[ECOLOG-L] Replies to my question "how to incorporate spatial autocorrelation in a multivariate GLM"
Hello, A few days ago I posted the following question on how to incorporate spatial autocorrelation in a multivariate GLM. I got some kind answers, which I thank and reproduce below. Regards, Alexandre Dear friends, I would like to ask for some advice. I am embarking in the analysis of 3,000 plant species occurrence data across biogeographic scales in South America. I am willing to try to jump from more traditional distance-based multivariate analysis (e.g., RDA on hellinger-transformed abundance data) to multivariate GLM as proposed by you (mvabund package) and also by Yee (VGAM package). However, distance-based methods have grown to incorporate spatial dependency through the development of MEM and AEM techniques, which model symmetric and asymmetric spatial relationships and can be included in the explanatory side of the analysis. Reading the multivariate GLM papers, however, I have not find exactly how to control or include spatial autocorrelation. I am thinking of including MEM and perhaps AEM variables simply as co-variables added to the explanatory environmental variables in the multivariate GLM. Is this a step I will regret later on? Is this ok? A second quick wondering: common GLM analyzes are carried out as a series of nested models in which we exclude variables from an initial full model based on anovas/AIC. I suppose this is also true for multivariate GLM. Is it? Can I compare successive models using the same approach used in common GLM? Thanks in advance for any thoughts, All the best, Alexandre Replies ** Hi Alex, Thanks for the e-mail, sounds like interesting stuff! Yes you could as you say use the MEM and AEM techniques with manyglm, while this is not the best of approaches for handling spatial data, it is the simplest and currently the best one given the current lack of code for an alternative. And yes you could use an AIC approach for model selection. *** Hi, the only thing i am aware of is the spatial autocorrection function available in the nlme package: for example: null.model <- lme(fixed = A~B, data = data, random = ~ 1 | dummy, method="ML") cor.model <- update(null.model, correlation = corExp(form = ~ x + y), method = "ML") argument "correlation" accepts several forms of spatial models based on variogram (here exponential based on xy coordinates). One can extract model goodness with extract.aic() or just summary(). However, this is univariate glm (but can be extended to interaction) and as far as i was told these procedures only exist for gaussian distributions, not for poisson/NB, which are better for species data most of the time. I was looking for the same, but in the end i went back to RDA with dbMEMs and used the aforementioned procedure only for highly correlated univariate pairs in the dataset. Please let me know, if you are more successful. *** Hi Alexandre, Not sure what the best solution is, but a few hacker ideas come to mind. First, you could create a spatially lagged variable from scratch. This would be created by deciding on a neighborhood size, say first order neighbors, and then creating a variable that was the average response (Y) value for the first order neighbors. Neighborhood size could be guestimated by looking at residual maps. This is similar to what happens in simultaneous autoregressive (SAR) lagged models. Then this lagged variable could be a fixed covariate in your model. You could test residuals from the lagged model to see if this removed your spatial autocorrelation. Since you mentioned a GAM approach, you could also do a spatial GAM, where Lat and Long variables are specified as smooth covariates with lots of knots to account for short range spatial structure. Again, you could test your residuals to see if this removed your spatial autocorrelation. If you are comfortable with Bayesian modeling, Banerjee et al. (2015, ‘Hierarchical modeling and analysis for spatial data’) have a chapter on multivariate spatial modeling, with a brief mention of generalized linear models. Some food for thought. *** Alexander, Any chance you might include spatial dependency (however you may choose to do it) as a random effect in a mixed-model structure? This way you can either run the model with the spatial dependency to test this explicitly or remove this effect from the model structure. And yes, you can use AIC to rank multivariate models. Just a quick note. *** Furthermore I received the suggestion to read the following papers: Spatial factor analysis: a new tool for estimating joint species distributions and correlations in species range James T. Thorson1*, Mark D. Scheuerell2, Andrew O. Shelton3, Kevin E. See4, Hans J. Skaug5 and Kasper Kristensen. Methods in Ecology and Evolution 2015 Geostatistical delta-generalized linear mixed models improve precision for estimated abundance indices for West Coast groundfishes. James T. Thorson1*,
[ECOLOG-L] How to incorporate spatial autocorrelation in multivariate GLM
Dear friends, I would like to ask for some advice. I am embarking in the analysis of 3,000 plant species occurrence data across biogeographic scales in South America. I am willing to try to jump from more traditional distance-based multivariate analysis (e.g., RDA on hellinger-transformed abundance data) to multivariate GLM as proposed by you (mvabund package) and also by Yee (VGAM package). However, distance-based methods have grown to incorporate spatial dependency through the development of MEM and AEM techniques, which model symmetric and asymmetric spatial relationships and can be included in the explanatory side of the analysis. Reading the multivariate GLM papers, however, I have not find exactly how to control or include spatial autocorrelation. I am thinking of including MEM and perhaps AEM variables simply as co-variables added to the explanatory environmental variables in the multivariate GLM. Is this a step I will regret later on? Is this ok? A second quick wondering: common GLM analyzes are carried out as a series of nested models in which we exclude variables from an initial full model based on anovas/AIC. I suppose this is also true for multivariate GLM. Is it? Can I compare successive models using the same approach used in common GLM? Thanks in advance for any thoughts, All the best, Alexandre
[ECOLOG-L] Which kind of species are logged?
Dear friends, Sorry for the timing of the posting, right before the hollidays. I would like to ask if any of you have ever read any reference to the successional status of the tree species preferentially cut by the timber logging industry. Are most of them pioneers, long-lived pioneers, of mature-climax species? Any help will do. Sincerely, Alexandre
[ECOLOG-L] Thoughts on the interpretation of Mass Effects x Neutral Dynamics?
Dear friends, I am thinking on the interpretation of the results of the variation partitioning of community composition by means of RDA. Despite all drawbacks of the approach, it continues as an important tool to access the global effects of environmental factors and space on the variance of species abundances in communities. However, I think there are two somehow different interpretations of the results. I would like to know what do you think about it, in order to make it clearer. The more classic interpreation for significant pure environmental and pure spatial effects (the most common result) is that the environmental effects represent species sorting (SS) by abiotic factors (niche related) while the spatial effects represent dispersal limitation, possibly linked to neutral dynamics, aside non-measured abiotic factors. In his review of these results, however, Cottenie (2005, Ecology Letters) proposed a classification of matacommunities based on variation partitioning results, and interpreted significant pure environmental + pure spatial fractions as indicative of Species Sorting + Mass Efffects metacommunity dynamics. Do you know why would it not be indicative of Species Sorting + Neutral Dynamics? What would be the reasoning for the differentiation between Mass effects and Neutral Dynamics? My first thought was that the pure spatial component would be indicative of dispersal limitation effects. This would be nearer neutrality than mass effects, since mass effects represent the opposite of dispersal limitation, wright? There is an overflow of dispersal not limitation. Thank you very much in advance for any thoughts, All the best, Alexandre Dr. Alexandre F. Souza Professor Adjunto II, Departamento de Ecologia Universidade Federal do Rio Grande do Norte (UFRN) Caixa Postal 1524, Campus Universitario Lagoa Nova CEP 59078-970 http://www.docente.ufrn.br/alexsouza Curriculo: lattes.cnpq.br/7844758818522706
[ECOLOG-L] Post on My Question on How to Estimate Soil Humidity from topographic or texture data
Dear friends, I have got some replies to my question on how to estimate soil humidity from topographic or texture data. Here they are: ** If you are interested in how much moisture the soil can hold than soil texture is your best proxy.Soil humidity is hard to measure anyway, since it fluctuates throughout the year depending on rain and temperature. Soil texture, however, can tell you more about the water retaining properties of the soil. If you know Organic Matter and Texture you can estimate Field Capacity based on equation by Saxton and Rawls (2006). I have the paper saved somewhere on my computer, I can send it to you if you like. Hope this helps. * There are a number of reasonably well documented equations and soil moisture model (or Topographic Wetness Index, TWI) approaches for calculating soil moisture parameters using a raster digital elevation model. The calculations (which I'm guessing have evolved over the past few years) are simply a function of the slope and the contributing upslope area; with the basic understanding that, all things (such as land cover and soil catenas) being equal, the bottom of a long slope slope will tend to be wetter than the top of a short slope. Some of the older hydrological software addons to ArcGIS (TauDEM, etc...) even have Soil Moisture functions included. The simplistic notation is: LN (a/S) or the natural log of the Upslope Area divided by the slope (for all points\pixels within the DEM). H. Mitasova has been working on the geospatial estimation of such estimates (which can easily be built into scripted functions within packages such as ArcGIS with Spatial Analyst (as well as Open Source software such as GRASS)... For more see: http://naldc.nal.usda.gov/download/44278/PDF http://skagit.meas.ncsu.edu/~helena/gmslab/reports/CerlErosionTutorial/denix/Advanced/ErosionRep97/rep97.html http://hydrology.usu.edu/RRP/userdata/4/87/Appendices.pdf http://ks.water.usgs.gov/pubs/reports/wrir.99-4242.html http://journalofmaps.com/student/10_01_Hardy.pdf
[ECOLOG-L] Soil humidity from topographic data?
Dear friends, I have a number of soil-related variables, and need to estimate soil humidity. Would someone know how? We have collected topographic (elevation, slope, convexity), soil (texture and nutrients), and vegetation cover data from 85 5 x 5 m plots scattered through a sandy coastal vegetation complex in northeastern Brazil, which include sand dunes as well as latosols. We did not have enough funding or time to measure soil humidity in the filed, but our analyses would greatly benefit from an estimation of it. Do you have an idea on how to do that? Many thanks, Alexandre
[ECOLOG-L] Would You Review My Manuscript?
Hello, I have just finished a manuscript on subtropical forest tree strategies in Brazil, and would like to hear a second opinion. If my work's theme falls within your field of expertise, would you like to review it? I offer to review your work in return. I would like to submitt the manuscript by the end of august, if possible. Find below the manuscript title and abstract, for you to have a better idea on what it is about. Best wishes, Alexandre Title In the lack of extreme pioneers: strategies, trait relationships, and ecological groups in subtropical trees Abstract We tested the general hypotheses (1) of the existence and nature of ecological strategies, (2) of the existence of a causal structure of traits that underlie such strategies, and (3) that species form discontinuities along trait variation gradients that form ecological groups. Data were collected in subtropical mixed conifer-hardwood forests in southern Brazilian. Leaf, size, and demographic traits were independent of the studied species’ phylogeny. Eleven tree trait variables measured for 66 large tree species were reduced to four principal components. The first axis had positive component loadings for mean growth, growth95, stem slenderness, leaf length, SLA, and a negative loading for mortality. The second axis had positive loadings for maximum height, crown depth, and wood density, and a negative loading for mortality. Seed size and seed dispersal mode appeared independently on the third and fourth axes, respectively. A path model fit the correlation structure between the trait and demographic variables. The model reinforced the two main ecological axes uncovered by the PCA because relationships between variables were segregated into two main sets of paths. The cluster analyses provided evidence of meaningful ecological types within the space of trait variance. Non-hierarchical k-means groups were more clearly and strongly related with the resource capture and height gradients depicted in the PCA than the groups formed by the hierarchical cluster analysis. We propose the recognition of seven ecological species groups in the studied forests. The lack of growth-mortality and wood density-stem slenderness trade-offs result from the fact that our studied forests lack extreme pioneers and have an over-representation of slow-growing hard-wood species. This means that the fastest-growing species do not grow so fast as to deeply incur in the costs of fast-growth and light wood, but enough to have the benefits of increased body size. Trait relationships and ecological groups may provide a useful approach for more realistically representing large and diverse sets of tree species in forest ecosystem models. Dr. Alexandre F. Souza Professor Adjunto II Departamento de Botanica, Ecologia e Zoologia Universidade Federal do Rio Grande do Norte (UFRN) http://www.docente.ufrn.br/alexsouza Curriculo: lattes.cnpq.br/7844758818522706
[ECOLOG-L] How to process 16-mm fisheye photographs
Dear ecofriends, We are studying the diversity and structure of sandy coastal vegetation in northeastern Brazil. This vegetation type is very complex and within a few hundred meters can change from forest to shrubland or just herbaceous cover. We took 16-mm fisheye photographs of 85 plots aiming at extracting canopy openess from them. The problem is that our lens is not hemispherical 8-mm. I would like to know whether we could extract more detailed information about sunfleck density, suntpath, diffuse radiation etc, that programs like GLA provide. These softwares work for hemispherical photographs. Can they be adjusted for 16-mm photographs? Perhaps through specific parameters? Many thanks in advance for any ideas. All the best, Alexandre Dr. Alexandre F. Souza Departamento de Botânica, Ecologia e Zoologia Universidade Federal do Rio Grande do Norte (UFRN) http://www.docente.ufrn.br/alexsouza Currículo Lattes: lattes.cnpq.br/7844758818522706
[ECOLOG-L] Post on Sand Erosion Measurement
Dear Ecofriends, Here are the answers I got about measuring sand erosion/deposition in coastal dunes. In summary I believe a combination of two approaches will suffice: 1 - Weighting sand deposition in plastic pots buried to sand surface 2 - Measuring the height of sand deposition against flat and graded surfaces like wood or aluminum rectangles. I will probably place one pot and one rectangle in each permanent plot and then model the resulting sand deposition surface in both grams and cm. Thank you for the quick response. All the best, Alexandre *** Hi Alexandre, I have heard that a cheap way to measure sand accumulation rates is to use 'sand traps'. These are plastic bottles of known volume placed level with the top of the sand. You can measure the rate of sand capture. * I don't know too much about it myself, but I have a few papers you could skim to get started. Zobeck is a good name to look up, as I believe he does a ton of that kind of work. Jason Field has also done a decent amount in an ecological context. I attached one from each. Best of luck! *** Dear Alexandre: I worked on demography of an herbaceous plant on the dunes around Lake Michigan, for my thesis many years ago. I used small aluminum tags to mark my plants. The tags were cut from soft drink cans with tin snips and numbered using a set of dies that I pounded into the top tab. The tags had a top tab region (numbered) which was bent at right angles to the main pointed stem. The stem was also bent into an angle of about 90 degrees long the vertical axis, to give it more strength as it was inserted into the ground. I found that, if I placed this tag at my plants with the bent flap right against the sand surface, that when I returned, months later, I could excavate the tag (which was marked by the presence of the plant that had continued to grow up through the accumulating sand). I could then measure the distance below the surface and determine how much sand deposition had occurred at the base of the plant in that interval. In my case, much of the sand deposition was due to ice along the edges of the lake, pushing sand up during the winter. I suppose if the dynamic is mostly deflation, this might not work very well. Good luck. Dr. Alexandre F. Souza Departamento de Botânica, Ecologia e Zoologia Universidade Federal do Rio Grande do Norte (UFRN) http://www.docente.ufrn.br/alexsouza Currículo Lattes: lattes.cnpq.br/7844758818522706
[ECOLOG-L] Question on Sand Dynamics on Dune Ecosystems
Dear friends, I am beginning to research vegetation ecology on coastal sandy dune ecosystems in northeastern Brazil and would like to ask you about ideas on how to measure the disturbance regime of such habitats. In our region dunes are fixed by shrubs, sometimes by trees, sometimes herbs only. I guess that the main disturbance factors are burial by sand transportation and drought. Do anyone know how to measure sand transportation by wind? Any other ideas or references? Thank you very much in advance, Alexandre Dr. Alexandre F. Souza Departamento de Botânica, Ecologia e Zoologia Universidade Federal do Rio Grande do Norte (UFRN) http://www.docente.ufrn.br/alexsouza Currículo Lattes: lattes.cnpq.br/7844758818522706
[ECOLOG-L] CSR x Forest Functional Groups
Hello, Regarding forest plant functional groups. Grime's CSR theory remains one of the most robust scheme of plant adaptation and formation of functional groups. Its attractiveness lies in that it is not an emergent grouping of properties, but derives from an ecological and evolutive theory. However, it is mostly applied and developed to herbaceous plants. Do anyone knows of any applications to trees/forest vegetation? I see potential problems with the utilisation of the concepts of competition as currently applied by the theory and how it is used by forest ecologists, who attribute it to shade-tolerant species (stress-tolerant in the CSR theory) which win competition over pioneer species (competitors in the CSR theory) in the long run. All the best, Alexandre
[ECOLOG-L] Post on: Reference on Ecological Management
Dear friends, I have recently posted a request for indications of books on ecological management. Here you have the list of suggestion I received. Savory's Holistic Management Meffe et al Ecosystem Management, Island Press reating a Forestry for the 21st Century: The science of ecosystem management. Kathryn Kohn and JF Franklin. Maintaining biodiversity in forest ecosystems Malcolm Hunter Primer-Conservation-Biology-Fourth-Edition (not really about management) Stankey, G.H., Clark, R.N., Bormann, B.T., 2005. Adaptive management of natural resources: theory, concepts, and management institutions. United States Department of Agriculture, Washington, D.C. This last one is available on the internet. All the best, Alexandre
[ECOLOG-L] Reference on Ecological Management
Dear friends, I am giving an undergraduate course on Natural Resource Management for the Ecology carreer nex semester. Despite extensive contact with conservation biology, I am not much familiar with management textbooks. Do anyone knows of good introductory textbooks on ecological management of plants, animals and ecosystems? Thank you in advance, Alexandre
[ECOLOG-L] Invitation to Review Manuscript on Regional Floristic Relationships in South America - Thank you
Dear friends, Just 24 hours ago I wrote to you asking if some of you would be willing to review a manuscript that I have just finished on floristic relationships between subtropical forests in Brazil. I would like to thank you very much for the many replies I received! I would like to consider my invitation fulfilled and ask for no new answers, in order to keep the whole thing manageable, considering that I offered to back-review a manuscript for each review I got. Best wishes, Alexandre Dr. Alexandre F. Souza Departamento de Botânica, Ecologia e Zoologia Universidade Federal do Rio Grande do Norte (UFRN) http://www.docente.ufrn.br/alexsouza Currículo Lattes: lattes.cnpq.br/7844758818522706
[ECOLOG-L] Invitation to Review Manuscript on Regional Floristic Relationships in South America
Dear friends, I have just finished a manuscript on floristic relationships between subtropical forests in Brazil. I wonder if some of you would like to review it and give me some feedback. I offer to review your manuscript in return. Sincerely, Alexandre Dr. Alexandre F. Souza Departamento de Botânica, Ecologia e Zoologia Universidade Federal do Rio Grande do Norte (UFRN) http://www.docente.ufrn.br/alexsouza Currículo Lattes: lattes.cnpq.br/7844758818522706