The Contemporary Scale and Context of Wildfire in Hawai‘iing wildi cy, extent, and long term trends across Hawai‘ jor lands– ‘ ‘ ‘ ‘ ‘ i lier re lematic when compiling

The Contemporary Scale and Context of Wildfire in Hawai lsquo i 1

Clay Trauernicht 26 Elizabeth Pickett 3 Chris tian P Giardina 4

Creighton M Litton 2 Susan Cordell 4 and Andrew Beavers 5

Abstract Wildfire is a major threat to natu ral resources and native species in Hawailsquo i but the fre quency and extent of wildfires across the archipel ago has not been well quantified Our objective was to summarize the available wildfi re data for Hawailsquo i and synthesize the social and ecologi cal dimensions of wildfi re drivshyers impacts and man age ment responses We constructed a 110- yr span of wildfi re records for the state of Hawailsquo i to examine histori cal trends (1904ndash 2011) and sum marized relationships between contemporary wildfi re occur rence (2005ndash 2011) and land use land cover types and human popu lation Total area burned state wide increased more than fourfold from 1904 to 1959 to peaks in the 1960sndash 1970s and mid - 1990s to present From 2005 to 2011 on average 1007 wildfi res were reported across the state per year (plusmn 77 SE) burning an av erage of 8427 ha yrminus1 (plusmn2394 SE) Most fi res (95 ) were lt4 ha while most area burned (93 ) was attributed to fi res ge40 ha Ignition frequency was posi tively corre lated with human popu lation across islands Wildfires were most fre quent in developed areas but most areas burned oc curred in dry nonnative grasslands and shrublands that currently compose 24 of Hawailsquo i rsquo s total land cov er These grass- domi nated landscapes allow wildfires to propa gate rapidly from areas of high ignition freshyquencies into the forested margins of the statersquo s watersheds placing native habshyi tat watershed integrity and human safety at risk There is an urgent need to better assess fi re risk and impacts at landscape scales and increase the integrashytion of prefi re planning and prevention into existing land manage ment goals

R esearch on the interaction between nonnashytive grasses and fire in Hawailsquo i has brought widespread attention to the impacts of novel wildfi re re gimes on native ecosystems globally (Drsquo Antonio and Vitousek 1992) Yet the extent to which wildfire has become a pervasive feashyture of Hawailsquo i rsquo s landscapes has not been well quantifi ed Wildfire in Hawailsquo i differs greatly from that of most of the continental United

1 This work was made possible by funding from the Joint Fire Science Program and the US Department of Agriculture Forest Service Pacifi c Southwest Research Station to support the Pacific Fire Exchange and the College of Tropical Agriculture and Human Resources University of Hawailsquo i at Mānoa via the USDA National Institute of Food and Agriculture Hatch (HAW00132- H and HAW01127- H) Smith- Lever Act Formula Grant

Pacific Science (2015) vol 69 no 4427ndash444 doi1029846941 copy 2015 by University of Hawailsquoi Press All rights reserved

States in that fi re occurs year- round (Chu et al 2002) and is supported by vast unbroshyken expanses of highly fi re- prone nonnative eco systems domi nated by invasive grasses and other fi re- adapted plants (Elmore et al 2005 Varga and Asner 2008 Ellsworth et al 2014) ( Figure 1) These nonnative species regu larly encroach into Hawailsquo i rsquo s native ecosystems which contain one of the worldrsquo s highest rates

and Renewable Resources Extension Act (RREA) proshygrams Manuscript accepted 22 February 2015

2 Department of Natural Resources and Environshymental Management College of Tropical Agriculture and Human Resources University of Hawailsquo i at Mānoa Honolulu Hawailsquo i 96822

3 Hawailsquo i Wildfire Management Organization Kashymuela Hawailsquo i 96743

4 Institute of Pacific Islands Forestry Pacifi c Southshywest Research Station USDA Forest Service Hilo Hawailsquo i 96785

5 Center for Environmental Management of Milishytary Lands Colorado State University Fort Collins Colorado 80523

6 Corresponding author (e- mail trauerni hawaiiedu)

427

428 PACIFIC SCIENCE middot Oc to ber 2015

Figure 1 Fires in nonnative grasslands in Hawailsquo i illustrating (a) a high intensity experi mental fire in guinea grass (Megathyrsus maximus) in which flame heights reached 4ndash 5 m (poles at right are 3 m) under rela tively benign envishyronmental conditions (ie 8 km hrminus1 winds 70 rela tive humidi ty) (photo by C Trauernicht) and ( b) high fuel bed continuity evi dent in an aerial photo of a rela tively low intensity wildfi re burning in fountain grass (Cenchrus setashyceus) (photo by E Moller)

of species ende mism and the larg est proporshytion of endan gered plant species in the United States where they sup press native plant re-genera tion increase eco system fl ammability and fi re fre quency and accelerate rates of habshyi tat loss (Hughes et al 1991 Smith and Tunishyson 1992 Ainsworth and Kauffman 2013 D rsquo Antonio et al 2011) There is growing conshycern among natu ral resource man ag ers fi re responders researchers and landown ers about wildfi re impacts (DOFAW 2010) Howshyever resources for pre fi re miti gation and wildfi re response in Hawailsquo i are limited and given the statersquo s geo graphic isolation Hawailsquo i lacks access to the impressive array of inter-agency sup port for fi re manage ment available on the mainland United States (DOFAW 2014)

The human dimensions of wildfi re in Hawailsquo i have also received little attention deshyspite the domi nance of human- caused ignishytions and the contribu tion of land use change to the current extent of fi re- prone nonnative eco systems Before human arrival wildfi re igshynition sources in Hawailsquo i are thought to have been limited to volcanic activ ity and infreshyquent dry lightning strikes In line with this soil charcoal evi dence sug gests wildfi res were infrequent with highly localized effects (Smith and Tunison 1992 Burney et al 1995) and as a result many native Hawaiian plants have

limited adaptations to fire as a fre quent ecoshylogi cal disturbance (LaRosa et al 2008 but see Vogl 1969) Humans have increased wildshyfi re occur rence across the archipel ago both by greatly increasing ignitions and by introducshying fi re- prone plant species espe cially nonnashytive grasses Especially in dry and mesic areas and in all eco system types during droughts (Dolling et al 2005 Cram et al 2013) these factors currently drive recur rent wildfi res in Hawailsquo i that incur ecologi cal and socioecoshynomic costs in terms of watershed function natu ral resource degradation community safety and emer gency response (DOFAW 2010 2014)

The first reported ldquo disas trousrdquo wildfi re in Hawailsquo i was an escaped agricultural fi re in 1901 that burned gt12000 ha of agricultural and forested lands over 3 months on the Hāmākua Coast of Hawailsquo i Island (Commisshysioner of Agriculture and Forestry 1903) The Hāmākua Fire directly led to the establishshyment of Hawailsquo i rsquo s Forest Reserve System the integration of wildfire into gov ern ment forest man age ment poli cy and the initia tion of anshynual wildfire reporting in 1904 Previous studies have summarized statewide wildfi re frequency and extent (Burgan et al 1974 Cuddihy and Stone 1990 Chu et al 2002 Weise et al 2010) however they have not covshyered the full range of records and until now

Wildfire in Hawai lsquo i middot Trauernicht et al 429

there has been no available information on spatial patterns of fi re occur rence Our objecshytives are to synthesize previously available wildfi re information along with a newly availshyable spa tially explicit wildfi re history for Hawailsquo i (HWMO 2013b) to describe (1) wildshyfi re fre quency extent and long- term trends (2) social ecologi cal and histori cal fac tors contributing to wildfire risk (3) ecologi cal and social impacts of wildfire (4) responses to re duce wildfi re impacts and (5) rec ommenshydations for fi re manage ment and science moving forward

mate ri als and meth ods

All ana lyses consider only wildfi res de fi ned here as unplanned wildland fi res as op posed to prescribed experi mental or struc ture fi res We drew on multiple sources of fi re history data to achieve the fi rst objective of summashyrizing wildfi re fre quency extent and longshyterm trends across Hawailsquo i rsquo s six major islandsndash Kaualsquo i Olsquo ahu Maui Molokalsquo i Lanalsquo i and Hawailsquo i Island The lack of reporting esshype cially of small wildfires among earlier reshycords can be problematic when compiling and ana lyzing fi re histories In addition the data sets do not include wildfires set by trainshying exer cises on mili tary lands that result in numerous ignitions on Olsquo ahu and Hawailsquo i Isshylands (Beavers et al 1999 Beavers and Burgan 2002) Therefore the available data provide conservative estimates of wildfi re occur rence

