Southeast Exotic Pest Plant Council

HABITAT SUSCEPTIBILITY TO INVASION BY COGONGRASS ON CAMP SHELBY TRAINING SITE, MS. Lisa Yager1, Deborah L. Miller2, and Jeanne Jones3, 1The Nature Conservancy, Camp Shelby, MS 39407, 2University of Florida/Milton Campus, Milton, FL 32583 and 3Mississippi State University, Mississippi State, MS 39762 (


Introduction -

Introduced in the early 1900's from Asia, cogongrass, Imperata cylindrica, has since become a serious weed problem throughout the southeastern United States. It occurs on Camp Shelby Training Site (CSTS) and Desoto National Forest in Mississippi in a wide variety of habitats (e.g. roadsides, training sites, upland forests, and wetlands) and is widespread throughout the local area along roadsides and in pastures and pine plantations. It often forms dense patches that can displace native vegetation, alter fire regimes and soil processes, reduce forage quality for domesticated animals and wildlife and cause other undesirable changes in natural and managed ecosystems. On CSTS, military firing operations deal with increased wildfire issues as a result of its high flammability. Its continued spread threatens military training operations, the habitat of the federally-listed-as-threatened gopher tortoise, and other biologically significant areas.

Most of CSTS is covered by pine forests which may be classified as pine/bluestem (pine overstory, sparse midstory, abundant bluestem-dominated groundcover) or pine/shrub habitat (pine overstory, well-developed shrub midstory, minimal herbaceous groundcover). Smaller tracts of hardwood communities with hardwood overstory and minimal understory vegetation also occur. Embedded within the forests are mowed herbaceous communities dominated by a mixture of planted grasses (Paspalum notatum, Lolium perenne) and native grasses and forbs. These areas include roadsides, rights of ways, and military training areas. As resources for cogongrass control are limited, knowledge of rates of spread and which habitats are more susceptible to invasion is needed to prioritize control efforts. To assess susceptibility of different habitats on Camp Shelby Training Site to invasion by cogongrass, we measured: 1) linear vegetative spread (tiller extension) on military training areas and roadsides adjacent to either pine/bluestem, pine/shrub, or hardwood communities and 2) spikelet dispersal into pine/bluestem and pine/shrub habitats.

Vegetative Spread -

Linear growth of cogongrass was estimated by mapping patches using a S.P. Constructor Total Station to collect UTM coordinates every 30-50 cm along patch edges during April 2002, 2003, and 2004. ArcView was used to calculate distances of 2003 points from the 2002 perimeters of each patch, and distances of 2004 points from the 2003 perimeters of each patch. Mean linear growth was determined by averaging the mean distances for each patch by averaging distances measured. Maximum growth was recorded as the maximum distance recorded for each habitat type.

During 2002 and 2003, patches of cogongrass which had not received soil disturbance showed greater mean linear growth in the more open environments (pine/bluestem, roadsides, military training areas) compared to the less open sites (pine/shrub, hardwoods) (Table 1). This pattern was less apparent during 2003/2004, although mean growth continued to be lowest in the hardwood forest habitat. Maximum tiller extension was not consistent among habitats between years. Not included in the table, but of interest, linear growth of greater than 10 m was recorded where heavy equipment had caused extensive soil disturbance in and near a cogongrass patch on a military training area.

Habitat conditions, such as type and intensity of disturbance, light levels, soil fertility, species composition, may influence rates of vegetative spread into different habitats. Light levels would generally be less in the hardwood forest and pine/shrub habitats compared to the other habitats and increased shade has been shown to reduce growth of cogongrass shoots and rhizomes (1). This may partially explain lower rates of vegetative growth in the hardwood forest and pine/shrub habitats. Soil disturbance enhanced growth of cogongrass seedlings in a coastal wet pine savanna (2) and this may be consistent for other habitats based upon the rapid growth observed after soil disturbance in a patch on a military training area.

Seed Dispersal -

In order to compare spikelet dispersal into pine/bluestem and pine/shrub habitats, we selected 3 sites along east/west roads where pine/bluestem and pine/shrub habitats occurred adjacent to each other. During May, 2004, we released 50 sterile cogongrass spikelets from 0.7 m platforms simultaneously into the 2 habitat types for each site on 5 different days. We observed the seeds until they landed upon an object or the ground and recorded landing location using a Trimble Pro-XR GPS unit. Arcview was used to determine dispersal distance from the release platforms. In a few instances in both habitat types, spikelets were lofted so high that we were unable to follow them to their landing location and we were unable to include these in the results. Maximum and mean wind speed was recorded at each release location. Mean and maximum distances dispersed and percentages of spikelets dispersing further than 5 and 10 m were calculated by day, treatment and road site location. Analyses of variance were performed using SAS general linear model with road site location treated as a block for dispersal distances and using SAS generalized linear model with road site location treated as a block for dispersal percentages.

Instances in which spikelets are carried high and then far by the wind may be important for long distance dispersal of cogongrass and creation of new isolated patches, but most seeds with the potential to create new patches will fall closer to the parent plant. As might be expected maximum wind speeds were greater in the more open habitats (1.7 m/sec pine/bluestem; 0.7 m/s pine/shrub) which have fewer shrubs to act as wind breaks. Most spikelets for both habitats fell within 5 m of the release platform and this is a typical pattern of seed dispersal. Mean dispersal distance was < 5 m for both habitats; however mean maximum dispersal distance was significantly greater into the pine/bluestem habitat type (37 m) as compared to the pine/shrub habitat (23 m). More spikelets dispersed further than 5 or 10 m into the more open pine/bluestem areas. Shrubs and other vegetation may act as a barrier to spikelets reaching the ground and thus prevent spikelets from reaching an adequate seedbed. Spikelets were more likely to land on bare ground or litter in the pine/bluestem habitats (34 %) compared to the pine/shrub habitats (16 %). These results indicate that, unless other factors within the habitat inhibit seedling establishment, encroachment by cogongrass through seed recruitment will proceed more rapidly into pine/bluestem or other open habitats compared to the more densely wooded pine/shrub habitats.

Conclusion -

Generally speaking, from conservation and military training perspectives, the pine/bluestem and training areas are of most interest and should receive priority for cogongrass control. Our results indicate that cogongrass encroachment may proceed at a faster rate in these habitats which suggests that these areas should receive priority for control.


  1. Macdicken, K.G., K. Hairiah, A. Otsamo, B. Duguma and N.M. Majid. 1997. Shade-based control of Imperata cylindrica: tree fallows and cover crops. Agroforestry Systems 36:131-149.
  2. King, S.E. and J.B. Grace. 2000. The effects of gap size and disturbance type on invasion of wet pine savanna by cogongrass, Imperata cylindrica (Poaceae). American Journal of Botany 87:1279-1286.

Table 1. Linear growth (tiller extension) of cogongrass from perimeter of the previous year's patch edge in 5 habitats on Camp Shelby Training Site, MS. Data is for patches which had not received obvious soil disturbance during study. The number of cogongrass patches surveyed is represented by n.

University of Georgia The Bugwood Network Forestry Images   The Bugwood Network - The University of Georgia
College of Agricultural and Environmental Sciences and Warnell School of Forest Resources
Copyright 2002. All rights reserved.       Page last modified: Monday, April 29, 2002
Questions and/or comments to: