Lovreglio R, Leone V, Giaquinto P, Notarnicola a Eries

Standard Article - doi: 10.3832/ifor0521-003 ©iForest – Biogeosciences and Forestry

Wildfire cause analysis: four case-studies in ficant prevention effort”. Similarly, it has been stated that “Until our southern Italy ability to determine the causes of forest fires improves, our efforts at prevention will es

(1) (1) (1) (2) sentially remain shots in the dark…” (Envi Lovreglio R , Leone V , Giaquinto P , Notarnicola A ronment Policy 2003). Causes are more diverse than is often as Forest fires in Italy are mainly caused by humans, and directly depend on so sumed to be the case, and fire initiation is cial behavior, whether voluntary (arson) or involuntary (negligence). Despite neither as random, nor, in some cases, as the progress in knowledge made with studying the physical facets of the phe meaningless as some analysis suggest (Le nomenon, causes and motives of human-related fire remain mostly unknown. one et al. 2003). Understanding the reasons This paper proposes the implementation of the Delphi method (an interactive why fires start is a crucial factor to prevent expert-questionnaire process) in order to assess why fires are ignited. In four or reduce their incidence. It is important to study cases, within a high fire incidence area (southern Italy), the Delphi tech avoid the piecemeal approach to fire which nique identified as major cause of negligence the use of fire in agriculture and, concentrates mainly on fire suppression as major causes of voluntary fires, motives related to seasonal labor. The main (Wilson 1976), and hence mainly focuses on results in terms of frequency are: (i) for involuntary events (negligence), ex fighting fires and developing the infrastruc perts unanimously identified the relevant importance of negligent use of agri ture to detect and facilitate access to fire, cultural fires, particularly stubble burning (13.99% of responses). For (ii) vo whereas inadequate attention is given to ad luntary fires (arson), results highlight the relative importance of fires ignited dressing the underlying causes of harmful by seasonal workers as an instrument to force or maintain employment (8.41% fires (Jackson & Fisher 2001). of responses). Databases today contain a high percentage of unknown causes of wildland fire. In many Keywords: Delphi method, Fire causes, Agricultural fires, Ignition motives, Ar countries the percentage of unknown fire son fires, Negligent fires, Panel of fire experts causes reaches up to 70% or 80% (UNECE 2008). No proper wildland fire planning can be done with such a degree of uncertainty. Introduction Forest fires are neither a natural occurrence Understanding the motive behind the cause Worldwide it has been determined that hu nor a natural disaster, with the exception of of fires may enhance investigative efforts mans are responsible for about 90% of bio those fires started by natural agents.They and focus on improving prevention through mass burning, with only a small percentage are, on the contrary, an anthropogenic phe social behaviour modifications. of natural fires contributing to the total nomenon which exclusively and directly de There is an ongoing need to improve the amount of vegetation burned. pends on social behavior, whether it be vo knowledge of this point, which is a pre-con The FAO Fire Management and Global As luntary or involuntary (Leone et al. 2002). dition for the implementation of suitable sessment 2006 Report offers some rough es Today we have knowledge of some aspects solutions. timations on fire causes in the world. Ac of fire: “what”, “when”, “where” it happens. The main causes of wildfires in Italy, as cording to this report, the Mediterranean re The knowledge of fire, as a complex physi described in official statistics, follow the ca gion accounts the larger proportion of human cal-chemical reaction, allows good predic tegories suggested by EEC Regulation No caused fires in the world (95%) followed by tion of its behavior, through the use of a 804/94 (no longer in effect) which esta South Asia (90%), South America (85%) and wide set of fire behavior/spread models that blished a Community system of information Northeast Asia (80% - FAO 2007). Surpris have been implemented in the United States, on forest fires. ingly, wildfires are still considered as natural Canada, Australia, Spain, Greece since the In 2001 the first complete survey of wild hazards, even by some authoritative sources 1950’s and 60’s. fires carried out by the State Forestry Service (NASA 2003); even in the EU, in different Remote sensing and satellite imagery assist assessed that the main cause of fires was ar Frame Programs such as FP6 and FP7, still to determine the time, place and extent of son, which caused 59.3% of the fires. The in progress, fires are regarded as natural ha active fires around the world, with special rest of the fires were a result of negligence zards. emphasis on their distributive and seasonal (17.8 %), natural (0.7 %) or unknown causes pattern, on energy, gas release, etc. (NASA (22.2 % - MIPAAF-CFS 2002). (1) Department of Crop Systems, Forestry 2009). On a longer period, the Italian fire database and Environmental Sciences, University of Yet we still do not know enough about who for the most recent period (1997-2007), Basilicata, v.le dell’Ateneo Lucano 10, I- starts wildfires and why. What we know when causality records are more reliable, 85100 Potenza (Italy); (2) Department of about the subject is mainly included in a list shows the following official percentage of Psychology, University of Bari, p.zza of credible (and sometimes obvious) motives causes: Umberto I 1, I-70122 Bari (Italy). for starting fires. What we do not have is the • negligence: 17.27 % capability of filling the gap between the ad • voluntary: 58.96% @ Raffaella Lovreglio vanced knowledge of the physical facets of • natural: 1.23% ([email protected]) the phenomenon and the very reasons of its • unknown: 22.54%. Received: Aug 29, 2009 - Accepted: Oct 28, genesis (Leone & Lovreglio 2003). The motives for arson were grouped into 2009 FAO (1999) notes that “It is next to im three categories, namely: profit-seeking, Citation: Lovreglio R, Leone V, Giaquinto P, possible to design specific fire prevention manifestations of protest, resentment or in Notarnicola A, 2010. Wildfire cause analysis: campaigns if one cannot identify the causes sensitivity toward forests. four case-studies in southern Italy. iForest 3: of wildfires in a systematic way. If critical The main category was revealed as “profit- 8-15 [online: 2010-01-22] URL: fire starting causes remain unknown, then it seeking”, to which 2.992 incidents have been http://www.sisef.it/iforest/show.php? becomes extremely difficult to mount a signi attributed, equivalent to 69.9% of arson id=521

