Spatial Ecology - Science topic

Hello!

I thank my research, as I will soon use platform R.

Another possible way is to use Spatial Feature Coverage Sampling, but with high number of sample points is not relevant. We are discussing its weakness in our ongoing paper.

I used R package GLATOS to build Random Walk simulated tracks constrained into a defined polygon

See applcation in our paper:

Hey, Iraida Redondo García sorry for the delay.

I think that the unit will depend on which unit is the coordinate, no? If it's decimal degrees, then the distance is in decimal degrees.

As far as the traps concern I recommend using completely randomized block design in setting your traps.

As for surveying the Lymantria dispar, my suggestion is to go for line transects.

As for burlap issue, I would suggest to go for some baits in parallel.

hi

if you want a example of the spline methods in R you can click this link: i think i will be heplfull for you

Alas, don't know that program. But there are a number of others that might serve your needs including the density analytics in, for example, the ESRI tools like ArcGIS and ArcPro.

Thanks for your interest. Look at my latest paper about dengue prevalence 2020. I have explained the method very well. Let me if you need help with your data. You can contact me at asad06@gmail.com.

Cheers,

Dear Nicolás, I can't recall specific studies on that to be honest... what I have seen are studies relating the plant-soil interaction with a focus on water budget, but I can't recall that being used for planning purposes.

I found one article that could be mostly related to what you are looking for, hope it helps.

For anyone working on this and similar problems, I have found the spdep R package very helpful.

https://cran.r-project.org/web/packages/spdep/spdep.pdf

Following web link may helps to meet your requirements.

Dear Sari Christine,

Classic ways to map birds in urban environment are the uses of remote sensing and photographs but you now can (if allowed) use drones to map roof-resting birds and transfer your data into QGIS.

Best regards,

Guy

Hi Mark,

My answer is probably a bit late for you, but have a look at this:

An inexpensive satellite-download GPS receiver for wildlife: field trial on black swans

Rebecca M. Lehrke, Lizzie McGregor, John Dyer, Margaret C. Stanley, and Todd E. Dennis. Wildlife Research 2017 44 (6-7), 558-564

Hope the citation helps!

"semivariograms and distance",there are two problems with Fararini's comment.

With only data values you can only estimate and model the variogram (using an empirical variogram). This is not an exact process, i.e. there is no way to be able to claim that you have chosen "the" model and have exactly estimated the parameters of the model. Moreover for a given (finite) data set the empirical variogram is not unique, you must choose the lag spacings and also you must distinguish between an omnidirectional empirical variogram and directional empirical variograms.

Secondly, the variogram is only indirectly related to correlation (unlike a covariance function or (auto)correlation function). A variogram model need not have a true range, e.g. the exponential and Gaussian models do not, moreover the Power models do not even have a sill. It is common to refer to a range for both the exponential and Gaussian models but this is not a theoretical result, rather it is the distance at which the value of the variogram model is 95% of the sill (95% is an arbitrary choice). If the variogram has a geometric anisotropy then the range depends on the direction (in 3D there are two direction parameters)

However the suggestion of using the empirical variogram to choose a maximum distance is not a bad one but recognize that you are choosing a distance, you are certainly NOT determining the "exact distance at which the autocorrelation stops/starts". It is critical to distinguish between an empirical variogram and a theoretical model.

Finally, "semivariogram" is long out dated terminology (since about 1988 when G. Matheron stopped using it)

To Chris Broekhoven, since you are using R packages, you might want to check references on the author of those packages. Do a search in Google for "R package dnearneigh", you will find a lot of documentation and some tutorial examples as well as for "R package knearneigh"

Both dnearestneighbor and knearestneighbor pertain to data sets where the interest is mostly or completely in the locations of the data points, variograms pertain to data sets where there is a value (for some variable, e.g. hydrologic parameter) at each data location and the interest is in how that value relates to the position coordinates.

