Click on the map for a larger image and zoom in to see details.

Distribution of Hawai'ian landcover based on LANDFIRE Biophysical Settings (BpS) data layer [31], http://www.landfire.gov/NationalProductDescriptions20.php.

Lowland dry and mesic communities: Lowland dry and mesic communities occur primarily below about 3,300 feet (1,000 m) on leeward-facing (generally south and west) slopes of the Hawai'ian islands [10,29].

Distribution of lowland dry and mesic communities based on LANDFIRE Biophysical Settings (BpS) data layer (LANDFIRE HI Refresh 2008), http://www.landfire.gov/NationalProductDescriptions20.php. Click on the map for a larger image and zoom in to see details.

Prior to human settlement, Hawai'ian lowland dry and mesic communities were predominantly comprised of dry forests, woodlands, and shrublands. It is likely that native lowland grasslands were infrequent or even nonexistent, and lowland forests, woodlands, and shrublands were widespread, probably extending to the coast in many places [10,26]. Descriptions of lowland dry and mesic communities are modified from Cuddihy and Stone [10] unless otherwise indicated.

Dry forests and woodlands: Dry forests and woodlands were thought to contain a diversity of tree species frequently dominated by 'ohi'a (Metrosideros polymorpha) or lama (Diospyros sandwicensis). Other trees present included: alahe'e (Canthium odoratum), kauila (Colubrina oppositifolia), 'aiea (Nothocestrum breviflorum), wiliwili (Erythrina sandwicensis), 'ohe makai (Reynoldsia sandwicensis), uhiuhi (Caesalpinia kavaiense), koki'o (Kokia drynarioides), koa (Acacia koa), koai'a (A. koaia), olopua (Nestegis sandwicensis), and iiulu (Sapindus oahuensis) [10]. As of 1990, much of the remaining dry forests and woodlands with native overstories occur steep slopes, in gulches, and on very rocky substrates (i.e., areas not suitable for agriculture) [10,52]

Shrublands: Native shrublands occurred in mixed stands of: 'a'ali'i (Dodonaea viscosa), 'iikia (Wikstroemia spp.), 'iiweoweo (Chenopodium oahuense), ko'oko'olau (Bidens menziesii), pukiawe (Styphelia tameiameiae), alahe'e, and low-growing 'ohi'a. Remnant stands of 'akoko (Chamaesyce spp.), nehe (Lipochaeta spp.), and kulu 'i (Nototrichium sandwicense) that currently occur on steep cliffs could represent shrub communities that were formerly more extensive.

Grasses: Native grasses found in lowland dry and mesic communities included 'emo-loa (Eragrostis variabilis), and kakonakona (Panicum torridum). Pili grass (Heteropogon contortus) is often considered a native grass. However, there is not a consensus on its nativity. Some botanists suggest pili grass is a Polynesian introduction (e.g., [38]), others believe it to be indigenous to Hawai'i (e.g., [10,46]), while others are undecided (e.g., [44,48]. The distinction of nativity may be considered important because pili grass played an important role in fire history during the Hawai'ian settlement period.

Upland dry and mesic communities: Upland dry and mesic communities that consist of montane, subalpine, and alpine vegetation occur on the leeward-facing sides of Maui and Hawai'i. The elevational range of the montane zone is approximately 3,300 to 6,600 feet (1,000-2,000 m); and that of the subalpine zone is approximately 6,600 to 9,200 feet (2,000-2,800 m). Vegetation above approximately 9,200 feet (2,800 m) is considered alpine.

Distribution of upland dry and mesic communities based on LANDFIRE Biophysical Settings (BpS) data layer (LANDFIRE HI Refresh 2008), http://www.landfire.gov/NationalProductDescriptions20.php. Click on the map for a larger image and zoom in to see details.

Descriptions of upland dry and mesic communities are modified from Cuddihy and Stone [10].

Montane vegetation: Prior to human settlement, forests were the primary natural vegetation of the leeward montane zone, and montane shrublands and grasslands were not prominent vegetation communities.

Native montane dry forests are dominated by koa, mamane (Sophora chrysophylla), 'ohi'a, and rarely, 'akoko (Chamaesyce olowaluana, C. celastroides). Montane mesic forests have a very restricted distribution in the upper montane zone on Maui and Hawai'i. Dominant trees are 'ohi'a and koa, but floristically they differ from wet forests because their understory is comprised of an abundance of shrubs, such as pukiawe, rather than tree ferns and epiphytes. On the island of Hawai'i, drier open woodlands composed of koa, mamane, native shrubs, and grasses were once common in the transition area between wet forest and the subalpine zone.

Montane dry shrublands may contain 'a'ali'i, na'ena'e (Dubautia linearis), ko'oko'olau, pukiawe, and 'ohi'a.

Native montane grasslands occur in a few dry and mesic areas on Hawai'i including the leeward slopes of Mauna Loa and Mauna Kea. The endemic grasses hardstem lovegrass (Eragrostis atropioides), alpine hairgrass (Deschampsia nubigena), and parkland panicgrass (Panicum tenuifolium) dominate these grasslands. The extensive nonnative pasturelands that are found in the montane zone on Maui and Hawai'i were converted from native forests and shrublands.

Subalpine vegetation: Extensive native subalpine vegetation occurs on both Hawai'i and Maui. Plant communities consist of forests, shrublands, and grasslands.

There are two main types of subalpine forests; both are relatively dry. On Hawai'i, open 'ohi'a forests grow on younger substrates, while open to closed mamane forests occur on older substrates on both Maui and Hawai'i. Naio (Myoporum sandwicense) is a common codominant of mamane forests.

