bioRxiv preprint doi: https://doi.org/10.1101/406025; this version posted August 31, 2018. The copyright holder for this preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made available under aCC-BY-NC 4.0 International license.
1 IMPENDING REGENERATION FAILURE OF THE IUCN VULNERABLE BORNEO IRONWOOD
2 (EUSIDEROXYLON ZWAGERI)
3 Lan Qie1, Alexander D. Elsy1, Ashley Stumvoll1, Magdalena Kwasnicka1, Anna L. Peel1, Joseph A. 4 Sullivan1, Maisie S. Ettinger1, Alasdair J. Robertson1, Jeanelle K. Brisbane1, Amber L. Sawyer1, 5 Yan N. Lui1, Siew Ngim Ow1, Matteo Sebastianelli1, Bartosz Majcher1, Muying Duan1, Hannah 6 Vigus1, Grace Pounsin2, Reuben Nilus3, and Robert Ewers1
7 1Department of Life Sciences, Silwood Park Campus, Imperial College London, Ascot SL5 7PY, UK
8 2Maliau Basin Conservation Area, Yayasan Sabah, 88817 Kota Kinabalu, Sabah, Malaysia
9 3Sabah Forestry Department Forest Research Centre, Mile 14 Jl. Sepilok, 90000 Sandakan, 10 Malaysia.
11
12 Abstract
13 The regeneration of many climax species in tropical forest critically depends on adequate seed
14 dispersal and seedling establishment. Here we report the decreased abundance and increased
15 spatial aggregation of younger trees of the Borneo ironwood (Eusideroxylon zwageri) in a
16 protected forest in Sabah Malaysia. We observed a high level of seedling herbivory with strong
17 density dependence, likely exacerbated by local aggregation and contributing to the
18 progressively shrinking size-distribution. We also note the largely undocumented selective
19 herbivory by sambar deer on E. zwageri seedlings. This study highlights the combined impact of
20 altered megafauna community on a tree population through interlinked ecological processes
21 and the need for targeted conservation intervention for this iconic tropical tree species.
22
23 Keywords
1
bioRxiv preprint doi: https://doi.org/10.1101/406025; this version posted August 31, 2018. The copyright holder for this preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made available under aCC-BY-NC 4.0 International license.
24 Seedling survival, neighborhood density, dispersal limitation, spatial clustering, threatened
25 species, wildings
2
bioRxiv preprint doi: https://doi.org/10.1101/406025; this version posted August 31, 2018. The copyright holder for this preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made available under aCC-BY-NC 4.0 International license.
26 INTRODUCTION
27 The loss of megafauna due to climate change, habitat alteration and hunting has a strong
28 impact on plants that are dependent on large-bodied animals for seed dispersal, and this effect
29 is particularly strong in tropical forests (Corlett 2010, Galetti et al. 2017). Many tree species
30 with large fruits have lost the animal dispersers they coevolved with, and consequently suffer
31 reduced seed dispersal and increased spatial aggregation of seedlings, leading to lower survival
32 and reduced gene flow (Harrison et al. 2013, Galetti et al. 2017). Such trees are typically late-
33 succession to canopy species (Harrison et al. 2016), and their regeneration is crucial to the
34 future of intact tropical forests and restoration efforts to accelerate succession in disturbed
35 forests.
36 Borneo ironwood (Eusideroxylon zwageri) is a classic example of rainforest climax tree species
37 facing regeneration challenges. E. zwageri is distributed in eastern and southern Sumatra,
38 Bangka, Belitung, Borneo, the Sulu archipelago and Palawan. In Borneo, E. zwageri was
39 formerly a common component of the main middle story in the mixed dipterocarp forest. It is a
40 prized timber species, renowned for its extraordinary strength, durability and rot resistance,
41 and it is a cultural keystone species to the indigenous people of the region (Franco et al. 2014).
