exploring carnivorous plant habitats based on images from social … · 2020. 5. 28. · exploring...
TRANSCRIPT
HAL Id: hal-02196764https://hal.inria.fr/hal-02196764
Submitted on 29 Jul 2019
HAL is a multi-disciplinary open accessarchive for the deposit and dissemination of sci-entific research documents, whether they are pub-lished or not. The documents may come fromteaching and research institutions in France orabroad, or from public or private research centers.
L’archive ouverte pluridisciplinaire HAL, estdestinée au dépôt et à la diffusion de documentsscientifiques de niveau recherche, publiés ou non,émanant des établissements d’enseignement et derecherche français ou étrangers, des laboratoirespublics ou privés.
Exploring carnivorous plant habitats based on imagesfrom social media
Rafael Blanco, Zujany Salazar, Tobias Isenberg
To cite this version:Rafael Blanco, Zujany Salazar, Tobias Isenberg. Exploring carnivorous plant habitats based on imagesfrom social media. IEEE VIS 2019: IEEE Conference on Visualization, Oct 2019, Vancouver, Canada.�hal-02196764�
Exploring Carnivorous Plant Habitats based on Images from Social MediaRafael Blanco*
Telecom SudParisZujany Salazar∗
Telecom SudParisTobias Isenberg∗
Inria
ABSTRACT
We visualize species habitat distribution information based on geo-lo-cated images posted on social media. For the example of carnivorousplants, we use published image data to produce interactive mapsof the spatial distribution of different species/genuses, histogramsof the elevations at which they grow, and plots of the temporaldistribution of the photographs. We further discuss the mismatchbetween our distribution maps and traditionally established maps aswell as further possibilities for research with our data.
1 MOTIVATION AND RELATED WORK
Many species of plants and animals are endangered today (e. g.,see the IUCN Red List 2012; https://www.iucnredlist.org/).This status also applies to many carnivorous plants, and past stud-ies have quantified the conservation threats of the different genusand species [4], providing information to prioritize certain areas ofconservation. Yet information about the distribution of the differentspecies is often difficult to obtain. Researchers use environment dataand current habitat sightings, to apply species distribution modeling(SDM), which uses statistical models to generate an estimate of theplant distribution around the world. In contrast to dedicated cam-paigns to document species distribution in the wild, we explore theuse of geo-located images posted in social media. While the use ofgeo-located social media images and other posts has been exploredin the past for the study of popular places [3] and events [1, 5], weuse the social images posted on Panoramio and Flickr not as singu-lar events but as evidence for the existence of a specific plant at alocation, providing evidence for the habitat of the depicted species.
2 DATA ACQUISITION
We collected a dataset by searching Panoramio1 and Flickr for geo-tagged images whose label or description included at least onefrom a series of search terms. These search terms included theLatin genus and family names, the terms “carnivorous plant(s) orsimilar, as well as a number of common names for different speciesin several languages including Chinese, Danish, Dutch, English,French, German, Italian, Japanese, Norwegian, Polish, Portuguese,Russian, Spanish, and Swedish. This resulted in more than 28,700candidate images. For each of the found images, we looked at boththe image itself and its location on the map using an online satellitemap, to manually determine whether the images showed a truehabitat and whether the location data was believable (e. g., Fig. 24).For example, we removed image locations in the middle of urbanareas. As some people posted many images of plants kept at home orused the same location that was not plausibly a habitat location, weremoved certain user IDs and locations from consideration to speed-up the inspection. We then recorded resulting locations, resultingin a list of more than 8,800 entries. Out of these, approx. 4,600are within a radius of approx. 250 m another location with the samespecies and can thus be considered to be duplicates. For each plantlocation, we recorded its scientific name,2 its geographic positionon the map, its elevation (based on an elevation look-up for the
*e-mail: {rblancog25 | zujany}@gmail.com, [email protected] Panoramio service has since been retired by Google.2We generally used the Latin names provided in the descriptions. If no
species name was provided or if it was wrong, we either classified the plantourselves if we knew the species well, or we only recorded the plant’s genus.
Figure 1: Screenshot of our data exploration interface, with markerscolored by genus and image thumbnails shown at the bottom.
geographic position), the time the picture was taken, the socialmedia service and the respective image ID, text description of thegeographic location such as area name and country, the name andID of the person who uploaded the images, and the image itself.
