Optimal Investment in Ecological Conservation and Restoration Proects under Climate Change: A Spatial Intertemporal Analysis Koel Ghosh James S. Shortle Pennsylvania State University Carl Hershner Virginia Institute of Marine Science Panel: Consortium for Atlantic Regional Assessment (CARA): Complex Coupled Systems Prepared for presentation at the Open Meeting of the Global Environmental Change Research Community, Montreal, Canada, 16-18 October, 2003
1. INTRODUCTION Proected climate change in this century is expected to pose significant threats to ecosystems and biodiversity. Climate change will affect fundamental ecological processes and the spatial distribution of terrestrial and aquatic species. The successful survival of the species will depend critically on the availability of migration corridors and the existence or emergence of suitable habitats. Because land use can affect these opportunities, a crucial issue in facilitating ecosystem adaptation to climate change is managing land use and landscapes to preserve migration corridors and potentially emergent habitats. This has important spatial implications for ecosystem management in general and conservation and restoration policy analysis and design in particular. Site-specific biotic conditions affect species ability to reproduce, a crucial determinant of successful restoration, thereby making returns to restoration effort site-specific. The returns to investment in ecological conservation and restoration will therefore be affected by the choice of site. Climate change is expected to alter the geographic distribution of the biotic conditions that impact species survival. There is gross uncertainty about where in space the biotic conditions most conducive to species conservation or restoration will emerge. The spatial uncertainty emerges as a key source of uncertainty in dealing with investment in ecological restoration proects. The fact that the locations of the most cost effective sites for conservation and restorations under a changed climate are uncertain raises two important ecologicaleconomic issues. The first has to do with evaluating the option of preserving alternative sites for cost effective restoration. This is important because, although a preserve all 2
strategy is infeasible, allowing for development poses the risk of physical irreversibility from altering the ecological basis of particular locations, thus eliminating them as future restoration sites and potentially decreasing the overall restoration opportunities. The second issue arises in context of the opportunity cost of excluding alternative land uses from current conservation or restoration sites that will no longer be useful for conservation purposes under an altered climate. These foregone current benefits are irreversibly lost forever and this is important if we consider heterogeneous preferences over time. These lost benefits, from the social perspective, constitute a part of the economic costs of conservation. This paper tries to explore the spatial implications of climate change for investment in ecological conservation and restoration proects. The paper concentrates on the design of economically optimal strategies for species conservation and restoration in an aquatic environment, using the Submerged Aquatic Vegetation (SAV) restoration program in the lower Chesapeake Bay as a case study. In context of the case study, the paper develops a methodological framework that addresses the key issues of uncertainty, irreversibility and space in climate change that is then used to determine optimal spatial allocation of restoration and competing land uses. The current paper is organized in the following manner. Section 2 introduces the case study and provides detailed background both on how climate change is expected to impact the returns to the restoration program and how existing restoration policies might cease to be optimal unless they are adapted accordingly. The section also highlights how the case study is a good fit for addressing the two ecological economic issues pointed out previously. Section 3 reports the spatial model that has been developed in context of the 3
case study and briefly comments on the data. Section 4 lays out the decision analysis that determines what the optimal restoration strategies and land use at the sites will be under the uncertainty of climate. The research is still an on going effort. Section 5 ends with future plans for the study. 2. CASE STUDY SAV s are ecologically important for the aquatic environment of the Chesapeake Bay. They are important natural resources that provide food and habitat for waterfowl, fish and invertebrates and mitigate shoreline erosion. They are also important for water quality, a concern in the Chesapeake Bay, as they produce oxygen, filter and trap sediments, and remove excess nutrients that can fire up unwanted algal growth. SAV abundance in the Chesapeake Bay regions was historically recorded to exceed 600,000 acres but by 1978 aerial surveys conducted by Virginia Marine Resources Commission documented only 41,000 acres (Moore and Orth, 1983). Declining water quality of the Bay characterized by high sediment and nutrient level were held primarily responsible in general. These prompted several diverse bay management and interest groups into planning and implementing SAV restoration programs throughout the bay. It has been established that the presence of SAV is good for the ecological health of the Bay and consequently SAV restoration is a top priority with the Chesapeake Bay Authority. Within the Hampton Roads area in Virginia, the case study site, it is the Virginia Marine Resource Commission that is tasked with restoration efforts. Map 1 shows the case study area, with the red areas on the map indicating existing SAV as of 2001. The SAV in this lower bay region is predominantly eelgrass (Zostera Marina) and they occur at water depths of 0.5 to 2 meters. The habitat requirements for SAV are listed as (1) 4
