Changes in final version:

All page numbers are the new page numbers with old page numbers in brackets.

1) p160 (158-159). added paragraph to discussion referencing Vergassola:

It was also found that an olfactory search was most efficient when combined with 
mid range sinuosities. This provides a similar message to that found in Vergassola 
et al. (2007) and suggested in Bell (1991) that having a random element to the move- 
ment can increase the efficiency of the search. In the case of Vergassola et al. (2007) 
the situation examined was one of locating a resource in a highly variable olfactory 
plume (relative to the scale of the organism’s perception) . In the case of this study, 
increasing the number of random moves allowed better orientation to the source 
which is a simple demonstration of the effect of random movement to aid orienta- 
tion toward a resource. 


2) p20 (20). added the value of the mean vector length to the graphs to try to show that it is mostly the scale of the graph which makes it look like one and that infact it is very close to one (r=0.946) as calculated using the circstats package in r.

3) P24 (24). "No analytical prediction can be made about the the linear displacement." Removed from paragraph 3. - Agreed with the comment that Benhamou 2004 does provide some tractability on this issue. 

4) P75 etc "Oviposition" rather than "foraging" - I have tried to search for all occurences of this and replace them. Agreed that it could be confusing. Added a paragraph on p76 to clarify:

Some clarification of the term ”foraging” may be required. Within this work, it refers 
to the general process of an animal moving around seeking resources. In this way 
”ovipositional” behaviour (i.e. searching for sites to lay eggs) represents a sub-set of 
the general behaviour of ”foraging”. Foraging may also be specifically foraging for 
food resources (in the case of butterflies, nectar). Because the simulation framework 
presented here is a general framework for movement, the two terms are often used 
interchangeably. It is hoped that this does not cause undue confusion.

5) p76. clarification of CRW + sensory input. Added the following line to end of paragraph 2:

Thus the CRW is the basic model of movement to which is added the potential for an individual step to be influenced by sensory input from the environment. 

6) p79 (78) "Agent-based" reworded the first paragraph in the box to:

Agent-based simulation or software refers to a specific level of encapsulation within the soft- 
ware architecture. In this case, an insect is represented within the software, referred as an 
agent. This representation can have both attributes (e.g. size, number of eggs) and behaviour 
(e.g. movement behaviour). Object-Oriented (OO) programming (Meyer, 2000) provides a 
natural way to build such a system and is used here. ”Agents” represent a somewhat higher 
level of aggregation than individual ”Objects”. An agent, for example may be composed of 
many ”Objects”. Adopting this strategy allows complex systems of interacting agents to be 
constructed whilst retaining a manageable high-level conceptual organisation.

7) p78. and 81 (80) description of Radius of attraction, removed "parameterisable" from the definition and re-worded the overlap of radii and edge effect description.

8) p83. (82) removed azimuth as a parameter, re-worded to say only has two parameters.

9) p113 (112) removed the first sentance of paragraph 3:

"Java is essentially an interpreted language, which means that new components are easy to include into an application whilst it is running" agreed that this is a bit confusing without more detail so better removed.

10) p122. (121) replaced Iteration with Treatment - agreed that iteration may have a specific meaning had some reservations about this in the first place but left it as iteration because thats whats in the code. however code easy to change compared to submitted thesis so made the change.

11) p126. (125) added a definition of the random number generator used by java:

The random number 
generators used were those provided by the default Java programming language, 
which uses a 48-bit seed, modified using a linear congruential formula (See Knuth, 
1998, Section 3.2.1.). 


12) p158. (132) Murray did not find it surprising that there was an effect introduced by the distance from the boundary. I added a paragraph at the end which describes a follow up experiment :

Subsequently, we performed informal experiments using the concept of simply cre- 
ating random steps that cross a boundary. These demonstrate that, infact, the orien- 
tation of steps which cross the boundary are not uniformly distributed with respect 
to the centre of the patch. The actual distribution shows that more of the steps will 
be headed directly for the centre of the patch. This is an effect of the fact that as 
the start position moves further from the boundary, the range of possible azimuth’s which would cause the step to cross the boundary is reduced, until at a distance 
equal to the step length, there is no chance of the step entering the boundary, even 
if it moves directly towards the centre. This is an underlying bias in the movement 
and thus it is not as surprising as initially thought that there is an effect on the dis- 
tribution of ”eggs”. 

13) p134. (133) replaced the first line of paragraph 2 which had refered to a "functional response" replaced with:

The second experiment explored the effect of movement patterns upon the ratio of 
eggs recieved by the outer versus inner resources.

14) p179. (177) some clarification about the arbitray nature of step length was added:

Here we should like to note that ”step length” is itself an arbitrary measure which 
depends upon the method of measurement. In some cases in the Jones (1977) paper, 
this is the result of measuring ”plant to plant” moves. In terms of calibrating the 
simulation, the end goal is to represent the movement in an abstract manner, using 
parameters which can be estimated from the field. The simulated paths themseleves 
will never be representational of actual flight paths, and so it is assumed that the 
abstraction of ”step length” is valid and useful in this circumstance. 

this is in response to murray's comment about step length being represented as a biologically significant parameter.

15) p180. added paragraph clarifying why a single value of B was used:

Notice that the release boundary distance (B) was constant for these experiments. 
Although as discussed in Chapter 5 the release boundary distance may have an ef- 
fect, it was felt that as long as a consistent boundary distance was set, any bias would 
be consistent accross results. In the event that results were shown to be particularly 
sensitive, it would have been possible to repeat experiments with different values 
for B, but this was not the case. 

16) p187.(185) last line of paragraph 2 was changed as it used to contain the phrase "very random" agreed that this is not very scientific now reads:

If F D > L, a forager having a high 
variance of turning angles (k = 0.5) with multiple eggs may even become ”trapped” 
and lay all its eggs on the same plant. 

17) p188, 189. (186) last part of paragraph 4 was changed to include mention that it might be a demographic response:

 This might be in reality a completely different process, 
perhaps not even behavioural in nature, for example it may be demographic with greater levels of recruitment within a large patch at a much greater scale.


18) p208. (205) changed line in paragraph two about "stabilising the results" hopefully to make it clearer what i meant:

the visual search reducing the differences between the results if compared visually.

19) P 75 - Fig 2.8 changed "blue" as diagram no longer in color

20) 126 fig 3.29 change 'blue" to 'line' in caption because of color.

21) P270 - 278 all bray curtis tables removed "blue"

22) P 101-102 reference to fig 3.19 and red circles - changed to "darker" also made picture greyscale in pdf so as not to confuse and added labels for the parameters.