ArticlePDF Available

Abstract and Figures

Insects are key components of urban ecological networks and are greatly impacted by anthropogenic activities. Yet, few studies have examined how insect functional groups respond to changes to urban vegetation associated with different management actions. We investigated the response of herbivorous and predatory heteropteran bugs to differences in vegetation structure and diversity in golf courses, gardens and parks. We assessed how the species richness of these groups varied amongst green space types, and the effect of vegetation volume and plant diversity on trophic- and species-specific occupancy. We found that golf courses sustain higher species richness of herbivores and predators than parks and gardens. At the trophic- and species-specific levels, herbivores and predators show strong positive responses to vegetation volume. The effect of plant diversity, however, is distinctly species-specific, with species showing both positive and negative responses. Our findings further suggest that high occupancy of bugs is obtained in green spaces with specific combinations of vegetation structure and diversity. The challenge for managers is to boost green space conservation value through actions promoting synergistic combinations of vegetation structure and diversity. Tackling this conservation challenge could provide enormous benefits for other elements of urban ecological networks and people that live in cities.
Content may be subject to copyright.
1
Scientific RepoRts | 7:40970 | DOI: 10.1038/srep40970
www.nature.com/scientificreports
Conserving herbivorous and
predatory insects in urban
green spaces
Luis Mata1, Caragh G. Threlfall2, Nicholas S. G. Williams2,3, Amy K. Hahs3, Mallik Malipatil4,
Nigel E. Stork5 & Stephen J. Livesley2
Insects are key components of urban ecological networks and are greatly impacted by anthropogenic
activities. Yet, few studies have examined how insect functional groups respond to changes to urban
vegetation associated with dierent management actions. We investigated the response of herbivorous
and predatory heteropteran bugs to dierences in vegetation structure and diversity in golf courses,
gardens and parks. We assessed how the species richness of these groups varied amongst green
space types, and the eect of vegetation volume and plant diversity on trophic- and species-specic
occupancy. We found that golf courses sustain higher species richness of herbivores and predators than
parks and gardens. At the trophic- and species-specic levels, herbivores and predators show strong
positive responses to vegetation volume. The eect of plant diversity, however, is distinctly species-
specic, with species showing both positive and negative responses. Our ndings further suggest that
high occupancy of bugs is obtained in green spaces with specic combinations of vegetation structure
and diversity. The challenge for managers is to boost green space conservation value through actions
promoting synergistic combinations of vegetation structure and diversity. Tackling this conservation
challenge could provide enormous benets for other elements of urban ecological networks and people
that live in cities.
Urbanisation has caused, and is forecasted to increasingly cause, global detrimental impacts on biodiversity1,2.
Yet, a mounting body of evidence suggests that urban environments can still support substantial levels of native
biodiversity including many threatened species3. Cities therefore provide unique opportunities to proactively
implement actions and strategies to conserve biodiversity. These actions may also have significant benefits
for people, as biodiverse urban ecosystems are known to improve the health and wellbeing of city-dwellers4.
However, successful conservation and management strategies relevant to a range of functionally dierent taxa are
yet to be devised, and much guidance is still required for conservation practice to realise the opportunities that
biodiverse urban areas could provide.
Insects are a key component of urban biodiversity5, and the ecological functions they perform translate into
a wide array of ecosystem services6,7 as well as disservices8,9. In forest ecosystems, Ewers et al.10 showed that the
contribution of insect and other invertebrate taxa to litter decomposition, seed predation and invertebrate pre-
dation was halved following logging because of a signicant decrease in invertebrate abundance. Furthermore,
insects play a key ecological role as prey for other taxa - especially insectivorous birds, reptiles and microbats.
In urban ecosystems, it is poorly understood how insect diversity is aected by changes to vegetation during
and aer urbanisation. Investigating how key functional groups, such as herbivores and predators11, respond to
changes in vegetation structure and diversity may help develop this understanding12 and help determine what
strategies may best conserve these functional groups and the ecosystem services they provide.
1Interdisciplinary Conservation Science Research Group, School of Global, Urban and Social Studies, RMIT University,
Melbourne 3000, Victoria, Australia. 2School of Ecosystem and Forest Sciences, Faculty of Science, The University
of Melbourne, Richmond 3121, Victoria, Australia. 3Australian Research Centre for Urban Ecology, Royal Botanic
Gardens Victoria c/o School of BioSciences, The University of Melbourne, Parkville 3010, Victoria, Australia.
4Department of Economic Development, Jobs, Transport and Resources, AgriBio, La Trobe University, Bundoora
3083, Victoria, Australia. 5Environmental Futures Research Institute, Griffith School of Environment, Griffith
University, Nathan 4111, Queensland, Australia. Correspondence and requests for materials should be addressed to
L.M. (email: luis.mata@rmit.edu.au)
Received: 09 September 2016
Accepted: 13 December 2016
Published: 19 January 2017
OPEN
www.nature.com/scientificreports/
2
Scientific RepoRts | 7:40970 | DOI: 10.1038/srep40970
Heteropteran bugs (Hemiptera: Heteroptera; henceforth bugs for brevity) comprise a hyperdiverse mono-
phyletic clade of insects distributed worldwide13,14. Bugs present a wide range of feeding strategies from strict
phytophagy and zoophagy to omnivory15,16, making them a suitable model taxon to better understand responses
of insect herbivores and predators to gradients in urban vegetation structure and diversity. Bugs also provide
important ecosystem services. For example, generalist predators such as damsel, pirate and assassin bugs (fam-
ilies: Nabidae, Anthocoridae and Reduviidae) are biological control agents in forest and agricultural ecosys-
tems15,17. Finally, some herbivorous bug species are highly host plant specic. For example, 60% of mirid bugs
(family Miridae) are associated exclusively with a single host plant and less than 20% occur in more than two
host plants18. e availability of suitable host plants is therefore an a priori requirement for the occurrence of bug
herbivore specialists within a given ecosystem.
e positive eects of complex vegetation structure and high plant diversity on insect diversity have been
previously documented19–21, and specically bug diversity22. Importantly however, the positive eects of complex
vegetation structure and high plant diversity may not be general across all insect taxa and functional groups23,
highlighting the relevance of incorporating trophic- and species-specic responses when investigating the gen-
erality of ecological patterns across dierent ecosystem types.
A mounting body of evidence indicates that managing vegetation structure and plant diversity in urban envi-
ronments can have positive eects on biodiversity at the landscape level24,25. Yet, in urban landscapes it is not
known which vegetation management actions can promote animal biodiversity in dierent green space types.
Urban green spaces, such as golf courses, public parks and residential gardens, play a crucial role in urban biodi-
versity conservation26–30. ese urban green spaces, however, are oen unintentionally managed and contain a
range of both early- and late-succession vegetation features (e.g. turf grass lawns, patches of unmanaged vegeta-
tion, trees, shrubs). Understanding how this diversity of habitat structures impacts insects and other animal taxa
will inform potential management practices that could promote biodiversity.
In this study we assess the impact of dierent vegetation management practices on herbivorous and predatory
insects by examining heteropteran bug responses to variation in vegetation structure and plant diversity in dier-
ent urban green space types (golf courses, residential gardens, public parks). Specically, we use multi-species site
occupancy models under a Bayesian mode of inference to:
(1) Assess how species richness of herbivorous and predatory bugs varies amongst green space types; and
(2) quantify the magnitude of the eect of vegetation volume and plant species diversity on bug trophic (i.e.,
herbivorous or predatory) and species-specic occupancy.
We also examined the role that habitat area has on bug diversity by assessing the t of the data to the power
function of species-area relationships31. e results suggest that area had a positive eect on bug diversity, but was
not the key driver of diversity. Instead, vegetation volume and plant diversity across habitat types were the major
factors in predicting bug diversity.
Results
e survey yielded 91 bug species (75 herbivores and 16 predators) from 19 families (Tables S1–S4). is rep-
resents approximately 20% of the total bug gamma diversity estimated for Victoria, Australia32, and agrees well
with the bug species richness found in other temperate urban areas33. As many as 98% of all species recorded were
native to the study area. Only two species, the green stink bug Nezara viridula and the Azalea lacebug Stephanitis
pyrioides, were non-native to Australia. Of the total number of species, 38 were unique to golf courses whilst only
six and eleven were unique to parks and gardens, respectively. Eleven herbivores were observed in all green space
types, with the most ubiquitous species being the alydid Mutusca brevicornis, occurring in 54% of all plots. By
contrast, only two predators were observed in all green space types, with the most ubiquitous species being the
Pacic damselbug Nabis kinbergii, occurring in 27% of all plots. An assessment of sample completeness showed
that further sampling would have resulted in little increase in sample coverage (Fig. S1).
Species-area model. Area (A) had a positive eect on bug species richness (S), with model estimates indi-
cating that the mean t of our data to the power function of the species-area relationship followed:
=. .
S026A
032
e mean estimate of the intercept parameter c (0.26) was associated with a wide 95% credible interval (hence-
forth CI95%) that ranged from 0.08 to 0.64, whereas the mean estimate of the slope parameter z (0.32) showed a
more accurate CI95% that ranged from 0.23 to 0.40. e observed vs. estimated t of our data to the power function
of the species-area relationship was at best intermediate (r2 = 0.43; Fig. S2). ere were more observed bug species
than those expected by the modelled species-area relationship in approximately 40% of golf course and park sites,
while in as much as 90% of garden sites there were less observed species than those predicted by the model.
