Ecological Archives E094-194-D1

Paula Casanovas, Heather J. Lynch, William F. Fagan, Ron Naveen. 2013. Understanding lichen diversity on the Antarctic Peninsula using parataxonomic units as a surrogate for species richness. Ecology 94:2110. http://dx.doi.org/10.1890/13-0237.1


Introduction

The biogeography of the Antarctic Peninsula is unique because of its pronounced latitudinal gradients and its geographic isolation (Fenton 1982). Lichens and bryophytes are the dominant and most diverse macrophytes in the terrestrial Antarctic ecosystem (Smith 1984). There are 404 species of lichens (Øvstedal and Smith 2001, 2004, 2009) and 111 species of mosses (Ochyra et al. 2008) described for the whole Antarctic Continent. In the Antarctic Peninsula region, crustose, fruticose, and foliose lichens comprise extensive cryptogamic communities, especially along the coast (Convey et al. 2008), with 264 species represented (Øvstedal and Smith 2001, 2004, 2009). In a recent study of lichen richness on the Antarctic Peninsula, it was shown that patterns of richness observed using museum records are highly scale-dependent and largely unexplained by the biogeographic variables found important in other systems (Casanovas et al. 2013). However, data available for the study were limited and exhibited significant heterogeneity in sampling frequency and sampling effort along the Antarctic Peninsula.

The input of nutrients by seabirds and seals on the islands where they breed is an important factor that influences lichen diversity (Ryan and Watkins 1989, Favero-Longo et al. 2001, Leishman and Wild 2001, Smykla et al. 2007, Smith 2008). It has been shown that penguin colonies affect the diversity of lichens at both the local scale (richness and composition of the lichen communities change as they approach the edge of a penguin colony, Smykla et al. 2007) and at the island scale (lichen richness increases with colony size up to 20,000 breeding pairs, Casanovas et al. 2013).

There are extensive areas of the Antarctic Peninsula where lichen specimens have never been collected (e.g., between 61°S and 62°S, Casanovas et al. 2013), and many sites frequently visited by tourists lack any floristic information (Peat et al. 2007, Chown and Convey 2007, Terauds et al. 2012, Casanovas et al. 2013). There are several ice-free terrestrial habitats in the Antarctic that are not represented in the current network of Antarctic protected areas (Teradus et al. 2012). The Antarctic Peninsula climate is changing dramatically (Convey 2011), and biodiversity inventories are fundamental for establishing baseline conditions against which to judge changes in floral abundance and composition.

Protecting the flora of the Antarctic Peninsula is important to the Antarctic Treaty Parties; Visitor Site Guidelines explicitly mention the vegetation of some landing sites as a priority for conservation (e.g., Barrientos Island), with trampling and damage of vegetation as one of the potential human impacts at several popular tourist landing areas (Tejedo et al. 2009, 2012). However, it is difficult to identify appropriate areas for lichen conservation because comprehensive data on lichens are neither available for most of Antarctica nor compiled for most Antarctic lichen species.

Expert collection of specimens in the field coupled with subsequent determination of species is the best method for determining species richness. However, the relative paucity of botanists working in Antarctica makes this approach impractical for broad-scale surveys of Antarctic floral biodiversity. In order to accelerate assessments of local biodiversity and help select areas for conservation, cost-effective survey methods and surrogate methods for the prediction of species richness are needed. A combination of a photographic "citizen scientist" approach for the collection of data, and the use of parataxonomic units (PUs; a visually distinguishable unit based only on external morphology, Krell 2004) as a basis for quantifying richness, might be a possible solution to effectively collect preliminary information and rapidly build databases on species diversity. Parataxonomy has been used successfully for the prediction of species richness in different communities, including aquatic macroinvertebrates (Clarke, Lake, and O’Dowd 2004), colonial reef corals (Budd, Johnson, and Potts 1994), vascular plants (Garrettson et al. 1998), and insects (Oliver and Beattie 1996a,b; Basset et al. 2000).

A "citizen scientist" is a non-expert person who helps collect or process data with a scientific purpose (Cohn 2008). The direct participation of citizen scientists in data collection efforts provides information on spatial and temporal scales that are impossible to collect using traditional methods (Silvertown 2009; Conrad and Hilchey 2011; Dickinson et al. 2012). The participation of lay persons in large-scale regional surveys has been successfully demonstrated in several projects (e.g., the Open Air Laboratories (OPAL) network in England, www.opalexplorenature.org; the North America Breeding Bird surveys and Christmas Bird counts, Butcher and Niven 2007).

In this paper, we show how such a diversity database can be built using both, parataxonomy and citizen scientist collaborations. We combine a photographic survey protocol used by non-scientist visitors to the Antarctic with parataxonomic classification to catalog lichens at 29 locations on the Antarctic Peninsula. To test the identification capabilities of the parataxonomic classifications, we developed a reference photographic data set using Antarctic lichen collection from the U.S. National Herbarium. We also compared lichen PU richness with species richness for the limited number of sites where historical information was available.

From photographs taken in the three austral summers between 2009/10 and 2011/12, different lichen PUs were isolated in the lab, and cataloged as "specimens". To date we have collected 1804 specimens for identification purposes. We estimated PU richness as a proxy for species and genus richness for each of the sites surveyed. Using these data, we provided three examples of data applications, from basic ecological questions on community compositions and relationships between lichens and other taxa, to specific questions directly related to the conservation of the Antarctic Peninsula flora. Additionally, these surveys provide preliminary information for identifying areas for protection and priorities for future research.

 

Metadata

Class I. Data set descriptors

A. Data set title: Database of Antarctic Peninsula lichen pictures and classifications of parataxonomic units.

B. Data set identification code: Antarctic_Lichens_db.zip

C. Data set updates: The data set will be updated as part of the Antarctic Site Inventory long term monitoring program. New data will be posted on the webpage of Oceanites (a nonprofit Antarctic research group) at www.oceanites.org; and on the Antarctic Biodiversity (ANTABIF) Data Portal at http://www.biodiversity.aq/data.html.

D. Principal investigators: Same as authors.

E. Abstract: Expert collection of specimens in the field and further determination of species is the best method for determining species richness. However, the relative paucity of botanists working in Antarctica makes this approach impractical for broad-scale surveys of Antarctic floral biodiversity. Lichens are the dominant macrophytes of terrestrial Antarctica and, as such, play a fundamental part of the ice-free terrestrial ecosystem. Many distinct ice-free terrestrial habitats in the Antarctic are not represented in the current network of Antarctic protected areas. However, it is difficult to identify appropriate areas for conservation because comprehensive data on distributional patterns of Antarctic flora are not available, and existing data for most Antarctic lichen species are not compiled. Consequently, cost-effective survey methods and surrogates for the prediction of species richness are needed to accelerate assessments of local biodiversity and help select areas for conservation. A combination of a photographic “citizen scientist” approach for the collection of data, and the use of parataxonomic unit (PU) richness as a surrogate for species richness, might be a possible solution to effectively collect preliminary information and rapidly build databases on species diversity. We have developed a database and gathered photographic information on lichen occurrences for sites that are frequently visited by tourists. We test the identification capabilities with a reference dataset of Antarctic lichen images from the U.S. National Herbarium, and showed that all species used in this test can be detected, and that for 74% of the images, all classifiers were able to identify the genus of the specimen. Twenty-nine sites were photographically surveyed by researchers and tourists between 2009/10 and 2011/12 in the Antarctic Peninsula region. We estimated PU richness as a proxy for species richness for each of the 29 sites surveyed, and provide two examples of potential applications. These surveys provide preliminary information for identifying areas for protection and priorities for future research.

F. Key words: Antarctic Peninsula; citizen science; detectability; historical data sets; lichens; parataxonomic units; penguin colonies.

Class II. Research origin descriptors

A. Overall project description: As part of the Antarctic Site Inventory (e.g., Lynch et al. 2012, Naveen and Lynch 2011), we have developed a database and gathered photographic information on lichen richness for sites that are frequently visited by tourists on the Antarctic Peninsula.

B. Period of Study: 2009–2010 austral Summer – to date.

C. Geography: Antarctic Peninsula

D. Research motivation

Lichens are the most diverse macrophytes on the Antarctic Peninsula. Their biodiversity and biogeography are particularly interesting in this area because suitable habitat is patchy at multiple spatial scales and may be changing rapidly due to climate change. It has been shown from museum records that patterns of lichen richness are highly scale-dependent and largely unexplained by biogeographic variables found to be important in other systems. However, previous studies have been hampered by significant heterogeneity in sampling frequency and sampling effort along the Antarctic Peninsula.

There are extensive areas of the Antarctic Peninsula where lichen specimens have never been collected (e.g., between 61°S and 62°S, Casanovas et al. 2013), as well as many sites frequently visited by tourists that lack any floristic information (Peat et al. 2007, Casanovas et al. 2013). The Antarctic Peninsula climate is changing dramatically (Convey 2011), and biodiversity inventories are fundamental for establishing baseline conditions against which to judge changes in floral abundance or composition. The protection of the Antarctic Peninsula flora is important to the Antarctic Treaty Parties; Visitor Site Guidelines explicitly mention the vegetation of some landing sites as a priority for conservation (e.g., Barrientos Island), with trampling and damage of vegetation as one of the potential human impacts at several popular tourist landing areas (Convey 2010, Stewart et al. 2005, Tejedo et al. 2009, 2012). However, it is difficult to identify appropriate areas for conservation because comprehensive data on flora are not available for most of Antarctica, nor are they compiled for most Antarctic lichen species.

E. Research methods

PHOTOGRAPHIC DATA SET

To date, 29 sites have been surveyed along the Antarctic Peninsula. Between 1 and 7 (mean = 3) non-scientist photographers participated in each survey, and between 1 and 7 (mean = 1) separate surveys were completed at each site. The backgrounds of the photographers were variable, but none of them were experts on lichens.

According to the study protocol, each participant took pictures of every macroscopic lichen seen at a given site, walking freely within the limits of the site in the time available on shore (usually 2–3 hours). Using a black and white control scale included in each photo, we standardized the white balance in each picture and digitally isolated all lichens from the photographs using the ImageJ image processing program (Abramoff et al. 2004, Ferreira and Rasband 2010). Each lichen was given a unique identification name and was considered a "digital specimen" for this paper. Once all specimens were isolated, three different research assistants (hereafter "classifiers") simultaneously and independently classified them using the photographs and descriptions in Øvstedal and Smith (2001, 2004, 2009) as a guide. As many lichen species are difficult or impossible to identify without a physical specimen, we consider determinations from photographers to represent parataxonomic units (PUs) and not true species. The number of specimens and PUs identified among them depended on the number of photographers, the quality of the photographs and specimens, and varied across sites.

 

REFERENCE DATA SET

With the objective of testing the identification capabilities of the classifiers (i.e., the similarity between the PU determination and the actual species being identified), we developed a reference photographic dataset for which the species photographed were known. Although many species cannot be determined using external morphology alone, the classifiers used the photographs and descriptions in Øvstedal and Smith (2001, 2004, 2009) as a guide; we expected that some species (and genera) would match the parataxonomic classifications. The images from this reference dataset were not used to compare them with the images taken in the field.

In collaboration with the Core Collections Management project at the Department of Botany at the Smithsonian Institution, we located, imaged, and created appropriate metadata for 39 Antarctic physical specimens representing 12 species and 9 genera of lichens. The U.S. National Herbarium hosts lichen collections from multiple trips by Mason Hale to the Antarctic (1980–1985), as well as duplicates of Antarctic lichens obtained from other herbaria. These specimens were identified by different lichenologists from the herbaria to the species level. For each physical specimen in the collection, we took photos of the sheet on which the collection is maintained and multiple close-ups. These data (e-records and images) were incorporated into the EMu catalog and are available to lichen researchers worldwide. For our reference data set, we called each of the close-up images a "digital specimen". Three different classifiers, acting independently and without knowing the actual species the digital specimen belongs to, classified the lichen digital specimens from this reference data set using Øvstedal and Smith (2001, 2004, 2009) as a guide.