We used annual area burned to examine histori cal trends (1904ndash 2011) in wildfi re ocshycur rence because it is less sensitive to undershyreporting of small fires as large fi res typ i cally account for gt 90 of total area burned (eg Niklasson and Granstr oumlm 2000) Annual summaries of area burned were drawn from three sources (1) a compilation of Hawailsquo i tershyritorial and state re ports of total area burned by wildfire for 1904ndash 1977 by Schmitt (1977 Schmitt data set hereafter) (2) Hawailsquo i Divishysion of Forestry and Wildlife annual reports for 1994ndash 2007 (available online at httpdlnr Hawaiigovforestryfi redata DOFAW data set hereafter) and (3) the fi rst spatially explicit (ie records with incident locations) fi re hisshytory for Hawailsquo i state for 1967ndash 2012 compiled

from agency records by the Hawailsquo i Wildfi re Management Organization (HWMO 2013b HWMO data set hereafter)

The Schmitt data set and the gov ern ment fi re re ports upon which it was based have been cited in previous pub lications (Burgan et al 1974 Cuddihy and Stone 1990) Over the 73- yr time span in the Schmitt data set 9 yr were missing data and 10 yr provided undershyestimates of total area burned as not all fi re re ports included fire size These underestishymated values are indicated and presented in the results (Figure 2) but were excluded from all ana lyses of trends in wildfires over time There were also eight instances in the Schmitt data set where total area burned was reported for a 2- yr time span (due to biennial reporting) in which case we split the value and al located half of the area reported to each of the 2 years The DOFAW data set proshyvided statewide annual summaries of total area burned and number of fires from 1994 to 2012 However data after 2007 were excluded as reports ceased to include county agency records after 2007 (W Ching pers comm)

In contrast to the Schmitt and DOFAW data sets in which only annual summaries were available the HWMO data set pro vided records of individual wildfi re incidents from all county state and fed eral fi re response agencies with the exception of the Department of Defense (Pierce and Pickett 2014) The HWMO data set contained state (DOFAW) records spanning the 47- yr period from 1967 to 2013 and National Park records from 1922 to 2012 but complete county fi re de partment records were avail able only from the early to mid- 2000s The HWMO data set included date of oc cur rence and fire size for 12906 wildfires with spatial locations of ignitions for 11109 reported fi res (86 of total) deshyrived from GPS coor di nates maps and or narrative accounts The cause of ignition was reported for 6218 reported fi res (48 of total)

To examine temporal trends in wildfi re ocshycur rence over the past century in Hawailsquo i (1904ndash 2011) we combined the Schmitt DOFAW and HWMO data sets and used a linear mixed model of area burned (transshyformed to area 02 to meet assumptions of hoshymo ge neity of vari ance) as a function of year


Figure 2 The available statewide record for annual area burned spanning 1904ndash 2011 using combined data sets from Schmitt (1977) online Hawailsquo i Division of Forestry and Wildlife (DOFAW) reports and the Hawailsquo i Wildfi re Management Organization (HWMO) Data were ana lyzed using a linear mixed model (dashed line) of area burned (transformed to area 02) as a function of year with data source as a random effect and assessed against the null model using Akaikersquo s information criterion (explained deviance [R2] = 020 Akaikersquo s weight [wi] = 091) In the case of biennial reporting (8 records in Schmittrsquo s data set) we split the bi ennial area burned value and allocated half of the area reported to each of the 2 yr Missing data points are indicated by ldquo xrdquo and years with under estimates of area burned are indicated by asterisks () and were excluded from the analy sis

with data source as a random effect Instead of arbitrarily excluding records for overlapshyping years we ana lyzed all available records for all years with data source as a random efshyfect and included a temporal corre lation term in our models (Pinheiro and Bates 2000) The random effect accounted for expected vari ashytion in the slope and inter cept among data sets while providing a pre dic tion of any overall temporal trend supported by the data whereas the corre lation term accounted for temporal au tocorre lation inherent in having multiple observations for overlapping years Years for which area burned was underestimated in the Schmitt database were excluded from the analy sis

To summarize cur rent wildfi re conditions state wide we used records from the HWMO data set spanning the years 2005ndash 2011 for which complete records from all agencies were available (N = 7054) We fi rst calcu lated mean values (plusmn1 SE) for number of ignitions and area burned per year for this 7- yr time span both across all records as well as across fi re

ignitions and area burned across months to characterize interan nual vari a tion We used information on the cause of ignition across all years in the HWMO data set to calcu late the percentage of fires resulting from natu ral causes arson or accidental human activ i ties

We also examined the spatial distribu tion of total wildfi re ignitions and total area burned

TABLE 1

Wildfires in Hawai lsquo i over the 2005 ndash 2011 Period Wildfire Size Classes Based on National Wildfi re

Coordinating Group Categories

Fire Size Class Area burned (ha) Fires (yrminus1) (ha yrminus1)

lt01 709 (plusmn 402) 274 (plusmn 27) 01ndash lt4 2479 (plusmn 393) 1506 (plusmn 223) 4 ndash lt40 323 (plusmn 58) 3875 (plusmn 693) 40ndash lt160 81 (plusmn 21) 4996 (plusmn 1288) 160ndashlt400 61 (plusmn 13) 11297 (plusmn 256) 400ndash lt2020 36 (plusmn 09) 30446 (plusmn 7374) gt2020 07 (plusmn 03) 31877 (plusmn 1465) Total 1007 (plusmn 771) 84270 (plusmn 23940)

size classes as de fined by the National Wild- Note From 2005 to 2011 most wildfi res (95 ) in Hawailsquo i were 4 ha or smaller whereas fi res gt40 ha accounted for 93 fi re Coordinating Group (Table 1) We also of the total area burned Values reported are statewide means

calcu lated mean values (plusmn1 SE) for number of (plusmn 1 SE)


across newly revised Hawailsquo i gap analy sis (HIshyGAP) land use land cover (LULC) types in a geographic information system (Gon et al 2006 unpubl revisions by J Jacobi) Imporshytantly because each record in the HWMO data set represents a sin gle ignition point atshytributed with area burned these summaries pro vide a rela tive description of area burned across LULC types as opposed to absolute values for the extent of each LULC affected by fire We also used the spatial distribu tion of wildfires from the HWMO data set to exshyamine the corre lation between ignition denshysity (fi res kmminus2) and pop u lation density per island for the six larg est islands in the archishypel a go In order to place Hawailsquo i rsquo s wildfi re summary in a national context we compared the percentage of total land area burned anshynually in Hawailsquo i with that of the entire conshytinental United States (including Alaska) and the 12 west ern - most states for each year from 2005 to 2011 using data from the National Inshyteragency Fire Center (WMFI 2012)

We used the results from these methods along with the available litera ture and discusshysions with vari ous fi re response and land man age ment pro fessionals to address the remaining objectives in the Discussion (2) contributing fac tors (3) impacts (4) responses and (5) future directions for fi re manage ment

in Hawailsquo i Nomenclature for all plant species re ferred to is derived from the Missouri Boshytanical Gardenrsquo s Tropicos database (http wwwtropicosorg)

re sults

Temporal patterns across the three available state wide fi re history data sets indicated a var-i able but overall increasing trend in annual area burned over time (Akaike weight [wi] gt 099 R2 = 028) (Figure 2) Fire activ ity reshymained rela tively consis tent for the fi rst half of the twentieth century after which there were two dra matic increases in area burned The fi rst occurred in the late 1960s and early 1970s peaking in 1969 at 18423 ha The secshyond began to increase in the late 1990s and peaked at 18752 ha in 2005

The HWMO data set for 2005ndash 2011 indishycated Hawailsquo i experienced an an nual average of 1007 (plusmn77 SE) wildfi res burning on av ershyage 8427 ha yrminus1 (plusmn2394 SE) Ninety- fi ve pershycent of these wildfi res were lt4 ha in size but 93 of the area burned annually resulted from wildfi res gt40 ha in size which accounted for only 18 of total ignitions (Table 1) Ignishytions occurred year- round with peaks in the summer (June July Au gust) and winter months (Janu ary December) whereas area

