We synthesize results from existing models of marine reserves to identify key theoretical issues that appear to be well understood, as well as issues in need of further exploration. Models of marine reserves are relatively new in the scientific literature; 32 of the 34 theoretical papers we reviewed were published after 1990. These models have focused primarily on questions concerning fishery management at the expense of other objectives such as conservation, scientific understanding, recreation, education, and tourism. Roughly one‐third of the models analyze effects on cohorts while the remaining models have some form of complete population dynamics. Few models explicitly include larval dispersal. In a fisheries context, the primary conclusion drawn by many of the complete population models is that reserves increase yield when populations would otherwise be overfished. A second conclusion, resulting primarily from single‐cohort models, is that reserves will provide fewer benefits for species with greater adult rates of movement. Although some models are beginning to yield information on the spatial configurations of reserves required for populations with specific dispersal distances to persist, it remains an aspect of reserve design in need of further analysis. Other outstanding issues include the effects of (1) particular forms of density dependence, (2) multispecies interactions, (3) fisher behavior, and (4) effects of concentrated fishing on habitat. Model results indicate that marine reserves could play a beneficial role in the protection of marine systems against overfishing. Additional modeling and analysis will greatly improve prospects for a better understanding of the potential of marine reserves for conserving biodiversity.

Abstract. Several schemes have been developed to help select the locations of marine reserves. All of them combine social, economic, and biological criteria, and few offer any guidance as to how to prioritize among the criteria identified. This can imply that the relative weights given to different criteria are unimportant. Where two sites are of equal value ecologically, then socioeconomic criteria should dominate the choice of which should be protected. However, in many cases, socioeconomic criteria are given equal or greater weight than ecological considerations in the choice of sites. This can lead to selection of reserves with little biological value that fail to meet many of the desired objectives. To avoid such a possibility, we develop a series of criteria that allow preliminary evaluation of candidate sites according to their relative biological values in advance of the application of socioeconomic criteria. We include criteria that, while not strictly biological, have a strong influence on the species present or ecological processes. Our scheme enables sites to be assessed according to their biodiversity, the processes which underpin that diversity, and the processes that support fisheries and provide a spectrum of other services important to people. Criteria that capture biodiversity values include biogeographic representation, habitat representation and heterogeneity, and presence of species or populations of special interest (e.g., threatened species). Criteria that capture sustainability of biodiversity and fishery values include the size of reserves necessary to protect viable habitats, presence of exploitable species, vulnerable life stages, connectivity among reserves, links among ecosystems, and provision of ecosystem services to people. Criteria measuring human and natural threats enable candidate sites to be eliminated from consideration if risks are too great, but also help prioritize among sites where threats can be mitigated by protection. While our criteria can be applied to the design of reserve networks, they also enable choice of single reserves to be made in the context of the attributes of existing protected areas. The overall goal of our scheme is to promote the development of reserve networks that will maintain biodiversity and ecosystem functioning at large scales. The values of ecosystem goods and services for people ultimately depend on meeting this objective.

Efforts to design monitoring regimes capable of detecting population trends can be thwarted by observational and economic constraints inherent to most biological surveys. Ensuring that limited resources are allocated efficiently requires evaluation of statistical power for alternative survey designs. We simulated the process of data collection on a landscape, where we initiated declines over 3 sample periods in species of varying prevalence and detectability. Changing occupancy levels were estimated using a technique that accounted for effects of false-negative errors on survey data. Declines were identified within a frequentist statistical framework, but the significance level was set at an optimal level rather than adhering to an arbitrary conventional threshold. By varying the number of sites sampled and repeat visits made, we show how managers can design an optimal monitoring regime that maximizes statistical power within fixed budget constraints. Results show that 2 to 3 visits/site are generally sufficient unless occupancy is very high or detectability is low. In both cases, the number of required visits increase. In an example of woodland bird monitoring in the Mt. Lofty Ranges, South Australia, we show that, although the budget required to monitor a relatively rare species of low detectability may be higher than that for a common, easily detectable species, survey design requirements for common species may be more stringent. We discuss implications for multi-species monitoring programs and application of our methods to more complex monitoring problems. JOURNAL OF WILDLIFE MANAGEMENT 69(2):473-482; 2005

Abstract. When viewed across long temporal and large spatial scales, severe disturbances in marine ecosystems are not uncommon. Events such as hurricanes, oil spills, disease outbreaks, hypoxic events, harmful algal blooms, and coral bleaching can cause massive mortality and dramatic habitat effects on local or even regional scales. Although designers of marine reserves might assume low risk from such events over the short term, catastrophes are quite probable over the long term and must be considered for successful implementation of reserves. A simple way to increase performance of a reserve network is to incorporate into the reserve design a mechanism for calculating how much additional area would be required to buffer the reserve against effects of catastrophes. In this paper, we develop a method to determine this ''insurance factor'': a multiplier to calculate the additional reserve area necessary to ensure that functional goals of reserves will be met within a given ''catastrophe regime.'' We document and analyze the characteristics of two relatively well-studied types of disturbances: oil spills and hurricanes. We examine historical data to characterize catastrophe regimes within which reserves must function and use these regimes to illustrate the application of the insurance factor. This tool can be applied to any reserve design for which goals are defined by a quantifiable measure, such as a fraction of shoreline, that is necessary to accomplish a particular function. In the absence of such quantitative measures, the concept of additional area as insurance against catastrophes may still be useful.

