The seven benefit functions—for biodiversity, carbon storage, flood control, forage production, pollination, recreation, and water provision—have distinctly different spatial distributions, although some areas are of high value to multiple services and other areas are of low value to many (). For example, the largest agricultural valley in the ecoregion, Salinas Valley, is characterized by the following: a wide swath of high-value pollination services driven by the high proportion of land under crops benefiting from animal pollination; a narrow area of high–flood control services due to riparian vegetation; and low values of other services. Similarly, the mountain ranges throughout the ecoregion are characterized by natural forest cover (accounting for carbon storage values), high precipitation (water provision), and proximity to major population centers and accessibility by road (recreation). Accordingly, the following areas share high values for carbon storage, recreation, and water provision: the Santa Cruz Mountains, the Santa Lucia Mountains along the Big Sur Coast, and the northern Diablo Range.
Spatial Analysis of Biodiversity and the Chosen Ecosystem Services
The spatial correlations between the ecosystems services are low (A), with nearly as many negative correlations as positive ones. The overall average correlation is positive but low (0.08). The average correlation between biodiversity and services is also low. The highest correlation is between carbon storage and water provision (0.58). Other relatively high correlations (>0.2) are between recreation and water provision (reflecting the importance of natural cover to both) and between recreation and flood control. The latter correlation reflects the combination of the benefits of natural cover, the accessibility of riparian areas by road (recreation), and the importance of those areas for flood control. Although riparian areas are also important for water quality and aquatic diversity, neither of these was treated explicitly, so the value of riparian areas is underrepresented in our analysis. Negative correlations are restricted to pollination and forage production with other features.
Pair-Wise Spatial Associations between Biodiversity and the Production of Ecosystem Services
The pair-wise overlap between the seven networks derived by MARXAN for individual benefit functions (individual networks) is displayed in B. In most comparisons, the number of shared planning units is more than expected, but in some cases (i.e., with pollination and recreation, and pollination and forage production), there are far fewer shared prioritized planning units than expected. The same overlaps are shown as percentages of the smaller of the two relevant networks in C. In many cases, these overlaps are considerable fractions of the smaller networks, exceeding 50% in several cases and 30% in most, so they are substantial in practical terms.
The seven individual networks are summed over space in , highlighting the distribution of areas selected for different numbers of benefits. Some highly urbanized areas are not selected by MARXAN for any benefits (e.g., San Francisco, San Jose, and the northwest corner of Santa Clara County, in gray), in part because we excluded highly developed lands from some service networks and in part because such lands simply do not provide high levels of services or are unsuitable for management for services.
Ecosystem Service and Biodiversity Hotspots
Other areas, such as San Luis Obispo County, are selected for only a few benefits (in light blue). Although much of San Luis Obispo County is agricultural, there are few high-value crops that benefit from animal pollination. Furthermore, because of the sparse forest cover, this county has relatively low values for carbon storage. Although livestock values are relatively high in San Luis Obispo County, neighboring Kern County has far higher values likely because of nearby feedlots, slaughterhouses, and transportation routes. Accordingly, Kern dominates the network for forage production (E). A portion of the planning units, for example in the northern Diablo Range (, in pink), were selected for multiple benefits. The value of this area for carbon, water, and recreation is explained above. In addition, the relatively intact oak woodlands are important to biodiversity, forage production, and—owing to the proximity to dense population in San Jose—flood control. Interestingly, both this hotspot of overlap and the hotspot in the North Santa Lucia Range are areas where considerable public land has already been protected.
The extent to which individual benefit-function targets are met by the four comprehensive networks is depicted in A: Biodiversity; Non-biodiversity; All; and Strategic (all except forage production and pollination, which removes all negative correlations and overlaps; see Methods
). The “Biodiversity” network would protect a considerable supply of ecosystem services. All four networks achieve the carbon storage targets, but none achieves the water provision target (set at only 40% of total water use), and only one (“Non-biodiversity”) achieves more than 60% of the pollination target.
