2  Herbarium Specimen References

This is a simple test of whether the use of a Large Language Model (e.g., Claude 3) can help organize botanical research materials. Here, we’re deconstructing the herbarium specimen data. The goal is to add visualizations that make the terse text into something that is more usable.

2.1 Source

The starting materials are two pages (252-253) from C. den Hartog’s 1970 book, The Sea-grasses of the World, North-Holland Publishing Company, Amsterdam.

The distribution of Halophila ovata is given in the text.

den Hartog, page 252

den Hartog, page 253

2.2 Initialize Libraries and Data

Show the code 
## Standard Libraries
library(tidyverse)
library(gt)
library(gtExtras)

## Map Libraries
library(sitemaps)
library(ggmap)

## Useful Libraries
library(lubridate)
library(stringr)

## LLM Libraries
library(claudeR)

## Test if Google Key is registered
if (!has_google_key()){

  ## Register the Google Maps API Key.
  register_google(key = My_Key, account_type = "standard")
  } ## end Google Key test

claude_key <- Sys.getenv("ANTHROPIC_API_KEY")

## Use two functions from sitemaps to initialize columneters
column <- site_styles()
hide  <- site_google_hides()

2.3 Start with the OCR Text

This is the manually cleaned-up results from the OCR text produced by Adobe Acrobat. There are still some errors in the text.

