Below is the uncorrected machine-read text of this chapter, intended to provide our own search engines and external engines with highly rich, chapter-representative searchable text of each book. Because it is UNCORRECTED material, please consider the following text as a useful but insufficient proxy for the authoritative book pages.
6Â TheÂ PathÂ ForwardÂ INTRODUCTION This chapter recommends a path forward to inform decisions on ballast dis- charge standards to reduce the probability of invasions. This path is bracketed by the regulatory climate, the myriad of variables that affect whether organisms in discharged ballast are likely to lead to an invasion, and the state of the science to evaluate the riskârelease relationship associated with ballast water discharge. It is abundantly clear that reducing propagule pressure will reduce the prob- ability of invasions, when controlling for all other variables. There is both strong theoretical and empirical support for this, across a diverse range of habi- tats, geographic regions, and types of organisms (Sax et al., 2005; Lockwood et al., 2007; Davis, 2009). The key issue, however, is the ability to characterize quantitatively the riskârelease relationship, with the goal of functionally describ- ing the incremental reduction in invasion probability achieved with declining propagule supply. Approaches to setting ballast discharge standards have relied primarily on expert opinion to evaluate the riskârelease relationship. The associated history and process (discussed in Chapters 2 and 5 respectively) have resulted in an ar- ray of different international, national, and state discharge guidelines and regula- tions that seek to reduce propagule supply below that of untreated ballast water. These differences result from both uncertainty about the riskârelease relation- ship and from the diverse approaches of different decision makers and stake- holders. Despite uncertainty about the riskârelease relationship (discussed fur- ther below), the IMO standards are clearly a first step forward, serving to reduce propagule pressure and thus the scale (number and rate) of invasions. In the initial (D-1) IMO phase, the use of open-ocean ballast water exchange (BWE) serves to routinely reduce the concentration of coastal organisms compared to unexchanged water (Ruiz and Reed, 2007; Bailey et al., 2011; Figure 1-6). In a second (D-2) phase, ballast water exchange is to be replaced by treatment tech- nologies to achieve specific discharge standards, in which allowable concentra- tions of organisms vary by taxonomic group and by size. For zooplankton greater than or equal to 50 Âµm, the D-2 standard provides a further reduction in Â 122Â
TheÂ PathÂ ForwardÂ 123Â Â coastal organism concentrations beyond that achieved with ballast water ex- change (Minton et al., 2005). Ballast water exchange has been implemented in the U.S. and many other countries (Chapter 2), but its quantitative effectiveness on reducing invasions is not yet known. To our knowledge, no new ballast-mediated invasions have been reported in the Great Lakes or San Francisco Bay in the five year period be- tween 2006 and 2010. This is encouraging and may suggest a decrease of inva- sions coincident with widespread use of ballast water exchange for these two locations; however, we urge great caution in any interpretation of a causal rela- tionship at the present time. Such conclusions are premature, because of limita- tions of the short and uneven baseline data. Mandatory ballast water exchange commenced in the United States only in 2007, and even though this began earli- er in the Great Lakes, such treatment was not applied to NOBOB vessels until more recently. Importantly, neither region has any coordinated or standardized set of measures designed explicitly to detect new invasions. There has been no formal analysis in the last five years for either location to evaluate time lags between establishment and detection, which can obfuscate reliable measures of changing invasion rates (Ruiz et al., 2000; Solow and Costello, 2004). Moreo- ver, there have been other periods in the past (pre-ballast water exchange) where no new nonindigenous species invasions related to ballast water have been de- tected in these two systems. In general, invasion probability increases with propagule pressure when conditions in the recipient system (such as a bay or port) are suitable for coloni- zation. As outlined in Chapter 4, however, the inflection point(s) in this rela- tionship are simply not known. These inflection points are required to predict invasion probability over the operational range of discharge concentrations be- ing considered. It is critical to also recognize that this relationship will vary with the nature of propagule pressure itself (e.g., species richness, abundance, frequency of inoculation, life stage, and genetic composition) as well as source region, recipient region, taxonomic group, season, voyage conditions, and many other potential variables (see Chapter 3). Such variation has important implica- tions for establishing discharge standards, in that invasion probability is ob- viously context dependent, with many potential influences beyond simply the number of propagules delivered. Looking forward, there are several approaches that can be used to predict the riskârelease relationship, using models described in Chapter 4. One or sev- eral of these models should be pursued in the months and years ahead to provide the foundation for a robust, scientific basis for setting and refining specific dis- charge standards. At the present time, none of the available models have been validated, due mainly to a lack of key data, creating too many untested assump- tions to provide confidence in the resulting estimates of invasion probability. Any successful approach also must consider the range of environmental and ecological conditions, geographic locations, and taxonomic groups (with differ- ent life cycle strategies, physiological capacities, habitat requirements) encom- passed in the global scale of shipping.
