Climate change impacts on connectivity in the ocean : Implications for conservation

Effective spatial management in the ocean requires a network of conservation areas that are connected by larval and adult dispersal. We propose a conceptual framework for including the likely impacts of a changing climate on marine connectivity, and synthesize information on the relationships between changing ocean temperature and acidification, connectivity and conservation tools. Our framework relies on concepts of functional connectivity, which depends on an organism’s biological and behavioral responses to the physical environment, and structural connectivity, which describes changes in the physical and spatial structure of the environment that affect connectivity and movement. Our review confirms that ocean climate change likely reduces potential dispersal distance and therefore functional connectivity. Structural connectivity in the ocean will inevitably change with the spatial arrangement of biogenic habitats resulting from disturbance as well as enhanced growth and mortality due to climate change. Climate change will also likely reduce the spatial scale of connectivity, suggesting that we will need more closely spaced protected areas.


INTRODUCTION
The rate of anthropogenic carbon dioxide (CO 2 ) emissions has increased from ;280 ppm prior to the industrial revolution to more than 400 ppm in 2013 and is predicted to increase to ;700 ppm in 2100 (Sabine et al. 2004).Such levels of atmospheric CO 2 influence global sea surface temperatures, which have already risen by 0.768C, on average, since pre-industrial times (IPCC 2007).Most areas of the ocean have warmed, albeit with considerable spatial variation in the magnitude of change (Hansen et al. 2006, Rayner et al. 2006, Hoegh-Guldberg et al. 2007).Increases in ocean temperature are accompanied by changes in ocean chemistry-specifically ocean acidification (OA) resulting from increasing dissolution of atmospheric CO 2 in the sea (Orr et al. 2005, Hoegh-Guldberg et al. 2007, Guinotte and Fabry 2008).Climate change has also been implicated as a primary cause of sea level rise, increased frequency and intensity of storms, and changing ocean circulation patterns (Meehl et al. 2005, IPCC 2007).In coming decades, the pace of these changes is expected to accelerate, which will alter the processes underlying connectivity in marine populations.However, the implications of ocean change for marine conservation planning are just beginning to be explored (Gaines et al. 2010, McLeod et al. 2012, Rau et al. 2012).In this paper, we propose a general framework to understand the impacts of warming and acidification on the connectivity of marine organisms.
Over the past decade, the number of studies on the effects of climate change on connectivity of marine organisms has increased dramatically, incorporating the effects of numerous physical and biological climatic stressors as well as multiple experimental, statistical, and modeling approaches.Most studies have focused on impacts on larval or juvenile stages of marine animal populations (e.g., 81% of 110 studies reviewed by Wernberg et al. [2012]), with benthic invertebrates being the most studied groups.Changes in ocean temperatures may influence biological processes such as dispersal and survival (Munday et al. 2009c, Johnson andWelch 2010).For example, O'Connor et al. (2007) demonstrated a relationship between temperature and a reduction in pelagic larval duration (PLD) across 69 marine species.In addition, ocean acidification has strong, negative effects on calcified organisms (e.g., calcified algae, corals, mollusks, and larval stages of echinoderms) while other taxa (e.g., crustaceans, and fleshy algae, seagrasses, and diatoms) are more resilient to ocean acidification (Kroeker et al. 2010, Kroeker et al. 2013).
Much of this work documents physiological effects at the individual level, and focuses less on ecological impacts at the population or community levels.Furthermore, 65% of experimental studies conducted between 2000 and 2009 involve single-factor manipulations of either warming or acidification (Wernberg et al. 2012), with most temperature studies showing significant negative effects of warming, but acidification responses exhibiting a great degree of variability (Kroeker et al. 2010, Wernberg et al. 2012, Kroeker et al. 2013).There have been fewer studies examining the effects of two or more climate change stressors at once (e.g., Wernberg et al.'s [2012] review in which 30% of the studies looked at two stressors and only ,5% of studies looked at more than two stressors), and these show even higher variation in responses.Increased ocean temperatures have been shown to both increase and decrease the severity of ocean acidification, and the potential interaction effects are likely significant (Gooding et al. 2009).
In this paper, we develop a conceptual framework based on the concepts of structural and functional connectivity and use this framework to evaluate cumulative impacts of warming and acidification on marine connectivity (Kindlmann and Burel 2008).We use this framework to synthesize existing data on the effects of temperature and ocean acidification stress on marine organisms.We then discuss how our findings relate to developing effective marine conservation strategies to mitigate climate change impacts on marine ecosystems.

