|
|
Development of an Integrated, Adaptive
Management Protocol for American Black Ducks
Scientific management for American black ducks (Anas rubripes) has been hampered by a lack of understanding regarding factors affecting the dynamics of black duck populations. This has resulted in a lack of agreement among stakeholders as to the potential for arresting the decline of black duck stocks through management intervention, especially through harvest regulations. Adaptive resource management (Walters 1986; ARM) is a process for making a decision, in the face of uncertainty, that results in an optimal value for some resource objective, while reducing that uncertainty through time. More narrowly defined, adaptive harvest management (AHM) (Johnson et al. 1993, 1997; Williams and Johnson 1995) seeks to reach a long term harvest objective, with the decisions being annual harvest regulations. In the discussion below we emphasize AMH, that is, the objective and the management decisions revolving around harvest. However, a broader discussion of ARM which explicitly includes both non-harvest objectives, as well as management decisions other than harvest regulations for reaching those objectives, certainly may be appropriate, as discussed below.
In a previous contract (Developing Models of Black Duck Populations in North America) 4 major alternative hypotheses potentially explaining variations in populations of American black ducks were identified: 1) Breeding ground habitat; 2) wintering ground habitat; 3) harvest , and 4) interactions with Mallards (Anas platyrhynchos). These hypotheses were used in conjunction with a life-cycle model for black ducks, summarizing each of the above hypotheses in terms of model parameters. Historical data were then used to estimate model parameters, and to develop a preliminary set of models, for further consideration as potentially useful in an adaptive management framework. The purpose of this proposed contract is to build upon that previous work, and to develop and evaluate a prototype for adaptive harvest management for American black ducks.
Objectives
1. Development of model stratified by 2-4 breeding stocks of black ducks; at a minimum these must capture essential differences between western and eastern black duck stocks, with respect to habitat conditions and the influence of mallards. Such a model almost surely will require explicit linkage to the existing model for eastern mallards.
2. Development of an appropriate objective function, possibly including explicit linkage between a black duck objective and a mallard objective.
3. Inclusion of relevant administrative units (see below).
4. Identification of key system states requiring monitoring for feedback into adaptive decision making, and the spatial and temporal scales at which monitoring is needed.
These advances must occur in tandem with the resolution of several policy issues:
5. Identification and clarification of goals and objectives of an adaptive management protocol. These might include (but are not limited to: a) sustainable black duck harvest; b) sustainable duck (i.e., mallard or black duck) harvest; c) incorporation of population (North American Waterfowl Management Plan [NAWMP]) objectives; d) incorporation of habitat objectives.
6. Identification of relevant units by which decisions (e.g., harvest) can or will be made. We assume that, at a minimum, these will involve >1 administrative units (e.g., regions, flyways) in each nation (Canada, U.S.).
Tasks
The following are specific tasks or steps that need to be taken to accomplish these objectives, and some preliminary results and/or thinking in each area.
1. Using the analyses completed under the previous work order, develop a model set for use in adaptive optimization.
AHM allows for the explicit incorporation of this uncertainty through the delineation of alternative models, which are thought to encompass the range of uncertainty. We have identified 8 alternative models that represent different combinations (presence/ absence) of the influence of the principal factors we hypothesize may be influencing black duck population dynamics: habitat conditions, mallard populations, and harvest rates. In the accompanying addendum to the Final Report for the black duck modeling project, we present some preliminary evidence for differential weighting among these alternatives, and perhaps for the exclusion of some alternatives, resulting in a smaller model set. We suggest that this (or a somewhat reduced) model set could very reasonably be used to initiate an AHM process, perhaps even with equal model weights. We recommend that further emphasis be devoted to initiating a very simple (e.g., single population, single decision variable) AHM model, and exploring the consequences of finer vs. coarser resolution in objective, decision space, and state space, and of alternative model weightings (e.g., equal vs. empirically derived) in reaching optimal decisions. We suggest that these steps will, in the earliest stages, be most useful as heuristic devices for familiarizing stakeholders with AHM, and for mapping our a more detailed, eventual implementation of AHM/ARM for American black ducks.
2. Working with the Black duck Joint Venture (BDJV) and other stakeholders, develop and incorporate objective functions.
Formulation of a suitable objective function is absolutely critical to the ARM process; it is also a matter for resolution among managers and policy makers, and not a scientific issue, per se. However, scientific resource management principles and theory do provide guidance for and AHM objective, in that the concept of "sustainability", is at the core of most natural resource management. We define a "sustainable" harvest one which does not, over some very long time horizon, result in a loss of a stocks ability to replenish itself. Presumably, the harvest of this resource (e.g., the annual take of black ducks) has inherent economic and other value, so it seems reasonable to formulate the objective in words as "the maximum harvest, over the long term, that can be taken while not diminishing the stock." In fact, harvest theory (e.g., Caughley 1977) permits us to express our harvest objective as
where
is size of the
harvest in year t. Because of the infinite time horizon, population
persistence is necessary, in order to provide harvest for future years;
short-term harvest at the expense of these later harvest opportunities is
precluded. However, it is true that this maximum could (at least in theory) be
achieved while still allowing the population to decline below levels that
would be acceptable to stakeholders. For this reason, earlier AHM objectives
were constrained so as to penalize the objective if mallard populations fall
below North American Waterfowl Management Plan (USFWS and CWS 1986) population
objectives (this constraint has since been removed).
