Dr. Maryna Kuzmenko

Bittersweet news, or perhaps it is more reasonable to say: salty news 🥺

The good thing is that the ‪@petiolepro‬ - Petiole Pro app was trusted once again to measure leaf area in an international consortium project in the largest continuous mangrove forest on Earth - Sundarbans 🌏

The bad news: we are losing the diversity of this priceless ecosystem.

Regarding salt.
In mangrove forests, salinity is silently but confidently reshaping the whole ecosystem.

The question is: what happens to mangrove functional diversity when salinity increases?

Not only “which species are present” but a bit deeper:

→ What traits do the leaves have?
→ How do plants adjust to salt stress?
→ Does the ecosystem become more functionally rich — or more similar, narrow, and vulnerable?

The researchers from Bangladesh, Canada, China, USA, UK, and Australia, measured eight foliar traits across mangrove plots, including
-> leaf area
-> specific leaf area
-> leaf dry matter content
-> chlorophyll
-> stomatal density
-> leaf shape
-> succulence
-> leaf carbon content.
-> and they also analysed functional diversity indicators.

The main message is serious: higher salinity reduced functional diversity, especially trait dissimilarity.

In simpler words it means that as the environment became saltier, mangrove communities became more similar in the way they function.

This is called trait convergence.

Under stress, nature starts to “select” species and traits that can survive. Plants with similar salt-tolerant strategies dominate. The community may still look like a forest, but functionally it becomes narrower.

The leaves also changed in predictable ways.

With higher salinity, the study found reductions in leaf area, leaf dry matter content, stomatal density, chlorophyll, and leaf carbon content.

And increases in:

→ leaf succulence
→ specific leaf area

This tells a story of adaptation. Plants shifting toward water storage, osmotic balance, and survival under salt stress.

But survival is not the same as resilience.

When many species begin to rely on similar strategies, the ecosystem may lose functional “options”.

And fewer functional options can mean lower capacity to adapt to future stress — especially under sea-level rise, freshwater reduction, and climate-driven salinisation.

___

One more important detail: species abundance also mattered.

High abundance of a few stress-tolerant species reduced trait dissimilarity and evenness. So the problem is not only salt itself, but also how salt reshapes dominance inside the plant community.

This is why trait-based ecology is so powerful and here is where technology can help.

👉 Remote sensing, computer vision, spectral imaging, and AI-assisted trait monitoring could support future mangrove conservation by detecting changes in canopy traits, greenness, chlorophyll, and stress signals at larger scales.

Not as a replacement for field ecology.
But as a way to scale it.
What do you think?

#Mangroves #ClimateChange #Biodiversity

Skip the Carousel

1 2 3 4 5 6 7 8 9 10

1 week ago | [YT] | 4

Dr. Maryna Kuzmenko

“There is no rain in the morning. The probability of rain at 12:00 PM is 75%.”

I asked two AI assistants and checked BBC Weather.
In one voice, all resources confirmed: cloudy, dry morning for Day 2 at @CerealsOfficial 2026.

“Perfect,” I thought. “I have at least three hours to do everything important before the heavy rain…”

-

However, at our final destination — Diddly Squat Farm, the weather was obvious.
The bus driver joked that I would probably need wellies 😂

“Absolutely agree,” I answered, while opening my umbrella and stepping into fresh mud created by thousands of legs and hundreds of tyres.

“But I will survive,” I said, navigating directly to the heart of cereal innovations 😉

_

And in this short muddy story, there is so much truth about AI weather forecasting in a changing climate.

And about AI coming to farms.
So, a few key thoughts from Cereals 2026.


1. Money-driven agriculture.
Subsidies are uncertain. Farming still needs to work and stay profitable.
That is why regenerative agriculture is now a strong background topic.
It is no longer only an ethical or environmental story.


2. Nitrogen-fixing crops are becoming more important
I liked seeing so many nitrogen-fixing crops.
Legumes, beans, peas, clover-type systems — not just “additional plants,” but part of the bigger conversation from Point 1.



3. Alternative crops are becoming more visible
It was good to see pseudocereals and beyond - buckwheat, linseed, sunflowers and lots options for companion cropping and soil-building systems

This matters because cereals are no longer only about dominating in the UK wheat, barley, rapeseed and oats.



