Before you can estimate how many species are being lost due to human induced global changes, you have to have some idea of how many species there are. The problem is that no one knows exactly how many species currently live on earth.

The most commonly quoted estimate is somewhere between 30 and 50 millions based on Erwin’s (1988, 1997) study of tropical insects. This estimate is controversial and politically charged because the larger your initial estimate, the larger the estimated species loss. You also have to take into account that Erwin himself did not present this as a definitive number, but presented his estimate in an effort to spur further research. Let’s look at how this number was arrived at.


Erwin initially centered his study in Panama. His method was knock down fogging of insects in the Tropical forest canopy. In other words, he selected certain trees and fogged them with insecticide. The dead insects falling out of the tree canopy were caught in plactic tarps and later identified to species.

In his initial study, Erwin sampled from 19 trees of only a single tree species Luehea seemanii . Sampling was carried out over the space of three seasons (early rainy, late rainy, early dry, late dry) in order to estimate seasonal changes in species diversity. [ASIDE: Some organisms subdivide their niche space temporally as well as physically. In Plattsburgh there are three species of mosquito which appear at different times of the year (spring – early summer, summer, late summer – fall). This serves to decrease competition and results in a greater species diversity. ]

Erwin found 955+ species of beetles, excluding weevils in his samples. That’s the hard data, now the estimating begins!

Based upon other evidence (studies from Brazil) he assumed that there should be as many weevils as there are leaf beetles. As a result, he added 206 assumed weevil species to his count (now 1161), and for simplicity rounded the number up to 1200 species of insect per tree species (remember, he only sampled the one tree species).

Now 1 hectare (10,000 square meters) of rich tropical rain forest can contain as many as 245 species of trees, but 40 to 100 species is a more common estimate, so Erwin used 70 tree species per hectare as an average. This is important, because we now have to consider the degree of host specificity in the ecology of insect species. [ASIDE: Host Specificity is defined as a case in which a species is in some way tied to the host tree species and cannot exist without it.] Now things get very interesting, because there is really no data available to allow us to judge the proportion of host specific insects per trophic (feeding) group! Erwin isn’t trying to pull a fast one, he’s simply using his best (educated) guess to fill a hole caused by a total lack of data (as I noted earlier, he’s trying to get people to do research to test his assumptions). His assumptions are listed in the table below.

erwinassume.jpg (56742 bytes)

Now remember, his total number of canopy species in Luehea is an estimate, his "% Host specific" is a guess, so obviously his total number of Host-specific species can be considered shakey at best.

Keeping all this in mind, he arrives at an estimate of roughly 163 host-specific species (on Luehea), or about 13.5% of all his beetle species. That means that the other 1,037 species (86.5%) are transient (possibly moving freely from tree to tree). Now if the number 163 host-specifics is about average per tree species, then with roughly 70 tree species per hectare he estimates a total of 11,410 host-specific species of beetles per hectare (163 host-specifics/tree species X 70 tree species/hectare). Add the 11,140 host-specifics to the 1,037 transient species and we arrive at roughly 12,448 beetle species per hectare in the canopy. Now by most estimates beetles make up about 40% of all insect species, so we can make it into an equation:

40% X (total number of Insect species in canopy) = 12,448.

If we solve this simple equation, we arrive at an estimate of 31,120 species of insect/hectare in the canopy. Erwin then estimates roughly 1 insect species living on the floor of the forest for every 3 species in the canopy, or roughly 10,269 species/hectare of forest floor. Add the canopy number to the floor number and we get 41,389 species of insect per hectare for this Panamanian seasonal forest.

If we use Erwin’s numbers for a world-wide estimate, we will get 162 host-specific beetle species times roughly 50,000 species of tropical trees equals 8,100,000 species of beetle. Remember that the beetles are roughly 40% of all insect species, so we need to add in an estimate of the other 60% which gives us 20,250,000 insect species in the canopy world-wide. We then add the estimate of forest floor living insects (remember, 1/3 of the number in the canopy) for a grand total of 30,000,000 species of insects!

More recently, based on work in Manaus, Brazil and Tambopata, Peru, Erwin has upped his estimate to 50,000,000 species of insects!

Do these numbers make sense? Erwin argues that they do. One of his arguments is that the number of new species he finds, as he continues to sample, shows no evidence of dropping off. This argument is based on the concept of a rarefaction curve (see figure below). Consider the following analogy.

wpe2.jpg (12695 bytes)

You have decided to start a trading card (sports, movie, whatever) collection, so you buy 1 pack of cards (contains 10 cards). In our example we assume that all the cards in the set have the same probability of being in any pack (in other words the manufacturer has not intentionally created rare cards, and that the cards are randomly mixed prior to packaging). When you look through the cards in your pack you find you have 10 different cards, so you buy another pack. This time you get 9 new cards, and a duplicate of 1 from your first pack. You have increased your diversity of cards by 90%. If the total "set" of cards is large (say 1,000 different cards), you will probably find, initially,  that you continue to rapidly increase the number of "new" cards (as compared to duplicates) in your collection as you continue to buy more packs of cards. However, after you have purchased a large number of packs you will eventually find that the number of "new" cards per pack begins to decrease, and eventually buying another pack will yield mostly duplicates and only rarely a "new" card. You have reached the "point of diminishing returns" where the cost of the pack does not justify the cost. Scientific sampling is basically similar. As long as your samples yield a large number of new species then there is a benefit to continued sampling. Once diminishing returns set in continued sampling will not add much to your data base. However, it is important to realize that by the time diminishing returns sets in you will have already collected a very large percentage of all the available species (or cards, as the case may be), especially if we maintain our initial assumption that there are no "rare" items.

