Biol. Lett.(2007)3, 344–347doi:10.1098/rsbl.2007.0043Published online6 March 2007Population ecologyDeforestation and apparentextinctions of endemicforest beetlesin MadagascarIlkka Hanski1,*, Helena Koivulehto1,Alison Cameron2and Pierre Rahagalala31Department of Biological and Environmental Sciences,PO Box 65, 00014 University of Helsinki, Finland2Department of Environmental Science, Policy & Management,137 Mulford Hall, University of California,Berkeley, CA 94720-3114, USA3De ́partement d’Entomologie, Faculte ́des Sciences, BP 906,Universite ́d’Antananarivo, Madagascar*Author for correspondence(ilkka.hanski@helsinki.fi).Madagascar has lost about half of its forest coversince 1953 with much regional variation, forinstance most of the coastal lowland forests havebeen cleared. We sampled the endemic forest-dwelling Helictopleurini dung beetles acrossMadagascar during 2002–2006. Our samplesinclude 29 of the 51 previously known species forwhich locality information is available. The mostsignificant factor explaining apparent extinctions(species not collected by us) is forest loss withinthe historical range of the focal species,suggesting that deforestation has already causedthe extinction, or effective extinction, of a largenumber of insect species with small geographicalranges, typical for many endemic taxa inMadagascar. Currently, roughly 10% of theoriginal forest cover remains. Species–areaconsiderations suggest that this will allow roughlyhalf of the species to persist. Our results areconsistent with this prediction.Keywords:Madagascar; forest loss;tropical extinctions; dung beetle;Helictopleurini1. INTRODUCTIONNearly two-thirds of all described species of animalsand plants are insects, of which roughly 45% arebeetles, but the level of threat imposed by the globalenvironmental changes to beetle and insect diversitiesremains poorly documented. Of the described speciesof insects and beetles, only 0.07 and 0.02% havebeen classified as globally extinct or threatened,respectively, compared with 24% for mammals and12% for birds (IUCN 2004;Birdlife 2006). Almostcertainly, the figures for insects reflect lack of knowl-edge rather than lack of threat to insects. In the partsof the world that are best studied, insects are equallyas threatened as vertebrates (Thomaset al. 2004)oreven more threatened (Stein & Flack 1997). InFinland, with a beetle fauna of 3640 species, 12.1%of the species are classified as nationally extinct(1.6%) or threatened (10.5%), compared with 15.3%in birds, 17.0% in mammals and 15.4% in vascularplants (Rassiet al. 2001).Knowledge of tropical insects is particularlylimited. One exception is the fauna of Singapore,which has been studied since 1819 (Brooket al.2003). The historically documented extinction rate is34% for birds, 43% for mammals, 26% for vascularplants and 38% for butterflies (Brooket al. 2003).Again, there is no difference between vertebrates,plants and butterflies.Here, we examine apparent extinctions in a well-studied group ofca60 endemic forest-inhabitingdung beetles in Madagascar. Our extensive samplingacross Madagascar during 2002–2006 yielded 54% ofthe described species, as well as four new ones. Weanalyse the factors that explain whether a previouslyknown species was included in our samples or not. Werefer to the latter species as ‘apparently extinct’ forshort. We are particularly interested in examiningwhether regional forest loss may explain apparentextinctions. The annual rate of deforestation has been1.4% between 1953 and 1993 and 2.0% between1993 and 1999 (Dufils 2003), and hence roughly halfof the forest cover has been lost in the past 50 years.The endemic forest dung beetles have small ranges(Viljanenet al. in preparation), as have many othertaxa in Madagascar (Wilme ́et al. 2006), and hence itis possible that many species have become (effectively)extinct due to a regionally high rate of forest loss.2. MATERIAL AND METHODSThe taxonomy of the endemic tribe Helictopleurini (Coprinae,Scarabaeidae) is well known (Lebis 1960;Paulian 1986and otherpapers;Montreuil 2005,in pressand other papers). We recordedthe sampling information for all specimens in the main collectionsof the Paris National Museum of Natural History. The collectionsinclude 51 species for which locality information is available,sampled from 126 distinct localities (figure 1a) during 1875–1990(3341 specimens). Half of the specimens have been collected priorto 1926 but almost none since the mid-1970s. The remaining ninespecies either lack sampling