Скачать 1.55 Mb.
The Action Plan seeks to significantly reduce the size of the Gulf of Mexico hypoxic zone by the year 2015, primarily through reductions in nitrogen (N) loadings from the MARB to the NGOM. Increases in N loads have clearly been occurring throughout the past decades, and there is ample evidence to conclude that N from the MARB is a driving force in determining, at least in part, the timing, severity and extent of the hypoxic zone. Since the mid-90s, N loadings from the MARB have decreased, although they are still much elevated over historic levels. Total phosphorus loadings, however, have not changed greatly during this period (Battaglin, 2006; Turner et al., in press; Section 2.1.9 of this report). This trend in nutrient loadings has led to reduced (albeit still very high by “Redfield” standards) N:P ratios. This evidence suggests that P is an additional nutrient of concern, in terms of input reductions. As conveyed in previous sections of this report, a number of investigators (Sylvan et al., 2006; Dagg et al., in press) have concluded that P is limiting primary production during key periods of high productivity and in zones of high biomass accumulation in the NGOM adjacent to hypoxic waters. Therefore, the role of P in the onset, extent, and duration of the hypoxic zone is worthy of additional consideration.
Many factors influence the cycling and ultimate fate of both N and P. As both play a significant role in driving primary production within the NGOM (and perhaps, in conjunction with Si, in the composition of the primary producers and the likely fate of produced organic carbon), it is logical to consider the potential for removal of either or both elements as a means to reducing hypoxia. The 2001 Action Plan focuses on N reductions but does not preclude either P reduction or dual removal strategies. For example, the most recent report of the Mississippi River/Gulf of Mexico Watershed Nutrient Task Force’s Management Action Review Team (MART, 2006a) concludes that most load reduction projects developed under the Clean Water Act Section 319 program have targeted both N and P for reduction. Indeed, Howarth et al. (2005) noted that some N control practices utilized in the U.S. effectively remove P as well, although the reverse is not always the case. However, not all control practices will be effective as a dual nutrient removal strategy; see specific discussion on this topic in Section 4.5.10.
Restoration plans that focus on N alone may not rapidly improve the situation in the MARB where many streams and river segments are degraded by excess P concentrations (Action Plan, MR/GMWNTF, 2001). Given recent discoveries concerning the importance of P in production of organic carbon within significant portions of the NGOM, focusing on N reduction alone may be insufficient to provide the desired reduction in the hypoxic zone. However, some plans being undertaken to reduce non-point sources of N [forested buffers, 319 programs, and others (see Section 4.4.2, for example)] will also lead to P reductions, as well. Reductions in P alone will alleviate some of the water quality issues facing freshwater regions of the MARB but are not likely, given our current state of understanding, to significantly address the over-enrichment of the NGOM. Therefore, greater emphasis on a dual nutrient removal strategy is warranted, a conclusion that has been reached in other instances (e.g., National Research Council, 2000; Boesch, 2002; Howarth and Marino, 2006).
Further work is necessary to examine how effectively current reduction strategies target both elements. There may be areas where shifts in removal techniques could improve P reduction. In addition, there is still much to be learned about the response of autotrophic and microbial communities to shifts in nutrient loading and ratios. A better understanding of how these communities have responded to the current loadings and predictions of how they will continue to adapt to nutrient reductions will greatly improve predictions of the likely response in the extent and duration of hypoxia to nutrient reductions in the future.
There are several types of modeling efforts working toward a better understanding of factors influencing the extent and duration of the Gulf of Mexico hypoxic zone. These vary from the simple to the complex and are based on empirically observed relationships, on mechanistic understanding, or some combination of both.
Empirical models are widely used in the aquatic sciences to establish relationships between variables, with the most well known being the correlation between spring P loading in lakes and summer chlorophyll concentrations (Vollenweider, 1976). This work has been widely used in a management context to justify reductions in anthropogenic phosphorus loading to lakes and to set goals for reductions for particular lakes. Nixon et al. (1996) developed a similar correlation between annual loading of DIN and rates of primary productivity for marine ecosystems. While establishment of empirical models has greatly enhanced understanding of the structure and functioning of aquatic ecosystems (Peters, 1986), the standard criticism of this approach is that correlation does not imply causation. Although correlations between variables exist, they do not explain why variables are correlated or the mechanisms of the relationship. They do, however, provide some very useful predictive capability. In addition, when ecosystem production is greatly different from that predicted, controls on productivity other than nutrients may be dominating, such as light limitation or limitation from rapid flushing (Howarth et al., 2006a).