Show the code 
"KENYA. Lamu, mud-fiats north-west of the town, exposed at low tides, locally extremely common and dense,female fl., fr., 1-7-1965, Mrs. F.M. Isaac A 21 (L). Mokowe, mud-flats in front of mangrove swamp, 30-6-1965, Mrs. F. M. Isaac A 100 (L). - Gazi, south of Ukunda, on mud-flats, uncovered at exceptionally low spring tides only, 11-12-1965, Mrs. F. M. Isaac A 116 (L).
INDIA. Pamban, October 1922, M. O. P. Iyengar 133 B (BM). 
THAILAND. Rawi, Satut, growing in sand, just exposed at low tide, together with Halodule uninervis, 13-1-1928, A. F. G. Kerr 14035 (BM).
HoNG KoNG. Cowloon Bay, on sandy bottom at ca. 2 m depth, Harland 282 (K).
PHILIPPINES. Luzon: Manila Bay, fr., May 1892, A. Loher 1595 (K, C, P); fl., fr., April 1905, E. D. Merrill 4112 (P, K, L, SYD); Malate Beach, fr., 8-5-1910, C. B. Robinson 9899 (C, P, L, BO); fr., March 1912, E. D. Merrill 1098 (C, U, WRSL). Albay Province, Albay Gulf, near Lubas Point, on sandy mud with Halimeda, below low-tide mark, 18-5-1958, M. Doty &, G. T. Velasquez 16848 (L, HAW). - Cebu: Mactan Island, together with Halophila ovalis and Halodule uninervis, January-February
1875, Moseley (BM).
MALAYAN PENINSULA. Singapore: Blakang Mati, 1892, Ridley 3780 (C, BM,
SING). Ponssol, on sandy mud, exposed at low tide, 5-2-1914, Holttum (SING). Labrador, 21-3-1928, Holttum (SING). Tandjong Behala Kuda, Pulau Pawai, sandy bottom, at half-tide level, growing with Halodule uninervis, 14-3-1950, J. Sinclair 38894 (SING, L). Pasir Laba, in a pure stand in a bed of Halophila ovalis, in the low eulittoral on mud, 27-12-1965, H. M. Burkill 3897 (L).
INDONESIA. Riouw Archipelago: Pulau Penyangat, 8-11-1930, Md. Nur 24603
(SING). - Java: Pasir Poetih, 27-8-1961, H. F. Neubauer 1601 (L). - Lesser Sunda Islands: Flores, near Bari, 12-7-1847, Zollinger 3334 (Type of Lemnopsis minor, P, L, BM, U); Bima, together with narrow-leaved Halodule uninervis, Zollinger 3431 (P). - Celebes: Salayer, Zollinger (L). - Moluccas: Amboina., October 1874, O. Beccari 11828 (L).
MARIANNE ISLANDS. Exact locality unknown, fl., Gaudichaud (Type, P, L). - Guam, Cocos Lagoon, Summer 1952, K. O. Emery 10699 (AHFH). Mana Bay
opposite Ypan, on east coast, inside barrier reef, in lagoon at ½-2 m depth, on sandy silty bottom, with scattered Enhalus acoroides, Caulerpa, Padina and other algae, 15-4-1962, B. O. Stone 4063 (L, HAW). - Saipan Island, in lagoon, 22-8-1952, N. 0. Bunker & R. Ocampo (AHFH).
WESTERN AusTRALIA. Carnarvon, Babbage Island, in creek through the mangrove swamp, 16-9-1967, 0. den Hartog 560 (L); on tidal flats and in creeks in front of the mangrove-swamp SW.of Carnarvon, 17-9-1967, 0. den Hartog 581 (L). Port Hedland, in small pools in the intertidal belt, on muddy bottom, 18-9-1967, O. den Hartog 588 (L).
QUEENSLAND. Thursday Island, a few specimens at low-water mark, 12-11-1967, 0. den Hartog 1016 (L); in very shallow, muddy littoral pools along the NE. side of the island, 12-11-1967, 0. den Hartog 1026 (L). Townsville, between Cape Pallarenda and Shelly Beach, in soft mud at half-tide level, 5-10-1967, 0. den Hartog 743 (L); Townsville, Cleveland Bay, flats of Crocodile Creek, 13-10-1967, 0. den Hartog 797 (L).
NEW CALEDONIA. Noumea, on the mud-flats, just below low-water mark, er and !j! fl., fr., April 1869, B. Balansa 1525 (P); Anse Vata near Noumea, in 1-2 m of water, 5-2-1960, M. Angot (L, HAW). Ouano (La Foa District), abundant, er fl., 20-1-1961, H. S. McKee 8229 (K). "