© SISEF http://www.sisef.it/iforest/ 8 iForest (2010) 3: 8-15 Lovreglio R et al. - iForest 3: 8-15 cases and 41.9% of total fires; 1 085 fires (Baughman 1989). fying real and perceived human motives; this (25.4% of arson cases and 15.3% of total The Delphi technique uses a panel of care latter feature correctly fits the analysis of fire fires) have been attributed to “manifestations fully selected experts who answer a series of motivations in areas where fires frequently of protest, resentment and insensitivity to questions, through either correspondence or occur (Linstone & Turoff 2002) but with ward forests”. face-to-face discussion. The accuracy of the little knowledge about their motives. Of the three categories taken into consider prediction depends on competence, expe Delphi has been used in many fields of En ation of fires caused by negligence, the most rience, objectivity and perception of the di vironmental Sciences: recreation (Anderson frequent was the category of fires caused by scerning judge. & Schneider 1993, habitat suitability evalu agricultural and forestry activities (1 492 in Each round of questioning is followed by ation (Crance 1987), quality of habitat cidents, equivalent to 60.8% of fires in feedback on the preceding round of replies, (Schuster et al. 1985), and fishery (Zuboy volving negligence), followed by the ca usually presented anonymously. The experts 1980). tegory of fires caused by cigarette stubs and are encouraged to revise their earlier answers In the field of forest fires (De Las Heras et matches (552 incidents or 22.5% of fires by in light of the other group member’s re al. 2007), the Delphi technique has been negligence). sponses. It is believed that during this pro used in a few cases in the United States to As for motivations behind causes, only cess the range of answers will decrease and explore the current state of wildland fire since 2004 have standard forms for forest the group will converge towards the “cor communication from the perspective of wild fire statistics been implemented to verify the rect” answer. After several rounds, the pro land fire communicators (Clute 2000); in proven or suspected motive of each fire, fol cess is complete and the average scores de Spain (FAB Consultores 1990) it has been lowing a list of official denominations which termine the final answers. used to analyze the causes of fires in the are reported in Annex 1, together with their The Delphi technique may be used in areas Balearic Islands, in a survey of arson fires four digit code. where there is an absence of sufficient data (ICONA 1995), in a survey on fire preven In any case assessment of causes, mainly of and/or an incomplete theory on cause and ef tion and communication (Ministerio de Me voluntary fires, is usually in the opinion of fect in regard to the phenomena under study. dio Ambiente 1997) and, more recently, in a the reporting forestry officer filling the fire Sitting between knowledge and speculation, survey on the perception of forest fires by statistic form and secure determinations are the informed deliberations of the panel of ex the Spaniards (APAS 2003, Dolzreuss & made in a minority of cases, as in every perts may best be considered an informed Irastorza 2005). country, only when culprits are brought to judgment. In Italy, the only known implementation of justice . This occurs in Italy in about 7% of The Delphi procedure has become a popu the technique is by the authors (Leone & the events (Corrado 2008). Many doubts re lar tool in technological forecasting and a Lovreglio 2003, Lovreglio et al. 2006, main therefore about reliability of the distri single definition is no longer appropriate for Lovreglio et al. 2008) firstly in the Fire Con bution of fire frequency by causes, mainly its numerous applications. trol Plan for the Gargano National Park, for voluntary fires (58.96% of total), which Distinguishing features of the Delphi te where the Delphi method and kernel density are probably overemphasized (Bovio 2005). chnique are: estimation technique permitted to analyze in Our study aims to contribute to a better • anonymity; depth the complex dynamics of an extreme knowledge of motives of fire causes, sug • iteration with controlled feedback; fire-prone territory. gesting the use of a rather simple but inter • statistical group response; The Delphi technique, for the assessment esting tool: the Delphi technique. • expert input. of fire motives in this paper, has been imple The Delphi technique generally includes mented in four different areas in Southern Materials and methods several steps: Italy, all of them marked by a more or less • the specification of a topic or subject be in severe fire occurrence (Fig. 1). The Delphi Technique vestigated; The Delphi technique was originally de • the construction of an ad hoc questionnaire Study area veloped in the 1950s by Olaf Helmer and for data collection; Study areas, which have been selected in Norman Dalkey, scientists at the Rand Cor • the selection of a panel of experts on the terms of increasing territorial surface, vary poration, as an iterative process for fore topic being investigated; ing from 100 000 to 1 000 000 hectares, are casting the likelihood that certain events will • the weighting of the opinions of the ex as reported below. occur; namely, the effects of a massive perts by means of the questionnaire; atomic attack on the United States. Project • the summary of the data resulting from the Area 1 Delphi was the name given to an Air Force- initial measurement; Comunità Montana Vallo di Diano (Moun sponsored Rand Corporation study, starting • the communication of the results of the ini tain Community are the Upland Develop in the early 1950’s, concerning the use of ex tial weighting of opinions as feedback to ment Authorities established by the Italian pert opinion to “obtain the most reliable con all the respondents; Law on the Development of Highlands, Law sensus of opinion of a group of experts by a • a re-evaluation of the opinions of the re 1102 of 1971), hereafter referred to simply series of intensive questionnaires inter spondents, as they have been informed and as C.M. Vallo di Diano, inside the National spersed with controlled opinion feedback.” may have been changed by their know Park of Cilento and Vallo di Diano, in the The subject of this study was the application ledge of earlier results including of other province of Salerno, region Campania. of “expert opinion to the selection, from the respondents’ supporting comments for C.M. consists of 15 municipalities, for a point of view of a Soviet strategic planner, of their opinions; total surface of 71 838 hectares, with me an optimal U.S. industrial target system and • an analysis, interpretation, and presentation dium high elevation (550 to 1.900 m a.s.l.) to the estimation of the number of A-bombs of the data and the writing of a final report. and an extreme climatic variability. required to reduce the munitions output by a While the Delphi is considered a forecas Forest cover is 27 884 ha, i.e., 39% of the prescribed amount” (Turoff 1970). ting procedure due to its significant use in total area; about half of them are transition It is a study method which generates ideas that area, there is a variety of other applica forest, developed from secondary succes and facilitates consensus among individuals tion areas among which we find developing sion. who have unique knowledge to share, de causal relationships in complex economic or Average number of fires per year (1998- signed as a means of scientific prediction social phenomena, distinguishing and clari 2004): 45.

