By Dr. Steven P. Courtney
Sustainable Ecosystems Institute
Summary
1. The scientific analyses are sufficiently powerful and sensitive to be useful in evaluating the proposed land exchange.
2. Modeling of the Murrelet population demonstrates that the effect of the proposed management will depend critically on the current vital rates of the population.
3. Survey data from the California population represent the best data for any portion of the species, and are consistent with a stable population.
4. The Humboldt Redwoods State Park contains a large amount of suitable habitat. In the long-term this, and the Headwaters Complex, will constitute large blocks of suitable habitat, which will support continued survival and recovery of the species.
5. The short-term effect of the land exchange will depend on the proportion of the population that will be disturbed by logging. Current information does not allow a good estimate of this quantity. Fieldwork in 1997 will clarify this issue. A current conservative estimate is that the State Park and the Headwaters Complex will preserve a majority of the existing population.
6. On the best available scientific information, the proposed land exchange is a feasible conservation measure for the Marbled Murrelet
Introduction
The proposed land exchange between the California and federal governments, and the Pacific Lumber Company is currently under discussion. The effects of the proposed agreement on the Marbled Murrelet are the subject of ongoing analyses. In this document I will first summarize the main findings of the studies to date. I will then evaluate the available evidence, and show how the results can predict the outcome of the proposed land exchange
Two main analyses are underway, together with several subsidiary studies. There will also be fieldwork in 1997 to establish key information. The first major analysis, carried out by Resit Akcakaya and Lev Ginzburg of Applied Biomathematics (NY) focuses on quantitative predictions of population trends under an array of assumptions about the Murrelet population. The goal of this analysis is to determine which factors will have the most impact on future status of the population. To the extent possible, this analysis will also determine the likely consequences for the Marbled Murrelet of different levels of future harvest on Pacific lumber lands.
The second major analysis is carried out by CJ Ralph and Sherri Miller and others at the Redwoods Sciences Laboratory (USDA Forest Service). The major goal of this analysis is to determine the amount of Marbled Murrelet habitat on Pacific Lumber Company and adjacent State Park lands. Ultimately, this analysis will yield an estimate of the proportion of habitat that will be lost locally under different harvest scenarios. Data collected at RSL over the past several years also includes estimates of population numbers for the Marbled Murrelet population, which will allow estimates of the current rate of decline.
Results
Quantitative analysis of risk
The report by Akcakaya (1997) : Ecological Risk Analysis for the Marbled Murrelet: The Sensitivity of Viability to the Parameters of the Zone 4 Metapopulation Model, sets out the results obtained thus far by the Applied Biomathematics team. It should be emphasized that this is the first report of several, and concerns an intermediate level of scale. Future analyses will consider smaller scales (restricted to the bioregion in the vicinity of Humbodlt Redwoods State Park, and Pacific Lumber lands), and larger scales (three state listed region).
Akcakaya's analysis essentially shows the risk of particular declines associated with different parameter values in a model. The majority of results are of the form shown in Figure 1. Here, there is a high risk (close to certainty) of a small decline in population numbers, and a lower probability of a more substantial decline. This is intuitive, in that natural populations fluctuate in abundance. However there are often limits set to such fluctuations, such that there is a smaller risk of a major decline. Figure 1 also shows the results under different logging scenarios. The horizontal bars indicate the point at which there is the maximal difference in risk between logging and no-logging scenarios. This is one way of evaluating the effects of logging. An alternative approach would be to determine the change in risk of a particular level of decline (e.g. 80%).
The major results can be characterized as follows:
1. The models successfully detect the effects of different factors, including logging.
2. The models are most sensitive to the assumptions made about vital rates (mortality, fecundity)
3. The model results are also affected by the form of density-dependence in the population
4. Other assumptions either have smaller effects or no discernible effects on model results
5. Logging is seen to increase the risk of smaller declines in scenarios where the population is otherwise in good condition. When the population is assumed to be declining rapidly for other reasons, logging does not substantially increase the rate of decline.
6. The effect of logging on model results is determined by the amount of habitat to be lost. Higher harvest levels increase the risk of small declines.
It is important that Applied Biomathematics 's models successfully detect the effects of logging (at different levels) as well as other factors. This result establishes that the models are at the right scale, and are sensitive enough to detect the effects of management. Decision-makers therefore can approach the results with confidence that the models are appropriate and applicable.