[1] "KENYA. Lamu, mud-fiats north-west of the town, exposed at low tides, locally extremely common and dense,female fl., fr., 1-7-1965, Mrs. F.M. Isaac A 21 (L). Mokowe, mud-flats in front of mangrove swamp, 30-6-1965, Mrs. F. M. Isaac A 100 (L). - Gazi, south of Ukunda, on mud-flats, uncovered at exceptionally low spring tides only, 11-12-1965, Mrs. F. M. Isaac A 116 (L).\nINDIA. Pamban, October 1922, M. O. P. Iyengar 133 B (BM). \nTHAILAND. Rawi, Satut, growing in sand, just exposed at low tide, together with Halodule uninervis, 13-1-1928, A. F. G. Kerr 14035 (BM).\nHoNG KoNG. Cowloon Bay, on sandy bottom at ca. 2 m depth, Harland 282 (K).\nPHILIPPINES. Luzon: Manila Bay, fr., May 1892, A. Loher 1595 (K, C, P); fl., fr., April 1905, E. D. Merrill 4112 (P, K, L, SYD); Malate Beach, fr., 8-5-1910, C. B. Robinson 9899 (C, P, L, BO); fr., March 1912, E. D. Merrill 1098 (C, U, WRSL). Albay Province, Albay Gulf, near Lubas Point, on sandy mud with Halimeda, below low-tide mark, 18-5-1958, M. Doty &, G. T. Velasquez 16848 (L, HAW). - Cebu: Mactan Island, together with Halophila ovalis and Halodule uninervis, January-February\n1875, Moseley (BM).\nMALAYAN PENINSULA. Singapore: Blakang Mati, 1892, Ridley 3780 (C, BM,\nSING). Ponssol, on sandy mud, exposed at low tide, 5-2-1914, Holttum (SING). Labrador, 21-3-1928, Holttum (SING). Tandjong Behala Kuda, Pulau Pawai, sandy bottom, at half-tide level, growing with Halodule uninervis, 14-3-1950, J. Sinclair 38894 (SING, L). Pasir Laba, in a pure stand in a bed of Halophila ovalis, in the low eulittoral on mud, 27-12-1965, H. M. Burkill 3897 (L).\nINDONESIA. Riouw Archipelago: Pulau Penyangat, 8-11-1930, Md. Nur 24603\n(SING). - Java: Pasir Poetih, 27-8-1961, H. F. Neubauer 1601 (L). - Lesser Sunda Islands: Flores, near Bari, 12-7-1847, Zollinger 3334 (Type of Lemnopsis minor, P, L, BM, U); Bima, together with narrow-leaved Halodule uninervis, Zollinger 3431 (P). - Celebes: Salayer, Zollinger (L). - Moluccas: Amboina., October 1874, O. Beccari 11828 (L).\nMARIANNE ISLANDS. Exact locality unknown, fl., Gaudichaud (Type, P, L). - Guam, Cocos Lagoon, Summer 1952, K. O. Emery 10699 (AHFH). Mana Bay\nopposite Ypan, on east coast, inside barrier reef, in lagoon at ½-2 m depth, on sandy silty bottom, with scattered Enhalus acoroides, Caulerpa, Padina and other algae, 15-4-1962, B. O. Stone 4063 (L, HAW). - Saipan Island, in lagoon, 22-8-1952, N. 0. Bunker & R. Ocampo (AHFH).\nWESTERN AusTRALIA. Carnarvon, Babbage Island, in creek through the mangrove swamp, 16-9-1967, 0. den Hartog 560 (L); on tidal flats and in creeks in front of the mangrove-swamp SW.of Carnarvon, 17-9-1967, 0. den Hartog 581 (L). Port Hedland, in small pools in the intertidal belt, on muddy bottom, 18-9-1967, O. den Hartog 588 (L).\nQUEENSLAND. Thursday Island, a few specimens at low-water mark, 12-11-1967, 0. den Hartog 1016 (L); in very shallow, muddy littoral pools along the NE. side of the island, 12-11-1967, 0. den Hartog 1026 (L). Townsville, between Cape Pallarenda and Shelly Beach, in soft mud at half-tide level, 5-10-1967, 0. den Hartog 743 (L); Townsville, Cleveland Bay, flats of Crocodile Creek, 13-10-1967, 0. den Hartog 797 (L).\nNEW CALEDONIA. Noumea, on the mud-flats, just below low-water mark, er and !j! fl., fr., April 1869, B. Balansa 1525 (P); Anse Vata near Noumea, in 1-2 m of water, 5-2-1960, M. Angot (L, HAW). Ouano (La Foa District), abundant, er fl., 20-1-1961, H. S. McKee 8229 (K). "