42 Despite being long-living (>1000 years), due to over-exploitation (Peluso 1992) and slow natural
43 regeneration (typically requires >100 years to reach 30 cm diameter) (Irawan 2005), E. zwageri
44 has become scarce across its distributional range and is classified as Vulnerable in the IUCN Red
45 List (1998) . The species produces large drupaceous fruits that measure 10-18 cm x 5-10 cm,
46 and contain a single large seed, measuring 7-15 cm x 4-7 cm, with a very hard seed coat (Irawan
47 2005). Water dispersal for the heavy seeds of E. zwageri is possible but limited as the species
3
bioRxiv preprint doi: https://doi.org/10.1101/406025; this version posted August 31, 2018. The copyright holder for this preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made available under aCC-BY-NC 4.0 International license.
48 often occur away from rivers (Irawan 2005). The large fruit with edible fleshy mesocarp and nut
49 are strong indications that E. zwageri is dispersed by animals, although evidence of this is
50 anecdotal and mainly refers to porcupines (Kostermans et al. 1994, Phillipps & Phillipps 2016).
51 Only megafauna (>44 kg in body mass) would be capable of long-distance dispersal of these
52 large seeds (Galetti et al. 2017). It has been speculated that E. zwageri seeds were dispersed by
53 the Sumatran rhinoceros (Dicerorhinus sumatrensis) (Phillipps & Phillipps 2016), formerly
54 occurring throughout Borneo but now virtually extinct in the wild in Sabah, and down to the
55 last few individuals, if any, still surviving in Borneo (Hance 2015). Despite the value and
56 threatened status of E. zwageri, we still have very limited understanding of its ecology and, in
57 particular, its natural regeneration from seeds.
58 Post germination, the seedling stage represents another bottleneck for the regeneration of a
59 tree species under survival pressures such as pathogens and seedling herbivory, which are
60 typically density dependent processes (Wang & Smith 2002). There have been anecdotal
61 reports that one of the common herbivores in Bornean forests, the sambar deer (Rusa
62 unicolor), preferentially browsed on E. zwageri seedlings. This would potentially exacerbate the
63 regeneration challenge already faced by E. zwageri. Once stems survive the dynamic stage and
64 reach 10 cm diameter the degree of spatial aggregation is expected to be stable across size
65 classes (Condit et al. 2000). Therefore in a long-living E. zwageri population (oldest trees may
66 be > 1000 years), spatial aggregation in younger versus older trees should reflect long-term
67 change in the strength of seed dispersal. We conducted a survey of E. zwageri across all life
68 stages from seedlings to adult trees in an old growth Bornean lowland forest and tested three
69 hypotheses: 1) reduced regeneration in this local population will be reflected in its tree size
4
bioRxiv preprint doi: https://doi.org/10.1101/406025; this version posted August 31, 2018. The copyright holder for this preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made available under aCC-BY-NC 4.0 International license.
70 distribution curve with a flat gradient at small size classes, or a unimodal shape (Halpin &
71 Lorimer 2017), 2) there is increased spatial aggregation in the younger sub-population
72 compared to the older sub-population of adult trees (≥ 10 cm diameter), and 3) seedlings
73 herbivory damage has density dependence.
74
75 MATERIALS AND METHODS
76 We conducted this study in January 2018 at Maliau Basin Conservation Area (4.74° N, 116.97° E,
77 256 m a.s.l.), Sabah, Malaysia. The survey area was in old growth forest on the south bank of
78 the Maliau River near the edge of the conservation area, in the vicinity of a large buffer zone
79 consisting of selectively logged and secondary forests. A 4-ha square plot was set up, within
80 which a grid of 10 by 10 m subplots was established for tree surveys. All E. zwageri trees with
81 DBH ≥ 1 cm were mapped and recorded to diameter classes in 10 cm intervals.
82 There were abundant E. zwageri seedlings in the area from the January 2017 fruiting event.
83 Many of these had noticeable herbivory damage with young leaves missing from the top,
84 characteristic of that by mammalian herbivores, and often resulting in complete defoliation. We
85 surveyed seedlings (all stages with DBH < 1 cm) in a 1.35 ha area in the centre of the 4-ha plot,
86 using 5 by 5 m subplots within the 10 by 10 m grid. In each seedling subplot, we counted E.