A FRAMEWORK FOR UNDERSTANDING IMPACTS OF CLIMATE CHANGE ON CONNECTIVITY
While there are many physical symptoms of climate change on ocean ecosystems (e.g., sea level rise, altered storm regimes [Meehl et al. 2005, Macreadie et al. 2011]), we focus specifically on changing temperature and pH for our review due to the increased uncertainty in their predictions and the lack of literature on their impacts.We relate empirical estimates of effect size of temperature and OA to functional connectivity, which describes the suite of biological or behavioral responses of individuals to the physical environment, and structural connectivity, which describes the physical and spatial structure of the environment (Kindlmann andBurel 2008, Turgeon et al. 2010).We use these concepts to characterize our theoretical and empirical understanding of the effects of climate on the biological and spatial processes underlying connectivity.This framework unites two very different effects of climate change: direct effects on the biology of dispersing individuals (functional connectivity), and emergent changes to the spatial structure of the landscape that influence the probability of population persistence (structural connectivity).By integrating individual and population-level responses to climate change, our framework can be populated with empirical data that are relevant to conservation objectives, such as probability of population persistence (Fig. 1).

FUNCTIONAL CONNECTIVITY: LINKING ENVIRONMENTAL CONDITIONS WITH INDIVIDUAL MOVEMENT
Functional connectivity may be used to describe an organism's biological and behavioral responses to individual elements in the physical environment as well as the spatial configuration of the entire landscape (Kindlmann and Burel 2008).This includes individual activity and performance (e.g., time available for dispersal and survival probability while dispersing), and the aggregate of individual activity at the population level.Ocean conditions influence the timing and location of cues for reproduction and settlement (Munday et al. 2009b)

Empirical examples
In marine systems, over 70% of species have a planktonic larval stage that is subject to dispersal v www.esajournals.org(Lockwood et al. 2002, Levin 2006, Munday et al. 2009c).Furthermore, most dispersal and mortality occurs during this pelagic development stage (Morgan 1995, Levin 2006).The eggs, larvae or spores of benthic adult fish, invertebrates or seaweeds may travel tens or hundreds of kilometers before settling and transitioning to juvenile life stages (Kinlan andGaines 2003, Siegel et al. 2003), yet mortality during the pelagic stage is also extremely high (Morgan 1995).Early life stages in the organisms' complex life cycle thus can be critical for the maintenance of functional connectivity in marine ecosystems.Studying individual larval activity is notoriously challenging because direct observations of dispersal (e.g., movement of larvae) for most species are rare (but see Mora andSale 2002, Planes et al. 2009).However, there has been a recent increase in research on larval retention (self-recruitment) within and dispersal (connectivity) among coral reef populations, as well as a wide range of techniques and approaches for their study, from larval tagging and paternity analyses at small scales, to biophysical models at larger scales (Jones et al. 2009).Table 1 summarizes a few examples of approaches with the potential to examine the effects of climate change on life history and functional connectivity.
Connectivity can also be mediated by behavioral traits that determine settlement time and location, and these traits can be affected by changing ocean conditions.The effect of ocean acidification on larval behavior and their sensory systems has been studied experimentally most extensively on tropical coral reef fish fishes.High CO 2 levels may have significant effects on the habitat selection process involved with attraction to olfactory cues, which can bring about changes in population connectivity patterns (Munday et al. 2009b).At near-future levels of CO 2 , orange clownfish (Amphiprion percula) tended to avoid positive settlement cues that would normally lead to the location of suitable adult settlement sites.Instead, larvae were attracted to negative stimuli that led to suboptimal locations for settlement or even locations devoid of settlement habitat, which they normally avoided in today's CO 2 conditions (Munday et al. 2009b).Exposure to elevated CO 2 has also been shown to affect the timing of settlement in damselfishes (Pomacentrus spp.) with different patterns of habitat use, potentially leading to detrimental effects on larval survival and population replenishment if settlement occurs at unfavorable times (Devine et al. 2012a).Acidification has also been shown to significantly affect adult fish homing behavior.Devine et al. (2012b) tested the effects of nearfuture levels of CO 2 on the ability of adult cardinal fish (Cheilodipterus quinquelineatus) to home to their diurnal resting site after nocturnal feeding, and they detected an impaired ability to distinguish between home and foreign sites as well as increased activity level.These studies provide evidence to suggest that ocean acidification has an effect on fish brain functioning and fish behaviors (larval and adult) that may negatively impact population connectivity patterns.

