BioOne sees sustainable scholarly publishing as an inherently collaborative enterprise connecting authors, nonprofit publishers, academic institutions,research libraries, and research funders in the common goal of maximizing access to critical research.

The Role of Nurse Trees in Mitigating Fire Effects on Tropical Dry ForestRestoration: A Case StudyAuthor(s): Ricardo J. Santiago-García, Sandra Molina Colón, Phillip Sollins, and Skip J. Van BloemSource: AMBIO: A Journal of the Human Environment, 37(7):604-608. 2008.Published By: Royal Swedish Academy of SciencesDOI: http://dx.doi.org/10.1579/0044-7447-37.7.604URL: http://www.bioone.org/doi/full/10.1579/0044-7447-37.7.604

BioOne (www.bioone.org) is a nonprofit, online aggregation of core research in the biological, ecological,and environmental sciences. BioOne provides a sustainable online platform for over 170 journals and bookspublished by nonprofit societies, associations, museums, institutions, and presses.

Your use of this PDF, the BioOne Web site, and all posted and associated content indicates your acceptance ofBioOne’s Terms of Use, available at www.bioone.org/page/terms_of_use.

Usage of BioOne content is strictly limited to personal, educational, and non-commercial use. Commercialinquiries or rights and permissions requests should be directed to the individual publisher as copyright holder.

Ricardo J. Santiago-Garcıa, Sandra Molina Colon, Phillip Sollins and Skip J. Van Bloem

The Role of Nurse Trees in Mitigating FireEffects on Tropical Dry Forest Restoration:A Case Study

The threat of fire is always a consideration whenestablishing a forest restoration program. Two wildfiresoccurred in 2006 and 2007 in an established dry forestrestoration project in Puerto Rico. The original goal of theproject was to determine differential growth responses ofnative trees under the nurse tree Leucaena leucocephalaversus in open sites. Tree species growth, mortality andresponse to the fires were evaluated according to theirleaf habit, successional status, and prefire tolerance toenvironmental conditions. Results showed that regard-less of a species’ leaf habit and successional status,trees attained greater height and lower mortality undernurse trees. In open sites, sprouting was the mostcommon fire response and mature-forest and evergreenspecies had greater postfire survival than pioneers anddeciduous species. Although nurse trees are typicallyused to help manage nutrient or light environments inreforestation projects, these trees also appear to providea secondary benefit of limiting fire damage by reducingfuel load.

INTRODUCTION

Fire is a serious risk in any ecosystem restoration project andcan result in the total loss of the restored site. Additionally, firepresents a particular threat to habitats where fires areuncommon, because the flora and fauna lack adaptivemechanisms that allow them to withstand a fire. This is thecase with seasonally dry tropical forests, an ecosystem in whichfire is not a part of the natural disturbance regime because fire isthe product of human activity (1). Some wildfires in dry forestsare intentionally set by farmers to prepare agricultural fieldsand to maintain pastures (2) while others are accidental or actsof vandalism (pers. obs.).

Periodic fires, intentional or accidental, can lead to a changein species composition by favoring fire-resistant species andeliminating common species in the otherwise undisturbedhabitats (3). Exotic grasses benefit from these repeated firesand can produce a grass-fire cycle that prevents forestregeneration (4). The exclusion of fire is the only means ofnatural dry forest regeneration in abandoned pastures (5).

Forest restoration projects practice fire control and exclusionby direct or indirect methods. Prescribed burns, fire breaks, andregular weeding are direct methods that reduce fuel load andminimize the spread of fires. An indirect method that may beconsidered for management is wildfire prevention by publicawareness programs. In addition, the use of nurse trees forreforestation can indirectly influence fire occurrence. Moreshade in the understory of nurse trees decreases the amount ofgrass cover around the planted trees, which results in a reducedfuel load should a wildfire occur. This particular effect isadvantageous in habitats where fires are common such asregenerating or degraded dry forests.

Nurse trees are adult plants that are frequently used toenhance seedling establishment by mitigating extreme environ-

mental factors (6). Plant growth adjacent to a mature tree canhave improved microclimatic conditions, protection againstherbivory, and increased water and nutrient availability (7). Amature tree improves microclimatic conditions in the understo-ry by reducing extreme solar radiation reaching seedlings andprotecting them by lowering air and soil temperature andreducing soil water evaporation (8).