Figure 3 Mean monthly wildfi re ignitions and area burned (plusmn1 SE) in Hawailsquo i over 2005ndash 2011 (N = 7054) (HWMO 2013a) illustrate year- round wildfi re occur rence with ignition peaks in July and December Jan u ary and peak area burned during drier summer months


Figure 4 In Hawailsquo i over the period 2005ndash 2011 the majority of wildfi re ignitions (light gray bars) occurred in developed areas (66 ) whereas the majority of area burned (dark gray bars) occurred in dry grasslands and shrub-lands domi nated by nonnative grasses (54 ) Land use land cover classes are from Gon et al (2006 with unpubl revisions by J Jacobi) and wildfi re records are from HWMO (2013a) (N = 7054)

burned showed a large increase during the summer months (Figure 3)

Among the 6218 fi res for which cause of ignition was reported 81 were listed as unshyde termined or un known 16 were attributed to human accidental causes (eg campfi res welding vehicle engines) 2 were listed as arson and 15 were attributed to natu ral causes (lightning and lava) The spatial distrishybu tion of wildfi re ignition and area burned across LULC types from 2005 to 2011 reshyvealed very high num bers of ignitions in deshyveloped areas and ignition density was highly corre lated with popu lation density across isshylands (Pearsonrsquo s ρ = 099) (Figure 5) This reshylationship still held (Pearsonrsquo s ρ = 086) when O lsquo ahu was excluded as a potential out lier due to its extremely high popu lation density Most of the area burned was attributed to dry nonshynative (ldquo alienrdquo ) ecosystems espe cially dry nonnative grasslands (Figure 4)

The mean an nual area burned in Hawailsquo i from 2005 to 2011 (8427 ha yrminus1) accounted for 048 of Hawailsquo i rsquo s total land area which was greater than the pro portion of land area burned across the entire US mainland (030 ) and even across the 12 states in the fi re- prone western United States (046 inshycluding Alaska) over this same time period (WMFI 2012) Even for two of the worst fi re years on record for the continental United States in terms of area burned Hawailsquo i had a larger per centage of its land area burned (WMFI 2012) (Figure 6)

dis cus sion

The synthesis of the Schmitt DOFAW and HWMO data sets provides the most complete summary of wildfi re occur rence for the state of Hawailsquo i to date The fi rst fi ve decades (1904ndash 1959) of records average only 800 ha


Figure 5 There is a strong corre lation (Pearsonrsquo s ρ = 099) between popu lation density per island (from 2010 US Census) and the total ignitions per square kilometer during 2005ndash 2011 (N =7054) (HWMO 2013a) The overall pattern remains intact when Olsquo ahu is removed from the analy sis although the corre lation is slightly weaker (Pearshysonrsquo s ρ = 086)

burned per year and sug gest incidents such as the 12000 ha Hāmākua Fire of 1901 were rare events Most available evi dence indicates that prior to human arrival fires were infrequent and largely restricted to active volcanic events (Smith and Tunison 1992 Burney et al 1995) The dra matic histori cal increase in area burned (Figure 2) and the current preva lence

of wildfi re in Hawailsquo i (Table 1)mdash espe cially rela tive to the archipel a gorsquo s limited land area (16636 km2) (Figure 6)mdash indicate that the transformation of Hawailsquo i rsquo s landscapes since human arrival has greatly increased wildfi re oc cur rence Fire was undoubtedly a principle agricultural tool of early Polynesian settlers in Hawailsquo i however the introduction of most


Figure 6 The annual area burned as a percentage of total land area in Hawailsquo i (black bars) was on par or exceeded the 12 western - most states of the United States including Alaska (light gray bars) and the United States as a whole (dark gray bars) for all years in the 2005ndash 2011 period except 2011 (WMFI 2012) The years 2005 2006 2007 2011 and 2012 (not shown) were the worst wildfire years on record for the continental United States in terms of total area burned

fi re- prone species (eg African pasture grasses) and large- scale conversion of native eco systems occurred well after Euro pean conshytact near the turn of the nineteenth century We therefore first draw on the spatial patterns of wildfi re from the HWMO data set to unshyderstand the conse quences of these changes in the context of ignitions fuel availabili ty and climatemdash the fundamental driv ers of wildfi re oc cur rence (Bond and Keeley 2005) We then consider the current ecologi cal and socioecoshynomic impacts of wildfi re Hawailsquo i rsquo s capacity for wildfi re response and man age ment and finally point to current efforts to improve wildfi re manage ment in the state

Contributing Factors Human - Caused Ignitions

People are the primary driver of wildfi res in Hawailsquo i Natural causes of ignition are rare with lava currently restricted to active fl ows on Hawailsquo i Island and lightning strikes acshycounting for lt 02 of attributed wildfi re causes The infl u ence of peo ple also supported

the high frequency of ignitions in developed areas (Figure 4) and the strong posi tive corshyre lation with island popu lation (Figure 5) The develop ment of Hawailsquo i rsquo s agricultural economy from the mid - nineteenth to early twentieth centuries accelerated popu lation growth and likely had significant impacts on the number of wildfi re ignitions However the arrival of Polynesians as recently as 800 yr ago (Wilmshurst et al 2011) likely marked the fi rst significant increase in wildfi re activ ity in Hawailsquo i (Kirch 1982 Cuddihy and Stone 1990) In addition to agricultural uses there are accounts of larger- scale intentional landshyscape burning to manage plant resources such as the native pili grass (Heteropogon contortus) for thatching and other plants for food and an i mal fodder (McEldowney 1979 Kirch 1982) The effects of re peated intentional burning on vege tation in Hawailsquo i were likely simi lar to patterns observed else where in the Pacific Burning by Maori in New Zealand (McWethy et al 2010) by Fijians on Viti Levu (King 2004) and by Micronesians in Yap and


Guam (Dodson and Intoh 1999 Athens and Ward 2004) wasmdash and on some islands conshytinues to bemdash used for farming and natu ral resource manipu lation and has resulted in the expansion of fi re- adapted vege tation types (eg savannas) Given the scarcity of dry lightning strikes it is possible that fi re manshyage ment by Hawaiians also maintained cershytain grassland and shrubland areas with native fi re- tolerant species such as pili grass (eg Daehler and Goergen 2005)

Contributing Factors Fuel Availability and Landscape Change

The cur rent pre domi nance of area burned among Hawailsquo i rsquo s nonnative grasslands and shrublands illustrates how changes in land cov er espe cially since Euro pean arrival have also contributed to increases in wildfi re ocshycur rence (Figure 4) Nonnative grass lands and mixed nonnative grasslands shrublands are cur rently the statersquo s most extensive veg eshytation type cov ering 24 of the total state land area (gt400000 ha) (revised HI- GAP Gon et al 2006) These areas are domi nated by highly invasive fi re- prone grasses introshyduced by Euro pe ans since at least the nineshyteenth century (Ripperton et al 1933) such as guinea grass (Megathyrsus maximus [Jacq] B K Simon amp S W L Jacobs) fountain grass (Cenchrus setaceus [Forssk] Morrone) molasshyses grass ( Melinis minutiflora P Beauv) and buffel grass (Cenchrus ciliaris L) These speshycies form continu ous nearly mono typic fuel beds ignite easi ly attain extremely high fi ne fuel loads and are capable of resprouting and or establishing from seed more vigorshyously in the postfi re en vironment than is the majority of native vege tation (Freifelder et al 1998 Beavers et al 1999 Castillo et al 2003 Ellsworth et al 2014) Nonnative grasses also readily invade existing native woodlands (Drsquo Antonio et al 2000 Litton et al 2006 D rsquo Antonio et al 2011) with some species cashypable of col o nizing young bar ren lava fl ows Grass invasions else where in both temper ate and tropi cal ecosystems have led to simi lar ecologi cal changes in terms of increasing fi ne fuel loads and continuity and increasing resulshytant fi re intensity (Platt and Gottschalk 2001

Rossiter et al 2003 Hoffmann et al 2004 Veldman and Putz 2011) In Hawailsquo i the exshytensive area covered by nonnative grasses (Elmore et al 2005 Varga and Asner 2008) (Figure 1) sup ports frequent wildfires that kill woody plants and erode the mar gins of forshyested areas leading to increased grass cov er for est contraction and often forest fragmenshytation (Blackmore and Vitousek 2000 Ellsshyworth et al 2014) This invasive grass - wildfi re cycle has had large impacts on native dry forshyest and increasingly threatens mesic forests as well (Drsquo Antonio and Vitousek 1992 Ellsworth et al 2014)

