Modeling Invasive Species Exposure at Adirondack Park Trailheads. Dan Rockefeller

Size: px
Start display at page:

Download "Modeling Invasive Species Exposure at Adirondack Park Trailheads. Dan Rockefeller"

Transcription

1 Modeling Invasive Species Exposure at Adirondack Park Trailheads Dan Rockefeller

2 Overview Invasive Species and Invasion Process The Role of Humans Invasive Species Management and Risk Assessment Objectives Methods Study Area Study Species Datasets Model Generation Data Analysis Mapping Results Discussion and Conclusions

3 Invasive species Species that when introduced to areas outside their native range by human activities, establish a selfsustaining population and spread, having significant negative impacts on biodiversity as well as human well-being. -Lockwood, Hoopes, & Marchetti, 2007 Impacts may include: Water quality Nutrient cycling Wildlife habitat Fire regimes Threatened and endangered species Industry (fisheries, agriculture, forestry, recreation) 3

4 Number of species The Invasion Process Ecological Uptake Native Range Barriers Ecological Transport Transport Pathway Barriers Establishment Recipient Ecosystem Ecological Barriers Spread

5 The Role of Humans Rapid movement over vast distances Globalized marketplace Intentional vs. Unintentional transport

6 The Role of Humans Rapid movement over vast distances Globalized marketplace Intentional vs. Unintentional transport Anthropogenic vs. Natural disturbance Photo Credit: Sam Beebe, Ecotrust

7 The Role of Humans Rapid movement over vast distances Globalized marketplace Intentional vs. Unintentional transport Anthropogenic vs. Natural disturbance NOT just the built environment Photo Credit: Friends of Brushy Hills, friendsofbrushyhills.wordpress.com Photo Credit: Steven Flint, adkinvasives.com Photo Credit: Chris Evans, River to River CWMA, Bugwood.org

8 The Role of Humans Rapid movement over vast distances Globalized marketplace Intentional vs. Unintentional transport Anthropogenic vs. Natural disturbance NOT just the built environment Nature-based tourism as dispersal vector Photo Credit: Trees that Please Nursery Blog 9/1/2012 Photo Credit: Southern Illionois University Photo Credit: PGT Nature Garden 9/2012

9 The Role of Humans Rapid movement over vast distances Globalized marketplace Intentional vs. Unintentional transport Anthropogenic vs. Natural disturbance NOT just the built environment Nature-based tourism as dispersal vector Management Photo Credit: Paul Rischmiller, adkinvasives.com/management

10 Invasive Species Management Labor intensive Expensive ($138 billion nationally per year) Prevention is the preferred management strategy Difficult Requires accurate risk assessments/edrr Image Credit: The invasion curve. Adapted from Invasive Plants and Animals Policy Framework, State of Victoria, Department of Primary Industries, (edis.ifas.ufl.edu)

11 Invasive Species Management: A Coordinated NYS Effort Legislation NYSISC (DEC, DAM) NYSISAC NYSIS Management Strategy Prevention/EDRR Partnerships/Coordination Partnerships for Regional Invasive Species Management (PRISM) Resource managers, NGOs, industry, state agencies, citizens Coordinating volunteers, education and outreach, EDRR, direct management of invasive species

12 Invasive Species Management: A Coordinated NYS Effort imapinvasives New York Natural Heritage Program Online IS database and mapping tool Trained citizens, professionals, scientists can collect and share spatially explicit data in real time Leverages the potential of many more eyes on the ground Image Credit: imapinvasives.org

13 Invasive Species Management: Adirondack Park and APIPP Infestations largely along roadways and waterbodies Recreation and tourism may be unintentionally facilitating invasive species spread APIPP has been coordinating IS management since 1998 Ultimate goal: Prevent the spread of invasive species 13

14 Invasive Species Management: Risk Assessment Risk assessments provide support for management decisions Allocating resources Target species Target locations Many focus on identifying species that may become invasive based on species traits Australian Weed Risk Assessment (WRA) program Typically include: species traits/habitat suitability and exposure assessments RISK = HABITAT SUITABILITY + EXPOSURE 14

15 Invasive Species Management: Exposure Analysis RISK = HABITAT SUITABILITY + EXPOSURE Extent and pattern of contact between source population and recipient ecosystem Ballast water exchange Well documented in shipping records Vital for the establishment of early detection systems and preventative management actions 15