3 VISUALIZATION SYSTEM
We designed a map-based (using Google’s Maps API) data explo-ration interface that indicates the location of each plant in the datasetwith a marker. We provide two visualization modes: a general datapoint exploration mode and a plant species or genus distributionmode. The exploration mode (e. g., Fig. 1) has a filter panel to filterfor plant characteristics (genus, species) and/or image data (location,date it was taken, social network). In this mode we provide an imagecarousel that shows thumbnails of the images of the currently dis-played data points. In the distribution mode (Fig. 12), we use circlesaround plant locations to generate distribution maps (e. g., Fig. 3),without applying any statistical methods. We use circles whose sizecan be changed based on the current map magnification. In addition,as such maps could potentially be published, we add a small randomoffset to obfuscate the specific locations to prevent potential poach-ing. Furthermore, we allow users to generate interactive plots of theimage dates and elevation histograms using Plotly (e. g., Fig. 2).
Our application relies on Flask for Python, HTML5, CSS3, andJavaScript. We use SQLAlchemy to manage the database that hasthe information of each image of a plant. Also, we use the FlickrAPI to allow the users to look up each entry directly on Flickr.
4 DATA EXPLORATION STRATEGIES AND CASE STUDIES
Our visualization system allows us to investigate the collected datain interesting ways. Primarily it is straight-forward to examine thegeographic distribution of different species and to compare this withknown distribution maps. For example, we can find that the distribu-tion of Drosera rotundifolia in our data matches with the distributionmap on Wikipedia (Fig. 18). However, we also see that our datadoes not support the full distribution on the established maps. Thisfact is likely due to a strong bias in our data: we only have picturesfrom places that people are likely to live at or travel to (Fig. 21,21(c)). We thus have no evidence for the distribution of Droserarotundifolia in Siberia, for example. In another example, we can alsosee that our data supports the distribution of Sarracenia purpureain North America (Fig. 19)—including the isolated population inthe southern United States but excluding the less populated areas
Jan. Feb Mar Apr May Jun Jul Aug Sep Oct Nov Dec0
50
100
150
200
Panoramio
Flickr
(a) 0 20 40 60 80 100 120
-100—00—100
100—200200—300300—400400—500500—600600—700700—800800—900900—10001000—11001100—12001200—13001300—14001400—15001500—16001600—17001700—18001800—19001900—20002000—21002100—22002200—23002300—24002400—25002500—26002600—27002700—28002800—29002900—30003000—31003100—32003200—33003300—34003400—35003500—36003600—37003700—38003800—39003900—40004000—4100
(b)
Figure 2: Examples of a generated data plots: (a) sightings of Sar-racenia purpurea by month and (b) elevation histogram for Nepenthes.
of central and northern Canada. Yet we can also see populations ofthe species in Europe—not only well known places where the plantwas introduced in Switzerland but also places in the UK, Scandi-navia, and the Netherlands. Similarly, our map of the distributionof Drosera arcturi (Fig. 3) shows habitats not only in Australia butalso in New Zealand and for Darlingtonia californica (Fig. 20) wesee also additional locations in western and central California.
In addition, we need to address the inherent unreliability of theposted data—in particular the geographic location could be intention-ally or unintentionally incorrect. We thus implemented verificationmechanisms that provide information about the reliability of thelocations. Specifically, we treat a location as more reliable if othersocial media users have posted pictures of other plants nearby. Basedon a user-selectable search threshold (we experimented, e. g., with1 km and 4 km; e. g., see Fig. 15 and 16), we search for nearby sitesposted by other users and depict sites with a species match in green,sites with a genus match in yellow, sites with only other-genus plantsnearby in orange, and the rest in red (other color maps for color-deficient users are possible). The process to compute these trust-worthiness values essentially uses a hashed comparison of the sitesusing 9-neighborhood “buckets” and thus takes about 10 seconds tocompute. We allow users to adjust the search radius and thus thetrust level they want to work with, which likely depends on a givensite and its number of data points. We also explored a verificationbased on the posts of a single person: We check if the same personposted multiple pictures from exactly the same GPS position—asign for manipulated coordinates—and color the markers based onhow many other pictures were posted a given location (Fig. 17).
In addition to the geographic distributions, we also analyzedabstract data. In particular, we support the analysis of the timethe picture was taken and of the elevation of the location (e. g.,Fig. 4 shows the number of entries in our dataset by year, and Fig. 5shows monthly detail since 2003). We also analyzed the dates ofthe pictures within a year (e. g., Fig. 6 for all entries), and saw thatsightings in the Northern Hemisphere with a peak in the middle ofthe calendar year are prevalent. Looking at specific plants, we canidentify the different growth periods between northern and SouthernHemisphere plants (e. g., Fig. 10). The elevation plots (e. g., Fig. 7,9) also show a quite different behavior of different species.