information (three species) or there areno specimens in Paris. We suspect that several of these latter speciesshould be synonymized with the better known species.During the years 2002–2006, we sampled Helictopleurini usingdung and carrion-baited traps. Our sample of 4880 specimens wascollected from 61 localities (figure 1b), including larger samples fromRanomafana National Park (NP), Masoala NP, Makira Reserve,Andasibe NP, Ambila-Lemaintso, Manombo reserve, Isalo NP,Zombitse-Vohibasia NP and Andahohelo NP. Smaller samples werecollected by the personnel of forest reserves in 52 localities acrossMadagascar using trapping kits provided by us. Our samples includefour new species (Montreuil 2005,in press).A single map of forest cover change between the years 1970,1990 and 2000 was provided by Conservation International (CI )at approximately 30 m resolution (Harperet al. 2005, unpublisheddata). This map was reclassified to single out forest cover for theyear 2000. The main sources of data were the Inventaire Ecolo-gique et Forestaire National ( IEFN) classification of LandsatThematic Mapper 5 data for the year 1993 and estimates of forestcover for the year 1999 carried out by the Joint Research Centre( JRC)–Space Application Institute, Ispra, Italy, using SPOT-4data. Where possible, CI used additional SPOT images to adddetails in the regions covered by dense cloud in the original IEFNand JRC images.CI also provided a digitized version of the 1953 forest covermap produced byHumbertet al. (1965), rasterized at the sameresolution as the 2000 map. The original map was produced usingaerial photographs and ground truthing. The 1953 study appears tohave focused on mapping major forest blocks, as the map does notcontain small fragments in remote areas that were present in thesatellite images. We assume that the additional small fragmentspresent in 1970 had not grown in the intervening years, and henceany forest cover present in 1970 but absent in the 1953 map wasadded to the latter.The maps for 1953 and 2000 were summarized as percentage offorest cover in grid cells of 0.18resolution (11.2 km at the equator).Using these maps, we calculated the extent of forest cover withinReceived23 January 2007Accepted5 February 2007344This journal isq2007 The Royal SocietyThis is an open-access article distributed under the terms of the Creative Commons Attribution License, which permits unrestricted use,distribution, and reproduction in any medium, provided the original work is properly cited.
the range of speciesxat timet(1953 or 2000) asFx;tZð1=nÞXXeKadijAj;ð2:1ÞwhereAjis the percentage of forest cover in cellj,dijis the distancebetween the sampling localityiand celljin degrees, and the secondsummation is over thensampling localities for speciesx. Thus,Fx,tmeasures the average amount of forest in the surroundings of thehistorical sampling localities for speciesx, giving decreasing weightto cells with increasing distance from the sampling localities. WeassumedaZ10, which gives substantial (greater than 0.05) weightto distances up to 33.6 km. The absolute and relative forest losseswere calculated asFx,1953KFx,2000andFx,2000/Fx,1953, respectively.As our sampling localities do not evenly cover all of Madagascar,it is possible that we have failed to sample a species because only afew or even none of our sampling localities were within its range. Toaccount for this, we calculated the average distance of thenhistoricalsampling localities for speciesxto our sampling localities asDxZð1=nÞXXeKadij;ð2:2Þwheredijis the distance (in degrees) from theith historical samplinglocality to thejth locality in our sampling. Since we used the valueaZ1 in this calculation, 18(112 km) distance has the weight 0.37.Other explanatory variables include body size, the last yearwhen the species was sampled prior to our sampling, the (log)number of historical sampling localities, the (log) number ofindividuals in the Paris collections and the range of the species,defined as the distance between the two most distant historicalsampling localities. As the latter three variables are all stronglycorrelated, we calculated the first principal component as a generalmeasure of past commonness ( PC1). PC1 accounted for 90% ofvariation in the three original variables.3. RESULTSOut of the 51 species sampled prior to our work and forwhich locality information is available, we have sampled29 species but failed to collect 22 other species. We ranstepwise logistic models with the explanatory variablesdescribed in§2 to