Roadside Tree Survey Summary.
For fun, I tried to see what would happen to photograph trees while driving along the roadside. Since I dont know trees, I added them to the https://www.inaturalist.org/projects/trees-of-southern-africa-id project
At present the status of these are:
June 2021 Kruger: mainly gravel speed 60km/h - 636 observations.
Status after 18 months
- 204 No ID (32%)
- 182 Plants (29%)
- 18 Dicots [almost all these "plants" are dicots]
- 5 Family-Tribe
- 46 Genus (7%)
- 181 Species (28%) - 169 RG (27%)
Kruger has relatively many active tree identifiers working in the lowveld.
@wynand_uys gave some useful ideas for improving the photographs (camera speeds, use oncoming trees, etc.).
August 2022 Caprivi tarred road 100km/h - 2,307 observations (lots of driving!)
Diverse: lots of tree species. Status after 4 months - not many tree specialists active on iNaturalist in this area.
- 2,035 No ID (88%)
- 65 Plants (3%)
- 10 Dicots [almost all these "plants" are dicots]
- 5 Family-Tribe
- 65 Genus (3%)
- 110 Species (5%) - 84 RG (4%)
August 2022 Kalahari (Botswana & Namibia) varied roads 100km/h - 1,822 observations (lots of driving!)
Thornveld: not so diverse. Status after 4 months - not many tree specialists active on iNaturalist in this area.
- 1,489 No ID (82%)
- 5 Plants (3%)
- 8 Dicots [almost all these "plants" are dicots]
- 17 Family-Tribe
- 31 Genus (2%)
- 272 Species (15%) - 163 RG (9%)
So all in all: not very fruitful, although in none of the above where people specifically approached to help identify the species. This has been entirely passive.
The question is - Is it worth it?
Want to help with IDs - i.e to Family, Genus or Species? Please click:
Comments
Which species were identified in these runs?
(5 or more observations).
Presumably these are the distinctive trees that are easily identified. Note that flowering and seeding may make some of these seasonally easy (or difficult).
KRUGER:
21 Sclerocarya birrea Marula
19 Senegalia nigrescens Knobthorn
17 Terminalia sericea Silver Terminalia
16 Philenoptera violacea Appleleaf
13 Combretum apiculatum Red Bushwillow
9 Euclea Guarris
9 Diospyros mespiliformis Jackalberry
8 Combretum imberbe Leadwood
8 Bolusanthus speciosus Tree-Wisteria
6 Cassia abbreviata Sjambok-Pod
5 Peltophorum africanum Weeping Wattle
5 Vachellia tortilis Umbrella Thorn
(52 species total)
CAPRIVI
35 Pterocarpus angolensis Kiaat
16 Hyphaene petersiana Makalani Palm
16 Senegalia nigrescens Knobthorn
11 Adansonia digitata African Baobab
10 Vachellia erioloba Camel Thorn
8 Strychnos Monkey Oranges
(26 species total)
KALAHARI
138 Senegalia mellifera Black Thorn
26 Vachellia erioloba Camel Thorn
21 Vachellia nebrownii Water Thorn
15 Vachellia hebeclada Candlepod Thorn
13 Prosopis glandulosa Honey Mesquite
13 Boscia albitrunca Shepherds Tree
11 Parkinsonia africana Greenhair Tree
5 Rhigozum brevispinosum Kalahari Gold
(31 species total)
@botswanabugs asked "Perhaps you could have posted some model tree pics, close up and distant so we could gain some expertise at IDing distant and blurry trees with more accuracy."
There are. The overnight stops are probably the best:
CAPRIVI
Ngepi
Namushasha
KALAHARI
Drotskys
Gobabis
There are a few more smaller sites, but the number of trees seen on tea stops is quite small (and finding them on iNaturalist is quite difficult! [fortunately they have multiple pictures])
e.g. Caprivi 1 - - Caprivi 2
Kalahari 1 - - Kalahari 2 - - Kalahari 3
An interesting exercise.
I had a quick peek at the Caprivi images. Less motion blur and generally better than the Kruger images but since I'm not familiar with that area I find it difficult to suggest ID's without some closer shots.
Is it worth it?
I suppose it depends on what you aim to achieve and whether a fair number of the observations get ID'd.
I noticed that the iNat Computer Vision is becoming more skilled with photos like these.
Some might regard it as a fun challenge to ID the drive-by shootings. I did, for a while but my enthusiasm has flagged a bit, sorry.
Yes, it is a job for the long cold, rainy, winter nights.
Many thanks for all the IDs.
I'll have another go with the Kalahari shoots.
Acknowledgements: Thanks for IDs (genus and species, top 5 or more than 10 IDs) at present to:
Kruger: @wynand_uys (246). @jaheymans (68), @sedgesrock (29), @troos (23) & @lotteryd (11)
Caprivi: @wielies (53), @alanhorstmann (11), wynand_uys (8), @charles_stirton (7) & sedgesrock (6)
Kalahari: @botswanabugs (115), @alexdreyer (30), alanhorstmann (13) & troos (8)
Many thanks.
i just went through 1000+ of your "unknown" plant pictures from august 2022 and identified them broadly, most of them as simply vascular plant, hopefully that helps them get ided! with some level of identification in the system, plant people on inat will have an easier time finding the observations and hopefully some of them will have some thoughts on what kinds of vascular plants they are. the ai recognition software has some ideas as well :)
the volume of observations is impressive!! hopefully someone goes through them that knows more ab what they are. it would do a lot to train the ai for the area.
@snake_smeuse thanks I guess they are all dicots. I need some HI recognition training ( Human Intelligence) but I am sure that could be done if we have pics of known trees from a distance.
All the trees are already in the Trees of Southern Africa project. Unfortunately our community is too small to bother with any plant observations that are above Family Level - and our tree specialists use the Tree project, so what they are identified as or not is irrelevant.
What this shows is two things
* our specialists are easily overwhelmed by volume
* trees are only identifiable to fine level by local specialists, and we have none in the arid Kalahari or Caprivi.
I only hope that it does not mean that observations of trees like this cannot be identified: that would be a real shame.
Working on getting the Nambians on board, but they have just launched their own CS site, so probably wont happen for a while
http://www.the-eis.com/atlas/
Thanks for trying.
@botswanabugs that's true, i reckon they're all dicots. not sure why i didn't go with that id. i am far too lazy to go back through them all, though.
@tonyrebelo thanks for the info! i suppose it does make sense for such a small community to have a project like that rather than using the inat system for finding trees to identify. maybe someone else that doesn't usually identify in your group will come along and know about the trees, though you'd have probably found them by now!
Add a Comment