We note that our data relies on a correct classification of thespecies. While due to our manual process we are certain that we usethe correct genus name for all entries, the same cannot be said for thespecies. Here we relied on the classification provided in the socialmedia post (if any), and only re-classified some specific specieswhich we knew well. A sizable portion of our entries (approx. 12%)thus only contains the genus of the respective plants locations. Anexpert botanist with a research background on these plants wouldcertainly be able to address these challenges. The abstract data plotsas well as the geographic maps are likely to support such an analysis,identifying outliers or non-classified plants in specific regions.
Figure 3: Example of a generated habitat map for Drosera arcturi.
5 CONCLUSION AND FUTURE WORK
Our work showcases an interesting and, to the best of our knowledge,previously not explored use of social media data for longer-term“events” such as plant grows in a specific habitat/region. We exploredboth geographic as well as abstract data analysis techniques for thisdata, and it would be interesting in the future to also explore space-time representations [2]. We are also planning to investigate machinelearning approaches to be able to identify flowers on the pictures,which would allow us to plot flowering periods of the plants. We alsoplan to investigate if ML approaches could support genus or speciesidentification to be able to further check the data or automate the datacollection process. The latter should also generally be integrated inthe visualization tool, so that we can merge the currently separateprocesses of data collection, correction, and visualization.
We collected our dataset and built our visual exploration tool dueto an interest in carnivorous plants, it can thus be used by enthusiaststo visit habitats. Our tool also has the potential to be used by plantbiologists or botanists to study the habitat information we collectedas well as by conservationists to become aware of unknown sites aswell as the evolution of known sites. For the latter of the two appli-cations, however, we point out that our dataset is limited becausefor some species we only have very few entries (e. g., currently 110species with 5 entries or fewer each). So users need to be aware ofsmall-number statistics issues when analyzing the data. In addition,the previously mentioned data biases need to be considered. Basedon the protection status of many carnivorous plants, however, wewill not make the tool or data publicly available to prevent poachingand will only share it with researchers in the application domain.
REFERENCES
[1] G. Andrienko, N. Andrienko, P. Bak, S. Kisilevich, and D. Keim. Analy-sis of community-contributed space- and time-referenced data (exampleof Flickr and Panoramio photos). In Proc. VAST (Posters), pp. 213–214.IEEE Computer Society, Los Alamitos, 2009. doi: 10.1109/VAST.2009.5333472
[2] B. Bach, P. Dragicevic, D. Archambault, C. Hurter, and S. Carpendale.A descriptive framework for temporal data visualizations based on gen-eralized space-time cubes. Computer Graphics Forum, 36(6):36–61,2017. doi: 10.1111/cgf.12804
[3] D. J. Crandall, L. Backstrom, D. Huttenlocher, and J. Kleinberg. Map-ping the world’s photos. In Proc. WWW, pp. 761–770. ACM, New York,2009. doi: 10.1145/1526709.1526812
[4] D. E. Jennings and J. R. Rohr. A review of the conservation threats tocarnivorous plants. Biological Conservation, 144(5):1356–1363, 2011.doi: 10.1016/j.biocon.2011.03.013
[5] S. Kisilevich, M. Krstajic, D. Keim, N. Andrienko, and G. Andrienko.Event-based analysis of people’s activities and behavior using Flickrand Panoramio geotagged photo collections. In Proc. Information Visu-alisation, pp. 289–296. IEEE Computer Society, Los Alamitos, 2010.doi: 10.1109/IV.2010.94
Exploring Carnivorous Plant Habitats based on Images from Social Media
Additional material
1975 1980 1985 1990 1995 2000 2005 2010 20150
200
400
600
800
1000Panoramio
Flickr
Figure 4: Date histogram by year and service, complete dataset. Interestingly, the number of posts per year declines after 2014, possibly to duethe fact that people post their pictures only after some time and not directly when taking them and/or due to Flickr’s changed upload policies whichapparently have let quite a number people to delete their images from the service (see Fig. 23 for a graph that supports this hypothesis).
