No tree an island: The plant-caterpillar food web of a secondary rain forest in New Guinea

ArticleinEcology Letters 7(11):1090 - 1100 · November 2004with59 Reads
Vojtech Novotny at The Czech Academy of Sciences
  • 41.01
  • The Czech Academy of Sciences
Scott E. Miller at Smithsonian Institution
  • 39.41
  • Smithsonian Institution
Jan Leps at University of South Bohemia in České Budějovice
  • 43.02
  • University of South Bohemia in České Budějovice
Abstract
We characterized a plant–caterpillar food web from secondary vegetation in a New Guinean rain forest that included 63 plant species (87.5% of the total basal area), 546 Lepidoptera species and 1679 trophic links between them. The strongest 14 associations involved 50% of all individual caterpillars while some links were extremely rare. A caterpillar randomly picked from the vegetation will, with ≥ 50% probability, (1) feed on one to three host plants (of the 63 studied), (2) feed on < 20% of local plant biomass and (3) have ≥ 90% of population concentrated on a single host plant species. Generalist species were quantitatively unimportant. Caterpillar assemblages on locally monotypic plant genera were distinct, while sympatric congeneric hosts shared many caterpillar species. The partitioning of the plant–caterpillar food web thus depends on the composition of the vegetation. In secondary forest the predominant plant genera were locally monotypic and supported locally isolated caterpillar assemblages.

Do you want to read the rest of this article?

  • ... Analyses were conducted in CANOCO 5 using the conservative approach of comparing the adjusted variation and p-value (999 random permutations, p adj. < 0.05) as recommended for datasets with lot of species and variables ( Šmilauer & Lepš, 2014). The level of specialisation of individual ant species to particular nest sites and tree species was assessed using two host-specificity measures, Llloyd´s index and HS k , since both have previously been used for analogous analyses of the interactions between herbivorous insects and their host trees ( Novotny et al., 2004;Wardhaugh et al., 2013). Both indices are frequency-based measures that depend on the distribution of individual ant species across the respective categories (nest sites or tree species), where higher values indicate a higher specificity of the ant species to the tree host species or nest site type. ...
  • ... Br€ andle & Brandl, 2006; Ribeiro & Borges, 2010; Johnson et al., 2012; Parker et al., 2012 Parker et al., , 2015 Whitfeld et al., 2012). For example, in a phylogenetically diverse forest, every species or genus of tree could be an island for many species of phytophagous insects (Southwood & Kennedy, 1983; Novotny et al., 2004). Not only do species differ in their phylogenetic distance from other species in communities, within a given species, individuals also differ in their average phylogenetic distance to their local neighbours, which we call 'degree of phylogenetic isolation of host individuals'. ...
  • ... The overall similar richness of specialist enemies in rare and common tree species was however surprising given the anticipated higher resource and evolutionary costs required to specialize on rare hosts (Jaenike 1990; Barrett and Heil 2012;Forister et al. 2012;Wardhaugh 2014). Enemies might have evolved specialized attributes to enable them to detect and overcome the defenses developed by rare hosts, as it is the case in some Lepidoptera species (Courtney and Courtney 1982), particularly in highly diverse ecosystems that exhibit high levels of enemy specialization (Novotny et al. 2004;Forister et al. 2015but see Morris et al. 2014). One potential hypothesis of enemy specialization on rare plants is that such strategy would allow enemies to escape their predators (Enemyfree space hypothesis, Jeffries and Lawton 1984). ...
  • ... The host plant explained 56 percent of the changes in lepidopteran species composition. Lepidopterans exhibit very high specificity to their host plants (Novotny et al. 2004, Dyer et al. 2007 ), as host plant selection is mediated by plant chemical compounds (reviewed in Thompson & Pellmyr 1991). For instance, Novotny et al. (2004) showed that the probability that caterpillars in tropical forests have a single host plant is more than 90 percent . ...
    ... Lepidopterans exhibit very high specificity to their host plants (Novotny et al. 2004, Dyer et al. 2007 ), as host plant selection is mediated by plant chemical compounds (reviewed in Thompson & Pellmyr 1991). For instance, Novotny et al. (2004) showed that the probability that caterpillars in tropical forests have a single host plant is more than 90 percent . Although caterpillars are affected by plant chemical compounds (Thompson & Pellmyr 1991, Coley & Barone 1996), changes in plant architecture (e.g., leaf shape) could also affect the performance (e.g., movement, oviposition) of individual lepidopterans (Brown & Lawton 1991 and references therein). ...
    ... The host plant explained 56 percent of the changes in lepidopteran species composition. Lepidopterans exhibit very high specificity to their host plants (Novotny et al. 2004, Dyer et al. 2007), as host plant selection is mediated by plant chemical compounds (reviewed inThompson & Pellmyr 1991). For instance,Novotny et al. (2004)showed that the probability that caterpillars in tropical forests have a single host plant is more than 90 percent. ...
  • ... Microhabitat specialization was assessed using a measure of specialization based on the host specificity index described by Novotny et al. (2004). The microhabitat specificity index (S m ) (Novotny and Basset 1998; Wardhaugh et al. 2013) is a frequency-based measure that account for variation in abundance between microhabitats, representing an alternative approach to specificity measurement (Novotny and Basset 2005). ...
    ... By using S m index, it is possible to separate beetle species according to their microhabitat preferences based on the proportion of the total number of individuals collected from microhabitats that supported the highest number of individuals and reducing the bias caused by rare records (Wardhaugh et al. 2013). Since S m is sensitive to sample size (Novotny et al. 2004 ), we set an arbitrary N [ 10 and restricted all analyses to the beetle species represented in the sample by at least 10 individuals. S m index is calculated as: The S m index ranged from 1 for specialist species to 1/n for species equitably distributed on all microhabitats. ...
    ... S m index is calculated as: The S m index ranged from 1 for specialist species to 1/n for species equitably distributed on all microhabitats. Thus, twig-girdling beetles were grouped into three categories: (1) microhabitat specialists: species with S m [ 0.9; (2) species with microhabitat preference: 0.5 \ S m \ 0.9 because most individuals were collected from single microhabitats but are not necessarily specialists; and (3) microhabitat generalists: species where S m \ 0.5 because no single microhabitat supported a majority of the population (Novotny et al. 2004; Wardhaugh et al. 2013). Comparisons of the mean microhabitat specificity (S m ) of beetles from different microhabitats were carried out using analysis of variance. ...
  • ... In antagonistic systems coevolutionary processes will tend to reinforce constraints against grand generalists (only a small proportion of herbivorous animals are true generalists), thus compartments should often be apparent in such networks [9]. Since host use by phytophagous insects tend to specialization [28], compartmentation or modularity [8] and may be influenced by resource availability [6], [29] we hypothesized that: i) strong seasonal effects would be detectable at the structural parameters of plant-herbivore networks at the studied site, with higher selectiveness (network-level specialization) and modularity during the wet season and forest habitats, ii) host availability and interaction frequency would vary among habitats and temporal periods, thus we may expect iii) higher connectance, interaction evenness and lower selectiveness during dry season periods and open habitats. This study is unique in the sense that, for the first time, the impacts of resource availability, habitat structural complexity and seasonality on the structure (parameters) of an antagonistic network were assessed. ...
    ... In general, in order to estimate sampling completeness within interaction networks, the number (richness) of interactions accumulated as sampling effort increased is frequently used [49]. Since in our study the assemblage of lepidopteran herbivores showed a high proportion of rare species, the result is similar to other tropical communities of herbivorous insects [3], [50] where asymptotic interaction richness is not reached [15], [29]. This non-asymptotic pattern of accumulation curves occurs as well for lepidopteran species richness hosted even by plant species of the same genus or family [3], [51].Table 1. Results of the repeated measures analysis of variance (ANOVA) for the effect of census and vegetation type on structural network parameters (A) and the plant community attributes (B).Figure 1. Vegetation type and seasonality effect on network parameters and plant community attributes. ...
  • ... Even when the resource pool was included in phylogenetic measures of specialisation ( Weiblen et al. 2006;Vamosi et al. 2014), the exclusion of resource commonness and of frequency of consumers may still affect the estimates of specialisation, as we showed in our data set. Likewise, previous attempts to consider species frequencies and abundances ( Novotny et al. 2004) did exclude relatedness and in turn did not compare the observed patterns with random expectations. Other specialisation metrics that do not measure relatedness have been applied in interaction network studies, such as Bl€ uthgen et al.'s (2006) d'. ...
    ... Even when the resource pool was included in phylogenetic measures of specialisation ( Weiblen et al. 2006;Vamosi et al. 2014), the exclusion of resource commonness and of frequency of consumers may still affect the estimates of specialisation, as we showed in our data set. Likewise, previous attempts to consider species frequencies and abundances ( Novotny et al. 2004) did exclude relatedness and in turn did not compare the observed patterns with random expectations. ...
  • ... (1) habitat specialists: species with Sm [ 0.9; (2) species with habitat preference: 0.5 \ Sm \ 0.9 because most individuals were collected from single habitats but are not necessarily specialists; and (3) habitat generalists: species where Sm \ 0.5 because no single microhabitat supported a majority of the population ( Novotny et al. 2004). ...
  • ... Since Sm is sensitive to sample size, we restricted all analyses to the butterfly species represented in the sample by at least 5 individuals. Sm index was calculated as: Author's personal copy fruit-feeding butterflies were grouped into three categories: (1) habitat specialists: species with Sm [ 0.9; (2) species with habitat preference: 0.5 \ Sm \ 0.9 because most individuals were collected from single habitats but are not necessarily specialists; and (3) habitat generalists: species where Sm \ 0.5 because no single microhabitat supported a majority of the population (Novotny et al. 2004). Regression trees were modeled using STATISTICA v7.0 (StatSoft). ...
  • ... Since Sm is sensitive to sample size, we restricted all analyses to the butterfly species represented in the sample by at least 5 individuals. Sm index was calculated as: Author's personal copy fruit-feeding butterflies were grouped into three categories: (1) habitat specialists: species with Sm [ 0.9; (2) species with habitat preference: 0.5 \ Sm \ 0.9 because most individuals were collected from single habitats but are not necessarily specialists; and (3) habitat generalists: species where Sm \ 0.5 because no single microhabitat supported a majority of the population (Novotny et al. 2004). Regression trees were modeled using STATISTICA v7.0 (StatSoft). ...
Project
My goal is to determine the species richness of the host trees growing within the grass plains of the lower Koiari lands outside of the Port Moresby city. We are currently collecting all insects fo…" [more]
Project
A baseline study looking at the presence of anthropogenic, Caesium-137 in one of the staple greens in people's diets throughout Papua New Guinea especially in light of the 2011 Fukushima disaster a…" [more]
Article
December 2003 · Ecological Entomology
    1. Caterpillar assemblages feeding on two alien plants, Piper aduncum and P. umbellatum, were studied in lowland rainforest in Papua New Guinea and compared with assemblages from 69 species of native woody hosts, including congeneric P. macropiper.2. Species richness of caterpillars feeding on P. aduncum (29 species per 1500 m2 of foliage) and P. umbellatum (36 species) was higher than the... [Show full abstract]
    Article
      This is the first in a series of papers providing taxonomic data in support of ecological and biogeographic studies of moths in New Guinea. The primary study is an extensive inventory of the caterpillar fauna of a lowland rainforest site near Madang, Papua New Guinea, from 1994–2001. The inventory focused on the Lepidoptera com-munity on 71 woody plant species representing 45 genera and 23... [Show full abstract]
      Article
        This is the first in a series of papers providing taxonomic data in support of ecological and biogeographic studies of moths in New Guinea. The primary study is an extensive inventory of the caterpillar fauna of a lowland rainforest site near Madang, Papua New Guinea, from 1994-2001. The inventory focused on the Lepidoptera community on 71 woody plant species representing 45 genera and 23... [Show full abstract]
        Article
        February 2004 · Conservation Biology
          Local species diversity of insect herbivores feeding on rainforest vegetation remains poorly known. This ignorance limits evaluation of species extinction patterns following various deforestation scenarios. We studied leaf-chewing insects feeding on 59 species of woody plants from 39 genera and 18 families in a lowland rainforest in Papua New Guinea and surveyed all plants with a stem diameter... [Show full abstract]
          Discover more