1"

INTERACTING EFFECTS OF PHYSICAL

DISTURBANCE AND OCEAN

ACIDIFICATION STRESS ON THE

CALIFORNIA MUSSELS, MYTILUS

CALIFORNIANUS

Andrew B Hines

Hopkins Marine Station

May 2014

2"

Interacting effects of physical disturbance and ocean acidification stress on the California mussel, Mytilus californianus

An Honors Thesis Submitted to

the Department of Biology in partial fulfillment of the Honors Program

STANFORD UNIVERSITY

Andrew Hines May 2013

3"

Acknowledgments Two years is a long time to work on an undergraduate thesis, and as the time

passed, it became increasingly easy to rationalize giving up, falling back on personal

hardship as an excuse to turn in, switch off, and hide. But for the grace and support of my

advisor, Fio Micheli, I would have never had the strength to finish this document. Every

word is a product of that support, of Fio’s tireless care and intelligent comments, keeping

me going through a two-year period that saw me finish this thesis and regain the joys of

being a scientist. Fio, together with the stats acumen of Jim Watanabe, the fieldwork of

Cheryl Butner, and the comments of Steve Palumbi, have made my time at Hopkins a

growing experience. I owe a debt of gratitude (an in one case, pastries) to all those who

have aided me in this endeavor, but it is with the utmost gratitude that I thank Fio, Steve,

Jim, and Cheryl.

To my friends Ana Guerra, Chelsea Wood and Jen Cole, whose stories and beers

and theories-on-the development-of-vampirism-in-reality are some of my fondest

memories, I offer thanks. I will miss our late-night conversations and early-morning

dives/dances. I would also like to thank Francesco Ferretti, Ashley Greenley, Kristy

Kroeker, Sarah Lummis, Steve Litvin, and Paul Leary; Fiolab has been as a family to me,

teaching me more than any classroom.

I would also like to acknowledge the wonderful staff of Hopkins Marine Station;

Doreen Zelles, Freya Sommer, John Lee, Chris Patton, Joe Wible, Don Kohrs, and of

course, Judy Thompson, who keep our little world spinning.

4"

Abstract One of the main impacts of human carbon dioxide emission is the decline of

oceanic pH. The concomitant reduction in carbonate ion concentration has proven

deleterious to range of calcareous marine species, many of which are also subject to

habitat loss and disturbance due to human exploitation of the coast. Here, I examined the

interaction of pH and dislodgement stress on the California mussel, Mytilus

californianus. First, pH was found to be the dominant stress affecting growth; however,

due to pseudoreplication, a second experimental run was conducted, which suggested that

pH and disturbance act synergistically to reduce growth rates. Further analysis

determined that pH and disturbance did not significantly differ between experiments, but

that potential compensatory effect had been acting in the tanks: increased pCO2 in tank

water may have allowed mussels to offset the metabolic demand of hypercapnia and

byssogenesis. These results highlight the difficulty of simulating the intertidal

environment in a controlled fashion and the necessity of continued study of compound

stress effects for the management of coastal resources under climate change.

5"

Table of Contents Introduction ....................................................................................................................... 6

Materials and Methods ..................................................................................................... 8

Tank setup ....................................................................................................................... 9

Analysis ........................................................................................................................... 9

Results .............................................................................................................................. 11

Discussion ......................................................................................................................... 12

Magnitude of treatment levels ....................................................................................... 12

Implications and Future work ....................................................................................... 14

References ........................................................................................................................ 16

Tables ................................................................................................................................ 19

Figures .............................................................................................................................. 20

6"

Introduction Recent research has shown that human activities are driving large-scale rapid

change in the oceans through multiple co-occurring stressors (Pachauri 2007, Halpern et

al. 2008, Rockström 2009). Rising atmospheric CO2 results in the oceanic uptake of CO2

(Doney 2010). This carbonate pump is governed by the following equations,

CO2(aq) + H2O ↔ H2CO3 ↔ H+ + HCO3- ↔ H+ + CO3

2- (1)

Ca2+ + 2HCO3- ↔ CaCO3 + CO2 + H20 (2)

An increase in atmospheric CO2 results in elevated oceanic bicarbonate ion concentration

(HCO3-) and hydrogen ions (H+) and a decrease in the concentration of carbonate ion

(CO32-) (Gattuso & Buddemeier 2000). Thus, influx of CO2 into seawater drives a shift in

the inorganic carbon equilibrium, from a CaCO3 precipitating environment, towards a

carbonate ion (CO32-) limited one (Zondervan et al. 2001).