01/200304/200307/200310/200301/200404/200407/200410/200401/200504/200507/200510/200501/200604/200607/200610/200601/200704/200707/200710/200701/200804/200807/200810/200801/200904/200907/200910/200901/201004/201007/201010/201001/201104/201107/201110/201101/201204/201207/201210/201201/201304/201307/201310/201301/201404/201407/201410/201401/201504/201507/201510/201501/201604/201607/201610/201601/201704/201707/201710/201701/201804/201807/201810/201801/201904/201907/2019
0
50
100
150
200
Panoramio
Flickr
Figure 5: Date histogram by month and service, complete dataset, ignoring pictures from before 2003.
Jan. Feb Mar Apr May Jun Jul Aug Sep Oct Nov Dec0
200
400
600
800
1000
1200
1400Panoramio
Flickr
Figure 6: Aggregated date histogram, by month, complete dataset.
0 500 1000 1500
-100—00—100
100—200200—300300—400400—500500—600600—700700—800800—900900—10001000—11001100—12001200—13001300—14001400—15001500—16001600—17001700—18001800—19001900—20002000—21002100—22002200—23002300—24002400—25002500—26002600—27002700—28002800—29002900—30003000—31003100—32003200—33003300—34003400—35003500—36003600—37003700—38003800—39003900—40004000—4100
Figure 7: Elevation histogram, complete dataset.
0^8
5^8
10^8
15^8
20^8
25^8
30^8
35^8
40^8
45^8
50^8
55^8
60^8
65^8
70^8
75^8
80^8
85^8
90^8
95^8
100^8
105^8
110^8
115^8
120^8
125^8
130^8
135^8
140^8
145^8
150^8
155^8
160^8
165^8
170^8
175^8
180^8
185^8
190^8
195^8
200^8
205^8
210^8
215^8
220^8
225^8
230^8
235^8
240^8
245^8
250^8
255^8
260^8
265^8
270^8
275^8
280^8
285^8
290^8
295^8
300^8
305^8
310^8
315^8
320^8
325^8
330^8
335^8
340^8
345^8
350^8
355^8
360^8
365^8
370^8
375^8
380^8
385^8
390^8
395^8
400^8
405^8
410^8
415^8
420^8
425^8
430^8
435^8
440^8
445^8
450^8
455^8
460^8
465^8
470^8
475^8
480^8
0
20
40
60
80
100
120
140
Figure 8: Histogram of Flickr ID bins for the identified plausible habitat images, complete dataset. Each bin contains 108 Flickr IDs. This plotshows some interesting patterns: We are not clear about the reason for the drop at around 120 · 108 and why fewer entries exist in the more recentbins since all bins are of equal size. These effects may be caused by the employed search strategies for the images using our set of keywords, orpotentially it is caused by when we conducted the searches for potential picture IDs on Flickr (which we then cached for later manual inspection).
0 50 100 150 200 250
-100—00—100
100—200200—300300—400400—500500—600600—700700—800800—900900—10001000—11001100—12001200—13001300—14001400—15001500—16001600—17001700—18001800—19001900—20002000—21002100—22002200—23002300—24002400—25002500—26002600—27002700—28002800—29002900—30003000—31003100—32003200—33003300—34003400—35003500—36003600—37003700—38003800—39003900—40004000—4100
(a)
0 5 10 15 20 25
-100—00—100
100—200200—300300—400400—500500—600600—700700—800800—900900—10001000—11001100—12001200—13001300—14001400—15001500—16001600—17001700—18001800—19001900—20002000—21002100—22002200—23002300—24002400—25002500—26002600—27002700—28002800—29002900—30003000—31003100—32003200—33003300—34003400—35003500—36003600—37003700—38003800—39003900—40004000—4100
(b)
0 1 2 3 4
-100—00—100
100—200200—300300—400400—500500—600600—700700—800800—900900—10001000—11001100—12001200—13001300—14001400—15001500—16001600—17001700—18001800—19001900—20002000—21002100—22002200—23002300—24002400—25002500—26002600—27002700—28002800—29002900—30003000—31003100—32003200—33003300—34003400—35003500—36003600—37003700—38003800—39003900—40004000—4100
(c)
Figure 9: Comparison of three elevation histograms with different characteristics for (a) Drosera rotundifolia, (b) Dionaea muscipula, and (c)Pinguicula aplina.