iForest (2010) 3: 8-15 9 © SISEF http://www.sisef.it/iforest/ Wildfire cause analysis in Italy

Area 2 Fig. 1 - Map of the stud National Park of Gargano, in the province ied areas. of Foggia (region Apulia). Gargano is the large limestone promontory which juts far out into the Adriatic, on Italy’s East coast. Land use is very complex, with many per manent meadows of primary origin vegeta ting in the upper part, intermingled with woods, mainly high forests and coppices, which cover about 50 640 hectares. The territory was originally devoted to pas ture, with croplands only in the peripheral strip. In the inner part, the calcareous basement is partially covered by deciduous broadleaved forests, ranging from high stands of Fagus sylvatica L. gradually succeeded by dense, Quercus cerris L. forests, to scrubby Mediterranean macchia, the most wide spread form of natural vegetation in the ty pically Mediterranean parts of the area. A wide peripheral strip of low elevation, high forest of native Pinus halepensis Mill. cha racterizes the area. The National Park of Gargano consists of 19 municipalities and about 262 000 hectares of land. The average number of fires per year (1974-2001): 152. Gargano is among the most severely fire affected areas in Italy and probably in the Mediterranean basin (Leone et al. 2002). As gion with a total area of 999 461 hectares, impact (forest burned area / year / forest a matter of fact, in the summer of 2007 the situated in southern Italy, with two short land). territory of the National Park once again was coastlines on both the Ionian and the Tyrrhe Sites 3 and 4 are in the Apulia Region, ravaged by a disastrous voluntary fire which nian Sea. Its western part, province of Po which has the minimum percentage of fore caused the loss of lives and enormous dam tenza, is roughly mountainous, with eleva sted land (forestry ratio) at the national level ages. tions over 2000 m. a.s.l. and is covered by (only 7.7% vs. a national value of 28.8%), about 270 000 hectares of forest. The eastern but a percentage of voluntary fires much Area 3 part is a rather flat province (Matera), with a above the national average (MIPAAF-CFS Province of Bari, in the Apulia region. On much less important forest surface (80 657 2002) and relatively to forestry ratio the the south eastern tip of Italy, the province of hectares) as a result of a historically dramatic highest percentage of burned forested land Bari, in the region Apulia, is a wide area of deforestation carried out in the XIXth cen among the Italian regions (Leone 1997b). 513 831 hectares. The terrain is mainly flat, tury, intended to recuperate agricultural land All areas are economically marginal, being with its highest elevation being the inner for extensive wheat cultivation purposes. included in the fourth quartile in the eco Murge Plateau, a wide and tabular (maxi The region includes two National Parks: nomic classifications of European Union mum elevation 686 m. a.s.l.) calcareous high the National Park of Val d’Agri and Lago (EU) Regions (NUTS II) and at the lower plateau, poorly developed and scarcely in negrese and the National Park of Pollino, the end of the national level. habited, originally devoted to pasture, now widest National Park in Italy; in the latter Having a Gross Domestic Product (GDP), included in the National Park of Alta Mur rare and precious endemic species grow such which is the basic measure for economic gia. as Pinus heldreichii H. Christ, 1863 (Bosni activity, <75% of EU 15 average, Campania, Forest cover of the Province is only 24 975 an pine). Broadleaved forests (mainly Quer Apulia and Basilicata were included in the so hectares, mainly degraded coppices of Quer cus spp. and Fagus sylvatica L.) represent called Objective 1 group of the EU together cus spp.; the majority of them are small sur about 51.8% of total forested area of the Re with Abruzzo (until 31/12/2000), Calabria, faces intermingled with cropland. gion. Molise, Sicily, and Sardinia. Plantations of Pinus halepensis and The average number of fires per year It leaves no doubt that Objective 1 Regions Cupressus made in the 1930’s and 50’s are (1992-2003): 372. can be described as “less developed”. frequent, with wide surfaces, in the inner All the study areas represent the typical Among them, the three regions considered in part of the province, on the higher sites of Mediterranean climate pattern, where hot this study in which the condition of margi the plateau. All of them were functional to and dry summers are a predisposing fire nality of the same sub-areas are accompa soil protection against erosion, mainly water factor. nied by the presence of some types of volun and wind erosion. With regard to wildfires, all sites are high tary wildfires, mainly related to labor con The average number of fires per year risk areas, in accordance to European Eco flicts. (1996-2001): 68. nomic Community (CEE) Reg. 2158/92, for A panel of experts is the cornerstone of the high frequency (number of fires per year), Delphi technique: in the case-studies, experts Area 4 high density (number of fires / year / forest in the particular field of forest fires were Basilicata region. Basilicata is a small re area), high burned area, medium or high fire found through the professionals working for

