BCI Canopy Database

The Barro Colorado 50-ha plot is a full census of trees and saplings down to 1 cm stem diameter (Hubbell and Foster 1985, Condit 1998). The census has been completed seven times since 1981, producing growth and mortality data of > 300 species. To support the demographic information, a survey of canopy structure was carried out annually across the 50 ha, recording vegetation density in several vertical layers at a horizontal scale of 5 m. The canopy survey can be used to produce maps of vegetation density, showing where there are gaps in the canopy at every depth. The three-dimensional map of canopy density was intended to help understand how tree growth and recruitment responds to canopy openness (Welden et al. 1991, Rüger et al. 2009).

Our purpose here is to document the methods employed in the canopy survey, describe the data, and provide links to permanent repositories where the data are publicly available. Since the complete census results over 30 years is also publicly available (Condit et al. 2012), other scientists thus have access to both demographic and canopy data. Detailed descriptions of the 50-ha plot methods and data are provided by Condit (1998) and Condit et al. (2013).

Tables are permanently archived at this Smithsonian archive URL. There is a single zip file including 23 tables in csv format. Methods are described in detail here and the tables illustrated and described.

Methods

Field technicians visited every one of 20301 corner posts of the 5x5 m grid across the plot (Condit [1998] desribes the grid). At each point, the density of vegetation in several discrete layers directly above was recorded. This census was done in all years from 1983-1993, 1995-1996, and 2003-2012. A simpler and quicker method was used in 2000-2001 due to a shortage of funds. The census was omitted in 1994, 1997-1999, and 2002-2003, and this was again solely due to logistics and financing.

Method 1: Vegetation layers

Call the coordinates of the corner post of one of the 5x5 m cells \( x, y \) . Six layers of vegetation directly above \( x, y \) were defined: \( [0,1); [1,2); [2,5); [5,10); [10,20); [20,30); \) and \( \ge30 \) m above the ground. This assigns a vertical coordinate, \( z \), with \( z=1 \) meaning height 0-1 m up to \( z=6 \) meaning the top category, height \( \ge30 \) m. A single datum is the presence of absence of vegetation at coordinates \( x, y, z\), with presence scored as 100 and absence as 0.

The exact method of determining whether vegetation was present or absent at coordinates \( x, y, z\) changed slightly after 1997. During the early period, 1983-1997, an imaginary vertical line was drawn upward from the ground past the tops of the trees. Verticality was assisted by using a plastic pole 7 m long with a leveling bubble attached, and height was estimated using a focusing range-finder. Vegetation was considered present at \( x, y, z\) if any leaves or branches crossed the imaginary vertical line within \( z \) 's height limits. Starting in 2003, this was changed to consider the entire 5x5 m cell north and east of coordinates \( x, y \). This meant that \( x, y, z\) defined a voxel, meaning a three-dimensional pixel. In each voxel, vegetation was scored as present if > 10% of the volume had branches and leaves; any less was scored as an absence.

Because of this difference, there are 20,301 sets of coordinates in 1983-2001 data, because all posts on plot boundaries could be included. But since 2003, the estimate could not be taken at the east and north boundaries, since the 5x5 m cell north and east of those points would lie outside the 50-ha plot. Without those boundaries, there are 20,000 sets of coordinates.

There was an additional small difference. Up to 1997, the two lowest height categories were collapsed into one, from 0-2 m. So there were five layers \( z \) in early censuses and six in later.

Method 2: Canopy height

In 2000 and 2001, abbreviated data were collected. At each \( x, y \) on the 5x5 m grid, the height of the highest vegetation vertically above was all that was recorded. The score was continuous, simply the estimated height based on the range-finder, though only integers were used and, in fact, the maximum value noted was usually exactly 40 m.

Data

Vegetation layers

All results from the estimates of vegetation density in layers are presented here in 23 tables, one per year. All tables have precisely the same format, with one row providing results for one location with coordinates \( x, y \). Thus tables for years prior to 1996 have 20,301 rows, but only 20,000 rows since 2003. Every row includes a column to indicate presence or absence of vegetation in every height category, always 0 or 100. Additional columns give the date and observers who collected data. For the earlier censuses, the latter two columns are empty because those data were not saved.

Here are the first 15 rows of data from 1985:

	x	y	date	ht0_2	ht0_1	ht1_2	ht2_5	ht5_10	ht10_20	ht20_30	ht30_	observers
1	0	0		100			0	100	100	100	0
2	0	5		100			0	100	100	100	0
3	0	10		100			0	100	100	100	100
4	0	15		100			0	0	100	100	100
5	5	0		0			0	0	100	100	0
6	5	5		100			0	0	100	100	0
7	5	10		100			0	100	0	100	0
8	5	15		100			0	0	100	0	0
9	10	0		100			0	100	100	100	0
10	10	5		100			0	100	100	100	0
11	10	10		0			0	0	100	0	0
12	10	15		100			0	100	100	0	0
13	15	0		0			0	100	100	100	0
14	15	5		100			0	100	100	0	0
15	15	10		100			0	100	100	0	0

and here from 2003:

	x	y	date	ht0_2	ht0_1	ht1_2	ht2_5	ht5_10	ht10_20	ht20_30	ht30_	observers
1	2	2	2003-08-05		100	100	100	100	100	100	100	Diogenes Ibarra,Irene Torres Tejada
2	2	7	2003-08-05		100	100	100	100	100	100	100	Diogenes Ibarra,Irene Torres Tejada
3	2	12	2003-08-05		100	100	100	100	100	0	100	Diogenes Ibarra,Irene Torres Tejada
4	2	17	2003-08-05		100	100	100	100	100	100	100	Diogenes Ibarra,Irene Torres Tejada
5	7	2	2003-08-05		100	100	100	0	0	100	100	Diogenes Ibarra,Irene Torres Tejada
6	7	7	2003-08-05		100	100	100	100	100	100	100	Diogenes Ibarra,Irene Torres Tejada
7	7	12	2003-08-05		100	100	100	100	100	100	100	Irene Torres Tejada,Diogenes Ibarra
8	7	17	2003-08-05		100	100	100	100	100	0	0	Diogenes Ibarra,Irene Torres Tejada
9	12	2	2003-08-05		100	100	100	100	100	0	100	Diogenes Ibarra,Irene Torres Tejada
10	12	7	2003-08-05		100	100	100	100	100	100	100	Diogenes Ibarra,Irene Torres Tejada
11	12	12	2003-08-05		100	100	100	100	100	0	100	Diogenes Ibarra,Irene Torres Tejada
12	12	17	2003-08-05		100	100	100	100	100	0	100	Diogenes Ibarra,Irene Torres Tejada
13	17	2	2003-08-05		100	100	100	0	0	0	0	Diogenes Ibarra,Irene Torres Tejada
14	17	7	2003-08-05		100	100	100	100	100	0	100	Irene Torres Tejada,Diogenes Ibarra
15	17	12	2003-08-05		100	100	100	100	100	0	100	Diogenes Ibarra,Irene Torres Tejada

The coordinates in the data tables from 2003 on are always displaced 2.5 m east and 2.5 m north of the corner posts. This reflects the shift in methods in those later censuses, when the vegetation scores referred to the entire 5x5 m cell. Thus, the first row in the 1983 table has \( x=0, y=0 \), since vegetation was assessed only at the vertical line extending above that point. But in the 2003 table, \( x=2.5, y=2.5 \), the midpoint of the 5x5 m cell, because the assessment covered the entire cell from \( x=0 \) to 5 and \( y=0 \) to 5.

Because of the change in methods after 1997, the vegetation density measures are not comparable between early and late period, with many more scores of 100 using the later method. This is evident in the first few rows above, and with a quantitative comparison of the number of sites with canopy density \( = 100 \).

	count_100
1983	4807
2003	15261

Use of canopy layers in estimating the light environment at every point in the 50-ha plot is presented in Condit (2013).

Canopy height

The estimates of height to the top of the canopy are collected in a single table, with \( x,y \) coordinates and height to the top of the canopy (meters) in two years. There are 20301 rows. The first 15 rows are illustrated here:

	x	y	Ht2000	Ht2001
1	0	0	28	28
2	0	5	40	26
3	0	10	40	33
4	0	15	0	35
5	5	0	30	27
6	5	5	29	28
7	5	10	32	28
8	5	15	32	27
9	10	0	22	20
10	10	5	18	18
11	10	10	17	29
12	10	15	36	28
13	15	0	31	27
14	15	5	31	30
15	15	10	35	30

Notice that though height was intended to be collected as a continuous measure, rounded to the nearest meter, there is a conspicuous tendency for observers to favor multiples of 5 m. The table can be downloaded in zipped ascii format from this Smithsonian archive.