 Let's get started with our second talk and learn about new technological advances in plant identification. And here to show us how they work and maybe how they don't work is Dr. Esther McGinnis. Esther is an extension horticulturist. She's the director of the NDSU Extension Master Gardener Program, and she's an associate professor within the Department of Plant Sciences. She and her graduate students conduct research on pollinator attraction, plants that can take cyclical saturation and drought, as well as native plants. And before coming to NDSU, Esther taught herbaceous plant identification to horticulture and landscape architecture students at the University of Minnesota. So Esther, welcome to the forums. All right, thank you very much. And we're gonna totally shift gears. You're from pollinator plants to using your cell phone and using those apps to identify plants. And in the title, you'll notice that it says round two. About five, six years ago, I did kind of a similar evaluation of cell phone apps to see just how accurate they were at identifying plants. And I have to tell you, I was not impressed at all that many years ago. In fact, I deleted the apps right off of my phone because I didn't think that they were very good. Well, we've got a new crop of cell phone apps that we are going to look at. But before we get there, I'm gonna tell you how things were in the old days. I guess I'm middle age, so I can say it's the old days. As Tom mentioned, I used to teach plant ID to horticulture students and to landscape architecture students. And it was a difficult class because in order to identify plants, we had to learn family characteristics, we had to learn characteristics of genius, of genre, and then even down to species. And that involved being very proficient in the vocabulary. So you had to be able to look at different characteristics when you dissected a flower or looked at the leaf and had to have the vocabulary to describe that. Here, I just have some of the vocabulary for describing different statements. So as you can see, there was a lot of terminology to learn. And can you imagine the terminology for describing leaves with all the different shapes and the serrations? I have to admit I flunked a few students. I mean, it was a lot to memorize. Now, fortunately, we do have some new cell phone apps out there and these are five of the more common apps and they all use image recognition. So a novice would be able to do this. I looked at five apps and I have to admit I just did this this past weekend. I downloaded five apps and for fair disclosure, I have an iPhone 8. So an ancient cell phone. But these all worked on my cell phone. And I'll go through the apps. iNaturalist, this was a joint venture between the California Academy of Sciences and the National and Nat Geo. We've got PlantNet, which is a citizen science project. And this is on a global basis. So it's cataloging plant diversity. So in addition to that, we have two apps that are not associated with science agencies or with National Geographic. We've got PlantSnap and PictureThis. And they're more like a commercial app. And PlantSnap was the only app that required me to create an account and even put a credit card on file once the free trial ended. There's PictureThis. And in addition to plant identification, it has an online plant encyclopedia. And then the fifth app that I evaluated was Seek. And Seek is supposed to be affiliated with iNaturalist. So how did I do this? Well, I tested this on 17 different plants. Considering this is wintertime, I had to use photos of those plants. So fortunately, I've got a huge photo library and was able to use that. I started off with kind of a softball, Monarda Punctata Spotted Bee Balm. And this is a native plant. It's actually quite a good pollinator plant. But I used this one because I had a good picture of it. And number two, it's very distinctive in its coloration. So you'll see that it has bracts underneath the flowers that are kind of in white, yellow, or white cream and pink. So I thought that would be an easy first one to try. And then I tried this with all five of the apps. So I used my cell phone, focused on the picture and then identified them. I gave points based on how well they did. If they got the genus right, I gave them half a point. If they got the specific epithet, or I'm gonna shorten that to species right, then I give them another half point. So for each correct identification, they could get a total of one point. Now, as to be expected, four of the five apps did very well on Spotted Bee Balm with it being so very distinctive. However, iNaturalist was out in left field. It thought it was cannabis sativa. So it thought it was marijuana. Now, next up, I wanted to do something that was a little bit harder. I chose Monarda Ditima Grand Parade. And this is a harder picture to use because instead of being able to see the leaves, all you can see are the flowers. And I wanted to see if that had an effect. Well, as you can see, plant net, plant snap and seek were all wrong. They didn't even get the genus right. However, iNaturalist, it said it wasn't confident in its identification, but it did give a