Statistical approaches
In addition to empirical approaches quantifying key aspects of functional connectivity, efforts to characterize and model the probability of dispersal can also be used to study potential changes in the scale of functional connectivity (Box 1).Dispersal kernel functions represent the probability of dispersing a particular distance from a source location over repeated dispersal events (Lockwood et al. 2002), and capture average patterns in pre-settlement processes, but not post settlement mortality and selection (Almany et al. 2007, Graham et al. 2008).Thus for many species, dispersal kernels relate the spatial scales of dispersal to those of demographic population connectivity.Dispersal kernels can also be used to relate distributions of dispersal distances to the larval development period (Fig. 2).The average dispersal distance from one population to another can determine the scale of demographic processes, such as recolonization after disturbance.In contrast, the long tails of a dispersal distribution may be the most important determinant of migration rate and genetic connectivity (Clark 1998, Neubert and Caswell 2000, Palumbi 2004).Generalized dispersal kernels have played a central role in the development of marine reserve theory (Botsford 2001, Lockwood et al. 2002, Botsford et al. 2003, Gerber et al. 2003) and are potentially a useful tool for relating the effects of climate change to population-level responses (Munday et al. 2009a, Munday et al. 2009c).Dispersal kernels can also be v www.esajournals.orgapplied to coupled circulation and population models (Neubert and Caswell 2000).

Analytical approaches: biophysical modeling
Biophysical modeling approaches are increasingly being used to understand the dispersal of single-species pelagic larval stages by coupling biological properties of the organism with realistic hydrodynamic conditions (Miller 2007, Werner et al. 2007, Treml et al. 2008, Cowen and Sponaugle 2009, Lett et al. 2009, Metaxas and Saunders 2009) (Box 1, Table 2).Such biophysical models typically represent transport, behavior, mortality, and settlement processes of larvae, and are particularly relevant to questions related to the effects of climate change on connectivity because connectivity patterns are usually determined by both physical and biological processes (Werner et al. 2007).A small number of these coupled models have been explicitly used to look at the effects of climate change on connectivity (Lett et al. 2010).For example, by using a biophysical modeling approach and dispersal kernels to describe spatial distributions of larvae, Huret et al. (2010) found that variability in PLD can have a major impact on dispersal at large scales and vertical behavior at smaller scales, with a reduction in PLD having the potential to reduce the dispersal distance.Other approaches have explicitly simulated larval behavior and focused on the effect of increased water temperatures accelerating development of swimming abilities by shortening the development time, thus increasing local retention and self-recruitment rate (Irisson 2008).
The breadth of modeling approaches that have been developed in the past 20 years represents a significant increase in our understanding of the Box 1

