A conceptual model of sprouting responses in relation to fire damage: an example with cork oak (Quercus suber L.) trees in Southern Portugal

A conceptual model of sprouting responses in relation to fire damage: an example with cork oak (Quercus suber L.) trees in Southern Portugal
of 9
All materials on our website are shared by users. If you have any questions about copyright issues, please report us to resolve them. We are always happy to assist you.
Similar Documents
  A conceptual model of sprouting responses in relation to firedamage: an example with cork oak ( Quercus suber  L.) treesin Southern Portugal Francisco Moreira   Filipe Catry   Ine ˆs Duarte   Vanda Aca´cio   Joaquim Sande Silva Received: 8 January 2008/Accepted: 21 July 2008/Published online: 10 August 2008   Springer Science+Business Media B.V. 2008 Abstract  The sprouting response types of 1,151cork oak ( Quercus suber  ) trees one and half yearsafter a wildfire in southern Portugal were character-ised. It was hypothesised that different response typesshould occur according to the following conceptualmodel: an increased level of damage (fire severity) ona sprouting tree that suffered a crown fire wasexpected to be reflected in a sequence of fouralternative events, namely (a) resprouting exclusivelyfrom crown, (b) simultaneous resprouting from crownand base, (c) resprouting exclusively from base and(d) plant death. To assess whether the level of expected damage was influenced by the level of protection from disturbance, we explored the rela-tionships between response types and tree size, bark thickness and cork stripping, using an information-theoretic approach. The more common response typewas crown resprouting (68.8% of the trees), followedby plant death (15.8%), simultaneous resproutingfrom crown and base (10.1%) and basal resprouting(5.3%). In agreement with the conceptual model,trees which probably suffered a higher level of damage by fire (larger trees with thinner bark;exploited for cork) died or resprouted exclusivelyfrom base. On the other hand, trees that were wellprotected (smaller trees with thicker bark notexploited for cork) were able to rebuild their canopythrough crown resprouting. Simultaneous resproutingfrom the crown and base was determined mainly bytree size, and it was more common in smaller trees. Keywords  Apical dominance    Mediterranean   Model    Mortality    Resource allocation   Resprouting    Severity Introduction Resprouting is an efficient mechanism through whichmany plants from the Mediterranean region recoverabove-ground biomass after they have suffered totalcrown consumption from a wildfire (Whelan 1995;Bond and van Wilgen 1996; Keeley 2006). Sprouting shoots can originate from dormant buds locatedabove ground (axillary, branch epicormic or stemepicormic) or from the base of the plant (i.e. from thecollar, roots or underground stems) (Bond and vanWilgen 1996; Miller 2000; Del Tredici 2001). Hereafter, these two sprouting modes will be referredto as ‘crown’ and ‘basal’ sprouting (Bond and vanWilgen 1996).Bellingham and Sparrow (2000) presented ageneral model of resprouting responses as a functionof increasing disturbance severity (severity defined as F. Moreira ( & )    F. Catry    I. Duarte    V. Aca´cio   J. S. SilvaCentro de Ecologia Aplicada ‘‘Prof. Baeta Neves’’,Instituto Superior de Agronomia, Universidade Te´cnicade Lisboa, Tapada da Ajuda, 1349-017 Lisbon, Portugale-mail:  1 3 Plant Ecol (2009) 201:77–85DOI 10.1007/s11258-008-9476-0  a measure of a plant 0 s perception of a disturbanceevent). This gradient of increasing severity wasexpected to create a sequence of hierarchical regen-erative responses ranging from crown (e.g. axillaryand branch epicormic) to basal sprouting, where theloss of one type of tissue (e.g. in twig) induces aregenerative response from the next level of hierar-chy (e.g. twig axil on the branch) (Bellingham andSparrow 2000). In their model, disturbance severity isexpressed as proportion of above-ground biomass lost(Bellingham and Sparrow 2000). For one particulartype of disturbance, wildfires, and in particular crownfires, often all the canopy foliage, buds and twigs areconsumed (crown consumption). When this happens,severity will depend mainly on the fire intensity andthe level of fire-protection mechanisms