Fire and vegetation dynamics on US Great Plains National Grasslands
Preliminary analysis
DAM
Version 2022-02-02
Overview
Objectives
- Describe spatial and temporal trends of fire across US National Grasslands in the Great Plains
- Describe vegetation patterns in response to fire
- Vegetation structure (bare ground, litter, woody vegetation)
- Annual forb & grass responses
- Perennial forb and grass cover and productivity (forage resource responses)
- Responses of woody encroachers
Approach
- Connect fire perimeters, remotely-sensed vegetation data, topoedaphic data, and grazing records via GIS
- Test for statistical patterns in general vegetation responses to fire
- Firstly, compare burned areas to neighboring unburned areas, controlling to the extent possible for potentially confounding variables (fires 2005-2015)
- Secondly, determine whether trends identified from the above subset of fire hold across the full dataset (1985-2020)
- Apply the same two-step process specifically to areas of high woody plant density prior to burns to determine role of fire in modulating woody plant dynamics
Methods
Study area
Location of US National Grasslands in the Great Plains.
Data sources
- Fire perimeters 1960-2020 from S_USA.FinalFirePerimeter
- Vegetation data from 1985-2020 are available on the Rangeland Analysis Platform (RAP)
- 30 m resolution
- Plant functional groups & herbaceous productivity (annual and perennial)
- Topoedaphic data (ecological site, slope, aspect) available via online GIS resources
- Grazing records 2005-2020 provided by USFS
Sampling
The statistically-rigorous approach to sampling proceeds as follows:
- Data are retrieved from points within paired sample areas
- Inside the fire perimeter vs. within 10 km of the fire boundary
- Same ecological site class
- Number of points assigned within dominant ecological sites to ensure at least 50% of area is sampled
- Slope and aspect data for each point calculated from Digital Elevation Map
- Aspect categorized as North- or South-facing
- Slope categorized as Flat, Steep, or Very Step
- Points inside and outside fire perimeters are matched by ecologial site, slope, and aspect
| National Grassland | Area (x 1000 ha) | Number of fires since 1985 |
|---|---|---|
| Little Missouri | 4149 | 310 |
| Buffalo Gap | 2650 | 212 |
| Oglala | 873.3 | 58 |
| Pawnee | 843.5 | 51 |
| Grand River | 626.1 | 48 |
| Comanche | 1798 | 36 |
| Fort Pierre | 846 | 32 |
| Sheyenne | 285 | 21 |
| Cimarron | 441.5 | 13 |
| Thunder Basin | 2504 | 10 |
| Black Kettle | 134 | 3 |
| Kiowa | 583.9 | 3 |
| Rita Blanca | 380.9 | 3 |
| Caddo | 68.24 | 0 |
| Lyndon B. Johnson | 467 | 0 |
| McClellan Creek | 5.673 | 0 |
| Cedar River | 27.18 | 0 |
Results
Fire trends
Fire trends on US Great Plains National Grasslands since 1970. Vertical breaks denote the beginning of remotely-sensed vegetation data available on the Rangeland Analysis Platform, and records on grazing allotment lease status.
| Assigned cause of fire | Number of fires caused |
|---|---|
| Lightning | 423 |
| Miscellaneous | 179 |
| Equipment use | 123 |
| Debris burning | 48 |
| Railroad | 22 |
| Campfire | 18 |
| Arson | 7 |
| Smoking | 4 |
| Children | 1 |
Vegetation trends
Recall that sample area refers to groups of data collection points inside and outside of a fire perimeter that have the same ecological site, slope, and aspect. The difference in the mean values within vs. outside the fire perimeter is used to represent the fire effect for that combination of environmental variables.
Fire effects on unleased allotments
Limiting analysis to allotments in and around fires that were not leased for grazing since 2015 includes between 9-17 sample areas per year across 4 fires from 4 US National Grasslands. Vegetation trend analysis on these data indicates that:
- Annual herbaceous cover (forbs + grasses) increased just 6% in the year after fire, and was not statistically-significantly different from pre-fire levels 5 yrs after fire. Annual herbaceous standing crop was statistically-significantly greater after fire, ranging from 20-65 addition tons per acre.
- Perennial herbaceous cover (forbs + grasses) declined as much as 20% in the year after fire but recovered by the fourth year after fire. Perennial herbaceous standing crop was 200 t/ac less in the year following fire, and 100 t/ac less than pre-fire levels 5 yrs after fire.
- Bare ground increased 1-2 years after fire and declined.
- Litter tended to decrease 1-2 years after fire and recover, but no changes were ever statistically significant.
- Woody plant cover changed very little in response to fire. Shrubs declined 2% in the year after fire, while trees never varied more than even half a percent.