temperature, (2) light penetration, (3) water currents and wave action, (4) bottom sediment, and (5) water depth (range: below low tide line to about 2 meters in depth). Map 1: The case study site-hampton Roads area, Virginia. In Virginia, restoration is implemented at sites in which the SAV had been historically abundant. The understanding here is that SAV can be restored in areas where they had been previously known to exist. Combining historic abundance and habitat conditions, historic coverage areas from 0.5 meters to 2 meters in water depth are targeted as SAV restoration sites in Virginia s waters. However, the sites in Virginia prioritized for SAV restoration coincides with locations that have been delineated suitable for aquaculture of oysters. Aquaculture in addition to bringing farmed seafood to the table and providing recreational opportunities 5
also assists in resource and habitat restoration. In particular aquaculture is ecologically and economically important for the restoration of the once renowned Bay oyster populations much of which was eliminated due to the onset of parasitic diseases like Dermo and MSX (Leffler, 1998). The decline of the fishery, once the nation's most prosperous, has impacted processing houses and supporting businesses throughout the Bay's waterside communities, especially on the Eastern Shore of Virginia resulting in social and economic pressures. Plants processing oysters declined in number from 80 in 1974 to 28 in 1990 in Virginia alone (Lipton, 2002). In addition, oysters provide important habitat services and also contribute to improving water quality in the Bay by filtering algae. The ecological services provided by the oysters are very similar to the ecological services provided by SAV. Current regulation in Virginia favors SAV over aquaculture in all conflict areas. Under the current policy, aquaculture is prohibited from the waters that house any SAV restoration sites. One of the most well understood impacts of climate change is sea level rise. The National Oceanic and Atmospheric Administration (NOAA) s National Ocean Service (NOS) Center for Operational Oceanographic Products and Services (CO-OPS) collects and distributes observations and predictions of water levels and currents. The rate of mean sea level rise or fall determined for the long-term water level stations in Virginia indicate a rising trend. An increase in the sea level implies that the water depth, an important habitat criterion for SAV, will increase at the current restoration sites by the amount of increase in sea level. Everything else remaining unchanged, this means that most suitable habitat conditions will emerge at sites where the appropriate water depth 6
occurs post the sea level rise. The current restoration sites will cease to be optimal for restoration because of the change in water depth at these sites. With rising sea level the ideal restoration sites will be migrating to shallower waters. Figure 1 illustrates how changing sea level might results in SAV relocating to shallower depth. The C indicates current sea level while the F indicates the future sea level. The SAV relocate to places where the water depth is appropriate. Figure 1: Migration of SAV in accordance with changing water depth. The maximum abundance of current SAV occurs at the 0.5 meters water depth. As the sea level rises and the sea moves inland, the SAV has restricted opportunities of moving onto land, primarily because it cannot grow on land. Therefore there is the risk of losing a substantial part of SAV that currently exists at the 0.5 meters water depth. One opportunity of conserving the existing SAV lies in the tidal marshes along the coast. The tidal marshes, on account of their biotic similarity with SAV habitat, have the potential for being conservation sites for the SAV to migrate into. There is, however, much uncertainty associated with this possibility of SAV conservation. The first source of uncertainty relates to the ability of the SAV to adopt the 7
