Wordless Wednesday posts are images from Ardnamurchan and the surrounding regions – Sunart, Morvern, Ardgour, Moidart and the Rough Bounds. They have no accompanying text or description. I will try and ensure they were photographed in the same month, though not necessarily the same year, that they appear online.
Ardnamurchan has excellent birdwatching. In a relatively small area it has a variety of habitats – mountain, moorland, all types of woodland, raised bog, coast, sea and freshwater lochs, machair and just about everything in between.
I’m an amateur, but enthusiastic, birdwatcher. Like many birdwatchers I try and keep a running list 1 of what I see locally (my patch) and on my travels.
I’m not a twitcher. I don’t chase rarities, I don’t keep monthly, annual or geographic records. I’m hopeless at discriminating between most of the Phylloscopus (leaf) warblers (and gulls 2). I do however have a vague idea of the – very extensive – list of UK birds I have never seen and enjoy reducing the number when I have the chance.
It’s always good to know what might be seen in a particular area i.e. what has been seen before and is therefore likely to turn up again.
Or that’s there all the time if I was only a bit more observant 😉
The BTO open mapping bird atlas data
Fortunately, thanks to thousands of volunteers and the British Trust for Ornithology, there are outstanding records of overwintering and breeding birds for the entire country. This dataset now contains ~1.4 million records covering ~450 species between 1968 and 2011. It is updated every couple of decades.
This data is at 10 km resolution (i.e. was a bird present / breeding in a particular 10 km square?). More recent surveys have been conducted at much higher resolution (based upon 2 km tetrads) and now include ~20 million records covering 99.9% of the country.
Until recently this data was only available in book form, but the 10km resolution data is now freely available online. For any species you can view a map of the UK showing if and where it is present in the winter, and whether it possibly, probably or definitely breeds here in the summer.
BTO bird atlas mapping
Inevitably, the maps are rather low resolution. You can just about see Ardnamurchan, but it’s not obvious. The BTO also offer a variety of other ways to view the data, such as providing lists of birds by searchable map reference or postcode.
Local lists of local birds
Not only can you query the dataset interactively, you can also download the bird distribution data as compressed ZIP format comma-separated variable (CSV) text files. Uncompressed the files total about 125Mb. The largest is the geographic and temporal distribution data which runs to ~1.5 million lines. Don’t try opening it in Excel (which is probably the default for this file type on your computer) as it can’t cope with more than ~1 million rows 3.
A wet spring weekend seemed like the ideal opportunity to generate an Ardnamurchan-specific bird list. CSV files are trivial to manipulate using a simple scripting language such as perl or python. What could possibly go wrong?
10 km resolution
Of those 1.5 million records, only a subset are relevant to Ardnamurchan. The rest of the country we can safely ignore.
The first challenge is to determine which 10 km squares overlay the ~30 km x ~10 km Ardnamurchan peninsula. The BTO uses the same grid overlay used by the Ordnance Survey and the latter provide a very useful tile locator map. If you select the relevant 100 km square (NM) you are taken to a more detailed view showing the individual labelled 10 km squares.
Ardnamurchan 10 km map squares
The numbering of these squares represent the first digits of the Eastings and Northings coordinates of the grid reference. For example, Acharacle primary school is at NM674 681 and is in the top right hand (north west) corner of NM66.
NM47, 57, 67, 46, 56, 66 covers Ardnamurchan in its entirety 4, together with bits of Muck, Moidart and a small wedge of Morvern. That’ll do.
For the moment the only two data files that are of relevance are the lookup table of species (species_lookup.csv) and their distribution (distributions.csv).
species_lookup.csv has four fields, the unique species code (an integer 5, the common and scientific name(s) and an indication whether it is a distinct species or an aggregate 6.
distributions.csv has eight fields including the survey period, season, species (the id number only), 10 km reference square and whether breeding was possible, probable or confirmed. I’ve ignored the additional fields for the moment. With ~1.5 million lines in the file there are multiple records for each 10 km square and for each species.
For example, there are 2082 records for the six 10 km squares covering Ardnamurchan and 9290 records for the Barn Owl (Tyto alba) covering the entire country.
Of these, only seven intersect. The Barn Owl is present in five squares during the winter, but breeding is only suspected or known in NM67 and NM56.
This doesn’t really qualify as ‘big data’ but untangling it is too much to handle with just a pen and paper 7. Instead, it’s a trivial exercise to spend a wet afternoon (when I couldn’t do any beekeeping) writing some perl scripts to extract the relevant records for presentation.