 

DATA ANALYSES

PU identifications and comparison with species identifications from the reference data set:

All images from the reference data set were identified to PU independently by three classifiers. A total of 12 species and 7 genera were represented in the original data set, and the PU results showed a mean of 18 PU "species" and 8 PU "genera" (using Øvstedal and Smith 2001, 2004, 2009 as a guide). The PU classifications were compared against the species classifications of the physical specimens made by expert lichenologists (which have been re-checked by experts at the museum). For 74 % of the images, all three classifiers identified the genus of the specimen correctly, and for 89 % of the images at least one classifier identified the genus correctly. All three classifiers identified the species correctly in only 13 % of the images, but at least one classifier identified the species correctly for 69 % of the images. All twelve species present in the data set were detected. This methodology shows to be effective for detecting the genera and species present in the lichen communities, and can be used for describing the diversity of genus in a given area with a useful level of accuracy.

Estimation of PU richness for different sites

Estimates of species richness are always dependent on sampling effort (Boulinier et al. 1998), and many PU will remain undetected due to the differential abundance and detection probability of different types of organisms (MacKenzie et al. 2002). Because of this, we estimated true species richness using the non-parametric estimator Chao2 (Chao et al. 2005). The estimations of lichen PU richness are listed in "PU richness by sites" under the Data structural descriptors section.

We also estimated the total PU richness for the South Shetland Islands and the entire Antarctic Peninsula (including all adjacent islands and the South Shetland Islands). Table 1 shows the observed and estimated number of "genera" and "species" PUs, as well as the number of actual genera and species known for those areas (from Øvstedal and Smith 2001).

Table 1. Comparison between estimated and observed PU numbers from the photographic documentation and known genera and species numbers, for the Antarctic Peninsula and the South Shetland Islands.

 

Observed "species"
PU numbers

Estimated "species"
PU numbers

Known species
numbers

Observed "genera"
PU numbers

Estimated "genera"
PU numbers

Known genera
numbers

South Shetland Islands

105.0 (± 5.5)*

170.3 (± 24.2)*

211

40.3 (± 0.5)*

49.3 (± 4.1)*

73

Antarctic Peninsula

129.6 (± 11.9)*

184.6 (± 39.3)*

264

44.6 (± 4.0)*

53.1 (± 4.9)*

89

*Mean among three classifiers, the number between parenthesis represent one standard deviation.


Comparison with historic data sets

Seven sites surveyed with the photodocumentation protocol had been surveyed before with estimations of species richness reported by Casanovas et al. (2013). Table 2 shows the observed number of species and PUs for each of these sites, as well as the number of physical specimens and digital specimens collected. Except for two cases (Half Moon Island and Petermann Island), the number of species from historical records and the number of PUs are very similar.

Table 2. Comparison between observed PU numbers from the photographic documentation and observed species numbers from historical records. The number of physical and digital specimens is given only as a reference.

Site (listed from north to south)

Observed "species"
PU numbers
(mean among three classifiers)

Observed species
numbers

Number of physical
specimens (historical data)

Number of digital
specimens (PU data)

Half Moon Island

33.0

13

21

273

Whalers Bay

42.6

49

107

652

Cuverville Island

31.6

34

62

307

Damoy Point

6.3

4

7

56

Brown Station

8.6

5

9

33

Petermann Island

10.0

35

74

59

Horseshoe Island

13.0

17

24

93

 

Examples of data applications

Detection model for estimating species richness and probabilities of detection and presence

Using the photodocumentation protocol and analysis of PUs, we used a detection model to estimate PU richness. This approach calculates the probability of each PU being detected in a given survey by a photographer and a classifier. We used a modified version of the hierarchical model proposed by Dorazio et al. (2006), using multiple classifiers as replicates for each photographer (analogous to multiple observers surveying during the same visit), and multiple photographers at the same site (in lieu of temporal replictions). We calculated the mean species richness, as well as the occupancy and detection probabilities for each of the observed PU genera. We implemented this procedure with two sites that had been visited the most times (Whalers Bay and Jougla Point, seven visits each). Model parameters were given vague normal prior distributions and the precisions associated with model synchrony and model error were given vague gamma distributions. Models were fit to the data using the software package WinBUGS (Lunn et al. 2000, Sturtz et al. 2005) using R (R Development Core Team 2010). We used a burn-in period of 1,000 samples and drew our posterior distribution from the following 9,000 samples in each of three overdispersed and randomly-initialized parallel Markov chain Monte Carlo chains. This was more than sufficient to achieve model convergence and adequate sampling of the posterior distribution (Appendix 1).

For the two different sites, PU occupancy and detection probabilities exhibited different patterns, and for the same PU these probabilities were different in the two sites in most cases (Fig. 1). However, some PUs exhibited similar probabilities of detection and occupancy at both sites (e.g., Acarospora and Turgidoscolum). Most PUs found in this study are common genera of macrolichens in the Antarctic Peninsula, and thus the probability of occurrence is similar for most PUs. The mean PU "genera" richness for Whalers Bay was 45.5 (± 10.4) and the mean PU "genera" richness for Jougla Point was 50.9 (± 11.5).

Penguin-lichen relationships

On the Antarctic Peninsula, penguins nest in colonies, and through their accumulated excreta, these penguins contribute significantly to the local nutrient status of the substratum (Myrcha et al. 1985, Smith 1985, Tatur et al. 1997). Penguin colonies vary in size from one site to another, from a few nests to >100,000 nests. At a local scale, the input of nutrients by colonies of penguins and other seabirds has a major influence on floristic richness (Smith 1978, Tatur 2002, Smykla et al. 2007).

