Introduction
Most plant species rely on dispersal by animals to escape density-dependent mortality, maintain genetic diversity, and colonize new sites (Traveset et al. Reference Traveset, Heleno, Nogales and Gallagher2014). Isolation of islands has resulted in ecosystems with lower species richness, higher endemicity, and fewer interactions between species than on continents (Fernández-Palacios et al. Reference Fernández-Palacios, Kreft, Irl, Norder, Ah-Peng, Borges, Burns, de Nascimento, Meyer, Montes and Drake2021). Mammals and birds are the main plant dispersers on continents (Ridley Reference Ridley1930). However, as the isolation of islands increases, mammals are rarer or even absent. Indeed, non-volant mammals are only able to cross tens to hundreds of kilometers of water depending on their size, and only volant mammals (i.e., bats) can colonize islands located thousands of kilometers off the coast (Carlquist Reference Carlquist1965, Whittaker & Fernández Palacios Reference Whittaker and Fernandez-Palacios2007). Therefore, plant species on islands often rely on a single or a few animal species for their dispersal, making dispersal highly vulnerable to extinctions (Schleuning et al. Reference Schleuning, Böhning-Gaese, Dehling and Burns2014). Understanding dispersal networks and identifying key dispersers is critical to conserving dispersal functions and biodiversity on islands (McConkey & Drake Reference McConkey and Drake2015, Nogales et al. Reference Nogales, McConkey, Carlo, Wotton, Bellingham, Traveset, González-Castro, Heleno, Watanabe, Ando, Rogers, Heinen and Drake2024).
Islands also experience higher extinction rates than continents (Fernández-Palacios et al. Reference Fernández-Palacios, Kreft, Irl, Norder, Ah-Peng, Borges, Burns, de Nascimento, Meyer, Montes and Drake2021). Extinctions particularly affect island birds, especially large ones (Fromm & Meiri Reference Fromm and Meiri2021), that can potentially disperse large seeds (McFadden et al. Reference McFadden, Fritz, Zimmermann, Pellissier, Kissling, Tobias, Schleuning and Graham2022). Loss of dispersal functions on islands due to the extinction of large native frugivores can entail changes in plant abundance, frequency of occurrence, succession trajectories, and even lead to plant species extinction (Albert et al. Reference Albert, Flores and Strasberg2020, Meehan et al. Reference Meehan, McConkey and Drake2002). Increased fruit or seed size and loss of dispersal ability have also been suggested as another characteristic of island plant species (Carlquist, Reference Carlquist1965). For instance, Carlquist (Reference Carlquist1965) suggested that in New Caledonia fruits from plants of the genus Zygogynum might have lost their dispersal ability because they became too large and cannot be swallowed by the native fauna anymore and now only rely on gravity to disperse.
Columbidae, which includes pigeons and doves, is one of the most threatened bird families (Walker Reference Walker2007) and one of the most important bird families for plant dispersal (Ridley Reference Ridley1930, Whittaker & Jones Reference Whittaker and Jones1994). Imperial pigeons (genus Ducula) are key dispersers in the Pacific islands where the dispersal functions of many plant species probably rely on one or a few of these species (Cibois et al. Reference Cibois, Thibault, Bonillo, Filardi and Pasquet2017). The key roles of extant and extinct Ducula species in the dispersal of plants have been particularly well studied in Tonga (e.g., McConkey & Drake Reference McConkey and Drake2002, McConkey et al. Reference McConkey, Meehan and Drake2004, Meehan et al. Reference Meehan, McConkey and Drake2002). In this archipelago, extinctions of the largest Ducula species have likely left many of the plant species without any birds able to disperse their seeds.
New Caledonia hosts one of the largest extant imperial pigeons, Ducula goliath Gray, 1859, locally known as Notou. A single published study has explored the key role of D. goliath in the dispersal of rainforest trees in New Caledonia (Carpenter et al. Reference Carpenter, Read and Jaffré2003). The authors suggest that the dispersal of trees with diaspores wider than 20 mm mainly relies on D. goliath and that trees with diaspores of this size represent about a quarter of the endozoochorous species. Nonetheless, this study suffers several shortcomings limiting the significance of its outcomes. The sample (123 tree species) represented a small portion of the over 1,000 native tree species known in New Caledonia (Birnbaum et al. Reference Birnbaum, Ibanez, Blanchard, Justeau-Allaire, Hequet, Eltabet, Vieilledent, Barbier, Barrière and Bruy2023) and was only representative of particular tree communities located on ultramafic soils which represent a third of the main island (Isnard et al. Reference Isnard, L’huillier, Rigault and Jaffré2016). More importantly, the lower threshold of 20 mm above which D. goliath was assumed to be the main disperser did not rely on observations of consumptions by D. goliath and other birds in New Caledonia. Furthermore, Carpenter et al. Reference Carpenter, Read and Jaffré2003 did not set an upper threshold above which the diaspores could not be swallowed by D. goliath.