The overall effect of this process is termed Ocean acidification, and has been

shown to affect marine species both positively and negatively, via several mechanisms

(adapted from Kroeker et al. 2010): (1) increased pCO2 and HCO3-, (2) decreased pH,

and (3) decreased CO32- concentration. An increase in pCO2 and HCO3

- concentrations is

expected to benefit some marine autrotrophs by allowing increased photosynthetic

efficiency (Hurd et al. 2009). Decreased pH is expected to induce hypercapnia in marine

heterotrophs, which in turn results in metabolic depression (Langenbuch & Pörtner

2004). The regulation of extracellular pH is an energetically intensive process, and OA-

induced hypercapnia may force energetic trade-offs between growth and homeostatic

maintenance (Pörtner et al. 2004). Finally, CO32- is required to form the calcium

carbonate (CaCO3) shells used by many marine species. In a carbonate ion limited

environment, calcification rates will likely decrease while the rate of dissolution of

calcium carbonate increases (Kleypas et al. 1999, Gattuso & Buddemeier 2000, Orr et al.

2005).

However, the responses of marine calcifiers to hypercapnic conditions are

intensely variable (Ries et al. 2009, Kroeker et al. 2010). Some species respond by

reducing growth rates while others, such as the blue mussel, Mytilus edulis thrive even in

acidic (hypercapnic) conditions (Kroeker 2010). Multiple hypotheses accounting for this

7"

variation in biological response have been proposed, but they all treat ocean acidification

(OA) as the dominant stressor. The resilience of Mytiloids to hypercapnia is due to their

metabolism and feeding ecology (Thomsen et al. 2010, Melzner et al. 2011) thus,

introducing disturbance as an additional component of stress creates a more realistic

stress regime, which targets mussel’s energy allocation mechanisms rather than their

shells alone. While lowered pH can be an important factor in the calcification rates of

marine mollusks (Gazeau et al. 2007), it is not the sole factor in the intertidal

environment.

Crain et al. [2008] found that stressors such as lowered pH and dislodgement can

be additive, synergistic, or antagonistic in combination. These categories, originally

described by Folt et al. [1999] are defined as follows: two stressors with negative

impacts, “A” and “B” are applied to an organism, and result in reduced amounts “a” and

“b”, respectively. This negative response is a reduction from the normal value being

measured, e.g. growth rate, and can be characterized as additive (response = a + b),

antagonistic (response < a + b), or synergistic (response > a + b) (Folt et al. 1999, Crain

et al. 2008). This classification provides a useful system for determining the forcing

stresses in an organism’s environment, and which are dominant. Given that the individual

effects of both acidification and trampling have been observed in marine intertidal

ecosystems (Thomsen et al. 2010, Melzner et al. 2011, Brosnan & Crumrine 1994), it is

important that we understand how they act together. Trampling mussels can remove

individuals from the substrate, and reattachment after dislodgment is an energetically

intensive process (Brosnan & Crumrine 1994). The combined impact of lowered pH on

calcification, and the energetic demand of byssogenesis, could result in altered energy

allocation, reduced growth rates, and decreased survival of individuals (Wood et al. 2008,

Kroeker et al. 2010). Furthermore, if lowered pH and dislodgement were found to

simultaneously depress the metabolic rate of a mussel, resulting in a reduced growth rate

(when compared to the effect of only pH, or only disturbance), they may be additive.

However, if the effect of pH alone overshadows that of dislodgement and pH combined,

pH stress would be the dominant force in mussel growth dynamics.

8"

This study tested the effects of physical dislodgement and acidification on

mussels. Dislodgment (from trampling and wave action) and ocean acidification have

tremendous impacts on coastal ecosystems singularly (Brosnan & Crumrine 1994,

Halpern et al. 2008), but are experienced by coastal organisms as discrete stressors.

California’s shores experience heavy traffic and, over the next 40 years, are projected to

experience a reduction in pH and CaCO3 to an average of 7.82 ± 0.04, and 1.26 ± 0.12,

respectively (Gruber et al. 2012) By subjecting the California mussel, Mytilus

californianus to both of these anthropogenic stressors, I attempted to gain a better

understanding of possible synergisms in nature, of which there is little information

(Darling & Coté 2008, Crain et al. 2008). I focused on two main questions, (1) how do

pH and dislodgement affect shell growth separately, and (2) when combined, are the

stressors additive, or is there a dominant mechanism that determines M. californianus

growth? I hypothesized that pH and disturbance treatments would have an additive

effect, reducing the somatic growth rate of mussels due to the energetic demands of

homeostasis.

