life, however, can also alter development to shift non-growth-related phenotypic traits [18]. Increases in environmental temperature may therefore not only affect body size but also alter other phenotype attributes important to fitness, such as behaviour [19–21], life history [22] and morphology [23,24].

Here, we document a decline in body size and shift in the morphology of a wild population of pupfish (Cyprinodon nevadensis amargosae) linked to a rapid, human-caused increase in temperature. The population-level differences in phenotype we observed mirror the developmental responses seen when this species is exposed to energetically stressful conditions during early life [23,25]. These findings illustrate how increasing temperatures linked to anthropogenic activities have the potential to alter fish size and morphology and highlight how understanding the links between environmental temperature and the timing and outcomes of developmental processes can help predict how ectotherms will respond to future climate warming.

2. Material and methods

Amargosa pupfish, C. n. amargosae, were studied in the Death Valley region of California, USA, from Tecopa Bore (35°53′08.4″ N, 116°14′03.0″ W) and the Amargosa River (35°50′56.9″ N, 116°13′49.9″ W). The Amargosa River supports populations of pupfish, speckled dace (Rhinichthys osculus) and non-native mosquitofish (Gambusia affinis) and has temperatures that vary seasonally from near freezing to greater than 40°C [26,27]. Tecopa Bore is a man-made, artesian spring where water emerges at approximately 47.5°C [26,28] and is home to pupfish and mosquitofish. Tecopa Bore was created when the Stauffer Chemical Company drilled a borehole for mineral exploration in March 1967. When drilling reached approximately 106–109 m, water emerged and drilling was abandoned. The borehole was left uncapped, leaving an artesian spring-fed marsh, and C. n. amargosae invaded the newly formed habitat within weeks from the nearby Amargosa River.

Adult C. n. amargosae pupfish were collected by minnow trap from Tecopa Bore (n = 83) and the Amargosa River (n = 40) between 1 and 5 June 2008 for a previous study [29]. Fish were anaesthetized (MS222; Argent Laboratories, Redmond, WA, USA), measured for standard length (SL) (± 0.05 mm) and weighed (± 0.01 g). Fish were also evaluated for the two paired pelvic fins and recorded as having either both fins (2), a single pelvic fin on the right side of the body (right), a single fin on the left (left) or no (0) pelvic fins. Pelvic fins function for trim correction during swimming, and the absence of these fins may affect manoeuvrability during social interactions like territorial defence.

Pupfish were collected again from Tecopa Bore and the Amargosa River on 12 September 2013 (bore, n = 298; river, n = 150), 9 May 2014 (bore, n = 512; river, n = 303) and 8 November 2014 (bore, n = 191; river, n = 181). Fish were measured, weighed and assessed for pelvic fins. Fish were also photographed (PowerShot SD990 IS; Canon, Melville, NY, USA) on the left side of the body, and linear measurements of eye diameter, head length and body depth were measured using ImageJ software (http://rsb.info.nih.gov/ij/).

Mass and length data were log₁₀(x + 1) transformed for normality and compared using ANOVA models to assess effects of population origin, collection date and sex. Variation in pelvic fin phenotype was evaluated by contingency table analyses. Since body dimensions varied with fish size, values were compared both (1) as percentage ratios normalized to body length to permit direct comparisons with previous studies on morphology in this pupfish [23,27], and also (2) with body length as a covariate to avoid shortcomings of using ratios to normalize for size variation [30]. Covariate analyses were conducted using corrected Akaike's information criterion to identify the best-fit model for each analysis, which involved log₁₀(x + 1) transformations of morphological values examined against log₁₀(x + 1) transformed values for ‘SL’ as the covariate and with ‘population’ and ‘sex’ as factors. Since measures of morphology can be correlated, we used a multiple analysis of covariance model (MANCOVA), followed by analysis of covariance (ANCOVA) models for each morphological measure. All analyses were performed using JMP Pro v. 11 (SAS, Cary, NC, USA).

