全球变化知识资源中心

Study sites.

The study encompassed two islands containing natural subtidal CO₂ vents and hereafter are referred to as ‘locations’ (Supplementary Fig. 2), where one was located in a seagrass-dominated embayment of an island in the Northern Hemisphere (a single vent at Vulcano Island, Mediterranean Sea) and one at a rock-dominated island in the Southern Hemisphere (a multiple-vent site at White Island, New Zealand).

At Vulcano Island, the vent is characterized by a single large CO₂ vent located in Levante Bay (Supplementary Fig. 2), with a slowly increasing gradient in pH with increasing distance from the vent. Two sites close to the vent and two control sites away from the vent were selected. The two vent sites represented approximate end-of-the-century projections in pH reduction of 0.28 units, on average, due to ocean acidification (Supplementary Table 5) based on the RCP8.5 scenario of greenhouse gas emissions³¹; water temperature did not differ between control and vent sites. Long-term variability of pH and other physico-chemical variables in the Bay are reported elsewhere³². Hydrogen sulphide released from the main bubbling area does not extend to the north-eastern part of the Bay and sulphate levels are typical of oceanic waters³². The substratum of the shallow parts of the Bay (1–5 m depth) at control sites was characterized by a mosaic of seagrass (mainly Cymodocea nodosa), turf-forming macroalgae rooted in the sediment (<10 cm in height; mainly Caulerpa prolifera), cobbles, small rocky reefs, and sandy substratum.

White Island is a volcanic island located in the Bay of Plenty of the North Island of New Zealand (Supplementary Fig. 2). Two independent vent and two control sites were identified along the north-eastern coast of the Island. The CO₂ plumes at vent sites were ~24 × 20 m in dimension and located at 6–8 m depth. The control sites were located adjacent to the vents (>~25 m away) where pH levels represented ambient oceanic conditions. pH levels at the two vent sites represented approximate end-of-the-century projections in pH reduction of 0.20 units, on average, and were not confounded by elevated temperatures (Supplementary Table 5). Measurements from different time periods showed similar pH values, suggesting the vents are relatively stable over time. Sulphate levels at the vent sites do not differ from control sites (mean ± s.e.m.: 1,157 ± 11 versus 1,154 ± 13 ppm, respectively³³) but are slightly higher than the regional and global oceanic averages (mean ± s.e.m.: 1,083 ± 4 versus ~904 ppm, respectively³³). The study area represents a rocky reef ecosystem, and the substratum at control sites was characterized by a mosaic of kelp (Ecklonia radiata), turf-forming macroalgae (<10 cm in height), and hard-substratum sea urchin barrens devoid of vegetation.

Vents are known to fluctuate in CO₂ release and this may potentially affect biological responses. Studies on fishes have shown that behavioural effects to elevated CO₂ manifest from one to four days after the onset of exposure, and recovery takes 8–48 h (ref. 34). Therefore, behavioural impairment to CO₂ operates at longer timescales than the typical short-term extremes that are observed at vents, including our study sites³². This is also true for the other processes we studied, as fish population dynamics, habitat shifts, and changes in prey abundances operate at even longer timescales. Hence, the similar responses observed in our study at two disparate vent systems are more likely to be the result of longer-term exposure to mean conditions rather than short-term extremes. Moreover, geochemical studies at our two vent locations both concluded that these vents are suitable for studies on ocean acidification^{32, 33}. Nevertheless, it should be noted that, for some traits or ecological processes, sporadic exposure to extreme conditions can also have an impact.

Carbonate chemistry measurement and analysis.

The CO₂ concentration in the water was calculated using the values of temperature, salinity, pH_NBS and total alkalinity (TA) measured in the field. The software CO2SYS was used to estimate seawater pCO₂ with constants K1 and K2 from Mehrbach³⁵ and refit by Dickson and Millero³⁶. Alkalinity was measured by dynamic endpoint titration using a Titrando (Metrohm) titrator. During the study, values for standards were successfully maintained within 1% accuracy from certified reference materials from A. Dickson (Scripps Institution of Oceanography). All samples were collected from 14 to 18 September 2013 at Vulcano Island and from 18 to 21 November 2013 at White Island; additional data was added for 2 May 2013 and 9–11 February 2015 at White Island. The pH_NBS was measured daily at Vulcano Island with a portable probe SG2-ELK SevenGo (Mettler Toledo) and at White Island with the multi-meter and logger Sonde 6600V2 (YSI), which was calibrated daily. TA samples were collected at each site on three different days, fixed with mercuric chloride, and preserved in Duran glass bottles (Schott) until analysis, according to standard operating procedures³⁷. Salinity was measured with a SR6 refractometer (Vital Sine).

Study species.

