Grillo O., Venora G. (eds.) Biodiversity Loss in a Changing Planet

Подождите немного. Документ загружается.

Destruction of the Forest Habitat in the Tatra National Park, Slovakia

271

similar conditions possible so that they were comparable in terms of properties and features

of the forest ecosystem (Fleischer & Matejka, 2009).

Tabular synthesis confirmed 10 types of phytocenoses, mostly secondary, identifiable in the

calamity area by mere visual observation (Fig. 11). They are: 1. Calamagrostis villosa, 2.

Chamerion angustifolium, 3. Calluna vulgaris, 4. Vaccinium myrtillus, 5. Avenella flexuosa, 6. Picea

abies, 7. Sphagnum magellanicum, 8. Carex rostrata, 9. Juncus effusus, and 10. Veronica officinalis

(Olšavská et al., 2009). Humid substrates are colonized by phytocenoses with the

dominating species of Sphagnum magellanicum and Carex rostrata. The second type of

phytocenose is that of associations with increased share of types requiring the higher N

level: Chamaerion angustifolium and Veronica officinalis. Associations with the dominant

Calamagrostis villosa, are the one most frequent and they form a distinct mosaic along with

overgrowths of Chamaerion angustifolium, in burnt down places. In future it is expected that

Chamaerion angustifolium, will be pushed out by Calamagrostis villosa. Associations of Lariceto-

Picetum especially Vaccinium myrtillus, Avenella flexuosa, Picea abies and the type of Calluna

vulgaris, bound to the most acid soil represent the climax stage of forest.

Fig. 11. Natural revitalization of the damaged area (Photo: M. Kopecká)

Some species may be perfectly capable of surviving in a remnant forest many others may

not. A forest patch is not the same as a piece of original forest: edge effects may now

encroach or even traverse the whole patch. For example, Repel (2008) analysed the breeding

bird assemblage structure; nesting, foraging and migrating guilds; bird and habitat

Biodiversity Loss in a Changing Planet

272

relationship and the seasonal dynamics of bird assemblages within four research plots

assigned by the management of the Tatra National Park.

 Reference stand, not affected by windstorm calamity

 Plot with extracted wood

 Post wild-fire plot

 Not extracted plot

The average density of breeding bird assemblages in reference stand was much higher than

in the plot with extracted wood and wildfire plot. The assemblages on the not extracted plot

had the highest average density. The structure of the breeding bird assemblages was most

influenced by the portion of the not disturbed forest stands in the plot, number of live

standing trees, proportion of dead wood in form of twig heaps, proportion of lying dead

wood, and proportion of stones/stone fields in research plots. Kocian et al. (2005) presume

that the activity of birds plays an especially significant role in foresting and restoration of

forest in the Tatras, as some species (jay, nuthatch nutcracker) propagate natural wood

species, such as beech, Swiss pine and hazel.

The use of forest fragmentation indices in the analysis of forest landscapes offers a great

potential for integration of spatial pattern information in the landscape-ecological

management processes, but requires understanding of the limitations and correct

interpretation of results. Further monitoring of forest fragmentation based on remote

sensing data together with the terrestrial monitoring of natural vegetation development and

dynamics of indicative plant and animal species is necessary to realize the possible

revitalization activities and to mitigate negative effects of the calamity windstorm in the

Tatra region.

4. Conclusion

Natural forest fragmentation is not a new phenomenon in the Tatras. Windthrows have

repeatedly happened in this region in the past (Zielonka et al., 2009) although in a much

smaller scale. Urbanization connected with the human-induced deforestation also played an

important role in the past because of tourism. The main difference between the old practices

and the current deforestation is the difference in scale and rate of increase. In the past, small

patches of pastures or damaged forest appeared in the large forested landscape and they

quickly grew back upon abandonment. What happened during the bora windstorm in 2004

in the Tatra National Park was precisely the opposite: remnant forest patches were left in the

“sea” of the degraded forest landscape.