Deven,

We have had really good results with Microwave Telemetry PTTs on both Whooping Cranes (leg mount) and Reddish Egrets (Microwave's service department has also been really easy to work with).  For the latter project, we attached 17 g transmitters (solar powered Argos/ GPS PTTs; http://www.microwavetelemetry.com/bird/solarArgosGPS.cfm) backpack style with a Teflon harness on Adult Reddish Egrets (I believe all >600 g).  We haven't had any transmitter failure with these (first ones deployed May 2016), and the lifespan on Whooping Crane transmitters was >3 years in most cases.  Early on we tried Northstar transmitters with Whooping Cranes, and we didn't have comparable lifespan.  Also, if you have good cell tower coverage, you may want to consider GSM/GPS transmitters because they work off of cell towers rather than communicating with satellites (http://www.microwavetelemetry.com/bird/GSM.cfm).  Using GSM/GPS transmitters can cut the annual data costs considerably over ARGOS/GPS PTT transmitters.  ARGOS satellite services is ~$100/month per individual (~$1200 per year), so you can see how this could get expensive pretty quickly with 20-30 birds tracked for 3 years.

I hope this is helpful.  If you would like more info, you are welcome to send me a message.

Best,

Will

Hello Sharon! In ArcGIS there is "Search toll" (Open ArcMap and set Ctrl+F), find by keywords to apply wherever... as you know what you want to do (clusters analysis, identify patterns... and so on...) It will not so difficult.

Or, other possibility is to use Geoda software. It works very well for spatial analysis.

Hope to help. 

Thanks for the feedback, I'll have a look at these suggestions.

The package below might be an option, but I agree with the recommendation above to test if your errors are independent, and then, if they are not, test if they are spatially auto-correlate, before trying to use a spatial model. https://cran.r-project.org/web/packages/MARSS/vignettes/UserGuide.pdf 

Arc/GIS and Erdas Imagine is best software

Hi!

You can perform spatial analysis on categorical data by using join count statistics.

A quick overview of the method can be found here: http://www.gitta.info/DiscrSpatVari/en/html/spat_depend_join_ct_stat.html

A more detailed description can be found herer: http://www.statsref.com/HTML/index.html?two_dimensional_spatial_autoco.html

Mathematical Functions of the Method: http://www.people.fas.harvard.edu/~zhukov/Spatial2.pdf

And here you can find a good discussion regarding modelling tools: http://gis.stackexchange.com/questions/49430/what-is-an-appropriate-statistic-to-measure-spatial-autocorrelation-of-points-wi

Hope this helped :)

Best,

Karolina

The gdalwarp utility is an image mosaicing, reprojection and warping utility. The program can reproject to any supported projection, and can also apply GCPs stored with the image if the image is "raw" with control information.

The gdal_translate utility can be used to convert raster data between different formats, potentially performing some operations like subsettings, resampling, and rescaling pixels in the process

Is it this reference? – M. Bunyan, S. Jose and R. Fletcher, "Edge Effects in Small Forest Fragments: Why More Is Better?," American Journal of Plant Sciences, Vol. 3 No. 7, 2012, pp. 869-878. doi: 10.4236/ajps.2012.37104.

They include multiple edges in their model. You could always write to them.

In contrast, my fruiting model used toroidal wrap, which meant there were no edges! The birds reappeared in the forest arena if they flew past the edge. This seemed more realistic. We used toroidal distances to account for this instead of euclidean distances. You could consider doing this?

Hi Oliver, it is hard to find out papers that exclusively discussed about only these three types of segmentation methods. The easiest way to find papers that might be of your interest to search papers with the key words, "survey" or "review" on segmentation methods used in remote sensing. However, I have here enlisted some papers that might be of your interest. Here they are:

1. A REVIEW ON IMAGE SEGMENTATION TECHNIQUES WITH REMOTE SENSING PERSPECTIVE by V. Dey a , Y. Zhang , M. Zhong.

2. A Survey: Image Segmentation Techniques by Waseem Khan, International Journal of Future Computer and Communication, Vol. 3, No. 2, April 2014.

3. EVALUATION OF REMOTE SENSING IMAGE SEGMENTATION QUALITY –

FURTHER RESULTS AND CONCEPTS 

4. A review on image segmentation techniques, Pattern Recognition, vol: 26, Issue:9, 1993

Regards

Chandrama

This additional information is very helpful.  The distinction between land-use and land-cover classifications is significant here in the USA.