Vegetation trends for 5 years after burning for 4 fires in US National Grasslands for which a sufficient number of unleased grazing allotments could be included for comparison. All changes (fire effects) compare sample locations within fire perimeters to sample locations outside the perimeter that are matched by ecological sites, slope, and aspect. Point colors reflect the results of a t-test comaring whether or not each value against 0, which represents no fire effect; if P ≥ 0.05, fire did not have a statistically-significantly different effect on that variable relative to pre-fire values in that year.
Fire effects across all allotments with known leasing history
When all allotments with a known leasing history since 2015 are included, the vegetation trend analysis is based on 53-107 sampling areas per year across 17 fire events in 8 US National Grasslands.
- Ecological site was the only term that was consistently identified as a statistically-significant factor in the effect of fires across all vegetation response variables.
- Whether the aspect was south-facing vs. north-facing mattered only to changes in annual herbaceous cover and production, and tree cover.
- There were significant interactions between face \(\times\) slope for perennial herbaceous, shrub, and tree cover.
| Category | Response | term | statistic | df | p.value |
|---|---|---|---|---|---|
| Cover | Annuals | site | 51.68 | 5 | <0.001 |
| Cover | Annuals | face | 7.472 | 1 | 0.006 |
| Cover | Bare ground | site | 103 | 5 | <0.001 |
| Cover | Bare ground | Year | 33 | 4 | <0.001 |
| Cover | Litter | site | 11.76 | 5 | 0.038 |
| Cover | Perennials | site | 151.9 | 5 | <0.001 |
| Cover | Perennials | Year | 31.26 | 4 | <0.001 |
| Cover | Perennials | face:slope | 6.378 | 2 | 0.041 |
| Cover | Shrubs | site | 44.88 | 5 | <0.001 |
| Cover | Shrubs | Year | 61.54 | 4 | <0.001 |
| Cover | Shrubs | face:slope | 6.435 | 2 | 0.04 |
| Cover | Trees | site | 35.5 | 5 | <0.001 |
| Cover | Trees | face | 20.75 | 1 | <0.001 |
| Cover | Trees | slope | 36.6 | 2 | <0.001 |
| Cover | Trees | face:slope | 37.5 | 2 | <0.001 |
| Production | Annuals | site | 63.77 | 5 | <0.001 |
| Production | Annuals | face | 11.18 | 1 | 0.001 |
| Production | Annuals | Year | 10.92 | 4 | 0.028 |
| Production | Perennials | site | 62.22 | 5 | <0.001 |
| Production | Perennials | Year | 18.29 | 4 | 0.001 |
Temporal trends in vegetation following fire over the larger dataset reflect those of the smaller dataset (17 vs 4 fires, respectively). Thus, trends appear consistent across leased and not-leased allotments, and across a geographical range of US National Grasslands. As indicated by the statistics above, there is often considerable variability among ecological sites and by aspect.
Vegetation trends for 5 years after burning for 17 fires in 8 US National Grasslands. All changes (fire effects) compare sample locations within fire perimeters to sample locations outside the perimeter that are matched by ecological sites, slope, and aspect. DIfferent colors and line types plot different ecological sites and aspects, respectively, while the black line, error bars, show the overall trend averaged across ecological sites and aspect. Point shading reflects the results of a t-test comaring whether or not each value against 0, which represents no fire effect; if P ≥ 0.05, fire did not have a statistically-significantly different effect on that variable relative to pre-fire values in that year.
Supplemental information
Limitations on controlled statistical comparisons
Rigorous statistical comparison of vegetation responses to fire–in which both management (grazing) and topoedaphic (ecological site, slope, aspect) factors are controlled for–is complicated by the limited number of fires for which grazing records indicate there are sufficient “controls” outside of the fire perimeter that were not leased for grazing immediately before a burn and for several years after.
Just 4 fires meet the above criteria, across 4 National Grasslands. Broken down by ecological site and year, there are even more gaps in the number of paired sample areas that meet statistically-rigorous vegetation sampling criteria:
| GY-1 | GY1 | GY2 | GY3 | GY4 | GY5 | Sum | |
|---|---|---|---|---|---|---|---|
| Comanche NG, Snakeweed Fire | 1 | 1 | 1 | 1 | 1 | 1 | 6 |
| Fort Pierre NG, War Creek Fire | 2 | 2 | 2 | 0 | 2 | 0 | 8 |
| Grand River NG, HayCreek Fire | 6 | 6 | 6 | 0 | 0 | 0 | 18 |
| Little Missouri NG, Brown Fire | 8 | 8 | 8 | 8 | 8 | 8 | 48 |
| Sum | 17 | 17 | 17 | 9 | 11 | 9 | 80 |
When the criterion that all sample areas not have an active grazing lease is dropped, 14 fires across 8 National Grasslands have sufficient data up to four years post-fire. Fires with only 3-4 years post-fire data likely occurred in 2016 or 2017 (e.g., the 2017 Magpie Fire). As of 1 Jan 2022 there is a new version of RAP data up, to which 2021 data will hopefully be added soon.