Dry shrublands of the upper mountain slopes are often dominated by pukiawe and 'ohelo (Vaccinium reticulatum). 'A'ali'i and na'ena'e (Dubautia ciliolata) may also be important. On the saddle between Mauna Kea and Mauna Loa, on the island of Hawaii, exists a unique shrubland composed of 'aweoweo. Mesic shrublands may occur between wet forests and mesic grasslands on the steep upper slopes of eastern Maui. They are often dominated by 'ohelo and 'ama'u fern (Sadleria cyatheoides).

Subalpine grasslands are dominated by the native bunchgrass, alpine hairgrass. These grasslands are relatively dry when growing in the cindery substrates of Haleakala crater or the upper slopes of Mauna Kea. When growing on windward slopes of Haleakala, they can be mesic.

Alpine vegetation: This vegetation type is comprised of very sparse native shrublands. Common species include widely-scattered, low-growing pukiawe and 'ohelo.

Lowland and upland wet communities: Wet communities occur on the windward sides (northeastern facing slopes) of the main islands except Ni'ihau and Kaho'olawe. Moisture-laden air carried with the prevailing trade winds is responsible for the wet climate.

Distribution of lowland and upland wet communities based on the LANDFIRE Biophysical Settings (BpS) data layer (LANDFIRE HI Refresh 2008), http://www.landfire.gov/NationalProductDescriptions20.php. Click on the map for a larger image and zoom in to see details.

Descriptions of wet communities are modified from Cuddihy and Stone [10] unless otherwise noted.

Lowland wet forests: Prior to human settlement, wet forests were the predominant vegetation of the windward lowlands. These forests were likely dominated by 'ohi'a or koa, and had an understory comprised of native trees, lianas, ferns, and tree ferns. Much of the original native lowland forest vegetation has been severely altered by cultivation, grazing, development and/or nonnative species. As of 1990, these forests occur mostly in protected areas on sites with rocky substrates or steep terrain [10,52].

Montane wet forests: The elevational boundary that distinguishes lowland from montane wet forests ranges from 1,500 to 3,300 feet (460-1,000 m) depending on the location and the classification system used. Wet montane forests still cover large areas of Hawai'i and Maui and are also found on steep windward slopes of Kaua'i, O'ahu, and Moloka'i. Most of these forests are dominated by 'ohi'a which often forms a closed canopy. Koa and 'olapa are often codominants. Native trees, shrubs, and tree ferns comprise the well developed understory.

Montane bogs: Bogs occur on most of the high Hawai'ian islands either near mountain summits or high-rainfall windward slopes. Bogs are generally small and are dominated by sedges, grasses, and stunted woody plants. Each bog contains a unique assemblage of native flora.

• Presettlement fire regime
• Ignition
• Fire season
• Inferring fire regime characteristics from fire effects
• Post-Hawai'ian settlement
• Post-European settlement

Presettlement fire regime: Evidence suggests that prior to human settlement (approximately 400 CE) fire was infrequent and unlikely to have played an important role in lowland dry and mesic communities [32,38]. Presettlement vegetation was likely comprised of forest, woodland, and shrubland communities (i.e., not grasslands) [10,26], which lacked a combustible understory of fine fuels capable of supporting fire [10,38,49]. While acknowledging lightning as a possible ignition source, Mueller-Dombois [38] argued that vegetation and fuel characteristics in dry forests made fires extremely unlikely (Table 2). He reasoned that dry forests were not likely to carry fire because they lacked a combustible understory or burnable organic layer. He referenced a remnant dry forest comprised of iiulu, alahe'e, and a'ali'i on the north shore of O'ahu that did not have an understory capable of carrying fire [38]. Similarly, in Hawai'i Volcanoes National Park, a seasonally dry 'ohi'a woodland with native shrubs had a discontinuous ground cover of lichens, bryophytes, sedges, and sparse bunchgrass [40,41]. In presettlement times, the only native grass in lowland dry and mesic communities that may have formed a matrix capable of carrying fire was 'emo-loa [49]. However, it is rarely present today, which suggests that it is not as well adapted to frequent fire as today's dominant nonnative grasses [38]. Mueller-Dombois [38] noted that in these communities, the native low-growing shrubs pukiawe and a'ali'i could have carried fire, but the fact that the dominant pukiawe rarely survives and does not recolonize after fire suggests that fires were not common.

The following summary includes information available on aspects of presettlement fire regimes in lowland dry and mesic communities. Information on these topics is sparse, and no information was available regarding presettlement fire frequency, fire type and severity, or fire pattern and size in these communities.

Ignition: Prior to human settlement, ignition sources included lava flows and lightning. Lava flows are intermittent and limited to localized active volcanic regions on the younger islands. The island of Hawai'i currently has a few active volcanoes including Kilauea, Mauna Loa, and Hualala [42]. In Hawai'i Volcanoes National Park, lava flows were the cause of 26% of the fires from 1924 to 1988 [10]. Haleakala, on Maui, is in the last stages of the volcanic cycle; the most recent eruption occurred in 1790. Older islands have not experienced active volcanic activity for over 1 million years [42]. Lightning frequency for all of the Hawai'ian islands has been reported at 30 to 50 lightning days/year [5,51] and 20 to 30 thunderstorms per year [50], with the winter months (October-April) having the most storms. Most thunderstorms are accompanied by heavy rain. Observations of 25 thunderstorms indicated only 3 without appreciable rain [51]. Thunderstorms are concentrated in the higher mountains of the island interiors and along windward slopes rather than in the lowlands [5]. Evidence for lightning-caused fire in native, unaltered vegetation is difficult to find in Hawai'i [45,51]. In recent years, lightning-caused fires have been observed in Hawai'i, but these fires occur in vegetation with a considerable nonnative grass component (e.g., [14,50,51]). Loope and others [33] state that there have been no documented cases of lightning-caused fires in native vegetation on Maui, because the combination of a dry lightning storm with highly ignitable (native) fuels is rare. Recent lava flows on the island of Hawai'i ignited fires fueled by an understory of nonnative, fire-adapted grasses [2,14,48].