Using benthic habitat data from the Florida Keys (USA), we demonstrate how siting algorithms can help identify potential networks of marine reserves that comprehensively represent target habitat types. We applied a flexible optimization tool-simulated annealing-to represent a fixed proportion of different marine habitat types within a geographic area. We investigated the relative influence of spatial information, planning-unit size, detail of habitat classification, and magnitude of the overall conservation goal on the resulting network scenarios. With this method, we were able to identify many adequate reserve systems that met the conservation goals, e.g., representing at least 20% of each conservation target (i.e., habitat type) while fulfilling the overall aim of minimizing the system area and perimeter. One of the most useful types of information provided by this siting algorithm comes from an ''irreplaceability analysis,'' which is a count of the number of times unique planning units were included in reserve system scenarios. This analysis indicated that many different combinations of sites produced networks that met the conservation goals. While individual 1-km 2 areas were fairly interchangeable, the ir-replaceability analysis highlighted larger areas within the planning region that were chosen consistently to meet the goals incorporated into the algorithm. Additionally, we found that reserve systems designed with a high degree of spatial clustering tended to have considerably less perimeter and larger overall areas in reserve-a configuration that may be preferable particularly for sociopolitical reasons. This exercise illustrates the value of using the simulated annealing algorithm to help site marine reserves: the approach makes efficient use of available resources, can be used interactively by conservation decision makers, and offers biologically suitable alternative networks from which an effective system of marine reserves can be crafted.

Disturbance events strongly influence the dynamics of plant and animal populations within nature reserves. Although many models predict the patterns of succession following a disturbance event, it is often unclear how these models can be used to help make management decisions about disturbances. In this paper we consider the problem of managing fire in Ngarkat Conservation Park (CP), South Australia, Australia. We present a mathematical model of community succession following a fire disturbance event. Ngarkat CP is a key habitat for several nationally rare and threatened species of birds, and because these species prefer different successional communities, we assume that the primary management objective is to maintain community diversity within the park. More specifically, the aim of management is to keep at least a certain fraction of the park, (e.g., 20%), in each of three successional stages. We assume that each year a manager may do one of the following: let wildfires burn unhindered, fight wildfires, or perform controlled burns. We apply stochastic dynamic programming to identify which of these three strategies is optimal, i.e., the one most likely to promote community diversity. Model results indicate that the optimal management strategy depends on the current state of the park, the cost associated with each strategy, and the time frame over which the manager has set his/her goal.

Abstract. A decision theory framework can be a powerful technique to derive optimal management decisions for endangered species. We built a spatially realistic stochastic metapopulation model for the Mount Lofty Ranges Southern Emu-wren (Stipiturus malachurus intermedius), a critically endangered Australian bird. Using discrete-time Markov chains to describe the dynamics of a metapopulation and stochastic dynamic programming (SDP) to find optimal solutions, we evaluated the following different management decisions: enlarging existing patches, linking patches via corridors, and creating a new patch. This is the first application of SDP to optimal landscape reconstruction and one of the few times that landscape reconstruction dynamics have been integrated with population dynamics. SDP is a powerful tool that has advantages over standard Monte Carlo simulation methods because it can give the exact optimal strategy for every landscape configuration (combination of patch areas and presence of corridors) and pattern of metapopulation occupancy, as well as a trajectory of strategies. It is useful when a sequence of management actions can be performed over a given time horizon, as is the case for many endangered species recovery programs, where only fixed amounts of resources are available in each time step. However, it is generally limited by computational constraints to rather small networks of patches. The model shows that optimal metapopulation management decisions depend greatly on the current state of the metapopulation, and there is no strategy that is universally the best. The extinction probability over 30 yr for the optimal state-dependent management actions is 50-80% better than no management, whereas the best fixed state-independent sets of strategies are only 30% better than no management. This highlights the advantages of using a decision theory tool to investigate conservation strategies for metapopulations. It is clear from these results that the sequence of management actions is critical, and this can only be effectively derived from stochastic dynamic programming. The model illustrates the underlying difficulty in determining simple rules of thumb for the sequence of management actions for a metapopulation. This use of a decision theory framework extends the capacity of population viability analysis (PVA) to manage threatened species.

We often need to estimate the size of wild populations to determine the appropriate management action, for example, to set a harvest quota. Monitoring is usually planned under the assumption that it must be carried out at fixed intervals in time, typically annually, before the harvest quota is set. However, monitoring can be very expensive, and we should weigh the cost of monitoring against the improvement that it makes in decision making. A less costly alternative to monitoring annually is to predict the population size using a population model and information from previous surveys. In this paper, the problem of monitoring frequency is posed within a decision-theory framework. We discover that a monitoring regime that varies according to the state of the system can outperform fixed-interval monitoring. This idea is illustrated using data for a red kangaroo (Macropus rufus) population in South Australia. Whether or not one should monitor in a given year is dependent on the estimated population density in the previous year, the uncertainty in that population estimate, and past rainfall. We discover that monitoring is important when a model-based prediction of population density is very uncertain. This may occur if monitoring has not taken place for several years, or if rainfall has been above average. Monitoring is also important when prior information suggests that the population is near a critical threshold in population abundance. However, monitoring is less important when the optimal management action would not be altered by new information.