Janzen–Connell effects and habitat-induced aggregation synergistically promote species coexistence

In 1987, in the first issue of Conservation Biology, Edward O. Wilson wrote about the "little things that run the world" – the importance and conservation of insects (Wilson, 1987). Readers of Insect Conservation and Diversity will no doubt be very familiar with the concept. Sadly, however, this perception is not as widely shared among the rest of the scientific community as it should be, and insects are still comparatively neglected as a prime focus of scientific investigations. For instance, if we look at the Thompson-Reuter impact factors (IF) of specialised scientific journals for 2017 (http://jcr.incites.thomsonreuters.com/JCRJournalHomeAction.action?year=&edition=&journal=#), the highest ranked journal dedicated to entomology, Annual Review of Entomology (IF = 13.860), is ranked 139th out of 122 271 journals. In comparison, our botanical colleagues fare somewhat better, with the highest ranked journal in plant sciences, Annual Review of Plant Biology (IF = 18.172), ranked 83rd overall. Insect Conservation and Diversity continues to be among the top journals in entomology (IF = 2.091; ranked 14th), but overall is ranked 4549th among the journals evaluated by Thompson-Reuter. There is certainly room for improvement, of course, but in general this reflects the large difference in the scale of endeavour across different scientific disciplines. Part of this challenge may be related to an imbalance in the ratio of funding afforded to invertebrate studies (Leather, 2009). We entomologists are acutely aware of inherent biases in conservation research. Vertebrate studies dominate the field (69% of papers versus 3% of described species) while invertebrate studies lag far behind (11% of papers versus 79% of species: Clark & May, 2002). This taxonomic chauvinism has been commented on and lamented upon many times (e.g. Leather, 2009 and references therein), including in one of our previous editorials (Leather et al., 2008). Moreover, current trends show no signs of improvement (Titley et al., 2017), and the imbalance against insect studies is becoming even more pronounced in tropical countries (Titley et al., 2017), where recent estimates suggest over 25 000 arthropod species occurring in just a few hectares of tropical rainforest (Basset et al., 2012). But these issues may not even be the most serious cause for concern. We argue here that the neglect of insects as study organisms has led to serious bias in our understanding of the functional ecology of ecosystems. In other words, ignorance of the identity and role of insects in ecosystems may seriously impede conclusions related to the true contribution that insects make to ecosystem functionality (Weisser & Siemann, 2008). We may be able to estimate indirectly the role of insects in ecological processes, but without a good knowledge of the identity and life history of the species responsible for these processes, our conclusions may be rather subjective. "Knowing the players" is therefore crucial for sound studies of the effects of insects on ecosystem functioning (Schmitz, 2008). This situation is particularly obvious in studies of insect-plant interactions (or should we say in this instance "plant-insect interactions"?), which represent a significant field of ecological research in its own right (Calatayud et al., 2018). Many plant science researchers in this field simply seem to ignore the identity and diversity of the types of insect species doing the work. For instance, given the difficulty in evaluating damage caused by sap-sucking insects, most studies of herbivory (leaf damage) only focus on the action of leaf-chewing insects. This is very evident in studies on herbivory carried out in tropical forests (e.g. Coley & Barone, 1996). Nevertheless, detailed studies have shown that the occurrence of sap-sucking insects on rainforest plants is by no means trivial (Novotny & Basset, 1998; Dem et al., 2013). Since these insects can be vectors of important plant diseases (Denno & Perfect, 2012), they could have a significant effect on rates of mortality of their hosts. In addition, most of the "plant-insect" literature has focused on insects feeding on leaves. Much less is known about the identities and roles of insects attacking other plant parts (e.g., flowers, fruits, roots, stems). Another important issue is the estimation of herbivory caused by leaf-chewing insects in tropical rainforests. Botanists have been keen to measure the area of holes in leaves (review in Coley & Barone, 1996) but few, if any, discuss the interpretation of their findings with regard to the identities and life histories of the main species responsible for leaf damage. Total leaf damage rates are often assumed to be correlated with insect species richness, abundance or biomass (e.g., Coley, 1983, discussing the spatial distribution of herbivory). The handful of studies that have, however, considered insect identity and associated variables (abundance, species richness, biomass) all concluded that leaf damage is likely to depend on the feeding behaviour of a few dominant leaf-chewing species and this may complicate the interpretation of results obtained in herbivory studies focusing on community-level patterns (e.g., Marquis, 1991; Basset & Höft, 1994). We know that the major impact of herbivores on plants, particularly in rainforests, is driven by relatively few insect species, because most of the rest are relatively rare and their action restricted in time (Owen, 1983; Bernays & Graham, 1988). Thus, while overall herbivory rates may be an important correlate of plant fitness, it gives us few clues about the distribution and feeding preferences of the species responsible for the leaf