Green space type model. e mean probability of occurrence for bugs was high in all green space types,
with species estimated to occur at 62% to 86% of sites. By contrast, the mean probability of detection was low, with
species estimated to be observed only 4% to 13% of the times when they were present. ese trends were consist-
ent across all green space types, as evidenced by the uncertainty associated with the mean responses (Table1).
e mean bug species richness was substantially higher in golf courses (57 spp.) than in parks (21 spp.)
and gardens (24 spp.). e golf course estimate was associated with a CI95% that did not overlap the CIs95% of
either parks or gardens (Table1). e mean estimated species richness for herbivores was higher in golf courses
(48 spp.) than in parks (18 spp.) and gardens (19 spp.) (Fig.1), with the golf course CI95% not overlapping the
CIs95% of either parks or gardens (Table1). Mean estimated species richness for predators was higher in golf
www.nature.com/scientificreports/
3
Scientific RepoRts | 7:40970 | DOI: 10.1038/srep40970
courses (9 spp.) than in parks (3 spp.) and gardens (5 spp.) (Fig.1), with the golf course CI95% not overlapping
the parks CI95% but slightly overlapping the gardens CI95% (Table2). Posterior estimates for the mean, standard
deviation and CI95% for the species-specic probabilities of occurrence and detection for each green space are
given in Table S1.
Trophic-level model. e mean probabilities of occurrence for herbivorous and predatory bugs were mod-
erately high, with herbivores and predators estimated to occur on average at 56% and 71% of sites, respectively.
In contrast, mean probabilities of detection were low for both trophic groups, with bug species estimated to be
observed on average only 3% (predators) to 4% (herbivores) of the times when they were present. ese trends
were consistent across all green space types, as evidenced by the uncertainty associated with the mean responses
(Table2).
e species-specic probabilities of occurrence for herbivores varied considerably, with individual species
estimated to occur in as few as 9% of sites (Cuspicona sp. 2) and in as many as 96% of sites (Nysius caledoniae).
In contrast, the species-specic probabilities of occurrence for predators varied only moderately, with individual
species estimated to occur between 54% (Dicrotelus prolixus) and 94% (N. kinbergii) of sites.
e species-specic probabilities of detection for herbivores also varied considerably, with individual species
estimated to be observed between 1% (N. caledoniae) and 70% (Melanocanthus scutellaris) of the times when
they were present. On the other hand, the variation in the species-specic probabilities of detection for predators
was much less pronounced, with species estimated to be observed between 1% (D. prolixus) and 29% (Gminatus
australis) of the times when they were present. Posterior estimates for the mean, standard deviation and CI95% for
the species-specic probabilities of occurrence and detection for herbivores and predators are given in Table S2.
Trophic-level eects of covariates. e mean eect of vegetation volume on the probability of occur-
rence for herbivorous and predatory bugs was positive, with posterior CIs95% that either contained only positive
values (herbivores) or with values that slightly overlapped zero (predators) (Table2). In contrast, the mean eect
of plant species diversity was negative for herbivorous bugs and positive for predatory bugs, with posterior CIs95%
that distinctly overlapped zero (Table2).
Predicted data derived from these eects showed that the predictive curves for the mean bug trophic-level
response to the vegetation volume gradient had positive slopes for both herbivores and predators (Fig.2a),
whereas the bug trophic-level response to plant species diversity was negative for herbivores and positive for pred-
ators (Fig.2b). When vegetation volume and plant species diversity were combined into a single environmental
space, a trend of high occupancy of herbivores was predicted for increasing levels of vegetation volume coupled
with decreasing levels of plant species diversity (Fig.3a). On the other hand, high occupancy of predators was
predicted for increasing levels of vegetation volume and plant species diversity (Fig.3b). In our study, high occu-
pancy levels of herbivores and predators (species occurring in more than 80% of sites) were almost exclusively
associated with the environmental space bounding the golf courses’ data points (solid rectangle in Fig.3a,b).
However, high occupancy levels of predators were also associated to an extent with the environmental space
Parameter Mean SD CI 2.5% CI 97.5%
Probability of occurrence
Golf courses 0.859 0.087 0.665 0.985
Parks 0.629 0.196 0.299 0.976
Gardens 0.755 0.125 0.492 0.975
Probability of detection
Golf courses 0.043 0.010 0.026 0.065
Parks 0.137 0.051 0.068 0.262
Gardens 0.077 0.022 0.043 0.130
Species richness
Tot a l
Golf courses 56.59 6.03 45.30 68.47
Parks 21.07 6.20 10.81 32.33
Gardens 24.18 3.79 16.76 31.33
Herbivores
Golf courses 48.10 5.15 38.55 58.17
Parks 18.04 5.26 9.33 27.34
Gardens 19.12 2.99 13.27 24.62
Predators
Golf courses 8.49 1.16 6.16 10.54
Parks 3.03 1.00 1.34 4.82
Gardens 5.06 0.93 3.18 6.69
Table 1. Posterior estimates for the probabilities of occurrence, probabilities of detection and species
richness as derived from the urban green space model. SD: Standard deviation; CI: Credible interval.
www.nature.com/scientificreports/
4
Scientific RepoRts | 7:40970 | DOI: 10.1038/srep40970
bounding the gardens’ data points (dashed rectangle in Fig.3b), which reects the strong positive relationship
between plant species diversity and occupancy by predators (Fig.2b).
Species-specic eects of covariates. e mean species-specic eects of vegetation volume on the
probabilities of occurrence for bugs were all positive, varying from 0.010 (S. pyrioides) to 1.682 (M. brevicornis)
in herbivores, and from 0.137 (Oechalia schellenbergii) to 1.994 (D. prolixus) in predators. In contrast, the mean
species-specic eects of plant species diversity were both positive and negative, varying from 2.112 (Dindymus
versicolor) to 2.702 (S. pyrioides) in predators, and from 2.868 (Orius sp.) to 3.169 (Chinoneides tasmaniensis).
Mean and standard deviation posterior estimates for the covariates’ species-specic eects on the probabilities
of occurrence for each herbivorous and predatory bug species are given in Tables S3–S4. We also provide in Tables
S3–S4 the posterior estimates for a series of quantiles, from which it is possible to derive a range of credible inter-
vals (99, 95, 75, 50, 25, 5 and 1%) to assess which species had the strongest responses to each of the explanatory
variables. We associated species with the three highest CIs (99, 95 and 75%) that did not overlap zero as having a
strong response to the given explanatory variable. As much as 59% (44 spp.) of herbivores showed strong positive
responses to the vegetation volume gradient (Fig.2c). e species with the strongest positive responses were
M. brevicornis and Stenophyella macreta, both with CIs99% that contained only positive values. Likewise, as much
as 44% (7 spp.) of predators showed strong positive responses to vegetation volume (Fig.2e), all with CI75% that
contained only positive values. Neither herbivores nor predators showed strong negative responses to vegetation
volume.
Figure 1. Estimated species richness of herbivorous (a,c,e) and predatory (b,d,f) bugs in gardens (a,b), parks
(c,d) and golf courses (e,f). Black lines indicate the mean response and coloured lines the posterior distribution
(i.e., 100% credible interval).
www.nature.com/scientificreports/
5
Scientific RepoRts | 7:40970 | DOI: 10.1038/srep40970
Responses to the plant diversity gradient were more complex, with dierent species showing either a positive
or a negative response. As few as 4% (3 spp.) of herbivores showed a strong positive response to plant species
diversity (Fig.2d). e species showing the strongest positive response was S. pyrioides, its CI99% containing only
positive values. On the other hand, 9% (7 spp.) of herbivores showed a strong negative response to plant species
diversity (Fig.2d), all with CI75% that contained only negative values. Only two predators showed strong responses
to the plant species diversity gradient (Fig.2f). ese species were C. tasmaniensis and Orius sp., which showed
positive and negative responses, respectively. Both species showed CI75% that did not overlap zero.
Discussion
Our study demonstrates that there are signicant dierences in the capacity of dierent green space types to
support diverse herbivorous and predatory bug communities. Vegetation structure has a positive eect on bug
diversity at the trophic- and species-specic levels (Fig.2a); while plant species diversity has a more variable
eect, generally increasing the diversity of predators, while reducing the diversity of herbivorous bugs (Fig.2b).
However, in both cases, these general responses to plant diversity are highly species specic, with individual spe-
cies within each trophic group displaying quite dierent responses to plant diversity. ese results indicate that
the changes to urban vegetation associated with dierent green space management practises, will have a distinct
eect on the predatory and herbivorous insects within these urban ecological networks.
In our examination of how bug species richness varied amongst green space types, we nd strong evidence
that golf courses are likely to sustain more herbivorous and predatory species than parks and gardens (Fig.1). is
nding is consistent with previous studies27. Given that in our study golf courses were the largest sites in surface
area, this nding is also consistent with the species-area relationship. In our study, however, as much as 40% of
golf course sites had more bug species than those predicted by the species-area relationship (Fig. S2). One poten-
tial explanation for golf courses presenting higher levels of herbivore and predatory bug species richness is that
the diversity of vegetation on golf courses provides a wider range of resource-rich habitats relative to parks and
gardens. For example, patches of low-intensity managed or unmanaged vegetation typical of tall grass or ‘rough’
areas of a golf course that contain ruderal grass and forb species may provide resources that support bug assem-
blages including granivores and grass specialists. Interestingly, analogous ruderal habitats in both agricultural
(e.g. oldelds) and urban (e.g. brownelds, derelict sites, vacant lots) environments have been documented to
be rich in rare and endangered insect biodiversity33,34. Additionally, in a parallel study the same golf course plots
used in this study had higher native plant species richness than residential and park plots35, further supporting
our ndings that increases in the proportion of native plants may also benet herbivore and predatory bugs in
this system.
Our study strengthens this understanding about the role of golf courses in supporting urban biodiversity in
two important ways. Firstly, our study demonstrates that vegetation structure and plant species diversity drive
bug diversity, regardless of functional group (e.g. herbivores and predators). Secondly, by examining heterop-
teran bugs, our study extends our understanding of urban insects to those that show incomplete metamorphosis.
Insects are divided into two clades, those that show incomplete metamorphosis, such as heteropteran bugs, and
those that show complete metamorphosis, such as butteries, bumble bees and ground beetles. Species that show
incomplete metamorphosis are reliant on similar resources throughout their life, and thus better reect local envi-
ronmental conditions than species that vary their resource requirements according to life stage. Future research
should focus on elucidating which vegetation features, or which plant species or group of species, contributes
most to increases in the biodiversity of herbivorous and predatory insect species in dierent green space types.
When examined at the trophic-level both herbivorous and predatory bug assemblages show positive responses
to vegetation volume (Fig.2a). Both herbivorous and predatory bug assemblages require vegetation-derived
resources to complete their life cycles. Herbivorous bugs, however, interact with vegetation resources directly,
utilising their piercing-sucking mouth specialisations to feed on the most nutritional-rich portions of plants
such as leaves, pollen, nectar, ower and leaf buds, and seed36. Most bug predators on the other hand interact
Parameter Mean SD CI 2.5% CI 97.5%
Probability of occurrence
Herbivores 0.559 0.631 0.334 0.804
Predators 0.709 0.789 0.277 0.984
Probability of detection
Herbivores 0.037 0.556 0.024 0.056
Predators 0.029 0.601 0.013 0.060
Eect of vegetation volume
Herbivores 1.032 0.305 0.535 1.738
Predators 1.203 0.921 0.095 3.610
Eect of plant species diversity
Herbivores 0.370 0.404 1.273 0.365
Predators 0.278 1.060 1.666 2.667
Table 2. Posterior estimates for the probabilities of occurrence, probabilities of detection and eects of
the vegetation volume and plant species diversity covariates as derived from the trophic-level model. SD:
Standard deviation; CI: Credible interval.
www.nature.com/scientificreports/
6
Scientific RepoRts | 7:40970 | DOI: 10.1038/srep40970
with plant resources indirectly, utilising vegetation elements as hunting and mating grounds. Most interestingly,
a few species, in the absence of suitable prey, will secure moisture and supplement their diets by feeding directly
from plant resources36. ese dierences are likely to explain why the trophic-level response of herbivores was
strictly positive, while the response of predators also included a negative component (Table2). Vegetation vol-
ume also has a strong positive eect on the species-specic probabilities of occurrence of most herbivorous and
predatory bug species, with many of the bugs in this study area predicted only to occur in sites with considerable
vegetation volume (Fig.2c,e). Our ndings concur with non-urban studies that have investigated the positive
relationship between vegetation structure and insect diversity20,22. However, our study shows that for both her-
bivorous and predatory insect species this relationship is species-specic, with most, but not all, species show-
ing strong responses to an increase in vegetation volume. For example, most herbivorous and predatory bugs
belonging in families traditionally associated with large-bodied species (e.g. Alydidae, Coreidae, Pentatomidae
and Reduviidae) show strong positive responses to vegetation volume, indicating perhaps that some of these
species have evolved in close association with large amounts of plant resources (e.g. the coreid Amorbus sp., a
herbivore closely associated with native Eucalyptus trees37). On the other hand, herbivorous bugs belonging in
families traditionally associated with small-bodied species with strong degrees of host specicity (e.g. Miridae
and Tingidae)18 show high site occupancy regardless of the amount of vegetation volume, suggesting that for these
species host-specicity is a more important driver of site occupancy than vegetation structure.
Figure 2. Predicted mean trophic-level (a,b) and species-specic (cf) responses of herbivorous (ad blue solid
lines) and predatory (a,b: red dashed lines; (e,f) red solid lines) bugs to the vegetation volume (a,c,d) and plant
species diversity (b,d,f) gradients. Species illustrated are limited to those that showed a strong response to the
covariates (i.e., those with 99, 95 and 75% CIs that did not overlap zero).
www.nature.com/scientificreports/
7
Scientific RepoRts | 7:40970 | DOI: 10.1038/srep40970
From a conservation standpoint, these ndings are both concerning and exciting. ey imply that most bug
species will tend to decline in urban areas if the vegetation structure of green spaces is simplied, for example, if
green spaces are managed predominantly as lawns or tree canopies. is simplication process could potentially
lead to local extinctions, if it were to occur over large areas. Yet, our ndings also imply that small-scale manage-
ment actions that increase the structure of urban vegetation undertaken by both public (e.g. local governments)
and private (e.g. homeowners) actors may encourage a greater diversity of heteropteran bugs. Local govern-
ments, for example, could boost insect diversity in their municipality by increasing the amount of mid-storey and
grassland-type vegetation, and promote the retention or planting of native plant species.
Our data do not indicate an effect of plant species diversity on herbivorous or predatory bugs at the
trophic-level. Rather, our results show that this eect is complex and distinctly species-specic, with species pre-
dicted to exhibit both positive and negative responses (Fig.2d,f). Although bug herbivorous and predatory assem-
blages are structured by a mix of specialists and generalists, most bug species, particularly predators, tend towards
a generalist diet, feeding on a wide array of host plants and arthropod prey15,36. Consequently, as our results have
shown, we will expect that the overall eect of plant diversity on bug communities should be small and that the
magnitude of this eect should be strong in only a small proportion of specialist species. For example, amongst
the herbivorous species showing a mean positive response to plant species diversity, less than 5% showed a strong
response, with the species showing the strongest response being the specialist non-native Azalea lacebug S. pyri-
oides. Azalea lacebugs, as their name implies, show strong host-specicity towards Rhododendron (Ericaceae), a
genus that in our study occurred exclusively in association with garden sites characterised by high levels of exotic
plant diversity35. Amongst the predators, only the stilt bug C. tasmaniensis showed a strong response to plant spe-
cies diversity. Interestingly, this species is a specialist predator, feeding exclusively on insect specialist herbivores
closely associated with a few native and non-native Geraniaceae. Positive relationships between plant and insect
diversity have been recorded in mechanistic experiments by Haddad et al.19,20. However, our study suggests that
in urban environments with much higher plant species richness and variability in plant species composition, the
relationship between plant diversity and the response of insect species and trophic groups is much less consistent,
and may be more strongly related with plant species identity than richness per se.
A potential explanation for this seemingly contradictory result is that in most urban environments, including
our study area, high plant diversity is predominantly associated with residential gardens. Overall, residential
gardens are thought to contribute substantially to insect diversity28, however, many management activities (e.g.
pesticide use and high-intensity management) can greatly reduce their contribution38. In this study, for example,
gardens supported signicantly lower bug species richness than the other green space types (Fig.1), perhaps due
to their lower vegetation structural complexity. Alternatively, the high diversity of plant species in many residen-
tial gardens, resulting primarily from idiosyncratic human preferences39, mean that they may be dominated by
non-native species35, which may not provide native insect specialists with the resources they need to thrive out-
side of their natural non-urban ranges. e question of why plant-diverse gardens in urban environments do not
attain the high levels of herbivorous and predatory insect species predicted by the non-urban literature remains
to be fully explored.
Our results suggest that high bug occupancy can be obtained in green spaces with specic combinations of
vegetation structure and plant diversity. When the vegetation volume and plant species diversity gradients were
combined into a single predicted environmental space, high occupancy of herbivores and predators was almost
exclusively associated with the environmental space bounding the data points of golf courses (Fig.3). We there-
fore conclude that large green spaces, such as golf courses, are more likely to support diverse herbivorous and
predatory insect assemblages due to their ability to provide a greater heterogeneity of vegetation structure and
plant diversity and favourable combinations of these green space attributes. e challenge we now face is under-
standing how we can boost the conservation value of all urban green spaces for herbivorous and predatory insects
through management strategies and actions aimed at promoting synergistic combinations of vegetation structure
Figure 3. Predicted combined eects of vegetation volume and plant species diversity on the occupancy of
herbivorous (a) and predatory (b) bugs. e superimposed rectangles represent the environmental space
dened by the vegetation volume and plant species diversity data points as quantied in each green space type
(golf courses: solid line; gardens: dashed line; parks: dotted line).
www.nature.com/scientificreports/
8
Scientific RepoRts | 7:40970 | DOI: 10.1038/srep40970
and plant diversity. is will be especially important in large green spaces with simple vegetation structure, and
in smaller green spaces such as public parks and residential gardens where it may be more dicult to intention-
ally achieve a heterogeneous mix of vegetation structure and diversity. Ultimately, tackling this conservation
challenge could provide enormous benets for all other elements of urban ecological networks, including human
city-dwellers.
Methods
Experimental design. e study was conducted in Melbourne, Australia’s second most populated city
(4 million inhabitants estimated for 2015) that supports diverse indigenous and introduced biodiversity.
Melbourne spans several bioregions, so to standardise underlying geology, climate and remnant vegetation asso-
ciations we limited the study area (Fig.4a) to the south-eastern suburbs within the Gippsland Plain bioregion.
e bioregion is characterised by sandy soils, average monthly maximum temperatures 13.5–25.9 °C, and average
monthly rainfall 47.3–66.1 mm. e dominant native vegetation communities are grassy woodland and heathy
woodland with a eucalypt overstorey40. e main types of urban green space include public parks, gardens, golf
courses and scattered patches of remnant native vegetation41.
Within the study area, we mapped all golf courses and identied all residential neighbourhoods and pub-
lic parks surrounding each golf course. We then identied triplets of green space sites (golf course, residential
gardens, public park; Fig.4b) established at approximately the same time (e.g. decade) as determined from his-
torical aerial imagery and municipal land release records. In all, 39 sites (Fig.4c) were randomly selected and
stratied by the three green space types. In our study, these sites constituted the units of inference – that is, the
spatial sample units in which we collected data to draw inferences on the ecological process (i.e. species site
occupancy)42. Patches of remnant vegetation were not included in the study as they are unevenly-distributed
and our study is focused on the eects of vegetation management in ‘human constructed’ green spaces. Triplets
of sites were located on average 14 km from each other, and sites within each triplet were located at least two
km from each other. is conguration was designed to generate a spatial distribution of sites that fullls the
requirement that observations should be drawn from spatially-independent units (i.e. each observation brings
one full degree of freedom43). We therefore expected no positive eects of spatial-autocorrelation associated with
our inferences. Site area ranged approximately across four orders of magnitude, varying from 6,712 to 862,022 m2
(mean = 273,193 m2).
Within each site we placed a series of sample plots (Fig.4d), which in our study constituted the unit of detec-
tion replication – that is, the sample spatial units in which we collected data to draw inferences on the observation
process (i.e. species detectability)42. We placed a minimum of two plots in sites < 50,000 m2 in size, with two
Figure 4. Schematic representation of our experimental design. Within the study area (a) we selected 13
triplet of sites (b), each of which consisted of a golf course, a park and a garden study site. In total, we surveyed
39 study sites (c), which represented the units of inference. Within each site we located a series of sample plots
(d), which represented the units of detection replication. e gure further indicates the average distance
between triplets of sites (Da) and study sites (Db), as well as the range of area values shown by the study sites
(Ass) and the average area value for plots within golf course and park (Agc/pk) and garden (Arg) sites.
www.nature.com/scientificreports/
9
Scientific RepoRts | 7:40970 | DOI: 10.1038/srep40970
additional plots being placed for every 50,000 m2 increase in site size. is conguration yielded a maximum of
eight plots in golf courses (for details of plot placements see relfall et al.29). Taken together, we collected sam-
ples in 182 plots: 104 in golf course, 52 in residential gardens and 26 in public parks. Within each golf courses we
placed half of the plots in ‘woodland rough, characterised by low-intensity managed shrub and tree vegetation,
and the other half in ‘long grassy rough, characterised by low-intensity managed or unmanaged herbaceous veg-
etation without trees. Golf course and park plots had an area of 600 m2 (20 × 30 m), while garden plots ranged
from 211 to 870 m2 (mean = 381 m2).
Insect sampling, sorting and identication. Insects were collected from 14 January to 12 March 2012
(Australian summer). At each plot, insects were collected from aboveground vegetation with 200 sweeps of an
entomological net (50 cm diameter), transferred to 70% ethanol-lled containers for storage and preservation,
and posteriorly sorted to order, and, whenever possible, bugs identied to species. Unlike many other insect taxa,
bugs are taxonomically tractable, allowing specimens to be sorted to morphospecies and oen to named species.
Explanatory variables. To measure vegetation structure within a plot, four parallel transects were estab-
lished and sampled at 5 m intervals. At each interval, we recorded the identity and growth form of any plant
species that intercepted a 2.5 m high pole at ve height intervals (0.0–0.2 m; 0.2–0.5 m; 0.5–1.0 m; 1.0–2.0 m; >
2.0 m). ese data were used to calculate the volume of vegetation within a given height band. To account for the
variable sizes of garden plots, we divided these volumes by the total available volume (i.e., area sampled multiplied
by height of relevant sampling interval) to generate a variable of vegetation volume (vvol: percentage of volume
occupied by vegetation). Additionally, the species identity data were used to generate a variable of plant diversity
(psd: plant species diversity). Vegetation variables at the plot level were then used to calculate the average for
a given green space site via averaging the values recorded for plots within each green space. In our study, vvol
ranged from < 1% to 40% (mean = 20%), and psd from 12 to 70 plant species per plot (mean = 27).
Modelling framework: Multi-species site occupancy models. Multi-species site occupancy mod-
els or community occupancy models are grounded in the idea that communities and metacommunities can be
described as a collection of individual species42,44. e hierarchical structure of multi-species site occupancy
models is composed of three levels: a level for the ecological process (e.g. species site occupancy), another for the
observation process (i.e. species detectability), and a third to account for the sampling of each species from its
(meta)community. e model is therefore a (meta)community hypermodel, in which the occupancy, detection
and eects parameters for each species are treated as random eects governed by hyperparameters that describe
the (meta)community42.
A key advantage of multi-species site occupancy models is that they allow inferences at both the species level,
such as the eects of covariates on the occupancy probability of each individual species, and community level,
such as community responses to random eects (e.g. green space type or trophic-level). Another key value of
this modelling framework is that the observation process hierarchical level reduces or even eliminates the bias
generated by the imperfect detection of species45,46. Treating each species as random eects is yet another key
feature of multi-species site occupancy models, particularly as this approach allows for the estimation of the
species richness for the whole observed community, as well as the number of species occurring at each specied
community level random eect (e.g. site, green space type, trophic-level)42,44. From the conservation point of
view, multi-species site occupancy models are exible analytical tools with the potential to improve assessments
of biodiversity responses to management-oriented actions47.
Statistical analyses. We analysed our data using two variations of the multi-species site occupancy
models provided by Zipkin et al.48 and Mata et al.33. In both models we condently assumed that the species
pool remained constant throughout the study, satisfying therefore an important assumption of the modelling
framework.
Our models included an extra hierarchical level that specied that the species-level random eects were gov-
erned by urban green space type (UGS-model) or trophic-level (TL-model) hyperparameters. e occurrence
model was specied as:
Ψ
~
zBernoulli (1)
tijtij,, ,,
where Ψ
t,i,j is the probability that, within green space type (UGS-model) or trophic-level (TL-model) t, species i
occurs at site j, and the detection model as:
Φ⋅
~
xBernoulli(z)(2)
tijk tijk tij,,, ,,, ,,
where Φ
t,i,j,k is, within green space type (UGS-model) or trophic-level (TL-model) t, the detection probability of
species i at site j at plot k. is specication satises the condition that the probability of detecting a species will
be zero when it is not present.
In the UGS-model, the linear predictor of the occupancy model on the logit-probability scale was specied as:
Ψ=logit( )occ (3)
tijtI,, ,
while in the TL-model the occupancy model linear predictor was specied as:
Ψ= +⋅ +⋅vvolpsdlogit( )occ xx (4)
tijtijti jti,, ,1(,)2(,)
www.nature.com/scientificreports/
10
Scientific RepoRts | 7:40970 | DOI: 10.1038/srep40970
where Ψ
t,i,j are the species-specic probabilities of occurrence for green space type (UGS-model) or trophic-level
(TL-model) t; occt,i the species-level random eects for green space type (UGS-model) or trophic-level (TL-model)
t; x1(t,i) and x2(t,i) the eects of covariates on the species-specic occurrence probabilities for trophic-level t; and
vvolj and psdj the mean = 0, sd = 1 standardised values for the vegetation volume (vvol) and plant species diversity
(psd) covariates for each site j. In the UGS-model, the species-level random eects occt,i were specied as:
~occNormal(mu ,tau )(5)
ti tt,
where mut ~ log(omega) log(1 − omega) and taut ~ Gamma (0.1, 0.1). us, the green space type occupancy
hyperparameters were also considered random-effects governed by the global occupancy hyperparameter
omega ~ Uniform (0, 1).
In the LP-model, the species-level random eects occt,i were specied as:
~occNormal(mu ,sigma )(6)
ti t,t
where mut ~ Normal (mu, sigma) and sigmat ~ Cauchy (0, 2.5). us, the trophic-level occupancy hyperparam-
eters were also considered random-eects governed by the global occupancy hyperparameters mu ~ Cauchy
(0, 2.5) and sigma ~ Cauchy (0, 2.5). is specication of normally-distributed hyperparameters with weakly
informative Cauchy (0, 2.5) priors follows Gelman et al.49 and Stan Development Team50.
e eects of the covariates on species-specic occupancy x1(t,i) and x2(t,i) (TL-model) were specied as:
..
.. .. ..
~
xNormal(mu x,sigmax )(7)
ti tt12(,)12( )12( )
where mu.x1..2(t) ~ Normal (mu.x1..2, sigma.x1..2) and sigma.x1..2(t) ~ Cauchy (0, 2.5). us, the trophic-level eect
hyperparameters were governed by the global effect hyperparameters mu.x1..2 ~ Cauchy (0, 2.5) and sigma.
x1..2 ~ Cauchy (0, 2.5). Finally, we assumed in both the UGS- and TL-model that the detection probability of spe-
cies i did not vary based on any measured covariate, and was thus determined by an unspecied species-level
eect dett,i as:
Φ=logit( )det (8)
tijk tI,,, ,
In the UGS-model, we estimated the total species richness of each green space type t, as well as the species
richness of herbivorous and predatory species, using the following summation structure:
∑∑
==
zij
(9)
i
St
i
Nt
11
where, within each green space t, St is the total number of sites, Nt the total number of detected species, and zi,j the
latent occurrence matrix. As these calculations were done as derived quantities within our Bayesian modelling
framework, we were able to report the species richness estimations with their full associated uncertainties.
Predictions. Using the model’s trophic-level hyperparameters, we predicted trophic-level occupancy for her-
bivores and predators for 500 values within the range of the vegetation volume (vvol) and plant species diversity
(psd) gradients. To guarantee predictions for a reasonable range of the gradients, we removed the 2.5% most
extreme values from each end of the vvol and psd original ranges. We used these predictions to graphically rep-
resent (1) the individual eects of vvol and psd on the occupancy of herbivores and predators (Fig.2), and (2) the
combined eect of vvol and psd on the occupancy of herbivores and predators (Fig.3). Into this latter predicted
environmental space, we superimposed three rectangles, which represent the environmental space bounding the
data points of the vvol and psd gradients as quantied in each green space type.
Species-area model. We modelled the eect of site area on bug species richness using the power function
of the species-area relationship31:
=ScA(10)
z
where S equals the number or bug species within a site, A the site’s area, and c and z are function parameters. In
the function’s linearised, logarithmically-transformed version, c equals the number of species in one unit of area
and z the slope of the species-area line.
e model for the power function of the species-area relationship was specied as:
λ~SPoisson(11)
ii
and the non-linear predictor as:
λ=cA(12)
ii
z
where, in each site i, Si and Ai are the number of observed bug species per site and the site’s area, respectively, and
λ
i the intensity parameter, which in a Poisson distribution equals both the mean and variance42. e latent varia-
bles c and z were given non-informative Uniform (0, 1) priors.
We used the model’s parameters c and z to predict the number of bug species per site given the empirical site
area data. Using a normally-distributed linear model (function lm in the R statistical environment51) we corre-
lated these estimations with the observed species richness to obtain a coecient of determination (r2). e value
www.nature.com/scientificreports/
11
Scientific RepoRts | 7:40970 | DOI: 10.1038/srep40970
of r2 indicated the proportion of the variance in the estimated number of bug species that is predictable from
our model – that is, the strength, or lack thereof, by which our data ts the power function of the species-area
relationship.
Bayesian inference implementation. Model parameters were estimated under a Bayesian mode of
inference. We used Markov Chain Monte Carlo (MCMC; Urban green spaces model and Species-area relation-
ship model) and Hamiltonian Monte Carlo (HMC; Trophic levels model) simulations to draw samples from the
parameters’ posterior distributions. MCMC algorithms were implemented in OpenBUGS52/JAGS53, accessed
through the R packages R2OpenBUGS54/jagsUI55. Our MCMC implementation used three chains of 50,000 itera-
tions, discarding the rst 5,000 in each chain as burn-in. HMC algorithms were implemented in Stan56, accessed
through the R package rstan57. Our HMC implementation used four chains of 5,000 iterations, discarding the rst
half iterations in each chain during warm-up. Visual inspections of MCMC and HMC chains plus values of the
Gelman-Rubin statistic (R-hat < 1.1) indicated acceptable convergence48,58.
e codes and data to re-run the analyses and generate plots are provided in the Supplementary Information.
References
1. Foley, J. A. et al. Global consequences of land use. Science. 309, 570–574 (2005).
2. Seto, . C., Güneralp, B. & Hutyra, L. . Global forecasts of urban expansion to 2030 and direct impacts on biodiversity and carbon
pools. Proc. Natl. Acad. Sci. USA 109, 16083–16088 (2012).
3. Ives, C. D. et al. Cities are hotspots for threatened species. Glob. Ecol. Biogeogr. 25, 117–126 (2016).
4. eniger, L. E., Gaston, . J., Irvine, . N. & Fuller, . A. What are the benets of interacting with nature ? Int. J. Environ. Health. es.
10, 913–935 (2013).
5. Sattler, T., Obrist, M., Duelli, P. & Moretti, M. Urban arthropod communities: added value or just a blend of surrounding
biodiversity ? Landsc. Urban Plan. 103, 347–361 (2011).
6. Scudder, G. e importance of insects In Insect biodiversity – Science and society (eds Foottit, . G. & Adler, P. H.) 7–32 (Wiley-
Blacwell, 2009).
7. Prather, C. M. et al. Invertebrates, ecosystem services and climate change. Biol. ev. 88, 327–348 (2013).
8. Dunn, . . Global mapping of ecosystem disservices: the unspoen reality that nature sometimes ills us. Biotropica. 42, 555–557
(2010).
9. ust, M. . & Su, N. Y. Managing social insects of urban importance. Annu. ev. Entomol. 57, 355–375 (2012).
10. Ewers, . M. et al. Logging cuts the functional importance of invertebrates in tropical rainforest. Nat. commun. 6, 6836, 10.1038/
ncomms7836 (2015).
11. Stor, N. E. Insect diversity: facts, ction and speculation. Biol. J. Linn. Soc. 35, 321–337 (1988).
12. Didham, . ., Ghazoul, J., Stor, N. E. & Davis, A. J. Insects in fragmented forests: a functional approach. Trends Ecol. Evol. 11,
255–260 (1996).
13. Henry, T. Biodiversity of Heteroptera in Insect biodiversity – Science and society (eds Foottit, . G. & Adler, P. H.) 223–263 (Wiley-
Blacwell, 2009).
14. Weirauch, C. & Schuh, . Systematics and evolution of Heteroptera: 25 years of progress. Annu. ev. Entomol. 56, 487–510 (2011).
15. Schaefer, C. W. & Panizzi, A. . Economic importance of Heteroptera: a general view In Heteroptera of economic importance (eds
Schaefer, C. W. & Panizzi, A. .) 3–8 (CC Press, 2000).
16. Eubans, M., Styrsy, J. & Denno, . e evolution of omnivory in heteropteran insects. Ecology. 84, 2549–2556 (2003).
17. Perdiis, D., Fantinou, A. & Lyouressis, D. Enhancing pest control in annual crops by conservation of predatory Heteroptera. Biol.
Control. 59, 13–21 (2011).
18. Cassis, G. & Schuh, . T. Systematics, biodiversity, biogeography, and host associations of the Miridae (Insecta: Hemiptera:
Heteroptera: Cimicomorpha). Annu. ev. Entomol. 57, 377–404 (2012).
19. Haddad, N. M., Tilman, D., Haarstad, J., itchie, M. & nops, J. M. N. Contrasting eects of plant richness and composition on
insect communities: a eld experiment. Am. Nat. 158, 17–35 (2001).
20. Haddad, N. M. et al. Plant species loss decreases arthropod diversity and shis trophic structure. Ecol. Lett. 12, 1029–1039 (2009).
21. aupp, M. J., Shrewsbury, P. M. & Herms, D. A. Ecology of herbivorous arthropods in urban landscapes. Annu. ev. Entomol. 55,
19–38 (2010).
22. őrösi, A., Batáry, P., Orosz, A., édei, D. & Báldi, A. Eects of grazing, vegetation structure and landscape complexity on grassland
leaoppers (Hemiptera: Auchenorrhyncha) and true bugs (Hemiptera: Heteroptera) in Hungary. Insect Conserv. Divers. 5, 57–66
(2012).
23. Lassau, S. A. & Hochuli, D. F. Eects of habitat complexity on ant assemblages. Ecography. 27, 157–164 (2004).
24. Beninde, J., Veith, M. & Hochirch, A. Biodiversity in cities needs space: a meta-analysis of factors determining intra-urban
biodiversity variation. Ecol. Lett. 18, 581–592 (2015).
25. Turrini, T. & nop, E. A landscape ecology approach identies important drivers of urban biodiversity. Glob. Chang. Biol. 21,
1652–1667 (2015).
26. Smith, . M., Gaston, . J., Warren, P. H. & ompson, . Urban domestic gardens (VIII): environmental correlates of invertebrate
abundance. Biodivers. Conserv. 15, 2515–2545 (2006).
27. Colding, J. & Fole, C. e role of golf courses in biodiversity conservation and ecosystem management. Ecosystems. 12, 191–206
(2009).
28. Sperling, C. D. & Lortie, C. J. e importance of urban bacgardens on plant and invertebrate recruitment: a eld microcosm
experiment. Urban Ecosyst. 13, 223–235 (2010).
29. relfall, C. G. et al. e conservation value of urban green space habitats for native bee communities. Biol. Conserv. 187, 240–248
(2015).
30. relfall, C. G., Williams, N. G. S., Hahs, A. . & Livesley, S. J. Approaches to urban vegetation management and the impacts on
urban bird and bat communities. Landsc. Urban Plan. 153, 28–39 (2016).
31. Tri antis, . A., Guilhaumon, F. & Whittaer, . J. e island species-area relationship: biology and statistics. J. Biogeogr. 39, 215–231
(2012).
32. Mata, L. Heteroptera ecology, biodiversity and conservation – PhD thesis (University of Barcelona, 2013).
33. Mata, L., Goula, M. & Hahs, A. . Conserving insect assemblages in urban landscapes: accounting for species-specic responses and
imperfect detection. J. Insect Conserv. 18, 885–894 (2014).
34. Gardiner, M. M., Burman, C. E. & Prajzner, S. P. e value of urban vacant land to support arthropod biodiversity and ecosystem
services. Environ. Entomol. 42, 1123–1136 (2013).
35. relfall, C. G. et al. Variation in vegetation structure and composition across urban green space types. Front. Ecol. Evol. doi:
10.3389/fevo.2016.00066 (2016).
www.nature.com/scientificreports/
12
Scientific RepoRts | 7:40970 | DOI: 10.1038/srep40970
36. Schuh, . T. & Slater, J. A. True bugs of the world (Hemiptera: Heteroptera) – Classication and natural history (Cornell University
Press, 1995).
37. Steinbauer, M. J., Yonow, T., eid, I. A. & Cant, . Ecological biogeography of species of Gelonus, Acantholybas and Amorbus in
Australia. Austral Ecol. 27, 1–25 (2002).
38. Matteson, . C. & Langellotto, G. A. Small scale additions of native plants fail to increase benecial insect richness in urban gardens.
Insect Conserv. Divers. 4, 89–98 (2011).
39. endal, D., Williams, . J. H. & Williams, N. S. G. Plant traits lin people’s plant preferences to the composition of their gardens.
Landsc. Urban Plan. 105, 34–42 (2012).
40. Bull, M. & Stolfo, G. Flora of Melbourne – A guide to the indigenous plants of the Grater Melbourne Area (Hyland House, 2014).
41. Australian esearch Centre for Urban Ecology. Biodiversity of Metropolitan Melbourne (Victorian Environmental Assessment
Council, 2009)
42. éry, M. & oyle, J. A. Applied hierarchical modeling in Ecology: analysis of distribution, abundance and species richness in  and
BUGS Volume 1: prelude and static models (Academic Press, 2016).
43. Legendre, P. Spatial autocorrelation: trouble or new paradigm. Ecology 74, 1659–1673 (1993).
44. oyle, J. A. & Dorazio, . M. Hierarchical modeling and inference in ecology. e analysis of data from populations, metapopulations
and communities (Academic Press, 2008).
45. Inayan, . J., Tingley, M. W., Furnas, B. J. & Beissinger, S. . Detecting diversity: emerging methods to estimate species diversity.
Trends Ecol. Evol. 29, 97–106 (2014).
46. Lahoz-Monfort, J. J., Guillera-Arroita, G. & Wintle, B. A. Imperfect detection impacts the performance of species distribution
models. Glob. Ecol. Biogeogr. 23, 504–515 (2014).
47. Zipin, E. F., oyle, J. A., Dawson, D. . & Bates, S. Multi-species occurrence models to evaluate the eects of conservation and
management actions. Biol. Conserv. 143, 479–484 (2010).
48. Zipin, E. F., DeWan, F. A. & oyle, J. A. Impacts of forest fragmentation on species richness: a hierarchical approach to community
modelling. J. Appl. Ecol. 46, 815–822 (2009).
49. Gelman, A., Carlin, J. B., Stern, H. S. & ubin, D. B. Bayesian data analysis (Taylor and Francis, 2014).
50. Stan Development Team. Stan modelling language – User’s guide and reference manual. Version 2.12.0. http://mc-stan.org/
documentation/ (Date of access: 10/11/2016) (2016).
51.  CoreTeam. : A language and environment for statistical computing.  Foundation for Statistical Computing, Vienna, Austria.
UL http://www.-project.org/ (2016).
52. Lunn, D., Spiegelhalter, D., omas, A. & Best, N. e BUGS project: evolution, critique and future directions. Stat. Med. 28,
3049–3067 (2009).
53. Plummer, M. JAGS: a program for analysis of Bayesian graphical models using Gibbs sampling in Proceedings of the 3rd International
Worshop on Distributed Statistical Computing (eds Horni, ., Leisch, F. & Zeileis, A.) (2003).
54. Sturtz, S., Ligges, U. & Gelman, A. 2WinBUGS: A pacage for running WinBUGS from . J. Stat. Sow. 12, 1–16 (2005).
55. ellner, . jagsUI: A Wrapper Around ‘rjags’ to Streamline ‘JAGS’ Analyses. Version 1.4.2. https://CAN.-project.org/pacage=
jagsUI (Date of access: 10/11/2016) (2016).
56. Carpenter, B. et al. Stan: a probabilistic programming language. J. Stat. Sow. In press (2016).
57. Guo, J. et al. rstan:  interface to Stan. Version 2.12.1. https://cran.r-project.org/web/pacages/rstan/ (Date of access: 10/11/2016)
(2016).
58. Gelman, A. & Hill, J. Data analysis using regression and multilevel/hierarchical models (Cambridge University Press, 2007).
Acknowledgements
is study was funded by the Australian Research Council (LP110100686), the Australian Research Centre for
Urban Ecology (ARCUE) and the Australian Golf Course Superintendents Association. Bob Carpenter provided
assistance with the probabilistic programing language Stan, Briony Norton and Frances Alexander with eldwork
and insect sorting, and Melinda Moir with lacebug species identication. LM and CGT wish to acknowledge
the support of funding from the Australian Government’s National Environmental Science Programme – Clean
Air and Urban Landscapes (NESP-CAUL) hub. LM also wishes to acknowledge the support of funding from the
Australian Research Council Centre of Excellence for Environmental Decisions (CEED). AKH was supported by
the Baker Foundation.
Author Contributions
A.K.H., C.G.T., L.M., N.E.S., N.S.G.W. and S.J.L. conceptualised the study. C.G.T., L.M., N.S.G.W. and S.J.L.
collected the data. L.M. and M.M. identied the species. L.M. conducted the data analyses. L.M. wrote the
manuscript with inputs from all authors. A.K.H., N.E.S., N.S.G.W. and S.J.L. secured funding.
Additional Information
Supplementary information accompanies this paper at http://www.nature.com/srep
Competing nancial interests: e authors declare no competing nancial interests.
How to cite this article: Mata, L. et al. Conserving herbivorous and predatory insects in urban green spaces.
Sci. Rep. 7, 40970; doi: 10.1038/srep40970 (2017).
Publisher's note: Springer Nature remains neutral with regard to jurisdictional claims in published maps and
institutional aliations.
is work is licensed under a Creative Commons Attribution 4.0 International License. e images
or other third party material in this article are included in the article’s Creative Commons license,
unless indicated otherwise in the credit line; if the material is not included under the Creative Commons license,
users will need to obtain permission from the license holder to reproduce the material. To view a copy of this
license, visit http://creativecommons.org/licenses/by/4.0/
© e Author(s) 2017
... Indeed, many studies have shown the negative relationship between urbanisation and invertebrate species richness (McKinney 2008;Faeth et al. 2011;Mata et al. 2014;Fenoglio et al. 2020;Piano et al. 2020) and trait composition change (Merckx et al. 2018;Fournier et al. 2020). While these studies explored response patterns at the community level, recent studies are increasingly highlighting that the response of invertebrates to urbanisation processes and filters are species-and trait-specific -that is, some species and trait values tend to be absent in extreme environmental conditions, whereas others are present across a wider environmental gradient (Magura et al. 2013;Mata et al. 2014Mata et al. , 2017Threlfall et al. 2017). Species-specific mechanisms driving the responses of invertebrates to urban filters are poorly understood. ...
Article
Full-text available
Invertebrates comprise the most diversified animal group on Earth. Due to their long evolutionary history and small size, invertebrates occupy a remarkable range of ecological niches, and play an important role as “ecosystem engineers” by structuring networks of mutualistic and antagonistic ecological interactions in almost all terrestrial ecosystems. Urban forests provide critical ecosystem services to humans, and, as in other systems, invertebrates are central to structuring and maintaining the functioning of urban forests. Identifying the role of invertebrates in urban forests can help elucidate their importance to practitioners and the public, not only to preserve biodiversity in urban environments, but also to make the public aware of their functional importance in maintaining healthy greenspaces. In this review, we examine the multiple functional roles that invertebrates play in urban forests that contribute to ecosystem service provisioning, including pollination, predation, herbivory, seed and microorganism dispersal and organic matter decomposition, but also those that lead to disservices, primarily from a public health perspective, e.g., transmission of invertebrate-borne diseases. We then identify a number of ecological filters that structure urban forest invertebrate communities, such as changes in habitat structure, increased landscape imperviousness, microclimatic changes and pollution. We also discuss the complexity of ways that forest invertebrates respond to urbanisation, including acclimation, local extinction and evolution. Finally, we present management recommendations to support and conserve viable and diverse urban forest invertebrate populations into the future.
... We used uninformative, uniform priors for all parameters and assessed convergence using the Gelman-Rubin statistic, where values <1.1 indicate convergence (Gelman et al., 2004) and by visual inspection of parameter traceplots. "Significant" differences were assessed using posterior probability, where "strong significant" effects were expressed as parameters with >95% posterior probability that did not overlap with 0 (i.e., our model estimates that the probability the parameter effect is more extreme than 0 is >95%), and "moderate significant" effects were expressed as parameters with >85% posterior probability that did not overlap with 0. We included the 85% threshold to capture species and community effects that exhibited a strong trend toward positive or negative association but were limited by sample size and high variation in the community hyperparameters, which can lead to "shrinkage" of species-specific effects toward the community mean (Kery & Royle, 2015;Mata et al., 2017;Suraci et al., 2021). ...
Article
Full-text available
Human development and roads threaten wildlife through distinct mechanisms and understanding the influence of these elements can better inform mitigation and conservation strategies. We used camera traps to quantify the effects of major roads, environmental factors, and human development on the mammalian community composition between sites north and south of a major interstate highway in northern Utah, USA. We found no significant differences in species richness nor community similarity across the north–south divide of the highway. Through Bayesian hierarchical modeling, we compared the effects of the distance to the highway, housing and human population density, normalized difference vegetation index (NDVI), and the human footprint index to changes in mammalian community composition and species‐specific habitat usage. Community occupancy response, similarity, and species richness were negatively affected by increased housing and human population densities and positively affected by increased NDVI and decreased human footprint, whereas their response to the highway was more inconclusive. We conclude that mammalian community composition in our study area is influenced by both environmental conditions and human development while the effect of the highway was more nuanced, possibly due to the presence of a newly constructed wildlife overpass. Taken together, the lack of differences in species richness or community composition across the highway suggests that it may not currently exacerbate the effects of other anthropogenic sources of habitat fragmentation and highlights the need for additional research into human–wildlife conflict mitigation strategies.
... It also adds to the growing number of studies documenting diversity-resistance relationships in urban environments (Clem & Held 2018;Doherty et al., Frank 2014). However, it conflicts with other results suggesting an increase in herbivore abundance with increasing plant diversity in urban environments (Mata et al., 2017), although the relationship between herbivore abundance and actual herbivory is not always positively correlated (Barbosa et al., 2009;Schueller et al., 2019). Tree diversity may have influenced the probability of focal trees being found and colonized by herbivores. ...
Article
Full-text available
Insect herbivory is an important component of forest ecosystems functioning and can affect tree growth and survival. Tree diversity is known to influence insect herbivory in natural forest, with most studies reporting a decrease in herbivory with increasing tree diversity. Urban ecosystems, on the other hand, differ in many ways from the forest ecosystem and the drivers of insect herbivory in cities are still debated. We monitored 48 urban trees from five species – three native and two exotic – in three parks of Montreal (Canada) for leaf insect herbivory and predator activity on artificial larvae, and linked herbivory with both predation and tree diversity in the vicinity of focal trees. Leaf insect herbivory decreased with increasing tree diversity and with increasing predator attack rate. Our findings indicate that tree diversity is a key determinant of multitrophic interactions between trees, herbivores and predators in urban environments and that managing tree diversity could contribute to pest control in cities.
... The usefulness of UGS is hence seen by how they enhance the quality of life in urban environments, provide biodiversity, and offer habitat corridors or refuge for numerous species (Cameron et al. 2020;Planchuelo et al. 2019;Wang et al. 2019;Mata et al. 2017;Mensah et al. 2016). However, despite these contributions of UGS, they are under threat due to population growth (Vargas-Hernández and Zdunek-Wielgołaska 2021) coupled with other urban activities such as farming, and construction (Ives et al. 2016;Norton et al. 2016). ...
Article
Full-text available
Urban green spaces (UGS) enhance the quality of life in urban environments and serve as habitat corridors or refuge for organisms, including beetles and spiders. The attributes of UGS allow them to harbour species that offer essential ecosystem services. However, the ability of UGS to provide services is limited by the extent to which they have been altered anthropogenically. We described the taxonomic richness and functional composition of arthropods in a mountainous urban ecosystem of Ghana by focussing on the activity of both beetles and spiders at the family level. Two main land-use types (woodlands and built-up areas) were identified and characterised based on the presence or absence of certain vegetation attributes. Sixteen plots in each land-use type with sizes 20 × 20 m were demarcated and fitted with four pitfall traps in each plot to sample continuously for eight weeks, the activity density of both beetles and spiders. Samples were sorted into families and functional groups (detritivores, fungivores, herbivory and predators). The taxonomic richness and activity density were both significantly higher in the woodlands than in the built-up areas. Similarly, all functional groups showed a higher affinity to the woodlands than the built-up areas. Habitat attributes defined by plant diversity and structural complexity were the underlying drivers explaining the differences in arthropod communities between the land-use types. Though the built-up areas seem degraded and open, the remaining small vegetation patches still support the activities of some taxa that should merit the protection of such remnant vegetation in urban ecosystems.
... The presence of contrasting microhabitats even within a range of a couple of meters allows the co-existence of ant species with different environmental requirements that can positively affect the Shannon diversity of ant assemblages 17,24,40 . On mounds characterised by high PC1 values litter accumulation can also be beneficial for several ant species and might increase ant diversity because it can provide both shelter and foraging sites for them 45 . ...
Article
Full-text available
Fragmented natural habitats within human-transformed landscapes play a key role in preserving biodiversity. Ants as keystone species are essential elements of terrestrial ecosystems; thus, it is important to understand the factors influencing their presence. In a large-scale multi-site study, we surveyed ant assemblages using sweep netting and D-vac sampling on 158 ancient burial mounds preserving grassland habitats in agricultural landscapes in East-Hungary. We asked the following questions: (1) How do habitat factors and landscape composition affect species richness and functional diversity of ants? (2) Which ant traits are affected by habitat factors and landscape composition? Despite their small sizes, mounds as permanent and relatively undisturbed landscape elements could provide safe havens for diverse ant assemblages even in transformed agricultural landscapes. The complex habitat structure of wooded mounds supported high species and functional diversity of ant assemblages. Ant species on wooded mounds had small or medium-sized colonies, enabling the co-existence of more species. The effect of landscape composition on ant assemblages was mediated by habitat factors: steep slopes buffered the negative effect of the cropland matrix and enabled higher ant diversity.
... Since ruderal grassland, independent of urban settlement, had a higher disturbance intensity, Sattler et al. (2011) suggested that such grasslands might have fewer, but more resilient arthropod species. Rangelands exhibited higher diversity than the ruderal and fragmented grasslands, probably due to less disturbance and fragmentation as well as the larger availability of structural habitat variation provided in more natural areas (Mata et al. 2017, Botha et al. 2018Koranyi et al. 2021). ...
Article
Full-text available
Predacious arthropods provide a valuable ecosystem service within urban environments by suppressing pest numbers. However, urban ecological studies largely ignore this functional group and its diversity and species composition patterns. Some studies have been published regarding these patterns, however they were mostly done in Australia, Europe and America, thus an African perspective is lacking. Our aim was to address the gap in African literature by quantifying the differences in predacious arthropod species richness and diversity within urban green space in varying urbanization intensities, habitat types of grassland and patch sizes in South Africa. Various indices were considered to examine the effect of urbanization on the diversity patterns of Chrysopidae (Neuroptera), Mantodea, Araneae and Coccinellidae (Coleoptera). Study sites included three levels of urbanization represented by human population density, two types of urban grasslands (i.e., ruderal and fragmented grassland) and a peri-urban rangeland grassland as control, and a wide range of patch sizes. Our results indicated that an increase in abundance of predacious arthropod taxa was associated with increased level of urbanization. Also, that urban and peri-urban grasslands had similar predacious arthropod species richness and diversity, but differed in species composition. No relationship was found between patch size and arthropod diversity or composition. Thus, predacious arthropod abundances are influenced by the level of urbanization and their species composition is influenced by the type of urban grassland (ruderal or fragmented), which are important considerations for future urban planning/management and conservation strategies. This study gives a South African perspective and indicates that despite the lack of assigned function of urban green spaces, they sustain diverse and distinct predacious arthropod communities, which in turn fulfil various roles in a functioning ecosystem.
... In addition to providing more precise estimates of species-specific effects (Zipkin et al., 2009), these models can estimate community-level parameters, as well as biodiversity metrics (Guillera-Arroita et al., 2019), all with associated uncertainty (Zipkin et al., 2010). Community occupancy models have led to improved insights into understanding species occurrence patterns for insects (Mata et al., 2017), birds (Zipkin et al., 2009), and mammals (Gallo et al., 2017). ...
Preprint
Full-text available
The occurrence and distributions of wildlife populations and communities are shifting as a result of global changes. To evaluate whether these shifts are negatively impacting biodiversity processes, it is critical to monitor the status, trends, and effects of environmental variables on entire communities. However, modeling the dynamics of multiple species simultaneously can require large amounts of diverse data, and few modeling approaches exist to simultaneously provide species and community level inferences. We present an "integrated community occupancy model" (ICOM) that unites principles of data integration and hierarchical community modeling in a single framework to provide inferences on species-specific and community occurrence dynamics using multiple data sources. We use simulations to compare the ICOM to previously developed hierarchical community occupancy models and single species integrated distribution models. We then apply our model to assess the occurrence and biodiversity dynamics of foliage-gleaning birds in the White Mountain National Forest in the northeastern USA from 2010-2018 using three independent data sources. Simulations reveal that integrating multiple data sources in the ICOM increased precision and accuracy of species and community level inferences compared to single data source models, although benefits of integration were dependent on data source quality (e.g., amount of replication). Compared to single species models, the ICOM yielded more precise species-level estimates. Within our case study, the ICOM had the highest out-of-sample predictive performance compared to single species models and models that used only a subset of the three data sources. The ICOM offers an attractive approach to estimate species and biodiversity dynamics, which is additionally valuable to inform management objectives of both individual species and their broader communities.
Article
Full-text available
1. The occurrence and distributions of wildlife populations and communities are shifting as a result of global changes. To evaluate whether these shifts are negatively impacting biodiversity processes, it is critical to monitor the status, trends and effects of environmental variables on entire communities. However, modelling the dynamics of multiple species simultaneously can require large amounts of diverse data, and few modelling approaches exist to simultaneously provide species and community‐level inferences. 2. We present an ‘integrated community occupancy model’ (ICOM) that unites principles of data integration and hierarchical community modelling in a single framework to provide inferences on species‐specific and community occurrence dynamics using multiple data sources. The ICOM combines replicated and nonreplicated detection–nondetection data sources using a hierarchical framework that explicitly accounts for different detection and sampling processes across data sources. We use simulations to compare the ICOM to previously developed hierarchical community occupancy models and single species integrated distribution models. We then apply our model to assess the occurrence and biodiversity dynamics of foliage‐gleaning birds in the White Mountain National Forest in the northeastern USA from 2010 to 2018 using three independent data sources. 3. Simulations reveal that integrating multiple data sources in the ICOM increased precision and accuracy of species and community‐level inferences compared to single data source models, although benefits of integration were dependent on the information content of individual data sources (e.g. amount of replication). Compared to single species models, the ICOM yielded more precise species‐level estimates. Within our case study, the ICOM had the highest out‐of‐sample predictive performance compared to single species models and models that used only a subset of the three data sources. 4. The ICOM provides more precise estimates of occurrence dynamics compared to multi‐species models using single data sources or integrated single‐species models. We further found that the ICOM had improved predictive performance across a broad region of interest with an empirical case study of forest birds. The ICOM offers an attractive approach to estimate species and biodiversity dynamics, which is additionally valuable to inform management objectives of both individual species and their broader communities.
Article
Full-text available
Global change, including urbanisation, threatens many of the >1400 bat species. Nevertheless, certain areas within highly urbanised cities may be suitable to harbour bat populations. Thus, managing urban habitats could contribute to bat conservation. Here, we wanted to establish evidence-based recommendations on how to improve urban spaces for the protection of bats. In a team effort with >200 citizen scientists, we recorded bat vocalisations up to six times over the course of two years at each of 600 predefined sites in the Berlin metropolitan area. For each species we identified the preferred and non-preferred landscape features. Our results show that artificial light at night (ALAN) had a negative impact on all species. For soprano pipistrelles and mouse-eared bats ALAN had the largest effect sizes among all environmental predictors. Canopy cover and open water were especially important for bat species that forage along vegetation edges and for trawling bats, respectively. Occurrence probability of species foraging in open space decreased with increasing distance to water bodies. On a larger scale, impervious surfaces tended to have positive effects on some species that are specialised on foraging along edge structures. Our study constitutes an important contribution to the growing body of literature showing that despite the many negative impacts of urbanization on wildlife, urban environments can harbour bat populations if certain conditions are met, such as access to vegetation and water bodies and low levels of ALAN. Our findings are of high relevance for urban planners and conservationists, as they allow inferences on how to manage urban spaces in a bat-friendly way. We recommend limiting ALAN to the minimum necessary and maintaining and creating uninterrupted vegetated corridors between areas with high levels of canopy cover and water bodies, in which ALAN should be entirely avoided.
Article
Urban green spaces vary in their local and regional attributes, and this variation can have contrasting effects on the ability of urban green spaces to support biodiversity. Here we assess the response of four insect taxa to urban development in Sydney, Australia, using a multiscale approach. We sampled four insect orders (Coleoptera, Diptera, Hemiptera and Hymenoptera) at 19 sites and employed a) multivariate analyses to test the responses of insect communities to environmental management regimes, and b) generalised linear models to determine taxa responses to urban green spaces at local and regional scales. We hypothesised that environmental management regimes (forest, urban forest remnant, and recreational park) affected the species composition of insect communities but found no evidence to support this hypothesis. Only species composition of hemipteran communities responded at this scale, with assemblages in urban forest remnants distinct to those in forests and recreational parks. Conversely, site attributes at local and regional scales had distinct effects on each insect group. While species richness of all taxa was positively related to local flower species richness, they exhibited variable responses to regional attributes such as greenspace proportion and site area. The taxon-specific response to local and regional attributes of urban green spaces suggest that there is no silver-bullet for managing local or regional traits to promote diversity of major insect groups in urban areas. These results highlight the opportunities for managers and planners to develop unique and targeted taxon-specific interventions aiming to enhance urban insect diversity.
Article
Full-text available
Stan is a probabilistic programming language for specifying statistical models. A Stan program imperatively defines a log probability function over parameters conditioned on specified data and constants. As of version 2.14.0, Stan provides full Bayesian inference for continuous-variable models through Markov chain Monte Carlo methods such as the No-U-Turn sampler, an adaptive form of Hamiltonian Monte Carlo sampling. Penalized maximum likelihood estimates are calculated using optimization methods such as the limited memory Broyden-Fletcher-Goldfarb-Shanno algorithm. Stan is also a platform for computing log densities and their gradients and Hessians, which can be used in alternative algorithms such as variational Bayes, expectation propagation, and marginal inference using approximate integration. To this end, Stan is set up so that the densities, gradients, and Hessians, along with intermediate quantities of the algorithm such as acceptance probabilities, are easily accessible. Stan can be called from the command line using the cmdstan package, through R using the rstan package, and through Python using the pystan package. All three interfaces support sampling and optimization-based inference with diagnostics and posterior analysis. rstan and pystan also provide access to log probabilities, gradients, Hessians, parameter transforms, and specialized plotting.
Article
Full-text available
The ecological sustainability and function of urban landscapes is strongly influenced by the composition and structure of the local plant community. Taxonomic composition generally refers to the identity of the species comprising the community, while we define structure as the presence of multiple canopy layers, as well as stems of varying diameter and age. These aspects of urban vegetation significantly influence the ecology of cities, yet they are generally poorly quantified across the range of natural and constructed plant communities present in urban landscapes. We quantified vegetation composition and structure to i) simultaneously assess their variation across four green spaces types (golf courses, public parks, residential neighbourhoods and patches of remnant vegetation) in Melbourne, Australia, and ii) investigate the relationship between vegetation composition and structure within these green spaces. The four green space types supported distinctly different plant communities. Vegetation composition in the residential neighborhoods differed significantly from the others (p
Chapter
Insects are important because of their diversity, ecological role, and influence on agriculture, human health, and natural resources. This chapter documents the dominance of insects in the living world and shows how they have been central to many advances in science. Insects create the biological foundation for all terrestrial ecosystems. They cycle nutrients, pollinate plants, disperse seeds, maintain soil structure and fertility, control populations of other organisms, and provide a major food source for other taxa. Most major insect pests in agriculture are non-native species that have been introduced into a new ecosystem, usually without their natural biological control agents. Insects have evolved unique features in the animal world that are a surprise to experts in biomechanics and bioengineering because many are recent inventions of humans. Insects have been in competition with humans for the products of our labor ever since cultivation of soil began.
Chapter
This chapter presents an overview of the seven infraorders and the 91 currently recognized families of Heteroptera. For each family, it provides a brief diagnosis, selected information on their habits and economic importance, references to the key literature, and known numbers of species. The chapter also discusses the importance of understanding heteropteran biodiversity from a phylogenetic and economic viewpoint, including their role in conservation biology and global warming. It then gives examples of the many true-bug attributes, including information on their morphology, distribution, numbers, host associations, and economic importance. Recent treatments of various families, such as the Coreidae, Lygaeoidea, Miridae, Pentatomidae, and other families presented in Schaefer and Panizzi (2000), have provided much-needed synthesis of widely scattered information that is important to better understand certain pest groups. Heteroptera's roles as plant feeders, hemolymph-sucking parasites, invertebrate predators, or water-quality indicators make them unquestionably important organisms in the environment.
Book
Applied Hierarchical Modeling in Ecology: Distribution, Abundance, Species Richness offers a new synthesis of the state-of-the-art of hierarchical models for plant and animal distribution, abundance, and community characteristics such as species richness using data collected in metapopulation designs. These types of data are extremely widespread in ecology and its applications in such areas as biodiversity monitoring and fisheries and wildlife management. This first volume explains static models/procedures in the context of hierarchical models that collectively represent a unified approach to ecological research, taking the reader from design, through data collection, and into analyses using a very powerful class of models. Applied Hierarchical Modeling in Ecology, Volume 1 serves as an indispensable manual for practicing field biologists, and as a graduate-level text for students in ecology, conservation biology, fisheries/wildlife management, and related fields. Provides a synthesis of important classes of models about distribution, abundance, and species richness while accommodating imperfect detection Presents models and methods for identifying unmarked individuals and species Written in a step-by-step approach accessible to non-statisticians and provides fully worked examples that serve as a template for readers' analyses Includes companion website containing data sets, code, solutions to exercises, and further information
Book
Insect Biodiversity: Science and Society brings together leading scientific experts to assess the impact insects have on humankind and the earth's fragile ecosystems. It examines why insect biodiversity matters and how the rapid evolution of insect species is affecting us all. Insects and related arthropods make up more than 50 percent of the known animal diversity globally, yet a lack of knowledge about insects is hindering the advance of science and society. This book explores the wide variety in type and number of insect species and their evolutionary relationships. Case studies offer assessments on how insect biodiversity can help meet the needs of a rapidly expanding human population, and also examine the consequences that an increased loss of insect species will have on the world. The book concludes that a better understanding of the biology and ecology of insects is the only way to sustainably manage ecosystems in an ever changing global environment.
Article
To balance the needs of people and biodiversity in cities, local governments are increasingly incorporating green spaces and urban greening initiatives into urban planning frameworks. Despite this, there is little information on which vegetation features or management actions are most useful in supporting biodiversity within these green spaces. We assess the effect of three vegetation management approaches that are commonly suggested to improve outcomes for urban biodiversity, including: (1) increasing the proportion of native vegetation; (2) increasing the density of trees, and (3) increasing the volume or complexity of understorey vegetation. We use a network of 39 urban green spaces (including golf courses, public parks and residential neighbourhoods) to assess how these vegetation management approaches impact urban bird and bat communities. The richness of both birds and bats increased with an increase in the proportion of native plants, bird species richness increased with increasing volume of understorey vegetation, and increasing large tree density led to increased bat activity. Our study suggests that increasing native vegetation composition and understorey vegetation volume, and the retention of large trees are practical vegetation management approaches that improve outcomes for urban birds and bats. Our data also suggests that as urbanisation and infill development proceeds, retention of large green spaces (such as golf courses) will be critical for urban biodiversity conservation, as these areas supported a component of the bird and bat fauna which was not recorded in other green space types.
Chapter
Distribution, abundance, and species richness are the fundamental state variables in ecology. Surprisingly, though, they are simply derived quantities of a more fundamental quantity, a point pattern, that is, the outcome of a stochastic process that (in the context of distribution and abundance) produces “points” (i.e., animal or plant locations) in a plane and where both the number and the locations are random variables. Presence/absence (the element of a species distribution) and abundance can only be defined once we discretize space; then, abundance is the number of points per spatial unit and occurrence denotes the event that there is at least one point within a site, as opposed to the absence of any point at the site. Thus, abundance is a simple aggregate of a point pattern, and occurrence is a deterministic function of abundance. In addition, species richness is the sum of occurring species, i.e., a quantity naturally derived from the point patterns of all occurring species at some site. We explain and illustrate these fundamental relationships with a simple simulation in program R. We also introduce “meta-population designs,” which is our term for the ubiquitous type of study design wherein distribution, abundance, or related demographic quantities are studied at a collection of spatial replicates. We show that this type of design is extremely common, both in ecology and especially in biodiversity monitoring, and that it comes in a large number of variants. It may include spatial subsampling, when each site is subdivided into nested, smaller sampling units, and of course with temporal replication, when all (or at least some) sites are surveyed multiple times over time, to study the dynamics of a distribution, abundance, or species richness. Applied Hierarchical Modeling for Ecologists (AHM) is all about the analysis of demographic data from meta-population designs using hierarchical models. Historically, there has been a fairly deep division line in ecology between models that focus on state parameters, such as abundance, distribution and species richness, and rate parameters, such as survival rate. Recently, however, and especially in the context of hierarchical modeling and with the remarkable power of BUGS software at your fingertips to fit almost arbitrarily complex models, this divide is being torn down in favor of more general models that may contain elements of both open and closed models, as your questions dictate and the available data allow. The AHM book is written very much in this spirit of a “grand unifying model.” When speaking about measurement error, most people have in mind a special type of process underlying measurements of continuous quantities such as body size or body mass, wherein positive and negative errors cancel out and the average measurement is therefore unbiased. In sharp contrast, the errors in the measurement of demographic quantities such as distribution, abundance or in the parameters driving their change, such as survival probability, are of a very different nature. For them, high and low errors do not normally cancel out, and a bias always remains in the mean regardless of how many times you replicate a measurement. We discuss false-positive and false-negative measurement errors for such discrete quantities. There is a long history of capture-recapture and distance sampling models to accommodate false-negative errors, but the mathematical treatment of false-positive errors is only in its infancy. Hierarchical models are ideally suited for inference about ecological quantities corrected for such measurement errors because it is straightforward to define one submodel for the true ecological state and another submodel that describes the measurement error process. These tremendously powerful models are the subject of this book.