A harvest objective as
above (possibly constrained by NAWMP goals) would be straightforward to
implement as part of AHM. However, several additional considerations may
result in further modification of the objective. First, is the issue of
harvest allocation. In theory, it would be possible to attain an
objective such (1) with very different distribution among countries, states or
provinces within each country, or other administrative/ political units. For
instance, the same maximum, long-term harvest (
) could be obtained
with 50% of 
each occurring in Canada and the U.S., or with 100% in 1
country or the other. Given the international nature of this shared resource,
there would likely be great dissatisfaction if 1 of the latter results were to
occur over any significant time horizon, or perhaps even for any year. Of
course, such a result is very unlikely to occur, given the migratory nature of
the species, unless hunting seasons were closed in either country.
Nevertheless, it might be desirable, in order to secure stakeholder support
for AHM in both countries, to impose a "parity constraint" on harvest
allocation, so that the objective would be penalized if the distribution of
harvest were to deviate substantially from recent historical distributions
(approximately equal between countries). Similar constraints could be imposed
at finer levels of resolution, but might prove either impractical, highly
suboptimal (from the overall objective standpoint), or both. Before any type
of harvest allocation constraint is imposed on a harvest objective, we would
strongly advise thorough investigation of the implications and costs of such
an approach.
An additional issue, tending to complicate the objective, is that of multiple species under management. For black ducks, mallards are obviously an important consideration, because of the potential that the state of mallard populations could directly influence the anticipated harvest of black ducks. Under various of our alternative models, it would certainly be important to model mallard populations, in order to incorporate this influence into the models predicting the harvest of black ducks under different black duck regulatory options. However, it seems a fair question to ask whether black ducks and mallards should be considered jointly in the objective function as well. This has interesting and important implications that should be explored through simulation trials. For instance, if mallards and black ducks were equally valued in the harvest, a composite objective might be expressed as
where
are the annual harvests for black ducks and
mallards, respectively. If on the other hand black ducks were deemed twice as
valuable (e.g., because of a tradition for hunting black ducks) then the
objective might be expressed as
For this objective to make sense, the decision problem would also have to be expanded to include harvest regulations for mallards as well as black ducks. Again, the consequences of including multiple species in a common objective, together with the effects of differing weighting factors, should be thoroughly explored.
Finally, although we have emphasized a harvest objective, the objective could be expanded to include non-harvest population objectives, or even to explicitly include a value for habitat (e.g., for nongame waterbirds). Difficulties could arise, however, in deciding how to value these objective components (e.g., what is the equivalence between numbers of harvested ducks, and acreages of wetland habitat?), and thus what objective weights to use. Nevertheless, non-harvest and non-harvest objectives may be worth considering.
3. Working with the Black duck Joint Venture (BDJV) and other stakeholders, develop alternatives.
Under AHM for black ducks, harvest rates are at least partially controllable by means of harvest regulations set in the U.S. and Canada. However, as suggested, the decision is not the rate of harvest (which is a partially random outcome), but rather some set of fairly complicated regulatory options. In practice, these can, and probably should, be grouped into discrete sets of regulations thought to have similar impact on harvest rates, and describable generically as "restrictive" (e.g., relatively short seasons and low bag limits), "moderate", and "liberal" (longer seasons, high bag limits), basically the approach taken with mallard AHM. In order to translate these regulatory options into harvest rates, an empirical relationship must be established that predicts an expected (average) harvest rate for each; inclusion of "partial controllability" also requires the modeling of a statistical distribution about that expected value. Historical data on harvest regulations and subsequent harvest rates (estimated through band recoveries) could be used to construct these relationships (e.g., see Johnson et al. 1997).
Again, this relatively simply approach is complicated by the fact that at a minimum, 2 countries currently set harvest regulations, mainly independent of one another. In addition, withing each country regulations are set on a flyway or regional basis, and actually implement at a state/ province or even finer spatial scale. Any AHM process developed for black ducks must take into account this complexity, and negotiate a compromise between overly coarse- or fine- resolution.
In addition, as alluded to above under the objective, it may be desirable and even necessary to link the harvest decision process for black ducks, with those of other species (notably mallards). For instance, a joint black duck -mallard objective (e.g., equ. 2-3 above) would require joint consideration of harvest regulations for each species (e.g., restrictive for both species, liberal for both, restrictive for one and liberal for the other). Under the simple scheme of 3 harvest levels for each species, this would result in 9 possible joint harvest strategies.