The Achievement of Alternative Strategies at Meeting Conservation Targets
Because target achievement was assessed at the stratification-unit level and then aggregated to the ecoregion level, there are services for which there are considerable surpluses despite unmet targets (e.g., water provision; B). In other cases, targets are well met, with surpluses (e.g., recreation, flood control) and without (e.g., carbon storage). The inability of “Non-biodiversity” and “All” to appropriately protect biodiversity demonstrates the risks to biodiversity associated with diluting the focus of conservation efforts without expanding the funds available for conservation. Such risks are greatly diminished when the ecosystem services targeted are chosen strategically (as in “Strategic”).
Coincidence and Side Benefits
There are major differences in the extent to which benefit-function targets could be met through the biodiversity network alone or with additions. The pollination targets are only 49% met by biodiversity, but they only need 10% additional land (). This additional 10% contributes relatively little to biodiversity targets. Yet if protection or restoration of natural habitat adjacent to farms pays off entirely through pollination-augmented agricultural profits [13
], these biodiversity benefits might come through strategic partnerships without the expenditure of conservation dollars.
Results from Adding Individual Ecosystem Service Targets to the Existing Biodiversity Network
Contrast this situation with recreation, for which targets are 82% met by the biodiversity network. To achieve the remaining 18%, we need 9% additional land, which has far greater benefits for biodiversity. Because so much recreation would be provided by the biodiversity network, additional recreation funding could potentially contribute to conservation.
Carbon targets are met entirely by the biodiversity network because sites with high carbon storage are crucial for forest conservation. Carbon credits applied to forests in California would offer new funding for key elements of the biodiversity network, just as they offer promise in developing nations [85
Finally, water provision targets are only 48% met by the biodiversity network, and the 21% additional land is highly valuable for biodiversity, both in total and per hectare. If this biodiversity value of lands valuable for water provision is a common phenomenon, great biodiversity benefits could accrue from the estimated 13% of terrestrial land that might be managed for urban water use [64
As human impacts on the environment expand in intensity and extent, there is a critical need to understand the degree of intersection between conservation priorities for biodiversity and for ecosystem services. This intersection of conservation priorities could achieve a measured and thoughtful balance between previously competing goals, while providing new sources of funding for its full-scale implementation.
The striking result of this preliminary analysis is the apparent contradiction between results of the spatial association and side benefit analyses. On the one hand are low correlations between the spatial distribution of the ecosystem-service benefit functions and relatively low levels of overlap between prioritized sites (). However, despite the generally low correlations, there are hotspots where high values of multiple benefits coincide (), although biodiversity protection was not strongly positively associated with any service (). Protecting regions selected for their biodiversity value is not likely to maximize protection of the full suite of benefits unless there are considerable changes to the process by which biodiversity priorities are determined.
On the other hand, the biodiversity network would protect impressive supplies of ecosystem services (). But networks configured to maximize the full suite of benefits could do even better (). The relatively low overlaps between sites that are most appropriate for different features when prioritized separately do not negate the possibility of considerable gains from simultaneous prioritization: sub-optimal but valuable sites may coincide. Developing methodologies for such combined planning analyses should be a top research priority.
We adapt a general planning framework for biodiversity to planning for ecosystem services and do not present detailed representations of individual ecosystem services. A much deeper analysis is warranted. The coarse scale of the readily available data for many of the ecosystem services hinders analysis of ecosystem services. The carbon pool and pollination analyses require data with finer resolution within relevant boundaries (C and D). For example, because most agricultural data are provided by political units such as counties, assessing pollination services requires interpolating fine-scale patterns from coarse-scale data, which likely introduces error. Despite the data limitations, this analysis yields five key insights for individual services, the relationships among them, and the exercise as a whole.
1. Suitability and demand are determining factors.
As with biodiversity planning, the network design process for ecosystem services is strongly influenced by factors other than patterns of biophysical supply. Prioritized sites generally have high feature values, but two other factors determine planning unit selection: suitability (lower in urban areas) and targets (intended to represent demand and specific to stratification units, outlined in dark gray in ). High feature-value sites are not selected for two reasons: low suitability for conservation and low pertinent demand or need. For example, the site-selection algorithm did not select some sites of high forage production in Kings and Kern counties, whereas it did select some low–forage-production sites in San Luis Obispo and Fresno Counties. The former sites have lower suitability due to agriculture and urban development, whereas the latter are more remote.