124Â Â PropaguleÂ PressureÂ andÂ InvasionÂ RiskÂ inÂ BallastÂ WaterÂ Â The sections below outline current limitations associated with models, in- cluding data needs, and a strategy to address these critical data gaps and quantify the riskârelease relationship. MODELS AND DATA GAPS All of the quantitative models reviewed in Chapter 4 seek to explain or pre- dict the number of invasions (probability of successful establishment) as a func- tion of propagule supply. These approaches assume that (a) reliable measures of propagule supply and invasion outcome are available over some meaningful spatial, temporal, and taxonomic range or (b) this relationship can be reliably derived from first principles (theory) in population biology. At the present time, neither of these criteria is met. There is a critical lack of data needed to populate empirical models or to validate theoretical ones. To directly estimate the riskârelease relationship in any of these approaches, one needs to know (or estimate reliably) the number of viable individuals of each species delivered in ballast water discharge to a particular location (e.g., a port), over some spatialâtemporal scale. There are few locations around the globe, let alone the U.S., for which direct measures of propagule supply in bal- last water are available. In North America, these include Port Valdez (Alaska), Coos Bay (Oregon), San Francisco Bay (California), Puget Sound (Washington), the Great Lakes, and Chesapeake Bay (see Table 4-2). However, most of these historical ballast data are collected for a very limited time horizon and sample size. Among the larger studies, for example, Carlton and Geller (1993) sampled 157 vessels arriving in Coos Bay from 1986â1991; Hines et al. (2000) sampled 196 vessels arriving in Port Valdez from 1999â2001, and Cordell et al. (2009) sampled 372 vessels arriving in Puget Sound from 2000â2007. While ballast studies have provided important insights into associated biota and concentrations, it is also critical to recognize that the taxonomic resolution has been insufficient to obtain occurrence and abundance measures for most species. This limitation exists because many ballast water organisms are larvae that cannot be identified to species based simply on morphology. Current mole- cular tools can permit identification of many but not all species, but these tools have largely been developed since most of the intensive ballast water sampling was conducted in the U.S., and even with genetic techniques quantitative esti- mates may still be a challenge. Thus, estimates for historical propagule supply from ballast can at best provide a temporal snapshot of concentrations of total organisms and selected taxa at only a few ports. As noted above, to determine the riskârelease relationship using historic in- formation it is also critical to know the number of invasions that have occurred and are attributed to ballast water at the same location and same time frame. There are many retrospective analyses that examine the invasions attributed to ballast water (and other vectors), underscoring the significance of this vector. However, there are also some important limitations in these data. First, the
TheÂ PathÂ ForwardÂ 125Â Â probability of detection has been highly variable in space and time, due to dif- ferences in search effort, expertise, and other factors (Ruiz et al., 2000; Ruiz and Hewitt, 2002). Remarkably, there is no current national program designed to provide standardized, field-based measures to detect new invasions and track status and trends (i.e., temporal and spatial patterns) for coastal invasions, de- spite the implementation of national programs to reduce invasions. Instead, ana- lyses are based largely on âby-catchâ data (invasions encountered in the course of other studies) or on localized and infrequent surveys. Thus, the available data result from independent and variable efforts, usually with different aims and methods, limiting inferences that make drawing comparisons through space or time difficult. Second, there is uncertainty about the vector(s) responsible for many coastal invasions, because multiple vectors are possible for some species. A large per- centage of species that arrive in estuarine and marine waters in the past decades is polyvectic, sharing ballast water, hull fouling, sea chest fouling, and other vectors as possible sources (Cohen and Carlton, 1995; Fofonoff et al., 2003; Ruiz et al., 2011). This occurs because species have one or more life stages that can be entrained by multiple vectors. For example, the same species of barnacle may conceivably be transported by a given vessel both in hull fouling as adults and in ballast water as larval stages. The data on ballast-mediated propagule supply are arguably best for holop- lankton, such as calanoid copepods, which have adult forms that can be identi- fied to species, unlike most meroplankton (e.g., Cordell et al., 2009). In addi- tion, holoplankton are less likely to be polyvectic and are attributed primarily to ballast water. However, even for this group it is challenging to estimate a quan- titative riskârelease relationship because of limitations (i.e., the patchy and in- complete nature) in the measures for propagule supply and/or invasion rate. It is therefore apparent that additional data are needed for directly estimating riskâ release relationships using any models, as discussed below. Proxy Variables Due to the current lack of direct, species-specific data for determining prop- agule pressure, some analyses have attempted to use proxy measures to examine riskârelease relationships. For example, one might use the number of people (or vehicles) visiting a national park as a proxy for propagule pressure of nonindi- genous species to that park and test for a relationship to the number of estab- lished nonindigenous species (Lonsdale, 1999; Stohlgren et al., 2003). In many cases, good statistical relationships between variables have been observed, but the practitioner must be careful that the employed proxy is in fact causal and not simply correlative. In the case of ballast-mediated invasions, several studies have used the number of ship arrivals as a proxy for propagule supply. Even if a relationship is found between vessel arrivals and invasions (in space or time), this relation-