Earlier we also alluded to the inclusion of non-harvest decisions, such as habitat management, into an ARM decision space. If habitat decisions were included, there are many issues that would have to be dealt with, such as (1) the spatial scale at which to describe (or summarize) habitat decision making, and (2) the different temporal scale at which habitat decisions tend to occur, compared to harvest decisions (longer term, vs. annual). We believe that an eventual, integrated harvest-habitat approach will prove valuable for waterfowl management in general, but that questions such as those posed above create significant logistical and technical challenges. As with the development of an objective, we recommend beginning with a simple decision space (e.g., an overall low, medium, or high rate of harvest), and then exploring the implications of geographic, multiple species, and harvest-habitat decision spaces through simulation.
4. Definition of state space and resolution of monitoring efforts
By "state space" we mean the array of state variables (principally, population size and habitat conditions) and the spatial and temporal resolution at which these are considered measured and represented in models. In order to make rationale decisions about how to manage the resource, we must understand what is the current system state (e.g., numbers of breeding black ducks, acres of breeding habitat). Given knowledge (or an estimate) of this state, a decision (e.g., a liberal harvest) and a dynamic model such as the one we have developed, one can then make a prediction as to the future state of that system (e.g., next year's numbers of breeding black ducks and acres of breeding habitat), and ultimately, of the expected return (e.g., long term harvest) for a series of decisions through time.
Our modeling effort should make clear that certain state variables,
at a minimum, must be available, in order to proceed with AHM. These are:
In our modeling effort, we described each of these using range-wide
estimates or indices. This was due both to the relative lack of long-term,
finer resolution data, as well as to the need to simplify modeling at this
stage of model development. However, as with the objective function and
decision space, finer resolution may be needed to adequately represent
the dynamic behavior of black duck populations, and may be possible as
more accurate information becomes available (e.g., from continued surveys
of black ducks and mallards in eastern Canada and the U.S. ). Differences
in black duck densities and the influence of mallards in western vs. eastern
Canada suggest a possible need for modeling 2 breeding stocks of black
ducks, with distinct responses to mallards, habitat conditions, and harvest.
In addition, it may be desirable to model distinct wintering (e.g., coastal
vs. interior) populations of black ducks, to the extent that reliable surveys
of these populations are feasible. As with the objective function and decision
space, investigations should be conducted into the extent to which spatial
or other model stratification produces better decisions, and whether the
gain in objective return (i.e., long term harvest) is worth the cost of
model complexity (and the concomitant increase of cost and loss of reliability,
for states that are measured at too fine a scale).
Obviously, population, habitat, and other state variables can and will
be collected at some arbitrary levels of spatial and temporal resolution,
and then aggregated as needed for model states. The process cannot work
in the other direction, i.e., if the decision is made to collect survey
data at an aggregated spatial scale or at >1 year intervals, it will not
be possible to re-create dis-aggregated data for later modeling use. We
therefore suggest that modeling and monitoring efforts work in concert
to achieve the best tradeoff.
Other Data Needs
Although not explicitly part of the state space of our models, our estimates of annual reproduction rates, harvest rates, and annual and seasonal (i.e., winter) survival rates were fundamental to model construction. Continued monitoring of these parameters is needed, both to enable revised projections of population growth rates, and to revised estimates of the functional relationships between demographic parameters and extrinsic and intrinsic variables. Again, our modeling effort was forced by the paucity of data to estimate these parameters and relationships on a range-wide basis. Finer scale information, e.g., corresponding to western vs. eastern black duck populations, is needed to establish whether significant spatial heterogeneity in functional relationships occurs, and to adjust models accordingly. These will require 1) continued banding efforts on population segments, to estimate harvest and survival rates; 2) harvest derivation analyses, to link harvest age ratios to breeding population segments, and 3) continued monitoring of black duck abundance, mallard abundance, and breeding habitats range wide.
5. Evaluate the potential impacts of model resolution and management
scale on optimal decision making, with respect to gain in objective (e.g.,
harvest) value, vs. costs (e.g., due to increased data requirements or
administrative burdens).
6. Depending on the results of 1-5, develop a working adaptive
management protocol for black ducks; possibly this effort will culminate
in a joint black duck - mallard ARM protocol.
7. At each step of the process, maintain close contact with the
BDJV and other stakeholders, communicating progress and soliciting feedback.
Participate (lead?) workshops to educate stakeholders as to the merits
of ARM and the technical issues involved.
Communication here means a 2-way exchange of ideas, not a "selling" of AHM. This exchange must include a component of information transfer and education, so that participants are aware of what AHM really is (not what they think it might be, or have heard that it is), but also so that those responsible for the technical aspects of AHM are aware of, and sensitive to, legitimate concerns about the AHM process.
Revised Timetable (June 1999- June 2003)
Project initiation June 1999
Preliminary model development June 2000
Interim report June 2000
Incorporate revised objective function Dec. 2000
Evaluate model resolution/ management scale Dec. 2001
Interim report June 2001
Development of working protocol Sept 2002
Interim report June 2002
Evaluation of protocol and Draft Report Mar. 2003
Final Report June 2003
This page last updated on 15 June, 2000.
Send e-mail to the webmaster