2.4 Decipher with Claude

The text in the previous chunk was fed to Claude 3 in an interactive session. The goal was to get a cleaned up file with the data arranged in useful categories. It took a few tries. Here are the prompts. (Along the way, Claude 3 was asked to put the results in a CSV-format table. This table was examined in R to check the progress. These “checking” steps are not shown here.)

  • The file lists the locations of herbarium specimens. These are arranged by country. Can you make a table that has columns for the levels of location, notes on habitat, collector’s name, date of collection, collection number and herbarium?

  • Can you divide the location column into city and location?

  • Here is a hint. The date is usually in the day-month-year format. The collector almost always has a number (sometimes with a letter or two) identifying the collection number. Can you look at the table that you just produced and do some adjustments?

  • If there is a name in the Date column, it probably means that the information that should be in the Specific Location or Habitat has spilled over into the following columns. Also, the Herbarium is usually a single letter, a few letters or several short letter strings separated by commas. Can you look at the table again, still keeping it in a CSV format?

The result is the file loaded in the following chunk.

Experience with this interactive session gives strong hints about how to craft a query that can be used with the API, thereby skipping the interactive phase and making this more a “production” system.

Show the code 
## Create a stub for titles used in the output products.
title <- "*Halophila ovata*"

## Identify the source for use with output products.
source <- "Den Hartog (1970) p 252-253"

## Paste in the results of the interactive Clause 3 session.
data <- read_csv(file = 
    "Country,City/Region,Specific Location,Habitat,Collection Number,Collector,Date,Herbarium
Kenya,Lamu,Mud-flats NW of town,Exposed at low tides locally common,A 21,Mrs. F.M. Isaac,01-07-1965,L
Kenya,Mokowe,Mud-flats,In front of mangrove swamp,A 100,Mrs. F.M. Isaac,30-06-1965,L
Kenya,Gazi,South of Ukunda,On mud-flats uncovered at low spring tides,A 116,Mrs. F.M. Isaac,11-12-1965,L
India,Pamban,-,-,133 B,M.O.P. Iyengar,10-1922,BM
Thailand,Rawi,Satut,In sand exposed at low tide with Halodule uninervis,14035,A.F.G. Kerr,13-01-1928,BM
Hong Kong,Kowloon Bay,-,Sandy bottom at ca. 2 m depth,282,Harland,-,K
Philippines,Luzon,Manila Bay,-,1595,A. Loker,05-1892,'K,C,P'
Philippines,Luzon,Manila Bay,-,4112,E.D. Merrill,04-1905,'P,K,L,SYD'
Philippines,Manila,Malate Beach,-,9899,C.B. Robinson,08-05-1910,'C,P,L,BO'
Philippines,Manila,-,-,1098,E.D. Merrill,03-1912,'C,U,WRSL'
Philippines,Albay Province,Albay Gulf near Lubas Point,Sandy mud with Halimeda below low-tide,16848,M. Doty & G.T. Velasquez,18-05-1958,'L,HAW'
Philippines,Cebu,Mactan Island,With Halophila ovalis and Halodule uninervis,-,Moseley,01-1875,BM
Malayan Peninsula,Singapore,Blakang Mati,-,3780,Ridley,1892,'C,BM,SING'
Malayan Peninsula,Singapore,Ponggol,Sandy mud exposed at low tide,-,Holttum,05-02-1914,SING
Malayan Peninsula,Singapore,Labrador,-,-,Holttum,21-03-1928,SING
Malayan Peninsula,Singapore,Tandjong Behala Kuda Pulau Pawai,Sandy bottom half-tide level with Halodule uninervis,38894,J. Sinclair,14-03-1950,'SING,L'
Malayan Peninsula,Singapore,Pasir Laba,Pure stand in Halophila ovalis bed low eulittoral on mud,3897,H.M. Burkill,27-12-1965,L
Indonesia,Riouw Archipelago,Pulau Penyengat,-,24603,Md. Nur,08-11-1930,SING
Indonesia,Java,Pasir Poetih,-,1601,H.F. Neubauer,27-08-1961,L
Indonesia,Lesser Sunda Islands,Flores near Bari,-,3334 (Type),Zollinger,12-07-1847,'P,L,BM,U'
Indonesia,Lesser Sunda Islands,Bima,With narrow-leaved Halodule uninervis,3431,Zollinger,-,P
Indonesia,Celebes,Salayer,-,-,Zollinger,-,L
Indonesia,Moluccas,Amboina,-,11828,O. Beccari,10-1874,L
Marianne Islands,-,Exact locality unknown,-,- (Type),Gaudichaud,-,'P,L'
Marianne Islands,Guam,Cocos Lagoon,-,10699,K.O. Emery,1952,AHFH
Marianne Islands,Guam,Mana Bay,Inside barrier reef 0.5-2 m sandy-silty bottom,4063,B.C. Stone,15-04-1962,'L,HAW'
Marianne Islands,Saipan Island,In lagoon,-,-,N.C. Bunker & R. Ocampo,22-08-1952,AHFH
Western Australia,Carnarvon,Babbage Island,In creek through mangrove swamp,560,C. den Hartog,16-09-1967,L
Western Australia,Carnarvon,SW of Carnarvon,Tidal flats and creeks in front of mangrove swamp,581,C. den Hartog,17-09-1967,L
Western Australia,Port Hedland,-,Small pools in intertidal belt muddy bottom,588,C. den Hartog,18-09-1967,L
Queensland,Thursday Island,-,Few specimens at low-water mark,1016,C. den Hartog,12-11-1967,L
Queensland,Thursday Island,NE side,Very shallow muddy littoral pools,1026,C. den Hartog,12-11-1967,L
Queensland,Townsville,Cape Pallarenda to Shelly Beach,Soft mud at half-tide level,743,C. den Hartog,05-10-1967,L
Queensland,Townsville,Cleveland Bay Crocodile Creek flats,-,797,C. den Hartog,13-10-1967,L
New Caledonia,Noumea,Mud-flats,Just below low-water mark,1525,B. Balansa,04-1869,P
New Caledonia,Noumea,Anse Vata,In 1-2 m of water,-,M. Angot,05-02-1960,'L,HAW'
New Caledonia,Ouano (La Foa District),-,Abundant,8229,H.S. McKee,20-01-1961,K")