Of all the parameters used in the models, assumptions concerning vital rates have the greatest effect. This is intuitively obvious. If we assume that the mortality and birth rates are such that the population is decreasing at 7% per year (Beissinger 1995), then it will inevitably decline in the model. If, on the other hand, we assume that the population is currently stable (birth rate equals death rate, no migration), there is a much smaller risk that the population will decline. This result remains true even under substantial levels of logging on Pacific Lumber lands. The model therefore focuses attention on a critical aspect of the biology of the bird, and at the assumptions made by biologists studying it. If we assume pessimistic values for vital rates, the population in Zone 4 is essentially doomed, irrespective of management (although forest harvest might accelerate the rate of decline). If on the other hand we assume that the population is now stabilized, and more optimistic values for vital rates, then the population is in good shape, and able to withstand some loss of habitat. The model therefore establishes that assumptions regarding vital rates are key to understanding the fate of the population. It is now important to establish whether the population is indeed currently declining, and which assumptions on vital rates are most realistic (see discussion).
The models are also quite sensitive to the effects of density dependence. If Marbled Murrelet populations are assumed to normally increase when below carrying capacity (Beverton-Holt models), then they have a greatly reduced risk of decline. However most of the models developed by Applied Biomathematics do not make this assumption: instead they assume that density-dependence only acts to set a limit to the population (ceiling models). This seems the more reasonable assumption (see discussion). Because the remaining analyses are made on the basis of this less optimistic, but more realistic assumption, we can essentially ignore the effects of density-dependence and focus on the remaining variables in the model.
Several other factors had detectable effects in the models: variance in vital rates; current carrying capacity; the risk of oil spills; the correlation between populations. These parameters are discussed below, in an attempt to evaluate which are the most realistic of the various assumptions. The models were not sensitive to initial conditions in the model, to Allee effects, or to population structure (number of populations).
Logging has discernible effects on the population under some conditions. There is an elevated risk of decline when the population is otherwise healthy. Greater logging impacts results in an increased risk of decline. If however the population is already declining, logging has only a small incremental, accelerating effect.
In summary the outcome of the model, and its predictions for decision-makers depend critically on two variables: vital rates, and the proportion of habitat that will be lost to logging. In the discussion I evaluate the current information on these issues.
Analysis of Marbled Murrelet population status and habitat
The results of the Redwood Science Lab team are summarized in the report " A preliminary evaluation of the impact of land management scenarios on the population of the Marbled Murrelet in southern Humboldt County, California'. The results in this report cover three main topics: an analysis of inland detection data throughout the region, to evaluate the use of detection data (Table 8); application of this approach to habitat data from Pacific Lumber Company lands, and the Humboldt Redwoods State Park (Tables 3 through 7); and data from offshore counts of Marbled Murrelets throughout northern California (Tables 1 & 2). The major results of these analyses can be characterized as follows:
1. There is no strong evidence for a trend in offshore counts. Overall the results suggest a stable population.
2. There are significant areas of potential Marbled Murrelet habitat on Humboldt State Park land (Table 3), in the Headwaters complex (Table 4) and elsewhere on Pacific Lumber Company lands (Table 5). The majority of potential habitat acreage is in the State Park.
3. The Headwaters complex area shows high occupancy levels at most stations, including both old-growth and residual stands (Table 7). The State Park shows lower occupancy levels in old-growth (Table 8). Coverage of the Park is far less complete than coverage on the Headwaters complex; there is essentially no coverage of residual stands in the Park.
4. The use of protocol survey data to estimate occupancy levels on this landscape appears well justified.
The data on offshore counts of Marbled Murrelets in northern California are important. They represent probably the best available information on population status anywhere in the range of the species. Comparable data are not available for any area in Oregon and Washington, where two or three years data are the most that are available. This data set therefore represents our best evidence on the current trends of the Marbled Murrelet population. Use of offshore survey counts is now recognized as the most effective method of monitoring Marbled Murrelet populations (Miller and Henson in press), and has been adopted by the Effectiveness Monitoring Group overseeing implementation of the President's Forest Plan.
Failure to detect a trend in population numbers could be due to two reasons: there could in reality be no trend; or the test and methods may be too weak to detect a trend that is actually present ('Type II error'). It is therefore essential to evaluate the strength of the results reported by the Redwood Sciences Team. Although these data are probably the best available for any part of the species' range they are not without fault: not all areas were surveyed in every year, and coverage was unequal. It is also difficult to evaluate the power of their regression analyses, which may be insufficient to detect all but the strongest trends. For instance the data in Table 2 show 11 of 22 regressions (2 positive, 9 negative) where the value of r exceeds 0.5 ; but in only 2 of these cases (1 positive, 1 negative) are the trends significant. It is plausible to argue that the low sample sizes (numbers of years) prevents all but the strongest correlations from being detected. One possible approach to dealing with this question is to reanalyze the data reported by the Redwood Sciences Team.
