Experimental restoration of an indigenous Hawaiian grassland after invasion by buffel grass (Cenchrus ciliaris)

Summary

Hawaiian Islands vegetation has been radically altered by the introduction of non-native plants. The spread of non-native grasses introduced as fodder crops is of particular concern. About half of the total land area of the Hawaiian Islands was at one time converted to grassland and although some pasturelands have been abandoned, introduced pasture grasses have continued to spread into managed nature reserves threatening native grassland communities. For example, introduced grasses have displaced native Hawaiian pili grass Heteropogon contortus in most dry, leeward habitats. The purpose of this study was to assess the feasibility of restoring an indigenous Heteropogon grassland at a site on the island of Hawaii where an introduced, drought-tolerant, African buffel grass Cenchrus ciliaris, is now dominant.

Study site: The Puukohola Heiau National Historic Site on the island of Hawaii was selected for restoration of Hawaiian pili grass Heteropogon contortus. The site is located along the leeward coast and is one of the driest parts of the island (average annual rainfall 211 mm). The site encompasses 34 ha, most of which is buffel grass Cenchrus ciliari grassland, with occasional introduced Kiawe Prosopis pallida trees.

Pili grass seeds: No seed sources were found within Puukohola Heiau NHS. Therefore in autumn 1997, pili grass seeds were collected from remnant patches along roadsides within 5 km of the site. Seeds were stored dry until the start of the experiment in February 1998. A second seed collection was made in autumn 1999 for use in the second restoration plot established in February 2000.

Burn plots: Two open grassland burn plots were established (one in 1998 and one in 2000), each covering roughly 15 × 25 m and lying 20 m apart. Prior to burning, vegetation within the plots averaged 95% Cenchrus (relative cover), with Waltheria indica and Sida fallax making up the remaining 5%. Absolute cover by Cenchrus was lower (averaging 83%) due to some openings due to scattered rocks. No Heteropogon was recorded.

In the 1998 plot, 28 replicated, 1-m², Cenchrus-dominated subplots spaced 1 m apart, were marked out with metal stakes. One of four treatments was randomly applied (number of replicates in parentheses):

1) burned (9)
2) Cenchrus hand pulled followed by burning (8)
3) burned followed by herbicide (glyphosate formulated as 2% Roundup Pro, applied to resprouting Cenchrus 2 weeks after the first burn using a backpack sprayer) (8)
4) unburned control (3)

Three weeks after the first burn, 50 Heteropogon seeds were added per subplot. Each seed was placed along a furrow scraped in the soil (0.25–0.5 cm deep) with a stick, and then covered with the disturbed soil. Heteropogon seeds were excluded from two subplots in the hand-pull and herbicide treatments to test for Heteropogon germination from a seed bank.

The 1998 plot was burned in February 1998, 2000, and 2002, but the seed addition and herbicide and hand-pull treatments were only applied in 1998. No information was available to indicate that Hawaiians traditionally burned Heteropogon at specific times of the year. February (winter) was chosen as there were better chances for suitable weather; in particular, low wind and higher relative humidity were necessary to decrease risks of an escaped fire.

Rainfall in 1998 (84 mm) and 1999 (87 mm) was unusually low (compared with 394 mm in 1997 and the 211 mm average). By March 1999 regrowth of plants in the subplots was restricted to initial resprouting of Cenchrus. Therefore, it was decided to water to the plot. Water was applied via a sprinkler system delivering approximately 127 L/minute to the entire plot, once per week for 30 minutes from April 1999 through May 1999 (i.e. 10 L of supplemental water per m² per week). The watering was turned off in the summer. The same watering regime was continued each November to April except in 2002, when the water was turned off in mid-March, 4 weeks after the final burn.

The second burn plot (initiated in 2000) included two of the 1998 treatments, assigned to 18, 1-m² subplots (replicates in parentheses): burning followed by herbicide (15) and unburned control (3). This plot was burned twice (February 2000 and February 2002) and twice as much water per week was added (20 L/m² ) following the same schedule as the 1998 plot. The purpose of the additional water was to see if Heteropogon growth and establishment could be accelerated. In both 1998 and 2000 plots.

Data collection: Plots were surveyed at monthly or quarterly intervals between February 1998 and June 2002. A 1-m² quadrat was laid over each subplot. Cover of bare ground, Cenchrus, Heteropogon, and any other species exceeding 5% cover was estimated with the assistance of a 0.25-m² quadrat divided into quarters. Heteropogon and Cenchrus seedlings were counted during the first year after water supplementation was initiated and other species recorded.

Heteropogon establishment: In treatments containing established Cenchrus, Heteropogon establishment was consistently poor (<10% cover). But in the burned plots where established Cenchrus had been removed, as many as 31 Heteropogon seedlings per square metre were recorded, and Heteropogon became the dominant cover, averaging 34% absolute cover (81% relative cover) after 4 years in the lower-water plot and 34% absolute cover (60% relative cover) after 2 years in the higher-water plot.

Water supplementation promoted growth of two other alien grasses from the seed bank, Digitaria insularis and Eragrostis spp. However, these quickly declined after watering of plots ended. Although initial suppression of Cenchrus was required, Heteropogon expanded quickly when seeds and fire were reintroduced, demonstrating that a Heteropogon-dominated grassland can be re-established in 2–4 years.

Seed supply: An important additional outcome of this restoration experiment has been that Park staff have been able to collect large quantities of seeds from Heteropogon established in the plots, which can be used in further restoration projects.

Conclusions: This study has shown that at a small scale, Heteropogon-dominated grasslands can be restored through removal of Cenchrus and addition of Heteropogon seeds. Biennial burning appeared to facilitate the restoration process, but alternative fire regimes should be explored. Supplemental watering probably increased that rate at which Heteropogon regained dominance (2–3 years). The biggest impediment to scaling up the restoration methods developed here is limited Heteropogon seed availability. With only small, remnant Heteropogon populations occurring, it was challenging to collect enough seeds even for these small restoration plots. Large quantities of Heteropogon seeds from abroad could have been purchased, but the preference was to use locally collected seeds – it has been suggested that Hawaiian Heteropogon may differ from that found in other regions in the Pacific and Africa, where it is also indigenous.


Note: If using or referring to this published study, please read and quote the original paper, this can be viewed at: http://www.blackwell-synergy.com/journal.asp?ref=1061-2971