Biology Letters
You have accessResearch article

Morphology-based differences in the thermal response of freshwater phytoplankton

Angel M. Segura

Angel M. Segura

Modelización y análisis de Recursos Naturales, Centro Universitario Regional Este, Universidad de la República, Ruta 9 y 15, Rocha, Uruguay

[email protected]

Google Scholar

Find this author on PubMed

,
Florencia Sarthou

Florencia Sarthou

Sección Limnología, Facultad de Ciencias, Universidad de la República, Iguá 4225, 11 400 Montevideo, Uruguay

Google Scholar

Find this author on PubMed

and
Carla Kruk

Carla Kruk

Modelización y análisis de Recursos Naturales, Centro Universitario Regional Este, Universidad de la República, Ruta 9 y 15, Rocha, Uruguay

Ecología Funcional de Sistemas Acuáticos, Centro Universitario Regional Este, Universidad de la República, Ruta 9 y 15, Rocha, Uruguay

Google Scholar

Find this author on PubMed

    Abstract

    The thermal response of maximum growth rate in morphology-based functional groups (MBFG) of freshwater phytoplankton is analysed. Contrasting an exponential Boltzmann–Arrhenius with a unimodal model, three main features were evaluated: (i) the activation energy of the rise (Er), (ii) the presence of a break in the thermal response and (iii) the activation energy of the fall (Ef). The whole dataset (N = 563) showed an exponential increase (Er ∼ 0.5), a break around 24°C and no temperature dependence after the breakpoint (Ef = 0). Contrasting thermal responses among MBFG were found. All groups showed positive activation energy (Er > 0), four showed no evidence of decline in growth rate (temperature range = 0–35°C) and two presented a breakpoint followed by a sharp decrease in growth rate. Our results evidenced systematic differences between MBFG in the thermal response and a coherent response significantly related to morphological traits other than size (i.e. within MBFG). These results provide relevant information for water quality modelling and climate change predictions.

    1. Introduction

    Temperature influences the ecological performance of organisms at multiple levels. Its effect on metabolic rates is well recognized [1,2] and has been modelled under different theoretical and empirical frameworks [14]. Most of the models assume an exponential temperature response known as the Boltzmann–Arrhenius function, formalized under the metabolic theory of ecology (MTE) as [2]:

    Display Formula
    1.1
    where k is the Boltzmann constant (eV K−1) and T is temperature in degrees Kelvin. A significant improvement from previous exponential models is the interpretation of E as an average activation energy similar to that observed for molecular reactions [2]. E was formerly introduced as a ‘universal’ constant but further systematic differences between heterotrophs (E ∼ 0.64) and autotrophs (E ∼ 0.32) were described [5]. In autotrophs, the activation energy was linked to the activity of the Rubisco enzyme (E ∼ 0.32). Recent research identified systematic deviations from the archetypal activation energy value, which questions its universality [6]. Irrespective of the particular formulation, the exponential model is useful to characterize the metabolic response in a wide thermal range. However, it does not describe the sharp decay observed above the optimum temperature (TBK) [7,8], which is generally included within the biologically relevant temperature ranges (e.g. [9]). Describing the full thermal response is critical in the face of anticipating the effect of increased thermal variability caused by climate change.

    The maximum growth rate (μ) of a population is tightly related to its metabolic activity [4]. Assuming a proportional relationship between metabolic rate and maximum growth rate, and taking the natural logarithms of both sides, equation (1.1) can be written as:

    Display Formula
    1.2
    where 1/kT is the explanatory variable, log(μ) the response variable, the activation energy (E) is the slope of the fitted relation and a is a normalization constant [10]. Whether log-maximum growth rates increase linearly with temperature—as predicted by the Arrhenius–Boltzmann function—or in a unimodal way within a biologically relevant temperature range is, however, least explored [7,8].

    The aquatic microscopic primary producers living in the pelagic realm (i.e. phytoplankton) produce half of the oxygen on Earth, operate as a sink of atmospheric CO2 and generate the biomass that fuels aquatic food webs. This community is an excellent model to study trait dependence of thermal responses. Trait-based classifications of phytoplankton organisms are well developed [9,11]. Particularly, morphology-based functional groups (MBFG) were defined based on individual traits (e.g. siliceous wall, aerotopes, surface area), clustering organisms in seven groups of freshwater phytoplankton with different physiology (e.g. growth rate, sinking rate; [11]), environmental preferences and community dynamics [12]. This classification might help to bridge the gap between individual species dynamics and bulk estimates of producers' biomass (e.g. chlorophyll a). The description of within-group coherent thermal responses would increase our ability to track ecosystem changes and help to explore the mechanism shaping deviations from ‘universal’ thermal responses. Here we evaluated the influence of morphological traits in the thermal response of freshwater phytoplankton using a morphology-based functional groups classification.

    2. Material and methods

    A systematic bibliographic search was conducted, aimed at gathering experimental information on how freshwater phytoplankton growth rate changes with temperature. This search covered peer reviewed journals reporting laboratory experiments exploring changes in population growth rate over a broad range of temperatures under nutrients replete and non-limiting light conditions [7].