In addition to increasing wildfire risk in reshymote areas fi re- prone grass species are now preva lent in much of the statersquo s ldquo wildland- urban interfacerdquo (WUI)mdash defined as the transition between developed and unoc cu-pied land (Butler 1974)mdash which in Hawailsquo i extends from areas where most of the statersquo s popu lation resides at lower ele vations upward into the lower mar gins of forested watersheds (HWMO 2013a) In con trast to WUI issues on the US mainland where the risk to valshyued resources posed by wildfire has increased as develop ment pushes into native fi re- prone habi tat the increasing wildfire risk in Hawailsquo i rsquo s WUI is more closely linked to agricultural abandonment and the expansion of fi re- prone vege tation primarily nonnative grasses Land use in Hawailsquo i underwent extensive change with the privatization of land in 1848 the disintegration of traditional Hawaiian sysshytems of watershed- scale land man age ment (Kamelsquo eleihiwa 1992) and the establishment of large- scale plantation agriculture and ranching By the mid - twentieth century active grazing lands had expanded to nearly 850000 ha (gt 50 of total state land area) and sugarshycane to more than 90000 ha (gt 5 of state land area) (Schmitt 1977) The sugar industry also introduced the pre harvest burning of cane fields which may have increased the risk of esshycaped fi res however the centrali zation of landholdings also simplifi ed access and comshymunication for wildfi re response via the coop-era tion of plantation fi re war dens with state agencies (DOFAW 2010)

More recently Hawailsquo i has under gone major declines in ranching and plantation


agriculture Between peak production estishymates in 1959ndash 1960 and surveys in 2012 acshytive crop land declined by 65 from gt 200000 ha to 70000 ha and active grazing lands deshyclined by 62 from 850000 ha to 324000 ha (Schmitt 1977 USDA 2012) It is estishymated that 90 of the gt760000 ha of suitshyable agricultural land in Hawailsquo i is currently fallow (Suryanata 2002) Unmanaged fallow agricultural lands in Hawailsquo i typ i cally beshycome domi nated by nonnative grasses (eg Cramer et al 2008 Veldman and Putz 2011) making agricultural abandonment a primary driver of the current domi nance of fi re-prone nonnative grasslands state wide (eg 24 of state land cover) Many of these lands are adja cent to housing and other infrastrucshytures which increases both the proba bility of ignitions and wildfire risk for communities Furthermore as agricultural lands go unshyman aged access roads and criti cal water reshysources (ie irrigation canals and reservoirs) be come degraded further compli cating wildshyfi re response Finally the division of agrishycultural lands into smaller par cels with the develop ment of new resi dential areas complishycates access for fi re responders and increases the proba bili ties of human- caused ignitions

Contributing Factors Climate

Climate is a central determinant of wildfi re oc cur rence and behavior (Krawchuk et al 2009) and climate change has been linked to re gionally specifi c increases in fi re activ ity (Moritz et al 2012) Fuel accu mulation and fuel moisture are driven by precipi tation temshyper a ture and humidity over short to medium time scales whereas wind speed and direction strongly infl u ence fi re intensity and the direcshytion and rate of spread at the time of fi re ocshycur rence Hawailsquo i experiences a wide range of climate conditions over rela tively small spashytial scales because of the islandsrsquo steep and complex topography domi nant northeast erly trade windndash driven weather patterns and asshysociated orographic precipi tation patterns Evidence sug gests these conditions have reshymained rela tively constant with cyclical inshycreases in pre cipi tation tied to decreased sea level during glacial periods over millennial

time scales (Gavenda 1992 Sheldon 2006) and to ldquo coolrdquo phases of the Pacific Decadal Oscilshylation on shorter time scales (Frazier et al 2012) However within the past century there is strong evi dence for both decreasing rainfall (Frazier et al 2011 Chen and Chu 2014 Kruk et al 2014 Elison Timm et al 2015) and inshycreasing temper a ture (Giambelluca et al 2008) The fi re history data demonstrate an increase in annual area burned and ignitions during the warmer drier conditions that preshyvail during sum mer months (Junendash Au gust) (Figure 3) In terms of interan nual vari abili ty larger fires have been previously corre lated to drought conditions in Hawailsquo i (Dolling et al 2005 Cram et al 2013) while Chu et al (2002) found a significant increase in area burned in the summers following El Nintildeo events These patterns sug gest that wildfi re occur rence may continue to increase across the state with fushyture warming and drying yet explicitly linkshying cli mate trends to past changes in wildfi re oc cur rence in Hawailsquo i is difficult due to the contemporaneous increases in both human popu lation and activ ity (ie ignitions) and the fl ammability of vege tation (ie nonnative grass expansion) across Hawailsquo i rsquo s landscapes over the past century

Ecological and Societal Impacts

The contemporary wildfi re re gime in Hawailsquo i affects natu ral and cultural resources on all isshylands with impli cations for human health and safety as well as economic develop ment Wildfire is a major cause of hab i tat degradashytion and native species decline in Hawailsquo i (see review by LaRosa et al 2008) and the potenshytial for wildfire to cause species extinctions is far greater in Hawailsquo i than in continental re gions The state contains 41 of the 845 federally listed threatened and endan gered plant species of which gt 50 are restricted to three or fewer pop u lations (US Fish and Wildlife Service unpubl data) making them extremely vulnera ble to a sin gle wildfi re Wildfi rersquo s complex interactions with nonnashytive plants and an i mals also pose unique challenges The status quo approach to conshyservation of bio logi cal diversity in Hawailsquo i is to fence areas and remove nonnative ungushy


lates such as feral sheep goats pigs and catshytle because these ani mals reduce native plant cov er re genera tion and survival (Mueller-Dombois and Spatz 1975 Cole and Litton 2014 Murphy et al 2014) However re movshying grazing ani mals may also release nonnashytive plants (Kellner et al 2011 Cole and Litton 2014) and could increase fine fuel loads and thus fi re intensity (eg inferred from greater flame lengths and rates of spread) when wildfi res occur (Blackmore and Vishytousek 2000) Ungulate removal is criti cal for native habi tat conservation however the trade- offs with wildfire risk need to also be asshysessed espe cially in drier ecosystems (eg Scowcroft and Conrad 1992 Thaxton and Jacobi 2009) to determine when and where fuels re duction measures are also re quired

In addition to the direct threat posed to native species wildfire can have nega tive impacts on criti cal ecosystem services that dishyrectly affect the quality of life for Hawailsquo i rsquo s resi dents and vis i tors The loss of veg e tation immediately following wildfire has been found to greatly increase soil erosion (Ice et al 2004) and the amount of sedi ment carried into streams and other water resources downslope in temper ate ecosystems (Neary et al 2003) In the Pacifi c ero sion from grass lands on Guam increased sixfold immediately after burning and remained twice that of unburned grasslands up to 18 months following wildfi re (Minton 2006) Over the lon ger term re ducshytions in woody plant cover due to re peated wildfires can decrease water table depth and in fi l tration into the soil both of which can dra mati cally increase discharge volumes and peak stormflow and hence downstream fl oodshying and sedi mentation during rainfall events (Le Maitre et al 1999) Postfi re impacts on landscape aes thetics have serious potential impli cations for Hawailsquo i rsquo s tourism industry whereas sedi ment loading in nearshore coral reef ecosystems threatens criti cal biodiversity food and recrea tional resources for res i dents and visi tors (Minton 2006 DOFAW 2010) Burned areas in Hawailsquo i re main closed to the pub lic for days to months (eg state lands folshylowing the 2007 Polipoli fire on Maui and the 2012 Hikimoe Ridge fire on Kauai) due to landslide and tree- fall danger thereby limitshy

ing access for hiking hunting gathering plants and tending of cultural sites Frequent fires in developed areas also impact power and communication infrastructure and lead to road closures and other munici pal challenges (Figure 4) Despite the spatial extent and dishyversity of resources currently affected by wildshyfire in Hawailsquo i there has been little integration of wildfire risk and impacts into assessments of ecosystem services community plan ning disas ter preparedness or climate change mitshyi gation