Fire season: Published literature about the presettlement fire season in Hawaii was not available (as of 2012). The earliest fire records–from 1906 to 1932 and 1943 to 1967–show that in general, on the islands of Hawai'i and O'ahu, the largest number of fires occurred during the drier summer months. These records include all fires, including human-caused fires [51]. Thunderstorms are more common in the wetter, winter months and in the higher mountains and along windward slopes [5], so during the drier summer months, when fuels are most susceptible to burning, there are fewer lightning ignitions.

Inferring fire regime characteristics from fire effects: Little is known about the fire history or the role of fire in the evolution and development of Hawai'ian ecosystems [2,38,44,51]. Some researchers have studied the responses of native flora to fire in order to gain insights into presettlement and historical fire regimes because plants evolved adaptations to their ancestral disturbance regimes [2]. While research indicates that many native plants can survive and/or reproduce after fire (e.g., [2,14,48]), it is unclear whether the traits that enable postfire survival are adaptations to fire, adaptations to some other local disturbance agent (e.g., volcanism, landslides, hurricanes, treefalls, etc.) [2], or the result of adaptations from their pre-Hawai'ian evolutionary environments [44].

Many Hawai'ian plants can survive and/or reproduce after fire [2,14,22,23,43,48]. Tunison and others [48] studied the responses of native Hawai'ian flora to lava and lightning caused fires in dry grasslands dominated by nonnative grasses and shrublands dominated by native shrubs with a nonnative grass understory in the eastern and central coastal lowlands on the island of Hawai'i. Varying degrees of postfire sprouting were observed for several native shrub species including 'a'ali'i, alahe'e, lama, and 'ulei (Osteomeles anthyllidifolia). In many sites 4 native species (3 shrubs, 1 grass) appeared to be stimulated by fire based on prolific postfire sprouting or seedling establishment. Other native species such as 'iikia (Wikstroemia phillyreifolia) were nearly eliminated from burned sites. No clear relationships between fire severity and postfire response of native plants were evident [48]. Ainsworth and Kauffman [2] studied the responses of native Hawai'ian flora to lava-ignited fires along an environmental gradient from dry shrub-dominated communities to mesic/wet 'ohi'a forest communities. They found that all of the 19 native tree, shrub, and tree fern species present in the prefire environment demonstrated some postfire persistence either by basal sprouting (majority), sprouting through epicormic tissues (rarely), or seedling establishment [2]. In contrast, Smith and Tunison [44] suggest that some species that occur in lowland dry areas such as lama and 'akia do not sprout after fire. It should also be noted that, despite the ability of some species to tolerate fire, they do not require fire to survive and reproduce, and fire may not be beneficial to the communities where they live. As of this writing (2012), fire has generally been shown to decrease the abundance of native woody plants because nonnative, invasive, fire-adapted plants out-compete natives for resources in the postfire environment and tend to dominate postfire communities [10,44] (see Post-European settlement).

In considering community-wide plant responses to fire, it appears that coastal lowland communities may be better adapted to fire than higher-elevation, submontane zone communities, suggesting that coastal lowland communities experienced fire more frequently than submontane communities. Differences in plant responses to fire between coastal communities and submontane communities in these lowlands may provide insights into differences in historic fire regimes between these plant communities. In a study of vegetation response to fire across an elevational gradient in Hawai'i Volcanoes National Park, D'Antonio and others [14] found that in the dry coastal lowlands (below about 1,300 feet (400 m)) fire had little effect on native plant cover and often stimulated native species regeneration; whereas in the seasonal submontane zone (1,300-3,900 feet (400-1,200 m)), burned sites had lower native species cover, and few native species were fire tolerant. Tunison and others [48] found that native vegetation in the eastern coastal lowlands of Hawai'i Volcanoes National Park typically averaged higher cover on burned (19.1-22.4%) than unburned (15.6%) sites (P<0.05); whereas in the submontane zone, Hughes and others [22] found that the average native vegetation cover was lower on burned sites (0.7-31.6%) than on unburned sites (117.0%) (P<0.05). Total cover of native shrubs was much greater in unburned sites (63.6%) than burned sites (0.6-31.2%) (P<0.05); and 4 of the dominant shrubs on unburned sites were absent from or had extremely low cover on burned sites. However, their failure to recolonize was likely due to the establishment of dense nonnative grasses soon after fire [22], not necessarily to their inability to survive and/or regenerate after fire. Pukiawe, the dominant shrub of the submontane zone, had low postfire cover and no postfire colonization [14]. In the coastal lowlands, pili grass and 'a'ali'i, accounted for much of the native postfire cover although cover of other native species also increased [14,48]. While overall cover of native woody plants was unchanged on burned sites, postfire responses of individual woody species varied, including the near elimination of 'iikia [48]. Several species appeared to be stimulated by fire, particularly when fires burned at low intensities. Aside from climatic differences between the communities, one explanation for the apparent fire-adaptation of vegetation in the coastal lowlands is that these sites were heavily influenced by Hawai'ian burning practices between approximately 400 and 1800 CE (see Post-Hawai'ian settlement), which may have shaped communities comprised largely of fire-tolerant species [44].

Post-Hawai'ian settlement: Between approximately 400 and 1800 CE, Hawai'ians settled, colonized, and reshaped much of the vegetation of the lowland dry and mesic communities on all the major Hawai'ian islands [26]. In these ecosystems, Hawai'ians used fire to clear land for agriculture, to support shifting agriculture (i.e., slash and burn), and for cultivating pili grass which was used to thatch houses (Table 2) [10,26,36]. Observations by early European explorers suggest periodic burning of the forest-supporting lowlands was frequent enough to exclude the native woody species [10]. The frequency of burning may have been intended to promote the growth of arrowoot or pia (Tacca leontopetaloide), morning glory (Ipomoea spp.), and other plants used as famine foods or pig feed [26,36].