Here we look at correlations between lichen PU richness and penguin colony size for the 22 sites (of 29) where sufficient information was available. To calculate penguin colony size at a site, we summed the per-visit abundances of three co-occurring penguin species (Pygoscelis antarctica, P. papua, and P. adeliae) and then averaged those abundance across visits. Average penguin colony sizes at the 22 sites ranged from 271 to 6260 nests. All the colonies are located at the same snow- and ice-free coastal areas delimited in this study. The information on penguin colony sizes comes from the Antarctic Site Inventory database (Naveen and Lynch 2011).

We found a significant correlation between the observed number of PUs present (using observed or estimated PU genera or PU species) and the size of the penguin colonies (Table 3). These results concurred with results on Casanovas et al. (2013) which found a significant correlation between lichen species richness and penguin colony sizes, using an independent dataset only derived from historical records.

Table 3. Correlation coefficient between PU richness and penguin colony size.

PU richness

Adjusted R²

p

Outliers?

Observed number of PU "genera"

0.23

0.01

none

Chao2 estimated number of PU "genera"

0.09

0.09

Georges Point

Observed number of PU "species"

0.26

0.006

none

Chao2 estimated number of PU "species"

0.27

0.006

none

Tourism-lichen relationships

Tourism in the Antarctic Peninsula has been increasing since the late 1980s. Environmental impact is now one of the most important issues surrounding tourism in Antarctica (Stewart et al. 2005).

Both tourists and wildlife occupy the relatively tiny fraction of Antarctica that is coastal and free of ice in the summer, leading to concern over concentration of environmental impact (Stewart et al. 2005). Furthermore, tourism occurs predominantly during the austral summer, coincident with the reproductive period of Antarctic flora and fauna. Trampling and damage of vegetation have been investigated by Tejedo et al. (2009, 2012) as one of the potential human impacts at tourist landing areas. Research has also highlighted the AP's vulnerability to human-mediated introduction of both native and alien species (Smith 1996).

Here, we look at correlations between lichen PU richness and the number of tourist visiting 20 sites (of 29) where sufficient information was available. The information on tourist visitation comes from the International Association of Antarctica Tour Operators (IAATO). Only the activities of IAATO members are included in this analysis, which account for approximately 95% of all of the commercial cruise ships operating and 90% of all the known visitors to the Peninsula. The number of tourists visiting each site was calculated as the mean of the total number of visitors at each site every season, from the 2003–2004 to the 2007–2008 seasons (IAATO 2006, 2007, 2008).

We found a significant correlation between the observed number of PUs present (using genera PUs or species PUs) and the number of visitors to a given site (Table 4). These are expected results because without correcting for sampling effort, the most popular sites are also the sites that have more sampling events in this database. However, we found no correlation between estimated number of PUs (using genera or species PUs) and the number of visitors to the sites studied. Table 5 shows a list of these sites, from most to least visited, and the estimated richness of lichens. This information could be used directly to delineate visitor guidelines for protection of lichen flora.

Table 4. Correlation coefficient between PU richness and number of tourists.

PU richness

Adjusted R²

p value

Observed number of PU "genera"

0.18

0.03

Chao2 estimated number of PU "genera"

-0.03

0.57

Observed number of PU "species"

0.21

0.02

Chao2 estimated number of PU "species"

0.11

0.08

 

Table 5. Lichen PU richness and tourism visitation.

Sites

Observed genus PUs

Estimated genus PUs

Observed species PUs

Estimated species PUs

Mean number of
total visitors per year,
from 2003 to 2008

Cuverville Island

15

19.5

33

117.5

15521.6

Whalers Bay

27

41.0

45

73.8

14612.6

Half Moon Island

17

18.5

28

48.2

13976.8

Almirante Brown

8

17.0

8

8.0

10287.2

Petermann Island

9

27.0

9

33.5

9627.0

Jougla Point

10

10.0

18

24.7

9329.0

Brown Bluff

10

18.0

17

29.1

6004.2

Aitcho Islands (Barrientos Island)

14

17.6

21

28.5

5878.2

Waterboat Point

14

18.9

29

62.3

5453.0

Pléneau Island

9

9.0

13

13.0

5355.8

Hannah Point

14

18.9

20

44.5

3650.0

Danco Island

9

9.00

12

12.0

2925.2

Baily Head

8

10.6

9

12.1

2678.4

Damoy Point

7

25.0

6

18.5

1479.0

Mikkelsen Harbor

14

26.5

24

74.0

1408.8

Torgersen Island

3

5.0

3

5.0

785.2

Booth Island

13

53.5

18

67.0

612.2

Georges Point

19

103.5

32

59.5

460.2

Spigot Peak

2

2.0

5

5.0

457.6

 

DATABASE CREATION

The data were digitized and checked for consistency.

 

Class III. Data set status and accessibility

A. Status

Latest update: 2012.

Metadata status: Metadata are complete.

Data verification: The data were checked for consistency. Parataxonomic unit names were thoroughly checked and corrected according to Øvstedal and Smith (2001, 2009).

B. Accessibility

Storage location and medium: The Ecological Society of America's Ecological Archives

C. Contact persons:

Paula Casanovas
Department of Biology
College of Computer, Mathematics and Natural Sciences
University of Maryland
College Park, MD 20742-4415 USA
E-mail: paulac@umd.edu

Heather J. Lynch
Department of Ecology and Evolution
Stony Brook University
Stony Brook, NY 11794 USA
E-mail: heather.lynch@stonybrook.edu

D. Copyright restrictions: Any paper using the data should cite this paper.

 

Class IV. Data structural descriptors

Description of sampling sites

A. Data Set File

Identity: Sites.csv

Size: 3,823 bytes

Format and storage mode: ASCII text, comma separated.