Materials and Methods M. californianus is commonly found on the west coast of North America,

occurring in rocky intertidal zones from Mexico to Alaska. Hopkins Marine Station of

Monterey Bay lies near the southern edge of the range, and provided a large closed-

access population of mussels for this study. 200 mussels were haphazardly sampled from

one of the northern beds near China point (36°37’18’N, 121º54’19”W), during late July-

early August 2012 (Supplement A). Within the bed, mussels were selected for similar

length. Each mussel was cleaned of epibionts, measured for initial length and width using

digital calipers (Supplement B) n=160, !! = 31mm SDL= 3.9mm, !! = 15mm,

SDW=1.9mm), then tagged. Using a random number generator, mussels were assigned to

one of 4 treatment bins, 20 mussels to a bin. An additional group of mussels was assigned

to mesh bags staked in the sampled plots to create an external control. All subjects were

kept in a fasting state during the experiment, with no food administered other than the

particulates generated in the tanks. As tanks were a common resource, with multiple

experiments and species sharing space, subjects were housed in modified open-top

9"

Tupperware containers (10x10x8cm, with grit-tape bottoms). Two mesh bags of 20

individuals were staked in the intertidal using zip ties and steel bolts, but were

subsequently found ripped open, with all mussels missing, on 10 August.

The entire experiment was replicated across two samples of mussels, for a total of

8 treatment bins, two of each pH and disturbance combination (Figure 8). After a one-

week acclimation period, experimental treatments began, and were administered weekly

(Supplement C). Disturbance treatments involved complete removal of an individual’s

byssus complex via wrenching from the substrate (plastic Tupperware tub with grit-tape),

while pH treatments were constant in the tanks. Measurements of mussel growth were

taken monthly (Supplement B and C) and required the exposure of each treatment group

to air for 30 minutes to 1 hour, during which time length and width was measured for

each individual. Immediately post-measurement, mussels were returned to their

treatments.

Tank setup

All experiments were conducted in the Aquaria complex of Hopkins Marine Station. Two

separate pH tanks were constructed (Figure 9) and filled with seawater from Monterey

Bay Aquarium to a depth of 8cm. Tank water temperature and dissolved oxygen were

measured continuously during the experiment using an AADI 4500a sensor. Tank pH

was measured as ion voltages using Honeywell durafet sensor, then converted to pH

using the following equation (calibrated 8/10/12 and 8/12/12 respectively by HMS staff):

!": !"!"#$%&' !!!! − !! (3)

where !"!"#$%&' !is the measured value, and ! and ! are probes-specific corrections.

Experimental pH values were set by the pH of incurrent water from the Monterey Bay

Aquarium (pH = 7.75 ± 0.05), and the approximated ambient pH of Monterey Bay (pH =

7.95).

Analysis

To determine the effect of pH-Disturbance treatment combinations, I compared the post-

treatment surface areas of each mussel to its pre-treatment value. As a baseline for

comparison, the Undisturbed and High pH (UH) treatment group was chosen. If pH and

10"

Disturbance had a negative effect on growth rate, the UH treatment should exhibit the

highest growth. Surface area measurements were taken monthly, and included removal

of subjects from treatment tubs, and recording of length and width. For a three-

dimensional shell, measurements of the longitudinal axis (length) and latitudinal axis

(width) were used to approximated the surface area of a mussel as the combined surface

area of two simple ellipses:

!"#$%&'!!"#$! !.!. = !2!"# (4)

Where ! is the latitudinal axis, and ! is the longitudinal axis. Lacking a third dimension,

and given the mathematical complexity of calculating the surface area of an ellipsoid, the

above method was deemed sufficient.

The effect of pH and disturbance was calculated as the incremental growth

response ratio !:

! = (!!!!!)!!

! (5)

Where !! is the final surface area, and !! is the initial (pre-treatment) surface area. The

response ratio quantified the incremental change resulting from experimental

manipulations. A positive response ratio indicated the treatment had a positive effect on

growth, while a negative value indicated a negative effect; a value of zero indicated no

effect.

Initial comparison of the variance in growth ratios was conducted using twin 2-

way Factorial ANOVAs, (Model I), with pH and Disturbance as fixed, orthogonal

factors, one for each experimental run. Variation in the growth rate among treatment

combinations across samples was tested with a 4-way Model II ANCOVA, with pH and

Disturbance as orthogonal fixed factors, initial surface area as a covariate, and sample

group as a random factor nested within pH and disturbance, after tests for normality and

homogeneity of variance. Due to the random nature of the sample groups, I used a Model

II (Mixed) effects model to calculate the overall mean effect for each stressor variable. A

mixed model accounted for the variation in treatment response due to the different

sample groups, as they were nested within tanks, while still allowing me to determine the

experimental variation due to my treatments. Mixed effects models account for this

variation by calculating the between group variance of the nested factor !"!"!"#$ and

weighting the significance of each experimental variable according to that term.

11"

Results The observed variation in growth rates of M. californianus exposed to low pH and

dislodgement did not support the hypotheses that these stressors have a negative impact

on mussels, nor did it provide conclusive evidence that ocean acidification and physical

dislodgment are synergistic or additive stressors. During the first trial (sample 1), growth

variation of experimentally manipulated mussels differed significantly among treatments

(Table 1). Analysis of variance of sample group 1 found only pH to be a significant factor

(2-way factorial ANOVA; pH: F(1,73)=5.75, P<0.01). Growth was greatest in the

Disturbed and Low pH group (Figure 1), suggestive of a synergistic effect of the two

stressors. Mussels in the Undisturbed treatments (undisturbed-low pH, UL, and

undisturbed-high pH, UH) and in the Disturbed and High pH treatment (DH) were found

to have a mean growth ratio significantly less than the Disturbed-low pH (DL) treatment,

but were not significantly different from each other. To determine if the positive effect of

pH in Sample group 1 was a treatment effect, and not an artifact of pseudo-replication of

pH tanks, a second experimental run was conducted (Figure 1, Sample 2). Sample Group

2 exhibited markedly different responses to the treatments compared to sample 1, with

the highest mean growth occurring in the UL treatment; lowest mean growth occurred in

the High pH treatments (UH and DH). However, growth rates were not significantly

different among these treatments. The interaction of pH and Disturbance was found to be

significant (Table 2); however, in contrast with sample 1, growth was lower when both

treatments were applied together, suggestive of an antagonistic effect (Figure 1).

The opposing trends found in each experimental run indicated that additional

confounding factors were affecting the growth of mussels. To account for those

confounding factors, Sample group was added to the model as a random, nested factor

(within pH and Disturbance treatments), with initial surface area included as a covariate.

However, because the effect of initial size was found not to be significant (3-way

ANCOVA F(1,4)=0.43, P=0.55), this factor was removed. ANOVA indicated that only

the nested term of sample group was significant (4-factor Model II ANCOVA;

F(4/156)=6.23, P<0.001; Table 3. When the two experiment runs were combined, the

effects of treatments were no longer significant. Due to the opposite effects of treatments

in the two experimental runs, interaction plots of treatments and growth rates were

12"

generated to determine whether treatments had a positive or negative overall effect,

within each sample group. Decreased pH was found to have a positive effect on growth

(Figure 2). The effect of Disturbance was both positive (sample 2) and negative (sample

1), and indicated an additional interaction (Figure 3). The interaction of pH and

Disturbance was found to have a positive effect in sample 1, and a negative effect in

sample 2; mussels in UH treatments grew less than mussels in DH treatments, while

mussels in DL treatments grew less than those in UL treatments (Figure 4). In addition,

the difference in mean growth ratio of DH and DL mussels in sample group 2 was five

times that of sample group 1.

Discussion My results do not offer conclusive evidence of a positive or negative effect of pH

or Disturbance. Furthermore, due to the highly variable treatment responses and the

resulting non-significance of each factor, it is impossible to determine whether or not pH

or Disturbance act synergistically, additively, or antagonistically. Thus, while low pH

appears to have an overall positive effect on mussel growth, further experiments are

needed to assess the significance of this effect and whether it interacts with dislodgement

disturbance. However, the variation of effects prompted by just two stressors illuminates

the complex nature of modeling forces in the intertidal. The results of this experiment are

limited due to pseudoreplication (only one ambient and one low pH tank were available).

Magnitude of treatment levels

The response of growth ratio to low pH appeared to be positive, and thus contradictory to

the established literature on ocean acidification, which contends that reduced pH shifts

the oceans from CaCO3 deposition to dissolution (Zondervan et al. 2001). Given the non-

significance of pH in my analysis, increased growth of mussels in the UL and DL

treatments suggests that the reduction in pH may have resulted in a separate,

compensatory reaction. While not directly measured, it is possible the increased CO2

concentration in the waters boosted algal growth (Ries et al. 2009, Kroeker et al. 2010)

providing an uncontrolled source of nutrients to offset the metabolic strain of

13"

hypercapnia. An increase in pyhtoplanktonic photosynthesis could buffer the negative

effect of ocean acidification on calcification (Ries et al. 2009, Kroeker et al. 2010).

However, growth in the High pH tanks was only marginally different from the Low,

indicating an even simpler explanation: pH treatments were biologically within M.

californianus tolerances. In particular, the pH values used in this experiment differed by

0.2 points, and were not constant (Figure 5).

