Extinction Risk and Probability of Decline as Metrics for Ranking Conservation Priority Species

Size: px

Start display at page:

Download "Extinction Risk and Probability of Decline as Metrics for Ranking Conservation Priority Species"

April Golden
5 years ago
Views:

1 Extinction Risk and Probability of Decline as Metrics for Ranking Conservation Priority Species Jessica C. Stanton & Wayne E. Thogmartin US Geological Survey Upper Midwest Environmental Sciences Center La Crosse, Wisconsin

2 Performance Indicators Critical for diagnosing problems, monitoring progress, and measuring success A good performance indicator should: Measure something important Be standardized across time (repeatable) Be applicable at different spatial and temporal scales Be easily comprehensible Be easily generated on a regular basis Have sound conceptual basis Be objective and data driven Improve communication

3 Performance Indicators For North American avifauna, metrics based on extinction risk meet all of these criteria Measure something important Be standardized across time (repeatable) Be applicable at different spatial and temporal scales Be easily comprehensible Be easily generated on a regular basis Have sound conceptual basis Be objective and data driven Improve communication

4 Extinction Risk Population Size Population Trend Variability

5 Partners in Flight The mission of PIF is three-fold: Helping species at risk Keeping common birds common Developing partnerships for birds, habitat, and people

6 North American Landbird Conservation Plan Regional Concern Scores based on 5 factors(1 [low] to 5 [high]) Population Size Breeding Distribution Relative Density (in region) Regional Threats Regional Population Trend Regional Concern Score = PS + BD + RD + TB_r + PT_r Max score = 20

7 North American Landbird Conservation Plan Regional Extinction Risk Population Size Population Trend Variability Regional Concern Score = PS + BD + RD + TB_r + PT_r Max score = 20

8 Role of Variability and Risk Even increasing populations are at risk Variability Each of these populations have overall trends of +5% growth/year

9 Monitoring populations through time Population-level processes (births, deaths, immigration, emigration) State of Population through time Year-to-year variability STATE PROCESS Observation Error Observation of population through time OBSERVATION PROCESS

10 North American Breeding Bird Survey Provides annual abundance observations for >420 species since 1966 Conducted on >4,000 routes distributed across the United States and Canada Summarized as abundance indices at various spatial scales

11 Sources of observation error Trend models accommodate observer differences Remaining potential sources include: Road traffic noise (Griffith et al. 2010) Day-of-year differences (Jones-Farrand et al. 2011) Time-of-day differences (Robbins 1981, Rosenberg and Blancher 2005) Changing land cover These factors add imprecision and potential bias in estimates of abundance and trend AND to measures of risk

12 How do we remove observation error? Population trend (u) Year-to-year variability (Q) State of Population through time (n t ) STATE PROCESS Observation Error (R) BBS Abundance Indices OBSERVATION PROCESS

13 Auto-Regressive State-Space Model We decompose, on a log scale, the annual abundance indices x t = ln(n t ) x t = x t-1 + u + w t where w t ~ MVN(0, Q) y t = x t + a + v t where v t ~ MVN(0, R) STATE PROCESS OBSERVATION PROCESS Observation error (R) held constant for each species across BCR s Population variability (Q) and trend (u) calculated for each BCR calculated for all diurnal terrestrial species at the resolution of Bird Conservation Regions (BCR) if the species was adequately sampled*

14 Auto-Regressive State-Space Model STATE OBSERVATION Image by khyri

15 Auto-Regressive State-Space Model STATE OBSERVATION

16 Extinction Risk Projection Using the results of the STATE PROCESS for each BCR, a simple population projection was estimated Projection based on Dennis et al. (1991) diffusion approximation x t = x t-1 + u + v t ; where v t ~ MVN(0, σ 2 ), σ 2 = R Calculate risk according to: Pr(decline to threshold) t = π(u) φ((-x d + u t)/(σ t)) + exp(2x d u /σ 2 ) φ((-x d - u ) t)/(σ t) Predicting future state requires unbiased estimates of Population Size, Trend, and Variability

17 Extinction Risk Projection Using the results of the STATE PROCESS for each BCR, a simple population projection was estimated Projection based on Dennis et al. (1991) diffusion approximation Two risk measures calculated: 50% Decline Probability population will decline to half of 2012 abundance index value Projected 20 years in future Quasi-Extinction Probability population will decline to index value of 0.01 Projected 50 years in future

18 Extinction Risk Projection Horned Lark, BCR 14

19 Extinction Risk Projection Horned Lark, BCR 6

20 Results Risk metrics calculated for 305 species over 33 BCRs for a total of 3,238 species x BCR combinations 29% of species x BCR combinations had significantly * negative population trends 110 species are predicted to decline to half of current abundance (with greater than 50% probability) in at least half of the BCR s they currently occupy 24 species have a greater than 20% probability of reaching the quasi-extinction threshold in at least one BCR currently occupied within the next 50 years

21 If historical patterns persist into the future, by the end of the century, 9 species may no longer be available for monitoring Species Lost from BBS

22 Near-term Declines of 30% Based on the trends of last decade ( ), 20 species expected to decline another 30% in the near-term

23 Risk of QE vs. PIF scores 1 Probability of Quasi-Extinction within 50 years PS + BD + RD + TB_r + PT_r

24 Underscored relative to risk

25 Half-life vs. PIF scores 1.0 Probability of 50% decline within 20 years PIF Regional Concern Score

26 Half-life vs. PIF scores

27 Performance Indicators Extinction Risk Metrics: Measure something important Can be standardized across time (repeatable) Are applicable at different spatial and temporal scales Are easily comprehensible Are easily generated on a regular basis Have sound conceptual basis Are objective and data driven Can improve communication

28 Potential Uses Incorporation into PIF regional and range-wide assessment scores Detection of regional patterns of multiple-species decline patterns

29 Potential Uses

30 Potential Uses Incorporation into PIF regional and range-wide assessment scores Detection of regional patterns of multiple-species decline patterns Detection of habitat or feeding guilds showing high collective risk patterns

31 Grassland Birds Potential Uses

32 Potential Uses Incorporation into PIF regional and range-wide assessment scores Detection of regional patterns of multiple-species decline patterns Detection of habitat or feeding guilds showing high collective risk patterns Tracking and monitoring progress of conservation activities

33 Acknowledgments Funding through the DOI Birds Forever Initiative The many dedicated volunteers who collect data for the North American Breeding bird survey J.R. Sauer, P.C. McKann, T. Will, and B.X. Semmens for consultation, assistance, and advice Image by Juan Pons

Conservation Objectives

Conservation Objectives Overall Conservation Goal: Sustain the distribution, diversity, and abundance of native landbird populations and their habitats in Ontario's Bird Conservation Regions High Level