## Add a column to identify the rows.
data$text <- c(1:nrow(data))

## Print a data confirmation table.
gt(data) |>
  tab_caption(caption = md(title)) |>
  tab_style(
    cell_text(v_align="top"),
    locations = cells_body()) |>
  tab_source_note(
    source_note = source)
Halophila ovata
Country City/Region Specific Location Habitat Collection Number Collector Date Herbarium text
Kenya Lamu Mud-flats NW of town Exposed at low tides locally common A 21 Mrs. F.M. Isaac 01-07-1965 L 1
Kenya Mokowe Mud-flats In front of mangrove swamp A 100 Mrs. F.M. Isaac 30-06-1965 L 2
Kenya Gazi South of Ukunda On mud-flats uncovered at low spring tides A 116 Mrs. F.M. Isaac 11-12-1965 L 3
India Pamban - - 133 B M.O.P. Iyengar 10-1922 BM 4
Thailand Rawi Satut In sand exposed at low tide with Halodule uninervis 14035 A.F.G. Kerr 13-01-1928 BM 5
Hong Kong Kowloon Bay - Sandy bottom at ca. 2 m depth 282 Harland - K 6
Philippines Luzon Manila Bay - 1595 A. Loker 05-1892 'K,C,P' 7
Philippines Luzon Manila Bay - 4112 E.D. Merrill 04-1905 'P,K,L,SYD' 8
Philippines Manila Malate Beach - 9899 C.B. Robinson 08-05-1910 'C,P,L,BO' 9
Philippines Manila - - 1098 E.D. Merrill 03-1912 'C,U,WRSL' 10
Philippines Albay Province Albay Gulf near Lubas Point Sandy mud with Halimeda below low-tide 16848 M. Doty & G.T. Velasquez 18-05-1958 'L,HAW' 11
Philippines Cebu Mactan Island With Halophila ovalis and Halodule uninervis - Moseley 01-1875 BM 12
Malayan Peninsula Singapore Blakang Mati - 3780 Ridley 1892 'C,BM,SING' 13
Malayan Peninsula Singapore Ponggol Sandy mud exposed at low tide - Holttum 05-02-1914 SING 14
Malayan Peninsula Singapore Labrador - - Holttum 21-03-1928 SING 15
Malayan Peninsula Singapore Tandjong Behala Kuda Pulau Pawai Sandy bottom half-tide level with Halodule uninervis 38894 J. Sinclair 14-03-1950 'SING,L' 16
Malayan Peninsula Singapore Pasir Laba Pure stand in Halophila ovalis bed low eulittoral on mud 3897 H.M. Burkill 27-12-1965 L 17
Indonesia Riouw Archipelago Pulau Penyengat - 24603 Md. Nur 08-11-1930 SING 18
Indonesia Java Pasir Poetih - 1601 H.F. Neubauer 27-08-1961 L 19
Indonesia Lesser Sunda Islands Flores near Bari - 3334 (Type) Zollinger 12-07-1847 'P,L,BM,U' 20
Indonesia Lesser Sunda Islands Bima With narrow-leaved Halodule uninervis 3431 Zollinger - P 21
Indonesia Celebes Salayer - - Zollinger - L 22
Indonesia Moluccas Amboina - 11828 O. Beccari 10-1874 L 23
Marianne Islands - Exact locality unknown - - (Type) Gaudichaud - 'P,L' 24
Marianne Islands Guam Cocos Lagoon - 10699 K.O. Emery 1952 AHFH 25
Marianne Islands Guam Mana Bay Inside barrier reef 0.5-2 m sandy-silty bottom 4063 B.C. Stone 15-04-1962 'L,HAW' 26
Marianne Islands Saipan Island In lagoon - - N.C. Bunker & R. Ocampo 22-08-1952 AHFH 27
Western Australia Carnarvon Babbage Island In creek through mangrove swamp 560 C. den Hartog 16-09-1967 L 28
Western Australia Carnarvon SW of Carnarvon Tidal flats and creeks in front of mangrove swamp 581 C. den Hartog 17-09-1967 L 29
Western Australia Port Hedland - Small pools in intertidal belt muddy bottom 588 C. den Hartog 18-09-1967 L 30
Queensland Thursday Island - Few specimens at low-water mark 1016 C. den Hartog 12-11-1967 L 31
Queensland Thursday Island NE side Very shallow muddy littoral pools 1026 C. den Hartog 12-11-1967 L 32
Queensland Townsville Cape Pallarenda to Shelly Beach Soft mud at half-tide level 743 C. den Hartog 05-10-1967 L 33
Queensland Townsville Cleveland Bay Crocodile Creek flats - 797 C. den Hartog 13-10-1967 L 34
New Caledonia Noumea Mud-flats Just below low-water mark 1525 B. Balansa 04-1869 P 35
New Caledonia Noumea Anse Vata In 1-2 m of water - M. Angot 05-02-1960 'L,HAW' 36
New Caledonia Ouano (La Foa District) - Abundant 8229 H.S. McKee 20-01-1961 K 37
Den Hartog (1970) p 252-253

2.5 Visualize the Locations

The location data need to be simplified by having just one row for each location. The geographic coordinates need to be added in order to make the map.

This chunk configures the data so that it can be used with a Claude 3 query.

Show the code 
## Extract just the rows needed and simplify the names.
loc <- data |>
  select(text,Country,`City/Region`) |>
  rename(city = `City/Region`)

## Confirm the extraction.
## gt(loc)

## Create the simplified block of data.
out <- with(loc, sprintf(paste(text, Country, city, "\n",
    sep = ', ', collapse = ' ')))

The next chunk uses Claude 3 to do the simplification and add the geographic coordinates.

The result of Claude 3 processing is written to a file. This is because there are periods of high-demand on the LLM. The result is a time-out due to a slow response to the API request. This, in turn, stops the following steps from running. By doing this step as a separate chunk and storing the result, the rest of the data flow works as expected.

This “storage strategy” is likely just a short-term problem.

Show the code 
## Create the request for Claude 3.
request <- "The attached text has three fields in each row. These are the number, country, and city. The locations need to be consolidated so there are no rows with duplicates of country and city. Then, each row needs to have the geographic coordinates of the place as separate lat and lon columns. Please return the result only as a CSV table (with header names) and without any comments. Here is the text: "

## Put the request together with the data.
query <- paste0(request,out)

## Send the request.
response <- claudeR(prompt = list(list(role = "user", 
                                       content = query)), 
                    model = "claude-3-opus-20240229", 
                    max_tokens =2000)

## Read the data table returned by Claude 3.
simple_loc <- read_csv(file=response)

## Store the results
write_csv(simple_loc, file = "temp/claude_response.csv")

The response is processed in the next chunk.

Note that the stored file is used (as explained in the previous chunk).