Prevention and Management of Wildfi re in Hawai lsquo i

The risks and impacts of wildfire can be reshyduced through expanded prevention more prefi re manage ment to improve the effectiveshyness of fi re sup pression and enhanced postshyfi re response Given that nearly all wildfi res in Hawailsquo i are human- caused outreach and ed-u cation are criti cal in Hawailsquo i and local proshygrams including DOFAW and HWMO have al ready adapted national resources such as Ready Set Go and Firewise Communities to local contexts (eg httphawaiiwildfi re orgready-set-go2html) Directly re ducing wildfire risk re quires managing vege tation to re duce the quantity and continuity of available fuels which has been shown to re duce costs and increase effectiveness of fi re sup pression elsewhere (Hurteau and North 2009 Syphard et al 2011) However evi dence for the beneshyfits of fuels manage ment is limited in Hawailsquo i (eg Beavers et al 1999 Castillo et al 2003 Ansari et al 2008) Traditional fuel breaks in which veg e tation is reduced or elimi nated by mechani cal and or chemical means can disshyrupt fuel continuity and pro vide fi re fi ght ers with access and defensible space but are costly to maintain in Hawailsquo i given the islandsrsquo difshyficult ter rain and rapid rates of year- round plant regrowth espe cially in nonnative grassshylands (Drsquo Antonio and Vitousek 1992)

These challenges have raised a call among Hawailsquo i rsquo s fi re manage ment community for cost- effective alternative fuels manage ment strat e gies (Pacifi cFireExchangeorg) Grazing has been prescribed to re duce fuels in grassshyland areas of Hawailsquo i although the trade- offs


between wildfire risk re duction and long- term forage sus tainability are not well quantifi ed (M Thorne pers comm) Still grazing has been shown to effectively reduce fi ne fuel loads through fi eld assessments (Blackmore and Vitousek 2000) experi mental prescriptive trials (Castillo et al 2003 Ansari et al 2008) and re mote sensing (Elmore et al 2005) Anshyother alternative is vege tated fuel breaks or ldquo greenstripsrdquo that integrate fi re- resis tant shrubs and grasses (St John and Ogle 2009) and or use trees to re duce her baceous fuels via shading (eg Trauernicht et al 2012) Howshyever plant species suitability and the effecshytiveness of greenstrip implementation in Hawailsquo i is only begin ning to be quantifi ed (Ellsworth 2012)

Postfire needs for soil stabilization and longer term revege tation have resulted in inshyvestments to increase seed storage capacity through both regional efforts such as the Hawailsquo i Island Native Seed Bank Cooperashytive and site- based projects such as the Pulsquo u Kukui Watershed Preserve on Maui (P Kaniaupio- Crozier pers comm) There is also increasing discussion among the manageshyment community about taking advantage of postfi re conditions to increase native species restoration and landscape resilience to future fi res (eg Loh et al 2009) Scaling up the cashypacity for ecologi cal restoration refor estation and watershed manage ment more gen erally can also inform postfi re strate gies The Aushywahi Dry Forest on Maui has excluded fi re-prone nonnative grasses on the order of hectares albeit through highly intensive weeding and planting of native trees and shrubs (Medeiros et al 2014) Larger- scale re-for estation projects such as those at Hawailsquo i Volcanoes National Park and Hakalau Forest National Wildlife Refuge on Hawailsquo i Island and the Kula Forest Reserve following the 2007 Polipoli fire on Maui have targeted tens to hundreds of hect ares (R Loh A Kikuta and L DeSilva pers comm) Reducing fi re risk however is more complex than sim ply ramping up restoration efforts For example high planting density and canopy closure are criti cal to re ducing fine fuel loads (McDaniel and Ostertag 2010 R Loh pers comm) Native plants used in restoration also vary in

their ability to recover from future fi res (Ainshysworth and Kauffman 2009 Loh et al 2009 Ammondt and Litton 2013) as well as their pro pen sity to burn The commonly used shrub species Dodonaea viscosa (L) Jacq for instance is extensively planted at dry for est sites but has been found to have extremely low live fuel moisture (Ellsworth 2012) Ulshytimately many ecologi cal restoration sites and criti cal watersheds in fi re- prone areas of Hawailsquo i will continue to depend on ldquo tradishytionalrdquo strat e gies such as fuel breaks and onshygoing coopera tion with fi re response agencies to protect these areas from wildfi re

Fire Suppression Challenges

Wildfires in Hawailsquo i rsquo s nonnative grasslands create difficult and danger ous con ditions for fi re sup pression personnel Rates of spread can be rapid and fi re intensities high (W Ching and M Nakahara pers comm) because of very high fine fuel loads (Beavers et al 1999 Castillo et al 2003 Ellsworth et al 2013) Furthermore Hawailsquo i rsquo s rug ged terrain proshyvides few op tions for safe ingress egress often limiting suppression efforts at a fi rersquo s leading edge to water drops from helicopters which sub stantially increases suppression costs Wildfi re response jurisdic tions in Hawailsquo i are split among vari ous agencies many of which maintain mutual aid agreements to share reshysources The four county fi re de partments (Hawailsquo i Kaualsquo i and Maui Counties and the City and County of Honolulu on Olsquo ahu) proshyvide initial response to the majority of wildshyfires in the state but they are primarily trained and equipped for structural fires The Hawailsquo i Division of Forestry and Wildlife is the prishymary fi re response agency for wildfi res on state lands (41 of state land area gt 640000 ha) and often assists county fi re de partments However DOFAW employs no full- time fi reshyfi ght ers acting instead as a reserve or militia-style response agency Federal agencies fund the only full- time wildfi re sup pression proshygrams in the state at Hawailsquo i Volcanoes Nashytional Park and the US Army garrisons on Hawailsquo i and Olsquo ahu Islands Importantly Hawailsquo i rsquo s ge ography also prevents the rapid mo bilization of heavy equipment such as


water trucks brush trucks and dozers among islands although personnel are typ i cally moshybilized in the event of large wildfi res

Future Directions

A significant challenge facing fi re manageshyment in Hawailsquo i is the lack of science- based re gionally rele vant information that is accesshysible to land manag ers and fi re responders (PacificFireExchangeorg) Our understandshying of wildfi re dynamics in Hawailsquo i and on tropi cal islands more gen eral ly is limited rel-a tive to the develop ment of fi re research and man age ment knowledge in continental ecoshysystems For instance established tools for fi re pre dic tion and opera tional response such as standard fuel and fire spread mod els and the National Fire Danger Rating System have uncertain appli cability in Hawailsquo i due to high environmental var i abili ty completely novel fuel types limited weather data and limited op portunities to conduct experi mental burns (Beavers et al 1999 Fujioka et al 2000 Benshyoit et al 2009 Weise et al 2010 Ellsworth et al 2013) Given the existing knowledge gaps and limited resources for projects it is imper a tive that efforts to fur ther develop fi re science in Hawailsquo i explicitly meet the needs of and be made accessible to on - the- ground practitioners To that end in 2011 the US Department of Agriculture Forest Service partnered with the nonprofit group Hawailsquo i Wildfire Management Organization and the College of Tropical Agriculture and Human Resources of the University of Hawailsquo i at Mānoa to create the Pacific Fire Exchange (Pacifi cFireExchangeorg) a member of the national Joint Fire Science Programrsquo s Fire Science Exchange Network (wwwfi rescience gov) The purpose of the Pacific Fire Exshychange is to improve communication between the science and practitioner communities in Hawailsquo i and US Affi liated Pacific Islands by increasing col lab o ration and develop ing and dissemi nating science- based best practices for fi re manage ment based on stakeholder needs

