Pacific Southwest Research Station

Partners in Flight Monitoring Working Group Bird/Habitat Subgroup Meeting:

This meeting took place Dec 2nd 1997 in Washington DC
Attendees - CJ Ralph, David Pashley, Bruce Peterjohn, John Sauer, Dave Mehlman, Don Catanzaro, Cindy Brown

Point Count Repository Status

In addition, a 2-year term position is being created for a person to provide part-time computer and programming support for the point count repository biologist. Patuxent has the necessary hardware and software in place for this effort.

Decisions as to the precise structure of the program and the repository have not been made; there are no substantive changes in this regard from previous discussions. Problems and costs will be explored gradually, and kinks worked out as they arise. The intent remains to start with a few data sets with which to begin, but the three most likely remain:

1. The USFWS Region 5 (Northeast) Nation wildlife Refuge point count set, carried out under Diane Pence. This includes the detailed work done on the Patuxent NWR, and consists of about 500-1000 counts conducted under reasonably rigorous guidelines.
2. The Tennessee data set accumulated by Mike Roedel at the Tennessee conservation League. This consists of 4-5000 counts at about 2000 station over 3-4 years using standard methodology.
3. CJ's Northwest California and s. Oregon data set, collected at about 5000 stations in a wide variety of habitats by a large number of cooperators over as many as 5-8 years for some stations. About 3/4 of this is georeferenced and in a common format.

It is generally felt that there is sufficient variation among these three data sets in terms of number of cooperators, degree of rigor, methodology, format, etc. to test the system.

Bruce thinks that the difficult issues in dealing with the bird data in this repository will be based much less on differences in methodology of data collection (5 vs. 10 minutes, fixed vs. variable radii, etc.) and more on the structure of the data bases.

Dave Mehlman said that the TNC bird- habitat project (part of the recent NFWF grant and contract with U. Arkansas) is meant to be a user of these data, and perhaps a catalyst in the effort. Ultimately, the intent is to aggregate bird-habitat relations on a national level and to link these relations to the National Vegetative Community Classification Scheme.

Don Catanzaro said that the Center for Advanced Spatial Technology at U. Ark. wants to work on providing linkage between birds and habitats, and offers expertise in vegetation issues and a capability to deal with large amounts of data combined with tremendous ornithological experience.

CJ felt that managers of the three data sets should work together very soon to massage their data to make it more manageable by Patuxent. It was felt that the person to be hired by Patuxent should be part of this process, so it should not begin until that person is in place. Bruce felt that pre-massaging data could impair the value of the test, in that most data are not going to benefit from that sort of treatment. CJ, on the other hand, suggested that the West Coast data were currently undergoing an organization for consistency, and that his work could be wasted if the organization chosen were ultimately to be considered incorrect. There is clearly a need for very early communication once the new Patuxent biologist is employed, but the exact procedure was not agreed upon.

What will probably come out of the trail process will be a set of instructions from Patuxent as to the structure of point count data that they will accept. Nobody expects Patuxent to accept any format and then go through the incredible amount of labor that would be required to standardize material. Once the new structure is available and publicized, it is hoped and expected that many people will gladly organize and report in that format in order to have their findings included in the national repository. It is also possible that many people will go back to existing data sets and reformat them to fit the new standards.

The issue of ownership of these data was brought up. As a federal agency, anything that Patuxent deals with is in the public domain. Patuxent has protocols and guidance that it will suggest to users, but it can do nothing legally to prevent access and use. Ethical standards must be enforced by editors and reviewers of journals. Potential contributors who want to zealously guard against others using their data should not submit those data to the national repository.

Vegetation data

One thing on which there was unanimous agreement was that point count stations must be rigorously georeferenced. With GPS capabilities, this is now pretty straightforward and is becoming standard practice.

Two of the tools, the classification scheme and remote sensing, are in existence and largely beyond the capacity or interest of PIF to fool with:

• The TNC Veg. Community Classification Scheme is not completely intact (even after years of work), but it is being accepted by the Ecological Society of America, is dictated for universal use in GAP projects, and is being integrated more and more into various federal agencies. There is no sense in PIF inventing something different. We need to use it carefully, however. Birds do not consistently respond to any given level within the hierarchy of the classification system, but given that the system is nested, intelligent use should preclude any confounding problems. Also, community classification does not measure some variables of obvious importance to birds, such as successional stage and structure. There is much more to bird-habitat relations than correlation between occurrence and , for example, alliance type, and we have to be careful to not delude ourselves into thinking that we are understanding more than we do.
• The world of remote sensing continues to make great strides, and through the services of U. Arkansas we have access to much of it. Remotely sensing data are being tied to the Veg. Class. Scheme. It is conceivable that any georeferenced point where bird data are collected could be assigned to a vegetative type on the basis of maps generated from remotely sensed data. However, it is not necessarily safe to assume a community type from a georeferenced spot without locally collected data. There are issues regarding imprecision and a lack of ground truthing. Furthermore, a problem inherent in tying bird occurrence to a community type is that most points are not unambiguously in a single vegetative type.