126Â Â PropaguleÂ PressureÂ andÂ InvasionÂ RiskÂ inÂ BallastÂ WaterÂ Â ship may be spurious and not causal. In fact, the number of ship arrivals is a poor proxy for propagule supply, in large part because of high variation in bal- last volume discharge among vessel arrivals. Overall, the majority of ships re- port that they do not discharge ballast water upon arrival (Falkner et al., 2009; Miller et al., 2010), stating that they are offloading cargo, do not have ballast water aboard, or simply do not need to discharge it. In contrast, some vessels can discharge > 50,000 metric tons of ballast upon arrival in the same port. Moreover, there are large differences among vessel types and arrival ports in per-capita ballast discharge (Carlton et al., 1995; Verling et al., 2005; Miller et al., 2010). Together, these factors result in a poor relationship between ballast discharge volume and ship arrivals across ports (see Chapter 1). Ballast water discharge volume is another proxy variable and an improve- ment over vessel arrivals, in that it captures one important source of variation among vessels. Combined with information on organism concentrations, this proxy could allow estimates of either total number of propagules per discharge or cumulative propagule supply over some temporal and spatial scale. Nonethe- less, volume by itself is unlikely to be representative of propagule supply be- cause significant variation occurs among vessels in the diversity and concentra- tion of biota in their ballast water and sediments. This variation reflects differ- ent source locations, seasons, age and management of the water, voyage condi- tions, and other variables (Smith et al., 1999; Minton et al., 2005; Verling et al., 2005). For example, total zooplankton concentration in ballast water is known to vary five orders of magnitude (Minton et al., 2005). Thus, there is a consi- derable range of uncertainty in translating ballast water volume to total propa- gule supply, even before considering species-level information. This may be one reason why a significant relationship between total ballast discharge volume and reported invasions was not observed among 24 estuaries in the U.S. (Lee et al., 2010). Finally, Lee et al. (2010) attempted to translate ballast volume into total or- ganism (zooplankton) discharge per vessel and estuary, using a pooled distribu- tion of organism concentrations observed for sampled ballast tanks across sever- al discreet locations. While this is conceptually a step forward, ballast volume effectively becomes the key (proxy) variable for propagule supply in the analy- sis, which cannot yet capture estuary-specific or regional differences that exist in the composition and concentrations of organisms delivered in ballast (as dis- cussed in Chapter 4). This approach represents the limits of current data that exist for ballast biota delivered to most ports (see above). Clearly, there is potential value in using proxy variables to characterize propagule supply associated with ballast discharge. If reliable proxies could be established, such as the actual volume of ballast water discharged, this would represent an efficient and cost-effective approach to tracking propagule delivery into the future, and it may also open up enormous potential to examine historical data. To date, the performance of available (proposed) proxies has been of li- mited utility, undermined by the highly variable biotic assemblages of ships operating on a truly global scale (see Chapter 3). Nonetheless, further research
TheÂ PathÂ ForwardÂ 127Â Â into development of possible proxies is warranted, given the potential pay-offs in tracking past, present, and future propagule pressure. Moreover, with the onset of new regulatory and monitoring regimes described below, the ability to use proxies to track future propagule pressure with more precision will be en- hanced. STRATEGIES FOR MOVING FORWARD: GATHERING OBSERVATIONAL AND EXPERIMENTAL DATA To advance the state of understanding about the riskârelease relationship requires the use of one or several modeling approaches (see Chapter 4) and meeting associated data requirements. Because of existing data gaps, past ana- lyses have relied on either proxy variables that have been of limited value or theoretical approaches that have not yet been validated. There are two sources of data that can be used to quantify the riskârelease relationship: observational (descriptive) data and experimental data. Both require a substantial commitment in the design, implementation and analysis stages, and sufficient replication to demonstrate broad applicability and generality of results. Generation of usable and meaningful descriptive data will also require a substantial time horizon. Importantly, the same data from either source can be used in a variety of the different models described in Chapter 4, as well as multiple models to test for concordance, as many models share some data requirements. However, the ear- lier a preferred model is identified, the earlier that data collection can commence to serve the needs of the model most efficiently. The sections below discuss the merits and constraints for each data type. Descriptive Data Descriptive data can be collected from analyses of field-based sampling to directly measure propagule supply delivered in ballast water and the associated (detected) rate of invasion. The advantage with this approach is that it could potentially describe the actual relationship that exists in the field. It implicitly includes all of the real-world factors (environmental and biological variables, as well as appropriate spatial and temporal scales) that may affect the outcome of inoculation and the probability of subsequent establishment, whether these are measured or not (Diamond, 1986). A descriptive approach requires two types of field-based surveysâship dis- charge monitoring and receiving system monitoringâto estimate the relation- ship between propagule delivery and invasion probability for one or several est- uaries. A well-designed ship discharge sampling program to measure propagule supply involves stratified random samples across ship type, source regions, and season that is repeated across years. For each ship sampled, the goal would be to characterize the abundance of individual species or taxonomic group of inter-