B. Header information

Site_ID: Unique ID for each site

Site_name: Site name according to the U.S. Board on Geographic Names

Latitude: Latitude of the site

Longitude: Longitude of the site

notes: A description of the site, or any other notes of relevance for the site.

 

Description of visits

A. Data Set File

Identity: Visits.csv

Size: 4,324 bytes

Format and storage mode: ASCII text, comma separated

B. Header information

Visit_ID: Unique number of a site of data collection

Site_ID: ID for each site. For detailed site description see file Sites.txt

Visit_day: Number between 01-31

Visit_month: Number of the month, between 01-12

Visit_year: Year number, using four digits

notes: Any notes relevant to the visit

 

List of the parataxonomic units on the database

A. Data Set File

Identity: Parataxonomy.csv

Size: 10,519 bytes

Format and storage mode: ASCII text, comma separated

B. Header information

Parataxon_name: Unique name for the parataxonomic unit

Parataxon_level: Indicates if the parataxonomic unit refers to a species or to a genus

Growth_form: Growth form of the lichen parataxonomic unit: crustose, foliose, fruticose

 

Description of photographs collected in each visit

A. Data Set File

Identity: Photographs.csv

Size: 93,428 bytes

Format and storage mode: ASCII text, comma separated

B. Header information

file_name: Unique name for the photograph

Visit_ID: ID for each visit. For detailed visit description see file Visit.txt

Photographer: ID for the photographer

Copyright_holder: The photographer or the Antarctic Site Inventory

Photo_quality: A qualitative measurement of the picture quality. Poor: the lichens in the image are not possible to distinguish; medium: the lichens in the image can be distinguished, but the resolution is very low, and no small structures are defined; good: the lichens in the image are well defined, but small structures cannot be distinguished; excellent: the lichens in the image are well defined, and small structures on the lichens can be observed.

Notes: Notes about the photograph

 

Description of specimens isolated from the photographs

A. Data Set File

Identity: Specimen.csv

Size: 145,034 bytes

Format and storage mode: ASCII text, comma separated

B. Header information

catalog_file_name: Unique name for the specimen

file_name: ID for each photograph from where the specimen was isolated. For detailed photograph description see file Photograph.txt

Notes: Notes about the specimen

 

Identification of specimens by different classifiers

A. Data Set File

Identity: Determinations.csv

Size: 509,961 bytes

Format and storage mode: ASCII text, comma separated

B. Header information

identification_ID: Unique number for the identification

catalog_file_name: ID for each specimen. For detailed specimen description see file Specimens.txt

Parataxon_name: Parataxonomic unit name. For detailed parataxonomic unit name description see file Parataxonomy.txt

Identification_quality: A qualitative measurement of the identification quality (poor, medium, good, excellent)

Determination_day: Number between 01-31

Determination_month: Number of the month, between 01-12

Determination_year: Year number, using four digits

Notes: Notes about the photograph

 

Description of PU richness by sites

A. Data Set File

Identity: Richness.csv

Size: 1,357 bytes

Format and storage mode: ASCII text, comma separated

B. Header information

Estimation_ID: Unique number for the estimation

Site_ID: ID for the site. For detailed site description see file Sites.txt

Observed_Genus_PU: Number of PUs observed

Estimated_Genus_PU: Estimated number of PUs using Chao2

Observed_Species_PU: Number of PUs observed

Estimated_Species_PU: estimated number of PUs using Chao2

Notes: Notes about the estimations.

 

 

Fig1

Fig. 1. Mechanistic statistical method for the estimation of PU richness. (a) and (b) PU specific probabilities of detection and occupancy for Whalers Bay and Jougla Point. (c) and (d) Posterior distribution of lichen PU richness for Whalers Bay and Jougla Point.


 

ACKNOWLEDGEMENTS

We thank all the field photographers and the undergrad researchers involve in the identification and preparation of the specimens, especially Beth Stevenson (for her work at the Smithsonian Institution) and Julien Buchbinder (for his work on identification of digital specimens and assistant on the development of the database). We also would like to thank Rusty Russell and Zuvayda Abdurahimova for their assistance at the U.S. National Herbarium.  H.J.L., R.N. and W.F.F gratefully acknowledge assistance from the US National Science Foundation Office of Polar Programs (Award No NSF/OPP – 0739515). This work was partially supported by NASA headquarters under the NASA Earth and Space Fellowship Program – grant NNX10AN55H to P.C.

 

Literature cited

Abramoff, M. D., P. J. Magalhaes, and S. J. Ram. 2004. Image Processing with ImageJ. Biophotonics International 11:36–42.

Basset, Y., V. Novotny, S. E. Miller and R. Pyle. 2000. Quantifying Biodiversity: Experience with Parataxonomists and Digital Photography in Papua New Guinea and Guyana. BioScience 50:899.

Boulinier, T., J. D.Nichols, J. R. Sauer, J. E. Hines, and K. H. Pollock. 1998. Estimating species richness: the importance of heterogeneity in species detectability. Ecology 79:1018–1028.

Budd, A. F., K. G. Johnson, and D. C. Potts. 1994. Recognizing Morphospecies in Colonial Reef Corlas: I. Landmark-Based Methods. Paleobiology 484–505.

Butcher, G. S., and D. K. Niven. 2007. Combining data from the Christmas Bird Count and the Breeding Bird Survey to determine the continental status and trends of North America birds. National Audubon Society.

Casanovas, P., H. J. Lynch, and W. F. Fagan. 2013. Multi-scale patterns of moss and lichen richness on the Antarctic Peninsula. Ecography 36:209–219.

Chao, A., R. L. Chazdon, R. K. Colwell, and T. J. Shen. 2005. A new statistical approach for assessing similarity of species composition with incidence and abundance data. Ecology Letters 8:148–159.

Chown, S. L., and P. Convey. 2007. Spatial and temporal variability across life’s hierarchies in the terrestrial Antarctic. Philosophical Transactions of the Royal Society B: Biological Sciences 362:2307–2331.

Clarke, A., P. S. Lake, and D. J. O’Dowd. 2004. Ecological impacts on aquatic macroinvertebrates following upland stream invasion by a ponded pasture grass (Glyceria maxima) in southern Australia. Mar. Freshwater Res. 55:709–713.

Cohn, J. P. 2008. Citizen Science: Can volunteers do real research? BioScience 58:192–197.

Conrad, C. C., and K. G. Hilchey. 2011. A review of citizen science and community-based environmental monitoring: issues and opportunities. Environmental monitoring and assessment 176:273–291.

Convey, P., J. A. E. Gibson, C. D. Hillenbrand, D. A. Hodgson, P. J. A. Pugh, J. L. Smellie, and M. I. Stevens. 2008. Antarctic terrestrial life-challenging the history of the frozen continent? Biological Reviews 83:103–117.

Convey, P. 2010. Terrestrial biodiversity in Antarctica - Recent advances and future challenges. Polar Science 4:135–147.

Convey, P. 2011. Antarctic terrestrial biodiversity in a changing world. Polar Biology 34:1629–1641.

Dickinson, J. L., B. Zuckerberg, and D. N. Bonter. 2012. Citizen Science as an Ecological Research Tool: Challenges and Benefits. Annual review of ecology, evolution, and systematics 41.

Dorazio, R. M., J. A. Royle, B. Söderström, and A. Glimskär. 2006. Estimating species richness and accumulation by modeling species occurrence and detectability. Ecology 87:842–854.

Favero-Longo, S. E., N. Cannone, M. Roger Worland, P. Convey, R. Piervittori, and M. Guglielmin. 2011. Changes in Lichen Diversity and Community Structure with Fur Seal Population Increase on Signy Island, South Orkney Islands. Antarctic Science 23:65–77.

Fenton, J. H. C. 1982. Vegetation re-exposed after burial by ice and its relationship to changing climate in the South Orkney Islands. British Antarctic Surrey Bulletin 51:247–255.

Ferreira, T. A., and W. Rasband. 2010. The ImageJ User Guide - Version 1.43.

Garrettson, M., J. F. Stetzel, B. S. Halpern, D. J. Hearn, B. T. Lucey, and M. J. McKone. 1998. Diversity and Abundance of Understorey Plants on Active and Abandoned Nests of Leaf-Cutting Ants (Atta cephalotes) in a Costa Rican Rain Forest. Journal of Tropical Ecology 14:17–26.

IAATO. 2006. Brief update on the Antarctic Peninsula landing site visits and site guidelines. Information Paper 66 for XXIX Antarctic Treaty Consultative Meeting.

IAATO. 2007. Brief update on the Antarctic Peninsula landing site visits and site guidelines. Information Paper 114 for XXX Antarctic Treaty Consultative Meeting.

IAATO. 2008. Update on the Antarctic Peninsula landing site use and site guidelines. Information Paper 82 for XXXI Antarctic Treaty Consultative Meeting.

Krell, F. T. 2004. Parataxonomy vs. taxonomy in biodiversity studies – pitfalls and applicability of "morphospecies" sorting. Biodiversity and Conservation 13:795–812.

Leishman, M. R., and C. Wild. 2001. Vegetation abundance and diversity in relation to soil nutrients and soil water content in Vestfold Hills, East Antarctica. Antarctic Science 13:126–134.

Lunn, D. J., A. Thomas, N. Best, and D. Spiegelhalter. 2000. WinBUGS-a Bayesian modelling framework: concepts, structure, and extensibility. Statistics and computing 10:325–337.

Lynch, H. J., R. Naveen, P. N. Trathan, and W. F. Fagan. 2012. Spatially integrated assessment reveals widespread changes in penguin populations on the Antarctic Peninsula. Ecology 93:1367–1377.

MacKenzie, D. I., J. D. Nichols, G. B. Lachman, S. Droege, A. J. Royle, and C. A. Langtimm. 2002. Estimating site occupancy rates when detection probabilities are less than one. Ecology 83:2248–2255.

Myrcha, A., S. J. Pietr, and A. Tatur. 1985. The role of pygoscelid penguin rookeries in nutrient cycles at Admiralty Bay, King George Island. Pages 156–162 in W. R. Siegfried, P. R. Condy, and M. R. Laws, editors. Antarctic nutrient cycles and food webs. Springer, New York, New York, USA.

Naveen, R., and H. J. Lynch. 2011. Antarctic Peninsula Compendium, Third edition.

Ochyra, R., H. Bednarek-Ochyra, and R. I. Smith. 2008. Illustrated Moss Flora of Antarctica. Cambridge University Press.

Oliver, I., and A. J. Beattie. 1996a. Invertebrate morphospecies as surrogates for species: a case study. Conservation Biology 10:99–109.

Oliver, I., and A. J. Beattie. 1996b. Designing a cost-effective invertebrate survey: a test of methods for rapid assessment of biodiversity. Ecological Applications 6 (2):594–607.

Oliver, I., and A. J. Beattie. 2002. A Possible Method for the Rapid Assessment of Biodiversity. Conservation Biology 7:562–568.

Øvstedal, D. O., and R. I. L. Smith. 2001. Lichens of Antarctica and South Georgia: a guide to their identification and ecology. Cambridge University Press, Cambridge, UK.

Øvstedal, D. O., and R. I. L. Smith. 2004. Additions and corrections to the lichens of Antarctica and South Georgia. Cryptogamie, Mycologie 25 (4):323–331.

Øvstedal, D. O., and R. I. L. Smith. 2009. Further additions to the lichen flora of Antarctica and South Georgia. Nova Hedwigia 88:157–168.

Peat, H. J., A. Clarke, and P. Convey. 2007. Diversity and biogeography of the Antarctic flora. Journal of Biogeography 34:132–146.