87 zwageri seedlings in two categories: “surviving” if the apical meristem was intact, and “fatally
88 damaged” if the apical meristem was eaten, including dead E. zwageri seedlings with evidence
89 of herbivory damage. Leafless and dead E. zwageri seedlings were identified by their
90 characteristic straight and reddish bare stem, often with the stony seed coat still attached at
5
bioRxiv preprint doi: https://doi.org/10.1101/406025; this version posted August 31, 2018. The copyright holder for this preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made available under aCC-BY-NC 4.0 International license.
91 root. For each subplot we also recorded ground vegetation cover as a proxy for interspecific
92 competition, estimated into four classes, 0-25%, 25-50%, 50-75% and 75-100%.
93 We quantified spatial aggregation of E. zwageri adult trees (DBH ≥ 10 cm) using the relative
94 neighborhood density Ωx, defined as the average density of neighbours in the annulus at
95 distance x for each focal tree, standardized by mean population density (Condit et al. 2000). We
96 calculated Ωx for the 0-100m distance range in 10 m steps, and compared this metric for the
97 older and younger sub-populations, categorized as those above or below the median tree size
98 respectively.
99 For E. zwageri seedlings we used a generalized linear model (GLM) with a binomial error
100 distribution and a logit link function to test the effects of conspecific seedling density and
101 vegetation cover on the proportion of fatal herbivory damage. All analyses were conducted in
102 the R statistical computing environment (R Core Team, 2016).
103
104 RESULTS
105 We recorded 90 E. zwageri trees (DBH ≥ 1 cm) in the 4 ha plot, among which 80 were adult
106 trees (DBH ≥ 10 cm). As predicted from the reduced regeneration hypothesis the adult tree size
107 distribution had a unimodal shape peaking at the 70 cm class (median) and decreasing towards
108 smaller classes (Fig. 1a). All adult trees were aggregated at the 0-10m scale (Ωx > 1), and the
109 degree of aggregation was significantly higher in the younger sub-population, consistent with
110 the expectation of increased dispersal limitation over time (Fig 1b).
6
bioRxiv preprint doi: https://doi.org/10.1101/406025; this version posted August 31, 2018. The copyright holder for this preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made available under aCC-BY-NC 4.0 International license.
111 A total of 495 E. zwageri seedlings were recorded in 197 of the 540 5 by 5 m subplots surveyed,
112 with an overall mean density of 367 seedlings ha-1. Nearly all appeared to be less than one year
113 old and resulting from the fruiting event in January 2017. Among these, 53.3% had fatal
114 herbivory damage, already dead or highly unlikely to survive. The proportion of fatal herbivory
115 damage increased rapidly with conspecific seedling density at subplot level (p < 0.001),
116 predicted to reach 0.97 at 1 seedling m-2 (Fig. 2). Fatal herbivory was not significantly associated
117 with vegetation cover (p = 0.09).
118
119 DISCUSSION
120 The remarkably low abundance of young E. zwageri trees indicates that the population of this
121 species is not sustainable in this forest stand (Halpin & Lorimer 2017). E. zwageri adult trees
122 were aggregated at the 0-10 m spatial scale, characteristic of tropical trees (Condit et al. 2000);
123 a similar aggregation pattern at this scale was observed at genotypical level in a recent study
124 (Shao et al. 2017). However, the increased spatial aggregation of E. zwageri in younger trees
125 supports the hypothesis of an increasing seed dispersal limitation over time due to the lack of
126 seed dispersers, as shown in another Bornean forest (Harrison et al. 2013). Most strikingly, E.
127 zwageri seedlings more than one year old (height > 1 m) were almost absent, suggesting
128 complete failure of seedling establishment in recent years. This can be explained largely by the
129 high rate of fatal seedling herbivory, increasing with local aggregation (Fig. 2). These patterns
130 highlight the effect of dispersal limitation and negative density-dependent seedling survival,
131 operating together to influence species spatial distribution (Shao et al. 2017).
7
bioRxiv preprint doi: https://doi.org/10.1101/406025; this version posted August 31, 2018. The copyright holder for this preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made available under aCC-BY-NC 4.0 International license.