Show the code 
## Read the stored response
simple_loc <- read_csv(file="temp/claude_response.csv")

## Process the response.
new_title <- paste0(title, "Distribution")

## Print a table to confirm the data.
gt(simple_loc) |>
  tab_caption(caption = md(new_title)) |>
  tab_style(
    cell_text(v_align="top"),
    locations = cells_body()) |>
  tab_source_note(
    source_note = source)
Halophila ovataDistribution
text country city lat lon
1 Kenya Lamu -2.2717 40.9031
2 Kenya Mokowe -2.2781 40.8536
3 Kenya Gazi -4.4227 39.5069
4 India Pamban 9.2667 79.2167
5 Thailand Rawi 6.5000 101.9000
6 Hong Kong Kowloon Bay 22.3226 114.2108
7 Philippines Luzon 16.0000 121.0000
8 Philippines Manila 14.6042 120.9822
9 Philippines Albay Province 13.1775 123.5165
10 Philippines Cebu 10.3157 123.8854
11 Malayan Peninsula Singapore 1.3521 103.8198
12 Indonesia Riouw Archipelago -0.9325 104.4492
13 Indonesia Java -7.6145 110.7122
14 Indonesia Lesser Sunda Islands -8.6500 121.0000
15 Indonesia Celebes -1.4531 120.5213
16 Indonesia Moluccas -3.0000 128.0000
17 Marianne Islands Guam 13.4443 144.7937
18 Marianne Islands Saipan Island 15.1825 145.7516
19 Western Australia Carnarvon -24.8672 113.6611
20 Western Australia Port Hedland -20.3102 118.5878
21 Queensland Thursday Island -10.5789 142.2207
22 Queensland Townsville -19.2589 146.8169
23 New Caledonia Noumea -22.2763 166.4572
24 New Caledonia Ouano (La Foa District) -21.7117 165.8446
Den Hartog (1970) p 252-253

2.6 Plot the distribution

Seeing the distribution of the collection sites is an important step in understanding the species.

Note that having the column names match the requirements in the sitemaps package makes this step straightforward.

The points are not labeled as there isn’t enough room.

Show the code 
## Create a basemap
basemap <- site_google_basemap(datatable = simple_loc)

## Plot the map.
## Note: No labels as there are too many points.
ggmap(basemap) +
  site_points(datatable = simple_loc)

2.7 List the Collectors

The history of specimen collection provides a fascinating view of the people involved in adding to our knowledge of this species and when they did their work.

Show the code 
## Extract the year from the date.
collectors <- data |>
  mutate(year = str_sub(Date, start= -4)) |>
  select(Collector, year)

## Get rid of duplicates (collector + date).
simplified_collectors <- unique(collectors)

## Sort by year.
simplified_collectors <- simplified_collectors |>
  arrange(year)

## Show the results
new_title <- paste0(title," Collectors")

gt(simplified_collectors) |>
  tab_caption(caption = md(new_title)) |>
  tab_style(
    cell_text(v_align="top"),
    locations = cells_body()) |>
  tab_source_note(
    source_note = source)
Halophila ovata Collectors
Collector year
Harland -
Zollinger -
Gaudichaud -
Zollinger 1847
B. Balansa 1869
O. Beccari 1874
Moseley 1875
A. Loker 1892
Ridley 1892
E.D. Merrill 1905
C.B. Robinson 1910
E.D. Merrill 1912
Holttum 1914
M.O.P. Iyengar 1922
A.F.G. Kerr 1928
Holttum 1928
Md. Nur 1930
J. Sinclair 1950
K.O. Emery 1952
N.C. Bunker & R. Ocampo 1952
M. Doty & G.T. Velasquez 1958
M. Angot 1960
H.F. Neubauer 1961
H.S. McKee 1961
B.C. Stone 1962
Mrs. F.M. Isaac 1965
H.M. Burkill 1965
C. den Hartog 1967
Den Hartog (1970) p 252-253

2.8 List the Herbarium Collections

This listing will show where the specimens are stored.

As done earlier, this chunk is run separately and the result is stored in a file. This is the work-around for use when the system is heavily loaded and a response is not received before a time-out.

Show the code 
## Extract the relevant data
herb_data <- data |>
  mutate(year = str_sub(Date, start= -4)) |>
  rename(number = `Collection Number`) |>
  select(Collector, number, year, Herbarium)

## Create the simplified block of data.
out <- with(herb_data, sprintf(paste(
                       Collector,number,year,Herbarium,"\n",
                       sep = ', ',
                       collapse = ' ')))

request <- "The text to process contains columns for the Collector, number, year, and Herbarium. The herbarium column may be a single item or a set of items separated with commas. The goal is to have one row in a table for each item. A set of items means that duplicate collections were placed in different herbaria. This means that additional rows need to be put in the table with each repeating the Collector, number, year and herbarium for each specimen. Please create a table in CSV format and return the result with column headers and without any comments. Here is the text to process: "

## Put the request together with the data.
query <- paste0(request,out)

## Send the request.
response <- claudeR(prompt = list(list(role = "user", 
                                       content = query)), 
                    model = "claude-3-opus-20240229", 
                    max_tokens =3000,
                    api_key = claude_key)

## Read the data table returned by Claude 3.
herb_list <- read_csv(file=response)

## Write table for temporary storage
write_csv(herb_list,file="temp/herbarium_data.csv")

Get a list of herbaria so that the abbreviations can be changed to full herbarium names.