R Development Core Team. 2010. R: A Language and Environment for Statistical Computing. R Foundation for Statistical Computing, Vienna, Austria.

Ryan, P. G., and B. P. Watkins. 1989. The influence of physical factors and ornithogenic products on plant and arthropod abundance at an Inland Nunatak group in Antarctica. Polar Biology 10:151–160.

Silvertown, J. 2009. A new dawn for citizen science. Trends in Ecology and Evolution 24:467–471.

Smith, R. I. L. 1984. Terrestrial plant biology of the sub-Antarctic and Antarctic. Pages 61–162 in R. M. Laws, editor. Antarctic Ecology, Volume 1. Academic Press, London, UK.

Smith, R. I. L. 1985. Nutrient cycling in relation to biological productivity in Antarctic and sub-Antarctic terrestrial ecosystems. Pages 156–162 in W. R. Siegfried, P. R. Condy, and M. R. Laws, editors. Antarctic nutrient cycles and food webs. Springer, New York, New York, USA.

Smith, R. I. L. 1996. Introduced plants in Antarctica: potential impacts and conservation issues. Biological conservation 76:135–146.

Smith, V. R. 1978. Animal-plant-soil nutrient relationships on Marion Island (Subantarctic). Oecologia 32:239–253.

Smith, V. R. 2008. Energy flow and nutrient cycling in the Marion Island terrestrial ecosystem: 30 years on. Polar Record 44:211–226.

Smykla, J., J. Wołek, and A. Barcikowski. 2007. Zonation of vegetation related to penguin rookeries on King George Island, maritime Antarctic. Arctic, Antarctic, and Alpine Research 39:143–151.

Stewart, E. J., D. Draper, and M. E. Johnston. 2005. A review of tourism research in the polar regions. Arctic 58:383–394.

Sturtz, S., U. Ligger, and A. Gelman. 2005. R2WinBUGS: A Package for Running WinBUGS from R. Journal of Statistical Software 12:1–16.

Tatur, A. 2002. Ornithogenic ecosystems in the maritime Antarctic - Formation, development and disintegration. Page 161 in L. Beyer and M. Bölter, editors. Geoecology of Antarctic Ice-free Coastal Landscapes. Springer Verlag, New York, New York, USA.

Tatur, A., A. Myrcha, and J. Niegodzisz. 1997. Formation of abandoned penguin rookery ecosystems in the maritime Antarctic. Polar Biology 17:405–417.

Tejedo, P., A. Justel, J. Benayas, E. Rico, P. Convey, and A. Quesada. 2009. Soil trampling in an Antarctic Specially Protected Area: tools to assess levels of human impact. Antarctic Science 21:229–236.

Tejedo, P., L. R. Pertierra, J. Benayas, P. Convey, A. Justel, and A. Quesada. 2012. Trampling on maritime Antarctica: can soil ecosystems be effectively protected through existing codes of conduct? Polar Research 31.

Terauds, A., S. L. Chown, F. J. Morgan, H. Peat, D. J. Watts, H. Keys, P. Convey, and D. M. Bergstrom. 2012. Conservation biogeography of the Antarctic. Diversity and Distributions 18:726–741.

Appendix 1: Description of the detection model used for the estimation of PU richness including model and parameters description, R code for its implementation, and corresponding data.

This modeling approach draws heavily on the model developed in Dorazio and Royle (2005); here we summarize that model in the context of our own application:

The probability that a PU is detected conditional that it is present is given by:

Equation

xij = the number of photographers that found PU i in plot j, as identified by K classifiers

θij = probability of detection of PU i given that occurs at plot j

Ψij = probability of occurrence of PU i at plot j

K = number of classifiers

I() = indicator function, equals one when its argument is true and is zero otherwise

 

The effects of the plot-specific and PU-specific differences in rates of occurrence and detection are modeled as:

 

logit(θij) = ui + αj

logit(Ψij)=  vi + βj

 

ui and vi denote species-level effects

αj and βj denote site-level effects, and it is assumed for this model that they have constant values.

 

R code for the model and implementation using Prince William Park as an example:

###Code modified from 
###http://www.mbr-pwrc.usgs.gov/site/communitymodeling/software-code/

library("reshape")
library("R2WinBUGS")

###################
#Write the model code to a text file (used to run WinBUGS)
cat("
  model{

#Define prior distributions for community-level model parameters
omega ~ dunif(0,1)

u.mean ~ dunif(0,1)  
mu.u <- log(u.mean) - log(1-u.mean)

v.mean ~ dunif(0,1)
mu.v <- log(v.mean) - log(1-v.mean)

tau.u ~ dgamma(0.1,0.1)  
tau.v ~ dgamma(0.1,0.1)

for (i in 1:(n+nzeroes)) {

#Create priors for PUs i from the community level prior distributions
    w[i] ~ dbern(omega)
    u[i] ~ dnorm(mu.u, tau.u)
    v[i] ~ dnorm(mu.v, tau.v)    

#Create a loop to estimate the Z matrix (true occurrence for PUs i at point j.      
   for (j in 1:J) {
       logit(psi[j,i]) <- u[i] 
       
  mu.psi[j,i] <- psi[j,i]*w[i]
  Z[j,i] ~ dbern(mu.psi[j,i])

#Create a loop to estimate detection for PUs i at point k during #sampling period k.      
     for (k in 1:K[j]) {  	
    	logit(p[j,k,i]) <-  v[i]	
       mu.p[j,k,i] <- p[j,k,i]*Z[j,i]
       X[j,k,i] ~ dbern(mu.p[j,k,i])
}  } }

#Sum all PUs observed (n) and unobserved PUs (n0) to find the total estimated richness
n0 <- sum(w[(n+1):(n+nzeroes)])	
N <- n + n0

#Finish writing the text file into a document called basicmodel.txt
}",file="basicmodel.txt")

###################
##PU lichen data
#Whalers bay data
WHAL <- read.csv("WHAL.csv")

