The PVA Oversight Panel reconvened on June 12 and 13 1997, in Arcata CA, to consider the results obtained by the scientific analyses. The initial impetus for bringing the panel back together was the suggestion of Tom Tuchmann that an impartial scientific evaluation would be useful to decision makers at this point.
The meeting was recorded on videotape. Copies of the tapes can be obtained from CJ's Studio (707-442-5939), or from Steven Courtney. These notes are prepared from the record.
The attached agenda shows the presentations and discussions as they occurred at the meeting.
Introduction
Courtney introduced the meeting, by describing the process that has been developed for the Pacific Lumber HCP, and for the Marbled Murrelet Population Viability Analysis. The original Advisory Group consisted of Cody, Kareiva, Murdoch and Raphael, and was convened to consider the design of the PVA. Kareiva could not be reached for this meeting, but Barry Noon of Redwoods Sciences Lab (soon to be of Colorado State University) has been added to the Team (Raphael arrived at the meeting on the evening of the 12th). Other Murrelet scientists had been invited to attend, but were unable to do so at the short notice necessary for this meeting.
The process that has been developed is to approach the Murrelet PVA in a cooperative and scientific manner. Courtney asked the group to maintain the approach, to acknowledge diversity of opinion where necessary, and to establish consensus where possible. Courtney described the goal of the meeting as a frank evaluation of the analysis to date. He stressed that the science was still in progress, and that new results were still arriving. An important objective of the meeting was to provide help and guidance over the last few months of this critical phase of the study.
Courtney also described the role of the PVA in the overall science analysis. Several other studies are ongoing in 1997, notably the various aspects of the Habitat Analysis. These other analyses have important roles in evaluating the level of 'take' under any HCP; the PVA will be most useful in examining 'jeopardy'. Of course, the data collected under the Habitat Analysis will also be important input to the PVA model. The Panel's main role is in advising the PVA scientists; however the advice to the Habitat Analysis scientists would also be useful.
Several studies are ongoing in 1997:
PVA Analysis: Final Product by late July/early August
Supported by Inland detection analysis (Hamer) by late July
Offshore surveys
Habitat Analysis:
Inland habitat surveys for Murrelets
Inland habitat description (including State Park)
Capture and radio-telemetry of Murrelets
At the request of the US Fish and Wildlife Service, Courtney also outlined the issues underlying the decision to list the species under the Endangered Species Act, and the goals of the Recovery Plan, which focus on issues of geographic distribution, and both short-and long-term survival. These issues were described in more detail by Paul Henson in a letter (attached) which was circulated at the meeting.
PVA Modeling
Ginzburg reported the results of the population modeling study, essentially unchanged since the May report. This study is continuing and data are arriving from RSL and other sources that will refine the models during summer 1997. This includes GIS data, to develop a spatially-explicit model. Focus is still on the mid-scale model (for Conservation Zone IV), as agreed at previous meetings, but larger (listed region) and smaller scale (bioregion) models will also be completed. As before, the basic design of the models is taken from the parameters developed in the model of Beisinger.
Discussion of the model focused on the assumptions used in construction. One question concerned the use of strongly divergent estimates of demographic parameters in the sensitivity analysis. It was argued that this approach will result in an over-estimate of the importance of these assumptions. Ginzburg agreed that as we develop a better understanding of vital rates, the analysis will focus our attention on the uncertainty associated with other parameters; however at this point this is still the major source of uncertainty.
Some technical suggestions were made. For instance, it was thought to be biologically unreasonable for density-dependent factors acting through habitat availability to affect the vital rates of sub-adults and juveniles. It was also pointed out that there might be subtle second-order interactions between parameters that have not been captured by the model to date: notably the effect of forest harvest is modeled simply by a reduction in carrying capacity (K), but might reasonably be argued to affect fecundity (increased predation on juveniles in fragmented forests). Ginzburg agreed to consider these issues.