Anthropogenic disruption of the natural development of the Tatra forest in the past caused

that the status of the forest before the calamity did not correspond to the natural

development at all. Wood species composition and structure on the greater part of the

affected area were not proper for the place. The majority of growths were mature and

resembled an economic forest prepared for harvesting. The growths consisted of slender

and tall trees with high-situated crowns, which are unstable and highly susceptible to the

wind and snow threats. Reasonable and consistent management should insist on growths of

different species and age on small areas, which will ensure ecological stability and

functionality of forest in an acceptable time horizon and simultaneously provide optimal

biotopes for all naturally occurring species. Revitalization of forest affected by the

Destruction of the Forest Habitat in the Tatra National Park, Slovakia

273

windstorm is a very complicated process in terms of expertise, organization and economy.

The aim of the present monitoring is to observe the process, to identify and assess the results

in individual stages. Successful revitalization calls for a new forest-economic concept that is

mosaic growths aided by natural processes and natural succession.

Habitat fragmentation not only reduces the area of available habitat but also can also isolate

populations and increase edge effects. Whatever the combination of biotic and abiotic

changes, the forest patches generally can no longer sustain the production of biodiversity it

once had as a part of a larger forest. Understanding of the possible consequences of forest

fragmentation remains of great concern to conservationists, biologists and landscape

ecologists. The use of forest fragmentation indices in the analysis of forest landscapes offers

a great potential for integration of spatial pattern information in the landscape-ecological

management processes, but also requires understanding of limitations and correct

interpretation of results.

5. Acknowledgment

This paper is one of the outputs of the VEGA Grant Agency Project No 2/0018/10 „Time-

spatial analysis of land use: dynamics of changes, fragmentation and stability assessments

by application of the CORINE land cover data layers“, pursued at the Institute of

Geography of the Slovak Academy of Sciences. Author thanks to K2 Studio for the

photographs.

6. References

Betts, M. G.; Forbes, G. J., Diamond; A.W.& Taylor, P.D. (2006). Independent Effects of

Fragmentation on Forest Songbirds: An Organism-Based Approach. Ecological

Applications, Vol. 16, No. 3, p. 1076-1089, ISSN: 1051-0761

Bruna, E.M. & Kress, J.W. (2002). Habitat Fragmentation and the Demographic Structure of

an Amazonian Understory Herb (Heliconia acuminata). Conservation Biology, Vol.

16, No. 5, p.1256-1266, ISSN: 1523-1739

Crofts, R.; Zupancic-Vicar, M.; Marghescu, T. & Tederko, Z., 2005. IUCN Mission to Tatra

National Park, Slovakia, April 2005. IUCN – the World Conservation Union: 43 pp.

04. 04. 2011 Available from:

http://www.wolf.sk/files/dokumenty/IUCN_EN_zaverecna_sprava_2005.pdf

Cunningham, M. & Moritz, C. (1998). Genetic effects of forest fragmentation on rainforest

restricted lizard (Scincidae: Gnypetoscincus queenslandiae). Biological Conservation,

Vol. 83, No. 1, pp. 19 – 30, ISSN 0006-3207

D'Eon, R. G.; Glenn, S. M.; Parfitt, I. & Fortin M. J. (2002). Landscape connectivity as function

of scale and organism vagility in a real forested landscape. Conservation Ecology.

Vol. 6, No.2, ISSN 1195-5449 31. 03. 2011 Available from:

http://www.ecologyandsociety.org/vol6/iss2/art10/print.pdf

Faaborg, J.; Brittingham, M.; Donovan, T. & Blake, J. (1993). Habitat Fragmentation in the

Temperate Zone: A perspective for managers. In: Status and management of

neotropical migratory birds. General Technical Report RM- 229, Finch, D.M., Stangel, P.