at the indi-vidual level (e.g. Bond and van Wilgen 1996).Although the hierarchical nature of sproutingresponses presented in Bellingham and Sparrow 0 smodel could also be expected in this situation, thefact that in a few studies sprouting responses atdifferent hierarchical levels were simultaneouslyregistered in the same individual plant (Trollope1984, current study) suggests that the factors under-lying response types will be more complex than justabove-ground biomass lost.In situations where wildfires caused total crownconsumption in sprouting trees we allege that distur-bances of differing levels of damage (severity), andcorresponding sprouting responses, not necessarilyorganised as an hierarchical model, can still berecognised. These responses will be determined bythe amount of bud damage in the twigs and branches,the level of damage to stem and root cambial tissueand the amount of below-ground reserves whichdetermines how much carbohydrate reserves can bemobilised to rebuild the lost biomass (Chapin et al.1990; Bond and van Wilgen 1996; Iwasa and Kubo 1997; Bellingham and Sparrow 2000) (Fig. 1). When the level of fire damage is low (e.g. caused by low fireintensity on trees with thicker bark, and where thestem cambium is not affected), the plant is expectedto resprout from crown buds that survived the fire(Fig. 1a). If the level of damage is extreme (e.g.caused by high fire intensity on trees with thinnerbark or where the stem cambium is damaged), themost likely outcome is plant death (Fig. 1d). Atintermediate levels of severity two response types canbe identified. If the level of damage is higher, allcrown buds will be killed, either directly through heator indirectly through the destruction of the vascularcambium in the stem, as the carbohydrate reservesthat support sprouting are primarily stored in below-ground structures (Del Tredici 2001). Furthermore,apical dominance will be suppressed directly throughbud destruction by heat or indirectly via damage tothe cambium (Kozlowski 1971; Kozlowski et al.1991; Miller 2000), and the tree is therefore expected to respond through basal resprouting (Fig. 1c).Alternatively, if the level of damage is not so severe,partial damage to the crown buds and cambium willcause weakened apical dominance (Kozlowski 1971)and at least some accessibility to below-groundreserves, thereby resulting in the simultaneous res-prouting of the crown and base (Fig. 1b). Since theamount of carbohydrate, nitrogen and phosphorusresources that can be used for growth also determinesthe extent to which plants can resprout (Chapin et al.1990), the observed resprouting patterns will there-fore also be influenced, and plants with depletedbelow-ground resources may suffer higher levels of damage since they are unable to allocate enoughenergy to restore the lost biomass. An example of these above-mentioned four types of responses can befound in a study of   Acacia karroo  savanna byTrollope (1984), where different responses wererelated to tree size and fire intensity. However, noother examples were found in the literature where the Level of damage / fire severity a b c Bud damage Cambial tissue damage Below-ground reserves  d Fig. 1  A conceptual model of post-fire responses of asprouting tree that suffered total crown consumption (combus-tion of leaves and twigs during a wildfire) in relation to agradient of increasing level of damage/fire severity. ( a ) Crownsprouting, ( b ) simultaneous sprouting from crown and base, ( c )basal sprouting, ( d ) plant death (for further explanations seetext)78 Plant Ecol (2009) 201:77–85  1 3  occurrence of different responses was registered andcharacterised for other tree species.The cork oak   Quercus suber   L. is a very importanttree species within the Mediterranean basin, bothfrom an economic and ecological perspective (Silvaand Catry 2006). The existence of a thick cork bark plays an important role in the capacity of this speciesto withstand the frequent occurrence of fire typical of Mediterranean climates (e.g. Pausas 1997; Moreiraet al. 2007). Another feature of cork oak trees is thecapacity of post-fire resprouting from the base andcrown after complete defoliation, hence the species isa good model for studying the different responsepatterns previously described.In general, there is