The present study evaluates juvenile tree species survivor-ship, growth, and response to fire when planted with andwithout Leucaena leucocephala (Lam.) de Wit as nurse trees.Leucaena is a common naturalized tree that grows onabandoned pasturelands, along roadsides, and in early second-ary dry forests in the arid zones of Puerto Rico (9). We alsoassess native species survival following multiple fires in openplots. The original intent of this project was not to study fireeffects per se, but to study native species growth in reforestationprojects. This study takes advantage of an unplanned experi-ment.

MATERIALS AND METHODS

This study occurred in the Sierra Bermeja hills of southwesternPuerto Rico in the municipality of Cabo Rojo, an area werefires are now common during the dry season. This area used tobe dominated by subtropical dry forest (sensu Holdridge [10])and was subsequently converted to pasture. The study site islocated at the top of a flat hill where the slope is nearly zero andtherefore aspect is not relevant to this study. Annual rainfallaverages 860 mm and is distributed bimodally into two seasons,a strong dry period from December to April and a lesser onefrom June to August (11) (Fig. 1).

Figure 1. Rainfall since establishment of the Sierra Bermejareforestation project. Rainfall data from Cabo Rojo, about 5 kmwest of the site. Bars are monthly means from 2002–2005 withstandard error. Filled symbols on lines indicate months of fireoccurrence. Annual rainfall was 455 mm in 2002, 1040 mm in 2003,930 mm in 2004, 1172 mm in 2005, 861 mm in 2006, and 515 mm in2007.

604 Ambio Vol. 37, No. 7–8, December 2008� Royal Swedish Academy of Sciences 2008http://www.ambio.kva.se

In 2002, as part of a larger study about forest restoration,two plots were established in an area that had beenpredominantly dominated by L. leucocephala. Grass, primarilyUrochloa maxima (Jacq.) R. D. Webster (guinea grass), iscommon in both plots. One of the plots had a L. leucocephalaoverstory (nurse tree plot) whereas the second plot hadLeucaena removed (open plot). Leucaena trees within 1–2 mof each sapling planted were cut at ground level in the openplot. Canopy closure was estimated at 100% for the nurse treeplot in 2006 but was not measured in the open plot since it hasno canopy. The plots are separated by about 25 m including adirt driveway and unmanaged ‘‘native’’ vegetation consisting ofgrass and L. leucocephala.

Twenty-four native tree species were planted in each plot tostudy the effects of nonnative forest cover on native speciesgrowth rates. A total 165 trees were initially planted in 2002.Sample sizes for species within plots varied due to theavailability of saplings, but were equal for each species betweentreatments when the plots were established. Species with greateravailability, and therefore more replication in the plots, tendedto be more common in Guanica State Forest, a mature dryforest nearby (11). Selected species were chosen based onavailability and not fire tolerance. Since there were fewindividuals per species, we classified them based on theirsuccessional status (pioneer or mature), leaf habit (deciduous or

evergreen) (Table 1) and tolerance to local prefire environmen-tal conditions based on growth and mortality before the firstfire. We were not able to group them by fire tolerance as thereare no data available for these species with the exception of thispaper.

Two accidental fires occurred in these plots, one in February2006 and another in March 2007, when high winds carriedburning embers over our fire breaks. Trees in both plots wereplanted in lanes and grass was trimmed approximately 6 monthsbefore the first fire hit. In the nurse tree plot, grass wasapproximately 40 cm in height before the first fire while in theopen plot it was about 60 cm in the lanes and 2 m in betweenlanes. Approximately 1 month before the second fire, grass wasbetween 60–70 cm in height and trimmed to ground level in a 1-m radius around the saplings in the open plot. The grass in thenurse tree plot was about 30 cm in height with no trimming. Theopen plot burned in both fires while the nurse tree plot onlyburned in the first fire. Fuel load was measured after both firesin the nurse tree plot, the open plot, and the open plot aftertrimming.