3. Results and discussion

In 2010, the culvert pipe for the outflow water of Tecopa Bore to pass under Tecopa Hot Springs Road was cleared of debris by the Public Works Department of Inyo County, California, USA. The pipe had previously been clogged, which raised the water level on the spring side of the road, creating a shallow marsh of approximately 0.0219 km² (estimated from satellite imagery, June 2009). Clearing the culvert increased flow through the pipe, and the marsh north of Tecopa Hot Springs Road drained to approximately 0.0015 km², leaving only deeper, channelized habitat (figure 1a,b). Between 1 and 5 June 2008, the mean temperature in Tecopa Bore marsh was 24.0 ± 5.3°C (±s.d.) (range: 17.5–33.3°C) [29] (figure 1c). From May 2014 to February 2016, however, the temperature averaged 33.0 ± 3.6°C (range: 19.3–42.8°C), indicating an approximately 9°C increase from 2008 (figure 1c). During this same period, the Amargosa River averaged 19.9 ± 5.7°C (range: 5.3–45.3°C). Salinity ranged between 1.6 and 2.6 ppt in Tecopa Bore and 2.1–2.7 ppt in the Amargosa River in 2014–2016. Temperature and salinity data were recorded at locations of the fish collection in the outflow channel and marsh of Tecopa Bore and main stem Amargosa River.

Figure 1.

(a) Map of Tecopa Bore in 2009 and 2013. Water emerges from the spring (arrowed) at approximately 47.5°C and flows into a shallow bulrush (Juncus spp.) marsh surrounded by saltgrass (Distichlis spicata). In 2010, clearing the culvert pipe under Tecopa Hot Springs Road increased culvert flow and drained the marsh. (b) Photos of the spring pool (1,2) and marsh (3,4) in June 2008 and May 2014. Photo locations indicated by red numbers in (a). (c) Temperature in Tecopa Bore increased between 2008 and 2014–2016, with Amargosa River temperatures (2014–2016) for comparison. Middle line of each box plot denotes median, box shows 25th and 75th percentiles and whiskers indicate maximum and minimum. Temperatures on 2 June 2008 in Tecopa Bore (black line), and 8 May 2014 in Tecopa Bore (red line) and the Amargosa River (blue line) illustrate the temperature increase in Tecopa Bore from 2008 to 2014. (Online version in colour.)

In 2008, pupfish from Tecopa Bore were similar in mass and length to same-sex individuals from the Amargosa River (figure 2a). In both populations, males were larger than females (mass: F_1,119 = 9.855, p = 0.0021; SL: F_1,119 = 4.872, p = 0.0292). However, in 2013–2014, male and female pupfish from Tecopa Bore averaged 33.4 and 39.0% less in mass and 7.2 and 7.9% shorter in length, respectively, than in 2008 (figure 2b) (females: mass, F_1,410 = 43.715, p < 0.0001; SL, F_1,473 = 11.701, p = 0.0007; males: mass, F_1,463 = 26.262, p < 0.0001; SL, F_1,521 = 8.337, p = 0.0040). Pupfish in Tecopa Bore in 2013–2014 were likewise smaller in mass and length than conspecifics from the Amargosa River both in 2008 and in 2013–2014 (p < 0.004 for all comparisons). The Amargosa River population, however, did not vary in size between 2008 and 2013–2014.

Figure 2.

(a) Box plots showing declines in mass (g) and standard length (SL) in the Tecopa Bore (TB) population but not the Amargosa River (AR) population, between 2008 and 2013–2014. Middle line of box plot indicates median, box denotes 25th and 75th percentiles, and whiskers indicate maximum and minimum. (b) Distributions of mass (i) and SL (ii) for Tecopa Bore pupfish. Mean (±s.d.) values indicated by dashed line and shaded boxes; asterisks indicate comparisons with 2008: *p < 0.05, **p < 0.001; n.d. denotes no difference. (c) Pelvic fin phenotypes in 2008 and 2013–2014. (d) Morphological measurements on pupfish collected in 2013–2014 followed Miller [27]. Eye diameter extended from the anterior to posterior eye edges, head length from the anterior of the lower jaw to the posterior of the operculum, and body depth as height at the posterior of the operculum. (e) Mean (±s.e.m.) values shown as relative % (i) and adjusted means (least-squares mean) (ii). *p < 0.05; **p < 0.0001. (f) Male body depth plotted against SL. Tecopa Bore males exhibited greater body depth as smaller fish, but Amargosa River males had deeper bodies when larger. Analysis shown for overlapping body sizes only. (Online version in colour.)

