69 5 Effects of Fire on the Diversity of Geometrid Moths on Mt. Kilimanjaro Jan C. Axmacher, Ludger Scheuermann, Marion Schrumpf, Herbert V.M. Lyaruu, Konrad Fiedler, and Klaus Müller-Hohenstein INTRODUCTION Fires often occur along the upper margin of the montane rain forest on Mt. Kilimanjaro (Beck et al. 1986; Blot 1999; Meyer 1890; Salehe 1997). This is especially true for exceptionally dry periods which regularly appear during the El Niño southern oscillation (ENSO) phases. As a consequence, the forest boundary has been profoundly lowered, and the forest has been replaced by heathland dominated by Erica excelsa , E. trimera , and E. arborea (Hemp and Beck 2001), with some small isolated patches of forest remaining above the current closed forest boundary. This change in the vegetation is accompa- nied by various changes in ecological condi- tions. The temperature drops more rapidly at nightfall outside the forest, and the microcli- mate is drier within the open heath vegetation, leading to the disappearance of many ferns and epiphytic plants outside the forest. The objective of our study was to explore how these fire-induced changes affect herbivo- rous insects along the upper forest margins and in forest remnants. We investigated changes in species composition and diversity of nocturnal geometrid moth communities along an altitudi- nal transect covering intact closed forests below 3000 m, a mosaic of forest remnants and heath- land at 3100 m, and heathland vegetation at 3300 m. The sites at 3100 m, therefore, enable a direct comparison of geometrid moth communities in heathland and forest fragments, whereas the lower and higher portions of the transect, albeit the altitudinal differences between the sites, pro- vide comparison of community structure and diversity of geometrid moths between large closed forest and heathland habitats. Geometrids were selected for four reasons: (1) They are one of the three most diverse families of Lepi- doptera. (2) They are taxonomically rather well known (Scoble 1999). (3) They proved to be suitable indicators of environmental change in a number of studies in tropical forests (e.g. Beck et al. 2002; Brehm 2002; Holloway 1984; Intachat et al. 1999; Kitching et al. 2000). (4) They regularly occur in quite large numbers at high altitudes at which most other insects are not very common (Brehm 2002). TEST SITES AND METHODS S TUDY A REA Mt. Kilimanjaro, the highest mountain in Africa (5892 m), is situated 300 km south of the equa- tor in Tanzania. It is a volcanic complex com- posed of three volcanic centers, covering an area of roughly 80 km × 40 km. Within the Mt. Kilimanjaro area, the study sites are located in the southwest part of the mountain along the Machame tourist route below the Shira plateau, along an altitudinal gradient from 2300 to 3300 m. The upper por- tions of this area were heavily affected by forest and heath fires during the last El Niño event (1996 to 1997), when large areas of the 3523_C005.fm Page 69 Wednesday, November 23, 2005 7:50 AM Copyright © 2006 Taylor & Francis Group, LLC 70 Land Use Change and Mountain Biodiversity ericaceous belt on the Shira plateau burned (Salehe 1997). These fires also affected the heath sites included in this study, whereas the forest remnants and forest sites remained intact. M ETHODS Geometrid moths were investigated at 13 sites. Eight of these sites were located within the intact forest belt in the ranges 2300 m, 2580 m, 2700 m, and 2900 m; two plots consisted of isolated patches of remnant forest at 3100 m; and three sites of heathland ranged from 3100 to 3300 m. Moths were attracted by an accumulator- powered weak UV-containing light source (Syl- vania blacklight-blue, F 15 W/BLB-TB) placed within a white reflective gauze cylinder (diam- eter 0.8 m, height 1.6 m). From the gauze, insects were sampled manually using plastic jars prepared with diethylether. The samples were taken from 7 to 10 P . M . local time between October 2001 and January 2002. Catches were restricted to periods