In looking at the Corine land cover codes, I see that while there are several that can easily be weighted at 0, and a few that can be weighted at 1, there are actually many that probably merit an intermediate weighting.  These intermediate ones merit such a weighting based on the probability that they include areas that might serve as habitat for heron guild as much as how suitable they tend to be for habitat.

I agree that there will need to be local input, especially regarding the actual local conditions of land that falls into the various Corine categories.

For some of the intermediate categories, I would actually suggest that instead of giving them a blanket weight you might instead buffer around the weight "1" areas where they adjoin these intermediate ones.  For example.  Instead of giving all woodlands an intermediate value, only give them a non-zero value based on their proximity to wetlands and watercourses (weight "1" areas).  One could similarly buffer at along the shorelines of large water bodies, since the shore areas are the actual habitat.

I'm giving you a straw-man starting point here for your local experts to refine.  This is a quick effort based on the nature of these land covers in the area I live.  I suspect that there are some important differences in the nature of the various land covers as they occur in the Po River valley and so some dramatic adjustments wouldn't surprise me.

I expect it will also elicit some comments here too, and that would be good.

1.1.1=0

1.1.2=0.1

1.2.1=0

1.2.2=0

1.2.3=0.1

1.2.4=0.1 (largely because they don't want birds around airports)

1.3.1=0

1.3.2=0

1.3.3=0

1.4.1=0.3

1.4.2=0.5

2.1.1=0.1

2.1.2=0.3

2.1.3=0.8

2.2.1=0.1

2.2.2=0.2

2.2.3=0.1

2.3.1=0.3

2.4.1=0.1

2.4.2=0.1

2.4.3=0.3

2.4.4=0.2

3.1.1=0.1

3.1.2=0.1

3.1.3=0.2

3.2.1=0.3

3.2.2=0.5

3.2.3=0.1

3.2.4=0.2

3.3.1=0.1

3.3.2=0

3.3.3=0.1

3.3.4=0.1

3.3.5=0

4.1.1=1

4.1.2=0.8

4.2.1=1

4.2.2=0.1

4.2.3=1

5.1.1=1

5.1.2=0.8

5.2.1=1

5.2.2=1

5.2.3=0

Evidence likelihood statistics make intuitive sense to me over and above the mainstream and Bayesian statistics I have used so far. The Achilles heel of the Bayesian paradigm seems to me the a priori probabilities required (as in attached 1).

The element of the testing of alternative hypotheses on predictors (environmental variables) based on the literature or comparing the SDM predictions with an alternative explanation (island theory) are both part of my set of SDM studies (attachment 2 and 3). These aspects I have not seen in other SDM papers and appreciate references.

However, I need a bit more background on the 'transformation' of categorical variables/predictors in this context. Concretely, I envisage  in the case of bear (attachment 2), that  the probability for a bear record in the 'beech forest' depends also on the relative coverage of this forest category within the research area. In other words, the statistical findings of an SDM depend on the delineation of the research area. An issue rarely questioned in SDMs. We have some insight in the impact of delineation for the bear study as we ran a model on the southern half of the research area (with proportionally more beech forest).  And maybe this spatial likelihood probability is inherent  in say the concept and algorithm of maxent.

Looking forward to some practical explanation on the 'transformation'.

Hi Ricardo,

The traditional D metric used for calculating the overlap/similarity of two or more species niche according to direct ecological observations/datasets (see Schoener, 1968, Ecology). Whereas, the traditional Hellinger distance metric (I) used for measuring the difference between two probability distributions (mathematical studies back to 1900s), and it was modified for ecological studies (see Warren, 2008, Evolution).

The ENMTools uses the (normalized) (modelled) suitability of occurrence of studied species in a defined space (i.e. study area) to calculate D and I metrics by comparing each grid in the study area between two species or time periods, etc.

D and I metrics (calculated from ENM) represent same ecological information. But, sure, these metrics are strictly linked to how you built distribution models. The difference between the calculated D and I is just because of their formulations. I can recommend you to read Rödder & Engler, 2011, Global Ecology and Biogeography.

