Some Comments on Spatial and Temporal Scale in Ecological Models Don DeAngelis, Mark Grismer, and Laurel Saito Acknowledgments Examples used here are due to • Lou Gross and his group (U. of Tennessee) • Wolf Mooij (Netherlands Institute of Ecology) • and others. Books on the Importance of Scale in Ecological Systems Giller, P. S., A. G. Hildrew, and D. G. Raffaelli (eds.) Aquatic Ecology: Scale, Pattern, and Process. Blackwell Scientific 1994 Harris, G. P., Phytoplankton Ecology: Structure, Function, and Fluctuation. Chapman and Hall 1986 Karlson, R. H. Dynamics of Coral Communities. Kluwer Academic 1999 MacCall, A. D. Dynamic Geography of Marine Fish Populations. Washington Sea Grant Program 1990 Mullin, M. M. Webs and Scales. Washington Sea Grant Program 1993 Okubo, A. Diffusion and Ecological Problems: Mathematical Models. Springer Verlag 1980 Seuront, L., and P. G. Strutton. Handbook of Scaling Methods in Aquatic Ecology. CRC Press 2004 Steele, J. H. (ed.) Spatial Pattern in Plankton Communities. Plenum Press 1978 What Ecological Factors Determine Natural Spatial Scales? We can differentiate spatial extent and spatial resolution. Spatial extent largely depends on the amount of area needed to encompass the ecological system (e.g., population or community) under consideration. This is the approximate range of the snail kite in Florida Reality Range of Florida panther population Extent of Geographic Range of Cape Sable Seaside Sparrow A much smaller area is needed to model the entire population of Cape Sable seaside sparrows. Even smaller areas encompass major subpopulations (outlined in red), which may be considered as individual ‘systems’ over short time periods. What Ecological Factors Determine Natural Spatial Scales? Spatial resolution necessary to accurately describe a process is determined by the scale over which some process or pattern in space varies according to some criterion (Levin and Pacala 1997). So we choose a scale of resolution small enough to capture that variation. What Ecological Factors Determine Natural Spatial Scales? Some of the main causes of such variation in a pattern or process that must be considered in modeling are: • Scale of variation in the abiotic conditions, including abiotc disturbances • Scale of variation in underlying biotic conditions • Individual organism size • Scale of ecological patterns emerging from model interactions Variation in Abiotic Conditions Example: We are attempting to model the spatial dynamics of numbers and biomass of small fishes across the entire Everglades. It is necessary to choose an appropriate spatial scale, or spatial ‘cell’ size to represent the elevation gradient across the Everglades, which will determine the spatio-temporal pattern of flooding and drying. The Modeled Functional Group The small fishes (Gambusia, mollies, killifishes, etc.) constitute one of the most important components of the Everglades system. This 'functional group' provides much of the prey-base for wading birds and other higher trophic level species. Sailfin molly (Poecilia latipinna) Least killifish (Heterandria formosa) Variation in Abiotic Conditions How fine do we need to go in spatial scale? The South Florida Water Management Model (SFWMM) produces output of daily water depths at a 2 x 2 mile scale. However, it is possible, through algorithms incorporating additional knowledge of vegetation, etc. (see atlss.org website), to refine this to a finer scale. Tree Island : Skinner’s Camp Variation in Abiotic Conditions How fine do we need to go in spatial scale? A resolution of 500 x 500 m may capture much of the biologically important change in mean elevation (and thus water depth) across large-scale gradients of the Everglades landscape. (We have confidence in refinement to this scale, as we can use USGS elevation data collected roughly at this scale.) Slough to prairie transition Variation in water depths Wet 0 Dry SFWMM HRH Variation in Abiotic Conditions However, the 500 x 500 m resolution fails to capture essential fine-scale heterogeneity within the cell. This heterogeneity is in the form of alligator ponds, solution holes, etc., that may form dry-season refuges for fish, aquatic macroinvertebrates (crayfish), etc. Unit Cell Layout Pond areas assumed permanently wet, marsh areas periodically dry Variation in Abiotic Conditions How do we deal with that heterogeneity? It is impractical to try to model explicitly at a finer spatial scale of resolution. But is it really necessary to do so? Probably not. The internal pattern can be dealt with