Here, we revised the role of D. goliath as a disperser for 460 native tree species representative of rainforest communities on both ultramafic and non-ultramafic soils. We set up lower and upper thresholds above and below which D. goliath would be the main bird disperser based on consumption observations gathered from multiple sources. We used fruit and seed morphology to estimate the proportion of native tree species relying on internal transportation by animals to disperse, and among these species, the proportion of species that would mainly rely on D. goliath. We also tested whether the maximum size of the diaspore a bird can disperse was associated with bird mass and beak width. We then discuss the potential alternative to D. goliath in the New Caledonian fauna for the dispersal of species with large fruits or seeds.
Materials and methods
Study site
New Caledonia is an archipelago located just above the tropic of Capricorn in the southwest Pacific ( Figure 1A). The main island of New Caledonia, the Grande Terre, is ∼16,650 km2 large and located ∼1,200 km east of Australia. About 30% of the island is covered by rainforests which are highly fragmented at lower elevations (Birnbaum et al. Reference Birnbaum, Ibanez, Blanchard, Justeau-Allaire, Hequet, Eltabet, Vieilledent, Barbier, Barrière and Bruy2023, Jaffré et al. Reference Jaffré, Bouchet and Veillon1998). New Caledonia is one of the first tropical biodiversity hotspots identified as a priority for conservation on a global scale (Myers Reference Myers1988). This archipelago exhibits the highest density of endemic plants in the world (Kier et al. Reference Kier, Kreft, Lee, Jetz, Ibisch, Nowicki, Mutke and Bearthlott2009). Its flora encompasses over 3,000 native vascular plant species and ∼75% endemics (Morat et al. Reference Morat, Jaffré, Tronchet, Munzinger, Pillon, Veillon, Chalopin, Birnbaum, Rigault, Dagostini, Tinel and Lowry2012). The avifauna encompasses 73 terrestrial native bird species (22 endemics) among which six belong to the Columbidae (two endemics) (Barré & Dutson Reference Barré and Dutson2000).
(A) Location of New Caledonia (in black) in the southwest Pacific. (B) Location of the studied sites on the main island of New Caledonia (Grande Terre).
Here, we studied a sample of 460 native tree species (76 families) representing ∼14% of the native vascular flora (Morat et al. Reference Morat, Jaffré, Tronchet, Munzinger, Pillon, Veillon, Chalopin, Birnbaum, Rigault, Dagostini, Tinel and Lowry2012) and ∼43% of the native tree species (Birnbaum et al. Reference Birnbaum, Ibanez, Blanchard, Justeau-Allaire, Hequet, Eltabet, Vieilledent, Barbier, Barrière and Bruy2023). These tree species occurred at least once in a network of twenty-one 1-ha forest plots located across the Grande Terre in which all trees with a diameter at breast height ≥ 10 cm were inventoried (Ibanez et al. Reference Ibanez, Blanchard, Hequet, Keppel, Laidlaw, Pouteau, Vandrot and Birnbaum2018). These plots were mostly located in the north of the Grande Terre (only one plot was located in the south) between 25 and 1,075 m elevation with six plots on ultramafic soils and 15 on non-ultramafic soils (Figure 1B).
Diaspore, dispersal syndrome, and limiting dimension
For each studied tree species, we compiled the type of fruit, whether or not the fruits are fleshy or fleshy appendages are present (e.g., aril around the seeds), whether or not flight appendages are present (e.g., wings on the seeds), and the size of the three dimensions (length, width, and thickness) of the fruits and seeds. Most data were compiled from the Flora of New Caledonia and published species descriptions. When these data were missing in the literature, we used photos on the Endemia website (www.endemia.nc) to determine the type of fruits and the presence of appendages. We also measured the dimensions of fruits and seeds on herbarium specimens located in the herbariums of Montpellier (MPU) and New Caledonia (NOU). Measurements were directly made on specimens or remotely using specimens available online (http://publish.plantnet-project.org/project/nou) and the ImageJ software.
We then identified which of the entire fruit (including single, multiple fruits developed from a single flower, or clusters of fruits developed from multiple flowers, i.e., infructescences) or the seed (and its appendage if present) was the diaspore. If the fruit was indehiscent (i.e., not splitting open), the fruit was considered as the diaspore, while if the fruit was dehiscent, the seed was considered as the diaspore. Dehiscence (i.e., the mode of opening) was not always straightforward to determine. Some fruits, such as those of Montrouziera species (Clusiaceae), do not exhibit dehiscence lines along which the fruit would split open. Instead, the outer layers of the fruit crack randomly when ripe and expose seeds surrounded by a viscous pulp attractive to dispersers (Figure 2D). In this particular case, the diaspore is the seed associated with its appendage, not the fruit.