16 Objective Characterize risk of exposure to invasive plant species at recreational trailheads (entry points) across the Adirondack Park by modeling relationship between visitors and invasive species occurrences. Aligned with APIPP Strategic Plan (2013) objectives: study of distribution trends related to vectors (i.e. hikers) use of modeling to predict where species are most likely to spread To accomplish this: Used several model forms to create indices of exposure for 7 terrestrial invasive species Produced risk maps using these indices 16

17 17

18 Methods Photo Credit: Lou Wagner, Umass Boston 18

19 Study Area: The Adirondack Park ~ 6 million acres ~ 48% Forest Preserve/52% Private 103 Municipalities 132,000 permanent residents 200,000 seasonal residents Within days drive of 85 million people ~300 trail registers 1,500 miles trail (nonmotorized) 840 miles snowmobile trail 5,000 miles public road Photo Credit: Dan Rockefeller

20 Study Species Photo Credits: DEC, DAM, NYS Prohibited and Regulated Plants

21 Study Species APIPP Priority APIPP Priority APIPP Priority Photo Credits: DEC, DAM, NYS Prohibited and Regulated Plants

22 Study Species APIPP Priority APIPP Species of Concern APIPP Priority APIPP Species of Concern APIPP Priority APIPP Species of Concern APIPP Species of Concern APIPP Species of Concern Photo Credits: DEC, DAM, NYS Prohibited and Regulated Plants

23 Datasets: imapinvasives All record types retained All records assumed accurate (ID & spatially) Records outside NYS removed Species Original Records Retained Records C. stoebe ssp. micranthos Cynanchum spp. 2,304 2,290 L. morrowii L. salicaria 3,554 2,864 M. vimineum 3,798 3,796 P. sativa R. cathartica 1,279 1,274 Total 13,168 12,392

24 Datasets: imapinvasives

25 Datasets: Trail Registry Database (TReD) Spatially referenced visitor information Origin (State/City) Destination Date Group size Length of stay Photo Credit: Abigail Larkin, TReD Project Coordinator Original Dataset Users from across globe 212 Trail Registers 153,432 records/428,275 users Photo Credit: Abigail Larkin, TReD Project Coordinator Filtered Dataset Only NYS users considered 21,971 records/61,812 users

26 Datasets: NYS Municipality Layer Alterations to match TReD Duplicate town names = Town + County Nassau & Suffolk County Municipalities merged = Long Island NYC boroughs merged = New York Saratoga/Saratoga Springs Towns adjacent to city of same name merged Original Data Layer = 994 Municipalities Altered Data Layer = 954 Municipalities

27 Methods: Modeling Approach

28 Methods: Modeling Approach Model Set A (16 models) Geometry (municipality) Raw town geometry Buffered town geometry (14.32 km) Geometry (IS observation) Raw IS point geometry Buffered IS point geometry (12.5m) Observation density Observations/km 2 Timing of visitation All months Months of August-November (Visitors per town X Observations per town) Log transformed Model Set B (4 models) Geometry (PRISM) Raw PRISM geometry Geometry (IS observation) Raw IS point geometry Buffered IS point geometry (12.5m) Timing of visitation All Months Months of August-November (% Visitation per PRISM X % total species observations per PRISM) Log transformed

29 Methods: Model Generation Set Model # Base Geometry Point Buffer (12.5m) Polygon Buffer (14.32km) Months 8-11 Set A 1 Town 2 Town X 3 Town X 4 Town X X 5 Town X 6 Town X X 7 Town X X 8 Town X X X 9 Town X 10 Town X X 11 Town X X 12 Town X X X 13 Town X X 14 Town X X X 15 Town X X X 16 Town X X X X Set B 17 PRISM 18 PRISM X 19 PRISM X 20 PRISM X X Observation Density

30 Methods: Sample Code

31 Methods: Analysis and Mapping Model Selection One-Way ANOVA (Tukey s HSD) aggregated model scores Between-model Spearman Rank Correlation Analysis Analysis of Selected Models Between-species Principle Components Analysis and Spearman Rank Correlation Analysis Mapping/Cartography Normalized scores to scale of 0-1 Trailheads displayed as colored points Green = Low Risk, Yellow = Moderate Risk, Red = High Risk Hot Spot Analysis Getis-Ord Gi* - measure of spatial autocorrelation Identifies probable clusters of high and low scoring trailheads Based on these results, Models 1 and 17 selected for further analysis 31

32 Methods: Model Generation Set Model # Base Geometry Point Buffer (12.5m) Polygon Buffer (14.32km) Months 8-11 Set A 1 Town 2 Town X 3 Town X 4 Town X X 5 Town X 6 Town X X 7 Town X X 8 Town X X X 9 Town X 10 Town X X 11 Town X X 12 Town X X X 13 Town X X 14 Town X X X 15 Town X X X 16 Town X X X X Set B 17 PRISM 18 PRISM X 19 PRISM X 20 PRISM X X Observation Density