Data mangling and untangling
The historical information in the BTO Bird Atlas data is of less interest to me (at the moment at least 8). I’m interested in the most recent reports from the winter and breeding season records … these are the birds I (and visitors to Ardnamurchan) are most likely to see now.
Maclean’s Nose … a good spot for white-tailed and golden eagles
Following the usual cycle of code, debug, coffee, code, debug, coffee pizza I had something that just about worked.
A total of 143 species overwinter, breed or both on Ardnamurchan. Of these, 26 (mostly overseas migrants like the cuckoo and willow warbler) are absent in winter. In contrast, 34 species are either absent in the summer or do not breed on Ardnamurchan.
Unsurprisingly considering the limited land area present in NM47 and NM57, these 10 km squares have the smallest number of summer or winter species:
NM47 – winter 59, breeding 56
NM57 – winter 49, breeding 48
NM67 – winter 83, breeding 89
NM46 – winter 66, breeding 89
NM56 – winter 77, breeding 81
NM66 – winter 103, breeding 90
It’s worth noting here that presence alone in the summer is not recorded for any species on Ardnamurchan. Just because a particular species does not breed does not mean there is no chance of seeing it 9. For example, I’ve regularly seen gannets in Loch Sunart though they don’t breed on the peninsula.
Having sliced and diced the BTO data it now needed to be presented in a readable format.
Rather than just present a ‘flat’ list of birds it seemed logical to include links to relevant information online. For example, to the BTO Bird Atlas maps of the entire UK for the winter and breeding season distributions and to the BTO Bird Facts data.
This turned out to be a little bit more complicated than it should have been. The BTO appear to use two different numbering schemes.
Species in the BTO Bird Atlas use a taxonomically-organised sequentially-numbered list of UK species. I’ve mentioned the red throated diver (#1) and Blue-crowned Parakeet (#1663) previously. In contrast, the Bird Facts pages are referenced using the 5 digit EURING code used by bird ringers. The EURING code for the Red throated diver is 00020 10.
Initial attempts to find the Blue-crowned Parakeet on the EURING list failed which highlighted a second problem 11. The scientific names (the only sort of names used) on the EURNG lists are sometimes different from the BTO lists. This happens when birds are reclassified. On the BTO the Blue-crowned Parakeet is Thectocercus acuticaudatus whereas EURING lists it as Aratinga acuticaudata. D’oh!
Entirely the wrong time of the year to see Blue-crowned parakeets …
I cross-referenced the two lists automagically using a short perl script and then manually corrected the dozen or so species from Ardnamurchan that disagreed between BTO and EURING systems 12.
Inevitably this manual bodge will break things in the future …
A picture is worth a thousand words
One of the problems with the BTO Bird Atlas maps is that the Ardnamurchan peninsula is rather difficult to see. I therefore created a series of small 3×2 gridded icons representing the six NM-prefixed squares that cover the peninsula. I used one icon for the winter distribution and one for the breeding distribution (with probability indicated by different sized ‘blobs’, just as they are on the BTO Bird Atlas maps). For example:
This icon represents a species that is recorded in all but NM57 in winter.
This icon indicates a species which possibly breeds in NM46, probably breeds in NM66 and NM67 and that breeding has been confirmed in NM56 and NM57.
There are ~132 different distributions of birds in the winter and breeding seasons. Rather than draw any of the icons by hand I used ImageMagick to create them automagically, called from a perl script that read the underlying data directly from the distributions.csv file.
I also added a notes column, but this has yet to be populated.
The finished list of Ardnamurchan birds in the most recent BTO census is linked from the menu above, or directly from here (and is also available in PDF format). It will be updated as and when new BTO census data is released.
It should be noted that for security reasons the breeding distribution of some rare birds is deliberately listed (by the BTO) at lower resolution.
There’s a second minor problem with the BTO Bird Atlas maps. They are displayed via a server which separates the delivery of the map from the menus that control the data displayed or its appearance. This means I could either link directly to the maps (and have no menus) or to link to the default map for a species and subsequently select the desired map (which has to be done interactively). The default map is the most recent breeding distribution, so this is linked from the ‘summer’ icons. Until I find a way to access the winter distributions and menus I’ve left the ‘winter’ icons linked directly to the relevant map alone.
The Blue-crowned Parakeet (Thectocercus acuticaudatus) is a small green South American parrot with a blue head. They inhabit Neotropical savanna-like habitats, woodland and forest margins from eastern Columbia to northern Argentina. Not Ardnamurchan 😉
Having discussed the Triffid-like rhododendrons that infest parts of Ardnamurchan and the tools needed to clear them, here’s a further instalment showing how to ensure they don’t re-grow.
