Fig 1.
Map shows the study area in Arizona and New Mexico, displaying boundaries of Ecological Management Units defined in USFWS (2012) and the extent of forested lands (National Land Cover Database). Inset shows the location of the study area relative to the range of the Mexican spotted owl within the U.S.
Table 1.
Desired minimum conditions for six structural attributes of Mexican spotted owl nesting/roosting habitat.
Fig 2.
Predicted-to-expected (P/E) ratio curve.
The P/E ratio curve [39] was used to evaluate, calibrate, and reclassify the forest vegetation cover type spectral similarity index model into four map classes. The solid black line is the mean P/E curve from bootstrapped replicates and dashed lines represent the 95% confidence intervals.
Fig 3.
Forest structural attributes of cover type spectral similarity map classes.
Dashed gray lines represent minimum desired conditions from Table 1. Height of bars represents mean and 95% confidence intervals are shown in error bars.
Table 2.
Results of pairwise comparisons between structural and compositional attributes and similarity class.
Fig 4.
Forest species composition of cover type spectral similarity map classes.
Height of bars represents mean and 95% confidence intervals are shown in error bars.
Fig 5.
Distribution on FIA plots by forest type and similarity class.
Numbers in columns indicate plot count of the specific forest type. Mixed-conifer and pine-oak were defined using descriptions in USFWS 2012. Ponderosa pine and Other were defined by FIA methodologies.
Fig 6.
Temporal trends in mapped spectral similarity classes over a 35-year period.
Dashed line shows a linear fit over the study period with equation and R2 values shown in each panel.
Fig 7.
“Bookend” changes in Mexican spotted owl cover type over a 35-year period.
The distribution of cover type classes for 1986 and 2020 are shown on the left, and the differenced map showing spatial changes in cover type classes over the study period is shown on the right. Cover type mapping is masked to limits of potential forested land according to the National Land Cover Database.
Fig 8.
Sankey diagram showing flow of cover type spectral similarity classes between 1985 and 2020.
The height of each segment (left and right columns) or flow (connections between columns) is proportional to the total land area in each class. Flows show changes in the spectral similarity class of classified pixels between the two time periods. Values were square root transformed to improve visualization.
Fig 9.
Effects of wildfire on cover type classes.
Change detection within the Rodeo-Chediski fire (2002) comparing a MTBS burn severity map (top right) and a differenced cover type map (bottom right).
Fig 10.
Area burned by large (≥400 ha) wildfires on our study area between 1986 and 2020.
The orange line shows total area burned within wildfire perimeters, whereas the green line shows the area within those wildfire perimeters that burned in forest-capable lands.