SEARCH

SEARCH BY CITATION

Keywords:

  • Bolivia;
  • Cyanocorax cyanomelas ;
  • disperser effectiveness;
  • frugivory;
  • Guettarda viburnoides ;
  • Pteroglossus castanotis ;
  • savanna

Abstract

  1. Top of page
  2. AbstractResumen
  3. Methods
  4. Results
  5. Discussion
  6. Conclusion
  7. Acknowledgments
  8. Literature Cited
  9. Supporting Information

For many tropical plants, birds are the most important seed dispersers. Not all birds, however, will provide equally effective dispersal services. Behavioral differences, during and after feeding, can result in different establishment probabilities of new individuals. During 3 yr, we examined species-specific quantitative and qualitative aspects of Guettarda viburnoides seed dispersal by avian frugivores, focusing on how these aspects modify seed dispersal effectiveness. Fruits of G. viburnoides were consumed by ten species of birds, two of which, Cyanocorax cyanomelas and Pteroglossus castanotis, removed 80 percent of the fruits. These two species differ in qualitative aspects of seed dispersal. First, they select for fruits of different sizes; C. cyanomelas feeds on larger fruits than P. castanotis, which results in the former dispersing larger endocarps than the latter. Second, they differ in their fruit handling treatment; C. cyanomelas are pulp consumers, whereas P. castanotis swallow the fruit whole, and are thus traditionally considered ‘legitimate’ dispersers. The probability of seedling emergence, the temporal pattern of emergence, the number of emerged seedlings per endocarp, and the probability of post-dispersal seed predation differs between endocarps dispersed by C. cyanomelas and P. castanotis; endocarps dispersed by the former have higher emergence probabilities, higher number of seedlings, faster emergence times, and lower predation probabilities than those dispersed by the latter. Finally, these birds differ in their landscape patterns of endocarp deposition; C. cyanomelas disperses endocarps to habitats with higher recruitment probabilities. Ultimately, the pulp consumer C. cyanomelas is a more effective disperser of G. viburnoides than P. castanotis.

Resumen

Para muchas plantas tropicales, la aves son los dispersores de semillas más importantes. Sin embargo, no todas las aves proveen servicios de dispersión igualmente efectivos. Diferencias en el comportamiento, tanto durante como después de la alimentación, pueden resultar en distintas probabilidades de reclutamiento de nuevos individuos. Durante tres años examinamos aspectos cuanti- y cualitativos de la dispersión de Guettarda viburnoides por aves frugívoras, enfocándonos en cómo esos aspectos modifican la efectividad de la dispersión. Los frutos de G. viburnoides son consumidos por 10 especies de aves, dos de las cuales, Cyanocorax cyanomelas y Pteroglossus castanotis, removieron el 80% de los frutos. Estas dos especies difieren en aspectos cualitativos de la dispersión de semillas. Primero, seleccionan frutos de tamaños distintos; C. cyanomelas se alimenta de frutos más grandes que P. castanotis, lo que resulta en que la primera disperse endocarpos más grandes que la segunda. Segundo, ambas especies difieren en el tratamiento que le dan al fruto; C. cyanomelas consume sólo la pulpa, mientras que P. castanotis traga el fruto entero y, por lo tanto, es considerado como un dispersor legítimo. La probabilidad de emergencia de plántulas, el número de plántulas que emergen por endocarpo, y la probabilidad de la depredación se semillas difiere entre endocarpos dispersados por C. cyanomelas y P. castanotis; los endocarpos dispersados por la primera especie tienen mayor probabilidad de emergencia, mayor número de plántulas, tiempos más rápidos de emergencia y menor probabilidad de depredación que aquellos dispersados por la segunda especie. Finalmente, estas aves difieren en los patrones espaciales de dispersión; C. cyanomelas dispersa los endocarpos a hábitats como mayor probabilidad de reclutamiento. En conclusión el consumidor de pulpa C. cyanomelas es un dispersor más efectivo de G. viburnoides que P. castanotis.

Vertebrate seed dispersal is a key process for the maintenance of plant populations (Howe & Smallwood 1982). From an individual plant perspective, however, not all dispersers will be equally important (Calviño-Cancela 2002, Figuerola et al. 2002, Wehncke et al. 2004, Bas et al. 2006). A disperser's effectiveness will depend on the number of seeds it disperses (quantity component), the condition of the dispersed seeds, and the probability that seeds will be dispersed to habitats where they can produce new recruits (quality components) (Schupp 1993, Schupp et al. 2010). Therefore, a frugivores’ dispersal effectiveness will be largely determined by its foraging behavior (Loiselle & Blake 1999, Jordano & Schupp 2000, Russo 2003).

A frugivore's feeding behavior can influence both the quantity and quality components of disperser effectiveness (Jordano & Schupp 2000, Russo et al. 2006, Carlo & Morales 2008). For example, swallowing the fruit whole versus ingesting only the pulp, and the number of fruits removed, among others, are factors that affect the quantity component of disperser effectiveness (Russo 2003, Carlo & Morales 2008).

Fruit treatment in the mouth or gut can also affect the quality component of disperser effectiveness. First, it can influence seed deposition patterns; although some pulp consumers (i.e., birds that ingest only the pulp) can take fruits for consumption to roosts or other locations, this behavior usually results in a large number of seeds dropped under the parent plant (Levey 1987, Jordano & Schupp 2000). In contrast, frugivores that swallow the fruit whole usually deposit a larger proportion of seeds away from the parent plant (Holbrook & Loiselle 2007). Through differential survival and growth associated with the habitats where seeds are dropped, seed deposition patterns may, in turn, determine the probability of plant recruitment (Schupp & Fuentes 1995, Nathan & Müller-Landau 2000). Second, fruit treatment can affect germination probabilities (Traveset & Wilson 1997, Traveset & Verdú 2002, Traveset et al. 2007), either negatively (e.g., Ellison et al. 1993, Domínguez-Domínguez et al. 2006), or positively (Webber & Woodrow 2004, Bas et al. 2006). Similarly, pulp removal (without seed ingestion) can increase germination, seedling emergence rates, seedling growth, and ultimately lead to a higher probability of establishment (Gross-Camp & Kaplin 2011, Fedriani et al. 2012).