without strong moonlight, avoiding the period from 5 d before to 5 d after full moon. For each site, two to four night catches were pooled for analysis. Subsequently, specimens were spread, sorted to morphospe- cies level, and taxonomically identified as far as possible at the Zoologische Staatssammlung in Munich. Data handling and statistical analysis were carried out using Microsoft Access, SPSS, Esti- mateS (Colwell 2000), and an Excel macro pro- vided by Meßner (1996) for calculating the nor- malized expected species shared (NESS) index and Sørensen index of similarity. RESULTS A total of 2158 individuals representing 92 morphospecies were sampled (Table 5.1); 41 (44%) of these morphospecies representing 1563 individuals (72%) could be identified to species level. P ROPORTIONS OF S UBFAMILIES WITHIN G EOMETRIDAE In the study area, three subfamilies of geometrid moths, Larentiinae, Ennominae, and Geometrinae, are largely represented and can be found along the whole altitudinal gradient. Two other subfamilies, Sterrhinae and Desmo- bathrinae, also occur but account for less than 3% of individuals as well as species at any site and disappear at sites above 2900 m. Concern- ing the number of individuals caught, Larenti- inae was the dominant subfamily throughout the whole study area, always accounting for more than 50% of the total catch. Within the closed forest, Ennominae proved to be the sec- ond most abundant subfamily, whereas they were much rarer in the heath, where the pro- portion of Geometrinae was very high. In the forest fragments, Geometrinae and Ennominae were caught in roughly equal proportions (Fig- ure 5.1). When considering the proportions of sub- families according to the occurrence of mor- phospecies on the plots, Larentiinae are domi- nant on all plots, followed by Ennominae. Geometrinae accounted for a smaller share of species richness, and the proportions seen in heathland are not different from those obtained at some lower forest sites (Figure 5.2). TABLE 5.1 Number of catches performed and number of geometrid moth individuals and morphospecies caught at each site Site 2300 F1 2300 F2 2580 F1 2580 F2 2700 F1 2700 F2 2900 F1 2900 F2 3100 Ff1 3100 Ff2 3100 E 3300 E1 3300 E2 Catches 4333223344222 Individuals 224 138 148 177 215 178 181 147 138 75 196 184 157 Morphospecies 32 31 28 33 31 37 38 32 34 24 26 19 21 3523_C005.fm Page 70 Wednesday, November 23, 2005 7:50 AM Copyright © 2006 Taylor & Francis Group, LLC Effects of Fire on the Diversity of Geometrid Moths on Mt. Kilimanjaro 71 A LPHA -D IVERSITY Fisher’s alpha proved to be a reliable measure of the diversity of moth communities in a num- ber of studies (e.g. Brehm 2002; Chey et al. 1997; Schulze 2000), as long as the samples fit a log-series distribution. This is true for all the sites sampled in this study. Once sample FIGURE 5.1 Proportions of subfamilies, based on numbers of individuals. Sites are sorted by altitude. (Site codes: altitude in meters, F = forest, Ff = forest fragment, E = heath vegetation). FIGURE 5.2 Proportions of subfamilies, based on numbers of morphospecies. Sites are sorted by altitude. (Site codes: altitude in meters, F = forest, Ff = forest fragment, E = heath vegetation). Forest fragment HeathForest 100 75 50 25 0 Percentage 2300 F1 2300 F2 2580 F1 2580 F2 2700 F1 2700 F2 2900 F1 2900 F2 3100 Ff1 3100 Ff2 3100 E 3300 E1 3300 E2 Site Desmobathrinae Sterrhinae Ennominae LarentiinaeGeometrinae Forest fragment HeathForest 100 75 50 25 0 Percentage 2300 F1 2300 F2 2580 F1 2580 F2 2700 F1 2700 F2 2900 F1 2900 F2 3100 Ff1 3100 Ff2 3100 E 3300 E1 3300 E2 Site Desmobathrinae Sterrhinae Ennominae LarentiinaeGeometrinae 3523_C005.fm Page 71 Wednesday, November 23, 2005 7:50 AM Copyright © 2006 Taylor & Francis Group, LLC 72 Land Use Change and Mountain Biodiversity size exceeds 120 individuals per site in spe- cies-poor communities as in the case of Mt. Kilimanjaro, Fisher’s alpha shows a greater independence of sample size than most other alpha diversity measures (Hayek and Buzas 1997). This