Bark and ambrosia beetles on native and transplanted dead wood along an altitudinal tropical forest gradient

Global warming is expected to shift the distribution ranges of many species towards higher latitudes and altitudes. This will rewire plant‐herbivore food webs as new combinations of herbivore species encounter novel host plants. We investigated the effects of a simulated altitudinal shift in an herbivore‐host community of bark and ambrosia beetles on fig trees in a tropical mountain forest on Mt. Wilhelm, Papua New Guinea. We reared beetles from wood baits cut from five local Ficus species in their natural altitudinal ranges, between 200 and 1700 m a. s. l. Further, we transplanted baits from three of these tree species 500 m below the limits of their natural distribution range, to simulate a mean temperature increase of 2.7°C. Beetle species richness declined, and their species composition changed with increasing elevation. Furthermore, while altitude explained a large proportion of variance in beetle composition, host tree species was more important for bark beetles than ambrosia beetles. Beetle communities that assembled on the transplanted baits were similar in diversity and host specificity to those on the fig trees native to the same elevation, but also contained a number of unique species. Overall, these results indicate that saproxylic beetles in this tropical forest are highly resilient and flexible to the potential effects of climate change.

relationships and food web structure (Dale et al., 2001;Hamann et al., 2021;Vergés et al., 2014).The impact of climate change on insects in the tropics is being intensely discussed as some studies report insect declines, possibly in response to the changing climate (Janzen & Hallwachs, 2021), while others have not found any signs of such decline (Basset et al., 2023).Altitudinal gradients have proved to be excellent models for the study of global warming as their steep temperature gradients across short geographic distances make them amenable to experiments (McCain & Garfinkel, 2021;Rasmann et al., 2014).
In this study, we investigated the effect of temperature increase on an herbivore-host plant system in a tropical forest using an altitudinal rainforest gradient.We conducted altitudinal transplant experiments on host trees beyond their native range and studied the response of their saproxylic (i.e., dead wood dependent; see Speight, 1989) beetles (Coleoptera).Field transplant experiments have been recognized as a powerful tool to assess the impact of climate change on communities (Nooten & Hughes, 2017) and previous studies have provided evidence for significant changes in transplanted butterflies (Merill et al., 2008), spiders (Krehenwinkel & Tautz, 2013), bees (McCabe et al., 2022), and epiphytes (Nadkarni & Solano, 2002).In cases where host plants were translocated, preceding experiments have shown shifts in their associated arthropod communities (Andrew & Hughes, 2007;Heimonen et al., 2014;Moir et al., 2011;Nooten et al., 2014) and changes in herbivore pressure (Garibaldi et al., 2011).However, transplant studies on herbivore-host systems in tropical forests are scarce.
Saproxylic insects play an important role in wood decomposition and therefore in the carbon cycle of forest ecosystems (Seibold et al., 2021).Among them, the most speciose group are saproxylic beetles (Lassauce et al., 2011;Müller et al., 2008).We focused on two groups of early colonizers of fresh dead wood: Bark beetles (Curculionidae: Scolytinae) and ambrosia beetles (Curculionidae: Scolytinae and Platypodinae).While bark beetle larvae feed directly on the phloem and cambium tissue of the host tree, ambrosia beetles excavate galleries into the sapwood and inoculate them with symbiotic fungi that serve as their food source (Hulcr et al., 2007).
Bark beetles are generally more host specific than ambrosia beetles (Novotny et al., 2010;Watanabe et al., 2014).This is particularly true, when they infest living trees and have to overcome the defenses of their hosts (Wermelinger, 2004).Both groups were selected as focus of this study for their diversity and ecological impact (Hulcr & Dunn, 2011) as well as their distribution along a wide altitudinal range.
We selected fig trees (Ficus, Moraceae) to investigate as hosts for the beetles.With over 750 species worldwide Ficus is a large pantropical plant genus and often a keystone species in tropical forest habitats (Harrison, 2005;Harrison et al., 2012).New Guinea is a diversity hotspot for Ficus, with numerous species coexisting in a single rainforest ecosystem.