Both plots were evaluated in February, July, and December2006, and March and October 2007. Prefire mortality wasrecorded from 2002–2006. We also evaluated prefire sitetolerance as species with the highest and lowest growth forthe period of 2002–2006. Postfire effects were analyzed bymeasurements of tree survivorship, growth, and response to fire.Growth and mortality of individuals were measured after theperiod of the first fire. Survivorship was assessed relative to thetotal number of individuals present before the first fire in eachplot (i.e., January 2006). Tree height was measured with aheight pole and growth was calculated as the difference ingrowth between the sampling periods. We classified survivalfrom a fire as to whether the main living stem of the treesurvived or if it resprouted from the base. Since the nurse treeplot only burned in the February 2006 fire, we only used theperiod from February to December 2006 for comparisonsbetween open and nurse tree plots and we only used the openplot to assess effects of multiple fires experienced at approxi-mately 1-y intervals.

RESULTS

Prefire Conditions and Seedling Establishment

Fuel loads were 400 kg ha�1 in the nurse tree plot, 6150 kg ha�1

in the untrimmed sections of the open plot, and 3150 kg ha�1 inthe open plot after trimming. We attribute this difference insurface fuel loads as the reason why the nurse tree plot did notburn in the second fire. During the first fire the open plotcompletely burned with no green vegetation remaining, whileapproximately 75% of the nurse tree plot burned. Mean charheight in the nurse tree plot was 73 cm and the maximum charheight was 205 cm. In the open plot, we were not able todetermine the mean or maximum char height because all thetrees burned, however the maximum sapling height andtherefore the minimum char height was 2.2 m. Of the 70individuals present at the moment of the first fire in the nursetree plot, 25 were completely dry with no leaves while 37individuals remained green and the remaining 5 had burntleaves but living stems. All of the 69 individuals present in theopen plot were completely dry and burned. After the secondfire, the open plot burned between lanes but it did not burnaround the surviving trees where grass was trimmed. However,the heat and flames from the fire reached the trees resulting indried out or burned leaves.

Prefire mortality resulted in 14 dead individuals from 7species in the open plot and 11 individuals from 10 species in the

Table 1. Species planted in both plots present at the time of thefirst fire. Species that did not survive the first fire in both plotswere not included (see results). Resprouts did not appear untilafter the first fire in February 2006 and were only present in theopen plot. Sample sizes are the numbers of individuals in eachplot before and after the first fire, and in the open plot the numberof individuals after the fire that resprouted or had surviving stems.Species have been grouped by their successional status (pioneer[P] or mature [M]) and leaf habit (deciduous [D] or evergreen [E]).Taxonomic nomenclature follows Little et al. (12, 13). Species inbold type had surviving saplings in the open site at the end of thestudy.

Amyris elemifera L. (M, E) Guaiacum sanctum L. (M, E)Nurse tree (n ¼ 4, 4) Nurse tree (n ¼ 6, 5)Open (n ¼ 4, 0) Open-resprouts (n ¼ 5, 3)

Bursera simaruba L. (P, D) Guettarda elliptica Sw. (M, D)Nurse tree (n ¼ 1, 1) Nurse tree (n ¼ 1, 1)Open (n ¼ 2, 0) Open-resprouts (n ¼ 1, 1)

Cassine xylocarpa Vent. (M, E) Gymnanthes lucida Sw. (M, E)Nurse tree (n ¼ 1, 1) Nurse tree (n ¼ 6, 6)Open-resprouts (n ¼ 4, 2)

Citharexylum fruticosum L.(P, D)

Nurse tree (n ¼ 1, 1)Open-resprouts (n ¼ 1, 1)

Open-surviving, resprouts(n ¼ 7, 3, 3)

Jacquinia arborea Vahl (M, E)Nurse tree (n ¼ 4, 4)Open-resprouts (n ¼ 3, 2)

Coccoloba diversifoliaJacq. (M, E)

Nurse tree (n ¼ 6, 6)Open-resprouts (n ¼ 5, 4

Ouratea littoralis Urb. (M, D)Nurse tree (n ¼ 1, 1)

Pictetia acuelata (Vahl)Urban (P, D)

Coccoloba kruggi Lindau (M, E) Nurse tree (n ¼ 3, 3)Open-resprouts (n ¼ 2, 2) Open-resprouts (n ¼ 3, 3)

Coccoloba uvifera (L.) L. (P, E)Nurse tree (n ¼ 1, 1)

Pisonia albida (Heimerl)Britton (P, D)

Open-surviving (n ¼ 1, 1) Nurse tree (n ¼ 6, 6)Colubrina arborescens (Sarg.)