number was exceeded in our study at all but one site, namely, the forest fragment 3100 Ff 2. When comparing geometrid diversity from different sites, two main results emerge. First, geometrid communities at Mt. Kiliman- jaro are not very diverse overall (values rang- ing from 9 to 15 in closed forests above 2250 m). Second, the values from the forest below the upper treeline below 3000 m and in the forest fragments at 3100 m do not differ sig- nificantly, whereas diversity in the heath both at 3100 m and 3300 m is significantly reduced (Figure 5.3) compared to all forested sites. B ETA -D IVERSITY To study the beta-diversity of the moth commu- nities in the study area, two similarity measures, the Sørensen index and the NESS index, were chosen. As an ordination method, nonlinear multidimensional scaling was applied (see Beck et al. 2002 and Schulze and Fiedler, in press, for methods). Figure 5.4 shows the ordi- nations of the samples for the Sørensen index and the NESS index for a low sample size parameter m = 1 to emphasize the similarity of moth ensembles with regard to the most abun- dant species and a higher parameter m = 37 with stress on the occurrence of rare species (Grassle and Smith 1976) In all three ordinations, the sites are primar- ily arranged according to their altitudinal posi- tion along the first axis. Apart from this, the forest plots are always positioned closely together and separated from the remaining sites by the first dimension. In all three ordinations, communities from forest fragments are posi- tioned closer to the heathland than to the forest sites. A clear distinction between heath and for- est fragments is only discernible with the NESS index, particularly with m = 1. According to the FIGURE 5.3 Diversity of Geometridae: Fisher’s alpha. The bars indicate the standard deviation, calculated using EstimateS. (From Colwell, R.K. [2000]. Sites are sorted by altitude. (Site codes: altitude in meters, F = forest, Ff = forest fragment, E = heath vegetation.) 2300 F1 2300 F2 2580 F1 2580 F2 2700 F1 2700 F2 2900 F1 2900 F2 3100 Ff1 3100 Ff2 3100 E 3300 E1 3300 E2 20 18 16 14 12 10 8 6 4 Fisher’s alpha Site Forest Heath Forest fragment 3523_C005.fm Page 72 Wednesday, November 23, 2005 7:50 AM Copyright © 2006 Taylor & Francis Group, LLC Effects of Fire on the Diversity of Geometrid Moths on Mt. Kilimanjaro 73 FIGURE 5.4A Two-dimensional nonlinear scaling of samples of Geometridae, based on matrices calculated with (A) Sørensen and NESS similarity indices, (B) small sample-size parameter m = 1, and (C) large sample- size parameter m = 37. Numbers correspond to the altitude in meters; = forest, ● = forest fragment, X = heath. Low stress values indicate an excellent goodness-of-fit of the ordinations to the initial similarity data. −2.5 −2 −1.5 −1.5 −1 −1 −0.5 −0.5 0 0 0.5 0.5 1 1 1.5 2 3300 3100 3100 2900 2700 2300 Stress: 0.04 2580 Dimension 2 (a) −2 −1.5 −1 −1 −0.5 −0.5 0 0 0.5 0.5 1 1 1.5 1.5 3300 3100 3100 2900 2700 2300 Stress: 0.04 2580 Dimension 2 (b) −2.5 −2 −1.5 −1.5 −1 −1 −0.5 −0.5 0 0 0.5 0.5 1 1 1.5 2 3300 3100 3100 2900 2700 2300 Stress: 0.06 2580 Dimension 2 (c) 3523_C005.fm Page 73 Wednesday, November 23, 2005 7:50 AM Copyright © 2006 Taylor & Francis Group, LLC 74 Land Use Change and Mountain Biodiversity Sørensen index, forest fragments are indistin- guishable from the heath plot at the same alti- tude (3100 m). Therefore, differentiation between heathland and forest fragments is more a matter of altered abundance relationships, whereas species composition in forest frag- ments cannot be distinguished clearly from heathlands at identical altitudes. DISCUSSION Our data reveal strong effects of fire-induced spread of heathland on geometrid moth com- munities at the three levels of subfamily rela- tionships, alpha-diversity, and species compo- sition. On the level of subfamilies, in addition to the complementary altitudinal increase of Larentiinae and decrease of Ennominae, Geometrinae appear to benefit from fire, as they are much more numerous