Example of diaspores and associated dispersal syndromes. (A) Samara (winged dry fruit) dispersed by wind (Malvaceae, Maxwellia lepidota Baill.), (B) dehiscent fruit with coloured seeds dispersed by internal animal transportation (Rhamnaceae, Jaffrea xerocarpa (Baill.) H.C. Hopkins & Pillon), (C) dehiscent fruit with fleshy appendage around the seeds (aril) dispersed by internal animal transportation (Sapindaceae, Lepidocupania lepidota (Radlk.) Buerki, Callm., Munzinger & Lowry), (D) fruit without dehiscence lines but the outer layers of the fruit crack randomly when ripe and expose seeds surrounded by a viscous pulp and dispersed by internal animal transportation (Clusiaceae, Montrouziera sphaeroidea Pancher ex Planch. & Triana with Rhacodactylus auriculatus Bavay, 1869), (E), berry (fleshy fruit) of Diospyros macrocarpa Hiern (Ebenaceae), the widest fruit that D. goliath is known to be able to swallow so far, (F) berry (fleshy fruit) of Syzygium neolaurifolium N.Snow & Byng (Myrtaceae), one of the biggest fruit, that cannot be swallowed by the Ducula goliath, (G) huge fruit of Planchonella thiensis Aubrév. (Sapotaceae), partially eaten by flying foxes (Pteropus spp.), (H) D. goliath eating Pandanus fruits. Photographs: (A-D, F) © Jérôme Munzinger, (E) © Vanessa Hequet, (G) © Christopher Davidson, (H) © Jean-Marc Meriot.
Finally, we determined the dispersal syndrome following the Pérez-Harguindeguy et al. (Reference Perez-Harguindeguy, Díaz, Garnier, Lavorel, Poorter, Jaureguiberry, Bret-Harte, Cornwell, Craine, Gurvich, Urcelay, Veneklaas, Reich, Poorter, Wright, Ray, Enrico, Pausas, de Vos, Buchmann, Funes, Quetier, Hodgson, Thompson, Morgan, ter Steege, van der Heijden, Sack, Blonder, Poschlod, Vaieretti, Conti, Staver, Aquino and Cornelissen2013) guidelines. Fleshy diaspores (e.g., Figure 2D–G) or diaspores with fleshy appendages (e.g., Figure 2C) or bright colour (e.g., Figure 2B) were considered as dispersed by internal animal transport (endozoochory), diaspores with flight appendages (e.g., Figure 2A) were considered as dispersed by wind (anemochory), while diaspore without any evidence of fleshy or flight appendages were considered to be dispersed by other means. For species dispersed by endozoochory, we determined the dimension of the diaspore that would limit its ability to be swallowed by birds (or other dispersers). For all spherical (length = width = thickness), ovoid (length ≠ width = thickness), and elliptical diaspores (length ≠ width ≠ thickness), we considered the second-longest dimension (i.e., the width) as the limiting dimension. When minimal (min) and maximal (max) values were provided we used the average value computed as [min+max]/2. When only one value was provided, we considered it as the average value.
Dispersal by birds
Twenty-two native terrestrial bird species consume fruits or seeds at least occasionally in New Caledonia (Table 1). Ten of these species are strictly frugivorous and/or granivorous, while the others are omnivorous or nectarivorous and consume fruits and seeds occasionally (Demoulins & Barré Reference Demoulins and Barré2005, Doughty et al. Reference Doughty, Day and Plant1999, Duston Reference Duston2011, Hannecart & Létocart Reference Hannecart and Létocart1980). For those 22 bird native species, we extracted morphological traits from the AVONET database (Tobias et al. Reference Tobias, Sheard, Pigot, Devenish, Yang, Sayol, Neate-Clegg, Alioravainen, Weeks, Barber, Walkden, MacGregor, Jones, Vincent, Phillips, Marples, Montaño-Centellas, Leandro-Silva, Claramunt, Darski, Freeman, Bregman, Cooney, Hughes, Capp, Varley, Friedman, Korntheuer, Corrales-Vargas, Trisos, Weeks, Hanz, Töpfer, Bravo, Remeš, Nowak, Carneiro, Moncada, Matysioková, Baldassarre, Martínez-Salinas, Wolfe, Chapman, Daly, Sorensen, Neu, Ford, Mayhew, Fabio Silveira, Kelly, Annorbah, Pollock, Grabowska-Zhang, McEntee, Carlos, Gonzalez, Meneses, Muñoz, Powell, Jamie, Matthews, Johnson, Brito, Zyskowski, Crates, Harvey, Jurado Zevallos, Hosner, Bradfer-Lawrence, Maley, Stiles, Lima, Provost, Chibesa, Mashao, Howard, Mlamba, Chua, Li, Gómez, García, Päckert, Fuchs, Ali, Derryberry, Carlson, Urriza, Brzeski, Prawiradilaga, Rayner, Miller, Bowie, Lafontaine, Scofield, Lou, Somarathna, Lepage, Illif, Neuschulz, Templin, Dehling, Cooper, Pauwels, Analuddin, Fjeldså, Seddon, Sweet, DeClerck, Naka, Brawn, Aleixo, Böhning-Gaese, Rahbek, Fritz, Thomas and Schleuning2022). Among the 11 traits available in the database, we kept those related to beak morphology (length, width, and depth), bird mass (which correlated with tarsus length, wing length, tail length, and Kipp’s distance, and Spearman’s Rho ≥ 0.80), and the hand-wing index. Species with higher hand-wing index have more elongated wings more efficient for long-distance flight, whereas species with lower hand-wing index have broader wings associated with weaker or shorter-distance flight (Sheard et al. Reference Sheard, Neate-Clegg, Alioravainen, Jones, Vincent, MacGregor, Bregman, Claramunt and Tobias2020).