suggestion. It suggested that monartifist gelosa was the name of this plant. And so this means it got the genus right, but not the species. So I gave it half a point for that. And then picture this, also thought this was monartifist gelosa. And so they earned half a point. Now, I have to admit, I'm not exactly disappointed in the app because this was a really tough picture to identify since the leaves were not showing. This is Aurelia cordata sun king. This is the perennial plant association plant of the year for 2020. And this is kind of a hard plant to identify, I thought for the app because it had yellow leaves and you didn't have flowers or interesting seed heads here that it could use to enhance its image recognition. Now, I'll have you notice that I covered up the sign that was there with a little text box so the app wouldn't be able to cheat. This was a tough one for the apps. I naturalist thought this was an elm tree. Plant net thought this was an angelica, plant snap thought this was a species of dogwood. Only picture this was able to identify it to genus and it received half a point for that. Now, going to more of the annual flowers that you would see this is Brawalia speciosa. This is kind of an uncommon bedding plant, but it's out there. I mean, it's a great bedding plant for shady situations and both plant net and picture this were able to identify it down to the species. All right, so I threw a little bit of a curve ball here. This is Baptisia American goldfinch. So I looked up the patent for this to determine the parentage on this. You'll notice that in the name, there is no specific epithet. And that's because this is a cross between species. The mother plant is a cross between Baptisia sphericarpa and Baptisia alba. And then the father plant, which provided the pollen for the cross is fully Baptisia sphericarpa. So three quarters of the genetic material would be that particular species. So before I started this, I determined that I would give a full point if the app was able to identify Baptisia sphericarpa. And we did have two. So starting to see a little bit of a trend here that plant net and picture this were able to identify it to the species level. And iNaturalist got down to a subfamily. So I knew it was in the bean and pea family, but it suggested a yellow sweet clover. So that this is definitely not yellow sweet clover folks. Next up is symbiotricum oblongifolium october skies. I chose this photo because my photo is not great. So I wanted to kind of simulate, you know, a picture from afar. There are over, there are about 100 species of symbiotricum. Symbiotricum encompasses some of the aster plants. So I wanted to use this photo because I figured there were people out there that would take a photo from a distance where you wouldn't have close-ups of the flower or close-ups of the leaves. And surprisingly, plant net and picture this did really well. They got it down to the species level. iNaturalist knew that this was an aster of some form, but plant net and seek didn't do so well. Interestingly, picture this had even more information. It listed some of the popular cultivars, including October skies. So that was, that was good to note that that particular app had cultivars, although it doesn't identify down to the cultivar level. So making a little more difficult, I wanted to try some sedges and some grasses. So I studied Kerak's radiota and Kerak's Pennsylvania when I was working on my master's degree. And these were tough plants to identify. They're about 175 different Kerak species in Minnesota. Minnesota seems to be the epicenter of diversity for Kerak species. So I was like, oh, none of the apps will get this one because I would spend a lot of time looking at the seeds. The seeds for Keraks are called perigenium. So I'd spend a lot of time looking at the seeds under microscopes, trying to get it down to species. Well, I'll tell you, plant net and picture this nailed it. All right, so we have grasses and this one is miscanthus purporescence, red flame grass, very popular here in North Dakota. And it's popular because of its fall coloration. This was a tough one. I wanted to include this because the photo is backlit by the sun. So it makes it a little bit harder to see. So I wanted to give a challenge here. Also it's experiencing some fall coloration. And then the third challenge is once again, this plant is a hybrid between two species. It's a cross between miscanthus sacrophlorus and miscanthus sinensis. So I predetermined before I tested the apps that I would give the apps a full point if they identified either of the parent species. So plant net, plant snap and picture this identified a miscanthus sinensis, I naturalist identified it or gave a suggestion that it was miscanthus sacrophlorus. So all four of those got points for this and then seek suggested that this was big blue stem which is erroneous. So this is probably the toughest picture that I threw at the app. This is prairie drop seed. We don't have any seed heads at this early stage. It's just a clump of grass. So I wanted to see what it could do at it. I honestly thought none of them would get it. I naturalist said it wasn't confident but it