at the heathland as compared to forest fragments situated at similar altitudes. However, this applies only to two species of the genus Comostolopsis (War- ren), which has become very abundant in this habitat. The altitudinal decrease of Ennominae is even more accentuated in open heathland, probably as a consequence of the rather strong link of most Ennominae to forested habitats. The overall predominance of Larentiinae has generally been observed at both high altitudes and latitudes that might reflect particular ther- mal adaptations of this moth subfamily (Brehm 2002). Within the intact montane rain forest occurring in the lower parts of the study area below 3000 m, there was no observed decrease in overall geometrid moth diversity with increasing altitude. The highest forest sites at 2700 and 2900 m had even slightly more diverse geometrid moth communities. Overall diversity in this tropical montane forest was nowhere higher than in temperate zone wood- lands and much lower than in other tropical mountain regions (Schulze 2000; Brehm 2002). Yet, species composition within the for- est belt revealed a continuous change from 2300 to 2900 m, regardless of the analytical methods chosen. Above 3000 m, the marked drop in geometrid alpha-diversity of the heath in com- parison to forest fragments in the direct vicinity (Plot 3100 Ff 1 and 2 in comparison to 3100 E) shows that the destruction of the forest along the upper forest line can trigger distinct effects on the diversity of these insects. Forest rem- nants do not show a similar impoverishment. However, with regard to species composition, forest remnants are more similar to the sur- rounding heathland than to intact forest. It is, therefore, not yet clear whether such forest rem- nants could serve as a permanent refuge for forest species, offering the possibility of recol- onization if the forest could encroach into the heathland again with time. Alternatively, these forest fragments might actually be sinks rather than sources, if the forest moth species are only present due to permanent immigration from the intact forest. CONCLUSION Fire-induced conversion of montane forest into heath scrubland severely reduces diversity of geometrid moths. If ENSO-triggered drought periods and related forest fires increase in fre- quency and severity, as is suggested by recent climate models, our data suggest massive con- sequences for large proportions of insects that are specifically dependent on the more humid and thermally less extreme and less variable conditions prevailing in the high-altitude closed forest. SUMMARY Due to reoccurring fires, the upper forest boundary at Mt. Kilimanjaro has become dis- tinctly lowered and fragmented. These large- scale and long-term vegetation changes are reflected in the changes to geometrid moth communities. These changes were studied on 13 sampling sites located within the intact for- est below 3000 m, as well as in forest fragments and in the heath vegetation replacing the forest above 3000 m. The overall diversity of geometrid moths decreases when the forest is replaced by heath. The species composition, especially the abundance pattern of the species, is dramatically altered. Forest fragments remaining within heathland areas might serve as refuge areas for some of the moth species, 3523_C005.fm Page 74 Wednesday, November 23, 2005 7:50 AM Copyright © 2006 Taylor & Francis Group, LLC Effects of Fire on the Diversity of Geometrid Moths on Mt. Kilimanjaro 75 but only to a limited extent, as is reflected by strong faunal dissimilarities between intact for- est and forest fragments. ACKNOWLEDGMENTS We would like to thank the German Research Foundation (DFG, Mu 364 14-1, 2 & 3), For- estry and Beekeeping Division (FBD), Kili- manjaro National Park (KINAPA), Tanzania National Parks (TANAPA), Tanzania Commis- sion for Science and Technology (COSTECH), the CITES-Office, Dr. Axel Hausmann at the Zoologische Staatssammlung, Munich, and numerous other helping hands and minds for their kind help and