Biodiversity Loss in a Changing Planet

274

W. (eds.), p. 331 – 338, Rocky Mountains Forest and Range Expert Station, U. S.

Department of Agriculture, Forest Service, Fort Collins, Colorado, USA

Fahrig, L. (2003). Effects of habitat fragmentation on biodiversity. Annual Rewiew of Ecology,

Evolution and Systematics, Vol. 34, (November 2003), pp. 487-515, ISSN 1546-2069

Feranec, J. & Oťaheľ, J. (2001). Land cover of Slovakia. Veda , ISBN 80-224-0663-5, Bratislava,

Slovakia

Fleisher, P. & Matejka, F. (2009.) Windfall research in TANAP-2008. Geophysical Institute of

the Slovak Academy of Sciences, Research Station of the TANAP, State Forest of

TANAP, ISBN 978-80-85754-20-9, Bratislava, Slovakia

Franklin, A. B.; Noon, B. R. & George L. T. (2002).What is habitat fragmentation? Studies in

Avian Biology, Vol. 25, pp. 20 – 29, ISSN 0197-9922

Gibbs, J. P. (2001). Demography versus habitat fragmentation as determinants of genetic

variation in wild populations. Biological Conservation, Vol. 100,No 1, pp.15 – 20,

ISSN 0006-3207

Izakovičová, Z.; Boltižiar M.; Celer, S.; David, S.; Ditě, D.; Gajdoš, P.; Hreško, J.; Ira, V.;

Grotkovská, L.; Kenderessy, P.; Kozová, M.; Oszlányi, J.; Petrovič, F.; Válkovcová,

Z. & Vološčuk, I.(2008). Krajinnoekologicky optimálne priestorové a funčné využitie

územia biosférickej rezervácie Tatry. Veda, ISBN 978-80-224-0998-8, Bratislava,

Slovakia

Jankovič, J.(2007). Projekt revitalizácie lesných ekosystémov na území Vysokých Tatier postihnutom

veternou kalamitou dňa 19. 11. 2004. National Forest Centre, Zvolen , Slovakia, 31. 3.

2011 Available from:

http://www.nlcsk.sk/files/57.pdf

Keitt, T. H.; Urban, D. L.& Milne, B. T. (1997). Detecting critical scales in fragmented

landscapes. Conservation Ecology, Vol. 1, No.1, ISSN 1195-5449 Available from:

http://www.ecologyandsociety/vol1/iss1/art4

Kocian, Ľ.; Topercer, J.; Baláž, E. & Fiala, J. (2005). Breeding birds and their habitat

requirements in the windthrown area of the Tatra National Park (Slovakia). Folia

faunistica Slovaca, Vol. 10, No. 9, pp. 37 – 43, ISSN 1336-4529

Kopecká, M. & Nováček, J. (2008). Hodnotenie fragmentácie krajinnej pokrývky na báze

dátových vrstiev CORINE Land Cover. Geografický časopis, Vol 60, No. 1, pp. 31-43,

ISSN 0016-7193

Kopecká, M. & Nováček, J. (2010). Natural forest fragmentation: An Example from the Tatra

Region, Slovakia. In: Land Use/Cover Changes in Selected Regions in the World. Bičík, I.,

Himiyama, Y., Feranec, J. (eds.), pp. 51- 56, International Geographical Union

Commission on Land Use and Land Cover Change, ISBN 978-4-907651-05-9

Asahikawa, Japan

Kummerle, T.; Radeloff, V. C.; Perzanowski, K. & Hostert, P. (2006). Cross-border

comparison of land cover and landscape pattern in Eastern Europe using a hybrid

classification technique. Remote Sensing of Environment, Vol. 103, No. 4, pp.449-464,

ISSN 0034-4257

Kurosawa, R. & Askins, R. A. (2003). Effects of Habitat Fragmentation on Birds in Deciduous

Forests in Japan. Conservation Biology, Vol. 17, No. 3, p.695 – 707, ISSN 1523-1739

Destruction of the Forest Habitat in the Tatra National Park, Slovakia

275

Lord, J. M. & Norton, D. A. (1990). Scale and the spatial concept of fragmentation.