Fig trees in New
Guinea also support rich insect herbivore communities (Basset et al., 1997).
Following previous studies on local faunal groups and other studies on the altitudinal distribution of saproxylic beetles (Rubin-Aguirre et al., 2015;Wu et al., 2008), we expected decreasing species richness with increasing altitude, as well as a change in community species composition.Further, we hypothesize that beetle communities on transplanted wood baits will be dominated by generalist ambrosia beetles and less diverse compared to those within their native altitudinal ranges.

| Study area
The study was conducted along an elevational gradient on Mt.
Wilhelm (4.509 m a. s. l.) in the Central Range of Papua New Guinea (PNG).The complete rainforest gradient extends from the lowland floodplains of the Ramu river (200 m a. s. l., 5°44′ S, 145°20′ E) to the timberline (3700 m a. s. l., 5°47′ S, 145°03′ E), with research sites spaced at 500 m elevational increments (see Figure 1a,b).Average annual precipitation is 3288 mm in the lowlands at 200-700 m a. s. l., rising above 4000 mm above the distinct condensation zone between 2500 and 2700 m a. s. l.Mean annual temperature (measured for 1 year by our data loggers) decreases from 27.4°C at the 200 m a. s.
l. site at a constant rate of 0.54°C per 100 m in elevation (see Sam et al., 2020).We used the four lowest elevational sites (200, 700, 1200, and 1700 m a. s. l.) comprising the main elevational range of Ficus for our study.

| Study design
There are 73 Ficus species distributed along the Mt.Wilhelm transect, with >30 species coexisting at each elevation between 200 and 1700 m a. s. l. (Segar et al., 2016).We selected five species for use as wood baits: Ficus hispidoides var.succosa, F. morobensis, F. nodosa, F. pungens, and F. umbrae.These species were chosen, since they were both easy to identify and sufficiently abundant at their respective sites.Species identification was confirmed by George Weiblen (University of Minnesota).The focal tree species occurred naturally at 1-4 sites within the 200-1700 m a. s. l. range.There were 12 species-site combinations in total (Table 1).We selected two mature trees of similar size per Ficus species at each of its sites, 24 trees in total, felled them and cut three wood baits out of each tree.
Further, for three fig species: Ficus morobensis, F. hispidoides, and F. umbrae, we felled additional trees and used them to create and transplant three baits per species to a new altitude, 500 m below the limits of their natural range in order to simulate novel herbivore-host pairs in response to climate change (Table 1).
Each bait was composed of the trunk section, branches (diameter 3-10 cm), and twigs (diameter <3 cm) tied together by wire, weighing approximately 20 kg (20.45 ± 0.89).Baits were suspended from a wooden construction, evenly spaced, and exposed for 3 weeks to allow infestation by saproxylic beetles (see Figure 1c).Subsequently, they were encased in individual rearing bags, consisting of 1 mm wire mesh covered with black cloth and fitted with a transparent plastic container filled with 70% ethanol (see Figure 1d).Beetles were sampled weekly for a period of 6 weeks.Specimens were sorted to morphospecies using the reference collection at the New Guinea Binatang Research Centre and identified to species wherever possible, using available literature (Hulcr & Cognato, 2013).Within the tribe Cryphalini, only clearly distinguishable morphospecies were considered, while the majority of morphologically uniform specimens were excluded from the analysis since the taxonomic information on this group is insufficient for correct determination.Furthermore, specimens were assigned to the guilds of ambrosia beetles or bark beetles, depending on their larval feeding habits.

| Data analysis
All analyses, unless specified otherwise, were performed using R 4.2.0 (R Core Team, 2022) and conducted on bark and ambrosia beetles together (hereafter referred to as "all beetles"), along with individual analyses for each group.The sample-based rarefaction and extrapolation (R/E) curves were created using the iNEXT package (Hsieh et al., 2020).Extrapolations were calculated to double the total sample sizes that were taken from the respective altitude level.
Accumulation curves were computed for species richness (Hill number: q = 0), which weighs all species evenly regardless of their relative abundance (Chao et al., 2014).Non-overlapping confidence intervals denote a significant difference at a level of 5% (Chao & Jost, 2012).
Additionally, indicator species were identified for each altitude as well as each tree species per altitude as described by Dufrêne and