Jan. Feb Mar Apr May Jun Jul Aug Sep Oct Nov Dec0
50
100
150
200
250
300
350
400Panoramio
Flickr
(a)
Jan. Feb Mar Apr May Jun Jul Aug Sep Oct Nov Dec0
2
4
6
8
10
12
14
16
18Panoramio
Flickr
(b)
Figure 10: Comparison of the monthly sightings of two sundew species: (a) Drosera rotundifolia which occurs on the Northern Hemisphereand (b) Drosera arcturi from the Southern Hemisphere. The two plots clearly show the different growth periods in the Northern and SouthernHemispheres, respectively, but could also be influenced by when people travel to visit the respective habitats.
Figure 11: Our tool in exploration mode.
Figure 12: Our tool in distribution mode, showing both our distribution maps and maps from Wikipedia or other sources.
Figure 13: In exploration mode, narrowed in onto a particular site, with pictures of the site shown below.
Figure 14: In exploration mode, looking at a particular picture from the site shown in Fig. 13. The image is Flickr image 28504089402 by JeremyRiel (cbn CC BY-NC 2.0).
Figure 15: In exploration mode with the trust mapping enabled, a sitein the United States that shows the different categories of verficationfor a 1 km search radius (red: not verified by other users; orange:other plants by other users; yellow: other plants of same genus byother users; and green: other plants of same species by other users).Same map section as in Fig. 16.
Figure 16: In exploration mode with the trust mapping enabled, a sitein the United States that shows the different categories of verficationfor a 4 km search radius (red: not verified by other users; orange:other plants by other users; yellow: other plants of same genus byother users; and green: other plants of same species by other users).Same map section as in Fig. 15.
Figure 17: Verification based on the images of a single poster. If four or more images of the same poster are at exactly the same GPS coordinates,we assume the GPS coordinates to be manipulated (red). If three images are at the same location, then the coordinates are likely manipulated(orange). If two images have the same location, then the coordinates are possibly manipulated (yellow). Otherwise the coordinates are probablynot manipulated or we cannot make a good judgment (green).
(a) (b)
Figure 18: Comparison of distribution maps for Drosera rotundifolia: (a) based on our social media data, (b) map from Wikipedia (p).
(a) (b)
Figure 19: Comparison of distribution maps for Sarracenia purpurea: (a) based on our social media data, (b) map from Wikipedia (p).
(a) (b)
Figure 20: Comparison of distribution maps for Darlingtonia californica: (a) based on our social media data, (b) map from Wikipedia (p).
(a) Heatmap of the geographic distribution of all 135,578 posts in the Ency-clopedia of Life group by 2012, providing an estimate of the expected globallikelihood for habitat image posts. Image by Roderic Page (cb CC BY 4.0).
(b) The same heatmap generated based on the currently available data of349,011 posts up to July 2019, generated with the same script by Roderic Page.The overall distribution pattern did not change compared to 2012.
(c) Same image as in (b), only with the less-than-ten-images-per-region class (formerlyyellow) shown transparently, to better emphasize the locations with multiple image posts.
Figure 21: Analysis of geographic distribution data based on the Encyclopedia of Life Flickr group which collects images and videos ofanimals, plants, fungi, protists, and bacteria: Since they cover a similar subject matter but for all species in general, the maps can beused as an indication of where it is more or less likely to get a good distribution coverage (see https://iphylo.blogspot.com/2012/06/where-is-in-crowdsourcing-mapping-eol.html). Color map: yellow: 1–9 images; light orange: 10–99 images; medium orange: 100–999images; dark orange: 1,000–9,999 images; red: more than 10,000 images. Notice that in those regions that have no color assigned (i. e., thebackground map is visible) not a single image was posted to the Encyclopedia of Life Flickr group.
Figure 22: Heatmap for our data, generated with the same script by Roderic Page. Same color map as in Fig. 21.
Figure 23: Flickr’s changed upload policies from January 2019 apparently have let quite a number people to delete their images from the service.This fact may have let us to “loose” some images in the process (see Fig. 4) since we do not continuously scape the Flickr database but insteadquery it only in irregular intervals. Flickr image 6855169886 by Franck Michel (cb CC BY 2.0).
(a) (b) (c)
Figure 24: Examples for images which our search found but which we manually excluded because they do not show plausible plant habitats: (a)fake/artificial plants (Flickr image 9479368060 by Craig Moe; cbn CC BY-NC 2.0), (b) botanical gardens or similar (Flickr image 5086658928 byJane Nearing; cbn CC BY-NC 2.0), and (c) plants kept at home (Flickr image 10753181585 by Mike Linksvayer; cz CC0 1.0). Other reasonsfor rejecting entries include name collisions with geographic place names, GPS locations at unsuitable locations (e. g., farmland, streets), andimages that do not show any or not the right plants (e. g., name collisions with common or scientific names, general textual habitat descriptions inthe comments of the postings).