Cheers,

Semra

Your question is about spatial variation of rainfall, but your further explanation is about temporal changes/ trends in the last decades. Could you please give some more information about the area and the precise changes and/ or variation in space and time you see in the data? Are you concerned about average rainfall, seasonal rainfall, monthly extremes, annual extremes?

Spatial variation of rainfall can be influenced by many factors like for instance elevation, slope, aspect and prevailing wind directions. This depends on the geographical and climatological characteristics of your study area.

Best regards,

Martijn

Hi all, thank you for your answers. I did use several similarity indices (i.e., sorensen, morisita etc.) to obtain somewhat comparable values between different site communities. I am basically looking at beta diversity. What I need is rather a test, that could help me compare these similarity values statistically.

Moreover, I've looked at 'species diversity versus number of individuals' graphs per site (at which point these sites saturate in their species numbers), and I have different values for each site. Based on these values, I would like to decide an optimum number of species, and hence decide which site is enough to give me that number.

This is important from bio-assessment point of view, so we don't have to sample 4 sites but maybe one site only. This is where I am stuck, and hence needed further opinions/suggestions.

I prefer using a correlation analysis, PCA and natural history of the species to delimit number of predictors. 

Dear Mauricio

Budke, J. C., Jarenkow, J. A., & de Oliveira-Filho, A. T. (2010). Intermediary disturbance increases tree diversity in riverine forest of southern Brazil. Biodiversity and Conservation, 19(8), 2371-2387.

Montero, J. C., Piedade, M. T. F., & Wittmann, F. (2014). Floristic variation across 600 km of inundation forests (Igapó) along the Negro River, Central Amazonia. Hydrobiologia, 729(1), 229-246.

Rosales, J., Blanco-Belmonte, L., & Bradley, C. (2008). Hydrogeomorphological and ecological interactions in tropical floodplains: The significance of confluence zones in the Orinoco Basin, Venezuela. Hydroecology and Ecohydrology: Past, Present and Future, 295-316.

Dear Nathan, I'd assume that most plant species employ both self- and cross-pollination. For these species the proportion between the two pollination methods may widely change, according to the ambient conditions. Like: self pollination may be preferred under harsh condition.

You too should be consider the species that you want modelling. The requirements for invertebrates (eg. insects) can be very different to the requirements of plants or vertebrates (eg. birds, reptiles).The distance threshold to use in spatial filtering (in the same habitat) can vary between some of these groups according to their ecological characteristics.

The Bhattacharyya distance is defined as the negative logarithm of the Bhattacharyya coefficient (affinity). The Bhattacharyya bound integrates over all positions in the domain and assumes that the sample belongs to only one of the two classes. This assumption is a major restriction on the scope of the method as it should strictly only be applied to simple two class problems where this is known to be the case. It is therefore not an absolute similarity measure but rather a relative separation measure.

To add, you may read this too.

Cheers!

Hi,

I've used mostly R packages, which is the best option (with packages such as ENiRG, biomod2, sdm and dismo). ENiRG has a user interface that makes it easier to use.

Outside of R I've tried these:

Biomapper: http://www2.unil.ch/biomapper/

ModEco: http://gis.ucmerced.edu/ModEco/

SAM: http://www.ecoevol.ufg.br/sam/

Open Modeller: http://openmodeller.sourceforge.net/

But I still think that using R is the better option, since you have a lot more control over your models and you can combine resources from several packages.

Cheers,

Frederico

Thank you!

Hi Dr. Zheng,

In my opinion, you can use species' latitudinal midpoints to distinguish them on broad spatial scale, e.g. polar vs temperate vs tropical. But it doesn't say anything about the width of the said distributions. While 2 species can have similar midpoints, one may present a narrow range whereas the other is ubiquist. So it is as important to determine the latitudinal width of the species ranges.

Technically speaking, just add the lowest latitude to the highest one and divide by 2.

You should also use percentile distribution to avoid using the most extreme records. And assessing the shape of the latitude data distribution might also be necessary, depending on how many points your species have.

Instead of midpoint, you should also try the median, but some studies have identified the midpoint as more robust (i.e. less sensitive to sampling effort and distribution data).