in a spatially implicit manner, by quantifying the fractions of cell area that are alligator holes, solution holes, etc., and that therefore can serve as dry season refuges. Their explicit locations are unimportant. Variation in Abiotic Conditions So we can ‘scale down’ by including finer scale information in a spatially implicit way. How do we ‘scale up’ the dynamics of the fish population to the whole landscape? We take into account processes of fish movement among cells. Landscape Layout and Movement White - movement from low water to high water areas Red - movement from high fish density to low density areas Variation in Abiotic Conditions Then the model for fish biomass and number dynamics can produce spatially explicit output on a 500 x 500 m scale across the landscape, on 5-day time steps over 31-year simulations (in this case comparing two scenarios). Average Fish density on February 15 from 1965 - 1995 Variation in Abiotic Conditions From this output, combined with other factors, such as the water depths that wading birds can utilize, we can estimate the habitat suitability (for reproduction) for long-legged and short-legged wading bird colonies across the Everglades, or for specific sub-regions for given years. SESI Output for Long-Legged Wading Birds in N. Taylor Slough: For 1993 In computing this, we introduced a further spatial scale, the wading birds’ foraging area. Wading Bird Nesting Colony Suitability is determined by foraging suitability of many 500 x 500 m pixels surrounding the colony The Wading Bird Foraging Index for a given pixel (potential site of a nesting colony) is determined by the ‘collective foraging suitability’ of the 500-m pixels in the ‘core’ area surrounding the colony pixel. The core foraging radius for long-legged and short-legged wading birds, ForagRadius, is currently 1.5 and 3.0 km, respectively. Black denotes Wading Bird colony Variation in Abiotic Conditions A scale of resolution of 500 x 500 m seems to be reasonable for describing the response of many Everglades biota to abiotic conditions, finer scale is used for the American crocodile, which responds to mangrove estuaries and short scale variations in depth. Fig 1 Open water Mangrove Dwarf Mangrove Marsh Nest sites R1 Fringe Road Small Landscape Non-habitat R2 R3 R4 R5 Variation in Underlying Biotic Conditions Biotic conditions, such as vegetation type (denotes by different colors in map), may also affect the dynamics of a population. In this case vegetation type is known to 30 x 30 m resolution, and so can be used explicitly in the model, even though water depth continues to be modeled at a resolution of 500 x 500. We have used this additional scale of resolution for vegetation in some specific cases, though not in general. So more than one spatial scale of variation can be used, depending on the availability of data at various scales of resolution and the importance of including it. Effect of Individual Body Size on Needed Scale of Resolution • In modeling competition of sessile benthic organisms (e.g. Robles and Descharnais, Ecology 2002) and terrestrial plants, organism size may dictate the scale of resolution. • This may apply as well to animal territories or home ranges Effect of Emergent Patterns on Needed Scale of Resolution Depending on the type of model, some spatial features of certain scales may emerge as a result of the interactions in a model. Two examples are: • Emergence of tree islands from models that combine hydrology, erosion/deposition, tree colonization, nutrient accumulation, to form tree islands. • Competition between species along an ecological gradient. Tree Islands in the Everglades Landscape Our model for tree island formation has a spatial scale of 10 x10 m Emergent Spatial Structure of Fish Community in the Everglades We are now attempting to model the spatial structure of communities of small fishes across elevation gradients. Fishes differ in their life history traits (fecundity/survival and movement ability are possible trade-offs). This suggests that there is some partitioning of space, at least on a temporary basis. An abstract model of fish competition in a spatio-temporally varying one-dimensional landscape attempts to capture that (DeAngelis, Trexler, and Loftus, CJFAS 2005) It is essential to choose a spatial resolution (200 m in this case) that allows emergence of patter in space. Elevation and Hydroperiod along Spatial Gradient CHARACTERISTICS OF TRANSECT ELEVATION AND HYDROPERIOD 2.5 2 ELEVATION: ARBITRARY UNITS 1.5 1 HYDROPERIOD: FRACTION OF YEAR FLOODED 