Allee effects at low population levels were discussed. Small scale effects will be missed in population models such as that described here. The group discussed the usefulness of individual-based models in studying this issue.
The panel commented that (under adverse survivorship conditions) logging has small effect on the model results, and some panel members were anxious that this not be misinterpreted as sanctioning large-scale logging. It was pointed out that if the population was indeed discovered to be declining rapidly, then it was unlikely that regulatory agencies would permit large-scale loss of habitat.
Courtney commented that the role of the PVA should not be to provide hard quantitative population projections, but to guide us on what factors are important. The models should also guide us on what we need to know more about. They have for instance focussed our attention on parameters that have the biggest effect, including survivorship. The PVA gives us better decision tools, and points us towards critical data needs. As emphasized in the first PVA workshop, we need to focus attention on what the available data actually say.
Population Trend Data
Ralph, Miller and Matsumoto presented data and preliminary analysis of at-sea
survey results carried out by RSL from 1989-present. These studies are ongoing,
and part of normal RSL research activity. 1996 data were presented for the
first time at the meeting (see appended results). 1996 was a uniformly poor
year for Murrelets in northern California with the lowest counts yet recorded
on several transects. Analysis of these data using simple regression analysis
on each transect mean densities yields some suggestions of downward population
trends. The overall pattern of these results is similar to those reported
in the earlier report, although that first analysis (on data up to 1995)
showed some positive trends that are no longer discernible.
The group agreed that the new data lend some support to the position that population numbers have declined over the period 1989-96. However it is not yet possible to determine the rate of decline, or whether this is ongoing. Neither is it yet possible to ascribe any loss of population to potential causative factors (forest harvest, ocean conditions, rising predator densities). Adjacent areas seem to fluctuate with little evidence of synchrony.
The group also discussed the additional insight that can be gained from use of adult-juvenile ratios. There was a diversity of opinion on the usefulness of these data, although the data collection method itself is well accepted. An important issue that is unresolved is the proportion of the at-sea population that is attempting to breed locally, and whether any of the population are non-breeders (perhaps derived from other populations to the north).
The group pointed out that population trends will be subject to time-lags; hence recent declines could be due to factors operating up to 10 years previously. Conversely, there may be a similar lag in future responses. This delayed response complicates trend analysis, and suggests a need for sufficient duration of study (10+ years) in order to reach firm conclusions.
Other points discussed by the group included the recognition that monitoring of the population is best carried out on the ocean; the need for power analyses; use of 2km transects as the basis of analysis; the need to include intensive survey data; the potential pseudo-replication problem associated with association of segments; estimates of marine productivity.
There was extensive discussion of statistical methods, including the need to state the ability to distinguish between alternative hypotheses (e.g. stability versus 6% annual decline). Burkett reported on the relation of these issues to those raised by Marine Survey workshops.
Habitat Analysis
Ralph et al reported on the latest data from the Habitat Analysis. The overall thrust of this work is to determine the relative importance of different stands for the conservation of the Murrelet. Alternative estimates of the importance of stands can be made on the basis of habitat area and quality, or by including information on Murrelet detection levels or levels of occupancy.
The basic process is to describe the vegetation types and landscape configurations in the bioregion, and to sample these for Murrelet occupancy. This will then result in a model of the value of different areas for the species.
Recent data have included information on the relative importance of Residual stands on the Pacific Lumber ownership (appended document). These data show that there is higher use of residual stands than had been predicted. Because this result was somewhat unexpected, the data were examined to determine whether the use of residual stands was over-estimated as a consequence of placement of residuals near old-growth, or vegetation types. The data seem to show that this is not the case. Residuals on Pacific Lumber lands are used at high levels even when distant from old-growth, and regardless of habitat type. However it should be noted that use of residuals is still less than use of old-growth. Moreover, it should be emphasized that these are detection levels, which may be elevated in more open (thinned) stands, with increased visibility. Detection levels also may not be a strong indicator of habitat quality or breeding success, which may be higher in undisturbed old-growth than in more open residual stands. It may also be that Murrelets are associated with only a few residual stands, but at high levels of activity.