Mytilus edulis, a close relative of M. californianus, was found to be unaffected by

pH as low as 7.4 (Berge et al. 2006). In designing my experiment, I referred to the work

of Michaelidis et al. [2005] who described declines in growth and metabolic activity

under hypercapnic conditions. During periods of hypercapnia, mussel shell growth was

reduced due to an increase in homeostatic energy expenditure (Thomsen & Melzner

2010, Matsuhiro & Miyashita 2004). In response to hypercapnia, energy expenditure

rose 58% above control level (Michaelidis et al. 2005, Thomsen et al. 2010). Thomsen et

al hypothesized that decreased shell growth was a consequence of higher energy demand

and existing energy re-allocation in a lowered pH system. The mussels must expend

more energy in ion regulation, and overcome essential nitrogen limitation brought on by

increased protein turnover. If the pH stress overcomes the mussels’ capacity to restore

somatic tissue, increased energy must be put into maintenance of existing tissue, resulting

in lower shell growth rate.

However, this work differed greatly from previous studies by measuring the

change in surface area rather than shell thickness or shell weight, both of which are

thought to be a more likely to respond to pH stress than overall reduction of growth rate

(Bamber 1990). In fact such mechanisms could still have been occurring in my

experiment, as a reduction in respiration rates (as measured by Michaelidis et al. [2005])

could imply a decrease in the interpallial pCO2, and therefore an increased ability to fix

CaCO3 (Gazeau et al. 2007). Thus, even mussels exposed to lowered pH could maintain

growth rate, if they reduced overall metabolic rate. This compensatory effect is

compounded if the mussels are able to feed, which would allow them to maintain their

metabolic rate and continue to grow (Thompsen & Melzner 2007). Even if the pH was

well below the norm, as long as they’re feeding, the mussels can maintain homeostasis

and continue to grow, as demonstrated by Thompsen & Melzner [2007].

14"

The compensatory effect of feeding, combined further obfuscates the effect of

disturbance. Disturbance treatments used in this study were determined from the work of

Brosnin & Crumrine [1994], who analyzed the effect of trampling and physical damage

on intertidal communities. I simulated the effect of human trampling and wave action on

mussels as a wrenching force that removed the mussels from the substrate, and destroyed

the plaque of the byssus complex. I reasoned that byssogenesis would be prioritized in

the intertidal, but neglected any size or age-based variation in byssal strength. To

maintain balance in my analysis, I avoided further damage to the shells, focusing only on

the energetics of byssogenesis while neglecting to standardize the force applied to the

mussels during treatment.

Experiments by Babbarro et al. [2007] suggest an additional metabolic strategy

available to Mytiloids in the management of byssogenesus, strategies that invalidate my

singular measurements of growth. Babarro and his colleagues found that fasting adult

mussels were able to maintain byssus secretion and attachment strength via the “transfer

of energy between soft tissues and byssus under stress,” resulting in rapid reattachment

and mobility. Since the secretion and strength of byssus threads is known to respond to

environmental flow rate (Lee 1990, Babbarro et al. 2007), and the water of the tanks

circulated without significant force, byssus destruction presented no real stress. Thus, the

manner and frequency of disturbance meted out in my treatments is also a poor estimate

of the conditions found in the intertidal. If the mussels could feed on particulates in the

tanks, and byssus secretion in a slow-flowing tank is not metabolically taxing, weekly

disturbances that left the shell intact could not compare to the millennia of waves and

storms survived by the mussel’s evolutionary precursors.

Implications and Future work

While non-significant on their own, the results of these experiments provide useful

preliminary data for future work on interacting stress effects. In particular, further work

should include a careful selection of stressors, both obvious, like water temperature

(Rodolfo-Metalpa et al. 2011), and nuanced, such as water flow-rate (Babarro et al.

2007) as well as greater depth of measurement. The molecular and physiological aspects

of stress responses offer important insights into organismal coping strategies, like those

15"

of Mytilus edulis (Melzner et al. 2010), and cannot be approximated with a single

measurement. Finally, the results of this experiment highlight the importance of

determining realistic stress parameters; my experiments were inconclusive due to

marginal levels of pH and disturbance, which proved inconsistent as stressors. Both

realistic and extreme values should be considered; recent experiments by Gruber et al.

[2012] have concluded that the California Current System is rapidly approaching pH and

aragonite (a form of CaCO3) saturation levels “well outside the natural range,” with

“frequent or even persistent under-saturation conditions.” The true effect of OA cannot

be accurately modeled without multiple pH levels that span the natural and worst-case

IPCC scenarios. In addition, while boosted algal growth may have sustained the mussels

in this experiment, the evidence is mounting that OA will have a negative impact on

global phytoplankton populations (Boyce et al. 2010, Hofmann et al. 2011). Marine

phytoplankton is the basis for the oceanic food web, and can provide a buffer against the

stress of life in the oceans. Without that buffer, mussels are fighting a losing battle; as go

the phytoplankton, so goes everyone else.