Remoteness is relevant for water provision also, where targets are linked to actual water use. Here, several planning units in the Santa Cruz Mountains (which have high precipitation; stratification unit 5,000, G) are not selected for water provision due to the relatively low demand compared to adjacent sites. Spatial mismatches between supply and demand complicate ecosystem-service provision and the planning for those services.
2. Spatial scale.
Two important points pertain to the issue of spatial scale. First, benefits vary in the scale of their operation and dependence on habitat, and this may dramatically affect simultaneous management for multiple services. Most strikingly, biodiversity conservation generally requires large intact landscapes, but crop pollination arises from small patches of (semi-) natural habitat within a human-dominated landscape (we did not consider long-term sustainability of pollinators, which might require larger patches). Not surprisingly, the two features are negatively associated spatially (), and each seems to greatly constrain the target achievement of the other in combined networks ().
Second, independent scales of supply and demand can affect relationships between target achievement and the total size of benefit demand and supply. Targets may be poorly met despite relatively high overall availability or they may be well met despite barely adequate availability. Targets are more easily met if demand occurs at broad scales and supply varies considerably at local to regional scales. For example, carbon storage demand is global, but supply varies greatly based on vegetation cover and climatic conditions; consequently, it was possible to meet the target of 50% of the ecoregion's carbon storage in all networks (). Although global demand makes it easier to meet regional targets, it also introduces artificiality: specific regional targets do not make much sense, because they ignore how well the global targets might be better met elsewhere. In contrast, when demand varies at smaller scales than supply, spatial mismatches are exacerbated and targets may be more difficult to achieve. For example, water demand accompanies agricultural use and residential development, which do not coincide spatially with areas of high water surpluses (precipitation minus evapotranspiration). Although water provision targets were easily met and exceeded in some stratification units (1,000, 5,000, and 6,000; see G), they could not be met or even approached in others (2,000 and 3,000, even with a relatively low target of 40% of total water use).
3. Population centers yield tensions.
For some ecosystem services, demand scales positively with the number of people in close proximity, whereas developed and agricultural lands are less productive or less suitable for management. These two factors result in a tension in planning, even for an individual service. For example, the demand and therefore the value of recreation opportunities is much greater close to cities (e.g., in the San Francisco Bay area, F). When people have alternative sources of outdoor recreation (e.g., South Bay area, in and around San Jose, F), the high value may be countered by low suitability (high costs of land management for recreation) such that high value sites are not selected by our method. When people have no other options (e.g., North of San Francisco, F), however, the high value supersedes low suitability, and the planning units are prioritized. Similarly, flood control and water provision services are more needed near cities, but are generally degraded by development.
4. Need new data, methods development.
To plan thoroughly for multiple ecosystem services, we need considerable advances in data and planning methodologies. Although there was sufficient data in this ecoregion for a first-pass analysis, planning for ecosystem services at smaller scales and in other ecoregions will likely require new research. For example, planning for crop pollination at finer scales requires an improved understanding of the contribution of individual pollinator species to particular crops, which is currently sparsely understood [86
]. In other places, we anticipate that the kinds of data collected by relevant government agencies in California does not yet exist.
Although the application of MARXAN yielded insights, the tool lacks several features that are required for ecosystem-service planning. First, a new tool should allow a single network to include different features with different suitability layers. Specific suitability would reflect the factors that affect that particular feature's management. This would allow simultaneous planning for terrestrial and aquatic diversity.