Frugivore preferences for certain fruit traits, such as fruit size, may also affect a frugivores’ foraging behavior and translate into differential dispersal efficiency. For example, Rey et al. (1997) found that an increase in fruit size caused a shift in the feeding behavior of Mediterranean birds from swallowing to pecking. In addition, plant species composition and patterns of community-wide fruit abundance can also influence a seed disperser's effectiveness (Kaplin & Moermond 1998).

Because most plants are dispersed by large assemblages of frugivores that have a low degree of specialization (Bascompte & Jordano 2007), it is generally argued that although frugivores affect plant fitness (Godínez-Alvarez et al. 2002, Howe & Miriti 2004, Loayza & Knight 2010), they are not strong selective agents in driving the evolution of fruit traits (Jordano 1995a,b). There is, however, increasing evidence that through differences in their behavior, frugivores can exert selective pressures on heritable plant traits that affect seed dispersal (e.g., Lord et al. 2002, Fadzly & Burns 2010, Lomáscolo et al. 2010). Traits such as fruit size are differentially selected for by frugivores depending on their feeding behavior (Jordano 1995a,b). For example, frugivores that swallow the fruit whole will choose smaller fruits, whereas pulp consumers can target larger fruits. If fruit size is related to seed size, selection of larger fruits may increase seed survival (Moles & Westoby 2004) and ultimately exert strong selection toward an increase in fruit size. Moreover, selective pressures can be coupled with the frugivores’ dispersal effectiveness. Thus, to understand the potential selective role of dispersers on plant traits and recruitment, a key step is to examine how species-specific components of frugivore behavior influence dispersal effectiveness.

Here, we used field observations and greenhouse experiments to identify quantitatively important dispersers (QID, hereon) of Guettarda viburnoides (Rubiaceae) in northeastern Bolivian savannas, and determine how their feeding behavior modifies key aspects of dispersal effectiveness. Specifically, we examined if differences in the feeding behavior of QID of G. viburnoides can lead to differences in (1) seed removal; (2) fruit choice based on size; (3) habitats of seed deposition; (4) seedling emergence probabilities; (5) the probability of post-dispersal seed predation by ants; and (6) whether they exert different selective pressures on seed size.

Methods

  1. Top of page
  2. AbstractResumen
  3. Methods
  4. Results
  5. Discussion
  6. Conclusion
  7. Acknowledgments
  8. Literature Cited
  9. Supporting Information

Study site and species

We conducted this study from 2006 to 2008 in the savannas of Beni Biological Station-Biosphere Reserve (BBS-BR) in Bolivia (14º30′–14º50′ S; 66º40′–65º50 asl). The area receives an average of 1900 mm of rain/year and is characterized by a marked seasonality, with a wet season between November and April, and a dry season during the rest of the year (Miranda 2000). The savannas lie between 130 and 235 m in elevation with local relief ranging from 2 to 6 m. This relief results in a very heterogeneous landscape, with permanent swamps, areas flooded from 4 to 10 mo, uplands that generally do not flood, forest islands, and small patches of woody vegetation that commonly form on termite or ant mounds (hereon, ‘woody patches’). Forest islands are isolated units of forest with a canopy height of up to 25 m (Moraes et al. 2000); at the study site, they range from 0.1 to 20 ha. Woody patches consist of small areas (2–175 m2) occupied by woody vegetation with a canopy height of up to 8 m. In general, woody patches will consist of a few—typically animal-dispersed—species. For this study, we distinguished between woody patches with or without an adult G. viburnoides tree. A map showing the distribution of forest islands and woody patches across the study site, as well as a more detailed description of these habitats, can be found elsewhere (Loayza & Knight 2010, Fig. 1).

image

Figure 1. Emergence curves for G. viburnoides. All curves end when the last seedling emerged. The experiment was terminated at 366 days.

Download figure to PowerPoint

Guettarda viburnoides Cham. and Schlecht. (Rubiaceae) are small trees (<6 m) that grow in semi-deciduous forests and grasslands from Brazil to Paraguay (Taylor et al. 2004). At BBS-BR, G. viburnoides produces subglobose drupes that are ripe from late March until early July. Each fruit encloses a single woody endocarp (7–15 mm, inline image9.67, SD = 1.68), which contains between three and seven seeds (inline image5, = 800). For the plant, the endocarp, and not the individual seeds within it, constitutes the unit of dispersal. Seeds are depredated mostly by Pheidole ants, which chew holes through the endocarp germination pores (Saavedra 2008). At the study site, the population of G. viburnoides consisted of approximately 140 adult individuals, which grow mostly in woody patches in the savanna, with a very small fraction of the population also occurring in forest islands. The majority of the woody patches have only a single adult G. viburnoides tree; hence, it is very easy to observe frugivore feeding behavior in the trees. From April to June, the fruits of G. viburnoides accounted for the large majority of the total fruit biomass available in this habitat.

Quantity components of seed dispersal

Frugivore observations were conducted by two to three independent observers from mid April to early May on a subset of 15, 12, and 5 large, visible trees in woody patches during 2006, 2007, and 2008, respectively. Observers were hidden at a distance of 10–20 m, and watched frugivore behavior with binoculars (10 × 50). Each tree was observed for up to 3 days, and frugivore activity was recorded for 3 h in the morning (0645–0945 h) and two in the afternoon (1600–1830 h). During each visit, we recorded (1) the identity of the frugivore species; (2) the number of fruits consumed per visit; (3) whether the frugivore swallowed the fruits whole or consumed only the pulp; and (4) whether frugivores defecated, dropped or regurgitated seeds before leaving the woody patch. We defined QID of G. viburnoides based on two criteria: (1) the bird species that accounted for highest percentage of fruits removed; and (2) movement of endocarps away from the feeding site. For this study, we considered a frugivore as a seed disperser only if, at least on certain occasions, it dispersed the endocarp out of the woody patch where the fruiting tree was located. Because woody patches are very small (inline image m2), and the canopy of an average G. viburnoides tree occupies a large portion of a patch, we did not consider birds as seed dispersers if they always dropped the endocarps in the home patch.

Quality components of seed dispersal

Post-feeding movements and endocarp deposition sites

To determine the habitats where QID disperse the endocarps, observers also recorded the post-feeding habitat immediately visited by the frugivore and, if possible, the actual site of endocarp deposition. Specifically, in the case of frugivores that swallowed the fruits whole and did not defecate or regurgitate the endocarps before leaving the tree, observers followed the bird with binoculars for up to 10 min and recorded the habitat of its first perch. We assumed that this habitat was where endocarps were most likely to be dispersed. In contrast, for pulp consumers, it was possible to see exactly were endocarps were dropped. In either case, the physical characteristics of the landscape allowed easy visualization and following of birds. We used a Chi-square goodness of fit test to determine whether QID dispersed seeds equally among the different habitats in the landscape.

