Warming increases the top-down effects and metabolism of a subtidal herbivore

Study system

The Galápagos Islands are located 965 km off the coast of Ecuador and are centered at the confluence of several oceanographic currents leading to a high degree of variability in ocean temperature and phytoplankton biomass. The Equatorial Undercurrent runs from west to east across the Pacific Basin and drives major upwelling as it collides with the western islands, resulting in cold and nutrient-rich waters around these islands (Houvenaghel, 1978; Houvenaghel, 1984). The cool Humboldt Current delivers nutrient-rich water to the southern edge of the archipelago (Kessler, 2006). The northeast region of the archipelago is strongly influenced by the warm, nutrient-poor waters of the North Equatorial Countercurrent (NECC, or the Panama Current) (Kessler, 2006). Both of these currents contribute to the South Equatorial Current (SEC), a westward flowing current that strongly influences the central region of the archipelago (Houvenaghel, 1984) (Fig. 1).

The dominant influence of the SEC changes seasonally, depending on the location of the Intertropical Convergence Zone (ITCZ) (Houvenaghel, 1984). The ITCZ is north of the equator during the Garùa (fine mist) season (May to December), and the Humboldt Current is the major contributor to the SEC. During the wet season (December to May) the ITCZ shifts towards the south and the dominant influence to the SEC is the NECC. This results in fluctuating gradients of temperature and resource availability throughout the central archipelago.

The maximum average SST across the archipelago occurs in February and March, with the minimum usually occurring in August or September (Houvenaghel, 1978; Schaeffer et al., 2008) (Fig. 2). Temperature data were collected from two sites on San Cristobal (the easternmost island in the archipelago). During 2011, there was a cold season low of 13°C in November 2011 and a high of 25°C in July 2011. In 2012, there was a warm season high of 30°C in February 2012 and a low of 20°C in May 2012 (Fig. 2 and File S1). This annual cycle is interrupted during ENSO events (Barber & Chavez, 1983; Chavez et al., 1999).

During the warm ENSO phase (El Niño), the easterly trade winds weaken resulting in a deeper EUC thermocline, warmer waters and decreased upwelling. This results in low standing stock of primary producers in the euphotic zone (Pennington et al., 2006); and, ultimately marine consumer populations, such as seabirds, marine iguanas, and sea lions, decrease in abundance (Valle et al., 1987; Laurie & Brown, 1990). In contrast, the cold ENSO phase (La Niña) occurs as the easterly trade winds strengthen, sea surface temperatures decrease, and upwelling intensifies resulting in higher standing stock of primary producers (Izumo, Picaut & Blanke, 2002). During La Niña, ocean temperature in the western islands can be as low as 11°C (Wellington, Strong & Merlen, 2001).

In addition to regional-scale spatiotemporal variability in temperature and resource availability, the upwelling and downwelling of internal waves result in extreme and rapid temperature changes over smaller spatial and temporal scales (Witman & Smith, 2003). For example, over a 53-week period on a rocky subtidal wall in the central archipelago at depths between 3 and 12 m, 20 cold water events were recorded where temperature dropped by 3–9°C over a 25 h period (Witman & Smith, 2003).

Study organisms

We measured the effect of temperature on the common subtidal herbivore Lytechinus semituberculatus (green sea urchin) and the green macroalga Ulva sp. Lytechinus, Tripneustes depressus (white sea urchin), and Eucidaris galapagensis (slate pencil urchin) are the three most common urchin species in the Galápagos Islands and together comprise 91% of urchin biomass (Brandt & Guarderas, 2002; Brandt, Witman & Chiriboga, 2012). In rocky subtidal habitats throughout the archipelago at depths between 1 and 5 m, Lytechinus and Eucidaris are the two most common urchins, while Tripneustes is rare (Table 1). Further, Lytechinus is a strong interactor and capable of converting algal turfs (brown filamentous turf, order Ectocarpales: Giffordia sp., Ectocarpus sp.) to urchin barrens in relatively short time periods, while Eucidaris does not have any detectable effect on the abundance of algal turfs (Irving & Witman, 2009).

Ulva sp. was chosen as a food item for Lytechinus for several reasons: (1) Ulva sp. are one of the most abundant macroalgal species, along with turf, crustose coralline algae, and Sargassum, in the Galápagos nearshore habitats (Vinueza et al., 2006; Vinueza, 2009); (2) ephemeral species, like Ulva, are highly palatable for herbivores (Carpenter, 1986); and (3) sea urchin fronts in the Galápagos appear to consume all macroalgal species except for brown species (e.g., Padina) (L. Carr personal observation) and damselfish turfs (Irving & Witman, 2009).

Urchins and Ulva were haphazardly collected from the southern part of San Cristobal Island (89°36′41.85″ W, 0°55′39.36″ S), and were immediately transported to the laboratory in buckets filled with seawater. All urchins were collected from a depth of ∼1.5 m. Study organisms were maintained in culture tanks indoors at 23°C (ambient sea water temperature) for two days prior to beginning water temperature adjustments. Assays were conducted in a shaded, outdoor facility at the joint UNC/USFQ Galápagos Science Center (San Cristobal Island, Galápagos).

Herbivore feeding rate

We conducted feeding rate assays in July 2012 to test the effect of two different temperatures (14°C or 28°C) on green urchin grazing rates on Ulva. Water temperature in the culture tanks was adjusted from ambient (23°C) to either 14°C or 28°C over a four-day period. This is within the time period of shallow subtidal temperature changes of this magnitude in the Galápagos (Witman & Smith, 2003).