There is an ongoing program of inland surveys in 1997. The main thrust of these surveys is to determine occupancy and activity levels in areas that have not received adequate coverage thus far: notably residual stands, the State Park, and some old-growth stands. The program for 1997 will establish a solid sample of all habitat classes (e.g. doubling effort in residuals), and of geographic coverage.
Discussion included: the need for a good measure of nest success and habitat quality; the number of nests that may be found in a single stand in a year; the rate of succession in Redwood forest (e.g. in residual stands); statistical analysis of data on occupancy vs. distance to old-growth; the use of a spatial analysis that would include marine and terrestrial factors. Stopher raised the importance of including distance from the ocean, and topographic features. Murdoch asked whether the number of Murrelets thought to be nesting in the Bioregion could be responsible for all the occupied detections (on 11 000 acres of habitat). Herman pointed out that surveys had been carried out in residuals to determine absence prior to harvest, and would have been selectively placed in 'better quality' habitat. Hence there might be a systematic bias to the surveys in these areas, which were not randomly selected.
Detrich emphasized that the conservation strategy will use data on habitat. There was also discussion of the opportunities raised by the continued presence of birds in residual stands after harvest.
The panel also noted that the importance of residual stands will be more easily evaluated when more data are available. These may also suggest alternative management options. For instance, it is possible that further information on nesting success and predator behavior will suggest that unentered old-growth can be thinned without harming nesting potential of the stand (these data are not yet available). Similarly, it is possible that residual stands are 'sinks' which attract nesting Murrelets, but which are poor breeding locations. Elimination of residuals would therefore have little negative effect on the population. These speculations must however await further data before they could be useful to decision-makers.
Courtney summarized by pointing out that there has been significant progress since the last meeting. Uncertainty has been narrowed considerably. The habitat analysis has already indicated that upper and lower estimates of habitat can be rejected.
Presentation by Ken Moore
There has been a diversity of opinion on the amount of Murrelet habitat
on Humboldt Redwoods State Park. This is an important issue, because it
determines the proportion of Murrelet habitat in the Bioregion that is protected,
or at risk from logging.
Moore, who has suggested that the amount of habitat in the State Park is low, presented the panel with a slide show illustrating his experience with the Park, and with Murrelet habitat in the region. He stressed that although he lacked quantitative data, and therefore was presenting 'guestimates' , he had extensive experience in delimitating habitat on the ground.
On the basis of this experience, Moore estimated that there was less than 10,000 acres of Murrelet habitat on the Park, of varying levels of 'quality'. He emphasized that ongoing research by Ralph, and the current Murrelet survey efforts would resolve any remaining differences, and supported this effort. A site-visit or separate meeting might be useful when these are complete.
Discussion
Predation and Fragmentation: Burkett emphasized that fragmentation studies from East Coast forests are being applied here, in very different forest types. Data are available from some Redwood forest work (by Suddjian etc), and show that harvest history does not seem to affect corvid (predator) numbers. Proximity to humans however does increase predator numbers. Courtney reported on work by himself and other SEI scientists that shows similar patterns.
Murdoch emphasized how important data on actual nest success could be to management of e.g. residual stands. This would also affect our understanding of, for instance, the effects of predation on vital rates.
There was extensive discussion of the possibility of increased mortality during dispersal from the natal nest site to new nesting areas. Mortality during such dispersal was not thought to be significantly greater than normal over-winter mortality.
Cyclicity of the marine environment was also discussed. Temporal correlation (runs of good or bad years) might be important, and affect population predictions form PVA models. It was pointed out that mortality in some seabirds under EL Nino events can be catastrophic. It might be that birds such as Murrelets cannot breed at all in El Nino years, given their 'marginal' life-history. Data on marine trends are being collected by D.Brosnan (SEI) and should be incorporated into models.
Cody emphasized the importance of using detection levels and occupancy levels to 'weight' the stands, such that more value is given to areas with many detections.