0 200 400 600 800 1000 1200
Drosera tomentosaUtricularia flaccidaUtricularia bremii
Drosera ascendensUtricularia nanaRoridula dentata
Drosera occidentalisDrosera alba
Drosera trinerviaUtricularia calycifida
Utricularia neottioidesDrosera parvula
Utricularia tenellaDrosera darwinensis
Nepenthes pilosaGenlisea filiformisUtricularia stygia
Utricularia involvensDrosera bulbosa
Nepenthes distillatoriaUtricularia endresii
Nepenthes ventricosaNepenthes mirabilis x rafflesiana
Nepenthes northianaNepenthes stenophylla x reinwardtiana
Nepenthes villosa x rajahNepenthes murudensisNepenthes hurrelliana
Nepenthes lowii x rajahDrosera androsacea
Nepenthes alataDrosera esterhuyseniae
Nepenthes reinwardtiana x stenophyllaDrosera villosaDrosera hilaris
Byblis (unclassified)Utricularia geminiscapa
Nepenthes angasanensisNepenthes beccarianaNepenthes sumatrana
Drosera murfetiiNepenthes stenophylla x fusca
Drosera natalensisNepenthes tobaica
Nepenthes spectabilisNepenthes platychila
Utricularia laciniataNepenthes veitchii x stenophylla
Nepenthes rajah x burbidgeaeDrosera ordensis
Utricularia kimberleyensisNepenthes x hookeriana
Byblis filifoliaNepenthes vogelii
Nepenthes suratensisDrosera chrysolepisUtricularia menziesii
Drosera barbigeraDrosera scorpioides
Nepenthes macrovulgarisDrosera madagascariensis
Utricularia odorataDrosera macrophylla
Sarracenia x catesbaeiCephalotus follicularis
Drosera miniataPinguicula villosa
Utricularia heterosepalaUtricularia erectifloraUtricularia geoffrayi
Utricularia delphinioidesNepenthes ampullaria x reinwardtiana
Nepenthes maxima x eymaeUtricularia minutissima
Utricularia unifloraNepenthes muluensis
Heliamphora (unclassified)Drosera brevicornisDrosera stoloniferaUtricularia tricolor
Nepenthes chanianaDrosera hyperostigma
Drosera cistifloraUtricularia floridana
Byblis linifloraSarracenia oreophila
Nepenthes hirsutaNepenthes ramispina
Utricularia micropetalaNepenthes gymnamphora
Utricularia amethystinaPinguicula balcanicaUtricularia longifolia
Drosera monticolaUtricularia arenaria
Genlisea (unclassified)Nepenthes hamata
Drosera stenopetalaUtricularia multifidaPinguicula longifolia
Drosera dilatato-petiolarisPinguicula macroceras
Nepenthes singalanaNepenthes pitopangiiUtricularia welwitschii
Drosera giganteaDrosera rosulataUtricularia alpina
Drosera aliciaeUtricularia striatula
Utricularia graminifoliaUtricularia asplundiiPinguicula parvifolia
Drosera linearisDrosera ultramafica
Nepenthes fuscaUtricularia leptoplectra
Drosera capensisUtricularia striata
Drosera pallidaUtricularia resupinataUtricularia laterifloraSarracenia x moorei
Nepenthes kampotianaNepenthes aristolochioidesNepenthes x kinabaluensis
Nepenthes pervilleiUtricularia chrysantha
Utricularia caeruleaDrosera fulva
Nepenthes maximaNepenthes faizaliana
Drosera unifloraDrosera roraimaeSarracenia rubraPinguicula mundiUtricularia fulva
Utricularia reticulataPinguicula moranensis
Pinguicula calyptrataPinguicula corsica
Nepenthes lowiiDrosera aberransUtricularia radiata
Utricularia jamesonianaDrosera erythrorhiza
Drosera montanaNepenthes madagascariensis
Utricularia purpurascensUtricularia unifolia
Pinguicula leptocerasUtricularia simulans
Drosera x obovataUtricularia pusilla
Utricularia macrorhizaNepenthes edwardsiana
Utricularia junceaUtricularia foliosa
Nepenthes sanguineaDrosera menziesiiDrosera petiolaris
Utricularia intermediaPinguicula elongataNepenthes mirabilis
Pinguicula nevadensisNepenthes stenophylla
Utricularia aureaPinguicula ionantha
Nepenthes albomarginataNepenthes veitchii