Second, a new tool should incorporate the possibility that targets will not be met with available resources or that they may be met from outside the planning region. Third, an ideal tool would incorporate some spatial and temporal dynamics to account for the potential impacts of management and threats on species and services. Ideally, conservation would target areas for protection based on the potential for loss of benefits, not simply for the benefits supplied under current land use as in this analysis. Such dynamics should also allow the representation of the dependence of ecosystem functions on changes in biodiversity; although these effects might not be generally strong for ecosystem stocks and fluxes, they are likely more important for stability [89
Fourth, such a tool should account for the fact that management for one purpose (e.g., threatened species) will be incompatible with management for another purpose (e.g., recreation). Fifth, a tool must account for the flow from particular ecosystems to particular beneficiaries. Site-selection software like MARXAN assigns value to a planning unit in the context of the larger stratification unit, without more specific accounting of spatial context or ecological processes [90
]. For example, in modeling the contribution of natural vegetation cover to flood control, we accounted for the proximity to the floodplain and for the population density in the relevant watershed's floodplain, but we could not specifically account for the population downstream that would be directly impacted by flood mitigation.
Finally, a tool should allow flexibility between the ends of benefit maximization (used by Naidoo and Ricketts [91
]) and suitability-maximizing target achievement (used here), which will each be appropriate for individual ecosystem services in different circumstances. Benefit maximization will be especially appropriate when services have substitutes whose appropriateness will also vary spatially; suitability maximization will be appropriate for features like biodiversity and perhaps recreation, for which the motivation for protection is principle rather than private preference [92
]. Such flexibility will allow more effective analysis and the increased potential for engaging partners whose interests in the full suite of ecosystem services will differ.
5. Need multidisciplinary and transdisciplinary teams.
Ecosystem-service planning must involve multidisciplinary and transdisciplinary teams. Interdisciplinarity (research between disciplines) is not sufficient, because ecosystem-service research and planning requires deep knowledge within—and across—multiple disciplines. Planning for ecosystem services requires expertise in biology, chemistry, physics, economics, finance, geosciences, geography, and particular analytical tools. The integration of theoretical understanding and empirical expertise from these diverse fields therefore requires a multidisciplinary team of experts working in close communication, spearheaded by transdisciplinary scholars and practitioners.
6. Consider trade-offs and side benefits.
Only by analyzing both the trade-offs and the side benefits for biodiversity of conserving ecosystem services and vice versa can we guide conservation efforts more effectively. Trade-off analyses will be applied most successfully when management for an ecosystem service cannot help to meet the targets of biodiversity conservation.
Analyses of the ancillary benefits of an ecosystem-service project to biodiversity conservation and vice versa have two purposes. Such analyses can reveal when an ecosystem-service project offers promise for attracting new conservation partners and funds for biodiversity projects, and when such projects are especially important for their biodiversity benefits.
By combining trade-off and side-benefit analyses with a thorough scoping of potential partnerships and new markets, we may achieve substantial increases in biodiversity conservation while conserving the ecosystem services critical for human well-being. For example, case studies of water regulation and delivery and flood control reveal that conservation of forests and wetlands are sometimes worthwhile from an ecosystem-service perspective alone (in the Yangtze River watershed, China [14
], around the Panama Canal [93
], and in the Catskills and Charles River watersheds, US [94
]). There are other places where such ecosystem-service values are undervalued or not quite sufficient to outweigh opportunity costs of conservation, but where the strategic investment of expertise and conservation funds could meet multiple goals simultaneously. If our results are representative of other places, lands for water provision and flood control may be particularly important for biodiversity conservation ().
The inclusion of ecosystem services in conservation planning has great potential to provide opportunities for biodiversity protection. This preliminary exercise seems to suggests that conservation planning for other services—either separately or in combinations with biodiversity—may result in considerable declines in the ability to meet biodiversity conservation targets, but this finding stems from assuming no new opportunities. Furthermore, strategic choices of particular services to include in conservation planning can yield considerable gains. Our strategic network of five benefits—biodiversity, carbon, flood control, recreation, and water provision—eliminated negative associations between features. This “Strategic” network met targets far better than did the “All” benefits network, both overall and especially for biodiversity protection ().
This study suggests that planning for ecosystem services would involve a major shift toward new geographies and a broadening of current conservation goals. The potential payoffs of such a shift are tremendous for both biodiversity conservation and human well-being [2
], promising to sustain critical services, open new revenue streams, and make conservation broad based and commonplace. The goal of simultaneously maximizing biodiversity conservation and ecosystem services critical to poverty alleviation and general human well-being is one that can be embraced by all.