Miller (P, E)Open (n ¼ 6, 0)

Reynosia uncinata Urban (P, E)Nurse tree (n ¼ 3, 2) Nurse tree (n ¼ 2, 2)Open (n ¼ 3, 0) Open-resprouts (n ¼ 1, 1)

Crossopetalum rhacomaCrantz. (M, E)

Tabebuia heterophylla (DC.)Britton (P, D)

Nurse tree (n ¼ 6, 6) Nurse tree (n ¼ 1, 1)Open-resprouts (n ¼ 6, 5) Open (n ¼ 1, 0)

Erythroxylon aerolatum L. (M, D)Nurse tree (n ¼ 5, 5)Open-resprouts (n ¼ 3, 2)

Thouinia portoricensisRadlk. (P, D)

Nurse tree (n ¼ 3, 3)Erythroxylon rotundifolium

Lunan (M, D)Open-resprouts (n ¼ 3, 3)

Nurse tree (n ¼ 3, 3)Open (n ¼ 4, 0)

Ambio Vol. 37, No. 7–8, December 2008 605� Royal Swedish Academy of Sciences 2008http://www.ambio.kva.se

nurse tree plot. All dead individuals in the open plot weremature forest species and only two of which were deciduous.Two species in the nurse tree plot with dead individuals werepioneers and 5 were evergreen.

In the nurse tree plot, the prefire general median growth forall species from 2002–2006 was 73.7 cm with Colubrinaarborescens presenting the greatest growth of 182.5 cm followedby Coccoloba diversifolia with 171.2 cm and Gymnanthes lucidawith 111.2 cm. All 3 species are evergreen and only C.arborescens is a pioneer species. We considered these 3 speciesas having high tolerance to environmental conditions present inthe nurse tree plot. The 3 species with the lowest growth (andtherefore lowest tolerance) in the nurse tree plot were Jacquiniaarborea with 26.6 cm followed by Crossopetalum rhacoma with40.8 cm and Amyris elemifera with 47.3 cm. All 3 are evergreen,mature-forest species. Pioneer species had a median growth of86.6 cm, mature-forest species grew 71.1 cm, evergreen speciesgrew 93 cm and deciduous species had a median growth of 73.5cm.

The open plot had a prefire 4-y median growth of 34.2 cm.Species that grew best in the plot and thus highly tolerant of thelocal environment were C. diversifolia with the greatest growth(171.4 cm) followed by Reynosia uncinata (100.2 cm), and G.lucida (72.7 cm). The 3 species are evergreen and only R.uncinata is a pioneer. In contrast, the 3 species with the lowestgrowth in the open plot and therefore having low tolerance toopen site reforestation were Guaiacum sanctum with 5.9 cmfollowed by A. elemifera with 9.7 cm and Pisonia albida with13.0 cm. P. albida is the only pioneer and is deciduous. Pioneerspecies had a median growth of 19.3 cm, mature-forest speciesgrew 43.7 cm, evergreen species grew 55.8 cm, and deciduousspecies grew 27.3 cm. Both high and low tolerance species in thenurse tree plot attained greater median heights than in the openplot.

Fire Effects and Postfire Response

Over the entire course of the study 81% of the saplings in theopen plot died. After the first fire, 9% of the saplings planted inthe nurse tree plot died compared to 42% in the open plot. Allindividuals of 17 of the 24 species (73%) present before the firstfire died in the open plot. In contrast, the nurse tree plot lost 3species of the original 24 (13%). Both plots only had a singleindividual of Capparis hastata, Eugenia foetida and Krugioden-dron ferreum. These individuals died due to the first fire andwere the only species that had 100% mortality in the nurse treeplot. Individuals from 3 other species with replication experi-enced mortality in the nurse tree plot: 17% of G. sanctum, 25%of A. elemifera, and 33% of C. arborescens.