Nepenthes bicalcarataDrosera glanduligeraUtricularia uliginosa
Sarracenia jonesiiNepenthes burbidgeae
Drosera pygmaeaPinguicula primulifloraDrosera neocaledonica
Drosophyllum lusitanicumUtricularia australis
Nepenthes macfarlaneiPinguicula crystallina
Utricularia minorDrosera whittakeriNepenthes gracilis
Utricularia dichotomaNepenthes reinwardtiana
Pinguicula caeruleaDrosera indica
Nepenthes tentaculataDrosera burmannii
Nepenthes vieillardiiPinguicula planifolia
Pinguicula vallisneriifoliaUtricularia gibba
Nepenthes villosaDrosera macrantha
Sarracenia (unclassified)Utricularia inflata
Utricularia purpureaNepenthes rajah
Pinguicula lusitanicaDrosera arcturi
Utricularia subulataDrosera binata
Utricularia cornutaDrosera brevifolia
Pinguicula grandifloraSarracenia alataPinguicula alpinaSarracenia minorDrosera capillarisPinguicula pumila
Pinguicula (unclassified)Pinguicula lutea
Nepenthes rafflesianaDrosera spatulata
Dionaea muscipulaSarracenia rosea
Nepenthes ampullariaUtricularia vulgaris
Sarracenia psittacinaUtricularia (unclassified)
Drosera peltataDrosera filiformis
Drosera anglicaSarracenia leucophylla
Sarracenia flavaNepenthes (unclassified)
Pinguicula vulgarisDarlingtonia californica
Drosera intermediaDrosera (unclassified)
Sarracenia purpureaDrosera rotundifolia
1111111111111111111111111111111111111111111111111111122222222222222222222233333333333333334444444555555556666666666777777777788888899999101010101111111112131414141415151616161616161617171717181818181920202020212122222324262626282828292930303234343435363839404142434450535355
6061626364
7475768080
939598
105107110
120141
157178
193205
244260
305322
468541
9701301
Figure 25: Species count in complete database of all species that we found, sorted by rank.
0 200 400 600 800 1000 1200
Utricularia welwitschiiUtricularia vulgarisUtricularia unifoliaUtricularia uniflora
Utricularia uliginosaUtricularia tricolorUtricularia tenella
Utricularia subulataUtricularia stygia
Utricularia striatulaUtricularia striata
Utricularia simulansUtricularia reticulata
Utricularia resupinataUtricularia radiataUtricularia pusilla
Utricularia purpureaUtricularia purpurascens
Utricularia odorataUtricularia neottioides
Utricularia nanaUtricularia multifida
Utricularia minutissimaUtricularia minor
Utricularia micropetalaUtricularia menziesii
Utricularia macrorhizaUtricularia longifolia
Utricularia leptoplectraUtricularia laterifloraUtricularia laciniata
Utricularia kimberleyensisUtricularia juncea
Utricularia jamesonianaUtricularia involvens
Utricularia intermediaUtricularia inflata
Utricularia heterosepalaUtricularia graminifolia
Utricularia gibbaUtricularia geoffrayi
Utricularia geminiscapaUtricularia fulva
Utricularia foliosaUtricularia floridanaUtricularia flaccida
Utricularia erectifloraUtricularia endresii
Utricularia dichotomaUtricularia delphinioides
Utricularia cornutaUtricularia chrysantha
Utricularia calycifidaUtricularia caerulea
Utricularia bremiiUtricularia australis
Utricularia aureaUtricularia asplundiiUtricularia arenaria
Utricularia amethystinaUtricularia alpina
Utricularia (unclassified)Sarracenia x moorei
Sarracenia x catesbaeiSarracenia rubraSarracenia rosea
Sarracenia purpureaSarracenia psittacinaSarracenia oreophila
Sarracenia minorSarracenia leucophylla
Sarracenia jonesiiSarracenia flavaSarracenia alata
Sarracenia (unclassified)Roridula dentata
Pinguicula vulgarisPinguicula villosa
Pinguicula vallisneriifoliaPinguicula pumila
Pinguicula primulifloraPinguicula planifoliaPinguicula parvifolia
Pinguicula nevadensisPinguicula mundi