Due to the scale of Hawailsquo i rsquo s fi re- prone eco systems the complexity of factors driving wildfi re occur rence and limited available reshy

sources partner- driven and col lab o rative apshyproaches will be criti cal to improving fi re man age ment Partnerships such as the Big Isshyland Wildfire Coordinating Group the West Maui Fire Task Force and the Oahu Wildfi re Information and Education Group pro vide excellent examples of coopera tion among fi re response agencies The key resources imshypacted by wildfire such as freshwater provishysioning native forest habi tat and nearshore eco systems also indicate that fi re manageshyment goals are directly aligned with objecshytives identified by land and watershed man ag ers more broadly (eg DOFAW 2010) This provides op portunities to link wildfi reshyfocused partnerships with existing collab o rashytive land manage ment efforts such as the Hawailsquo i Association of Watershed Partnershyships to expand the scale and impact of wildshyfi re pre paredness and miti gation efforts The explicit relationship between wildfi re and human activ ity in Hawailsquo i (Figures 4 and 5) also strongly sug gests that reducing wildfi re impacts will re quire bridging the ecologi cal di mensions of land manage ment with social issues such as human behavior public safe ty edu cation eco nomics and disas ter response HWMO has played a piv otal role in this area through on- the- ground wildfi re ed u cation and the facili tation of Community Wildfi re Protection Plans across the state These efshyforts have brought to gether the interests and expertise of fi re responders large land manshyag ers and the general pub lic and pro vided impetus for collab o rative community- based wildfi re miti gation projects Additional opshyportunities to address and better understand the human dimensions of wildfire in Hawailsquo i include collab o rations with place- based ed u-cation programs that foster environmental and cultural stewardship and projects that adshydress landscape- scale issues of land developshyment and plan ning

con clu sions

Increases in wildfi re occur rence in Hawailsquo i (Figure 2) are likely to continue given a growshying human popu lation expanding invasive grass cov er and projected temper a ture inshycreases and precipi tation declines (Giambelluca


et al 2008 Frazier et al 2011 Elison Timm et al 2015) Wildfi re threatens human comshymunities and a multitude of nat u ral and cultural resources There are formidable chalshylenges facing fi re manage ment today and into the future in Hawailsquo i However aware ness of Hawailsquo i rsquo s wildfi re problem is spreading from those directly involved with fi re response into the wider land manage ment community polshyicy makers and the general pub lic Due to its disre gard for property lines and the diversity of resources it affects wildfi re pro vides an imshymediate and explicit incentive to bring fi re responders land manag ers researchers and local communities together to pro mote the pro tection of valued natu ral and cultural resources

ac knowl edg ments

Discussions with multiple fire response and land man age ment pro fessionals contributed to this work particu larly Wayne Ching (Fire Management Officer DOFAW) Miles Nakashyhara (DOFAW and HWMO re tired) Chief Eric Moller (US Army Garrison at Pohakushyloa Training Area) Chief Scott Freeman (US Army Garrison at Schofield Barracks) Ryan Peralta (DOFAW) Lance DeSilva (DOFAW) Jay Hatayama (DOFAW) Pat Porter (DOshyFAW) Jack Minassian (Hawailsquo i Community College) and members of the Big Island Wildshyfire Coordinating Group We would also like to thank the diverse array of per spectives proshyvided by the Pacific Fire Exchange Advisory Panel Thanks also to HWMO staff espe cially Orlando Smith and Pablo Beimler for data updates and quality con trol to the HWMO data set J B Friday Brett Murphy David Bowman Lisa Ellsworth and two anon yshymous reviewers pro vided constructive comshyments on ear lier versions of this manu script

Literature Cited

Ainsworth A and J B Kauffman 2009 Reshysponse of native Hawaiian woody species to lava- ignited wildfires in trop i cal forests and shrublands Plant Ecol 201197ndash 209

mdash mdash mdash 2013 Effects of re peated fires on nashytive plant community develop ment at

Hawaii Volcanoes National Park Int J Wildl Fire 221044ndash 1054

Ammondt S and C M Litton 2013 Restoshyration of native plant communities in a Hashywaiian dry low land eco system domi nated by the invasive grass Megathyrsus maxi-mus Appl Veg Sci 1629ndash 39

Ansari S H Hirsh and T Thair 2008 Reshymoval of invasive fi re- prone grasses to inshycrease training lands in the Pacifi c Department of Defense Legacy Project 07shy362 SWCA Environmental Consultants Honolulu Hawailsquo i

Athens J and J Ward 2004 Holocene veg-e tation savanna ori gins and human settleshyment of Guam Rec Aust Mus Suppl 2915ndash 30

Beavers A and R E Burgan 2002 Analysis of fi re history and manage ment con cerns at Pohakuloa Training Area Center for Enshyvironmental Management of Military Lands Fort Collins Colorado

Beavers A R E Burgan F M Fujioka R Laven and P Omi 1999 Analysis of fi re man age ment con cerns at Makua Military Reservation Center for Environmental Management of Military Lands Fort Colshylins Colorado

Benoit J F M Fujioka and D R Weise 2009 Modeling fi re behavior on tropi cal isshylands with high- reso lu tion weather data Pages 321 ndash 330 in A Gonzaacutelez- Cab aacuten ed Proceedings of the third international sympo sium on fi re eco nomics planning and pol i cy common prob lems and apshyproaches Gen Tech Rep PSW- GTR- 227 US Department of Agriculture Forest Service Pacific Southwest Research Stashytion Albany California

Blackmore M and P M Vitousek 2000 Cattle grazing forest loss and fuel loadshying in a dry for est eco system at Pursquo u Warsquo awarsquo a Ranch Hawaii Biotropica 32625ndash 632

Bond W J and J E Keeley 2005 Fire as a global ldquo herbivorerdquo The ecology and evoshylu tion of fl ammable ecosystems Trends Ecol Evol 20387ndash 394

Burgan R E F M Fujioka and G H Hishyrata 1974 A fi re dan ger rating system for Hawaii Fire Technol 10275ndash 281


Burney D A R V DeCandido L P Burshyney F N Kostel- Hughes T W Stafford and H F James 1995 A Holocene record of climate change fi re ecology and human activ ity from mon tane Flat Top Bog Maui J Paleolimnol 13209ndash 217

Butler C P 1974 The urban wildland fi re inshyterface Pages 1ndash 17 in Proceedings of Western States Section Combustion Instishytute papers 74 Western States Section Combustion Institute Pittsburgh Pennshysylvania

Castillo J G Enriques M Nakahara D R Weise L Ford R Moraga and R Vihshynanek 2003 Effects of cattle grazing glyphosphate and prescribed burning on fountain grass fuel loading in Hawaii Pages 230ndash 239 in Proceedings of 23rd Tall Timbers fi re ecology confer ence Fire in grassland and shrubland ecosystems Tall Timbers Research Station Tallahassee Florida

Chen Y R and P S Chu 2014 Trends in precipi tation extremes and return levels in the Hawaiian Islands under a changing clishymate J Climatol 343915ndash 3925

Chu P S W Yan and F M Fujioka 2002 Fire- climate relationships and long- lead seasonal wildfi re pre dic tion for Hawaii Int J Wildl Fire 1125ndash 31

Cole R and C M Litton 2014 Vegetation response to removal of non - native feral pigs from Hawaiian trop i cal montane wet for est Biol Invasions 16125ndash 140

Commissioner of Agriculture and Forestry 1903 Territory of Hawaii report of the Commissioner of Agriculture and Forestry for the biennial period end ing December 31st 1902 Gazette Print Honolulu Hawai lsquo i

Cram D S Cordell J Friday C P Giarshydina C M Litton E Moller and E Pickett 2013 Fire and drought in parashydisemdash say it isnrsquo t so smokey Rural Conshynect 719ndash 21

Cramer V A R J Hobbs and R J Standish 2008 Whatrsquo s new about old fi elds Land abandonment and ecosystem assembly Trends Ecol Evol 23104ndash 112

Cuddihy L W and C Stone 1990 Alterashytion of native Hawaiian veg e tation Effects of humans their activ i ties and introducshy

tions University of Hawailsquo i Press Honoshylulu

D rsquo Antonio C M R F Hughes and J T Tunison 2011 Long- term impacts of invashysive grasses and subse quent fire in seasonshyally dry Hawaiian woodlands Ecol Appl 211617ndash 1628

D rsquo Antonio C J Tunison and R Loh 2000 Variation in the impact of exotic grasses on native plant compo sition in relation to fi re across an ele vation gradient in Hawaii Austral Ecol 25507ndash 522

D rsquo Antonio C M and P M Vitousek 1992 Biological invasions by exotic grasses the grass fi re cycle and global change Annu Rev Ecol Syst 2363ndash 87

Daehler C C and E M Goergen 2005 Experimental restoration of an indigeshynous Ha waiian grass land after invasion by Buffel grass (Cenchrus ciliaris) Restor Ecol 13380ndash 389

Dodson J and M Intoh 1999 Prehistory and palaeoecology of Yap Federated States of Micronesia Quat Int 5917ndash 26

DOFAW 2010 Hawaii statewide assessment of forest conditions and trends State of Hashywaii Department of Land amp Natural Reshysources Division of Forestry and Wildlife Honolulu

mdash mdash mdash 2014 Five- Year Fire Protection Plan State of Hawaii Department of Land amp Natural Resources Division of Forestry and Wildlife Honolulu

Dolling K P S Chu and F M Fujioka 2005 A climatologi cal study of the Keetch Byram drought index and fi re acshytiv ity in the Hawaiian Islands Agric For Meteorol 13317ndash 27