In 2008, all pupfish collected from Tecopa Bore had both paired pelvic fins. However, in 2013–2014, pelvic fin phenotypes in Tecopa Bore had changed (χ² = 65.432, d.f. = 3, p < 0.0001) and 34.0% of fish lacked one or both pelvic fins (figure 2c), with the frequencies of fin phenotypes similar between sexes (p = 0.421). The absence of one or both pelvic fins in Tecopa Bore pupfish in 2013–2014 differed from the nearby Amargosa River (χ² = 332.172, p < 0.0001), where only 6 of 634 pupfish (0.946%) collected in 2013–2014 had fewer than 2 fins (figure 2c) and fin phenotypes did not differ between 2008 and 2013–2014.

Since pupfish collected from Tecopa Bore in 2008 were not photographed, we could not directly test for changes in body shape between 2008 and 2013–2014. Nonetheless, we predicted that fish collected from Tecopa Bore in 2013–2014 would exhibit shape characteristics similar to those of captive-reared pupfish experiencing energetic stressors during early life [23]. MANCOVA analysis of eye diameter, head length and body depth (figure 2d) indeed indicated shape variation between the Tecopa Bore and Amargosa River populations (table 1) (Wilks' lambda approx. F = 478.75, p < 0.0001). Further analyses using either relative percentage values or body length as a covariate (table 1) showed that pupfish from Tecopa Bore had a larger eye diameter (F_1,1523 = 27.702, p < 0.0001) and head length (population × SL, F_1,1523 = 10.484, p = 0.0012) than Amargosa River fish (figure 2e). These results were consistent when analysing all pupfish collected, or when analysing only fish that overlapped in SL (table 1) to reduce any bias arising from different size distributions of the populations.

Table 1.

Analyses of morphological measures in pupfish collected from Tecopa Bore and the Amargosa River in 2013–2014.

		population		sex		SL		pop. × sex		pop. × SL		sex × SL
factor	adjusted R2	F-value	p-value	F-value	p-value	F-value	p-value	F-value	p-value	F-value	p-value	F-value	p-value
MANCOVA (whole model)		15.916	<0.0001	97.304	<0.0001	7683.422	<0.0001	4.125	0.0063	2.472	0.0602	55.133	<0.0001
ANCOVAs
using all fish
eye diameter	0.7283	27.702	<0.0001	8.897	0.0029	3258.36	<0.0001	4.308	0.0447	0.036	0.8444	6.221	0.0127
head length	0.9117	26.340	<0.0001	39.913	<0.0001	12085.03	<0.0001	0.179	0.6725	10.484	0.0012	1.197	0.2741
body depth	0.9610	0.453	0.5011	366.899	<0.0001	26823.41	<0.0001	8.506	0.0036	15.320	<0.0001	167.878	<0.0001
using overlapping size fish only
eye diameter	0.7063	26.467	<0.0001	8.491	0.0036	2874.506	<0.0001	4.564	0.0328	0.153	0.6959	5.282	0.0217
head length	0.9039	27.058	<0.0001	38.678	<0.0001	10842.27	<0.0001	0.097	0.7551	7.906	0.0050	1.597	0.2065
body depth	0.9577	0.383	0.5362	359.897	<0.0001	24240.56	<0.0001	7.262	0.0071	6.534	0.0107	159.435	<0.0001

These analyses also revealed pupfish from the Amargosa River to be larger in relative body depth (% of SL) than pupfish from Tecopa Bore (figure 2e). Adjusted means from the ANCOVA model, however, indicated the opposite: that Tecopa Bore males were deeper bodied (figure 2e) (population × sex, F_1,1523 = 8.506, p = 0.0036; sex × SL, F_1,1523 = 167.878, p < 0.0001). That discrepancy between the relative percentage and ANCOVA approaches arises from population-level variation in how body depth changes with fish size (population × SL, F_1,1523 = 15.320, p < 0.0001). Among smaller fish, Tecopa Bore males tend to have greater body depths; but, among larger fish, Amargosa River males have deeper bodies (figure 2f). That allometric difference indicates that despite Tecopa Bore males being deeper bodied early in life, they do not develop the deep body sexual dimorphism typical of large male pupfish from the Amargosa River and other populations [27,31].