| GY-1 | GY1 | GY2 | GY3 | GY4 | GY5 | Sum | |
|---|---|---|---|---|---|---|---|
| Buffalo Gap NG, Cactus Fire, Clayey | 4 | 4 | 4 | 4 | 4 | 4 | 24 |
| Buffalo Gap NG, Cactus Fire, Loamy | 2 | 2 | 2 | 2 | 2 | 2 | 12 |
| Buffalo Gap NG, Cottonwood CA Fire, Badlands | 3 | 3 | 3 | 3 | 3 | 0 | 15 |
| Buffalo Gap NG, Cottonwood CA Fire, Clayey | 5 | 5 | 5 | 5 | 5 | 0 | 25 |
| Buffalo Gap NG, Cottonwood CA Fire, Loamy | 5 | 5 | 5 | 5 | 5 | 0 | 25 |
| Buffalo Gap NG, Cottonwood CA Fire, Shallow | 1 | 1 | 1 | 1 | 1 | 0 | 5 |
| Buffalo Gap NG, FREEMAN Fire, Clayey | 2 | 2 | 2 | 2 | 2 | 0 | 10 |
| Buffalo Gap NG, WOLF Fire, Clayey | 1 | 1 | 1 | 1 | 0 | 0 | 4 |
| Cimarron NG, Greenwood Fire, Limy | 2 | 2 | 2 | 2 | 2 | 2 | 12 |
| Cimarron NG, Greenwood Fire, Loamy | 1 | 1 | 1 | 1 | 1 | 1 | 6 |
| Cimarron NG, Steeler Fire, Sandy | 3 | 0 | 0 | 0 | 0 | 0 | 3 |
| Cimarron NG, Tunner Fire, Sandy | 3 | 0 | 3 | 3 | 3 | 3 | 15 |
| Comanche NG, Snakeweed Fire, Limy | 1 | 1 | 1 | 1 | 1 | 1 | 6 |
| Fort Pierre NG, War Creek Fire, Clayey | 4 | 2 | 2 | 2 | 2 | 0 | 12 |
| Grand River NG, HayCreek Fire, Loamy | 4 | 2 | 2 | 2 | 2 | 2 | 14 |
| Grand River NG, HayCreek Fire, Sandy | 2 | 1 | 1 | 1 | 1 | 1 | 7 |
| Grand River NG, HayCreek Fire, Shallow | 6 | 3 | 3 | 3 | 3 | 3 | 21 |
| Grand River NG, Pautre Fire, Loamy | 4 | 4 | 4 | 4 | 4 | 4 | 24 |
| Grand River NG, Pautre Fire, Sandy | 6 | 6 | 6 | 6 | 6 | 6 | 36 |
| Grand River NG, Pautre Fire, Shallow | 3 | 3 | 3 | 3 | 3 | 3 | 18 |
| Little Missouri NG, Brown Fire, Badlands | 8 | 4 | 4 | 4 | 4 | 4 | 28 |
| Little Missouri NG, Brown Fire, Limy | 8 | 4 | 4 | 4 | 4 | 4 | 28 |
| Little Missouri NG, Magpie Fire, Badlands | 2 | 2 | 2 | 2 | 0 | 0 | 8 |
| Little Missouri NG, Magpie Fire, Limy | 1 | 1 | 1 | 1 | 0 | 0 | 4 |
| Little Missouri NG, RidgeRoad Fire, Loamy | 10 | 5 | 5 | 5 | 5 | 5 | 35 |
| Little Missouri NG, RidgeRoad Fire, Sandy | 8 | 4 | 4 | 4 | 4 | 4 | 28 |
| Little Missouri NG, RidgeRoad Fire, Shallow | 2 | 1 | 1 | 1 | 1 | 1 | 7 |
| Pawnee NG, Gold Art Fire, Loamy | 2 | 2 | 2 | 2 | 2 | 2 | 12 |
| Pawnee NG, Road 114 Fire, Loamy | 3 | 3 | 3 | 3 | 3 | 0 | 15 |
| Sheyenne NG, Gregor Fire, Sandy | 1 | 1 | 1 | 1 | 1 | 1 | 6 |
| Sum | 107 | 75 | 78 | 78 | 74 | 53 | 465 |