Conservation Design for Sustainable Avian Populations

Slides:



Advertisements
Similar presentations
Community and gradient analysis: Matrix approaches in macroecology The world comes in fragments.
Advertisements

Krishna Pacifici Department of Applied Ecology NCSU January 10, 2014.
Spatial Structure & Metapopulations. Clematis fremontii Erickson 1945.
Maximum likelihood estimates What are they and why do we care? Relationship to AIC and other model selection criteria.
Species interaction models. Goal Determine whether a site is occupied by two different species and if they affect each others' detection and occupancy.
458 Estimating Extinction Risk (the IUCN criteria) Fish 458; Lecture 24.
Stepping Forward Population Objectives Partners in Flight Conservation Design Workshop April 2006 and Delivering Conservation.
Announcements Added a README file re: VORTEX HW3 due Wednesday First draft due April 16 (Changed from April 13)!
A COMPARISON OF APPROACHES FOR VERIFYING SOUTHWEST REGIONAL GAP VERTEBRATE-HABITAT DISTRIBUTION MODELS J. Judson Wynne, Charles A. Drost and Kathryn A.
Bio432 Topic(s) for 2nd paper: Mating system evolutionary ecology Kin selection Non-kin cooperation Cultural evolution See references on course webpage.
Purposes of protected areas protect focal sp. / spp. –umbrella species protect biodiversity (spp. richness, endemism) protect large, functioning ecosystems.
An overview of a few of the methods used in landscape ecology studies.
Problem Definition Exercise. U.S. Fish & Wildlife Service General Summary Responses from ½ of those surveyed (n=14/31) Broad and narrow in scope Narrow.
Introduction Evaluating Population-Habitat Relationships of Forest Breeding Birds at Multiple Scales Using Forest Inventory and Analysis Data Todd M. Fearer.
Using Birds to Guide Post-fire Management in the Plumas & Lassen National Forests Ryan D. Burnett, Nathaniel Seavy, and Diana Humple 4/21/2011.
Ecoregion typing Ecological classification or typing will allow the grouping of rivers according to similarities based on a top-down nested hierarchical.
Commonly referred to as MIS.  From the 1982 planning regulations 36 CFR (a)(1)- “… certain vertebrate and/or invertebrate species present in the.
USGS Global Change Science National Climate Change & Wildlife Science Center and SE Regional Hub Sonya Jones USGS Southeast Area NIDIS Planning Meeting.
POPULATION ECOLOGY. ECOLOGY Study of living organisms as groups Interactions between living organisms (predator-prey, parasitism etc) Interactions between.
MANAGEMENT AND ANALYSIS OF WILDLIFE BIOLOGY DATA Bret A. Collier 1 and T. Wayne Schwertner 2 1 Institute of Renewable Natural Resources, Texas A&M University,
Landscape Ecology Questions Current regulations in Massachusetts and other states tend to leave landscapes rich in wetlands but lacking diverse and extensive.
Using historic data sources to calibrate and validate models of species’ range dynamics Giovanni Rapacciuolo University of California Berkeley
Resource Selection Functions and Patch Occupancy Models: Similarities and Differences Lyman L. McDonald Senior Biometrician Western EcoSystems Technology,
Habitat Evaluation Procedures – an enlightened Congress passes conservation legislation Affecting management of fish & wildlife resources NEPA.
EEES4760/6760 Landscape Ecology Jiquan chen Feb. 25, Fragmentation 2.Island Biogeographic Theory (IBT)
Landscape Ecology: Conclusions and Future Directions.
Patch Occupancy: The Problem
Habitat Fragmentation. Many times, natural habitats show a “patchy” distribution. This affects the organisms that live there.
Spatial ecology I: metapopulations Bio 415/615. Questions 1. How can spatially isolated populations be ‘connected’? 2. What question does the Levins metapopulation.
Population dynamics across multiple sites Multiple populations How many populations are needed to ensure a high probability of survival for a species?
From Expert-based to Data-based Decision Support for Strategic Habitat Conservation Ashton Drew & Jaime Collazo NCSU Biology Department Biodiversity &
