Analysis

Data Integration and Data Mining
Ecological Research Design
GIS Analysis
Statistical Analysis

Data Integration and Data Mining

Designed and implemented a data warehouse to contain environmental data common to multiple research projects at Jug Bay Wetlands Sanctuary, a component of the Chesapeake Bay National Estuarine Research Reserve system. The initial data mining applications are to examine factors associated with salamander migration and box turtle movements. Intuitively the weather conditions are a factor in predicting movement of the animals, but the exact relationship and the specific factors are unknown.

Data in the warehouse includes weather reports at varied intervals from four different sources/locations, three types of water quality measurements from varied locations and sources, tide magnitude and timing, times of sun and moon rise and set, phase of the moon and spatial locations of research sites and data sources. Enhanced techniques are being developed to identify data problems. An easy to use method for entering or downloading current data has been implemented.

Photo of marbled salamander, Ambystoma opacum

Marbled salamander Ambystoma opacum that started the data warehouse project.

Ecological Research Design

Jeff has been significantly involved in research design issues for two projects.

Amphibians

Vernal pool survey

He led the re-design of the 2008 vernal pool survey at Jug Bay Wetlands Sanctuary to collect data on egg masses and tadpoles/larvae abundance. The study site is a large (up to 1.8 ha) vernal pool which is too large to survey from the perimeter as many published protocols describe. The new design uses a carefully specified procedure along random transects to support statistically valid estimates of the population of the entire pool. This new protocol along with supporting materials were presented as a poster at the Maryland Water Monitoring Council Vernal/Seasonal Pool Workshop in March 2008. The poster and related materials are also available here.

Plants

Wild rice experiment

He was a core participant in a 2007 controlled experiment to measure the impact of geese grazing on wild rice (Zizannia aquatica) productivity. He participated in the initial set up of the experimental plots and periodic measuring of plant density and height. He performed on-going statistical analysis to verify that validity was being maintained (e.g. that there was no statistically significant difference between the control and treatment areas before the treatments were performed and that there was no significant difference between the measurements obtained by the various people assisting with the project). He played an active role in the redesign of the experiment as unexpected field conditions reduced the feasibly of collecting the data that was initially planned.

GIS Analysis
	Jeff has obtained a wide variety of GIS layers to provide geospatial context for the Jug Bay data warehouse. Initial GIS work has included comparison of wild rice locations to historical river shorelines, mapping of invasive plants as part of a monitoring and removal success study, and a time series of vernal pool boundaries for the 2007 and 2008 years. GIS analysis includes delineation of subwatershed, particularly the area draining into the vernal pool. Other work with the digital elevation data clearly illustrated the presence of deltas associated with small stream valleys producing a small increase in marsh elevation (with corresponding vegetation changes).
GPS track data (white dots) shows wild rice currently growing in marsh that the historical shoreline data shows did not exist in 1944 but has been present since at least 1993.

Statistical Analysis

In addition to the on-going statistical analysis of the wild rice data described above, Jeff has performed various types of statistical analysis.

A dry period in 2008 prevented the vernal pool from filling until flooding rains occurred in May. This was long after wood frog and spotted salamander mating season, so there was no egg mass data to analyze. However, the heavy rains filled the pool unusually full and the water depth was measured periodically using the random transects in addition to GPS tracks of the perimeter. Analysis of this data to estimate the pools volume and the frequency distribution of depths is currently underway and will be completed before the 2009 season..
He assisted with the home range analysis of mud turtle movement.
A pending update to the fish survey database will include analysis of the fish index of biotic integrity.
He frequently performs statistical analysis (typically ANOVA) of data from his human-computer interaction research studies, e.g. a study of the usability of collaborative GIS work presented as a poster at the American Geophysical Union 2008 Fall Meeting.