###################
#Data preparation
WHALM <- melt(WHAL)

#Reshape the tables and select the information needed
#Rep = the classifiers that have identifiers each of the digital specimens
WHALT <- data.frame(cast(WHALM, visit + Photographer + species_PU + value ~ a))
WHALT <-WHALT[,-5]
names(WHALT)<- c("Point","Photographer","Species","Rep")
WHALdata<-na.omit(WHALT)
WHALdata$Occ <- rep(1, dim(WHALdata)[1])

#How many citings for each species
WHALtotal.count = tapply(WHALdata$Occ, WHALdata$Species, sum)

#Find the number of unique species
WHALuspecies = as.character(unique(WHALdata$Species))

#n is the number of observed species
WHALn=length(WHALuspecies)

#Find the number of unique sampling locations
WHALupoints = as.character(unique(WHALdata$Point))

#J is the number of sampled points
WHALJ=length(WHALupoints)

#The detection/non-detection data is reshaped into a three dimensional 
#array X where the first dimension, j, is the plots; the second 
#dimension, k, is the rep (classifier); and the last dimension, i, is the PU.
# the array element xijk is the number of photographers that found PU i in plot j, as 
#identified by classifier k

WHALjunk.melt=melt(WHALdata,id.var=c("Species", "Point", "Rep"), measure.var="Occ")
WHALX=cast(WHALjunk.melt, Point ~ Rep ~ Species)

#Create all zero encounter histories to add to the detection array X 
#as part of the data augmentation to account for additional 
#PUs (beyond the n observed PUs).

#nzeroes is the number of all zero encounter histories to be added
  WHALnzeroes = 50
#X.zero is a matrix of zeroes
  WHALX.zero = matrix(0, nrow=7, ncol=3)

#Xaug is the augmented version of X.  The first n PUs were actually observed
#and the n+1 through nzeroes PUs are all zero encounter histories  
  WHALXaug <- array(0, dim=c(dim(WHALX)[1],dim(WHALX)[2],dim(WHALX
  [3]+WHALnzeroes))
  WHALXaug[,,(dim(WHALX)[3]+1):dim(WHALXaug)[3]] = rep(WHALX.zero, WHALnzeroes)
  dimnames(WHALX)=NULL
  WHALXaug[,,1:dim(WHALX)[3]] <-  WHALX

#K is a vector of length J indicating the number of reps at each point j  
  WHALKK <- WHALX.zero
  WHALa=which(WHALKK==0); WHALKK[WHALa] <- 1
  WHALK=apply(WHALKK,1,sum, na.rm=TRUE)
  WHALK=as.vector(WHALK)


#################
#Create the necessary arguments to run the bugs() command 
#Load all the data
WHALsp.data = list(n=WHALn, nzeroes=WHALnzeroes, J=WHALJ, K=WHALK, X=WHALXaug)

#Specify the parameters to be monitored
WHALsp.params = list('u', 'v', 'mu.u', 'mu.v', 'tau.u', 'tau.v', 'omega', 'N')

#Specify the initial values
WHALsp.inits = function() {
  omegaGuess = runif(1, WHALn/(WHALn+WHALnzeroes), 1)
  psi.meanGuess = runif(1, .25,1)
  list(omega=omegaGuess,w=c(rep(1, n), rbinom(WHALnzeroes, size=1, prob=omegaGuess)),
       u=rnorm(WHALn+WHALnzeroes), v=rnorm(WHALn+WHALnzeroes),
       Z = matrix(rbinom((WHALn+WHALnzeroes)*J, size=1, prob=psi.meanGuess), 
                  nrow=WHALJ, ncol=(WHALn+WHALnzeroes))
  )
}
n<-WHALn
J<-WHALJ

#Run the model and call the results ?fit?
WHALfit = bugs(WHALsp.data, WHALsp.inits, WHALsp.params, "basicmodel.txt", debug=TRUE, n.chains=3, n.iter=1000, n.burnin=500, n.thin=5)

#See baseline estimates of PUs-specific occupancy and detection in one of 
#the habitat types (PRWI)
WHALspecies.occ = WHALfit$sims.list$u
WHALspecies.det = WHALfit$sims.list$v

#Show occupancy and detection estimates for only the observed PUs (1:n)
WHALpsi = plogis(WHALspecies.occ[,1:n]) 
WHALp   = plogis(WHALspecies.det[,1:n]) 

WHALocc.matrix <- cbind(apply(WHALpsi,2,mean),apply(WHALpsi,2,sd))
colnames(WHALocc.matrix) = c("mean occupancy", "sd occupancy")
rownames(WHALocc.matrix) = WHALuspecies

WHALdet.matrix <- cbind(apply(WHALp,2,mean),apply(WHALp,2,sd))
colnames(WHALdet.matrix) = c("mean detection", "sd detection")
rownames(WHALdet.matrix) = WHALuspecies

WHALresults <- data.frame(WHALuspecies,round(WHALocc.matrix, digits=2),round(WHALdet.matrix, digits=2))

WHALresults$WHALuspecies = with(WHALresults, factor(WHALuspecies, levels = WHALresults$mean.occupancy))

WHALresults <- WHALresults[order(WHALresults$mean.detection) , ]

plot(WHALresults$mean.occupancy, ylim=c(0,1),pch=19, axes = FALSE, xlab = "", ylab = "Probability")
points(WHALresults$mean.detection, ylim=c(0,1),col="grey",pch=19)
axis(1, at=c(1:length(WHALuspecies)), lab=F)
text(c(1:length(WHALuspecies)),-0.1, labels=WHALuspecies, xpd=T, srt=40, adj=1,cex=0.5)
axis(side = 2, cex.axis = 1)
box()

#See estimates of total richness (N)
WHALN ? WHALfit$sims.list$N
mean(WHALN) 
summary(WHALN) 
table(WHALN)
plot(table(WHALN))

[Back to E094-194]