© SISEF http://www.sisef.it/iforest/ 10 iForest (2010) 3: 8-15 Lovreglio R et al. - iForest 3: 8-15 governmental and non-governmental organ fire and social problems of the territory, par hierarchical, military-like organization. izations; namely, the foresters of CFS (State ticularly those working in stations that are Among the experts, we did not include pro Forestry Department) - areas 1.2.4 - and Pro commanded by a Warrant Officer who is fessionals from the National Fire Brigade, vincial rangers (area 3). usually at the end of his/her career. It leaves, since structural firefighting (and not wild Number of expert for each area is given in therefore, no doubt that their experience in fires) is their functional responsibility in ac Tab. 1. The four panels of experts therefore the specific area of forest fires causes and in- cordance with the Law, Act 353/2000. This included a total of 156 respondents. depth knowledge of their territory, is con organization only provides assistance in Foresters belong to CFS, Corpo Forestale firmed. structure protection on an emergency basis dello Stato (State Forestry Department), a Provincial rangers belong to a police to save lives in the event of a wildfire. National Police Agency in Italy, responsible agency as well, but on the local level. Structured questioning was achieved for protecting Italy’s natural resources, the They are trained as forestry rangers and through the use of ad hoc questionnaires environment, countryside and ecosystems, provide hunting and fishing control services, where all official motives of forest fires, re particularly National Parks and National enforcement of environmental laws; rangers cognized by the State Forestry Department Forests. Its best known protection duty is are ex officio watersheds, fish, game and fire (MIPAAF-CFS 2002) and which are fighting wildfires. guardians and collaborate with CFS in fire routinely used by the same experts in forest Foresters stationed at Forest Stations are fighting, mainly alarm and first attack. fire statistics (see Annex 1), are reported. required to keep records of forest fire statist Personnel involved in the experts’ panel is The use of the anonymous questionnaire ics. After each fire, they must fill out a therefore directly and permanently involved provided the group members the freedom to standard statistic form, explaining (among in forest fire control and statistic activities. express their opinion without feeling pres other things) the motivation and cause of For this reason, they must be considered well sured by the wider group or dominant mem fire. knowledgeable and experienced in the sub bers. They all have considerable experience in ject. In both cases, experts are organized in a One of the main problems with such a

Tab. 1 - Study-areas and fire motives in order of decreasing frequency.

C. M. Vallo di Diano National Park of Gargano Province of Bari Region Basilicata Number of experts Number of experts Number of experts Number of experts N=8 N=34 N=58 N=56 Freq. Freq. Freq. Freq. Main motives Main motives Main motives Main motives (%) (%) (%) (%) Fire set for cleaning on bor 12.50 “Fire industry” 7.55 Fire set for stubble burn 13.99 Fire set for stubble burn 11.16 ders of croplands or in aban ing ing doned agrarian plots. Fire set by farmers for planta 10.94 Pyromania and myth 7.55 “Fire industry” 10.56 Fire set for cleaning on 8.48 tion cleaning after harvest mania borders of croplands or in abandoned agrarian plots. Fire set for stubble burning 10.94 Fire set for stubble 7.19 Fire set by the auxiliary 8.41 Pyromania and myth 7.97 burning workers who are retained mania by national forest fire ser vices Carelessly discarded cigarette 9.38 Fire set by the auxili 5.76 Fires started by lightning 6.44 Fire set for pasture renov 7.62 butts or matches on road banks ary workers who are ation retained by national forest fire services Fire set to obtain products de 9.38 Fire caused as protest 5.76 Fire caused with the intent 5.39 Fire caused by conflicts 6.68 riving from fire passage against limitations im of earning from the re between or revenge posed in conservation moval of vegetation for against owners, ownership areas the purpose of building controversies speculation Fire caused by the creation or 7.81 Fire caused by con 5.40 Pyromania and myth 4.53 Fire set by farmers for 6.25 renewal of pastures at the ex flicts between or re mania plantation cleaning after pense of forests venge against owners, harvest ownership controver sies Fire caused by conflicts 6.25 Fire set as revenge or 5.40 Fire set for pasture renov 4.31 Carelessly discarded ci 5.82 between or revenge against retaliation against pub ation garette butts or matches owners, ownership controver lic administration on road banks sies Carelessly discarded cigarette 4.69 Fire set for cleaning on 4.68 Fire caused by conflicts 4.31 Fires started by lightning 3.91 butts or matches in coun borders of croplands between or revenge tryside or in abandoned against owners, ownership agrarian plots controversies others 28.11 others 50.35 others 42.06 others 42.11 iForest (2010) 3: 8-15 11 © SISEF http://www.sisef.it/iforest/ Wildfire cause analysis in Italy

Tab. 2 - Decreasing frequency of most relevant motives (all areas included). cussing responses among the panel members of areas 1, 2, 3 and Delphi modulator. Max. observed Code of Responses were processed and results Motive frequency (%) motives presented to experts. The panel members were then asked if this information had Stubble burning 13.99 3104 changed their opinion in any way and if they Agricultural uses 12.50 3101 wanted to modify their responses to any Pruning rests burning 10.94 3102 question. Fire-fighters 10.56 4008 None of the participants reconsidered their Cigarettes 9.38 3001 responses after feedback. Experts in area 4, Mushrooming 9.38 4010 who were contacted by mail, preliminarily Agricultural uses 8.48 3101 stated that they would not accept any further meetings or questionnaire sending. Fire-fighters 8.41 4006 Far from being a breakdown, we consider Pyromaniacs 7.97 4108 that lack of reconsidering responses avoided Renewal of pastures 7.81 4001 a tendency to force a middle-of-the-road Agricultural uses 7.62 3103 consensus or give responses which experts All others <7.55 - think the monitoring group want to hear. The risk in the method is actually that of structure is actually to avoid negative group to 8, 1 being the most important and so on. generating an artificial consensus. In addi dynamics that may emerge, such as domina A list of motives, with their four-digit iden tion, we cannot exclude the so called tion by key individuals, prestige of a certain tification code, is presented in Annex 1 (MI “Hawthorn Effect”, which is the psycholo participant, shyness of certain participants PAAF-CFS 2002): it contains the 43 pos gical response in which subjects alter their and above all, superior in rank. sible official motives behind the cause of fire behavior because they are aware they are The panel members met face-to-face with in Italy and is divided into five groups (nat participating in a study. the monitor (or modulator) of the Delphi in ural, accidental, negligent, deliberate, doubt). quiry (areas 1, 2, 3) or were contacted via Cause categories are those officially adopted Results mail (only for area 4): this latter option is in Italy by the Italian State Forestry Depart Tab. 1, Tab. 2 and Tab. 3 report respec considered acceptable (Dunham 1998). ment for forest fire statistics. tively the most frequent motives, as identi Experts, who usually identify fire causes The motives maintain their original state fied by the experts and their rank-ordering, for their activity, as already mentioned, were ment, but explanatory notes are added. which is the modal value or mode of rank- given the questionnaire, reporting the list of Experts were consequently asked questions scores. motives and identification codes. They were that specifically refer to their respective We did not use their arithmetic mean, asked to “vote” for the eight most important areas of expertise when filling out the forest which is not suitable for use with nominal or and/or relevant motives in the territory fire statistic forms, since wildfire motives ordinal data (the scores are ordinal value where they act and then to rank them in or and relative codes are very familiar to them. only representing position of order, not der of decreasing importance, scoring from 1 Controlled feedback was achieved by dis quantity).