By the time of European contact in the late 1700s, most of the native dry and mesic communities below 2,500 feet (760 m) were converted from dry forest and open woodland to open grasslands [10,26,36]. Kirch [26] stated that it is generally recognized that most, if not all, of the lowland grasslands in Hawai'i are anthropogenic in origin. He further stated that "the primary tool that affected these great modifications of the prehuman vegetation was undoubtedly fire". Studies of soil layers reveal burn layers and/or subfossil land snail assemblages indicating that the original vegetation was generally forest, and these open grasslands were anthropogenic rather than the native community [8,9,10,27]. Christensen and Kirch [9] hypothesized that the shift in land snail assemblages at Barbers Point, O'ahu, coincident with Hawai'ian occupation, reflects a reduction in vegetation cover and moisture availability. Snail assemblages indicated that the original vegetation in these leeward areas consisted of dry forest and woodland, rather than open grassland [9,10,26]. While sediment cores from both lowland leeward and windward sites on O'ahu revealed a shift from forests dominated by lo'ulu palm (Pritchardia spp.) to communities dominated by grasses, shrubs, herbs, and ferns, the researchers did not attribute this vegetation shift primarily to fire. They gave 3 reasons why fires were unlikely to cause of the loss of lo'ulu palm: 1) they found only trace levels of charcoal particles in sediment cores; 2) they doubted that fire could account for the near eradication of the palm; and 3) the decline in lo'ulu began prior to the widespread growth and expansion of human populations (i.e., before Hawai'ian burning) [4]. Although there is some discrepancy regarding the degree to which burning caused the shift in vegetation, there is general agreement that much of the lowland dry and mesic zone was converted from forests and woodlands to unwooded grasslands.

Observations by early European explorers confirm the presence of extensive fire-maintained pili grasslands (rather than native forest and woodlands) occupying coastal lowlands, plains, and hillsides of leeward areas [10,17,26,51]. In 1792, Menzies observed a "large fire kindled a few miles to the eastward of Waimea, Kaua'i, and spreading over the face of that plain country, which was mostly covered with dry, rank grass (pili grass) that burnt with great rapidity". The inhabitants explained that the fire had been set so that the next crop of grass would grow up clear and free of stumps, and would therefore be better for thatching their houses [26]. Others noted the "predominance of course spiry grass that had the appearance of having undergone the action of fire" [26], and Egler [17] describes the sport of rolling burning barrels of tar down the grass-covered hillsides. These pili-dominated grasslands produce abundant, continuous, fine fuels which are dry most of the year, readily carry fire, and are promoted and maintained by annual burning [51].
	Remnant stand of dry forest surrounded by nonnative grasses, located at Auwahi, Maui. Photo courtesy of Arthur Medeiros.
	Pili grasslands on Lahaina Pali, Maui. Photo courtesy of Forest and Kim Starr.

Post-European settlement: Despite an imprecise understanding of presettlement and post-Hawai'ian settlement fire regimes, there is a general consensus that in many Hawai'ian ecosystems, including lowland dry and mesic communities, fire regimes have been altered from typically rare fire events in the presettlement era to more frequent and often more severe fires today [6,32,49]. Following European contact, fire frequency greatly increased in the lowlands [6,26,32]. Charcoal deposits in a sediment core from a high-elevation bog on Maui confirm a marked increase in fire frequency after European settlement over the last 2 centuries. Pollen records and the origin of the charcoal particles indicate that these fires burned in grasses rather than forests or shrublands. Although the European settlement period was a continuation of the transition from native forests and shrublands to nonnative pasture grasses, the researchers suggested that the charcoal may have been derived from burning sugar cane rather than grasses [6].

Throughout the state, fire records from the past century indicate a more than 5-fold increase in the average area burned and the average number of fires per year; and even more pronounced increases in Hawai'i Volcanoes National Park (Table 3) [10,32]. Changes in fuel composition and spatial distribution coupled with increased human-caused ignitions are responsible for much of the increase [32]. Throughout much of the state, particularly in lowland and mesic communities (e.g., dry 'ohi'a woodland and coastal grasslands-scrub), highly flammable nonnative grasses have replaced less flammable native communities. Native forests and shrublands with discontinuous fine surface fuels that did not support fire spread [38] were invaded by nonnative grasses that form a continuous fuel bed highly capable of carrying fire [32,49]. Within Hawai'i Volcanoes National Park, the 3-fold increase in fire frequency and 60-fold increase in fire size since the early 1970s (Table 3) coincides with increased volcanic activity from Mauna Ulu and Pu'u O'o, the spread of nonnative invasive grasses, and the removal of nonnative, domestic goats (Capra hircus) [32,48,49], which had reduced fine fuels by grazing. Tunison and others [48] reported that in the coastal lowlands of Hawai'i Volcanoes National Park, fire frequency increased sharply following the removal of nonnative goats, the subsequent increase in grass biomass, and the spread of fire-tolerant nonnative species.