The Gulf of California as an Illustrative Case Study
To examine how temperature and pH change may influence conservation planning, we model directional changes in functional connectivity with climate change by linking an abiotic change (e.g.temperature, pH) to the rate of a biological process (larval development and related dispersal).We examine the implications of a change in mean dispersal distance resulting from climate change for spatial prioritization of conservation actions in the Gulf of California (GoC).We focus on the GoC because (1) data on larval spawning locations are available (Soria et al. 2010), (2) a circulation model is available (Marinone et al. 2008), and (3) a spatial prioritization effort has been conducted to identify conservation sites (Ulloa et al. 2006), but this effort did not account for connectivity or climate change.Here we explore whether our framework suggests that future adaptation of the prioritization scheme might be necessary.
We focus on how climate change may impact dispersal and connectivity of a key species in the Northern Gulf, the rock scallop Spondylus calcifer.Rock scallops are highly sought after in artisanal fisheries, and have been recently protected by a network of marine reserves (Cudney-Bueno et al. 2009).Planktonic larvae of S. calcifer drift with ocean currents for several weeks following spawning and fertilization (Soria et al. 2010).Using our understanding of the potential magnitude and spatial variability in changes in connectivity and abundance resulting from potential changes in pH-value and temperature, we map potential changes in functional connectivity that could result from local climate change using a published circulation model (Marinone et al. 2008).We do not consider changes in structural connectivity because we have no reason to forecast change to habitat for rock scallop.In the absence of a downscaled general circulation model (GCM) for this region, we apply general climate change projections of 38C warming and À0.4 pH units as a simple illustration of our approach (Soria et al. 2010).We consider these changes in the context of the current configuration of priority areas identified in the Gulf of California Ecoregional Assessment (Ulloa et al. 2006), specifically focusing on the Rocas Consag priority area (Fig. 3).
Our results suggest that the current spatial extent of potential settling larvae is larger than the range for future settling with ocean warming and acidification (Fig. 3).In the northern areas, current settlement extends further south and in the eastern areas, settlement stretches further west than in a þ38C warming scenario (Fig. 3).In addition, there is greater larval density within a smaller area in the future temperature scenario.How the impacts of climate change on functional connectivity affect the potential success of conservation areas depends on whether a single priority area or a network is considered (Fig. 3).Rocas Consag (Graham et al. 2008) is the biggest priority area in the Northern Gulf.Our climate change scenarios dramatically affect scallop settlement in Rocas Consag, reducing it by nearly half.However, when the entire priority area network in the Northern Gulf is considered, the impact of warming on settlement locations is negligible (Fig. 3, Table 2).This suggests that the greater coverage and spatial distribution of priority areas in the network would be more resilient to changes in mean dispersal distance of rock scallop.interaction between the characteristics of the physical environment (e.g., temperature and circulation) and changes in organismal life history traits that are ultimately responsible for survival and dispersal (e.g., PLD).However, only a few studies have explicitly considered changes in oceanic currents as an additional variable (but see Munday et al. 2009c).Nonetheless, the modeling studies described in this section (see Table 1) have proven useful tools to test hypotheses about how climate change may impact connectivity patterns (Lett et al. 2010).Furthermore, an increasing number of approaches solely focusing on how modeling population connectivity can be modified by ocean currents (i.e., under current conditions) are becoming available for specific regions (Marinone et al. 2008, Treml et al. 2008).Despite their widespread application, however, using such approaches remains difficult since assessing climate change impacts still requires extensive computing power and sophisticated models (e.g., coupled climate ocean general circulation models).The capacity to run them at the appropriate spatial resolution and in conjunction with relevant biological processes is still in its infancy (Munday et al. 2009c).The success of most of these models has strongly relied on approaches that focus at small, very local scales, which makes the use of these models heavily site-specific.

Synthesis of functional connectivity framework
As seen from the few examples mentioned here, increasing evidence suggests direct effects of changes in ocean climate on individual dispersal biology.Increasing ocean temperatures are likely to accelerate larval and egg development, leading to reduced time in the water column (Hirst andLo ´pez-Urrutia 2006, O'Connor et al. 2007) allowing earlier habitat-seeking behavior (Munday et al. 2009c), changes in dispersal distance (Lett et al. 2009, Ayata et al. 2010, Huret et al. 2010, Lett et al. 2010), changes in transport processes and thus settlement location (Vikebø et al. 2007, Munday et al. 2009c, Huret et al. 2010, Tracey et al. 2012), and modifications of local versus regional connectivity (Irisson 2008, Ayata et al. 2010).Additionally, even though there remains considerable variability in responses among taxa, the combined effect of increased ocean temperatures and acidification shows a trend towards lower survival, growth, and development (Findlay et al. 2010, Lischka et al. 2011, Kroeker et al. 2013).These relationships between temperature, acidification and mortality could be mediated by direct physiological impacts, or indirectly through changes in size at age, morphology or feeding efficiency (Kurihara 2008, Dupont and Thorndyke 2009, Kroeker et al. 2010, Kroeker et al. 2013).This combination of stressors can also potentially impact other biological sources of mortality, such as predation and starvation, and changes to mortality could substantially affect total settlement.Variation in responses could also reflect species-specific sensitivities and local adaptation to different environmental conditions (Kroeker et al. 2013).Nonetheless, the apparent trend of decreased mean dispersal distance and reduced survival probability remains crucial and useful when considering the framework proposed here (Fig. 1a).Variability may be incorporated into estimates of functional connectivity for particular cases using dispersal kernels.
Despite the fact that approaches quantifying particular trajectories for the study of functional connectivity are still under development, we can nonetheless continue to study changes to the dispersal process with the use of dispersal kernels (Fig. 2).Even without specifically quantifying dispersal trajectories, we can consider how changes to the ocean influence functional connectivity by relating properties of the ocean environment to outcomes of individual movement such as dispersal distance and survival.