A few more thoughts regarding predator prey / resource consumer dynamics and population regulation: Food webs From: Bolen and Robinson (2003)
The science of conservation planning Course objective: a free-ranging examination of some key scientific principles and research needs pertaining to conservation.
Identifying Species Targets at the Landscape/ Seascape Scale.
Landscape ecology methods
Global Analyzing community data with joint species distribution models abundance, traits, phylogeny, co-occurrence and spatio-temporal structures Otso.
Introduction to Models Lecture 8 February 22, 2005.
Workshop on Applied Hierarchical Modeling in BUGS and unmarked Patuxent Wildlife Research Center November 2015.
Multiple Season Model Part I. 2 Outline  Data structure  Implicit dynamics  Explicit dynamics  Ecological and conservation applications.
Estimation of Animal Abundance and Density Miscellaneous Observation- Based Estimation Methods 5.2.
OUTLINE FOR THIS WEEK Lec 11 – 13 METAPOPULATIONS concept --> simple model Spatially realistic metapopulation models Design and Implementation Pluses/minuses.
Statistical Methods. 2 Concepts and Notations Sample unit – the basic landscape unit at which we wish to establish the presence/absence of the species.
 1 Species Richness 5.19 UF Community-level Studies Many community-level studies collect occupancy-type data (species lists). Imperfect detection.
Spatially Explicit Capture-recapture Models for Density Estimation 5.11 UF-2015.
Emergence of Landscape Ecology Equilibrium View Constant species composition Disturbance & succession = subordinate factors Ecosystems self-contained Internal.
Inferences About Animal Populations. Why Estimate Population Attributes? Science Understand ecological systems Learn stuff Management/Conservation Apply.
 Integrated Modelling of Habitat and Species Occurrence Dynamics.
Monitoring and Estimating Species Richness Paul F. Doherty, Jr. Fishery and Wildlife Biology Department Colorado State University Fort Collins, CO.
Single Season Model Part I. 2 Basic Field Situation From a population of S sampling units, s are selected and surveyed for the species. Units are closed.
Single Season Occupancy Modeling 5.13 UF Occupancy Modeling State variable is proportion of patches that is occupied by a species of interest.
1 Occupancy models extension: Species Co-occurrence.
 1 Modelling Occurrence of Multiple Species. 2 Motivation Often there may be a desire to model multiple species simultaneously.  Sparse data.  Compare/contrast.
 Multi-state Occupancy. Multiple Occupancy States Rather than just presence/absence of the species at a sampling unit, ‘occupancy’ could be categorized.
Multiple Season Study Design. 2 Recap All of the issues discussed with respect to single season designs are still pertinent.  why, what and how  how.
Single Season Study Design. 2 Points for consideration Don’t forget; why, what and how. A well designed study will:  highlight gaps in current knowledge.
 Occupancy Model Extensions. Number of Patches or Sample Units Unknown, Single Season So far have assumed the number of sampling units in the population.
Population Ecology and Conservation Population Ecology and Conservation A Conceptual Framework 2.1 UF-2015.
Multi-state Occupancy. Multiple Occupancy States Rather than just presence/absence of the species at a sampling unit, ‘occupancy’ could be categorized.
Biological structure of Fisheries Resources In Space And Time.
Chloe Boynton & Kristen Walters February 22, 2017
Identification of Restoration Sites for  a Fire-dependent Bird in an Urbanizing Environment Bradley A. Pickens North Carolina Cooperative Fish and Wildlife.
Communities and the Landscape Lecture 15 April 7, 2005
Landscape dynamics in the Southern Atlantic Coastal Plain in response to climate change, sea level rise and urban growth Todd S. Earnhardt, Biology Department,
Quantifying Scale and Pattern Lecture 7 February 15, 2005
Integrated Monitoring in Bird Conservation Regions
Large-scale Ecology Interacting ecosystems
Analysis to Inform Management
Construction of nature reserves
Delivering Conservation
Estimating mean abundance from repeated presence-absence surveys
Presentation transcript:

Conservation Design for Sustainable Avian Populations Modeling species occurrence dynamics at landscape levels NC Cooperative Fish & Wildlife Research Unit, NCSU Patuxent Wildlife Research Center AL Cooperative Fish & Wildlife Research Unit, Auburn University Atlantic Coast Joint Venture Gap Analysis Program (BaSIC)

Abundance…and vital rates… Why not abundance as a starting point? BBS-derived abundance estimates cannot account/adjust for factors that affect the detection process Availability and perception Multiple species and habitats Background noise (Simons et al. 2007) Discrepancy between adjusted and unadjusted estimates might be substantial. Thus there could be profound implications for Modeling species dynamics Conservation design Vital rates… Available for game and endangered species Remaining community members--fragmentary at best. Simons, T.R., M.W. Alldredge, and K.H. Pollock. 2007. Experimental analysis of the auditory detection process on avian point counts. Auk 124:986-999. Nichols, J. D., L. Thomas, and P. B. Conn. In Press. Inferences about landbird abundance from count data: recent advances and future directions. Environmental and Ecological Statistics. 2

Patch Occupancy Models Patch Occupancy offers an alternative approach. Survey counts are re-tallied as “presence-absence” (detection-nondetection), Assumptions and inferences are more tractable, Preserves ability to estimate selected vital rates and community-level metrics over time. Patch Occupancy (MacKenzie et al. 2006) is an appropriate framework for this project given the number of species and habitats in question, and the issues with detection probability concerning BBS raw data. It generates parameters of interest (e.g., vital rates) and is appropriate to evaluate range dynamics potentially associated with Global Climate Change (e.g., Royle and Kéry 2007; Ecology 88:1813-1823). MacKenzie, D. I., J. D. Nichols J. A. Royle, K. H. Pollock, L. L. Bailey, and J. E. Hines. 2006. Occupancy estimation and modeling: inferring patterns and dynamics of species occurrence. Academic Press. 324 pp.

Patch Occupancy Patch Occupancy (Psi) is defined as the probability that a site is occupied. It is conditioned by fact that species is not always detected with certainty, even when present (p < 1) Notation: - probability site i is occupied - probability of detecting the species in site i at time j, given species is present The model framework permits relating  and p to site and/or sampling characteristics via the logistic model (or logit link). Most applicable to this project will be: Site-specific: model  and/or p e.g., habitat type, patch size, patch isolation Newer versions of PRESENCE generate Psi-conditional. The Psi-conditional is the estimate of occupancy, given that the species was never detected at a site. So, Psi is the probability that any site is occupied, some of which you will detect the species. And Psi-conditional is the probability that sites where you had no detections are occupied.

Patch Occupancy and Conservation Design Abundance – revisited? It is possible to estimate abundance from presence-absence data (Royle and Nichols 2003). Two assumptions need to be met: the probability of detecting an animal at a site is a function of how many animals are actually at that site, the spatial distribution of the animals across the survey sites follows a specified prior distribution, such as the Poisson distribution, BUT approach based on temporal replication, not spatial. If review of literature or data suggests that discrepancy between adjusted and unadjusted counts is deemed acceptable W. Thogmartin’s (2004) provides a comprehensive approach to estimate abundance from BBS data Incorporates multiple factors influencing abundance estimates including possible changes in detection due to changes in observers over time. It would be very useful to further qualify (rank) landscape contexts (patches) with low extinction rates using abundance estimates generated via Patch Occupancy framework. Abundance estimates would also provide a link with population objectives, a metric used in most conservation planning documents. There are options but they might require relaxing assumptions (e.g., equal detection across a segment or primary sampling unit) in the case of BBS-based data. But other issues are being examined before formally proposing using BBS data. Such assumptions might not be needed for species like the Bobwhite Quail where temporal and spatial replication is part of routine surveying protocols. Royle, J.A. and J.D. Nichols. 2003. Estimating abundance from repeated presence-absence data or point counts. Ecology 84(3):777-790. Thogmartin, W. E., J. R. Sauer, and M. G. Knutson. 2004. A hierarchical spatial model of avian abundance with application to Cerulean Warblers. Ecological Applications 14:1766-1779.