Elison Timm O T W Giambelluca and H F Diaz 2015 Statistical downscaling of rainfall changes in Hawailsquo i based on the CMIP5 global model projections J Geo-phys Res Atmospheres 120 (1) 92ndash 112

Ellsworth L M 2012 Improved wildfi re man age ment in Megathyrsus maximus domshyi nated eco systems in Hawaii PhD diss University of Hawailsquo i at Mānoa

Ellsworth L M C M Litton A P Dale and T Miura 2014 Invasive grasses change landscape structure and fi re behavior in Hawaii Appl Veg Sci 17680ndash 689


Ellsworth L M C M Litton A Taylor and J B Kauffman 2013 Spatial and temporal vari ability in Megathyrsus maximus grassshylands on Oahu Hawaii Int J Wildl Fire 221083ndash 1092

Elmore A J G P Asner and R F Hughes 2005 Satellite moni toring of vege tation phenol ogy and fire fuel conditions in Hashywaiian dry lands Earth Interact 91ndash 21

Frazier A G H F Diaz and T W Giamshybelluca 2011 Rainfall in Hawailsquo i Spatial and temporal changes since 1920 Amerishycan Geophysical Union Fall meeting 2011 abstract GC21B- 0900

Frazier A G T W Giambelluca and H F Diaz 2012 Spatial rainfall patterns of ENSO and PDO in Hawailsquo i Ameri can Geophysical Union Fall meeting 2012 abshystract B13B- 0502

Freifelder R R P M Vitousek and C M D rsquo Antonio 1998 Microclimate change and effect on fi re following forest- grass convershysion in season ally dry trop i cal woodland Biotropica 30286ndash 297

Fujioka F M D R Weise and R E Burshygan 2000 A high res o lu tion fi re dan ger rating system for Hawaii 3rd Symposium on Fire and Forest Meteorology 9ndash 14 Jan-u ary Long Beach California

Gavenda R 1992 Hawaiian quaternary pashyleoenvironments A review of geologi cal pedo logi cal and botan i cal evi dence Pac Sci 46295ndash 307

Giambelluca T W H F Diaz and M S A Luke 2008 Secular temper a ture changes in Hawailsquo i Geophys Res Lett 35L12702

Gon S M A Allison R J Cannarella J D Jacobi K Y Kaneshiro M H Kido M Lane- Kamahele and S E Miller 2006 The Hawaii gap analy sis project US Geoshylogical Survey Honolulu Hawailsquo i

Hoffmann W A V Lucatelli F J Silva I N C Azeuedo M D Marinho A M S Albuquerque A D Lopes and S P Moreira 2004 Impact of the invasive alien grass Melinis minutifl ora at the savanna- forshyest ecotone in the Brazilian Cerrado Divshyers Distrib 1099ndash 103

Hughes F P M Vitousek and J T Tunison 1991 Alien grass invasion and fire in the

seasonal submontane zone of Hawailsquo i Ecology 72743ndash 747

Hurteau M and M North 2009 Fuel treatshyment effects on tree- based forest carbon storage and emissions under mod eled wildshyfi re sce narios Front Ecol Environ 7409ndash 414

HWMO 2013a Hawaii Wildfi re Manageshyment Organization Communities at risk from wildfire map Accessed 15 August 2014 httpwwwscribdcomdoc22834 7006

mdash mdash mdash 2013b Hawaii Wildfi re Management Organization Hawaii state wildfi re history data set Accessed 15 August 2014 http gisctahrhawaiieduWildfi reHistory

Ice G G D G Neary and P W Adams 2004 Effects of wildfire on soils and watershyshed processes J For 10216ndash 20

Kamelsquo eleihiwa L 1992 Native land and forshyeign desires Pehea la e pono ai Bishop Museum Press Honolulu Hawailsquo i

Kellner J G Asner and K Kinney 2011 Reshymote analy sis of bio logi cal invasion and the impact of enemy re lease Ecol Appl 62094ndash 2104

King T 2004 Fire on the land Livelihoods and sustainability in Navosa Fiji PhD diss Massey University New Zealand

Kirch P 1982 The impact of the pre historic Polynesians on the Hawaiian ecosystem Pac Sci 361ndash 14

Krawchuk M A M A Moritz M A Pashyrisien J Van Dorn and K Hayhoe 2009 Global pyrogeography The current and future distribu tion of wildfire PLoS ONE 4e5102

Kruk M A Lorrey G Griffiths M Lander E Gibney H Diamond and J J Marra 2014 On the state of the knowledge of rainfall extremes in the western and northshyern Pacifi c basin Int J Climatol 35 (3) 321ndash 336

LaRosa A J Tunison A Ainsworth J B Kauffman and R Hughes 2008 Fire and nonnnative invasive plants in the Hawaiian Islands biore gion Pages 225ndash 241 in K Zouhar J Smith S Sutherland and M Brooks eds Wildland fire in ecosystems Fire and nonnative invasive plants Gen


Tech Rep RMRS- GTR- 42 USDA Forest Service Rocky Mountain Research Stashytion Fort Collins Colorado

Le Maitre D C D F Scott and C Colvin 1999 A review of information on interacshytions between vege tation and groundwater Water SA 25137ndash 152

Litton C M D Sandquist and S Cordell 2006 Effects of nonnative grass invasion on aboveground car bon pools and tree popu lation structure in a trop i cal dry forshyest of Hawaii For Ecol Manag 231103ndash 113

Loh R A Ainsworth J T Tunison and C M Drsquo Antonio 2009 Testing native speshycies response to fi remdash a first step towards building fi re resilient plant communities at Hawaii Volcanoes National Park Tech Rep 167 Pacific Cooperative Studies Unit University of Hawailsquo i Honolulu

McDaniel S and R Ostertag 2010 Strateshygic light manipu lation as a restoration strat egy to reduce alien grasses and enshycourage native regenera tion in Hawaiian mesic forests Appl Veg Sci 13280ndash 290

McEldowney H 1979 Archaeological and histori cal litera ture search and research deshysign lava fl ow control study Hilo US Army Corps of Engineers manuscr 050879 Anthropology Department Bishop Mushyseum Honolulu Hawailsquo i

McWethy D B C Whitlock J M Wilmshurst M S McGlone M Fromont X Li A Dieffenbacher- Krall W O Hobbs S C Fritz and E R Cook 2010 Rapid landscape trans formation in South Island New Zealand following initial Polynesian settlement Proc Natl Acad Sci 10721343ndash 21348

Medeiros A C E I von Allmen and C G Chimera 2014 Dry for est restoration and unas sisted native tree seedling recruitment at Auwahi Maui Pac Sci 6833ndash 45

Minton D 2006 Fire erosion and sedi menshytation in the Asan- Piti watershed and War in the Pacific NHP Guam Tech Rep 150 Pacifi c Cooperative Studies Unit Univershysity of Hawailsquo i Honolulu

Moritz M A M A Parisien E Batllori M A Krawchuk J Van Dorn D J Ganz

and K Hayhoe 2012 Climate change and disruptions to global fi re activ i ty Eco-sphere 3 art 49

Mueller- Dombois D and G Spatz 1975 The in fl u ence of fe ral goats on the low land vege tation in Hawaii Volcanoes National Park Phytocoenologica 31ndash 29

Murphy M F Inman - Narahari R Ostertag and C M Litton 2014 Invasive fe ral pigs impact native tree ferns and woody seedshylings in Hawaiian for est Biol Invasions 1663ndash 71

Neary D G Gottfried and P Ffolliott 2003 Post- wildfi re watershed fl ood reshysponses Section 1B- 7 pages 1ndash 8 in Proshyceedings of the 2nd international wildland fi re ecology and fi re manage ment congress and 5th sympo sium on fire and for est meshyte o rolo gy Ameri can Meteorological Socishyety Boston Massachusetts

Niklasson M and A Granstr oumlm 2000 Numbers and sizes of fi res Long- term spashytially explicit fi re history in a Swedish boshyreal landscape Ecology 811484ndash 1499

Pierce A and E Pickett 2014 Building a spatial database of recent fi re occur rence for manage ment and research in Hawaii Fire Manag Today 7437ndash 42

Pinheiro J C and D M Bates 2000 Mixed effects models in S and S- Plus Springer New York

Platt W J and R M Gottschalk 2001 Efshyfects of exotic grasses on potential fi ne fuel loads in the groundcover of south Florida slash pine savannas Int J Wildl Fire 10155ndash 159