Seedling emergence of endocarps dispersed by qid

To determine if endocarps dispersed by pulp consumers and by birds that swallow the fruit whole have different rates of seedling emergence, we established a greenhouse experiment in 2006, and set up 96 replicates each of three treatments: (1) gut-passed endocarps (i.e., dispersed by birds that swallow the fruit whole); (2) endocarps with the pulp removed (i.e., dispersed by pulp-consumers); and (3) endocarps in intact fruits (control). Individual endocarps were planted in 18-cell seedling trays, where each cell was randomly assigned an endocarp treatment. Gut-passed endocarps and endocarps with the pulp removed were collected from forest islands and woody patches. Intact fruits were collected from >25 trees in the population at the end of the fruiting season. We followed seedling emergence for 1 yr in replicates placed under identical soil, light, and moisture conditions.

We used several analyses to determine whether endocarp treatment affects seedling emergence. First, to examine if the temporal pattern of seedling emergence differed among treatments, we conducted a Cox's proportional hazards model. Here, the dependent variable is the hazard function, which describes how the chance of emerging (i.e., hazard) changes with time with respect to endocarp treatment. For this analysis, the hazard functions of gut-passed endocarps, and endocarps with the pulp removed were compared to the hazard function of endocarps in intact G. viburnoides fruits. Second, to test whether the proportion of endocarps with emerged seedlings at the end of the experiment differed among treatments, we conducted a proportions test. Finally, because each endocarp contains on average five seeds, there can be variation in the final number of emerged seedlings per endocarp. Thus, to examine if the number of seedlings that emerged per endocarp differed among treatments, we used a Kruskal–Wallis test (excluding endocarps with zero emergence), and conducted non-parametric multiple comparisons for unequal sample sizes to determine differences between groups.

Fruit size selected by QID

We examined if frugivores with different feeding behavior select fruits of different size in two steps. First, we established how fruit size varies across the G. viburnoides population and then we examined differences in mean endocarp size between endocarps dispersed by pulp consumers and by birds that swallow the fruit whole. Fruit size variation was established from 15 adult, similar-sized trees from the population. For each tree, we measured the width of 25 to 34 haphazardly selected fruits to the nearest mm with a digital caliper (Mitutoyo, Absolute Digimatic Caliper Series 500, accuracy ±0.01 mm). We also measured the width of the endocarp of each fruit to determine if endocarp size can be predicted from fruit size. We examined fruit size variation among trees with a one-way ANOVA. For each tree, the relationship between fruit and endocarp size was examined with a linear regression. For both analyses, original data were log-transformed to meet normality requirements.

To examine variation in endocarp size between endocarps dispersed by pulp consumers and by birds that swallow the fruit whole, in 2008, we randomly selected 199 gut-passed endocarps and 210 endocarps with only the pulp removed, and measured them to the nearest mm with a digital caliper. All endocarps were collected in the study area from multiple locations, and belonged to a larger pool of endocarps collected in 2008. Differences in endocarp size between the two treatments were examined with a t-test using log-transformed data to meet normality requirements.

Relationship between endocarp size and risk of predation

To examine if the risk of being depredated by ants was a function of endocarp size (and ultimately fruit choice by QID), we randomly selected 200 endocarps from a pool of endocarps collected at the study site in 2008 and, for each endocarp, measured its width and recorded whether or not it was predated. The relationship between endocarp size and the probability of predation was examined using a logistic regression. All statistical analyses were conducted using the R statistical environment (R Development Core Team 2012).

Seed dispersal effectiveness (sde)

We used the data on the quantitative and qualitative components to calculate the seed dispersal effectiveness of each QID. Specifically, the SDE was calculated as the product between the values of the quantity component (proportion of visits by each QID*proportion of visits that result in dispersal outside the woody patch*mean number of fruits consumed per visit) and the values of the quality component (mean probability of endocarp surviving predation*proportion of emerged endocarps*mean number of seedlings emerged per endocarp). Note that, although the SDE provides a framework to compare the contributions of different dispersal agents on plant recruitment, it is not without caveats. Specifically, the calculation of SDE assumes that all the important factors have been included and that all the factors multiplied together are equally important. Moreover, it is important to consider that the SDE can be context-dependent across temporal and spatial scales (Schupp et al. 2010).

Results

  1. Top of page
  2. AbstractResumen
  3. Methods
  4. Results
  5. Discussion
  6. Conclusion
  7. Acknowledgments
  8. Literature Cited
  9. Supporting Information

Quantity components of seed dispersal

We observed ten species of birds handling and/or consuming fruits of G. viburnoides trees (402 h of observation) (Table 1). On the basis of the total number of visits, the total number of fruits removed, and the feeding behavior, we identified two species, Cyanocorax cyanomelas (Purplish Jay) and Pteroglossus castanotis (Chestnut-eared Aracari) as QID of G. viburnoides at the study site (Table 1). Two other species, Cyanocorax chrysops (Plush-crested Jay) and Ara severa (Chestnut-fronted macaw), also accounted for a relatively high percentage of visits over the 3 yr, but they were not considered QID for two reasons. Ara severa consumed the pulp of G. viburnoides fruits without destroying the endocarp, but dropped 100 percent of the endocarps under the parent tree. Cyanocorax chrysops is a visitor almost as frequent as P. castanotis, but it consumes very few fruits per visit in comparison with P. castanotis. Therefore, overall, it accounts for a small percentage of the total number of fruits removed each year.

Table 1. Percent of visits by fruit-eating birds to G. viburnoides trees in woody patches at Beni Biological Station-Biosphere Reserve between 2006 and 2008.
SpeciesFrugivore typeaMean number of fruits handled per visit ± 1 SDDispersed seeds out of a woody patchPercent of total visits
2006 (170 h)2007 (173 h)2008 (59 h)3-yr mean
  1. a

    SD, seed disperser; SP, seed predator; PC, pulp consumer; PCSD, pulp consumer–disperser (sensu Jordano & Schupp 2000).

  2. b

    This species swallows the fruit whole, but likely destroys the endocarp in the gizzard.

