cheap bathroom basins

good morning everybody. this is don major, fire and landscape ecologist with blmboise idaho. i just want to kind of um welcome everybody the cheatgrass dieoff in the great basin webinar where we're going to be talking about some of the work thats been done by um usgs, eros lab, stinger ghaffarian technologies, and the um rocky mountain research shrub science lab.i just wanted to give everyone a little bit of background.

about a year and a half ago um it kind of came to our attention this issue of cheatgrass dieoff occurring in portions of nevada and then also in utah. so the great basin restoration initiative alongwith the great basin research management partnership and the landscape conservation cooperative came together and tried to identify a way to explore this issue of dieoff kind of in an integrated fashion. so oneof the things we did

to start everything off was to um employ the knowledge and expertiseof the eros lab and trying to help us betteridentify where dieoff may be occurring across theportions of the great basin. we followed that with a more directed integrated science group where we brought together researchers totry to uh address aspects of causal mechanisms that may be responsible for the dieoff. so um what we're going to do today is kind of run-through

presentations by stephen boyte looking atthe spatial extent uh question and i don't have the actual, where's the title, well you'll see it in just a minute, uh but we're gonnafollow that with about a 10-minute um question period and uh then that will conclude and thenwe will move on to susan meyer talking about some of the causal mechanisms. so with that i think if we can turn it over to stephen. << anddon if you don't mind if interupt for a second. << sure. << if anyone has any questions, if any of the audience has any questions um you are welcome to type your questions into the question

pane in your control panel on the top right of your screen at any point during the webinar and ican, i can um type responses to you. but if they're questions for the presenter then as don said we'll, we'll take those after the presentation. << excellent. << okay so i will make steve the presenter now. << okay can you guys still hear me? << yes! << alright good deal. we're ready to go. um i will start over. alright there we go.

um my name steve boyte and um i am with stinger ghaffarian technologies and contracted through the eros data center. with me here today is bruce wylie. he is a co-author as is don major with the great basin restoration initiative out of boise, idaho. we're going to talk about mapping our inter-annual cheatgrass production and dieoff using remote sensing and ecological models. so i'll go ahead and get started. this is a map of our study area. it um is in the, in the northern nevada and

southeastern oregon, southwestern idaho area. our objectives over this past year have been to track cheatgrass abundance and extent spatiallyand temporally. we wanted to also identify cheatgrass dieoff areas and then we're going to develop a dieoff probability map. okay, um the remote sensing product that we'reusing is emodis ndvi expedited moderate resolution imagingspectroradiometer at 250 meters. this is a 7-day composite where they take

a picture of, or an image of the earth, the coterminous us i should say, each day for a week and then theytake the best pixel out of those seven days and then they composite it and make those available in weekly time steps. now ndvi is a normalized different vegetation index where they take the nir infrared band and subtract out the red infra, or the red band and then they divide that by the nir infrared band plus the red band. so that gives us a vegetaion index where we canmeasure

um or use it as a proxy for vegetation dynamics, map biomass, or even use it as a proxy for net primary production. now the one thing about cheatgrass is it's spring phenology is earlier than most other vegetation types in the greatbasin, therefore it produces a profile that's profiledistinguishable from these other vegetation types. and thisgraph here that we're showing um is, is showing two pixels, theprofile of two pixels, in close proximity. they'reprofiles

during early spring and for 2000, 2001 and 2002. now the red line shows a area that is dominated by big sagebrush and showing its profile, and then is being contrasted with the profile that is an area that's dominated bycheatgrass. and as you can see there's a, a definite difference between these profiles in early spring. the rest of the year they kind of follow along um the same pattern, but in the spring the cheatgrass

profile is significantly higher. it shows higher values than the big sagebrush does. now what we did was we selected a cheatgrass growing season period for spring, in this case it was from the middle of march to the middle of may, and we useda period where cheatgrass would have senescenced. in this case it was early to late june into mid july and so we then created an index that contrasted the spring and summer difference, spectral differences. now this image that i'm showing here is a picture of the spring period image. okay. the bright areas are areas

where vegetation growth is higher and darker areas are where vegetation growth is lower. now if you look at the summer image, you can see kind of a reverse. the areasthat were brighter in the spring are typically darker during the summerimage, and the area's that were dark in the summer are now brighter in the summer, or dark in the spring are now brighter in the summer image. and when we take and generate our index, where we take the spring image and subtract out the

summer image, and divide by the spring image plus the summer image, we get our cheatgrass index now again, in this cheatgrass index image, the brighter areas show areas where wethink there's cheatgrass, where there should be cheatgrass growing in the spring, and the darker areas are areas where cheatgrass would not be. so we put that data into a regressiontree model, and i'm gonna go through this quickly, because it can take a lot of time if i don't.

but we have our satellite inputs andthen we have our cheatgrass index, and we use the growing season ndvi or our spring image index for 2001 and2006, because what we've done is we've trainedour model on eric peterson data sets from 2001 and 2006. where peterson said there, cheatgrass existed we trained our model using 9,000, or about 9,000 training points. and so we took those data sets from 2001 and 2006

and we took the spring image from those two years, and we took the summer image from those two years, and we took the cheatgrass index image from those two years, and we added to them some site specific variables like ssurgo, or the state soil geographicsurvey data from the available water capacity, and the elevation data set, and a wetness index, along with nrcs's major land resourse area. we take these data sets, we put them through a rule-based

piecewise regression model and that generates a model output. and then we take that model output and we start adding annual growing season ndvi from 2000-2010, the summer image from 2000-2010, and the annual cheatgrass index that we generated from 2000-2010. we added those to those same four site-specific variables,and we put it into a mapping application. and this mapping application then uses all of theseraster datasets

to generate, to extrapolate um the data from our training points to theentire study area, and we're able to generate um maps for each year, through our entire study area. this is themodel detail. its a strong model. the training data on 8953 cases. our r^2 = 0.77 and the test data data the r^2 = 0.71. so we're happy with strength of the model. basically what this

table is showing you is how each dataset is used in the development of the algorithms, that we're used to, in the regressionequations as well as how frequently each dataset was used inthe stratification of the rules. now what that does is, like i said, it generates maps from 2000 through 2010, cheatgrass maps. these are cheatgrass extent and abundance maps. areas that are black are areas where

we did not estimate that there was, or themodel did not estimate that there was cheatgrass production for that particular year. now what we did when we have this time series of cheatgrass production, we can compare it to things like precipitation, and that's what this chart is doing here. this black line is our mean cheatgrass production percent through our time series.

the dashed gray line is our precipitation for each year that we got from octoberthrough may, and then we have the average cheatgrassproduction for the entire study area, for the11 years, as well as the precipitation average for our study area. as you can see as cheatgrass or as precipitationincreases here cheatgrass increases as well. when precipitation drops cheatgrass drops. again in 2003, from 2002-2003, precipitation increased and

or precipitation increased and cheatgrass increased and this pattern goes up until 2005. in 2005 precipitation hit an all, a high here. in fact it was a hundredmillimeters higher than the average but cheatgrass production declined slightly and it's a very average, at a very average um percent. and in 2006 precipitation declines but, but its still higher, significantly higher, thanthe average and we see a slight increase in themean cover

of cheatgrass in this area. finally in 2007 cheatgrass starts to go up and that is also at the time whenprecipitation starts to drop. so what ishappening here in between 2005 and 2007? um when we look at our maps uh later on you'll see that 2005, this iswhen we're estimating that there is significant dieoff this area, significant dieoff of cheatgrass in this

area. i want to show you another chart where again we are, are um paring cheatgrass production and precipitation. this is a scatter plot and when we look at all data points, and compare all data points, there's reallyno relationship at all. in fact our r^2 = 0.01. so um when we compare all years norelationship. so when we look a little deeper, cut up the data a little bit, we findcohorts

that seem to make sense and match, andthis is 2000-2003 data. and when we compare the cheatgrass production to um precipitation, and then we add 2008 and 2009 data points to these data points, we have an r^2 of 0.82. so we have a strong positiverelationship here. however, when we look at 2005-2007, there's also a very strongrelationship, r^2 = 0.98. however, thisis a negative relationship was