Ripperton J R Goff D Edwards and W Davis 1933 Range grasses of Hawaii Hashywaii Agric Exp Stn Bull 65 University of Hawailsquo i Honolulu

Rossiter N S Setterfield M Douglas and L Hutley 2003 Testing the grass- fi re cyshycle alien grass invasion in the trop i cal sashyvannas of northern Australia Divers Distrib 9169ndash 176

Schmitt R C 1977 Historical statistics of Hawaii University of Hawailsquo i Press Hoshynolulu

Scowcroft P and C Conrad 1992 Alien and native plant response to release from fe ral

444

sheep brows ing on Mauna Kea Pages 625ndash 665 in C P Stone C W Smith and J T Tunison eds Alien plant invasions in native eco systems of Hawailsquo i Management and research University of Hawailsquo i Cooperative National Park Resources Unit Honolulu

Sheldon N D 2006 Quaternary gla cial- inshytergla cial cli mate cycles in Hawaii J Geol 114367ndash 376

Smith C and J Tunison 1992 Fire and alien plants in Hawaii Research and manageshyment impli cations for native ecosystems Pages 394ndash 408 in C P Stone C W Smith and J T Tunison eds Alien plant invashysions in native ecosystems of Hawailsquo i Management and research University of Hawailsquo i Cooperative National Park Reshysources Unit Honolulu

St John L and D Ogle 2009 Greenstrips or vege tated fuel breaks TN Plant Mateshyrials 16 USDA Natural Resources Consershyvation Service Boise Idaho

Suryanata K 2002 Diversifi ed agriculture land use and agrofood networks in Hawaii Econ Geogr 788ndash 10

Syphard A D J E Keeley and T J Brenshynan 2011 Factors affecting fuel break efshyfectiveness in the control of large fi res on the Los Padres National Forest California Int J Wildl Fire 20764ndash 775

Thaxton J and J Jacobi 2009 Assessment of fuels potential fi re behavior and manageshyment op tions in subalpine vege tation on Mauna Kea Volcano Hawaii Hawailsquo i Coshyoperative Studies Unit Tech Rep 13 Unishyversity of Hawailsquo i at Hilo Hilo

PACIFIC SCIENCE middot Oc to ber 2015

Trauernicht C B P Murphy T E Portner and D M J S Bowman 2012 Tree covshyerndash fi re interactions pro mote the persisshytence of a fi re- sensitive conifer in a highly fl ammable savanna J Ecol 100958ndash 968

USDA 2012 USDA Census of Agriculture Accessed 15 March 2014 httpwwwnass usdagovData_and_Statisticsindexasp

Varga T and G Asner 2008 Hyperspectral and LiDAR re mote sensing of fi re fuels in Hawaii Volcanoes National Park Ecol Appl 18613ndash 623

Veldman J W and F E Putz 2011 Grassshydomi nated vege tation not species- diverse natu ral savanna replaces de graded tropishycal forests on the southern edge of the Amazon Basin Biol Conserv 1441419ndash 1429

Vogl R 1969 The role of fire in the evolu tion of the Hawaiian flora and veg e tation Pages 5 ndash 60 in Proceedings of the Tall Timbers fi re ecology confer ence Tallahassee Florida

Weise D R S L Stephens F M Fujioka T J Moody and J Benoit 2010 Estimashytion of fi re dan ger in Hawailsquo i using limited weather data and simu lation 1 Pac Sci 64199ndash 220

Wilmshurst J M T L Hunt C P Lipo and A J Anderson 2011 High- precision radiocarbon dating shows recent and rapid initial human colo nization of East Polyneshysia Proc Natl Acad Sci 1081815ndash 1820

WMFI 2012 WMFI fi re reporting Accessed 15 August 2014 https wwwnifcblmgov cgiFireReportingcgi


Figure 1 Fires in nonnative grasslands in Hawailsquo i illustrating (a) a high intensity experi mental fire in guinea grass (Megathyrsus maximus) in which flame heights reached 4ndash 5 m (poles at right are 3 m) under rela tively benign envishyronmental conditions (ie 8 km hrminus1 winds 70 rela tive humidi ty) (photo by C Trauernicht) and ( b) high fuel bed continuity evi dent in an aerial photo of a rela tively low intensity wildfi re burning in fountain grass (Cenchrus setashyceus) (photo by E Moller)

of species ende mism and the larg est proporshytion of endan gered plant species in the United States where they sup press native plant re-genera tion increase eco system fl ammability and fi re fre quency and accelerate rates of habshyi tat loss (Hughes et al 1991 Smith and Tunishyson 1992 Ainsworth and Kauffman 2013 D rsquo Antonio et al 2011) There is growing conshycern among natu ral resource man ag ers fi re responders researchers and landown ers about wildfi re impacts (DOFAW 2010) Howshyever resources for pre fi re miti gation and wildfi re response in Hawailsquo i are limited and given the statersquo s geo graphic isolation Hawailsquo i lacks access to the impressive array of inter-agency sup port for fi re manage ment available on the mainland United States (DOFAW 2014)

The human dimensions of wildfi re in Hawailsquo i have also received little attention deshyspite the domi nance of human- caused ignishytions and the contribu tion of land use change to the current extent of fi re- prone nonnative eco systems Before human arrival wildfi re igshynition sources in Hawailsquo i are thought to have been limited to volcanic activ ity and infreshyquent dry lightning strikes In line with this soil charcoal evi dence sug gests wildfi res were infrequent with highly localized effects (Smith and Tunison 1992 Burney et al 1995) and as a result many native Hawaiian plants have

limited adaptations to fire as a fre quent ecoshylogi cal disturbance (LaRosa et al 2008 but see Vogl 1969) Humans have increased wildshyfi re occur rence across the archipel ago both by greatly increasing ignitions and by introducshying fi re- prone plant species espe cially nonnashytive grasses Especially in dry and mesic areas and in all eco system types during droughts (Dolling et al 2005 Cram et al 2013) these factors currently drive recur rent wildfi res in Hawailsquo i that incur ecologi cal and socioecoshynomic costs in terms of watershed function natu ral resource degradation community safety and emer gency response (DOFAW 2010 2014)

The first reported ldquo disas trousrdquo wildfi re in Hawailsquo i was an escaped agricultural fi re in 1901 that burned gt12000 ha of agricultural and forested lands over 3 months on the Hāmākua Coast of Hawailsquo i Island (Commisshysioner of Agriculture and Forestry 1903) The Hāmākua Fire directly led to the establishshyment of Hawailsquo i rsquo s Forest Reserve System the integration of wildfire into gov ern ment forest man age ment poli cy and the initia tion of anshynual wildfire reporting in 1904 Previous studies have summarized statewide wildfi re frequency and extent (Burgan et al 1974 Cuddihy and Stone 1990 Chu et al 2002 Weise et al 2010) however they have not covshyered the full range of records and until now
















TABLE 1











re sults









dis cus sion









































Future Directions




con clu sions




ac knowl edg ments


Literature Cited



















































































444
































TABLE 1











re sults









dis cus sion









































Future Directions




con clu sions




ac knowl edg ments


Literature Cited



















































































444























TABLE 1











re sults









dis cus sion









































Future Directions




con clu sions




ac knowl edg ments


Literature Cited



















































































444





















re sults









dis cus sion









































Future Directions




con clu sions




ac knowl edg ments


Literature Cited



















































































444






















dis cus sion









































Future Directions




con clu sions




ac knowl edg ments


Literature Cited



















































































444
























































Future Directions




con clu sions




ac knowl edg ments


Literature Cited



















































































444




















































Future Directions




con clu sions




ac knowl edg ments


Literature Cited



















































































444














































Future Directions




con clu sions




ac knowl edg ments


Literature Cited



















































































444







































Future Directions




con clu sions




ac knowl edg ments


Literature Cited



















































































444
































Future Directions




con clu sions




ac knowl edg ments


Literature Cited



















































































444

























Future Directions




con clu sions




ac knowl edg ments


Literature Cited



















































































444



















Future Directions




con clu sions




ac knowl edg ments


Literature Cited



















































































444



















ac knowl edg ments


Literature Cited



















































































444























































































444
































































444








































444

















444

















Documents

The Contemporary Scale and Context of Wildfire in Hawai‘iing wildi cy, extent, and long term trends across Hawai‘ jor lands– ‘ ‘ ‘ ‘ ‘ i lier re lematic when compiling