16"

References 1. Babarro, et al., Secretion of byssal threads and attachment strength of Mytilus

galloprovincialis: the influence of size and food availability, J. Mar. Bio. Assoc.

UK, 88.04 (2008)

2. Bamber, R. N., The effects of acidic seawater on 3 species of lamellibranch

mollusc, J. Exp. Mar. Biol. Ecol., 143 (1990)

3. Berge, et al., Effects of increased sea water concentrations of CO2 on growth of

the bivalve Mytilus edulis L, Chemosphere, 62 (2006)

4. Boyce, D. G., et al. Global phytoplankton decline over the past century.

Nature 466 (2012)

5. Brosnan, D, Crumrine, L., Effects of human trampling on marine rocky shore

communities, J. Exp. Mar. Bio. Ecol., 177, (1994)

6. Crain, C. Kroeker, K. & Halpern, B. Interactive and cumulative effects of

multiple human stressors in marine systems, Ecol. Lett, 11, (2008)

7. Darling, S. & Coté, I. Quantifying the evidence for ecological synergies, Ecol.

Lett. 11, (2008)

8. Doney, S.C. et al., Ocean acidification: the other CO2 problem, Annu. Rev. Mar.

Sci., 1, (2009)

9. Doney, S.C., The Growing Human Footprint on Coastal and Open-Ocean

Biogeochemistry, Science, 328, (2010)

10. Folt, C.L., et al. Synergism and antagonism among multiple stressors. Limnol.

Oceanogr., 44, (1999)

11. Turley, C.M. et al., Novel microcosm system for investigating the effects of

elevated carbon dioxide and temperature on intertidal organisms, Aqua Bio, 3,

(2008) DOI:10.3354/ab00061

12. Gazeau, F. et al. Impact of elevated CO 2 on shellfish calcification. Geophysical

Research Letters, 34, (2007).

13. Gattuso J.P. & Buddemeier R.W., Ocean biogeochemistry: Calcification and

CO2. Nature, 407 (2000)

14. Gruber, Nicolas, et al., Rapid progression of ocean acidification in the California

Current System." Science 337 (2012)

17"

15. Halpern, B. et al., A Global Map of Human Impact on Marine Ecosystems,

Science, 319, (2008) DOI: 10.1126/science.1149345

16. Hoegh-Guldberg, O. & Bruno, J., The Impact of Climate Change on the World’s

Marine Ecosystems, Science, 328, (2010) DOI: 10.1126/science.1189930

17. Hoegh-Guldberg, O. et al., Coral reefs under rapid climate change and ocean

acidification, Science, 318, (2007) DOI: 10.1126/science.1152509

18. Hofmann et al., Declining ocean chlorophyll under unabated anthropogenic

CO2 emissions, Environ. Res. Lett 6 (2011) doi:10.1088/1748-9326/6/3/034035

19. Hurd C.L., et al. (2009). Testing the effects of ocean acidification on algal

metabolism: Considerations for experimental designs. J. Phyc., 45 (2009)

20. Kleypas J.A., et al. Geochemical consequences of increased atmospheric CO2 on

coral reefs. Science 284 (1999)

21. Kroeker, K. J. et al., Meta-analysis reveals negative yet variable effects of ocean

acidification on marine organisms, Eco. Lett., (2010). DOI: 10.1111/j.1461-

0248.2010.01518.x.

22. Langenbuch M. & Pörtner H.O. High sensitivity to chronically elevated CO2

levels in a eurybathic marine sipunculid. Aquatic Toxicology, 7 (2004).

23. Lee C.Y., Shirley S.L. and Owen M.D., The rate and strength of byssal

reattachment by blue mussels (Mytilus edulis L.), Can. J. Zoo., 68, (1990)

24. Matsushiro A., Miyashita T., Evolution of hard-tissue mineralization: comparison

of the inner skeletal system and the outer shell system, J. Bone. Miner. Metab. 22

(2004) 25. Michaelidis B. et al., Effects of long-term moderate hypercapnia on acid–base balance and growth

rate in marine mussels Mytilus galloprovincialis, Mar Ecol Prog Ser 293 (2005)

26. Melzner F, et al., Food Supply and Seawater pCO2 Impact Calcification and

Internal Shell Dissolution in the Blue Mussel Mytilus edulis, PLoS ONE 6 (2011)

27. Orr J.C., et al., Anthropogenic ocean acidification over the twenty-first century

and its impact on calcifying organisms. Nature, 437 (2005)

28. Pachauri, R., Climate Change 2007: Synthesis Report, IPCC Secretariat, Geneva,

(2007)

29. Thomsen, J. & Melzner, F., Calcifying invertebrates succeed in a naturally CO2

enriched coastal habitat but are threatened by high levels of future acidification,

18"

Biogeosci. Dis. 7 (2010)

30. R Core Team, R: A language and environment for statistical computing. R

Foundation for Statistical Computing, Vienna,Austria. (2014), URL

http://www.R-project.org/.