Conservation Biology , Vol. 4, No. 2, pp. 197 –202, ISSN 1523-1739

Madera, P. & Zimová, E. (2004). Metodické postupy a projektování lokálního ÚSES. Ústav

lesnické botaniky, denrologie a geobiocenologie, Brno, Czech republic , 31. 03. 2011,

Available from:

http://www2.zf.jcu.cz/public/departments/kpu/vyuka/tvok/metodika_uses.

pdf

McGarigal, K. & Marks., B. J. (1995). FRAGSTATS: Spatial Pattern Analysis Program for

Quantifying Landscape Structure. United States Department of Agriculture, Forest

Service, Pacific Northwest Research Station, Portland, USA , 31. 03. 2011, Available

from: http://www.fs.fed.us/pnw/pubs/pnw_gtr351.pdf

Olšavská, G.; Križová E. & Šoltés, R. (2009). Pokalamitý vývoj vegetácie na trvalo

monitorovacích plochách vo Vysokých Tatrách. In: Windfall research in TANAP-

2008. Fleisher, P. & Matejka, F. (eds.), pp. 172 – 182, Geophysical Institute of the

Slovak Academy of Sciences, Research Station of the TANAP, State Forest of

TANAP, ISBN 978-80-85754-20-9, Bratislava, Slovakia

Parker, T. H.; Stanberry, B. M.; Becker, C. D. & Gipson, P. S. (2005). Edge and Area Effects

on the occurrence of Migrant Forest Songbirds. Conservation Biology, Vol. 19, No. 4,

pp. 1157-1167, ISSN 1523-1739

Parent, J., & Hurd, J. (2008). An improved method for classifying forest fragmentation. CLEAR, 02.

04. 2011 Available from:

http://clear.uconn.edu/publications/research/presentations/parent_nearc2008_ff

.ppt

Repel, M. (2008). Diverzita, denzita a potravné vzťahy zoskupení vtákov vo Vysokých Tatrách

postihnutých vetrovou kalamitou. PhD. thesis, Technická univerzita, Zvolen, Slovakia

Riitters, K. H.; Wickham, J. D.; O'Neil, R.V.; Jones, K. B.; Smith, E. R.; Coulston, J. W.; Wade,

T. G. & Smith, J. H. (2002). Fragmentation of Continental United States Forests.

Ecosystems, Vol. 5, No. 8, pp. 815 – 822, ISSN 1435-0629

Riitters, K. H. (2005). Downscaling indicators of forest habitat structure from national

assessments. Ecological Indicators , Vol. 5, No. 4, pp. 273 – 279, ISSN 1470-160X

Trzcinski, M. K.; Fahrig, L.& Merriam, G. (1999). Independent effects of forest core and

fragmentation on the distribution of forest breeding birds. Ecological Applications,

Vol. 9, No. 2., pp. 586-593, ISSN 1051-0761

Valladares, G.; Salvo, A. & Cagnolo, L. (2006). Habitat Fragmentation Effects on Trophic

Processes of Insect-Plant Food Webs. Conservation Biology, Vol. 20, No.1, pp. 212-

217, ISSN 1523-1739

Vogt, P., Riitters, K. H.; Estreguil C.; Kozak, J.; Wade, T. G. & Wickham, J. D. (2007).

Mapping spatial patterns with morphological image processing. Landscape Ecology,

Vol. 22, No. 2, pp. 171 – 177, ISSN 1572-9761

Zielonka, T.; Holeksa, J.; Malcher, P. & Fleischer, P., (2009). A two-hundred year history of

spruce – larch stand in the Slovakian High Tatras damaged by windstorm in 2004.

In: Windfall research in TANAP-2008. Fleisher, P. & Matejka, F. (eds.), pp. 269 - 274,

Geophysical Institute of the Slovak Academy of Sciences, Research Station of the

TANAP, State Forest of TANAP, ISBN 978-80-85754-20-9, Bratislava, Slovakia

Biodiversity Loss in a Changing Planet

276

Zipperer, W., C. (1993). Deforestation patterns and their effects on forest patches. Landscape

Ecology, Vol. 8, No. 3, pp. 177–184, ISSN 1572 - 9761

Modern Methods of Estimating Biodiversity from

Presence-Absence Surveys

Robert M. Dorazio

, Nicholas J. Gotelli

and Aaron M. Ellison

U.S. Geological Survey, Southeast Ecological Science Center, Gainesville, Florida