Structural connectivity: relating environmental conditions to habitat area
Structural connectivity reflects the spatial structure of the environment and the influence of spatial structure on the likelihood of individuals from one habitat reaching another habitat (Kindlmann and Burel 2008).In contrast to functional connectivity, which can be used to describe the effect of ocean conditions, such as temperature, on the dispersal potential of individuals via physiological mechanisms, structural connectivity describes the effect of habitat size, proximity, and quality on the connectivity of individuals inhabiting that habitat (Kindlmann and Burel 2008).For some marine species, the spatial arrangement of their habitat is determined by other species, referred to as foundation species.In such biogenic habitats, species like coral reefs, kelp, marine algae, and seagrasses host a vast biodiversity of associated invertebrates, fish, and seaweeds (Thompson et al. 2002, O'Hara et al. 2008, Hewitt and Thrush 2010, Perry et al. 2011).Structural connectivity in the ocean will inevitably respond to changes in the spatial rearrangement of biogenic habitats resulting from climate change.The following section summarizes a few examples of the impacts of increased ocean temperatures and ocean acidifi-cation on foundation species that maintain structural connectivity.Although we focus mostly on coral reef systems given the breadth of evidence available for them, the framework could be applied to other systems with major foundation species.

Empirical examples
Ocean warming may impact habitat patch area directly through effects on foundation species' growth, survival, and resilience potential.For example, climate change is expected to increase the frequency and severity of extreme temperature events, which may lead to increased rates of coral bleaching and subsequent decreases in coral cover (McWilliams et al. 2005, Vivekanandan et al. 2009, Anthony et al. 2011, Hoegh-Guldberg 2011, Hoeke et al. 2011).Even when corals acclimatize to increasing temperatures, they experience reduced growth rates (Cantin et al. 2010, Jones andBerkelmans 2010).The combination of mortality from bleaching and slower growth rates can lead to an overall reduction in coral habitat (Alvarez-Filip et al. 2009, Cantin et al. 2010, Pandolfi et al. 2011).Reduced habitat area implies increased isolation between some remaining habitats, and these changes in area and configuration could reduce dispersal, colonization and alter community structure.
Much of the research on the effects of acidification on foundation species has focused on corals.The general mechanism by which corals or other calcifying organisms are affected by ocean acidification is through the increase in atmospheric CO 2 altering the current dissolved inorganic carbon distribution in seawater and reducing its pH (Cohen andHolcomb 2009, Findlay et al. 2010).The reduction of carbonate ions (CO 3 2À ) slows the rate at which calcifying organisms produce CaCO 3 (calcification) to a point where rates of erosion exceed rates of skeletal accretion, most likely due to the fact that calcification is very energetically costly (Cohen and Holcomb 2009).Quantitative syntheses of the available experimental evidence on the sensitivity of coral calcification to ocean acidification have suggested that under business-asusual emissions scenarios, coral calcification could decline by ;15-22% on average by the end of the century (Chan and Connolly 2013).