Ara severa PC15.74 ± 3.71No15.993.306.43
Columba cayannensis SPb2No0.670.22
Cyanocorax cyanomelas PCSD4.02 ± 2.98Yes64.6783.5277.7875.32
Cyanocorax chrysops PCSD3.08 ± 0.99Yes5.332.207.414.98
Ortalis motmot SD10.50 ± 3.53No2.200.73
Pipile pipile SD12.50 ± 2.12No1.330.44
Psarocolius decumanus PC1.40 ± 0.55No2.673.702.12
Pteroglossus castanotis SD18.06 ± 9.02Yes6.675.497.416.52
Ramphastos toco SD14.57 ± 3.50Yes2.673.301.99
Tyrannus melancholicus SD1No3.701.23

Fruit handling techniques

Cyanocorax cyanomelas ingest only the pulp without swallowing the endocarp. They frequently drop the endocarp under the fruiting tree in the woody patch, but occasionally disperse it to other woody patches (see next section); thus, they are considered as pulp consumer–dispersers (PCSD; Jordano & Schupp 2000). In contrast, P. castanotis swallows the fruit whole and always defecates intact endocarps away from the woody patch (see next section); thus, it would be traditionally considered as a ‘legitimate’ seed disperser (SD; Jordano & Schupp 2000). Cyanocorax cyanomelas and P. castanotis accounted for approximately 75 percent and 6.5 percent, respectively, of the total visits to G. viburnoides from 2006 to 2008 (Table 1). During an individual visit, however, P. castanotis removed almost four times more fruits than C. cyanomelas (= 15.18, < 0.001) (Table 2).

Table 2. Quantity and quality components of G. viburnoides seed dispersal by C. cyanomelas and P. castanotis. These values were used for the SDE calculations.
  C. cyanomelas P. castanotis
  1. a

    Based on the total number of visits recorded over 3 yr.

  2. b

    Estimated as 1 − Pi, where Pi is the probability that endocarps measuring 10.37 and 9.02 mm (estimated from the logistic regression curve) and is calculated as:

    • display math
  3. c

    Note that the values per se have no biological meaning, and should only be used for comparative purposes; that is, that C. cyanomelas is a more effective seed disperser than P. castanotis.

Quantitative components
Proportion of total visitsa0.7500.065
Proportion of visits that result in dispersal outside the woody patch0.0971.000
Mean # fruits consumed per visit4.6218.06
Product 0.34 1.17
Qualitative components
Mean size of dispersed endocarps (mm)10.379.02
Mean probability of survivalb0.730.58
Proportion of emerged endocarps0.650.19
Mean number of seedlings emerged/endocarp1.881.25
Product 0.89 0.14
SDEc 0.30 0.16

Post-feeding movements and endocarp deposition sites

Patterns of habitat selection following fruit consumption differed among the two QID of G. viburnoides. During this study, all individuals of P. castanotis observed feeding from G. viburnoides in woody patches (= 17), immediately flew and perched in a forest island following fruit consumption. In all instances, P. castanotis left the woody patch without regurgitating or defecating the endocarps (Table 2). Consequently, it is likely that this species disperses all of the G. viburnoides endocarps to forest islands. This result is supported by seed trap data across the study site in 2006 and 2008; endocarps collected in seed traps in forest islands had been defecated > 95 percent of the time, while those collected in seed traps in woody patches had only the pulp removed.

On average, 9.7 percent of the total visits by C. cyanomelas resulted in the dispersal of an endocarp away from the feeding site (Table 2). This translated into a total of 19 endocarps (2.1% of fruits consumed) dispersed by C. cyanomelas to other woody patches. No visits by C. cyanomelas resulted in dispersal to forest islands. There was no difference in the proportion of endocarps dispersed to woody patches either with or without G. viburnoides2 = 0.053, > 0.05).

Seedling emergence of endocarps dispersed by QID

The temporal pattern of seedling emergence differed significantly among endocarps dispersed by birds that swallow the fruit whole, pulp consumers and non-dispersed endocarps (χ2 = 48.3, df = 2, < 0.001). Compared with endocarps in intact fruits, time to emergence was 1.87 times faster in endocarps with the pulp removed (β = 0.87, βSE = 0.21, < 0.001), whereas 1.22 times slower in gut-passed endocarps (β = −0.78, βSE = 0.29, < 0.01) (Fig. 1).

After 1 yr, the proportion of endocarps with emerged seedlings differed among treatments (χ2 = 28.35, df = 2, < 0.001). The proportion of endocarps with emerged seedlings was highest for endocarps with the pulp removed, followed by endocarps in intact fruits, and lowest for gut-passed endocarps (Fig. 1). Finally, the number of seedlings emerged per endocarp also differed according to endocarp treatment (H’ = 16.19, df = 2, < 0.001). Specifically, more seedlings emerged per endocarp in endocarps with the pulp removed (1.88, SE = 0.14) than in either gut-passed endocarps (1.25, SE = 0.12) or endocarps in intact fruits (1.19, SE = 0.08) (Table 2; Fig. 2); there was no difference in the final number of seedlings emerged between gut-passed endocarps and endocarps in intact fruits.

image

Figure 2. Frequency distribution of the number of emerged seedlings per endocarp in each of the treatments.

Download figure to PowerPoint

Fruit size selected by qids

Mean fruit size differed significantly among trees (F4,370 = 69.47, < 0.0009), ranging from 10 to 21 mm among the sampled individuals. Fruit size predicted endocarp size in 14 of the 15 selected trees (Appendix A). When performing the analysis for all 15 trees combined, fruit size explained 73 percent of the variation in endocarp size (r2 = 0.73, = 384, < 0.001); therefore, we used endocarp size as a reliable surrogate of fruit size.

Endocarp size differed between gut-passed endocarps and endocarps with the pulp removed (= 11.3, df = 406, < 0.001). Endocarps processed by pulp consumers (most likely C. cyanomelas) were on average 1.35 mm larger than gut-passed endocarps processed by ‘legitimate’ dispersers (most likely P. castanotis) (Table 2). This suggests that, on average, birds that swallow the fruit whole feed on smaller fruits than do pulp consumers. By considering only the size of defecated endocarps and the results from the regression of fruit versus endocarp size, we can infer that P. castanotis feeds mostly on fruits between 11.4 and 13.4 mm, and can swallow fruits up to approximately 15 mm, whereas C. cyanomelas feeds predominantly on fruits ranging from 15.5 to 17.5 mm, and is able to feed on fruits up to 25 mm (Fig. 3).

image

Figure 3. Distribution of mean fruit sizes among sampled trees. Error bars indicate 1 ±  SE. Box with diagonal lines indicates predicted range of fruit sizes that P. castanotis mostly selects (based on sizes of gut-passed endocarps). Box with crosshatch shows the predicted range of fruit sizes on which C. cyanomelas most commonly feeds (based on sizes of endocarps with the pulp removed). Dashed and dotted vertical lines indicate the maximum predicted fruit size P. castanotis and C. cyanomelas, respectively, feed on (based on the maximum size of gut-passed endocarps and endocarps with the pulp removed).