University of Vermont, Department of Biology, Burlington, Vermont

Harvard University, Harvard Forest, Petersham, Massachusetts

USA

1. Introduction

Communities of species are often sampled using so-called “presence-absence” surveys,

wherein the apparent presence or absence of each species is recorded. Whereas counts of

individuals can be used to estimate species abundances, apparent presence-absence data are

often easier to obtain in surveys of multiple species. Presence-absence surveys also may

be more accurate than abundance surveys, particularly in communities that contain highly

mobile species.

A problem with presence-absence data is that observations are usually contaminated by zeros

that stem from errors in detection of a species. That is, true zeros, which are associated with the

absence of a species, cannot be distinguished from false zeros, which occur when species are

present in the vicinity of sampling but not detected. Therefore, it is more accurate to describe

apparent presence-absence data as detections and non-detections, but this terminology is

seldom used in ecology.

Estimates of biodiversity and other community-level attributes can be dramatically affected

by errors in detection of each species, particularly since the magnitude of these detection

errors generally varies among species (Boulinier et al. 1998). For example, bias in estimates

of biodiversity arising from errors in detection is especially pronounced in communities

that contain a preponderance of rare or difﬁcult-to-detect species. To eliminate this source

of bias, probabilities of species occurrence and detection must be estimated simultaneously

using a statistical model of the presence-absence data. Such models require presence-absence

surveys to be replicated at some – but not necessarily all – of the locations selected for

sampling. Replicate surveys can be obtained using a variety of sampling protocols, including

repeated visits to each sample location by a single observer, independent surveys by different

observers, or even spatial replicates obtained by placing clusters of quadrats or transects

within a sample location. Information in the replicated surveys is crucial because it allows

species occurrences to be estimated without bias by using a model-based speciﬁcation of the

observation process, which accounts for the errors in detection that are manifest as false zeros.

Several statistical models have been developed for the analysis of replicated, presence-absence

data. Each of these models includes parameters for a community’s incidence matrix (Colwell

2 Will-be-set-by-IN-TECH

et al. 2004, Gotelli 2000), which contains the binary occupancy state (presence or absence)

of each species at each sample location. The incidence matrix is only partially observed

owing to species- and location-speciﬁc errors in detection; however, the incidence matrix

can be estimated by ﬁtting these models to the replicated, presence-absence data. Therefore,

any function of the incidence matrix – including species richness, alpha diversity, and beta

diversity (Magurran 2004)– also can be estimated using these models.

Models for estimating species richness – and other measures of biodiversity – from replicated,

presence-absence data were ﬁrst developed by Dorazio & Royle (2005) and Dorazio et al.

(2006). By including spatial covariates of species occurrence and detection probabilities

in these models, Kéry & Royle (2009) and Royle & Dorazio (2008) estimated the spatial

distribution (or map) of species richness of birds in Switzerland. Similarly, Zipkin et al. (2010)

showed that this approach can be used to quantify and assess the effects of conservation

or management actions on species richness and other community-level characteristics.

More recently, statistical models have been developed to estimate changes in communities

from a temporal sequence of replicated, presence-absence data. In these models the

dynamics of species occurrences are speciﬁed using temporal variation in covariates of

occurrence (Kéry, Dorazio, Soldaat, van Strien, Zuiderwijk & Royle 2009) or using ﬁrst-order

Markov processes (Dorazio et al. 2010, Russell et al. 2009, Walls et al. 2011), wherein

temporal differences in occurrence probabilities are speciﬁed as functions of species- and

location-speciﬁc colonization and extinction probabilities. The latter class of models, which

includes the former, is extremely versatile and may be used to confront alternative theories

of metacommunity dynamics (Holyoak & Mata 2008, Leibold et al. 2004) with data or to

estimate changes in biodiversity. For example, Dorazio et al. (2010) estimated regional levels

of biodiversity of butterﬂies in Switzerland using a model that accounted for seasonal changes

in species composition associated with differences in phenology of ﬂight patterns among

species. Russell et al. (2009) estimated the effects of prescribed forest ﬁre on the composition

and size of an avian community in Washington.