Terrestrial bird species native to New Caledonia and consuming at least occasionally fruits or seeds. Bird mass, beak dimensions, and tail length have been extracted from the AVONET database (Tobias et al. 2022)
Origin, N = native, E = endemic species, (E) = endemic subspecies. Status (IUCN status), LC = least concern, NT = near threatened. Diet (main diet), F = frugivorous, G = granivorous, O = omnivorous, N = nectarivorous. Birds’ full names, Chalcophaps longirostris sandwichensis Ramsay, EP, 1878; Columba vitiensis hypoenochroa Quoy & Gaimard, 1830; Drepanoptila holosericea Temminck, 1809; Drepanoptila holosericea Temminck, 1809; Ducula goliath Gray, 1859; Ducula pacifica Gmelin, 1789; Ptilinopus greyi Bonaparte, 1857; Coracina analis Verreaux & Des Murs, 1860; Coracina caledonica Gmelin, 1788; Lalage leucopyga montrosieri J. Verreaux & Des Murs, 1860; Corvus moneduloides Lesson, 1830; Erythrura psittacea Gmelin, 1789; Glycifohia undulata Sparrman, 1787; Lichmera incana incana Latham, 1790; Philemon diemenensis Lesson, 1931; Aplonis striata striata Gmelin, 1788; Turdus poliocephalus xanthopus Forster, 1844; Zosterops lateralis griseonata Gray, 1859; Zosterops xanthochroa Gray 1859; Charmosyna diadema Verreaux & Des Murs, 1860; Cyanoramphus saisseti Verreaux & Des Murs, 1860; Eunymphicus cornutus Gmelin, 1788; Trichoglossus haematodus deplanchei Verreaux & Des Murs, 1860.
We also compiled observations of fruits and seeds consumption from various sources including published papers (Barré et al. Reference Barré, Garine-Wichatitsky, Lecoq and Maillard2003, Duron et al. Reference Duron, Garcia-Iriarte, Brescia and Vidal2017, Tassin et al. Reference Tassin, Boissenin and Barré2010), gray literature (Boissenin et al. Reference Boissenin, Gomez, Barré, Chambrey and Tassin2006, Colombo Reference Colombo2008, Condamin Reference Condamin1978, Létocart Reference Létocart1998), books (Desmoulins & Barré 2005, Hannecart & Letocart Reference Hannecart and Létocart1980), online photo databases (Endemia and iNaturalist), and personal communications from naturalists (Ardimani, Chartendrault, Colombo, Gailhbaud, Lebreton, Meandu Poveu, Mériot, and Sabatier). Those observations either consist of observations in the field (a bird species was observed consuming the fruits or seeds of a plant species), feeding experiments (fruits were offered to birds in captivity) or seeds found in the digestive tracts of hunted birds or the droppings of birds (Table S1).
Data analysis
All analyses were performed using R version 4.4.1 (R Core Team 2024). We computed the percentage of plant species represented in the different dispersal syndrome categories (i.e., endozoochory, anemochory, and others) within twenty-one 1-ha forest plots to assess the importance of endozoochory in New Caledonian rainforests. Then, we graphically compared the frequency distribution of the limiting dimension of diaspores dispersed by internal animal transport and the frequency distribution of beak widths of the 22 native terrestrial bird species that could potentially swallow and disperse these diaspores. Finally, using Spearman’s correlation tests and observed consumption of diaspores by birds, we tested the relationship between the limiting dimension of the diaspores, the beak width, and the body size of these birds.
Finally, we focused on the specific role of D. goliath in dispersing diaspores of New Caledonian rainforest tree species. We performed a principal component analysis (PCA) on the morphological traits of the bird species using the ade4 R package (Dray & Dufour Reference Dray and Dufour2007). This PCA allows us to analyze how birds’ traits correlate to each other and how the morphology of D. goliath stands out from the morphology of the other birds. Based on observations of diaspores consumption by birds, we determined a range of diaspore sizes that could only be swallowed by D. goliath among birds. We used a null model approach to test whether the diaspores observed to be dispersed by D. goliath were significantly wider than those observed to be dispersed by the other bird species. We randomly shuffled the plant species observed to be dispersed by the bird species and then computed the median and the 95% quantile of the width of diaspores dispersed by D. goliath. We repeated these operations a thousand times and then calculated the probability of observing the actual median and 95% quantile values under the null model assuming a normal distribution.