31. Ries, J. B., Cohen, a. L., & McCorkle, D. C., Marine calcifiers exhibit mixed

responses to CO2-induced ocean acidification, Geology 37 (2009)

DOI:10.1130/G30210A

32. Rockstrom, J. et al., A safe operating space for humanity, Nature, 461, (2009).

URL: http://dx.doi.org/10.1038/461472a

33. Rodolfo-Metalpa, R. et al. Coral and mollusc resistance to ocean acidification

adversely affected by warming. Nature Climate Change 1.6 (2011)

34. Wood H.L., Spicer J.I. & Widdicombe S., Ocean acidification may increase

calcification rates, but at a cost. Proc. Roy. Soc. B. 275, (2008)

35. Wootton, T.J et al. Dynamic patterns and ecological impacts of declining ocean

pH in a high-resolution multi-year dataset Proc. Nat. Acad. Sci. 105 (2008)

36. Zondervan, Ingrid, et al., Decreasing marine biogenic calcification: A negative

feedback on rising atmospheric pCO2, Glo. Biogeochem. Cyc. 15.2 (2001)

19"

Tables

Table 1: 2-way Factorial ANOVA, group 1 Source df SS MS F value P pH_1 1 0.001750 0.0017496 5.746 0.0191 Disturbance_1 1 1 0.000224 0.0002240 0.736 0.3938 Interaction (P:D) 1 0.000520 0.0005201 1.708 0.1953 Within 73 0.022227 0.0003045

Table 2: 2-way Factorial ANOVA, group 2 Source df SS MS F value P pH_2 1 0.00355 0.003552 7.739 0.5418 Disturbance_2 1 0.00003 0.000034 0.075 0.9455

Interaction (P:D) 1 0.00462 0.004624 10.077 0.0022 Within 76 0.03488 0.000459 Table 3: 3 way Mixed Model ANOVA, groups 1&2 Source df SS MS F value P Initial Surface Area 1 0.00103 0.00103 0.43130 0.54589

pH 1 0.00509 0.00509 2.13293 0.21795

Disturbance 1 4.67E-05 4.67E-05 0.01966 0.89554

Interaction (P:D) 1 0.00102 0.00102 0.42740 0.54895

Sample group S(P:D) 4 0.00955 0.00239 6.23206 0.00012

Within 149 0.05710 0.00038 Total 156

20"

Figures

0

0.01

0.02

0.03

0.04

0.05

0.06

0.07

UL UH DH DL UL UH DH DL

Sample 1 Sample 2

Gro

wth

Rat

io

Treatment and Experimental Groups

Incremental Growth ratio in M. californianus

Figure 1: Mean Incremental Growth Ratio across all treatments and sample groups, with calculated Error bars (mean ± 1 SE). Treatment combinations are color-coded by Disturbance level and pH combination: blue for the Undisturbed/Low (UL); red for Undisturbed/High (UH); green for Disturbed/Low (DH); purple for Disturbed/Low (DL). Experimental runs identified by sample group number, e.g. 1 = sample group 1, 1st experiment. Y-axis is the Incremental growth rate, a dimensionless ratio given by equation 5.

21"

0.025

0.030

0.035

0.040

0.045

0.050

∆Growth ratio by pH treatment

pH level

Mea

n G

row

th ra

tio

High Low

Figure 2: Interaction plot of the change in Mean growth ratio across pH Treatments, with error bars (mean ± 1 SE). Experimental runs indicated by line type, e.g. hashed line is Sample 1, while the, solid line is Sample 2. Here, the lines indicate the difference in growth ratio between pH treatments for each sample run, while slope indicates the magnitude of the change, e.g. the positive slope indicates a higher growth ratio in the Low pH treatments of both samples 1 and 2. However, mussels of sample 2 had higher mean growth ratios (than sample 1) for both treatments, and exhibited a greater magnitude of positive response to pH than those of sample 1.