In the present paper we analyze a set of replicated, presence-absence data that previously

was analyzed using statistical models that did not account for errors in detection of each

species (Gotelli & Ellison 2002). Our objective is to illustrate the inferential beneﬁts

of using modern methods to analyze these data. In the analysis we model occurrence

probabilities in assemblages of ant species as a function of large-scale, geographic covariates

(latitude, elevation) and small-scale, site covariates (habitat area, vegetation composition, light

availability). We ﬁt several models, each identiﬁed by a speciﬁc combination of covariates, to

assess the relative contribution of these potential sources of variation in species occurrence

and to estimate the effect of these contributions on geographic differences in ant species

richness and other measures of biodiversity. We also provide the data and source code used in

our analysis to allow comparisons between our results and those obtained using alternative

methods of analysis.

2. Study area and sampling methods

2.1 Ant sampling

The data in our analysis were obtained by sampling assemblages of ant species found in

New England bogs and forests. The initial motivation for sampling was to determine the

278

Biodiversity Loss in a Changing Planet

Modern Methods of Estimating Biodiversity from Presence-Absence Surveys 3

extent of the distribution of the apparent bog-specialist, Myrmica lobifrons, in Massachusetts

and Vermont. Bogs are not commonly searched for ants, but in 1997 we had identiﬁed

M. lobifrons as a primary component of the diet of the carnivorous pitcher plant, Sarracenia

purpurea, at Hawley Bog in western Massachuestts. This was the ﬁrst record for M. lobifrons

in Massachusetts. At the time the taxonomic status of this species was being re-evaluated

(Francoeur 1997), and it was largely unknown in the lower (contiguous) 48 states of the United

States. In addition to our interest in M. lobifrons, we also wanted to explore whether bogs

harbored a distinctive ant fauna or whether the ant faunas of bogs were simply a subset of the

ant species found in the surrounding forests. Thus, at each of the sites selected for sampling,

we surveyed ants in the target bog and in the upland forest adjacent to the bog (Gotelli &

Ellison 2002).

At each of 22 sample sites, we established two 8

× 8 m sampling grids, each containing 25

evenly spaced pitfall traps. One sampling grid was located in the center of the bog; the other

was located within intact forest 50-500 m away from the edge of the bog. Each pitfall trap

consisted of a 180-ml plastic cup (95 mm in diameter) that was ﬁlled with 20 ml of dilute

soapy water. Traps were buried so that the upper lip of each trap was ﬂush with the bog

or forest-soil surface, and left in place for 48 hours during dry weather. At the end of the

48 hours, trap contents were collected, immediately ﬁxed with 95% ethanol, and returned to

the laboratory where all ants were removed and identiﬁed to species. Traps were sampled

twice in the summer of 1999, and the time between each sampling period was 6 weeks (42

days); therefore, we consider the two sampling periods as early- and late-summer replicates.

Locations of traps were ﬂagged so that pitfall traps were placed at identical locations during

the two sampling periods.

2.2 Measurement of site covariates

The geographic location (latitude (LAT) and longitude (LON)) and elevation (ELEV, meters

above sea level) of each bog and forest sample site was determined using a Trimble Global

Positioning System (GPS). At each forest sample site we also estimated available light levels

beneath the canopy using hemispherical canopy photographs, which were taken on overcast

days between 10:00 AM and and 2:00 PM at 1 m above ground level with an 8 mm ﬁsh-eye

lens on a Nikon F-3 camera. Leaf area index (LAI, dimensionless) was determined from

the subsequently digitized photographs using HemiView software (Delta-T, Cambridge, UK).

Because there was no canopy over the bog, the LAI of each bog was assigned a value of zero.

