At what point do you stop searching for new species in a defined area?

Some context:
In Western Sydney, there’s a critically endangered vegetation type called Cooks River Clay Plain Scrub Forest, of which there’s less than 1% of its original extent. All the remaining patches are heavily fragmented and surrounded by very urban areas. The biggest remaining patch (~11 ha, so pretty small) is a 2 minute drive from my house, so I’ve started to survey it and record everything I can find in there.

In 1979, a local resident called Tony Price did a similar thing, but for plant species only; he spent 3 years surveying the area (not just to get a species count, but also looking at flowering times, fire ecology, etc. --> I’m only doing the species count). Since then, there have been zero major surveys of the patch, with 1 or 2 surveys in the 41 years since merely confirming the top 10% common/dominant species were still there, and then just assuming everything else had stayed the same since 1979…

So I’m recording everything (all the non-plant stuff will be ‘baseline’ data given it hasn’t been recorded there before), but I’m specifically comparing my plant species count with his to see which species have persisted over time, which have disappeared, which new ones have appeared, etc.

To cut to the chase, at what point do I stop searching for new plant species and assume I’ve found ‘everything’ that is still there. In 1979, Price recorded 467 species; after ~42 hours so far, I’ve found 267:


The figure is slightly dodgy because a) I don’t spend the entire time each survey looking only for plants and b) I don’t walk the exact same route through the reserve each time, but it’s clear that relatively soon, the curve will start to plateau, and it will become harder and harder for me to find new species each trip.

Quite a number of the species Price found will no longer be there (e.g. rare orchids smothered by weeds, garden escapees that appeared once or twice but never established), so I’m assuming the ‘true’ number of current species there is less than 467 (although this is probably slightly counterbalanced as I’ve found a number of new invasive species that weren’t there in 1979). Given I’m at 267 right now, at what point do I pack up shop and say I’ve found essentially everything, or at the very least it would not be feasible to continue to search. The dilemma of course is that if you survey over a long time and don’t find something, there are two possibilities: 1) it genuinely isn’t there or 2) you just haven’t looked in the right spot yet/have overlooked it each time. As long as you continue to not find that species, it is impossible to tell which of these 2 scenarios applies.

This has become a very long and rambling post, so I guess to sum up:

At what point on a species accumulation curve, for a relatively small area, do you stop surveying and decide you have found ‘everything’ within reason?


I guess you can say you’re close to stop only if the species count for the spot is close to the overall species count for region, even if the spot doesn’t contain all kinds of biomes unless you’re totally sure there’s no small patch of something different as it often is. Maybe if there’s less than 200 species left to find it could help creating a plan of which spots you’re more likely to find those. I don’t think you should stop if you still have an interest in it.
Personally I prefer to not stop, now I found 283 plant species right by my house, though if looking more into mosses there will be much more, considering you’re researching something of 11 hectars it sounds like a cool thing to do.


Если бы всё было так просто. Вот обследовали вы какую-то территорию. А потом ситематики часть видов объединили, часть раздробили, а то и вовсе, заявили, что этот вид тут в принципе не встречается, а встречается совсем другой, очень похожий.

1 Like

The 2019 International Congress for Conservation Biology conference in Kuala Lumpur had a plenary talk, Rarefaction and Extrapolation: Standardizing Samples to Make Fair Comparisons of Biodiversity among Multiple Assemblages on nearly this exact subject:

In this talk, two types of standardization methods are reviewed: (1) Sample-size-based rarefaction and extrapolation methods aim to compare diversity estimates for equally-large samples determined by samplers. (2) Coverage-based rarefaction and extrapolation methods aim to compare diversity estimates for equally-complete samples; the sample completeness in this method is measured by sample coverage (the proportion of the total number of individuals that belong to the species detected in the sample), a concept originally developed by Alan Turing and I. J. Good in their cryptographic analysis during World War II. Contrary to intuition, sample coverage for the observed sample, rarefied samples, and extrapolated samples can be accurately estimated by the observed data themselves. These two types of standardization methods allow researchers to efficiently use all available data to make robust and detailed inferences about the sampled assemblages, and also to make objective comparisons among multiple assemblages. Hypothetical and real examples are presented for illustrating the use of the online software iNEXT (iNterpolation/EXTrapolation) to compute and plot seamless rarefaction/extrapolation sampling curves based on several diversity measures.

Look up Anne Chao’s work.


Yeah I’ve started drawing up plans based on which families I’m still missing. I’ve got 30-40 eucalypts I haven’t done yet that I have to wait for flowers/fruit before I can ID them.

I’ll definitely keep going indefinitely, but I’m intending to publish my survey; there’s a bit of an imperative to get the data out given urban development + pollution is continuing around the area.

(direct translation: “And then the systematics combined some of the species, some shattered”)
Definitely something I’ve had to account for, many of the species names he used are now synonymised, have moved into different genera, etc.

Awesome, cheers for this! Looks perfect for my question.


@earthknight I assume you’ve encountered similar situations in your SE Asia fieldwork, where you’ve had to make a decision when to stop surveying an area? Although I’m lucky to be working against a checklist for a small area, which I assume doesn’t exist for many places in the tropics.

For us the situation is a bit different. We aren’t really doing a lot of surveys of over-all biodiversity, generally it’s for specific species and it’s a limitation on time/funds/manpower more than anything else (other than a few times where the reason for a survey is more one of political brinksmanship between NGOs than anything else).