completely unexpected. we wouldn't expect that precipitation would be higher than cheatgrass production and there would be a negativerelationship, because we know that in most years cheatgrass production is highly dependent on the amount of precipitation that um, that it receives. 2004 represented by this circle here and 2010 are both anomalous years. when we add those datasets to the othercohorts, they significantly drop the r^2's. so therelationship

gets weaker. now when we compare cheatgrass production with elevation, we get this chart here, and i think thatthere's a lot of data here so i'm gonna be short and just focus on some of the, thehighlights. and um one highlight is that at the lower elevations um cheatgrass production is at its highest. and um this top-line, this

dashed black line, is showing 2001 data, our mean cheatgrass percent coverage for 2001. that was our, our most um productive year for cheatgrass. the dotted black line shows cheatgrass production um in 2009, which is the lowest productive year that we had for cheatgrass. and this solid black line is the mean for all eleven years in our dataset. even though

this is the, the highest percent cheatgrass that falls into this elevation category of less than 1137 meters, only about 1% of the land area um as you can see from this y-axis. only, only about 1% actually falls in this land area. um right here between the 1137 and 1637 meters uh elevation cohorts or classes, this is where um a lot of

the land area falls and is where most of the cheatgrasses will lie. and there's not much difference in any of these uh mean, mean cheatgrass percent cover for that area. moving on to comparing cheatgrass production and aspect, um this chart shows that there's not agreat difference in um the mean cheatgrass percent cover, um

regardless of which aspect we're looking at, or even if the there is no realaspect, or no real slope. and um so pretty much everything ranges from about5.5-6.5% here. now when we look at cheatgrassproduction and percent slope, um most of this area has a slope of 0-10%. in fact, almost 80% of the land area falls into that um

class and it's also the area that has thehighest percent of cheatgrass. in fact the more um steep the slope, the less, the less percent cheatgrass that we are going to find. now when we do develop these cheatgrass time series productivity maps,what we can do is we can compare them and create difference maps. so what we've done, is we've taken the maximum value for each pixel for all years, and we made a map, and then wecompared

each individual year to that maximumvalue map. and here i've displayed 3 of those maps. on the left hand side, the far left, we have our 2001 difference map, this is a 2006 difference map in the middle, and then 2010 on the right. and sothe areas that are consistently black in all of the maps, are areas where we'd expect zero cheatgrass percentage or a model estimated cheatgrass, zero cheatgrass percent cover. and so when you take zero,

subtact zero, you get zero. now also in um the other areas that are black, ones that are variable. that is, for example, in 2001, these are areas where 2001 would have had the highest um mean cheatgrass percent value for each pixel that is black. um when we go over to the 2006 you start seeing more red areas and these red areas indicate that, or even in this

area over here there's some, some um fushia and purple colors, these are indicating that there are,there's less cheatgrass productivity than there was in our maximum here, which is in many cases going to be 2001. and in 2010 we see even more um red areas. so the productivity, the cheatgrassproductivity in 2010 was lower than what we see in 2006 and it's significantly lower than what we see in 2001. however, the one thing that we thinkwe can do is we can improve on this.

this is a difference map, it doesn't take intoaccout weather effects and we know that weatherdrives vegetation growth, in particular precipitation, with cheatgrass. so what we're going to do is i'm gonna show you where we've improved, what we think we've improved on these difference maps, so the data is more meaningful. we're doing that by developing production, cheatgrass production maps have been normalized for weather, for annual weather, and then we're comparing it to ourproduction maps that have not

been normalized for annual weather. and this allows us to highlight ecosystem performance anomalies, and these performance anomalies, whenthey're underperforming anomalies, they're going to be areas that we think are cheatgrass dieoffs. so again, another model schematic, this side of the model i've already showed you so we're not going to worry about that, but this side, side of the model is where we normalize for the weather. in fact, we have prism weather inputs, we've taken precipitation, temperature minimums, temperature maximums,

and we've aggregated those for these 5 seasons. these for each year, 2000-2010, are gonna be put into ourrule based piecewise regression model as an independent, asindependent variables. we also have a site potential, which is what we're calling our long-term cheatgrass um percent cover average. and this is wherewe took every pixel for every year, and we calculated the median from our ecosystem

performance, or our cheatgrass extent maps, that were generated over here. sowe take that pixel median and for each year, and thenwe calculate the pixel mean above that pixel median. that becomes another dependent variable in our, in our regression tree model. we run that through our regression tree model, that generates our modeled or our expectedecosystem performance. so this right down here is our nor, our, our, our model that's been normalized for annual weather. we compare it to the model, or the

cheatgrass extent that were not normalized for annual weather and we compare them. and we generate ecosystem performance anomalies um at the 90%confidence interval level. and again, we generate 11 years of these ecosystem performance or cheatgrass dieoff maps. now this is a scattergram, or scatterplot, that shows um the data points for all eleven years. and so we have,

represented by the green diamonds, wehave overperforming pixels, the red diamonds are representingunderperforming pixels, and the tan diamonds are representingnormal performing pixels at the 90% confidence interval. so what would cause a pixel to underperform when we, when we normalize it for the weather? andthe one thing is is the disturbance will cause it to underperform. so in this case, with cheatgrass it could

be something like, um a fire could have burned through the year before. or um like we suspect with the cheatgrass dieoff, there's a a pathogen that is attacking the cheatgrass and killing of the cheatgrass. and so it's causing it to underperform. now one thing i want to again reiterate is that this is weatherbased, that as these pixels move up and down this regression line, they're doing so based on um, on the weather input. and when we look at

2 pixels, one that's over, overperforming here and one that is underperforming here, these theoretically could be the very same pixel just from different years. so if this is pixel 1 from 2003 and this is pixel 1 from 2005, you know even though this pixel is is shown as over-performing by our model it has a lower expected performance score than this pixel that underperforming two years later. it's

got a higher actual score than this one does, but the model expects um better performance ourt of this one, than itdoes ourt of this one. and so that's why we have the under-performance here and the, or the over-performance in this pixel and the under-performance in this pixel. and again, the anomaly threshhold is gonna vary with our weather conditions. so what happens when we get a year where precipitation isvery high, like

2005, significantly higher than the average, but are cheatgrass percent cover is less than what we wouldexpect. okay, so in 2005 the cheatgrass mean was, our percent mean was 6.5%, when over the 11 years it was 6.25%. but yet the precipitation was 100millimetres higher than the average for our 11 years. what does our map look like then? it looks

like this. this is our 2005 dieoff map or our ecosystem performance anomaly map, which shows a lot of under-performance. the red indicates dieoff areas and again thismodel is normalized for weather. in fact, 26% of the area is, our model estimates, is dieoff orunder-performance. this made us pause because this map is much reder, shows a significantly great, greater um amount of dieoff than

any of the other maps that we generated. i was reading um one of peterson's publications, one of eric peterson's publications, and he talked about, in thatpublication, where in 2005 he had field teams out in the field and they revisited 30 areas that they had visited in 2002 and/or 2003. and they found in 2005 that in those 30 areas there was an 8.5% reduction inmean cover of annual

grasses, where most annual grasses are cheatgrass. they found that 8.5% reduction in 2005 cheatgrass, or annual grass, than there in 2002 and/or 2003. so we felt better about this. we'd gone back in and checked our data and made sure we didn't do something wrong and we're comfortable that we didn't. so again uh we're pointing to a 2005 map that looks likes something significant uh was going on and and in regards to cheatgrass dieoff. so what happens in