did suggest sprawolus heteroleppus which is prairie drop seed and then picture this identified it. So both of them got points. I wanted to try a shrub. This is staghorn sumac and but it's not like the species. This is a plant that's either been bred or it's been selected because it has dissected leaves which are not typical for the species. And you can see we got all sorts of answers for this. I naturalist said it wasn't confident and its top suggestion was this is common ragweed. Folks, this is not common ragweed. Plant net came up with a species I'd never heard of. Plant snap thought this was a lady fern. Only picture this identified this as ruse typhina. So you can see there's kind of a wide variety. Some of the apps are good with some things and some of the apps are better with others. How about an invasive? In fact, a noxious weed. So common buckthorn is a noxious weed in Minnesota and I know it's spreading in Minnesota. This is kind of a different type of picture because it was just a branch that was cut off and brought to an extension agent's office. Here we can see that it's kind of a messy sample. We've got rust on it. If you look closely at the stem you can see we've got soybean aphids on it. So I mean, this is kind of a nasty looking plant here but even so four of the five apps identified it as common buckthorn. What about for berries? Extension offices get loads and loads of questions asking for plant ID of shrubs that have berries because people wanna know can I make jam with this or is this gonna kill me? Is this poisonous? So we get lots of questions like that. So I happen to have photos of three different berry plants on my photo library. So here I'm using Hascap which is La Nisra cerulea and once again, we've got three of the apps identified it to species, plant snaps at La Nisra Japonica and Seek only had the genus. So pretty good there. I wanted to do a native berry. This is Saskatoon service berry. This is native to North Dakota and absolutely beloved by various Native American tribes throughout North Dakota and boy oh boy, does it make a good pie. So this was probably a little bit too easy. All five of the apps did in fact identify this to species but however, notice this is a good picture. We've got a nice photo of the leaves so the app can see the leaf shape and we've got some very distinctive berries that are showing up here. Here you can see that, here let me get my laser pointer. You can see that it's got this distinctive end where the sepals used to be suggesting it's in the Rose A.C.A family. All right, since that previous photo was too easy, I thought I would go with black chokeberry having immature berries that were green and hard. So this knocked back the success of it, plant net and picture this, we're able, so plant net, picture this and seek, we're the only three that were able to identify black chokeberry. Sorry, one final one before we tabulate the results. This is mosaic plant and this really is way out in left field. This is a plant native to Brazil, it's an aquatic plant. So can it identify an aquatic plant? And furthermore, this is a plant that will only live in water that is 72 degrees or higher. So definitely kind of a unique plant. Well, we were able to get two of the apps to identify it to species. The inaturalist came in and then you'll notice that I've had all these asterisks at the bottom. So I looked at inaturalist top suggestions and when the top suggestions were right, I gave them points and plant net also identified it down to species. Now picture this, which was doing so well, thought this was a water chestnut. Now there's some similarity to water chestnut, but still it's wrong. All right, so let's tally this up. I showed you 15 of the trials, I actually had 17 of them and picture this came in first. They got 14 out of 17 right or 82%. Coming in a close second was plant net had 13 out of 17 right, 76.5%. The remaining apps had fewer than 50% of their answers right and plant snap had fewer than one quarter right. All right, so some tips from my weekend experiment here. Accuracy of the identification depends upon the quality of the photo. So make sure that when you're doing this that you are taking a good photo. Furthermore, it's in good lighting. That seems to really help it a lot. One thing I didn't try here is maybe take more than one photo. So take a picture of the whole plant, maybe do a close-up of the leaf, a close-up of the flower, close-up if they have a seed head on it. And then compare the results there. You might actually find you get a little bit more accurate result from that effect. Considering that none of the apps were 100% accurate, I think it's helpful to have more than one app on your cell phone to do a cross-check. Okay, so maybe you can get into the ballpark and then see if another app agrees. However, don't rely on the app to determine if a plant has edible berries. I mean, considering that we're not at 100% accuracy here, I would hate for somebody to have to go to the hospital and have their stomach pumped because they ate something that was toxic. So in that situation, we would suggest consulting with an expert or maybe consulting a suite of