In the Weeds: Mapping Change, Managing for Resilience
Utilizing Stream Weed Surveys for Rapid Assessment and Monitoring in Mauka Watershed Management Areas on Kauaʻi
Protecting the biocultural resources of native forests for resilience in the face of climate change and invasion from introduced species is complex and often requires novel approaches. Quantifiable metrics of forest health and habitat change are extremely valuable for conservation prioritization and as measures for effective management actions. Streams may serve as corridors of ingress for habitat-modifying weeds like strawberry guava (Psidium cattleyanum), Himalayan ginger (Hedychium gardnerianum) and Australian tree fern (Cyathea cooperi) and thus also as indicators of looming changes in habitat health.
Anecdotal detection of these and other weed species along the streams of the Alakaʻi Plateau has been common, though no systematic sampling had been conducted. To aid on-going watershed protection work, The Nature Conservancy’s Kauaʻi Forest Program collected detailed baseline information for streams across newly-fenced and planned fenced areas. Field staff conducted thorough surveys of ten stream stretches totaling 26.8 kilometers of linear stream corridor. Analysis of individual weed species was completed in ESRI ArcGIS using the kernel density tool and a 20-meter buffer to spatially delimit weed density along streams.
Results showed a general pattern of decreasing weed densities with increased elevation along the streams. This provided repeatable metrics that can be used to measure habitat changes by area and evaluate management actions, and the analysis illuminated areas of high density for immediate weed control work by field staff. In addition to the area-specific results, this presentation will focus on providing the field methodology and analysis steps required for implementation by practitioners with similar projects.
INTRODUCTION
Protecting the biocultural resources of native forests for resilience in the face of climate change and invasion from introduced species is complex and often requires novel approaches. Quantifiable metrics of forest health and habitat change are extremely valuable for conservation prioritization and as measures for effective management actions. Streams may serve as corridors of ingress for habitat-modifying weeds like strawberry guava (Psidium cattleyanum), Himalayan ginger (Hedychium gardnerianum) and Australian tree fern (Cyathea cooperi) and thus also as indicators of looming changes in habitat health.
Anecdotal detection of these and other weed species along the streams of the Alakaʻi Plateau has been common, though no systematic sampling had been conducted. To aid on-going watershed protection work, The Nature Conservancy’s Kauaʻi Forest Program collected detailed baseline information for streams across newly-fenced and planned fenced areas. Field staff conducted thorough surveys of ten stream stretches totaling 26.8 kilometers of linear stream corridor. Analysis of individual weed species was completed in ESRI ArcGIS using the kernel density tool and a 20-meter buffer to spatially delimit weed density along streams.
Results showed a general pattern of decreasing weed densities with increased elevation along the streams. This provided repeatable metrics that can be used to measure habitat changes by area and evaluate management actions, and the analysis illuminated areas of high density for immediate weed control work by field staff. In addition to the area-specific results, this presentation will focus on providing the field methodology and analysis steps required for implementation by practitioners with similar projects.
.png)
Field Data Collection
.png)
- We divided the streams into sections and developed a datasheet to use as we walked the stream with our GPS. We took points on priority weed species, either as number of mature and immature or as square meters of cover, visible within the stream corridor (between 5 to 10 meters on both sides).
- We also took observations of other non-native plants and signs of ungulate disturbance for use in an overall description.
- Weed points were entered into a single shapefile and a “POPULATION” field was added to equal the sum of immature and mature, or area.
Office Data Processing
- Shapefiles were made from each weed to run kernel density (i.e. psicat, cyacoo and hedgar)
- We made classification thresholds with 5-6 classes, keeping in mind that a zero value set as hollow eliminates non-values in spatial representation.
- We then created a 20 meter buffer around the stream survey shapefile
- We used the raster clip tool for each of the kernel density layers, clipping it with the 20 meter stream buffer
- Weed points were clipped with the 20 meter stream buffer to isolate only points within the stream corridor, and were named to distinguish them from the complete set of points (strm_psicat, strm_cyacoo and strm_hedgar); and then we ran the kernel density again.
.png)
Compare the CyaCoo density clip from the stream corridor points (image 1) with the clip from all the CyaCoo points (image 2). Note density value differences represented with colors.