The effects of acidification could be modified by concurrent warming.Some species of coralline crustose algae have been shown to be particularly sensitive to ocean acidification under warm conditions in manipulative experiments (Kleypas andYates 2009, Doropoulos et al. 2012).In general, acidification is expected to reduce calcification rates in crustose coralline algae, potentially reducing the amount of suitable coral settlement habitat (Kleypas andYates 2009, Doropoulos et al. 2012).Predicted end of the century CO 2 levels have also been demonstrated through experimental manipulations to significantly reduce crustose coralline algal cover and consequently coral settlement, suggesting acidification may also reduce coral population recovery (Doropoulos et al. 2012).These combined impacts will likely affect structural connectivity for species dependent on biogenic habitats like coral reefs (Hoegh-Guldberg et al. 2007).
Many habitat-forming species are also likely to be affected by ocean acidification through shifts in relative abundances that could lead to changes in community composition.Acidification is expected to increase productivity of autotrophic foundation species such as seagrasses, especially in cold climates (Guinotte and Fabry 2008), with positive effects on meadow size, the number of meadows, and their isolation.In tropical systems, natural in situ field studies detected an eightfold increase in seagrass cover and a doubling in noncalcareous macroalgae (Guinotte and Fabry 2008).Although structural connectivity of coral reefs may be reduced, connectivity among seagrass meadows seems likely to increase.
This emerging evidence suggests both positive and negative responses of marine foundation species to ocean change, with the direction of effect depending in part on local environmental conditions, species composition, acclimation and adaptation capacity, physiological responses to interacting factors such as temperature, acidification, and nutrients (Pandolfi et al. 2011, Chan andConnolly 2013).The effects of temperature increases and acidification on the spatial arrangement of foundation species will contribute to changes in structural connectivity.Reduced habitat area could decrease structural connectivity, reducing immigration or metapopulation dynamics that might enable patch size to be maintained.In addition, chances for successful v www.esajournals.orgrecruitment and population persistence could be reduced because of an increase in the number of smaller habitat patches and increased distance between them (structural connectivity; Fig. 1b; Guinotte et al. 2003, Munday et al. 2009a, Hendriks et al. 2010).Decreases in suitable habitat may not only reduce overall coral reef area, but also decrease resilience to future disturbances because new larval recruits will be less likely to reach new areas to settle.
Metapopulation theory suggests that changes to habitat area will have disproportionate effects on population size and persistence (Kininmoth et al. 2010, Kininmonth et al. 2011).With less available habitat, particularly of foundation species, there will not only be less larvae produced, but also potentially longer distances for settlers to travel.Recent work has used regionally downscaled climate change models that include ocean circulation together with models that include habitat type and foundation species to estimate changes in connectivity (Mumby 1999, Mumby et al. 2011).In the future, these modeling exercises should incorporate changes in functional connectivity for a more complete estimation of connectivity changes due to climate change (Fig. 1).

IMPLICATIONS OF CHANGING CONNECTIVITY FOR MARINE CONSERVATION
Understanding how to incorporate climate change into marine conservation strategies such as marine reserve design and marine spatial planning is critical to sustainably managing marine populations.One common approach to marine conservation is the establishment of marine reserves, which can be planned based on three general principles: (1) reserve size and spacing must be scaled to mean dispersal distance of target species, (2) reserves must capture an adequate fraction of larvae of target species, and (3) networks of reserves have distinct advantages over single reserves (Botsford et al. 2003, Gerber et al. 2003, Gruss et al. 2011).Here we discuss implications of our conceptual framework and synthesis for marine reserve theory.