2006 and 2007 with those dieoff maps? we have 4% dieoff in 2006 and 3% dieoff in 2007. what it tells me is that there is, thatcheatgrass dieoff is highly variable, both spatially and temporally, much like cheatgrass productivity is. one of the things we wanted to do is some validation with our dieoff maps. and so the bureau of land management,fortunately for us, had gone out in 2010 and they had flown areas

in 2010, that they had recognized on theground that were dieoff areas. and they got up ina helicopter, it flew over this area and they created these dieoff polygons for us. we over-laid them onto our 2009 um dieoff map and they matched almost 60% of the time. 60% of the time the pixels inside the polygons were um red, representing that dieoff. many of them were a dark brown well

and that is getting very close to a dieoff. now considering the fact that yes we have some modeling erring considering the factthat, you know, when you're looking at a peice of grass it could as small as 4-6 inches high and you're looking at that from a helicopter you know there's going to be most likely some error in the mapping. i think that we're very happy with with um, we're very happy with our, with our results here. the final thing thati want to talk about is a dieoff probability map. now we havedeveloped a dieoff probability map where we took our 11 year time series of dieoff

maps and um used again some attributes here that areshown in this graph or in this table here. and a quick aside here is we havetemperature maximum, long-term temperature maximum, we have long-term precipitation data, and we havelong-term temperature minimum data that we were, that we downloaded and processed from prism. the one thing that we can do

is we can take and substitute these threedatasets with future climate data from 2020, 2030,all the way up to 2100, and we can input those future climate datasets in here and we can them project out where, where dieoff is likely to occur under different climate cenarios. but back to this probability map and um what we have is we're we'rewe're happy because if you look down here at our training data, more than 9,000

training points, we have an error, training error of 12.12% so um our accuracy here is about 88%. in this case with our test data our accuracy is about 86% and when we lookover here at our confusing, confusion matrix we haveuser um accuracies of 88% and 85% in our predicted class or our training data, and 87% and 61% for our test data. now the 61%, there's only 28 data points so

um the low score isn't um too terribly concerning. now our overall acccuracy for our training data um was high as well at 88%. so i want to move on now to the dieoff probability map. um the the white areas are areas where we mask these areas out because they are not shrublands or grasslands. the black areas are where we would, our model estimate that there's zero probability

of cheatgrass dieoff, and then we work our way from cooler colors up to the warmer colors, going from low to high probability of cheatgrass dieoff. up here in the owyhee uplands, in these two areas in the north and northeast sections um is where we see the highest probability uh of cheatgrass dieoff, along with this section right down here just south of interstate 80 here. winnemucca is

here, locate you, and elko, nevada is here, and this is us route 950 leading north out of winnemucca. now i zoomed in on this map here um so that we could look atthose dieoff polygons that the blm generated and gave us to use. and as you can see here those bluepolygons, much of the area inside of those blue polygons are the warmer colors. so it looks like um those are higher incidences or higher probability of dieoff occurring.

there's certainly areas outside of those polygons that show high probability of dieoff occurring but we're happy again with um the amount of the warmer colors that are existing in the dieoff polygons. and to finish up, i want to talk about our next steps. um we want to or we're working right now on completing the development of cheatgrass production and dieoff datasets for the northern greatbasin. um they'll look very much like the the datasets that i just showed you.

we're also, as i alluded to, we're going toapply future climate data to the dieoff probability model, so that we can predict probability of future dieoff areas. we're also going to compare the time series of cheatgrass production maps with the monitoring and trends burn severity fire data to see if we can see any correlations between high cheatgrass areas on our production maps with um fires and then what happens after um a fire occurs. how does cheatgrass um

come back in those areas and then we're also finally going to compare dieoff areas to um topographic and edaphic datasets. so we can try to pinpoint where um dieoff are going to occur based on those those two datasets or two type of datasets. and that's all i have so if you have questions now um i guess it's the time to ask them. << so again if you have questions please type them into your questions box and i'll field them to steve.

<< ask here if she can still hear you. << can you still hear me genie? << uh huh, yes. << alright. << mike pellant asks, how do you account forlivestock removal of cheatgrass? << how do we account for livestock? how do, how am i supposed to answer this? can he still hear me? << yeah, uh huh. << okay um well if livestock removes cheatgrass um that would, could fall into a disturbance category kind of like fire or a pathogen. if cheatgrass

is is grazed heavily um compared toprevious years, um then it would show up as a disturbanceand it would be mapped as one of the under performing areas. << uh this is bruce. also you know that that cheatgrass growing period is prettyshort uh and so it's, i don't know, mike would know betterthan i i guess but you know it's hard to get a lot ofhigh-intensity grazing uh in a short period of time, uh but i guess mike would be more moreknowledgable about

than i am on that. but it seems like it would be hard to get very high intensity grazing during a short uh period of growing when you're stocking rates are usually pretty light anyway. a lot of uh hectares per aum. << okay, thanks. that's the only, oh we have another one. dennis davis asks, hold on a second i lost it, here we go. do you have any preliminary data toindicate the causal factors climate,

disease, other? << how the die-off, um as far as disease i think that's um, that's what what the presentation next isgoing to be talking about. that susan meyer will be talking about um a pathogen. as far as climate, we don'thave um, oh we're normalizing our datasets for for weather, not climate. um but we're still in the process of analyzingthis data and like i said

one of the future plans we have, in the very near future, is to compare our dieoff areas with topographic and edaphic datasets. so hopefully that will give us some idea of of where it's occurring and go ahead bruce. << certainly the probability of dieoff map relied heavily on climate. << yes. << yeah so we have identified certain climatic conditions where, whichseem to be vulnerable for cheatgrass dieoffs. << right. << and do you want to elaborate on whatthat was?

what makes cheatgrass vulnerable? << um we it's we it's a really complicated model at this point. << okay, that's fine. << that what climate variables were important, uh but we really can't tell you what the thresholds were. we would have to diginto these uh uh decision tree model and see what thresholds were used to separate dieoffs off.

<< okay, thank you. << you bet. << um dan, i hope i pronounce your last name right, it's either manier or maniay asks, an early graphic indicatedstatistical out liars in 2004 and 2010, i think these were theyears, could you revisit the environmentalconditions, ie of the predictor variables during those years? are they real outliars ecologically or does exclusion simplymake the regressions better? well

exclusions certainly made the regressionbetter. they they um what we were looking for was, no no here let me see. i'm gonna go to that and see if we can't, well you probably, they may not be able to see it. << yeah we can, we can still see your screen. << oh, you can? okay then let me go....<< i mean you're presentation isn't up yet. << yeah. let me go ahead and bring this presentation up again. i i think he's talking about this slide here. um

certainly in 2010 you can there was um, there was high, higher than average precipitation andlower, significantly lower than average meancheatgrass cover. okay, so that's kind of a an anomaly in and of itself. we would expect that with the precipitation that we're seeing, we would expect that more cheatgrass production. in 2004 um i think that that's not nearly as

as surprising as 2010. um certainly 2000 really is is an unusual year as well and one of the things that, you know, as i look at this data i think is is this another another trend where we're moving into an area where things don't make sense, kind of like 2005, 2006, and 2007. um i know last year in in the in thiswinnemucca um area, there was a lot ofprecipitation and there was a lot of cheatgrass growth. however, um until we get that data and we're able to normalize our cheatgrass production for the annual weather that occurred last year,

we won't really know if if those 2011 data um make sense, or if it's under producing oreven over producing. so um yeah the 2010 data, the predictorvariables, um would indicate that there shouldhave been higher cheatgrass production. << okay, thank you. susan meyer asks, what would an over performing area mean? << we're thinking on that one here. <

you know, theoretically speaking could itbe an area that maybe some cattle had got in there earlyin the season and grazed earlier um in a few years and and in a particular year maybe there was no grazing that occurred. um that could that could generate over performance. << another thing is previous years seed stock may have been very high and so you may have a lot of available seed source for very productiveyear uh on following year. the other thingthat

steve has found is that sometimes uh cheatgrass will germinate in the fall and go throughthe winter as a basal rosette. and then in in the next spring it comes onlike gangbusters. so it may be that our model is not quite capturing all of that previous fall germination event and then theexponential growth that occurs in the spring related to that. so that may be a partially explanation. does that sound okay? << yeah that does and the other thing that could be happening is is simplythe timing

of um precipitation may be optimal in those years where we're getting over performance or in those areas where we're getting over performance. so while the while the totals may be um similar to other years, the timing uh is optimal. << okay, thank you. um dan manier who asked the previous question about outliers um asks, so do other variables help explain these differences/anomalies? oh he,

never mind he said you can ignore my follow up question. okay we'll move on. so brian watts asks many of these areas have had successful re-vegetation treatments within the 2001-2010 time frame, was this accounted for as a possiblereason for change in cheatgrass? <

certainly validate our models. and yes um if those datasets, if somebody has those datasets, um that would be a a nice thing for us to be able to get our hands on. << okay and um.... << this is don. just to kind of add to that, um one opportunity is that with the information that we have available that they've presented, um being able to kind of step it down with some of that local field knowledge will help better inform

um where are areas that likely are associated with some of the past re-veg work. where as also outside those areas or in other location where you have you know, might be able to use local knowledge to better filter things out and focus attention, management attention on those other areas. << okay and brian watts adds, in other words could the perceiveddieoff be attributed to rehab success and not a natural dieoff? << right. << yes, yes it could. << okay.