Best,

In resume here are apparently the number of points used to determine the Maxent distribution (background points and presence points):

 1 covariate&10 replicates:46373 each run

1 covariate&no replicate:46373

7 covariates&10 replicates:135727 each run

7 covariates&no replicate:145651

I fell to see the patern knowing that the total number of presence points is 99518 and that the total number of background points is 46373 although I ask for a max of 30.000 but wonder if Maxent listen if you use background csv file instead of grid (maybe the option of max background point is just for grid file)?

To me it seems a fairly simple mixed model would do the job here:

Spider abundance as a function of Rain + Temp + Geophys data and a random factor Trap ID. You can include interaction between the two types of explanatory variables.

Since each trap is presumably sampled just once in each year, I don't see why Year would need to be a random effect. A year effect would be the same for all traps. It's the traps that really are sampled repeatedly and Trap ID would therefore be included as a random effect. If traps are sampled multiply in each year, you could include Year as a random effect as well, with Trap ID nested within it.

Adding a variable does not necessarily have to improve a model. This can be due to lack of sufficient data to learn the extra parameters your model is estimating. But, it can be also that you are in a problem where data is very noisy. Realize that the data is not reality, it is an observation of it and it has its own measurement error. Hope is useful. It not necessarily means the variable you are adding is not important for a mechanistic point of view. There can be additional issues like that this extra variable is meaningful at different spatial or temporal scales your model is not considering. An example is if the timing of spawning during the year is important, but your model forecast annual values using annual variables.

Hi Prakash Pradhan jee,

if you are looking for Moran's I test in spatial statistics through ARCGIS then there is a separate method where and how to handle this issue. If you are handing  community data then I recommend you have a look vif (variance inflation factor) in vegan package. If you are working with multiple regression I suggest you have a look at step function and AIC values in glmm package.

Goodluck.

Chitra

Dear Yan,

Fragstats is a stand-alone program so you don't need any extension. You can calculate metrics also with R, by using the SDMTools package and the ClassStat function!

Hello,

We use Microwave Telemetry's 22g Solar Argos/GPS PTT for our lesser-prairie chicken work and we've been very impressed with the transmitters and the company. I'm sure 22g is too big for your species, but I know in maybe the past month or so they came out with a 5g solar PTT that might work for you.  I would contact them and ask about it.  I think they're only producing a limited number of 5g PTTs for the first year, but they also make a 9.5g solar PTT. Here's the website:

If you call in, ask for Cathy.  She's been incredibly helpful in answering my many questions.  When I first called in, she also gave me the names of several researchers who were using their product in similar projects.  Talking to them was very helpful and I'm sure she could point you in the same direction.

Best of luck,

Ashley 

Clearly - in those cases - climate is not the primary driver of distribution. Other, proximate deterministic factors may be more important. Or, randomness (past dispersal events). There is always the possibility too that there are local genetic adaptations to either environment, as yet undetected. Such disjuncts offer interesting opportunities to examine what shapes a distribution. For example, the wood frog of North America has primarily a montane distribution in the southern end of its range (and in the North goes all the way to Alaska). But, in the south there is a documented population, disjunct, on the Atlantic coastal plain. Why?

Thanks. I am interested in this issue as I'm currently looking at spatial prioritization for single species recovery programmes, to inform conservation policy. There are several spatial questions related to this - for example is it better to direct resources in a species core or edge of range? I have found very few tests of conservation interventions applied right across a species range (i.e. covering a theoretical, or measured, core and edge).

I am also interested in examples of where a conservation intervention has been applied on the edge or outside a current range but where the dispersal ecology of the species, or range shift under climatic change, means colonisation or sustained population persistence is unlikely (i.e. where resources have been directed with little prospect of success).

Cross-site analyses of LTER data can be difficult. LTER investigations are aimed at understanding a certain system, and thus focus on differing questions and differing methodologies. This has been a source of criticism levied toward LTER with respect to answering regional to global scale questions and is at least part of the argument for NEON.

The difficulty here is deciding whether you are comparing apples to apples: Are the variables measuring the same thing and thus comparable. It would probably be helpful to know what you are planning. Is there a specific question you have in mind? You might be able to select a subset of LTERs that have certain types of data that would be appropriate for answer a focused question. For example, if you were focused on soil water chemistry, there are 27 soil lysimeter studies in the LTER network at nine or ten sites (based on a quick search of the LTER Network Data Portal). If you just want to look for interesting patterns, I would be skeptical of the utility of cross-site comparisons of LTER to show you much.