tidal marshes as their new habitat. The second source of uncertainty surrounds the possibility of current land use changes along the coast eliminating the tidal marshes making them unavailable for conservation purposes. The potential migration of SAV in response to sea level rise has policy implications for resource management in the lower bay area. Under existing policy, the resource manager will continue to restore SAV at current sites. As the habitat conditions decline at these sites with rising sea level, the returns to restoration effort at these sites will be impacted negatively. As the current sites become unsuitable for restoration, they can be opened up to aquaculture. If the bay manager does not anticipate the possibility of change and fails to adapt resource management policies accordingly, it will result in inefficient spatial allocation of SAV conservation and restoration. In addition, there will be the lost aquaculture benefits that will arise out of excluding aquaculture from current restoration sites as per existing policy even when the water depth change makes aquaculture the optimal land use at these sites. Current policy also does not address the possibility of preserving the tidal marshes as future habitat for the SAV. The option of having an alternative site available for SAV conservation provides flexibility to resource managers in terms of where they would chose to conserve the SAV. The flexibility is of some value to the resource manager, as it opens up the opportunity to lowering the costs of obtaining the targeted abundance of SAV. The option of an alternative site therefore also needs to be considered while determining cost effective strategies for SAV restoration. 8
3. SPATIAL MODEL & DATA There are two time periods, t = T, T+50 denoting now and fifty years hence respectively. Areas within the entire SAV restoration region of the bay (water area from the coast to the 2 meters depth) having similar water depth are grouped together as sites in the model. Let = -1,-2,-3, -4, -5 be the sites. The negative sign in front of the numbers indicate a site that is currently under water. Let there be an additional site, =0. The site 0 models the coastal strip of land which is not under water in time period T but has the possibility of going under if the sea level rises. Let the water depth at site at time t be indicated by D t. If y unit is the current water depth at site and we consider a x unit rise in sea level, then D y and D = y x. Then, for any x > 0, site = 0 goes under water in T+50 T = T +50 + and the water column at the remaining sites will increase by x units. Figure 2 illustrates how the water depth at each site increases by 0.5 meters in response to x = 0.5 meters of sea level rise on a 50 years time horizon The numbers reported within each site represents the relevant water depth within that site at that time period. Thus, corresponding to =0,- 1,-2,-3,-4,-5 and say current water depth given by = 0,0.5,1,1.5,2, 2 +, the water depth D T will be D T +50 = 0.5,1,1.5,2,2.5,2 +.5. The top half of the picture shows the depth details of the spatial layout in time T while the bottom half shows the altered depth details of the same layout at time T+50. The cluster of spots on the 0 th site indicates the presence of tidal marshes in that site. The tidal marshes are potential places for growing SAV in site 0 when it gets submerged in water in time period T+50. 9
Figure 2: Diagrammatic representation of the spatial model for SAV migration. As the sea level rises and the water depth increases, the SAV is expected to migrate from site to +1. Let SAV t and HSAV t be the SAV and historic SAV abundance (measured in units of area) present at site at time period t. Then SAV T and SAVT + 50 will be the SAV abundance at site in time T and T+50 respectively. 10
Modeling Assumptions Water depth is the only habitat criteria that will change at the sites. None of the other criteria alter between the sites. The areas that come under historical SAV coverage are targeted for restoration. The historical SAV area within each site is not affected by the water depth change at the sites. Consequently, there is no difference in the historical SAV area at each site across time i.e. HSAV = HSAV 50 T T + The shellfish aquaculture suitability areas within each site ( ) are also not affected by sea level rise. The water depth is not a criterion for practicing aquaculture. This implies = A. AT T + 50 Actual or existing SAV abundance at a site alters with changing water depth at the sites i.e. SAV. T SAVT + 50 Only one land use, either SAV restoration (R) or aquaculture (A), can be implemented at a site in any time period. It is possible to switch between water uses instantaneously at each site-there is no lag period involved. At time T, the land use at each site for =-1,-2,-3, -4, -5 is restoration. The current distribution of existing SAV is known. The SAV at site in time period t will depend upon the current water depth at that site, previous SAV presence within that site SAVt 1, and the restoration effort implemented at that site R t. Equation 1 captures this functional relationship, (1) SAV = f D, SAV, R ) t D t ( t t 1 t A t 11
R is a binary variable denoting that the land use is restoration site in time t for =1 t R t and otherwise for R t =0. This relationship will be used to proect existing SAV in each of the sites in T+50 time period. Intuitively, one would expect a net loss of SAV abundance for = -4 and -5 as the water depth at these sites will be greater than the depth that can be tolerated by SAV. The SAV abundance in the other sites will depend upon how previous SAV and restoration can contribute to existing SAV abundance at these sites. The positive and negative signs at the bottom of Figure 1 indicate intuitive expectations of gains and losses in SAV abundance at corresponding sites. The (+/0) sign at the bottom of the 0 th band indicate a gain of SAV in this band if the tidal marshes are available and can successfully host the SAV or no