132 A survey of 13 other Bornean canopy species found a seedling survival rate of approximately
133 50% over ten years (Delissio et al. 2002). In contrast, we observed an overall survival rate of
134 46.7% in E. zwageri seedlings in less than one year since germination. The actual survival rate
135 may be even lower as we only accounted for dead stems with detectable remains. Using our
136 model as a conservative approximation of the density dependent annual mortality rate and
137 applying it to the surviving E. zwageri seedlings at subplot level, we predict that after six years
138 seedling density would drop to merely 3 individuals ha-1, leading to a progressively shrinking
139 demography (Fig. 1a). Seedling height may not reach 2 m in this time (Irawan 2005), and thus
140 they remain susceptible to ground level herbivory.
141 The heavy herbivory observed here was apparently associated with a single mammalian
142 herbivore, the sambar deer, based on characteristics of the stem damage and reports by local
143 rangers. Adapted to surviving shade conditions for relatively long period, E. zwageri seedlings
144 allocate a high proportion of resources towards defense in the form of concentrated tannin (CT)
145 and lignin (Kurokawa et al. 2004). This is an evolved mechanism against invertebrate herbivores
146 and pathogens, but mammalian herbivores such as sambar deer have co-evolved to produce a
147 tannin-binding protein in their saliva, and in fact show a preference towards plant species high
148 in CT and lignin (Semiadil et al. 1995).
149 It is possible that at local scale, browsing pressure from sambar deer can influence E. zwageri
150 regeneration. A generalist grazer and browser, the sambar deer appears to benefit from
151 intermediate forest disturbance, preferring forest edges and selectively logged forest due to
152 the increased availability of phytomass on the ground (Corlett 2007, Meijaard & Sheil 2007). A
153 recent study in our research area found that mammalian herbivores showed a 19% increase in
8
bioRxiv preprint doi: https://doi.org/10.1101/406025; this version posted August 31, 2018. The copyright holder for this preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made available under aCC-BY-NC 4.0 International license.
154 abundance in logged forest compared to unlogged forest and, in particular, the abundance of
155 sambar deer more than doubled (Wearn et al. 2017). There are few large predators in Borneo,
156 and predation pressure on this species here predominantly comes from hunting (Bennett et al.
157 2000). It is likely that in protected areas where hunting is prohibited the sambar deer has
158 enjoyed predator release. Our study site is in a protected area adjacent to a large buffer zone
159 consisting of selectively logged forests. The sambar deer population here therefore have
160 probably benefited from both favorable habitat and the absence of hunting. As an interesting
161 contrast, sambar deer population was low in another Sabah forest (Kabili-Sepilok) (Ross et al.
162 2013), and greater densities of young E. zwageri trees were recorded, on average 49 stems with
163 DBH 5 – 10 cm per 4-ha (Qie and Nilus, unpublished data).
164 Our results shine a spotlight on an iconic tree species E. zwageri. We show the intensified
165 interaction between this native tree and a native mammalian herbivore resulting from changes
166 in the megafauna community and forest landscape in Borneo. Anthropogenic forces on the
167 forest ecosystem in recent centuries seem to have eradicated its long-distance seed dispersers
168 and in some places favored its seedling herbivores. Where these impacts co-occur they may
169 lead to an impending regeneration failure of E. zwageri at local scale, as we are starting to see
170 in this population.
171 E. zwageri requires active conservation intervention and efforts should focus on its most
172 vulnerable seedling stage. Once survived through this recruitment bottleneck, the tall saplings
173 and trees of E. zwageri are known to be exceptionally resilient, including being drought- and
174 fire- tolerant (Delmy 2001, Van Nieuwstadt & Sheil 2005). Ex situ propagation of this species
175 has only been carried out on small scales partly due to inadequate supply of seeds and
9
bioRxiv preprint doi: https://doi.org/10.1101/406025; this version posted August 31, 2018. The copyright holder for this preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made available under aCC-BY-NC 4.0 International license.
176 seedlings (Azani et al. 2001, Irawan 2012), yet we show that a great number of E. zwageri
177 seedlings in natural forests are being lost to high mortality. Our results suggest that at local
178 density of < 0.2 m-2 E. zwageri seedlings may have optimal survival. Therefore, moderate
179 harvest of E. zwageri seeds and seedlings in natural forests (i.e. wildlings) for ex situ cultivation
180 and re-introduction once they reach sapling stage should be feasible. This approach may
181 simultaneously increase E. zwageri regeneration and population gene flow, providing a cost-
182 effective way of assisted forest regeneration.
183
184 ACKNOWLEDGEMENT
185 This study was supported by the Imperial College London MRes Tropical Forest Ecology field
186 course program. We thank the Sabah Biodiversity Centre for access license and the excellent
187 support of staff members at the Maliau Basin Studies Centre.
188
189 REFERENCES
190 ASIAN REGIONAL WORKSHOP. 1998. (Conservation & Sustainable Management of Trees, Viet Nam, August
191 1996). Eusideroxylon zwageri. The IUCN Red List of Threatened Species 1998:
192 e.T31316A9624725. http://dx.doi.org/10.2305/IUCN.UK.1998.RLTS.T31316A9624725.en.
193 Downloaded on 29 May 2018.
194 AZANI, A. M., et al. 2001. Rehabilitation of Tropical Rainforests Based on Indigenous Species for Degraded
195 Areas in Sarawak, Malaysia. In S. Kobayashi, J. W. Tunbull, T. Toma, T. Mori and N. M. N. A.
10
bioRxiv preprint doi: https://doi.org/10.1101/406025; this version posted August 31, 2018. The copyright holder for this preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made available under aCC-BY-NC 4.0 International license.
196 Majid (Eds.). Rehabilitation of Tropical Degraded Forest Ecosystems, pp. 141-148. Center for
197 International Forestry Research, Bogor, Indonesia.
198 BENNETT, E., et al. 2000. Saving Borneo's bacon: the sustainability of hunting in Sarawak and Sabah. In J.
199 Robinson and E. Bennett (Eds.). Hunting for sustainability in tropical forests, pp. 305–324.
200 Columbia University Press, New York.
201 CONDIT, R., et al. 2000. Spatial Patterns in the Distribution of Tropical Tree Species. Science 288: 1414-
202 1418.
203 CORLETT, R. T. 2007. The Impact of Hunting on the Mammalian Fauna of Tropical Asian Forests. Biotropica
204 39: 292-303.
205 CORLETT, R. T. 2010. Megafaunal extinctions and their consequences in the tropical Indo-Pacific. Terra
206 Australis 32: 117-131.
207 DELISSIO, L. J., et al. 2002. A decade of canopy-tree seedling survival and growth in two Bornean rain
208 forests: persistence and recovery from suppression. J. Trop. Ecol. 18: 645-658.
209 DELMY, A. 2001. Fire Resistance of Tree Species in Bukit Soeharto Education Forest, East Kalimantan,
210 Indonesia. In S. Kobayashi, J. W. Tunbull, T. Toma, T. Mori and N. M. N. A. Majid (Eds.).
211 Rehabilitation of Tropical Degraded Forest Ecosystems, pp. 27-34. Center for International
212 Forestry Research, Bogor, Indonesia.
213 FRANCO, F., et al. 2014. Terras (Eusideroxylon zwageri Teijsm. & Binn.), a Cultural Keystone Species of the
214 Berawan People of Sarawak, Malaysia. Pertanika Journal of Social Sciences and Humanities 22:
215 891-902.
216 GALETTI, M., et al. 2017. Ecological and evolutionary legacy of megafauna extinctions. Biol. Rev.
217 HALPIN, C., AND C. LORIMER. 2017. A Demographic Approach to Evaluating Tree Population Sustainability.
218 Forests 8: 46.
219 HANCE, J. 2015. Officials: Sumatran rhino is extinct in the wild in Sabah. Mongabay.
11
bioRxiv preprint doi: https://doi.org/10.1101/406025; this version posted August 31, 2018. The copyright holder for this preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made available under aCC-BY-NC 4.0 International license.
220 HARRISON, R. D., et al. 2016. Impacts of hunting on tropical forests in Southeast Asia. Conserv. Biol. 30:
221 972-981.
222 HARRISON, R. D., et al. 2013. Consequences of defaunation for a tropical tree community. Ecol. Lett. 16:
223 687–694.