.png)
.png)
METHODOLOGY
Field Data Collection
- We divided the streams into sections and developed a datasheet to use as we walked the stream with our GPS. We took points on priority weed species, either as number of mature and immature or as square meters of cover, visible within the stream corridor (between 5 to 10 meters on both sides).
- We also took observations of other non-native plants and signs of ungulate disturbance for use in an overall description.
- Weed points were entered into a single shapefile and a “POPULATION” field was added to equal the sum of immature and mature, or area.
Office Data Processing
- Shapefiles were made from each weed to run kernel density (i.e. psicat, cyacoo and hedgar)
- We made classification thresholds with 5-6 classes, keeping in mind that a zero value set as hollow eliminates non-values in spatial representation.
- We then created a 20 meter buffer around the stream survey shapefile
- We used the raster clip tool for each of the kernel density layers, clipping it with the 20 meter stream buffer
- Weed points were clipped with the 20 meter stream buffer to isolate only points within the stream corridor, and were named to distinguish them from the complete set of points (strm_psicat, strm_cyacoo and strm_hedgar); and then we ran the kernel density again.
Compare the CyaCoo density clip from the stream corridor points (image 1) with the clip from all the CyaCoo points (image 2). Note density value differences represented with colors.
.png)
.png)
.png)
.png)
.png)
.png)
.png)
HALEHAHA RESULTS
.png)
.png)
.png)
.png)
.png)
.png)
.png)
DRINKING GLASS RESULTS
Method and Analysis
Density classes can be shared across all units (image 1) or tailored to each unit (image 2) depending on what you are trying to compare. You can also use square meters instead of individuals, as with blackberry; or use this method to quantify ungulate disturbance through points (scat) or square meters (wallows and trails).
.png)
.png)
Management Implications
Location and density through the surveyed units support our plan of general suppression for Australian tree fern and strawberry guava, and eradication of Himalayan ginger. Morella faya, carried along streams, will continue to be spot treated in stream corridors, and benign non-natives will be tolerated and monitored.
Further Development:
This process could be adapted into a GIS tool that doesn’t require the user to set the symbology or run the density. You could input the location of the data file, choose a buffer, specify the weed species and thresholds and automatically generate a map. We plan to repeat these surveys every 3 to 5 years and monitor changes. Data collected over time can help guide and evaluate our management practices, provide answers to questions such as the influence of deer browsing on non-native species like blackberry, and quantify the effects of ungulate removal on disturbance colonizers and native plant regrowth.
DISCUSSION
Method and Analysis
Density classes can be shared across all units (image 1) or tailored to each unit (image 2) depending on what you are trying to compare. You can also use square meters instead of individuals, as with blackberry; or use this method to quantify ungulate disturbance through points (scat) or square meters (wallows and trails).
Management Implications
Location and density through the surveyed units support our plan of general suppression for Australian tree fern and strawberry guava, and eradication of Himalayan ginger. Morella faya, carried along streams, will continue to be spot treated in stream corridors, and benign non-natives will be tolerated and monitored.
Further Development:
This process could be adapted into a GIS tool that doesn’t require the user to set the symbology or run the density. You could input the location of the data file, choose a buffer, specify the weed species and thresholds and automatically generate a map. We plan to repeat these surveys every 3 to 5 years and monitor changes. Data collected over time can help guide and evaluate our management practices, provide answers to questions such as the influence of deer browsing on non-native species like blackberry, and quantify the effects of ungulate removal on disturbance colonizers and native plant regrowth.
Support for this work, as well as continued comprehensive protection of mauka watershed forests on Kauaʻi, came from a number of sources.
Mahalo to the field crew Marcela Brimhall, Nicolai Barca, Andrea Rey and Cody Statler for their work collecting stream survey data, to Marcela Brimhall and Andrea Rey for iterating endlessly on the kernel density analysis, and to Alicia Hedlesky for pulling it all together.
ACKNOWLEDGEMENTS
Support for this work, as well as continued comprehensive protection of mauka watershed forests on Kauaʻi, came from a number of sources.
Mahalo to the field crew Marcela Brimhall, Nicolai Barca, Andrea Rey and Cody Statler for their work collecting stream survey data, to Marcela Brimhall and Andrea Rey for iterating endlessly on the kernel density analysis, and to Alicia Hedlesky for pulling it all together.