Show the code 
herb <- read_csv(file="herbarium_names.csv")

new_title <- "Herbaria with seagrass specimens"
source2 <- "den Hartog (1970) p 7-8."

## Put out a table for reference
gt(herb)|>
  tab_caption(caption = md(new_title)) |>
  tab_style(
    cell_text(v_align="top"),
    locations = cells_body()) |>
  tab_source_note(
    source_note = source2)
Herbaria with seagrass specimens
Herbarium Name
A Arnold Arboretum, Cambridge, Mass., U.S.A.
AD State Herbarium of South Australia, Adelaide, S.A., Australia
ADW Waite Agricultural Research Institute, Adelaide, S.A., Australia
AHFH Herbarium of the Allan Hancock Foundation, University of Southern California, Los Angeles, Cal., U.S.A.
AMD Hugo de Vries Laboratorium, Amsterdam, the Netherlands
BM British Museum (Natural History), London, England
BO Herbarium Bogoriense, Bogar, Indonesia.
BRI Botanic Museum and Herbarium, Brisbane, Qld., Australia
C Botanisk Museum & Herbarium, Copenhagen, Denmark
CAL Central National Herbarium, Calcutta., India
CANTY Canterbury Museum, Christchurch, New Zealand
CN Laboratoire de Botanique, Faculte des Sciences, Caen, France
FI Herbarium Universitatis Florentinae, Istituto Botanico, Firenze, Italy
G Conservatoire et Jardin botaniques, Geneva, Switzerland
GH Gray Herbarium, Harvard University, Cambridge, Mass., U.S.A.
HAW University of Hawaii, Botany Department, Honolulu, Hawaii, U.S.A.
J Moss Herbarium, University of the Witwatersrand, Johannesburg, South Africa
K Herbarium of the Royal Botanic Gardens, Kew, Richmond, England
L Rijksherbarium, Leyden, the Netherlands
LAE Division of Botany, Department of Forest, Lae, New Guinea.
MEL National Herbarium of Victoria, Royal Botanic Gardens, South Yarra, Vict., Australia
MICH University Herbarium, University of Michigan, Ann Arbor, Mich., U.S.A.
P Museum national d'histoire naturelle, Laboratoire de Phanerogamie, Paris, France
PBVM Laboratoire de Biologie vegetale marine, Paris, France
PERTH PERTH Western Australian Herbarium, Perth, W.A., Australia
PERTH-U University of Western Australia, Department of Botany, Nedlands, W.A., Australia
PRE Botanical Research Institute, National Herbarium, Pretoria, South Africa
S Botanical Department of the Naturhistoriska Riksmuseet, Stockholm, Sweden
SING Herbarium of the Botanic Gardens, Singapore
SYD National Herbarium of New South Wales, Sydney, N.S.W., Australia
TI Faculty of Science, University of Tokyo, Tokyo, Japan
U Botanical Museum & Herbarium, Utrecht, the Netherlands.
UC Herbarium of the University of California, Berkeley, Cal., U.S.A.
US U.S. National Museum, Department of Botany, Washington, D.C., U.S.A.
W Naturhistorisches Museum, Vienna, Austria
WRSL Instytut Botaniczny, Universytetu Wroclawskiego, Wroclaw, Poland
Z Botanischer Garten und Institut fur Systematische Botanik der Universitat, Zurich, Switzerland
den Hartog (1970) p 7-8.

The list is processed in the following chunk.

Here is a list of specimens, organized by Herbarium.

Show the code 
## Read the stored data.
specimens <- read_csv(file="temp/herbarium_data.csv")

## Merge with the herbarium names
complete <- merge(specimens, herb, by="Herbarium")

complete <- complete |>
  group_by(Name) |>
  dplyr::arrange(year) |>
  select(Collector, number, year)