Pinguicula moranensisPinguicula macroceras
Pinguicula luteaPinguicula lusitanicaPinguicula longifolia
Pinguicula leptocerasPinguicula ionantha
Pinguicula grandifloraPinguicula elongata
Pinguicula crystallinaPinguicula corsica
Pinguicula calyptrataPinguicula caerulea
Pinguicula balcanicaPinguicula alpina
Pinguicula (unclassified)Nepenthes x kinabaluensis
Nepenthes x hookerianaNepenthes vogelii
Nepenthes villosa x rajahNepenthes villosa
Nepenthes vieillardiiNepenthes ventricosa
Nepenthes veitchii x stenophyllaNepenthes veitchiiNepenthes tobaica
Nepenthes tentaculataNepenthes suratensisNepenthes sumatrana
Nepenthes stenophylla x reinwardtianaNepenthes stenophylla x fusca
Nepenthes stenophyllaNepenthes spectabilisNepenthes singalana
Nepenthes sanguineaNepenthes reinwardtiana x stenophylla
Nepenthes reinwardtianaNepenthes ramispina
Nepenthes rajah x burbidgeaeNepenthes rajah
Nepenthes rafflesianaNepenthes platychilaNepenthes pitopangii
Nepenthes pilosaNepenthes pervillei
Nepenthes northianaNepenthes murudensis
Nepenthes muluensisNepenthes mirabilis x rafflesiana
Nepenthes mirabilisNepenthes maxima x eymae
Nepenthes maximaNepenthes madagascariensis
Nepenthes macrovulgarisNepenthes macfarlaneiNepenthes lowii x rajah
Nepenthes lowiiNepenthes kampotianaNepenthes hurrelliana
Nepenthes hirsutaNepenthes hamata
Nepenthes gymnamphoraNepenthes gracilis
Nepenthes fuscaNepenthes faizaliana
Nepenthes edwardsianaNepenthes distillatoria
Nepenthes chanianaNepenthes burbidgeaeNepenthes bicalcarataNepenthes beccariana
Nepenthes aristolochioidesNepenthes angasanensis
Nepenthes ampullaria x reinwardtianaNepenthes ampullaria
Nepenthes albomarginataNepenthes alata
Nepenthes (unclassified)Heliamphora (unclassified)
Genlisea filiformisGenlisea (unclassified)
Drosophyllum lusitanicumDrosera x obovataDrosera whittakeri
Drosera villosaDrosera uniflora
Drosera ultramaficaDrosera trinervia
Drosera tomentosaDrosera stolonifera
Drosera stenopetalaDrosera spatulata
Drosera scorpioidesDrosera rotundifolia
Drosera rosulataDrosera roraimaeDrosera pygmaeaDrosera petiolaris
Drosera peltataDrosera parvulaDrosera pallida
Drosera ordensisDrosera occidentalis
Drosera neocaledonicaDrosera natalensis
Drosera murfetiiDrosera monticolaDrosera montanaDrosera miniata
Drosera menziesiiDrosera madagascariensis
Drosera macrophyllaDrosera macrantha
Drosera linearisDrosera intermedia
Drosera indicaDrosera hyperostigma
Drosera hilarisDrosera glanduligera
Drosera giganteaDrosera fulva
Drosera filiformisDrosera esterhuyseniae
Drosera erythrorhizaDrosera dilatato-petiolaris
Drosera darwinensisDrosera cistiflora
Drosera chrysolepisDrosera capillarisDrosera capensis
Drosera burmanniiDrosera bulbosa
Drosera brevifoliaDrosera brevicornis
Drosera binataDrosera barbigera
Drosera ascendensDrosera arcturiDrosera anglica
Drosera androsaceaDrosera aliciae
Drosera albaDrosera aberrans
Drosera (unclassified)Dionaea muscipula
Darlingtonia californicaCephalotus follicularis
Byblis linifloraByblis filifolia
Byblis (unclassified)
5110
143
213
153
167
1510
711
1643
142115
228
32
16477
11
1612
117
4226
3821
1016
3121
292
608
18
126
186446
1417
29
105970
1203
74205
21244
6341
1305
236
7623
356
18910
580
505
1519
6217
281110
304
6480
8121
3934
11
201
342111
1815
161
303
144
9315
18
113
118
28
142
261
117
134
329
79
1613
2220
17
12
10720
1260
31
426
1628
19
61135
952
130169
2217
1571
711
2411
414
217
22
406
46832
31
2068
1781
135
132
757
341
613
5521
53193
16
111
54198
32223
11
Figure 26: Species count in complete database of all species that we found, sorted by species name.