Tab. 3 - Rank-ordering (1 to 8) of most frequent fire motives in study-areas.

Rank- C. M. Vallo di Diano National Park of Gargano Province of Bari Region Basilicata Order 1 Agricultural uses, Rests Stubble burning Stubble burning, Fire industry Stubble burning, burning Pyromania and myth mania 2 Stubble burning Renewal of pastures Stubble burning, Auxiliary Agricultural uses, Pasture Auxiliary workers workers, Ownership controver renovation, Stubble burning sies 3 Agricultural uses, Rests Auxiliary workers, Fire Stubble burning, Cleaning of Agricultural uses Stubble burn burning, Stubble burning, Re industry, Retaliation against road/railroad, Auxiliary ing newal of pastures, Mushroom public administration workers ing, Ownership controversies 4 Cigarettes, Agricultural uses Stubble burning, Fire industry Pasture renovation, Fire industry Cigarettes, Pasture renovation 5 Rests burning, Mushrooming, Renewal of pastures Building speculation, Auxiliary Pasture renovation, Ownership controversies workers Ownership controversies 6 Fire-crackers and Auxiliary workers Hot vehicle exhaust pipes, Cigarettes, Retaliation against bottle-rockets Building speculation, public administration, Mush Ownership controversies rooming 7 Renewal of pastures Pyromaniacs Lightning, Hot vehicle exhaust Lightning, Sparks from brake pipes, Auxiliary workers, Reta shoes of trains, Auxiliary liation against public admini workers, Ownership controver stration sies 8 Cigarettes Lightning Cigarettes, Stubble Lightning, Pyromania and myth Lightning, Power lines burning, Forestry operations mania Ownership controversies, Pyro mania and myth mania