characteristics. Say for example, you actually cut into the berry and see how many seeds are there. Are there sepals on the bottom of the fruit or how is the fruit attached to the plant? So it's good to actually do a further confirmation using a suite of characteristics that we know is true of the plant that you hope that it is. So other observations I made seek was hard to use for me, but I'm wondering if that had to do with the fact that my cell phone is ancient, you know, an iPhone 8. I think they've got iPhone 14s that are out now. Now I have to admit I've got an old phone because I'm at the stage where I'm saving up for money to send my kid to college. So maybe some of you that your kids are already through college and you've got a nicer cell phone, maybe seek will work better for you. I know seek is actually built into the iPhone picture app on some of the newer phones. And then finally, one thing I noticed is that picture this has 360 identify. So I think this means that you can use picture this to try and get more of a panoramic effect or to try and actually circle around the plant. So I'm anxious to try that on a plant this summer and see if in fact that improves the accuracy of that app. So just a couple more slides here before I take questions. If you want to verify whether your iPhone or whether your cell phone apps are accurate, you know, you can certainly go to websites. I'm just gonna show you my favorite website. It's Minnesota Wildflowers. This is best if you're trying to identify native plants, wildflowers, plants you find in the ditch, you know, or weeds. So it's really good. If you have, if you've got it down to a genus or a species on your cell phone, you can just pop it into the plant name box, which is up here and see if it matches. So Minnesota Wildflowers has photos. If you click on the photos, you'll get them to blow up and you'll see if they've got the same characteristics as what showed up on your cell phone. Furthermore, they have a nice search function. So if you're not a fan of using a cell phone app but would rather do something like this, you could type in characteristics. Okay, you're looking at a wildflower. What's the color? Maybe you can identify the flower shape, maybe put in the leaf attachment. Are the leaves attached where the leaves are opposite from one another or do the leaves alternating down the stem? You know, so put in some of these characteristics, put in the bloom season and see if you can come up with the right plant. So hopefully here, I've given you some tips here that you can start identifying plants without having to take a full semester long class from me and without having to learn a bunch of vocabulary. It seems like things have really improved over the last five years. Okay, that's good. And I agree that Minnesota Wildflower site is great. It's very thorough and it's from Minnesota so we have very similar plants. Okay, we invite people's questions here. You know, Esther, you mentioned that one was very hard use. Picture this, was that the hard one? No, picture this was the most accurate. Seek was the hard one. It just had a hard time focusing. It was glitchy. I had to shut down my phone. But I don't know, I can't say for certain that was my experience but maybe some of you have had a better experience with it. Which one had the best experience? Which one was the most user friendly? Picture this, it was just so easy. It was just point and photograph. Now keep in mind, I was photographing pictures that were on my computer screen is what I was doing. And it was just easy and just popped up. I didn't have to do a lot. So I like that. I really liked it. I didn't have to get too professional with it. And that's a free one. Totally free. So that looks promising. And the one plant net costed money, right? Plant net? No, plant net. Which one costed money? It was plant snap, I think. I think it was plant snap. And it didn't look like it's a good deal here. That one didn't look like a good deal. I'm going to cancel my subscription before it kicks in. Oh, you get a trial version. I got a trial version for seven days. I'm going to cancel that one. Smart. People are commenting about Google Lens. And this person used Google Lens for 10 plants, and they got 80% accuracy. Have you ever heard of Google Lens or tested that? I haven't tested it for plants. I think Google has some feature. I know we were using it for my mother-in-law's China and dishes that we're going to be selling at a garage sale. So I think that there are some good things that are going on through Google. But I didn't have a chance to test that. OK, this person did the paid version of picture this for two years, and they love it. The only thing they've got an issue with is telling the difference between a philodendron and apothos. How about that? Details, details. That's right. Another person has comments comparing or discussing plant snap and picture. Your research mirrors their experience. And so it's not just your iPhone 8. It goes for other phones too. How about have you ever thought about doing or using bug identification apps? Or maybe you and Jan should work on one of those. Well, iNaturalist does that. OK, all