Our results showed that both immediate postfire and delayedmortality of functional groups were always higher in open sitesthan in the nurse tree plot (Fig. 2). In the open plot, deciduousspecies had greater mortality (64%) than evergreen (32%), andpioneer mortality (62%) exceeded mature forest species (38%).In the nurse tree plot, no functional group had greater than 7%mortality (evergreen species) (Fig. 2). Delayed mortality in theopen plot after the first fire was greater for mature forest species(9% of the prefire saplings) than pioneers (5%) and all delayedmortality was among evergreen species. During the year inbetween fires, the 50% of the species planted in the open plotsuffered 100% mortality. After the second fire, from March toOctober 2007, 5 more species were eliminated. By October 2007,only 7 species and 19% of the individual saplings remained inthe open plot: C. diversifolia, C. rhacoma, Guettarda elliptica, G.lucida (mature and evergreen), Citharexylum fruticosum, Picte-tia acuelata, and Thouinia portoricensis (pioneer and deciduous).

The average height decreased for saplings of each species ineach plot after the first fire. However, the decrease was muchgreater in the open plot where the median height of all saplingsdecreased by over 90 cm while in the nurse tree plot the decreasewas 50 cm (Table 2). Throughout the study, the saplings weretaller in the nurse tree plot (Table 2). In the nurse tree plotdeciduous and pioneer species were taller than evergreen andmature species by Oct 2007. During the first year’s postburnperiod (March to December 2006) species in the open plot had ahigher median growth rate (17 cm) than in the nurse tree plot(10 cm) (Table 2).

In the open plot, deciduous species were slightly taller afterthe fire but evergreen species overtook the height of deciduousspecies due to being less affected by the second fire (Table 2).Pioneers had greater median height than mature species in theopen plot throughout the study. The stems of only fourindividuals survived the first fire while the remaining 25survivors resprouted from the base.

There were differences in growth between resprouters andtrees in which the main living stem survived in the open plot(survivors). Only 2 species, both evergreen, had survivors afterthe first fire: Coccoloba uvifera and G. lucida. Coccoloba uviferasurvivors died by December 2006 leaving only G. lucida. G.lucida survivors grew 34 cm during the year after the first fireand 17 cm during the second postfire period from March toOctober 2007. Thirteen species resprouted after the first fire,having a median average growth for the period betweenFebruary and December 2006 of 17 cm. After the second firein 2007 most of the 2006 sprouts died and resprouters presenteda median average growth of only 12 cm on newly generatedsprouts.

Generally, species that showed high tolerance to siteconditions before the fire also did well after the fire. C.diversifolia and G. lucida were among the top 3 for the amountof postfire growth. In the nurse tree plot, C. diversifolia’smedian height increase was 58 cm (rank 2) and G. lucida 34 cm(rank 3). These species each grew 33 cm in the open plot (Rank1-tie). Pisona albida had the highest median growth (66 cm) inthe nurse tree plot despite being at the median for prefiregrowth. The remaining high-tolerant species in the open plot, R.uncinata, was killed by the first fire and replaced by Thouiniaportoricensis with the third-most growth.

Those species exhibiting low tolerance to prefire siteconditions in the open plot were all eliminated by the end ofthe study, however G. sanctum did survive the first fire. The 3species with the lowest growth in the nurse-tree plot during the

Figure 2. Cumulative mortality by species functional group in eachplot (O ¼ open, NT ¼ nurse tree) by sampling period. High and lowtolerance are the 3 species with the greatest and least growth ineach plot prior to the first fire (see text). There was no mortality ofdeciduous saplings in the nurse tree plot.

606 Ambio Vol. 37, No. 7–8, December 2008� Royal Swedish Academy of Sciences 2008http://www.ambio.kva.se

same period were R. uncinata with�5 cm, C. arborescens (8 cm;the species with the highest prefire growth), and C. rhacoma (11cm). It is notable that J. arborea and A. elemifera had belowmedian overall postfire growth in the nurse tree plot, but hadthe second and fourth highest relative growth (postfire growth/February 2006 height). These 2 species, along with P. albida,clearly grew better after the fire in the nurse tree plot.

DISCUSSION

We attribute the difference in mortality between plots to adifference in fuel load and consequently different fire intensities.Although both plots were trimmed periodically (always at thesame time), grass grew back more slowly and was less denseunder the shade of the nurse tree plot, resulting in a lowerintensity fire. This conclusion is supported by findings from dryforests of Nicaragua, where low intensity fires have been shownto have high postfire survivorship (2).

Greater growth and survival are benefits of nurse trees in ourdry forest ecosystems. Prior to the first fire, only 2 species were.10% taller in the open plot than the nurse tree plot (J. arboreaand C. rhacoma). All other species were taller in the nurse treeplot, probably due to the better microclimatic conditionscompared to the open site. In addition to mitigating fire effects,nurse trees provide shade, causing a decrease in soil and airtemperatures and improving the availability of water and soilnutrients (8).

The 2 survival mechanisms, resprouting and main stemsurvivorship, presented marked contrasts in their response tomultiple fires. Resprouting was the most common mechanismand these sprouts tended to have rapid relative growth due tohigh root masses available to support low stem biomass.Resprouting was also the most common fire response in dryforest species in Nicaragua after 1 fire (2) and sproutingindividuals dominated regeneration after low-intensity fires in alowland tropical dry forest in Bolivia (14). However, thedisadvantage to basal resprouting is that stem height is resetto ground level, making these stems much more susceptible tosubsequent fires. This was clearly shown by the high mortalityin our open plot after the second fire.

Once a sapling survives a fire, or multiple fires, the postfireenvironment also contains a collection of potential stresses.Immediately after fires, soil temperatures are higher which

would decrease soil moisture. Although sapling-grass competi-tion may be temporarily decreased, grass reestablishes quicklyand would be a stronger competitor for light with resproutingsaplings (4, 15) as opposed to those with surviving main stems.

While our sample sizes are too small for definitiveconclusions, they do provide a hypothetical list of species thatwould be useful (or problematic) for reforestation in fire-proneCaribbean dry forest. The 7 species that remained alive after 2fires (Table 1) may be particularly suitable for reforestation inopen, fire-prone areas. Those species that failed to survive thefirst fire in the nurse tree plot may indicate species that arepoorly adapted to fire disturbance. In open sites we found thatmature-forest and evergreen species had about half themortality of pioneers and deciduous species after a single fire.Evergreen species may have been able to respond better to thefire stress by producing new leaves to provide photosynthateneeded for recovery. Mature dry forest species have to tolerateunpredictable patterns in severity and length of drought. Theymay therefore be better able to withstand a single fire thanpioneer species that are genetically inclined for rapid coloniza-tion and not stress tolerance. However, it was clear in our casethat our native species do not survive multiple fires in annualsuccession.

CONCLUSIONS

The results of this case study suggested that nurse treereforestation could be a promising tool for protecting saplingsin fire-prone habitats like disturbed dry forests. Furthermore,saplings from mature forest species survived a single fire ingreater proportions than pioneers, so reforestation programsthat use mature forest species to speed up succession do notappear to be more susceptible to fire. Limited sample size andplot replication preclude us from being able to definitively listfire tolerant and intolerant species, however, the benefits ofreforestation under nurse trees were clear and it appeared thatevergreens survived better than deciduous saplings. In the eventof a wildfire nurse trees reduce mortality and increase growth ofnative trees compared to open-site reforestation. Planting underL. leucocephala limited grass growth, reduced fuel loads, andresulted in greater postfire stem survival. We found similarpatterns to other burned tropical dry forests (2), whereresprouting was a more common fire-response mechanism than

Table 2. Median heights of deciduous, evergreen, pioneer, and mature species by plot. The number of species in each plot after the first fire arein parentheses. Bold columns are measurements taken immediately after fires. There was no fire in March 2007 in the nurse tree plot.