Table 3. Average acres burned and number of fires per year in Hawai'i and Hawai'i Volcanoes National Park from 1904-1995
Area	State of Hawai'i		Hawai'i Volcanoes National Park
Time period	1904 to 1939	1940 to 1976	1920 to ~1970	~1970 to 1995
Average area burned (acres)	1,044	5,740	<2.5	430
Average number of fires/year	4	24	11	39
Table modified from LaRosa and others [32]

A grass/fire cycle fueled by nonnative grasses and/or ferns has been identified in many 'ohi'a woodlands and lama forests [22,32,48,49]. Prior to the establishment and spread of nonnative grasses, 'ohi'a woodlands consisted of open stands of trees and shrubs with few native grasses. The sparse and discontinuous fine fuels were generally not conducive to fire spread [32,39,49]. In contrast, as of 1998, nonnative grasses such as broomsedge (Andropogon virginicus), bush beardgrass (Schizachyrium condensatum) and molasses grass (Melinis minutiflora) comprised 30% of the understory biomass and 65% to 80% of the understory cover in the 'ohi'a woodlands of Hawai'i Volcanoes National Park. These grasses form a continuous surface layer of fine fuels mixed with the native vegetation [12,15,49]. Similarly, Asian sword fern (Nephrolepis multiflora) has invaded mesic 'ohi'a woodlands [1]. Unlike the native vegetation, these nonnative grasses and ferns have characteristics that encourage fire spread and perpetuate repeated fires. They have high standing biomass with a year-round high dead-to-live fuel ratio, they can burn under high relative humidities (85-90%) and high fuel moistures (20-25%), and they have a continuous spatial distribution that facilitates fire spread. Because they can quickly regenerate, reproduce, and spread after fires, these nonnative species tend to dominate the postfire community, increasing the likelihood of repeated fires [1,14,22,32]. For additional information on the grass/fire cycle and the effects of nonnative grassland conversion in Hawai'i, see the following publications: [11,13,14,16,18,22,32,34,35,44].

• Presettlement fire regime
• Inferring fire regime characteristics from fire effects
• Post-Hawai'ian settlement
• Post-European settlement

Presettlement fire regime: Prior to human settlement (approximately 400 CE and later), fires may have occurred infrequently in the montane dry and mesic communities of Maui and Hawai'i (the two islands where these higher-elevation communities occur) [20,25,37,38,44]. Communities in this upland zone often have a pronounced dry season and are characterized by a mosaic of koa or 'ohi'a stands, shrublands, and grasslands. Infrequent fire occurrence is implied from 2 studies of soil profiles on the island of Hawai'i. On the southeast flank of Mauna Loa, in Hawai'i Volcanoes National Park, researchers found charcoal approximately 2,170 years old in 2 soil pits in forests at 4,000-4,300 feet (1,200-1,300 m), indicating that a fire burned forest vegetation in the area at around 220 BC [37] (Table 2). Soil pits dug in nearby savanna vegetation did not contain charcoal, but analysis of soil properties led the researchers to suggest that the savanna originated from the fire. At the Pohakuloa Training Area, Hawai'i, researchers found 18 pieces of macroscopic charcoal in 5 soil pits in dry montane 'a'ali'i and grass-dominated shrublands. The ages of charcoal fragments ranged from modern (2 pieces) to 7,730 years BP (1 piece); 13 pieces were older than 1,500 years BP [20,25] (Table 2). Fragments were dated at 8 distinct ages implying at least 8 distinct fires occurred prior to human colonization. Four fragments were identified as either 'akoko or 'ohi'a, but neither of these species occur there as of 2012. Kinney [25] suggests that present-day communities may be responses to "both recent anthropogenic fires and the deeply embedded volcanic history of the landscape". Ages of these charcoal particles predate the arrival of Polynesians to Hawai'i indicating that repeated fires occurred prior to human settlement. While these 2 studies indicate the occurrence of infrequent fire in montane communities, the scope of literature remains inadequate to characterize fire regimes in these communities.

Presettlement fires may have been important in the seasonal montane zone. Mueller-Dombois [38,41] based this inference on ecological characteristics of the native plant communities. The understory vegetation in the seasonal montane zone consists of stand-forming native ferns and grasses that could easily carry fire, and many of the native species in these communities recover rapidly after fire. Native stand-forming understory species present include decomposition brackenfern (Pteridium aquilinum var. decompositum), alpine hairgrass, parkland panicgrass (Panicum tenuifolium), and large Hawai'i lovegrass (Eragrostis grandis).

Literature describing presettlement fire regimes of alpine and subalpine communities was not available. However, alpine communities are characterized by a sparse to open cover of low shrubs and patches of grass [10], which suggests that they would be unlikely to carry fire.

Montane-subalpine shrublands on Haleakala, Maui. Photo courtesy of Amelia Leubscher.

No information was available regarding presettlement fire type and severity, or fire pattern and size in upland dry and mesic communities. For general information about presettlement ignition patterns and fire season in Hawai'i see: Ignition and Fire season, above.

Inferring fire regime characteristics from fire effects: While contemporary observations of fire in upland dry and mesic zones indicate that many native plants can survive and/or reproduce after fire (e.g., [38,51]), it is unclear whether the traits that enable survival are adaptations to fire, to some other local disturbance agent (e.g., volcanism, landslides, hurricanes, treefalls, etc.) [2], or to the species' pre-Hawai'ian evolutionary environments [44]. Koa forests of the montane mesic zone may have developed in the presence of fire. These forests do not appear to be substantially altered by fire [49], based on observed postfire recovery of koa, the dominant tree, and other native species. Several researchers have noted that shade-intolerant koa sprouts from root suckers and reproduces profusely from seed after fires [23,39,43,51]. Similarly, mamane, the other dominant tree in the montane zone, sprouts and recovers quickly after fire [48]. Tunison and others [49] note that plant community composition in shrublands and grasslands exhibits only subtle changes after fire, with a shift from dominance of the less fire-tolerant pukiawe to that of fire-tolerant 'a'ali'i; it is unclear whether this applies in the montane zone.