damage. In this essay, we briefly explore the implications of ignoring the identity and traits of insects in the context of another research topic popular among our botanical colleagues, the Janzen–Connell hypothesis, JCH (Janzen, 1970; Connell, 1971). The JCH proposes an explanation for the coexistence of tree species in diverse tropical forests. Seeds are most likely to disperse to sites close to their parent trees, but this is also where they are likely to be most frequently attacked by host-specific enemies such as insects and pathogens that might aggregate near the parent trees. By contrast, seeds and seedlings that do manage to disperse further away from the parent tree are more likely to survive due to escape from enemies. In other words, conspecific negative density-dependent survival results from the proliferation of species-specific herbivores and pathogens on hosts in areas of high conspecific plant densities, giving a negative correlation between relative pest attack rate and distance from parent trees to their nearby offspring (Janzen, 1970; Connell, 1971; Comita et al., 2010; Bagchi et al., 2014). In the seminal paper by Janzen (1970), few examples of insect species responsible for negative density dependence among rainforest plants are provided, but this information may be gathered from subsequent papers, along with more recent studies (Table 1). Most of the studies concerned with Janzen–Connell effects pay little attention to the identity of insects potentially able to induce such effects (reviews in Clark & Clark, 1984; Hammond & Brown, 1998; Carson et al., 2008; Comita et al., 2014: 63 studies considered). The compilation in Table 1 indicates that most studies that have assessed the role of specific insect species in causing patterns consistent with the predictions of the JCH were performed in the Neotropics (only one study originated from the Old World tropics), in rather open forests, savanna or even open pastures, targeted seeds over seedlings, often included palm or leguminous trees (64% of cases) and the main species responsible for Janzen–Connell effects were often bruchine beetles. One might be tempted to think that many of these study systems were perhaps selected for the ease of studying large seed crops attacked by noticeable seed predators. What is clear, is that more studies targeting closed tall forests, and trees from other plant families and their seedlings are urgently needed before we can make sweeping conclusions about the generality of Janzen–Connell effects induced specifically by insects. Another bias that is obvious from the studies listed in Table 1 is the almost exclusive focus on chewing insects attacking either seeds or seedlings. The only exception is an influential paper by Janzen in which he reports on the effects of an external-feeding sap-sucking bug on seeds of Sterculia apetala (Janzen, 1972a). Seed bugs (Lygaeidae and related families) are renowned as potentially important seed predators in the tropics (Slater, 1972 and references therein). Hence, it is also clear that if we are serious about evaluating potential Janzen–Connell effects induced by insects, it is imperative to pay more attention to the guild of externally seed- and fruit-sucking insects in rainforests. Janzen's study on seed mortality by seed-sucking bugs on Sterculia apetala also illustrates another potentially important point. Since the externally sap-sucking bug studied by Janzen may transmit a pathogenic fungus to the host tree (Janzen, 1972a), the ultimate cause of seed mortality might appear to be caused by a seed pathogen rather than by an insect. This illustrates the need to consider the synergy between insects and pathogens. As discussed by Carson et al. (2008), the JCH is ultimately a plant community-level hypothesis, but all the studies reported in Table 1 targeted a single plant species. While research within the framework of the JCH has mostly been conducted on enemies that attack seeds and seedlings that have already dispersed from the mother plant, Janzen (1970) also suggested that coexistence of plant species in tropical forests could also be promoted by pre-dispersal seed enemies (i.e., enemies attacking developing or mature seeds in the canopy). Gripenberg (2018), in stressing the need to pay attention to attack by pre-dispersal seed enemies, reviewed the studies that have assessed the pattern of insect seed predation in tropical forest plant communities. To date, this includes only 15 studies world-wide, from which just two thirds provide hard data about insects. Again, currently available data are so limited that we lack the necessary insect background to discuss adequately the contribution of insects to Janzen–Connell effects in tropical rainforests. What can we gain from knowing the identity and ecology of insects in studies of negative-density dependence in tropical rainforests? Primarily this includes information on patterns of host use (specificity) by specific insect species; information on whether the same insect species tend to feed on adult foliage and seedlings; and spatial patterns of foraging by insects. To address some of these issues briefly, we need to consider the separate effects of insects feeding on seeds versus seedlings. We know that most insects attacking seeds in rainforests are highly host specific (Janzen, 1980; Ctvrtecka et al., 2014; Gripenberg, 2018), in accordance with the expectations of the JCH. What is less well known is the degree of spatial contagion of seed predators near parent trees, which may depend on the ecology of species considered. For example, Janzen (1975b) reported that two species of bruchine beetles are host specific to the seeds of Guazuma ulmifolia in Costa Rica, with one being a pre-dispersal seed predator attacking the seeds on the tree, while the other exclusively attacks the mature seeds after they have fallen to the ground. Hence, the identity and ecology of insect species is crucial to fully understand patterns of pre- and post-dispersal seed attack and any resulting effects on plant fitness and patterns of recruitment. Even if the assumptions of host specificity and contagion near the parent trees are met, this does not imply that Janzen–Connell effects related to seeds may be pervasive. Insects need to subsist at minimum densities on their hosts in order to induce significant plant mortality. For example, in the forests of New Guinea 95% of the woody plant species sampled for seed-eating weevil and lepidopteran assemblages had low rates of seed infestation (Ctvrtecka et al., 2014; Sam et al., 2017). Here, a recognition of the main insect species and estimation of their infestation rates in seeds are needed before assessing possible Janzen–Connell effects induced by insects. Overlooking even the higher taxa of insects responsible for seed damage may lead to ambiguous interpretation of results. For example, Bruchinae are often host-specific on seeds of Fabaceae in the Neotropics (Janzen, 1980), whereas they are almost totally replaced by several less host-specific weevil subfamilies in the Old World (Ctvrtecka et al., 2014; Basset et al., 2018). The potential for Bruchinae to induce Janzen–Connell on their fabaceaous hosts is thus much higher than for weevils of the Old World, as suggested by Table 1. Furthermore, botanists pay considerable attention to plant phylogeny in studies of JCH, but they should also take note of plant traits that may explain oviposition patterns of insects attacking seeds, which are not necessarily related to plant phylogeny. One of the most important traits in this regard may be the degree of fleshiness of the fruit (Sam et al., 2017; Basset et al., 2018; C. Dahl et al., unpubl. data). When assessing the contributions of insects to Janzen–Connell effects, it is also important to have good insights into the feeding ecology of different taxa. Even in relatively well-known Lepidoptera, it can be difficult to separate the seed predator species from pulp eaters or scavengers. Several taxa that are often considered to be scavengers also contain lineages with other life history strategies, such as in the Tineidae (Robinson, 2009), so precise identification of insects reared from seeds or fruits is crucial. If we now turn our attention to seedlings, there are very few community-wide studies of insect herbivores attacking seedlings in tropical rainforests. Twenty years ago, one study in Guyana concluded that free living species attacking seedlings persisted at very low densities, were often generalists, and that Janzen–Connell effects mediated by insects feeding on seedlings were, consequently, unlikely to exist in the system studied (Basset, 1999). We now know that the lack of host specificity (particularly for insects feeding on seedlings) does not necessarily invalidate their potential contribution to plant species coexistence, as negative density dependence may also be generated by the action of generalist herbivores if they tend to be attracted to areas of high conspecific plant density (Lewis & Gripenberg, 2008). Regarding contagion from parent trees, we have noted that insect species responsible for Janzen–Connell effects were often studied in rather open forest or pastures (Table 1), and less so in closed tall forests. In fact, in these forests, where presumably Janzen–Connell effects induce high local diversity of trees (Janzen, 1970; Connell, 1971), contagion of insect herbivores from the parent trees to seedlings has rarely been demonstrated. This may be because the biotic and abiotic conditions experienced in the canopy versus understorey of forests are strikingly different, resulting in different suites of free-living herbivores attacking plants in these two strata. These differences have been observed both at the level of host plant species (e.g., Basset, 2001) and the plant community as a whole (Basset et al., 2015). There may of course be exceptions and they are more likely to involve endophagous insects (stem borers, gallers, miners) than ectophagous insects, because external conditions induced by the forest strata may be buffered to some extent by microclimatic conditions inside the host tissues. Nevertheless, the proportion of host tree species studied that supported the same insect species of either gallers or miners in both the canopy and understorey in one Panamanian wet forest was low and amounted to only 6% (out of 18 species: Medianero et al., 2003). Under these conditions, contagion of insect herbivores from parent trees to seedlings is likely to be rather uncommon in closed tall rainforests. Despite claims that in some instances signs of leaf damage can be unequivocally assigned to particular insect species (Barone, 2000; Downey et al., 2018), in our experience it is nearly impossible to do so for the vast majority of the diverse insect species feeding on the leaves of tropical trees and seedlings, particularly in the case of generalist species. This greatly impedes our ability to investigate the causal mechanisms of negative density dependence in seedlings of tropical rainforests. Moreover, one recent study suggested that the amount and categories of herbivore damage on rainforest seedlings may even differ between continents. For example, the percentage of damage on seedlings that could be assigned to insects represented 56%, 78%, and 85% of observations in rainforests in Panama, Thailand, and Papua New Guinea, respectively (Y. Basset et al., unpubl. data). Identifying the main herbivore species responsible for such variation in herbivory (at least leaf-chewing herbivory) is crucial. And, of course, the degree to which seedlings of different plant species can tolerate differing levels of herbivory before Janzen–Connell effects are triggered is an open question. If we do entertain the idea that at least some insect species are responsible for some examples of negative-density dependence observed in rainforests (review in Comita et al., 2010), then which taxa are most likely to be responsible for these effects? If we consider post-dispersal attack of seeds fallen on the ground, then highly host-specific Bruchinae (Janzen, 1980) and perhaps certain Curculionidae (Pinzón-Navarro et al., 2010) may fit the bill, although many species may only be involved in pre-dispersal attack. We should also not underestimate ants as seed removers in rainforests (Ruzi et al., 2017), and therefore as possible engineers of Janzen–Connell effects. Insect herbivores attacking seedlings in rainforests involve many taxa (Basset & Charles, 2000). Leaf-chewing insects are often represented by Chrysomelidae, leaf-feeding weevils (Entiminae), but Lepidoptera larvae are relatively rare on seedlings (e.g., 6% of the total insect individuals collected in Basset & Charles, 2000). Orthoptera and Phasmatodea are also rather infrequent, at least during day-time censuses (Basset & Charles, 2000). The low incidence of most of these insects on seedlings (Basset, 1999) makes them unlikely candidates to successfully induce Janzen–Connell effects, but exceptions may exist. Further cases of insects notoriously dangerous for the survival of seedlings are worth discussing briefly. First, the action of potential vectors of phytopathogens needs to be quantified and understood. This includes, for example, xylem-feeding and generalist Cicadellinae, which are common as nymphs and adults in the understorey of tropical rainforests, and are able to transmit phytopathogenic viruses (Nielson, 1968). Additionally, this may involve adult weevils (for example Conotrachelus spp.) or bark beetles, which attack seeds at the larval stage and perform maturation feeding on seedlings as adults (Basset & Charles, 2000). In this situation, they may transmit pathogenic fungi, as for example in the case of Dutch elm disease (Martín et al., 2018). Second, insects damaging meristems may be particularly threatening, such as one erebid moth decapitating seedlings in Costa Rica (Janzen, 1971b). In Panama, this category of damage represents nearly 20% of all observations of seedlings damaged in a community study (Y. Basset et al., unpubl. data). Lepidopterous stem borers may also damage meristems but this group is far less diverse than free-feeding caterpillars, so it may be relatively easy to quantify their effects on particular host species (e.g., Sullivan, 2003). Last, insects able to completely defoliate seedlings are also of concern. This may include outbreaks of host-specific Lepidoptera (Barone, 2000), but this situation is rather rare in tropical rainforests. Large generalist caterpillars such as Saturniidae (Hartnett et al., 2012) may be worth investigating in this context. In conclusion, Janzen–Connell effects mediated by insects in tropical rainforests appear to be less likely by contagion of host-specific species from parent trees to seedlings, but more likely via a combination of escape of seeds from pre-dispersal attack (Lawson et al., 2012), and attack of seedlings by generalist herbivores in the forest understorey, possibly aggravated by transmission of diseases by insect vectors. To collect and identify the culprits of damage is challenging, particularly on seedlings, because generalists may subsist at low densities (Basset, 1999) or specialists may have elusive behaviours. For example, Janzen (1971b), estimated that on average just 10 minutes were necessary for an erebid moth to decapitate one seedling before walking off, rendering any direct census of caterpillars in this study system very difficult. Elegant experiments with insecticide or exclusion of insect herbivores may help us to quantify the action of insect herbivores more effectively (e.g., Bagchi et al., 2014) and those results should be coupled with good old-fashioned natural history observations, or with observations acquired with new technologies. For example, the metabarcoding of the gut of potential insect herbivores (e.g., García-Robledo et al., 2013) or automatic detection of insect activity (e.g., Reynolds & Riley, 2002) on seedlings, particularly at night, appear to be promising opportunities in this context. Further, such studies may be performed at locations where extensive vegetation data, including the basal area, spatial location, and seed production of parent trees, may be available, such as in the ForestGEO network of permanent forest plots (Anderson-Teixeira et al., 2015; Basset et al., 2018). New tools, such as DNA barcoding, are now available to assist with rapid and accurate identification of insect species (Miller, 2014), including the BIN clustering algorithm and interim nomenclature system, which facilitates forming putative species concepts and communicating about them (Schindel & Miller, 2010; Ratnasingham & Hebert, 2013). We hope that we may have convinced our non-entomologist readers, perhaps curious about the title of this essay, of the value of paying attention to the identity of insects potentially responsible for Janzen–Connell effects in rainforests, and, to this effect, to collaborate with entomologists. Hopefully, some of our regular readers may also see better scope for collaboration with botanists or forest ecologists regarding this fascinating topic. The ideas advanced in this essay were shaped by a project supported by the Czech Science Foundation (GAČR 16-20825S) and a grant from the US National Science Foundation (DEB 0841885).

Janzen–Connell effects (JCEs), specialised predation of seeds and seedlings near conspecific trees, are hypothesised to maintain species richness. While previous studies show JCEs can maintain high richness relative to neutral communities, recent theoretical work indicates JCEs may weakly inhibit competitive exclusion when species exhibit interspecific fitness variation. However, recent models make somewhat restrictive assumptions about the functional form of specialised predation—that JCEs occur at a fixed rate when offspring are within a fixed distance of a conspecific tree. Using a theoretical model, I show that the functional form of JCEs largely impacts their ability to maintain coexistence. If predation pressure increases additively with adult tree density and decays exponentially with distance, JCEs maintain considerably higher species richness than predicted by recent models. Loosely parameterising the model with data from a Panamanian tree community, I elucidate the conditions under which JCEs are capable of maintaining high species richness.

Dispersal vacuum in the seedling recruitment of a primate-dispersed Amazonian tree

Janzen–Connell effects are negative effects on the survival of a plant's progeny at high conspecific densities or close to its conspecifics. Although the role of Janzen–Connell effects on the maintenance of plant diversity was frequently studied, only few studies targeted Janzen–Connell effects via postdispersal seed predation in temperate grassland systems. We examined effects of conspecific density (abundance of conspecific adult plants) on postdispersal seed predation by invertebrates of three grassland species (Centaurea jacea, Geranium pratense, and Knautia arvensis) in experimental plant communities. Additionally, we examined the impact of plant species richness and different seed predator communities on total and relative seed predation (= seed predation of one plant species relative to others). We offered seeds in an exclusion experiment, where treatments allowed access for (1) arthropods and slugs, (2) arthropods only, (3) small arthropods only, and (4) slugs only. Treatments were placed in plots covering a gradient of abundance of conspecific adults at different levels of plant species richness (1, 2, 3, 4, 8 species). Two of the plant species (C. jacea and K. arvensis) experienced higher rates of seed predation and relative predation with increasing abundance of conspecific adults. For C. jacea, this effect was mitigated with increasing plant species richness. Differences in seed predator communities shifted seed predation between the plant species and changed the magnitude of seed predation of one plant species relative to the others. We exemplify density‐dependent increase in seed predation via invertebrates in grassland communities shaping both the total magnitude of species‐specific seed predation and seed predation of one species relative to others. Further differences in seed predator groups shift the magnitude of seed predation between different plant species. This highlights the importance of invertebrate seed predation to structure grasslands via density‐dependent effects and differing preferences of consumer groups.

Abstract:A community of frugivorous weevils was studied by quantitative rearing of 57 weevil species represented by 10485 individuals from 326 woody plant species in lowland rain forest in Papua New Guinea. Only fruits from 35% of plant species were attacked by weevils. On average, weevils were reared from only 1 in 33 fruits and 1 kg of fruit was attacked by 2.51 individuals. Weevil host specificity was relatively high: 42% of weevil species fed on a single plant genus, 19% on a single plant family and only 16% were reared from more than one family. However, monophagous specialists represented only 23% of all reared individuals. The average 1 kg of fruits was infested by 1.84 individuals of generalist weevils (feeding on allogeneric or allofamilial host species), 0.52 individual of specialists (feeding on a single or several congeneric species), and 0.15 individual of unknown host specificity. Large-seeded fruits with thin mesocarp tended to host specialist species whereas those with thick, fleshy mesocarp were often infested with both specialists and generalists. Weevils tended to avoid small-seeded, fleshy fruits. The low incidence of seed damage (3% of seeds) suggests that weevils are unlikely to play a major role in regulating plant populations via density-dependent mortality processes outlined by the Janzen–Connell hypothesis.

Density‐dependent recruitment is fundamental to understanding species diversity and community dynamics in plants. Although there is compelling evidence that seeds and seedlings die from conspecific negative density dependence (CNDD) as predicted by the Janzen–Connell hypothesis, characterising adult recruitment remains a challenge for long‐living trees. Previous studies have used the decrease in fine‐scale spatial genetic structure (FSGS) across life stages to indicate CNDD; however, this has not been tested rigorously. We addressed these challenges by integrating dispersal kernels and FSGS. To establish links between density dependence and FSGS, we simulated seedlings based on the estimated dispersal kernels from parentage analyses, and further simulated adults under various seedling‐to‐adult recruitment scenarios, using an individual‐based spatially explicit model. We tested this method in an isolated Cyclobalanopsis glauca population on China's Dajinshan Island. We detected significant FSGS in the seedlings and weaker, though also significant, FSGS in the adults. As expected, the observed FSGS of seedlings was well predicted by the simulated seedlings, with observations falling inside the 95% confidence envelopes over all distance classes. However, the simulations showed that CNDD enhanced the FSGS while positive density dependence dampened it during the seedling‐to‐adult transition. The adult FSGS of our population was therefore explained by positive rather than negative density‐dependent adult recruitment. Synthesis. Our study demonstrates that the change of FSGS in conjunction with dispersal‐based model tests can offer a valuable insight into density‐dependent adult recruitment. The results indicate that the transitions from seedlings to adults in C. glauca are dominantly regulated not by Janzen–Connell effects, but by processes of positive density dependence. More broadly, the findings may provide a caution against extrapolations of widespread Janzen–Connell effects in seeds and seedlings to adult recruits, underscoring a critical gap between mechanisms at early stages and long‐term population and community dynamics.

Summary Although plants can reduce the impacts of herbivory in multiple ways, these defensive traits are often studied in isolation and an understanding of the resulting strategies is incomplete. In the study reported here, empirical evidence was simultaneously evaluated for the three main sets of traits available to plants: (i) resistance through constitutive leaf traits, (ii) tolerance to defoliation and (iii) escape in space, for three caesalpiniaceous tree species Microberlinia bisulcata, Tetraberlinia bifoliolata and T. korupensis, which co‐dominate groves within the lowland primary rain forest of Korup National Park (Cameroon). Mesh cages were placed around individual wild seedlings to exclude insect herbivores at 41 paired canopy gap and understorey locations. After following seedling growth and survival for c. 2 years, caged and control treatments were removed, leaves harvested to determine nutrient and phenolic concentrations, leaf mass per area estimated, and seedling performance in gaps followed for a further c. 2 years to quantify tolerance to the leaf harvesting. The more nutrient‐rich leaves of the weakly shade‐tolerant M. bisulcata were damaged much more in gaps than the two strongly shade‐tolerant Tetraberlinia species, which had higher leaf mass per area and concentrations of total phenols. Conversely, the faster‐growing M. bisulcata was better able to tolerate defoliation in terms of height growth (reflushing capacity), but not at maintaining overall leaf numbers, than the other two species. Across gaps, insect‐mediated Janzen–Connell effects were most pronounced for M. bisulcata, less so for T. korupensis, and not detectable for T. bifoliolata. The three species differed distinctly in their secondary metabolic profiles. Taken together, the results suggested a conceptual framework linking the three sets of traits, one in which the three co‐dominant species adopt different strategies towards herbivore pressure depending on their different responses to light availability. This study is one of the first in a natural forest ecosystem to examine resistance to, tolerance of, and escape from herbivory among a group of co‐occurring tropical tree species.

Seed-predator satiation and Janzen-Connell effects vary with spatial scales for seed-feeding insects.

The spatial placement of recruits around adult conspecifics represents the accumulated outcome of several pattern-forming processes and mechanisms such as primary and secondary seed dispersal, habitat associations or Janzen–Connell effects. Studying the adult–recruit relationship should therefore allow the derivation of specific hypotheses on the processes shaping population and community dynamics. We analysed adult–recruit associations for 65 tree species taken from six censuses of the 50 ha neotropical forest plot on Barro Colorado Island (BCI), Panama. We used point pattern analysis to test, at a range of neighbourhood scales, for spatial independence between recruits and adults, to assess the strength and type of departure from independence, and its relationship with species properties. Positive associations expected to prevail due to dispersal limitation occurred only in 16% of all cases; instead a majority of species showed spatial independence (≈73%). Independence described the placement of recruits around conspecific adults in good approximation, although we found weak and noisy signals of species properties related to seed dispersal. We hypothesize that spatial mechanisms with strong stochastic components such as animal seed dispersal overpower the pattern-forming effects of dispersal limitation, density dependence and habitat association, or that some of the pattern-forming processes cancel out each other.

The Janzen–Connell hypothesis proposes that specialized herbivores maintain high numbers of tree species in tropical forests by restricting adult recruitment so that host populations remain at low densities. We tested this prediction for the large timber tree species, Swietenia macrophylla, whose seeds and seedlings are preyed upon by small mammals and a host‐specific moth caterpillar Steniscadia poliophaea, respectively. At a primary forest site, experimental seed additions to gaps – canopy‐disturbed areas that enhance seedling growth into saplings – over three years revealed lower survival and seedling recruitment closer to conspecific trees and in higher basal area neighborhoods, as well as reduced subsequent seedling survival and height growth. When we included these Janzen–Connell effects in a spatially explicit individual‐based population model, the caterpillar's impact was critical to limiting Swietenia's adult tree density, with a > 10‐fold reduction estimated at 300 years. Our research demonstrates the crucial but oft‐ignored linkage between Janzen–Connell effects on offspring and population‐level consequences for a long‐lived, potentially dominant tree species.

Tropical forest community ecology

The loss of large frugivores leads to seed dispersal loss and regeneration failure of numerous large-seeded trees near mother trees. Although Janzen–Connell effects are considered as the primary underlying cause, other factors remain understudied. Here, we used a field experiment to test the impact of flesh persistence on the recruitment of two large-seeded Sapotaceae species that lost their dispersers. In the rainforest of Mare Longue (Réunion), we sowed 3840 seeds in a four-factor design: seed treatment (seed cleaning; flesh persistence), canopy closure (understory; gap), year of sowing (01/2018; 11/2019) and species (Labourdonnaisia calophylloides, Mimusops balata). We also used camera traps to evaluate the impact of extant vertebrates. Seed treatment was by far the most influential factor: flesh persistence led to seedling recruitment divided by 3,2 on average, mainly due to failure of germination or seedling emergence. There were also significant variations in recruitment between species, years and canopy closure levels, notably due to the behaviour of the invasive fauna, especially giant snails that could unexpectedly restore recruitment by feeding on fruit flesh. Together, our results demonstrate strongly depleted recruitment due to flesh persistence and the importance of field experiments to understand the processes at work in complex ecosystems with novel plant–animal interactions.

The Janzen–Connell hypothesis proposes that density dependent seed and seedling mortality, combined with increasing seed and seedling survival away from the parent tree, together promote regular spacing of species and thus α diversity. This hypothesis has rarely been tested in tropical Africa, and rarely in montane forests anywhere. We tested this hypothesis using a combination of field experiments and observations in the most floristically diverse dry submontane forest in Nigeria. We investigated distance effects on seedling herbivory, seedling survival and seedling height growth. We found a significant decrease in herbivory with distance from conspecific adult trees for all three species of experimentally planted seedlings (Entandrophragma angolense, Deinbollia pinnata and Sterculia setigera), and also for naturally occurring seedlings of Pouteria altissima but not of Newtonia buchananii or Isolona pleurocarpa. The relative density of large seedlings/saplings of P. altissima, N. buchananii and I. pleurocarpa increased significantly at greater distance from conspecific adult trees; however, we found no significant distance effect on survival or height growth over 3 mo for all three experimentally planted species. Taken together, our results are some of the first to show that Janzen–Connell effects occur on the African continent and in particular montane tropical forest and suggest that such effects may be pantropical.

The Janzen–Connell Hypothesis (JCH) predicts that density‐responsive and host‐specific natural enemies limit the population sizes of abundant species. Importantly, these interactions help to maintain local community diversity through time. While ample evidence exists for the demographic predictions of the JCH, it remains unclear which natural enemies drive these dynamics across different plant communities. While large mammalian herbivores are often assumed to lack the specialized diet needed to drive Janzen–Connell effects, they do show a degree of host‐preference that could drive density‐dependent plant demography. However, the potential role of large mammalian herbivores in Janzen–Connell interactions has only rarely been investigated. Using 204 seedling transects (1 m × 10 m) at 51 sites across a 900‐ha forested reserve in southwestern Pennsylvania (USA), we examined the role that large mammals play in driving conspecific negative density dependence (CNDD) in temperate tree seedlings. Individual fences were erected around half of the transects ( n = 102) to exclude large mammals, and were paired with adjacent unfenced transects. Within transects, a total of ~15,000 individual seedlings were monitored over three growing seasons. Demographic neighbourhood models were constructed to examine the influence of neighbourhood composition and density on seedling survival and growth. An interaction term between conspecific neighbour density and fencing treatment was included to test the hypothesis that large herbivores cause CNDD. We found that seedling survival was influenced by both conspecific neighbour density and fencing. CNDD was strongest when large mammals were allowed access to seedlings, and these results were driven by two abundant taxa ( Prunus serotina and Fraxinus spp). Despite evidence that large mammals mediate CNDD, we found no effect of fencing on rarified species richness or evenness in seedling transects during the study. Synthesis . Understanding the specific natural enemies driving conspecific negative density dependence remains vital for understanding the maintenance of forest diversity across the globe. Our results indicate that large mammalian herbivores are capable of driving CNDD in temperate tree species. These results suggest that large mammals may be an important and generally overlooked agent contributing to Janzen–Connell interactions in forest communities. We expect that further research examining large mammals in other systems will be important in the future.

The Janzen-Connell (J-C) hypothesis suggests that specialised natural enemies cause distance- or density-dependent mortality among host plants and is regarded as an important mechanism for species coexistence. However, there remains debate about whether this phenomenon is widespread and how variation is structured across taxa and life stages. We performed the largest meta-analysis of experimental studies conducted under natural settings to date. We found little evidence of distance-dependent or density-dependent mortality when grouping all types of manipulations. Our analysis also reveals very large variation in response among species, with 38.5% of species even showing positive responses to manipulations. However, we found a strong signal of distance-dependent mortality among seedlings but not seed experiments, which we attribute to (a) seedlings sharing susceptible tissues with adults (leaves, wood, roots), (b) seedling enemies having worse dispersal than seed enemies and (c) seedlings having fewer physical and chemical defences than seeds. Both density- and distance-dependent mortality showed large variation within genera and families, suggesting that J-C effects are not strongly phylogenetically conserved. There were no clear trends with latitude, rainfall or study duration. We conclude that J-C effects may not be as pervasive as widely thought. Understanding the variation in J-C effects provides opportunities for new discoveries that will refine our understanding of J-C effects and its role in species coexistence.