    Species were classified into seven MBFG [11]. Group I clusters small, high surface/volume unicells and filaments (e.g. picoplankton); group II includes siliceous flagellates (e.g. Mallomonas); group III are long filaments, with high surface/volume and aerotopes (e.g. Cylindrospermopsis); group IV represent medium size unicells and colonies without other traits (e.g. aerotopes, flagella, Coelastrum); group V includes medium to large size flagellates (e.g. Peridinium); group VI is composed of medium to large unicells and colonies with siliceous walls (mainly diatoms); and finally group VII represents large mucilaginous colonies (e.g. Microcystis). Few instances of group II along a temperature gradient were observed in the literature and the group was excluded from the individual analysis.

    We evaluated whether a linear (exponential; equation (1.2)) or a segmented model [7,13] could better explain the relationship between natural logarithm of maximum growth rate (μ) and temperature (T; in degrees Kelvin). We chose segmented regression to model nonlinear responses because it enables an objective estimate of breakpoint and provides a direct estimation of activation energy for the fall region [7]. This model is simple (only 1 more parameter than a quadratic model) and captures the asymmetry observed between the activation energy in the rise and fall while retaining its direct interpretation from the slopes. Quadratic models can only fit symmetric responses, and parameters have no direct interpretation [14]. More complex models based on physiological grounds were proposed [8], but available data generally preclude appropriate fit. We selected between linear and segmented using Akaike information criteria (AIC; [15]) for the whole dataset and for each MBFG separately. We defined differences in thermal performance as being significant when the 95% confidence interval (CI) of fitted parameter did not overlap. All models were fitted in the statistical program R [13,16].

    3. Results and discussion

    We compiled data from 563 experiments from prokaryotes and eukaryotes over nine major phylogenetic groups (see electronic supplementary material, table S1 for original data). The linear model (equation (1.2)) fitted to the whole dataset gives an estimation of activation energy (E[CI-95%]) = 0.31[0.25–0.38]) similar to previous values estimated for freshwater and marine organisms [5,7]. The segmented model improved the fit with respect to the linear model (ΔAIC = 13.5; figure 1) with a larger estimated activation energy for the rise (Er = 0.51[0.38–0.64]). This suggests that the fitting of a linear model (equation (1.2)) generally underestimates the average activation energy of the rise, as found before [7]. The breakpoint was estimated at 21°C (294 Kelvin), and the activation energy after the breakpoint (Ef) was not significantly different from zero (Ef [95% CI] = 0 [−0.2 to 0.2]). The unexpected independence of maximum growth rate with temperature after the breakpoint might result from combining MBFG with different thermal responses across all species.

    Figure 1.

    Figure 1. Temperature dependence of maximum growth rate in freshwater phytoplankton. Continuous lines represent the best fitted linear and segmented (bold) models. See table 1 for model statistics.

    Table 1.Model statistics for the thermal response of freshwater phytoplankton. The general linear (L) model (y = a + b1x) relates the inverse of temperature (x = 1/(Tk)) with the logarithm of maximum growth rate (y = log μ) following equation (1.2), where T is temperature in degrees Kelvin and k is the Boltzmann constant; thus the first slope (b1) represents the activation energy of the rise (Er) and the second slope (b2) the activation energy of the fall (Ef). The segmented (S) model is fully defined by the slopes (b1 and b2), one intercept (a) and a breakpoint (bkp). 95% confidence intervals are presented for the relevant parameters. Differences in Akaike information criteria (ΔAIC) between the linear model and segmented model are presented. Most of the growth rates of MBFG II were registered at 20°C (26 out of 36) and thus its individual thermal response was not characterized.

    MBFG N R2 model breakpoint (°C) b1 (Er) b2 (Ef) intercept a ΔAIC
    all groups 563 0.14 S 24.0 (20.3 to 27.8) 0.45 (0.31 to 0.57) 0 (−0.2 to 0.2) 1.5 12
    I 30 0.35 L 0.38 (0.16 to 0.60) n.a. 14.5 −3
    III 181 0.25 S 32.0 (34.4 to 29.6) 0.40 (0.27 to 0.52) 1.7 (0.7 to 2.7) 64 35
    IV 137 0.43 L 0.63 (0.51 to 0.75) n.a. 24.5 2
    V 89 0.04 L 0.23 (0.02 to 0.48) n.a. 8.4 4
    VI 48 0.48 S 27.0 (29.8 to 24.2) 0.49 (0.27 to 0.71) 2.9 (1.0 to 4.8) 112 23
    VII 42 0.21 L 0.45 (0.25 to 0.67) n.a. 17.5 0

    Contrasting patterns in shape (linear and segmented) and activation energy (Er, Ef) in the thermal dependence of maximum growth rate among MBFG were found. The group's models increased the average explained variance (R2 from 0.16 to average of 0.3; table 1). Four MBFG showed a linear response (I, IV, V and VII) and two presented a segmented relationship (III and VI; table 1; figure 2). Some of the groups did not reach their optimum growth rate at elevated environmental temperatures (>35°C). More research exploring the full physiological thermal range (PTR) for these groups is necessary.

    Figure 2.

    Figure 2. Thermal dependence of maximum growth rate for each morphology-based functional group (MBFG) and the best fitted model. For details on model parameters see table 1. Models are presented as a function of temperature for ease of interpretation.

    Differences in activation energy were found among MBFG (table 1; figure 2). The mechanistic basis of such a difference is not fully understood. We propose two groups of mechanisms to account for the thermal response: (i) changes in resource use [17] and (ii) specific metabolic pathway limitations. The first one is related to differential effects of temperature on resource acquisition and homeostatic maintenance [18]. The different thermal responses of respiration and photosynthesis impose a restriction on phytoplankton growth rate that could select for different morphologies at different temperatures, as has been shown previously. The second group correspond to different activation energy for specific metabolic pathways. For example, the activation energy of enzymes in the metabolic pathway of silica wall formation in MBFG VI can define its thermal response, while the formation of carbon-rich mucilage envelope might determine the response of organisms in MBFG VII. Body size is an important variable for defining MBFGs, but it cannot account for the differences observed in the activation energy among groups. For example, MBFG III and IV are similar sized (approx. 1540 µm3 [15]) but present different thermal responses (Er = 0.4 and 0.63, respectively).

    The residual variability observed in the thermal response differs among groups (figure 2). The variability of MBFG I, VI and VII was low, while for MBFG V it was high (R2 = 0.04; p = 0.07). MBFG V includes phagotrophic and mixotrophic organisms with contrasting changes in feeding strategies along a temperature gradient, which can explain the observed variability [19]. Another source of variability relies in the experimental set-up. Nutrient conditions were non-limiting, but light conditions varied among experimental set-ups (electronic supplementary material, table S1). A compromise was reached between obtaining representative data for each MBFG and homogeneity in the experimental design. The residual variability is expected to be caused by differential enzymatic limitations of the metabolic pathways not reflected by the morphologic traits used to construct the MBFG.

    The finding of breakpoints in two groups (III, VI) within biologically relevant temperature ranges implies strong departures from predictions generated under linear models (equation (1.2)) and expands the paradigm of the PTR used previously [9].

    The uneven distribution of the number of data points among groups can decrease the power to detect a breakpoint, especially in MBFG I (N = 30), II (N = 36) and VII (N = 42). However, MBFG VI presented a moderate number of data points (N = 48) but a marked breakpoint. Some of the criticisms to modelling thermal responses as a log-linear pattern are not applicable here [14]. First, the number of data points used for modelling MBFG thermal dependence was larger (N = 30 to 180) than the number used in single species evaluation (median value = 6; [14]). Second, a large range of temperature values was used (<5°C to >35°C) covering the PTR for most groups (see electronic supplementary material, table S1 for specific data). Third, there was no evidence of a curvilinear relationship in the fitted models (figure 2). Thus, present evaluation is representative of the thermal response of MBFGs. Future exploration of variability among species within MBFG should be performed that would allow the exploration of intra-group patterns.

    In summary, the present results demonstrate coherent thermal responses associated to morphological traits in freshwater phytoplankton with important consequences for theoretical and applied ecology. First, we have made advances in the evaluation of MBFG as a tool to model and understand community assembly. Second, we provide a comprehensive list of maximum growth rates and temperatures for freshwater phytoplankton that can be analysed by other researchers. Third, we provide a model for the temperature response of maximum growth rate for seven MBFG that can be easily included in freshwater quality models to analyse ecosystems' response to climate change. The search for thermal dependence patterns at intermediate organization levels (functional groups) will provide better ways to understand community and ecosystem changes fostered by temperature changes. Further work on theoretical aspects and refined experiments to explore mechanisms are required.

    Data accessibility

    Data presented here were obtained from published or digitized sources and are fully available for the scientific community. We include electronic supplementary material with raw data.

    Authors' contributions

    A.M.S. and C.K. designed research; F.S. compiled and analysed data. A.M.S., C.K. and F.S. contributed to data interpretation. A.M.S. wrote a first draft and all authors contributed to the writing and revising of the manuscript. All authors approve the final version and agree to be held accountable for the content therein and approve the final version of the manuscript.

    Competing interests

    Authors declare no competing interest.

    Funding

    This work was partially funded by CSIC i+d 2016 id 197 project conducted by C.K. and an ANII postgraduate fellowship to F.S. (POS_NAC_2011_1_3297). C.K. and A.M.S. thank SNI-ANII.

    Acknowledgements

    C.K. thanks Carmela Carballo, Lucía Nogueira and Carolina Cabrera for helping in data curation. We are indebted to Jean Philippe Gibert and two anonymous referees whose comments on a previous version greatly improved the manuscript.

    Footnotes

    Electronic supplementary material is available online at https://doi.org/10.6084/m9.figshare.c.4080164.

    Published by the Royal Society. All rights reserved.