explain apparent extinctions. Relativeforest loss (Fx,2000/Fx,1953) entered the model first(table 1). The other two variables that were selectedwere distance to our sampling localities (Dx)andPC1.An equally good model was obtained if PC1 wasdropped from the candidate variables, in which case thelast year when the species had been sampled wasselected at 5% level.Since the uncollected species cannot be included inour molecular phylogeny (Koivulehtoet al.inprep-aration), we cannot critically assess possible phyloge-netic bias in apparent extinctions. However, we mayuse the eight taxonomic species groups ofLebis (1960)as a proxy, as these groups match the clades in themolecular phylogeny reasonably well. There is nodifference in the fraction of apparently extinct speciesamong the morphological groups (pZ0.46).Most of the 22 species that we have not collectedhave been previously collected from only one (nZ9)or two (nZ6) localities, widely scattered acrossMadagascar (figure 1c). The most striking exception isHelictopleurus undatus, which has been collected from27 localities across much of Madagascar. A closerexamination shows that since 1950 this species has beenrestricted to a small region in the northeast (figure 1d),giving the impression that it gradually disappearedfrom its former range during the twentieth century.4. DISCUSSIONHelictopleurini are mostly relatively large, manyspecies have colour-patterned elytra and are diurnaland easy to sample with dung and carrion-baitedtraps. For these reasons, Helictopleurini have beenrelatively well collected in the past. In our specializedsampling, only four new species were discovered, allof which appear to be very localized and rare. Fourspecies compose only 7% of the described species,which is a small percentage for tropical insects in aslarge and diverse an area as Madagascar. We concludethat Helictopleurini have been sufficiently well col-lected in the past to warrant this analysis.The best predictor of whether a species wascollected by us or not was relative forest loss withinits past range. The estimated remaining forest coverfrom 1953 to 2000 ranges from 10 to 60% fordifferent species (figure 2). Species for which lessthan one-third of the 1953 forest cover remainstended to be apparently extinct. The other factorsthat had a significant effect on species’ occurrence(a)(b)(c)(d)Figure 1. Maps showing (a) the historical sampling localities during 1875–1990, (b) our sampling localities during 2002–2006,(c) the sampling localities of 21 apparently extinct species apart fromH. undatusand (d) the localities forH. undatusprior to1900 (grey circles), during 1901–1950 (open triangles) and during 1951–1973 (black squares). Localities are mostly shown withresolution of 0.58.Table 1. Stepwise logistic regression model explainingwhetherwehavesampledaspeciesofHelictopleurini(nZ29)or not (nZ22). (p-value for the full modelZ0.38; d.f.Z47.)explanatory variabledeviancedifferencepconstant69.74relative forest loss61.528.210.004distance to our traps53.248.280.004past commonness49.244.000.046Deforestation and extinctions in MadagascarI. Hanskiet al.345Biol. Lett.(2007)
during 2002–2006 were PC1, the last year when thespecies had been collected prior to our sampling andthe distances from the past sampling localities to oursampling sites. None of these effects is unexpected.Most of the apparently extinct species have not beencollected for 50 years or longer (figure 2).It is perhaps surprising that the past or current extentof forest within the range of the species had nosignificant effect on whether we collected the species ornot. Admittedly, the forest maps depict a crude pictureof forest cover and the past sampling localities provideinaccurate estimates of species’ ranges. In this situation,long-term change rather than one-time state of theenvironment may better reflect species’ responses.Furthermore, in many regions forest loss from 1953 to2000 was a continuation of a longer process. Somespecies may have been on decline already in 1953 dueto deforestation, and the occurrence of species istracking forest loss and fragmentation with a time lag(Hanski & Ovaskainen 2002).Forest loss and fragmentation is associated with anincreasing pressure on lemurs, the most importantdung producers in Madagascar. According to the fossilrecord, 16 large-bodied lemurs have gone extinct sincehuman colonizationca2300 years ago (Burneyet al.2004). The long-term and dramatic decline ofH. undatus(figure 1d) may be due to high degree ofresource