little information available onthe relative frequency of the different response typesas well as the factors influencing these responses incork oak. Previous studies (e.g. Cabezudo et al. 1995;Pausas 1997; Barberis et al. 2003; Catry et al. 2007; Moreira et al. 2007) focused mainly on the factorsinfluencing post-fire survival, and showed the keyrole of cork stripping, cork thickness and tree size ondetermining oak survival. In this article, we hypoth-esise that these three variables also influence otherpost-fire response patterns (as described in Fig. 1)besides death, since they are expected to influence thelevel of resistance to fire and, consequently, the levelof damage.Cork stripping is a common operation that isnormally performed after the tree attains a certaincircumference at breast height (70 cm in Portugal).Cork is a valuable raw material for industry and isperiodically removed with an axe by manuallycutting along vertical and horizontal lines on thestem and thicker branches and stripping off cork planks (Pereira and Tome´ 2004). After each cork stripping, the tree has the capacity to produce newcork bark by adding new layers of cork every year(Pereira and Tome´ 2004), Moreira et al. (2007) showed that unstripped trees (with unharvested virgincork) had higher survival rates than trees that hadbeen exploited for cork (i.e. trees debarked at leastonce). These authors suggested that the highersurvival rates of unstripped trees may be explainedby the higher insulating properties of virgin cork (fora given bark thickness) and the absence of stresscaused by cork extraction. In fact, cork extraction is adisturbance that has negative effects on tree healthand growth (Costa et al. 2004). Thus, stripping probably requires a greater allocation of below-ground energy reserves that will subsequently not beavailable for investment in resprouting. Conse-quently, unstripped trees are expected to showlower levels of damage when compared to exploitedtrees since their buds are more protected and theirbelow-ground reserves may be better preserved.Cork thickness depends on the harvesting cycle andthe time elapsed between harvesting events. Cork canonlybeharvestedevery9–15 years(minimum9 yearsaccording to Portuguese legislation), and severalstudies have shown that cork age (and thus thickness)is inversely related to post-fire mortality (e.g. Lamey1893; Pampiro et al. 1992; Cabezudo et al. 1995; Pausas 1997; Barberis et al. 2003; Catry et al. 2007; Moreiraetal.2007).Thethickerthebark,thelowertheexpected level of post-fire damage (again, buds andcambium are more protected from fire).Barberis et al. (2003) and Moreira et al. (2007)provided evidence that trees with larger diameter atbreast height (DBH) had a lower probability of survival. Possible explanation for this pattern includea likely higher amount of stripping damages, highersusceptibilitytostressordiseasesandhigherfrequencyof poor management practices (e.g. deep ploughing,excessive pruning) in older trees (Costa et al. 2004;Moreira et al. 2007). A bigger tree that has sufferedseveral damage events across its lifespan is thereforeprone to higher levels of post-fire damage, mainlybecauseofthelackofcarbohydratereservestoinvestinresprouting (Iwasa and Kubo 1997). The aim of this article is to explore the importanceof tree size, bark thickness and cork stripping indetermining the whole range of post-fire responsetypes in cork oak. In particular we aimed to: (a)quantify the relative frequency of four different post-fire responses in burned cork oak trees 1.5 years afteran intense wildfire and (b) explore whether stripping,bark thickness and tree size influenced each of theobserved types of post-fire responses as hypothesised. Methods Study area and plot definitionThe study area is located in ‘‘Serra do Caldeira˜o’’, amountain range in the Algarve province, southernPortugal. The climate is Mediterranean with an Plant Ecol (2009) 201:77–85 79  1 3  average annual temperature and rainfall of 16.6  Cand 900 mm, respectively. The altitude ranges from150 to 580 m above sea level. Soils consist mainlyof shallow schist lithosols that have a low fertilityand are prone to erosion. The landscape is charac-terised by vast expanses of cork oak forests rangingfrom areas with high tree cover, to ‘‘montados’’ thathave scattered