128Â Â PropaguleÂ PressureÂ andÂ InvasionÂ RiskÂ inÂ BallastÂ WaterÂ Â est, creating a species- or taxon-specific estimate of propagule supply. If the goal is a comprehensive assessment of sampled ballast, including larval forms, this would also require molecular analyses to identify propagules to species. It should be noted that the nature of propagules in ballast tanks is changing, as more and more vessels arrive in the U.S. without intact, original ballast water from a foreign port or estuary, but rather (due to ballast exchange requirements) with largely oceanic (high seas) holo- and meroplankton. Because concentra- tions of coastal plankton are expected to decrease through time with increasingly stringent ballast management, it is important to consider the effects of temporal changes in evaluation of propagule supply. In parallel to ballast water analyses, standardized field surveys are needed for the same estuaries, repeated over time, to estimate the occurrence of new invasions for the same taxa or species found in arriving ballast. Ideally, these surveys would involve many taxa across habitats, since it is not possible at the present time to select a subset of âindicatorâ species that are known to be repre- sentative of invasions in general. Analyses of survey results should also consid- er the occurrence of polyvectic species in evaluating new invasions, to address whether ballast is the probable mode of introduction. While obtaining such measures of propagule supply and invasion outcome for a single estuary (port system) is necessary for evaluating the riskârelease relationship in the field, it is not sufficient to address generality. That is, an as- sessment in any one estuary is not likely to be representative. The riskârelease relationship is expected to differ greatly among locations, both due to differenc- es in propagule supply and also to invasion susceptibility (Lonsdale, 1999). For example, this relationship may be very different among the Great Lakes, San Francisco Bay, Chesapeake Bay, or Tampa Bay (Florida). To achieve a robust understanding, it would be necessary to replicate this descriptive approach in several estuaries, ideally in different geographic regions, considering differences in (a) source region of vessels and associated propagules, (b) recipient region environmental conditions, and (c) recipient region hydrographic regime and retention. A critical feature and disadvantage of such descriptive measures is the time horizon. Under the best of circumstances in terms of survey intensity, it is esti- mated to take a decade to (a) estimate propagule supply and (b) detect some of the resultant invasions, accounting for time lags in discovery of nonindigenous species in the field. The nature of this approach is a retrospective analysis, re- quiring several years to accumulate sufficient data to adequately represent in- oculationâestablishment outcomes (the relationships between propagule supply and invasion rate).
TheÂ PathÂ ForwardÂ 129Â Â Experimental Data Experimental measurements have advantages and disadvantages that are somewhat complementary to those for descriptive measures. With experiments, it is possible to know and control precisely the propagule pressure for individual or multiple species, exposing target laboratory or mesocosm communities under many different conditions, and measure the outcome. As with descriptive meas- ures, it is critical to replicate experiments across a range of taxonomic groups and environmental conditions, to provide robust data. Even with a large number of experiments, the time horizon here to acquire these data is likely to be much shorter than what would be necessary for descriptive data. The principal limitation with an experimental approach is the potential ina- bility to capture all of the relevant variables, spatial scales, and taxonomic groups. Mesocosms are artificial settings that cannot precisely mimic real-world conditions, despite every effort to do so. For example, the relatively large sur- face-to-volume ratio in mesocosms may influence plankton behavior and sur- vival in ways not found in nature; the larger the mesocosm, the more one can reduce artifacts associated with walls and differential light penetration and the lower the propagule density of the inoculum that one can achieve, thereby pro- viding greater reality to ballast discharges. In addition, it is not known a priori which species or taxonomic groups are most representative for examining inva- sion dynamics (i.e., the riskârelease relationship) in general. Thus, caution is warranted in interpreting the data from such experiments relative to translating the results to real-world scenarios (see Diamond, 1986). Terrestrial ecologists have long conducted colonization experiments at comparatively large scales (e.g., mangroves studied by Simberloff and Wilson, 1969), and have manipulated propagule pressure and other variables (Vila et al., 2008; MacDougall and Wilson, 2007). In general, less experimental progress has been made in aquatic colonization studies (but see Chadwell and Engelhardt, 2008). Typically either the volume used in experiments is too small and/or the density of colonists too high to realistically simulate the dynamics of ballast release (although see Bailey et al., 2009). Given that large-scale experiments have been used to assess ecological effects of deforestation, lake acidification, and cultural eutrophication (Ricklefs and Miller, 1999), there is no reason to believe that similar studies cannot be conducted to assess riskârelease relation- ships in freshwater or marine communities, using large-scale experimental me- socosms. These would need to either be closed (quarantine) systems or use local biota to prevent any accidental release, but this is not technologically challeng- ing to achieve. Ideally such studies would be conducted with an array of species with different life histories, reproductive modes, sizes, and source regions to represent the range historically and currently observed in ballast discharges to the U.S. Even developing extensive mesocosm facilities necessary for this ap- proach, and with extensive replication, such experiments can be implemented with a fraction of the effort, cost, and time needed for descriptive measures.