specialization. Unfortunately, nothing isknown of its biology and by now this once exceptionallywidespread species may already be extinct.Madagascar has large numbers of species withnarrow geographical ranges (Wilme ́et al. 2006). In thissituation, and taking into account that Madagascar hasalready lost most of its forests, very large numbers ofinsects and other poorly known taxa may already beextinct, effectively extinct or rapidly heading towardsextinction due to past and current deforestation. Thecurrent plans to expand the protected area network tosix million hectares will amount toca10% of theoriginal forest cover. Species–area considerationssuggest that this will protect roughly half of the species(MacArthur & Wilson 1967). Our results are consistentwith this prediction.We thank the personnel of forest reserves and parks inMadagascar for their help in sampling, Evgeniy Meyke forcalculations of forest loss, Navjot Sodhi and an anonymousreferee for their comments and the Academy of Finland forsupport (grant nos 64350 and 20286, Finnish Centre ofExcellence Programme 2000–2005).Birdlife2006(http://www.birdlife.org/datazone/species/index.html)Brook, B. W., Sodhi, N. S. & Ng, P. K. L. 2003 Cata-strophic extinctions follow deforestation in Singapore.Nature424, 420–423. (doi:10.1038/nature01795)Burney, D., Burney, L., Godfrey, L., Jungers, W.,Goodman, S., Wright, H. & Jull, T. 2004 A chronologyfor late prehistoric Madagascar.J. Hum. Evol.47, 25–63.(doi:10.1016/j.jhevol.2004.05.005)Dufils, J.-M. 2003 Remaining forest cover. InThe naturalhistory of Madagascar(eds S. M. Goodman & J. P.Benstead), pp. 88–96. Chicago, IL: The University ofChicago Press.Hanski, I. & Ovaskainen, O. 2002 Extinction debt atextinction threshold.Conserv. Biol.16, 666–573. (doi:10.1046/j.1523-1739.2002.00342.x)Humbert, H., Cours Darne, G., Besaire, H., Blasco, F., Legris,P. & Riquier, J. 1965 Carte International du Tapis Vegetalet conditions e ́cologies a`1/1.000.000. Extrait des Traveuxde la Section Scientifique et Technique de L’InstituteFranc ̧ais de Pondiche ́ry, France. Digitized by the Centerfor Applied Biodiversity Science, Washington, DC.IUCN 20042004 IUCN Red List of Threatened Species.Seehttp://www.iucnredlist.org.Koivulehto, H., Orsini, L. & Hanski, I. In preparation.Adaptive radiation of tropical forest dung beetles inMadagascar.Lebis, E. 1960. Helictopleurina. InInsectes Cole ́opte`resScarabaeidae(ed. D. Tananarive). Tsimbazaza: L’Institutde la Recherche Scientifique Tananarive.MacArthur, R. H. & Wilson, E. O. 1967The theory of islandbiogeography. Princeton, NJ: Princeton University Press.Montreuil, O. 2005 NewHelictopleurusd’Orbigny, 1915from Madagascar and review of the “semivirensgroup”sensu Lebis 1960 ( Insecta, Coleoptera, Scarabaeidae,Oniticellini).Zoosystema27, 123–135.Montreuil, O. 2007 NouveauxHelictopleurusd’Orbigny,1915.Bulletin de la Socie ́te ́entomologique de France.past commonness–2–101234last year when collected1880 1900 1920 1940 1960 1980 2000remaining forest cover0.60.70.100.20.40.30.5(a)(b)Figure 2. Plots showing whether a species has been collected by us (open symbols) or not (closed symbols, ‘apparentlyextinct’ species), depending on relative forest loss within the range of the species plotted (a) against the last year when thespecies was collected prior to our sampling and (b) against the past commonness of the species ( PC1).346 I. Hanskiet al. Deforestation and extinctions in MadagascarBiol. Lett.(2007)
Paulian, R. 1986 Catalogue des Cole ́opte`res Scarabaeidaede Madagascar.Bulletin d’Academie Malgache62,89–111.Rassi, P., Alanen, A., Kanerva, T. & Mannerkoski, I.(eds) 2001Suomen Lajien Uhanalaisuus 2000. Helsinki,Finland: Edita.Stein, B. A. & Flack, S. R. 19971997 Species report card: thestate of U.S. plants and animals. Arlington, Virginia: TheNature Conservancy.Thomas, J. A., Telfer, M. G., Roy, D. B., Preston, C. D.,Greenwood, J. J. D., Asher, J., Fox, R., Clarke, R. T. &Lawton, H. H. 2004 Comparative losses of Britishbutterflies, birds, and plants and the global extinctioncrisis.Science303, 1879–1881. (doi:10.1126/science.1095046)Viljanen, H., Koivulehto, H., Montreuil, O., Rahagalala, P.,Randriamanaitra, T., Johnson, S. & Hanski, I.In preparation. An endemic forest dung beetle commu-nity in Madagascar.Wilme ́, L., Goodman, S. M. & Ganzhorn, J. U. 2006Biogeographic evolution of Madagascar’s microendemicbiota.Science312, 1063–1065. (doi:10.1126/science.1122806)Deforestation and extinctions in MadagascarI. Hanskiet al.347Biol. Lett.(2007)