trees and an understory of crops orpastures. In the 2004 summer, an intense wildfireburned ca. 25,000 ha in this region. A 1  9  1 km 2 grid of points covering part of the burned area wasused to define a 50 m-radius circle (sampling plot)around each point. Plots were checked in the fieldfor accessibility and to confirm whether they hadburned and were dominated by cork oak trees. Atotal of 40 plots were ultimately selected. Largewithin-plot variability in tree size and cork age (andconsequently bark thickness) was common sincecork debarking was not carried out simultaneouslyon all individual trees (for further details seeMoreira et al. 2007).Tree variablesIndividual tree evaluation in the plots took placebetween December 2005 and April 2006, approxi-mately 1.5 years after the fire. Trees were assessedalong four 50-m strip transects departing from theplot centres at right angles. Given the very highdensity of young trees in many plots, only trees largerthan ca. 9 cm DBH were measured. Approximately30 trees per plot were assessed (mean  ±  s.e. of 28.8  ±  0.51, range  =  14–30,  n  =  40) yielding a totalof 1,151 individuals. For each tree, several variableswere measured (see Moreira et al. 2007 for details);however, for the purposes of this article only thefollowing variables are presented: (a) tree size (DBH,cm), taken as the average of two measurements at1.3 m above ground level, (b) bark thickness (averagethickness, cm) at breast height, calculated from fourmeasurements using a bark gauge and (c) presence/ absence of cork stripping in order to distinguishunstripped trees with virgin cork from exploited treeswhere cork debarking (stripping) had occurred atleast once. The types of post-fire responses were alsoassessed and classified into four mutually exclusivecategories: (a) dead trees (no resprouting from thebase or crown), (b) trees that resprouted exclusivelyfrom the crown, (c) trees that resprouted exclusivelyfrom the base (thus with a dead stem) and (d) treesthat resprouted from both the crown and base.Data analysisTo examine the influence of tree variables on post-fireresponse types, an information theoretic approach wasused based on the Akaike information criterioncorrected for small sample sizes (AIC c ) (Burnhamand Anderson 2002). This approach starts with theformulation of a series of models that rely on anunderstandingofthesystembeingstudied,followedbyan assessment of how different putative modelscompare to the reality (Rushton et al. 2004). The suite of candidate models is compared using AIC c , and thesmallertheAIC c valuethebetterthemodelfitsthedata.Each of the four response types was modelledseparately using a binary variable taking the value 1for the specific response type and 0 for the remainingtypes. A generalised linear model with binomial errorstructure and a logit link function (McCullagh andNelder 1989) was used to test a group of biologicallyplausible models, including separate models for eachof the three variables (stripping, bark thickness,DBH) assumed to be biologically significant, and allpossible combinations of these variables. Two inter-action terms were also added to this list of variables:stripping  9  bark thickness, as previous analysesshowed that we could expect different responses,for a given bark thickness, of unstripped or exploitedtrees (Moreira et al. 2007); and stripping  9  DBH, asthe effects of tree size could also vary according tostripping status. This yielded four groups (one groupper response type) of 27 models each, resulting fromall combinations of these five variables. The smallerAIC c  among the models in each group was used toidentify the more parsimonious model (Burnham andAnderson 2002) for each response type.The fit and predictive performance of the modelswith smaller AIC c  was evaluated through the likeli-hood ratio statistic (full model  v 2 ) and by calculatingthe area under the receiver operating characteristics(ROC) curve (Saveland and Neueschwander 1990;Pearce and Ferrier 2000). This has the advantage of assessing model performance in a threshold-indepen-dent fashion, being independent of the prevalence of the several response types. The AUC varies between0.5 (no discrimination ability) to 1 (perfect discrimi-nation ability) (Pearce and Ferrier 2000). Usually, 80 Plant Ecol (2009) 201:77–85  1 3  AUC values of 0.5–0.7 are taken to indicate lowaccuracy, values of 0.7–0.9 indicate useful applica-tions and values above 0.9 indicate high accuracy(Swets1988).ThecalculationoftheAUCandstandard error wasbased ona non-parametric assumption. Forabetter visualization of the expected probabilities of thefitted models, data from bark thickness and tree sizeweregroupedintoclasses.Theformerwasdividedintothree classes:  B 2 cm (33.8% of the trees), 2–4 cm(54.4%) and [ 4 cm (11.8%). Tree size was alsodivided into three DBH categories: B 20 cm (28.8% of the trees), 20–40 cm (58.5%) and [ 40 cm (12.7%).TherewasnocorrelationbetweenbarkthicknessandDBH ( r   =  0.021,  n  =  1151,  P  =  0.487). However,exploited trees ( n  =  859) had significantly larger DBHthan unstripped ones ( n  =  292) (mean  ±  s.e. of 30.7 ±  0.406 cm and 16.5  ±  0.253 cm, respectively, t  -test,  t   =  29.6,  P \ 0.001), and had a slightly thinnerbark (mean  ±  s.e. of 2.39  ±  1.289 cm and2.93 ±  0.835, respectively,  t  -test,  t   =  8.1,  P \ 0.001). Results Response typesFor the 1,151 sampled trees, the most commonresponse type was resprouting exclusively fromcrown (68.8%,  n  =  792 trees), followed by death(15.8%,  n  =  182), simultaneous resprouting from thecrown and base (10.1%,  n  =  116) and lastly, respro-uting exclusively from the base (5.3%,  n  =  61).Influence of predictor variables on response typesThe more parsimonious model for tree death, amongthe set of models compared, is shown in Table 1 andFig. 2. The probability of a tree dying increased if ithad been exploited and had a larger DBH. Bark thickness was also a key variable but only if treeswere exploited, in this case the probability of deathincreased as bark thickness decreased. Similarly todeath, the model with the lowest AIC for resproutingonly from base showed that this response type wasalso more likely in stripped trees (Table 1; Fig. 2). Bark thickness was an important variable in the caseof stripped trees, and was negatively correlated tobasal resprouting probability. The more parsimoniousmodel for simultaneous resprouting from the base andcrown (Table 1; Fig. 2) included only DBH, with larger trees being less likely to show this responsetype. Finally, resprouting exclusively from the crownwas more likely in unstripped trees (Table 1; Fig. 2). For stripped trees, this resprouting type increasedwith bark thickness and decreased with DBH. Over-all, model performance was low to moderate withAUC values ranging from 0.64 to 0.82. Discussion Differences in sprouting behaviour are important forunderstanding vegetation dynamics, extinction risksfor threatened species and for defining managementregimes for woody plants (Bond and Midgley 2003). Table 1  Generalized linear models with the lowest AIC c  among the set of models compared, for each of the four post-fire responsetypes in cork oak (death, resprouting exclusively from crown, resprouting exclusively from base, resprouting from both crown andbase).Variable Death Base only Crown and base Crown onlyStripping 1.645  ±  0.280 2.955  ±  0.440  - 1.464  ±  0.278Bark thicknessDBH 0.031  ±  0.007  - 0.055  ±  0.012Stripping  9  bark thickness  - 0.688  ±  0.086  - 1.272  ±  0.182 0.809  ±  0.075Stripping  9  DBH  - 0.016  ±  0.007Constant  - 2.722  ±  0.229  - 3.570  ±  0.358  - 0.842  ±  0.283 0.940  ±  0.130Model  v 2 101.95 81.03 27.79 153.65AUC 0.71  ±  0.022 0.82  ±  0.026 0.64  ±  0.026 0.70  ±  0.017The variables entering each model (linear predictor), their coefficients ( ± s.e.), the model  v 2 and the area under the ROC curve(AUC  ±  s.e.) are shown for each response type. See Fig. 2 for model visualization. All model  v 2 , variable coefficients and AUCvalues are significant ( P \ 0.05)Plant Ecol (2009) 201:77–85 81  1 3
Related Search
We Need Your Support
Thank you for visiting our website and your interest in our free products and services. We are nonprofit website to share and download documents. To the running of this website, we need your help to support us.

Thanks to everyone for your continued support.

No, Thanks

We need your sign to support Project to invent "SMART AND CONTROLLABLE REFLECTIVE BALLOONS" to cover the Sun and Save Our Earth.

More details...

Sign Now!

We are very appreciated for your Prompt Action!