Conservation Design Project - Approach SAMBI Region Focal Regions SAMBI Eastern United States Assessments based on BBS and remotely-sensed data. Analytical Approach - MacKenzie, D. I., J. D. Nichols J. A. Royle, K. H. Pollock, L. L. Bailey, and J. E. Hines. 2006. Occupancy estimation and modeling: inferring patterns and dynamics of species occurrence. Academic Press. 324 pp. Mackenzie, D.I., J.D. Nichols, J.E. Hines, M.G. Knutson, and A.B. Franklin. 2003. Estimating site occupancy, colonization and local extinction probabilities when a species is not detected with certainty. Ecology 84:2200-2207. Patch Occupancy MacKenzie et al. 2003 and 2006 Monica Iglecia, MS Student Post-Doctoral Research Associate 6

Data Source and Sampling Units Motivation Reduces extent and number of habitat classes within sampled unit Minimizes heterogeneity in detection probability Occupancy-habitat relationship “tighter” Improved interpretation Potential to increase sample size If deemed necessary, spatial correlation can be incorporated into models Primary Sampling Units BBS route segments/year Each BBS route split into 4 segments Secondary Sampling Units Each segments contains 8 stations (spatial replicates) “tighter” refers to increasing the chance that occupancy can be more closely linked with a habitat class. Also, measures of context are more interpretable as a smaller sample unit is buffered. Sample sizes can certainly be doubled (extreme segments). Whether all 4 segments are replicates is still a matter of debate. There are ways to incorporate spatial correlation, including a recent development that deals between-station correlation. It might be applicable if we believe that birds can move between stations within a segment during a survey. Finally, the detection process among spatial replicates is not as confounded as compared to the standard 50 mile BBS route. Reducing such heterogeneity may allow estimating abundance as per Royle and Nichols (2003)—still being evaluated.

Multi-Season Data Framework Local Extinction Colonization Year 1990 2001 2008 Surveys/ sampled rt/ year The MSG project draws from 3 primary landcover dates, not equally spaced. Landcover from 1992 needs to be “cross-walked” to be compatible with 2001 and 2006. The idea is to draw inferences about interval predictions. Examples of inferences that can be drawn from this approach (Pollock’s robust design) will be illustrated using 2001 and 2006. 1 2 ... k8 1 2 ... k8 1 2 ... k8 Closure

Multi-season Models Modeling dynamics or changes in occupancy over time (occupancy as a state variable) Parameters of interest for this work: t = t+1/ t = rate of change in occupancy t = P(absence at time t+1 | presence at t) = patch extinction probability t = P(presence at t+1 | absence at t) = patch colonization probability The multi-season framework will be the “workhorse” in this project. To go beyond “pattern” (static), we need to quantify the dynamics of patch occupancy. This framework offers that capability. Here I list three parameters of interest. Colonization and local extinction probabilities (t to t+1) could be functions of the number of ‘neighboring’ occupied sites in the season t (autologistic function). Nichols et al. (Patuxent researchers) will be developing analytical approach to deal with situations where occupancy of neighboring locations is not known. Nichols highlights some applications: Direct modeling of extinction probability as function of proportion of patches that are suitable: direct estimation of “extinction threshold” relationships Incorporation of succession and habitat alteration into occupancy modeling Changes in extinction/colonization in response to changes in habitat: much stronger inferences than from static habitat-occupancy patterns Multi-species extension permits tests of Caswell-Cohen ideas about predator-mediated and disturbance-mediated coexistence of competing species in patch systems.

Patch Dynamics – Multiple Seasons (Not Ext.) (Occupied) (Ext.) (Col.) Patch dynamics over 3 seasons (years)…using model notation (words parenthetically). The idea is to envision patches ”winking in and out” over time, hopefully, as a function of covariates distilled from the literature review process, consultation with regional biologists (workshops), and ecological theory. (Unoccupied) (Not Col.)

Community Dynamics Multi-species, multi-season Occurrence Local Turnover Probability that a species selected at random from the community in year j was not present in year i Local Extinction or Colonization Probability that a species present in year i is not present in some later year j (i < j) The number of species not present at time i that colonize and are present at time j Rate of change in species richness Ratio of estimated richness in successive time periods Co-occurrence Model local rates of extinction and colonization as functions of occupancy of other species The community-level parameters and inferences are borne from the multi-season framework. Examples of rates of interest are listed in the slide. 11

Essentials… Formulate a priori hypotheses Explicit statements about processes and predictions reflecting knowledge from the literature, expert opinion, and ecological theory Useful to think in terms of the following question: What is the ecological (landscape) basis for sensitivity of a species? Life History What biological process or requirement is (are) a determinant driver of the species’ continued survival?! Habitat classes (states) need to be kept to a minimum and defined keeping in mind: The scale at which it can be modeled and the interplay/relevance to biological processes