matt germino asks, i may have missed this in the powerpoint but can you clarify what cheatgrass percent is? is is percent of pixels dominated by cheatgrass? why do values only reach 20%? groundcover of cheat can become close to 80% or more of ground area. << um first of all we're looking at the mean, overall, the overall study area when we're looking at the mean we're looking at the average value of

every pixel um in the study area that is is identified as as shrub or grasslands. um as far as, we had some areas, and and i didn't show the individual um maps where you can see every piece ofthe map, i was just showing it as an example. i can tell you however that we saw um in some years areas that had 87%cheatgrass cover. um especially up in that in that northeast and north central areas.

so there is there is significantly morecheatgrass um cover than than we showed in ourmaps. << but basically what we're trying topredict are the peterson estimates of percent cover of cheatgrass out of peterson right? << yes, that's our training set. << and so uh you said to me that you, there was uh not large areas thathad higher than 30% cheatgrass in those peterson maps. is that right? << um as i remember, yes that is true. umthere are areas

though in peterson's map that do getpretty high free and that is reflected in our individual, like i said, our individual annual map of cheatgrass production. << okay, thanks. david tart asks, doescompetition from other plant species play a role in dieoff? is that something for the next presentation? <

buttercups and stuff coming in. but i don't think there's much that can competeeffectively against cheatgrass. << okay. kevin kilbride asks, this is very interesting, what onthe ground data is known about the response of natives in areas with cheatgrass dieoff? looking for management and restoration considerations. << well uh again susan might be better to address this uh but inthe field that that we were there, we were seeing uh poa secunda

uh coming back in some areas, some dieoff areas. uh that was the one that stuck out uh to me as coming back. << wasn't there an area that was uh devoid of vegetation in 2010 and then in 2011 there was a lot of tumble mustard that came back? << oh, yeah i think so. << okay. okay, thanks. david tart asks, is a dieoff a low production year or a complete absence of cheatgrass? << it can be defined as a partial orcomplete

stand failure. that's how we define it. << okay. um let's......... << for um, if i can just interject, not to cut anything short but kind of in uh in time interest um is therea way we can capture these questions and be able to address them specifically back to the folks and look hopefully airlift that yes wehave all across? << well hopefully they're listening um but yes we have all the questions. um, the go to system saves all of the questions. << okay i was just trying to kind of make sure we had enough time for susan. << oh yeah, yeah of course. there's just one more question. do you want me

to ask it? << okay. << shaun taylor asks, since 2010 uh is an outlier for precipitation versus cheatgrass, oh uh since 2010 is an outlier for precipitation versus cheatgrass, is it also a year of high cheatgrass dieoff? << is an outlier for, so if we're comparing cheatgrass to precipitation? << uh, huh. <

it has a larger um let's see here, i don't know exactly what the percent is off the top of my head. but this is my 2010 map here, and there is, you know, there's there'sthere's there's not a, it's not like 2005. um but there is a fairly significantamount of cheatgrass dieoff, especially down over in this area here and up in this area here as well. and um i

think as far as uh, as um dieoff, 2009 might have been alittle more of a, well no i shouldn't say that. i will, i will leave it at what i was saying right there saying right there. << okay alright thank you very much. don would you like me to make susan the presenter now? << yes please. that would be great. << okay. << susan are you mic'd on. << i am mic'd on. thank you very much. now i just have to make sure i press the right

buttons here. (laughing) and let me make my slideshow come on before i put the show my screen up. this is only part that makes menervous. (laughing) okay and there show my screen. everybodycan see me okay? << yep. << yep. << okay um our work started quite a bit later thanthe work of the folks at the eros lab. we only gotour money at the end of the summer so we've only been hitting the groundrunning for about six months. so if things

seem a little sketchier at our end, that's oneof the reasons. um our research uh group includes folksfrom four, four institutions. uh myself at the shrub sciences lab in provo, and then we have peter weisberg and inelizabeth ledger at university of nevada reno, and peter is our remote sensing guy and elizabeth isour restoration ecologist, and then we have brad geary atbyu who's a plant pathologist, zach aanderudwho's a soil microbial ecologist, and julie beckstead who's

just kind of our all purpose ecologist, whoalso has a lot of experience with plant pathogen work. so i pulled these people together totry and uh, i, we were asked to address everypossibility about what might be causing these dieoffs. oops, doesn't want to advance my slides, that's interesting. lets try an arrow. oh cute. why won't my slide advance? um i just want to acknowledge

uh the the fine efforts of mike pellant anddon major to obtain some funding for our part of this research. you all know last year was a very tough funding year and we were uh very impressed withtheir ability to to get us some money to getstarted on this research and we thank them for that. this is going to slow me down a little but my mine my next buttons don't work. <

on what we're calling a dieoff. uhdieoff is kind of a misnomer because it its sort of a hangover from the 80's when we had the shrub dieoff. some of you've been around long enough to remember that. where we had stand loss of shrubs over large areas inthe great basin, and that was really a dieoff becauseperennial plants were dying. whereas what's reallyhappening with cheatgrass is stand failure because cheatgrassdies every year right? it sets seed and dies. it's an

annual plant. so so what we're really looking at isestablishment or stand failure, and as was just mentioned, this is amatter of degree but usually people don't notice it on theground unless it's 100%. but probably it happens, it would be like stand thinning. it mayhappen every year to some degree but you don't really notice it unlessit's 100% and also, oops i just lost my train of thought let me,

my little button is worrying me here. sowhat are our research objectives? what we were charged with initially was to determine the causesof these dieoffs and that's obviously gonna need to be understood in order to be able to manage post dieoff. and then we we decided to emphasize twoother, two other aspects of the research. one was to determine what did what what factors affecthow quickly cheatgrass recovers after a die-off. andyou can see from these maps that were

just presented that sometimes thisrecovery happens very quickly, within a year, sometimes it takes many years, so we need to understand what's controlling that. and then in relative, relevant to the questionsthat were asked at the end to the last presentation, do dieoffs represent restorationopportunities? can native plants grow in a die-off? and i will say that we haveno evidence that die, that whatever it is that causes dieoffs has any effect atall on established native plants, they seem to be fine.

