Monitoring, Modeling, and Management

An adaptive management cycle of Monitoring, Modeling, and Management is outlined in the proceedings of the 2002 Partners in Flight Conference (LINK) and is being applied in the following two projects:

1.  Managing Landbird Populations in National Forests of the Pacific Northwest (LINK)

2.    Management Strategies for Reversing Declines in Landbirds of Conservation Concern on Military Installations (LINK)

In this adaptation of the adaptive management cycle we first monitored bird populations and collected baseline demographic data for up to ten years.  From these data and high-resolution land cover data we constructed species-landscape models in which demographics (e.g. mean annual numbers of adults or young) were expressed as a function of landscape characteristics.  From the set of species for which such models were acceptable we chose a suite of species of conservation/management concern and formulated management guidelines to maintain or create “source” populations.

In the next step of the cycle we rearranged the network of stations to better monitor birds of concern by a) keeping stations that already effectively monitor the species as controls, b) moving “slow” stations that catch few birds to suitable breeding habitat (managed or natural) for species of concern.  At some stations we implemented management actions within the boundaries of (or vicinity of) stations in a manner expected to benefit the species of conservation concern.

Now we are monitoring the effectiveness of those actions over five or more years.

Monitoring background demographics

Modeling the relationships between local avian demographics and the characteristics of surrounding landscape requires a time series of demographic data a) such that declining or increasing populations can be recognized and, b) to measure and subsequently account for the potential bias of spatio-temporal variation in weather conditions.  To gain acceptable precision in survival rate estimates and have enough statistical power to detect trends a group of stations must operate for a number of years.  Estimates of time-independent survival rates require a minimum of four years.  In the Pacific Northwest study we collected baseline data for ten years (1992-2001) and in the study on military lands collected baseline data for nine years.

Modeling bird and landscape data

One assumption of constant-effort bird banding protocols is that they sample birds from within the boundaries of stations and birds whose territories contain a net.  Furthermore, they sample post fledgling dispersal such that many young and adults captured late in the season nested or were reared in the surrounding landscape.  Other studies (IBP unpublished) show that relationships between avian demographics and landscape characteristics (derived from National Land Cover Dataset [1992]) increase with radius from the banding station, and that these relationships may differ by species and region.  For the DoD Legacy-funded project we chose a scale of 5km but reduced this to 2km for the USFS Pacific Northwest Region study.

Parameter selection

Modeling landscape data is notoriously difficult due to the high level of covariation between variables.  First we selected a subset of statistics that biological sense for a species.  For instance, a grassland bird would not be expected to respond to detailed characteristics of forest types, but might respond to the areal coverage or edge characteristics of deciduous or coniferous forest.

Model selection

Once a set of variables was chosen we conducted a multiple linear regressions for a chosen demographic (e.g. numbers of young) using model selection based on information theory Bozdogans ICOMP statistic.  This statistic penalizes models for covariance and colinearity among its variables and therefore tends to chose models with a minimum number of terms.  Such relationships are easy to interpret and utilize in order to predict the effects of proposed management on a population.  Alternate modeling techniques such as principal components analysis are less easy to interpret and difficult to use as a predictive tool.

Managing for “source” populations

Any local population of a species of conservation concern should be managed as a “source” population such that the species is both abundant and, annual variation notwithstanding, productive enough to more than replace its annual losses such that the excess disperse to existing unoccupied territories and newly created but unoccupied habitat.  Because MAPS monitoring stations sample birds from the adjacent landscape, especially during the post-fledging phase of the season, management does not have to directly impact the station footprint to effect change in the alpha diversity and numbers of birds captured.

Monitoring management effectiveness

Once management is implemented then “effectiveness monitoring” is conducted to compare post-management demographics to pre-management demographics.  Again it is important to monitor for as many years as is needed to obtain demographic parameter estimates with acceptable precision.  Also, some effects may be lagged several years during the period of community relaxation that often occurs after a major disturbance.  If the management goals are not met (e.g. productivity remained low) after some time then the management will be reconsidered and adjusted in another attempt to meet the goals.

[BACK]