A previous study (Carpenter et al. Reference Carpenter, Read and Jaffré2003) used a lower threshold of 20 mm above which D. goliath would be the only bird able to disperse the diaspores. Still, this value was based on observations in Australian rainforests in Queensland fixing to 20 mm the maximum size of diaspores that can be dispersed by volant birds (Willson et al. Reference Willson, Irvine and Walsh1989). In this previous study, no upper threshold was set up to determine a diaspore size over which D. goliath would not be able to swallow the diaspore. An experimental study (Munzinger, unpublished data), during which fruits were measured before being offered to birds in captivity showed that D. goliath was able to swallow fruits up to 33 mm wide (Diospyros macrocarpa Hiern, Ebenaceae, Figure 2E). We used the observations of diaspore consumption by birds and the average diaspore widths to check whether 20 mm would be a proper threshold above which D. goliath would be the only bird able to disperse the diaspores or if this threshold would have to be revised. Similarly, we checked whether or not D. goliath was observed swallowing diaspores larger than 33 mm. Then, we used these thresholds (or the revised thresholds) to compute the percentage of tree species that could only be swallowed by D. goliath among the endozoochorous species. This analysis was also carried out at the family level, for the 13 plant families most represented in terms of the number of endozoochorous tree species (at least 10 species) in the 21 study plots.
Results
Among the 460 studied native tree species, 377 (82.0%) were identified as dispersed by internal animal transportation (endozoochory), 12.4% by wind (anemochory), and 5.7% by other means. For most of the endozoochorous tree species (76.3% of the species), the diaspore consisted in the fruit and in the seed with the fleshy appendage for the remaining. We found diaspore sizes for 341 of the 377 plant species dispersed by internal animal transportation. The averaged width of the diaspores ranged from less than 2 mm (seeds of Jaffrea xerocarpa (Baill.) H.C.Hopkins & Pillon, Rhamnaceae, Figure 2B) to 55 mm (fruits of Planchonella thiensis Aubrév., Sapotaceae, Figure 2G) and its distribution was highly skewed toward small sizes with a median width of 10 mm (Figure 3A). The average beak width of terrestrial birds consuming at least occasionally fruits or seeds ranged from 2.5 mm (Zosterops xanthochrous G. R. Gray, 1859, Zosteropidae) to 14 mm (Coracina caledonica Gmelin, 1788, Campephagidae). The widths of the diaspores were smaller than the widest beak for 65.7% of the endozoochorous plant species.
(A) Distribution of diaspore width and beak size for endozoochorous plant species (N=341) and birds (N=22). (B) Number of plant (bird) species with a diaspore (beak) smaller than (larger than or equal to) a given width. In (A) and (B), the vertical dotted lines represent the widest beak size (14 mm) and the width of the beak of D. goliath (7.4 mm). The hatched areas represent areas where sizes are above the widest beak width.
Based on beak and diaspore sizes alone, the number of birds that can potentially disperse seeds by internal transportation decreased sharply with the increasing size of the limiting dimension of the diaspore (Figure 3B). For instance, 14 bird species had a beak wider than 5 mm while only five bird species had a beak wider than 10 mm. The average beak width of D. goliath (7.4 mm) was wider than the limiting dimension of 34.9% of the studied endozoochorous plant species (119 among 341 species). Seven other bird species had wider beaks than the limiting dimension of these species. More than a third of the studied endozoochorous plant species (34.3%) had diaspores wider than the average beak width of the species with the widest beak.
We found evidence of observations of diaspore consumption for only 10 of the 22 native terrestrial bird species consuming fruits or seeds at least occasionally (Figure S2). More than half of these observations (i.e., 59%) reported diaspore consumption by D. goliath. D. goliath was the only bird species observed consuming fruits or seeds wider than 16 mm on average (excluding large fruits pecked by Zosterops spp.). The median of the averaged width of the diaspores consumed by D. goliath was 19.3 mm and the 95% quantile was 34.1 mm (Figure 4A). These values were significantly higher (p.value < 0.05) than expected under a null model where observations of diaspore consumption were randomly shuffled between bird species.
(A) Density plot highlighting the diaspore width preference for D. goliath (N = 50) compared to all other species pooled together (N = 30, excluding large fruits pecked by Zosterops spp). (B) Principal component analysis (PCA) highlighting the morphology of the bird species (BM = body mass, HWI = hand-wing index, BD, BL, and BW = beak depth, length, and width). In (B), each square refers to an individual species and the arrows represent trait values. The first and second axes of the PCA explain 53.3 and 28.2% of the observed variance in bird species traits.
The first and second axes of the PCA explained together 81.5% of the observed variance in bird morphology (Figure 4B). D. goliath stands out from the other species on the second axis of the PCA. It was the heaviest bird (600 g on average versus 10–394 g on average for the other species) and had the second longest beak (40.8 mm on average versus 11.3–43.5 mm on average for the other species). On the first axis, D. goliath didn’t have a wider beak than the other species (7.4 mm versus 2.8–14.0 mm on average for the other species). It had an intermediate hand-wing index (28.5 versus 17.3–47.5 for the other species). We didn’t find significant relationships between beak width or the body size of the birds and the average width of the larger diaspore they were observed to consume (Spearman’s rank test, P > 0.1). D. goliath was observed to consume diaspore reaching on average more than four times the average width of its beak.