Table 1 shows an analysis of population changes between years across all sites, not as a correlation analysis (with its problems of small sample size) but as a test of change in numbers from one year to the next. Several points should be made from this analysis. Firstly, there is no pervasive pattern of either decrease or increase (on 16 occasions the population increased by 20% or more, on 19 occasions it decreased by the same amount, and in 11 cases there was no major change). Secondly, there is considerable variance from one year to the next: in 35 of 46 cases, the population changed by greater than 20%. Thirdly, there appears to be little if any correlation in the changes in numbers between adjacent sites. Note however that the data within a site (collected at 800 or 1400 m from shore) are well correlated together (34 of 43 cases show congruent changes). This argues that the survey methods are accurate and dependable.
In summary, this analysis supports the suggestion of the Redwood Sciences Team, that there is no ongoing trend in Marbled Murrelet numbers in northern California. This is consistent with a population that is relatively stable in numbers.
The second major result of the study is that there is a large amount of potential habitat in the Humboldt State Park. The acreage of potential habitat in the State Park is a critical variable, because this determines the proportion of the potential habitat that is at risk under various logging scenarios. The Redwood Sciences Teams results suggest that a majority of potential Marbled Murrelet habitat in the bioregion is already protected in the State Park. However this result should be carefully evaluated before it is used by decision-makers in assessing the proposed land exchange. The State Park lands have not been completely inventoried, and additional ground-truthing may be desirable.
The third major result of the study concerns the relative levels of occupancy of different stands in the Bioregion. Marbled Murrelets are known to nest in higher density in some preferred areas of habitat. Such high use sites may then harbor a disproportionately large part of the local population. Ralph and Miller suggest that the Headwaters Complex is such a high density area, and the Humboldt State Park has lower use. Marbled Murrelets are present at all old-growth sites in the Headwaters Complex, and in most residual stands there. On the State Park lands there is a similarly high level of detection in old-growth stands (but lower occupancy levels); residual stands are essentially unsurveyed on the Park.
This result is important because it suggests that the analysis of the effect of different logging scenarios should incorporate differences in habitat quality. It may be that the Humboldt Redwoods State Park contains a majority of the acreage of potential, available Marbled Murrelet habitat in the Bioregion, but that the proportion of the actual Murrelet population is not as large. At this point, the results of Ralph and Miller suggest that, although the State Park has the majority of acreage of habitat, it contains less acreage of actual occupied stands than does the Headwaters Complex (2,242 acres compared with 3,246). However these results should be strongly qualified: the Park has been insufficiently surveyed to date to determine occupancy rate with acceptable levels of certainty. It is also essential to include the other stands on Pacific Lumber lands before a final determination of the proportion of Marbled Murrelets that will be protected under different scenarios. Discussion
Although these analyses are ongoing, and can be expected to yield important new information in the coming weeks, there are already some major results that can be used to guide decision-makers. The results obtained have not indicated compelling reasons to reject the proposed land exchange as clearly unsuitable for the Marbled Murrelet. The proposed agreement would protect a significant proportion of the high quality habitat in the bioregion. In my opinion, the proposed land exchange would increase the level of protection for the Marbled Murrelet in northern California. I therefore support the continued negotiation between the parties, and continued analysis of the effects of the exchange, including fieldwork in 1997.
The results are also useful in addressing the critical question of 'jeopardy'. Even if the proposed land exchange benefits conservation of Murrelet habitat, it is not yet determined what constitutes 'enough' habitat to ensure continued survival and recovery. The results of the analyses to date support the following position:
Under certain assumptions of Marbled Murrelet demography, and the impact of logging on potential habitat, the proposed land exchange will not lead to a substantially increased risk of a major decline in the population over the next fifty years. Beyond fifty years, succession will lead to development of new 'recovery' habitat. Therefore, under these assumptions, the proposed land exchange and logging will not constitute jeopardy for the species.
The analysis of the current proposed exchange therefore hinges critically on the stated assumptions regarding the Marbled Murrelet population in northern California, and the probable impact of logging. If an optimistic view of these assumptions is taken, the proposed exchange is defensible as it stands. If a more pessimistic view is taken, then the proposed exchange may not be adequate for protection. These assumptions can be stated as follows:
It has normally been assumed that the Marbled Murrelet, throughout the listed region of California, Oregon, and Washington, is in a serious, ongoing decline. Beissinger (1995) has modeled the population using a similar model to that of the Applied Biomathematics team, but using estimates of mortality from other alcids, and estimates of fecundity from at-sea adult-juvenile ratios. He predicted that the species, throughout its range, was declining, and was likely to continue to decline. The Fish and Wildlife Service has followed this reasoning in designating Critical Habitat, including lands covered in the proposed land exchange. The Recovery Team and others have also adopted this approach in making recommendations or administrative decisions. Ralph et al (1995) also call attention to this evidence, but point out that the adult-juvenile methodology is not well developed, and that the data were collected in years that may be atypical (because of ocean conditions). Most recently participants in a workshop on survey methods for the species called attention to the many unresolved problems with the method, which is known to give spurious results in some circumstances (Courtney in Miller and Henson). Adult-juvenile ratios in the bioregion have typically been low, suggesting poor breeding success (results of Ralph and Miller).