130Â Â PropaguleÂ PressureÂ andÂ InvasionÂ RiskÂ inÂ BallastÂ WaterÂ Â While experiments offer a powerful and relatively rapid approach to under- standing invasion dynamics, they cannot capture all of the variables and scales that may affect ballast-mediated invasions. One approach to minimize this drawback may be to conduct experiments under âbest-caseâ conditions that are favorable to establishment and that would represent a conservative estimate and a precautionary approach. For example, conditions might simulate a small la- goon or port with high retention, favorable abiotic conditions (e.g., temperature and oxygen), ample food, and low predation or competition. In addition, target species for experiments could also be selected from this perspective, to include a range of taxa with frequent invasions around the globe or with attributes (e.g., parthenogenic, fast-growing, trophic or habitat generalist) considered especially conducive to establishment. CONCLUSIONS AND RECOMMENDATIONS The current state of science does not allow a quantitative evaluation of the relative merits of various discharge standards in terms of invasion probability (see the Statement of Task in Chapter 1). Of the approaches suggested to date and reviewed in this report, descriptive statistical modeling with proxy variables (such as the per capita invasion probability approach; see Chapter 4) is currently available to empirically examine the riskârelease relationship because there are data available for ballast volume (derived indirectly from vessel arrival data) and historical invasion rates across estuaries. However, it must be cautioned that these are extraordinarily coarse-level data because (a) as noted above, vessel arrival data often do not directly translate into a measure of ballast water actual- ly discharged, (b) when actual ballast volume data are available, these do not translate well into known propagule supply, such as species richness or abun- dance, and, further, (c) there is no significant relationship between ballast vo- lume and invasions. In addition, the actual number of historical invasions in all estuaries is considered to be underestimated, perhaps significantly so (Carlton, 2009). Thus, while statistical modeling has been applied to current datasets, the data are not sufficient in present form to characterize a biologically meaningful relationship, much less estimate the associated uncertainty, to be able to identify with confidence the invasion probabilities associated with particular discharge standards (propagule supplies). Several actions are needed to advance a robust understanding of the riskâ release relationship in order to inform decisions about ballast water discharge standards. As a logical first step, a benchmark discharge standard should be established that clearly reduces concentrations of coastal organisms below cur- rent levels resulting from ballast water exchange (such as the IMO D-2 stan- dard). This will serve to reduce the likelihood of invasion in coastal ecosystems beyond that at the present time. Following the setting of an initial benchmark, a riskârelease model or mod- els should be selected as the foundation for the data gathering and analysis ef-
TheÂ PathÂ ForwardÂ 131Â Â fort. What model or models is ultimately chosen will reflect the available re- sources, in terms of time, data, and personnel. Indeed, using multiple models with the same data could be valuable to test for concordance. This would also allow one to assess the range of outcomes that would result from proposed bal- last water discharge standards. Furthermore, there is considerable worth in tran- sitioning from simple single-species models to more complex multi-species me- chanistic models as more data become available. Finally, a two-track approach should be pursued to obtain both experimental and field-based (descriptive) data. Experiments can be used to evaluate the riskârelease relationship and should deliver results over the next three to five years. Field-based descriptive data should also be collected and analyzed to parameterize the same types of models, providing real-world validation of expe- rimental data. Results from such efforts would be expected to materialize in about ten years. For both experimental and descriptive measures, there is a clear need for a highly directed and coordinated effort. Existing data are limited, and there is no reason to expect a different outcome from future research without a concerted focus. There is a critical need for an explicit sampling design, standardized me- thods and analyses, and data integration across multiple sites and times. A con- sistent framework and sufficient oversight and coordination will be necessary to produce the high-quality data specifically needed to populate riskârelease mod- els, especially where multiple locations or organizations are involved. Regard- less of the structure, any such program must include a centralized data reposito- ry, along with active periodic audits on data gathering efforts over time to ensure a standardized methodology is being employed. There are many potential designs for the structure, implementation, and ac- tual measures that would make up each descriptive and experimental study to quantify the riskârelease relationship. Elaboration of the specific designs and protocols is certainly beyond the scope of this report and would require a sepa- rate decision making process about particular models to be used and the level of effort that would be pursued. The following recommendations are intended to provide guiding principles for that purpose. Recommendations for Experiments Experiments should be used to estimate the effect of propagule pressure on establishment success, using statistical and probabilistic models (Chapter 4). The experiments should (a) be conducted in large-scale mesocosms designed specifically to simulate field conditions, (b) include a diverse range of taxa, en- compassing different life-histories and species from known source regions of potential invasions, and (c) include different types of environments (e.g., fresh water, estuarine, and marine water) where ballast discharge may occur. While it is possible to identify a diverse range of experiments that would be highly informative, initial experimental efforts should focus especially on sin-