right. And in fact, we're doing, I think, a bubble bee count through iNaturalist, which is kind of a citizen science opportunity for our master gardeners, or actually for anybody. OK, you got another comment here that picture this will only work free for a while, and then it costs $29.99. And they say picture this is the best app for identification. Michigan State University has researched the apps. And picture this has won four years in a row. So there you go. I'm not sure I paid. That's the only app that I had. I had heard about picture this last fall and downloaded it. And I know I don't think I've ever had to pay for it. Oh, how long have you had it? So that's the only app that I had prior to this weekend. I got that when I think I downloaded it last August or September, because that's when I heard about it. OK. And I'm cheap. I'm cheap. So I'm pretty sure I didn't pay for it. Um, there's also an app called the Minnesota Wildflower Search. And it looks just like their website. And what else we got here? Well, what do you think about that? The first one, the B-Ball, I thought was interesting that they thought was marijuana. Was that that California-based website? That's. It was the California-based website. That makes a lot of sense. I have to admit, I did test all the plants on hemp. So Harleen's doing research on hemp, so I have a photo on hemp. And every one of them was able to identify it. That's got to be the most popular one. It's got to be. A popular, submitted photo, I would guess. OK, what else we got here? This person just tried to download, picture this. And there is a premium plant. Yep. The cost. And then another comment says, if the user goes to the upper left-hand corner and hits Restore, then you can still use the app without paying. And this person, this agent's been doing it for years. Another cheap person. I like it. Does anybody have any other comments or questions about these apps? OK, I have to say, I'm very surprised how accurate they are. Don't you have the general, you've got to be pleased? I mean, if you're right 80% of the time and you were not an easy tester there, you give hard tests, I got to say. Now, why don't you flunk some people? Man, you're not an easy-going teacher. You're pretty tough on these people. And water plants from Brazil and crazy grasses from Minnesota. Man, there's everything, no plants with no leaves on them. Wow. And you still get 82%, not bad. That's not bad. I'm afraid we're going to be out of a job someday. I know, I didn't want to say that. But no, they may be old identification, but they can't give that special comment or that special expertise that only a person can provide. How about you think they could tell the difference between poison ivy and poison oak? Sure. Yeah, I think so. Can you use these plant apps for weeds to identify a weed? Oh, yes. Absolutely. That's beautiful. Pretty much, if you can tell broadly from a grass, you're halfway there, I think with weed control. OK, that's all good comments. Wow, we learned about some modern technology that was useful. That's very good. So thank you, Esther, for that. Oh, you're welcome. Thank you. OK, we're going to take about a five-minute break now. But let's also, Jan had it. I wanted to answer a question about going back to the monarchs. And somebody had a question about, like you mentioned, sooty black mold. And so I don't know if Jan, if you're still there, or you left. You were here till just now. Yeah, she left. It looks like. Well, maybe you can answer this, Esther, since you're better than a plant identification app. I'd rather trust you than that app. But this person, like they mentioned about sooty mold and aphids causing that on the weeds here, on the velvet weed. No, milkweed. Milkweed, sorry, man. We had a lot of velvet weed in my leaf in my life. Is it coutrelle? OK, so if this person uses soapy water to kill the aphids, will that soapy water also harm the caterpillars and the eggs of the monarchs? That's a great question. So we use insecticidal soap to wash off the cuticle unsophabodyed insects, such as aphids and such. Now, if you're noticing that there are no caterpillars around, you're OK. Because this is one of those things where it has to make contact with the insect. So when you spray it, make sure that you're not spraying it on the caterpillars. But they're going to be a little bit tough. As they get older, they're going to be much tougher than an aphid. So that aphid is really a soft bodied insect. Now, I would assume that the caterpillars at the very beginning might be a little sensitive to it. So I would be very careful and check first before you use insecticidal soap. But once they're bigger, just avoid them. Just try and hit the aphids instead. And I mean, there are milkweed aphids out there. And I think they're bright red. They're kind of an interesting looking thing. So pretty easy to spot. With soapy water, just target the aphids. And soapy water doesn't have any residual to worry about. So just target the aphids. You'll be OK. Sounds good. OK, Esther. Time to go home so you can prepare for the blizzard. There you go. You get to have dinner. There you go. Thank you, Esther. Thank you, Tom.