- The Marbled Murrelet population in northern California is not now declining, and vital rates are such that the population can be expected to normally remain stable.
- The proposed levels of logging on Pacific Lumber lands would not remove a majority of remaining habitat in the bioregion.
To summarize: based on the indirect evidence of adult-juvenile ratios, most biologists have argued that the Marbled Murrelet population is in an ongoing decline. This includes the population in the bioregion. The new data of the Redwood Sciences Team are the first direct evidence to bear on population trends. Their data by contrast suggest a stable population.
Decision-makers must therefore determine whether the available evidence supports the optimistic or the pessimistic position on the population in northern California. The data do strongly suggest a stable adult population. Ralph and Miller's data show no evidence for a systematic decline from 1990 to 1995; reanalysis of their data confirms this view (Table 1 of this report). However there are also data suggesting poor breeding success (low adult-juvenile ratios). What possible scenarios might explain these contrasting results? Either data set might be based on flawed methodology - it is possible that adult-juvenile data are indeed subject to systematic error, but is difficult to see how the standard survey data of the Redwood Sciences Team could be seriously flawed. Alternatively, there may be biological explanations for the results. The Recovery Team, Beissinger (1995) and others have suggested that, because Murrelets are long-lived, there would be a lag in the population's response to loss of habitat. Under this scenario, with few juveniles entering the population, the population would decline at a rate close to the adult mortality rate. However this scenario is not supported by the directly observed data of Ralph and Miller. Assuming even a high survival rate of adults (95% annually), the population should decline by 23% over 5 years. The data of the Redwood Sciences Team do not support such a decline. A last possibility is that breeding success of Murrelets is strongly determined by ocean conditions, and that recently observed poor breeding success (1992 on) is a temporary state. This scenario is both compatible with the results, and supported by other data on seabird breeding success in California.
At this point, the best available evidence suggests that the population in northern California is relatively stable. The best supported position is that the vital rates for the species more closely resemble the optimistic than the pessimistic scenario.
Some data is also available for evaluating other assumptions of the models. Density-dependence is unlikely to follow the optimistic Beverton-Holt formulation, and more probably is of the pessimistic ceiling type. Since this pessimistic assumption was built into most of Akcakaya's models, this does not affect the predictions for different logging and conservation scenarios. Table 1 of this report suggests that numbers of Murrelets in a specified small area vary dramatically from year to year - this more probably results from migration than from variance in vital rates. Higher migration rates generally favor persistence in metapopulation models, but the Murrelet model is not sensitive to such migration. Table 1 also shows no evidence of a strong correlation in numbers between sites at this small scale. Low correlation would favor Murrelet persistence. To summarize, the other available information do not support pessimistic scenarios, and offer some support for optimistic assumptions.
Lastly, we must evaluate the proposed land transfer and logging activity. At this point, it appears that protection of the Headwaters Complex alone would protect a large part of the Marbled Murrelet population. This new reserve, together with the Humboldt Redwoods State Park would constitute a large portion of presently suitable and future recovery habitat. However we cannot yet determine the proportion of currently occupied habitat that would be lost to harvest of all remaining Pacific Lumber habitat. Ongoing analyses, including fieldwork in 1997, will be essential to resolving this issue.
Recommendations for decision-makers
1. The results of these studies should not be used as the sole basis for making management decisions. However the results clarify which issues are important in the proposed land exchange. Decision-makers should recognize that two factors will have most importance: the current vital rates of Marbled Murrelets in northern California, and the proportion of the population whose habitat is conserved or logged.
2. The best available scientific information is consistent with the Marbled Murrelet population in northern California being currently stable. Given a currently stable population, the Marbled Murrelet in northern California is at a relatively low risk of major decline. Under these circumstances, a land exchange which involves some logging of habitat is unlikely to lead to jeopardy of the species.
3. There are large amounts of potential habitat in Humboldt Redwoods State Park. In the long term the park, and the Headwaters Complex will provide large blocks of high-quality habitat. This will include significant areas of regrowth or 'recovery' habitat.
4. The short-term effect of the proposed land exchange depends on the proportion of the population whose habitat will be lost to logging. We do not yet have sufficient information to determine the proportion of the population in different areas of forest. This information will be available during 1997. If we assume that the Headwaters Complex contains 50% of the population on Pacific Lumber lands, and relatively low occupancy rates in the park, then the proposed conservation strategy would still preserve a substantial majority of the existing population.
5. The best available information suggests that the proposed land exchange is a feasible conservation strategy for the Marbled Murrelet.