Some new forecast modeling work has been completed since the Integrated Assessment. Turner et al. (2006) developed simple linear and multiple regression models to examine hypoxia in the NGOM. Empirical models require important decisions regarding the choice of variables and of the time scales of model operation. Turner et al. (2006) tested many different nutrient loading lag times and concluded that the best relationship was obtained two months (May) prior to the maximum observed extent of hypoxia (July), with significant correlations for nitrate+nitrite, total nitrogen (TN), ortho-P and total phosphorus (TP) (r2 values of 0.50, 0.27, 0.54, and 0.60, respectively). A multiple regression analysis was also developed incorporating nutrient load and a new variable “Year” to account for the increase in carbon in surface sediments after the 1970s causing significantly more sediment oxygen demand. A lag of two months of nutrient loading was, again, the most significant variable to describe hypoxic area with r2 values of 0.82, 0.80, 0.69, and 0.64 obtained with nitrate+nitrite, TN, ortho-P and TP, respectively. Turner et al. (2006) then used the nitrate+nitrite model to extrapolate beyond the data range used to construct their models to predict hypoxic area prior to available measurements. When the hindcasted values became negative, they were plotted as zero values. In general, it is considered incorrect to extrapolate model results in this manner beyond the range of the data supporting the model, as other mechanisms and relationships may exist that may not be included in the regression analysis. Further, the SAB Panel believes that the addition of the variable “Year” in the multiple regression analysis is inappropriate as the addition of one more year will cause prediction of a positive increase in hypoxia with time.
Among models that address Gulf of Mexico hypoxia and include some consideration of processes and mechanisms, that of Scavia et al. (2003) is one of the simplest. Their model uses a relationship between the nitrogen loading from the MARB and the decay of oxygen “downstream” (i.e., in the NGOM - within the plume and the nearshore reaches to the west of the Mississippi and Atchafalaya River outflows). When used in a forecast mode, this model is able to only explain approximately 45-55% of the variability in hypoxic length and area. This model explicitly addressed uncertainty in prediction. The SAB Panel found this approach to be very useful. Recently, in combination with a watershed model, the model of Scavia et al. (2003) has been used to address how climatic variability and change may affect Gulf hypoxia (Donner and Scavia, 2007). A similar model has also been applied very successfully to understand hypoxia and anoxia in Chesapeake Bay (Scavia et al., 2006). The Scavia et al. (2003) model focused on N loading and did not consider P. Consideration of P would seem to be a timely addition to the model, and a manuscript including P recently was accepted for publication by Scavia and Donnelly (Scavia and Donnelly, in press). This model approach, and the modeling efforts of Bierman and colleagues and Justic and colleagues (see below) all provide reasonably consistent guidance and suggest similar levels of N reduction that might be required to reduce the extent of the hypoxic zone.
Other process-based models are more complex and attempt to model both physical and biological controls occurring in the hypoxic region. Examples include those of Bierman et al. (1994), Justic et al. (1996, 2002), and Green et al. (2006b). The Bierman et al. (1994) model is the most complex of these approaches and simulates the steady-state summertime conditions for the hypoxic area using three-dimensional modeling of the physics as well as interactions between food web processes, nutrients, and oxygen. The model of Justic et al. (1996, 2002) simulates oxygen dynamics at one location within the hypoxic zone using a simple model that has two vertical layers and meteorological conditions and nitrogen loads as drivers. The Green et al. (2006b) surface mixed layer model is based on food web dynamics and relatively simple two-dimensional physics (no vertical dimensionality) of the Mississippi River plume. This model predicts, among other things, the relationship between carbon sources and bottom-water oxygen depletion; the model does not include changes to either N or P inputs or dynamics. None of these more complex models explicitly presented analysis of uncertainty or sensitivity analysis of potential biasing terms. As with the Scavia et al. (2003) model, Bierman et al. (1994) and Justic et al. (1996, 2002) do not consider P loads or dynamics.