33 Methods: Analysis and Mapping Model Selection One-Way ANOVA (Tukey s HSD) aggregated model scores Between-model Spearman Rank Correlation Analysis Analysis of Selected Models Between-species Spearman Rank Correlation Analysis and Principle Components Analysis Mapping/Cartography Normalized scores to scale of 0-1 Trailheads displayed as colored points (quantiles) Green = Low Risk, Yellow = Moderate Risk, Red = High Risk Hot Spot Analysis Getis-Ord Gi* - measure of spatial autocorrelation Identifies probable clusters of high and low scoring trailheads Based on these results, Models 1 and 17 selected for further analysis 33

34 Results Photo Credit: Steven Flint, adkinvasives.com 34

35 One-Way ANOVA (Tukey s HSD) Model 1 Mean (Scaled) Model 17 Mean (Scaled) Lythrum A 2.92 (0.61) Lythrum A 3.20 (0.85) Centaurea A B 2.73 (0.57) Centaurea A 3.16 (0.84) Cynanchum B C 2.47 (0.52) Rhamnus B 3.02 (0.80) Rhamnus B C 2.46 (0.52) Pastinaca B 2.96 (0.78) Lonicera C 2.35 (0.49) Cynanchum C 2.85 (0.75) Pastinaca C D 2.18 (0.46) Lonicera C 2.84 (0.75) Microstegium D 1.91 (0.40) Microstegium D 2.10 (0.55) Species that do not share a letter are significantly different. 35

36 Results: Spearman Rank Correlation Analysis Model 1 Strong, positive betweenspecies Spearman rank correlations (coefficients >.780) Strong, positive Spearman rank correlation with total trail visitation (coefficients >.806) Model 17 Between-species Spearman rank correlation varied C. stoebe, Cynanchum = C. stoebe, R. cathartica = 0.735

37 Results: Principle Components Analysis Model 1 Model 17 37

38 Exposure Scores (Model 1)

39 Exposure Scores (Model 1) High Peaks Management Unit Lake George Management Unit

40 Exposure Scores (Model 1)

41 Exposure Scores (Model 1) High Peaks Management Unit Lake George Management Unit

42

43 Exposure Scores (Model 1) 25 trail registers in highest exposure quantile for all species 14 located within HP and LG Management Units Management Unit (State land only) Trail Register

44 Exposure Scores (Model 17)

45 Exposure Scores (Model 17) Highlights trail registers that receive a large percentage of use from regions (PRISMs) where a given species is prominent

46 Exposure Scores (Model 17)

47 Exposure Scores (Model 17) Highlights trail registers that receive a large percentage of use from regions (PRISMs) where a given species is prominent

48

49 Discussion Photo Credit: Bill Johnson, nps.gov 49

50 Discussion: Overview Objective: provide assessment of IS exposure risk for Adirondack Park trailheads by modeling the relationship between visitors and statewide invasive species occurrences. Assessments like this allow managers to make targeted, and justifiable management decisions. Previous studies have not combined visitor and IS location information to assess exposure across a landscape-scale network of recreational access points 50

51 Discussion: Management Implications Model 1 suggests management focus on frequently traveled trails Model 17 highlights potential species-specific focal areas Proactive Management Species-specific management plans (geographic/species prioritization) Photo Credit: Adirondack Mountain Club, Adirondack Explorer High risk trailheads and focal areas Which species to monitor for? High exposure species in Park OR low exposure species NOT in Park? Boot-brushing stations ADK Mountain Club Maps Regionally/species specific Targeted public education program (similar to Summit Steward or Boat Launch Steward programs) 51

52 Limitations NYS users ONLY Did not account for possible contact with IS along route to Adirondacks Did not consider differences among trail registers Did not consider differences in species dispersal characteristics

53 Conclusions Hikers may be an important dispersal vector Trailheads are access points for hikers and potential invaders This research may be used to support decision making High risk trailheads Species-specific priority areas Models may be modified for other recreational access points Campgrounds Photo Credit: Abigail Larkin, TReD Project Coordinator Boat launches 53

54 Acknowledgements Colin Beier (ESF) Chris Nowak (ESF) Jennifer Dean (NYNHP) Brendan Quirion (APIPP) NYSDEC Division of Lands and Forests Great South Woods Regional Planning Project The imapinvasives Team O Brien and Gere Abbie Larkin Jesse Caputo Sara Velardi 54

55 Thank You! Please Register 55