224 IRAWAN, B. 2005. Ironwood (Eusideroxylon zwageri Teijsm. & Binn.) and its varieties in Jambi, Indonesia.
225 Cuvillier Verlag.
226 IRAWAN, B. 2012. Growth Performance of One Year Old Seedlings of Ironwood (Eusideroxylon zwageri
227 Teijsm. & Binn.) Varieties. Jurnal Manajemen Hutan Tropika 18.
228 KOSTERMANS, A. J. G. H., et al. 1994. Eusideroxylon Teisjm. and Binnend. In S. l. and L. R.H.M.M. (Eds.).
229 Plant Resources of South-East Asia 5 (1). Timber trees: major commercial timbers. PROSEA,
230 Borgor, Indonesia.
231 KUROKAWA, H., et al. 2004. Allocation to defense or growth in dipterocarp forest seedlings in Borneo.
232 Oecologia 140: 261-270.
233 MEIJAARD, E., AND D. SHEIL. 2007. The persistence and conservation of Borneo’s mammals in lowland rain
234 forests managed for timber: observations, overviews and opportunities. Ecol. Res. 23: 21.
235 PELUSO, N. L. 1992. The Ironwood Problem: (Mis)Management and Development of an Extractive
236 Rainforest Product. Conserv. Biol. 6: 210-219.
237 PHILLIPPS, Q., AND K. PHILLIPPS. 2016. Phillipps' Field Guide to the Mammals of Borneo and Their Ecology:
238 Sabah, Sarawak, Brunei, and Kalimantan. Princeton University Press.
239 R CORE TEAM. 2016. R: A language and environment for statistical computing. R Foundation for Statistical
240 Computing, Vienna, Austria https://www.R-project.org/
241 ROSS, J., et al. 2013. Activity patterns and temporal avoidance by prey in response to Sunda clouded
242 leopard predation risk. Journal of Zoology 290: 96-106.
12
bioRxiv preprint doi: https://doi.org/10.1101/406025; this version posted August 31, 2018. The copyright holder for this preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made available under aCC-BY-NC 4.0 International license.
243 SEMIADIL, G., et al. 1995. Dietary preferences of sambar (Cervus unicolor) and red deer (Cervus elaphus)
244 offered browse, forage legume and grass species. The Journal of Agricultural Science 125: 99-
245 107.
246 SHAO, X., et al. 2017. Intra‐specific relatedness, spatial clustering and reduced demographic
247 performance in tropical rainforest trees. Ecol. Lett.
248 VAN NIEUWSTADT, M. G. L., AND D. SHEIL. 2005. Drought, fire and tree survival in a Borneo rain forest, East
249 Kalimantan, Indonesia. J. Ecol. 93: 191-201.
250 WANG, B. C., AND T. B. SMITH. 2002. Closing the seed dispersal loop. Trends Ecol. Evol. 17: 379-386.
251 WEARN, O. R., et al. 2017. Mammalian species abundance across a gradient of tropical land-use intensity:
252 A hierarchical multi-species modelling approach. Biol. Conserv. 212: 162-171.
253
254
13
bioRxiv preprint doi: https://doi.org/10.1101/406025; this version posted August 31, 2018. The copyright holder for this preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made available under aCC-BY-NC 4.0 International license.
a
b
255
256 FIGURE 1. Demographic and spatial distribution of E. zwageri adult trees (DBH ≥ 10 cm) in the
257 4-ha plot. a) Population size distribution with a unimodal shape and a mode at the 70 cm class.
258 b) Spatial aggregation measured as the mean relative neighborhood density in 10m annuli each
259 focal tree, standardized by mean population density. The population was divided into the older
14
bioRxiv preprint doi: https://doi.org/10.1101/406025; this version posted August 31, 2018. The copyright holder for this preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made available under aCC-BY-NC 4.0 International license.
260 (DBH ≥70cm) and younger (DBH < 70 cm) sub-populations. Bars represent 95% bootstrapped
261 confidence intervals.
262
263 FIGURE 2. Relationship between the rate of fatal herbivory of E. zwageri seedlings and
264 conspecific seedling density at 5x5m subplot level. Line was fitted using a GLM model. Shading
265 areas represent confidence intervals.
266
15