it does not kill established grasses or shrubs. whether it would kill their seedlings is a moot point. that's something we needto look at. so to try and get our brains around how, whatthe possible mechanisms, causal mechanisms die-offs, we madeourselves a little flow chart and the this is the way it works. down here wehave the life stages of this ann, of this annual plant. it goes from a seed, the seed germinates, you get a germinant thathasn't emerged yet, then you get an emerge seedling, then you get an established plant, and then you get

see production on the mature plant. andthe death, that that looks like a die-off in thespring when there should be mature plants there and there aren't, that deathcan happen at any of these life stages. and these life stages, this this bottom panel here shows when these life stages are expected to be present. so the seeds can be present year roundbecause cheatgrass does form a persistent seedbank. so there can be seeds present year-round. usually there's no growing plantspresent in the summer

but they could be present in autumn,winter, or spring. and then the mature plants would be present in the, in the spring, in the late spring, and inthe summer when they're producing seed. so that's just to get you some perspectiveon when these different variables might operate. so here's the way we set up ourour model. we start with macro-climate or weather, seasonal temperature and precipitationpatterns. they interact to generate indirect affects on theplants that are these various stressors or some of them could be positive. it couldbe extra water or it could be

not as cold. so these these maybe shouldn't have been expressed as stresses. but these are variables, at theplant level, that result from the interactions of theseweather variables. and then they too can interact with eachother and they can affect the abiotic conditionsdirectly, that directly influence the plants. theycan also affect agents that might be involved with die-off, including pathogens and what we call predators and herbivores. and what we mean by predators there,

would be seed predators. so these abiotic stressors or variables are gonna directly affect the plants, or they are going to affect thesebiotic factors, which then are gonna affect the plants. and these three things can potentially interact with each other andthen as they interact they may affect thesedifferent life stages of the plant. and depending onwhat what the pathway is, we would have some insight into what the cause of the death or

the die-off would have been. so that's uh,that's a lot fine print but i want to walk you througha couple examples of how this how this might work. okaysometimes, as we know, there are drought years and cheatgrasssimply doesn't establish. this is not commonlyreferred to as a die-off because this is explained by the weather, right? the thing that caught people'sattention about die-offs was that they happen independent of dry weather. if there's a drought year and there's a,and there's no cheatgrass, people aren't too

surprised, or if this lower productivity. but if it's awet year and there's hardly any cheatgrass then that gets people's attention. but in fact you could call this a die-off because it's stand failure. so here's theway it might work, and i mean you can work it through any of thesevariables. but this particular example says, okay we had this unusual kind ofa year, whereby the the the winter was very open, the snow coverwas very low, so we had high coal stress

and high freeze-thaw dynamics. the theseedlings were getting shredded up by freeze-thaw in the surface horizon and that could directly affect theseedlings such that they just die without having tohave, be influenced by pathogens or predators.you could have a year where the the cold temperature and freeze-thaw dynamics were so bad that they could just prevent seedling establishment. so that would be an example of a die-off mechanism that

operated only through this abioticpathway. another possible die-off mechanism that'sreceived quite a bit of attention and is, undoubtedly operates in somecases is herbivory. and our experience withherbivory involved the grasshopper herbivory. this was out in skull valley in 1999. we just happened to be doing agrasshopper study out there that year, largely because there was so manygrasshoppers that destroyed some of our studies, so we decided to study grasshoppers. but wehad these exclusion cages out at our

study site and on the white rocks road. and they contained the only green cheatgrass for severalsquare miles because the grasshopper outbreak was sosevere that year that they literally ate every blade of grass. there wasabsolutely no seed production over the, over really quite a large areain the central part of skull valley. so this is an example of a die-off that happened through an herbivore action right, but whatcauses

grasshopper outbreaks? i don't actuallyknow too much about this, but one of the theories is that if you have the seasonal weather and precip. patternsinteracting in such a way that you have very high snow-cover, then you can have very high survival of grasshopper eggs, and then you can getoutbreaks of grasshoppers. so what we have is aninteraction of this this this winter weather conditions withgrasshoppers to create this outbreak that thendemolished all the plants over tens

of square miles and caused a dieoff. so that's an example of an herbivore caused die-off and there are other examples. there's some folks who've done a lot work or did did a few years ago with fall armywormsas a possible cause of die-offs. and that certainly hasn't beenexcluded as a possibility. these things havemultiple causes okay. there's no simple answer to what causes these things but we're trying to look at all the options. the one we're the mosthung up on though, just because of our training, is disease. these are cheatgrass seeds

and they have been killed by a plantpathogen called a fusarium, which is an ascomycete uh facultative pathogen that's extremely common in just about every ecosystem in theworld. these are very well studied organisms,very complex, but they are out there. so we're lookingat whether they might have something to do with this. but the way, one way we think that this might work with the fusariums is if you have the weather patterns that create

early-season water stress. now that's notto say no precip, but supposing you have a yearwhen there's always little false start storms in the early fall that, where the seeds are sitting there imbibe but they can't, they don't stay wet longenough to germinate. but that abiotic condition at the sea levelmight interact with a pathogen to to enable the pathogen to kill seeds in high numbers. so you'd have thateffect in a year when you had this particular weather scenario.

but in a year when you didn't have thatweather scenario the pathogen might still be present but you would not getthe die-off effect. okay the reason we think that might be going onis because a few years ago we did the classic die-off study for soil pathogens, where we gotdie-off soils and not die-off soils. and we put them, we put them in the greenhouse. we autoclaved half of them. we planted are cheatgrass and, you know,the hypothesis was

more cheatgrass will fail to emerge from die-off soils and autoclaving will make this effect goaway. right? the classic test of whether it's a plantpathogen. what we got was no effect. the cheatgrass plants grew perfectly well whether it was a die-off soil or not. they grew perfectly well whether we autoclaved or not. we basically had no, we had no effect on emergence from any of these treatments. so this wasfrustrating to us since we believed that a pathogen was involved. so we isolated abunch

of of of potential pathogens off of the seeds from this greenhouseexperiment, which was carried out by an undergrad student of ours owen bachmann, who is now working with beth as a graduate student on this project. he did a yeoman's job on this. it was ahuge experiment. he basically got a nul result. so we we we isolated a bunch of organisms off of these seeds that did not die and identified them using moleculartechniques.

we picked the ones that belong togenero(?) that are potential pathogens. wecultured them up to produced conidial inoculum, and then we inoculated cheatgrass seeds withthem in the laboratory and scored germination and mortality. and we had two two pairs treatments. we either did this with dormant or non-dormant cheatgrass seeds, or we did it with or without an initial water stress period before the seedswere transferred to free water.

so what we were testing here is, aredormant seeds more susceptible or are non-dormant seeds more susceptible? and we felt it should be non-dormant seeds because if the pathogens is going to prevent emergence of a current year stand, it has to be able to kill non-dormant seeds. and we also predicted that the water stressperiod should should give the pathogen an advantagebecause it, our hypothesis was it will be able togrowing and infect at low water potentials that prevent the the germination of non-dormant seeds. sothat's what we're investigating. this is

what fusarium looks like when itkills cheatgrass seeds in in a laboratory experiment. it forms these little tufts of mycelium on the radical end. so it is somehowsensing the radical end of the seed. it knows that's where theradical is going to come out. it is sensing some chemical signal that that puts out that will uh cue it. that its, that that seed is now susceptible to be infected. pardon the names of these isolates.

this this this these isolates, the student who's worked on this for us, her name is janna lynn frank and shehas an army of undergraduate research assistance. in this particularcycle the experiments she allowed these kids to name their own isolates. so this guy named his isolate night fury,which i believe is a dragon from some movie. i don't, i don't know, i was, they had to explain this to me. but anyway this is the pattern we got with night fury in this experiment. either way the background here, these arenight fury spores.

it's a very prolific spore producer, so we like night fury a lot. but look at the pattern we got. the triangles are dormant seeds and they essentially were not impacted bythe pathogen at all. very very low mortality on dormant seeds whether we did it in purewater, which is the blue, or whether we did it with stresstreatment, which is purple. when we put the pathogen on non-dormantseeds in pure water, this particular isolate

could kill about 40% of them. however when we, we had, we gave it thewater stress treatment first it slammed the cheatgrass seeds, the non-dormant cheatgrass seeds. it killed almost 100% of the non-dormant cheatgrass seeds. so thisfit our model right? our model was dormant seeds should escape cause they're not putting out the cue that says kill me. we should, they should, most of the seeds should be able to escape in water cuz they germinate so quickly.

but if we give that pathogen a head startby incubating the seeds at low water potential or under water stress, then the pathogen should be able to win and kill a lot seeds. okay, we didn't just do this with nightfury, we did this with quite a few different strains, and they all show a similar patternpretty much, it's just a matter degree. and pleaseagain i apologize for these names i did not make up these names but someof them are very apt.