Taken as a whole, these data reveal that the Tecopa Bore population of C. n. amargosae underwent a reduction in size and shift in morphology following anthropogenic changes to its habitat. Prior studies with C. n. amargosae under controlled experimental conditions point to the increase in temperature—and more generally elevated energetic stress—as likely initiating this phenotypic shift [23,25]. Pelvic fins develop in teleost fishes as part of a suite of post-embryonic morphological changes during the larval-to-juvenile transition [32]. In pupfish, exposure to energetically stressful conditions during this transition impedes pelvic fin development, and laboratory studies have found that fewer than 40% of C. n. amargosae raised at approximately 34°C develop both pelvic fins [23]. Several lines of evidence point to these changes in morphological development being underlain by environmental impacts on thyroid hormone signalling [23,25], which regulates the larval-to-juvenile metamorphic transition in fish [33].

While elevated temperature can trigger this alternative developmental trajectory even under conditions of abundant food and normoxia [23], food availability and low oxygen conditions in Tecopa Bore may have augmented energetic stresses to contribute to the observed morphological changes. Pupfish raised in the laboratory under restricted food, for instance, also fail to develop pelvic fins and exhibit a larger head and eye size and smaller body depth [23]. In Tecopa Bore, the diet of C. n. amargosae is approximately 85% algae and detritus [28,34], in line with the generally omnivorous habits of pupfishes [35,36]. Although food assimilation efficiency does not vary in adult C. nevadensis between 20 and 32°C [37], food intake rates by pupfish in Tecopa Bore could be inadequate to both meet elevated basal metabolic demands and support growth under that habitat's high temperature.

Measurements of dissolved O₂ (DO) in Tecopa Bore between 2014 and 2016 indicate DO is considerably lower in Tecopa Bore (1.82–3.63 mg l⁻¹; 25.8–40.8% saturation) than in the Amargosa River (6.32–11.85 mg l⁻¹; 62.6–68.6%). While it is not known if DO conditions changed in Tecopa Bore between 2008 and 2013–2014, or if consistent exposure to low DO alone can induce the development of morphological traits mirroring those observed for pupfish in Tecopa Bore, it is well established that hypoxia can depress metabolic rate [38] and increase the expression of less-efficient metabolic pathways involving anaerobic enzymes such as lactate dehydrogenase (LDH) [39,40], with responses varying both with hypoxia duration and taxonomically [41,42]. Muscle LDH activity measured in pupfish collected from Tecopa Bore in 2014 was approximately 2× higher than in conspecifics from the Amargosa River, suggesting heightened anaerobic capacity [43]. Interestingly, C. nevadensis acclimated to 33°C can even cease O₂ consumption entirely for more than 2 h, during which time anaerobic pathways producing ethanol appear to be used [44]. While specific mechanisms of anaerobism were not explored here, exposure of pupfish to the continuously high temperatures of Tecopa Bore may diminish metabolic efficiency and slow growth to, ultimately, give rise to smaller-sized fish [45]. That smaller size may also confer lower reproductive output, given that fecundity in fish often relates to body size [46–48].

Together, these findings reveal that warming habitats can reduce fish size [7–10,13,14], while also providing evidence that broader phenotypic changes may ensue for wild populations via alterations in development that emerge either from direct impacts of elevated temperatures or from wider ecological changes (i.e. reduced food availability, reductions in DO) that can accompany the rapid, human-induced warming of aquatic habitats.

Ethics

Approved by the Institutional Animal Care and Use Committee of California Polytechnic State University (Protocol no. 1507). Fish were collected under California Department of Fish and Wildlife permit no. SC-4793.

Data accessibility

Data are available in the Dryad Digital Repository: https://dx.doi.org/10.5061/dryad.1n0jh42 [49].

Authors' contributions

S.C.L. conceived the study, and all the authors contributed to data collection. S.C.L. drafted the manuscript, which was then evaluated and revised by all authors. All authors gave final approval for publication and agreed to be held accountable for the work described therein.

Competing interests

The authors have no competing interests.

Funding

Supported by a New Investigator Award from CSU Program for Education and Research in Biotechnology (CSUPERB) to S.C.L., by a U.S. EPA GRO Fellowship (grant no. 91776901-0) to S.L.B. and by the William and Linda Frost Fund in the Cal Poly College of Science and Mathematics to M.M.M. and E.A.E.