This document is an initial attempt to characterize and visualize the wintering waterfowl data collected during winter from 1964 to present at Tennessee National Wildlife Refuge (TNNWR). We focus on ducks and geese.

The specific objectives of this document are:

  1. To characterize the timing and frequency of surveys during winters 1964-2014.

  2. Conduct some basic quality control and assessment of the data via built-in R functions and visualization.

  3. Explore species composition of the data to identify those species or groups amenable to further analysis.

  4. Create preliminary phenology charts for select species/groups and explore changes in phenology over time.

Study area

This is a crude map of the refuge with polygons labeled by Unit Code (click on a polygon). I have not attempted to reconcile the subunit codes from the shapefiles with those found in the survey data.

Data exploration

Survey history

We first visualize the distribution of surveys over time. Each point represents a survey and the number in the upper left is the number of surveys in that calendar year. Surveys are increasingly less frequent(from ~ weekly to ~ biweekly) and the survey window is becoming increasingly narrow (most recently November through February).

Species composition and count characteristics

Over the survey period (1964 - 2014), TNNWR has conducted 1120 surveys with complete count data; 901 occurred between Sepember and April, inclusive. To get a feel for the species most likely to contribute meaningfully to an analysis, the following table summarizes these 901 “winter” surveys:

Table 2. Number of ‘winter’ surveys (of 901) with detection and median and maximum count totals.
Species # surveys present Median count Maximum count
Mallard 889 29125 259542
Wood Duck 871 3150 9300
Canada Goose 860 7200 131000
American Black Duck 797 2995 59222
American Wigeon 788 3077 83000
Gadwall 734 1500 33304
Northern Pintail 728 1100 18000
Ring-Necked Duck 686 900 16020
Bufflehead 672 70 2000
Hooded Merganser 638 120 5000
Lesser Snow Goose 609 24 4300
Scaup 584 200 8300
Canvasback 568 100 6200
Northern Shoveler 555 25 14000
Ruddy Duck 526 10 3310
American Green-Winged Teal 476 20 12968
Common Goldeneye 441 0 2076
Redhead 415 0 1600
Blue-Winged Teal 343 0 6510
Common Merganser 316 0 7000
Greater White-Fronted Goose 200 0 2390
Red-Breasted Merganser 130 0 600
Ross’ Goose 50 0 21
Unidentified Duck 43 0 20680
Cackling Goose 25 0 900
Brant 18 0 12
Barnacle Goose 8 0 1
Oldsquaw 8 0 10
Cinnamon Teal 2 0 1
Black Scoter 1 0 6
Eurasian Wigeon 1 0 1
Surf Scoter 1 0 20

A more complete picture of the distribution of counts for those species present on at least 25% (> 225) of the winter surveys is available:

Species phenology

No glaring remaining errors are apparent, but it’s really difficult to diagnose possible mistaken entries when counts span such a wide range. So we may have to focus on particular species to look for anomalies. We’ll focus henceforth on the 901 October through March surveys and the 20 species present on \(\geq\) 25% of the surveys. NOTE: we extend the time period of interest to September and April for the figures because, although September and April surveys essentially ceased in 1982 and 1978, respectively, we’d like to know that what we see in October - March is a fair characterization/continuation of the fall/winter period when we subsequently drop September and March for analysis.

We identified six duck species with adequate data to evaluate changes in phenology from 1964-2014: four species of dabbling ducks (Mallard, American Black Duck, American Wigeon, and Northern Pintail) and two species of diving ducks (Ring-necked Duck and Canvasback). For these six species, we create two figures to identify problems: the first (top) illustrates the temporal patterns of refuge-wide counts (1) within a winter and (2) over the 51 years of surveys. Survey year is represented by the color of each line: early surveys are depicted by blue lines transitioning to red lines for more recent surveys. This plot gives an overall impression of the data for each species. The second figure (bottom) is more helpful for identifying potential problems as it separates the survey data for every winter.

Mallard

American Black Duck

American Wigeon

Northern Pintail

Ring-Necked Duck

Canvasback