4. Precision farming is becoming very practical
Because of Point 1, everyone is trying to optimise inputs and get not only yield, but better soils too.

So this is where drones, satellite imagery, sensors, variable-rate application, camera-guided sprayers, soil maps, and AI decision support start to make sense.

But... Point 5


5. Technology must survive real farm conditions
Most of what I saw was machinery.
Probably because walking between giant tractors and equipment was less muddy for me :)

Drones were not flying because of the weather.
I heard there was a robodog — not by its barking, but through crowd rumours.

There were also animal-tech innovations for herd management and animal health, which is a really good trend.

Many arable farmers also keep animals as a second revenue stream when cereal production is not as profitable as we wish.

Farmers are getting used to technology on farms.
But the real pain is bringing a working demo into real life — where wellies are needed.

_

PS: At 12:00 PM, there was NO rain at Diddly Squat Farm & a big crowd was listening to Britain’s Got Talent finalists, The Hawkstone Farmers’ Choir.

PS-2: The biggest losers of this story are my pink wellies.
Such a missed opportunity.

But it brings me to one clear idea: I need to come back to Cereals 2027.
Preparation has started :)

#cereals2026

Skip the Carousel

1 2 3 4 5 6 7 8 9 10

2 weeks ago | [YT] | 25

Dr. Maryna Kuzmenko

I'm terrible at football.
I don't even properly know the rules 😂

But there is something in this game that makes it an interesting thing.

It's the spirit.
The feeling of belonging to some team — even if I don't know their names and numbers, not even speaking about their roles in the game 😁

But to me it's an amazing feeling.
Particularly once every four years, when it's the World Cup.

--

That's why I just cannot figure out when to have my YouTube livestreams now.

They coincide with so many matches, and I understand that I should not disturb this natural life flow of events.

My AI jokes about Old MacDonald’s farm and technologies simply do not fit into the football calendar right now.

I hope the very smart YouTube algo-rithms will forgive me for my bad behaviour, particularly, when I will come back :).

--

Hence, as of today — full CANCELLATION.

But if you have already purchased the ticket for my Saturday livestream, please keep it.
I will come back.
Promise 💯👌🤝

2 weeks ago | [YT] | 14

Dr. Maryna Kuzmenko

Apologies, my dear friends and YouTube algorithms. I’m not sure about today’s Livestream 🫣

I’m currently lost in the clouds and navigating only by the summerish mood. Not sure if I have a chance to escape back to my traditional routine today.

But we can try to do the Cross-Border synchronization even without Livestream.

Brief instruction:

1. Go outside.
2. Lift your head.
3. Look up into your clouds 🤩
4. I’m also checking the clouds right now.
5. Is your sky the same as mine?
Any abnormally looking clouds?
Any misshapen pieces?
Any discoloration?

Please, let me know in the comments 🤗
I'll come with AI and we'll double check 🤝

___

PS: according to my hyper local weather forecast, tomorrow I will have a rainy evening. What a perfect time to discuss old McDonald's transition from traditional farming to his vertical farm facilities 😂

1 month ago | [YT] | 5

Dr. Maryna Kuzmenko

Necrotic leaf area damage needs to be measured.
This was the topic of our latest LinkedIn Live on Friday, 22 May.

So let's sum up the practical side of the discoveries.

1. On production level, assessment in percentage is OK.
At least two indices can be enough to understand the dynamics.
Then it is possible to get a combined index of healthy / damaged leaf area and use it as a practical indicator for decision-making.

2. But these indices must be calculated within a specific timeframe.
For example, every 5–7 days, depending on the crop, disease pressure and environmental conditions.
In any case, one number is not enough.
To get real insight, there is a need for two or more measurements in a row.
Only then can you compare.
Only then can you see the direction.

3. For science, there is a need to use both relative and absolute measurements.
Relative measurement tells us percentage.
For example: 32% or 49% of the leaf area is damaged.
Absolute measurement tells us the real area.
For example: how many cm² of the leaf are necrotic.
And this is a completely different level of data quality.