While individual native plants in upland dry and mesic communities may survive fire, the vegetation community may be negatively affected. Observations of a modern (1984) fire in the Kula Forest Reserve on Maui indicate that, despite postfire sprouting of many upland native shrubs, the postfire vegetation was substantially altered from its prefire composition. Varying degrees of postfire sprouting were observed in several upland native shrubs including: a'ali'i, mamane, 'aiakanene (Coprosma ernodeoides), sandalwood (Santalum haleakalae), 'ohelo, alpine mirrorplant (Coprosma montana), moutain dubautia (Dubautia menziesii), and pukiawe. However, not all of the individual shrubs, even the most resilient species, sprouted after fire. For example, the most common shrub, pukiawe, rarely sprouted. Additionally, thick postfire growth of nonnative common velvetgrass (Holcus lanatus) appeared to inhibit the growth and sprouting of native shrubs [33]. In the absence of competing nonnative species (e.g., in the presettlement and possibly pre-European settlement eras), native shrubs might recover and postfire upland communities would more closely resemble prefire communities (see Post-European Settlement).

Post-Hawai'ian settlement: Unlike the lowland communities of the Hawai'ian islands, upland vegetation was relatively undisturbed by Hawai'ian practices [26]. Fire frequency in upland communities may not have differed considerably from the presettlement period to the Hawai'ian settlement period. Dated charcoal from the montane shrublands of the Pohakuloa Training Area does not reveal a change in fire frequency from the presettlement to post-Hawai'ian settlement eras [20,25]. Human-ignited fires in lowland vegetation could have burned into upland communities, resulting in higher fire frequencies than the presettlement era, although this has not been documented (see Post-Hawai'ian Settlement in the section on fire regimes in lowland dry and mesic communities) [33].

Post-European settlement: Literature describing post-European fire regime characteristics in montane and subalpine communities is scarce, and available fire records are inadequate to describe changes from presettlement and Hawai'ian settlement conditions [32]. High-elevation, human-caused fires have been documented in and around Haleakala National Park, Maui [33], and in or around Mauna Kea Forest Reserve (MKFR), Hawai'i [47]. On Haleakala, physical evidence and anecdotal fire records from the past century indicate that most fires were small (up to 12 acres (5 ha)), but fires larger than 1,000 acres (400 ha) also occurred. These larger fires generally started in lower elevation ranchlands (probably comprised of nonnative grasses) and moved into subalpine shrublands [32,33]. Similarly, human-caused fires in or near MKFR tended to be small with the exception of some large fires (>1,000 acres (400 ha)) burning in pasture lands [47]. Although recent fires have been reported throughout the year, most upland fires occurred in the dry summer months or during periods of drought [33,47]. Within both Haleakala National Park and MKFR, nonnative grasses comprise a substantial part of the biomass and fuels. In communities that have been invaded by nonnative grasses, fire behavior, pattern, and size are likely to be altered from their pre-invaded states. While fires may have become more frequent over the past 2 centuries due to increases in human-caused ignitions and nonnative grass cover, insufficient information is available to characterize past fire regimes and thus detect changes in fire regime characteristics for upland dry and mesic communities [32].

• Presettlement fire regime
• Inferring fire regime characteristics from fire effects
• Post-Hawai'ian settlement
• Post-European settlement

Presettlement fire regime: Presettlement fire regime characteristics in wet communities are poorly understood [32,45]. Some scientists suggest that it is unlikely that fire played an important role in the development of Hawai'ian wet communities [38]. Prevailing climate and ecological conditions in these systems make fires unlikely or very infrequent. Rain forest vegetation is almost always wet. These environments have high precipitation rates and high relative humidities, and litter recycles rapidly [10] so fuels are generally not adequate to support fire. On the other hand, even the wettest communities experience drought [51], and uluhe (Dicranopteris linearis), a dominant native fern, can carry fire after extended dry periods in typically wet rain forests [10]. Although Vogl [51] was not able to find evidence of fire in wet montane environments, such as charcoal or lightning-struck trees, he suggests that fires likely occurred in these wet communities, albeit with long fire-return intervals. He reasons that rain forest can burn in drought conditions, dominant species can survive fire, and lightning occurs periodically. Vogl further contends that fires may be more frequent in windward wet areas than leeward sites because many leeward areas are occupied by barren lava flows or semi-desert, sparse vegetation cover [51].

Studies of sediment cores from wet regions indicate that fires occurred infrequently over the last 9,000 years (Table 2) [6,24,38,44]. A sediment core from Flat Top Bog, a high-elevation bog on Haleakala, Maui, dating back to over 9,000 years BP, shows microscopic charcoal particles throughout the core. Analysis of charcoal particles dated from the early-to-mid-Holocene suggests that they were derived from woody plants rather than graminoids. The researchers suggest that the charcoal particles were dispersed from fires in montane forests, but they do not specify whether these were rain forests or dry/mesic forests. The presence of pyroclastics and active volcanic vents on Haleakala implies that these fires were volcano-ignited, although it is possible that they were caused by lightning [6]. A soil core from a cinder cone bog in the Kohala Mountains on the island of Hawai'i, dating back approximately 1,600 radiocarbon years, had charcoal throughout it, implying periodic fires in the area [24]. Smith and Tunison [44] also indicate the probable occurrence of fire prior to human settlement. They report on carbon found "at depth" in a soil core from an O'ahu bog revealing "a 26,000-year record of fire occurrence", and undated charcoal from another soil core from Laupahohoe Forest Reserve on the island of Hawai'i. The location of Laupahoehoe Forest Reserve suggests that the charcoal was derived from wildfire rather than volcanic activity [44]. Mueller-Dombois [38] suggests that fire did not play an important role in the evolution of tropical wet communities; however, he provides evidence that fires occurred infrequently in Hawai'ian rain forest sites subjected to volcanism. Unpublished research from Lipman and others, cited by Mueller-Dombois [38], reported charcoal in 3 strata in a soil pit near the Ola'a Forest Tract in Hawai'i Volcanoes National Park. These charcoal layers were dated at approximately 340, 1,040, and 2,080±200 years old, yielding a fire-return interval of 700 to 1,000 years. Mueller-Dombois [38] correlates these fires with volcanism rather than drought, suggesting that rain forest vegetation will burn in lava-ignited fires but not in free-burning fires. Despite evidence that fires occurred infrequently in wet regions prior to human settlement, detailed information on fire regime characteristics including fire frequency and ignition sources is lacking.