gain at all if they are gone. The question marks are for the case where there is no clear intuition of what the SAV abundance will be for that site. Virginia Institute of Marine Science (VIMS) obtained bathymetry information from the Chesapeake Bay Program. The information was then used to create bathymetry bands at incremental depth of 0.5 meters from the coastline all the way to the 2 meters water depth. VIMS also created coverage of current (2001) SAV and historic SAV (1971-2001). The coverage for suitable shellfish aquaculture was obtained from the oyster and hard clam models developed for the VIMS Shallow Water Use Conflict Proect. The Tidal Marsh Inventory data had been completed in 1992 by the Wetlands Research Program at VIMS. The above coverages were unioned together using GIS software ArcInfo. A frequency was run to determine areas of SAV, wetlands, and shellfish aquaculture within each bathymetry bands. The information was then compiled 12
on an Excel spreadsheet that lists the area in square meters within each bathymetry band and the acreage of historic SAV, current SAV, and shellfish aquaculture within the bands. The bathymetry bands, ignoring their precise spatial configuration or shape, are fitted as the rectangular sites in the spatial model. Within each bathymetry band we have acreage data on historic SAV ( HSAV t ), currently (2001) existing SAV ( SAV T ), and shellfish aquaculture ( A t ). Restoring a site or implementing aquaculture land use at a site means using up the entire area available within the site as HSAV t and A t respectively for those land use purposes. Detailed maps of the bathymetry bands and the various coverages within the bands are available upon request 1. We can consider only one scenario of sea level rise (0.5 meters in 50 years of time) because we are restricted to bathymetry information at intervals of 0.5 meters. This might be deemed a higher end figure but technical limitations prevent us from considering any other scenario of sea level rise. 4. LAND USE DECISION In order to understand the spatial implications of climate change for investment in ecological conservation and restoration proects optimal strategies, there is need to first proect what the spatial distribution of the SAV will be under sea level rise. Equation (1) provides direction for predicting the SAV abundance at a site in time T+50 under different scenarios of sea level rise and restoration. The functional relationship is not precisely known. Proection of SAV for site would involve proportionately weighing existing (2001) SAV corresponding to the water depth at that site by the acres of SAV that previously existed at that site and acres that are restored within that site. There is no 1 Correspondence: kug2@psu.edu 13
prior information to provide guidance on the choice of the weights. Sensitivity analysis of the weights will be conducted to determine the robustness of the proections. The analysis has been broken down into three distinct parts (1, 2, and 3) that differ from one another in the sites and the restoration opportunities that they consider. Part (1) focuses on what the optimal land use practices are at the current restoration sites ( = -1,-2,-3, -4, -5) under the possibility of sea level rise. To do so, it first has to be determined what the optimal land use (A or R) is at these sites post sea level rise. This raises the question of what the decision rule should be for deciding on optimality of land use. The standard way of doing so would be cost benefit analysis. The challenge of doing cost benefit analysis in context of the case study arose in assigning dollar values to the benefits of SAV. Studies on valuation of SAV are non existent. SAV are non marketable goods and there is no price information for it. Indirect values of SAV can be either obtained from market prices of the resources SAV supports and the services it provides or assessed from contingent valuation studies. The latter would require much time while the former would have its share of controversy. The SAV and shellfish benefits cannot be compared using dollar values owing to the lack of information on SAV values. A decision rule that does not depend on valuation of SAV will have to be applied. The Chesapeake Bay Program and other agencies associated with SAV restoration program appear to take the acres of SAV presence as a measure of the success of the SAV restoration program. It seemed appropriate that the final analysis should use acreage/expected acreage of SAV presence as a numeraire for evaluating the restoration option. In addition, there is a consensus among ecologists and bay managerial authorities 14