It should be pointed out that complex water quality models that could be very useful in the NGOM have been developed and used in other environmentally stressed regions like the Chesapeake Bay system (Cerco and Cole, 1993), Long Island Sound (St. John et al., 2007), the New York/New Jersey Harbor/New York Bight complex (Landeck-Miller and St. John, 2006), and the Massachusetts/Cape Cod Bays system (Besitkepe et al. 2003). These models include a coupling to three-dimensional and time-dependent hydrodynamics, a water column eutrophication submodel and a sediment diagenesis/nutrient flux submodel. The water column eutrophication submodel includes state-variables for three functional phytoplankton groups; dissolved inorganic nutrients (ammonium, nitrate+nitrite, ortho-phosphate, and silica);, and labile and refractory forms of dissolved and particulate organic nitrogen and phosphorus, biogenic silica; labile and refractory forms of particulate and dissolved organic carbon; and dissolved oxygen. The sediment nutrient flux submodel includes state-variables for labile, refractory, and inert organic carbon, nitrogen, and phosphorus, as well as biogenic silica. Inorganic substances tracked include ammonium, nitrate+nitrite, ortho-phosphate, silica, sulfide, and methane. Processes tracked in the sediment flux model include: organic matter deposition; sediment diagenesis; burial; the flux of inorganic nutrients between the water column and the sediment bed; and the generation of sediment oxygen demand (SOD).
There is an inherent trade off between model simplicity (where many potentially important factors are not considered) and complexity (where many coefficients and a great amount of data are required). More complex models may have value to help devise effective management strategies, especially if N reductions alone will not be sufficient to control hypoxia and if the more complex models can reasonably capture the importance of P. However, with complexity comes greater numbers of estimated parameters and the uncertainty associated with them. Hence this type of model may not improve forecasting capabilities dramatically. The development of more complex models is likely to prove extremely valuable for understanding the physical factors controlling water and carbon (C) transport, the dynamics of nutrient interactions with primary producers, and the recycling and loss of C and nutrients from the system. There is also great value in refining and further developing simple models, which may, in the end, prove most valuable for making management decisions. Scavia et al. (2004) explicitly compared the models of Scavia et al. (2003), Biermann et al. (1994), and Justic et al. (1996, 2002) for use in managing Gulf of Mexico hypoxia and showed that all three models gave broadly consistent guidance.
The physics of the NGOM region is complex, and there is clear value in developing more complex models of physical processes for this region. Improved three-dimensional models with finer grid structure than present models would have many uses. These uses include assisting the interpretation of monitoring data and serving as platforms upon which improved models of biogeochemistry and ecological response could be built. However, the level of complexity in the biogeochemistry and ecology need not match the complexity of the physical models (Hetland and DiMarco, 2007). Complex physical models could be very valuable in constructing simple box mass-balance accounting models for C, N, P, Si, and O, for example. The importance of developing such budget-based models is discussed further below.
In addition to statistical and simulation models, another modeling format that should be considered involves construction and evaluation of material budgets or mass balance models. These are basically quantitative input-output budgets with additional complexity added by consideration of internal processes of production, recycling and loss. These relatively simple budgets provide a quantitative mass balance framework to test the understanding of how the systems work. These budgets should be developed on a seasonal basis (e.g., summer hypoxic season) and evaluated for distinctive areas (e.g., Mississippi River Plume). These budgets are largely based on empirical observations and are not simulated through time, although data used in a budget analysis are needed in simulation models for both calibration and verification. As an example, an oxygen budget (Equation 1) would involve DO inputs/outputs from air-sea diffusion, horizontal advective/dispersive transport, and vertical transport between euphotic and sub-pycnocline zones. In addition, DO is added through daytime photosynthesis and lost through water column and sediment respiration. Evaluation of these pathways indicates especially important processes, and imbalances in the budget point to areas where understanding or measurements are inadequate. We suggest that conceptual mass balance models also be used to provide a checklist of needed measurements for future NGOM hypoxia research/monitoring.
Other general points regarding modeling efforts are summarized in Section 3.4 of this report. An important conclusion for both models of the response of the NGOM to nutrient inputs and watershed models generating estimates of nutrient loads is that a diverse ensemble of models is needed, including both relatively simple and more complex ones. No one best approach to modeling can be identified, and management of Gulf hypoxia is best served by having multiple models with multiple outputs. The SAB Panel suggests that modeling efforts, ranging from the simple to complex, be conducted in parallel wherein there is the opportunity for cross-testing of results among model formats. When predictions tend to agree, managers can have more confidence in deciding upon courses of action. When models do not agree, dissecting the reasons for divergence can lead to better understanding and, ultimately, better management.
|Board of Elections – Advisory Opinion 1398 – 1399||Panel Report on the draft Replacement London Plan|
|Draft: August 2July xxanuary x19, 20065 abstract [For presentation at the Advisory Group on apec financial System Capacity Building meeting at Seattle, February 28, 2007.]||Draft – do not quote or distribute|
|Ranch hand advisory committee||Medical Devices Advisory Committee|
|Positions, training & advisory roles||Veterinary medicine advisory committee|
|Advisory committee on immunization practices||National Vaccine Advisory Committee (nvac)|