cooperation. References Beck, E., Scheibe, R., and Schulze, E.D. (1986). Recovery from fire: observations in the alpine vegetation of western Mt. Kiliman- jaro (Tanzania). Phytocoenologia , 14: 55–77. Beck, J., Schulze, C.H., Linsenmair, K.E., and Fiedler, K. (2002). From forest to farmland: diversity of geometrid moths along two hab- itat gradients on Borneo. Journal of Tropical Ecology , 17: 33–51. Blot, J. (1999). The incidence of forest fire in Kili- manjaro. French Institute for Research in Africa, IFRA Les Cahier 16: 85–86. Brehm, G. (2002). Diversity of Geometrid Moths in a Montane Rainforest in Ecuador. Ph.D. dis- sertation, University of Bayreuth. Chey, V.K., Holloway, J.D., and Speight, M.R. (1997). Diversity of moths in forest planta- tions and natural forests in Sabah. Bulletin of Entomological Research , 87: 371–385. Colwell, R.K. (2000). EstimateS: Statistical Estima- tion of Species Richness and Shared Species from Samples, Version 6.0 b1. http://vice- roy.eeb.uconn.edu/estimates. Grassle, J.F. and Smith, W. (1976). A similarity mea- sure sensitive to the contribution of rare spe- cies and its use in investigation of variation in marine benthic communities. Oecologia , 25: 13–22. Hayek, L.A. and Buzas, M.A. (1997). Surveying Nat- ural Populations . Columbia University Press, New York. Hemp, A. and Beck, E. (2001). Erica excelsa as a fire-tolerating component of Mt. Kiliman- jaro’s forests. Phytocoenologia , 31: 449–475. Holloway, J.D. (1984). Moths as indicator organisms for categorizing rain-forest and monitoring changes and regeneration processes. In Chadwick, A.C. and Sutton, S.L. (Eds.). Tropical Rain-Forests: The Leeds Sympo- sium . Leeds Philosophical and Literary Society, Leeds, 235–242. Intachat, J., Holloway, J.D., and Speight, M.R. (1999). The effects of logging on geometroid moth populations and their diversity in low- land forests of Peninsular Malaysia. Journal of Tropical Forest Science , 11: 61–78. Kitching, R.L., Orr, A.G., Thalib, L., Mitchell, H., Hopkins, M.S., and Graham, A.W. (2000). Moth assemblages as indicators of environ- mental quality in remnants of upland Aus- tralian rain forest. Appl Ecol , 37: 284–297. Meßner, S. (1996). Untersuchungen zur Biodiversität der Myrmecofauna (Formicidae) im Parc National de la Comoe (Elfenbeinküste) . Diploma thesis, University of Würzburg. Meyer, H. (1890). Ostafrikanische Gletscherfahrten. Forschungsreisen im Kilimandscharo- Gebiet . Duncker und Humblot, Leipzig. Salehe, J. (1997). Preliminary Assessment of the Mount Kilimanjaro Forest Fires of February and March 1997. Consultant’s Report, IUCN, Nairobi. Schulze, C.H. (2000). Auswirkungen anthropogener Störungen auf die Diversität von Herbi- voren. Ph.D. dissertation, University of Bayreuth. Schulze, C.H. and Fiedler, K. (2003). Vertical and temporal diversity of a species-rich moth taxon in Borneo. In Basset, Y., Novotny, V., Miller, S., and Kitching, R. (Eds.). Arthro- pods of Tropical Forests: Spatio-Temporal Dynamics and Resource Use in the Canopy . Cambridge University Press, Cambridge, pp. 68–85. Scoble, M. (1999). Geometrid Moths of the World: A Catalogue (Lepidoptera, Geometridae). 2 Volumes. The Natural History Museum, London. 3523_C005.fm Page 75 Wednesday, November 23, 2005 7:50 AM Copyright © 2006 Taylor & Francis Group, LLC . 2 (a) −2 −1 .5 −1 −1 −0 .5 −0 .5 0 0 0 .5 0 .5 1 1 1 .5 1 .5 3300 3100 3100 2900 2700 2300 Stress: 0.04 258 0 Dimension 2 (b) −2 .5 −2 −1 .5 −1 .5 −1 −1 −0 .5 −0 .5 0 0 0 .5 0 .5 1 1 1 .5 2 3300 3100 3100 2900 2700 2300 Stress:. goodness-of-fit of the ordinations to the initial similarity data. −2 .5 −2 −1 .5 −1 .5 −1 −1 −0 .5 −0 .5 0 0 0 .5 0 .5 1 1 1 .5 2 3300 3100 3100 2900 2700 2300 Stress: 0.04 258 0 Dimension 2 (a) −2 −1 .5 −1 −1 −0 .5 −0 .5 0 0 0 .5 0 .5 1 1 1 .5 1 .5 3300 3100 3100 2900 2700 2300 Stress:. 2 3300 3100 3100 2900 2700 2300 Stress: 0.06 258 0 Dimension 2 (c) 352 3_C0 05. fm Page 73 Wednesday, November 23, 20 05 7 :50 AM Copyright © 2006 Taylor & Francis Group, LLC 74 Land Use Change and Mountain Biodiversity