Patch Dynamics…and Conservation Design 2001 Landcover covariates 2006 Landcover covariates Landcover Change Analysis Initial set of a priori models Interval predictions and models 2000 2001 2002 2003 2004 2005 2006 2007 2008 E/C1 E/C2 E/C3 E/C4 E/C5 E/C6 E/C7 E/C8 We set ourselves to assess patch dynamics, and hence, address questions regarding predicted impacts and expressions of persistence (also see possible applications as per Nichols on previous multi-season slide). Before analyses begin, a change analysis should be conducted to help us define expectations. Two paths are possible with regards to dynamics. The first would focus around available data on landcover from 2001 and 2006 (we can generalized this approach to include 1992). The second is to use the full time series data set from BBS. This approach is preferred because it “bridges” two intervals defined by landcover data. Benefits: 1) We create an opportunity to identify alternative explanations to account for patch extinction or colonization rates, and 2) foundation to understand “state transition probabilities between intervals”, crucial to predict potential impacts on bird occurrence (ultimately populations) due to changes in landcover predicted by habitat and global climate models. Noteworthy-approach incorporates “history” – Markov process. NOTE: MSG project deals with only 1992, 2001 and 2006 landcover data, but for global climate, LandSat data will enable working with multi-year landcover data over 3 decades (B. Grand new project). Continous landcover data might allow us to develop a more rigorous understanding of transitions probabilities over time intervals of various lengths. Transition probabilities between patch occupancy and changes in habitat “states” Expressions of Persistence

Inferential benefits from time series…    Plausible outcomes and uses of patch dynamics based on all available BBS data between three widely spaced landcovers (illustrated with 2001 and 2006). Correlational - two figures on left: a) an increase in patch occupancy with increasing suitable habitat classes (holding context constant); b) the opposite situation. Hopefully hypotheses will leave us with something more than a mere correlation--strong support for a biologically sound model. The scenario depicted on the top-right figure suggest that something other than habitat is influencing patch occupancy. In other words, other processes are operating. The bottom right figure depicts annual changes in patch occupancy. Insights from its variation are invaluable. For example, variation might be large relative to detectable changes in landcover stemming from a change analysis—not good! If that were not an issue, then the time series helps us discern when the “down-turn” might have started. One could seek out auxiliary information to determine IF the change in landcover, detected in 2006, occurred before. Alternatively, and as a competing explanation, we could assess alternative biological processes (e.g., change in avian community composition). Such analysis also serves to determine whether patterns on one species are a community-wide phenomenon (co-variation). The interpretational benefits of patch dynamics apply to the work on global climate as per Nichols et al. Inferences (e.g., transitional probabilities) will likely be stronger because landsat change analyses over 30 years will generate more continuous landcover data, and hence, a nearly “ideal” match between estimates of patch occupancy and corresponding landcover.  14

Patch Occupancy and Conservation Design Modeling Species-Habitat Relationships Brown-headed Nuthatch Evolved in the southeast Mature, pine forests. Local extinctions. Notable culprits short-age rotations, fire suppression. Poor disperser Secondary Cavity Nester – snags Types of data/information on life history that could be modeled or used to generate a priori hypotheses.

Brown-headed Nuthatch Structural and Functional Covariates Colonization Process Community – Presence of primary Cavity Nesters (+) Inter-Patch Distance – Poor Disperser (-) Inter-Patch Matrix – Distance*Composition Extinction Process Habitat – Mature Pine (-) Fire Rotation – Increasing time since last burn (-) Community – Presence of Cavity Competitors (+) For the nuthatch, a priori models might reflect the interplay between habitat class (type, age) with isolation or distance (due to poor dispersal capabilities), isolation might be further modeled as a function of the type of “matrix” (dominant type - urban or agric. between pine forest patches). Community-level analyses (co-occurrence) can also be invoked as presence/absence of “primary cavity nesters” might be hypothesized to be essential element of nuthatch persistence. Conversely, presence/absence of nest cavity competitors, as suggested in BNA acct., could be a competing model to assess persistence. Some predicted effects in parenthesis.