Tab. 4 - Percentage of different forest causes in the study regions (2001) from the official does not mean it ought to be considered as statistics. criminal use of fire. Respondents give minor emphasis, in rank Region ordering, to the obvious and rather banal Fire causes motive of cigarettes carelessly discarded (%) Apulia Basilicata Campania which is, on the contrary, one of the most re Natural 0.7 0.3 0.4 current in the mentioned MIPAAF-CFS sur Accidental 0.2 0.6 1.2 vey for 2001 in Italy (Matches and cigarette stubs abandoned or imprudently thrown Negligent 43.0 36.1 28.6 along trails, roadways and train tracks ac Deliberate 54.1 58.9 66.5 count for 7.8% of total negligent fires, Unknown 2.0 4.1 3.4 whereas agricultural accounts for 60.8%, after that source). The reduced number of respondents is the prompts a specific behaviour, similar conclu Taking a look at deliberate fires, motives cause of frequent multiple modal values. sions can be drawn from the results of rank closely related to labour conflicts (groups Tab. 1, in particular, provides the percen ordering. 4006-4008) appear to be the most relevant tage with which each motivation, as identi Instead of the usual summary of the over ones in our study. These motives confirm the fied in Annex 1, was considered by the whelming majority of causes as arsonist or importance and presence of the so called group of experts for its area. It is important unknown, the panel of experts resist to “fire industry” (Leone & Vita 1982, Leone to note that the most frequent motives (fre mundane explanations and attribute the ma et al. 1988, Leone & Saracino 1990, Leone quency > 10%) are related to agricultural use jority of fire ignitions to cultural motives (for 1997a, Leone et al. 2003), i.e., voluntary of fire, followed by fire started as a result of instance traditional use of fire, such as in the fires lit by seasonal workers as an instrument labor conflicts. case of stubble burning) institutional beha for forcing/maintaining employment and/or viours (for instance protest against declara creating new job opportunities (CFS 1992). Discussion tion of protected area, revenge against Public Clearly, it is a case of fire being ignited as a A number of interesting conclusions can be Administrations or Public Entities) or social means of subsistence, intentionally set by drawn from the results of the study. tensions (e.g., ownership conflicts, labour fire-fighters, as a way to maintain their job The majority of respondents converge to conflicts) and give answers which appear to and increase their revenue (WWF 1993). wards a rather limited number of motives be rather homogeneous and convergent, des Deliberate fires of this kind are well correl (Tab. 2). More than 80% of the respondents pite the obvious difference of study areas. ated with the low level of income (Leone & actually refer to involuntary motives; Answers referring to high ranked motives Vita 1982). The lower the level of income, namely, agricultural use of fire (codes 3101, are actually focused on a rather restricted, the higher will be the number of provoked 3102, 3104). homogeneous group (Tab. 3) where a few fires (Vélez 1986, Vélez 2000); the condition In terms of frequency, responses by experts motives cover more than 50% of possibilit of marginality in areas where such events are (even though operating in different areas) ies. In detail, only 22 out of 43 possible common is confirmed by the economic clas converge, reporting negligent fires as the motives (51.16 %) are considered by experts. sification of study-cases, all included in Ob most frequent set of motives (15 out of 32) On the contrary, the less frequent and less jective 1 Regions, as already mentioned. and therefore confirming doubts expressed important motives appear rather scattered, The motive referring to conflict with au by some authors, as in the introduction, with few cases for each; results confirm, in thorities (code 4103) is observed only in the about the distribution of the percentage of any case, the absolutely scarce importance of National Park of Gargano, and is therefore causes (Bovio 2005, Corrado 2008). natural and accidental fires. not reported in Tab. 3 because of its local Among voluntary motives, the most fre In the first rank order, i.e., scored 1, the importance. This could be interpreted as a quent one (gathering more than 50% of re most relevant group is referred to as “agri sort of reaction against land use restrictions spondents and a frequency of 10.56 %), is cultural use of fire”, whereas a less impor in protected areas, which sometimes ex referred to as “fire caused with the intent of tant group of motives refers to deliberate fire plodes in very violent ways. Confrontation being included in fire-fighting efforts” setting (Tab. 3). can therefore occur, of which the forest fire (motive code 4008), followed by motives re Expanding the comments to the first four is a symptom (FAO 2005), an unorthodox ferring to: ranks, motives connected to negligent fires way of affirming rights of use (WWF 1993). • harvesting mushrooms and edible sprouts, are referred to as “careless use of agricultur The motive referring to conflicts with Public issued after passage of fire; al fires” (codes 3101 to 3105), i.e., always Administration (code 4101) could be inter • protest of seasonal fire-fighters; with cultural motives. preted in a similar way. • pasture renewal; The most frequent motive seems to be The motives which refer to ownership con • and pyromaniacs. stubble burning (code 3104), i.e., systematic flicts (code 4102) are ranked relatively low The results, apparently in contrast with the burning of stubble on wheat lands, for the in terms of importance but have, in fact, im current public opinion that wants the majo preparation of the agricultural land for new portance in some limited areas, such as the rity of fires to be deliberate, are perfectly fit sowing-purposes and the elimination of Gargano National Park, as conflicts between ting with the point of view of international residue or thickets that restrict exploitation. shifting shepherds, not owners of grazing experts on the frequent overemphasis of vo The principal aim of agricultural burnings land but only landless occupants, and stable luntary causes of fire. is functional, i.e., they are not started simply rural dwellers, indicating a social problem in For the Regions involved in this study, for destructive purposes. These burnings are the local cattle-farming system (Leone et al. Tab. 4 reports the percentage of fire causes legally authorized as long as they comply 2002, Leone et al. 2003, Leone & Lovreglio from an in-depth official survey carried out with certain conditions, but become illegal as 2003). in 2001 by the State Forestry Department a result of failure to comply with preventive The popular and abused motive which usu (MIPAAF-CFS 2002). measures laid down by the regional laws. In ally refers to fire used as a tool to convert With regard to motives, defined as the in any case, although the agricultural or shrub rural land into urban land, i.e. building spec ner drive, reason or incentive that induces or land burning may have started a fire, this ulation (WWF 1993), is absent in the results.