According to the available observations of consumption and diaspore width, D. goliath would be the unique bird disperser for 21.8% of the native tree species having diaspore 16–33 mm wide (Figure 5A). The diaspores of only 3.4% of these species would be on average too wide (>33 mm) to be swallowed and dispersed by D. goliath. At the family level, 11 of the 13 most species-rich families had at least some species that could only be dispersed by D. goliath. It would be the unique bird disperser for 27.3, 30.0, 33.3, 42.9, and up to 57.1% for the Ebenaceae, Rubiaceae, Clusiaceae, Sapotaceae, and the Lauraceae, respectively.
Diaspore distribution in three width classes highlighting the potential role of D. goliath in the dispersal. (A) For all endozoochorous tree species (N=341) and (B) for tree species within the most species-rich families (families represented by at least 10 species in our dataset), separately. Based on consumption observations, diaspores with widths ranging between 16 and 33 mm could only be dispersed by D. goliath among birds.
Discussion
We found that 82% of the 460 studied native tree species representative of communities found in New Caledonian rainforests are likely dispersed by internal animal transportation (endozoochory). This proportion is substantially higher than the 72% found by Carpenter et al. (Reference Carpenter, Read and Jaffré2003) in New Caledonia over 123 studied species. This smaller proportion of endozoochory to the benefit of a higher proportion of tree species dispersed by wind (anemochory, 12% in our study versus 22% in Carpenter et al. Reference Carpenter, Read and Jaffré2003) could be explained by the fact that the study of Carpenter et al. (Reference Carpenter, Read and Jaffré2003) was biased towards more disturbed forests (Read et al. Reference Read, Jaffré, Godrie, Hope and Veillon2000), and anemochorous tree species are often the first to establish in disturbed areas. Another study in New Caledonia (Ititiaty Reference Ititiaty, Brescia, Bordez, Gensous, McCoy and Fogliani2020) found that only 42% of the 407 studied taxa were dispersed by internal animal transportation (and 14–17% by wind), but this study focused on taxa found in maquis, a shrubland-type vegetation mainly maintained by disturbances (Jaffré Reference Birnbaum, Ibanez, Blanchard, Justeau-Allaire, Hequet, Eltabet, Vieilledent, Barbier, Barrière and Bruy2023). The proportion of endozoochorous species increases along the succession from maquis to forest (Ititiaty et al. Reference Ititiaty, Brescia, Bordez, Gensous, McCoy and Fogliani2020) and from forest edge to forest interior (Ibanez et al. Reference Ibanez, Hequet, Chambrey, Jaffré and Birnbaum2017).
The average width of the widest diaspore dispersed by endozoochory was 55 mm, but 65.7% of the endozoochorous native tree species had diaspores with average widths < 14 mm, i.e., the averaged width of the widest beak among the 22 terrestrial bird species consuming at least occasionally fruits or seeds in New Caledonia. This supports the idea that internal transportation by birds is likely one of the main dispersal means for trees in New Caledonian rainforests and that the dispersal success of the diaspores might be limited by the width of the diaspore compared to the width of the beak of the potential bird dispersers. The important role of birds in the dispersal success of diaspore in New Caledonia is further supported by the fact that the passage of seeds through the digestive tract of birds does not prevent or can even promote seed germination (Boissenin et al. Reference Boissenin, Gomez, Barré, Chambrey and Tassin2006, Duron et al. Reference Duron, Garcia-Iriarte, Brescia and Vidal2017, Tassin et al. Reference Tassin, Boissenin and Barré2010).