Median height of species (cm)

January February July December March October2006 2006 2006 2006 2007 2007

All species pooledOpen plot (15) 98 0 6 17 11 26

Surviving stems (2) 90 29 36 62 52 69Resprouts (14) 0 0 5 12 11 23

Nurse tree plot (21) 144 94 100 104 112 125Nurse tree plot

Deciduous 166 99 107 124 154 154Evergreen 115 92 100 98 83 100Pioneer 165 103 117 114 135 155Mature 127 88 95 103 111 118High tolerance 169 103 119 137 157 173Low tolerance 53 27 40 43 42 50

Open plotDeciduous 114 0 6 17 13 26Evergreen 85 0 3 12 20 31Pioneer 122 0 16 24 21 30Mature 89 0 4 8 10 20High tolerance 144 0 10 33 40 55Low tolerance 26 0 3 7 4 0

Ambio Vol. 37, No. 7–8, December 2008 607� Royal Swedish Academy of Sciences 2008http://www.ambio.kva.se

having the main stem survive. However, stem survival ofsaplings under nurse trees was higher than reprouting rates inopen sites and resulted in trees attaining and maintaininggreater heights faster than they would if they were forced toresprout after each fire. Although not fire-specific, highersurvivorship of seedlings has been shown under nurse tree sitesin dry forests (16). Use of nurse trees is a practical restorationmethod that not only reduces the risk of losing a reforestationproject to fire but also allows land managers and ecologists totake advantage of the trees present at a restoration site insteadof having to remove them. Future studies that explicitly studyinteraction between fire dynamics, woody species establishment,and nurse tree benefits would be useful. Furthermore, nativeCaribbean dry forest species have not been subjected to naturalfire sources (17) and therefore may be less likely to survive evenlow intensity fires than dry forest species elsewhere. Because ofthe prevalence of highly flammable nonnative grasses in oldfields and other restoration sites, susceptibility of native woodyspecies to fire should be determined before widespread planting.

References and Notes

1. Janzen, D.H. 1988. Management of habitat fragments in a tropical dry forest: growth.Ann. Mo. Bot. Gard. 75, 105–116.

2. Otterstrom, S.M., Schwartz, M.W. and Velazquez-Rocha, I. 2006. Responses to fire inselected tropical dry forest trees. Biotropica 38, 592–598.

3. Viera, D.L. and Scariot, A. 2006. Principles of natural regeneration of tropical dryforests for restoration. Restor. Ecol. 14, 11–20.

4. D’Antonio, C.M. and Vitousek, P.M. 1992. Biological invasions by exotic grasses, thegrass/fire cycle, and global change. Ann. Rev. Ecol. System. 23, 63–87.

5. Janzen, D.H. 2002. Tropical dry forest: area de conservacion Guanacaste, northwesternCosta Rica. In: Handbook of Ecological Restoration: Restoration in Practice, Vol. 2.Perrow, M.R. and Davy, A.J. (eds). Cambridge University Press, Cambridge, pp. 559–583.

6. Cavieres, L.A., Badano, E.I., Sierra-Almeida, A., Gomez-Gonzalez, S. and Molina-Montenegro, M.A. 2006. Positive interactions between alpine plant species and the nursecushion plant Laretia acaulis do not increase with elevation in the Andes of centralChile. New Phytol. 169, 59–69.

7. Callaway, R.M. and Pugnaire, F.I. 1999. Facilitation in plant communities. In:Handbook of Functional Plant Ecology. Pugnaire, F.I. and Valladares, F. (eds). MarcelDekker, New York, pp. 623–649.

8. Padilla, F.M. and Pugnaire, F.I. 2006. The role of nurse plants in the restoration ofdegraded environments. Front. Ecol. Environ. 4, 196–202.

9. Parotta, J.A. 1992. Leucaena leucocephala (Lam.) de Wit Leucaena, tantan. In: Silvics ofNative and Exotic Trees of Puerto Rico and the Caribbean Islands. SO-ITF-SM-52. USDepartment of Agriculture, Forest Service, Southern Forest Experiment Station, NewOrleans, Louisiana, pp. 308–316.