Rhododendron (Rhododendron ponticum) are relatively shallow-rooted woody shrubs. They can spread by wind-dispersed seed and by suckering. Suckers are more correctly termed basal shoots and emerge from the root system.
Dense rhododendron growth with no understorey
Each of these suckers essentially has the ability to form a new plant.
If you cut a rhododendron to near ground level – leaving just enough of the woody stump to trip over 🙁 – it will usually produce new growth.
Even if it doesn’t produce growth from the stump, it will almost certainly generate suckers from which large new plants will eventually grow.
Consequently, simply cutting rhododendrons to the ground and leaving the roots untouched results in dense multi-stemmed thickets forming.
Multi-stemmed rhododendron from suckering
Therefore to ensure the rhododendron does not re-grow you must killthe roots of the plant. There are a variety of methods that can be used to achieve this, some faster than others.
You can cut and clear the rhododendron and then periodically return to the stumps and, using something like a hammer, smash down the re-growth. Over a period (perhaps a long period) the stump will be killed.
There is an approach called the ‘lever and mulch’ method in which the shrub is cut down and the amputated top growth is used to cover the stump, effectively starving any new growth of light.
I’ve not tried either of these methods – I don’t have the time and want the site cleared – and so cannot comment further on them.
Environmentally unfriendly methods
A very common approach is to cut down the plant and then return after it has re-sprouted to spray with a systemic weedkiller such as glyphosate (in commercial products Roundup or Gallup).
Spraying requires a dry relatively calm day. If it’s wet and/or windy the weedkiller will either be less effective or may damage other plants.
Ardnamurchan has fabulous weather but it’s not always suitable for spray treatment 😉
I’m also not keen on the relatively indiscriminate application of weedkillers so have not used this method.
Drill and drop
What’s needed is a method that applies a weedkiller such as concentrated glyphosate in small amounts precisely where it is needed.
This can be easily achieved using stem injection.
The method is simplicity itself. A 10mm hole is drilled near the base of the trunk and 1-2ml of strong weedkiller is applied directly into the hole. The weedkiller is transported throughout the plant and, over a period of weeks or a few months, the plant is killed.
Once the plant is dead it can be cut down and burnt.
For pretty-obvious reasons this method is known as drill and drop.
Birds-eye view of a multi-stemmed rhododendron drilled for glyphosate application
Notice the holes an inch or two above the leaf litter in all the major stems. Why, if ‘… the weedkiller is transported throughout the plant …’ does every stem have to be drilled?
That’s because I oversimplified the method. The weedkiller appears to only kill the top growth and root of the individual sucker. It’s not entirely clear why. I presume that the new growth establishes its own root system over time (some of these stems are 2-3″ in diameter and 15-18 feet long) and that the weedkiller only spreads systemically within this new growth.
Consequently every stem must be drilled … and some will be missed 1.
Therefore, there is another method that can be used …
Drop and drill
You cut the plant down first, drill the remaining stump and apply the weedkiller. This approach has the advantage of immediacy. There’s no waiting for the rhododendron to die. It also has the advantage of thoroughness. Since you’ve cut all the top growth down you cannot inadvertently miss any of the stems.
Drop and drill
The additional advantage is that you are generally drilling from the top into the stump. It’s easier to generate a clean hole and there’s even less chance of spilling weedkiller.
A battery-powered portable drill with a 10mm wood bit is needed. If the rhododendron growth is very dense or very extensive or – in my case – very dense and extensive 🙁 then it’s worth having a spare charged battery or two so you aren’t forced to stop just as you’re beginning to have fun.
The hole needs to be 2-5 cm deep. It is best drilled at an acute angle to reduce the chance of excess weedkiller running down the stem onto other plants. One hole is generally sufficient, even for the largest stems.
Stems as narrow as 2 cm in diameter can also be drilled. Choose a place where the growth is horizontal – close to the original stump – and drill a 1 cm deep hole, filling it with weedkiller to just below the brim.
Choose your poison
The weedkiller I use is Gallup glyphosate. Bought in bulk this costs about £20 for 2 litres. Because you’re applying a very small amount to a single, potentially very large, plant it needs to be used quite concentrated. I’ve used it at 20-25% i.e. mixed 1 in 4 or 1 in 5 with water.
Prepare a stock solution in a well-labelled plastic bottle. Don’t make up much more than you need as I’ve read it degrades over time and so loses efficacy. I make up 250ml which is enough for a very rewarding afternoon of rhodo-poisoning.