Arun, "best" entirely depends on what geometric characteristics you need to maintain for your analysis. Conformality (maintenance of local angles of bearing) and equivalence (maintenance of relative area) are mutually exclusive properties for any map projection, so you must choose one, or a compromise. At 1:50,000 you're pretty large scale (i.e., zoomed in), so you can go with something that compromises area somewhat safely - as long as exact areas aren't crucial to your analysis! The Himalayan area being wider than it is tall, a conic projection, say the Lambert Conformal Conic, sounds good, as long as you've chosen the central meridian and the two standard parallels well for the region (some info on that projection here: http://resources.arcgis.com/en/help/main/10.1/index.html#//003r00000034000000).

The main thing is to be aware of the certain kinds of geometric distortion different projections introduce, and to choose a projection such that you avoid or minimize distortion that would introduce error into your analysis. For example, if direction (i.e., azimuthality) is important to your analysis, you'd need to be sure your map maintains that property. Also, many projections are named for properties they maintain, but they may not maintain that property everywhere. For instance, an azimuthal map typically maintains direction from one point to all other points, but a line drawn between two points that don't include that one point will not have the correct, real-Earth azimuth.

The best discussion about projections in general I've ever come across is Chapter 3 of the GeoCart software users manual, essentially ten pages of map projection education. I highly recommend reading it: http://www.mapthematics.com/Downloads/Geocart_Manual.pdf

I used ER Mapper to have statistics of pre-arranged polygons (average in multispectral space, medians, covariance matrices) as well as to do cluster analysis based on these stats. If I can be at your help, don't hesitate to contact me.

Not sure I fully understand your question, but almost all spatial analysis is going to require that you keep track CONSISTENTLY of the entities. So I'd say yes, you need to preserve the ordering. In fact, one of the ways that we might do a simple simulation test, would be to compute for the OBSERVED labels, and then scramble them (say reorder them 100 times) and see how DIFFERENT the observed value is from the distribution of values from the variations? Maybe this does not make sense from my poor explanation, but take a look at  a book like EXPLORATORY SPATIAL DATA ANALYSIS by Bailey and Gatrell. I think it does a good job on the idea of simulating random variations on the observed data.

Hello!

Yes, you can do network functions with linearK() for one dimension. And LinearKcross() for two dimensional Ripley.  Now, in order to calculate envelopes in a way that the geographic position is fixed, but the labels (sick, not-sick, or male, Female) are randomly shifted among individuals you need to include an option in the envelope command:

envelope(..., simulate=expression(rlabel(X)))  

where X is the ppp object (which has labels).  In this expression rlabel() randomly assigns one of the labels to the points.  That way, the position of the points is kept but the labels are shifted. I think this is what you want.

Hope this help

eric,

Take a look at the following article. I think that methods based on full mapping of individuals could better suit your case.

Hi Israel,

To control for the effects of spatial autocorrelation, you can use spatial eigenvector mapping (SEVM). This approach removes all significant autocorrelation from the residuals of your model. Have a look at Griffith & Peres-Neto (2006) Spatial modeling in ecology: the flexibility of eigenfunction spatial analyses. Ecology 87, 2603–2613.

Good luck

I don't think you can do it, unless you know what is the proportion of animals you have telemetry data on. Otherwise, there is no information on resights of a sample of tagged animals on the proportion of untagged animals, and hence you can't get density/abundance. At best, if you have a large and representative sample of animals for which you have telemetry, you can construct a RELATIVE density surface, which essentially would represent habitat use (assuming you get locations  independently of habitat, which might or not be true!). But you have no information on the telemetry data alone to inform about the height of said function. I believe the references provided by @Rishi Kumar Sharma are only useful if you have BOTH telemetry data and additional information (e.g. captures and recaptures of both telemetry and non telemetry animals), which is clear form the wording in the title ("....radiotelemetry and replicated mark-resight techniques..." and "....radiotelemetry and replicated mark-resight techniques...") 