RESERVE SIZE AND SPACING
Determining adequate reserve size depends on management goals.One approach is to make the reserve larger than the mean dispersal distance to ensure that the reserve persists independent of dynamics in other reserves or fished areas (Botsford et al. 2001, White et al. 2010).With increasing reserve size there is eventually a tradeoff in terms of reduced spillover (Gaines et al. 2010).Our review suggest that with reduced structural connectivity, reserves may need to be larger to compensate for habitat loss (Fig. 1b), consistent with previous suggestions (McLeod et al. 2009).
In terms of spacing, network designs must consider the mean larval dispersal distance of target species (Botsford et al. 2001, Lockwood et al. 2002, Botsford et al. 2003).Marine reserves are thought to benefit fisheries primarily through adult spillover to unprotected areas and larval export from protected areas to other protected areas as well as unprotected areas (Gaines and Bertness 1992, Botsford et al. 2001, Botsford et al. 2003, Gerber et al. 2003, Gaines et al. 2010).If reserves are spaced so that there is sufficient connectivity among them, then persistence of the entire network is possible even if no reserve is self-persistent (Kaplan 2006).With variability in dispersal distances, reserve spacing is less important than size (Kaplan 2006); yet spacing must be close enough relative to dispersal distance to allow network persistence (Moffitt et al. 2009).Ocean warming and acidification may alter the spatial scale of connectivity, potentially increasing connectivity among nearby habitats and reducing connectivity among already distant habitats.Thus, marine reserve networks may need to be designed either with a spatial arrangement that can accommodate changes in the scale of connectivity (McLeod et al. 2009), or with future flexibility that could allow for the creation of additional reserves to serve as stepping stones (Treml et al. 2008) to connect distant habitats that are increasingly isolated.Optimal reserve size and spacing will clearly depend on the dispersal distance of the target species, which may be related to PLD.Shorter PLDs will require reserves to be spaced closer together, but may also mean that reserve size can be relatively small and still maintain export.Longer PLDs will likely require larger reserves, but the space between them can also be greater.

CAPTURING AN ADEQUATE FRACTION OF LARVAE
Our review suggests that warming and OA could reduce larval survival, changing total larval output for several taxa (Anthony et al. 2008, Kroeker et al. 2010, Talmage and Gobler 2010).Botsford et al. (2001) propose a numerical threshold of 35% of natural larval settlement in a population that must be captured by reserves.This value was derived for a case of no reproduction outside reserves and for a settlerrecruit relationship that has a slope of 1/0.35 at the origin (reflecting groundfish stocks off the California and Oregon coasts).While 35% is not a universal or absolute number (White et al. 2010), reduced larval survival from increasing pH could represent a threat for any minimum threshold based on a healthy, non-exploited or otherwise impacted population.To accommodate climatedriven increases in mortality, targets for minimum larval retention should be greater than 35%.Although larger reserves could compensate for reduced larval retention, there would be tradeoffs in the amount of larval export.Reserves designed to protect key life stages where the target species aggregate may be sufficient.
Identifying where these places may persist in a changing climate is critical to future reserve design.During spatial planning exercises, particular attention could be given to protecting places that support high levels of larval export as the climate changes (i.e., ''core areas'').Including smaller, but spatially closer reserves may ensure connectivity even to more distant habitats.In places where warming is expected to be greater, these distances may need be relatively short.In Box 1, we use a coupled bio-physical model for the Gulf of California to illustrate implications for reserve design (Table 2).