22"

0.025

0.030

0.035

0.040

0.045

∆Growth ratio by Disturbance treatment

Disturbance level

Me

an

Gro

wth

ra

tio

Disturbed Undisturbed

Figure 3: Interaction plot of change in Mean Growth ratio across Disturbance treatments, error bars (mean ± 1 SE). Experimental runs indicated by line type, here the hashed line is Sample 1, and the solid line is Sample 2. The interaction plot visualizes broad trends across treatments, and is useful in primary comparisons among treatments, and in the determination of numerical treatment effects. Here, the Disturbed treatment had a positive effect on growth ratio in Sample 2, but a negative effect in Sample 1; however, the magnitude of the positive effect on growth seen in sample 2 is less than the magnitude of the negative effect observed in sample 1.

23"

0.02

0.03

0.04

0.05

0.06

Sample group 1

pH level

Mea

n G

row

th ra

tio

High Low

0.02

0.03

0.04

0.05

0.06

Sample group 2

pH level

Mea

n G

row

th ra

tio

High Low

Effects of pH and Disturbance on Growth ratio

Figure 4: Interaction plot of the change in Mean Growth ratio across pH levels, by Disturbance treatment (Sample runs 1 & 2) with error bars (mean ± 1 SE). Disturbed treatments indicated by hashed line; undisturbed treatments given by solid line. Here, the effect of pH and Disturbance resulted positive growth in sample 1, as well as positive and marginally negative growth in sample 2. In sample 1, growth ratio was higher (meaning greater net growth) in the disturbed and low pH treatments, though both disturbance/pH combinations resulted in growth. In contrast, sample 2 exhibited a greater growth ratio increase in low pH and undisturbed treatments, while disturbed treatments across pH levels were not significantly different.

24"

Figure 5: pH variation during the experiment. Ambient pH is in blue, low pH is in red. Due to the type of pH manipulation (the incurrent water was acidic, and raised to the ambient level via bubbling) the variation in High (ambient) pH is negligible, while the pH of MBA water fluctuated semi-monthly.

25"

High pH Undisturbed

Sample group 1 Undisturbed

Sample group 2 Disturbed

Sample group 1 Disturbed

Sample group 2

Low pH

Undisturbed Sample group 1

Undisturbed Sample group 2

Disturbed Sample group 1

Disturbed Sample group 2

Figure 6: Diagram of the Experimental Setup. Large boxes are pH tanks, color-coded boxes represent factorial treatment combinations and replicate experiments.

26"

Figure 9: pH tank construction. Water flow is set to two liters per minute, through two spigots; one spigot sends the water through 30m of drip irrigation tubing, which is submerged in the reservoir tank of the water from the other system forming a simple heat exchanger. The other spigot sends the water to a 1.2 X 0.8 X 1.6 meter reservoir where an air pump connected to a soaker hose bubbles air into the tank. The surface water from this tank then gravity feeds into a 1.2 X 0.7 X 1 meter reservoir; the water is pumped at 3.8 liters per minute to the top of a column 2.4m tall by 0.27m diameter filled with BioSpheres. This column drains back into the same reservoir. The surface water from this reservoir then gravity feeds to the experiment trough.

27"

Supplement A: Sampling Dates Group 1: originally sampled 16/7/12, finish 17/9/12 Group 2: originally sampled 10/8/12, finish 14/9/12 Group 3: originally sampled 17/8/12, finish 21/9/12

removed from experiment due to loss of subjects B. Measurement Protocol Before recording any measurements, 3 mussels were measured to calibrate the calipers. Length

Align the less-curved side of mussel parallel to the calipers, and slowly close the calipers until they just touch the mussel. This measurement attempts to capture the distance from umbo to edge of the mussel, and requires that the umbo touch the base of the calipers (i.e. instead of measuring from the hinge, measure from one-side of the hinge, to get the absolute length.

Width

Position the mussel such that the overall curve of the shell us upward (away from you). Clear the shell of debris and position the lower arm of the caliper on the leading edge of the shell. Close the calipers until the top arm touches the shell, to get absolute width.

28"

C. Treatment/measurement timeline Date(dd/m/yy) Event Treatments Color code

27/8/12 Disturb Group 1 Disturbed/ High No colors Disturbed/ Low

31/8/12 Disturb Group 2 Disturbed/ High Red/white Disturbed/ Low

3/9/12 Disturb Group 1 Disturbed/ High No colors Disturbed/ Low

7/9/12 Disturb Group 2 Disturbed/ High Red/white Disturbed/ Low

10/9/12 Disturb Group 1 Disturbed/ High No colors Disturbed/ Low

14/9/12 Measure Group 2 Dist./ High Red/white Dist. /Low Undist./ High Undist./ Low

17/9/12 Measure Group 1 Dist./ High No colors Dist./ Low Undist./ High Undist./ Low

21/9/12 Measure Group 3 Both Bags (red bags, pick 20 and measure them)

No colors