Hello Ali, I´m glad to hear you that your analysis works fine.

The outputs represents every run in your GARP analysis, a simple and robust way to get a better (and a few representative results) result, is better turn on the "best subset options".  When you have only the best subset results (for example 15 results of 100 runs) you can make a map algebra and make an average map for the potential distribution of the specie. Several authors state that OpenModeller outputs are different of the DesktopGarp software because OpenModeller don´t allow you to modify all the relevant options in the algorithm, this is commonly called as "Tailored software", please use the OpenModeller platform only if you can´t do it in DesktopGARP.

If you have any trouble please don´t hesitate to ask, best regards from Colombia.

Luís Enrique.

PD: In order to get good results in your potential distribution modeling, remember that you need a good quantity of points (about 80, there are no a magical number) for the best subset analysis because you use a dataset of points to create and test the GARP models, and another dataset of points for select the best subset models. 

Hallo Yolandi,

There are many ways of doing this. It would be fun working through them to see what they revealed !

One is to calculate the euclidian distance between the pair of points for each year. That is, take the point for, say, 2005 in plot a and that for 2005 in plot b.  Then calculate the distance beween these - by basic trigonometry.  If the trajectories were identical then you'd expect the distances to all be zero.  In practice they'd be a distribution of values around a mean. You can then test whether the mean obtained is significantly different from zero. (I would do this by repeatedly randomizing the plot b values against a stationary column of the a values and each time calculate the distances and their mean.)

Another possibility is to calculate similarity values between all pairs of points (that is beween the point for year 2005 in plot a and that for the same year in plot b). You'd use the mean and variance values for each year as the basic data. (I would suggest that you might also include kurtosis and skew of the distributions of NDVI values. All these are essentially multivariate measures of the characteristics of the NDVI values for each year in the two localities.

Once you have the similarity values you can try techniques such as clustering to see whether the values for plot a cluster separately from those for plot b.  If there's no separation the trajectories are the same.

There's many more possibilities !  best wishes, Andrew

Short answer is yes. It's simple physics of diffusion. When cell abundance continues to increase through growth, the bloom will spread out. This will happen with or without water movement, and with or without cell motility. You can do a simple experiment to demonsrate that: Add an aliquot of dense phytoplankton very gently to a container of water without any stirring. Observe over time how that patch of phytoplankton spreads out.

Sorry, I didn't finished my answer. For ussing occupancy analysis you can use several programs

"Presence": http://www.mbr-pwrc.usgs.gov/software/presence.html

"Mark" http://www.phidot.org/software/mark/

Or you can use library "unmarked" within R or bayesian approaches with Bugs or jags also within R.

You just need to see which kind of data you have and which constrains you need, and then see if it is implemented in Presence or unmarked or try to implement it with Bugs or jags

Thanks. Your tip for establishing a base station and post-processing is a good one. I'll do that.

The problem is accuracy, in my opinion 15-30m - usually even less (it is common accuracy GPS) for this problem is enough. The second problem is the time (how long schould devices work - battery problem). The next problem is the frequency of information you want have on your computer. (for example you can use GPS-GSM system (or modem) and send positions via GPRS (figh frequency) or SMS (low - in packages) or simply store this information "on carrier - animal" and read after reaserch). First term the specifaction.

Larger open lakes will still be used, certainly by the Red-knob Coot, but only if they have some shorelines with suitable vegetation. Rather than setting a min or max size or trying to use P/A ratios, it would be better to just take out the central pixels (e.g., reclassify all LC pixels classified as water for which all 8 nearest neighbours are also water). The only challenge here would be that they would still use small vegetated islands. As suggested previously, for min size I would think they would use any body of water that shows up as water in a LCC map at 30x30m resolution. Also, you might have a look at some of the work on modelling mosquito habitat with remote sensing - it's a very similar problem - there are other tools and freely available RS data sets that you could use to get a better estimate than the GAP data are capable of.

Hi!

I had to measure the diameter of dolines for my thesis and I think for measuring any areal landform one could use the same methods. So I recommend this paper: Bondesan, A., Meneghel, M., Sauro, U., 1992: Morphometric Analysis of Dolines, International Journal of Speleology 21, 1-55 (http://scholarcommons.usf.edu/ijs/vol21/iss1/1/).