In all honesty, we are still building the biodiversity checklist for the region. There wasn’t one when I got here, so I assembled one from various sources, and found some enormous gaps in it, primarily in arthropods. We’ve been filling those gaps mainly as an aside in our other work as it doesn’t really have too much of an effect on the over all conservation priorities of the region. The species that are driving the conservation work we have a good handle on the numbers, distributions, changes, pressures, etc of.

1 Like

Wow… amazing project!

A difference Tony Price may have encountered during his survey is that over 3 years there may have been different growing conditions from year-to-year (due rainfall, heat spells, cold snaps, other factors) that influenced the growth of different plants in different years.


I’ve often thought about this myself, but I don’t think I’ve explored enough areas in general to feel like I’ve spent enough time in any one place to stop going there.
Perhaps year 2-3 will get a bit more stale when I stop finding new bridges to pull over at or something

I would stop surveying once the curve really starts to flatten out, which has not yet happened.

There will be a long “tail” of rare species you can still find, one by one, but nonetheless the curve should get flatter than this.


Keep looking.
This is technically extinct … but found again!


I concur with @teellbee’s comment: One aspect I don’t see in your data or protocol is allowance for seasonal and annual variation in species presence/absence. I’m not familiar with the ecology of your particular biome, but in those that I’ve worked in and studied (e.g. diverse California and Texas habitats), there is inevitably not only seasonal variability in the presence and detectability of some plant species but also remarkable year-to-year variation due to variance in rainfall patterns, etc. This can be particularly evident among annual species and those adapted to fire regimes, for instance. So how has your effort been distributed over the seasons and through the years so far? How have weather patterns varied during your efforts? These aspects might affect the asymptotic “line to gain” for your species numbers.


Really great points, I’ll certainly give myself a full set of seasons as the bare minimum. At the moment I’ve covered Autumn --> 1/3 of spring. A lot of stuff is just starting to flower now, so hoping to bump up to at least 300 within the next couple of trips.


I’m starting to reach a point where I’m measuring grass seeds to see if they’re 2-4 mm, or 4-6mm long…


Yes, good, you need to do that kind of thing. But another thing to consider is:

The law of diminishing returns tends to mean that at some point you will just lose interest in putting in the many hours, or many days, of searching, research and attempted ID-ing that become necessary in order to find one more new thing.

But if your enthusiasm is still strong, you can keep going for quite a lot longer.


This is awesome! I’m so glad to see these kinds of discussions on the forum.
I hope that iNat will eventually get the kind of funding/money necessary to expand the “Trips” utility to do more polygon-based, abundance-quanitifed, and effort-standardized data recording.

To respond to your actual request before I ramble tangentially, I would only suggest that you ponder a way to “get creative” about changing up your search pattern – for my purposes (see below) it’s important to do the same protocol repeatedly, but if your goal is saturation more than standardization, you might consider ways to change your focus/alternative search strategies…

I have been doing a similar project on macrofungi in California’s redwood forests, but have not yet taken quite the same initiative on a species accumulation curve (complicated of course by the presence of many undescribed taxa).

But I have gotten some nice graphs of diversity over the course of a season.
This is species diversity of macrofungi on ten transects plotted over 16 visits during the winter of 2018-2019.


Some species won’t show up every year, or they are very scarce in some and not others. It’s always worth monitoring if you have the chance to. Especially after different seasonal weather timing, or major events like fires. Sometimes species operate on 2-year or more cycles, so they just simply won’t be there period until the next “emergence year”.


Can you get another botanist in? A different perspective often produces new species in a well-worked area and will give you some idea of how much you have missed.

For the reasons others have mentioned, if you are aiming for a complete plant species list, I think you need to survey for several years. But don’t let that delay your preliminary report if the area is under threat. But if you survey for too many years, it is no longer a snapshot - species will be arriving and disappearing during the period of the survey. In other words, your species accumulation curve will never become horizontal.

I tried to do a similar study with water beetles in the UK. I kept going back to the same water body 6 times a year, 2 hours collecting each time. I decided I would stop when I had gone a full year without adding any new species. But after about 7 years I stopped just out of boredom.


This is a great project! And of course, how long you persist has to be at least partly determined by how long you find the process enjoyable. Others have mentioned some of these factors, but if your goal is to achieve a relatively complete species count you might want to consider:

  • Making observations throughout a full year (or in all seasons). You mention that it’s just 1/3 of the way into spring, so I would expect to see a lot more plants emerging over the next couple of months, plus a wide range of invertebrates associated with them. Right now, 85% of your observations are from September.
  • Making observations in less accessible areas of the reserve. It’s not a big place, but it does seem there are two or three patches of scrub away from the paths that you might be able to investigate (with care).
  • Assuming that annual rainfall is quite variable, making observations across several years to catch those plant species that really only emerge in wetter years.
  • Looking at how your species list compares against Tony Price’s records for specific microenvironments and using that info to determine where the biggest gaps seem to be. This could just mean that those habitats are the ones where species have had most trouble surviving, but it would be worth focusing on those areas to see if you’re able to rediscover some of those species.
  • And then there’s fire, which you said was part of Price’s study. Here in California, it’s pretty common for lost plant populations and occasionally lost species to emerge from the seedbank following wildfires. In between times, they just aren’t present other than as seeds or some other dormant form. Given the urban surroundings of your reserve, I assume there’s a heavy focus on suppressing fire at all costs. In the unfortunate event that fire does affect any part of the reserve, it would definitely be worth resurveying those areas during the next spring to see what comes up.
    Looking forward to seeing your results as they progress!

I think my next step is either get into the creek for algae, or start bringing a ladder for treetop epiphytes :D