In any case building on areas swept by fire is et al. 2003): the first issue of the study is that NUTS 3 level or less, such as groups of mu not permitted for ten years by Italian Law culture determines incendiary behaviour; the nicipalities; official statistics carried out by 353/2000. main action is, therefore, to change from a CFS in Italy are, on the contrary, usually re Special mention must be made of pyroma parochial, pragmatic approach and view the ferred at NUTS1 and NUTS 2 level and give niacs (code 4108), which received a rather wildfire problem from a wider perspective, only frequency of motivations. high frequency, but ranked low in terms of more careful of the society living in a terri For this, Delphi technique is a precious tool importance in two study cases. As a matter tory. in the field of fire prevention and fire control of fact, the term “pyromaniac” is largely For instance, in the case of fire set by planning. misused as a synonym of arsonist, not only means of stubble burning, it would be pos in Italy but also in other countries (APAS sible to allow fire use through regulation, Acknowledgments 2003, Dolzreuss & Irastorza 2005). zoning and periods of prohibition, by requi The Authors thank prof. Emilio Chuvieco, Psychologists, on the contrary, agree that ring burning licenses, together with the University of Alcalà de Henares, Spain, for pyromaniacs - people with a mental illness adoption of a set of measures to prevent fire careful review and precious suggestions to resulting in an uncontrollable urge to start escapes, such as peripheral plowing lanes improve text. A special thank is due to fires - account for only a small minority of (FAO 2005). Kristy Schuster, from USA, for careful revi arson. On the contrary, it would be much more sion of English text. High scoring of this motive (but only in difficult to establish actions against motives one area), could be interpreted as a case of referring to voluntary fires set by fire- References being misinformed or a lack of information, fighters as a way to maintain their job and Anderson DH, Schneider IE (1993). Using the and is further confirmation of confusion and increase their revenue, the so called “fire in Delphi process to identify significant recreation common improper use of term, even by the dustry”. research-based innovations. Journal of Park and experts. This calls for a more intense and im In Southern Italy, the use of labour in Recreation Administration 11 (1): 25-36. proved training of people who are involved activities involving the cultivation and the APAS (2003). Estudio Sociòlogico sobre la Per in forest fire statistics. protection of woods is mainly allocated to ceptión de la Poblacion Española hacia los Incen the maintenance of a minimum level of ma dio Forestales. Associatión para la Promocciòn Conclusions nual labour. de Actividades Socioculturales [online] URL: Knowing what motive induced or prom Wildfires voluntarily lit by member of sea http://www.idem21.com/descargas/pdfs/Incendi pted a destructive behavior is critical to sonal firefighting crews in search of jobs, osForestales.pdf mount significant prevention efforts, which create and maintain work under the pressure Baughman MJ (1989). Effective use of the Delphi embrace measures that modify fuels and hu of the events; authors take advantage of the process. In: “Discovering new knowledge about man behavior, so that the initiation, spread, emotional reaction of concerned communi trees and forests” (Leary RE ed). USDA, Forest and intensity of fires are reduced to such an ties who ask reinforcement of extinguishing Service, North Central Forest Experiment Sta extent that they can be controlled by the activities: fire therefore creates new job op tion, Gen. Techn. Rep. NC-135. technical means available. portunities, first in extinguishing the re Bovio G (2005). Incendi dolosi? In: “Foreste Causes of fires arise undoubtedly from peated fire events and then in replanting or ricerca cultura. Scritti in onore di Orazio Cian many complex social, environmental, poli restoring the burnt forests. cio” (Corona et al. eds). Accademia Italiana di tical, organizational and economic forces, The results are more fires, more emotive Scienze Forestali, Florence, Italy, pp. 43-49. whose importance is likely to vary by coun attention, more people called to extinguish, CFS (1992). Aspects sociaux, economiques et cul try or region and over time. but also more fires to maintain jobs, more turels des incendies de foréts en Italie. In: Pro Results of this study confirm this variety fires to foster attention, turning into a per ceeding of the Seminar on “Forest Fire Preven from one area to the next, though some mo manent, vicious cycle. tion, Land Use and People”. ECE/FAO/O1T, tivations are a recurrent and common core. Breaking this vicious cycle could be Athens (Greece) April 1987, pp. 45-59. Understanding the motive behind the cause achieved by giving people alternative job op Corrado G (2008). Le azioni politiche per la difesa may enhance investigative efforts and focus portunities in a more continuative way, for del bosco dagli incendi. In: Atti del III Congresso on improving prevention and social behavior instance in prevention activities, such as pre di Selvicoltura per la conservazione e il migli modification. Less emphasis should there ventive sylvicultural tending (thinning, pru oramento dei boschi, pp. 345-354. fore be given to interventional actions and ning, prescribed burning) mainly making Clute O (2000). A perspective from wildland fire fire exclusion, which are mediatically im salary indifferent to repeated or provoked communicators. Wildland fire: Communicator’s pressive, whereas more attention and re fire presence. guide, The National Interagency Fire Center and sources should be devoted to the analysis of In conclusion, due to the increasing num the National Wildfire Coordinating Group. [on complex causes of such social phenomenon ber of fires on the Mediterranean and natio line] URL: http://www.nifc.gov/preved/comm_ (Folkman 1976) with the hopes of behavior nal scale of the so called “fire club” coun guide/wildfire/fire_14.html modification of people. tries, prevention must receive higher priority Crance JH (1987). Guidelines for using the Delphi Delphi, which is ultimately a structured and greater attention than the improvement technique to develop habitat suitability index process for collecting and “distilling” expe of fire suppression resources, which do limit curves. Biological report 82 (10.134), Fish and rience from a group of experts, reveals use damages but at such a cost that possibilities wildlife service national ecology center, Fort ful for this scope, since it exploits the expe to increase those resources are nearly ex Collins, CO, USA. rience and in-depth accumulation of know hausted (FAO 2005). De Las Heras J, Salvatore R, Rodrigues MJ, ledge of territory by professionals in a rather In this context, the Delphi technique proves Lovreglio R, Leone V, Giaquinto P, Notarnicola fast and simple way. to be useful in helping fire managers in lis A (2007). Wildfire motivation survey through the In the study areas of Southern Italy, the ting, identifying and in some way measuring Delphi method. In: “Actas de la IV Conferencia Delphi method allowed for placing the right the most frequent and relevant human caused Internacional sobre Incendios Forestales”, Sevilla emphasis on some motives, which can help ignition risks and sources. (España) 13-18 May 2007. [online] URL: to pinpoint appropriate preventive actions for In addition, it permits to give a sort of http://www.fire.unifreiburg.de/sevill2007/contri the specific reality. Results confirm that “weight” to fire motivations and, what is far butions/doc/SESIONES_TEMATICAS/ST4/Her wildfires are site and culture specific (Leone more important, to lay down statistics at as_et-AL_SPAIN_ITALY.pdf

Dolzreuss ML, Irastorza FI (2005). State of the art Leone V (1997b). Forest fires in Italy in 1995 and - doi: 10.1016/j.foreco.2006.08.034 of forest fire causes in Spain. In: Proceedings of 1996. International forest fire news 17: 5-9. Lovreglio R, Rodrigues MJ, Silletti G, Leone V the II International Conference on “Prevention Leone V, Koutsias N, Martinez Fernandez J, (2008). Applicazione del metodo Delphi per l’an strategies of fires in Southern Europe”. Centre Allgawer B, Vega Garcia C, Lovreglio R (2003). alisi delle motivazioni degli incendi: il caso Tecnologic Forestal de Cataluña, Barcelona The human factor in fire danger assessment. In: Taranto. L’Italia Forestale e Montana 5: 427- (Spain) 9-11 may 2005, pp. 325-331. “Wildland fire danger estimation and mapping. 447. - doi: 10.4129/IFM.2008.5.04 Dunham RB (1998). The Delphi technique. Wis The role of remote sensing data” (Chuvieco Sa Ministerio de Medio Ambiente (1997). Diseño y consin School & Business web site, Winsconsin, linero E ed). Series in Remote Sensing vol. 4, planificación de acciones de comunicación para USA. [online] URL: http://instruction.bus.wis World Scientific Publishing Co., New Jersey, la prevención de incendios forestales. FAB Con c.edu/obdemo/readings/delphi.htm USA, pp. 143-194. sultores, S. A., Madrid, Spain. Environment Policy (2003). Putting out the fire: Leone V, Lovreglio R (2003). Parco nazionale del MIPAAF-CFS (2002). Indagine conoscitiva in saving Greece’s forests. Environment Policy. Gargano. Piano di previsione e prevenzione con cendi boschivi. Ministero delle Politiche Agri [online] URL: http://www.greece.gr/ENVIRON tro gli incendi boschivi (Unpublished internal re cole e Forestali, Corpo Forestale Dello Stato, MENT/EnvironmentalPolicy/AttackingTheRoot . port) [online] URL: http://www.minambiente.it/ Stampa New Graphic snc, Roma, Italy. stm Last Accessed 18.2.2003 Leone V, Lovreglio R (2003). Human fire causes: NASA (2003). Natural Hazards. NASA web site, FAB Consultores (1990). Estudio sobre la causal a challenge for modelling. In: “Innovative con Wasshington, DC, USA. [online] URL: idad de los incendios forestales en la Comunidad cepts and methods in fire danger estimation” http://earthobservatory.nasa.gov/NaturalHazards/ Autonoma de las Islas Baleares.Conselleria (Chuvieco E, Martin P, Justice C eds). Proceed NASA (2009). Earthobservatory. [online] URL: d’Agricultura I Pesca. Govern Balear, Impresrap ings of the “4th International Workshop on re http://earthobservatory.nasa.gov/GlobalMaps/vie it. mote sensing and GIS applications to forest fire w.php?d1=MOD14A1_M_FIRE FAO (1999). FAO meeting on public policies af management”. Ghent (Belgium) June 2003, pp. Schuster EG, Frissell SS, Baker EE, Loveless RS fecting forest fires. FAO Forestry Paper 138, UN 149-170. [online] URL: http://www.fire.uni- (1985). The Delphi method: application to elk Food and Agriculture Organization, Rome, Italy. freiburg.de/fwf/Fire%20Danger%20Ghent%20 habitat quality. USDA Forest Service, Inter FAO (2005). Community based fire management Proceedings.pdf mountain Research Station, Research Paper INT- in Spain, April 2005. Forest protection working Leone V, Lovreglio R, Martinez Fernandez J 353. Ogden, Utah, USA. papers, working paper FFM/4/E. Forest resources (2002). Forest fires and anthropic influences: a Turoff M (1970). The Design of a policy Delphi. development service, forest resources division. study case (Gargano National Park, Italy). In: Technological forecasting and social change 2: FAO, Rome, Italy. [online] URL: http://www. “Forest fire research & wildland fire safety” 149-171. - doi: 10.1016/0040-1625(70)90161-7 fao.org/docrep/009/ag045e/ag045e00.HTM (Viegas DX ed). Millipress, Rotterdam, The UNECE (2008). Forest fire statistics. UNECE FAO (2007). Fire management global assessment. Netherland, pp. 11-28. Timber Committee, United Nations, Report A thematic study prepared in the framework of Leone V, Mazzucca D, Vita F (1988). Hypothesis ECE/TIM/BULL/2002/4. [online] URL: the Global forest resources assessment 2005. as to the causes of forest fires in Puglia. In: Pro http://www.unece.org/trade/timber/ff-stats.html FAO Forestry Paper 151. FAO, Rome, Italy. [on ceedings of the FAO/ECE/ILO “Seminar on Wilson CC (1976). Detection and control of forest line] URL: http://www.fao.org/forestry/fra2005/ methods and equipments for the prevention of fires for the protection of the human environ en/ forest fires”. Valencia (Spain) Sept 3 - Oct 10 ment. Project 0206-74-003, Publ. M/LO788/ Folkman WS (1976). Vandalistic forest fire set 1986. ICONA, Madrid, Spain, pp. 190-198. E/1.78/1/70, FAO, Rome, Italy. ting. In: Proceedings of the symposium “Vandal Leone V, Saracino A (1990). Arson and forest fire WWF (1993). Fires in the Mediterranean. [online] ism and outdoor recreation” (Alfano SS, Magill industry: the state of the art in Italy. In: Proceed URL: http://www.env-edu.gr/Documents/Forest AW eds). USDA For. Serv. Gen. Tech. Rep. ings of the First International Conference on %20fires%20in%20the%20Mediterranean.doc PSW-17. “Forest fire research”. Coimbra (Portugal) 19-23 Vélez R (1986) Incendios forestales y su relación ICONA (1995). Motivaciones de los incendios November 1990, pp. 13.1-13.9. con el medio rural. Revista de Estudios Agroso forestales intencionados. ICONA Ministerio de Leone V, Vita F (1982). Incendi boschivi e mar ciales. 136:195-224. Agricultura, Pesca y Alimentación, Madrid, ginalità economica: il caso della Puglia. Cellu Vélez R (2000) La defensa contra incendios fore Spain. losa e Carta 7/8: 41-57. stales. Fundamentos y experiencias. McGraw- Jackson JW, Fisher RJ (2001). Keynote address, Linstone HA, Turoff M (2002). The Delphi Hill/Interamericana de España S.A.U., Madrid, communities in flames conference, Balikpapan, method techniques and applications (digital ver pp. 2242. Indonesia. [online] URL: http://www.fao.org/ sion). [online] URL: http://www.is.njit.edu/ Zuboy JR (1980). The Delphi technique: a poten docrep/005/AC798E/ac798e00.htm#Contents pubs/delphibook/ tial methodology for evaluating recreational fish Leone V (1997a). Sociological aspects in the phe Lovreglio R, Leone V, Giaquinto P, Notarnicola A eries. NOAA technical memorandum NMFS- nomenology of forest fires. In: “The Forest and (2006). New tools for the analysis of fire causes SEFC-19, U.S. Dept. of commerce, National Man” (Ciancio O ed). Accad. Ital. Sc. Forest., and their motivations: the Delphi technique. oceanic and atmospheric administration, National Florence, Italy, pp. 305-324. Forest Ecology and Management 234 (1): 18-33. marine fisheries service, Washington, DC, USA.