We did not find a significant relationship between the average width of the beak of bird species and the maximum width of the diaspore they have been observed to consume as suggested by other studies (Burns Reference Burns2013, McFadden et al. Reference McFadden, Fritz, Zimmermann, Pellissier, Kissling, Tobias, Schleuning and Graham2022). Observations highlight that Columbidae, i.e., D. goliath, the metallic pigeon (Columba vitiensis hypoenochroa Quoy & Gaimard, 1830), the red-bellied fruit dove (Ptilinopus greyi Bonaparte, 1857), and the cloven-feathered dove (Drepanoptila holosericea Temminck, 1809), can swallow diaspores much wider than their beak (see also Tassin et al. Reference Tassin, Boissenin and Barré2010). D. goliath, who has a mean beak width of 7.4 mm, have been observed to swallow a 33 mm wide fruit of Diospyros macrocarpa (Ebenaceae) and defecate the seed (J. Munzinger, unpublished data). The Pacific imperial-pigeon (Ducula pacifica Gmelin, 1798), which also occurs in New Caledonia (on the Loyalty Islands), is a bit smaller and has the same beak width as D. goliath (Tobias et al. Reference Tobias, Sheard, Pigot, Devenish, Yang, Sayol, Neate-Clegg, Alioravainen, Weeks, Barber, Walkden, MacGregor, Jones, Vincent, Phillips, Marples, Montaño-Centellas, Leandro-Silva, Claramunt, Darski, Freeman, Bregman, Cooney, Hughes, Capp, Varley, Friedman, Korntheuer, Corrales-Vargas, Trisos, Weeks, Hanz, Töpfer, Bravo, Remeš, Nowak, Carneiro, Moncada, Matysioková, Baldassarre, Martínez-Salinas, Wolfe, Chapman, Daly, Sorensen, Neu, Ford, Mayhew, Fabio Silveira, Kelly, Annorbah, Pollock, Grabowska-Zhang, McEntee, Carlos, Gonzalez, Meneses, Muñoz, Powell, Jamie, Matthews, Johnson, Brito, Zyskowski, Crates, Harvey, Jurado Zevallos, Hosner, Bradfer-Lawrence, Maley, Stiles, Lima, Provost, Chibesa, Mashao, Howard, Mlamba, Chua, Li, Gómez, García, Päckert, Fuchs, Ali, Derryberry, Carlson, Urriza, Brzeski, Prawiradilaga, Rayner, Miller, Bowie, Lafontaine, Scofield, Lou, Somarathna, Lepage, Illif, Neuschulz, Templin, Dehling, Cooper, Pauwels, Analuddin, Fjeldså, Seddon, Sweet, DeClerck, Naka, Brawn, Aleixo, Böhning-Gaese, Rahbek, Fritz, Thomas and Schleuning2022), but usually regurgitates diaspore wider than 20 mm (McConkey et al. Reference McConkey, Meehan and Drake2004). D. pacifica could swallow diaspores up to 33 mm (Meehan et al. Reference Meehan, McConkey and Drake2002) and seems to prefer to consume medium size diaspores (15–20 mm) over smaller ones (McConkey et al. Reference McConkey, Meehan and Drake2004). It may also be true for D. goliath with a preference of diaspores skewed towards even larger ones.
Based on consumption observations and diaspore dimensions, our results suggest that D. goliath would be the only bird that can disperse diaspores with a width ranging between 16 mm and 33 mm. Diaspores of this range of size represented 21.8% of the 341 studied endozoochorous tree species with known diaspore dimensions and were particularly frequent in the Lauraceae, Sapotaceae plant families (57.1 and 42.9% of the studied Lauraceae and Sapotaceae, respectively). A previous study found that D. goliath would be the only bird able to disperse the diaspore of ∼24% of the 88 endozoochorous species they studied (Carpenter et al. Reference Carpenter, Read and Jaffré2003). We suggest that this proportion was overestimated because the authors did not set up a maximum size over which D. goliath would not be able to swallow the diaspore and thus considered that D. goliath could swallow diaspores up to 50 mm. On the other side, they set up the minimum width above which D. goliath would be the only bird able to disperse the diaspore to 20 mm. This value was based on observations in Australian rainforests in Queensland fixing to 20 mm the maximum size of diaspores that can be dispersed by volant birds (Willson et al. Reference Willson, Irvine and Walsh1989), we used a lower threshold of 16 mm.
D. goliath thus appears as a key disperser for a significant number of tree species. Some extinct bird species could also have dispersed trees with relatively large diaspores in New Caledonia. The Kanaka pigeon (Caloenas canacorum Balouet & Olson, Reference Balouet and Olson1989, Columbidae) would have been about the size of D. goliath (Balouet & Olson Reference Balouet and Olson1989). Other extinct (Gallicolumba longitarsus Balouet & Olson, Reference Balouet and Olson1989) or locally extinct (Pampusana stairi Gray, 1856) Columbidae were much smaller than D. goliath and thus would not have been able to disperse large diaspores. Sylviornis neocaledoniae Poplain, 1980, a giant flightless bird reaching 1.3 m in height could have played an important role in dispersing large diaspores, but a recent isotopic analysis suggests that this bird was carnivorous (based on marine and terrestrial molluscs) (Riamon Reference Riamon2022). Likewise, the extinct giant horned turtle (Meiolania mackayi Anderson, 1925, Meiolaniidae) could have played an important role in dispersing large diaspores, but turtles of this genus were most likely herbivorous (Brown & Moll Reference Brown and Moll2019).