Murdoch stated that, if the population is declining to a new equilibrium at a low K, then new logging will lead to further decline if K is reduced further. The length of the lag of this response cannot be longer than the average life-span of the bird.
Estimates of survivorship and breeding success are critical, and data
on these factors should be collected if at all possible. The panel again
emphasized the difficulty of working with this species, and the need to
narrow down opinion on critical variables.
Presentation of Agency Questions
The USFWS (through P.Henson) and the CDFG (through M.Stopher) had, prior
to the meeting, developed a list of questions for the panel (see attached
documents). Courtney had developed a precis of these questions, which was
presented to the panel as the final agenda item for the first day of work
(see attached documents). The panel were asked to consider these questions
as a guide to the needs of the parties, but not to be limited in their comments.
The panel were urged to pass comment on the process as followed thus far,
on the quality of the information and the analysis, and on what reasonably
could be deduced at this point. Guidance for future decisions and for future
analyses was also sought. The panel was also asked to set limits on how
the data could be used. Courtney pointed out the consequences of making
scientific errors in different directions, and asked for advice on the most
appropriate interpretation under these circumstances. The panel was also
asked for advice on design features to be incorporated in conservation plans.
Detrich asked for the panel's assessment of the use of the PVA and Habitat
Analysis process in the regulatory arena.
Panel Response
The panel met independently of the larger group to develop their responses.
The panel had some consensus, and went as far as they could in answering
the questions that were posed. Their main points can be summarized as follows:
1. The Panel is impressed with the process developed by SEI to address this 'classical conflict situation' of groups with competing goals. The process appears to be working well.
2. The Panel feels well qualified to evaluate the work so far.
3. The Panel is impressed with the progress made since the first meeting of the group. Many excellent data have been collected. Continuing on this path is appropriate. Much better data is now being collected. There will be a qualitative increase in our abilities to understand and manage the species, as more information becomes available.
4. The prognosis for making a good decision is very good, with a few more data.
5. Nevertheless, the Panel is not yet in a position to give definitive answers to agencies' questions. The panel cannot increase the comfort level for those making decisions now.
6. The time-frame for making better decisions is not long (perhaps 5 years).
7. It is likely that a decrease in breeding habitat will lead to further decline of Murrelets, but a definitive answer to this question is difficult at this time.
8. There is more information to be extracted from existing data, which can be analyzed relatively swiftly.
9. Decline of the population is the most defensible hypothesis on population change, especially with the 1996 data. However the panel is not ready to defend this hypothesis. Although the RSL marine data are the best available anywhere in the range of the species, they are still not enough to understand population trends. More data are important. Analytical improvements would help the study.
10. The cause of any decline is unknown. Future trends are also unknown. A priori we might expect to be in a transitional phase as the population adjusts to a new K following timber harvest in the 1980's.
11. We also don't know the breeding success of the population - there may be many non-breeders present. The panel questioned the value of adult-juvenile ratios. Models that predict trends based on demography, habitat history, predators, marine factors should be examined to determine which best predict the apparent trend.
12. The panel did not wish to make any call on 'Jeopardy' of the species or population, at any level (bioregion or listed range). The panel saw no way to add to comfort levels on this decision right now.
13. The panel was supportive of adaptive management. Information gathered in the next few years would be critical, and should be incorporated into management.
14. Some panel members advocated that the conservation management process should explicitly incorporate information needs and uncertainty. Models of the management process should be updated through an ongoing monitoring program with decisions that are conditioned by the results of the monitoring.
15. The adaptive management process should, as far as possible, avoid irreversible changes
16. Some changes to the PVA were suggested.
17. Additional statistical support should be provided to the study of population trends. This should include a weighted analysis, and an estimate of power. This process would also help in the future allocation of survey effort.
18. The process is making significant headway. This is good science. The gaps between divergent opinions have narrowed. For instance, it is now apparently agreed that Pacific Lumber does not own 50% of the available Murrelet habitat (outside of the Headwaters); neither does the Company own as little as 12%. The true figure is probably in the range of 25 to 33%.