Contemporary records of fires in native vegetation (i.e., records from the past century) indicate that fires in Hawai'ian rain forests generally occur during periods of drought and burn with high intensity [21,51]. Fuels in Hawai'ian rain forests include dense, continuous vegetation with abundant ladder fuels, suggesting that under dry conditions, fires can burn with high intensity. Hall [21] described a large (2-4 miles wide and 15 miles long) fire that occurred in 1902 and burned continuously, leaving only occasional unburned patches. The fire occurred after a several months of drought and burned rapidly through a dense forest of 'ohi'a and koa with a dense understory of ferns in southern Hamakua, Hawai'i. He also described another intense fire that occurred approximately 50 years earlier in the same area. He states the fire "killed practically all the forest and undergrowth, and consumed the humus. Its heat must have been intense, for it baked the soil to such an extent that at the present time it shows as a brick-like layer from 2 to 6 inches thick. In many cases it burned the roots of trees several feet below the surface". Although Hall [21] did not specify ignition sources, the vegetation description implies that these fires likely burned in native vegetation, although it is also possible that these forests contained nonnative plants that might have affected fire behavior and spread. Vogl [51] reports several fires on the windward slopes of Maui, O'ahu, and Kaua'i that occurred in the last century; however, information on ignition sources, fuel characteristics, and vegetation is not given. Because it is likely that these fires were human-caused and burned nonnative vegetation, inferences on presettlement fire regimes cannot be made. Similarly, recent lava-ignited fires in the wet region of the island of Hawai'i burned in communities with a considerable component of nonnative plants [2]. Thus, fire behavior, size, and pattern are likely different than under presettlement conditions.

'Ohi'a dominated rain forest. Photo courtesy of Forest and Kim Starr.

For general information about presettlement ignition patterns and fire season in Hawai'i see: Ignition and Fire season, above.

Details regarding presettlement fire frequency, type and severity, and pattern and size are not available for Hawai'ian wet communities.

Inferring fire regime characteristics from fire effects: While contemporary observations of fire in lowland and upland wet areas indicate that fires can burn in rain forests and that some native plants can survive fire (e.g., [2,51]), it is unclear whether the traits that enable survival are adaptations to fire, to some other local disturbance agent (e.g., volcanism, landslides, hurricanes, treefalls, etc.) [2], or to their pre-Hawai'ian evolutionary environments [44]. Several rain forest dominants possess traits that could be considered adaptations to fire [51]. 'Ohi'a can sprout from the base following fire [2]. Tree ferns including hapu'u (Cibotium glaucum) often refoliate rapidly after fire [2]. Uluhe climbs to form tall, dense, tangled mats that easily dry out and become highly flammable [51]. When uluhe rain forest is intact (i.e., not invaded by nonnative plants), uluhe and other pioneering native species can recover "rapidly" after fire [49]. Tunison and others [49] postulated that forests containing uluhe regenerated successfully after fire because the heat produced by the fire remained high in the uluhe canopy and did not affect the seed bank or fern rhizomes. On the other hand, native regeneration was disrupted on sites that contained nonnative invasive plants before fire because these nonnative plants sprouted, spread, and dominated the postfire community [49].

Post-Hawai'ian settlement: Between approximately 400 and 1800 CE, Hawai'ians settled, colonized, and reshaped much of the vegetation in Hawai'ian wet communities in low-elevation regions [26], while higher-elevation wet communities were little influenced by settlement.

The native vegetation in windward areas was wet forest [10], but by the time of European contact, virtually all valley bottoms with permanent stream flow had been transformed into networks of ponded and marshy habitat suitable for cultivating taro (Coslocasia esculenta) [26]. Areas 4 to 5 miles inland and below approximately 1,500 feet (460 m) elevation were transformed to unwooded grasslands, while forest remained at higher elevations [36]. Fire was probably used to clear low-elevation wet areas for cultivation [26,36]. Early visitors described an "expanse of unwooded grasslands or a 'plain' behind Hilo and in a band from Kea'au to Mountain View” on the windward side of the island of Hawai'i [36]. These grasslands were estimated to reach from 4 or 5 miles above the coastal village at Hilo Bay to approximately 1,500 feet elevation (460 m), the lower forest edge. McEldowney [36] stated that “the cumulative effects of shifting agricultural practices, prevalent among Polynesian and Pacific peoples, probably created and maintained this open grassland mixed with pioneering species and species that tolerate light and regenerate after a fire” (Table 2). Descriptions of burning and planting on the lower windward slopes of Mauna Loa and Kilauea on the island of Hawai'i are available, but the effects of these agricultural practices are less clear [36]. McEldowney's [36] claims that Hawai'ian land use practices converted formerly wet forests into open grasslands and maintained them as such are based on written descriptions of slash and burn agriculture, the presence of relict and second growth forests down to the coast, and evidence that rain forests can burn after dry winters. Evidence of Hawai'ian burning and forest clearing in low-elevation wet regions is also provided by a study of charcoal and subfossil snail deposits in Halawa Valley, Moloka'i . An excavation at the base of an alluvial fan indicated that upland slopes were burned and associated land snails and charcoal were deposited; soon afterward an alluvial mud containing abundant charcoal was deposited. The charcoal was dated at 772 ±90 years BP (~1100-1320 CE), during the period of rapid Hawai'ian settlement and expansion. Additionally, the native land snail assemblage indicated that the original vegetation in the lower slopes of the valley was comprised of native forest, while the subsequent absence of snails suggests barren upslope vegetation [27,28].