10. Holdridge, L.R. 1967. Life Zone Ecology. Tropical Science Center, San Jose, Costa Rica,206 pp.

11. Murphy, P.G. and Lugo, A.E. 1986. Structure and biomass of a subtropical dry forest inPuerto Rico. Biotropica 18, 89–96.

12. Little, E.L., Wadsworth, F.H. and Marrero, J. 2001. Common Trees of Puerto Rico andthe Virgin Islands. (2nd revised ed). Editorial de la Universidad de Puerto Rico, RioPiedras, 764 pp. (In Spanish).

13. Little, E.L., Woodbury, R.O. and Wadsworth, F.H. 1988. Trees of Puerto Rico and theVirgin Islands. Second Volume. US Department of Agriculture Handbook 449-S. RioPiedras, 1177 pp. (In Spanish).

14. Kennard, D.K., Gould, K., Putz, F.E., Fredericksen, T.S. and Morales, F. 2002. Effectof disturbance intensity on regeneration mechanisms in a tropical dry forest. For. Ecol.Manag. 162, 197–208.

15. Bond, W.J. and Midgley, J.J. 2001. Ecology of sprouting in woody plants: thepersistence niche. Trends Ecol. Evol. 16, 45–51.

16. Sanchez-Velazquez, L.R., Quintero-Gradilla, S., Aragon-Cruz, R. and Pineda-Lopez,MaR. 2004. Nurses for Brosimum alicastrum reintroduction in secondary tropical dryforest. For. Ecol. Manag. 198, 404–404.

17. Murphy, P.G. and Lugo, A.E. 1986. Ecology of tropical dry forest. Ann. Rev. Ecol.System. 17, 67–88.

18. We would like to thank F. O. Perez Martinez, B. T. Wolfe, and J. G. MartinezRodriguez for their help in the field; two anonymous reviewers and W. A. Gould forhelpful comments on previous drafts of this paper. This research was funded by theUSDA Forest Service McIntire Stennis grant PR00-MS-014 and International Instituteof Tropical Forestry, cooperative agreement 05-CA-11120101-022 to SJV.

Ricardo J. Santiago-Garcıa is a graduate student in theAgronomy program at the University of Puerto Rico atMayaguez. His research is focused on the role of nurse treesin improving reforestation success. He plans to continue to usebasic ecological science in practical applications for environ-mental management. His address: Department of Agronomyand Soils, University of Puerto Rico-Mayaguez, PO Box 9030,Mayaguez, PR 00681.E-mail: aquamanpr@gmail.com

Sandra Molina Colon is the Coordinator of the EnvironmentalSciences Program and Professor of Biology at the PontificalCatholic University of Ponce, Puerto Rico. Her research hasexplained the long term effects of land use practices in PuertoRican dry forests. From this experience, she has become aleading voice for local conservation of the few remainingmature dry forests. Her address: Programa de CienciasAmbientales Pontificia Universidad Catolica de Puerto Rico2250, Las Americas Suite 606, Ponce, PR 00717-0777.E-mail: smolinacolon@email.pucpr.edu

Phillip Sollins is Professor Emeritus from the Department ofForest Ecosystems and Society, Oregon State University. Hehas a long and distinguished research career with importantcontributions to our understanding of soil carbon, soilchemistry, and forest-soil feedbacks. He maintains an activeresearch program while splitting his time between Oregon andPuerto Rico. His address: Department of Forest Ecosystemsand Society Oregon State University, 321 Richardson Hall,Corvallis, OR 97331.E-mail: phil.sollins@oregonstate.edu

Skip J. Van Bloem is an Associate Professor in theDepartment of Agronomy and Soils at the University of PuertoRico at Mayaguez. Skip is an ecologist whose research hasfocused on nutrient cycling and physiological responses todisturbance, and how these responses influence ecosystemstructure. His interests include how natural and manmadedisturbances occurring at different periods interact on ecosys-tem structure and stability. His address: Department ofAgronomy and Soils, University of Puerto Rico-MayaguezPO Box 9030, Mayaguez, PR 00681.E-mail: svanbloem@uprm.edu

608 Ambio Vol. 37, No. 7–8, December 2008� Royal Swedish Academy of Sciences 2008http://www.ambio.kva.se