q-tip was a wus, it couldn't kill anything. it it probably isn'treally even a pathogen. it killed less than 10% of the seeds, even under water stress non-dormant seeds. again these these pathogens wereisolated out of out of owen's experiment. that doesn't prove, and they are fusariums, but it doesnot prove they're all pathogens. this one doesn't appear to be very uh virulent at the very least. and

we really stacked the deck in favor ofthe pathogen in these tests because we did the test at a temperature that thepathogen really likes. it doesn't do as well at lower temperatures. we also put a lotof spores on these seeds. so if it could do it it ought to be able to do it in this test. so we sort of have a hierarchy of less and less wussy. we get up here, fluffy was uh was able to kill almost all the seedswith the water stress treatment but it could kill anything

in in this in unless there was a waterstress treatment. same with wine spill, very lethal as long as it got a head start,but no head start, no kill. and then we had these three that could kill a fairly high percentage of seeds,even without giving the pathogen a head start. so whatdoes this mean? we don't actually know that any of theseorganisms are involved with the die-off yet. one thing that's interesting is that some of these isolates came from die-off soils and some of them came from control

soils. so we think that the the these organisms are endemic in thissystem. they don't just suddenly magically appear when there's a die-off. what happens is something environmental. some, it's the perfect storm for fusariums, some some set of environmental conditions occurs that permits them to go from endemic toepidemic. and whether its particular strains thatbecome epidemic

we don't know yet, but at least we havesome candidate pathogens. these are isolated from die-offareas and they can kill cheatgrass seeds. so here's our take home's. dormant seeds arelargely unaffected, probably because the pathogen requires acue from a germinating seed to trigger its infection process. most of the non-dormant seeds in freewater can escape, although some strains can kill some ofthe seeds in free water, at least in the conditions of our experiment. if you hold the non-dormant seeds under water stress

that suppress germination they cansubsequently suffer very high mortality, depending on the strain. so how does this relate to the field? i mean constant water stress for seven days is not really gonna happen in the field, but we think that's a surrogate treatment that would be sort of like this intermittent wetting and drying, with long periods a partial seedhydration that would favor the pathogen, that we might get in a year when therewere false start storms early in the season. we we these soils that were in owen's greenhouse experiment, came both from

the winnamucca area, eden valley, and from skull valley, utah. we haven'tsorted out yet which strains are from which area, but basically this this this phenomenon occurred in bothareas. again they were found in both thedie-off and the control field soils. whether they could cause diseasedepended on the physiological status of the seeds, whether they were dormant or not,and also on the presence of environmentatl conditions conducive to infection. we've got a very complex geneticsituation out there with different

tolerances to the environmental conditions and alsodifferent levels virulence. so we are, we are seeing the tip of theiceberg on this. so what are we gonna do to sort this out? we've already started some of this. we'regonna use bait, bait seed experiments in the field. insteadof bringing these soils into the greenhouse and planting into them, we're just gonna put seeds in the fieldand then go back and retrieved them, and see whose on them, and whether theyhave killed the seeds. and julie beckstead is

already working on this. we're, she'sactually working on a die-off area up in washington, that's a little bit closerto her. she's in spokane but we can all, it'salready clear that we've got a lot of the same types oforganisms that are operating up there. so we're gonna, that's how we'regoing to obtain more pathogens and we're gonna do that also in skull valley and winnemucca this spring. we're going to identify uh and screen additional candidate pathogens from those seed bait experiments and also from seed bank samples,

or if we get lucky and have a die-offand can actually finds dying seeds and seedlings, we'llisolate off of those. we'll take a harder look at theenvironmental component of the disease triangle. ya'll know the disease triangle and it involves the pathogen, the host, but a very important component is environment. for instance we want, we have manipulated water potential a little bit, but we want to manipulate temperature,we want to do hydration-dehydration cycles, different uh, different environmentalconditions, to see

if that will clarify how this mightactually operate in the field. and then to move to the field in a in amanipulative way, we need to develop durable dry inoculum.so far what we do is we grow up the inoculum on petri dishes. we rinse the petri dishes with sterile water and we get this, the spores in a suspension of sterile water and then we, we control the concentrations of spores in the sterile water and we poor those over the seeds in thepetri dishes. but that's not very practical way to doa field experiment . so we're working on

ways to get dry inoculum and probably this night fury that produces thiswonderful abundance of these what we call banana spores. these are the durable spores of fusarium. it'll be thefirst one we work on to try to develop this dryinoculum. and once we have dry inoculum then we can uh caryout manipulative experiments to elucidate these mechanisms by by trying to create mini die-offs withinoculated seeds. inoculate the seeds, we put themin the field, and then we manipulate

water. we do it a different seasons to manipulate temperature and see if we cancreate mini die-offs in our experimental plots. we also, and this is zach aandarud's,one of his main pieces, is to figure out how to quantify howmuch of these different pathogens are present in the field and in the bait seeds. because we've got awhole raft of candidates but just isolating them off owen's seeds gaveus no indication of there abundance but there are really cool molecular

tools for quantifying how much of different pathogens there are in the field. in the environmental samples basically. and then lastly we're gonna, assuming we get some good die-offs this year, that's we don't get to control thatunfortunately, we will be planting cheatgrass into die-off and control soils, followingseed and seedling fate, and doing some manipulations likefungicide treatments with different classes of fungicides totry to figure out

uh exactly who the culprit might be. so this is a lot of work we have plannedfor the next year or two. i'm gonna talk about peter's stuff. peter weisberg is our landscape ecologist working on thisproject. i got these slides from him and he kindof jumped in the deep and he didn't paddle around in the shallow endvery much. so i just want to tell you why are we, why do we have a remote sensingcomponent and how was it different from what the eros peopleare doing?

okay, their their approach is to say how much cheatgrass cover shouldthere be on this year based on the weather, and then if it's less than that, maybe that's a die-off or that's caused by disturbance and one of the disturbance variables is a die-off. whereas peter's approach isjust going to be to compare across years in a given spot. so is there less cheatgrass or morecheatgrass in a particular spot this year than lastyear?

because what what we what what what whatmy fan is, the reason i wanted to have this remote sensing thing, is my fantasy isthis wonderful animated sequence of maps that show how the dynamics of die-offs in space and time across years and in a small geographicarea. not this very large scale stuff theeros people are doing. what we want peter to do is say, okay here's a spotwhat's the die of history of this spot? when was it a die-off,

and when was it not to die-off? and afterit became a die-off, how long did it stay one? how how long did it take for cheatgrassto come back? so the idea is to get up close andpersonal with these die-offs ratherthan, rather than back off and look at these average indices and stuff. which that's obviouslyvery powerful and these people have come a long way. i compliment them. even the differencebetween that talk and the one at the range meetings last

month was was impressive. but i thinkthere's still a place for this uh, using a remote sensing approachto to to to to try and address causality more directly and that'swhat peters gonna try to do. he's basically using a similar approach on the, on the image analysis, which is this ndvi approach that that's already been explained prettywell so i won't dwell on that. but we're going to work in the winnemucca area and also in the skull valley areaand we're going to use these differences between years

to quantify the changes we see as die-off, new invasion, productivity increase, burn, whatever it might have been. so we, again we're just gettingstarted on this. we uh, the first thing we did is we wentout last fall while the cheat grass was stillpresent as standing litter, and attempted to estimate the previous springs cover based onthe fall standing litter. so already we have, wehave a

a disconnect because that may notbe very accurate. but we had to get started on somethingso in both the skull valley area and the winnamucca area, we went out and collectedthis training data for peter's model. and he's gotten it to work fairlywell for winnemucca, he still hasn't gotten it to work toowell for skull valley. also the other problem he's having bigtime is that using fall 2011 data as your training data means your using, your trying to use cover data from the satellite

imagery from spring 2011, and we all knowwhat spring 2011 was like. it was probably the cloudiest spring inthe last thirty years. so so far he is struggling with cloudocclusion. so in his maps, all this grey stuff is stuff that he just cannotsee. and it turned that a lot of histraining points were actually in places that were cloud occluded because wecouldn't get the cart and the horse going in the same order very well tostart with. but he's definitely got a start on this. and again this is just to show 2010,

a few clouds, 2011, amazing cloud cover. and his criterion was he had to have three pixels that he could really see during the period ofmaximum spring growth of cheatgrass, before he waswilling to say it wasn't cloud covered. so he used a very strict criteria and he, their they their thinking they could relaxthat and maybe improve improve this a little bit. but the redareas are at least, we just made an arbitrary uh elevation threshold of 1700 meters that we're not considering.