that the SAV community is very important and therefore SAV restoration is more important than any other water use in the shallow waters. The preference on part of the policy makers for SAV restoration must be reflected in the final analysis. We use Multicriteria Analysis (MCA) as the relevant decision rule. Multicriteria methods revolve around preference of decision makers and make way for simultaneous consideration of multiple conflicting obectives. They are also extremely helpful in methodically combining qualitative and quantitative types of information-the type of information we have in our case study. For its many advantages, we opt to use multicriteria decision analysis for determining the optimal land use at the sites post sea level rise. J Let A denote the finite set of n feasible actions a k ( k = 1,2,3..., n) or alternatives at site. For = -1,-2, -3,-4,-5, A = { R, A}, the set consisting of the two possible land use choices available for the water sites. G is the set of m evaluation criteria g i ( i = 1,2,3,..., m) considered relevant to the decision problem in the case study. For the case study, 5 relevant criteria (m=5) have been established and they are (1) Ecological services like habitat (g1) and water quality (g 2 ), (2) Cost of implementing that alternative (g 3 ), (3) Opportunity cost of excluding aquaculture under each alternative (g 4 ), and (4) the resultant SAV abundance at each site accruing out of the alternative choice for that site (g 5 ). Any land use (say a 1 ) will be rated better than another land use (say a 2 ) according to the i th criterion, if g ( a 1 ) g ( a 2 ). i > i The land use choice decision problem for each site for = -1,-2, -3,-4,-5 can be represented in a tabular form or in a 5 2 P matrix whose typical elements g ( a )( i = 1,2,..,5; k = 1,2) i k represents the evaluation of the k th alternative against the i th 15
criterion. The matrix P is the evaluation table for site. g i is a qualitative evaluation criterion measured on a nominal or ordinal scale for i = 1, 2,3 and a quantitative evaluation criterion measured on an interval or ratio scale for i =4 and 5. Qualitative evaluation methods like the regime method will then be used for producing a ranking among the alternatives. Weights will be assigned to the different criteria that reflect prioritization among the criteria by the decision makers. Decision trees combining available strategies and probabilities will be created. The payoff against each strategy will be stated in terms of (i) expected gains or losses in SAV acres, (ii) costs of restoration incurred, and (iii) opportunity cost of aquaculture. The land use choice that performs the best using some decision rule (MCA or expected acreage of SAV per unit cost) will be the optimal current land use choice at any given site under the uncertainty of sea level rise. Part (2) considers the optimal strategies for conserving/restoring SAV under sea level rise, given the availability of an alternate site, =0. If sea level does rise, SAV will vanish from bands =-4, -5 at one end of the restoration region but will be able to grow in the tidal marshes at site =0. Thus there arises the possibility of trading restoration opportunities between these sites. The strategies available to the decision maker under this set up are (a) restore site = -4, -5 but do nothing about protecting site =0, (b) restore = -4, -5 and also protect site =0, and (c) do not restore site = -4, -5 (equal to opening it to aquaculture) but protect site =0. The methodology here involves construction of decision tree and determining the payoffs (expected SAV acreage against the costs of obtaining it) for each strategy. The next step in this analysis would be to either impose a SAV target constraint and look 16
for cost effectiveness among the strategies or alternatively impose a budget constraint and look for the strategy that is both feasible and maximizes expected acreage of SAV. In part (3), the obective is to value the option of having an extra site available for growing SAV. If this site was not available, resource managers would have no choice but to restore the extreme sites given by = -4, -5. The possibility of having the other site available for restoration allows the managers the freedom of not restoring the extreme sites. This freedom will exist as long as the tidal marshes at site 0 are available. If due to land uses along the coast, they were to be eliminated, then no SAV can be gained in case of sea level rise. The option of the alternative site has value and resource managers need to evaluate that to know how much is it worth to protect the tidal marshes. In (3) the decision tree in part (2) will be modified by the inclusion of another chance node that represents whether site 0 will be available or not. The aim here is to treat strategies (a) and (b) listed in (2) above as investment choices. Some notion of acreage versus costs will be used to capture the aspect of value of or payoff against the strategies. The difference in value under strategy (a) from the value under strategy (b), if positive, will be the option value of preserving the site. 5. FUTURE PLANS Immediate future plans consist of achieving the empirical part of the analysis. There are data issues and informational gaps that need to be worked around. Long term future plans of the CARA proect for the case study include conducting value of information studies improved forecasts of sea level rise or proection of land uses along the coast. Climate change will also impact the temperature of the waters in which SAV thrive. 17
Temperature change along with sea level rise can be considered to fully understand the impacts of climate change on SAV restoration. Reference: Leffler, M. (1998). Aquaculture and Restoration. Maryland Marine notes, Education and Outreach from Maryland Sea Grant 16. Lipton, D. (2002). Chesapeake Bay Oyster Economics. Orth, R. J. and K. A. Moore. 1983. Chesapeake Bay: An Unprecedented Decline in Submerged Aquatic Vegetation. Science. 222:51-53 18