Our research focuses on the population dynamics of plants and how they are influenced by impacts of natural disturbances and global environmental change. We are particularly interested in the interactive effects of fire, grazing and drought in grasslands and woodlands in southern Australia, and how climate change, fragmentation and shrub encroachment affect ecosystems.

Sunday, 12 January 2014

New analyses (thankfully) confirm a long-held notion in grassland ecology

Kangaroo Grass grassland at Evans St
in Sunbury, Victoria,  with lots of
native forbs. This type of grassland
has always been thought to be associated
with areas that have a long history of
frequent burning and little stock grazing.
One of the most important papers on the ecology and management of temperate native grasslands in southern Australia was published by John Stuwe and Bob Parsons in 1977. It's a great paper for two main reasons.

First, it has a very simple objective. To answer the question: how do long-term management regimes affect the floristics of grasslands. This was a paper way ahead of its times in some respects because it recognised that grassland 'condition' (or 'state' to used the modern jargon) was likely a function of its disturbance regime. In this case, frequently burned areas versus grazed versus 'unmanaged'.

Bob Parsons
Second, despite 35 yrs of research since then, the overarching premise of the paper still stands. There hasn't been a magic new understanding of grasslands, just better resolution of the mechanisms that underpin the reponses observed by Stuwe & Parsons (S&P). For example, my own research has shown that grassland canopy gap size positively affects recruitment success  and survival of many grassland herbs because it allows high light penetration to the soil surface. Gaps can be maintained by frequent burning, hence maintaining species coexistence through ramet survival and maximising infrequent (but potentially very successful) recruitment opportunities.

Much of the important insight that was gained from the S&P paper was based on collecting floristic data from areas with different management history (lists of species and their abundance using 4 x 5 m quadrats) and then using a technique called 'hand sorting' to create a Two-Way Table. From this, they identified 'character species' that typify each management area, as well as community metrics such as species richness. They found that frequently burned areas contained more native species than unmanaged areas, and that daisies were more common in burned areas compared to grazed areas. But, when you eyeball the data, there appears to be a lot of overlap in the species composition of grasslands with different management (both in native and exotic species). When I gave the dataset to my Honours student to look at, they were not convinced that S&P had such clear-cut outcomes as appear in their paper. Could S&P have 'over-emphasised' their data, rendering one of the great papers in Australian grassland ecology misleading????

With modern analytical methods like classification and ordination, there is an opportunity to check whether the basic results of the original study were correctly manipulated and interpreted. We can add rigour to the data analysis rather than relying on subjective sorting as was originally done.

With this in mind, I transcribed the original data (which is both found in the original thesis and in the published paper, and shows the value of archiving ecological data in accessible locations that can be retrieved despite changes in technology). I then used a common data analysis package to examine floristic data (Primer) and undertook both a non-metric multidimensional scaling ordination (NMDS) and a cluster analysis. If S&P were correct - that management history affects the floristic composition of grasslands - then we should see "groupings/clusters" in the resultant outputs that are reflective of the a priori identified management regimes. And, using analyses such as ANOSIM and SIMPER (I won't go into the details), we can see if management causes significant differences in composition, and which species are responsible for the differences.

Lightly grazed grasslands have fewer
native daisies than frequently burned
With a little trepidation, I ran the analysis. Either way, the outcome was always going to be interesting. But I didn't really want to 'prove' S&P got it wrong.

Much to my relief, the ordination and cluster analysis confirm that management history in native grasslands DOES affect grassland composition, and that the differences are due to those identified by S&P - more daisies in frequently burned grasslands, and more weeds in unmanaged and grazed grasslands. Phew! That was somewhat of a relief given the importance this paper has had for a generation of grassland ecologists, many of whom have trained under Bob Parsons.

Here is the main figure I generated showing the outcomes of this analysis.

In this ordination, based on abundance data (so species are weighted according to how much cover they project), we can see that RA (frequently burned railway remnants) are floristically distinct from both RO (unmanaged roadside remnants) and PG (remnants partially grazed by stock). RO and PG remnants do not differ significantly from one another in composition (indicating there is a lot of overlap in the species that occur in these remnant types). The way to read the figure is such: coloured symbols indicate quadrats that were sampled in areas with different management histories. How far apart each symbol is from other symbols indicates how similar the composition of any two quadrats is. Hence, if two symbols are near each other, they are similar in composition while the most dissimilar quadrats will be further apart from each other.

Clearly the analysis lends support to the idea that the composition of grasslands (both species present, and the abundance of those species) is affected by management history.

You can see also which species mostly give rise to these stark differences. The lines you see on the ordination that radiate from the centre of the image, with species names at their end, tell us which species have a high contribution in quadrats. We can see that most of the lines point towards the green symbols (the frequently burned railway remnants); this can be equally read as they are pointing away from the RO and PG remnants. These species are mostly native forbs (including the daisies Leptorhynchos and Helichrysum (now Chrysocephalum)) and we interpret this as suggesting that RA sites are characterised by a suite of native (and some exotic) species that tend not to be found as much in PG and RO sites (i.e. they contribute little there). By contrast, the unmanaged RO sites tend to be characterised more by exotic species such as Holcus and Hypochoeris.

Hence, S&P seem to have got it right; frequently burned roadsides have a composition that is different from other native grasslands, and this is due to management history promoting some species over others. In essence, daisies are promoted in frequently burned areas, but tend to disappear in grazed and unmanaged areas. The concept of 'habitat segregation', first coined by S&P, has been at the heart of conservation management of native grasslands for 35 years and continues to resonate even today. I wonder what other ecological and conservation ideas will stand the test of time.

Here's to S&P.

Further Reading

Stuwe, J. & Parsons, R.F. (1977) Themeda australis grasslands on the Basalt Plains, Victoria: floristics and management effects. Australian Journal of Ecology 2, 467-476.

Morgan, J.W. (1998) Importance of canopy gaps for recruitment of some forbs in Themeda triandra-dominated grasslands in south-eastern Australia. Australian Journal of Botany 46, 609-627.

No comments: