Temperature sensitivity of the interspecific interaction strength of coastal marine fish communities

Abstract

The effects of temperature on interaction strengths are important for understanding and forecasting how global climate change impacts marine ecosystems; however, tracking and quantifying interactions of marine fish species are practically difficult especially under field conditions, and thus, how temperature influences their interaction strengths under field conditions remains poorly understood. We herein performed quantitative fish environmental DNA (eDNA) metabarcoding on 550 seawater samples that were collected twice a month from 11 coastal sites for 2 years in the Boso Peninsula, Japan, and analyzed eDNA monitoring data using nonlinear time series analytical tools. We detected fish-fish interactions as information flow between eDNA time series, reconstructed interaction networks for the top 50 frequently detected species, and quantified pairwise, fluctuating interaction strengths. Although there was a large variation, water temperature influenced fish-fish interaction strengths. The impact of water temperature on interspecific interaction strengths varied among fish species, suggesting that fish species identity influences the temperature effects on interactions. For example, interaction strengths that Halichoeres tenuispinis and Microcanthus strigatus received strongly increased with water temperature, while those of Engraulis japonicus and Girella punctata decreased with water temperature. An increase in water temperature induced by global climate change may change fish interactions in a complex way, which consequently influences marine community dynamics and stability. Our research demonstrates a practical research framework to study the effects of environmental variables on interaction strengths of marine communities in nature, which would contribute to understanding and predicting natural marine ecosystem dynamics.

Keywords: community dynamics; ecology; environmental DNA; fish; information flow; interaction network; none; quantitative environmental DNA analysis.

Figure 1.. Study sites and overall dynamics of environmental DNA (eDNA) concentrations and the number of fish species detected.

(a) Study sites in the Boso Peninsula. The study sites are influenced by the Kuroshio Current (red arrow; left panel) and distributed along the coastal line in the Boso Peninsula (right panel). (b) Total eDNA copy numbers estimated by quantitative eDNA metabarcoding (see Methods for detail). (c) Fish species richness detected by eDNA metabarcoding. Points and lines indicate raw values and LOESS lines, respectively. The line color indicates the sampling site. Warmer colors generally correspond to study sites with a higher mean water temperature.

Figure 1—figure supplement 1.. Dynamics of environmental DNA (eDNA) copy numbers of Japanese black seabream (Acanthopagrus schlegelii) that was used as an internal standard.

Gray horizontal line indicates the minimum eDNA copy number of Japanese black seabream detected. Approximately 83.3% of samples contain detectable concentrations of the standard DNA, which was used to convert sequence reads to eDNA copy numbers. For samples which do not contain the standard DNA, we assume that it contains the minimum amount of eDNA copy numbers (i.e., 0.69 copies/2 extracted DNA; gray horizontal line).

Figure 1—figure supplement 2.. Dynamics of water temperature and the relationships between water temperature and total environmental DNA (eDNA) concentration and fish species richness.

(a) Dynamics of sea surface water temperature at the sampling sites. (b) The relationship between water temperature and total eDNA copy numbers, and (c) the relationship between water temperature and fish species richness. Colors indicate sampling sites. For (b) and (c), dashed lines indicate standardized major axis regression.

Figure 2.. Interaction networks of the fish community in the Boso Peninsula coastal region.

The ‘average’ interaction network reconstructed by quantifying information transfer between environmental DNA (eDNA) time series. Transfer entropy (TE) was quantified by leveraging all eDNA time series from multiple study sites to draw this network. Only information flow larger than 80% quantiles (i.e., strong interaction) was shown as interspecific interactions for visualization. The edge color indicates scaled TE values, and fish illustration colors represent their ecology (e.g., habitat and feeding behavior). Node colors and node sizes indicate the fish family and fish abundance (total eDNA copy numbers of the fish species), respectively.

Figure 2—figure supplement 1.. The relationships between network properties and environmental variables.

Diagonal panels show density distributions of the data, and lower triangle area shows scattered plots. Each point represents one water sample at each study site, meaning that ‘Mean interaction strength’ is a mean value at the community level. Solid red and gray lines indicate statistically clear (p < 0.05) and unclear (p > 0.05) standardized major axis regression, respectively.

Figure 3.. Dependence of interaction strengths on biotic and abiotic variables (50 dominant fish species and 11 study sites were leveraged).

The panels show the overall effects of biotic and abiotic variables on interaction strengths of the 50 dominant fish species: Effects of (a, d) water temperature, (b, e) species richness, and (c, f) total environmental DNA (eDNA) copy numbers. The y-axis indicates the effects of the variables on fish–fish interaction strengths quantified by the MDR S-map method. (a–c) show the effects on the species interactions that a focal species receives (i.e., in-strength), and (d–f) show the effects on the species interactions that a focal species gives (i.e., out-strength). The line indicates the average effects estimated by the general additive model (GAM), and the gray region indicates 95% confidential intervals. LME and GAM indicate the statistical clarity of the linear mixed model portion and GAM portion, respectively. Detailed statistical results and raw data are shown in Supplementary file 1d and Figure 3—figure supplement 1, respectively.