Erwin is basically saying that despite having collected a very large number of insects, the number of new species represented by those insects shows no evidence of leveling off (see below), so there is little evidence that he has reached the point of diminishing returns. In other words there are still large numbers of species yet to be discovered, which supports his massive estimate of species diversity.

erwinrare.jpg (66028 bytes)


Needless to say, many scientists dispute Erwin’s estimates. They generally suggest a range of from 5 to 15 millions of species on earth.


Erwin’s is not the only attempt to estimate species diversity. Most others rely on some proportion or assumed relationship to estimate diversity.

Stork and Gaston used the ratio of butterflies to other insects to arrive at their estimate. They started by assuming that the very well studied British insect fauna would be representative of all insect faunas. In Britain there are 67 species of butterfly for every 22,000 insect species. They then estimated that there are between 15,000 and 22,000 species of butterfly worldwide, which would result in between 4.9 and 6.6 million insect species worldwide.

May took the view that the number of all animal species worldwide could be estimated using a simple relationship between body size and number of species (consider the number of species of butterflies compared to that of cats, or some other large vertebrate). Using the equation

S ~ L-x (where S = # of species, L = body length, and x is a factor between 1.5 and 3),

He arrived at an estimate of between 10 and 50 million species worldwide (see graph below).

sizevs#.jpg (84749 bytes)


Before we briefly examine attempts to determine modern extinction rates, we need a frame of reference. We can get that from the fossil record, especially in terms of what are called backround extinction rates – the constant rate of extinction through time. Raup (in our text) estimates that individual species last about 10 million years before going extinct. If we use this estimate, and assume 10 million species on earth today (you can adjust this upwards if you would rather use 30 or 50 millions), then we would expect to lose somewhere between 1 and 10 species per year just due to natural causes. This comes out to between 0.00001% and 0.0001% of our total number of species per year, or 0.001% to .01% per century. We would however expect, under natural conditions that this would be balanced by the evolution of new species, leading to little net loss. For comparison, it is interesting to note that some observers have claimed a current extinction rate of ~ 1% per century which is anywhere from 100 to 1000 times greater than Raup’s estimated backround extinction rate!

Extinction Estimate Using the Species/Area Relationship

Based on classic studies of Island Biogeography (which we will discuss in greater detail later in the semester), the number of species can be directly related to habitat area using the equation

S = cAz Where S = number of species, A = area, and c and z are constants.

The constant c will depend on both the taxon (animal group), and the biogeographic provence in question; and z is a fitted constant which usually falls into the range 0.15 – 0.35.

For example, if we use c = 1 and z =0.25 for a series of islands with areas of 10, 100, 1,000, and 10,000 sq. km., the equation above would predict that these islands would have 2, 3, 6, and 10 species respectively. Note that while the area of the islands increases exponentially, the increase in the number of species is a much more modest arithmetic increase. We can, of course, also use the equation to predict the loss of species due to habitat loss. If we were to apply this to the rain forest habitat, and assume 10 million species (much lower than Erwin’s estimate); we would then predict that the loss of 1% of rain forest area each year would result in a species loss of 0.2% to 0.3% per year, or roughly 20,000 to 30,000 species.

Below you can see two examples of projections using the species/area relationship. Both cite a "worst-case end point" which represents a projected loss of almost all tropical forest. In the Land Bird Projection the term "Pleistocene refugia" refers to estimates of bird species numbers during the Pleistocene when tropical rain forest was severly limited due to the climate during the Ice Age.

plantextinctions.jpg (247935 bytes)

Amazbrdext.jpg (294792 bytes)

One argument against using the species/area relationship to predict extinction rate is that the observed rate of extinction does not appear to be anywhere near as extreme as predicted (see table 7.1 below). The problem with this argument is that it assumes an instantaneous response time between change in area and species number. Most scientists assume that there will be a lag in response time which will be dependent on the life span of the species in question. A population of ten giant tortoises is in effect extinct if all ten are of the same sex, yet depending on the age structure of this small population, they may last 100 years before the last one dies, marking the final extinction of the species.  Such species are considered to be "committed to extinction".

extinctable.jpg (97339 bytes)

While the data clearly illustrates that the rate of extinction is increasing, the graph below appears to indicate a recent major drop-off. This illustrates how a change in definition can cause misleading results. Recently the IUCN (International Union for the Conservation of Nature) has redefined "extinct" as not having definitely been sited in the wild for 50 years. As a result, many species that in all likelihood are extinct will not be classified as such for a number of years.

extinctcurve.jpg (95728 bytes)



Erwin, Terry L., 1988, The Tropical Forest Canopy: The Heart of Biotic Diversity, in, E.O.Wilson, ed., Biodiversity, National Academy Press, Washington, D.C., pp.123-129.

Erwin, Terry L., 1997, Biodiversity at its utmost: Tropical Forest Beetles, in, Reaka-Kudla, M.L., D.E. Wilson, and E.O.Wilson (eds.), Biodiversity II, Joseph Henry Press, Washington, D.C., pp.27-40.

Reid, W.V., and K.R.Miller, 1989, Keeping Options Alive: The Scientific Basis for Conserving Biodiversity.  World Resources Institute, Washington, D.C.

Smith, F.D.M., R.M. May, R. Pellew, T.H. Johnson and K.R. Walter, 1993, How much do we know about the current extinction rate? Trends in Ecology and Evolution, 8:375-378.

Stork, N.E., 1997, Measuring Global Biodiversity and Its Decline, in, Reaka-Kudla, M.L., D.E. Wilson, and E.O.Wilson (eds.), Biodiversity II, Joseph Henry Press, Washington, D.C., pp.41-68.