TA B L E 1
The distribution of wood baits of the five Ficus tree species along the altitudinal gradient (X, natural distribution range of tree species; t, transplanted outside the natural range).
Bipartite food web graphs were created for the herbivore-host interactions at each altitude using the bipartite package (Dormann et al., 2008).Species specialization indices d′ (Blüthgen et al., 2006) were calculated for tree species in each network.This index is robust regarding network size and therefore suitable for the comparison of the tree species' position in the saprotrophic network between altitudes.The value of the index ranges from 0 (most generalized) to 1 (most specialized).The Shannon diversity of interactions was calculated for each tree species per network.Every observed index value was corrected by testing it against null-model generated values with 1000 repetitions.
To assess the difference in beetle communities along the altitudinal gradient a distance matrix with the Bray-Curtis dissimilarity index was computed after species abundance data was log-transformed.
The dissimilarity was partitioned into two components using the betapart package (Baselga et al., 2022): Balanced variation in abundance (d BC-bal ), whereby individuals of some species in one site are substituted by the same number of individuals of different species in another site, and abundance gradients (d BC-gra ), whereby some individuals are lost form one site to another (Baselga, 2013) To investigate the influence of wood bait properties on differences in species composition, the Bray-Curtis distance matrix was used in a PERMANOVA performed with the adonis2 function of the vegan package (Oksanen et al., 2022).The explanatory variables in this analysis were altitude (as a second-order polynomial), tree species and transplant status of the baits.Their significance was derived by a permutation-based test with 9999 permutations.Finally, the same dissimilarity matrix was used to perform non-metric multidimensional scaling (NMDS) for a visual representation of the differences in species composition between wood baits.One bait with insufficient beetle samples was excluded from this and the following analysis.A canonical correspondence analysis (CCA) with logtransformed species abundances as response variables and altitude and tree species as explanatory variables was performed and the significance of ordination axes was assessed with a Monte Carlo Tests with 999 permutations.The CCA was performed using CANOCO 5.12 (Ter Braak & Šmilauer, 2012).

| RE SULTS
In total we recorded 32,845 beetle specimens belonging to 86 species (Table S1).Of these, 3963 specimens belonging to 57 species were identified as ambrosia beetles and 28,882 specimens belonging to 29 species as bark beetles.Notably, approximately 90% (26,068) of bark beetles were represented by a single species, Ficicis despectus (Walker, 1859).
The rarefaction/extrapolation curves show a mid-elevation peak of beetle species richness as the highest number of species was sampled at 1200 m, closely followed by 700 m (Figure 2).When only accounting for bark beetles, the 1200 m elevation also yielded the most bark beetle species, while ambrosia beetles reached their maximum at 700 m.However, in all cases only 1700 m showed a significantly lower species richness than the other altitudes.Sample coverage was high for the three lowest altitudes (0.90 for 200 m; 0.92 for 700 m; 0.86 for 1200 m) but notably lower for 1700 m (0.74) with no significant difference occurring (Figure S1).A detailed look at the beetle species occurring on the transplanted baits shows that a considerable proportion of them are unique to the transplants (see Figure S2).The proportion of unique species is slightly higher for ambrosia beetles compared to bark beetles and far higher on F. hispidoides and F. morobensis at 200 m than on F. umbrae at 1200 m.
The 700 m elevation hosts the highest number of ambrosia and bark beetle indicator species (Figure S3).When looking at the number of indicator species at the tree species level per altitude (Table S2), it shows that F. morobensis hosts the same number of ambrosia bee- see Figure S4).The species specialization index d′ shows similar values for both the transplanted and native Ficus species at 200 m.The d′ value of F. umbrae transplants at 1200 m is lower than that of native Ficus, but it is almost identical to the species' value within its native range at 1700 m.On the other hand, for the transplanted species at 200 m, the Shannon diversity of interactions was higher for F. morobensis and comparable for F. hispidoides to that of the native species.At 1200 m the diversity value was comparable to the native species.
The partitioning of the Bray-Curtis distance matrix shows that the d BC-bal generally accounted for a larger part of the dissimilarity between samples than d BC-gra (Figure 4).This denotes that differences in community composition between wood baits are predominantly caused by a balanced variation of species rather than a loss/ increase of abundance.It is also apparent that dissimilarity between samples of the same tree species increases with altitudinal distance in most cases.Furthermore, the dissimilarity patterns of transplants are not markedly different from those of native baits.
The analysis of the distance matrix through PERMANOVA revealed an influence of all three explanatory variables on changes in community composition (Table 2): Altitude was highly significant for all beetles (F = 8.5030; p < .001),ambrosia beetles (F = 8.7097; p < .001)and bark beetles (F = 4.8654; p < .001).The tree species identity was highly significant for all beetles (F = 2.2715; p < .001) and bark beetles (F = 3.4989; p < .001)but not for ambrosia beetles (F = 1.4146; p = .059).It is notable that altitude explained the highest amount of variance in all beetles and ambrosia beetles, but not in bark beetles where tree species explained the highest amount.
By contrast, transplant status explained a much lower amount of variance but was still significant for all beetles (F = 3.0088; p < .01) and ambrosia beetles (F = 3.2799; p < .01)but not for bark beetles (F = 1.9328; p = .055).
The NMDS plot of the Bray-Curtis distance matrix shows a high similarity of community composition of beetle samples from the same altitude with the 200 and 700 m separated from the 1200 and 1700 m along the first axis of the plot (Figure 5a).This pattern also applies to the communities from transplanted baits, which are considerably more similar to those from wood baits from the same altitude than to congeneric baits from the altitude of their origin.The same patterns can be seen in the ambrosia beetle communities, even though the separation by altitude is less pronounced (Figure 5b).
The bark beetle communities show only a slight separation between wood baits from different altitudes (Figure 5c).Instead, there is a stronger separation by tree species as the samples from F. pungens host a more distinct community than the other species.This is further supported by the position of the tree species' centroids in the CCA analysis (Figure 5d).The communities on the F. umbrae baits are separated from those of the other species along the first axis which reflects altitude.Ficus punges separates from the other species along the second axis, and this seems to be mainly driven by a small number of bark beetle species on this tree species.