A major drawback of using natural CO₂ vents as an experimental area representing future acidified ecosystems is that many animals move in and out of the vent areas and are therefore not continuously exposed to high CO₂. To avoid this limitation we focused on site-attached species that occupy a territory directly after settlement, show little movement, and have small home ranges. For both vent locations this was an acceptable approach, as the benthic fish communities were dominated by gobies (Gobiidae) and triplefins (Tripterygiidae). All of these species maintain territories of a few m² and are highly site-attached^{38, 39}. We focused on the benthic species that showed highest abundances at each study location (F. lapillum, common triplefin and G. bucchichi, Bucchich’s goby). The common triplefin is a habitat generalist found commonly in most habitats⁴⁰.

In addition, we determined densities of all other triplefin species at White Island, as potential competitors of the common triplefin.

We also quantified potential predators at both locations in roving transects using a video camera. At Vulcano, the only conspicuous predators were juveniles (~20 cm total length) of the relatively site-attached grouper Serranus scriba, which appeared to associate with rocks in seagrass habitat. No other free swimming predators were observed during any snorkel surveys. At White Island, the only conspicuous predators were relatively site-attached scorpionfish (Scorpaena spp.) and hiwi hiwi (Chironemus marmoratus); roving predators were not seen during these transects or other SCUBA surveys.

Fish escape behaviour and performance.

We designed a device that mimicked the approach of a potential threat while recording the fish’s escape behaviour (Supplementary Fig. 3). The device was a cubical frame made of white PVC pipes. A GoPro camera was attached to the top of the frame. A black iron rod was attached to the top of the frame and extended ~60 cm forward from the camera. A metal ruler of 30 cm was attached to the end of the rod, pointing downwards so that the bottom half of the ruler was in view of the camera’s field. The recordings were taken at a speed of 30 s per frame.

To elicit an escape response by the fishes, the tip of the ruler was lowered vertically towards the head of a randomly selected individual until the ruler reached the substratum. The camera was recording continuously and captured the entire threat approach and escape process. Both species showed very similar behaviour towards the approaching threat. The response of a fish was to first direct its eyes towards the approaching ruler, followed by a fast jump with a few tail flips when the ruler approached too close, before the fish settled back onto the substratum several centimetres away. The test was performed under animal ethics approval #S-2013-150.

A total of 209 individuals were tested at Vulcano Island (control: 107 fish; vent: 102 fish; 14–15 September 2013) and 146 individuals at White Island (control: 73 fish; vent: 73 fish; 20 November 2013). For the video recording of each individual fish we used the program VLC media player 2.0.1 to quantify: distance from the approaching ruler at which the fish initiated its escape response, distance covered during the escape, and duration of the escape, which was transformed to escape speed by dividing escape distance by escape duration. In addition, the habitat in which the fish resided during the mimicked attack was recorded as bare, non-vegetated substrate, turf algae, or small rocks. The escape was defined as from the moment at which the fish started its jump until it landed back onto the substratum. Because the fish were always approached from their side, their forward escape response was generally in a direction parallel to the line of sight (that is, escaping either towards the left- or right-hand side of the camera’s view, rather than towards or away from the camera). The distance at which the fish initiated their escape was measured as the distance between the top of the head and the tip of the ruler, using the gridded ruler as a reference for the magnification. Likewise, the distance moved during the escape was measured from the recording. All measurements were done by forwarding the recording frame by frame (1/30 s). The escape speed was calculated by dividing the escape distance by the number of frames to complete the escape.

Fish population structure.

Abundance and total body length of the two species (common triplefin at White Island and Bucchich’s goby at Vulcano Island) and other triplefin species (only at White Island) were visually quantified in replicate 2 × 10 m belt transects on snorkel at Vulcano Island and SCUBA at White Island (n = 3 transects at each of the two control and two vent sites at each of the two locations, for a total of six control and six high-CO₂ transects at each location).

Predators were quantified at both locations in roving transects (~4–5 min each) using a video camera. At Vulcano Island, two replicate transects at each site were performed at control (four total) and vent (four total) sites. At White Island, six video transects were performed at control sites and six transects at vent sites.

To determine whether CO₂ had a direct effect on fish habitat association, we quantified abundances of the common triplefin at White Island and Bucchich’s goby at Vulcano Island in 1 m² quadrats in different microhabitats. At each control and vent site at each location, ten quadrats (five per site) were randomly deployed in each of three microhabitats (30 quadrats in total at each site at each location): rocky barrens, turf macroalgae and kelp at White Island, and fleshy macroalgae (mainly C. prolifera), seagrass and sand at Vulcano Island.

Fish ages of the common triplefin were determined from otolith analyses. Otoliths were embedded in resin, sectioned transversely and placed on microscope slides. Sectioned otoliths were viewed under a compound microscope (Leica DMLB) and age estimated by counts of growth increments.

Fish diet composition and food abundance.

Gut content analysis was performed for the common triplefin collected at control (ten fish) and vent sites (nine fish) at White Island. Abundance of their prey was estimated from a total of 40 cores of turf at White Island (n = 10 at each replicate vent and control site). A small circular core (diameter 4.25 <