0.5 0 1 3 5 7 9 11 13 15 17 19 SPATIAL CELL NUMBER 21 23 25 27 29 31 Hydrologic Driving Function in ODFISH HYDROPERIODS OF SPATIAL CELLS DEPTH OF WATER IN METERS 2 1.5 1 0.5 0 0 50 100 150 200 -0.5 -1 -1.5 -2 TIME IN DAYS 250 300 350 400 Biomass Distributions of Hypothetical Small Fish Species Populations on a Given Day Across an Elevation Gradient BIOMASSES (G/M2) 1.4 1.2 1 0.8 0.6 0.4 0.2 0 1 6 11 16 21 26 31 SPATIAL CELL NUMBER Spatial gradient represented by 31 cells, each 200 m in length Temporal Scales of Variation Temporal resolution in the model must be chosen on a fine enough scale to capture short time scale events that are important to population dynamics. In particular, rapid changes in water depth must be represented. Local drying events can happen within a few weeks, forcing snail kites to move. Wetland Dry Sudden Increases in Water Depth Can Interrupt Breeding Cycle of Cape Sable Seaside Sparrow Increase in water depth to 15 cm will cause any nesting to cease. It can commence if water depths decrease. The figure shows 2 possible nesting cyles of 45 days. USGS 30 DEPTH OF WATER IN CELL Nesting can start in a cell as soon as water depths decrease to 5 cm. Nesting can start 25 72 days 20 1 cycle 45 days 15 1 cycle 45 days 10 5 0 1 15 29 43 57 71 85 99 113 127 141 155 169 DAY OF REPRODUCTIVE SEASON Water Depth May Rapidly Become Unfavorable for Wading Birds 60 WATER DEPTH IN CELL At any given time during the breeding season, only pixels that are in certain water depth ranges are usable and contribute to the foraging suitability of the core area of the colony. WaterDepth 50 40 30 Range 5 to 35 cm 20 10 Long-legged wading birds: MINDEPTH to MAXDEPTH: Currently 5 to 35 cm DAY OF REPRODUCTIVE SEASON Short-legged wading birds: MINDEPTH to MAXDEPTH: Currently 1 to 20 cm 148 141 134 127 120 113 106 99 92 85 78 71 64 57 50 43 36 29 22 15 8 1 0 Temporal Scales of Variation Other important processes operate on much longer time scales. This includes vegetation succession processes, which are modeled in a rule-based model. e Years Since Last Hydroperiod Disturbance Herbaceous/Forested Ve getation Succession 3 Dry Prairie [29, 39] 30–60 3 Dry Prairie [29] 30–60 Graminoid Dry Prairie EC [39] 20–50 3 {.53} Mesic–Hydric Forest [16] {.22} Slash Pine Woodland [25] {.16} Gallberry/ Palmetto CG [30] 30–60 {.08} Ham mock [2] 10–45 {0.996} Muhly Grass Marsh[45] {0.004} Sparsely Wooded Wet Prairie CG [52] 60–120 {0.996} Muhly Grass Marsh[45] {0.004} Sparsely Wooded Wet Prairie CG [52] 60–120 {0.996} Muhly Grass Marsh[45] {0.004} Sparsely Wooded Wet Prairie CG [52] 60–120 {.25} Graminoid Marsh CG [42] 120–270 {.7} Cladium [43] 130–330 {.03} Eleocharis [44] 150–300 {.03} Typha [46] 180–280 Forb Emergent Marsh [56] 230–360 {.25} Graminoid Marsh CG [42] 120–270 {.7} Cladium [43] 130–330 {.03} Eleocharis [44] 150–300 {.03} Typha [46] 180–280 Forb Emergent Marsh [56] 230–360 {.25} Graminoid Marsh CG [42] 120–270 {.7} Cladium [43] 130–330 {.03} Eleocharis [44] 150–300 {.03} Typha [46] 180–280 Forb Emergent Marsh [56] 230–360 {.01} Live Oak [5] 0–60 Decid. Shrub [37] 110–320 10 [16, 25, 30] 30–60 Trop. Hammock [2] 10–45 Live Oak [5] 0–60 0 [16, 25, 30] 30–60 10 [2] 10–45 [5] 0–60 {.91} Swamp Forest CG [17] 120–290 {.09} Brd Lvd/Mixed Evergreen Shrub [28] 120–150 Forb Emergent Marsh [56] 230–360 Float ing Leaved Veg. [57] 330–360 {.77} Decid/Trop. Swamp Forest [3] 60–180 {.23} Brd Lvd/Mixed Evergreen Shrub [28] 120–150 Decid. Shrub [37] 110–320 Swamp Forest CG [17] 120–290 Float ing Leaved Veg. [57] 330–360 0 [16, 25, 30] 30–60 [2] 10–45 20 Decid/Trop. Swamp Forest [3] 60–180 Decid. Shrub [37] 110–320 [5] 0–60 0 40 {.55} Trop. Hammock [2] 10–45 {.04} Live Oak [5] 0–60 {.4} Xeric–Mesic Live Oak EC [4] 0–30 Float ing Leaved Veg. [57] 330–360 Swamp Forest CG [17] 120–290 {.35} Bay/Gum/Cypress EC [6] 60–160 {.65} Swamp Forest [3] 60–180 Float ing Leaved Veg. [57] 330–360 Decid. Shrub [37] 110–320 Swamp Forest CG [17] 120–290 Cypress Forest CG [18] 200–340 Decid/Trop. Swamp Forest [3] 60–180 50 0 30 60 90 120 No Data 150 180 210 240 270 300 330 Conclusions The empirical study suggests that management models of fishes must incorporate rapid recovery from drought events. Some species may experience greater population growth rates in short-hydroperiod habitats, although ultimate population sizes may be limited by periodic dry down, site productivity, and other factors. Exploratory models, such as ODFISH, may help understand the factors underlying rapid recovery and help improve management models, such as ALFISH.