| DISCUSS ION
Altitudinal trends in rainforest species diversity typically vary between a monotonous decrease with increasing altitude, and a F I G U R E 3 Food webs displaying the tropic interactions between beetle species (orange: ambrosia beetles; blue: bark beetles) and tree species (red colored boxes indicate transplants) at individual altitudes.The box width indicates the abundance of beetle species, and the thickness of the connection (gray) indicates the relative weight of the interaction (abundance of the interaction).Species specialization index (d′) and Shannon diversity of interactions (H) are also displayed for all tree species including a denotation of significant difference from the null-model generated values (***p < .001;**p < .01;*p < .05).For information on the identity of beetle species see Figure S4.
mid-elevation maximum.At Mt. Wilhelm, birds, fruit flies, and butterflies exhibited a monotonous decrease, while ants, geometrid moths, and ferns showed a mid-elevation maximum (Beck et al., 2017;Colwell et al., 2016;Finnie et al., 2021).These patterns are driven by abiotic conditions, such as temperature, biotic conditions, such as predation pressure, and the resource availability (Beck et al., 2017;Colwell et al., 2016;Finnie et al., 2021).In our Ficus-beetle system, the mid-elevation maximum in bark beetle diversity may be driven by a mid-elevation maximum in resources, represented by Ficus trees (Segar et al., 2016).To uncover the drivers behind the overall trends of bark and ambrosia beetles on the entire vegetation further research is required.
The results of the PERMANOVA confirm our expectations that dissimilarity among ambrosia beetles is largely explained by altitudinal bait position, while host tree species has the highest explanatory value in bark beetles.This was to be expected since ambrosia beetles are regarded as far broader generalists that are mainly reliant on their symbiotic fungi, as opposed to bark beetles, who feed directly on the woody tissues of their host trees (Hulcr et al., 2007).Aside from the fact that ambrosia beetles feed on their host trees "indirectly" they also tend to colonize them at a slightly later point than bark beetles, namely when the trees are freshly dead or dying (Hulcr et al., 2007).For this reason, they do not have to contend as much with the hosts chemical defenses as the latter which are heavily TA B L E 2 Results of the permutational multivariate analysis of variance (PERMANOVA) for the effects of wood bait properties (altitude, tree species, and transplant status) on between-bait Bray-Curtis distance of beetle communities.affected by these substances (Byers, 1995).Additionally, their fungal symbionts are also known to often be generalists in terms of their plant host species (Veselská et al., 2019).On a similar note, Novotny et al. ( 2010) discovered a lower degree of specialization in fungalchewing beetles compared to phloem-feeding species in a lowland rainforest in PNG.
The bark and ambrosia beetle communities that formed on the transplanted wood baits showed little overall difference from the communities on the native Ficus hosts at the same elevation, and/ or the communities on the conspecific hosts at an elevation 500 m higher.Second, the NMDS plot shows that the beetle community composition on transplanted wood baits is far more similar to those of baits at the same altitude than those at their altitude of origin.
These results give further credence to the notion that specialization in saproxylic beetles regarding their choice of host trees is generally lower in tropical habitats (Beaver, 1979;Tavakilian et al., 1997).The reason behind this trend could likely be the high tree species density in tropical forests, which would discourage narrow specialization (Basset, 1992), especially regarding an ephemeral resource like dead wood.On the other hand, Novotny et al. (2002), employing a phylogenetic instead of a taxonomic approach, postulates that the low host specificity of tropical leaf-chewing insects is rooted in the fact that tropical flora is dominated by few large genera in which monophagy is rare.The same might also be the case for tropical saproxylic arthropods (see also Milberg et al., 2014).The notion of high generality is further supported by the overall low d′ values of tree species recorded in this study.Regarding the host tree, Basset and Novotny (1999) similarly discovered a high faunal overlap in leafchewing and sap-sucking insects on Ficus species in PNG.
The transplantation did, however, have a significant impact on the beetle communities, as revealed by the PERMANOVA analysis, particularly on the ambrosia beetles.The cause for this result is most certainly the number of beetle species that occur only on the transplants and not on the native baits, which is also higher in the case of ambrosia beetles.A likely explanation for this phenomenon is that the suitability of fig tree wood as a resource for the beetles varies to bark beetles at all stages of their life cycle (Raffa et al., 2005).On the other hand, wood density has been shown to decrease with elevation in a wide variety of tropical tree species (Chave et al., 2006).
Since a major defensive trait of the xylem stems from the fact that it is composed of digestion-resistant polymers (Hulcr & Dunn, 2011), ambrosia beetles and their symbiotic fungi might be able to colonize the lower-density tissue of wood transplants more easily.Similarly, Cerambycidae, another family of early-colonizing saproxylic beetles, have been found to prefer wood of lower density in tropical forests (Lanuza-Garay & Barrios, 2018;Torres et al., 2023).
Our results partly diverge from those of previous experiments on transplanted host plants and their associated insects: Significant differences in the community composition of herbivores have been detected not only between transplants and their source population, but also between transplants and congeneric plant species (Nooten et al., 2014) or conspecific genotypes (Heimonen et al., 2014) at the same sites.However, it is important to note that these transplantation experiments took place across a larger scale latitudinal gradient, and examined a wider variety of leaf feeding taxa which are likely to be more host specific than saproxylic beetles.Conversely, Moir et al. (2011) found that insect communities on plants that were translocated along a latitudinal gradient were more similar to those on closely related species in their surroundings, which is more in line with the outcome of our study.
In conclusion, the results of our study suggest a positive outlook on the future of the Ficus-ambrosia-bark-beetle system on the slopes of Mt.Wilhelm.Beetle species of both trophic groups seem to be readily able to utilize congeneric host tree species that they encounter.
Together with the vast altitudinal distribution of Ficus, which greatly exceeds the range covered in this experiment, large scale extinctions among these saproxylic groups due to climate change induced range shifts seem unlikely.However, this does not rule out the possibility of substantial impacts on competition, predation and other processes in these assemblages which were not investigated in this study.
Furthermore, it has been observed that saproxylic beetle communities exhibit a greater turnover along larger geographic and phylogenetic distances (Seibold et al., 2022).Therefore, it is likely that beetles that are specialized on tree genera with a considerably narrower altitudinal range than Ficus might be significantly more negatively affected by such shifts than the groups investigated in this study.