Figure 3—figure supplement 1.. Dependence of interaction strengths on additional abiotic variables.

The panels show the overall effects of additional abiotic variables on interaction strengths of the 50 dominant fish species: Effects of (a, d) salinity, (b, e) tide level (cm), and (c, f) wave (m). The y-axis indicates the effects of environmental variables on fish–fish interaction strengths quantified by the MDR S-map method. (a–c) show the effects on the species interactions that a focal species receives (i.e., in-strength), and (d–f) show the effects on the species interactions that a focal species gives (i.e., out-strength). The line indicates the average effects estimated by the general additive model (GAM), and the gray region indicates 95% confidential intervals. LME and GAM indicate the statistical clarity of the linear mixed model portion and GAM portion, respectively. Detailed statistical results are shown in Supplementary file 1d.

Figure 3—figure supplement 2.. The relationship between interaction strengths and water temperature, species richness, and total DNA concentrations.

The panel shows the overall relationship between interaction strengths of the 50 dominant fish species and (a, d) water temperature, (b, e) species richness, and (c, f) total DNA concentration. The y-axis indicates the interaction strength between fish species quantified by the MDR S-map method. Note that the MDR S-map enables quantification of interaction strengths at each time point, and thus the number of data points is large (also true in Figure 3—figure supplement 3 and Figure 4—figure supplements 1 and 2). (a–c) Points indicate the species interactions that a focal species receives (i.e., in-strength), and (d–f) points indicate the species interactions that a focal species gives (i.e., out-strength). The line indicates the nonlinear regression line estimated by the general additive model. Colors of points and lines indicate the study site.

Figure 3—figure supplement 3.. The relationship between interaction strengths and salinity, tide level, and wave.

The panel shows the overall relationship between interaction strengths of the 50 dominant fish species and (a, d) salinity, (b, e) tide level (cm), and (c, f) wave (m). The y-axis indicates the interaction strength between fish species quantified by the MDR S-map method. Note that the MDR S-map enables quantifications of interaction strengths at each time point, and thus the number of data points is large. (a–c) Points indicate the species interactions that a focal species receives (i.e., in-strength), and (d–f) points indicate the species interactions that a focal species gives (i.e., out-strength). The line indicates the nonlinear regression line estimated by the general additive model. Colors of points and lines indicate the study site.

Figure 4.. Temperature dependence of fish species interactions at the species level.

(a and b) show temperature effects on fish species interactions quantified by the MDR S-map method. Note that the MDR S-map enables quantifications of interaction strengths at each time point, and thus the number of data points is large. (a) Points indicate the species interactions that a focal species (indicated by the strip label and fish image) receives (i.e., in-strength). (b) Points indicate the species interactions that a focal species (indicated by the strip label and fish image) gives (i.e., out-strength). For (a) and (b), only fish species of which interactions are statistically clearly affected by water temperature are shown (to exclude fish species with relatively weak temperature effects, p < 0.0001 was used as a criterion here). Point color indicates the study site. Gray line is drawn by general additive model (GAM; the study sites were averaged for visualization purpose).

Figure 4—figure supplement 1.. Dependence of fish species interactions on species richness at the fish species level.

(a and b) show effects of species richness on fish species interactions quantified by the MDR S-map method. (a) Points indicate the species interactions that a focal species (indicated by the strip label and fish image) receives (i.e., in-strength). (b) Points indicate the species interactions that a focal species (indicated by the strip label and fish image) gives (i.e., out-strength). For (a) and (b), only fish species of which interactions are statistically clearly affected by water temperature are shown (to exclude fish species with relatively weak temperature effects, p < 0.0001 was used as a criterion here). Point color indicates the study site. Gray line is drawn by general additive model (GAM; the study sites were averaged for visualization purpose).

Figure 4—figure supplement 2.. Dependence of fish species interactions on the total DNA concentration (an index of total fish abundance) at the species level.

(a and b) show effects of the total DNA concentrations on fish species interactions quantified by the MDR S-map method. (a) Points indicate the species interactions that a focal species (indicated by the strip label and fish image) receives (i.e., in-strength). (b) Points indicate the species interactions that a focal species (indicated by the strip label and fish image) gives (i.e., out-strength). For (a) and (b), only fish species of which interactions are statistically clearly affected by water temperature are shown (to exclude fish species with relatively weak temperature effects, p < 0.0001 was used as a criterion here). Point color indicates the study site. Gray line is drawn by general additive model (GAM; the study sites were averaged for visualization purpose).