Ulva and urchins were placed in 4-L plastic container mesocosms and received a fresh supply of temperature-conditioned seawater every 12 h. Temperature treatments were maintained with either Visi-Therm submersible individual heaters (Marineland, Blacksburg, Virginia, USA) or ice baths. Feeding assays were replicated twice (n = 5 replicates for each trial). Herbivore presence and absence treatments were randomly assigned in water tables. Each mesocosm was equipped with an iButton Thermochron datalogger (Dallas semiconductor, Dallas, Texas, USA) and water temperature was recorded every 5 min. Eight mesocosms were equipped with a HOBO Pendant temperature/light sensor (HOBO, Bourne, Massachusetts, USA) and relative light intensity was measured every 5 min.

Starting conditions for each mesocosm were 2.50 ± 0.004 g of wet mass Ulva tissue and either three urchins or no urchins (control to test for autogenic loss). The average test size for the urchins in the mesocosms was 3.55 ± 0.08 cm, which is representative of the green sea urchin populations in southern San Cristobal (n = 120 from two sites measured in May and June 2011: minimum 2.75 cm, maximum 5.8 cm. Mean ± 1 SE of 3.79 ± 0.61 cm).

Assays were terminated and final algal biomass was measured after 48 h (when ∼50% of algal tissue was consumed (Tomas et al., 2011)). Biomass consumption was estimated as ([H_i × C_f/C_i]−H_f), where H_i and H_f were the initial and final wet weights of algal tissue in the presence of herbivores, and C_i and C_f were initial and final wet weights of the controls. Relative light intensity levels did not vary between mesocosms and was 326.06 ± 19.7 lumens/ft². These light levels are less than the average relative light levels at 1.5 m depth in the field (886.86 ± 74.07 lumens/ft²), but are within the range of light conditions experienced throughout the tidal cycle at the southern sites of San Cristobal Island during the month of June (Table 1).

Feeding rate assays were initially analyzed using a mixed model ANCOVA with one level of nesting. The analysis tested for one fixed effect (temperature treatment), covariate (urchin test size), and one random effect (temporal block). Consumption data were log transformed to meet the assumption of homogeneity of variances. The random effect was not significant (p = 0.183). Therefore, results were pooled and the random effect was dropped for final analysis of treatment effects. All statistical analyses were performed in R (v. 2.15.2).

Respiration and photosynthesis

To estimate the temperature response of metabolic pathways (net photosynthesis and respiration), we measured oxygen production and consumption rates for Ulva and green urchins in 0.6 L containers under conditions identical to the feeding rate assays. Initial and final oxygen concentrations were measured for Ulva (5 ± 0.08 g of leaf tissue) and paired blanks (seawater only) (n = 20 replicates) using a YSI-200 oxygen sensor (Yellow Springs Instruments, Yellow Springs, Ohio, USA). Samples of Ulva tissue (5 ± 0.08 g) were obtained by using three Ulva “rosettes” plucked from the substrate by the holdfast. Rosettes used were similar sizes and no cutting or tearing was necessary. After the initial measurement, aquaria were covered with plastic to minimize oxygen exchange with the air and left for 2 h. Net photosynthesis rates were estimated by subtracting measurements of dark oxygen consumption from light oxygen production.

Lytechinus oxygen consumption rates were measured according to methods described in Siikavuopio, Mortensen & Christansen (2008). Urchins were held at 23°C prior to oxygen consumption trials. One randomly selected urchin was placed into an airtight, closed-system respirometry chamber (0.6 L) and water temperature was gradually adjusted over a 10 h period to 14°C and then back up to 28°C over a 30 h period. Within the chamber was a mounted YSI 200 dissolved oxygen probe to measure oxygen concentrations (mg/L) and a pump for water circulation to prevent the development of strong oxygen and temperature gradients. The chamber was placed into a water bath to maintain temperature treatments. An individual urchin was then placed into the chamber and oxygen concentrations were measured every five minutes for one hour at each temperature treatment (14°C and 28°C). Trials were repeated for 11 urchins. The mean weight specific oxygen consumption rate (Q, mg O₂ kg⁻¹ h⁻¹) was calculated with the equation of Karamushko & Christiansen (2002): ${\displaystyle Q=(C_0-C_t)V/WT}$C₀ and C_t are the initial and final oxygen concentration (mg O₂ߙl⁻¹), respectively. V is the volume (l) of the chamber minus the test urchin volume (test urchin volume was estimated from their biomass). W is the biomass of the urchin in kg. T is the measurement time in hours.

Oxygen consumption and production test were analyzed with a t test on change in O₂. All statistical analyses were conducted using R (version 2.15.2).

Herbivore feeding rate

Temperatures in the cold mesocosms were maintained at 14.01 ± 0.08°C and 14.03 ± 0.07°C for trials 1 and 2, respectively. Warm mesocosms for trials 1 and 2 were 28.06 ± 0.09°C and 28.00 ± 0.04°C, respectively. The range of temperatures maintained across both trials was 13.54–14.47°C for the cold treatment and 27.56–28.5°C for the warm treatment (Fig. 3). Green urchin consumption was 46% higher at the warmer temperature (p < 0.0001) (Fig. 4A). Urchin test size was not a significant covariate (p = 0.87).

Respiration and photosynthesis

Green urchin metabolism was significantly higher at 28°C than at 14°C (p < 0.001, Fig. 4B). Ulva oxygen consumption was greater at 28°C (at 14°C, 2.55 ± 0.11 g O₂• g tissue^-1• h^-1; at 28°C, 3.24 ± 0.13 g O₂• g tissue ^-1• h^-1; p = 0.004). Oxygen production was also greater at 28°C (at 14°C, 3.88 ± 0.09 g O₂• g tissue^-1• h^-1; at 28°C, 4.74 ± 0.15 g O₂• g tissue ^-1• h^-1; p = 0.01). However, net photosynthesis rates did not vary with temperature (p = 0.45, Fig. 4C).