Download figure to PowerPoint

Relationship between endocarp size and risk of predation

The probability of an endocarp being depredated was related to its size (estimate = −0.5, z = −4.3, < 0.0001). Specifically, smaller endocarps (i.e., those dispersed by birds that swallow the fruit whole) have a ca 2.2 higher risk of predation than larger endocarps (Table 2; Fig. 4).

image

Figure 4. Fitted logistic regression curve showing that the probability of post-dispersal predation (predated and not predated) of G. viburnoides is dependent on endocarp size. Histograms represent the observed data and the line is the predicted probability of predation. Lines below species name represent the predicted range of endocarp size on which each species most commonly feeds.

Download figure to PowerPoint

Seed dispersal effectiveness

Overall, pulp consumers (i.e., C. cyanomelas) provide more effective seed dispersal to G. viburnoides than ‘legitimate’ dispersers (i.e., P. castanotis) (Table 2). For P. castanotis, in spite of the large value of the quantity component, the SDE was relatively low because of the low-quality dispersal it provides to G. viburnoides. Conversely, the low values of the quantity component of dispersal are compensated by large values of dispersal quality by C. cyanomelas.

Discussion

  1. Top of page
  2. AbstractResumen
  3. Methods
  4. Results
  5. Discussion
  6. Conclusion
  7. Acknowledgments
  8. Literature Cited
  9. Supporting Information

Results from this study show that only a few species from a tropical frugivore assemblage visiting a plant can account for most of its seed dispersal. Moreover, we show that as a result of differences in feeding behavior, bird species differ in the seed dispersal benefits they provide to G. viburnoides. Using a spatially explicit model that combined data on habitat-specific variation in demographic rates of G. viburnoides and on the sites where C. cyanomelas and P. pteroglossus disperse the endocarps, Loayza and Knight (2010) showed how seed dispersal by these species could theoretically affect the population dynamics of G. viburnoides. Here, we provide empirical evidence, of how the individual components of feeding and post-feeding behavior interact to determine a dispersers’ effectiveness.

Quantity components of seed dispersal

The overall assemblage of frugivorous birds feeding on G. viburnoides was relatively constant in successive years. The most frequent and reliable visitor to G. viburnoides was C. cyanomelas (>75% visits). Yet, visit frequency alone was not a valid predictor of the relative quantity of fruits removed, as P. castanotis removed over 30 percent of the number of fruits taken by C. cyanomelas, but accounted for only 6.5 percent of the visits.

The number of fruits removed by each QID resulted from a combined function of their foraging behavior and their relative abundance in the landscape. When visiting a tree, an individual of C. cyanomelas removed only a few fruits per visit; this species’ contribution to the quantity component of dispersal effectiveness is given by the frequency of visits. Cyanocorax cyanomelas is a common bird in all habitats at the study site (Brace et al. 1997), therefore the number of fruits removed is likely a consequence of its numerical dominance. In contrast, P. castanotis is less common (Brace et al. 1997), but during a single visit, an individual will consume greater quantities of fruits than C. cyanomelas. Consequently, its contribution to the number of seeds dispersed is given by its foraging behavior, not its abundance. Given that these quantitative differences determine in part the SDE of P. castanotis and C. cyanomelas, the relative importance of these species for the recruitment of G. viburnoides may change if there are spatial and temporal changes in the relative abundances of these birds.

Quality components of seed dispersal

Cyanocorax cyanomelas and P. castanotis differed significantly in five aspects of the qualitative components of seed dispersal: (1) they selected fruits of different sizes; (2) they had different endocarp deposition patterns among the habitats in the landscape; (3) their endocarp treatment in the beak or gut had contrasting effects on seedling emergence probabilities; and (4) endocarps processed by each QID had different probabilities of post-dispersal seed predation.

Cyanocorax cyanomelas fed on larger fruits than P. castanotis. When feeding from one plant species, fruit selection will presumably be determined by the behavior of the frugivore, as well as fruit traits such as ripeness, secondary compounds, and size. For example, pulp consumers may be able to feed on larger fruits than birds that swallow the fruit whole because gape size constrains the size of fruits that can be swallowed (Wheelwright 1985). The result of size differences between endocarps with the pulp removed and gut-passed endocarps was surprising because gape size would not be a factor limiting P. castanotis from consuming larger fruits. This species is not the only one that swallows the fruit whole in the study area; hence, some of the gut-passed endocarps measured could have been consumed by Ramphastos toco and/or Pipile pipile. However, these species are so infrequent that they would not account for any differences in the analysis. Although unlikely, it is also possible that smaller defecated endocarps may be a result of scarification during the passage through the gut (e.g., Fedriani & Delibes 2009, Delibes et al. 2012) rather than by fruit selection by P. castanotis.

Endocarp size of G. viburnoides is related to number of seeds, with larger endocarps containing more seeds (Saavedra 2008). Therefore, fruit size selection may have consequences on emergence probabilities. If the differences we found between endocarps consumed by birds that swallow the fruit whole and pulp consumers do represent differences in the fruit selection process, then these results would suggest that P. castanotis is either consistently selecting the smallest fruits within a crop, or that this species is feeding more commonly on trees that produce smaller fruits. By selecting fruits of different sizes, frugivores have potentially different roles as agents of selection on fruit size (Jordano 1995a,b, Lord 2004). In comparison with P. castanotis, C. cyanomelas is selecting larger fruits and dispersing larger endocarps that have higher emergence probabilities and a lower predation risk. Therefore, this species may be acting as an important selective agent toward an increase in seed size. Moreover, differences in selection pressures exerted by both species are coupled with their seed dispersal effectiveness, with C. cyanomelas being also the most effective disperser of the two QID. Whether these patterns translated into an evolutionary response toward an increase in seed size, however, has yet to be evaluated.

Endocarp deposition habitats differed between the two QID; C. cyanomelas deposited almost all of the endocarps under the parent tree, and dispersed 2 percent of the endocarps to other woody patches in the savanna. In contrast, P. castanotis presumably dispersed all endocarps in forest islands, which was consistently the habitat of the first perch used after feeding by this species. Although we cannot be certain that P. castanotis always dispersed the endocarps in forest islands, our seed trap data taken concurrently during this study support this hypothesis. While indirect, this suggests that birds that swallow the fruit whole disperse endocarps into forest islands.

Differences in the seed rain pattern generated among frugivores are common for many systems (Alcántara et al. 2000a, Wenny 2000, Calviño-Cancela 2002). These differences can be important depending on the extent to which habitats differ in suitability for recruitment (Schupp 1993). At the study site, woody patches were more suitable for recruitment than forest islands (Loayza & Knight 2010). Consequently, the spatial patterns of seed deposition by QID will likely determine the probability of recruitment, and ultimately influence plant fitness. Habitat suitability, however, is context-dependent (Schupp 2007, Schupp et al. 2010, Loayza et al. 2011); thus, the role of QID for plant recruitment may change in successive years or in different populations.

QID also differed in whether they swallowed fruits whole or consumed only the pulp. Cyanocorax cyanomelas consumed only the pulp and dropped the endocarp; conversely, P. castanotis swallowed the fruit and presumable defecated the endocarp in forest islands. Endocarps processed by each QID had different seedling emergence probabilities. Gut-passed endocarps had not only the lowest emergence probabilities, and number of emerged seedlings per endocarp, but also the slowest emergence rate in comparison with endocarps that were either in intact fruits or that had only the pulp removed. Although slower emergence rates do not necessarily translate into a lower probability of recruitment (e.g., Abbott & Roundy 2003), these results suggest that endocarp ingestion may have a negative effect on emergence, and ultimately plant fitness. Similar results have also been observed for other bird species in which germination rates were higher for seeds cleaned of their pulp versus those in toucan feces (Domínguez-Domínguez et al. 2006), and in primates, where seeds with the pulp removed (i.e., spit seeds) had higher establishment rates than those that were defecated (Gross-Camp & Kaplin 2011). Lastly, because gut-passed endocarps were smaller than those with the pulp removed, the lower emergence probabilities and number of emerged seedlings per endocarp may result from the fewer number of seeds in endocarps that had been ingested and defecated, rather than from a negative effect of endocarp ingestion per se.

We show that larger endocarps have a higher probability to survive post-dispersal predation. In general, larger seeds have higher predation rates (e.g., Willson & Whelan 1990, Brewer 2001, Gómez 2004); however, this is not a rule, and a negative relationship has also been observed (e.g., Kollmann et al. 1998, Alcántara et al. 2000b). In the case of G. viburnoides, larger endocarps may have a thicker cover, which would take predators a longer time to reach the seed (Kaufman & Collier 1981, Alcántara et al. 2000b, Zhang & Zang 2008), and thus become less energetically rewarding than smaller endocarps.

Conclusion

  1. Top of page
  2. AbstractResumen
  3. Methods
  4. Results
  5. Discussion
  6. Conclusion
  7. Acknowledgments
  8. Literature Cited
  9. Supporting Information

Understanding the roles of individual dispersers for plant recruitment requires dissecting the components of dispersal effectiveness. In this study, we found an assemblage of ten species of birds feeding on G. viburnoides fruits from trees in woody patches within the savanna. Only two of these species, however, were frequent enough visitors to the trees to be considered quantitatively important dispersers of G. viburnoides. These two species, C. cyanomelas and P. castanotis, differed in several aspects of the quality components of seed dispersal. In contrast to other studies, here we show that although C. cyanomelas are pulp consumers, rather than ‘legitimate’ dispersers, they are more effective dispersers than P. castanotis because the former removes larger quantities of fruits, and because their feeding behavior leads to both the highest seedling emergence probabilities, and to dispersal in habitats where seeds have a higher probability of recruiting. Finally, because C. cyanomelas selects larger fruits, and because population growth at the study site responds mainly to the activity of this species, it is possible that through its behavior, C. cyanomelas exerts a selective pressure toward an increase in maternal fruit size. Clearly, however, more research is needed to understand how frugivore behavior may translate into fitness differences in plants and ultimately drive evolution of fruit traits in response to variation in selective pressures.

Acknowledgments

  1. Top of page
  2. AbstractResumen
  3. Methods
  4. Results
  5. Discussion
  6. Conclusion
  7. Acknowledgments
  8. Literature Cited
  9. Supporting Information

We are grateful to the people who assisted with fieldwork, particularly Francisco Saavedra, Nataniel Burgos, Renato Balderrama, Ariel Terán, Oliver Burgos, Alejandro Yarari, and Mélanie Houard. Fieldwork was facilitated by logistic support of the Instituto de Ecología in La Paz, Bolivia and the Estación Biológica Beni. We also thank Bette Loiselle, Tiffany Knight, John Blake, Eugene Schupp, José Fedriani, Douglas Levey and Beth Kaplin, and the beer-review group at Universidad de La Serena (Chile) for comments that helped improve this study. This study was funded by grants awarded to A.P.L. from the National Science Foundation (DEB-0709753), the Rufford Small Grants Foundation, the Scott Neotropical Fund from the Cleveland Metropolitan Zoo, the Neotropical Grassland Conservancy, the Webster Groves Nature Study Society, Sigma Xi, and the Whitney R. Harris World Ecology Center at the University of Missouri-St. Louis.

Literature Cited

  1. Top of page
  2. AbstractResumen
  3. Methods
  4. Results
  5. Discussion
  6. Conclusion
  7. Acknowledgments
  8. Literature Cited
  9. Supporting Information
  • Abbott, L. B., and B. A. Roundy. 2003. Available water influences field germination and recruitment of seeded grasses. J. Range Manag. 56: 5664.
  • Alcántara, J. M., P. J. Rey, A. M. Sanchez-Lafuente, and F. Valera. 2000b. Early effects of rodent post-dispersal seed predation on the outcome of the plant –seed disperser interaction. Oikos 88: 362370.
  • Alcántara, J. M., P. J. Rey, F. Valera, and A. M. Sanchez-Lafuente. 2000a. Factors shaping seedfall pattern of a bird-dispersed plant. Ecology 81: 19371950.
  • Bas, J. M., P. Pons, and C. Gómez. 2006. Exclusive frugivory and seed dispersal of Rhamnus alaternus in the bird breeding season. Plant Ecol. 183: 7789.
  • Bascompte, J., and P. Jordano. 2007. Plant-animal mutualistic networks: The architecture of biodiversity. Annu. Rev. Ecol. Evol. Syst. 38: 567593.
  • Brace, R. C., J. Hornbuckle, and J. W. Pearce-Higgins. 1997. The avifauna of the Beni Biological Stations, Bolivia. Bird Conserv. Int. 7: 117139.
  • Brewer, S. W. 2001. Predation and dispersal of large and small seeds of a tropical palm. Oikos 92: 245255.
  • Calviño-Cancela, M. 2002. Spatial patterns of seed dispersal and seedling recruitment in Corena album (Empetraceae): The importance of unspecialized dispersers for regeneration. J. Ecol. 90: 775784.
  • Carlo, T. A., and J. M. Morales. 2008. Inequalities in fruit removal and seed dispersal: Consequences of bird behaviour, neighborhood density and landscape aggregation. J. Ecol. 96: 609618.
  • Delibes, M., C. Corbacho, G. Calvo, and J. M. Fedriani. 2012. Agriculture as matchmaker of an unexpected mutualism: Great bustard disperses and enhances of domestic olive seeds. Basic Appl. Ecol. 13: 125131.
  • Domínguez-Domínguez, L. E., J. E. Morales-Mávil, and J. Alba-Landa. 2006. Germinación de semillas de Ficus insipida (Moraceae) defecadas por tucanes (Ramphastos sulfuratus) y monos araña (Ateles geoffroyi). Revista de Biología Tropical 54: 387394.
  • Ellison, A. M., J. S. Denslow, B. A. Loiselle, and M. D. Brenes. 1993. Seed and seedling ecology of Neotropical Melastomataceae. Ecology 74(1733–1749): 1.
  • Fadzly, N., and K. Burns. 2010. What weta want: Colour preferences of a frugivorous insect. Arthropod Plant Interact. 4: 267276.
  • Fedriani, J. M., and M. Delibes. 2009. Functional diversity in fruit-frugivore interactions: A field experiment with Mediterrenean mammals. Ecography 32: 983992.
  • Fedriani, J. M., M. Zywiec, and M. Delibes. 2012. Thieves or mutualists? Pulp feeders enhance endozoochore local recruitment. Ecology 93: 575587.
  • Figuerola, J. A., J. Green, and L. S. Santamaria. 2002. Comparative dispersal effectiveness of wigeongrass seeds by waterfowl wintering in south-west Spain: Quantitative and qualitative aspects. J. Ecol. 90: 9891001.
  • Godínez-Alvarez, H., A. Valiente-Banuet, and A. Rojas-Martínez. 2002. The role of seed dispersers in the population dynamics of the columnar cactus Neobuxbauma tetetzo. Ecology 83: 26172629.
  • Gómez, J. M. 2004. Bigger is not always better: Conflicting selective pressures on seed size in Quercus ilex. Evolution 58: 7180.
  • Gross-Camp, N. D., and B. A. Kaplin. 2011. Differential seed handling by two African primates affects seed fate and establishment of large-seeded trees. Acta Oecol. 37: 578586.
  • Holbrook, K. M., and B. A. Loiselle. 2007. Using toucan-generated dispersal models to estimate seed dispersal in Amazonian Ecuador. In A. J. Dennis, E. W. Schupp, R. J. Green, and D. A. Westcott (Eds). Seed dispersal: Theory and its application in a changing world, pp. 391406. CABI, Oxfordshire, UK.
  • Howe, H. F., and M. N. Miriti. 2004. When seed dispersal matters. Bioscience 54: 651660.
  • Howe, H. F., and J. Smallwood. 1982. Ecology of seed dispersal. Annu. Rev. Ecol. Syst. 13: 201228.
  • Jordano, P. 1995a. Angiosperm fleshy fruits and seed dispersers: A comparative analysis of adaptation and constraints in plant-animal interactions. Am. Nat. 145: 163191.
  • Jordano, P. 1995b. Frugivore-mediated selection on fruit and seed size: Birds and St. Lucie's cherries, Prunus mahaleb. Ecology 76: 26272639.
  • Jordano, P., and E. W. Schupp. 2000. Seed disperser effectiveness: The quantity component and patterns of seed rain for Prunus mahaleb. Ecol. Monogr. 70: 591615.
  • Kaplin, B. A., and T. C. Moermond. 1998. Variation in seed handling by two species of forest monkeys in Rwanda. Am. J. Primatol. 45: 83101.
  • Kaufman, L. W., and G. Collier. 1981. The economics of seed handling. Am. Nat. 118: 4660.
  • Kollmann, J., D. A. Coomes, and S. M. White. 1998. Consistencies in post-dispersal seed predation of temperate fleshy-fruited species among seasons, years and sites. Funct. Ecol. 12: 683690.
  • Levey, D. 1987. Seed Size and Fruit-Handling Techniques of Avian Frugivores. Am. Nat. 129: 471485.
  • Loayza, A. P., and T. Knight. 2010. Seed dispersal by pulp consumers, not “legitimate” seed dispersers, increases Guettarda viburnoides population growth. Ecology 91: 26842695.
  • Loayza, A. P., B. A. Loiselle, and R. S. Rios. 2011. Context-dependent recruitment of Guettarda viburnoides in a neotropical forest–savanna mosaic. Am. J. Bot. 98: 13171326.
  • Loiselle, B. A., and J. G. Blake. 1999. Dispersal of melastome seeds by fruit eating-birds of tropical understory. Ecology 80: 330336.
  • Lomáscolo, S. B., D. J. Levey, R. T. Kimball, B. M. Bolker, and H. T. Alborn. 2010. Dispersers shape fruit diversity in Ficus (Moraceae). Proc. Natl. Acad. Sci. USA 107: 1466814672.
  • Lord, J. M. 2004. Frugivore gape size and the evolution of fruit size and shape in southern hemisphere floras. Austral Ecol. 29: 430436.