It provides the guiding principles for measuring length, diameter, perimeter, area etc. of various shapes including V shape (of Molat island).

This looks like a question that can be answered with spatial autocorrelation calculation. This can be done in ArcGIS, you can find detailed help here: http://resources.arcgis.com/en/help/main/10.1/index.html#/How_Spatial_Autocorrelation_Global_Moran_s_I_works/005p0000000t000000/

Best regards

Did you see, the uicn courses?

Dear Neven, you can read

Landscape Ecology: The Effect of Pattern on Process

Monica Goigel Turner

Page 171 of 171-197

Annual review of ecology and systematics, 1989

or one from 2005

Landscape ecology: what is the state of the science?

MG Turner -

Annual Review of Ecology, Evolution, and Systematics, 2005

And

Spatial ecology and conservation of seabirds facing global climate change: a review

D Grémillet, T Boulinier

Marine Ecology Progress Series, 2009

Levan,

I'd say that without exploring a number of factors, including sampling, geometric constraints, species area relationships, biogeographical history, and both phylogenetic and functional diversity, you cannot really have an understanding of the mechanisms underlying the richness pattern in your data. Just looking at species richness is insufficient, in my opinion, to say very much about the relative importance of competition, environmental filtering, convergent evolution, or niche conservatism has on the patterns in richness, though there is much research out there that tries to do just that. Christie McCain's meta-analysis "Global analysis of bird elevational diversity" (2009) describes four of the most commonly occurring trends in richness along a gradient (for birds): 1. decreasing, 2. low plateau 3. low plateau with mid peak, and 4. mid peak. Within her paper are numerous citations of papers that have found these different richness patterns. For a paper that addresses species richness, functional diversity, and phylogenetic diversity along a gradient, check out Devictor et al. 2010. Spatial mismatch and congruence between taxonomic, phylogenetic and functional diversity: the need for integrative conservation strategies in a changing world.

Best,

Kevin

Hi Patric. Thanks so much!

I think what your are getting at is that you can have spatial autocorrelation for some variables and not for others, depending on the spatial scale in question. So, for example, imagine your response was tree height and one of your predictors was the slope of the terrain where the trees were growing. If you sampled in an area that is relatively flat with regularly spaced ridges, you might see a pattern of spatial autocorrelation between the slope predictor at the scale of the distance between ridges. Similarly, you may see spatial autocorrelation between the response, tree height, in the flat areas at a smaller spatial scale. You could resample your data to remove the spatial autocorrelation based on a semivariogram. Semivariiograms can be easily calculated in R using the "variog" commend in the "geoR" package.

hi Tom,

I hope the following code might be of some help to you.

###R-code###

library(spatstat)

X <- simdat

Xenv<-envelope(X, Kest)

plot(Xenv)

min(Xenv$lo)

max(Xenv$hi)

range(Xenv$lo)

range(Xenv$hi)

###END R-code###

best wishes

Hi David,

Firstly thanks for this question since I have been trying to deal with this problem myslef. I have to agree with Luis and the only way to deal with this is to use a geographicically weighted regression model. I found that the easiest way to get a handle on this was to download the freeware Spatial Analysis for Macroecology (SAM) - it is really user friendly, contains many different types spatial anlayses, explorative and diagnostic tools and an easy to follow manual. I ended up using generalised least squares, which is like OLS but the spatial structure of the data are described by a spatial autocorrelation function.

Hope this helps.

In their paper Kaczensky et al. (2008, Resource selection by sympatric wild equids in the Monglian Gobi) dealt with temporal autocorrelation of GPS data in a different way: all locations separated by a time interval lower than 72h were given a weight less than 1 based on a specific power law. You may want to have a look at this. Good luck!

There was some discussion related to this topic on the following thread

I don't think there is consensus since spatial autocorrelation can take many forms (eg. different scales simultaneously). Maybe you could clarify your question with specific examples of the dataset?

Hi Nathan,

Presumably, life-history strategy is going to play the biggest role in how fast juveniles leave the nursery areas. If your lifespan is 3 years, then you want to grow up, get out and start bree