## Show the results
new_title <- paste0(title," Herbarium Collections")

gt(complete, groupname_col = "Name") |>
  tab_caption(caption = md(new_title)) |>
  tab_style(
  style = cell_text(weight = "bold"),
  locations = cells_row_groups()) |>
  tab_style(
    cell_text(v_align="top"),
    locations = cells_body()) |>
  tab_source_note(
    source_note = source)
Halophila ovata Herbarium Collections
Collector number year
Herbarium of the Royal Botanic Gardens, Kew, Richmond, England
Harland 282 -
A. Loker 1595 1892
E.D. Merrill 4112 1905
H.S. McKee 8229 1961
Rijksherbarium, Leyden, the Netherlands
Zollinger - -
Gaudichaud - (Type) -
Zollinger 3334 (Type) 1847
O. Beccari 11828 1874
E.D. Merrill 4112 1905
C.B. Robinson 9899 1910
J. Sinclair 38894 1950
M. Doty & G.T. Velasquez 16848 1958
M. Angot - 1960
H.F. Neubauer 1601 1961
B.C. Stone 4063 1962
Mrs. F.M. Isaac A 21 1965
Mrs. F.M. Isaac A 100 1965
Mrs. F.M. Isaac A 116 1965
H.M. Burkill 3897 1965
C. den Hartog 588 1967
C. den Hartog 1016 1967
C. den Hartog 797 1967
C. den Hartog 560 1967
C. den Hartog 581 1967
C. den Hartog 1026 1967
C. den Hartog 743 1967
Museum national d'histoire naturelle, Laboratoire de Phanerogamie, Paris, France
Zollinger 3431 -
Gaudichaud - (Type) -
Zollinger 3334 (Type) 1847
B. Balansa 1525 1869
A. Loker 1595 1892
E.D. Merrill 4112 1905
C.B. Robinson 9899 1910
British Museum (Natural History), London, England
Zollinger 3334 (Type) 1847
Moseley - 1875
Ridley 3780 1892
M.O.P. Iyengar 133 B 1922
A.F.G. Kerr 14035 1928
Botanical Museum & Herbarium, Utrecht, the Netherlands.
Zollinger 3334 (Type) 1847
E.D. Merrill 1098 1912
Botanisk Museum & Herbarium, Copenhagen, Denmark
Ridley 3780 1892
A. Loker 1595 1892
C.B. Robinson 9899 1910
E.D. Merrill 1098 1912
Herbarium of the Botanic Gardens, Singapore
Ridley 3780 1892
Holttum - 1914
Holttum - 1928
Md. Nur 24603 1930
J. Sinclair 38894 1950
National Herbarium of New South Wales, Sydney, N.S.W., Australia
E.D. Merrill 4112 1905
Herbarium Bogoriense, Bogar, Indonesia.
C.B. Robinson 9899 1910
Instytut Botaniczny, Universytetu Wroclawskiego, Wroclaw, Poland
E.D. Merrill 1098 1912
Herbarium of the Allan Hancock Foundation, University of Southern California, Los Angeles, Cal., U.S.A.
N.C. Bunker & R. Ocampo - 1952
K.O. Emery 10699 1952
University of Hawaii, Botany Department, Honolulu, Hawaii, U.S.A.
M. Doty & G.T. Velasquez 16848 1958
M. Angot - 1960
B.C. Stone 4063 1962
Den Hartog (1970) p 252-253

2.9 Consolidate the Habitat Comments

Even though most of the comments are brief, they may offer some insight into the habitat of the species.

Here, again, the results are read into a file to avoid problems of a busy server.

Show the code 
habitat_comments <- data |>
  dplyr::select(Habitat)

## Create the simplified block of data.
out <- with(habitat_comments, sprintf(paste(
                       Habitat,"\n",
                       sep = ', ',
                       collapse = ' ')))

request <- "Many of the specimens contain comments about habitat of the seagrass species. Can you use just these comments and synthesize some general statements about the habitat? This overview should be arranged as regular sentences and paragraphs. Here is the text to process: "

## Put the request together with the data.
query <- paste0(request,out)


## Send the request.
response <- claudeR(prompt = list(list(role = "user", 
                                       content = query)), 
                    model = "claude-3-opus-20240229", 
                    max_tokens =3000,
                    api_key = claude_key)

writeLines(response, "temp/habitat_response.txt")

Read and print the response.

Show the code 
## Read the data returned by Claude 3.
Habitats <- readLines("temp/habitat_response.txt")

Habitats <- data.frame(Habitats)

new_title <- paste0(title," Habitat Synthesis")

library(gt)
## Print the response in a neat format.
gt(Habitats)|>
  tab_caption(caption = md(new_title))
Halophila ovata Habitat Synthesis
Habitats
Based on the habitat comments provided, the seagrass species can be found in various coastal environments, particularly in intertidal zones and shallow waters. They often grow on sandy or muddy substrates, which are exposed during low tides. These seagrasses are commonly found in front of mangrove swamps, on mud-flats, and in tidal creeks.
The seagrasses can form pure stands or grow in association with other species, such as Halodule uninervis and Halophila ovalis. They are also found in shallow littoral pools with muddy bottoms and can occur at depths ranging from the low-water mark to around 2 meters, often in sandy-silty bottoms inside barrier reefs.
In some locations, the seagrass species are described as locally common or abundant, indicating their significant presence in specific habitats. Overall, these seagrasses appear to be well-adapted to the dynamic conditions of intertidal and shallow coastal environments, where they play important ecological roles.