Trickle 2 container for delivery of glyphosate
Administering 1-2 ml to a small hole in the base of a drilled stem requires a steady hand and a bottle with a nozzle that doesn’t drip or leak. I’ve used what a beekeeper would know as a Trickle 2 oxalic acid bottle. These probably have a proper name, but I’ve no idea what it is.
The bottle has a 100 ml reservoir and a small upper chamber that takes 5 ml. It has a twist-lock nozzle and doesn’t drip. It needs to be refilled every 50 or so stems. Remember you’re probably using it on your hands and knees in dense undergrowth. A refill is a welcome opportunity to stand up and stretch.
Drill and drop vs. Drop and drill
In practice I’ve found I have to use a mix of the two approaches. Each has advantages and disadvantages, but both are usually needed.
Drill and drop is great because the killed shrubs are much lighter to cut down, easier to transport and burn faster. It takes about 3 months for the plants to die – the upper growth withers and finally drops off.
The disadvantage of drill and drop is that you need to get to the base of the stems in very heavily-overgrown areas. Almost inevitably this means crawling about, working in confined dingy spaces and involves encounters with brambles, wasps nests and sharp pointy things.
Wear safety glasses.
New growth retrospectively drilled and killed
Almost inevitably you’ll miss some stems. Drop and drill has the advantage of immediacy. You cut the plant down and then drill and poison the stump. You gradually work your way through a dense thicket, but the drilling and glyphosate delivery is done in an already cleared area so is much easier.
Slash and burn
I prefer to drill the plants, let them die back, cut them all down and return to ‘mop up’ the remainders I missed the first time round. I usually carry a pen to mark stumps cut down ‘live’ that need retrospective treatment 2. This avoids double-dosing.
If you cut the trunk near but below the previously drilled site the woody interior of the stem is usually stained distinctively, so you can avoid re-dosing it.
Waiting for the inferno
The side branches and top growth are removed with a billhook, stacked up and subsequently burnt.
The main stems are cut up and dried for firewood.
Something for the winter
That lot is about half what I prepared from the 1600 m2 (i.e. 40 m x 40 m) area I’ve cleared so far.
Glyphosate – mode of action and toxicity
Glyphosate is a broad-spectrum (i.e. indiscriminate) weedkiller that inhibits the plant enzyme 5-enolpyruvylshikimate-3-phosphate synthase. This enzyme is critical for the synthesis of the aromatic amino acids tyrosine, tryptophan and phenylalanine. Production of these amino acids is important during plant growth which explains why glyphosate only kills growing plants.
Glyphosate was discovered by Monsanto, given the trade name Roundup and first used for agricultural purposes in 1974.
It is adsorbed through the leaves, through drilled or cut stems but only poorly through the roots. Stem drilling, or stem injection, is a much more effective way of administering glyphosate than simply painting it onto a cut stem; there is less wastage and much less potential for collateral damage.
Glyphosate is extremely widely used. Consequently, residual amounts are sometimes found in food products. The levels are very low, with over 99.5% of nearly 7000 food products tested in 2016 having less than the maximum allowed residue levels.
However, the very widespread use and potential long-term exposure to low levels of glyphosate mean that it has repeatedly been associated with causing human cancers. Monsanto (now Bayer) have recently lost a court case when sued by a an individual who had developed non-Hodgkin lymphoma, claiming it was caused by long-term exposure to high levels of glyphosate while working as a gardener. Monsanto is appealing against the decision.
The scientific evidence (a very different level of ‘proof’ to a law court) supporting glyphosate being a carcinogen is very limited and equivocal.
A long-term (decades) study of 55,000 people showed that there was no increased risk of cancer in those exposed to higher levels of glyphosate.
Very recently a meta-analysis (a statistical study of other studies) has linked exposure to very high levels of glyphosate to non-Hodgkin lymphoma. Despite the headline figure being a scary 41% increased risk 3 of developing non-Hodgkin’s lymphoma the absolute risk was only about 0.5%. And this was only for people exposed to very high levels over a very long period.
There’s wasa good account of this subject on the More or Lesspodcast recently.
Nevertheless, I always take care when using glyphosate. I wear disposable nitrile gloves and safety glasses. I administer small doses (1-2 ml) to the drilled holes, avoid spilling it and wash my tools and hands carefully after use.
A blank canvas
Ardnamurchan or Borneo?
Cleared and ready for native tree planting
There’s still more to do (measured in acres not m2) but after a bit more clearing this area will be planted up with native trees in the 2019/20 winter.