The issue that we should deal with directly is that of the vegetative data collected in association with point counts. One issue is development of a means of tying locally collected data to the Veg. Class. Scheme. Point-collected data are not a good way to directly bird-habitat relationships. Rather, they are a means of tying bird data to a better scale at which to measure these relationships, which is essentially the scale measured through remote sensing. This is a requirement in order to make any extrapolation beyond what is occurring at a single point- that is, models to explain bird- habitat relationships in a general manner.

In order to even get to that point, however, greater consistency in the types of vegetative information collected at a site must be achieved. CJ suggests that there are 30 generic techniques in use. The national Monitoring WG suggested in its Sacramento meeting that a project be undertaken to look through the literature, determine the range of techniques in use, and search for areas of commonality among them. The first question to be answered is whether coalescence around a limited number of types of data is possible or if mammoth problems of mutual incompatibility preclude this possibility. The second question is whether those components that are shared are actually useful, to assess bird-habitat relationships, to tie with the Veg. Class. Scheme, to tie to remotely sensed data, etc. We can begin to formulate means of achieving consistency once we know the answers to these two questions. If commonality is significant and shared components are useful, an consistent system should be devised into which existing data can be manipulated and inserted. If this is not the case, many existing data sets may be orphaned and unusable in any national or continental analysis. Either way, the end result should be a system (probably nested regarding level of detail) that will thereafter be available for universal use in association with bird monitoring.

Mike Scott has expressed an interest in pursuing this as part of his Ornithological Council monitoring survey, but is generally felt that this is a large job that is beyond his logistical capacity. The bird-habitat literature has been pulled together, so the time needed to complete a literature search is not a problem. CJ suggests that an ornithologist and a plant ecologist (perhaps graduate students?) need to work together on this, and that the initial phase (describing the range of techniques) could be done in perhaps 3-4 months. $30,000 is a guess as to what it might cost. Because nobody present at the meeting currently has or knows of anyone else that has the money or the staff time to do this work, it will be necessary to generate a proposal and some funding. With money, however, any number of groups could do the work (Mike Scott, U. Arkansas, CJ at Humboldt, etc.). The next step, coming up with an agreed-upon consistent national system, could perhaps be part of this proposal (with additional associated costs.) or it could be the responsibility of the national Monitoring Working Group. If it is the former, it would have to be assigned to a larger group of experienced individuals.

A suggested potential funder for this project is the Multi-resource Land Consortium (MRLC), which is housed at the EROS Data Center in Souix Falls, SD. This is part of USGS and is headed by Pete Campbell. This group is looking for users of NASA generated remotely sensed data, in part to help justify the huge amounts of money being spent on satellites, etc. A small grant like this may be looked upon quite favorably. Not only would we be using their remote data, but they need huge amounts of ground-truthing for GAP projects, etc. If all people in the country doing bird point counts were contributing consistently gathered vegetative data that could be tied to a Veg. Class. Scheme that is used universally in GAP projects, we would provide them a service that would be incredibly expensive to otherwise acquire.

It is necessary that ground-collected data be translatable into Veg. Class. Scheme (D. Mehlman is going to start assembling information regarding this transfer). That is necessary but it is not sufficient. Data not needed for translation to a scheme that are also relevant to bird use, such as structural information, should also be included. Assignation to an alliance can be done with a subset of the data that need to be collected. The proposal should be forward-looking; it should include an experimental test component (which could probably be done using the Patuxent NWR data set); and the results should be publishable.

How are we going to develop such a proposal? D. Pashley will finish these minutes with a short proposed outline (below). All participants at the meeting agreed to then think about the outline and then to start filling in some of the pieces. CJ will distribute these notes more widely through the Monitoring WG. Everyone should send their thoughts on this to CJ and David Pashley.

Research Contact:

• C.J. Ralph