  • Lord, J. M., A. S. Markey, and J. Marshall. 2002. Have frugivores influenced the evolution of fruit traits in New Zealand? In D. J. Levey, W. R. Silva, and M. Galetti (Eds). Seed dispersal and frugivory: Ecology, evolution, and conservation, pp. 5568. CABI Publishing, Wallingford, Oxfordshire, UK.
  • Miranda, C. L. 2000. La Reserva de la Biósfera Estación Biológica del Beni, Bolivia. In O. Herrera-MacBryde, F. Dallmeier, B. MacBryde, J. A. Comiskey, and C. Miranda (Eds). Biodiversity, conservation and management in the region of the Beni Biological Station Biosphere Reserve, Bolivia, pp. 2125. Smithsonian Institution/Monitoring and Assessment of Biodiversity Program (SI/MAB), Rockville, Maryland.
  • Moles, A. T., and M. Westoby. 2004. Seedling survival and seed size: A synthesis of the literature. J. Ecol. 92: 372381.
  • Moraes, M. R., M. O. Ribera, and E. Villanueva. 2000. La vegetación de la Reserva de la Biosfera Estación Biológica del Beni y su importancia para la conservación. In O. Herrera-MacBryde, F. Dallmeier, B. MacBryde, J. A. Comiskey, and C. Miranda (Eds). Biodiversity, conservation and management in the region of the Beni Biological Station Biosphere Reserve, Bolivia, pp. 5973. Smithsonian Institution/Monitoring and Assessment of Biodiversity Program (SI/MAB), Rockville, Maryland.
  • Nathan, R., and H. C. Müller-Landau. 2000. Spatial patterns of seed dispersal, their determinants and consequences for recruitment. Trends Ecol. Evol. 15: 278285.
  • R Development Core Team. 2012. R: A language and environment for statistical computing. R Foundation for Statistical Computing, Vienna, Austria. ISBN 3-900051-07-0, URL http://www.R-project.org.
  • Rey, P. J., J. E. Gutierrez, J. Alcántara, and F. Valera. 1997. Fruit Size in Wild Olives: Implications for avian seed dispersal. Funct. Ecol. 11: 611618.
  • Russo, S. E. 2003. Responses of dispersal agents to tree and fruit traits in Virola calophylla (Myristicaceae): Implications for selection. Oecologia 136: 8087.
  • Russo, S. E., S. Portnoy, and C. K. Augspurger. 2006. Incorporating animal behavior into seed dispersal models: Implications for seed shadows. Ecology 87: 31603174.
  • Saavedra, F. 2008. Patrones de reclutamiento – Dispersión, sobrevivencia y establecimiento – de Guettarda viburnoides (Rubiaceae) en las sabanas de la Reserva de la Biósfera Estacion Biológica del Beni (RE-EBB). Licenciatura Thesis, Universidad Mayor de San Andrés, La Paz, Bolivia.
  • Schupp, E. W. 1993. Quantity, quality, and the effectiveness of seed dispersal by animals. Vegetatio 107(108): 1529.
  • Schupp, E. W. 2007. The suitability of a site for seed dispersal is context-dependent. In A. J. Dennis, E. W. Schupp, R. J. Green, and D. A. Westcott (Eds). Seed dispersal: Theory and its application in a changing world, pp. 445462. CABI International, Wallingford, Oxfordshire.
  • Schupp, E. W., and M. Fuentes. 1995. Spatial patterns of seed dispersal and the unification of plant ecology. EcoScience 2: 267275.
  • Schupp, E. W., P. Jordano, and J. M. Gómez. 2010. Seed dispersal effectiveness revisited: A conceptual review. New Phytol. 188: 333353.
  • Taylor, C., P. G. Delprete, A. Vincentini, R. Cortés, D. Zappi, C. Persson, C. Bestetti Costa, and E. Araujo de Anunciação. 2004. Rubiaceae. In J. A. Steyermark, P. E. Berry, K. Yatskievych, and B. K. Holst (Eds). Flora of the Venezuelan Guayana: Volume 8 – Poaceae-Rubiaceae, pp. 497847. Missouri Botanical Garden Press, St. Louis.
  • Traveset, A., A. W. Robertson, and J. Rodriguez-Pérez. 2007. A review of the role of endozoochory in seed germination. In A. J. Dennis, E. W. Schupp, R. J. Green, and D. A. Westcott (Eds). Seed dispersal: Theory and its application in a changing world, pp. 78103. CABI, Oxfordshire, UK.
  • Traveset, A., and M. Verdú. 2002. A meta-analysis of the effect of gut treatment on seed germination. In D. J. Levey, W. R. Silva, and M. Galetti (Eds). Seed dispersal and frugivory: Ecology, evolution, and conservation, pp. 339350. CABI Publishing, Wallingford, Oxfordshire, UK.
  • Traveset, A., and M. F. Wilson. 1997. Effect of birds and bears on seed germination of fleshy-fruited plants in temperate rainforests of southeast Alaska. Oikos 80: 8995.
  • Webber, B. L., and E. Woodrow. 2004. Cassowary frugivory, seed defleshing and fruit fly infestation influence the transition from seed to seedling in the rare Australian rainforest tree, Ryparosa sp. nov. 1 (Achariaceae). Funct. Plant Biol. 31: 505516.
  • Wehncke, E. V., C. N. Valdez, and C. A. Dominguez. 2004. Seed dispersal and defecation patterns of Cebus capucinus and Alouatta palliata: Consequences for seed dispersal effectiveness. J. Trop. Ecol. 20: 535543.
  • Wenny, D. G. 2000. Seed dispersal, seed predation, and seedling recruitment of a neotropical montane tree. Ecol. Monogr. 70: 331351.
  • Wheelwright, N. 1985. Fruit size, gape width and the diet of fruit-eating birds. Ecology 66: 808818.
  • Willson, M. F., and C. J. Whelan. 1990. Variation in postdispersal survival of vertebrate-dispersed seeds: Effects of density, habitat, location, season, and species. Oikos 57: 191198.
  • Zhang, H., and Z. Zang. 2008. Endocarp thickness affects seed removal speed by small rodents in a warm-temperate broad-leafed deciduous forest, China. Acta Oecol. 34: 285293.

Supporting Information

  1. Top of page
  2. AbstractResumen
  3. Methods
  4. Results
  5. Discussion
  6. Conclusion
  7. Acknowledgments
  8. Literature Cited
  9. Supporting Information
FilenameFormatSizeDescription
btp12070-sup-0001-AppendixS1.docxWord document2118KAPPENDIX S1. Relationship between fruit and endocarp size in each of the 15 selected trees. Endocarp size can be predicted from fruit size in all but one of the trees (T15).

Please note: Wiley Blackwell is not responsible for the content or functionality of any supporting information supplied by the authors. Any queries (other than missing content) should be directed to the corresponding author for the article.