Author: Grant McKown (james.mckown@unh.edu)
Purpose: Conduct the univariate analysis portion of the broader NAMASTE project to identify and visualize trends in vegetation communitiy of the tidal marshes in National Estuarine Research Reserves. R code is broken down into scripts that tackle portions of each analysis. See below for Analysis Overview and Script Overview.
Analysis Overview
-
The univariate analysis was approached from three unique directions: (1) National Analysis over Time, (2) Compilation analysis of site-level regressions, and (3) Regressions against select site metrics
-
Vegetation metrics that were included in the analyses were: (1) Abiotic Cover, (2) Live Cover, (3) Halophyte Cover, (4) Freshwater Cover, (5) Ecotone Migration Index, (6) Salt Ratio, (7) Species Richness, and (8) Shannon-Weiner Diversity
-
Vegetation metrics were calculated for each plot for each monitoring season.
-
Site characteristics that were classified for each site were: (1) Region, (2) Salinity regime, (3) Tidal range, and (4) Geomorphology
-
For the national time analysis, the plot-level dataframe was averaged by (1) Vegetation Zone and (2) Site to conduct analysis at the vegetation zone or site-level.
- Mixed linear regressions (Site nested within Region as random factors) were conducted with Time as a fixed factor and either Vegetation Zone or Site Characteristics as the second fixed factor. Interactions were always included in models.
- Additionally, a general national mixed linear model for each vegetation metric was conducted with just Time as a fixed factor.
-
For the compilation analysis, the simple linear regressions were conducted for each site. The compilation analysis differed between Ecotone Migration Index and the other vegetation metrics.
- For the EMI, compilation scores were caclulated accoring to the figure to the right and were based on the magnitude of the p-value and direction of the slope
- For the other vegetation metrics, the slopes of the site-level regressions (whether significant or not) were averaged +/- 1 standard error by vegetation zone and site characteristics.
- Additionally, the percentage of sites with significant regressions for each vegetation metric were calculated and visualized.
-
For the third analysis, national linear regressions were conducted on site-level regression slopes and two selected site metrics.
- Sea level rise (last 19 yrs) and the NERRS Landscape Condition were pre-selected as potential drivers or explainations for changes in vegetation community
- Slopes were generated by Kim Cressman (Catbird Stats) and may be different from slopes calculated in the compilation analysis's site-level regressions
- ANOVA tables and slopes were generated and reported.
Script Overview
Script 1 - Dataset Formatting: Import of the plot-level data and summarize by site or vegetation zone for future analyses
Script 2 - Site level regressions - EMI Compilation Analysis: Conduct site-level regressions to calculate p-values and slopes. Calculates compilation analysis scores of Ecotone Migration Index.
Script 3: Site level regessions & Compilation - Conducts site-level regressions on remaining vegetation metrics, calculates average +/- standard error of slopes and percent of sites with significant regressions.
Script 4: National Regressions - Vegetation Zone: National time mixed linear models with vegetation zone as the unit of replication.Models include vegetation zone, time, and interaction as fixed factors.
Script 5: National Regressions - Site Characteristics: National time mixed linear models with Site as the unit of replication. Models consider site characteristics, time, and interaction as fixed factors.
Script 6: National Regressions - Whole Study: National time mixed linear models with Site as the unit of replication. Models consider time as only fixed factor.
Script 7 - National Slope Regressions - SLR & Landscape Condition: National regressions of vegetation metric slopes from site-level regressions against Sea Level Rise and NERR Landscape Condition. Slope is unit of replication.
Script 8 - National Slope Regressions - Veg Zone - SLR & Landscape Condition: National regressions of vegetation metric slopes from site-level regressions against Sea Level Rise and NERR Landscape Condition. Vegetation Zone is unit of replication.
Script 9 - Vegetation Metrics Correlation Plot: Creation of correlation plot and matrix between vegetation metrics with the plot level detaframe.