19. The panel cannot give sensible answers to many questions (e.g. 3,4,5,6, on list provided) at this point.
20. It will be difficult to make good decisions now with the available information. However if the decisions can be deferred, they may be good decisions. The data being collected and developed now are qualitatively better. It is critical to keep this process going. A FEW more years will make a big difference to statistical power.
Cody argued that extra effort, and continued data collection is needed now because: A. the analysis is at a 'qualitative crux' statistically, and B. the population itself is probably at a critical transition phase (this will not be true in 5 more years). Data collection needs to continue. This will not need to be an open-ended process. A few more years will make a much stronger case - beyond that additional effort would be pointless. The current data collection is proceeding well.
The panel also generally agreed that early conservation/management decisions should be tentative, and that later decisions need be less so, because the quality of the data will be far superior.
There was some discussion of the concept of an 'interim strategy', involving harvest of stands of least conservation value, and continued data collection over the next 5 years, using adaptive management. Courtney pointed that the Butano Creek THP provides a model and precedent of this. The panel suggested prioritizing of stands, perhaps on the basis of the number of Murrelets detected on surveys. It would be difficult to prioritize at this point. Raphael suggested incorporating information on nest predation into the process. Reid suggested setting up alternative prioritization models, and determining of the same results are predicted using different criteria.
Adaptive management and Information needs
The panel had strong opinions on the need for further information. They argued that the inferences that could be drawn now were weak, and would not be of much help in making critical decisions. It was emphasized that this was not a simple case of scientists asking for more and more research. Instead the data are at a critical point, where a few more years data will make a large difference in the quality of data and of decisions. Regression analysis in particular is at a critical juncture, where 3 to 5 years' data will resolve the population trend issue.
Noon emphasized the need to incorporate uncertainty into the decision-making process. Monitoring is an essential part of the adaptive management process; otherwise monitoring is a pointless activity. A five year time frame seems appropriate.
The panel identified additional data to be collected:
Other needs:
Better trend estimate techniques
Check raw data
Data go-between for PVA and Habitat Analysis
Possibly, a behavioral/ individual-based model
There may also be opportunities that develop from analytical results - for instance the use of residual stands as potential recovery habitat.
Concluding remarks
Detrich pointed out that in November he had 'put his foot in it' and suggested that sensitivity analysis would be valuable. It has been really valuable, even if the ambiguity is frustrating. We have more clearly defined what we know and don't know. We have also demonstrated that we have done the best science we can; even if at this juncture the science can't help, some hope that it will in future.
Stopher reiterated these points and emphasized that existing data, and upcoming analysis on trends and on State Park habitat will be very valuable.
Courtney closed the meeting by pointing out the progress that had been made, prior to and during the meeting. There was now general agreement that the data indicated some decline in the population. There was also consensus that the amount of habitat on Pacific Lumber lands (outside of the Headwaters Complex) constituted c.25 to 33% of the habitat in the Bioregion. The PVA models have successfully identified that assumptions on life-history parameters determine the outcome of population projections. Perhaps most importantly, there is general recognition that the best possible science has been brought to this issue, but that important questions are unresolved, and will remain so in the short-term.
Given the difficulties and the emotive power of these issues, the group has been extraordinarily successful. The cooperative attitude of all participants has led to important progress.
Agenda
June 12/13 1997
Mad River Saloon, Arcata
June 12
10.30 Introduction Courtney
The HCP Process
10.50 Reports on Results
PVA Ginzburg
11.20 Population Trend data Ralph et al
1.30 Habitat Analysis Ralph et al
2.50 Humboldt Redwoods Moore
State Park
3.30 Discussion
4.30 Presentation of agency S.Courtney
questions
June 13
Panel Meeting 8.00 to 9.30
9.30 Response to questions Panel
11.30 Adaptive management
12.30 Meeting ended