it's not really cheatgrass country and we just arbitrarily said we're only considering what's below 1700 meters. sothe red is stuff that's been mask out and the the grey is stuff that was so cloud covered he couldn't see it. so youcan see we're not there yet on this, but i mean he's only had the training day uhfor two months. so we've got a long ways to go on this.the training data this is the predicted cover fromthe training day, this is the field, i'm sorry this is the field measured cover, this is the predicted cover from

his model, and it's not as pretty as bruce's butit's pretty good, r^2= 0.53. and so our plan isto collect a lot more training data this spring and hopefully it won't be socloudy, and do a better job of of training peter's model. he did get some results though. he wasable to show that there was a lot more cheatgrass out there in the areas that weren't cloud cover in2011,

than there was in 2010. this is, 2011was a very wet year. it's not in, its not in theother dataset yet but but if they put it in they'll they'll they might see some stuff. so he could detect differences in cheatgrass cover across years, which he then interpreted in terms of, of uh change. and what he didn't get very much ofbetween 2010 and 2011 was die-off. if anything we got increase, and again this is confounded with the weather

as mentioned earlier, we got an increasebecause it was a more productive year. there are some areas that look likedie-offs but he has told me that many of these areas areartifactual, they're not really die-offs. he's still working on figuring out howto interpret these images. he did find one that looked like itmight really be a die-off, it just wasn't a die-off year for one. so he zoomed in on that and uh to makea long story short, it might be a die-off but but based onthe uh google

imagery, it looks more like a burn, and hehas not followed up on this yet this could be a very powerful way to mapburns as well cause i know there aren't good burn maps. this, and and peter's intention is to take,once he gets the technique down, is to take this all the wayback through the landsat imagery to, the thirty years of landsat imagery, and do year year year to year ,year to year,year-to-year comparisons to track the the appearance and thepersistence of die-offs through time and get me that neat

movie that i want to watch. but at least in the winnemucca area, in the area that he could look at thatwas in cloud occluded, he didn't find anything that looked like a realdie-off. so again this is his. the clouds but we may not need to haveevery pixel, so he thinks maybe we will switch to some modis imagery to combine as well to make our maps were complete. uh we need to do a lot more validationin skull valley, or excuse me training data

collection in skull valley cause the model didn't fit very well in skullvalley. could be due to the inexperience of the people i had collecting the coverdata. maybe the cover data wasn't collected in the same way that it was collected in winnemucca. so we need to work on that and get people, get some group training. make sure we're all doing this the same way. but by, by summer we should have better, a better procedure. the good news is the ndvi seems towork okay. if

you get good good good uh good training data, you should be able to convert that to cover values pretty well. and he did this compositing approachtaking a maximum value across months within the season, and that worked. so he's pretty happy since he's only been working on this for six weeks. so here's his laundry list ofeverything we have left to do and his little note on his powerpoint was a lot steps. so we're not even close on this yet.

he does have a student to work on it. we we are short on money on this project. we have some, which were grateful for,but we did we definitely don't have what we thought, what we thought we needed. so he does not have a dedicated studentto do this remote sensing analysis, he's doing a lot of it himself. thestudent we have at reno, owen bachmann, will be working with beth on these uh restoration experiments, more of the field aspect. so peter ismaking progress.

so why, again the reason we want peter'sstuff is so that we can follow these die-offs through time on a, on a fairly local scale. i've been working out in skull valley forover 20 years and we've had die-offs before, and i sort of remember them, and theydidn't really penetrate my consciousness as notdrought related for the first few years. there are no records. i can't rememberwhere what where when what died of where.

even in our exclosure, we have had die-offs in our study exclosure. it's a big exclosure, it's many hectares. and there have been die-offs in there, but we don't have any records of that. so it's very hard to get your brain around what the processis. so hopefully peter is gonna be able totell me what is the history of die-offs in the white rocks exclosure. when did they happen? how long did they persist? when they wentaway, what was the process? because the process by which they

recover is as important as when theyoccur particularly from the point of view of soilerosion, or restoration, or any of a lot of things or or the invasion of secondary weeds, you name it. we need to know what's controlling how long these things persist. this is one of owen's studysites. it was a new die-off in may, based on hisseed bank data, may 2008, may 2009 it was still there pretty strong, by 2010 it had filled back in with cheatgrass.

why didn't it do that in 2009? that's the kind of questions we're trying tofigure out. and it went straight to cheatgrass without an intermediate uh annual dicot phase. so here's our little picture that we're, we like to make these schematics to sort of focus our attention on the things weneed to look at, either in the field or as in our causal mechanism experiments. so here we have the stand failure. so oneof two things happens, either

it reestablishes from the carryover seedbank and you're right back where you started.and in that grasshopper die-off i showed you,where the grasshoppers ate everything over 10's of square miles and there wasn't a blade of grass except in our exclusion cages, what do you think was there the nextyear? the next year it looked like nothing had ever happened.it was a solid, full stand of cheatgrass. so itre-established from the seed bank. that's common,

that a commonly happens. we had another one in rush valley where we observed the same thing, one year cheatgrass was gone, the next year bingo, it was back again. but sometimes it doesn't do that,sometimes there's a stand failure and then it fails toestablish from the seed bank. and there are some possible causes forthat. it could be that whatever caused the die-off, if it's a pathogen, is still operating and is killing the second year's seedlings too, or it could be that there'sno carryover seed bank.

i'm not talking about black fingers ofdeath today, i i i didn't think i would havetime to go into that components of things. but this is another pathogen that's out there in large quantities that alsokills cheatgrass seeds, but it kills dormant seeds. so it killsthe seeds that don't germinate and emerge and if it's very abundant it can, it can take out all the carryover seed bank, in which casethat would be a reason why cheatgrass would fail to establish from acarryover seed bank,

because there isn't one, because bfodkilled all those ungerminated seeds. it's usually not that extreme but it could be potentially. and lastly if itstays a not, if it stays a die-off for more than acouple years, if you have bad bad recruitment years, saythere's a drought the next year, then the site can deteriorate to the point, mainly by the loss of litter, that cheatgrass is going to have a tough time recruiting back in. so we think there's two scenarios forthis persistent die-off,

where it doesn't re-establish from the seedbank the next year. if there's an annual dicot seed bank,this is those tumble mustards and bur buttercup, and annual kochia, and tumbleweed, that is russian thistle, uh halogeton, a cast of characters who very from innocuous to awful right, but those are the annual dicots. they recruit from the seed bank and dominate the site. they permitthe litter to build up again and then cheatgraass can re-invade. we know itmoves around readily, it

disperses readily, so it can get back inthere even if there's no seed bank, if the site is favorable. but suppose there's no annual dicot seed bank, then the thing is just gonna sit there andhave no no plants on it for a couple of years. it's gonna lose all the litter and be veryunfriendly for establishment of cheatgrass. it's gonna take time for the annual dicots to disperse in, build the litter up again so that cheatgrass can re-invade. so why can the annuals......