F
Location of the study area in the Madang Province of Papua New Guinea (PNG) (a) and the four study sites along the Mt.Wilhelm elevational gradient (b).Photos of the exposed wood baits (c) and the baits enclosed in rearing bags (d).
tle species, when transplanted to 200 m as it does within its native range at 1200 m.One of these species, Xylosandrus difficilis, occurs exclusively on the transplanted baits.F. hispidoides hosts an ambrosia beetle species at its transplant at 200 m and a bark beetle species at 700 m, but neither within its native range at 1200 m.Ficus pungens hosts indicator bark beetle species at every level of its distribution range with the most being found at 700 m.Ficus umbrae did not host any indicator species when transplanted to 1200 m, but hosted a bark beetle species within its native range at 1700 m.Ficus nodosa is the only tree species that does not host any indicator species at all.The food web graphs show fewer trophic connections per tree species at the highest altitude of 1700 m, when compared to the three altitudes below (Figure3; for information on the identity of beetle species F I G U R E 2 Rarefaction/extrapolation curves for the number of species of all beetle species (a), ambrosia beetles (b), and bark beetles at all altitudes (c).

F
Heatmap-plot of the dissimilarity partitioning of the Bray-Curtis distance matrix of beetle communities from all wood baits into the components of balanced variation and abundance gradient (HIS = F. hispidoides; MOR = F. morobensis; NOD = F. nodosa; PUN = F. pungens; UMB = F. umbrae; red framing indicating transplants).

F
I G U R E 5 NMDS plot of communities on wood baits (red circles indicating transplants) for all beetle species (a), ambrosia beetle species (b), and bark beetle species (c); CCA plot of beetle species (orange: ambrosia beetles; blue: bark beetles; for full beetle names see TableS1) in relation to tree species (HIS = F. hispidoides; MOR = F. morobensis; NOD = F. nodosa; PUN = F. pungens; UMB = F. umbrae) and altitude (d).at different altitudes.Volf et al. (2020) discovered increased levels of chemical defenses in the leaves of tropical Ficus trees at higher elevations.The same might be true for lignin and other defensive compounds in the woody tissues, like monoterpenes which are toxic