1. Species Interactions Drive Fish Biodiversity Loss in a High-CO2 World
Ivan Nagelkerken, Silvan U. Goldenberg, Camilo M. Ferreira, Bayden D. Russell, Sean D. Connell 
Current Biology 
Volume 27, Issue 14, Pages e4 (July 2017)
DOI: /j.cub
Copyright © 2017 Elsevier Ltd Terms and Conditions

2. Current Biology 2017 27, 2177-2184.e4DOI: (10.1016/j.cub.2017.06.023)
Copyright © 2017 Elsevier Ltd Terms and Conditions

3. Figure 1
Fish Species Richness, and Density of Fish, Predators, and Food at CO2 Vent and Control Sites
Species richness (A) and total fish density (B), both quantified using visual surveys during 2013, 2015, and 2016; predator density (per 10 min video recordings) in 2016, as quantified by baited remote underwater video (C); and (D) density of food items (invertebrates) in Data are shown as the mean + SE. ∗p < Details of the statistical analyses are shown in Table S1. See also Figures S2 and S3.
Current Biology  , e4DOI: ( /j.cub )
Copyright © 2017 Elsevier Ltd Terms and Conditions

4. Figure 2
Species Interactions and Densities of Four Fish Species at CO2 Vent and Control Sites
Differences in (A) density, (B) competitive dominance, (C) attraction to novel food, and (D) predator-escape behavior among four fish species at CO2 vent and control sites. The inset in (D) shows the in situ response of the dominant species (common triplefin) to a live predator (moray eels; Figure S1A). Letters above the bars indicate significant differences if letters are not shared. ∗p < 0.05. ns, not significant. Details of the statistical analyses are shown in Table S2. Densities were recorded during 2015 and 2016; behaviors were recorded during See also Figure S3.
Current Biology  , e4DOI: ( /j.cub )
Copyright © 2017 Elsevier Ltd Terms and Conditions

5. Figure 3
Experimental and In Situ Habitat Use by a Dominant and Subordinate Competitor Species from CO2 Vent and Control Sites
Habitat preference, as tested in an experimental on-board choice arena, of (A) dominant alone (common triplefin), (B) subordinate alone (Yaldwin’s triplefin), (C) dominant in the presence of a subordinate, and (D) subordinate in the presence of a dominant. The choice experiments were performed on a boat on-site (in 2016) using freshly collected fish and water from the CO2 vent and control areas, respectively. MANOVA (multivariate analysis of variance): turf versus bare habitat use 1 differs from use 2, and turf versus bare habitat use 2 differs from use 3. In situ relative habitat use is depicted for the (E) dominant species and (F) subordinate species at CO2 vent and control sites, respectively. Letters above the bars indicate significant differences (p < 0.05) if letters are not shared. Details of the statistical analyses are shown in Table S3. Error bars represent the SE. See also Figures S3 and S4.
Current Biology  , e4DOI: ( /j.cub )
Copyright © 2017 Elsevier Ltd Terms and Conditions

6. Figure 4
Overview of Our Empirical Findings Illustrating How Neutral, Negative, and Positive Direct and Indirect Effects of Ocean Acidification Can Interact to Reduce Local Species Richness and Create Novel Community Compositions in the Absence of Intergenerational Adaptation
The width of the arrows indicates the relative strength of the process (described within the white rounded rectangles) under ambient CO2 (control) versus elevated CO2 (CO2 vent) conditions. Artwork by Tullio Rossi ( symbols courtesy of the Integration and Application Network, University of Maryland Center for Environmental Science ( Under ambient CO2 conditions, the most abundant species showed more intense exploratory behavior and was also the dominant competitor in terms of competitive encounters won in the presence of food and eagerness in approaching novel food. Elevated CO2 did not alter this competitive dominance (neutral direct effect), as would be expected from other studies [28]. In contrast, the dominant species suffered from a reduced anti-predator response under elevated CO2 (negative direct effect 1), and we are the first to provide such evidence from in situ acidified communities with real predators (see the inset in Figure 2D). The dominant species also became bolder by switching its habitat use from a mix of turf and bare habitat (when alone) to preferentially open bare habitat in the presence of a competitor (negative direct effect 2). Weakening of such critical behaviors usually leads to increased mortality from predation [29, 30], but with fewer predators present at vents (negative indirect effect via predators), populations of the most common and dominant species could thrive. Loss of predators has also been observed at vents in the northern hemisphere [27] and is most likely driven by loss of shelter due to habitat shifts, where structurally complex vegetation like kelp is replaced by low-relief turf algae that show increased dominance under elevated CO2 [31]. Only one uncommon species (crested blenny) showed a significant density increase at vents. Similar mechanisms as for the dominant species may be responsible for this, including a higher competitive dominance, a higher attraction to novel food, and a reduction in predator-escape behavior as opposed to the other subordinates. In contrast, all other uncommon species showed a decrease in their densities at vents. These behavioral subordinates showed increased shelter behavior under elevated CO2 (negative indirect effect 1 via competitors; as also observed in some other fish species [32, 33]), proportionally higher use of turf shelter in situ (negative indirect effect 2 via competitors), and complete loss of attraction toward novel food presented in situ (negative indirect effect 3 via competitors). Additionally, the loss of kelp habitat and consequent reduced niche diversity at the vents may have impacted uncommon species, as specialist species suffer disproportionally from habitat loss compared to generalists [19]. Finally, the increased density of invertebrate food and increase in habitat availability supported an increase in carrying capacity of the system (positive indirect effect via resources), which benefitted the competitively dominant species most. Overall, this led to novel community compositions, density reduction in uncommon species, and loss of local species diversity.
Current Biology  , e4DOI: ( /j.cub )
Copyright © 2017 Elsevier Ltd Terms and Conditions