4. To get absolute measurements, calibration is not optional.
Calibration is must have.
Our standard practice in Petiole Pro is to use a calibration plate.
However, if you work with ImageJ, MATLAB or any other tool, you still need some calibration object present.
Without calibration, you can rely only on percentages & relative measurements.
You can say “this leaf is 40% damaged”.
But you cannot correctly say how many cm² are damaged.

5. Necrotic leaf damage must be clearly decided and explained.
This is the part where human knowledge is still very valuable.
Some lesions introduce discoloration, but they are not necrotic areas yet.
So, if we care about precision of terminology, we cannot call every yellow, pale or brownish area necrosis.
Before measuring necrosis, we need to define what exactly we call necrotic tissue.

6. RGB indices do not diagnose disease. They classify colour patterns.
This is important.
AGRI, NGRDI or similar colour-based indices do not “see” the biology directly.
They analyse the image colour signal.
Healthy green tissue usually behaves differently from yellow, brown, dark or dry tissue.
So the algorithm separates pixels according to colour characteristics.
This is powerful.
But it is not magic.

7. Different indices may give different necrosis estimates.
In our report, AGRI estimated a higher damaged area than NGRDI.
This is not a mistake by default.
It means that each index is sensitive to slightly different colour behaviour.
One method may be stricter.
Another may be more conservative.
That is why the combined result can be very useful: it reduces dependence on one single index.

8. The most practical result is not just a number, but a repeatable workflow.
What does this mean?
Same crop + lighting logic + calibration approach + timeframe.
And the SAME definition of necrosis.
Only then the numbers become useful.

What do you think?
#necrosis

Skip the Carousel

1 2 3 4 5 6 7 8 9 10

1 month ago | [YT] | 16

Dr. Maryna Kuzmenko

Old McDonald had a smart farm,
Ee i ee i oh!
And on that smart farm he had an AI Assistant,
Ee i ee i oh!


With a prompt-prompt here
And a prompt-prompt there
Here prompt,
There prompt,
Everywhere prompt-prompt


Old McDonald had a smart farm,
Ee i ee i oh!


Please, learn the lyrics, prepare yourself for an hour of AI-in-agriculture fun and see you on Saturday, 23 May, 8.00PM British Summer Time
👍🌱😄

1 month ago | [YT] | 2

Dr. Maryna Kuzmenko

Automatic assessment of herbivory leaf damage is finally moving from visual scoring to measurable data.

Yes, finally! 🥳



I lost count of how many times we were asked about this functionality in Petiole Pro.

In fact, if the hole is large enough, and if border or edge feeding is present, the current version of our mobile app can already count it.

But today I am celebrating a different milestone: automatic batch processing of leaves with herbivory damage.

It means not one perfect demo leaf.
It means many leaves, with different types of chewing damage — internal holes, edge feeding, and mixed damage patterns.



Why does this matter?


Because herbivory assessment is still often described in very human language:
“low damage”
“moderate damage”
“severe damage”



Useful? Yes.
Repeatable? Not always.



One person’s “low damage” may be another person’s “moderate damage”. And when we talk about research trials, breeding programmes, pest monitoring or crop protection experiments, this subjectivity becomes a real problem.


With our latest workflow, Petiole Pro can now automatically process damaged leaves and extract measurable traits:
1. Detect the total leaf area.
2. Detect the area of internal holes.
3. Exclude hole area from the measured leaf area — automated, with zero manual touch.
4. Calculate the percentage of eaten area versus total leaf area.
5. Save captured images showing what has actually been measured, together with a
CSV file of measurements — as usual.



So instead of saying:
“This leaf has moderate damage.”



We can say:
→ this leaf has X holes
→ the damaged area is X cm²
→ missing tissue represents X% of the leaf
→ damage distribution is mostly internal, edge-related, or mixed


And that difference matters.


For plant scientists, it means more repeatable phenotyping.
For breeders, it means clearer comparison between resistant and susceptible varieties.
For crop protection trials, it means better evidence of treatment effects.
For growers and agronomists, it means moving from “looks worse” to “measured change”.


Of course, not all herbivory damage is equally simple.



Internal holes are easier because the missing tissue is enclosed within the leaf boundary. Edge feeding is more difficult because the original leaf shape has been eaten away and needs to be estimated.



But this is exactly why I am excited about this step 🤩.