IMPLICATIONS FOR MARINE SPATIAL PLANNING
In the context of a framework based on structural and functional connectivity, the evidence we have reviewed suggests that connectivity could be reduced in warmer, more acidic oceans.We illustrate how these changes could impact the effectiveness of the design of marine conservation areas for a benthic marine invertebrate in the Gulf of California (Box 1, Fig. 3).Understanding the precise impacts of warming and acidification on marine connectivity will require detailed studies aimed at quantitatively determining scaling relationships and thresholds between climate factors and critical demographic parameters.It is important to quantitatively consider the potential for general changes in the Fig. 3. Weighted mean density of larvae at (A) current temperature, (B) current temperature þ38C, and (C) þ38C and acidification of À0.4 pH units.Passive larval trajectories were modeled using a three-dimensional, baroclinic numerical model, and final location was determined using a temperature-dependent larval duration (Eq.1).Conservation priority areas including Rocas Consag (Graham et al. 2008) are indicated by black lines.Values were calculated on a 1-km grid with a 5 3 5 km smoothing algorithm.v www.esajournals.orgsize and spacing of reserves (Almany et al. 2009, McLeod et al. 2010), particularly for foundation or target species that are experiencing changes in ocean temperature and acidification.While it is well-established that a network of marine protected areas is needed to manage marine species with a bipartite life cycle, reserves should be closer together and larger in size to mitigate climate change impacts (McLeod et al. 2009).In fact, climate change could lead to some predictable (directional) changes in patterns.For example, our review suggests that conservation areas may need to be closer together and larger in area than previously thought.Increasing the proportion of habitats covered by reserves by increasing reserve size and putting reserves closer together would increase the likelihood for success.Ensuring that connectivity is maintained through stepping stones that may not necessarily increase the total proportion of area under reserves is also possible.
Protection of areas of greater resilience or resistance is often cited as an element of mitigating the effects of climate change for marine ecosystems (West andSalm 2003, Grimsditch andSalm 2005), although identifying these areas is challenging (Chollett et al. 2010).Furthermore, because many fish species migrate to warmer water for reproduction, shortened PLDs could actually increase MPA retention and therefore benefit single MPA persistence.
In recent years, Marine Spatial Planning (MSP) has become a popular approach to marine conservation (Foley et al. 2010) because of its focus on sustainable use for a wide range of stakeholders.Most MSP efforts are spatially explicit to account for a range of biological, social, and economic factors (Crowder andNorse 2008, Ogden 2010).Some efforts include connectivity in the design framework (Beger et al. 2010a, Beger et al. 2010b), but they rarely consider how connectivity may change in coming decades.Given that the potential effects of climate change and ocean acidification on connectivity are significant, ignoring these variations may compromise the long-term effectiveness of management and conservation actions.In light of uncertainty about how ecosystem services associated with multi-use marine zones are going to change with climate change, flexible zoning will be essential in a changing world.In the spirit of adaptive management, regular (e.g., decadal) assessment of social and ecological patterns should be conducted and spatial management can be modified based on the best available data.
In addition to considering habitat and species in conservation planning, spatial planning should also include areas that are predicted to exhibit different responses to climate change (Ackerly et al. 2010).Incorporating this type of ''climate heterogeneity'' into marine spatial planning offers the opportunity to zone for higher levels of protection in key areas that are not predicted to change while enforcing some restriction in places that may become key larval settling areas over time.Considering the full range of uses that are typically included in MSP is important to understand the relevance of climate change and ocean acidification for conservation prioritization.

CONCLUSION
Even in the absence of desired data, developing a framework for predicting changes in the spatial distribution of marine organisms not only highlights insights from available data and models (Fig. 1), but also underscores the importance of improving our fundamental understanding of climate change impacts on marine systems.Data on the interactive effects of ocean acidification and temperature as well as habitat data at fine-scales may enable us to better assess how changes in functional and structural connectivity may affect marine populations and ultimately, design conservation strategies aimed at maintaining their persistence.Our framework can be used to incorporate other dimensions of climate change, such as altered circulation patterns, sealevel rise, and frequency and intensity of storms as these data become available.Integrating fundamental insights of climate change impacts on marine organisms into conservation planning is critical to the future of marine biodiversity.
, potential dispersal distances and survival (O'Connor et al. 2007, Munday et al. 2009c), and feeding and predation opportunities during dispersal periods.Here we summarize empirical examples and research approaches to studying the effects of ocean temperature and acidification on functional connectivity.

Fig. 1 .
Fig. 1.Climate change impacts on marine connectivity with implications for Marine Spatial Planning (MSP); (a) Climate change could reduce functional connectivity by altering the duration of, trajectory and survival during the larval life cycle.Bold arrows represent organism dispersal; (b) Climate change could reduce structural connectivity by reducing habitat (dotted arrows).Grey patches represent pre-climate change habitat extent and blue patches represent habitat after climate-related change.

Fig. 2 .
Fig. 2. Basic properties of dispersal kernels (A) vary predictably with the planktonic larval duration (PLD) and mortality (Siegel et al. 2003).(B) Mean absolute displacement of larvae relative to parental source increases with PLD, (C) variance in dispersal distances increases nonlinearly with PLD, (D) survival at the modal dispersal distance declines with PLD and (E) daily mortality rate.

Table 1 .
Review on the current state of knowledge of the effects of temperature and ocean acidification stress on marine organisms with respect to functional and structural connectivity.

Table 2 .
Proportion of larval trajectoriesthat settle in conservation priority areas.Results from analysis of effects of temperature and pH on potential larval dispersal distance in the Gulf of California (Fig.3).