132Â Â PropaguleÂ PressureÂ andÂ InvasionÂ RiskÂ inÂ BallastÂ WaterÂ Â gle-species riskârelease relationships. Ideally, these would include taxa and conditions that are selected as âbest caseâ scenarios, seeking to maximize inva- sion success and provide a conservative estimate of invasion probability (as dis- cussed previously). Thus, rather than experiments that examine complex and interactive effects of many different environmental and biological variables, a premium is placed on relatively simple initial experiments that provide a signifi- cant amount of data across âmodelâ taxa and conditions in a short amount of time. This approach should be applied to multiple species, and serious consider- ation should be given (and explicit criteria used) to select the appropriate organ- isms and conditions. The experiments should be advanced aggressively, in a directed fashion, to yield results in a three- to five-year time horizon. While this represents a signif- icant investment in effort and resources, it is the most cost- and time-efficient path to obtaining critical data needed to parameterize riskârelease models com- pared to field-based measures (see below). Experiments could potentially iden- tify a solid interim basis for discharge standards, noting the inherent challenges in working with a limited number of species and the assumptions that these would be representative of a broad array of potential invasions. Importantly, these data may also have direct application to other vectors, in addition to ballast water, as they test basic questions about establishment that are relevant to prop- agule pressure arising from all vectors. Recommendations for Descriptive Studies In addition to experiments, descriptive field-based measures are recom- mended to ground-truth the models, providing a critical validation step to con- firm that (a) riskârelease relationships are consistent with experimental results and (b) observed invasion rates are consistent with these predictions. Imple- menting such an effort at one location is not sufficient. This should occur at selected sentinel estuaries (e.g., San Francisco Bay, Chesapeake Bay, and Tam- pa Bay), chosen to include different coasts, ship traffic patterns, source regions, and environmental conditions. For each sentinel estuary, measures of propagule supply (in shipsâ ballast) and invasion rate would be made repeatedly over a minimum of a ten-year time horizon to provide a data set for independent analy- sis and validation of experimental results. The specific design of data collection needs to be defined explicitly, consi- dering the model(s) being used and making sure that the output will represent the riskârelease relationship and directly translate to a discharge standard. While it may be reasonable to explore potential proxy variables as one compo- nent, it is critical to not focus extensively on proxies or other variables that may not represent the riskârelease relationship. Also critical is an a priori estimate of the uncertainty explicit at all scales, as well as sampling effort (number and frequency of measures), in order to properly design measures and interpret and compare predictions. The same data could be used for statistical and probabilis-
TheÂ PathÂ ForwardÂ 133Â Â tic models, moving toward increasing resolution (e.g., hierarchical probability models (HPM) as described in Chapter 4) if and as appropriate data are availa- ble. However, collecting the data required to examine multiple species across multiple estuaries with HPM would require a large effort, sustained over a con- siderable period of time. As outlined previously, a comprehensive model would require sampling many vessels (stratified by vessel type, season, and source) and quantifying the concentration of each species present in discharged ballast (as well as volume per discharge event). Field surveys to detect invasions of these species would also need to be conducted coincident with ballast measures. One possible strategy would be to focus on a subset of target species dis- charged in ballast water to multiple estuaries. This would reduce considerably the effort required for analysis of ballast water, compared to characterizing the entire community. It may also serve to reduce the sampling effort, and increase the probability of detection, of the target species in field surveys. Intuitively, it would make sense to focus particular attention not only on species that can be identified and counted in ballast samples, but also on species that are unlikely to be polyvectic (such as copepods and mysids), providing the clearest signal (and least noise) for analysis of riskârelease relationships associated with ballast wa- ter. With this strategy, selection of taxa is critical and should take into consider- ation biological and environmental requirements (especially whether suitable conditions exist in the specific estuary). As discussed earlier, a challenge is how generally representative any such species would be. Nonetheless, this would result in single-species models (for multiple species) in parallel to the experi- mental approach outlined previously. While collection of field-based descriptive data required for a meaningful analysis of the riskârelease relationship is somewhat daunting in scope, recent developments make this more feasible than in the past, with continuing rapid improvements in the sensitivity and efficiency of analytical methods. First, pending international and national regulations require commercial vessels to install sampling ports that provide representative and standardized samples of ballast discharge (see Chapter 2); a similar requirement already exists under California state law (California State Lands Commission, 2010). This will pro- vide an important platform for ready access and standardized, comparable sam- ples across vessels and locations. Methods for compliance testing and analyses are already being developed to further standardize data that results from any sampling. Second, the implementation of ballast water treatment systems will reduce the concentrations, and possibly the diversity, of organisms in ballast water (Chapters 1, 3, 4). This may serve to simplify the sampling, having less biological material to process for quantitative analysis. Third, the use of mole- cular genetic tools has dramatically expanded the capacity (and reduced the time, effort, and cost) to detect species, based on DNA. Several methods al- ready exist that provide high sensitivity in detecting target species in very large volumes of water, including the ability to estimate abundance with quantitative PCR, and the capacity of Next Generation Sequencing is now available for spe- cies detection and analyses at the scale of entire communities (Mardis, 2008a,b;
134Â Â PropaguleÂ PressureÂ andÂ InvasionÂ RiskÂ inÂ BallastÂ WaterÂ Â Darling and Mahon, 2011). Importantly, these methods are also rapidly improv- ing in terms of sensitivity, speed, and cost. Although motivated by understanding the riskârelease relationship, it is perhaps useful to point out that field-based measures outlined above would also serve a broader range of applications. These data can provide critical feedback for adaptive management, identifying performance of discharge standards to reduce invasions (see Ruiz and Carlton, 2003). Field-based surveys provide an important baseline, now lacking, to track invasions from other vectors concur- rently. In addition, such surveys would align with recommendations from the Interagency Ocean Policy Task Force (2010). The Task Force plan subdivides the United States into nine regional planning areas and recommends the federal government support âdisciplinary and interdisciplinary science, research, moni- toring, mapping, modeling, forecasting, exploration, and assessment to conti- nually improve understanding of ocean, coastal, and Great Lakes ecosystemsâ (Interagency Ocean Policy Task Force, 2010). *** To date, there has been no concerted effort to collect and integrate the data necessary to provide a robust analysis of the riskârelease relationship needed to evaluate invasion probability associated with particular ballast water discharge standards. Existing experimental and field data are of very limited scope. There is currently no program in place to implement either ship-based ballast sampling or field surveys to detect new invasions across sites. On-going research pro- vides confidence that this approach is feasible, but it is scattered across sites and usually short-term in nature. Several models exist which can quantify the riskâ release relationship, given sufficient data that are now lacking. This report out- lines the paths, using multiple methods over different time frames, that could address these data gaps, and thus provide a robust foundation for framing scien- tifically supportable discharge standards for ballast water. REFERENCES Bailey, S. A., L. A. Velez-Espino, O. E. Johannsson, M. A. Koops, and C. J. Wiley. 2009. Estimating establishment probabilities of Cladocera introduced at low densi- ty: an evaluation of the proposed ballast water discharge standards. Canadian Jour- nal of Fisheries and Aquatic Sciences 66:261â276. Bailey, S. A., M. G. Deneau, L. Jean, C. J. Wiley, B. Leung, and H. J. MacIsaac. 2011. Evaluating efficacy of an environmental policy to prevent biological invasions. En- vironmental Science and Technology 45:2554â2561. Carlton, J., and J. Geller. 1993. Ecological roulette: The global transport of nonindigen- ous marine organisms. Science 261:78â82. Carlton, J. T., D. M. Reid, and H. van Leeuwen. 1995. Shipping Study. The role of shipping in the introduction of non-indigenous aquatic organisms to the coastal wa- ters of the United States (other than the Great Lakes) and an analysis of control op-