Lizards, especially large geckos and skinks, could also be important dispersers on islands (Valido & Olesen Reference Valido and Olesen2019). New Caledonia hosts over 85 species of geckos (Diplodactylidae and Gekkonidae) and skinks (Scincidae) including some of the largest species described worldwide (Bauer & Sadlier Reference Bauer and Sadlier2000, Sadlier et al. Reference Sadlier, Swan, Astrongatt and McCoy2018). The diet of those species is not well known, but evidence shows that Lacertoides pardalis Sadlier, Shea & Bauer, 1997, an endemic skink living in maquis and reaching more than 100 mm long (without the tail) (Bauer & Sadlier Reference Bauer and Sadlier2000), eats Myrtastrum rufopunctatum (Pancher ex Brongn. & Gris) Burret and Uromyrtus ngoyensis (Schltr.) Burret (Myrtaceae) fruits (Sadlier et al. Reference Sadlier, Shea, Jourdan, Whitaker and Bauer2014, Reference Sadlier, Swan, Astrongatt and McCoy2018). The latter, U. ngoyensis, can be 15 mm wide. Rhacodactylus leachianus Cuvier, 1829 (Diplodactylidae), the second largest described gecko (up to 255 mm long without the tail) (Bauer & Sadlier Reference Bauer and Sadlier2000), also eat fruits including figs (Bauer et al., Reference Bauer, Bauer and Sadlier2001) and Elaeodendron curtipendulum Endl. (Celastraceae) (Bauer & Sadlier Reference Bauer and Sadlier2000) whose fruits can be 10 mm wide. Thus, although large geckos and skinks could be important dispensers for the New Caledonian flora, evidence is lacking for their ability to disperse large diaspores dispersed by D. goliath (larger than 16 mm).
Fruit bats (or flying foxes) belonging to the genus Pteropus are also known as important dispersers in the Indo-Pacific islands (e.g., Banack Reference Banack1998, Fall et al. Reference Fall, Drezner and Franklin2007, Florens et al. Reference Florens, Baider, Marday, Martin, Zmanay, Oleksy, Krivek, Vincenot, Strasberg and Kingston2017). Three Pteropus species are present in New Caledonia (two endemics P. vetulus Jouan, 1863, and P. ornatus Gray, 1870, and one native P. tonganus Quoy & Gaimard, 1830). The diet of the New Caledonian species of Pteropus has been poorly studied (Boissenin et al. 2016, Wiles and Fujita, Reference Wiles and Fujita1992), but Pteropus are known to feed on and disperse the seeds of many forest trees including species with large diaspores such as Diospyros spp. (Ebenaceae), Syzygium spp. (Myrtaceae), or Planchonella spp. (Sapotaceae). While fruit bats can only swallow and defecate relatively small seeds (≤ 6 mm), they disperse large-seeded fruits by carrying (i.e., stomatochory) the whole fruits away from the parent trees (Aziz et al. Reference Aziz, McConkey, Tanalgo, Sritongchuay, Low, Yong, Mildenstein, Nuevo-Diego, Lim and Racey2021, McConkey et al. Reference McConkey, Sushma and Sengupta2024). Fruit bats could notably be the only disperser of the juicy and odorous Planchonella spp. fruits that are too large to be swallowed by D. goliath (e.g., P. sphaerocarpa (Baill.) and P. thiensis Aubrév., 50 and 55 mm wide on average).
Conclusion
Our study supports that D. goliath is a key disperser for the New Caledonian native flora. We estimated that D. goliath would be the only bird disperser for ∼20% of the native tree species dispersed by internal animal transportation (endozoochory) or ∼15% of all the native tree species growing in New Caledonian rainforests. Besides birds, the only alternative native dispersers for those species could be the three flying fox species. The IUCN level of threat of extinction of D. goliath has been lowered from vulnerable (in 1994) to near threatened since 2014. D. goliath as well as the three flying fox species (classified as near threatened for P. vetulus or vulnerable for P. ornatus and P. tonganus) are still threatened by overhunting and habitat shrinkage (e.g., Oedin et al. Reference Oedin, Brescia, Vidal and Millon2022). Conservation actions and hunting regulations should continue to be improved to further lower the threats on these species and the key functions they provide in New Caledonian ecosystems.
Our study also highlights that relatively little is known in New Caledonia and more largely in the Pacific islands (Nogales et al. Reference Nogales, McConkey, Carlo, Wotton, Bellingham, Traveset, González-Castro, Heleno, Watanabe, Ando, Rogers, Heinen and Drake2024) about the dispersal ecology of the native flora and the feeding ecology of the native fauna. Consumption observations are scarce in New Caledonia and are biased towards more observations for D. goliath and more generally for pigeons and doves than for the other bird taxa. Our study also focused on endozoochory; therefore, the role of other birds such as the parrots that can carry large diaspores far away from the parent trees might have been overlooked (McConkey et al. Reference McConkey, Sushma and Sengupta2024). Improving our knowledge of interactions between the flora and fauna, including all modes of dispersal, would help us better understand the functioning of these unique ecosystems.
Supplementary material
The supplementary material for this article can be found at https://doi.org/10.1017/S026646742500001X
Acknowledgments
We are grateful to David Bruy, Porter P Lowry II, Vanessa Hequet, Richard MK Saunders, and Neil Snow for helping with the fruits and seeds description and to Philippe Birnbaum for providing forest plot data. We thank Vanessa Hequet, Jean-Marc Meriot, and the late Christopher Davidson for providing field pictures. We would like to thank Pierric Gailhbaud and Raphaël Colombo who contributed some of the data as part of their unpublished master’s degrees. We thank Fabrice Brescia for commenting on the manuscript. We thank two anonymous reviewers for their helpful comments.
Financial support
T Ibanez has been supported by the Hermon Slade Foundation (grant number HSF 19105).
Competing interests
The authors declare none.
