Habitat Selection - Science topic

Since your presence data are biased along transects, one could generate the pseudo absence data along the transects too, within a certain width. This is akin to the "target group" method sensu Phillips

Sample selection bias and presence-only distribution models: implications for background and pseudo-absence data

See this paper uses absence points along transects for herbivores

Okay, that question finally broke my brain hehe :)

You can use 'repeated measures ANOVA', which compares means across one or more variables that are based on repeated observations. Scores for the same individual are dependent, whereas the scores for different individuals are independent. Or just state the assumption that the probability of a repeated measure is equal to an individual sighting, then go ahead with standard analysis

Hi Daniel,

This attached article by Julian et al. shows the oxygen consumption rate of some electric fish species. I hope it helps.

Best regards,

Nonato Mendes Jr.

Hi,

I would suggest the following:

1) Combine your landcover data. Ideally, you have one categorial data set that has your prefered (=biological meaningful for your targeted species) landcover classes. As I read your data originates from the same landcover, so this should not be a problem. But avoid NoData in your data. If there are a lot of cells that you dont have data for, then the MaxEnt algorythm also cannot calculate occurence probability for those cells.

A solution might be for example to include a additional landcover classes which fills those NoData cells, but be very cautious with that to not alter your landscape.

Also dont forget variables with potential negative effect on species, like maybe human infrastructure.

2) After running some models with a sinle landcover data set, you should get knowledge about which landcover category is prefered and which one is avoided by your species. Then you can go on and included additional "distance to good landcover" and "distance to bad landcover" variables. Those are continuous data spanning your whole study area that represents the positive and negative effects of landcover on its suroundings.

3) I am no expert on shorebirds, but in general I start the modeling process by including topographical variables, landcover variables and climatic variables. You can try to include meaningful climatic variables; MaxEnt will decide itself if those are explaining or not. Dont forget to exclude highly correlated variables though.

4) Dont forget to check "categorial" before running the model. This is trivial, but happens quite often.

Cheers, Florian

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

https://www.jstor.org/stable/1366963?seq=1#page_scan_tab_contents

Hi Jeremy Orange. There is a simple and elegant example of how can you perform these comparisons between habitat availability and use in these articles:

Neu, C. W., Byers, C. R. and J. M. Peek. 1974. A technique for analysis of utilization-availability data. Journal of Wildlife Management 38:541-545.

Byers, C. R., Steinhorst, R. K. and P. R. Krausman. 1984. Clarification of a technique for analysis of utilization-availability data. Journal of Wildlife Management 48:1050-1053.

With GPS track data and quantification of habitat availability this is easy to do.

All the best,

Alexander

I like this discussion

Hi Muhammad yes in that case each individual would be a separate point or buffer. Maybe the best way of incorporating the fission fusion part is to have a 1 fission or 0 fusion variable for each observation. This can then be entered into the model as an interaction to understand if or how each habitat variable is related to grouping behavior, this may be easier than separating it into two analyses (Fis and Fus) until you find you need to. The random available points should be sampled from the surveyed study area, i.e. only where you have sampled. Unsure how you are collecting the data (GPS collar?) but you may also want to put individual ID in as a random effect. You should use a mixed effects model with a logit link function (logistic regression). There is a lot I don't know about your study and a lot to consider with RSFs to get them correct, hope this helps you make some progress. I have found some of the most helpful RSF papers are by Mark Boyce, Chris Johnson and Scott neilson, and their colleges.

The most common methods are angling and electrofishing.

Recent efforts to sample fish assemblages in deeper rivers by using multimesh gillnets developed for lotic environments (Fjälling et al. 2015), i.e. the Norden multimesh Stream Survey Net (NSSN).

Fish abundance or density is ideally expressed as numbers or biomass per area or volume of habitat sampled.

First option is to determine the absolute abundance. However, it is costly and time-consuming. It requires extensive data collection, such as a precise estimate of density at sampling sites and a probability-based array of those sites.

For many populations, indices of relative abundance are sufficient for assessment. I think that this is the best choice for you. Catch per unit effort (C/f or CPUE) is a relative abundance index, which is often directly related, though not always in a linear fashion, to absolute abundance. CPUE is more specifically expressed as numbers or biomass per unit effort (NPUE or BPUE).

You can find detailed description here:

https://www.researchgate.net/publication/292460108_Relative_abundance_and_catch_per_unit_effort .

Precise relative abundance estimates should be collected from several areas (randomly distributed samples) in order to be representative of an entire system. In addition, the distribution of some species may need to be understood before undertaking relative abundance estimation, particularly for species that are rare or poorly sampled.

Abundance can be expressed separately for each species in the catch or summed over all species to represent the total fish community. Alternative abundance metrics may be calculated for groups of individuals according to e.g. size, age or functional traits as mentioned above.

onnectable","webrtcAudioPrivate","webrtcDesktopCapturePrivate","system.cpu"],"manifest_permissions":[]},"commands":{},"content_settings":[],"creation_flags":1,"events":["runtime.onConnectExternal","runtime.onMessageExternal"],"from_bookmark":false,"from_webstore":false,"granted_permissions":{"api":["desktopCapture","processes","webConnectable","webrtcAudioPrivate","webrtcDesktopCapturePrivate","system.cpu"],"manifest_permissions":[]},"incognito_con

Most fish live in local places during the dry season. During the wet season, for example cat fish migrates up river to desovates up there while for example trout remins in local places by avoiding the force of the river discharge living in steady ponds usually found besides the main river channel. In this way trout does not spend too much energy dealing with the force of the river discharge during the wet season. There are of course very many differences of the local fish species and the magrating fishes in different places.

I agree with you Ronny. We have to accept some dependence (literal meaning) between animal location at the time of a previous fix, where it is now, and where it can potentially go before the next fix. But I understand mathematical people focus on something different, statistical independence of points as a basic assumption of kernel density estimation. Nevertheless, in many previous animal tracking studies it was acceptable to use kernel methods with long intervals between fixes (reducing the issue of auto-correlation), and could okay for your fixes 3 times per night with 2 h intervals. Newer methods like aKDE from Fleming et al and Brownian Bridge movement models I think are more important for people who get high frequency fixes. Might be interesting to try both with your owl data. It sounds a very interesting study, best wishes for your analysis.

Via the processing toolbox you can find a tool called Sky view factor. It is based on this work

and calculates, among others, the visible sky as a percentage of the unobstructed hemisphere above a certain location.

I tried it with your data and it could work for your owl habitat mapping:

Thanks Dr. Sahadeven Payyadakath, It will be very helpful in my studies.

Regards

Sahar

One thing I know is that it also is dependent on the read length. This is also dependent on the brand/type of sequencer (ion torrent, Illumina etc) you will use.

Fecal samples are tough matter to do DNA analyis on, thats for sure. 

You might also want to make sure that the way you isolate and analyze does not create some kind of bias towards plants with large amounts of DNA or DNA that is easily degraded because of leaf structure.

Well, you have to put your transects where the birds are! No bird present = no bird counted lol. So put the transects where the birds are because you simply want to know what their density is in those locations. There's nothing wrong with doing that. Sampling doesn't have to be random remember :)

> notonecta.dbmem1 <- dbmem(pond_coord, thresh=0.01707804, MEM.autocor = "non-null", store.listw = TRUE, silent = FALSE)

[1] 0.14567211 0.04095702 -0.03654004 -0.09026366 -0.09084915 -0.09090909 -0.09090909

[8] -0.09090915 -0.14911688 -0.18256433

Characteristics of weights list object:

Neighbour list object:

Number of regions: 11

Number of nonzero links: 78

Percentage nonzero weights: 64.46281

Average number of links: 7.090909

Weights style: B

Weights constants summary:

n nn S0 S1 S2

B 11 121 76.8873 151.6504 2420.918

> test

class: krandtest lightkrandtest

Monte-Carlo tests

Call: moran.randtest(x = notonecta.dbmem1, nrepet = 99)

Number of tests: 10

Adjustment method for multiple comparisons: none

Permutation number: 99

Test Obs Std.Obs Alter Pvalue

1 MEM1 0.22924883 5.0022965 greater 0.01

2 MEM2 0.06445536 2.9170811 greater 0.02

3 MEM3 -0.05750422 0.8315211 greater 0.19

4 MEM4 -0.14205079 -0.6149102 greater 0.68

5 MEM5 -0.14297220 -0.7215775 greater 0.74

6 MEM6 -0.14306653 -0.7440757 greater 0.77

7 MEM7 -0.14306653 -0.6722448 greater 0.80

8 MEM8 -0.14306663 -0.8240201 greater 0.82

9 MEM9 -0.23466999 -2.6063308 greater 0.99

10 MEM10 -0.28730731 -3.0581972 greater 1.00

Best,

Instead you can use resource selection function (RSF) for use-availability data, e.g., package “ResourceSelection” in R.

Best,

Andrzej

Only chlorophyll abundance cannot be used for figuring out the potential fishing zones. It has to be used in conjunction with temperature data. 

So, you need to have chlorophyll map and temperature map - Digitize the temperature gradient and chlorophyll levels - overlay both these vector files - The region of intersection of these vectors will give you the potential fishing zone. 

The validity of such maps produced will be only 2 to 3 days from the date of acquisition of the satellite imagery.

Regards,

Nithyapriya B.

Remember that for stationary random functions the sill is equal to the variance of the sample data.

I would be inclined not to restrict the simulated values to a range, but to specify a variance (= sill)  such that some high percentage of simulated values lie within the upper and lower limits, say 99%.  Standard tables of the normal distribution can be used to calculate that variance. 

In my opinion, by using a GLMM you will overcome many problems originated by other methods. Additionally, GLMMs are much more flexible than other methods.

Example with radio-tagged individuals:

Weisshaupt, N., R. Arlettaz, T.S. Reichlin, A. Tagmann-Ioset & M. Schaub. 2011. Habitat selection by foraging Wrynecks Jynx torquilla during the breeding season: identifying the optimal habitat profile. Bird Study 58: 111-119.  

Hi, have you heard about occupancy modelling as a tool for habitat selection?

I think that this analysis will be useful to you for estimate habitat selection.

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 Donald, 

Thanks for your response. We have converted our tracking data into a spatial count of presences / unit area metric, however, we have since decided to shift our analysis as we thought the complexity of this model wasn't necessary given our core questions. 

Hi Melanie and Hein,

Thanks for joining in on the discussion (and for the papers Hein, I look forward to reading them). Hopefully the following text answers both of your questions.

When we fit a conditional logistic regression using the glm, the response data are all 1's and the independent covariates are the differences between the used and available habitat values (i.e., within a 1:1 paired design). We remove the intercept to fit a no-intercept model which forces the relationship between relative probability of use and the independent variables to go through 0.50. The R syntax is:

glm(point ~ -1 + Difference1 + Difference2, data=data, family=binomial)

When I fit the same model using coxph, I do not difference the used and available data and the response variable is a an R object of class "Survival" (see the Survival package for details) of syntax:

Surv(time, point)

where time is always one (time-to-event = 1) and point is 0/1 indicating use/available. Additional, the strata argument must be used to identify matched (i.e., paired) points. The complete syntax for fitting a conditional logistic regression with coxph is:

coxph( Surv(time, point) ~ X1 + X2 + strata(pair), data=data)

These two formulas give identical beta estimates, AIC model rankings (but different absolute AIC values), standard errors, and p-values. I think this is because the syntax for coxph "tricks" the function into maximizing a likelihood that is identical to the likelihood function for a conditional logistic regression. I think that coxph differences observations within a pair "behind the scenes" but I am not positive.

Now regarding collinearity, if you calculate a correlation matrix for Difference1 and Difference 2 and then another matrix X1 and X2, you get different correlation coefficients. This makes sense because Difference1 = X1used - X1available and, in our study design, availability can vary widely from point to point even if used is fairly similar. These differences in collinearity means that different covariates are included in the final model set, depending on the cutoff for determining collinearity. Hence my original question, if these two different formulas which use different structures for the response variable give identical results, which data structure should be used to assess collinearity?

Hope this helps and thanks again for your questions,

Javan

Thanks!

Not sure how you sampled through time, but you could, potentially, use a GLM, and test for differences through time, across treatments, and as an interaction of time and treatment. You can do this pretty easily in SPSS and R. That would help shape an understanding of how changes through time affected each community as well as just the treatment. When you talk about community composition, what metric are you using for that? Is it like one index value or is it a consideration of richness, diversity, evenness all as separate indices? If all you did was sample at the end of the 55 days, you can't really map changes through time, then just comparing the treatments through ANOVA should probably work. Dissimilarity indices are useful too, and applying them in concert with a GLM might help you better understand what drove some of the community changes.

Hope that helps.

There is abundant published research on that general topic, both empirical and theoretical. Just as some examples that could serve as introduction to previously used methods and theoretical frameworks:

Russo, Sabrina E., Stephen Portnoy, and Carol K. Augspurger. Incorporating animal behavior into seed dispersal models: implications for seed shadows. Ecology 87.12 (2006): 3160-3174.

Herrera, Jose M., Juan M. Morales, and Daniel García. Differential effects of fruit availability and habitat cover for frugivore‐mediated seed dispersal in a heterogeneous landscape. Journal of Ecology 99.5 (2011): 1100-1107.

Spiegel, O. and Nathan, R. (2012), Empirical evaluation of directed dispersal and density-dependent effects across successive recruitment phases. Journal of Ecology, 100: 392–404.

Aukema, Juliann E., and Carlos Martínez del Rio. "Where does a fruit-eating bird deposit mistletoe seeds? Seed deposition patterns and an experiment." Ecology 83.12 (2002): 3489-3496.

Santamaría L, Rodríguez-Pérez J, Larrinaga AR, Pias B: Predicting spatial

patterns of plant recruitment using animal-displacement kernels.

PLoS One 2007, 2:e1008.

Schupp, Eugene W., Pedro Jordano, and José María Gómez. Seed dispersal effectiveness revisited: a conceptual review. New Phytologist 188.2 (2010): 333-353.

Rodríguez-Pérez J, Wiegand T, Santamaría L: Frugivore behavior

determines plant distribution: a spatially explicit analysis of a

plant-disperser interaction. Ecography 2012, 35:113–123.

Morales JM, Carlo TA: The effects of plant distribution and frugivore

density on the scale and shape of dispersal kernels. Ecology 2006,

87:1489–1496.

Russo SE, Portnoy S, Augspurger CK: Incorporating animal behavior into

seed dispersal models: implications for seed shadows. Ecology 2006,

87:3160–3174.

Robledo-Arnuncio JJ, Klein EK, Muller-Landau HC, Santamaría L (2014) Space, time and complexity in plant dispersal ecology. Movement Ecology 2:16. (see section 2 in particular)

Hello André,

I hope these (please find below) articles can help you:

Saab, V. (1999). Importance of spatial scale to habitat use by breeding birds in riparian forests: a hierarchical analysis. Ecological applications, 9(1), 135-151.

Orians, G. H. and J. F. Wittenberger. 1991. Spatial and Temporal Scales in Habitat Selection. The American Naturalist 137:S29-S49.

Rettie, W. J., & Messier, F. (2000). Hierarchical habitat selection by woodland caribou: its relationship to limiting factors. Ecography, 23(4), 466-478.

Byrne, M. E., J. Clint McCoy, J. W. Hinton, M. J. Chamberlain, and B. A. Collier. 2014. Using dynamic Brownian bridge movement modelling to measure temporal patterns of habitat selection. Journal of Animal Ecology 83:1234-1243. 

Bleach, I., C. Beckmann, G. P. Brown, and R. Shine. 2014. Effects of an invasive species on refuge-site selection by native fauna: The impact of cane toads on native frogs in the Australian tropics. Austral Ecology 39:50-59.

Blaustein, L., M. Kiflawi, A. Eitam, M. Mangel, and J. E. Cohen. 2004. Oviposition habitat selection in response to risk of predation in temporary pools: mode of detection and consistency across experimental venue. Oecologia 138:300-305.

Best regards

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 

Dear Ning Li,

Non foraging habitat use is often explained after some of the following factors (they are not mutually exclusive): thermal insulation (especially in cold areas), protection against predators, access to sexual mates, proximity to preferred feeding areas, and in some birds you could also consider territorial landmarks. Other factor could be of importance depending of the species. Depending on the species and its natural history, you could develop some hypothesis (i.e. after foraging move to secluded places, or males go to landmarks and female to sunny warm areas to help digestions) and test that against your observations.

You could also expect some kind of relationship between post-feeding areas and habitat requirements of preferred food species seeds.

I do not know if that is of any help.

Regards,

Jabi

Thank u sir.could you refer me to any paper or book?

Expanding on answers by Williams and by Zollner above, wolves are highly territorial and use scat and urine for scent marking territorial boundaries.  Consequently, use of scat location data for wolf habitat selection studies will tend to have a bias towards territorial boundary areas and the core areas may be under-represented.  I have found that a single monitoring approach is rarely adequate for studying presence and habitat use by wolves and thus multiple approaches (e.g., scat/track, telemetry, camera traps) are often necessary to balance the strengths and weaknesses of each approach. 

Positive selection: also called (Darwinian selection) variants that increase in frequency until they fix in the relevant population. The selective pressure that leads to this fixation is termed positive selection.

Negative selection: Also called purifying selection, it means that selection is purging changes that cause deleterious impacts on the fitness of the host.

I think you would benefit from considering both the human population size and the spatial distribution of human development. For instance, a clumped pattern of development usually have a lower impact in the environment than a scattered pattern. There are several spatially explicit indicators that allow you quantifying these dimensions. Please, let me know if you need further information on these indicators and provide more details about your study and available data.

As Fathul suggests spatial processes can mediate competition in a number of different ways to facilitate coexistence. There is an excellent review of theory and data here: http://onlinelibrary.wiley.com/doi/10.1046/j.1461-0248.2003.00530.x/full, free pdf here: http://izt.ciens.ucv.ve/ecologia/Archivos/ECOLOGIA_DE%20_POBLACIONES_Hasta%202004/ECOL_POBLAC_Hasta%202004_(A-G)/Amarasekare%202003.pdf.

I think that you can use R where there are a lot of application

Hi Rubén,

I didn't  know about this publication. Thank you very much.

Ciao Guido. Do you want to compare transects in term of bird community indices i.e. species richness or model species density or just presence absence?

I think you could use the lenght of your transect as an offset in poisson GLMs.  or  the article Spatial Models for Line Transect Sampling by Sharon L. Hedley and Stephen T. Buckland (Journal of Agricultural, Biological, and Environmental Statistics, Vol. 9, No. 2 (Jun., 2004), pp. 181-199) could help you (see link)

By the way, as long as each single transect is your sample unit you should correct your bird community indicies by the length of the transect and be more worried about between transects autocorrelation.

I would be worried about spatial autocorrelation within trasects only if you want to model each presence point inside a transect and compare it with other transects.

I hope it helps.

Enrico

Is this reptile's movement restricted to the riverbed, and is that why you are forcing your random points to be within the riverbed? As for choosing a random point on the ground to conduct your vegetation registration....your GPS is already generating that randomness for you. I suggest that you simply walk to the point where the GPS indicates you should be (i.e, when the "distance to waypoint = 0" for the first time). This GPS location is the actual (true) location + random error (<=5 +- meters). I don't see a need to introduce additional randomness. If you felt that you had to add more randomness then I simply suggest that you a priori generate a list of random distances from a uniform distribution bounded between 0 and 5 (R code: runif(n=20,min=0,max=5)) and similarly generate a list of random compass bearings. Then when you get to the GPS location you use the first random distance and bearing from the list to find your sampling location. 

Either of the two approaches I suggested is less subjective and less open to criticism then the 'throw an object into the air' sampling approach.

Cheers,

Jason

On Mauritius there are several examples of  exotic birds that were flourishing, that declined and were replaced by a subsequent introduction.  The Java Sparrow was replaced by the House Sparrow, the Fire Finch was replaced by the Common Waxbill, and the Cape Canary replaced by the Village Weaver.  Also less convincingly the Madgascar Lovebird was replaced by the Ring-necked Parakeet.  These cases are published in a paper I wrote a long while ago, and summarised in Lost Land of the Dodo, by Cheke and Hume.

Habitat occupancy is the population limit of a given habitat. Habitat preference is the habitat most likely to be chosen by a species given the opportunity or which habitat the species is best suited for. Habitat usage is how a species manipulates its surroundings to better its odds of survival, how it interacts with its habitat. Habitat selection is the process by which a species chooses its habitat.

Thank you very much, Spanish is ok.

I think Federico Morelli observation is very important, define the resolution and scale for the input variables and subsequently the models are crucial. Nowadays many HDM models use Global data sets in particular for climate data. Providing the coarse resolution of these data, they cannot be used to produce good models at local or regional scales. In some regions of the world there is a lack of more accurate environmental data, and then the use of these databases may be acceptable nevertheless in such cases the number of variable need to be keep to a minimum avoiding redundancy so not to carry on with the intrinsic errors produced by the lack of accuracy of such variables. In mountain regions and complex landscape this is a delicate issue and decisions in terms of data quality are crucial as this change the whole outputs of the models. I advise also the use of platforms like BIOMOD that will help to analyze the performance of different models according to the variables used and the object to be modeled. We found very different results in terms of the different model performances for animals vs. plants. It is clear that adding movement trajectories to the animal models provide much better results but obtaining the data is also more expensive.