Patch Occupancy and Conservation Design Predicting the possible impacts of urban growth and global climate are central themes of the project. Projected impacts 5, 10, out to 100yrs Global Climate – Range Dynamics Temperature and landscape changes over past 30 yrs Urban growth – Community Dynamics Past sprawl on the SAMBI area and projected in 2 focal areas

Range Dynamics of North American Landbirds Range Boundaries – transition zones - demographic flux With changing temperatures populations on boundaries are hypothesized to exhibit differing rates of extinction and colonization. Some predictions by regions are: Rate of local extinction: S>EW>N,C Rate of local colonization: N,C>EW>S Spatial Comparisons: Mean change in rate of local extinction: S>EW,C>N Assessment will rely on BBS data (1996-2008) and landcover assessments since the 1970s (landsat). Nichols et al. will lead this work. Benefits extend to the MSG project, both in terms of analytical tools and applications, but also in terms of shared objectives to predict impacts stemming from global climate. (background from Nichols et al.) For selected landbird species, we make specific predictions about locations near northern portions of historical ranges (N), near southern portions of historical ranges (S), near eastern and western portions of historical ranges (EW), and in central areas of historical ranges (C). Some predictions involve differences among these groups of locations during the last 2 decades. Other predictions involve differences in relative changes in dynamic rate parameters between 1966-1985 and 1986-2007 among these groups of locations. 18

Community Dynamics in Urban Landscapes Evokes biological integrity (Karr & Karr 1996) We will frame hypotheses as per life history/functional traits (McGill et al. 2006, Croci et al. 2008). Example from Croci et al. (2008) Urban Adapters Urban Avoiders Forested, shrubs Open landscapes Resident Migratory Longer Life Expectancy Shorter Enclosed nesters Open Widely Distributed Narrower > 2 clutches 1-2 clutches Croci, S., A. Butet, and P. Clergeau. 2008. Does urbanization filter birds on the basis of their biological traits?! Condor 110:223-240. McGill, B. J., B. J. Enquist, E. Weiher, and M. Westoby. 2006. Rebuilding community ecology from functional traits. TREE 21:178-192. McKinney, M. L. 2006. Urbanization as a major cause of biotic homogenization. Biological Conservation 127:247-260. Karr, J. 1996. Karr, J. R. 1996. Ecological integrity and ecological health are not the same. In P.C. Schulze (ed.), Engineering Within Ecological Constraints, pp. 97–109. National Academy Press, Washington, D.C.

Groupings by 4 life history traits…10 species Survival No. Broods Nest Structure Migratory Status An useful approach is to follow the framework and findings outlined by several authors on the previous slide. We could formulate hypotheses regarding the ability of birds to sustain/adapt to continued encroachment and/or habitat alteration based on their life history traits. The dendrogram illustrates how 10 species, 5 migratory and 5 resident, were grouped based on 3 other life history features (arbitrary values for adult survival rates, number of broods per season, and nest structure (enclosed or open cup). We could make predictions about which of the emerging groups will be the “adapter” and which will be the “avoider.” Analyses of turnover rates, species richness, etc. will provide the basis to confront our hypothesized predictions and identify which species fit or not within the hypothesized scheme. 20

Community Dynamics in Urban Landscapes Species Turnover – predictions over past 18 years based on life history…who comes in and leaves? Species Richness – predictions about a “homogenized” avian community—any evidence? Co-occurrence – predictions about functional groups, brood parasite-hosts We will be looking for “thresholds” …patterns in the dynamic process of extinction and colonization under hypothesized prediction about The basic question is not only what’s the “surviving species assemblage”, but why? In other words, what are the life history attributes of the “winners” as the landscape change. In other words, test whether Psi (constrained patterns) have anything to do with hypothesized predictions based on life history traits. What are colonization or extinction rates as a function of road density, edge index, number of forested patches (frequency), average size of forested patches, distance among forested patches, and matrix (proportion of hostility-dominated by urban). As benchmark, we could frame hypotheses as per Croci et al. (2008).

Summary of Expected Results Patch Occupancy and Conservation Design Summary of Expected Results Understand interplay between pattern and process in the SAMBI region, Understand range and community dynamics of avian species in Southern United States, Enable the development of decision-support tools for conservation design Incorporate transition probabilities between patch occupancy and habitat “states” given projected changes on landscapes 22