To compute a global site factor (GSF, total solar radiation) for each forest sample site (Rich

et al. 1993), we summed weighted values of direct site factor (DSF, total direct beam solar

radiation) and indirect site factor (total diffuse solar radiation). GSF values are expressed as a

percentage of total possible solar radiation (i.e., above the canopy) during the growing season

(April through October), corrected for latitude and solar track. The GSF of each bog was

assigned a value of one.

Digital aerial photographs were obtained for each sampled bog from state mapping

authorities, or, when digital photographs were unavailable (ﬁve sites), photographic prints

(from USGS-EROS) were scanned and digitized. Aerial photographs were used to construct a

set of data layers (Arc-View GIS 3.2) from which bog area (AREA) was calculated. The area of

the surrounding forests was not measured, as the forest was generally continuous for at least

several km

around each bog.

279

Modern Methods of Estimating Biodiversity from Presence-Absence Surveys

4 Will-be-set-by-IN-TECH

3. Statistical analysis

We analyzed the captures of ant species observed at our sample sites using a modiﬁcation

of the multi-species model of occurrence and detection that includes site-speciﬁc covariates

(Kéry & Royle 2009, Royle & Dorazio 2008). This modiﬁcation allows a ﬁnite set of candidate

models to be speciﬁed and ﬁt to the data simultaneously such that prior beliefs in each model’s

utility can be updated (using Bayes’ rule) to compute the posterior probability of each model.

The resulting set of posterior model probabilities can be used to select a single (“best”) model

for inference or to estimate scientiﬁcally relevant quantities while averaging over the posterior

uncertainty of the models (Draper 1995).

To compare our results with previous analyses (Gotelli & Ellison 2002), we analyzed the

data observed in bogs and forests separately. These two habitats are sufﬁciently distinct

that differences in species occurrence – and possibly capture rates – are expected a priori.

Furthermore, the potential covariates of occurrence differ between the two habitats, adding

another reason to analyze the bog and forest data separately.

3.1 Hierarchical model of species occurrence and capture

We summarize here the assumptions made in our analysis of the ant captures. Let y

∈

{

0, 1, . . . , J

} denote the number of pitfall traps located at site k that contained the ith of n

distinct species of ants captured in the entire sample of R

= 22 sites. At each site 25 pitfall

traps were deployed during each of 2 sampling periods (early- and late-season replicates);

therefore, the total number of replicate observations per site was constant (J

= 50). While

constant replication among sites simpliﬁes implementation of the model, it is not required.

However, it is essential that J

> 1 for some (ideally all) sample sites because information

from within-site replicates allows both occurrence and detection probabilities to be estimated

for each species. In the absence of this replication these two parameters are confounded.

The observed data form an n

× R matrix Y

obs

of pitfall trap frequencies, so that rows are

associated with distinct species and columns are associated with distinct sample sites. Note

that n, the number of distinct ant species observed among all R sample sites, is a random

outcome. In the analysis we want to estimate the total number of species N that are present

and vulnerable to capture. Although N is unknown, we know that n

≤ N, i.e., we know that

the number of species observed in the samples provides a lower bound for an estimate of N.

To estimate N, we use a technique called parameter-expanded data augmentation (Dorazio

et al. 2006, Royle & Dorazio 2011), wherein rows of all-zero trap frequencies are added to the

observed data Y

obs

and the model for the observed data is appropriately expanded to analyze

the augmented data matrix Y

=(Y

obs

, 0). The technical details underlying this technique are

described by Royle & Dorazio (2008, 2011), so we won’t repeat them here. Brieﬂy, however,

the idea is to embed the unobserved, all-zero trap frequencies of the N

− n species in the

community within a larger data set of ﬁxed, but known size (say, M species, where M

> N)

for the purpose of simplifying the analysis. The conventional model for the community of

N species is necessarily modiﬁed so that each of the M

− n rows of augmented data can be

estimated as either belonging to the community of N species (and containing sampling zeros)

or not (and containing structural zeros). In particular, we add a vector of parameters w

(

,...,w

) to the model to indicate whether each species is a member of the community

= 1) or not (w = 0). The elements of w are assumed to be independentally and identically

280

Biodiversity Loss in a Changing Planet