While observations from early explorers and anecdotal paleoecological evidence suggest widespread land clearing, agriculture, and burning in low-elevation wet regions, details of fire frequency or other fire regime characteristics during this period are lacking. In contrast, a sediment core from a high elevation bog on Maui did not show an increase in charcoal during Hawai'ian settlement, but this is likely due to the fact that Hawai'ians had little impact on high-elevation areas [6]. The lack of fire history evidence or observations in upland wet areas suggests that Hawai'ians did not actively burn upland wet communities.

Post-European settlement: Fire records from the past century indicate that few fires occurred in Hawai'ian wet forests prior to the 1990s [32]. However, these fire records only cover ~100 years, a small fraction of the (probably) much longer fire-return intervals of Hawai'ian wet communities. Burney and others [6] found consistently greater amounts of charcoal from the last 2 centuries than from earlier periods in a soil core from a high-elevation bog on Maui, but this is evidence of broad-scale trends rather than a definitive indication of fire frequency specific to montane rain forest and bog communities. In fact, the researchers suggest that the increase in charcoal in the post-European era is due to burning of sugar cane in lowland fields. It is likely that fire frequency in windward areas has increased in the past few centuries due to increased human-caused ignitions and altered fuel characteristics, but these fires are probably not burning in predominantly native communities. Since the 1990s, at least 13 small- to medium-sized (<30 to 5,000 acres (<12-2000 ha)) fires occurred in lowland and montane wet forests on the island of Hawai'i. These fires typically occurred following prolonged droughts and were fueled by native and nonnative plant species [32].

Nonnative plants have invaded most of the remaining Hawai'ian montane and lowland rain forest communities. Although many native plants in wet forests have traits that enable them to survive fire, their postfire recovery may be inhibited by the presence of nonnative species [1,2,7]. Research on the recovery of a wet forest 1 and 2 years after fire in the southeastern region of the island of Hawai'i indicated that many nonnative trees sprouted after fire, and nonnative grasses had higher cover on burned sites than unburned sites [1,2,3]. Burned sites had a higher proportion of nonnative plants than neighboring unburned sites, and cover of many nonnative species was significantly higher on burned than unburned sites [1,3]. Because nonnative grasses and ferns can rapidly invade or recolonize burned sites, recovery of native species may be reduced [3,7]. The 1983 East Rift eruptions on the island of Hawai'i ignited fires that burned several thousand acres of rain forest. Recovery of burned areas and colonization of new lava fields by native plant communities was inhibited by nonnative plants, which are generally more successful pioneers in postfire environments [7].

Despite limited information about fire regime characteristics, fire frequency appears to differ among presettlement, Hawai'ian settlement, and contemporary times in many Hawai'ian communities. This makes describing reference fire regimes in Hawai'i extremely difficult. Estimated reference conditions based on Biophysical Settings (BpS) models describe fire regime characteristics including fire regime group, mean fire-return interval (MFRI), and percent replacement fire (i.e., fire severity) (Table 1) [29]. It is unclear what time periods and vegetation conditions were used in estimating reference conditions for Hawai'ian communities because "LANDFIRE uses BpS to depict reference conditions of vegetation across landscapes. The actual time period for this data set is a composite of both the historical context provided by the fire regime and vegetation dynamics models and the more recent field and geospatial data used to create it" [30].

Information limitations in lowland dry and mesic communities: Archeological and paleoecological studies and anecdotal observations indicate that Hawai'ians employed widespread burning practices throughout much of the lowland dry and mesic communities (Table 2), but detailed descriptions of fire frequency or other fire regime characteristics are not available.

Reference conditions for lowland dry and mesic communities are extremely difficult to describe because fire frequency and vegetation characteristics differ among presettlement, Hawai'ian settlement, and contemporary periods. Reference conditions seem to fit some time periods better than others. For instance, according to LANDFIRE BpS model descriptions, MFRI ranges from 12 to 21 years for lowland dry communities (Table 1), a depiction of frequent fire more applicable to the post-Hawai'ian settlement period and/or current trends. This short MFRI contradicts Mueller-Dombois' widely cited review [38] that, prior to human settlement, dry lowland communities lacked the fine fuels necessary to carry fire and that fires would have been very unlikely.

Information limitations in upland dry and mesic communities: A few studies indicate that fire occurred infrequently in upland dry and mesic communities (Table 2) [20,25,37], but detailed descriptions of fire frequency or other fire regime characteristics are not available. Additionally, while contemporary fire records suggest that fire frequency in some upland communities may have increased over the past century, there are insufficient data to characterize presettlement and historic fire regimes and thus detect a change.

Recent research from Kinney [25] and Hall and others [20] may inform the MFRI for the Hawaii Montane-Subalpine Dry Shrubland BpS, which currently does not have estimated reference conditions (Table 1). This research was conducted in the dry montane-subalpine 'a'ali'i and grass-dominated shrublands of the Pohakuloa Training Area, Hawai'i, and revealed charcoal fragments dating back to 7,730 years BP, including fragments dated from at least 8 distinct fires (see Presettlement Fire Regime in the upland dry and mesic fire regime section).

Information limitations in lowland and upland wet communities: A few studies indicate that fire occurred infrequently in wet communities prior to human settlement (Table 2) [6,24,38,44]; and a few indicate that, during the Hawai'ian settlement period, Hawai'ians used fire as an agricultural tool in low-elevation wet areas [28,36]. Specific details of fire frequency, size, pattern, season, or severity are not available. Contemporary fire records and observations can rarely be used to describe historic fire regimes because recent fires behave differently due to the presence of nonnative plants. Recent fire records suggest that fire frequency in some wet communities may have increased over the past century, but few data are available to help evaluate this hypothesis.

Reference conditions for wet communities are extremely difficult to describe because of sparse information on fire occurrence in native communities and because fire frequency and vegetation characteristics differ among presettlement, Hawai'ian settlement, and contemporary periods. Due to this lack of information, fire regime reference conditions are only available for 2 of the wet community LANDFIRE BpS models.