We are not just detecting pretty leaves.
We are starting to quantify messy, real biological damage.



And here we are again — with our love for leaves in all shapes and sizes, but with numbers 💚🌍🤝



#pestmanagement #entomology #herbivory #plantscience

Skip the Carousel

1 2 3 4 5 6 7 8 9 10

1 month ago | [YT] | 5

Dr. Maryna Kuzmenko

The lifetime of my YouTube channel in screenshots.
I registered the channel on the 5th of November 2020, but it took ages to understand what I should do on YouTube...
However, the actual start of my YouTube channel was provoked by a conversation.

Someone asked me in 2025:
“Maryna, are you on YouTube?”

I answered:
“No”

And simultaneously asked myself silently:
“But why?”

Since then, I haven’t been watching YouTube.
I’ve been creating it 🤗

I have some interesting plans ahead.
It will be my pleasure to share them with you soon 🤝

Skip the Carousel

1 2 3 4 5 6 7 8 9 10

1 month ago | [YT] | 7

Dr. Maryna Kuzmenko

Spring is in full swing, but I’m busy thinking about a real problem:
How to put an elephant into the room?
The elephant — as usual — is AI in agriculture 😉



And my room is just 15 minutes of speaking time at one of the most promising learning events in Turkey 🇹🇷



The Kırşehir Ahi Evran University 1st Artificial Intelligence Days in Mucur, Kırşehir, is also open to global participants who want to get insights into AI across different directions.


And agriculture is one of the priorities 👍

So now I have the following directions for using artificial intelligence in crop production, considering some of the top-priority crops in Turkey:



👉 AI vs. Tuta absoluta: How Computer Vision Is Changing the Fight Against Tomato Leafminer
📆 March 2026
🔗 lnkd.in/erxajJFq


👉 AI for Grains: What’s “hot” in 2026 for wheat kernel quality control?

📆 March 2026
🔗 lnkd.in/eHQ2TA9h


👉 AI for Sugar Beet: Digital Monitoring of Salt Stress Responses
📆 March 2026
🔗 lnkd.in/evQxnAW4


👉 Corn: Predicting Critical Nitrogen Dilution Curve
📆 April 2026
🔗 lnkd.in/eZi9syd5


👉 Hazelnuts (Cobnuts): Drone-Based Tree-Level Health Monitoring
📆 January 2026
🔗 lnkd.in/eHR8FKSX


👉 Onion: How Drones and Machine Learning Are Changing Yield Prediction
📆 April 2026
🔗 lnkd.in/en2nxnjd


👉 Citrus: Detection of Leaf and Fruit Diseases
📆 April 2025
🔗 lnkd.in/ezUA2pBC


👉 Grapes: Non-Invasive Estimation of Ripeness and Sensory Quality
📆 June 2025
🔗 lnkd.in/erCSGgCr


👉 Olives: Yield Prediction on Trees with Deep Learning
📆 May 2025
🔗 lnkd.in/em57vd8R


👉 Potato: Monitoring Storage Microclimates to Prevent Late Blight Losses
📆 January 2026
🔗 lnkd.in/eQRMY3HC


As you can see, the list is long.

But I have already discovered the most practical way to put this AI elephant into my 15-minute presentation on Wednesday, 6th May, between 8:00–10:00 BST👌

It is:
1. Open the Zoom link.
2. Put my AI-in-phenotyping elephant into the room.
3. Deliver the presentation, answer the questions.
4. Have a great time with the global audience of the 1st Artificial Intelligence Days 🌱
5. Say “thank you” to Ömer Ertuğrul for the opportunity 🙏


Are you with me?

Register here: lnkd.in/eV-HWw2v


Participation is free of charge, but full of benefits (including certificates) 🎓🌱📚

#aiinagriculture #turkey #onlineconference

Skip the Carousel

1 2 3 4 5 6 7 8 9

1 month ago | [YT] | 22

Dr. Maryna Kuzmenko

Just in case - if you wish to hear AI jokes about tractor, join me today :)
I cannot promise lots of fun (these AI agents are so unpredictable!) but as we've seen last time - some basic sense of humour is still present 🤣

#IAskedAI

1 month ago | [YT] | 2