TheÂ PathÂ ForwardÂ 135Â Â tions. The National Sea Grant College Program/Connecticut Sea Grant Project R/ES-6. Department of Transportation, United States Coast Guard, Washington, DC and Groton, Connecticut. Report Number CG-D-11-95. Government Accession Number AD-A294809. Carlton, J. T. 2009. Deep invasion ecology and the assembly of communities in histori- cal time. Pp. 13â56 In: Biological Invasions in Marine Ecosystems. G. Rilov and J. A. Crooks, editors. Berlin: Springer-Verlag. Chadwell, T. B., and K. A. M. Engelhardt. 2008. Effects of pre-existing submerged vegetation and propagule pressure on the invasion success of Hydrilla verticillata. Journal of Applied Ecology 45:515â523. Cohen, A. N., and J. T. Carlton. 1995. Biological Study. Nonindigenous Aquatic Spe- cies in a United States Estuary: A Case Study of the Biological Invasions of the San Francisco Bay and Delta. A Report for the United States Fish and Wildlife Service, Washington, DC, and The National Sea Grant College Program, Connecticut Sea Grant, NTIS Report Number PB96-166525. Cordell J., D. Lawrence, N. Ferm, L. Tear, S. Smith, and R. Herwig. 2009. Factors in- fluencing densities of non-indigenous species in the ballast water of ships arriving at ports in Puget Sound, Washington, United States. Aquatic Conservation: Marine and Freshwater Ecosystems 19:322â343. Darling, J. A., and A. R. Mahon. 2011. From molecules to management: Adopting DNA-based methods for monitoring biological invasions in aquatic environments. Environmental Research doi:10.1016/j.envres.2011.02.001. Diamond, J. 1986. Overview: laboratory experiments, field experiments, and natural experiments. Pp 3â23 In: Community Ecology. J. Diamond, and T. J. Case, editors. New York: Harper and Row. Fofonoff, P. W., G. M. Ruiz, B. Steves, and J. Carlton. 2003. Ships or On Ships? Me- chanisms of transfer and invasion for non-native species to the coasts of North America. Pp. 152â182 In: Invasive Species: Vectors and Management Strategies. G. M. Ruiz, and J. T. Carlton, editors. Washington, DC: Island Press. Hines, A. H., G. M. Ruiz, and L. S. Godwin. 2000. Assessing the risk of non-indigenous species invasion in a high latitude ecosystem: ballast water treatment facility in Port Valdez, Alaska. Pp. 81â88 In: Marine Bioinvasions, Proceeding of the First Nation- al Conference, MIT Sea Grant College Program, Cambridge. J. Pederson, editor. Cambridge, MA: Massachusetts Institute of Technology. Interagency Ocean Policy Task Force. 2010. Final recommendations of the Interagency Ocean Policy Task Force. The White House Council on Environmental Quality, Washington, DC. Lonsdale, W. M. 1999. Global patterns of plant invasions and the concept of invasibili- ty. Ecology 80:1522â1536. MacDougall, A. S., and S. D. Wilson. 2007. Herbivory limits recruitment in an old-field seed addition experiment. Ecology 88:1105â1111. Mardis, R. E. 2008a. Next-Generation DNA Sequencing Methods. Annual Review of Genomics and Human Genetics 9:387â402. Mardis, R. E. 2008b. The impact of next-generation sequencing technology on genetics. Trends in Genetics 24(3):133â141. Miller, A. W., T. Huber T, M. S. Minton, and G. M. Ruiz. 2010. Status and trends of ballast water management in the United States: Third Biennial Report to the United States Coast Guard. Edgewater, MD: National Ballast Information Clearinghouse, Smithsonian Environmental Research Center (SERC). Minton, M., E. Verling, A. Miller, and G. Ruiz. 2005. Reducing propagule supply and
136Â Â PropaguleÂ PressureÂ andÂ InvasionÂ RiskÂ inÂ BallastÂ WaterÂ Â coastal invasions via ships: effects of emerging strategies. Frontiers in Ecology and the Environment 3:304â308. Reusser, D. 2010. Biogeography of Nonindigenous Species: From Description to Pre- diction. Corvallis, OR: Oregon State University. Ricklefs, R. E., and G. Miller. 1999. Ecology. Fourth edition. San Francisco, CA: W. H. Freeman. Ruiz, G. M., and J. T. Carlton. 2003. Invasion vectors: a conceptual framework for management. Pp. 459â504 In: Invasive Species: Vectors and Management Strate- gies. G. M. Ruiz, and J. T. Carlton, editors. Washington, DC: Island Press. Ruiz, G. M., and C. L. Hewitt. 2002. Toward understanding patters of coastal marine invasions: A prospectus. Pp. 529â547 In: Invasive aquatic species of Europe. E. Leppakoski, S. Olenin, and S. Gollasch, editors. Dordrecht: Kluwer Academic Pub- lishers. Ruiz, G. M., and D. F. Reid (editors). 2007. Current state of understanding about the effectiveness of ballast water exchange (BWE) in reducing aquatic nonindigenous species (ANS) introductions to the Great Lakes Basin and Chesapeake Bay, USA: Synthesis and analysis of existing information. NOAA Technical Memorandum GLERL-142, National Ocean and Atmospheric Administration, Ann Arbor. Ruiz, G. M., P. Fofonoff, J. T. Carlton, M. J. Wonham, and A. H. Hines. 2000. Inva- sions of Coastal Marine Communities in North America: Apparent Patterns, Proc- esses, and Biases. Ann. Rev. Ecol. Syst. 31:481â531. Ruiz, G. M., P. W. Fofonoff, B. Steves, S. F. Foss, and S. N. Shiba. 2011. Marine inva- sion history and vector analysis of California: A hotspot for western North America. Diversity and Distributions. 17: 362â273. Smith, L. D., M. Wonham, L. McCann, G. Ruiz, A. Hines, and J. Carlton. 1999. Inva- sion pressure to a ballast-flooded estuary and an assessment of inoculant survival. Biological Invasions 1:67â87. Solow, A., and C. Costello. 2004. Estimating the rate of species introductions from the discovery record. Ecology 85:1822â1825. Stohlgren, T. J., D. Barnett, and J. Kartesz. 2003. The rich get richer: pattern so plant invasions in the United States. Frontiers in Ecology and Environment 1:11â14. Verling, E., G. M. Ruiz, L. D. Smith, B. Galil, A. W. Miller, and K. Murphy. 2005. Supply-side invasion ecology: characterizing propagule pressure in coastal ecosys- tems. Roy. Soc London B 272:1249â1256. Vila, M., A. S. D. Siamantziour, G. Brundu, I. Camarda, P. Lambdon, F. Medail, E. Mo- ragues, C. Sueh, A. Traveset, A. Y. Troumbis, and P. E. Hulme. 2008. Widespread resistance of Mediterranean island ecosystems to the establishment of three alien species. Diversity and Distributions 14:839â851.