<< did we lose you susan? << can you still hear me? << okay now i can hear you. << i'm good, i'm good again? << yes. << i just got a little note that said i didn't have audio for second. so we think the annual dicots, and this this has been known for a zillion years, from early studies up in idaho from40-50 years ago that there's a severe disturbance, annual dicots come in first, the weedy annual dicots, and then cheatgrasscan re-invade.

so here some pictures what that mightlook like. this is a die-off that owen studied, and he knew from the seed bank characteristics that this was the first year of the die-off. and you can see there was very thick litter from the previous and that litter was full of seeds, and the next year it was business as usual. this die-off didnot persist, it was a transient die-off, that's that's one kind. then you can have this kind. this is a die-off that based on the seed bank characteristics, he determine

was more than one year old. it had manyfewer seeds in the seed bank and you can see the litters are alreadystarting to go, its going patchy. we came back to this die-off two years later and it was still a die-off. the litter wasalmost completely lost and annual dicots presumably hadrecruited in. they they they probably weren'tthere as a seed bank, they probably had to recruit in, and i mean these things are very good at recruiting in. i mean isn't tumbleweed the classic disperser,or tumble mustard. why do you think they tumble like that?they go looking for

severe disturbances to recruit into. another thing that makes a die-off a really happy place for these dicots is that their nitrogen enriched, because when cheatgrass fails togrow, it leaves a pool of available nitrate in the soil andthat makes these annual dicots happy as littleclams. so they go looking for these nitrogen enriched, cheatgrass free the areas to uh recruited into, and they do recruit into them.

so this is 2010 and i can tell you, i'msorry i don't have a picture, but the cheatgrass is now, in 2011, it hadstarted to move back in to this area. so this die-off will be five,it'll take, it will have taken it five years to recover. now why it didn't recover the firstyear, we're not sure, but once that litter starts to go then that changes the whole dynamicof this successional trajectory. okay last, i, we don't have any date on this yet, so ionly have one slide.

so the question is, if there's adie-off that means cheatgrass just controlled itself right? so you have cheatgrass control. so does that mean you can seed? is that, is that a restoration opportunity? and it depends on several things right? ifnative species can establish from seed under the conditions that prevail after the die-off has occurred. ifthere's a cheatgrass seed bank present and it just recruits back ,it's atransient die-off, you probably will have thrown your seedaway. if

the pathogen is still operating and if itcan attack the natives that's not going to work. ifthis site has deteriorated so much in itsphysical conditions cause of loss of litter that the natives are going to havea really hostile environment, that could also prevent your nativeseeding from taking. so it's not immediately apparent that die-offsare going to be restoration opportunities, but we want toinvestigate this in detail because this naturallyoccurring control happens. we want to know what we, whether itpresents an opportunity. so the way we're

going to approach this, obviously first we're gonna find out ifthese natives we want to seed are susceptible to these causal agents, especially the pathogens, and thats gonnabe fairly easy to do. and chances are fairly high that yes, native grasses are susceptible tothese fusariums. there's is no reason to think these fusariums are host specific, but we don't know that yet. and thenassuming we get some good die-offs were gonna do these precision seedingexperiments,

where we plant these natives into die-offs and controls with and without controlling cheatgrass some other way, use treatments designed to protect theseeds from different die-off agents, use litter, use litter mulching techniques to ameliorate the seedbedconditions. in other words manipulate everything we think might have to dowith it. and then these are going to be thesetoothpick experiments that beth has pioneered, so that we can track individual seedsand seedling fate.

and this also means that if the seeddoesn't come up, we can fing out what killed it. and we can find out whichpathogen killed it using zachs uh molecular techniques. and lastly if we get to thestage where we're going to do these mini die-off experiments, we willput native seeds in the mini die-off experiments as well. so in summary we have a lot ofdifferent approaches we could take and its gonnadepend a lot on what kind of a die-off year we get. wecan't afford to sit around

for two or three years waiting for thatplace in the graph that's the magic perfect storm year for high die-off to study this phenomenon. we've go to approach it from as many angles as we can. so that's that's where we'reat with this and this is where we hope it will all go. you should have seen the picture of owen after he did this. he has very large red afro of curly hair and it was completelycovered with cheatgrass seeds, and he had a very grumpy expression on his face.but he's not giving me that's slide to

share with you. okay let me see how i give myself back to you. << well if you have questions you can leave it on just in case people want to refer back. <

we uh, we will take that into account. we've done those kind of experiments with our other little critter, the black fingers of death. we've looked at how well it survives burning and it has very high, it actually survives burning better thancheatgrass seeds do. yeah that's a good question. it'd be niceto know if there's a way to take it out for onething, because you know, you see something that can killcheatgrass this well, if your kind of a biocontrol nut, you think gee maybe we could use this for biocontrol. i'm not even going there yet

but we would definitely need to understandhow it, how it persists in the field and we will be doing those kinds of studies assuming we can get enough funding. that's a whole other layer ofthingto do, to look at the fire side of things. but yes, we are interested in that. << great, thanks. dirk metz asks, how many of the uh fusarium isolates cause mortalityover 50%? have there been any experimentsconducted to see response created by these isolates and the

presense other organisms that occurin the rhizosphere? << that is also a very goodquestion. we we are just barely getting started onthis okay. we've tested maybe a dozen, well let's see maybe 15 isolates and probably half of them can cause mortality with the water stresstreatment of over 50%. but this is areally artificial situation right? we're, we're at the absolute baby steps of figuringthis out. julie had a student, an undergrad a fewyears ago, who actually put

different , he put combinations of differentorganisms off of cheatgrass seeds in experiments together, to see how theywould affect each other, to see if they would compete, and he did get some interestingstuff. he had a fusarium in his study, whichinterestingly did not cause any mortality on his his non-dormant or dormant cheatgrass seeds in water, but they did interfere with blackfingers. when he put the fusarium and the blackfingers together on seeds,

black fingers wasn't able to kill as manyseeds. the fusarium didn't kill any anyway so it wasn't a fair test for the fusarium. but yes, that's a good question and there's, the the whole, and this is where zack can help us too, because the soil microbial community is vast and complex. and to think you can reduceit to a half a dozen pathogens is pretty simplistic. we're just hoping we get lucky and thatthere's one or two major players out there. but yes, we're gonna be,

his his technique consists, he can, he canactually take environmental samples and find outeverybody whose in them and how much each one. so we have some verypowerful tools to look at that. thanks. <

unit uh have been doing work on, withnatives actually, looking at the, looking at uh, they did this very elaborate experiment where they controlled the dept of the snow in artificial manipulations and then they had the things censored up outthe wahzoo, and then they had seeds and they looked at seed fate. but i don't know the details of how they measure all that, we'd have totalk to zach and i don't think he's on today.

but that's a good question, is not simple. << okay, thank you. don major asks, if folks in the field think they have found a die-off, what info should they collect? << that would be very helpful to know. they should if they think they found the die-off, itwould be helpful to know whether it's, whetherthey'd, whether this was the first year it was observed? is anew one? how much litter is there? how big is it?

um are there any, are there any remnantperennials? are there any annual dicots present? whatwould the annual dicots be that are present? um a good location, so we can go have alook at it or find it on peter's maps. that's what comes to my mind but i, don i think you and me need to sit down and be more systematic about thatbecause it'd be neat if we could just come up with a datasheet for people to fill out. we could put it onsome website somewhere and say help the, help us figure out die-offs. if you findone you think is a die-off,

fill out this form and send it tosomebody. we, i think that would be really helpfulbecause like i say, it is impossible for even the human brain, the marvelouscomputer that it is, to keep track of all the stuff in yourhead. you really need people out there helping you find things. << yeah, this is don just to kind of add to that a little bit. um hopefully before spring comes on us um working with matt germino andusgs through the re-net i believe it is, uh we're hoping to maybe have

at least a location where folks could,um you know once we kind of flesh out what uh, what is kind of the bare-bones information we need, uh have a place that that could be provided. << that would be great. <

die-offs that i didn't even mention today,that's soon going to be going to press. so that,that information will become more available. but the whole black fingers interaction thing is very interesting and you can tellfrom the seed bank samples how long it's been since the die-off occurred. if you have a reference area that you're comparing it with. so yeah we need to teach people how toat least take these samples and what the minimum number would be thatwould be useful. << yeah. << great, thanks. well that was the last question so you can

take it from here don. << alright, well i just wanted to thank everyone for attending andand special thanks to the presenters and facilitator. i just want to say thank you very much and if you have any questionsfeel free to direct them at the presenters as well as send me an email. and i think with that we will say goodbye. << okay. << thanks owen. <

<< can um, can we get the, can you send me the stuff on the presentations so we can get that out. << will do. << thank you. << oh, ok. << just the record stuff. << yeah okay, bye. << awesome, bye. << do you want these uh powerpoints don? << i think it would beawesome if we could archive them in a location that folks that would want to go see could go see. << i tried to make them so they have enough text they make some kind sense right. but not so much that people can't stand it. it's very hard to draw the happy medium there.

<< okay thanks talk to you soon. << yep bye. << thanks everybody, great job.

Author : indra

Share this

Related Posts